From 65449753791a9dd259a32ac289cb6f191b7f63e1 Mon Sep 17 00:00:00 2001
From: alisdair sullivan <alisdairsullivan@yahoo.ca>
Date: Wed, 28 Mar 2012 23:46:06 -0700
Subject: [PATCH] add more comprehensive checking to json_escape

---
 src/jsx_utils.erl | 136 +++++++++++++++++++++++++++++++++++++++++++---
 1 file changed, 127 insertions(+), 9 deletions(-)

diff --git a/src/jsx_utils.erl b/src/jsx_utils.erl
index aea689c..7c688f9 100644
--- a/src/jsx_utils.erl
+++ b/src/jsx_utils.erl
@@ -592,15 +592,6 @@ binary_escape_test_() ->
                 <<"\\/Date(1303502009425)\\/">>
             )
         },
-        {"bad utf8",
-            ?_assertError(badarg, json_escape(<<32, 64, 128, 255>>, #opts{}))
-        },
-        {"bad utf8 ok",
-            ?_assertEqual(
-                json_escape(<<32, 64, 128, 255>>, #opts{loose_unicode=true}),
-                <<32, 64, 16#fffd/utf8, 16#fffd/utf8>>
-            )
-        },
         {"bad surrogate", ?_assertError(badarg, json_escape(<<237, 160, 127>>, #opts{}))},
         {"bad surrogate ok",
             ?_assertEqual(
@@ -616,4 +607,131 @@ binary_escape_test_() ->
         }
     ].
 
+
+surrogates_test_() ->
+    [
+        {"surrogates - badjson",
+            ?_assertEqual(check_bad(surrogates()), [])
+        },
+        {"surrogates - replaced",
+            ?_assertEqual(check_replaced(surrogates()), [])
+        }
+    ].
+
+
+good_characters_test_() ->
+    [
+        {"acceptable codepoints",
+            ?_assertEqual(check_good(good()), [])
+        },
+        {"acceptable extended",
+            ?_assertEqual(check_good(good_extended()), [])
+        }
+    ].
+
+
+reserved_test_() ->
+    [
+        {"reserved noncharacters - badjson",
+            ?_assertEqual(check_bad(reserved_space()), [])
+        },
+        {"reserved noncharacters - replaced",
+            ?_assertEqual(check_replaced(reserved_space()), [])
+        }
+    ].
+
+
+noncharacters_test_() ->
+    [
+        {"noncharacters - badjson",
+            ?_assertEqual(check_bad(noncharacters()), [])
+        },
+        {"noncharacters - replaced",
+            ?_assertEqual(check_replaced(noncharacters()), [])
+        }
+    ].
+
+
+extended_noncharacters_test_() ->
+    [
+        {"extended noncharacters - badjson",
+            ?_assertEqual(check_bad(extended_noncharacters()), [])
+        },
+        {"extended noncharacters - replaced",
+            ?_assertEqual(check_replaced(extended_noncharacters()), [])
+        }
+    ].
+
+
+check_bad(List) ->
+    lists:dropwhile(fun({_, {error, badjson}}) -> true ; (_) -> false end,
+        check(List, #opts{}, [])
+    ).
+
+
+check_replaced(List) ->
+    lists:dropwhile(fun({_, <<16#fffd/utf8>>}) -> true
+            ; (_) -> false 
+        end,
+        check(List, #opts{loose_unicode=true}, [])
+    ).
+
+
+check_good(List) ->
+    lists:dropwhile(fun({_, _}) -> true ; (_) -> false end,
+        check(List, #opts{}, [])
+    ).
+
+
+check([], _Opts, Acc) -> Acc;
+check([H|T], Opts, Acc) ->
+    R = escape(to_fake_utf(H, utf8), Opts),
+    check(T, Opts, [{H, R}] ++ Acc).
+
+
+escape(JSON, Opts) ->
+    try json_escape(JSON, Opts)
+    catch error:badarg -> {error, badjson}
+    end.
+    
+
+noncharacters() -> lists:seq(16#fffe, 16#ffff).
+
+
+extended_noncharacters() ->
+    [16#1fffe, 16#1ffff, 16#2fffe, 16#2ffff]
+        ++ [16#3fffe, 16#3ffff, 16#4fffe, 16#4ffff]
+        ++ [16#5fffe, 16#5ffff, 16#6fffe, 16#6ffff]
+        ++ [16#7fffe, 16#7ffff, 16#8fffe, 16#8ffff]
+        ++ [16#9fffe, 16#9ffff, 16#afffe, 16#affff]
+        ++ [16#bfffe, 16#bffff, 16#cfffe, 16#cffff]
+        ++ [16#dfffe, 16#dffff, 16#efffe, 16#effff]
+        ++ [16#ffffe, 16#fffff, 16#10fffe, 16#10ffff].
+
+
+surrogates() -> lists:seq(16#d800, 16#dfff).
+
+
+reserved_space() -> lists:seq(16#fdd0, 16#fdef).
+
+
+good() -> lists:seq(16#0000, 16#d7ff) ++ lists:seq(16#e000, 16#fdcf) ++ lists:seq(16#fdf0, 16#fffd).
+            
+            
+good_extended() -> lists:seq(16#100000, 16#10fffd).
+
+
+%% erlang refuses to encode certain codepoints, so fake them all
+to_fake_utf(N, utf8) when N < 16#0080 -> <<N:8>>;
+to_fake_utf(N, utf8) when N < 16#0800 ->
+    <<0:5, Y:5, X:6>> = <<N:16>>,
+    <<2#110:3, Y:5, 2#10:2, X:6>>; 
+to_fake_utf(N, utf8) when N < 16#10000 ->
+    <<Z:4, Y:6, X:6>> = <<N:16>>,
+    <<2#1110:4, Z:4, 2#10:2, Y:6, 2#10:2, X:6>>;
+to_fake_utf(N, utf8) ->
+    <<0:3, W:3, Z:6, Y:6, X:6>> = <<N:24>>,
+    <<2#11110:5, W:3, 2#10:2, Z:6, 2#10:2, Y:6, 2#10:2, X:6>>.
+
+
 -endif.
\ No newline at end of file