umgeher/cowboy
0
Fork 0
mirror of https://github.com/ninenines/cowboy.git synced 2025-07-14 12:20:24 +00:00
Code Issues Wiki Activity
cowboy/src/cowboy_req.erl

1190 lines
40 KiB
Erlang
Raw Normal View History

Add and document the new multipart code The old undocumented API is removed entirely. While a documentation exists for the new API, it will not be considered set in stone until further testing has been performed, and a file upload example has been added. The new API should be a little more efficient than the old API, especially with smaller messages.
2014-02-06 19:36:25 +01:00
%% Copyright (c) 2011-2014, Loïc Hoguin <essen@ninenines.eu>
Make HTTP req accessors call themselves with their default default value Use lists:keyfind/3 in qs_val/3, binding/3 and header/3 and make qs_val/2, binding/2 and header/2 call those.
2011-03-29 13:49:48 +02:00
%% Copyright (c) 2011, Anthony Ramine <nox@dev-extend.eu>
Initial commit.
2011-03-07 22:59:22 +01:00
%%
%% Permission to use, copy, modify, and/or distribute this software for any
%% purpose with or without fee is hereby granted, provided that the above
%% copyright notice and this permission notice appear in all copies.
%%
%% THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
%% WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
%% MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
%% ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
%% WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
%% ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
%% OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
Rename cowboy_http_req to cowboy_req
2012-08-27 13:28:57 +02:00
-module(cowboy_req).
Move the reply function to cowboy_http_req.
2011-03-18 22:38:26 +01:00
Have only one -export and -export_type per line This should make easier spotting additions and removals in commits.
2012-08-27 12:16:07 +02:00
%% Request API.
Remove cowboy_req:fragment/1 Clients do not send it. We skip the value if we receive it now, as it shouldn't happen, and won't for all the mainstream clients.
2013-05-15 15:17:33 +02:00
-export([new/14]).
Have only one -export and -export_type per line This should make easier spotting additions and removals in commits.
2012-08-27 12:16:07 +02:00
-export([method/1]).
-export([version/1]).
-export([peer/1]).
-export([host/1]).
-export([host_info/1]).
-export([port/1]).
-export([path/1]).
-export([path_info/1]).
Rename cowboy_req:raw_qs/1 to cowboy_req:qs/1
2012-09-17 11:46:45 +02:00
-export([qs/1]).
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
-export([parse_qs/1]).
-export([match_qs/2]).
Introduce cowboy_req:host_url/1 to remove more duplicate code
2012-09-16 01:13:44 +02:00
-export([host_url/1]).
Add cowboy_req:url/1 to return the full request URL Use it in cowboy_websocket for hixie76, replacing http by ws.
2012-09-15 22:03:00 +02:00
-export([url/1]).
Have only one -export and -export_type per line This should make easier spotting additions and removals in commits.
2012-08-27 12:16:07 +02:00
-export([binding/2]).
-export([binding/3]).
-export([bindings/1]).
-export([header/2]).
-export([header/3]).
-export([headers/1]).
-export([parse_header/2]).
-export([parse_header/3]).
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
-export([parse_cookies/1]).
-export([match_cookies/2]).
Have only one -export and -export_type per line This should make easier spotting additions and removals in commits.
2012-08-27 12:16:07 +02:00
-export([meta/2]).
-export([meta/3]).
Introduce cowboy_req:set_meta/3 to set request metadata
2012-09-15 22:51:37 +02:00
-export([set_meta/3]).
Have only one -export and -export_type per line This should make easier spotting additions and removals in commits.
2012-08-27 12:16:07 +02:00
%% Request body API.
-export([has_body/1]).
-export([body_length/1]).
Initial commit with connection/streams Breaking changes with previous commit. This is a very large change, and I am giving up on making a single commit that fixes everything. More commits will follow slowly adding back features, introducing new tests and fixing the documentation. This change contains most of the work toward unifying the interface for handling both HTTP/1.1 and HTTP/2. HTTP/1.1 connections are now no longer 1 process per connection; instead by default 1 process per request is also created. This has a number of pros and cons. Because it has cons, we also allow users to use a lower-level API that acts on "streams" (requests/responses) directly at the connection process-level. If performance is a concern, one can always write a stream handler. The performance in this case will be even greater than with Cowboy 1, although all the special handlers are unavailable. When switching to Websocket, after the handler returns from init/2, Cowboy stops the stream and the Websocket protocol takes over the connection process. Websocket then calls websocket_init/2 for any additional initialization such as timers, because the process is different in init/2 and websocket_*/* functions. This however would allow us to use websocket_init/2 for sending messages on connect, instead of sending ourselves a message and be subject to races. Note that websocket_init/2 is optional. This is all a big change and while most of the tests pass, some functionality currently doesn't. SPDY is broken and will be removed soon in favor of HTTP/2. Automatic compression is currently disabled. The cowboy_req interface probably still have a few functions that need to be updated. The docs and examples do not refer the current functionality anymore. Everything will be fixed over time. Feedback is more than welcome. Open a ticket!
2016-02-10 17:28:32 +01:00
-export([read_body/1]).
-export([read_body/2]).
Have only one -export and -export_type per line This should make easier spotting additions and removals in commits.
2012-08-27 12:16:07 +02:00
-export([body/1]).
Check the length before reading the body in body/1 and body_qs/1
2013-03-06 08:50:45 -05:00
-export([body/2]).
Have only one -export and -export_type per line This should make easier spotting additions and removals in commits.
2012-08-27 12:16:07 +02:00
-export([body_qs/1]).
Check the length before reading the body in body/1 and body_qs/1
2013-03-06 08:50:45 -05:00
-export([body_qs/2]).
Add and document the new multipart code The old undocumented API is removed entirely. While a documentation exists for the new API, it will not be considered set in stone until further testing has been performed, and a file upload example has been added. The new API should be a little more efficient than the old API, especially with smaller messages.
2014-02-06 19:36:25 +01:00
%% Multipart API.
-export([part/1]).
Add cowboy_req:part/2 export
2014-08-12 00:51:46 -07:00
-export([part/2]).
Add and document the new multipart code The old undocumented API is removed entirely. While a documentation exists for the new API, it will not be considered set in stone until further testing has been performed, and a file upload example has been added. The new API should be a little more efficient than the old API, especially with smaller messages.
2014-02-06 19:36:25 +01:00
-export([part_body/1]).
-export([part_body/2]).
Have only one -export and -export_type per line This should make easier spotting additions and removals in commits.
2012-08-27 12:16:07 +02:00
%% Response API.
-export([set_resp_cookie/4]).
-export([set_resp_header/3]).
-export([set_resp_body/2]).
Add cowboy_req:set_resp_body_fun/2 This allows streaming a body without knowing the length in advance. Also allows {stream, StreamFun} response body in the REST code.
2013-01-05 23:35:30 +01:00
-export([set_resp_body_fun/2]).
Have only one -export and -export_type per line This should make easier spotting additions and removals in commits.
2012-08-27 12:16:07 +02:00
-export([set_resp_body_fun/3]).
-export([has_resp_header/2]).
-export([has_resp_body/1]).
Add cowboy_req:delete_resp_header/2 Both requested and better for making Req opaque.
2012-09-16 01:55:40 +02:00
-export([delete_resp_header/2]).
Have only one -export and -export_type per line This should make easier spotting additions and removals in commits.
2012-08-27 12:16:07 +02:00
-export([reply/2]).
-export([reply/3]).
-export([reply/4]).
Initial commit with connection/streams Breaking changes with previous commit. This is a very large change, and I am giving up on making a single commit that fixes everything. More commits will follow slowly adding back features, introducing new tests and fixing the documentation. This change contains most of the work toward unifying the interface for handling both HTTP/1.1 and HTTP/2. HTTP/1.1 connections are now no longer 1 process per connection; instead by default 1 process per request is also created. This has a number of pros and cons. Because it has cons, we also allow users to use a lower-level API that acts on "streams" (requests/responses) directly at the connection process-level. If performance is a concern, one can always write a stream handler. The performance in this case will be even greater than with Cowboy 1, although all the special handlers are unavailable. When switching to Websocket, after the handler returns from init/2, Cowboy stops the stream and the Websocket protocol takes over the connection process. Websocket then calls websocket_init/2 for any additional initialization such as timers, because the process is different in init/2 and websocket_*/* functions. This however would allow us to use websocket_init/2 for sending messages on connect, instead of sending ourselves a message and be subject to races. Note that websocket_init/2 is optional. This is all a big change and while most of the tests pass, some functionality currently doesn't. SPDY is broken and will be removed soon in favor of HTTP/2. Automatic compression is currently disabled. The cowboy_req interface probably still have a few functions that need to be updated. The docs and examples do not refer the current functionality anymore. Everything will be fixed over time. Feedback is more than welcome. Open a ticket!
2016-02-10 17:28:32 +01:00
-export([send_body/3]).
Have only one -export and -export_type per line This should make easier spotting additions and removals in commits.
2012-08-27 12:16:07 +02:00
-export([chunked_reply/2]).
-export([chunked_reply/3]).
-export([chunk/2]).
Add request body reading options The options were added to allow developers to fix timeout issues when reading large bodies. It is also a cleaner and easier to extend interface. This commit deprecates the functions init_stream, stream_body and skip_body which are no longer needed. They will be removed in 1.0. The body function can now take an additional argument that is a list of options. The body_qs, part and part_body functions can too and simply pass this argument down to the body call. There are options for disabling the automatic continue reply, setting a maximum length to be returned (soft limit), setting the read length and read timeout, and setting the transfer and content decode functions. The return value of the body and body_qs have changed slightly. The body function now works similarly to the part_body function, in that it returns either an ok or a more tuple depending on whether there is additional data to be read. The body_qs function can return a badlength tuple if the body is too big. The default size has been increased from 16KB to 64KB. The default read length and timeout have been tweaked and vary depending on the function called. The body function will now adequately process chunked bodies, which means that the body_qs function will too. But this means that the behavior has changed slightly and your code should be tested properly when updating your code. The body and body_qs still accept a length as first argument for compatibility purpose with older code. Note that this form is deprecated and will be removed in 1.0. The part and part_body function, being new and never having been in a release yet, have this form completely removed in this commit. Again, while most code should work as-is, you should make sure that it actually does before pushing this to production.
2014-06-02 23:09:43 +02:00
-export([continue/1]).
Simpler code for sending errors following crashes
2013-08-24 11:20:14 +02:00
-export([maybe_reply/2]).
Remove duplicate code for ensure_response
2012-09-15 22:19:39 +02:00
-export([ensure_response/2]).
Have only one -export and -export_type per line This should make easier spotting additions and removals in commits.
2012-08-27 12:16:07 +02:00
Add the private function set_host/4 used by cowboy_protocol
2012-09-16 16:04:16 +02:00
%% Private setter/getter API.
Better handle socket closing with loop handlers We now read from the socket to be able to detect errors or TCP close events, and buffer the data if any. Once the data receive goes over a certain limit, which defaults to 5000 bytes, we simply close the connection with an {error, overflow} reason.
2013-02-11 09:03:13 +01:00
-export([append_buffer/2]).
Add cowboy_req:get/2 and :set/2 private functions They should not be used unless you have a very special need, which generally involves interacting with the internals of Cowboy.
2012-09-29 11:08:59 +02:00
-export([get/2]).
-export([set/2]).
Add the private set_bindings/4 function used by cowboy_protocol
2012-09-17 00:39:29 +02:00
-export([set_bindings/4]).
Add a function to lock request responses instead of inlining Just more cleanup in order to have req() opaque.
2012-09-16 13:57:27 +02:00
-export([lock/1]).
Add a cowboy_req:to_list/1 function Removes a lot of duplicate code when printing error messages.
2012-09-15 20:33:57 +02:00
-export([to_list/1]).
Add function cowboy_http_req:compact/1 to regain memory. It removes all the non-essential data from the HTTP request record. It allows some applications to make better use of their memory, for example websockets which do not need to keep all the headers information and can simply discard it using this function.
2011-06-01 12:49:03 +02:00
Move cookie parsing/building code to cowlib The code for parsing has also been rewritten to be more efficient and to be able to handle cookie values with space inside them properly. Update cowlib to 0.2.0.
2013-10-23 11:21:31 +02:00
-type cookie_opts() :: cow_cookie:cookie_opts().
Make cookies use universal time instead of local time Includes: * cowboy_clock:rfc2109/1 now expects UTC datetime * Rewrite of the cookie code to cowboy_http * Removal of cowboy_cookies * Add type cowboy_req:cookie_opts/0 Cookies should now be set using cowboy_req:set_resp_cookie/3. Code calling cowboy_cookies directly will need to be updated.
2012-12-07 14:54:45 +01:00
-export_type([cookie_opts/0]).
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
-type content_decode_fun() :: fun((binary()) -> binary()).
Improve the specs for cowboy_req:init_stream/4
2013-05-16 15:54:30 +02:00
-type transfer_decode_fun() :: fun((binary(), any())
Update cowlib to 0.6.1
2014-03-27 11:39:09 +01:00
-> cow_http_te:decode_ret()).
Improve the specs for cowboy_req:init_stream/4
2013-05-16 15:54:30 +02:00
Initial commit with connection/streams Breaking changes with previous commit. This is a very large change, and I am giving up on making a single commit that fixes everything. More commits will follow slowly adding back features, introducing new tests and fixing the documentation. This change contains most of the work toward unifying the interface for handling both HTTP/1.1 and HTTP/2. HTTP/1.1 connections are now no longer 1 process per connection; instead by default 1 process per request is also created. This has a number of pros and cons. Because it has cons, we also allow users to use a lower-level API that acts on "streams" (requests/responses) directly at the connection process-level. If performance is a concern, one can always write a stream handler. The performance in this case will be even greater than with Cowboy 1, although all the special handlers are unavailable. When switching to Websocket, after the handler returns from init/2, Cowboy stops the stream and the Websocket protocol takes over the connection process. Websocket then calls websocket_init/2 for any additional initialization such as timers, because the process is different in init/2 and websocket_*/* functions. This however would allow us to use websocket_init/2 for sending messages on connect, instead of sending ourselves a message and be subject to races. Note that websocket_init/2 is optional. This is all a big change and while most of the tests pass, some functionality currently doesn't. SPDY is broken and will be removed soon in favor of HTTP/2. Automatic compression is currently disabled. The cowboy_req interface probably still have a few functions that need to be updated. The docs and examples do not refer the current functionality anymore. Everything will be fixed over time. Feedback is more than welcome. Open a ticket!
2016-02-10 17:28:32 +01:00
-type body_opts() :: [{continue, boolean()} %% doesn't apply
Add request body reading options The options were added to allow developers to fix timeout issues when reading large bodies. It is also a cleaner and easier to extend interface. This commit deprecates the functions init_stream, stream_body and skip_body which are no longer needed. They will be removed in 1.0. The body function can now take an additional argument that is a list of options. The body_qs, part and part_body functions can too and simply pass this argument down to the body call. There are options for disabling the automatic continue reply, setting a maximum length to be returned (soft limit), setting the read length and read timeout, and setting the transfer and content decode functions. The return value of the body and body_qs have changed slightly. The body function now works similarly to the part_body function, in that it returns either an ok or a more tuple depending on whether there is additional data to be read. The body_qs function can return a badlength tuple if the body is too big. The default size has been increased from 16KB to 64KB. The default read length and timeout have been tweaked and vary depending on the function called. The body function will now adequately process chunked bodies, which means that the body_qs function will too. But this means that the behavior has changed slightly and your code should be tested properly when updating your code. The body and body_qs still accept a length as first argument for compatibility purpose with older code. Note that this form is deprecated and will be removed in 1.0. The part and part_body function, being new and never having been in a release yet, have this form completely removed in this commit. Again, while most code should work as-is, you should make sure that it actually does before pushing this to production.
2014-06-02 23:09:43 +02:00
| {length, non_neg_integer()}
Initial commit with connection/streams Breaking changes with previous commit. This is a very large change, and I am giving up on making a single commit that fixes everything. More commits will follow slowly adding back features, introducing new tests and fixing the documentation. This change contains most of the work toward unifying the interface for handling both HTTP/1.1 and HTTP/2. HTTP/1.1 connections are now no longer 1 process per connection; instead by default 1 process per request is also created. This has a number of pros and cons. Because it has cons, we also allow users to use a lower-level API that acts on "streams" (requests/responses) directly at the connection process-level. If performance is a concern, one can always write a stream handler. The performance in this case will be even greater than with Cowboy 1, although all the special handlers are unavailable. When switching to Websocket, after the handler returns from init/2, Cowboy stops the stream and the Websocket protocol takes over the connection process. Websocket then calls websocket_init/2 for any additional initialization such as timers, because the process is different in init/2 and websocket_*/* functions. This however would allow us to use websocket_init/2 for sending messages on connect, instead of sending ourselves a message and be subject to races. Note that websocket_init/2 is optional. This is all a big change and while most of the tests pass, some functionality currently doesn't. SPDY is broken and will be removed soon in favor of HTTP/2. Automatic compression is currently disabled. The cowboy_req interface probably still have a few functions that need to be updated. The docs and examples do not refer the current functionality anymore. Everything will be fixed over time. Feedback is more than welcome. Open a ticket!
2016-02-10 17:28:32 +01:00
| {read_length, non_neg_integer()} %% to be added back later as optimization
| {read_timeout, timeout()} %% same
| {transfer_decode, transfer_decode_fun(), any()} %% doesn't apply
| {content_decode, content_decode_fun()}]. %% does apply
Add request body reading options The options were added to allow developers to fix timeout issues when reading large bodies. It is also a cleaner and easier to extend interface. This commit deprecates the functions init_stream, stream_body and skip_body which are no longer needed. They will be removed in 1.0. The body function can now take an additional argument that is a list of options. The body_qs, part and part_body functions can too and simply pass this argument down to the body call. There are options for disabling the automatic continue reply, setting a maximum length to be returned (soft limit), setting the read length and read timeout, and setting the transfer and content decode functions. The return value of the body and body_qs have changed slightly. The body function now works similarly to the part_body function, in that it returns either an ok or a more tuple depending on whether there is additional data to be read. The body_qs function can return a badlength tuple if the body is too big. The default size has been increased from 16KB to 64KB. The default read length and timeout have been tweaked and vary depending on the function called. The body function will now adequately process chunked bodies, which means that the body_qs function will too. But this means that the behavior has changed slightly and your code should be tested properly when updating your code. The body and body_qs still accept a length as first argument for compatibility purpose with older code. Note that this form is deprecated and will be removed in 1.0. The part and part_body function, being new and never having been in a release yet, have this form completely removed in this commit. Again, while most code should work as-is, you should make sure that it actually does before pushing this to production.
2014-06-02 23:09:43 +02:00
-export_type([body_opts/0]).
Add experimental and incomplete SPDY support The SPDY connection processes are also supervisors. Missing: * sendfile support * request body reading support
2013-05-30 20:21:01 +02:00
-type resp_body_fun() :: fun((any(), module()) -> ok).
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
-type send_chunk_fun() :: fun((iodata()) -> ok).
Add chunked response body fun Adds a new type of streaming response fun. It can be set in a similar way to a streaming body fun with known length: Req2 = cowboy_req:set_resp_body_fun(chunked, StreamFun, Req) The fun, StreamFun, should accept a fun as its single argument. This fun, ChunkFun, is used to send chunks of iodata: ok = ChunkFun(IoData) ChunkFun should not be called with an empty binary or iolist as this will cause HTTP 1.1 clients to believe the stream is over. The final (0 length) chunk will be sent automatically - even if it has already been sent - assuming no exception is raised. Also note that the connection will close after the last chunk for HTTP 1.0 clients.
2013-02-18 21:20:36 +00:00
-type resp_chunked_fun() :: fun((send_chunk_fun()) -> ok).
Use a single Transport:send/2 call to send the reply if possible This gives a huge boost in performance when replies are small.
2012-09-26 10:08:43 +02:00
Make cowboy_req:req() opaque Remove http.hrl in the process. If you feel furious about this change after reaching this commit, don't worry! You probably just need a few access or modification functions to be added in cowboy_req for your needs. Problem is, I don't know these. I'll be very thankful if you can open a ticket and tell me all about it so the best solution can be swiftly implemented to make you happy again.
2012-09-17 01:11:44 +02:00
-record(http_req, {
%% Transport.
Add experimental and incomplete SPDY support The SPDY connection processes are also supervisors. Missing: * sendfile support * request body reading support
2013-05-30 20:21:01 +02:00
socket = undefined :: any(),
Make cowboy_req:req() opaque Remove http.hrl in the process. If you feel furious about this change after reaching this commit, don't worry! You probably just need a few access or modification functions to be added in cowboy_req for your needs. Problem is, I don't know these. I'll be very thankful if you can open a ticket and tell me all about it so the best solution can be swiftly implemented to make you happy again.
2012-09-17 01:11:44 +02:00
transport = undefined :: undefined | module(),
connection = keepalive :: keepalive | close,
%% Request.
pid = undefined :: pid(),
Don't use decode_packet/3 for parsing the request-line First step in making all methods and header names binaries to get rid of many inconsistencies caused by decode_packet/3. Methods are all binary now. Note that since they are case sensitive, the usual methods become <<"GET">>, <<"POST">> and so on.
2012-09-20 06:22:51 +02:00
method = <<"GET">> :: binary(),
Move cowboy_http:version() to cowboy:http_version()
2013-05-16 16:06:52 +02:00
version = 'HTTP/1.1' :: cowboy:http_version(),
Make cowboy_req:req() opaque Remove http.hrl in the process. If you feel furious about this change after reaching this commit, don't worry! You probably just need a few access or modification functions to be added in cowboy_req for your needs. Problem is, I don't know these. I'll be very thankful if you can open a ticket and tell me all about it so the best solution can be swiftly implemented to make you happy again.
2012-09-17 01:11:44 +02:00
peer = undefined :: undefined | {inet:ip_address(), inet:port_number()},
host = undefined :: undefined | binary(),
Remove cowboy_dispatcher Types and code are moved to cowboy_router. The match/3 export from cowboy_dispatcher isn't available anymore as it is called internally.
2013-01-22 18:25:16 +01:00
host_info = undefined :: undefined | cowboy_router:tokens(),
Make cowboy_req:req() opaque Remove http.hrl in the process. If you feel furious about this change after reaching this commit, don't worry! You probably just need a few access or modification functions to be added in cowboy_req for your needs. Problem is, I don't know these. I'll be very thankful if you can open a ticket and tell me all about it so the best solution can be swiftly implemented to make you happy again.
2012-09-17 01:11:44 +02:00
port = undefined :: undefined | inet:port_number(),
path = undefined :: binary(),
Remove cowboy_dispatcher Types and code are moved to cowboy_router. The match/3 export from cowboy_dispatcher isn't available anymore as it is called internally.
2013-01-22 18:25:16 +01:00
path_info = undefined :: undefined | cowboy_router:tokens(),
Rename cowboy_req:raw_qs/1 to cowboy_req:qs/1
2012-09-17 11:46:45 +02:00
qs = undefined :: binary(),
Remove cowboy_dispatcher Types and code are moved to cowboy_router. The match/3 export from cowboy_dispatcher isn't available anymore as it is called internally.
2013-01-22 18:25:16 +01:00
bindings = undefined :: undefined | cowboy_router:bindings(),
Move cowboy_http:headers() to cowboy:http_headers()
2013-05-16 16:16:32 +02:00
headers = [] :: cowboy:http_headers(),
Make cowboy_req:req() opaque Remove http.hrl in the process. If you feel furious about this change after reaching this commit, don't worry! You probably just need a few access or modification functions to be added in cowboy_req for your needs. Problem is, I don't know these. I'll be very thankful if you can open a ticket and tell me all about it so the best solution can be swiftly implemented to make you happy again.
2012-09-17 01:11:44 +02:00
meta = [] :: [{atom(), any()}],
%% Request body.
Improve the specs for cowboy_req:init_stream/4
2013-05-16 15:54:30 +02:00
body_state = waiting :: waiting | done | {stream, non_neg_integer(),
transfer_decode_fun(), any(), content_decode_fun()},
Make cowboy_req:req() opaque Remove http.hrl in the process. If you feel furious about this change after reaching this commit, don't worry! You probably just need a few access or modification functions to be added in cowboy_req for your needs. Problem is, I don't know these. I'll be very thankful if you can open a ticket and tell me all about it so the best solution can be swiftly implemented to make you happy again.
2012-09-17 01:11:44 +02:00
buffer = <<>> :: binary(),
Add and document the new multipart code The old undocumented API is removed entirely. While a documentation exists for the new API, it will not be considered set in stone until further testing has been performed, and a file upload example has been added. The new API should be a little more efficient than the old API, especially with smaller messages.
2014-02-06 19:36:25 +01:00
multipart = undefined :: undefined | {binary(), binary()},
Make cowboy_req:req() opaque Remove http.hrl in the process. If you feel furious about this change after reaching this commit, don't worry! You probably just need a few access or modification functions to be added in cowboy_req for your needs. Problem is, I don't know these. I'll be very thankful if you can open a ticket and tell me all about it so the best solution can be swiftly implemented to make you happy again.
2012-09-17 01:11:44 +02:00
%% Response.
Add optional automatic response body compression This behavior can be enabled with the `compress` protocol option. See the `compress_response` example for more details. All tests are now ran with and without compression for both HTTP and HTTPS.
2013-01-07 22:42:16 +01:00
resp_compress = false :: boolean(),
Add a workaround to disable chunked transfer-encoding This is an undocumented workaround to disable chunks when using HTTP/1.1. It can be used when the client advertises itself as HTTP/1.1 despite not understanding the chunked transfer-encoding. Usage can be found looking at the test for it. When activated, Cowboy will still advertise itself as HTTP/1.1, but will send the body the same way it would if it was HTTP/1.0.
2013-06-26 18:04:20 +01:00
resp_state = waiting :: locked | waiting | waiting_stream
| chunks | stream | done,
Move cowboy_http:headers() to cowboy:http_headers()
2013-05-16 16:16:32 +02:00
resp_headers = [] :: cowboy:http_headers(),
Add cowboy_req:set_resp_body_fun/2 This allows streaming a body without knowing the length in advance. Also allows {stream, StreamFun} response body in the REST code.
2013-01-05 23:35:30 +01:00
resp_body = <<>> :: iodata() | resp_body_fun()
Add chunked response body fun Adds a new type of streaming response fun. It can be set in a similar way to a streaming body fun with known length: Req2 = cowboy_req:set_resp_body_fun(chunked, StreamFun, Req) The fun, StreamFun, should accept a fun as its single argument. This fun, ChunkFun, is used to send chunks of iodata: ok = ChunkFun(IoData) ChunkFun should not be called with an empty binary or iolist as this will cause HTTP 1.1 clients to believe the stream is over. The final (0 length) chunk will be sent automatically - even if it has already been sent - assuming no exception is raised. Also note that the connection will close after the last chunk for HTTP 1.0 clients.
2013-02-18 21:20:36 +00:00
| {non_neg_integer(), resp_body_fun()}
| {chunked, resp_chunked_fun()},
Make cowboy_req:req() opaque Remove http.hrl in the process. If you feel furious about this change after reaching this commit, don't worry! You probably just need a few access or modification functions to be added in cowboy_req for your needs. Problem is, I don't know these. I'll be very thankful if you can open a ticket and tell me all about it so the best solution can be swiftly implemented to make you happy again.
2012-09-17 01:11:44 +02:00
%% Functions.
Fix an issue for replies in onresponse mishandling headers Reported and fixed over email by Adrian Roe.
2013-01-20 18:57:42 +01:00
onresponse = undefined :: undefined | already_called
Move cowboy_protocol:onresponse_fun() to cowboy:onresponse_fun()
2013-05-16 17:01:38 +02:00
| cowboy:onresponse_fun()
Make cowboy_req:req() opaque Remove http.hrl in the process. If you feel furious about this change after reaching this commit, don't worry! You probably just need a few access or modification functions to be added in cowboy_req for your needs. Problem is, I don't know these. I'll be very thankful if you can open a ticket and tell me all about it so the best solution can be swiftly implemented to make you happy again.
2012-09-17 01:11:44 +02:00
}).
-opaque req() :: #http_req{}.
Introduce a cowboy_req:req/0 type Cleanup the typespecs of the cowboy_req module following this.
2012-08-27 14:27:41 +02:00
-export_type([req/0]).
Move the reply function to cowboy_http_req.
2011-03-18 22:38:26 +01:00
%% Request API.
Initial commit.
2011-03-07 22:59:22 +01:00
Add experimental and incomplete SPDY support The SPDY connection processes are also supervisors. Missing: * sendfile support * request body reading support
2013-05-30 20:21:01 +02:00
-spec new(any(), module(),
Fix crashes when creating the request We now obtain the peer address before creating the Req object. If an error occurs, then something went wrong, we close the connection nicely directly.
2013-02-06 23:32:31 +01:00
undefined | {inet:ip_address(), inet:port_number()},
Remove cowboy_req:fragment/1 Clients do not send it. We skip the value if we receive it now, as it shouldn't happen, and won't for all the mainstream clients.
2013-05-15 15:17:33 +02:00
binary(), binary(), binary(),
Move cowboy_http:headers() to cowboy:http_headers()
2013-05-16 16:16:32 +02:00
cowboy:http_version(), cowboy:http_headers(), binary(),
Add optional automatic response body compression This behavior can be enabled with the `compress` protocol option. See the `compress_response` example for more details. All tests are now ran with and without compression for both HTTP and HTTPS.
