json-maglev- 1.5.4 → 1.6.1

data/ext/json/ext/parser/parser.h

@@ -3,7 +3,7 @@
 
 #include "ruby.h"
 
-#if HAVE_RE_H
+#ifndef HAVE_RUBY_RE_H
 #include "re.h"
 #endif
 
@@ -36,15 +36,16 @@ typedef unsigned char UTF8;   /* typically 8 bits */
 typedef struct JSON_ParserStruct {
     VALUE dwrapped_parser;
     VALUE Vsource;
-    const char *source;
+    char *source;
     long len;
-    const char *memo;
+    char *memo;
     VALUE create_id;
     int max_nesting;
     int current_nesting;
     int allow_nan;
     int parsing_name;
     int symbolize_names;
+    int quirks_mode;
     VALUE object_class;
     VALUE array_class;
     int create_additions;
@@ -63,13 +64,13 @@ typedef struct JSON_ParserStruct {
 
 static UTF32 unescape_unicode(const unsigned char *p);
 static int convert_UTF32_to_UTF8(char *buf, UTF32 ch);
-static const char *JSON_parse_object(JSON_Parser *json, const char *p, const char *pe, VALUE *result);
-static const char *JSON_parse_value(JSON_Parser *json, const char *p, const char *pe, VALUE *result);
-static const char *JSON_parse_integer(JSON_Parser *json, const char *p, const char *pe, VALUE *result);
-static const char *JSON_parse_float(JSON_Parser *json, const char *p, const char *pe, VALUE *result);
-static const char *JSON_parse_array(JSON_Parser *json, const char *p, const char *pe, VALUE *result);
-static VALUE json_string_unescape(VALUE result, const char *string, const char *stringEnd);
-static const char *JSON_parse_string(JSON_Parser *json, const char *p, const char *pe, VALUE *result);
+static char *JSON_parse_object(JSON_Parser *json, char *p, char *pe, VALUE *result);
+static char *JSON_parse_value(JSON_Parser *json, char *p, char *pe, VALUE *result);
+static char *JSON_parse_integer(JSON_Parser *json, char *p, char *pe, VALUE *result);
+static char *JSON_parse_float(JSON_Parser *json, char *p, char *pe, VALUE *result);
+static char *JSON_parse_array(JSON_Parser *json, char *p, char *pe, VALUE *result);
+static VALUE json_string_unescape(VALUE result, char *string, char *stringEnd);
+static char *JSON_parse_string(JSON_Parser *json, char *p, char *pe, VALUE *result);
 static VALUE convert_encoding(VALUE source);
 static VALUE cParser_initialize(int argc, VALUE *argv, VALUE self);
 static VALUE cParser_parse(VALUE self);

data/ext/json/ext/parser/parser.rl

@@ -76,8 +76,9 @@ static VALUE mJSON, mExt, cParser, eParserError, eNestingError;
 static VALUE CNaN, CInfinity, CMinusInfinity;
 
 static ID i_json_creatable_p, i_json_create, i_create_id, i_create_additions,
-          i_chr, i_max_nesting, i_allow_nan, i_symbolize_names, i_object_class,
-          i_array_class, i_key_p, i_deep_const_get, i_match, i_match_string, i_aset, i_leftshift;
+          i_chr, i_max_nesting, i_allow_nan, i_symbolize_names, i_quirks_mode,
+          i_object_class, i_array_class, i_key_p, i_deep_const_get, i_match,
+          i_match_string, i_aset, i_leftshift;
 
 %%{
     machine JSON_common;
@@ -115,7 +116,7 @@ static ID i_json_creatable_p, i_json_create, i_create_id, i_create_additions,
 
     action parse_value {
         VALUE v = Qnil;
-        const char *np = JSON_parse_value(json, fpc, pe, &v);
+        char *np = JSON_parse_value(json, fpc, pe, &v);
         if (np == NULL) {
             fhold; fbreak;
         } else {
@@ -129,7 +130,7 @@ static ID i_json_creatable_p, i_json_create, i_create_id, i_create_additions,
     }
 
     action parse_name {
-        const char *np;
+        char *np;
         json->parsing_name = 1;
         np = JSON_parse_string(json, fpc, pe, &last_name);
         json->parsing_name = 0;
@@ -138,16 +139,17 @@ static ID i_json_creatable_p, i_json_create, i_create_id, i_create_additions,
 
     action exit { fhold; fbreak; }
 
-    a_pair  = ignore* begin_name >parse_name
-        ignore* name_separator ignore*
-        begin_value >parse_value;
+    pair  = ignore* begin_name >parse_name ignore* name_separator ignore* begin_value >parse_value;
+    next_pair   = ignore* value_separator pair;
 
