json 3.0.0.rc1 → 3.0.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 297c71ea2d3c5ecf3b479b5b3708cb36ff532bd700e05e7eea48c53c0e79d96f
-  data.tar.gz: ca9386834406f49385b0e6e091f3f9db7ef66ccfe2632b3c87d340d212872d4d
+  metadata.gz: b4a457db7bd0eba155736808c796e81741e7ea72438768e183a7b7a795e8de4e
+  data.tar.gz: 61bb013e78224c5c91f620d3097986735411befd5139f33eaecb2c3c510ababe
 SHA512:
-  metadata.gz: 616c27a9a4ca01fce4bc13dc85da3e38bfbc8c13daa57c56396b77e61dcc215b1a6cf2994e1b7ba3c7bdbacf9c6b21ce7b7a147a057c9968d447ebe2f19039ba
-  data.tar.gz: 3f79549799744cfcc517fedcd722e1644fe44e8497176b5864f882174b48247e2b5c89e14f68acae527c3ee807908b20dc799daf04071acdf3ce363f235b492c
+  metadata.gz: 6e0b4b4a8160235dcef28c7924ecda754a7ae792a714e6b5fbc5baa810fd358129c4340dd1056d31df823bf139f2b42bccb9c2d400c7282079d3dcf3b2f8c2c8
+  data.tar.gz: e67782447e1934dcada56b848982e6bb15708997b9eca220f7b631470734e0b14b2b227ac82612b03b8eb6b3d54b7686d4627a6d551d05b8252da40203f07087

data/CHANGES.md

@@ -2,6 +2,11 @@
 
 ### Unreleased
 
+### 2026-09-07 (3.0.0)
+
+* Add `JSON::ParserError#json_path` to locate parse errors in the document as a JSONPath-style string (e.g. `$.foo[0].bar`). For duplicate key errors it points at the duplicated key itself.
+* Fix the parser to also reject lone trailing UTF-16 surrogates (`\uDCxx` with no leading partner), symmetric to the leading-surrogate case. The Java parser already rejected these; this closes the CRuby/JRuby parity gap.
+
 ### 2026-08-11 (3.0.0.rc1)
 
 With the removal of the insecure `create_additions` option, `JSON.load` and `JSON.dump` are
@@ -24,8 +29,8 @@ JavaScript comments in documents are no longer supported by default.
 Numerous rarely used aliases have been removed.
 
 * `JSON.load` defaults are now safe to use.
-* All unknown options are now cause an `ArgumentError` rather than to be ignored.
-* The `allow_comments` parsing option now default to `false`.
+* All unknown options will now cause an `ArgumentError` rather than to be ignored.
+* The `allow_comments` parsing option now defaults to `false`.
 * The `allow_duplicate_key` option now defaults to `false`, for both parsing and generating JSON.
 * Removed the `limit` positional argument of `JSON.dump`.
 * Removed the `escape_slash` alias of `script_safe`.

data/ext/json/ext/generator/generator.c

@@ -1124,15 +1124,6 @@ static void generate_json_fallback(FBuffer *buffer, struct generate_json_data *d
     }
 }
 
-static inline void generate_json_symbol(FBuffer *buffer, struct generate_json_data *data, VALUE obj)
-{
-    if (data->state->strict) {
-        generate_json_string(buffer, data, rb_sym2str(obj));
-    } else {
-        generate_json_fallback(buffer, data, obj);
-    }
-}
-
 static void generate_json_null(FBuffer *buffer, struct generate_json_data *data, VALUE obj)
 {
     fbuffer_append(buffer, "null", 4);
@@ -1218,7 +1209,13 @@ start:
         } else if (RB_FLONUM_P(obj)) {
             generate_json_float(buffer, data, obj);
         } else if (RB_STATIC_SYM_P(obj)) {
-            generate_json_symbol(buffer, data, obj);
+            if (data->state->strict) {
+                obj = rb_sym2str(obj);
+                JSON_ASSERT(RBASIC_CLASS(obj) == rb_cString);
+                goto generate_string;
+            }
+
+            generate_json_fallback(buffer, data, obj);
         } else {
             goto general;
         }
@@ -1239,6 +1236,7 @@ start:
             case T_STRING:
                 if (fallback && klass != rb_cString) goto general;
 