2013-01-07 22:42:16 +01:00
inet:port_number() | undefined, binary(), boolean(), boolean(),
Move cowboy_protocol:onresponse_fun() to cowboy:onresponse_fun()
2013-05-16 17:01:38 +02:00
undefined | cowboy:onresponse_fun())
Add a private cowboy_req function to create a Req object Private means you must not used it. It's meant for internal use.
2012-09-16 15:51:15 +02:00
-> req().
Remove cowboy_req:fragment/1 Clients do not send it. We skip the value if we receive it now, as it shouldn't happen, and won't for all the mainstream clients.
2013-05-15 15:17:33 +02:00
new(Socket, Transport, Peer, Method, Path, Query,
Optimize cowboy_protocol * #state{} changes are avoided where possible * #state{} is now smaller and use less memory * the Req object is created only after the whole request is parsed * parsing makes use of a single binary match context * external calls are avoided in the critical path * URL fragment is now extracted properly (retrieval API next commit) * argument orders to local functions modified to avoid extra operations * dispatching waits as long as possible before tokenizing host/path * handler opts are no longer shown in the error messages except in init The code may not look as beautiful as it was before. But it really is, for parsing code. The parsing section of the file may be skipped if your eyes start to burn.
2012-09-26 14:11:53 +02:00
Version, Headers, Host, Port, Buffer, CanKeepalive,
Add optional automatic response body compression This behavior can be enabled with the `compress` protocol option. See the `compress_response` example for more details. All tests are now ran with and without compression for both HTTP and HTTPS.
2013-01-07 22:42:16 +01:00
Compress, OnResponse) ->
Make sure the peer name is always available
2013-01-30 23:38:51 +01:00
Req = #http_req{socket=Socket, transport=Transport, pid=self(), peer=Peer,
Remove cowboy_req:fragment/1 Clients do not send it. We skip the value if we receive it now, as it shouldn't happen, and won't for all the mainstream clients.
2013-05-15 15:17:33 +02:00
method=Method, path=Path, qs=Query, version=Version,
Optimize cowboy_protocol * #state{} changes are avoided where possible * #state{} is now smaller and use less memory * the Req object is created only after the whole request is parsed * parsing makes use of a single binary match context * external calls are avoided in the critical path * URL fragment is now extracted properly (retrieval API next commit) * argument orders to local functions modified to avoid extra operations * dispatching waits as long as possible before tokenizing host/path * handler opts are no longer shown in the error messages except in init The code may not look as beautiful as it was before. But it really is, for parsing code. The parsing section of the file may be skipped if your eyes start to burn.
2012-09-26 14:11:53 +02:00
headers=Headers, host=Host, port=Port, buffer=Buffer,
Add optional automatic response body compression This behavior can be enabled with the `compress` protocol option. See the `compress_response` example for more details. All tests are now ran with and without compression for both HTTP and HTTPS.
2013-01-07 22:42:16 +01:00
resp_compress=Compress, onresponse=OnResponse},
Support ad-hoc keep-alive for HTTP/1.0 Clients Only go for keep-alive if they submit a 'connection: keep-alive' header in the request, keep behaviour the same otherwise. The new RFC 7230 (http://tools.ietf.org/html/rfc7230#section-6.3) states: If the received protocol is HTTP/1.0, the "keep-alive" connection option is present, the recipient is not a proxy, and the recipient wishes to honor the HTTP/1.0 "keep-alive" mechanism, the connection will persist after the current response; Even though clients are discouraged from doing so in Appendix A.1.2 (http://tools.ietf.org/html/rfc7230#appendix-A.1.2)
2014-01-14 13:30:58 -05:00
case CanKeepalive of
Optimize cowboy_protocol * #state{} changes are avoided where possible * #state{} is now smaller and use less memory * the Req object is created only after the whole request is parsed * parsing makes use of a single binary match context * external calls are avoided in the critical path * URL fragment is now extracted properly (retrieval API next commit) * argument orders to local functions modified to avoid extra operations * dispatching waits as long as possible before tokenizing host/path * handler opts are no longer shown in the error messages except in init The code may not look as beautiful as it was before. But it really is, for parsing code. The parsing section of the file may be skipped if your eyes start to burn.
2012-09-26 14:11:53 +02:00
false ->
Req#http_req{connection=close};
true ->
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
case parse_header(<<"connection">>, Req) of
undefined ->
Support ad-hoc keep-alive for HTTP/1.0 Clients Only go for keep-alive if they submit a 'connection: keep-alive' header in the request, keep behaviour the same otherwise. The new RFC 7230 (http://tools.ietf.org/html/rfc7230#section-6.3) states: If the received protocol is HTTP/1.0, the "keep-alive" connection option is present, the recipient is not a proxy, and the recipient wishes to honor the HTTP/1.0 "keep-alive" mechanism, the connection will persist after the current response; Even though clients are discouraged from doing so in Appendix A.1.2 (http://tools.ietf.org/html/rfc7230#appendix-A.1.2)
2014-01-14 13:30:58 -05:00
case Version of
'HTTP/1.1' -> Req; %% keepalive
'HTTP/1.0' -> Req#http_req{connection=close}
end;
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
Tokens ->
Optimize Connection header parsing Still optimizing the critical path. Removes cowboy_http:connection_to_atom/1.
2012-09-27 01:31:19 +02:00
Connection = connection_to_atom(Tokens),
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
Req#http_req{connection=Connection}
Optimize cowboy_protocol * #state{} changes are avoided where possible * #state{} is now smaller and use less memory * the Req object is created only after the whole request is parsed * parsing makes use of a single binary match context * external calls are avoided in the critical path * URL fragment is now extracted properly (retrieval API next commit) * argument orders to local functions modified to avoid extra operations * dispatching waits as long as possible before tokenizing host/path * handler opts are no longer shown in the error messages except in init The code may not look as beautiful as it was before. But it really is, for parsing code. The parsing section of the file may be skipped if your eyes start to burn.
2012-09-26 14:11:53 +02:00
end
end.
Add a private cowboy_req function to create a Req object Private means you must not used it. It's meant for internal use.
2012-09-16 15:51:15 +02:00
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
-spec method(req()) -> binary().
Initial commit with connection/streams Breaking changes with previous commit. This is a very large change, and I am giving up on making a single commit that fixes everything. More commits will follow slowly adding back features, introducing new tests and fixing the documentation. This change contains most of the work toward unifying the interface for handling both HTTP/1.1 and HTTP/2. HTTP/1.1 connections are now no longer 1 process per connection; instead by default 1 process per request is also created. This has a number of pros and cons. Because it has cons, we also allow users to use a lower-level API that acts on "streams" (requests/responses) directly at the connection process-level. If performance is a concern, one can always write a stream handler. The performance in this case will be even greater than with Cowboy 1, although all the special handlers are unavailable. When switching to Websocket, after the handler returns from init/2, Cowboy stops the stream and the Websocket protocol takes over the connection process. Websocket then calls websocket_init/2 for any additional initialization such as timers, because the process is different in init/2 and websocket_*/* functions. This however would allow us to use websocket_init/2 for sending messages on connect, instead of sending ourselves a message and be subject to races. Note that websocket_init/2 is optional. This is all a big change and while most of the tests pass, some functionality currently doesn't. SPDY is broken and will be removed soon in favor of HTTP/2. Automatic compression is currently disabled. The cowboy_req interface probably still have a few functions that need to be updated. The docs and examples do not refer the current functionality anymore. Everything will be fixed over time. Feedback is more than welcome. Open a ticket!
2016-02-10 17:28:32 +01:00
method(#{method := Method}) ->
Method.
Initial commit.
2011-03-07 22:59:22 +01:00
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
-spec version(req()) -> cowboy:http_version().
Initial commit with connection/streams Breaking changes with previous commit. This is a very large change, and I am giving up on making a single commit that fixes everything. More commits will follow slowly adding back features, introducing new tests and fixing the documentation. This change contains most of the work toward unifying the interface for handling both HTTP/1.1 and HTTP/2. HTTP/1.1 connections are now no longer 1 process per connection; instead by default 1 process per request is also created. This has a number of pros and cons. Because it has cons, we also allow users to use a lower-level API that acts on "streams" (requests/responses) directly at the connection process-level. If performance is a concern, one can always write a stream handler. The performance in this case will be even greater than with Cowboy 1, although all the special handlers are unavailable. When switching to Websocket, after the handler returns from init/2, Cowboy stops the stream and the Websocket protocol takes over the connection process. Websocket then calls websocket_init/2 for any additional initialization such as timers, because the process is different in init/2 and websocket_*/* functions. This however would allow us to use websocket_init/2 for sending messages on connect, instead of sending ourselves a message and be subject to races. Note that websocket_init/2 is optional. This is all a big change and while most of the tests pass, some functionality currently doesn't. SPDY is broken and will be removed soon in favor of HTTP/2. Automatic compression is currently disabled. The cowboy_req interface probably still have a few functions that need to be updated. The docs and examples do not refer the current functionality anymore. Everything will be fixed over time. Feedback is more than welcome. Open a ticket!
2016-02-10 17:28:32 +01:00
version(#{version := Version}) ->
Version.
Initial commit.
2011-03-07 22:59:22 +01:00
Fix cowboy_req:peer/1
2016-06-20 17:28:59 +02:00
-spec peer(req()) -> {inet:ip_address(), inet:port_number()}.
peer(#{peer := Peer}) ->
Peer.
Initial commit.
2011-03-07 22:59:22 +01:00
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
-spec host(req()) -> binary().
Initial commit with connection/streams Breaking changes with previous commit. This is a very large change, and I am giving up on making a single commit that fixes everything. More commits will follow slowly adding back features, introducing new tests and fixing the documentation. This change contains most of the work toward unifying the interface for handling both HTTP/1.1 and HTTP/2. HTTP/1.1 connections are now no longer 1 process per connection; instead by default 1 process per request is also created. This has a number of pros and cons. Because it has cons, we also allow users to use a lower-level API that acts on "streams" (requests/responses) directly at the connection process-level. If performance is a concern, one can always write a stream handler. The performance in this case will be even greater than with Cowboy 1, although all the special handlers are unavailable. When switching to Websocket, after the handler returns from init/2, Cowboy stops the stream and the Websocket protocol takes over the connection process. Websocket then calls websocket_init/2 for any additional initialization such as timers, because the process is different in init/2 and websocket_*/* functions. This however would allow us to use websocket_init/2 for sending messages on connect, instead of sending ourselves a message and be subject to races. Note that websocket_init/2 is optional. This is all a big change and while most of the tests pass, some functionality currently doesn't. SPDY is broken and will be removed soon in favor of HTTP/2. Automatic compression is currently disabled. The cowboy_req interface probably still have a few functions that need to be updated. The docs and examples do not refer the current functionality anymore. Everything will be fixed over time. Feedback is more than welcome. Open a ticket!
2016-02-10 17:28:32 +01:00
host(#{host := Host}) ->
Host.
Initial commit.
2011-03-07 22:59:22 +01:00
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
-spec host_info(req()) -> cowboy_router:tokens() | undefined.
Initial commit with connection/streams Breaking changes with previous commit. This is a very large change, and I am giving up on making a single commit that fixes everything. More commits will follow slowly adding back features, introducing new tests and fixing the documentation. This change contains most of the work toward unifying the interface for handling both HTTP/1.1 and HTTP/2. HTTP/1.1 connections are now no longer 1 process per connection; instead by default 1 process per request is also created. This has a number of pros and cons. Because it has cons, we also allow users to use a lower-level API that acts on "streams" (requests/responses) directly at the connection process-level. If performance is a concern, one can always write a stream handler. The performance in this case will be even greater than with Cowboy 1, although all the special handlers are unavailable. When switching to Websocket, after the handler returns from init/2, Cowboy stops the stream and the Websocket protocol takes over the connection process. Websocket then calls websocket_init/2 for any additional initialization such as timers, because the process is different in init/2 and websocket_*/* functions. This however would allow us to use websocket_init/2 for sending messages on connect, instead of sending ourselves a message and be subject to races. Note that websocket_init/2 is optional. This is all a big change and while most of the tests pass, some functionality currently doesn't. SPDY is broken and will be removed soon in favor of HTTP/2. Automatic compression is currently disabled. The cowboy_req interface probably still have a few functions that need to be updated. The docs and examples do not refer the current functionality anymore. Everything will be fixed over time. Feedback is more than welcome. Open a ticket!
2016-02-10 17:28:32 +01:00
host_info(#{host_info := HostInfo}) ->
HostInfo.
Implement path_info feature The dispatcher now accepts '...' as the leading segment of Host and the trailing segment of Path, this special atom matches any remaining path tail. When given "cowboy.bugs.dev-extend.eu", host rule ['...', <<"dev-extend">>, <<"eu">>] matches and fills host_info with [<<"cowboy">>, <<"bugs">>]. When given "/a/b/c/d", path rule [<<"a">>, <<"b">>, '...'] matches and fills path_info with [<<"c">>, <<"d">>].
2011-05-09 14:31:06 +02:00
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
-spec port(req()) -> inet:port_number().
Initial commit with connection/streams Breaking changes with previous commit. This is a very large change, and I am giving up on making a single commit that fixes everything. More commits will follow slowly adding back features, introducing new tests and fixing the documentation. This change contains most of the work toward unifying the interface for handling both HTTP/1.1 and HTTP/2. HTTP/1.1 connections are now no longer 1 process per connection; instead by default 1 process per request is also created. This has a number of pros and cons. Because it has cons, we also allow users to use a lower-level API that acts on "streams" (requests/responses) directly at the connection process-level. If performance is a concern, one can always write a stream handler. The performance in this case will be even greater than with Cowboy 1, although all the special handlers are unavailable. When switching to Websocket, after the handler returns from init/2, Cowboy stops the stream and the Websocket protocol takes over the connection process. Websocket then calls websocket_init/2 for any additional initialization such as timers, because the process is different in init/2 and websocket_*/* functions. This however would allow us to use websocket_init/2 for sending messages on connect, instead of sending ourselves a message and be subject to races. Note that websocket_init/2 is optional. This is all a big change and while most of the tests pass, some functionality currently doesn't. SPDY is broken and will be removed soon in favor of HTTP/2. Automatic compression is currently disabled. The cowboy_req interface probably still have a few functions that need to be updated. The docs and examples do not refer the current functionality anymore. Everything will be fixed over time. Feedback is more than welcome. Open a ticket!
2016-02-10 17:28:32 +01:00
port(#{port := Port}) ->
Port.
Add cowboy_http_req:port/1. Returns the port given in the Host header if present, otherwise the default port of 443 for HTTPS and 80 for HTTP is returned.
2011-05-04 12:05:57 +02:00
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
-spec path(req()) -> binary().
Initial commit with connection/streams Breaking changes with previous commit. This is a very large change, and I am giving up on making a single commit that fixes everything. More commits will follow slowly adding back features, introducing new tests and fixing the documentation. This change contains most of the work toward unifying the interface for handling both HTTP/1.1 and HTTP/2. HTTP/1.1 connections are now no longer 1 process per connection; instead by default 1 process per request is also created. This has a number of pros and cons. Because it has cons, we also allow users to use a lower-level API that acts on "streams" (requests/responses) directly at the connection process-level. If performance is a concern, one can always write a stream handler. The performance in this case will be even greater than with Cowboy 1, although all the special handlers are unavailable. When switching to Websocket, after the handler returns from init/2, Cowboy stops the stream and the Websocket protocol takes over the connection process. Websocket then calls websocket_init/2 for any additional initialization such as timers, because the process is different in init/2 and websocket_*/* functions. This however would allow us to use websocket_init/2 for sending messages on connect, instead of sending ourselves a message and be subject to races. Note that websocket_init/2 is optional. This is all a big change and while most of the tests pass, some functionality currently doesn't. SPDY is broken and will be removed soon in favor of HTTP/2. Automatic compression is currently disabled. The cowboy_req interface probably still have a few functions that need to be updated. The docs and examples do not refer the current functionality anymore. Everything will be fixed over time. Feedback is more than welcome. Open a ticket!
2016-02-10 17:28:32 +01:00
path(#{path := Path}) ->
Path.
Initial commit.
2011-03-07 22:59:22 +01:00
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
-spec path_info(req()) -> cowboy_router:tokens() | undefined.
Initial commit with connection/streams Breaking changes with previous commit. This is a very large change, and I am giving up on making a single commit that fixes everything. More commits will follow slowly adding back features, introducing new tests and fixing the documentation. This change contains most of the work toward unifying the interface for handling both HTTP/1.1 and HTTP/2. HTTP/1.1 connections are now no longer 1 process per connection; instead by default 1 process per request is also created. This has a number of pros and cons. Because it has cons, we also allow users to use a lower-level API that acts on "streams" (requests/responses) directly at the connection process-level. If performance is a concern, one can always write a stream handler. The performance in this case will be even greater than with Cowboy 1, although all the special handlers are unavailable. When switching to Websocket, after the handler returns from init/2, Cowboy stops the stream and the Websocket protocol takes over the connection process. Websocket then calls websocket_init/2 for any additional initialization such as timers, because the process is different in init/2 and websocket_*/* functions. This however would allow us to use websocket_init/2 for sending messages on connect, instead of sending ourselves a message and be subject to races. Note that websocket_init/2 is optional. This is all a big change and while most of the tests pass, some functionality currently doesn't. SPDY is broken and will be removed soon in favor of HTTP/2. Automatic compression is currently disabled. The cowboy_req interface probably still have a few functions that need to be updated. The docs and examples do not refer the current functionality anymore. Everything will be fixed over time. Feedback is more than welcome. Open a ticket!
2016-02-10 17:28:32 +01:00
path_info(#{path_info := PathInfo}) ->
PathInfo.
Implement path_info feature The dispatcher now accepts '...' as the leading segment of Host and the trailing segment of Path, this special atom matches any remaining path tail. When given "cowboy.bugs.dev-extend.eu", host rule ['...', <<"dev-extend">>, <<"eu">>] matches and fills host_info with [<<"cowboy">>, <<"bugs">>]. When given "/a/b/c/d", path rule [<<"a">>, <<"b">>, '...'] matches and fills path_info with [<<"c">>, <<"d">>].
2011-05-09 14:31:06 +02:00
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
-spec qs(req()) -> binary().
Initial commit with connection/streams Breaking changes with previous commit. This is a very large change, and I am giving up on making a single commit that fixes everything. More commits will follow slowly adding back features, introducing new tests and fixing the documentation. This change contains most of the work toward unifying the interface for handling both HTTP/1.1 and HTTP/2. HTTP/1.1 connections are now no longer 1 process per connection; instead by default 1 process per request is also created. This has a number of pros and cons. Because it has cons, we also allow users to use a lower-level API that acts on "streams" (requests/responses) directly at the connection process-level. If performance is a concern, one can always write a stream handler. The performance in this case will be even greater than with Cowboy 1, although all the special handlers are unavailable. When switching to Websocket, after the handler returns from init/2, Cowboy stops the stream and the Websocket protocol takes over the connection process. Websocket then calls websocket_init/2 for any additional initialization such as timers, because the process is different in init/2 and websocket_*/* functions. This however would allow us to use websocket_init/2 for sending messages on connect, instead of sending ourselves a message and be subject to races. Note that websocket_init/2 is optional. This is all a big change and while most of the tests pass, some functionality currently doesn't. SPDY is broken and will be removed soon in favor of HTTP/2. Automatic compression is currently disabled. The cowboy_req interface probably still have a few functions that need to be updated. The docs and examples do not refer the current functionality anymore. Everything will be fixed over time. Feedback is more than welcome. Open a ticket!
2016-02-10 17:28:32 +01:00
qs(#{qs := Qs}) ->
Qs.
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
-spec parse_qs(req()) -> [{binary(), binary() | true}].
Initial commit with connection/streams Breaking changes with previous commit. This is a very large change, and I am giving up on making a single commit that fixes everything. More commits will follow slowly adding back features, introducing new tests and fixing the documentation. This change contains most of the work toward unifying the interface for handling both HTTP/1.1 and HTTP/2. HTTP/1.1 connections are now no longer 1 process per connection; instead by default 1 process per request is also created. This has a number of pros and cons. Because it has cons, we also allow users to use a lower-level API that acts on "streams" (requests/responses) directly at the connection process-level. If performance is a concern, one can always write a stream handler. The performance in this case will be even greater than with Cowboy 1, although all the special handlers are unavailable. When switching to Websocket, after the handler returns from init/2, Cowboy stops the stream and the Websocket protocol takes over the connection process. Websocket then calls websocket_init/2 for any additional initialization such as timers, because the process is different in init/2 and websocket_*/* functions. This however would allow us to use websocket_init/2 for sending messages on connect, instead of sending ourselves a message and be subject to races. Note that websocket_init/2 is optional. This is all a big change and while most of the tests pass, some functionality currently doesn't. SPDY is broken and will be removed soon in favor of HTTP/2. Automatic compression is currently disabled. The cowboy_req interface probably still have a few functions that need to be updated. The docs and examples do not refer the current functionality anymore. Everything will be fixed over time. Feedback is more than welcome. Open a ticket!
2016-02-10 17:28:32 +01:00
parse_qs(#{qs := Qs}) ->
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
cow_qs:parse_qs(Qs).
Initial commit.
2011-03-07 22:59:22 +01:00
Reverse the order of arguments of match_* functions Wasn't following the same order as the rest of the module.
2014-10-04 13:21:16 +03:00
-spec match_qs(cowboy:fields(), req()) -> map().
match_qs(Fields, Req) ->
filter(Fields, kvlist_to_map(Fields, parse_qs(Req))).
Initial commit.
2011-03-07 22:59:22 +01:00
Introduce cowboy_req:host_url/1 to remove more duplicate code
2012-09-16 01:13:44 +02:00
%% The URL includes the scheme, host and port only.
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
-spec host_url(req()) -> undefined | binary().
Initial commit with connection/streams Breaking changes with previous commit. This is a very large change, and I am giving up on making a single commit that fixes everything. More commits will follow slowly adding back features, introducing new tests and fixing the documentation. This change contains most of the work toward unifying the interface for handling both HTTP/1.1 and HTTP/2. HTTP/1.1 connections are now no longer 1 process per connection; instead by default 1 process per request is also created. This has a number of pros and cons. Because it has cons, we also allow users to use a lower-level API that acts on "streams" (requests/responses) directly at the connection process-level. If performance is a concern, one can always write a stream handler. The performance in this case will be even greater than with Cowboy 1, although all the special handlers are unavailable. When switching to Websocket, after the handler returns from init/2, Cowboy stops the stream and the Websocket protocol takes over the connection process. Websocket then calls websocket_init/2 for any additional initialization such as timers, because the process is different in init/2 and websocket_*/* functions. This however would allow us to use websocket_init/2 for sending messages on connect, instead of sending ourselves a message and be subject to races. Note that websocket_init/2 is optional. This is all a big change and while most of the tests pass, some functionality currently doesn't. SPDY is broken and will be removed soon in favor of HTTP/2. Automatic compression is currently disabled. The cowboy_req interface probably still have a few functions that need to be updated. The docs and examples do not refer the current functionality anymore. Everything will be fixed over time. Feedback is more than welcome. Open a ticket!
2016-02-10 17:28:32 +01:00
host_url(#{port := undefined}) ->
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
undefined;
Initial commit with connection/streams Breaking changes with previous commit. This is a very large change, and I am giving up on making a single commit that fixes everything. More commits will follow slowly adding back features, introducing new tests and fixing the documentation. This change contains most of the work toward unifying the interface for handling both HTTP/1.1 and HTTP/2. HTTP/1.1 connections are now no longer 1 process per connection; instead by default 1 process per request is also created. This has a number of pros and cons. Because it has cons, we also allow users to use a lower-level API that acts on "streams" (requests/responses) directly at the connection process-level. If performance is a concern, one can always write a stream handler. The performance in this case will be even greater than with Cowboy 1, although all the special handlers are unavailable. When switching to Websocket, after the handler returns from init/2, Cowboy stops the stream and the Websocket protocol takes over the connection process. Websocket then calls websocket_init/2 for any additional initialization such as timers, because the process is different in init/2 and websocket_*/* functions. This however would allow us to use websocket_init/2 for sending messages on connect, instead of sending ourselves a message and be subject to races. Note that websocket_init/2 is optional. This is all a big change and while most of the tests pass, some functionality currently doesn't. SPDY is broken and will be removed soon in favor of HTTP/2. Automatic compression is currently disabled. The cowboy_req interface probably still have a few functions that need to be updated. The docs and examples do not refer the current functionality anymore. Everything will be fixed over time. Feedback is more than welcome. Open a ticket!
2016-02-10 17:28:32 +01:00
host_url(#{scheme := Scheme, host := Host, port := Port}) ->
PortBin = case {Scheme, Port} of
{<<"https">>, 443} -> <<>>;
{<<"http">>, 80} -> <<>>;
Drop R15 support
2014-07-12 14:19:29 +02:00
_ -> << ":", (integer_to_binary(Port))/binary >>
Add cowboy_req:url/1 to return the full request URL Use it in cowboy_websocket for hixie76, replacing http by ws.
2012-09-15 22:03:00 +02:00
end,
Initial commit with connection/streams Breaking changes with previous commit. This is a very large change, and I am giving up on making a single commit that fixes everything. More commits will follow slowly adding back features, introducing new tests and fixing the documentation. This change contains most of the work toward unifying the interface for handling both HTTP/1.1 and HTTP/2. HTTP/1.1 connections are now no longer 1 process per connection; instead by default 1 process per request is also created. This has a number of pros and cons. Because it has cons, we also allow users to use a lower-level API that acts on "streams" (requests/responses) directly at the connection process-level. If performance is a concern, one can always write a stream handler. The performance in this case will be even greater than with Cowboy 1, although all the special handlers are unavailable. When switching to Websocket, after the handler returns from init/2, Cowboy stops the stream and the Websocket protocol takes over the connection process. Websocket then calls websocket_init/2 for any additional initialization such as timers, because the process is different in init/2 and websocket_*/* functions. This however would allow us to use websocket_init/2 for sending messages on connect, instead of sending ourselves a message and be subject to races. Note that websocket_init/2 is optional. This is all a big change and while most of the tests pass, some functionality currently doesn't. SPDY is broken and will be removed soon in favor of HTTP/2. Automatic compression is currently disabled. The cowboy_req interface probably still have a few functions that need to be updated. The docs and examples do not refer the current functionality anymore. Everything will be fixed over time. Feedback is more than welcome. Open a ticket!
2016-02-10 17:28:32 +01:00
<< Scheme/binary, "://", Host/binary, PortBin/binary >>.
Introduce cowboy_req:host_url/1 to remove more duplicate code
2012-09-16 01:13:44 +02:00
Remove cowboy_req:fragment/1 Clients do not send it. We skip the value if we receive it now, as it shouldn't happen, and won't for all the mainstream clients.
2013-05-15 15:17:33 +02:00
%% The URL includes the scheme, host, port, path and query string.
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
-spec url(req()) -> undefined | binary().
Initial commit with connection/streams Breaking changes with previous commit. This is a very large change, and I am giving up on making a single commit that fixes everything. More commits will follow slowly adding back features, introducing new tests and fixing the documentation. This change contains most of the work toward unifying the interface for handling both HTTP/1.1 and HTTP/2. HTTP/1.1 connections are now no longer 1 process per connection; instead by default 1 process per request is also created. This has a number of pros and cons. Because it has cons, we also allow users to use a lower-level API that acts on "streams" (requests/responses) directly at the connection process-level. If performance is a concern, one can always write a stream handler. The performance in this case will be even greater than with Cowboy 1, although all the special handlers are unavailable. When switching to Websocket, after the handler returns from init/2, Cowboy stops the stream and the Websocket protocol takes over the connection process. Websocket then calls websocket_init/2 for any additional initialization such as timers, because the process is different in init/2 and websocket_*/* functions. This however would allow us to use websocket_init/2 for sending messages on connect, instead of sending ourselves a message and be subject to races. Note that websocket_init/2 is optional. This is all a big change and while most of the tests pass, some functionality currently doesn't. SPDY is broken and will be removed soon in favor of HTTP/2. Automatic compression is currently disabled. The cowboy_req interface probably still have a few functions that need to be updated. The docs and examples do not refer the current functionality anymore. Everything will be fixed over time. Feedback is more than welcome. Open a ticket!
2016-02-10 17:28:32 +01:00
url(Req) ->
url(Req, host_url(Req)).
url and host_url may return undefined this can happen when the request parsing fails and onresponse needs the url, its perfectly possible that the url has not yet been defined
2012-11-27 13:22:38 -06:00
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
url(_, undefined) ->
undefined;
Initial commit with connection/streams Breaking changes with previous commit. This is a very large change, and I am giving up on making a single commit that fixes everything. More commits will follow slowly adding back features, introducing new tests and fixing the documentation. This change contains most of the work toward unifying the interface for handling both HTTP/1.1 and HTTP/2. HTTP/1.1 connections are now no longer 1 process per connection; instead by default 1 process per request is also created. This has a number of pros and cons. Because it has cons, we also allow users to use a lower-level API that acts on "streams" (requests/responses) directly at the connection process-level. If performance is a concern, one can always write a stream handler. The performance in this case will be even greater than with Cowboy 1, although all the special handlers are unavailable. When switching to Websocket, after the handler returns from init/2, Cowboy stops the stream and the Websocket protocol takes over the connection process. Websocket then calls websocket_init/2 for any additional initialization such as timers, because the process is different in init/2 and websocket_*/* functions. This however would allow us to use websocket_init/2 for sending messages on connect, instead of sending ourselves a message and be subject to races. Note that websocket_init/2 is optional. This is all a big change and while most of the tests pass, some functionality currently doesn't. SPDY is broken and will be removed soon in favor of HTTP/2. Automatic compression is currently disabled. The cowboy_req interface probably still have a few functions that need to be updated. The docs and examples do not refer the current functionality anymore. Everything will be fixed over time. Feedback is more than welcome. Open a ticket!