-    main := begin_object
-          (a_pair (ignore* value_separator a_pair)*)?
-          ignore* end_object @exit;
+    main := (
+      begin_object
+      (pair (next_pair)*)? ignore*
+      end_object
+    ) @exit;
 }%%
 
-static const char *JSON_parse_object(JSON_Parser *json, const char *p, const char *pe, VALUE *result)
+static char *JSON_parse_object(JSON_Parser *json, char *p, char *pe, VALUE *result)
 {
     int cs = EVIL;
     VALUE last_name = Qnil;
@@ -178,6 +180,7 @@ static const char *JSON_parse_object(JSON_Parser *json, const char *p, const cha
     }
 }
 
+
 %%{
     machine JSON_value;
     include JSON_common;
@@ -208,13 +211,13 @@ static const char *JSON_parse_object(JSON_Parser *json, const char *p, const cha
         }
     }
     action parse_string {
-        const char *np = JSON_parse_string(json, fpc, pe, result);
+        char *np = JSON_parse_string(json, fpc, pe, result);
         if (np == NULL) { fhold; fbreak; } else fexec np;
     }
 
     action parse_number {
-        const char *np;
-        if(pe > fpc + 9 && !strncmp(MinusInfinity, fpc, 9)) {
+        char *np;
+        if(pe > fpc + 9 - json->quirks_mode && !strncmp(MinusInfinity, fpc, 9)) {
             if (json->allow_nan) {
                 *result = CMinusInfinity;
                 fexec p + 10;
@@ -231,7 +234,7 @@ static const char *JSON_parse_object(JSON_Parser *json, const char *p, const cha
     }
 
     action parse_array {
-        const char *np;
+        char *np;
         json->current_nesting++;
         np = JSON_parse_array(json, fpc, pe, result);
         json->current_nesting--;
@@ -239,7 +242,7 @@ static const char *JSON_parse_object(JSON_Parser *json, const char *p, const cha
     }
 
     action parse_object {
-        const char *np;
+        char *np;
         json->current_nesting++;
         np =  JSON_parse_object(json, fpc, pe, result);
         json->current_nesting--;
@@ -261,7 +264,7 @@ main := (
         ) %*exit;
 }%%
 
-static const char *JSON_parse_value(JSON_Parser *json, const char *p, const char *pe, VALUE *result)
+static char *JSON_parse_value(JSON_Parser *json, char *p, char *pe, VALUE *result)
 {
     int cs = EVIL;
 
@@ -282,10 +285,10 @@ static const char *JSON_parse_value(JSON_Parser *json, const char *p, const char
 
     action exit { fhold; fbreak; }
 
-    main := '-'? ('0' | [1-9][0-9]*) (^[0-9] @exit);
+    main := '-'? ('0' | [1-9][0-9]*) (^[0-9]? @exit);
 }%%
 
-static const char *JSON_parse_integer(JSON_Parser *json, const char *p, const char *pe, VALUE *result)
+static char *JSON_parse_integer(JSON_Parser *json, char *p, char *pe, VALUE *result)
 {
     int cs = EVIL;
 
@@ -313,10 +316,10 @@ static const char *JSON_parse_integer(JSON_Parser *json, const char *p, const ch
     main := '-'? (
               (('0' | [1-9][0-9]*) '.' [0-9]+ ([Ee] [+\-]?[0-9]+)?)
               | (('0' | [1-9][0-9]*) ([Ee] [+\-]?[0-9]+))
-             )  (^[0-9Ee.\-] @exit );
+             )  (^[0-9Ee.\-]? @exit );
 }%%
 
-static const char *JSON_parse_float(JSON_Parser *json, const char *p, const char *pe, VALUE *result)
+static char *JSON_parse_float(JSON_Parser *json, char *p, char *pe, VALUE *result)
 {
     int cs = EVIL;
 
@@ -342,7 +345,7 @@ static const char *JSON_parse_float(JSON_Parser *json, const char *p, const char
 
     action parse_value {
         VALUE v = Qnil;
-        const char *np = JSON_parse_value(json, fpc, pe, &v);
+        char *np = JSON_parse_value(json, fpc, pe, &v);
         if (np == NULL) {
             fhold; fbreak;
         } else {
@@ -365,7 +368,7 @@ static const char *JSON_parse_float(JSON_Parser *json, const char *p, const char
           end_array @exit;
 }%%
 