+            generate_string:
                 if (RB_LIKELY(valid_json_string_p(obj))) {
                     raw_generate_json_string(buffer, data, obj);
                 } else if (as_json_called) {
@@ -1250,7 +1248,13 @@ start:
                 }
                 break;
             case T_SYMBOL:
-                generate_json_symbol(buffer, data, obj);
+                if (data->state->strict) {
+                    obj = rb_sym2str(obj);
+                    JSON_ASSERT(RBASIC_CLASS(obj) == rb_cString);
+                    goto generate_string;
+                }
+
+                generate_json_fallback(buffer, data, obj);
                 break;
             case T_FLOAT:
                 if (fallback && klass != rb_cFloat) goto general;

data/ext/json/ext/parser/parser.c

@@ -5,7 +5,7 @@
 static VALUE mJSON, eNestingError, eParserError, Encoding_UTF_8;
 static VALUE CNaN, CInfinity, CMinusInfinity, JSON_empty_string;
 
-static ID i_new, i_try_convert, i_encode, i_at_line, i_at_column;
+static ID i_new, i_try_convert, i_encode, i_at_line, i_at_column, i_at_json_path;
 #ifndef HAVE_RB_STR_TO_INTERNED_STR
 static ID i_uminus;
 #endif
@@ -651,11 +651,42 @@ static VALUE build_parse_error_message(const char *format, JSON_ParserState *sta
     return rb_enc_sprintf(enc_utf8, format, ptr);
 }
 
+static VALUE json_path_new(JSON_ParserState *state, VALUE duplicate_key)
+{
+    VALUE path = rb_ary_new_capa(state->current_nesting);
+
+    json_frame_stack *frames = state->frames;
+    rvalue_stack *values = state->value_stack;
+
+    for (long depth = 1; depth < frames->head; depth++) {
+        json_frame *frame = &frames->ptr[depth];
+
+        bool innermost = depth == frames->head - 1;
+        long child_head = innermost ? values->head : frames->ptr[depth + 1].value_stack_head;
+        long count = child_head - frame->value_stack_head;
+
+        if (frame->type == JSON_FRAME_ARRAY) {
+            rb_ary_push(path, LONG2NUM(frame->phase == JSON_PHASE_ARRAY_COMMA ? count - 1 : count));
+        } else if (innermost && !UNDEF_P(duplicate_key)) {
+            rb_ary_push(path, duplicate_key);
+        } else if (count & 1) {
+            rb_ary_push(path, values->ptr[child_head - 1]);
+        } else if (frame->phase == JSON_PHASE_OBJECT_COMMA && count >= 2) {
+            rb_ary_push(path, values->ptr[child_head - 2]);
+        } else {
+            break;
+        }
+    }
+
+    return path;
+}
+
 static VALUE parse_error_new(JSON_ParserState *state, VALUE message, long line, long column, bool eos)
 {
     VALUE exc = rb_exc_new_str(eParserError, message);
     rb_ivar_set(exc, i_at_line, LONG2NUM(line));
     rb_ivar_set(exc, i_at_column, LONG2NUM(column));
+    rb_ivar_set(exc, i_at_json_path, json_path_new(state, Qundef));
     return exc;
 }
 
@@ -1010,6 +1041,9 @@ NOINLINE(static) VALUE json_string_unescape(JSON_ParserState *state, JSON_Parser
                         raise_syntax_error_at("incomplete surrogate pair at %s", state, p);
                         break;
                     }
+                } else if ((ch & 0xFC00) == 0xDC00) {
+                    raise_syntax_error_at("unpaired trailing surrogate at %s", state, p);
+                    break;
                 }
 
                 int unescape_len = convert_UTF32_to_UTF8(buffer, ch);
@@ -1199,14 +1233,17 @@ NORETURN(static) void raise_duplicate_key_error(JSON_ParserState *state, VALUE d
     );
 