2016-02-10 17:28:32 +01:00
url(#{path := Path, qs := QS}, HostURL) ->
Add cowboy_req:url/1 to return the full request URL Use it in cowboy_websocket for hixie76, replacing http by ws.
2012-09-15 22:03:00 +02:00
QS2 = case QS of
<<>> -> <<>>;
_ -> << "?", QS/binary >>
end,
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
<< HostURL/binary, Path/binary, QS2/binary >>.
Add cowboy_req:url/1 to return the full request URL Use it in cowboy_websocket for hixie76, replacing http by ws.
2012-09-15 22:03:00 +02:00
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
-spec binding(atom(), req()) -> any() | undefined.
binding(Name, Req) ->
Make the default value the last parameter of HTTP req accessors This makes cowboy_http_req's qs_val/3, binding/3 and header/3 behave like proplists:get_value/3.
2011-03-29 13:57:21 +02:00
binding(Name, Req, undefined).
Make HTTP req accessors call themselves with their default default value Use lists:keyfind/3 in qs_val/3, binding/3 and header/3 and make qs_val/2, binding/2 and header/2 call those.
2011-03-29 13:49:48 +02:00
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
-spec binding(atom(), req(), Default) -> any() | Default when Default::any().
Initial commit with connection/streams Breaking changes with previous commit. This is a very large change, and I am giving up on making a single commit that fixes everything. More commits will follow slowly adding back features, introducing new tests and fixing the documentation. This change contains most of the work toward unifying the interface for handling both HTTP/1.1 and HTTP/2. HTTP/1.1 connections are now no longer 1 process per connection; instead by default 1 process per request is also created. This has a number of pros and cons. Because it has cons, we also allow users to use a lower-level API that acts on "streams" (requests/responses) directly at the connection process-level. If performance is a concern, one can always write a stream handler. The performance in this case will be even greater than with Cowboy 1, although all the special handlers are unavailable. When switching to Websocket, after the handler returns from init/2, Cowboy stops the stream and the Websocket protocol takes over the connection process. Websocket then calls websocket_init/2 for any additional initialization such as timers, because the process is different in init/2 and websocket_*/* functions. This however would allow us to use websocket_init/2 for sending messages on connect, instead of sending ourselves a message and be subject to races. Note that websocket_init/2 is optional. This is all a big change and while most of the tests pass, some functionality currently doesn't. SPDY is broken and will be removed soon in favor of HTTP/2. Automatic compression is currently disabled. The cowboy_req interface probably still have a few functions that need to be updated. The docs and examples do not refer the current functionality anymore. Everything will be fixed over time. Feedback is more than welcome. Open a ticket!
2016-02-10 17:28:32 +01:00
binding(Name, #{bindings := Bindings}, Default) when is_atom(Name) ->
case lists:keyfind(Name, 1, Bindings) of
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
{_, Value} -> Value;
false -> Default
Initial commit with connection/streams Breaking changes with previous commit. This is a very large change, and I am giving up on making a single commit that fixes everything. More commits will follow slowly adding back features, introducing new tests and fixing the documentation. This change contains most of the work toward unifying the interface for handling both HTTP/1.1 and HTTP/2. HTTP/1.1 connections are now no longer 1 process per connection; instead by default 1 process per request is also created. This has a number of pros and cons. Because it has cons, we also allow users to use a lower-level API that acts on "streams" (requests/responses) directly at the connection process-level. If performance is a concern, one can always write a stream handler. The performance in this case will be even greater than with Cowboy 1, although all the special handlers are unavailable. When switching to Websocket, after the handler returns from init/2, Cowboy stops the stream and the Websocket protocol takes over the connection process. Websocket then calls websocket_init/2 for any additional initialization such as timers, because the process is different in init/2 and websocket_*/* functions. This however would allow us to use websocket_init/2 for sending messages on connect, instead of sending ourselves a message and be subject to races. Note that websocket_init/2 is optional. This is all a big change and while most of the tests pass, some functionality currently doesn't. SPDY is broken and will be removed soon in favor of HTTP/2. Automatic compression is currently disabled. The cowboy_req interface probably still have a few functions that need to be updated. The docs and examples do not refer the current functionality anymore. Everything will be fixed over time. Feedback is more than welcome. Open a ticket!
2016-02-10 17:28:32 +01:00
end;
binding(Name, _, Default) when is_atom(Name) ->
Default.
Initial commit.
2011-03-07 22:59:22 +01:00
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
-spec bindings(req()) -> [{atom(), any()}].
Initial commit with connection/streams Breaking changes with previous commit. This is a very large change, and I am giving up on making a single commit that fixes everything. More commits will follow slowly adding back features, introducing new tests and fixing the documentation. This change contains most of the work toward unifying the interface for handling both HTTP/1.1 and HTTP/2. HTTP/1.1 connections are now no longer 1 process per connection; instead by default 1 process per request is also created. This has a number of pros and cons. Because it has cons, we also allow users to use a lower-level API that acts on "streams" (requests/responses) directly at the connection process-level. If performance is a concern, one can always write a stream handler. The performance in this case will be even greater than with Cowboy 1, although all the special handlers are unavailable. When switching to Websocket, after the handler returns from init/2, Cowboy stops the stream and the Websocket protocol takes over the connection process. Websocket then calls websocket_init/2 for any additional initialization such as timers, because the process is different in init/2 and websocket_*/* functions. This however would allow us to use websocket_init/2 for sending messages on connect, instead of sending ourselves a message and be subject to races. Note that websocket_init/2 is optional. This is all a big change and while most of the tests pass, some functionality currently doesn't. SPDY is broken and will be removed soon in favor of HTTP/2. Automatic compression is currently disabled. The cowboy_req interface probably still have a few functions that need to be updated. The docs and examples do not refer the current functionality anymore. Everything will be fixed over time. Feedback is more than welcome. Open a ticket!
2016-02-10 17:28:32 +01:00
bindings(#{bindings := Bindings}) ->
Bindings;
bindings(_) ->
[].
Initial commit.
2011-03-07 22:59:22 +01:00
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
-spec header(binary(), req()) -> binary() | undefined.
Don't use decode_packet/3 for parsing the headers Header names are now binaries. Since header names are case insensitive they are all converted to lowercase. For example: <<"content-length">>. The max_line_length option was removed. Three new options have been added instead: * max_request_line_length (defaults to 4096) * max_header_name_length (defaults to 64) * max_header_value_length (defaults to 4096)
2012-09-21 09:18:56 +02:00
header(Name, Req) ->
Make the default value the last parameter of HTTP req accessors This makes cowboy_http_req's qs_val/3, binding/3 and header/3 behave like proplists:get_value/3.
2011-03-29 13:57:21 +02:00
header(Name, Req, undefined).
Make HTTP req accessors call themselves with their default default value Use lists:keyfind/3 in qs_val/3, binding/3 and header/3 and make qs_val/2, binding/2 and header/2 call those.
2011-03-29 13:49:48 +02:00
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
-spec header(binary(), req(), Default) -> binary() | Default when Default::any().
Initial commit with connection/streams Breaking changes with previous commit. This is a very large change, and I am giving up on making a single commit that fixes everything. More commits will follow slowly adding back features, introducing new tests and fixing the documentation. This change contains most of the work toward unifying the interface for handling both HTTP/1.1 and HTTP/2. HTTP/1.1 connections are now no longer 1 process per connection; instead by default 1 process per request is also created. This has a number of pros and cons. Because it has cons, we also allow users to use a lower-level API that acts on "streams" (requests/responses) directly at the connection process-level. If performance is a concern, one can always write a stream handler. The performance in this case will be even greater than with Cowboy 1, although all the special handlers are unavailable. When switching to Websocket, after the handler returns from init/2, Cowboy stops the stream and the Websocket protocol takes over the connection process. Websocket then calls websocket_init/2 for any additional initialization such as timers, because the process is different in init/2 and websocket_*/* functions. This however would allow us to use websocket_init/2 for sending messages on connect, instead of sending ourselves a message and be subject to races. Note that websocket_init/2 is optional. This is all a big change and while most of the tests pass, some functionality currently doesn't. SPDY is broken and will be removed soon in favor of HTTP/2. Automatic compression is currently disabled. The cowboy_req interface probably still have a few functions that need to be updated. The docs and examples do not refer the current functionality anymore. Everything will be fixed over time. Feedback is more than welcome. Open a ticket!
2016-02-10 17:28:32 +01:00
header(Name, #{headers := Headers}, Default) ->
maps:get(Name, Headers, Default).
Initial commit.
2011-03-07 22:59:22 +01:00
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
-spec headers(req()) -> cowboy:http_headers().
Initial commit with connection/streams Breaking changes with previous commit. This is a very large change, and I am giving up on making a single commit that fixes everything. More commits will follow slowly adding back features, introducing new tests and fixing the documentation. This change contains most of the work toward unifying the interface for handling both HTTP/1.1 and HTTP/2. HTTP/1.1 connections are now no longer 1 process per connection; instead by default 1 process per request is also created. This has a number of pros and cons. Because it has cons, we also allow users to use a lower-level API that acts on "streams" (requests/responses) directly at the connection process-level. If performance is a concern, one can always write a stream handler. The performance in this case will be even greater than with Cowboy 1, although all the special handlers are unavailable. When switching to Websocket, after the handler returns from init/2, Cowboy stops the stream and the Websocket protocol takes over the connection process. Websocket then calls websocket_init/2 for any additional initialization such as timers, because the process is different in init/2 and websocket_*/* functions. This however would allow us to use websocket_init/2 for sending messages on connect, instead of sending ourselves a message and be subject to races. Note that websocket_init/2 is optional. This is all a big change and while most of the tests pass, some functionality currently doesn't. SPDY is broken and will be removed soon in favor of HTTP/2. Automatic compression is currently disabled. The cowboy_req interface probably still have a few functions that need to be updated. The docs and examples do not refer the current functionality anymore. Everything will be fixed over time. Feedback is more than welcome. Open a ticket!
2016-02-10 17:28:32 +01:00
headers(#{headers := Headers}) ->
Headers.
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
-spec parse_header(binary(), Req) -> any() when Req::req().
parse_header(Name = <<"content-length">>, Req) ->
Use cowlib for parsing headers; remove cowboy_http
2015-02-04 16:18:28 +01:00
parse_header(Name, Req, 0, fun cow_http_hd:parse_content_length/1);
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
parse_header(Name = <<"cookie">>, Req) ->
Use cowlib for parsing headers; remove cowboy_http
2015-02-04 16:18:28 +01:00
parse_header(Name, Req, [], fun cow_cookie:parse_cookie/1);
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
parse_header(Name = <<"transfer-encoding">>, Req) ->
Use cowlib for parsing headers; remove cowboy_http
2015-02-04 16:18:28 +01:00
parse_header(Name, Req, [<<"identity">>], fun cow_http_hd:parse_transfer_encoding/1);
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
parse_header(Name, Req) ->
parse_header(Name, Req, undefined).
-spec parse_header(binary(), Req, any()) -> any() when Req::req().
Use cowlib for parsing headers; remove cowboy_http
2015-02-04 16:18:28 +01:00
parse_header(Name, Req, Default) ->
parse_header(Name, Req, Default, parse_header_fun(Name)).
parse_header_fun(<<"accept">>) -> fun cow_http_hd:parse_accept/1;
parse_header_fun(<<"accept-charset">>) -> fun cow_http_hd:parse_accept_charset/1;
parse_header_fun(<<"accept-encoding">>) -> fun cow_http_hd:parse_accept_encoding/1;
parse_header_fun(<<"accept-language">>) -> fun cow_http_hd:parse_accept_language/1;
parse_header_fun(<<"authorization">>) -> fun cow_http_hd:parse_authorization/1;
parse_header_fun(<<"connection">>) -> fun cow_http_hd:parse_connection/1;
parse_header_fun(<<"content-length">>) -> fun cow_http_hd:parse_content_length/1;
parse_header_fun(<<"content-type">>) -> fun cow_http_hd:parse_content_type/1;
parse_header_fun(<<"cookie">>) -> fun cow_cookie:parse_cookie/1;
parse_header_fun(<<"expect">>) -> fun cow_http_hd:parse_expect/1;
parse_header_fun(<<"if-match">>) -> fun cow_http_hd:parse_if_match/1;
parse_header_fun(<<"if-modified-since">>) -> fun cow_http_hd:parse_if_modified_since/1;
parse_header_fun(<<"if-none-match">>) -> fun cow_http_hd:parse_if_none_match/1;
parse_header_fun(<<"if-unmodified-since">>) -> fun cow_http_hd:parse_if_unmodified_since/1;
parse_header_fun(<<"range">>) -> fun cow_http_hd:parse_range/1;
parse_header_fun(<<"sec-websocket-extensions">>) -> fun cow_http_hd:parse_sec_websocket_extensions/1;
parse_header_fun(<<"sec-websocket-protocol">>) -> fun cow_http_hd:parse_sec_websocket_protocol_req/1;
parse_header_fun(<<"transfer-encoding">>) -> fun cow_http_hd:parse_transfer_encoding/1;
parse_header_fun(<<"upgrade">>) -> fun cow_http_hd:parse_upgrade/1;
parse_header_fun(<<"x-forwarded-for">>) -> fun cow_http_hd:parse_x_forwarded_for/1.
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
parse_header(Name, Req, Default, ParseFun) ->
Parse 'Connection' headers as a list of tokens Replaces the 'Connection' interpretation in cowboy_http_protocol from raw value to the parsed value, looking for a single token matching close/keep-alive instead of the whole raw value (which could contain more than one token, for example with Firefox 6+ using websocket). Introduce the functions cowboy_http_req:parse_header/2 and /3 to semantically parse the header values and return a proper Erlang term.
2011-10-05 03:17:13 +02:00
case header(Name, Req) of
Use cowlib for parsing headers; remove cowboy_http
2015-02-04 16:18:28 +01:00
undefined -> Default;
Value -> ParseFun(Value)
Add 'Accept' header parsing Rework the cowboy_http_req:parse_header/2 function while I was at it.
2011-10-26 04:07:08 +02:00
end.
Parse 'Connection' headers as a list of tokens Replaces the 'Connection' interpretation in cowboy_http_protocol from raw value to the parsed value, looking for a single token matching close/keep-alive instead of the whole raw value (which could contain more than one token, for example with Firefox 6+ using websocket). Introduce the functions cowboy_http_req:parse_header/2 and /3 to semantically parse the header values and return a proper Erlang term.
2011-10-05 03:17:13 +02:00
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
-spec parse_cookies(req()) -> [{binary(), binary()}].
parse_cookies(Req) ->
parse_header(<<"cookie">>, Req).
Implement cookies in cowboy_http_req
2011-07-08 13:41:30 -05:00
Reverse the order of arguments of match_* functions Wasn't following the same order as the rest of the module.
2014-10-04 13:21:16 +03:00
-spec match_cookies(cowboy:fields(), req()) -> map().
match_cookies(Fields, Req) ->
filter(Fields, kvlist_to_map(Fields, parse_cookies(Req))).
Implement cookies in cowboy_http_req
2011-07-08 13:41:30 -05:00
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
-spec meta(atom(), req()) -> any() | undefined.
Add meta/2 and meta/3 to cowboy_http_req to save useful protocol information * cowboy_http_protocol now defines 'websocket_version' as metadata. * cowboy_http_rest now defines 'media_type', 'language', 'charset' as metadata.
2011-12-19 09:44:24 +01:00
meta(Name, Req) ->
meta(Name, Req, undefined).
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
-spec meta(atom(), req(), any()) -> any().
Add meta/2 and meta/3 to cowboy_http_req to save useful protocol information * cowboy_http_protocol now defines 'websocket_version' as metadata. * cowboy_http_rest now defines 'media_type', 'language', 'charset' as metadata.
2011-12-19 09:44:24 +01:00
meta(Name, Req, Default) ->
case lists:keyfind(Name, 1, Req#http_req.meta) of
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
{Name, Value} -> Value;
false -> Default
Add meta/2 and meta/3 to cowboy_http_req to save useful protocol information * cowboy_http_protocol now defines 'websocket_version' as metadata. * cowboy_http_rest now defines 'media_type', 'language', 'charset' as metadata.
2011-12-19 09:44:24 +01:00
end.
Introduce cowboy_req:set_meta/3 to set request metadata
2012-09-15 22:51:37 +02:00
-spec set_meta(atom(), any(), Req) -> Req when Req::req().
set_meta(Name, Value, Req=#http_req{meta=Meta}) ->
Remove outdated comments, all edoc, plus a few minor tweaks
2014-03-26 19:05:59 +01:00
Req#http_req{meta=lists:keystore(Name, 1, Meta, {Name, Value})}.
Introduce cowboy_req:set_meta/3 to set request metadata
2012-09-15 22:51:37 +02:00
Skip the request body if it hasn't been read by the handler.
2011-03-21 17:26:00 +01:00
%% Request Body API.
No need for fully qualified call to the same module
2013-04-11 14:22:16 +02:00
-spec has_body(req()) -> boolean().
Initial commit with connection/streams Breaking changes with previous commit. This is a very large change, and I am giving up on making a single commit that fixes everything. More commits will follow slowly adding back features, introducing new tests and fixing the documentation. This change contains most of the work toward unifying the interface for handling both HTTP/1.1 and HTTP/2. HTTP/1.1 connections are now no longer 1 process per connection; instead by default 1 process per request is also created. This has a number of pros and cons. Because it has cons, we also allow users to use a lower-level API that acts on "streams" (requests/responses) directly at the connection process-level. If performance is a concern, one can always write a stream handler. The performance in this case will be even greater than with Cowboy 1, although all the special handlers are unavailable. When switching to Websocket, after the handler returns from init/2, Cowboy stops the stream and the Websocket protocol takes over the connection process. Websocket then calls websocket_init/2 for any additional initialization such as timers, because the process is different in init/2 and websocket_*/* functions. This however would allow us to use websocket_init/2 for sending messages on connect, instead of sending ourselves a message and be subject to races. Note that websocket_init/2 is optional. This is all a big change and while most of the tests pass, some functionality currently doesn't. SPDY is broken and will be removed soon in favor of HTTP/2. Automatic compression is currently disabled. The cowboy_req interface probably still have a few functions that need to be updated. The docs and examples do not refer the current functionality anymore. Everything will be fixed over time. Feedback is more than welcome. Open a ticket!
2016-02-10 17:28:32 +01:00
has_body(#{has_body := HasBody}) ->
HasBody.
Add chunked transfer encoding support and rework the body reading API Introduces 3 low level functions and updates the existing higher levels functions. The new primitives are has_body/1, body_length/1 and stream_body/1. In addition to that, a helper function init_stream/4 has been added. Streaming a body implies to decode the Transfer-Encoding and Content-Encoding used for the body. By default, Cowboy will try to figure out what was used and decode them properly. You can override this if you want to disable this behavior or simply support more encodings by calling the init_stream/4 function before you start streaming the body.
2012-03-29 01:14:44 +02:00
%% The length may not be known if Transfer-Encoding is not identity,
%% and the body hasn't been read at the time of the call.
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
-spec body_length(req()) -> undefined | non_neg_integer().
Initial commit with connection/streams Breaking changes with previous commit. This is a very large change, and I am giving up on making a single commit that fixes everything. More commits will follow slowly adding back features, introducing new tests and fixing the documentation. This change contains most of the work toward unifying the interface for handling both HTTP/1.1 and HTTP/2. HTTP/1.1 connections are now no longer 1 process per connection; instead by default 1 process per request is also created. This has a number of pros and cons. Because it has cons, we also allow users to use a lower-level API that acts on "streams" (requests/responses) directly at the connection process-level. If performance is a concern, one can always write a stream handler. The performance in this case will be even greater than with Cowboy 1, although all the special handlers are unavailable. When switching to Websocket, after the handler returns from init/2, Cowboy stops the stream and the Websocket protocol takes over the connection process. Websocket then calls websocket_init/2 for any additional initialization such as timers, because the process is different in init/2 and websocket_*/* functions. This however would allow us to use websocket_init/2 for sending messages on connect, instead of sending ourselves a message and be subject to races. Note that websocket_init/2 is optional. This is all a big change and while most of the tests pass, some functionality currently doesn't. SPDY is broken and will be removed soon in favor of HTTP/2. Automatic compression is currently disabled. The cowboy_req interface probably still have a few functions that need to be updated. The docs and examples do not refer the current functionality anymore. Everything will be fixed over time. Feedback is more than welcome. Open a ticket!
2016-02-10 17:28:32 +01:00
body_length(#{body_length := Length}) ->
Length.
Add chunked transfer encoding support and rework the body reading API Introduces 3 low level functions and updates the existing higher levels functions. The new primitives are has_body/1, body_length/1 and stream_body/1. In addition to that, a helper function init_stream/4 has been added. Streaming a body implies to decode the Transfer-Encoding and Content-Encoding used for the body. By default, Cowboy will try to figure out what was used and decode them properly. You can override this if you want to disable this behavior or simply support more encodings by calling the init_stream/4 function before you start streaming the body.
2012-03-29 01:14:44 +02:00
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
-spec body(Req) -> {ok, binary(), Req} | {more, binary(), Req} when Req::req().
Add request body reading options The options were added to allow developers to fix timeout issues when reading large bodies. It is also a cleaner and easier to extend interface. This commit deprecates the functions init_stream, stream_body and skip_body which are no longer needed. They will be removed in 1.0. The body function can now take an additional argument that is a list of options. The body_qs, part and part_body functions can too and simply pass this argument down to the body call. There are options for disabling the automatic continue reply, setting a maximum length to be returned (soft limit), setting the read length and read timeout, and setting the transfer and content decode functions. The return value of the body and body_qs have changed slightly. The body function now works similarly to the part_body function, in that it returns either an ok or a more tuple depending on whether there is additional data to be read. The body_qs function can return a badlength tuple if the body is too big. The default size has been increased from 16KB to 64KB. The default read length and timeout have been tweaked and vary depending on the function called. The body function will now adequately process chunked bodies, which means that the body_qs function will too. But this means that the behavior has changed slightly and your code should be tested properly when updating your code. The body and body_qs still accept a length as first argument for compatibility purpose with older code. Note that this form is deprecated and will be removed in 1.0. The part and part_body function, being new and never having been in a release yet, have this form completely removed in this commit. Again, while most code should work as-is, you should make sure that it actually does before pushing this to production.
2014-06-02 23:09:43 +02:00
body(Req) ->
body(Req, []).
Initial commit with connection/streams Breaking changes with previous commit. This is a very large change, and I am giving up on making a single commit that fixes everything. More commits will follow slowly adding back features, introducing new tests and fixing the documentation. This change contains most of the work toward unifying the interface for handling both HTTP/1.1 and HTTP/2. HTTP/1.1 connections are now no longer 1 process per connection; instead by default 1 process per request is also created. This has a number of pros and cons. Because it has cons, we also allow users to use a lower-level API that acts on "streams" (requests/responses) directly at the connection process-level. If performance is a concern, one can always write a stream handler. The performance in this case will be even greater than with Cowboy 1, although all the special handlers are unavailable. When switching to Websocket, after the handler returns from init/2, Cowboy stops the stream and the Websocket protocol takes over the connection process. Websocket then calls websocket_init/2 for any additional initialization such as timers, because the process is different in init/2 and websocket_*/* functions. This however would allow us to use websocket_init/2 for sending messages on connect, instead of sending ourselves a message and be subject to races. Note that websocket_init/2 is optional. This is all a big change and while most of the tests pass, some functionality currently doesn't. SPDY is broken and will be removed soon in favor of HTTP/2. Automatic compression is currently disabled. The cowboy_req interface probably still have a few functions that need to be updated. The docs and examples do not refer the current functionality anymore. Everything will be fixed over time. Feedback is more than welcome. Open a ticket!
2016-02-10 17:28:32 +01:00
-spec read_body(Req) -> {ok, binary(), Req} | {more, binary(), Req} when Req::req().
read_body(Req) ->
read_body(Req, []).
-spec read_body(Req, body_opts()) -> {ok, binary(), Req} | {more, binary(), Req} when Req::req().
read_body(Req=#{pid := Pid, streamid := StreamID}, Opts) ->
%% @todo Opts should be a map
Length = case lists:keyfind(length, 1, Opts) of
false -> 8000000;
{_, ChunkLen0} -> ChunkLen0
end,
ReadTimeout = case lists:keyfind(read_timeout, 1, Opts) of
false -> 15000;
{_, ReadTimeout0} -> ReadTimeout0
end,
Ref = make_ref(),
Pid ! {{Pid, StreamID}, {read_body, Ref, Length}},
receive
{request_body, Ref, nofin, Body} ->
{more, Body, Req};
{request_body, Ref, {fin, BodyLength}, Body} ->
{ok, Body, set_body_length(Req, BodyLength)}
after ReadTimeout ->
exit(read_body_timeout)
end.
set_body_length(Req=#{headers := Headers}, BodyLength) ->
Req#{
headers => Headers#{<<"content-length">> => integer_to_binary(BodyLength)},
body_length => BodyLength
}.
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
-spec body(Req, body_opts()) -> {ok, binary(), Req} | {more, binary(), Req} when Req::req().
Add request body reading options The options were added to allow developers to fix timeout issues when reading large bodies. It is also a cleaner and easier to extend interface. This commit deprecates the functions init_stream, stream_body and skip_body which are no longer needed. They will be removed in 1.0. The body function can now take an additional argument that is a list of options. The body_qs, part and part_body functions can too and simply pass this argument down to the body call. There are options for disabling the automatic continue reply, setting a maximum length to be returned (soft limit), setting the read length and read timeout, and setting the transfer and content decode functions. The return value of the body and body_qs have changed slightly. The body function now works similarly to the part_body function, in that it returns either an ok or a more tuple depending on whether there is additional data to be read. The body_qs function can return a badlength tuple if the body is too big. The default size has been increased from 16KB to 64KB. The default read length and timeout have been tweaked and vary depending on the function called. The body function will now adequately process chunked bodies, which means that the body_qs function will too. But this means that the behavior has changed slightly and your code should be tested properly when updating your code. The body and body_qs still accept a length as first argument for compatibility purpose with older code. Note that this form is deprecated and will be removed in 1.0. The part and part_body function, being new and never having been in a release yet, have this form completely removed in this commit. Again, while most code should work as-is, you should make sure that it actually does before pushing this to production.
2014-06-02 23:09:43 +02:00
body(Req=#http_req{body_state=waiting}, Opts) ->
%% Send a 100 continue if needed (enabled by default).
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
case lists:keyfind(continue, 1, Opts) of
Add request body reading options The options were added to allow developers to fix timeout issues when reading large bodies. It is also a cleaner and easier to extend interface. This commit deprecates the functions init_stream, stream_body and skip_body which are no longer needed. They will be removed in 1.0. The body function can now take an additional argument that is a list of options. The body_qs, part and part_body functions can too and simply pass this argument down to the body call. There are options for disabling the automatic continue reply, setting a maximum length to be returned (soft limit), setting the read length and read timeout, and setting the transfer and content decode functions. The return value of the body and body_qs have changed slightly. The body function now works similarly to the part_body function, in that it returns either an ok or a more tuple depending on whether there is additional data to be read. The body_qs function can return a badlength tuple if the body is too big. The default size has been increased from 16KB to 64KB. The default read length and timeout have been tweaked and vary depending on the function called. The body function will now adequately process chunked bodies, which means that the body_qs function will too. But this means that the behavior has changed slightly and your code should be tested properly when updating your code. The body and body_qs still accept a length as first argument for compatibility purpose with older code. Note that this form is deprecated and will be removed in 1.0. The part and part_body function, being new and never having been in a release yet, have this form completely removed in this commit. Again, while most code should work as-is, you should make sure that it actually does before pushing this to production.
2014-06-02 23:09:43 +02:00
{_, false} ->
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
ok;
Add request body reading options The options were added to allow developers to fix timeout issues when reading large bodies. It is also a cleaner and easier to extend interface. This commit deprecates the functions init_stream, stream_body and skip_body which are no longer needed. They will be removed in 1.0. The body function can now take an additional argument that is a list of options. The body_qs, part and part_body functions can too and simply pass this argument down to the body call. There are options for disabling the automatic continue reply, setting a maximum length to be returned (soft limit), setting the read length and read timeout, and setting the transfer and content decode functions. The return value of the body and body_qs have changed slightly. The body function now works similarly to the part_body function, in that it returns either an ok or a more tuple depending on whether there is additional data to be read. The body_qs function can return a badlength tuple if the body is too big. The default size has been increased from 16KB to 64KB. The default read length and timeout have been tweaked and vary depending on the function called. The body function will now adequately process chunked bodies, which means that the body_qs function will too. But this means that the behavior has changed slightly and your code should be tested properly when updating your code. The body and body_qs still accept a length as first argument for compatibility purpose with older code. Note that this form is deprecated and will be removed in 1.0. The part and part_body function, being new and never having been in a release yet, have this form completely removed in this commit. Again, while most code should work as-is, you should make sure that it actually does before pushing this to production.
