add config options for escape_slash

Author: Francois Chagnon

ext/json/ext/generator/generator.c

@@ -16,7 +16,7 @@ static ID i_to_s, i_to_json, i_new, i_indent, i_space, i_space_before,
           i_object_nl, i_array_nl, i_max_nesting, i_allow_nan, i_ascii_only,
           i_quirks_mode, i_pack, i_unpack, i_create_id, i_extend, i_key_p,
           i_aref, i_send, i_respond_to_p, i_match, i_keys, i_depth,
-          i_buffer_initial_length, i_dup;
+          i_buffer_initial_length, i_dup, i_escape_slash;
 
 /*
  * Copyright 2001-2004 Unicode, Inc.
@@ -124,7 +124,7 @@ static void unicode_escape_to_buffer(FBuffer *buffer, char buf[6], UTF16
 
 /* Converts string to a JSON string in FBuffer buffer, where all but the ASCII
  * and control characters are JSON escaped. */
-static void convert_UTF8_to_JSON_ASCII(FBuffer *buffer, VALUE string)
+static void convert_UTF8_to_JSON_ASCII(FBuffer *buffer, VALUE string, char escape_slash)
 {
     const UTF8 *source = (UTF8 *) RSTRING_PTR(string);
     const UTF8 *sourceEnd = source + RSTRING_LEN(string);
@@ -171,12 +171,14 @@ static void convert_UTF8_to_JSON_ASCII(FBuffer *buffer, VALUE string)
                         case '\\':
                             fbuffer_append(buffer, "\\\\", 2);
                             break;
-                        case '/':
-                            fbuffer_append(buffer, "\\/", 2);
-                            break;
                         case '"':
                             fbuffer_append(buffer, "\\\"", 2);
                             break;
+                        case '/':
+                            if(escape_slash) {
+                                fbuffer_append(buffer, "\\/", 2);
+                                break;
+                            }
                         default:
                             fbuffer_append_char(buffer, (char)ch);
                             break;
@@ -225,7 +227,7 @@ static void convert_UTF8_to_JSON_ASCII(FBuffer *buffer, VALUE string)
  * characters required by the JSON standard are JSON escaped. The remaining
  * characters (should be UTF8) are just passed through and appended to the
  * result. */
-static void convert_UTF8_to_JSON(FBuffer *buffer, VALUE string)
+static void convert_UTF8_to_JSON(FBuffer *buffer, VALUE string, char escape_slash)
 {
     const char *ptr = RSTRING_PTR(string), *p;
     unsigned long len = RSTRING_LEN(string), start = 0, end = 0;
@@ -271,14 +273,16 @@ static void convert_UTF8_to_JSON(FBuffer *buffer, VALUE string)
                     escape = "\\\\";
                     escape_len = 2;
                     break;
-                case '/':
-                    escape = "\\/";
-                    escape_len = 2;
-                    break;
                 case '"':
                     escape =  "\\\"";
                     escape_len = 2;
                     break;
+                case '/':
+                    if(escape_slash) {
+                        escape = "\\/";
+                        escape_len = 2;
+                        break;
+                    }
                 default:
                     {
                         unsigned short clen = trailingBytesForUTF8[c] + 1;
@@ -631,6 +635,8 @@ static VALUE cState_configure(VALUE self, VALUE opts)
     state->ascii_only = RTEST(tmp);
     tmp = rb_hash_aref(opts, ID2SYM(i_quirks_mode));
     state->quirks_mode = RTEST(tmp);
+    tmp = rb_hash_aref(opts, ID2SYM(i_escape_slash));
+    state->escape_slash = RTEST(tmp);
     return self;
 }
 
