oj 3.16.16 → 3.17.3

data/ext/oj/parser.c

@@ -75,7 +75,7 @@ enum {
 
 /*
 0123456789abcdef0123456789abcdef */
-static const char value_map[257] = "\
+static const char value_map[258] = "\
 X........ab..a..................\
 a.i..........f..ghhhhhhhhh......\
 ...........................k.m..\
@@ -85,7 +85,7 @@ a.i..........f..ghhhhhhhhh......\
 ................................\
 ................................v";
 
-static const char null_map[257] = "\
+static const char null_map[258] = "\
 ................................\
 ............o...................\
 ................................\
@@ -95,7 +95,7 @@ static const char null_map[257] = "\
 ................................\
 ................................N";
 
-static const char true_map[257] = "\
+static const char true_map[258] = "\
 ................................\
 ............o...................\
 ................................\
@@ -105,7 +105,7 @@ static const char true_map[257] = "\
 ................................\
 ................................T";
 
-static const char false_map[257] = "\
+static const char false_map[258] = "\
 ................................\
 ............o...................\
 ................................\
@@ -115,7 +115,7 @@ static const char false_map[257] = "\
 ................................\
 ................................F";
 
-static const char comma_map[257] = "\
+static const char comma_map[258] = "\
 .........ab..a..................\
 a.i..........f..ghhhhhhhhh......\
 ...........................k....\
@@ -125,7 +125,7 @@ a.i..........f..ghhhhhhhhh......\
 ................................\
 ................................,";
 
-static const char after_map[257] = "\
+static const char after_map[258] = "\
 X........ab..a..................\
 a...........o...................\
 .............................m..\
@@ -135,7 +135,7 @@ a...........o...................\
 ................................\
 ................................a";
 
-static const char key1_map[257] = "\
+static const char key1_map[258] = "\
 .........ab..a..................\
 a.p.............................\
 ................................\
@@ -145,7 +145,7 @@ a.p.............................\
 ................................\
 ................................K";
 
-static const char key_map[257] = "\
+static const char key_map[258] = "\
 .........ab..a..................\
 a.p.............................\
 ................................\
@@ -155,7 +155,7 @@ a.p.............................\
 ................................\
 ................................k";
 
-static const char colon_map[257] = "\
+static const char colon_map[258] = "\
 .........ab..a..................\
 a.........................q.....\
 ................................\
@@ -165,7 +165,7 @@ a.........................q.....\
 ................................\
 ................................:";
 
-static const char neg_map[257] = "\
+static const char neg_map[258] = "\
 ................................\
 ................O---------......\
 ................................\
@@ -175,7 +175,7 @@ static const char neg_map[257] = "\
 ................................\
 ................................-";
 
-static const char zero_map[257] = "\
+static const char zero_map[258] = "\
 .........rs..r..................\
 r...........u.t.................\
 .............................H..\
@@ -185,7 +185,7 @@ r...........u.t.................\
 ................................\
 ................................0";
 
-static const char digit_map[257] = "\
+static const char digit_map[258] = "\
 .........rs..r..................\
 r...........u.t.NNNNNNNNNN......\
 .....w.......................H..\
@@ -195,7 +195,7 @@ r...........u.t.NNNNNNNNNN......\
 ................................\
 ................................d";
 
-static const char dot_map[257] = "\
+static const char dot_map[258] = "\
 ................................\
 ................vvvvvvvvvv......\
 ................................\
@@ -205,7 +205,7 @@ static const char dot_map[257] = "\
 ................................\
 .................................";
 
-static const char frac_map[257] = "\
+static const char frac_map[258] = "\
 .........rs..r..................\
 r...........u...vvvvvvvvvv......\
 .....w.......................H..\
@@ -215,7 +215,7 @@ r...........u...vvvvvvvvvv......\
 ................................\
 ................................f";
 
-static const char exp_sign_map[257] = "\
+static const char exp_sign_map[258] = "\
 ................................\
 ...........x.x..yyyyyyyyyy......\
 ................................\
@@ -225,7 +225,7 @@ static const char exp_sign_map[257] = "\
 ................................\
 ................................x";
 
-static const char exp_zero_map[257] = "\
+static const char exp_zero_map[258] = "\
 ................................\
 ................yyyyyyyyyy......\
 ................................\
@@ -235,7 +235,7 @@ static const char exp_zero_map[257] = "\
 ................................\
 ................................z";
 
-static const char exp_map[257] = "\
+static const char exp_map[258] = "\
 .........rs..r..................\
 r...........u...yyyyyyyyyy......\
 .............................H..\
@@ -245,7 +245,7 @@ r...........u...yyyyyyyyyy......\
 ................................\
 ................................X";
 
-static const char big_digit_map[257] = "\
+static const char big_digit_map[258] = "\
 .........rs..r..................\
 r...........u.D.CCCCCCCCCC......\
 .....J.......................H..\
@@ -255,7 +255,7 @@ r...........u.D.CCCCCCCCCC......\
 ................................\
 ................................D";
 