2014-06-02 23:09:43 +02:00
_ ->
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
ExpectHeader = parse_header(<<"expect">>, Req),
Add request body reading options The options were added to allow developers to fix timeout issues when reading large bodies. It is also a cleaner and easier to extend interface. This commit deprecates the functions init_stream, stream_body and skip_body which are no longer needed. They will be removed in 1.0. The body function can now take an additional argument that is a list of options. The body_qs, part and part_body functions can too and simply pass this argument down to the body call. There are options for disabling the automatic continue reply, setting a maximum length to be returned (soft limit), setting the read length and read timeout, and setting the transfer and content decode functions. The return value of the body and body_qs have changed slightly. The body function now works similarly to the part_body function, in that it returns either an ok or a more tuple depending on whether there is additional data to be read. The body_qs function can return a badlength tuple if the body is too big. The default size has been increased from 16KB to 64KB. The default read length and timeout have been tweaked and vary depending on the function called. The body function will now adequately process chunked bodies, which means that the body_qs function will too. But this means that the behavior has changed slightly and your code should be tested properly when updating your code. The body and body_qs still accept a length as first argument for compatibility purpose with older code. Note that this form is deprecated and will be removed in 1.0. The part and part_body function, being new and never having been in a release yet, have this form completely removed in this commit. Again, while most code should work as-is, you should make sure that it actually does before pushing this to production.
2014-06-02 23:09:43 +02:00
ok = case ExpectHeader of
expect header parses to an atom (continue)
2015-05-29 15:07:22 -07:00
continue -> continue(Req);
Add request body reading options The options were added to allow developers to fix timeout issues when reading large bodies. It is also a cleaner and easier to extend interface. This commit deprecates the functions init_stream, stream_body and skip_body which are no longer needed. They will be removed in 1.0. The body function can now take an additional argument that is a list of options. The body_qs, part and part_body functions can too and simply pass this argument down to the body call. There are options for disabling the automatic continue reply, setting a maximum length to be returned (soft limit), setting the read length and read timeout, and setting the transfer and content decode functions. The return value of the body and body_qs have changed slightly. The body function now works similarly to the part_body function, in that it returns either an ok or a more tuple depending on whether there is additional data to be read. The body_qs function can return a badlength tuple if the body is too big. The default size has been increased from 16KB to 64KB. The default read length and timeout have been tweaked and vary depending on the function called. The body function will now adequately process chunked bodies, which means that the body_qs function will too. But this means that the behavior has changed slightly and your code should be tested properly when updating your code. The body and body_qs still accept a length as first argument for compatibility purpose with older code. Note that this form is deprecated and will be removed in 1.0. The part and part_body function, being new and never having been in a release yet, have this form completely removed in this commit. Again, while most code should work as-is, you should make sure that it actually does before pushing this to production.
2014-06-02 23:09:43 +02:00
_ -> ok
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
end
Add request body reading options The options were added to allow developers to fix timeout issues when reading large bodies. It is also a cleaner and easier to extend interface. This commit deprecates the functions init_stream, stream_body and skip_body which are no longer needed. They will be removed in 1.0. The body function can now take an additional argument that is a list of options. The body_qs, part and part_body functions can too and simply pass this argument down to the body call. There are options for disabling the automatic continue reply, setting a maximum length to be returned (soft limit), setting the read length and read timeout, and setting the transfer and content decode functions. The return value of the body and body_qs have changed slightly. The body function now works similarly to the part_body function, in that it returns either an ok or a more tuple depending on whether there is additional data to be read. The body_qs function can return a badlength tuple if the body is too big. The default size has been increased from 16KB to 64KB. The default read length and timeout have been tweaked and vary depending on the function called. The body function will now adequately process chunked bodies, which means that the body_qs function will too. But this means that the behavior has changed slightly and your code should be tested properly when updating your code. The body and body_qs still accept a length as first argument for compatibility purpose with older code. Note that this form is deprecated and will be removed in 1.0. The part and part_body function, being new and never having been in a release yet, have this form completely removed in this commit. Again, while most code should work as-is, you should make sure that it actually does before pushing this to production.
2014-06-02 23:09:43 +02:00
end,
%% Initialize body streaming state.
CFun = case lists:keyfind(content_decode, 1, Opts) of
false ->
Use cowlib for parsing headers; remove cowboy_http
2015-02-04 16:18:28 +01:00
fun body_content_decode_identity/1;
Add request body reading options The options were added to allow developers to fix timeout issues when reading large bodies. It is also a cleaner and easier to extend interface. This commit deprecates the functions init_stream, stream_body and skip_body which are no longer needed. They will be removed in 1.0. The body function can now take an additional argument that is a list of options. The body_qs, part and part_body functions can too and simply pass this argument down to the body call. There are options for disabling the automatic continue reply, setting a maximum length to be returned (soft limit), setting the read length and read timeout, and setting the transfer and content decode functions. The return value of the body and body_qs have changed slightly. The body function now works similarly to the part_body function, in that it returns either an ok or a more tuple depending on whether there is additional data to be read. The body_qs function can return a badlength tuple if the body is too big. The default size has been increased from 16KB to 64KB. The default read length and timeout have been tweaked and vary depending on the function called. The body function will now adequately process chunked bodies, which means that the body_qs function will too. But this means that the behavior has changed slightly and your code should be tested properly when updating your code. The body and body_qs still accept a length as first argument for compatibility purpose with older code. Note that this form is deprecated and will be removed in 1.0. The part and part_body function, being new and never having been in a release yet, have this form completely removed in this commit. Again, while most code should work as-is, you should make sure that it actually does before pushing this to production.
2014-06-02 23:09:43 +02:00
{_, CFun0} ->
CFun0
end,
case lists:keyfind(transfer_decode, 1, Opts) of
false ->
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
case parse_header(<<"transfer-encoding">>, Req) of
[<<"chunked">>] ->
body(Req#http_req{body_state={stream, 0,
Add request body reading options The options were added to allow developers to fix timeout issues when reading large bodies. It is also a cleaner and easier to extend interface. This commit deprecates the functions init_stream, stream_body and skip_body which are no longer needed. They will be removed in 1.0. The body function can now take an additional argument that is a list of options. The body_qs, part and part_body functions can too and simply pass this argument down to the body call. There are options for disabling the automatic continue reply, setting a maximum length to be returned (soft limit), setting the read length and read timeout, and setting the transfer and content decode functions. The return value of the body and body_qs have changed slightly. The body function now works similarly to the part_body function, in that it returns either an ok or a more tuple depending on whether there is additional data to be read. The body_qs function can return a badlength tuple if the body is too big. The default size has been increased from 16KB to 64KB. The default read length and timeout have been tweaked and vary depending on the function called. The body function will now adequately process chunked bodies, which means that the body_qs function will too. But this means that the behavior has changed slightly and your code should be tested properly when updating your code. The body and body_qs still accept a length as first argument for compatibility purpose with older code. Note that this form is deprecated and will be removed in 1.0. The part and part_body function, being new and never having been in a release yet, have this form completely removed in this commit. Again, while most code should work as-is, you should make sure that it actually does before pushing this to production.
2014-06-02 23:09:43 +02:00
fun cow_http_te:stream_chunked/2, {0, 0}, CFun}}, Opts);
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
[<<"identity">>] ->
case body_length(Req) of
Add request body reading options The options were added to allow developers to fix timeout issues when reading large bodies. It is also a cleaner and easier to extend interface. This commit deprecates the functions init_stream, stream_body and skip_body which are no longer needed. They will be removed in 1.0. The body function can now take an additional argument that is a list of options. The body_qs, part and part_body functions can too and simply pass this argument down to the body call. There are options for disabling the automatic continue reply, setting a maximum length to be returned (soft limit), setting the read length and read timeout, and setting the transfer and content decode functions. The return value of the body and body_qs have changed slightly. The body function now works similarly to the part_body function, in that it returns either an ok or a more tuple depending on whether there is additional data to be read. The body_qs function can return a badlength tuple if the body is too big. The default size has been increased from 16KB to 64KB. The default read length and timeout have been tweaked and vary depending on the function called. The body function will now adequately process chunked bodies, which means that the body_qs function will too. But this means that the behavior has changed slightly and your code should be tested properly when updating your code. The body and body_qs still accept a length as first argument for compatibility purpose with older code. Note that this form is deprecated and will be removed in 1.0. The part and part_body function, being new and never having been in a release yet, have this form completely removed in this commit. Again, while most code should work as-is, you should make sure that it actually does before pushing this to production.
2014-06-02 23:09:43 +02:00
0 ->
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
{ok, <<>>, Req#http_req{body_state=done}};
Len ->
body(Req#http_req{body_state={stream, Len,
Add request body reading options The options were added to allow developers to fix timeout issues when reading large bodies. It is also a cleaner and easier to extend interface. This commit deprecates the functions init_stream, stream_body and skip_body which are no longer needed. They will be removed in 1.0. The body function can now take an additional argument that is a list of options. The body_qs, part and part_body functions can too and simply pass this argument down to the body call. There are options for disabling the automatic continue reply, setting a maximum length to be returned (soft limit), setting the read length and read timeout, and setting the transfer and content decode functions. The return value of the body and body_qs have changed slightly. The body function now works similarly to the part_body function, in that it returns either an ok or a more tuple depending on whether there is additional data to be read. The body_qs function can return a badlength tuple if the body is too big. The default size has been increased from 16KB to 64KB. The default read length and timeout have been tweaked and vary depending on the function called. The body function will now adequately process chunked bodies, which means that the body_qs function will too. But this means that the behavior has changed slightly and your code should be tested properly when updating your code. The body and body_qs still accept a length as first argument for compatibility purpose with older code. Note that this form is deprecated and will be removed in 1.0. The part and part_body function, being new and never having been in a release yet, have this form completely removed in this commit. Again, while most code should work as-is, you should make sure that it actually does before pushing this to production.
2014-06-02 23:09:43 +02:00
fun cow_http_te:stream_identity/2, {0, Len},
CFun}}, Opts)
end
end;
{_, TFun, TState} ->
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
body(Req#http_req{body_state={stream, 0,
Add request body reading options The options were added to allow developers to fix timeout issues when reading large bodies. It is also a cleaner and easier to extend interface. This commit deprecates the functions init_stream, stream_body and skip_body which are no longer needed. They will be removed in 1.0. The body function can now take an additional argument that is a list of options. The body_qs, part and part_body functions can too and simply pass this argument down to the body call. There are options for disabling the automatic continue reply, setting a maximum length to be returned (soft limit), setting the read length and read timeout, and setting the transfer and content decode functions. The return value of the body and body_qs have changed slightly. The body function now works similarly to the part_body function, in that it returns either an ok or a more tuple depending on whether there is additional data to be read. The body_qs function can return a badlength tuple if the body is too big. The default size has been increased from 16KB to 64KB. The default read length and timeout have been tweaked and vary depending on the function called. The body function will now adequately process chunked bodies, which means that the body_qs function will too. But this means that the behavior has changed slightly and your code should be tested properly when updating your code. The body and body_qs still accept a length as first argument for compatibility purpose with older code. Note that this form is deprecated and will be removed in 1.0. The part and part_body function, being new and never having been in a release yet, have this form completely removed in this commit. Again, while most code should work as-is, you should make sure that it actually does before pushing this to production.
2014-06-02 23:09:43 +02:00
TFun, TState, CFun}}, Opts)
end;
body(Req=#http_req{body_state=done}, _) ->
{ok, <<>>, Req};
body(Req, Opts) ->
ChunkLen = case lists:keyfind(length, 1, Opts) of
false -> 8000000;
{_, ChunkLen0} -> ChunkLen0
end,
ReadLen = case lists:keyfind(read_length, 1, Opts) of
false -> 1000000;
{_, ReadLen0} -> ReadLen0
end,
ReadTimeout = case lists:keyfind(read_timeout, 1, Opts) of
false -> 15000;
{_, ReadTimeout0} -> ReadTimeout0
end,
body_loop(Req, ReadTimeout, ReadLen, ChunkLen, <<>>).
Use cowlib for parsing headers; remove cowboy_http
2015-02-04 16:18:28 +01:00
%% Default identity function for content decoding.
%% @todo Move into cowlib when more content decode functions get implemented.
body_content_decode_identity(Data) -> Data.
Add request body reading options The options were added to allow developers to fix timeout issues when reading large bodies. It is also a cleaner and easier to extend interface. This commit deprecates the functions init_stream, stream_body and skip_body which are no longer needed. They will be removed in 1.0. The body function can now take an additional argument that is a list of options. The body_qs, part and part_body functions can too and simply pass this argument down to the body call. There are options for disabling the automatic continue reply, setting a maximum length to be returned (soft limit), setting the read length and read timeout, and setting the transfer and content decode functions. The return value of the body and body_qs have changed slightly. The body function now works similarly to the part_body function, in that it returns either an ok or a more tuple depending on whether there is additional data to be read. The body_qs function can return a badlength tuple if the body is too big. The default size has been increased from 16KB to 64KB. The default read length and timeout have been tweaked and vary depending on the function called. The body function will now adequately process chunked bodies, which means that the body_qs function will too. But this means that the behavior has changed slightly and your code should be tested properly when updating your code. The body and body_qs still accept a length as first argument for compatibility purpose with older code. Note that this form is deprecated and will be removed in 1.0. The part and part_body function, being new and never having been in a release yet, have this form completely removed in this commit. Again, while most code should work as-is, you should make sure that it actually does before pushing this to production.
2014-06-02 23:09:43 +02:00
body_loop(Req=#http_req{buffer=Buffer, body_state={stream, Length, _, _, _}},
ReadTimeout, ReadLength, ChunkLength, Acc) ->
{Tag, Res, Req2} = case Buffer of
<<>> ->
body_recv(Req, ReadTimeout, min(Length, ReadLength));
_ ->
body_decode(Req, ReadTimeout)
end,
case {Tag, Res} of
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
{ok, Data} ->
Add request body reading options The options were added to allow developers to fix timeout issues when reading large bodies. It is also a cleaner and easier to extend interface. This commit deprecates the functions init_stream, stream_body and skip_body which are no longer needed. They will be removed in 1.0. The body function can now take an additional argument that is a list of options. The body_qs, part and part_body functions can too and simply pass this argument down to the body call. There are options for disabling the automatic continue reply, setting a maximum length to be returned (soft limit), setting the read length and read timeout, and setting the transfer and content decode functions. The return value of the body and body_qs have changed slightly. The body function now works similarly to the part_body function, in that it returns either an ok or a more tuple depending on whether there is additional data to be read. The body_qs function can return a badlength tuple if the body is too big. The default size has been increased from 16KB to 64KB. The default read length and timeout have been tweaked and vary depending on the function called. The body function will now adequately process chunked bodies, which means that the body_qs function will too. But this means that the behavior has changed slightly and your code should be tested properly when updating your code. The body and body_qs still accept a length as first argument for compatibility purpose with older code. Note that this form is deprecated and will be removed in 1.0. The part and part_body function, being new and never having been in a release yet, have this form completely removed in this commit. Again, while most code should work as-is, you should make sure that it actually does before pushing this to production.
2014-06-02 23:09:43 +02:00
{ok, << Acc/binary, Data/binary >>, Req2};
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
{more, Data} ->
Add request body reading options The options were added to allow developers to fix timeout issues when reading large bodies. It is also a cleaner and easier to extend interface. This commit deprecates the functions init_stream, stream_body and skip_body which are no longer needed. They will be removed in 1.0. The body function can now take an additional argument that is a list of options. The body_qs, part and part_body functions can too and simply pass this argument down to the body call. There are options for disabling the automatic continue reply, setting a maximum length to be returned (soft limit), setting the read length and read timeout, and setting the transfer and content decode functions. The return value of the body and body_qs have changed slightly. The body function now works similarly to the part_body function, in that it returns either an ok or a more tuple depending on whether there is additional data to be read. The body_qs function can return a badlength tuple if the body is too big. The default size has been increased from 16KB to 64KB. The default read length and timeout have been tweaked and vary depending on the function called. The body function will now adequately process chunked bodies, which means that the body_qs function will too. But this means that the behavior has changed slightly and your code should be tested properly when updating your code. The body and body_qs still accept a length as first argument for compatibility purpose with older code. Note that this form is deprecated and will be removed in 1.0. The part and part_body function, being new and never having been in a release yet, have this form completely removed in this commit. Again, while most code should work as-is, you should make sure that it actually does before pushing this to production.
2014-06-02 23:09:43 +02:00
Acc2 = << Acc/binary, Data/binary >>,
case byte_size(Acc2) >= ChunkLength of
true -> {more, Acc2, Req2};
false -> body_loop(Req2, ReadTimeout, ReadLength, ChunkLength, Acc2)
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
end
Add request body reading options The options were added to allow developers to fix timeout issues when reading large bodies. It is also a cleaner and easier to extend interface. This commit deprecates the functions init_stream, stream_body and skip_body which are no longer needed. They will be removed in 1.0. The body function can now take an additional argument that is a list of options. The body_qs, part and part_body functions can too and simply pass this argument down to the body call. There are options for disabling the automatic continue reply, setting a maximum length to be returned (soft limit), setting the read length and read timeout, and setting the transfer and content decode functions. The return value of the body and body_qs have changed slightly. The body function now works similarly to the part_body function, in that it returns either an ok or a more tuple depending on whether there is additional data to be read. The body_qs function can return a badlength tuple if the body is too big. The default size has been increased from 16KB to 64KB. The default read length and timeout have been tweaked and vary depending on the function called. The body function will now adequately process chunked bodies, which means that the body_qs function will too. But this means that the behavior has changed slightly and your code should be tested properly when updating your code. The body and body_qs still accept a length as first argument for compatibility purpose with older code. Note that this form is deprecated and will be removed in 1.0. The part and part_body function, being new and never having been in a release yet, have this form completely removed in this commit. Again, while most code should work as-is, you should make sure that it actually does before pushing this to production.
2014-06-02 23:09:43 +02:00
end.
body_recv(Req=#http_req{transport=Transport, socket=Socket, buffer=Buffer},
ReadTimeout, ReadLength) ->
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
{ok, Data} = Transport:recv(Socket, ReadLength, ReadTimeout),
body_decode(Req#http_req{buffer= << Buffer/binary, Data/binary >>}, ReadTimeout).
Add request body reading options The options were added to allow developers to fix timeout issues when reading large bodies. It is also a cleaner and easier to extend interface. This commit deprecates the functions init_stream, stream_body and skip_body which are no longer needed. They will be removed in 1.0. The body function can now take an additional argument that is a list of options. The body_qs, part and part_body functions can too and simply pass this argument down to the body call. There are options for disabling the automatic continue reply, setting a maximum length to be returned (soft limit), setting the read length and read timeout, and setting the transfer and content decode functions. The return value of the body and body_qs have changed slightly. The body function now works similarly to the part_body function, in that it returns either an ok or a more tuple depending on whether there is additional data to be read. The body_qs function can return a badlength tuple if the body is too big. The default size has been increased from 16KB to 64KB. The default read length and timeout have been tweaked and vary depending on the function called. The body function will now adequately process chunked bodies, which means that the body_qs function will too. But this means that the behavior has changed slightly and your code should be tested properly when updating your code. The body and body_qs still accept a length as first argument for compatibility purpose with older code. Note that this form is deprecated and will be removed in 1.0. The part and part_body function, being new and never having been in a release yet, have this form completely removed in this commit. Again, while most code should work as-is, you should make sure that it actually does before pushing this to production.
2014-06-02 23:09:43 +02:00
Add chunked transfer encoding support and rework the body reading API Introduces 3 low level functions and updates the existing higher levels functions. The new primitives are has_body/1, body_length/1 and stream_body/1. In addition to that, a helper function init_stream/4 has been added. Streaming a body implies to decode the Transfer-Encoding and Content-Encoding used for the body. By default, Cowboy will try to figure out what was used and decode them properly. You can override this if you want to disable this behavior or simply support more encodings by calling the init_stream/4 function before you start streaming the body.
2012-03-29 01:14:44 +02:00
%% Two decodings happen. First a decoding function is applied to the
%% transferred data, and then another is applied to the actual content.
%%
%% Transfer encoding is generally used for chunked bodies. The decoding
%% function uses a state to keep track of how much it has read, which is
%% also initialized through this function.
%%
%% Content encoding is generally used for compression.
Add request body reading options The options were added to allow developers to fix timeout issues when reading large bodies. It is also a cleaner and easier to extend interface. This commit deprecates the functions init_stream, stream_body and skip_body which are no longer needed. They will be removed in 1.0. The body function can now take an additional argument that is a list of options. The body_qs, part and part_body functions can too and simply pass this argument down to the body call. There are options for disabling the automatic continue reply, setting a maximum length to be returned (soft limit), setting the read length and read timeout, and setting the transfer and content decode functions. The return value of the body and body_qs have changed slightly. The body function now works similarly to the part_body function, in that it returns either an ok or a more tuple depending on whether there is additional data to be read. The body_qs function can return a badlength tuple if the body is too big. The default size has been increased from 16KB to 64KB. The default read length and timeout have been tweaked and vary depending on the function called. The body function will now adequately process chunked bodies, which means that the body_qs function will too. But this means that the behavior has changed slightly and your code should be tested properly when updating your code. The body and body_qs still accept a length as first argument for compatibility purpose with older code. Note that this form is deprecated and will be removed in 1.0. The part and part_body function, being new and never having been in a release yet, have this form completely removed in this commit. Again, while most code should work as-is, you should make sure that it actually does before pushing this to production.
2014-06-02 23:09:43 +02:00
%%
Use the new chunked and identity transfer decoding code from cowlib
2014-03-24 14:14:43 +01:00
%% @todo Handle chunked after-the-facts headers.
%% @todo Depending on the length returned we might want to 0 or +5 it.
Add request body reading options The options were added to allow developers to fix timeout issues when reading large bodies. It is also a cleaner and easier to extend interface. This commit deprecates the functions init_stream, stream_body and skip_body which are no longer needed. They will be removed in 1.0. The body function can now take an additional argument that is a list of options. The body_qs, part and part_body functions can too and simply pass this argument down to the body call. There are options for disabling the automatic continue reply, setting a maximum length to be returned (soft limit), setting the read length and read timeout, and setting the transfer and content decode functions. The return value of the body and body_qs have changed slightly. The body function now works similarly to the part_body function, in that it returns either an ok or a more tuple depending on whether there is additional data to be read. The body_qs function can return a badlength tuple if the body is too big. The default size has been increased from 16KB to 64KB. The default read length and timeout have been tweaked and vary depending on the function called. The body function will now adequately process chunked bodies, which means that the body_qs function will too. But this means that the behavior has changed slightly and your code should be tested properly when updating your code. The body and body_qs still accept a length as first argument for compatibility purpose with older code. Note that this form is deprecated and will be removed in 1.0. The part and part_body function, being new and never having been in a release yet, have this form completely removed in this commit. Again, while most code should work as-is, you should make sure that it actually does before pushing this to production.
2014-06-02 23:09:43 +02:00
body_decode(Req=#http_req{buffer=Data, body_state={stream, _,
TDecode, TState, CDecode}}, ReadTimeout) ->
case TDecode(Data, TState) of
Add chunked transfer encoding support and rework the body reading API Introduces 3 low level functions and updates the existing higher levels functions. The new primitives are has_body/1, body_length/1 and stream_body/1. In addition to that, a helper function init_stream/4 has been added. Streaming a body implies to decode the Transfer-Encoding and Content-Encoding used for the body. By default, Cowboy will try to figure out what was used and decode them properly. You can override this if you want to disable this behavior or simply support more encodings by calling the init_stream/4 function before you start streaming the body.
2012-03-29 01:14:44 +02:00
more ->
Add request body reading options The options were added to allow developers to fix timeout issues when reading large bodies. It is also a cleaner and easier to extend interface. This commit deprecates the functions init_stream, stream_body and skip_body which are no longer needed. They will be removed in 1.0. The body function can now take an additional argument that is a list of options. The body_qs, part and part_body functions can too and simply pass this argument down to the body call. There are options for disabling the automatic continue reply, setting a maximum length to be returned (soft limit), setting the read length and read timeout, and setting the transfer and content decode functions. The return value of the body and body_qs have changed slightly. The body function now works similarly to the part_body function, in that it returns either an ok or a more tuple depending on whether there is additional data to be read. The body_qs function can return a badlength tuple if the body is too big. The default size has been increased from 16KB to 64KB. The default read length and timeout have been tweaked and vary depending on the function called. The body function will now adequately process chunked bodies, which means that the body_qs function will too. But this means that the behavior has changed slightly and your code should be tested properly when updating your code. The body and body_qs still accept a length as first argument for compatibility purpose with older code. Note that this form is deprecated and will be removed in 1.0. The part and part_body function, being new and never having been in a release yet, have this form completely removed in this commit. Again, while most code should work as-is, you should make sure that it actually does before pushing this to production.
2014-06-02 23:09:43 +02:00
body_recv(Req#http_req{body_state={stream, 0,
TDecode, TState, CDecode}}, ReadTimeout, 0);
{more, Data2, TState2} ->
{more, CDecode(Data2), Req#http_req{body_state={stream, 0,
TDecode, TState2, CDecode}, buffer= <<>>}};
{more, Data2, Length, TState2} when is_integer(Length) ->
{more, CDecode(Data2), Req#http_req{body_state={stream, Length,
TDecode, TState2, CDecode}, buffer= <<>>}};
{more, Data2, Rest, TState2} ->
{more, CDecode(Data2), Req#http_req{body_state={stream, 0,
TDecode, TState2, CDecode}, buffer=Rest}};
{done, TotalLength, Rest} ->
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
{ok, <<>>, body_decode_end(Req, TotalLength, Rest)};
Add request body reading options The options were added to allow developers to fix timeout issues when reading large bodies. It is also a cleaner and easier to extend interface. This commit deprecates the functions init_stream, stream_body and skip_body which are no longer needed. They will be removed in 1.0. The body function can now take an additional argument that is a list of options. The body_qs, part and part_body functions can too and simply pass this argument down to the body call. There are options for disabling the automatic continue reply, setting a maximum length to be returned (soft limit), setting the read length and read timeout, and setting the transfer and content decode functions. The return value of the body and body_qs have changed slightly. The body function now works similarly to the part_body function, in that it returns either an ok or a more tuple depending on whether there is additional data to be read. The body_qs function can return a badlength tuple if the body is too big. The default size has been increased from 16KB to 64KB. The default read length and timeout have been tweaked and vary depending on the function called. The body function will now adequately process chunked bodies, which means that the body_qs function will too. But this means that the behavior has changed slightly and your code should be tested properly when updating your code. The body and body_qs still accept a length as first argument for compatibility purpose with older code. Note that this form is deprecated and will be removed in 1.0. The part and part_body function, being new and never having been in a release yet, have this form completely removed in this commit. Again, while most code should work as-is, you should make sure that it actually does before pushing this to production.
2014-06-02 23:09:43 +02:00
{done, Data2, TotalLength, Rest} ->
{ok, CDecode(Data2), body_decode_end(Req, TotalLength, Rest)}
Add chunked transfer encoding support and rework the body reading API Introduces 3 low level functions and updates the existing higher levels functions. The new primitives are has_body/1, body_length/1 and stream_body/1. In addition to that, a helper function init_stream/4 has been added. Streaming a body implies to decode the Transfer-Encoding and Content-Encoding used for the body. By default, Cowboy will try to figure out what was used and decode them properly. You can override this if you want to disable this behavior or simply support more encodings by calling the init_stream/4 function before you start streaming the body.
2012-03-29 01:14:44 +02:00
end.
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
body_decode_end(Req=#http_req{headers=Headers}, TotalLength, Rest) ->
Don't use decode_packet/3 for parsing the headers Header names are now binaries. Since header names are case insensitive they are all converted to lowercase. For example: <<"content-length">>. The max_line_length option was removed. Three new options have been added instead: * max_request_line_length (defaults to 4096) * max_header_name_length (defaults to 64) * max_header_value_length (defaults to 4096)
2012-09-21 09:18:56 +02:00
Headers2 = lists:keystore(<<"content-length">>, 1, Headers,
Drop R15 support
2014-07-12 14:19:29 +02:00
{<<"content-length">>, integer_to_binary(TotalLength)}),
Add chunked transfer encoding support and rework the body reading API Introduces 3 low level functions and updates the existing higher levels functions. The new primitives are has_body/1, body_length/1 and stream_body/1. In addition to that, a helper function init_stream/4 has been added. Streaming a body implies to decode the Transfer-Encoding and Content-Encoding used for the body. By default, Cowboy will try to figure out what was used and decode them properly. You can override this if you want to disable this behavior or simply support more encodings by calling the init_stream/4 function before you start streaming the body.
2012-03-29 01:14:44 +02:00
%% At this point we just assume TEs were all decoded.
Don't use decode_packet/3 for parsing the headers Header names are now binaries. Since header names are case insensitive they are all converted to lowercase. For example: <<"content-length">>. The max_line_length option was removed. Three new options have been added instead: * max_request_line_length (defaults to 4096) * max_header_name_length (defaults to 64) * max_header_value_length (defaults to 4096)
2012-09-21 09:18:56 +02:00
Headers3 = lists:keydelete(<<"transfer-encoding">>, 1, Headers2),
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
Req#http_req{buffer=Rest, body_state=done, headers=Headers3}.
-spec body_qs(Req) -> {ok, [{binary(), binary() | true}], Req}
| {badlength, Req} when Req::req().
Add request body reading options The options were added to allow developers to fix timeout issues when reading large bodies. It is also a cleaner and easier to extend interface. This commit deprecates the functions init_stream, stream_body and skip_body which are no longer needed. They will be removed in 1.0. The body function can now take an additional argument that is a list of options. The body_qs, part and part_body functions can too and simply pass this argument down to the body call. There are options for disabling the automatic continue reply, setting a maximum length to be returned (soft limit), setting the read length and read timeout, and setting the transfer and content decode functions. The return value of the body and body_qs have changed slightly. The body function now works similarly to the part_body function, in that it returns either an ok or a more tuple depending on whether there is additional data to be read. The body_qs function can return a badlength tuple if the body is too big. The default size has been increased from 16KB to 64KB. The default read length and timeout have been tweaked and vary depending on the function called. The body function will now adequately process chunked bodies, which means that the body_qs function will too. But this means that the behavior has changed slightly and your code should be tested properly when updating your code. The body and body_qs still accept a length as first argument for compatibility purpose with older code. Note that this form is deprecated and will be removed in 1.0. The part and part_body function, being new and never having been in a release yet, have this form completely removed in this commit. Again, while most code should work as-is, you should make sure that it actually does before pushing this to production.