@@ -666,6 +672,7 @@ static VALUE cState_to_h(VALUE self)
     rb_hash_aset(result, ID2SYM(i_ascii_only), state->ascii_only ? Qtrue : Qfalse);
     rb_hash_aset(result, ID2SYM(i_quirks_mode), state->quirks_mode ? Qtrue : Qfalse);
     rb_hash_aset(result, ID2SYM(i_max_nesting), LONG2FIX(state->max_nesting));
+    rb_hash_aset(result, ID2SYM(i_escape_slash), state->escape_slash ? Qtrue : Qfalse);
     rb_hash_aset(result, ID2SYM(i_depth), LONG2FIX(state->depth));
     rb_hash_aset(result, ID2SYM(i_buffer_initial_length), LONG2FIX(state->buffer_initial_length));
     return result;
@@ -799,9 +806,9 @@ static void generate_json_string(FBuffer *buffer, VALUE Vstate, JSON_Generator_S
     obj = rb_funcall(obj, i_encode, 1, CEncoding_UTF_8);
 #endif
     if (state->ascii_only) {
-        convert_UTF8_to_JSON_ASCII(buffer, obj);
+        convert_UTF8_to_JSON_ASCII(buffer, obj, state->escape_slash);
     } else {
-        convert_UTF8_to_JSON(buffer, obj);
+        convert_UTF8_to_JSON(buffer, obj, state->escape_slash);
     }
     fbuffer_append_char(buffer, '"');
 }
@@ -1252,6 +1259,31 @@ static VALUE cState_max_nesting_set(VALUE self, VALUE depth)
     return state->max_nesting = FIX2LONG(depth);
 }
 
+/*
+ * call-seq: escape_slash
+ *
+ * If this boolean is true, the forward slashes will be escaped in
+ * the json output.
+ */
+static VALUE cState_escape_slash(VALUE self)
+{
+    GET_STATE(self);
+    return state->escape_slash ? Qtrue : Qfalse;
+}
+
+/*
+ * call-seq: escape_slash=(depth)
+ *
+ * This sets whether or not the forward slashes will be escaped in
+ * the json output.
+ */
+static VALUE cState_escape_slash_set(VALUE self, VALUE enable)
+{
+    GET_STATE(self);
+    state->escape_slash = RTEST(enable);
+    return Qnil;
+}
+
 /*
  * call-seq: allow_nan?
  *
@@ -1384,6 +1416,9 @@ void Init_generator(void)
     rb_define_method(cState, "array_nl=", cState_array_nl_set, 1);
     rb_define_method(cState, "max_nesting", cState_max_nesting, 0);
     rb_define_method(cState, "max_nesting=", cState_max_nesting_set, 1);
+    rb_define_method(cState, "escape_slash", cState_escape_slash, 0);
+    rb_define_method(cState, "escape_slash?", cState_escape_slash, 0);
+    rb_define_method(cState, "escape_slash=", cState_escape_slash_set, 1);
     rb_define_method(cState, "check_circular?", cState_check_circular_p, 0);
     rb_define_method(cState, "allow_nan?", cState_allow_nan_p, 0);
     rb_define_method(cState, "ascii_only?", cState_ascii_only_p, 0);
@@ -1439,6 +1474,7 @@ void Init_generator(void)
     i_object_nl = rb_intern("object_nl");
     i_array_nl = rb_intern("array_nl");
     i_max_nesting = rb_intern("max_nesting");
+    i_escape_slash = rb_intern("escape_slash");
     i_allow_nan = rb_intern("allow_nan");
     i_ascii_only = rb_intern("ascii_only");
     i_quirks_mode = rb_intern("quirks_mode");

ext/json/ext/generator/generator.h

@@ -50,8 +50,8 @@ static const UTF32 halfMask = 0x3FFUL;
 static unsigned char isLegalUTF8(const UTF8 *source, unsigned long length);
 static void unicode_escape(char *buf, UTF16 character);
 static void unicode_escape_to_buffer(FBuffer *buffer, char buf[6], UTF16 character);
-static void convert_UTF8_to_JSON_ASCII(FBuffer *buffer, VALUE string);
-static void convert_UTF8_to_JSON(FBuffer *buffer, VALUE string);
+static void convert_UTF8_to_JSON_ASCII(FBuffer *buffer, VALUE string, char escape_slash);
+static void convert_UTF8_to_JSON(FBuffer *buffer, VALUE string, char escape_slash);
 static char *fstrndup(const char *ptr, unsigned long len);
 