-static const char big_dot_map[257] = "\
+static const char big_dot_map[258] = "\
 ................................\
 ................IIIIIIIIII......\
 ................................\
@@ -265,7 +265,7 @@ static const char big_dot_map[257] = "\
 ................................\
 ................................o";
 
-static const char big_frac_map[257] = "\
+static const char big_frac_map[258] = "\
 .........rs..r..................\
 r...........u...IIIIIIIIII......\
 .....J.......................H..\
@@ -275,7 +275,7 @@ r...........u...IIIIIIIIII......\
 ................................\
 ................................g";
 
-static const char big_exp_sign_map[257] = "\
+static const char big_exp_sign_map[258] = "\
 ................................\
 ...........K.K..LLLLLLLLLL......\
 ................................\
@@ -285,7 +285,7 @@ static const char big_exp_sign_map[257] = "\
 ................................\
 ................................B";
 
-static const char big_exp_zero_map[257] = "\
+static const char big_exp_zero_map[258] = "\
 ................................\
 ................LLLLLLLLLL......\
 ................................\
@@ -295,7 +295,7 @@ static const char big_exp_zero_map[257] = "\
 ................................\
 ................................Z";
 
-static const char big_exp_map[257] = "\
+static const char big_exp_map[258] = "\
 .........rs..r..................\
 r...........u...LLLLLLLLLL......\
 .............................H..\
@@ -305,7 +305,7 @@ r...........u...LLLLLLLLLL......\
 ................................\
 ................................Y";
 
-static const char string_map[257] = "\
+static const char string_map[258] = "\
 ................................\
 RRzRRRRRRRRRRRRRRRRRRRRRRRRRRRRR\
 RRRRRRRRRRRRRRRRRRRRRRRRRRRRARRR\
@@ -315,7 +315,7 @@ RRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRR\
 MMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMM\
 PPPPPPPPPPPPPPPPQQQQQQQQ........s";
 
-static const char esc_map[257] = "\
+static const char esc_map[258] = "\
 ................................\
 ..B............B................\
 ............................B...\
@@ -325,7 +325,7 @@ static const char esc_map[257] = "\
 ................................\
 ................................~";
 
-static const char esc_byte_map[257] = "\
+static const char esc_byte_map[258] = "\
 ................................\
 ..\"............/................\
 ............................\\...\
@@ -335,7 +335,7 @@ static const char esc_byte_map[257] = "\
 ................................\
 ................................b";
 
-static const char u_map[257] = "\
+static const char u_map[258] = "\
 ................................\
 ................EEEEEEEEEE......\
 .EEEEEE.........................\
@@ -345,7 +345,7 @@ static const char u_map[257] = "\
 ................................\
 ................................u";
 
-static const char utf_map[257] = "\
+static const char utf_map[258] = "\
 ................................\
 ................................\
 ................................\
@@ -355,7 +355,7 @@ SSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSS\
 ................................\
 ................................8";
 
-static const char space_map[257] = "\
+static const char space_map[258] = "\
 .........ab..a..................\
 a...............................\
 ................................\
@@ -365,7 +365,7 @@ a...............................\
 ................................\
 ................................S";
 
-static const char trail_map[257] = "\
+static const char trail_map[258] = "\
 .........ab..a..................\
 a...............................\
 ................................\
@@ -375,7 +375,7 @@ a...............................\
 ................................\
 ................................R";
 
-static const byte hex_map[256] = "\
+static const byte hex_map[257] = "\
 ................................\
 ................\x00\x01\x02\x03\x04\x05\x06\x07\x08\x09......\
 .\x0a\x0b\x0c\x0d\x0e\x0f.........................\
@@ -594,7 +594,7 @@ static void big_change(ojParser p) {
     }
 }
 