-static const char *JSON_parse_array(JSON_Parser *json, const char *p, const char *pe, VALUE *result)
+static char *JSON_parse_array(JSON_Parser *json, char *p, char *pe, VALUE *result)
 {
     int cs = EVIL;
     VALUE array_class = json->array_class;
@@ -386,9 +389,9 @@ static const char *JSON_parse_array(JSON_Parser *json, const char *p, const char
     }
 }
 
-static VALUE json_string_unescape(VALUE result, const char *string, const char *stringEnd)
+static VALUE json_string_unescape(VALUE result, char *string, char *stringEnd)
 {
-    const char *p = string, *pe = string, *unescape;
+    char *p = string, *pe = string, *unescape;
     int unescape_len;
 
     while (pe < stringEnd) {
@@ -490,17 +493,17 @@ match_i(VALUE regexp, VALUE klass, VALUE memo)
     return ST_CONTINUE;
 }
 
-static const char *JSON_parse_string(JSON_Parser *json, const char *p, const char *pe, VALUE *result)
+static char *JSON_parse_string(JSON_Parser *json, char *p, char *pe, VALUE *result)
 {
     int cs = EVIL;
+    VALUE match_string;
+
     *result = rb_str_buf_new(0);
     %% write init;
     json->memo = p;
     %% write exec;
 
-    if (json->create_additions) {
-      VALUE match_string = json->match_string;
-      if (RTEST(match_string)) {
+    if (json->create_additions && RTEST(match_string = json->match_string)) {
           VALUE klass;
           VALUE memo = rb_ary_new2(2);
           rb_ary_push(memo, *result);
@@ -509,7 +512,6 @@ static const char *JSON_parse_string(JSON_Parser *json, const char *p, const cha
           if (RTEST(klass)) {
               *result = rb_funcall(klass, i_json_create, 1, *result);
           }
-      }
     }
 
     if (json->symbolize_names && json->parsing_name) {
@@ -522,34 +524,6 @@ static const char *JSON_parse_string(JSON_Parser *json, const char *p, const cha
     }
 }
 
-
-%%{
-    machine JSON;
-
-    write data;
-
-    include JSON_common;
-
-    action parse_object {
-        const char *np;
-        json->current_nesting = 1;
-        np = JSON_parse_object(json, fpc, pe, &result);
-        if (np == NULL) { fhold; fbreak; } else fexec np;
-    }
-
-    action parse_array {
-        const char *np;
-        json->current_nesting = 1;
-        np = JSON_parse_array(json, fpc, pe, &result);
-        if (np == NULL) { fhold; fbreak; } else fexec np;
-    }
-
-    main := ignore* (
-            begin_object >parse_object |
-            begin_array >parse_array
-            ) ignore*;
-}%%
-
 /*
  * Document-class: JSON::Ext::Parser
  *
@@ -564,7 +538,7 @@ static const char *JSON_parse_string(JSON_Parser *json, const char *p, const cha
 
 static VALUE convert_encoding(VALUE source)
 {
-    const char *ptr = RSTRING_PTR(source);
+    char *ptr = RSTRING_PTR(source);
     long len = RSTRING_LEN(source);
     if (len < 2) {
         rb_raise(eParserError, "A JSON text must at least contain two octets!");
@@ -603,12 +577,15 @@ static VALUE convert_encoding(VALUE source)
     return source;
 }
 
-static inline void parser_iv_set(JSON_Parser *json, const char* iv_name, VALUE v)
-{
-  // store reference to v in a Ruby instVar to keep v alive
-  //  without using a gc_mark function in Data_Wrap_Struct calls
-  rb_iv_set(json->dwrapped_parser, iv_name, v);
-}
+#if defined MAGLEV
+// Maglev doesn't support the mark function, keep a reference in the object
+#define QUOTE(x) #x
+#define PARSER_SET_REFERENCE(json, field, val)                          \
+    (json)->field = (val);                                              \
+    rb_iv_set(json->dwrapped_parser, QUOTE(@field), (json)->field);
+#else
+#define PARSER_SET_REFERENCE(json, field, val) (json)->field = (val);
+#endif
 
 /*
  * call-seq: new(source, opts => {})
@@ -636,26 +613,15 @@ static inline void parser_iv_set(JSON_Parser *json, const char* iv_name, VALUE v
  * * *object_class*: Defaults to Hash
  * * *array_class*: Defaults to Array
  */
-
-static int init_count = 0;
 static VALUE cParser_initialize(int argc, VALUE *argv, VALUE self)
 {
-    const char *ptr;
-    long len;
     VALUE source, opts;
-
     GET_PARSER_INIT;
-    init_count += 1;
 