 /* ruby api and some helpers */
@@ -74,6 +74,7 @@ typedef struct JSON_Generator_StateStruct {
     char allow_nan;
     char ascii_only;
     char quirks_mode;
+    char escape_slash;
     long depth;
     long buffer_initial_length;
 } JSON_Generator_State;
@@ -145,6 +146,8 @@ static VALUE cState_allow_nan_p(VALUE self);
 static VALUE cState_ascii_only_p(VALUE self);
 static VALUE cState_depth(VALUE self);
 static VALUE cState_depth_set(VALUE self, VALUE depth);
+static VALUE cState_escape_slash(VALUE self);
+static VALUE cState_escape_slash_set(VALUE self, VALUE depth);
 static FBuffer *cState_prepare_buffer(VALUE self);
 #ifndef ZALLOC
 #define ZALLOC(type) ((type *)ruby_zalloc(sizeof(type)))

lib/json/common.rb

@@ -358,12 +358,14 @@ class << self
     #  :max_nesting: false
     #  :allow_nan:   true
     #  :quirks_mode: true
+    #  :escape_slash: true
     attr_accessor :dump_default_options
   end
   self.dump_default_options = {
     :max_nesting => false,
     :allow_nan   => true,
     :quirks_mode => true,
+    :escape_slash => true,
   }
 
   # Dumps _obj_ as a JSON string, i.e. calls generate on the object and returns
@@ -443,7 +445,7 @@ module ::Kernel
   # one line.
   def j(*objs)
     objs.each do |obj|
-      puts JSON::generate(obj, :allow_nan => true, :max_nesting => false)
+      puts JSON::generate(obj, :allow_nan => true, :max_nesting => false, :escape_slash => true)
     end
     nil
   end
@@ -452,7 +454,7 @@ def j(*objs)
   # indentation and over many lines.
   def jj(*objs)
     objs.each do |obj|
-      puts JSON::pretty_generate(obj, :allow_nan => true, :max_nesting => false)
+      puts JSON::pretty_generate(obj, :allow_nan => true, :max_nesting => false, :escape_slash => true)
     end
     nil
   end

lib/json/pure/generator.rb

@@ -40,18 +40,20 @@ module JSON
   # Convert a UTF8 encoded Ruby string _string_ to a JSON string, encoded with
   # UTF16 big endian characters as \u????, and return it.
   if defined?(::Encoding)
-    def utf8_to_json(string) # :nodoc:
+    def utf8_to_json(string, escape_slash = true) # :nodoc:
       string = string.dup
       string.force_encoding(::Encoding::ASCII_8BIT)
-      string.gsub!(/[\/"\\\x0-\x1f]/) { MAP[$&] }
+      string.gsub!(/["\\\x0-\x1f]/) { MAP[$&] }
+      string.gsub!('/') { MAP[$&] } if escape_slash
       string.force_encoding(::Encoding::UTF_8)
       string
     end
 
-    def utf8_to_json_ascii(string) # :nodoc:
+    def utf8_to_json_ascii(string, escape_slash = true) # :nodoc:
       string = string.dup
       string.force_encoding(::Encoding::ASCII_8BIT)
-      string.gsub!(/[\/"\\\x0-\x1f]/n) { MAP[$&] }
+      string.gsub!(/["\\\x0-\x1f]/n) { MAP[$&] }
+      string.gsub!('/') { MAP[$&] } if escape_slash
       string.gsub!(/(
                       (?:
                         [\xc2-\xdf][\x80-\xbf]    |
@@ -79,12 +81,15 @@ def valid_utf8?(string)
     end
     module_function :valid_utf8?
   else
-    def utf8_to_json(string) # :nodoc:
-      string.gsub(/[\/"\\\x0-\x1f]/n) { MAP[$&] }
+    def utf8_to_json(string, escape_slash = true) # :nodoc:
+      string = string.gsub(/["\\\x0-\x1f]/n) { MAP[$&] }
+      string.gsub!('/') { MAP[$&] } if escape_slash
+      string
     end
 
