From 776ee8a8a29f7dd456e51aadc9224d5d6a94381a Mon Sep 17 00:00:00 2001
From: alisdair sullivan <alisdairsullivan@yahoo.ca>
Date: Wed, 3 Aug 2011 00:45:49 -0700
Subject: [PATCH] two new tests for bad low pairs of json surrogate pairs

---
 src/jsx_decoder.hrl                        | 18 +++++-------------
 test/cases/bad_low_surrogate.json          |  1 +
 test/cases/bad_low_surrogate.test          |  3 +++
 test/cases/bad_low_surrogate_replaced.json |  1 +
 test/cases/bad_low_surrogate_replaced.test |  4 ++++
 5 files changed, 14 insertions(+), 13 deletions(-)
 create mode 100644 test/cases/bad_low_surrogate.json
 create mode 100644 test/cases/bad_low_surrogate.test
 create mode 100644 test/cases/bad_low_surrogate_replaced.json
 create mode 100644 test/cases/bad_low_surrogate_replaced.test

diff --git a/src/jsx_decoder.hrl b/src/jsx_decoder.hrl
index ac40cfa..537fb84 100644
--- a/src/jsx_decoder.hrl
+++ b/src/jsx_decoder.hrl
@@ -568,7 +568,8 @@ escaped_unicode(Bin, Stack, Opts, String, Acc) ->
 
 low_surrogate(<<?rsolidus/?utfx, Rest/binary>>, Stack, Opts, String, High) ->
     low_surrogate_u(Rest, Stack, Opts, String, High);
-%% not an escaped codepoint, our high codepoint is illegal
+%% not an escaped codepoint, our high codepoint is illegal. dispatch back to
+%%   string to handle
 low_surrogate(<<S/?utfx, Rest/binary>> = Bin, Stack, Opts, String, _) ->
     case Opts#opts.loose_unicode of
         true ->
@@ -608,15 +609,10 @@ low_surrogate(<<D/?utfx, Rest/binary>>, Stack, Opts, String, [C, B, A], H)
     case erlang:list_to_integer([A, B, C, D], 16) of
         X when X >= 16#dc00, X =< 16#dfff ->
             V = surrogate_to_codepoint(H, X),
-            case V rem 16#10000 of
-                Y when Y == 16#fffe; Y == 16#ffff ->
+            case V rem 16#10000 of Y when Y == 16#fffe; Y == 16#ffff ->
                     case Opts#opts.loose_unicode of
                         true ->
-                            string(Rest,
-                                Stack,
-                                Opts,
-                                [16#fffd] ++ String
-                            )
+                            string(Rest, Stack, Opts, [16#fffd] ++ String)
                         ; false ->    
                             {error, {badjson, <<D/?utfx, Rest/binary>>}}
                     end
@@ -627,11 +623,7 @@ low_surrogate(<<D/?utfx, Rest/binary>>, Stack, Opts, String, [C, B, A], H)
         ; _ ->
             case Opts#opts.loose_unicode of
                 true ->
-                    string(Rest,
-                        Stack,
-                        Opts,
-                        [16#fffd, 16#fffd] ++ String
-                    )
+                    string(Rest, Stack, Opts, [16#fffd, 16#fffd] ++ String)
                 ; false ->    
                     {error, {badjson, <<D/?utfx, Rest/binary>>}}
             end
diff --git a/test/cases/bad_low_surrogate.json b/test/cases/bad_low_surrogate.json
new file mode 100644
index 0000000..4b22b57
--- /dev/null
+++ b/test/cases/bad_low_surrogate.json
@@ -0,0 +1 @@
+["\ud801\u0032"]
\ No newline at end of file
diff --git a/test/cases/bad_low_surrogate.test b/test/cases/bad_low_surrogate.test
new file mode 100644
index 0000000..9f36e55
--- /dev/null
+++ b/test/cases/bad_low_surrogate.test
@@ -0,0 +1,3 @@
+{name, "bad_low_surrogate"}.
+{jsx, {error, badjson}}.
+{json, "bad_low_surrogate.json"}.
diff --git a/test/cases/bad_low_surrogate_replaced.json b/test/cases/bad_low_surrogate_replaced.json
new file mode 100644
index 0000000..4b22b57
--- /dev/null
+++ b/test/cases/bad_low_surrogate_replaced.json
@@ -0,0 +1 @@
+["\ud801\u0032"]
\ No newline at end of file
diff --git a/test/cases/bad_low_surrogate_replaced.test b/test/cases/bad_low_surrogate_replaced.test
new file mode 100644
index 0000000..2fc7b3d
--- /dev/null
+++ b/test/cases/bad_low_surrogate_replaced.test
@@ -0,0 +1,4 @@
+{name, "bad_low_surrogate_replaced"}.
+{jsx, [start_array,{string, [16#fffd, 16#fffd]},end_array,end_json]}.
+{json, "bad_low_surrogate_replaced.json"}.
+{jsx_flags, [loose_unicode]}.