-#ifndef MAGLEV
-    if (json->Vsource) {
+    if (RTEST(json->Vsource)) {
         rb_raise(rb_eTypeError, "already initialized instance");
     }
-#endif
     rb_scan_args(argc, argv, "11", &source, &opts);
-    source = convert_encoding(StringValue(source));
-    ptr = RSTRING_PTR(source);
-    len = RSTRING_LEN(source);
     if (!NIL_P(opts)) {
         opts = rb_convert_type(opts, T_HASH, "Hash", "to_hash");
         if (NIL_P(opts)) {
@@ -685,6 +651,13 @@ static VALUE cParser_initialize(int argc, VALUE *argv, VALUE self)
             } else {
                 json->symbolize_names = 0;
             }
+            tmp = ID2SYM(i_quirks_mode);
+            if (option_given_p(opts, tmp)) {
+                VALUE quirks_mode = rb_hash_aref(opts, tmp);
+                json->quirks_mode = RTEST(quirks_mode) ? 1 : 0;
+            } else {
+                json->quirks_mode = 0;
+            }
             tmp = ID2SYM(i_create_additions);
             if (option_given_p(opts, tmp)) {
                 json->create_additions = RTEST(rb_hash_aref(opts, tmp));
@@ -693,52 +666,78 @@ static VALUE cParser_initialize(int argc, VALUE *argv, VALUE self)
             }
             tmp = ID2SYM(i_create_id);
             if (option_given_p(opts, tmp)) {
-                json->create_id = rb_hash_aref(opts, tmp);
+                PARSER_SET_REFERENCE(json, create_id, rb_hash_aref(opts, tmp));
             } else {
-                json->create_id = rb_funcall(mJSON, i_create_id, 0);
+                PARSER_SET_REFERENCE(json, create_id, rb_funcall(mJSON, i_create_id, 0));
             }
-            parser_iv_set(json, "@create_id", json->create_id);
             tmp = ID2SYM(i_object_class);
             if (option_given_p(opts, tmp)) {
-                json->object_class = rb_hash_aref(opts, tmp);
-                parser_iv_set(json, "@object_class", json->object_class);
+                PARSER_SET_REFERENCE(json, object_class, rb_hash_aref(opts, tmp));
+            } else {
+                PARSER_SET_REFERENCE(json, object_class, Qnil);
             }
             tmp = ID2SYM(i_array_class);
             if (option_given_p(opts, tmp)) {
-                json->array_class = rb_hash_aref(opts, tmp);
-                parser_iv_set(json, "@array_class", json->array_class);
+                PARSER_SET_REFERENCE(json, array_class, rb_hash_aref(opts, tmp));
+            } else {
+                PARSER_SET_REFERENCE(json, array_class, Qnil);
             }
             tmp = ID2SYM(i_match_string);
             if (option_given_p(opts, tmp)) {
                 VALUE match_string = rb_hash_aref(opts, tmp);
-                json->match_string = RTEST(match_string) ? match_string : Qnil;
-                parser_iv_set(json, "@match_string", json->match_string);
+                PARSER_SET_REFERENCE(json, match_string, RTEST(match_string) ? match_string : Qnil);
+            } else {
+                PARSER_SET_REFERENCE(json, match_string, Qnil);
             }
         }
     } else {
         json->max_nesting = 19;
         json->allow_nan = 0;
         json->create_additions = 1;
-        json->create_id = rb_funcall(mJSON, i_create_id, 0);
-        parser_iv_set(json, "@create_id", json->create_id);
+        PARSER_SET_REFERENCE(json, create_id, rb_funcall(mJSON, i_create_id, 0));
+        json->object_class = Qnil;
+        json->array_class = Qnil;
+    }
+    if (!json->quirks_mode) {
+      source = convert_encoding(StringValue(source));
     }
     json->current_nesting = 0;
-    json->len = len;
-    json->source = ptr;
-    json->Vsource = source;
-    parser_iv_set(json, "@vsource", json->Vsource);
+    json->len = RSTRING_LEN(source);
+    json->source = RSTRING_PTR(source);;
+    PARSER_SET_REFERENCE(json, Vsource, source);
     return self;
 }
 