2014-06-02 23:09:43 +02:00
body_qs(Req) ->
body_qs(Req, [
{length, 64000},
{read_length, 64000},
{read_timeout, 5000}]).
-spec body_qs(Req, body_opts()) -> {ok, [{binary(), binary() | true}], Req}
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
| {badlength, Req} when Req::req().
Add request body reading options The options were added to allow developers to fix timeout issues when reading large bodies. It is also a cleaner and easier to extend interface. This commit deprecates the functions init_stream, stream_body and skip_body which are no longer needed. They will be removed in 1.0. The body function can now take an additional argument that is a list of options. The body_qs, part and part_body functions can too and simply pass this argument down to the body call. There are options for disabling the automatic continue reply, setting a maximum length to be returned (soft limit), setting the read length and read timeout, and setting the transfer and content decode functions. The return value of the body and body_qs have changed slightly. The body function now works similarly to the part_body function, in that it returns either an ok or a more tuple depending on whether there is additional data to be read. The body_qs function can return a badlength tuple if the body is too big. The default size has been increased from 16KB to 64KB. The default read length and timeout have been tweaked and vary depending on the function called. The body function will now adequately process chunked bodies, which means that the body_qs function will too. But this means that the behavior has changed slightly and your code should be tested properly when updating your code. The body and body_qs still accept a length as first argument for compatibility purpose with older code. Note that this form is deprecated and will be removed in 1.0. The part and part_body function, being new and never having been in a release yet, have this form completely removed in this commit. Again, while most code should work as-is, you should make sure that it actually does before pushing this to production.
2014-06-02 23:09:43 +02:00
body_qs(Req, Opts) ->
Initial commit with connection/streams Breaking changes with previous commit. This is a very large change, and I am giving up on making a single commit that fixes everything. More commits will follow slowly adding back features, introducing new tests and fixing the documentation. This change contains most of the work toward unifying the interface for handling both HTTP/1.1 and HTTP/2. HTTP/1.1 connections are now no longer 1 process per connection; instead by default 1 process per request is also created. This has a number of pros and cons. Because it has cons, we also allow users to use a lower-level API that acts on "streams" (requests/responses) directly at the connection process-level. If performance is a concern, one can always write a stream handler. The performance in this case will be even greater than with Cowboy 1, although all the special handlers are unavailable. When switching to Websocket, after the handler returns from init/2, Cowboy stops the stream and the Websocket protocol takes over the connection process. Websocket then calls websocket_init/2 for any additional initialization such as timers, because the process is different in init/2 and websocket_*/* functions. This however would allow us to use websocket_init/2 for sending messages on connect, instead of sending ourselves a message and be subject to races. Note that websocket_init/2 is optional. This is all a big change and while most of the tests pass, some functionality currently doesn't. SPDY is broken and will be removed soon in favor of HTTP/2. Automatic compression is currently disabled. The cowboy_req interface probably still have a few functions that need to be updated. The docs and examples do not refer the current functionality anymore. Everything will be fixed over time. Feedback is more than welcome. Open a ticket!
2016-02-10 17:28:32 +01:00
case read_body(Req, Opts) of
Add request body reading options The options were added to allow developers to fix timeout issues when reading large bodies. It is also a cleaner and easier to extend interface. This commit deprecates the functions init_stream, stream_body and skip_body which are no longer needed. They will be removed in 1.0. The body function can now take an additional argument that is a list of options. The body_qs, part and part_body functions can too and simply pass this argument down to the body call. There are options for disabling the automatic continue reply, setting a maximum length to be returned (soft limit), setting the read length and read timeout, and setting the transfer and content decode functions. The return value of the body and body_qs have changed slightly. The body function now works similarly to the part_body function, in that it returns either an ok or a more tuple depending on whether there is additional data to be read. The body_qs function can return a badlength tuple if the body is too big. The default size has been increased from 16KB to 64KB. The default read length and timeout have been tweaked and vary depending on the function called. The body function will now adequately process chunked bodies, which means that the body_qs function will too. But this means that the behavior has changed slightly and your code should be tested properly when updating your code. The body and body_qs still accept a length as first argument for compatibility purpose with older code. Note that this form is deprecated and will be removed in 1.0. The part and part_body function, being new and never having been in a release yet, have this form completely removed in this commit. Again, while most code should work as-is, you should make sure that it actually does before pushing this to production.
2014-06-02 23:09:43 +02:00
{ok, Body, Req2} ->
{ok, cow_qs:parse_qs(Body), Req2};
{more, _, Req2} ->
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
{badlength, Req2}
Add chunked transfer encoding support and rework the body reading API Introduces 3 low level functions and updates the existing higher levels functions. The new primitives are has_body/1, body_length/1 and stream_body/1. In addition to that, a helper function init_stream/4 has been added. Streaming a body implies to decode the Transfer-Encoding and Content-Encoding used for the body. By default, Cowboy will try to figure out what was used and decode them properly. You can override this if you want to disable this behavior or simply support more encodings by calling the init_stream/4 function before you start streaming the body.
2012-03-29 01:14:44 +02:00
end.
Add and document the new multipart code The old undocumented API is removed entirely. While a documentation exists for the new API, it will not be considered set in stone until further testing has been performed, and a file upload example has been added. The new API should be a little more efficient than the old API, especially with smaller messages.
2014-02-06 19:36:25 +01:00
%% Multipart API.
Add multipart support
2011-11-08 00:51:49 +01:00
Add and document the new multipart code The old undocumented API is removed entirely. While a documentation exists for the new API, it will not be considered set in stone until further testing has been performed, and a file upload example has been added. The new API should be a little more efficient than the old API, especially with smaller messages.
2014-02-06 19:36:25 +01:00
-spec part(Req)
-> {ok, cow_multipart:headers(), Req} | {done, Req}
when Req::req().
part(Req) ->
Add request body reading options The options were added to allow developers to fix timeout issues when reading large bodies. It is also a cleaner and easier to extend interface. This commit deprecates the functions init_stream, stream_body and skip_body which are no longer needed. They will be removed in 1.0. The body function can now take an additional argument that is a list of options. The body_qs, part and part_body functions can too and simply pass this argument down to the body call. There are options for disabling the automatic continue reply, setting a maximum length to be returned (soft limit), setting the read length and read timeout, and setting the transfer and content decode functions. The return value of the body and body_qs have changed slightly. The body function now works similarly to the part_body function, in that it returns either an ok or a more tuple depending on whether there is additional data to be read. The body_qs function can return a badlength tuple if the body is too big. The default size has been increased from 16KB to 64KB. The default read length and timeout have been tweaked and vary depending on the function called. The body function will now adequately process chunked bodies, which means that the body_qs function will too. But this means that the behavior has changed slightly and your code should be tested properly when updating your code. The body and body_qs still accept a length as first argument for compatibility purpose with older code. Note that this form is deprecated and will be removed in 1.0. The part and part_body function, being new and never having been in a release yet, have this form completely removed in this commit. Again, while most code should work as-is, you should make sure that it actually does before pushing this to production.
2014-06-02 23:09:43 +02:00
part(Req, [
{length, 64000},
{read_length, 64000},
{read_timeout, 5000}]).
Add and document the new multipart code The old undocumented API is removed entirely. While a documentation exists for the new API, it will not be considered set in stone until further testing has been performed, and a file upload example has been added. The new API should be a little more efficient than the old API, especially with smaller messages.
2014-02-06 19:36:25 +01:00
Add request body reading options The options were added to allow developers to fix timeout issues when reading large bodies. It is also a cleaner and easier to extend interface. This commit deprecates the functions init_stream, stream_body and skip_body which are no longer needed. They will be removed in 1.0. The body function can now take an additional argument that is a list of options. The body_qs, part and part_body functions can too and simply pass this argument down to the body call. There are options for disabling the automatic continue reply, setting a maximum length to be returned (soft limit), setting the read length and read timeout, and setting the transfer and content decode functions. The return value of the body and body_qs have changed slightly. The body function now works similarly to the part_body function, in that it returns either an ok or a more tuple depending on whether there is additional data to be read. The body_qs function can return a badlength tuple if the body is too big. The default size has been increased from 16KB to 64KB. The default read length and timeout have been tweaked and vary depending on the function called. The body function will now adequately process chunked bodies, which means that the body_qs function will too. But this means that the behavior has changed slightly and your code should be tested properly when updating your code. The body and body_qs still accept a length as first argument for compatibility purpose with older code. Note that this form is deprecated and will be removed in 1.0. The part and part_body function, being new and never having been in a release yet, have this form completely removed in this commit. Again, while most code should work as-is, you should make sure that it actually does before pushing this to production.
2014-06-02 23:09:43 +02:00
-spec part(Req, body_opts())
-> {ok, cow_multipart:headers(), Req} | {done, Req}
when Req::req().
part(Req, Opts) ->
Initial commit with connection/streams Breaking changes with previous commit. This is a very large change, and I am giving up on making a single commit that fixes everything. More commits will follow slowly adding back features, introducing new tests and fixing the documentation. This change contains most of the work toward unifying the interface for handling both HTTP/1.1 and HTTP/2. HTTP/1.1 connections are now no longer 1 process per connection; instead by default 1 process per request is also created. This has a number of pros and cons. Because it has cons, we also allow users to use a lower-level API that acts on "streams" (requests/responses) directly at the connection process-level. If performance is a concern, one can always write a stream handler. The performance in this case will be even greater than with Cowboy 1, although all the special handlers are unavailable. When switching to Websocket, after the handler returns from init/2, Cowboy stops the stream and the Websocket protocol takes over the connection process. Websocket then calls websocket_init/2 for any additional initialization such as timers, because the process is different in init/2 and websocket_*/* functions. This however would allow us to use websocket_init/2 for sending messages on connect, instead of sending ourselves a message and be subject to races. Note that websocket_init/2 is optional. This is all a big change and while most of the tests pass, some functionality currently doesn't. SPDY is broken and will be removed soon in favor of HTTP/2. Automatic compression is currently disabled. The cowboy_req interface probably still have a few functions that need to be updated. The docs and examples do not refer the current functionality anymore. Everything will be fixed over time. Feedback is more than welcome. Open a ticket!
2016-02-10 17:28:32 +01:00
case maps:is_key(multipart, Req) of
true ->
{Data, Req2} = stream_multipart(Req, Opts),
part(Data, Opts, Req2);
false ->
part(init_multipart(Req), Opts)
end.
Add request body reading options The options were added to allow developers to fix timeout issues when reading large bodies. It is also a cleaner and easier to extend interface. This commit deprecates the functions init_stream, stream_body and skip_body which are no longer needed. They will be removed in 1.0. The body function can now take an additional argument that is a list of options. The body_qs, part and part_body functions can too and simply pass this argument down to the body call. There are options for disabling the automatic continue reply, setting a maximum length to be returned (soft limit), setting the read length and read timeout, and setting the transfer and content decode functions. The return value of the body and body_qs have changed slightly. The body function now works similarly to the part_body function, in that it returns either an ok or a more tuple depending on whether there is additional data to be read. The body_qs function can return a badlength tuple if the body is too big. The default size has been increased from 16KB to 64KB. The default read length and timeout have been tweaked and vary depending on the function called. The body function will now adequately process chunked bodies, which means that the body_qs function will too. But this means that the behavior has changed slightly and your code should be tested properly when updating your code. The body and body_qs still accept a length as first argument for compatibility purpose with older code. Note that this form is deprecated and will be removed in 1.0. The part and part_body function, being new and never having been in a release yet, have this form completely removed in this commit. Again, while most code should work as-is, you should make sure that it actually does before pushing this to production.
2014-06-02 23:09:43 +02:00
Initial commit with connection/streams Breaking changes with previous commit. This is a very large change, and I am giving up on making a single commit that fixes everything. More commits will follow slowly adding back features, introducing new tests and fixing the documentation. This change contains most of the work toward unifying the interface for handling both HTTP/1.1 and HTTP/2. HTTP/1.1 connections are now no longer 1 process per connection; instead by default 1 process per request is also created. This has a number of pros and cons. Because it has cons, we also allow users to use a lower-level API that acts on "streams" (requests/responses) directly at the connection process-level. If performance is a concern, one can always write a stream handler. The performance in this case will be even greater than with Cowboy 1, although all the special handlers are unavailable. When switching to Websocket, after the handler returns from init/2, Cowboy stops the stream and the Websocket protocol takes over the connection process. Websocket then calls websocket_init/2 for any additional initialization such as timers, because the process is different in init/2 and websocket_*/* functions. This however would allow us to use websocket_init/2 for sending messages on connect, instead of sending ourselves a message and be subject to races. Note that websocket_init/2 is optional. This is all a big change and while most of the tests pass, some functionality currently doesn't. SPDY is broken and will be removed soon in favor of HTTP/2. Automatic compression is currently disabled. The cowboy_req interface probably still have a few functions that need to be updated. The docs and examples do not refer the current functionality anymore. Everything will be fixed over time. Feedback is more than welcome. Open a ticket!
2016-02-10 17:28:32 +01:00
part(Buffer, Opts, Req=#{multipart := {Boundary, _}}) ->
Add and document the new multipart code The old undocumented API is removed entirely. While a documentation exists for the new API, it will not be considered set in stone until further testing has been performed, and a file upload example has been added. The new API should be a little more efficient than the old API, especially with smaller messages.
2014-02-06 19:36:25 +01:00
case cow_multipart:parse_headers(Buffer, Boundary) of
more ->
Add request body reading options The options were added to allow developers to fix timeout issues when reading large bodies. It is also a cleaner and easier to extend interface. This commit deprecates the functions init_stream, stream_body and skip_body which are no longer needed. They will be removed in 1.0. The body function can now take an additional argument that is a list of options. The body_qs, part and part_body functions can too and simply pass this argument down to the body call. There are options for disabling the automatic continue reply, setting a maximum length to be returned (soft limit), setting the read length and read timeout, and setting the transfer and content decode functions. The return value of the body and body_qs have changed slightly. The body function now works similarly to the part_body function, in that it returns either an ok or a more tuple depending on whether there is additional data to be read. The body_qs function can return a badlength tuple if the body is too big. The default size has been increased from 16KB to 64KB. The default read length and timeout have been tweaked and vary depending on the function called. The body function will now adequately process chunked bodies, which means that the body_qs function will too. But this means that the behavior has changed slightly and your code should be tested properly when updating your code. The body and body_qs still accept a length as first argument for compatibility purpose with older code. Note that this form is deprecated and will be removed in 1.0. The part and part_body function, being new and never having been in a release yet, have this form completely removed in this commit. Again, while most code should work as-is, you should make sure that it actually does before pushing this to production.
2014-06-02 23:09:43 +02:00
{Data, Req2} = stream_multipart(Req, Opts),
part(<< Buffer/binary, Data/binary >>, Opts, Req2);
Add and document the new multipart code The old undocumented API is removed entirely. While a documentation exists for the new API, it will not be considered set in stone until further testing has been performed, and a file upload example has been added. The new API should be a little more efficient than the old API, especially with smaller messages.
2014-02-06 19:36:25 +01:00
{more, Buffer2} ->
Add request body reading options The options were added to allow developers to fix timeout issues when reading large bodies. It is also a cleaner and easier to extend interface. This commit deprecates the functions init_stream, stream_body and skip_body which are no longer needed. They will be removed in 1.0. The body function can now take an additional argument that is a list of options. The body_qs, part and part_body functions can too and simply pass this argument down to the body call. There are options for disabling the automatic continue reply, setting a maximum length to be returned (soft limit), setting the read length and read timeout, and setting the transfer and content decode functions. The return value of the body and body_qs have changed slightly. The body function now works similarly to the part_body function, in that it returns either an ok or a more tuple depending on whether there is additional data to be read. The body_qs function can return a badlength tuple if the body is too big. The default size has been increased from 16KB to 64KB. The default read length and timeout have been tweaked and vary depending on the function called. The body function will now adequately process chunked bodies, which means that the body_qs function will too. But this means that the behavior has changed slightly and your code should be tested properly when updating your code. The body and body_qs still accept a length as first argument for compatibility purpose with older code. Note that this form is deprecated and will be removed in 1.0. The part and part_body function, being new and never having been in a release yet, have this form completely removed in this commit. Again, while most code should work as-is, you should make sure that it actually does before pushing this to production.
2014-06-02 23:09:43 +02:00
{Data, Req2} = stream_multipart(Req, Opts),
part(<< Buffer2/binary, Data/binary >>, Opts, Req2);
Add and document the new multipart code The old undocumented API is removed entirely. While a documentation exists for the new API, it will not be considered set in stone until further testing has been performed, and a file upload example has been added. The new API should be a little more efficient than the old API, especially with smaller messages.
2014-02-06 19:36:25 +01:00
{ok, Headers, Rest} ->
Initial commit with connection/streams Breaking changes with previous commit. This is a very large change, and I am giving up on making a single commit that fixes everything. More commits will follow slowly adding back features, introducing new tests and fixing the documentation. This change contains most of the work toward unifying the interface for handling both HTTP/1.1 and HTTP/2. HTTP/1.1 connections are now no longer 1 process per connection; instead by default 1 process per request is also created. This has a number of pros and cons. Because it has cons, we also allow users to use a lower-level API that acts on "streams" (requests/responses) directly at the connection process-level. If performance is a concern, one can always write a stream handler. The performance in this case will be even greater than with Cowboy 1, although all the special handlers are unavailable. When switching to Websocket, after the handler returns from init/2, Cowboy stops the stream and the Websocket protocol takes over the connection process. Websocket then calls websocket_init/2 for any additional initialization such as timers, because the process is different in init/2 and websocket_*/* functions. This however would allow us to use websocket_init/2 for sending messages on connect, instead of sending ourselves a message and be subject to races. Note that websocket_init/2 is optional. This is all a big change and while most of the tests pass, some functionality currently doesn't. SPDY is broken and will be removed soon in favor of HTTP/2. Automatic compression is currently disabled. The cowboy_req interface probably still have a few functions that need to be updated. The docs and examples do not refer the current functionality anymore. Everything will be fixed over time. Feedback is more than welcome. Open a ticket!
2016-02-10 17:28:32 +01:00
{ok, Headers, Req#{multipart => {Boundary, Rest}}};
Add and document the new multipart code The old undocumented API is removed entirely. While a documentation exists for the new API, it will not be considered set in stone until further testing has been performed, and a file upload example has been added. The new API should be a little more efficient than the old API, especially with smaller messages.
2014-02-06 19:36:25 +01:00
%% Ignore epilogue.
{done, _} ->
Initial commit with connection/streams Breaking changes with previous commit. This is a very large change, and I am giving up on making a single commit that fixes everything. More commits will follow slowly adding back features, introducing new tests and fixing the documentation. This change contains most of the work toward unifying the interface for handling both HTTP/1.1 and HTTP/2. HTTP/1.1 connections are now no longer 1 process per connection; instead by default 1 process per request is also created. This has a number of pros and cons. Because it has cons, we also allow users to use a lower-level API that acts on "streams" (requests/responses) directly at the connection process-level. If performance is a concern, one can always write a stream handler. The performance in this case will be even greater than with Cowboy 1, although all the special handlers are unavailable. When switching to Websocket, after the handler returns from init/2, Cowboy stops the stream and the Websocket protocol takes over the connection process. Websocket then calls websocket_init/2 for any additional initialization such as timers, because the process is different in init/2 and websocket_*/* functions. This however would allow us to use websocket_init/2 for sending messages on connect, instead of sending ourselves a message and be subject to races. Note that websocket_init/2 is optional. This is all a big change and while most of the tests pass, some functionality currently doesn't. SPDY is broken and will be removed soon in favor of HTTP/2. Automatic compression is currently disabled. The cowboy_req interface probably still have a few functions that need to be updated. The docs and examples do not refer the current functionality anymore. Everything will be fixed over time. Feedback is more than welcome. Open a ticket!
2016-02-10 17:28:32 +01:00
{done, Req#{multipart => done}}
Add multipart support
2011-11-08 00:51:49 +01:00
end.
Add and document the new multipart code The old undocumented API is removed entirely. While a documentation exists for the new API, it will not be considered set in stone until further testing has been performed, and a file upload example has been added. The new API should be a little more efficient than the old API, especially with smaller messages.
2014-02-06 19:36:25 +01:00
-spec part_body(Req)
-> {ok, binary(), Req} | {more, binary(), Req}
when Req::req().
part_body(Req) ->
Add request body reading options The options were added to allow developers to fix timeout issues when reading large bodies. It is also a cleaner and easier to extend interface. This commit deprecates the functions init_stream, stream_body and skip_body which are no longer needed. They will be removed in 1.0. The body function can now take an additional argument that is a list of options. The body_qs, part and part_body functions can too and simply pass this argument down to the body call. There are options for disabling the automatic continue reply, setting a maximum length to be returned (soft limit), setting the read length and read timeout, and setting the transfer and content decode functions. The return value of the body and body_qs have changed slightly. The body function now works similarly to the part_body function, in that it returns either an ok or a more tuple depending on whether there is additional data to be read. The body_qs function can return a badlength tuple if the body is too big. The default size has been increased from 16KB to 64KB. The default read length and timeout have been tweaked and vary depending on the function called. The body function will now adequately process chunked bodies, which means that the body_qs function will too. But this means that the behavior has changed slightly and your code should be tested properly when updating your code. The body and body_qs still accept a length as first argument for compatibility purpose with older code. Note that this form is deprecated and will be removed in 1.0. The part and part_body function, being new and never having been in a release yet, have this form completely removed in this commit. Again, while most code should work as-is, you should make sure that it actually does before pushing this to production.
2014-06-02 23:09:43 +02:00
part_body(Req, []).
Add and document the new multipart code The old undocumented API is removed entirely. While a documentation exists for the new API, it will not be considered set in stone until further testing has been performed, and a file upload example has been added. The new API should be a little more efficient than the old API, especially with smaller messages.
2014-02-06 19:36:25 +01:00
Add request body reading options The options were added to allow developers to fix timeout issues when reading large bodies. It is also a cleaner and easier to extend interface. This commit deprecates the functions init_stream, stream_body and skip_body which are no longer needed. They will be removed in 1.0. The body function can now take an additional argument that is a list of options. The body_qs, part and part_body functions can too and simply pass this argument down to the body call. There are options for disabling the automatic continue reply, setting a maximum length to be returned (soft limit), setting the read length and read timeout, and setting the transfer and content decode functions. The return value of the body and body_qs have changed slightly. The body function now works similarly to the part_body function, in that it returns either an ok or a more tuple depending on whether there is additional data to be read. The body_qs function can return a badlength tuple if the body is too big. The default size has been increased from 16KB to 64KB. The default read length and timeout have been tweaked and vary depending on the function called. The body function will now adequately process chunked bodies, which means that the body_qs function will too. But this means that the behavior has changed slightly and your code should be tested properly when updating your code. The body and body_qs still accept a length as first argument for compatibility purpose with older code. Note that this form is deprecated and will be removed in 1.0. The part and part_body function, being new and never having been in a release yet, have this form completely removed in this commit. Again, while most code should work as-is, you should make sure that it actually does before pushing this to production.
2014-06-02 23:09:43 +02:00
-spec part_body(Req, body_opts())
Add and document the new multipart code The old undocumented API is removed entirely. While a documentation exists for the new API, it will not be considered set in stone until further testing has been performed, and a file upload example has been added. The new API should be a little more efficient than the old API, especially with smaller messages.
2014-02-06 19:36:25 +01:00
-> {ok, binary(), Req} | {more, binary(), Req}
when Req::req().
Add request body reading options The options were added to allow developers to fix timeout issues when reading large bodies. It is also a cleaner and easier to extend interface. This commit deprecates the functions init_stream, stream_body and skip_body which are no longer needed. They will be removed in 1.0. The body function can now take an additional argument that is a list of options. The body_qs, part and part_body functions can too and simply pass this argument down to the body call. There are options for disabling the automatic continue reply, setting a maximum length to be returned (soft limit), setting the read length and read timeout, and setting the transfer and content decode functions. The return value of the body and body_qs have changed slightly. The body function now works similarly to the part_body function, in that it returns either an ok or a more tuple depending on whether there is additional data to be read. The body_qs function can return a badlength tuple if the body is too big. The default size has been increased from 16KB to 64KB. The default read length and timeout have been tweaked and vary depending on the function called. The body function will now adequately process chunked bodies, which means that the body_qs function will too. But this means that the behavior has changed slightly and your code should be tested properly when updating your code. The body and body_qs still accept a length as first argument for compatibility purpose with older code. Note that this form is deprecated and will be removed in 1.0. The part and part_body function, being new and never having been in a release yet, have this form completely removed in this commit. Again, while most code should work as-is, you should make sure that it actually does before pushing this to production.
2014-06-02 23:09:43 +02:00
part_body(Req, Opts) ->
Initial commit with connection/streams Breaking changes with previous commit. This is a very large change, and I am giving up on making a single commit that fixes everything. More commits will follow slowly adding back features, introducing new tests and fixing the documentation. This change contains most of the work toward unifying the interface for handling both HTTP/1.1 and HTTP/2. HTTP/1.1 connections are now no longer 1 process per connection; instead by default 1 process per request is also created. This has a number of pros and cons. Because it has cons, we also allow users to use a lower-level API that acts on "streams" (requests/responses) directly at the connection process-level. If performance is a concern, one can always write a stream handler. The performance in this case will be even greater than with Cowboy 1, although all the special handlers are unavailable. When switching to Websocket, after the handler returns from init/2, Cowboy stops the stream and the Websocket protocol takes over the connection process. Websocket then calls websocket_init/2 for any additional initialization such as timers, because the process is different in init/2 and websocket_*/* functions. This however would allow us to use websocket_init/2 for sending messages on connect, instead of sending ourselves a message and be subject to races. Note that websocket_init/2 is optional. This is all a big change and while most of the tests pass, some functionality currently doesn't. SPDY is broken and will be removed soon in favor of HTTP/2. Automatic compression is currently disabled. The cowboy_req interface probably still have a few functions that need to be updated. The docs and examples do not refer the current functionality anymore. Everything will be fixed over time. Feedback is more than welcome. Open a ticket!
2016-02-10 17:28:32 +01:00
case maps:is_key(multipart, Req) of
true ->
part_body(<<>>, Opts, Req, <<>>);
false ->
part_body(init_multipart(Req), Opts)
end.
Add request body reading options The options were added to allow developers to fix timeout issues when reading large bodies. It is also a cleaner and easier to extend interface. This commit deprecates the functions init_stream, stream_body and skip_body which are no longer needed. They will be removed in 1.0. The body function can now take an additional argument that is a list of options. The body_qs, part and part_body functions can too and simply pass this argument down to the body call. There are options for disabling the automatic continue reply, setting a maximum length to be returned (soft limit), setting the read length and read timeout, and setting the transfer and content decode functions. The return value of the body and body_qs have changed slightly. The body function now works similarly to the part_body function, in that it returns either an ok or a more tuple depending on whether there is additional data to be read. The body_qs function can return a badlength tuple if the body is too big. The default size has been increased from 16KB to 64KB. The default read length and timeout have been tweaked and vary depending on the function called. The body function will now adequately process chunked bodies, which means that the body_qs function will too. But this means that the behavior has changed slightly and your code should be tested properly when updating your code. The body and body_qs still accept a length as first argument for compatibility purpose with older code. Note that this form is deprecated and will be removed in 1.0. The part and part_body function, being new and never having been in a release yet, have this form completely removed in this commit. Again, while most code should work as-is, you should make sure that it actually does before pushing this to production.
2014-06-02 23:09:43 +02:00
Initial commit with connection/streams Breaking changes with previous commit. This is a very large change, and I am giving up on making a single commit that fixes everything. More commits will follow slowly adding back features, introducing new tests and fixing the documentation. This change contains most of the work toward unifying the interface for handling both HTTP/1.1 and HTTP/2. HTTP/1.1 connections are now no longer 1 process per connection; instead by default 1 process per request is also created. This has a number of pros and cons. Because it has cons, we also allow users to use a lower-level API that acts on "streams" (requests/responses) directly at the connection process-level. If performance is a concern, one can always write a stream handler. The performance in this case will be even greater than with Cowboy 1, although all the special handlers are unavailable. When switching to Websocket, after the handler returns from init/2, Cowboy stops the stream and the Websocket protocol takes over the connection process. Websocket then calls websocket_init/2 for any additional initialization such as timers, because the process is different in init/2 and websocket_*/* functions. This however would allow us to use websocket_init/2 for sending messages on connect, instead of sending ourselves a message and be subject to races. Note that websocket_init/2 is optional. This is all a big change and while most of the tests pass, some functionality currently doesn't. SPDY is broken and will be removed soon in favor of HTTP/2. Automatic compression is currently disabled. The cowboy_req interface probably still have a few functions that need to be updated. The docs and examples do not refer the current functionality anymore. Everything will be fixed over time. Feedback is more than welcome. Open a ticket!