     rb_str_concat(message, build_parse_error_message("", state));
+    VALUE exc;
     if (state->parser) { // line and columns can't be accurate in resumable
-        rb_exc_raise(parse_error_new(state, message, 0, 0, false));
+        exc = parse_error_new(state, message, 0, 0, false);
     } else {
         long line, column;
         cursor_position(state, &line, &column);
         rb_str_catf(message, " at line %ld column %ld", line, column);
-        rb_exc_raise(parse_error_new(state, message, line, column, false));
+        exc = parse_error_new(state, message, line, column, false);
     }
+    rb_ivar_set(exc, i_at_json_path, json_path_new(state, duplicate_key));
+    rb_exc_raise(exc);
 }
 
 NOINLINE(static) void json_on_duplicate_key(JSON_ParserState *state, JSON_ParserConfig *config, size_t count, const VALUE *pairs)
@@ -2122,6 +2159,12 @@ static VALUE cParser_parse(JSON_ParserConfig *config, VALUE src)
     // the rvalue stack.
     VALUE result = complete ? *rvalue_stack_peek(state->value_stack, 1) : Qundef;
 
+    if (complete) {
+        json_ensure_eof(state, config);
+    } else {
+        raise_eos_error("unexpected end of input", state);
+    }
+
     // This may be skipped in case of exception, but
     // it won't cause a leak.
     rvalue_stack_eagerly_release(value_stack_handle);
@@ -2130,12 +2173,6 @@ static VALUE cParser_parse(JSON_ParserConfig *config, VALUE src)
     RB_GC_GUARD(frame_stack_handle);
     RB_GC_GUARD(Vsource);
 
-    if (complete) {
-        json_ensure_eof(state, config);
-    } else {
-        raise_eos_error("unexpected end of input", state);
-    }
-
     return result;
 }
 
@@ -2871,6 +2908,7 @@ void Init_parser(void)
     i_encode = rb_intern("encode");
     i_at_line = rb_intern("@line");
     i_at_column = rb_intern("@column");
+    i_at_json_path = rb_intern("@json_path");
 
     binary_encindex = rb_ascii8bit_encindex();
     utf8_encindex = rb_utf8_encindex();

data/lib/json/common.rb

@@ -142,7 +142,56 @@ module JSON
 
   # This exception is raised if a parser error occurs.
   class ParserError < JSONError
-    attr_reader :line, :column
+    # Line number where the parser encountered an error.
+    # Is <tt>nil</tt> when raised by JSON::ResumableParser.
+    attr_reader :line
+
+    # Column number where the parser encountered an error.
+    # Is <tt>nil</tt> when raised by JSON::ResumableParser.
+    attr_reader :column
+
+    # Returns a best effort JSONPath string representing where in the document
+    # the parser encountered an error:
+    #
+    #   begin
+    #     JSON.parse('{"articles": [ { "title": invalid } ]}')
+    #   rescue JSON::ParserError => error
+    #     error.json_path # => "$.articles[0].title"
+    #   end
+    def json_path
+      return @json_path if String === @json_path
+
+      if Array === @json_path
+        path = build_json_path(@json_path)
+        @json_path = path unless frozen?
+        return path
+      end
+    end
+
+    private
+
+    def build_json_path(segments)
+      error = false
+      path = segments.filter_map do |segment|
+        next if error
+
+        case segment
+        when Integer
+          "[#{segment}]"
+        when String, Symbol
+          if segment.match?(/\A[a-zA-Z\$\_][a-zA-Z\$\_0-9]*\z/)
+            ".#{segment}"
+          else
+            segment = segment.to_s.gsub(/["\\]/, { '"' => '\\"', '\\' => '\\\\' })
+            %{["#{segment}"]}
+          end
+        else
+          error = true
+          nil
+        end
+      end.join
+      "$#{path}".freeze
+    end
   end
 
   # This exception is raised if the nesting of parsed data structures is too

data/lib/json/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module JSON
-  VERSION = '3.0.0.rc1'
+  VERSION = '3.0.0'
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: json
 version: !ruby/object:Gem::Version
-  version: 3.0.0.rc1
+  version: 3.0.0
 platform: ruby
 authors:
 - Florian Frank