-/*
- * call-seq: parse()
- *
- *  Parses the current JSON text _source_ and returns the complete data
- *  structure as a result.
- */
-static VALUE cParser_parse(VALUE self)
+%%{
+    machine JSON;
+
+    write data;
+
+    include JSON_common;
+
+    action parse_object {
+        char *np;
+        json->current_nesting = 1;
+        np = JSON_parse_object(json, fpc, pe, &result);
+        if (np == NULL) { fhold; fbreak; } else fexec np;
+    }
+
+    action parse_array {
+        char *np;
+        json->current_nesting = 1;
+        np = JSON_parse_array(json, fpc, pe, &result);
+        if (np == NULL) { fhold; fbreak; } else fexec np;
+    }
+
+    main := ignore* (
+            begin_object >parse_object |
+            begin_array >parse_array
+            ) ignore*;
+}%%
+
+static VALUE cParser_parse_strict(VALUE self)
 {
-    const char *p, *pe;
+    char *p, *pe;
     int cs = EVIL;
     VALUE result = Qnil;
     GET_PARSER;
@@ -756,6 +755,62 @@ static VALUE cParser_parse(VALUE self)
     }
 }
 
+
+%%{
+    machine JSON_quirks_mode;
+
+    write data;
+
+    include JSON_common;
+
+    action parse_value {
+        char *np = JSON_parse_value(json, fpc, pe, &result);
+        if (np == NULL) { fhold; fbreak; } else fexec np;
+    }
+
+    main := ignore* (
+            begin_value >parse_value
+            ) ignore*;
+}%%
+
+static VALUE cParser_parse_quirks_mode(VALUE self)
+{
+    char *p, *pe;
+    int cs = EVIL;
+    VALUE result = Qnil;
+    GET_PARSER;
+
+    %% write init;
+    p = json->source;
+    pe = p + json->len;
+    %% write exec;
+
+    if (cs >= JSON_quirks_mode_first_final && p == pe) {
+        return result;
+    } else {
+        rb_raise(eParserError, "%u: unexpected token at '%s'", __LINE__, p);
+        return Qnil;
+    }
+}
+
+/*
+ * call-seq: parse()
+ *
+ *  Parses the current JSON text _source_ and returns the complete data
+ *  structure as a result.
+ */
+static VALUE cParser_parse(VALUE self)
+{
+  GET_PARSER;
+
+  if (json->quirks_mode) {
+    return cParser_parse_quirks_mode(self);
+  } else {
+    return cParser_parse_strict(self);
+  }
+}
+
+
 static JSON_Parser *JSON_allocate()
 {
     JSON_Parser *json = ALLOC(JSON_Parser);
@@ -770,7 +825,14 @@ static JSON_Parser *JSON_allocate()
     return json;
 }
 
-/* deleted JSON_mark function */
+static void JSON_mark(JSON_Parser *json)
+{
+    rb_gc_mark_maybe(json->Vsource);
+    rb_gc_mark_maybe(json->create_id);
+    rb_gc_mark_maybe(json->object_class);
+    rb_gc_mark_maybe(json->array_class);
+    rb_gc_mark_maybe(json->match_string);
+}
 
 static void JSON_free(JSON_Parser *json)
 {
@@ -780,9 +842,7 @@ static void JSON_free(JSON_Parser *json)
 static VALUE cJSON_parser_s_allocate(VALUE klass)
 {
     JSON_Parser *json = JSON_allocate();
-    VALUE data_obj = Data_Wrap_Struct(klass, NULL, JSON_free, json);
-    json->dwrapped_parser = data_obj;
-    return data_obj;
+    return json->dwrapped_parser = Data_Wrap_Struct(klass, JSON_mark, JSON_free, json);
 }
 
 /*
@@ -797,6 +857,18 @@ static VALUE cParser_source(VALUE self)
     return rb_str_dup(json->Vsource);
 }
 
+/*
+ * call-seq: quirks_mode?()
+ *
+ * Returns a true, if this parser is in quirks_mode, false otherwise.
+ */
+static VALUE cParser_quirks_mode_p(VALUE self)
+{
+    GET_PARSER;
+    return json->quirks_mode ? Qtrue : Qfalse;
+}
+
+
 void Init_parser()
 {
     rb_require("json/common");
@@ -809,6 +881,7 @@ void Init_parser()
     rb_define_method(cParser, "initialize", cParser_initialize, -1);
     rb_define_method(cParser, "parse", cParser_parse, 0);
     rb_define_method(cParser, "source", cParser_source, 0);
+    rb_define_method(cParser, "quirks_mode?", cParser_quirks_mode_p, 0);
 
     CNaN = rb_const_get(mJSON, rb_intern("NaN"));
     CInfinity = rb_const_get(mJSON, rb_intern("Infinity"));
@@ -822,6 +895,7 @@ void Init_parser()
     i_max_nesting = rb_intern("max_nesting");
     i_allow_nan = rb_intern("allow_nan");
     i_symbolize_names = rb_intern("symbolize_names");
+    i_quirks_mode = rb_intern("quirks_mode");
     i_object_class = rb_intern("object_class");
     i_array_class = rb_intern("array_class");
     i_match = rb_intern("match");