2016-02-10 17:28:32 +01:00
part_body(Buffer, Opts, Req=#{multipart := {Boundary, _}}, Acc) ->
Add request body reading options The options were added to allow developers to fix timeout issues when reading large bodies. It is also a cleaner and easier to extend interface. This commit deprecates the functions init_stream, stream_body and skip_body which are no longer needed. They will be removed in 1.0. The body function can now take an additional argument that is a list of options. The body_qs, part and part_body functions can too and simply pass this argument down to the body call. There are options for disabling the automatic continue reply, setting a maximum length to be returned (soft limit), setting the read length and read timeout, and setting the transfer and content decode functions. The return value of the body and body_qs have changed slightly. The body function now works similarly to the part_body function, in that it returns either an ok or a more tuple depending on whether there is additional data to be read. The body_qs function can return a badlength tuple if the body is too big. The default size has been increased from 16KB to 64KB. The default read length and timeout have been tweaked and vary depending on the function called. The body function will now adequately process chunked bodies, which means that the body_qs function will too. But this means that the behavior has changed slightly and your code should be tested properly when updating your code. The body and body_qs still accept a length as first argument for compatibility purpose with older code. Note that this form is deprecated and will be removed in 1.0. The part and part_body function, being new and never having been in a release yet, have this form completely removed in this commit. Again, while most code should work as-is, you should make sure that it actually does before pushing this to production.
2014-06-02 23:09:43 +02:00
ChunkLen = case lists:keyfind(length, 1, Opts) of
false -> 8000000;
{_, ChunkLen0} -> ChunkLen0
end,
case byte_size(Acc) > ChunkLen of
true ->
Initial commit with connection/streams Breaking changes with previous commit. This is a very large change, and I am giving up on making a single commit that fixes everything. More commits will follow slowly adding back features, introducing new tests and fixing the documentation. This change contains most of the work toward unifying the interface for handling both HTTP/1.1 and HTTP/2. HTTP/1.1 connections are now no longer 1 process per connection; instead by default 1 process per request is also created. This has a number of pros and cons. Because it has cons, we also allow users to use a lower-level API that acts on "streams" (requests/responses) directly at the connection process-level. If performance is a concern, one can always write a stream handler. The performance in this case will be even greater than with Cowboy 1, although all the special handlers are unavailable. When switching to Websocket, after the handler returns from init/2, Cowboy stops the stream and the Websocket protocol takes over the connection process. Websocket then calls websocket_init/2 for any additional initialization such as timers, because the process is different in init/2 and websocket_*/* functions. This however would allow us to use websocket_init/2 for sending messages on connect, instead of sending ourselves a message and be subject to races. Note that websocket_init/2 is optional. This is all a big change and while most of the tests pass, some functionality currently doesn't. SPDY is broken and will be removed soon in favor of HTTP/2. Automatic compression is currently disabled. The cowboy_req interface probably still have a few functions that need to be updated. The docs and examples do not refer the current functionality anymore. Everything will be fixed over time. Feedback is more than welcome. Open a ticket!
2016-02-10 17:28:32 +01:00
{more, Acc, Req#{multipart => {Boundary, Buffer}}};
Add request body reading options The options were added to allow developers to fix timeout issues when reading large bodies. It is also a cleaner and easier to extend interface. This commit deprecates the functions init_stream, stream_body and skip_body which are no longer needed. They will be removed in 1.0. The body function can now take an additional argument that is a list of options. The body_qs, part and part_body functions can too and simply pass this argument down to the body call. There are options for disabling the automatic continue reply, setting a maximum length to be returned (soft limit), setting the read length and read timeout, and setting the transfer and content decode functions. The return value of the body and body_qs have changed slightly. The body function now works similarly to the part_body function, in that it returns either an ok or a more tuple depending on whether there is additional data to be read. The body_qs function can return a badlength tuple if the body is too big. The default size has been increased from 16KB to 64KB. The default read length and timeout have been tweaked and vary depending on the function called. The body function will now adequately process chunked bodies, which means that the body_qs function will too. But this means that the behavior has changed slightly and your code should be tested properly when updating your code. The body and body_qs still accept a length as first argument for compatibility purpose with older code. Note that this form is deprecated and will be removed in 1.0. The part and part_body function, being new and never having been in a release yet, have this form completely removed in this commit. Again, while most code should work as-is, you should make sure that it actually does before pushing this to production.
2014-06-02 23:09:43 +02:00
false ->
{Data, Req2} = stream_multipart(Req, Opts),
case cow_multipart:parse_body(<< Buffer/binary, Data/binary >>, Boundary) of
{ok, Body} ->
part_body(<<>>, Opts, Req2, << Acc/binary, Body/binary >>);
{ok, Body, Rest} ->
part_body(Rest, Opts, Req2, << Acc/binary, Body/binary >>);
done ->
{ok, Acc, Req2};
{done, Body} ->
{ok, << Acc/binary, Body/binary >>, Req2};
{done, Body, Rest} ->
{ok, << Acc/binary, Body/binary >>,
Initial commit with connection/streams Breaking changes with previous commit. This is a very large change, and I am giving up on making a single commit that fixes everything. More commits will follow slowly adding back features, introducing new tests and fixing the documentation. This change contains most of the work toward unifying the interface for handling both HTTP/1.1 and HTTP/2. HTTP/1.1 connections are now no longer 1 process per connection; instead by default 1 process per request is also created. This has a number of pros and cons. Because it has cons, we also allow users to use a lower-level API that acts on "streams" (requests/responses) directly at the connection process-level. If performance is a concern, one can always write a stream handler. The performance in this case will be even greater than with Cowboy 1, although all the special handlers are unavailable. When switching to Websocket, after the handler returns from init/2, Cowboy stops the stream and the Websocket protocol takes over the connection process. Websocket then calls websocket_init/2 for any additional initialization such as timers, because the process is different in init/2 and websocket_*/* functions. This however would allow us to use websocket_init/2 for sending messages on connect, instead of sending ourselves a message and be subject to races. Note that websocket_init/2 is optional. This is all a big change and while most of the tests pass, some functionality currently doesn't. SPDY is broken and will be removed soon in favor of HTTP/2. Automatic compression is currently disabled. The cowboy_req interface probably still have a few functions that need to be updated. The docs and examples do not refer the current functionality anymore. Everything will be fixed over time. Feedback is more than welcome. Open a ticket!
2016-02-10 17:28:32 +01:00
Req2#{multipart => {Boundary, Rest}}}
Add request body reading options The options were added to allow developers to fix timeout issues when reading large bodies. It is also a cleaner and easier to extend interface. This commit deprecates the functions init_stream, stream_body and skip_body which are no longer needed. They will be removed in 1.0. The body function can now take an additional argument that is a list of options. The body_qs, part and part_body functions can too and simply pass this argument down to the body call. There are options for disabling the automatic continue reply, setting a maximum length to be returned (soft limit), setting the read length and read timeout, and setting the transfer and content decode functions. The return value of the body and body_qs have changed slightly. The body function now works similarly to the part_body function, in that it returns either an ok or a more tuple depending on whether there is additional data to be read. The body_qs function can return a badlength tuple if the body is too big. The default size has been increased from 16KB to 64KB. The default read length and timeout have been tweaked and vary depending on the function called. The body function will now adequately process chunked bodies, which means that the body_qs function will too. But this means that the behavior has changed slightly and your code should be tested properly when updating your code. The body and body_qs still accept a length as first argument for compatibility purpose with older code. Note that this form is deprecated and will be removed in 1.0. The part and part_body function, being new and never having been in a release yet, have this form completely removed in this commit. Again, while most code should work as-is, you should make sure that it actually does before pushing this to production.
2014-06-02 23:09:43 +02:00
end
Add multipart support
2011-11-08 00:51:49 +01:00
end.
Add and document the new multipart code The old undocumented API is removed entirely. While a documentation exists for the new API, it will not be considered set in stone until further testing has been performed, and a file upload example has been added. The new API should be a little more efficient than the old API, especially with smaller messages.
2014-02-06 19:36:25 +01:00
init_multipart(Req) ->
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
{<<"multipart">>, _, Params} = parse_header(<<"content-type">>, Req),
Add and document the new multipart code The old undocumented API is removed entirely. While a documentation exists for the new API, it will not be considered set in stone until further testing has been performed, and a file upload example has been added. The new API should be a little more efficient than the old API, especially with smaller messages.
2014-02-06 19:36:25 +01:00
{_, Boundary} = lists:keyfind(<<"boundary">>, 1, Params),
Initial commit with connection/streams Breaking changes with previous commit. This is a very large change, and I am giving up on making a single commit that fixes everything. More commits will follow slowly adding back features, introducing new tests and fixing the documentation. This change contains most of the work toward unifying the interface for handling both HTTP/1.1 and HTTP/2. HTTP/1.1 connections are now no longer 1 process per connection; instead by default 1 process per request is also created. This has a number of pros and cons. Because it has cons, we also allow users to use a lower-level API that acts on "streams" (requests/responses) directly at the connection process-level. If performance is a concern, one can always write a stream handler. The performance in this case will be even greater than with Cowboy 1, although all the special handlers are unavailable. When switching to Websocket, after the handler returns from init/2, Cowboy stops the stream and the Websocket protocol takes over the connection process. Websocket then calls websocket_init/2 for any additional initialization such as timers, because the process is different in init/2 and websocket_*/* functions. This however would allow us to use websocket_init/2 for sending messages on connect, instead of sending ourselves a message and be subject to races. Note that websocket_init/2 is optional. This is all a big change and while most of the tests pass, some functionality currently doesn't. SPDY is broken and will be removed soon in favor of HTTP/2. Automatic compression is currently disabled. The cowboy_req interface probably still have a few functions that need to be updated. The docs and examples do not refer the current functionality anymore. Everything will be fixed over time. Feedback is more than welcome. Open a ticket!
2016-02-10 17:28:32 +01:00
Req#{multipart => {Boundary, <<>>}}.
Add and document the new multipart code The old undocumented API is removed entirely. While a documentation exists for the new API, it will not be considered set in stone until further testing has been performed, and a file upload example has been added. The new API should be a little more efficient than the old API, especially with smaller messages.
2014-02-06 19:36:25 +01:00
Initial commit with connection/streams Breaking changes with previous commit. This is a very large change, and I am giving up on making a single commit that fixes everything. More commits will follow slowly adding back features, introducing new tests and fixing the documentation. This change contains most of the work toward unifying the interface for handling both HTTP/1.1 and HTTP/2. HTTP/1.1 connections are now no longer 1 process per connection; instead by default 1 process per request is also created. This has a number of pros and cons. Because it has cons, we also allow users to use a lower-level API that acts on "streams" (requests/responses) directly at the connection process-level. If performance is a concern, one can always write a stream handler. The performance in this case will be even greater than with Cowboy 1, although all the special handlers are unavailable. When switching to Websocket, after the handler returns from init/2, Cowboy stops the stream and the Websocket protocol takes over the connection process. Websocket then calls websocket_init/2 for any additional initialization such as timers, because the process is different in init/2 and websocket_*/* functions. This however would allow us to use websocket_init/2 for sending messages on connect, instead of sending ourselves a message and be subject to races. Note that websocket_init/2 is optional. This is all a big change and while most of the tests pass, some functionality currently doesn't. SPDY is broken and will be removed soon in favor of HTTP/2. Automatic compression is currently disabled. The cowboy_req interface probably still have a few functions that need to be updated. The docs and examples do not refer the current functionality anymore. Everything will be fixed over time. Feedback is more than welcome. Open a ticket!
2016-02-10 17:28:32 +01:00
stream_multipart(Req=#{multipart := done}, _) ->
{<<>>, Req};
stream_multipart(Req=#{multipart := {_, <<>>}}, Opts) ->
{_, Data, Req2} = read_body(Req, Opts),
Add request body reading options The options were added to allow developers to fix timeout issues when reading large bodies. It is also a cleaner and easier to extend interface. This commit deprecates the functions init_stream, stream_body and skip_body which are no longer needed. They will be removed in 1.0. The body function can now take an additional argument that is a list of options. The body_qs, part and part_body functions can too and simply pass this argument down to the body call. There are options for disabling the automatic continue reply, setting a maximum length to be returned (soft limit), setting the read length and read timeout, and setting the transfer and content decode functions. The return value of the body and body_qs have changed slightly. The body function now works similarly to the part_body function, in that it returns either an ok or a more tuple depending on whether there is additional data to be read. The body_qs function can return a badlength tuple if the body is too big. The default size has been increased from 16KB to 64KB. The default read length and timeout have been tweaked and vary depending on the function called. The body function will now adequately process chunked bodies, which means that the body_qs function will too. But this means that the behavior has changed slightly and your code should be tested properly when updating your code. The body and body_qs still accept a length as first argument for compatibility purpose with older code. Note that this form is deprecated and will be removed in 1.0. The part and part_body function, being new and never having been in a release yet, have this form completely removed in this commit. Again, while most code should work as-is, you should make sure that it actually does before pushing this to production.
2014-06-02 23:09:43 +02:00
{Data, Req2};
Initial commit with connection/streams Breaking changes with previous commit. This is a very large change, and I am giving up on making a single commit that fixes everything. More commits will follow slowly adding back features, introducing new tests and fixing the documentation. This change contains most of the work toward unifying the interface for handling both HTTP/1.1 and HTTP/2. HTTP/1.1 connections are now no longer 1 process per connection; instead by default 1 process per request is also created. This has a number of pros and cons. Because it has cons, we also allow users to use a lower-level API that acts on "streams" (requests/responses) directly at the connection process-level. If performance is a concern, one can always write a stream handler. The performance in this case will be even greater than with Cowboy 1, although all the special handlers are unavailable. When switching to Websocket, after the handler returns from init/2, Cowboy stops the stream and the Websocket protocol takes over the connection process. Websocket then calls websocket_init/2 for any additional initialization such as timers, because the process is different in init/2 and websocket_*/* functions. This however would allow us to use websocket_init/2 for sending messages on connect, instead of sending ourselves a message and be subject to races. Note that websocket_init/2 is optional. This is all a big change and while most of the tests pass, some functionality currently doesn't. SPDY is broken and will be removed soon in favor of HTTP/2. Automatic compression is currently disabled. The cowboy_req interface probably still have a few functions that need to be updated. The docs and examples do not refer the current functionality anymore. Everything will be fixed over time. Feedback is more than welcome. Open a ticket!
2016-02-10 17:28:32 +01:00
stream_multipart(Req=#{multipart := {Boundary, Buffer}}, _) ->
{Buffer, Req#{multipart => {Boundary, <<>>}}}.
Add and document the new multipart code The old undocumented API is removed entirely. While a documentation exists for the new API, it will not be considered set in stone until further testing has been performed, and a file upload example has been added. The new API should be a little more efficient than the old API, especially with smaller messages.
2014-02-06 19:36:25 +01:00
Move the reply function to cowboy_http_req.
2011-03-18 22:38:26 +01:00
%% Response API.
Fix cowboy_http:cookie_to_iodata/3 No more trying to quote, this is still completely broken everywhere.
2012-12-17 12:32:17 +01:00
%% The cookie name cannot contain any of the following characters:
%% =,;\s\t\r\n\013\014
%%
%% The cookie value cannot contain any of the following characters:
%% ,; \t\r\n\013\014
Make cookies use universal time instead of local time Includes: * cowboy_clock:rfc2109/1 now expects UTC datetime * Rewrite of the cookie code to cowboy_http * Removal of cowboy_cookies * Add type cowboy_req:cookie_opts/0 Cookies should now be set using cowboy_req:set_resp_cookie/3. Code calling cowboy_cookies directly will need to be updated.
2012-12-07 14:54:45 +01:00
-spec set_resp_cookie(iodata(), iodata(), cookie_opts(), Req)
-> Req when Req::req().
set_resp_cookie(Name, Value, Opts, Req) ->
Move cookie parsing/building code to cowlib The code for parsing has also been rewritten to be more efficient and to be able to handle cookie values with space inside them properly. Update cowlib to 0.2.0.
2013-10-23 11:21:31 +02:00
Cookie = cow_cookie:setcookie(Name, Value, Opts),
Initial commit with connection/streams Breaking changes with previous commit. This is a very large change, and I am giving up on making a single commit that fixes everything. More commits will follow slowly adding back features, introducing new tests and fixing the documentation. This change contains most of the work toward unifying the interface for handling both HTTP/1.1 and HTTP/2. HTTP/1.1 connections are now no longer 1 process per connection; instead by default 1 process per request is also created. This has a number of pros and cons. Because it has cons, we also allow users to use a lower-level API that acts on "streams" (requests/responses) directly at the connection process-level. If performance is a concern, one can always write a stream handler. The performance in this case will be even greater than with Cowboy 1, although all the special handlers are unavailable. When switching to Websocket, after the handler returns from init/2, Cowboy stops the stream and the Websocket protocol takes over the connection process. Websocket then calls websocket_init/2 for any additional initialization such as timers, because the process is different in init/2 and websocket_*/* functions. This however would allow us to use websocket_init/2 for sending messages on connect, instead of sending ourselves a message and be subject to races. Note that websocket_init/2 is optional. This is all a big change and while most of the tests pass, some functionality currently doesn't. SPDY is broken and will be removed soon in favor of HTTP/2. Automatic compression is currently disabled. The cowboy_req interface probably still have a few functions that need to be updated. The docs and examples do not refer the current functionality anymore. Everything will be fixed over time. Feedback is more than welcome. Open a ticket!
2016-02-10 17:28:32 +01:00
%% @todo Nah, keep separate.
Make cookies use universal time instead of local time Includes: * cowboy_clock:rfc2109/1 now expects UTC datetime * Rewrite of the cookie code to cowboy_http * Removal of cowboy_cookies * Add type cowboy_req:cookie_opts/0 Cookies should now be set using cowboy_req:set_resp_cookie/3. Code calling cowboy_cookies directly will need to be updated.
2012-12-07 14:54:45 +01:00
set_resp_header(<<"set-cookie">>, Cookie, Req).
Add cowboy_http_req:set_resp_cookie/4 Pretty much just an alias for a combination of set_resp_header and cookie.
2011-12-07 11:54:57 +01:00
Don't use decode_packet/3 for parsing the headers Header names are now binaries. Since header names are case insensitive they are all converted to lowercase. For example: <<"content-length">>. The max_line_length option was removed. Three new options have been added instead: * max_request_line_length (defaults to 4096) * max_header_name_length (defaults to 64) * max_header_value_length (defaults to 4096)
2012-09-21 09:18:56 +02:00
-spec set_resp_header(binary(), iodata(), Req)
Improve consistency of return types for cowboy_req API The signature of parse_header, body_qs, multipart_data and the set_resp_* functions has changed. See the cowboy_req module edoc for more details.
2012-09-16 03:51:07 +02:00
-> Req when Req::req().
Initial commit with connection/streams Breaking changes with previous commit. This is a very large change, and I am giving up on making a single commit that fixes everything. More commits will follow slowly adding back features, introducing new tests and fixing the documentation. This change contains most of the work toward unifying the interface for handling both HTTP/1.1 and HTTP/2. HTTP/1.1 connections are now no longer 1 process per connection; instead by default 1 process per request is also created. This has a number of pros and cons. Because it has cons, we also allow users to use a lower-level API that acts on "streams" (requests/responses) directly at the connection process-level. If performance is a concern, one can always write a stream handler. The performance in this case will be even greater than with Cowboy 1, although all the special handlers are unavailable. When switching to Websocket, after the handler returns from init/2, Cowboy stops the stream and the Websocket protocol takes over the connection process. Websocket then calls websocket_init/2 for any additional initialization such as timers, because the process is different in init/2 and websocket_*/* functions. This however would allow us to use websocket_init/2 for sending messages on connect, instead of sending ourselves a message and be subject to races. Note that websocket_init/2 is optional. This is all a big change and while most of the tests pass, some functionality currently doesn't. SPDY is broken and will be removed soon in favor of HTTP/2. Automatic compression is currently disabled. The cowboy_req interface probably still have a few functions that need to be updated. The docs and examples do not refer the current functionality anymore. Everything will be fixed over time. Feedback is more than welcome. Open a ticket!
2016-02-10 17:28:32 +01:00
set_resp_header(Name, Value, Req=#{resp_headers := RespHeaders}) ->
Req#{resp_headers => RespHeaders#{Name => Value}};
set_resp_header(Name,Value, Req) ->
Req#{resp_headers => #{Name => Value}}.
Add set_resp_header/3 and set_resp_body/2 to cowboy_http_req These functions allow to set response headers and body in advance, before calling any of the reply functions. Also add has_resp_header/2 and has_resp_body/1 to check if the given response headers have already been set.
2011-11-28 09:09:41 +01:00
Initial commit with connection/streams Breaking changes with previous commit. This is a very large change, and I am giving up on making a single commit that fixes everything. More commits will follow slowly adding back features, introducing new tests and fixing the documentation. This change contains most of the work toward unifying the interface for handling both HTTP/1.1 and HTTP/2. HTTP/1.1 connections are now no longer 1 process per connection; instead by default 1 process per request is also created. This has a number of pros and cons. Because it has cons, we also allow users to use a lower-level API that acts on "streams" (requests/responses) directly at the connection process-level. If performance is a concern, one can always write a stream handler. The performance in this case will be even greater than with Cowboy 1, although all the special handlers are unavailable. When switching to Websocket, after the handler returns from init/2, Cowboy stops the stream and the Websocket protocol takes over the connection process. Websocket then calls websocket_init/2 for any additional initialization such as timers, because the process is different in init/2 and websocket_*/* functions. This however would allow us to use websocket_init/2 for sending messages on connect, instead of sending ourselves a message and be subject to races. Note that websocket_init/2 is optional. This is all a big change and while most of the tests pass, some functionality currently doesn't. SPDY is broken and will be removed soon in favor of HTTP/2. Automatic compression is currently disabled. The cowboy_req interface probably still have a few functions that need to be updated. The docs and examples do not refer the current functionality anymore. Everything will be fixed over time. Feedback is more than welcome. Open a ticket!
2016-02-10 17:28:32 +01:00
%% @todo {sendfile, Offset, Bytes, Path} tuple
Improve consistency of return types for cowboy_req API The signature of parse_header, body_qs, multipart_data and the set_resp_* functions has changed. See the cowboy_req module edoc for more details.
2012-09-16 03:51:07 +02:00
-spec set_resp_body(iodata(), Req) -> Req when Req::req().
Add set_resp_header/3 and set_resp_body/2 to cowboy_http_req These functions allow to set response headers and body in advance, before calling any of the reply functions. Also add has_resp_header/2 and has_resp_body/1 to check if the given response headers have already been set.
2011-11-28 09:09:41 +01:00
set_resp_body(Body, Req) ->
Initial commit with connection/streams Breaking changes with previous commit. This is a very large change, and I am giving up on making a single commit that fixes everything. More commits will follow slowly adding back features, introducing new tests and fixing the documentation. This change contains most of the work toward unifying the interface for handling both HTTP/1.1 and HTTP/2. HTTP/1.1 connections are now no longer 1 process per connection; instead by default 1 process per request is also created. This has a number of pros and cons. Because it has cons, we also allow users to use a lower-level API that acts on "streams" (requests/responses) directly at the connection process-level. If performance is a concern, one can always write a stream handler. The performance in this case will be even greater than with Cowboy 1, although all the special handlers are unavailable. When switching to Websocket, after the handler returns from init/2, Cowboy stops the stream and the Websocket protocol takes over the connection process. Websocket then calls websocket_init/2 for any additional initialization such as timers, because the process is different in init/2 and websocket_*/* functions. This however would allow us to use websocket_init/2 for sending messages on connect, instead of sending ourselves a message and be subject to races. Note that websocket_init/2 is optional. This is all a big change and while most of the tests pass, some functionality currently doesn't. SPDY is broken and will be removed soon in favor of HTTP/2. Automatic compression is currently disabled. The cowboy_req interface probably still have a few functions that need to be updated. The docs and examples do not refer the current functionality anymore. Everything will be fixed over time. Feedback is more than welcome. Open a ticket!
2016-02-10 17:28:32 +01:00
Req#{resp_body => Body}.
%set_resp_body(Body, Req) ->
% Req#http_req{resp_body=Body}.
Add set_resp_header/3 and set_resp_body/2 to cowboy_http_req These functions allow to set response headers and body in advance, before calling any of the reply functions. Also add has_resp_header/2 and has_resp_body/1 to check if the given response headers have already been set.
2011-11-28 09:09:41 +01:00
Add cowboy_req:set_resp_body_fun/2 This allows streaming a body without knowing the length in advance. Also allows {stream, StreamFun} response body in the REST code.
2013-01-05 23:35:30 +01:00
-spec set_resp_body_fun(resp_body_fun(), Req) -> Req when Req::req().
add guards on set_resp_body_fun for streamlen and streamfun
2013-02-01 10:40:00 -06:00
set_resp_body_fun(StreamFun, Req) when is_function(StreamFun) ->
Add cowboy_req:set_resp_body_fun/2 This allows streaming a body without knowing the length in advance. Also allows {stream, StreamFun} response body in the REST code.
2013-01-05 23:35:30 +01:00
Req#http_req{resp_body=StreamFun}.
%% If the body function crashes while writing the response body or writes
%% fewer bytes than declared the behaviour is undefined.
Use a single Transport:send/2 call to send the reply if possible This gives a huge boost in performance when replies are small.
2012-09-26 10:08:43 +02:00
-spec set_resp_body_fun(non_neg_integer(), resp_body_fun(), Req)
Add chunked response body fun Adds a new type of streaming response fun. It can be set in a similar way to a streaming body fun with known length: Req2 = cowboy_req:set_resp_body_fun(chunked, StreamFun, Req) The fun, StreamFun, should accept a fun as its single argument. This fun, ChunkFun, is used to send chunks of iodata: ok = ChunkFun(IoData) ChunkFun should not be called with an empty binary or iolist as this will cause HTTP 1.1 clients to believe the stream is over. The final (0 length) chunk will be sent automatically - even if it has already been sent - assuming no exception is raised. Also note that the connection will close after the last chunk for HTTP 1.0 clients.
2013-02-18 21:20:36 +00:00
-> Req when Req::req();
(chunked, resp_chunked_fun(), Req)
Use a single Transport:send/2 call to send the reply if possible This gives a huge boost in performance when replies are small.
2012-09-26 10:08:43 +02:00
-> Req when Req::req().
Merge branch 'master' of git://github.com/tsloughter/cowboy
2013-02-06 18:35:07 +01:00
set_resp_body_fun(StreamLen, StreamFun, Req)
when is_integer(StreamLen), is_function(StreamFun) ->
Add chunked response body fun Adds a new type of streaming response fun. It can be set in a similar way to a streaming body fun with known length: Req2 = cowboy_req:set_resp_body_fun(chunked, StreamFun, Req) The fun, StreamFun, should accept a fun as its single argument. This fun, ChunkFun, is used to send chunks of iodata: ok = ChunkFun(IoData) ChunkFun should not be called with an empty binary or iolist as this will cause HTTP 1.1 clients to believe the stream is over. The final (0 length) chunk will be sent automatically - even if it has already been sent - assuming no exception is raised. Also note that the connection will close after the last chunk for HTTP 1.0 clients.
2013-02-18 21:20:36 +00:00
Req#http_req{resp_body={StreamLen, StreamFun}};
set_resp_body_fun(chunked, StreamFun, Req)
when is_function(StreamFun) ->
Req#http_req{resp_body={chunked, StreamFun}}.
Add cowboy_http_req:set_resp_body_fun/3.
2011-12-28 18:00:27 +01:00
Don't use decode_packet/3 for parsing the headers Header names are now binaries. Since header names are case insensitive they are all converted to lowercase. For example: <<"content-length">>. The max_line_length option was removed. Three new options have been added instead: * max_request_line_length (defaults to 4096) * max_header_name_length (defaults to 64) * max_header_value_length (defaults to 4096)
2012-09-21 09:18:56 +02:00
-spec has_resp_header(binary(), req()) -> boolean().
Initial commit with connection/streams Breaking changes with previous commit. This is a very large change, and I am giving up on making a single commit that fixes everything. More commits will follow slowly adding back features, introducing new tests and fixing the documentation. This change contains most of the work toward unifying the interface for handling both HTTP/1.1 and HTTP/2. HTTP/1.1 connections are now no longer 1 process per connection; instead by default 1 process per request is also created. This has a number of pros and cons. Because it has cons, we also allow users to use a lower-level API that acts on "streams" (requests/responses) directly at the connection process-level. If performance is a concern, one can always write a stream handler. The performance in this case will be even greater than with Cowboy 1, although all the special handlers are unavailable. When switching to Websocket, after the handler returns from init/2, Cowboy stops the stream and the Websocket protocol takes over the connection process. Websocket then calls websocket_init/2 for any additional initialization such as timers, because the process is different in init/2 and websocket_*/* functions. This however would allow us to use websocket_init/2 for sending messages on connect, instead of sending ourselves a message and be subject to races. Note that websocket_init/2 is optional. This is all a big change and while most of the tests pass, some functionality currently doesn't. SPDY is broken and will be removed soon in favor of HTTP/2. Automatic compression is currently disabled. The cowboy_req interface probably still have a few functions that need to be updated. The docs and examples do not refer the current functionality anymore. Everything will be fixed over time. Feedback is more than welcome. Open a ticket!
2016-02-10 17:28:32 +01:00
has_resp_header(Name, #{resp_headers := RespHeaders}) ->
maps:is_key(Name, RespHeaders);
has_resp_header(_, _) ->
false.
Add set_resp_header/3 and set_resp_body/2 to cowboy_http_req These functions allow to set response headers and body in advance, before calling any of the reply functions. Also add has_resp_header/2 and has_resp_body/1 to check if the given response headers have already been set.
2011-11-28 09:09:41 +01:00
Introduce a cowboy_req:req/0 type Cleanup the typespecs of the cowboy_req module following this.