-    def utf8_to_json_ascii(string) # :nodoc:
+    def utf8_to_json_ascii(string, escape_slash = true) # :nodoc:
       string = string.gsub(/[\/"\\\x0-\x1f]/) { MAP[$&] }
+      string.gsub!('/') { MAP[$&] } if escape_slash
       string.gsub!(/(
                       (?:
                         [\xc2-\xdf][\x80-\xbf]    |
@@ -167,6 +172,7 @@ def initialize(opts = {})
           @ascii_only            = false
           @quirks_mode           = false
           @buffer_initial_length = 1024
+          @escape_slash          = false
           configure opts
         end
 
@@ -198,6 +204,10 @@ def initialize(opts = {})
         # :stopdoc:
         attr_reader :buffer_initial_length
 
+        # If this attribute is set to true, forward slashes will be escaped in
+        # all json strings.
+        attr_accessor :escape_slash
+
         def buffer_initial_length=(length)
           if length > 0
             @buffer_initial_length = length
@@ -239,6 +249,11 @@ def quirks_mode?
           @quirks_mode
         end
 
+        # Returns true, if forward slashes are escaped. Otherwise returns false.
+        def escape_slash?
+          @escape_slash
+        end
+
         # Configure this State instance with the Hash _opts_, and return
         # itself.
         def configure(opts)
@@ -262,6 +277,7 @@ def configure(opts)
           @depth                 = opts[:depth] || 0
           @quirks_mode           = opts[:quirks_mode] if opts.key?(:quirks_mode)
           @buffer_initial_length ||= opts[:buffer_initial_length]
+          @escape_slash          = !!opts[:escape_slash] if opts.key?(:escape_slash)
 
           if !opts.key?(:max_nesting) # defaults to 100
             @max_nesting = 100
@@ -450,9 +466,9 @@ def to_json(state = nil, *args)
                 string = encode(::Encoding::UTF_8)
               end
               if state.ascii_only?
-                '"' << JSON.utf8_to_json_ascii(string) << '"'
+                '"' << JSON.utf8_to_json_ascii(string, state.escape_slash) << '"'
               else
-                '"' << JSON.utf8_to_json(string) << '"'
+                '"' << JSON.utf8_to_json(string, state.escape_slash) << '"'
               end
             end
           else
@@ -462,9 +478,9 @@ def to_json(state = nil, *args)
             def to_json(state = nil, *args)
               state = State.from_state(state)
               if state.ascii_only?
-                '"' << JSON.utf8_to_json_ascii(self) << '"'
+                '"' << JSON.utf8_to_json_ascii(self, state.escape_slash) << '"'
               else
-                '"' << JSON.utf8_to_json(self) << '"'
+                '"' << JSON.utf8_to_json(self, state.escape_slash) << '"'
               end
             end
           end

tests/test_json.rb

@@ -410,10 +410,15 @@ def test_backslash
     assert_equal data, JSON.parse(json)
     #
     data = [ '/' ]
-    json = '["\/"]'
+    json = '["/"]'
     assert_equal json, JSON.generate(data)
     assert_equal data, JSON.parse(json)
     #
+    data = [ '/' ]
+    json = '["\/"]'
+    assert_equal json, JSON.generate(data, :escape_slash => true)
+    assert_equal data, JSON.parse(json)
+    #
     json = '["\""]'
     data = JSON.parse(json)
     assert_equal ['"'], data

tests/test_json_generate.rb

@@ -142,6 +142,7 @@ def test_pretty_state
       :buffer_initial_length => 1024,
       :quirks_mode           => false,
       :depth                 => 0,
+      :escape_slash          => false,
       :indent                => "  ",
       :max_nesting           => 100,
       :object_nl             => "\n",
@@ -159,6 +160,7 @@ def test_safe_state
       :buffer_initial_length => 1024,
       :quirks_mode           => false,
       :depth                 => 0,
+      :escape_slash          => false,
       :indent                => "",
       :max_nesting           => 100,
       :object_nl             => "",
@@ -176,6 +178,7 @@ def test_fast_state
       :buffer_initial_length => 1024,
       :quirks_mode           => false,
       :depth                 => 0,
+      :escape_slash          => false,
       :indent                => "",
       :max_nesting           => 0,
       :object_nl             => "",