2012-08-27 14:27:41 +02:00
-spec has_resp_body(req()) -> boolean().
Initial commit with connection/streams Breaking changes with previous commit. This is a very large change, and I am giving up on making a single commit that fixes everything. More commits will follow slowly adding back features, introducing new tests and fixing the documentation. This change contains most of the work toward unifying the interface for handling both HTTP/1.1 and HTTP/2. HTTP/1.1 connections are now no longer 1 process per connection; instead by default 1 process per request is also created. This has a number of pros and cons. Because it has cons, we also allow users to use a lower-level API that acts on "streams" (requests/responses) directly at the connection process-level. If performance is a concern, one can always write a stream handler. The performance in this case will be even greater than with Cowboy 1, although all the special handlers are unavailable. When switching to Websocket, after the handler returns from init/2, Cowboy stops the stream and the Websocket protocol takes over the connection process. Websocket then calls websocket_init/2 for any additional initialization such as timers, because the process is different in init/2 and websocket_*/* functions. This however would allow us to use websocket_init/2 for sending messages on connect, instead of sending ourselves a message and be subject to races. Note that websocket_init/2 is optional. This is all a big change and while most of the tests pass, some functionality currently doesn't. SPDY is broken and will be removed soon in favor of HTTP/2. Automatic compression is currently disabled. The cowboy_req interface probably still have a few functions that need to be updated. The docs and examples do not refer the current functionality anymore. Everything will be fixed over time. Feedback is more than welcome. Open a ticket!
2016-02-10 17:28:32 +01:00
has_resp_body(#{resp_body := {sendfile, Len, _}}) ->
Len > 0;
has_resp_body(#{resp_body := RespBody}) ->
iolist_size(RespBody) > 0;
has_resp_body(_) ->
false.
%has_resp_body(#http_req{resp_body=RespBody}) when is_function(RespBody) ->
% true;
%has_resp_body(#http_req{resp_body={chunked, _}}) ->
% true;
%has_resp_body(#http_req{resp_body={Length, _}}) ->
% Length > 0;
%has_resp_body(#http_req{resp_body=RespBody}) ->
% iolist_size(RespBody) > 0.
Add set_resp_header/3 and set_resp_body/2 to cowboy_http_req These functions allow to set response headers and body in advance, before calling any of the reply functions. Also add has_resp_header/2 and has_resp_body/1 to check if the given response headers have already been set.
2011-11-28 09:09:41 +01:00
Don't use decode_packet/3 for parsing the headers Header names are now binaries. Since header names are case insensitive they are all converted to lowercase. For example: <<"content-length">>. The max_line_length option was removed. Three new options have been added instead: * max_request_line_length (defaults to 4096) * max_header_name_length (defaults to 64) * max_header_value_length (defaults to 4096)
2012-09-21 09:18:56 +02:00
-spec delete_resp_header(binary(), Req)
Add cowboy_req:delete_resp_header/2 Both requested and better for making Req opaque.
2012-09-16 01:55:40 +02:00
-> Req when Req::req().
Initial commit with connection/streams Breaking changes with previous commit. This is a very large change, and I am giving up on making a single commit that fixes everything. More commits will follow slowly adding back features, introducing new tests and fixing the documentation. This change contains most of the work toward unifying the interface for handling both HTTP/1.1 and HTTP/2. HTTP/1.1 connections are now no longer 1 process per connection; instead by default 1 process per request is also created. This has a number of pros and cons. Because it has cons, we also allow users to use a lower-level API that acts on "streams" (requests/responses) directly at the connection process-level. If performance is a concern, one can always write a stream handler. The performance in this case will be even greater than with Cowboy 1, although all the special handlers are unavailable. When switching to Websocket, after the handler returns from init/2, Cowboy stops the stream and the Websocket protocol takes over the connection process. Websocket then calls websocket_init/2 for any additional initialization such as timers, because the process is different in init/2 and websocket_*/* functions. This however would allow us to use websocket_init/2 for sending messages on connect, instead of sending ourselves a message and be subject to races. Note that websocket_init/2 is optional. This is all a big change and while most of the tests pass, some functionality currently doesn't. SPDY is broken and will be removed soon in favor of HTTP/2. Automatic compression is currently disabled. The cowboy_req interface probably still have a few functions that need to be updated. The docs and examples do not refer the current functionality anymore. Everything will be fixed over time. Feedback is more than welcome. Open a ticket!
2016-02-10 17:28:32 +01:00
delete_resp_header(Name, Req=#{resp_headers := RespHeaders}) ->
Req#{resp_headers => maps:remove(Name, RespHeaders)}.
Add cowboy_req:delete_resp_header/2 Both requested and better for making Req opaque.
2012-09-16 01:55:40 +02:00
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
-spec reply(cowboy:http_status(), Req) -> Req when Req::req().
Initial commit with connection/streams Breaking changes with previous commit. This is a very large change, and I am giving up on making a single commit that fixes everything. More commits will follow slowly adding back features, introducing new tests and fixing the documentation. This change contains most of the work toward unifying the interface for handling both HTTP/1.1 and HTTP/2. HTTP/1.1 connections are now no longer 1 process per connection; instead by default 1 process per request is also created. This has a number of pros and cons. Because it has cons, we also allow users to use a lower-level API that acts on "streams" (requests/responses) directly at the connection process-level. If performance is a concern, one can always write a stream handler. The performance in this case will be even greater than with Cowboy 1, although all the special handlers are unavailable. When switching to Websocket, after the handler returns from init/2, Cowboy stops the stream and the Websocket protocol takes over the connection process. Websocket then calls websocket_init/2 for any additional initialization such as timers, because the process is different in init/2 and websocket_*/* functions. This however would allow us to use websocket_init/2 for sending messages on connect, instead of sending ourselves a message and be subject to races. Note that websocket_init/2 is optional. This is all a big change and while most of the tests pass, some functionality currently doesn't. SPDY is broken and will be removed soon in favor of HTTP/2. Automatic compression is currently disabled. The cowboy_req interface probably still have a few functions that need to be updated. The docs and examples do not refer the current functionality anymore. Everything will be fixed over time. Feedback is more than welcome. Open a ticket!
2016-02-10 17:28:32 +01:00
reply(Status, Req) ->
reply(Status, #{}, Req).
Add shortcuts to reply functions New functions are reply/2, reply/3, chunked_reply/2 in cowboy_http_req.
2011-10-13 16:16:53 +02:00
Move cowboy_http:status() to cowboy:http_status()
2013-05-16 16:29:24 +02:00
-spec reply(cowboy:http_status(), cowboy:http_headers(), Req)
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
-> Req when Req::req().
Initial commit with connection/streams Breaking changes with previous commit. This is a very large change, and I am giving up on making a single commit that fixes everything. More commits will follow slowly adding back features, introducing new tests and fixing the documentation. This change contains most of the work toward unifying the interface for handling both HTTP/1.1 and HTTP/2. HTTP/1.1 connections are now no longer 1 process per connection; instead by default 1 process per request is also created. This has a number of pros and cons. Because it has cons, we also allow users to use a lower-level API that acts on "streams" (requests/responses) directly at the connection process-level. If performance is a concern, one can always write a stream handler. The performance in this case will be even greater than with Cowboy 1, although all the special handlers are unavailable. When switching to Websocket, after the handler returns from init/2, Cowboy stops the stream and the Websocket protocol takes over the connection process. Websocket then calls websocket_init/2 for any additional initialization such as timers, because the process is different in init/2 and websocket_*/* functions. This however would allow us to use websocket_init/2 for sending messages on connect, instead of sending ourselves a message and be subject to races. Note that websocket_init/2 is optional. This is all a big change and while most of the tests pass, some functionality currently doesn't. SPDY is broken and will be removed soon in favor of HTTP/2. Automatic compression is currently disabled. The cowboy_req interface probably still have a few functions that need to be updated. The docs and examples do not refer the current functionality anymore. Everything will be fixed over time. Feedback is more than welcome. Open a ticket!
2016-02-10 17:28:32 +01:00
reply(Status, Headers, Req=#{resp_body := Body}) ->
reply(Status, Headers, Body, Req);
reply(Status, Headers, Req) ->
reply(Status, Headers, <<>>, Req).
Add shortcuts to reply functions New functions are reply/2, reply/3, chunked_reply/2 in cowboy_http_req.
2011-10-13 16:16:53 +02:00
Move cowboy_http:status() to cowboy:http_status()
2013-05-16 16:29:24 +02:00
-spec reply(cowboy:http_status(), cowboy:http_headers(),
Fix reply/4 spec There is wrong -spec of reply/4 function 3rd parametr type must be same as [#http_req spec](https://github.com/ninenines/cowboy/blob/master/src/cowboy_req.erl#L140)
2015-01-14 18:43:11 +03:00
iodata() | resp_body_fun() | {non_neg_integer(), resp_body_fun()}
| {chunked, resp_chunked_fun()}, Req)
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
-> Req when Req::req().
Initial commit with connection/streams Breaking changes with previous commit. This is a very large change, and I am giving up on making a single commit that fixes everything. More commits will follow slowly adding back features, introducing new tests and fixing the documentation. This change contains most of the work toward unifying the interface for handling both HTTP/1.1 and HTTP/2. HTTP/1.1 connections are now no longer 1 process per connection; instead by default 1 process per request is also created. This has a number of pros and cons. Because it has cons, we also allow users to use a lower-level API that acts on "streams" (requests/responses) directly at the connection process-level. If performance is a concern, one can always write a stream handler. The performance in this case will be even greater than with Cowboy 1, although all the special handlers are unavailable. When switching to Websocket, after the handler returns from init/2, Cowboy stops the stream and the Websocket protocol takes over the connection process. Websocket then calls websocket_init/2 for any additional initialization such as timers, because the process is different in init/2 and websocket_*/* functions. This however would allow us to use websocket_init/2 for sending messages on connect, instead of sending ourselves a message and be subject to races. Note that websocket_init/2 is optional. This is all a big change and while most of the tests pass, some functionality currently doesn't. SPDY is broken and will be removed soon in favor of HTTP/2. Automatic compression is currently disabled. The cowboy_req interface probably still have a few functions that need to be updated. The docs and examples do not refer the current functionality anymore. Everything will be fixed over time. Feedback is more than welcome. Open a ticket!
2016-02-10 17:28:32 +01:00
reply(Status, Headers, Stream = {stream, undefined, _}, Req) ->
do_stream_reply(Status, Headers, Stream, Req);
reply(Status, Headers, Stream = {stream, Len, _}, Req) ->
do_stream_reply(Status, Headers#{
<<"content-length">> => integer_to_binary(Len)
}, Stream, Req);
reply(Status, Headers, SendFile = {sendfile, _, Len, _}, Req) ->
do_reply(Status, Headers#{
<<"content-length">> => integer_to_binary(Len)
}, SendFile, Req);
reply(Status, Headers, Body, Req) ->
do_reply(Status, Headers#{
<<"content-length">> => integer_to_binary(iolist_size(Body))
}, Body, Req).
do_stream_reply(Status, Headers, {stream, _, Fun}, Req=#{pid := Pid, streamid := StreamID}) ->
Pid ! {{Pid, StreamID}, {headers, Status, response_headers(Headers, Req)}},
Fun(),
ok.
do_reply(Status, Headers, Body, Req=#{pid := Pid, streamid := StreamID}) ->
Pid ! {{Pid, StreamID}, {response, Status, response_headers(Headers, Req), Body}},
ok.
Fix warnings
2016-03-06 18:09:43 +01:00
-spec send_body(iodata(), fin | nofin, req()) -> ok.
Initial commit with connection/streams Breaking changes with previous commit. This is a very large change, and I am giving up on making a single commit that fixes everything. More commits will follow slowly adding back features, introducing new tests and fixing the documentation. This change contains most of the work toward unifying the interface for handling both HTTP/1.1 and HTTP/2. HTTP/1.1 connections are now no longer 1 process per connection; instead by default 1 process per request is also created. This has a number of pros and cons. Because it has cons, we also allow users to use a lower-level API that acts on "streams" (requests/responses) directly at the connection process-level. If performance is a concern, one can always write a stream handler. The performance in this case will be even greater than with Cowboy 1, although all the special handlers are unavailable. When switching to Websocket, after the handler returns from init/2, Cowboy stops the stream and the Websocket protocol takes over the connection process. Websocket then calls websocket_init/2 for any additional initialization such as timers, because the process is different in init/2 and websocket_*/* functions. This however would allow us to use websocket_init/2 for sending messages on connect, instead of sending ourselves a message and be subject to races. Note that websocket_init/2 is optional. This is all a big change and while most of the tests pass, some functionality currently doesn't. SPDY is broken and will be removed soon in favor of HTTP/2. Automatic compression is currently disabled. The cowboy_req interface probably still have a few functions that need to be updated. The docs and examples do not refer the current functionality anymore. Everything will be fixed over time. Feedback is more than welcome. Open a ticket!
2016-02-10 17:28:32 +01:00
send_body(Data, IsFin, #{pid := Pid, streamid := StreamID}) ->
Pid ! {{Pid, StreamID}, {data, IsFin, Data}},
ok.
Add optional automatic response body compression This behavior can be enabled with the `compress` protocol option. See the `compress_response` example for more details. All tests are now ran with and without compression for both HTTP and HTTPS.
2013-01-07 22:42:16 +01:00
Initial commit with connection/streams Breaking changes with previous commit. This is a very large change, and I am giving up on making a single commit that fixes everything. More commits will follow slowly adding back features, introducing new tests and fixing the documentation. This change contains most of the work toward unifying the interface for handling both HTTP/1.1 and HTTP/2. HTTP/1.1 connections are now no longer 1 process per connection; instead by default 1 process per request is also created. This has a number of pros and cons. Because it has cons, we also allow users to use a lower-level API that acts on "streams" (requests/responses) directly at the connection process-level. If performance is a concern, one can always write a stream handler. The performance in this case will be even greater than with Cowboy 1, although all the special handlers are unavailable. When switching to Websocket, after the handler returns from init/2, Cowboy stops the stream and the Websocket protocol takes over the connection process. Websocket then calls websocket_init/2 for any additional initialization such as timers, because the process is different in init/2 and websocket_*/* functions. This however would allow us to use websocket_init/2 for sending messages on connect, instead of sending ourselves a message and be subject to races. Note that websocket_init/2 is optional. This is all a big change and while most of the tests pass, some functionality currently doesn't. SPDY is broken and will be removed soon in favor of HTTP/2. Automatic compression is currently disabled. The cowboy_req interface probably still have a few functions that need to be updated. The docs and examples do not refer the current functionality anymore. Everything will be fixed over time. Feedback is more than welcome. Open a ticket!
2016-02-10 17:28:32 +01:00
response_headers(Headers, Req) ->
RespHeaders = maps:get(resp_headers, Req, #{}),
maps:merge(#{
<<"date">> => cowboy_clock:rfc1123(),
<<"server">> => <<"Cowboy">>
}, maps:merge(RespHeaders, Headers)).
%reply(Status, Headers, Body, Req=#http_req{
% socket=Socket, transport=Transport,
% version=Version, connection=Connection,
% method=Method, resp_compress=Compress,
% resp_state=RespState, resp_headers=RespHeaders})
% when RespState =:= waiting; RespState =:= waiting_stream ->
% Req3 = case Body of
% BodyFun when is_function(BodyFun) ->
% %% We stream the response body until we close the connection.
% RespConn = close,
% {RespType, Req2} = if
% true ->
% response(Status, Headers, RespHeaders, [
% {<<"connection">>, <<"close">>},
% {<<"date">>, cowboy_clock:rfc1123()},
% {<<"server">>, <<"Cowboy">>},
% {<<"transfer-encoding">>, <<"identity">>}
% ], <<>>, Req)
% end,
% if RespType =/= hook, Method =/= <<"HEAD">> ->
% BodyFun(Socket, Transport);
% true -> ok
% end,
% Req2#http_req{connection=RespConn};
% {chunked, BodyFun} ->
% %% We stream the response body in chunks.
% {RespType, Req2} = chunked_response(Status, Headers, Req),
% if RespType =/= hook, Method =/= <<"HEAD">> ->
% ChunkFun = fun(IoData) -> chunk(IoData, Req2) end,
% BodyFun(ChunkFun),
% %% Send the last chunk if chunked encoding was used.
% if
% Version =:= 'HTTP/1.0'; RespState =:= waiting_stream ->
% Req2;
% true ->
% last_chunk(Req2)
% end;
% true -> Req2
% end;
% {ContentLength, BodyFun} ->
% %% We stream the response body for ContentLength bytes.
% RespConn = response_connection(Headers, Connection),
% {RespType, Req2} = response(Status, Headers, RespHeaders, [
% {<<"content-length">>, integer_to_list(ContentLength)},
% {<<"date">>, cowboy_clock:rfc1123()},
% {<<"server">>, <<"Cowboy">>}
% |HTTP11Headers], stream, Req),
% if RespType =/= hook, Method =/= <<"HEAD">> ->
% BodyFun(Socket, Transport);
% true -> ok
% end,
% Req2#http_req{connection=RespConn};
% _ when Compress ->
% RespConn = response_connection(Headers, Connection),
% Req2 = reply_may_compress(Status, Headers, Body, Req,
% RespHeaders, HTTP11Headers, Method),
% Req2#http_req{connection=RespConn};
% _ ->
% RespConn = response_connection(Headers, Connection),
% Req2 = reply_no_compress(Status, Headers, Body, Req,
% RespHeaders, HTTP11Headers, Method, iolist_size(Body)),
% Req2#http_req{connection=RespConn}
% end,
% Req3#http_req{resp_state=done, resp_headers=[], resp_body= <<>>}.
%reply_may_compress(Status, Headers, Body, Req,
% RespHeaders, HTTP11Headers, Method) ->
% BodySize = iolist_size(Body),
% try parse_header(<<"accept-encoding">>, Req) of
% Encodings ->
% CanGzip = (BodySize > 300)
% andalso (false =:= lists:keyfind(<<"content-encoding">>,
% 1, Headers))
% andalso (false =:= lists:keyfind(<<"content-encoding">>,
% 1, RespHeaders))
% andalso (false =:= lists:keyfind(<<"transfer-encoding">>,
% 1, Headers))
% andalso (false =:= lists:keyfind(<<"transfer-encoding">>,
% 1, RespHeaders))
% andalso (Encodings =/= undefined)
% andalso (false =/= lists:keyfind(<<"gzip">>, 1, Encodings)),
% case CanGzip of
% true ->
% GzBody = zlib:gzip(Body),
% {_, Req2} = response(Status, Headers, RespHeaders, [
% {<<"content-length">>, integer_to_list(byte_size(GzBody))},
% {<<"content-encoding">>, <<"gzip">>},
% |HTTP11Headers],
% case Method of <<"HEAD">> -> <<>>; _ -> GzBody end,
% Req),
% Req2;
% false ->
% reply_no_compress(Status, Headers, Body, Req,
% RespHeaders, HTTP11Headers, Method, BodySize)
% end
% catch _:_ ->
% reply_no_compress(Status, Headers, Body, Req,
% RespHeaders, HTTP11Headers, Method, BodySize)
% end.
%
%reply_no_compress(Status, Headers, Body, Req,
% RespHeaders, HTTP11Headers, Method, BodySize) ->
% {_, Req2} = response(Status, Headers, RespHeaders, [
% {<<"content-length">>, integer_to_list(BodySize)},
% |HTTP11Headers],
% case Method of <<"HEAD">> -> <<>>; _ -> Body end,
% Req),
% Req2.
Add optional automatic response body compression This behavior can be enabled with the `compress` protocol option. See the `compress_response` example for more details. All tests are now ran with and without compression for both HTTP and HTTPS.
2013-01-07 22:42:16 +01:00
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
-spec chunked_reply(cowboy:http_status(), Req) -> Req when Req::req().
Add shortcuts to reply functions New functions are reply/2, reply/3, chunked_reply/2 in cowboy_http_req.
2011-10-13 16:16:53 +02:00
chunked_reply(Status, Req) ->
Initial commit with connection/streams Breaking changes with previous commit. This is a very large change, and I am giving up on making a single commit that fixes everything. More commits will follow slowly adding back features, introducing new tests and fixing the documentation. This change contains most of the work toward unifying the interface for handling both HTTP/1.1 and HTTP/2. HTTP/1.1 connections are now no longer 1 process per connection; instead by default 1 process per request is also created. This has a number of pros and cons. Because it has cons, we also allow users to use a lower-level API that acts on "streams" (requests/responses) directly at the connection process-level. If performance is a concern, one can always write a stream handler. The performance in this case will be even greater than with Cowboy 1, although all the special handlers are unavailable. When switching to Websocket, after the handler returns from init/2, Cowboy stops the stream and the Websocket protocol takes over the connection process. Websocket then calls websocket_init/2 for any additional initialization such as timers, because the process is different in init/2 and websocket_*/* functions. This however would allow us to use websocket_init/2 for sending messages on connect, instead of sending ourselves a message and be subject to races. Note that websocket_init/2 is optional. This is all a big change and while most of the tests pass, some functionality currently doesn't. SPDY is broken and will be removed soon in favor of HTTP/2. Automatic compression is currently disabled. The cowboy_req interface probably still have a few functions that need to be updated. The docs and examples do not refer the current functionality anymore. Everything will be fixed over time. Feedback is more than welcome. Open a ticket!
2016-02-10 17:28:32 +01:00
chunked_reply(Status, #{}, Req).
Add shortcuts to reply functions New functions are reply/2, reply/3, chunked_reply/2 in cowboy_http_req.
2011-10-13 16:16:53 +02:00
Move cowboy_http:status() to cowboy:http_status()
2013-05-16 16:29:24 +02:00
-spec chunked_reply(cowboy:http_status(), cowboy:http_headers(), Req)
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
-> Req when Req::req().
Initial commit with connection/streams Breaking changes with previous commit. This is a very large change, and I am giving up on making a single commit that fixes everything. More commits will follow slowly adding back features, introducing new tests and fixing the documentation. This change contains most of the work toward unifying the interface for handling both HTTP/1.1 and HTTP/2. HTTP/1.1 connections are now no longer 1 process per connection; instead by default 1 process per request is also created. This has a number of pros and cons. Because it has cons, we also allow users to use a lower-level API that acts on "streams" (requests/responses) directly at the connection process-level. If performance is a concern, one can always write a stream handler. The performance in this case will be even greater than with Cowboy 1, although all the special handlers are unavailable. When switching to Websocket, after the handler returns from init/2, Cowboy stops the stream and the Websocket protocol takes over the connection process. Websocket then calls websocket_init/2 for any additional initialization such as timers, because the process is different in init/2 and websocket_*/* functions. This however would allow us to use websocket_init/2 for sending messages on connect, instead of sending ourselves a message and be subject to races. Note that websocket_init/2 is optional. This is all a big change and while most of the tests pass, some functionality currently doesn't. SPDY is broken and will be removed soon in favor of HTTP/2. Automatic compression is currently disabled. The cowboy_req interface probably still have a few functions that need to be updated. The docs and examples do not refer the current functionality anymore. Everything will be fixed over time. Feedback is more than welcome. Open a ticket!
2016-02-10 17:28:32 +01:00
chunked_reply(Status, Headers, Req=#{pid := Pid, streamid := StreamID}) ->
Pid ! {{Pid, StreamID}, {headers, Status, response_headers(Headers, Req)}},
Req. %% @todo return ok
% ok.
Add chunked reply support. Send the status line and headers using cowboy_http_req:chunked_reply/3, and individual chunks with cowboy_http_req:chunk/2.
2011-05-08 17:26:21 +02:00
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
-spec chunk(iodata(), req()) -> ok.
Initial commit with connection/streams Breaking changes with previous commit. This is a very large change, and I am giving up on making a single commit that fixes everything. More commits will follow slowly adding back features, introducing new tests and fixing the documentation. This change contains most of the work toward unifying the interface for handling both HTTP/1.1 and HTTP/2. HTTP/1.1 connections are now no longer 1 process per connection; instead by default 1 process per request is also created. This has a number of pros and cons. Because it has cons, we also allow users to use a lower-level API that acts on "streams" (requests/responses) directly at the connection process-level. If performance is a concern, one can always write a stream handler. The performance in this case will be even greater than with Cowboy 1, although all the special handlers are unavailable. When switching to Websocket, after the handler returns from init/2, Cowboy stops the stream and the Websocket protocol takes over the connection process. Websocket then calls websocket_init/2 for any additional initialization such as timers, because the process is different in init/2 and websocket_*/* functions. This however would allow us to use websocket_init/2 for sending messages on connect, instead of sending ourselves a message and be subject to races. Note that websocket_init/2 is optional. This is all a big change and while most of the tests pass, some functionality currently doesn't. SPDY is broken and will be removed soon in favor of HTTP/2. Automatic compression is currently disabled. The cowboy_req interface probably still have a few functions that need to be updated. The docs and examples do not refer the current functionality anymore. Everything will be fixed over time. Feedback is more than welcome. Open a ticket!
2016-02-10 17:28:32 +01:00
chunk(_Data, #{method := <<"HEAD">>}) ->
Fix the handling of the HEAD requests Responses to the HEAD requests used to include an response body.
2011-09-06 12:11:44 +02:00
ok;
Initial commit with connection/streams Breaking changes with previous commit. This is a very large change, and I am giving up on making a single commit that fixes everything. More commits will follow slowly adding back features, introducing new tests and fixing the documentation. This change contains most of the work toward unifying the interface for handling both HTTP/1.1 and HTTP/2. HTTP/1.1 connections are now no longer 1 process per connection; instead by default 1 process per request is also created. This has a number of pros and cons. Because it has cons, we also allow users to use a lower-level API that acts on "streams" (requests/responses) directly at the connection process-level. If performance is a concern, one can always write a stream handler. The performance in this case will be even greater than with Cowboy 1, although all the special handlers are unavailable. When switching to Websocket, after the handler returns from init/2, Cowboy stops the stream and the Websocket protocol takes over the connection process. Websocket then calls websocket_init/2 for any additional initialization such as timers, because the process is different in init/2 and websocket_*/* functions. This however would allow us to use websocket_init/2 for sending messages on connect, instead of sending ourselves a message and be subject to races. Note that websocket_init/2 is optional. This is all a big change and while most of the tests pass, some functionality currently doesn't. SPDY is broken and will be removed soon in favor of HTTP/2. Automatic compression is currently disabled. The cowboy_req interface probably still have a few functions that need to be updated. The docs and examples do not refer the current functionality anymore. Everything will be fixed over time. Feedback is more than welcome. Open a ticket!
2016-02-10 17:28:32 +01:00
chunk(Data, #{pid := Pid, streamid := StreamID}) ->
Do not send empty chunks User code may sometimes send an empty value which gets understood by the client as being the end of the stream while this was not intended. Ignoring empty values allow making sure the stream isn't ended by mistake.
2015-02-16 19:49:01 +01:00
case iolist_size(Data) of
0 -> ok;
Initial commit with connection/streams Breaking changes with previous commit. This is a very large change, and I am giving up on making a single commit that fixes everything. More commits will follow slowly adding back features, introducing new tests and fixing the documentation. This change contains most of the work toward unifying the interface for handling both HTTP/1.1 and HTTP/2. HTTP/1.1 connections are now no longer 1 process per connection; instead by default 1 process per request is also created. This has a number of pros and cons. Because it has cons, we also allow users to use a lower-level API that acts on "streams" (requests/responses) directly at the connection process-level. If performance is a concern, one can always write a stream handler. The performance in this case will be even greater than with Cowboy 1, although all the special handlers are unavailable. When switching to Websocket, after the handler returns from init/2, Cowboy stops the stream and the Websocket protocol takes over the connection process. Websocket then calls websocket_init/2 for any additional initialization such as timers, because the process is different in init/2 and websocket_*/* functions. This however would allow us to use websocket_init/2 for sending messages on connect, instead of sending ourselves a message and be subject to races. Note that websocket_init/2 is optional. This is all a big change and while most of the tests pass, some functionality currently doesn't. SPDY is broken and will be removed soon in favor of HTTP/2. Automatic compression is currently disabled. The cowboy_req interface probably still have a few functions that need to be updated. The docs and examples do not refer the current functionality anymore. Everything will be fixed over time. Feedback is more than welcome. Open a ticket!
2016-02-10 17:28:32 +01:00
_ ->
Pid ! {{Pid, StreamID}, {data, nofin, Data}},
ok
Do not send empty chunks User code may sometimes send an empty value which gets understood by the client as being the end of the stream while this was not intended. Ignoring empty values allow making sure the stream isn't ended by mistake.
2015-02-16 19:49:01 +01:00
end.
Add chunked reply support. Send the status line and headers using cowboy_http_req:chunked_reply/3, and individual chunks with cowboy_http_req:chunk/2.
2011-05-08 17:26:21 +02:00
Remove outdated comments, all edoc, plus a few minor tweaks
2014-03-26 19:05:59 +01:00
%% If ever made public, need to send nothing if HEAD.
Add experimental and incomplete SPDY support The SPDY connection processes are also supervisors. Missing: * sendfile support * request body reading support
2013-05-30 20:21:01 +02:00
-spec last_chunk(Req) -> Req when Req::req().
last_chunk(Req=#http_req{socket=Socket, transport=Transport}) ->
_ = Transport:send(Socket, <<"0\r\n\r\n">>),
Req#http_req{resp_state=done}.
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
-spec continue(req()) -> ok.
Add request body reading options The options were added to allow developers to fix timeout issues when reading large bodies. It is also a cleaner and easier to extend interface. This commit deprecates the functions init_stream, stream_body and skip_body which are no longer needed. They will be removed in 1.0. The body function can now take an additional argument that is a list of options. The body_qs, part and part_body functions can too and simply pass this argument down to the body call. There are options for disabling the automatic continue reply, setting a maximum length to be returned (soft limit), setting the read length and read timeout, and setting the transfer and content decode functions. The return value of the body and body_qs have changed slightly. The body function now works similarly to the part_body function, in that it returns either an ok or a more tuple depending on whether there is additional data to be read. The body_qs function can return a badlength tuple if the body is too big. The default size has been increased from 16KB to 64KB. The default read length and timeout have been tweaked and vary depending on the function called. The body function will now adequately process chunked bodies, which means that the body_qs function will too. But this means that the behavior has changed slightly and your code should be tested properly when updating your code. The body and body_qs still accept a length as first argument for compatibility purpose with older code. Note that this form is deprecated and will be removed in 1.0. The part and part_body function, being new and never having been in a release yet, have this form completely removed in this commit. Again, while most code should work as-is, you should make sure that it actually does before pushing this to production.
2014-06-02 23:09:43 +02:00
continue(#http_req{socket=Socket, transport=Transport,
version=Version}) ->
HTTPVer = atom_to_binary(Version, latin1),
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
ok = Transport:send(Socket,
Add request body reading options The options were added to allow developers to fix timeout issues when reading large bodies. It is also a cleaner and easier to extend interface. This commit deprecates the functions init_stream, stream_body and skip_body which are no longer needed. They will be removed in 1.0. The body function can now take an additional argument that is a list of options. The body_qs, part and part_body functions can too and simply pass this argument down to the body call. There are options for disabling the automatic continue reply, setting a maximum length to be returned (soft limit), setting the read length and read timeout, and setting the transfer and content decode functions. The return value of the body and body_qs have changed slightly. The body function now works similarly to the part_body function, in that it returns either an ok or a more tuple depending on whether there is additional data to be read. The body_qs function can return a badlength tuple if the body is too big. The default size has been increased from 16KB to 64KB. The default read length and timeout have been tweaked and vary depending on the function called. The body function will now adequately process chunked bodies, which means that the body_qs function will too. But this means that the behavior has changed slightly and your code should be tested properly when updating your code. The body and body_qs still accept a length as first argument for compatibility purpose with older code. Note that this form is deprecated and will be removed in 1.0. The part and part_body function, being new and never having been in a release yet, have this form completely removed in this commit. Again, while most code should work as-is, you should make sure that it actually does before pushing this to production.
2014-06-02 23:09:43 +02:00
<< HTTPVer/binary, " ", (status(100))/binary, "\r\n\r\n" >>).
Simpler code for sending errors following crashes
2013-08-24 11:20:14 +02:00
%% Meant to be used internally for sending errors after crashes.
Reply with 400 on header parsing crash This is a first step to improve the HTTP status codes returned by Cowboy on crashes. We will tweak it over time. Also fixes a small bug where two replies may have been sent when using loop handlers under rare conditions.
2014-07-12 12:09:43 +02:00
-spec maybe_reply([{module(), atom(), arity() | [term()], _}], req()) -> ok.
maybe_reply(Stacktrace, Req) ->
Simpler code for sending errors following crashes
2013-08-24 11:20:14 +02:00
receive
{cowboy_req, resp_sent} -> ok
after 0 ->
Reply with 400 on header parsing crash This is a first step to improve the HTTP status codes returned by Cowboy on crashes. We will tweak it over time. Also fixes a small bug where two replies may have been sent when using loop handlers under rare conditions.
2014-07-12 12:09:43 +02:00
_ = do_maybe_reply(Stacktrace, Req),
Simpler code for sending errors following crashes
2013-08-24 11:20:14 +02:00
ok
end.
Use cowlib master
2015-02-01 19:36:13 +01:00
do_maybe_reply([{erlang, binary_to_integer, _, _}, {cow_http_hd, parse_content_length, _, _}|_], Req) ->
cowboy_req:reply(400, Req);
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
do_maybe_reply([{cow_http_hd, _, _, _}|_], Req) ->
Reply with 400 on header parsing crash This is a first step to improve the HTTP status codes returned by Cowboy on crashes. We will tweak it over time. Also fixes a small bug where two replies may have been sent when using loop handlers under rare conditions.
2014-07-12 12:09:43 +02:00
cowboy_req:reply(400, Req);
do_maybe_reply(_, Req) ->
cowboy_req:reply(500, Req).
Move cowboy_http:status() to cowboy:http_status()
2013-05-16 16:29:24 +02:00
-spec ensure_response(req(), cowboy:http_status()) -> ok.
Remove duplicate code for ensure_response
2012-09-15 22:19:39 +02:00
%% The response has already been fully sent to the client.
ensure_response(#http_req{resp_state=done}, _) ->
ok;
%% No response has been sent but everything apparently went fine.
%% Reply with the status code found in the second argument.
Add a workaround to disable chunked transfer-encoding This is an undocumented workaround to disable chunks when using HTTP/1.1. It can be used when the client advertises itself as HTTP/1.1 despite not understanding the chunked transfer-encoding. Usage can be found looking at the test for it. When activated, Cowboy will still advertise itself as HTTP/1.1, but will send the body the same way it would if it was HTTP/1.0.
2013-06-26 18:04:20 +01:00
ensure_response(Req=#http_req{resp_state=RespState}, Status)
when RespState =:= waiting; RespState =:= waiting_stream ->
Remove duplicate code for ensure_response
2012-09-15 22:19:39 +02:00
_ = reply(Status, [], [], Req),
ok;
%% Terminate the chunked body for HTTP/1.1 only.
Add a workaround to disable chunked transfer-encoding This is an undocumented workaround to disable chunks when using HTTP/1.1. It can be used when the client advertises itself as HTTP/1.1 despite not understanding the chunked transfer-encoding. Usage can be found looking at the test for it. When activated, Cowboy will still advertise itself as HTTP/1.1, but will send the body the same way it would if it was HTTP/1.0.
2013-06-26 18:04:20 +01:00
ensure_response(#http_req{method= <<"HEAD">>}, _) ->
Remove duplicate code for ensure_response
2012-09-15 22:19:39 +02:00
ok;
Add experimental and incomplete SPDY support The SPDY connection processes are also supervisors. Missing: * sendfile support * request body reading support
2013-05-30 20:21:01 +02:00
ensure_response(Req=#http_req{resp_state=chunks}, _) ->
_ = last_chunk(Req),
Add a workaround to disable chunked transfer-encoding This is an undocumented workaround to disable chunks when using HTTP/1.1. It can be used when the client advertises itself as HTTP/1.1 despite not understanding the chunked transfer-encoding. Usage can be found looking at the test for it. When activated, Cowboy will still advertise itself as HTTP/1.1, but will send the body the same way it would if it was HTTP/1.0.
2013-06-26 18:04:20 +01:00
ok;
ensure_response(#http_req{}, _) ->
Remove duplicate code for ensure_response
2012-09-15 22:19:39 +02:00
ok.
Add the private function set_host/4 used by cowboy_protocol
2012-09-16 16:04:16 +02:00
%% Private setter/getter API.
Better handle socket closing with loop handlers We now read from the socket to be able to detect errors or TCP close events, and buffer the data if any. Once the data receive goes over a certain limit, which defaults to 5000 bytes, we simply close the connection with an {error, overflow} reason.
2013-02-11 09:03:13 +01:00
-spec append_buffer(binary(), Req) -> Req when Req::req().
append_buffer(Suffix, Req=#http_req{buffer=Buffer}) ->
Req#http_req{buffer= << Buffer/binary, Suffix/binary >>}.
Add cowboy_req:get/2 and :set/2 private functions They should not be used unless you have a very special need, which generally involves interacting with the internals of Cowboy.
2012-09-29 11:08:59 +02:00
-spec get(atom(), req()) -> any(); ([atom()], req()) -> any().
get(List, Req) when is_list(List) ->
[g(Atom, Req) || Atom <- List];
get(Atom, Req) when is_atom(Atom) ->
g(Atom, Req).
g(bindings, #http_req{bindings=Ret}) -> Ret;
g(body_state, #http_req{body_state=Ret}) -> Ret;
g(buffer, #http_req{buffer=Ret}) -> Ret;
g(connection, #http_req{connection=Ret}) -> Ret;
g(headers, #http_req{headers=Ret}) -> Ret;
g(host, #http_req{host=Ret}) -> Ret;
g(host_info, #http_req{host_info=Ret}) -> Ret;
g(meta, #http_req{meta=Ret}) -> Ret;
g(method, #http_req{method=Ret}) -> Ret;
g(multipart, #http_req{multipart=Ret}) -> Ret;
g(onresponse, #http_req{onresponse=Ret}) -> Ret;
g(path, #http_req{path=Ret}) -> Ret;
g(path_info, #http_req{path_info=Ret}) -> Ret;
g(peer, #http_req{peer=Ret}) -> Ret;
g(pid, #http_req{pid=Ret}) -> Ret;
g(port, #http_req{port=Ret}) -> Ret;
g(qs, #http_req{qs=Ret}) -> Ret;
g(resp_body, #http_req{resp_body=Ret}) -> Ret;
Add support for the webkit deflate frame extension
2013-04-15 18:36:33 +02:00
g(resp_compress, #http_req{resp_compress=Ret}) -> Ret;
Add cowboy_req:get/2 and :set/2 private functions They should not be used unless you have a very special need, which generally involves interacting with the internals of Cowboy.
2012-09-29 11:08:59 +02:00
g(resp_headers, #http_req{resp_headers=Ret}) -> Ret;
g(resp_state, #http_req{resp_state=Ret}) -> Ret;
g(socket, #http_req{socket=Ret}) -> Ret;
g(transport, #http_req{transport=Ret}) -> Ret;
g(version, #http_req{version=Ret}) -> Ret.
-spec set([{atom(), any()}], Req) -> Req when Req::req().
set([], Req) -> Req;
set([{bindings, Val}|Tail], Req) -> set(Tail, Req#http_req{bindings=Val});
set([{body_state, Val}|Tail], Req) -> set(Tail, Req#http_req{body_state=Val});
set([{buffer, Val}|Tail], Req) -> set(Tail, Req#http_req{buffer=Val});
set([{connection, Val}|Tail], Req) -> set(Tail, Req#http_req{connection=Val});
set([{headers, Val}|Tail], Req) -> set(Tail, Req#http_req{headers=Val});
set([{host, Val}|Tail], Req) -> set(Tail, Req#http_req{host=Val});
set([{host_info, Val}|Tail], Req) -> set(Tail, Req#http_req{host_info=Val});
set([{meta, Val}|Tail], Req) -> set(Tail, Req#http_req{meta=Val});
set([{method, Val}|Tail], Req) -> set(Tail, Req#http_req{method=Val});
set([{multipart, Val}|Tail], Req) -> set(Tail, Req#http_req{multipart=Val});
set([{onresponse, Val}|Tail], Req) -> set(Tail, Req#http_req{onresponse=Val});
set([{path, Val}|Tail], Req) -> set(Tail, Req#http_req{path=Val});
set([{path_info, Val}|Tail], Req) -> set(Tail, Req#http_req{path_info=Val});
set([{peer, Val}|Tail], Req) -> set(Tail, Req#http_req{peer=Val});
set([{pid, Val}|Tail], Req) -> set(Tail, Req#http_req{pid=Val});
set([{port, Val}|Tail], Req) -> set(Tail, Req#http_req{port=Val});
set([{qs, Val}|Tail], Req) -> set(Tail, Req#http_req{qs=Val});
set([{resp_body, Val}|Tail], Req) -> set(Tail, Req#http_req{resp_body=Val});
set([{resp_headers, Val}|Tail], Req) -> set(Tail, Req#http_req{resp_headers=Val});
set([{resp_state, Val}|Tail], Req) -> set(Tail, Req#http_req{resp_state=Val});
set([{socket, Val}|Tail], Req) -> set(Tail, Req#http_req{socket=Val});
set([{transport, Val}|Tail], Req) -> set(Tail, Req#http_req{transport=Val});
set([{version, Val}|Tail], Req) -> set(Tail, Req#http_req{version=Val}).
Remove cowboy_dispatcher Types and code are moved to cowboy_router. The match/3 export from cowboy_dispatcher isn't available anymore as it is called internally.
2013-01-22 18:25:16 +01:00
-spec set_bindings(cowboy_router:tokens(), cowboy_router:tokens(),
cowboy_router:bindings(), Req) -> Req when Req::req().
Add the private set_bindings/4 function used by cowboy_protocol
2012-09-17 00:39:29 +02:00
set_bindings(HostInfo, PathInfo, Bindings, Req) ->
Req#http_req{host_info=HostInfo, path_info=PathInfo,
bindings=Bindings}.
Add a function to lock request responses instead of inlining Just more cleanup in order to have req() opaque.
2012-09-16 13:57:27 +02:00
-spec lock(Req) -> Req when Req::req().
lock(Req) ->
Req#http_req{resp_state=locked}.
Add a cowboy_req:to_list/1 function Removes a lot of duplicate code when printing error messages.
2012-09-15 20:33:57 +02:00
-spec to_list(req()) -> [{atom(), any()}].
to_list(Req) ->
lists:zip(record_info(fields, http_req), tl(tuple_to_list(Req))).
Initial commit.
2011-03-07 22:59:22 +01:00
%% Internal.
Fix a small edoc issue
2012-09-29 13:16:48 +02:00
%% We don't match on "keep-alive" since it is the default value.
Optimize Connection header parsing Still optimizing the critical path. Removes cowboy_http:connection_to_atom/1.
2012-09-27 01:31:19 +02:00
-spec connection_to_atom([binary()]) -> keepalive | close.
connection_to_atom([]) ->
keepalive;
connection_to_atom([<<"close">>|_]) ->
close;
connection_to_atom([_|Tail]) ->
connection_to_atom(Tail).
Move cowboy_http:status() to cowboy:http_status()
2013-05-16 16:29:24 +02:00
-spec status(cowboy:http_status()) -> binary().
Switch the HTTP protocol to use binary packets instead of lists. The server now does a single recv (or more, but only if needed) which is then sent to erlang:decode_packet/3 multiple times. Since most requests are smaller than the default MTU on many platforms, we benefit from this greatly. In the case of requests with a body, the server usually read at least part of the body on the first recv. This is bufferized properly and used when later retrieving the body. In the case of pipelined requests, we can end up reading many requests in a single recv, which are then handled properly using only the buffer containing the received data.
2011-05-05 14:03:39 +02:00
status(100) -> <<"100 Continue">>;
status(101) -> <<"101 Switching Protocols">>;
status(102) -> <<"102 Processing">>;
status(200) -> <<"200 OK">>;
status(201) -> <<"201 Created">>;
status(202) -> <<"202 Accepted">>;
status(203) -> <<"203 Non-Authoritative Information">>;
status(204) -> <<"204 No Content">>;
status(205) -> <<"205 Reset Content">>;
status(206) -> <<"206 Partial Content">>;
status(207) -> <<"207 Multi-Status">>;
status(226) -> <<"226 IM Used">>;
status(300) -> <<"300 Multiple Choices">>;
status(301) -> <<"301 Moved Permanently">>;
status(302) -> <<"302 Found">>;
status(303) -> <<"303 See Other">>;
status(304) -> <<"304 Not Modified">>;
status(305) -> <<"305 Use Proxy">>;
status(306) -> <<"306 Switch Proxy">>;
status(307) -> <<"307 Temporary Redirect">>;
status(400) -> <<"400 Bad Request">>;
status(401) -> <<"401 Unauthorized">>;
status(402) -> <<"402 Payment Required">>;
status(403) -> <<"403 Forbidden">>;
status(404) -> <<"404 Not Found">>;
status(405) -> <<"405 Method Not Allowed">>;
status(406) -> <<"406 Not Acceptable">>;
status(407) -> <<"407 Proxy Authentication Required">>;
status(408) -> <<"408 Request Timeout">>;
status(409) -> <<"409 Conflict">>;
status(410) -> <<"410 Gone">>;
status(411) -> <<"411 Length Required">>;
status(412) -> <<"412 Precondition Failed">>;
status(413) -> <<"413 Request Entity Too Large">>;
status(414) -> <<"414 Request-URI Too Long">>;
status(415) -> <<"415 Unsupported Media Type">>;
status(416) -> <<"416 Requested Range Not Satisfiable">>;
status(417) -> <<"417 Expectation Failed">>;
status(418) -> <<"418 I'm a teapot">>;
status(422) -> <<"422 Unprocessable Entity">>;
status(423) -> <<"423 Locked">>;
status(424) -> <<"424 Failed Dependency">>;
status(425) -> <<"425 Unordered Collection">>;
status(426) -> <<"426 Upgrade Required">>;
Add new HTTP status codes from RFC 6585 See http://tools.ietf.org/html/rfc6585 for further details.
2012-05-06 13:11:21 +02:00
status(428) -> <<"428 Precondition Required">>;
status(429) -> <<"429 Too Many Requests">>;
status(431) -> <<"431 Request Header Fields Too Large">>;
Switch the HTTP protocol to use binary packets instead of lists. The server now does a single recv (or more, but only if needed) which is then sent to erlang:decode_packet/3 multiple times. Since most requests are smaller than the default MTU on many platforms, we benefit from this greatly. In the case of requests with a body, the server usually read at least part of the body on the first recv. This is bufferized properly and used when later retrieving the body. In the case of pipelined requests, we can end up reading many requests in a single recv, which are then handled properly using only the buffer containing the received data.
2011-05-05 14:03:39 +02:00
status(500) -> <<"500 Internal Server Error">>;
status(501) -> <<"501 Not Implemented">>;
status(502) -> <<"502 Bad Gateway">>;
status(503) -> <<"503 Service Unavailable">>;
status(504) -> <<"504 Gateway Timeout">>;
status(505) -> <<"505 HTTP Version Not Supported">>;
status(506) -> <<"506 Variant Also Negotiates">>;
status(507) -> <<"507 Insufficient Storage">>;
status(510) -> <<"510 Not Extended">>;
Add new HTTP status codes from RFC 6585 See http://tools.ietf.org/html/rfc6585 for further details.
2012-05-06 13:11:21 +02:00
status(511) -> <<"511 Network Authentication Required">>;
Switch the HTTP protocol to use binary packets instead of lists. The server now does a single recv (or more, but only if needed) which is then sent to erlang:decode_packet/3 multiple times. Since most requests are smaller than the default MTU on many platforms, we benefit from this greatly. In the case of requests with a body, the server usually read at least part of the body on the first recv. This is bufferized properly and used when later retrieving the body. In the case of pipelined requests, we can end up reading many requests in a single recv, which are then handled properly using only the buffer containing the received data.
2011-05-05 14:03:39 +02:00
status(B) when is_binary(B) -> B.
Move the reply function to cowboy_http_req.
2011-03-18 22:38:26 +01:00
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
%% Create map, convert keys to atoms and group duplicate keys into lists.
%% Keys that are not found in the user provided list are entirely skipped.
%% @todo Can probably be done directly while parsing.
Reverse the order of arguments of match_* functions Wasn't following the same order as the rest of the module.
2014-10-04 13:21:16 +03:00
kvlist_to_map(Fields, KvList) ->
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
Keys = [case K of
{Key, _} -> Key;
{Key, _, _} -> Key;
Key -> Key
end || K <- Fields],
Reverse the order of arguments of match_* functions Wasn't following the same order as the rest of the module.
2014-10-04 13:21:16 +03:00
kvlist_to_map(Keys, KvList, #{}).
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
Reverse the order of arguments of match_* functions Wasn't following the same order as the rest of the module.
2014-10-04 13:21:16 +03:00
kvlist_to_map(_, [], Map) ->
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
Map;
Reverse the order of arguments of match_* functions Wasn't following the same order as the rest of the module.
2014-10-04 13:21:16 +03:00
kvlist_to_map(Keys, [{Key, Value}|Tail], Map) ->
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
try binary_to_existing_atom(Key, utf8) of
Atom ->
case lists:member(Atom, Keys) of
true ->
case maps:find(Atom, Map) of
{ok, MapValue} when is_list(MapValue) ->
Reverse the order of arguments of match_* functions Wasn't following the same order as the rest of the module.
2014-10-04 13:21:16 +03:00
kvlist_to_map(Keys, Tail,
Use map syntax instead of maps:put/3
2015-07-27 17:32:05 +02:00
Map#{Atom => [Value|MapValue]});
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
{ok, MapValue} ->
Reverse the order of arguments of match_* functions Wasn't following the same order as the rest of the module.
2014-10-04 13:21:16 +03:00
kvlist_to_map(Keys, Tail,
Use map syntax instead of maps:put/3
2015-07-27 17:32:05 +02:00
Map#{Atom => [Value, MapValue]});
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
error ->
Reverse the order of arguments of match_* functions Wasn't following the same order as the rest of the module.
2014-10-04 13:21:16 +03:00
kvlist_to_map(Keys, Tail,
Use map syntax instead of maps:put/3
2015-07-27 17:32:05 +02:00
Map#{Atom => Value})
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
end;
false ->
Reverse the order of arguments of match_* functions Wasn't following the same order as the rest of the module.
2014-10-04 13:21:16 +03:00
kvlist_to_map(Keys, Tail, Map)
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
end
catch error:badarg ->
Reverse the order of arguments of match_* functions Wasn't following the same order as the rest of the module.
2014-10-04 13:21:16 +03:00
kvlist_to_map(Keys, Tail, Map)
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
end.
%% Loop through fields, if value is missing and no default, crash;
%% else if value is missing and has a default, set default;
%% otherwise apply constraints. If constraint fails, crash.
Reverse the order of arguments of match_* functions Wasn't following the same order as the rest of the module.
2014-10-04 13:21:16 +03:00
filter([], Map) ->
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
Map;
Reverse the order of arguments of match_* functions Wasn't following the same order as the rest of the module.
2014-10-04 13:21:16 +03:00
filter([{Key, Constraints}|Tail], Map) ->
filter_constraints(Tail, Map, Key, maps:get(Key, Map), Constraints);
filter([{Key, Constraints, Default}|Tail], Map) ->
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
case maps:find(Key, Map) of
{ok, Value} ->
Reverse the order of arguments of match_* functions Wasn't following the same order as the rest of the module.
2014-10-04 13:21:16 +03:00
filter_constraints(Tail, Map, Key, Value, Constraints);
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
error ->
Use map syntax instead of maps:put/3
2015-07-27 17:32:05 +02:00
filter(Tail, Map#{Key => Default})
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
end;
Reverse the order of arguments of match_* functions Wasn't following the same order as the rest of the module.
2014-10-04 13:21:16 +03:00
filter([Key|Tail], Map) ->
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
true = maps:is_key(Key, Map),
Reverse the order of arguments of match_* functions Wasn't following the same order as the rest of the module.
2014-10-04 13:21:16 +03:00
filter(Tail, Map).
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
Reverse the order of arguments of match_* functions Wasn't following the same order as the rest of the module.
2014-10-04 13:21:16 +03:00
filter_constraints(Tail, Map, Key, Value, Constraints) ->
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
case cowboy_constraints:validate(Value, Constraints) of
true ->
Reverse the order of arguments of match_* functions Wasn't following the same order as the rest of the module.
2014-10-04 13:21:16 +03:00
filter(Tail, Map);
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
{true, Value2} ->
Use map syntax instead of maps:put/3
2015-07-27 17:32:05 +02:00
filter(Tail, Map#{Key => Value2})
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
end.
Add cowboy_req:url/1 to return the full request URL Use it in cowboy_websocket for hixie76, replacing http by ws.
2012-09-15 22:03:00 +02:00
%% Tests.
-ifdef(TEST).
url_test() ->
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
undefined =
url and host_url may return undefined this can happen when the request parsing fails and onresponse needs the url, its perfectly possible that the url has not yet been defined
2012-11-27 13:22:38 -06:00
url(#http_req{transport=ranch_tcp, host= <<>>, port= undefined,
Remove cowboy_req:fragment/1 Clients do not send it. We skip the value if we receive it now, as it shouldn't happen, and won't for all the mainstream clients.
2013-05-15 15:17:33 +02:00
path= <<>>, qs= <<>>, pid=self()}),
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
<<"http://localhost/path">> =
Add cowboy_req:url/1 to return the full request URL Use it in cowboy_websocket for hixie76, replacing http by ws.
2012-09-15 22:03:00 +02:00
url(#http_req{transport=ranch_tcp, host= <<"localhost">>, port=80,
Remove cowboy_req:fragment/1 Clients do not send it. We skip the value if we receive it now, as it shouldn't happen, and won't for all the mainstream clients.
2013-05-15 15:17:33 +02:00
path= <<"/path">>, qs= <<>>, pid=self()}),
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
<<"http://localhost:443/path">> =
Add cowboy_req:url/1 to return the full request URL Use it in cowboy_websocket for hixie76, replacing http by ws.
2012-09-15 22:03:00 +02:00
url(#http_req{transport=ranch_tcp, host= <<"localhost">>, port=443,
Remove cowboy_req:fragment/1 Clients do not send it. We skip the value if we receive it now, as it shouldn't happen, and won't for all the mainstream clients.
2013-05-15 15:17:33 +02:00
path= <<"/path">>, qs= <<>>, pid=self()}),
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
<<"http://localhost:8080/path">> =
Add cowboy_req:url/1 to return the full request URL Use it in cowboy_websocket for hixie76, replacing http by ws.
2012-09-15 22:03:00 +02:00
url(#http_req{transport=ranch_tcp, host= <<"localhost">>, port=8080,
Remove cowboy_req:fragment/1 Clients do not send it. We skip the value if we receive it now, as it shouldn't happen, and won't for all the mainstream clients.
2013-05-15 15:17:33 +02:00
path= <<"/path">>, qs= <<>>, pid=self()}),
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
<<"http://localhost:8080/path?dummy=2785">> =
Add cowboy_req:url/1 to return the full request URL Use it in cowboy_websocket for hixie76, replacing http by ws.
2012-09-15 22:03:00 +02:00
url(#http_req{transport=ranch_tcp, host= <<"localhost">>, port=8080,
Remove cowboy_req:fragment/1 Clients do not send it. We skip the value if we receive it now, as it shouldn't happen, and won't for all the mainstream clients.
2013-05-15 15:17:33 +02:00
path= <<"/path">>, qs= <<"dummy=2785">>, pid=self()}),
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
<<"https://localhost/path">> =
Add cowboy_req:url/1 to return the full request URL Use it in cowboy_websocket for hixie76, replacing http by ws.
2012-09-15 22:03:00 +02:00
url(#http_req{transport=ranch_ssl, host= <<"localhost">>, port=443,
Remove cowboy_req:fragment/1 Clients do not send it. We skip the value if we receive it now, as it shouldn't happen, and won't for all the mainstream clients.
2013-05-15 15:17:33 +02:00
path= <<"/path">>, qs= <<>>, pid=self()}),
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
<<"https://localhost:8443/path">> =
Add cowboy_req:url/1 to return the full request URL Use it in cowboy_websocket for hixie76, replacing http by ws.
2012-09-15 22:03:00 +02:00
url(#http_req{transport=ranch_ssl, host= <<"localhost">>, port=8443,
Remove cowboy_req:fragment/1 Clients do not send it. We skip the value if we receive it now, as it shouldn't happen, and won't for all the mainstream clients.
2013-05-15 15:17:33 +02:00
path= <<"/path">>, qs= <<>>, pid=self()}),
Breaking update of the cowboy_req interface Simplify the interface for most cowboy_req functions. They all return a single value except the four body reading functions. The reply functions now only return a Req value. Access functions do not return a Req anymore. Functions that used to cache results do not have a cache anymore. The interface for accessing query string and cookies has therefore been changed. There are now three query string functions: qs/1 provides access to the raw query string value; parse_qs/1 returns the query string as a list of key/values; match_qs/2 returns a map containing the values requested in the second argument, after applying constraints and default value. Similarly, there are two cookie functions: parse_cookies/1 and match_cookies/2. More match functions will be added in future commits. None of the functions return an error tuple anymore. It either works or crashes. Cowboy will attempt to provide an appropriate status code in the response of crashed handlers. As a result, the content decode function has its return value changed to a simple binary, and the body reading functions only return on success.
2014-09-23 16:43:29 +03:00
<<"https://localhost:8443/path?dummy=2785">> =
Add cowboy_req:url/1 to return the full request URL Use it in cowboy_websocket for hixie76, replacing http by ws.
2012-09-15 22:03:00 +02:00
url(#http_req{transport=ranch_ssl, host= <<"localhost">>, port=8443,
Remove cowboy_req:fragment/1 Clients do not send it. We skip the value if we receive it now, as it shouldn't happen, and won't for all the mainstream clients.
2013-05-15 15:17:33 +02:00
path= <<"/path">>, qs= <<"dummy=2785">>, pid=self()}),
Add cowboy_req:url/1 to return the full request URL Use it in cowboy_websocket for hixie76, replacing http by ws.
2012-09-15 22:03:00 +02:00
ok.
Optimize Connection header parsing Still optimizing the critical path. Removes cowboy_http:connection_to_atom/1.
2012-09-27 01:31:19 +02:00
connection_to_atom_test_() ->
Tests = [
{[<<"close">>], close},
{[<<"keep-alive">>], keepalive},
{[<<"keep-alive">>, <<"upgrade">>], keepalive}
],
[{lists:flatten(io_lib:format("~p", [T])),
fun() -> R = connection_to_atom(T) end} || {T, R} <- Tests].
Add cowboy_req:url/1 to return the full request URL Use it in cowboy_websocket for hixie76, replacing http by ws.
2012-09-15 22:03:00 +02:00
-endif.
Reference in a new issue Copy permalink