-static void parse(ojParser p, const byte *json) {
+static void parse(ojParser p, const byte *json, bool more) {
     const byte *start;
     const byte *b = json;
     int         i;
@@ -635,6 +635,9 @@ static void parse(ojParser p, const byte *json) {
             if ('"' == *b) {
                 p->map = colon_map;
                 break;
+            } else if ('\0' == *b && !more) {
+                parse_error(p, "quoted string not terminated");
+                break;
             }
             b--;
             p->map      = string_map;
@@ -659,6 +662,9 @@ static void parse(ojParser p, const byte *json) {
                 p->funcs[p->stack[p->depth]].add_str(p);
                 p->map = (0 == p->depth) ? value_map : after_map;
                 break;
+            } else if ('\0' == *b && !more) {
+                parse_error(p, "quoted string not terminated");
+                break;
             }
             b--;
             p->map      = string_map;
@@ -668,6 +674,10 @@ static void parse(ojParser p, const byte *json) {
             p->cur = b - json;
             p->funcs[p->stack[p->depth]].open_object(p);
             p->depth++;
+            if ((int)sizeof(p->stack) <= p->depth) {
+                parse_error(p, "too deeply nested");
+                break;
+            }
             p->stack[p->depth] = OBJECT_FUN;
             p->map             = key1_map;
             break;
@@ -690,6 +700,10 @@ static void parse(ojParser p, const byte *json) {
             p->cur = b - json;
             p->funcs[p->stack[p->depth]].open_array(p);
             p->depth++;
+            if ((int)sizeof(p->stack) <= p->depth) {
+                parse_error(p, "too deeply nested");
+                break;
+            }
             p->stack[p->depth] = ARRAY_FUN;
             p->map             = value_map;
             break;
@@ -1176,6 +1190,8 @@ extern void oj_set_parser_validator(ojParser p);
 extern void oj_set_parser_saj(ojParser p);
 extern void oj_set_parser_usual(ojParser p);
 extern void oj_set_parser_debug(ojParser p);
+extern void oj_set_parser_safe(ojParser p, VALUE options);
+extern void oj_safe_init(VALUE parser_class);
 
 static int opt_cb(VALUE rkey, VALUE value, VALUE ptr) {
     ojParser    p   = (ojParser)ptr;
@@ -1354,6 +1370,34 @@ static VALUE parser_missing(int argc, VALUE *argv, VALUE self) {
     return p->option(p, key, rv);
 }
 
+static void validate_primitives_are_complete(ojParser p) {
+    if (0 >= p->ri) {
+        return;
+    }
+
+    switch (p->map[256]) {
+    case 'N': parse_error(p, "expected null"); break;
+    case 'F': parse_error(p, "expected false"); break;
+    case 'T': parse_error(p, "expected true"); break;
+    }
+}
+
+static void validate_non_primitives_are_complete(ojParser p) {
+    if (0 >= p->depth) {
+        return;
+    }
+    if (OBJECT_FUN == p->stack[p->depth]) {
+        parse_error(p, "Object is not closed");
+    } else {
+        parse_error(p, "Array is not closed");
+    }
+}
+
+static void validate_document_end(ojParser p) {
+    validate_primitives_are_complete(p);
+    validate_non_primitives_are_complete(p);
+}
+
 /* Document-method: parse(json)
  * call-seq: parse(json)
  *
@@ -1363,21 +1407,21 @@ static VALUE parser_missing(int argc, VALUE *argv, VALUE self) {
  */
 static VALUE parser_parse(VALUE self, VALUE json) {
     ojParser    p;
-    const byte *ptr = (const byte *)StringValuePtr(json);
+    int         frozen = OBJ_FROZEN(json);
+    const byte *ptr;
+
+    if (!frozen) {
+        rb_str_freeze(json);
+    }
+    ptr = (const byte *)StringValuePtr(json);
 
     TypedData_Get_Struct(self, struct _ojParser, &oj_parser_type, p);
 
     parser_reset(p);
     p->start(p);
-    parse(p, ptr);
+    parse(p, ptr, false);
+    validate_document_end(p);
 
-    if (0 < p->depth) {
-        if (OBJECT_FUN == p->stack[p->depth]) {
-            parse_error(p, "Object is not closed");
-        } else {
-            parse_error(p, "Array is not closed");
-        }
-    }
     return p->result(p);
 }
 
@@ -1396,7 +1440,7 @@ static VALUE load(VALUE self) {
     while (true) {
         rb_funcall(p->reader, oj_readpartial_id, 2, INT2NUM(16385), rbuf);
         if (0 < RSTRING_LEN(rbuf)) {
-            parse(p, (byte *)StringValuePtr(rbuf));
+            parse(p, (byte *)StringValuePtr(rbuf), true);
         }
         if (Qtrue == rb_funcall(p->reader, oj_eofq_id, 0)) {
             if (0 < p->depth) {
@@ -1466,7 +1510,7 @@ static VALUE parser_file(VALUE self, VALUE filename) {
     while (true) {
         if (0 < (rsize = read(fd, buf, size))) {
             buf[rsize] = '\0';
-            parse(p, buf);
+            parse(p, buf, true);
         }
         if (rsize <= 0) {
             if (0 != rsize) {
@@ -1577,6 +1621,28 @@ static VALUE parser_validate(VALUE self) {
     return validate_parser;
 }
 
+static VALUE parser_safe(int argc, VALUE *argv, VALUE self) {
+    VALUE options;
+
+    if (1 == argc) {
+        options = argv[0];
+
+        Check_Type(options, T_HASH);
+    } else {
+        options = rb_hash_new();
+    }
+
+    ojParser p = OJ_R_ALLOC(struct _ojParser);
+
+    memset(p, 0, sizeof(struct _ojParser));
+    buf_init(&p->key);
+    buf_init(&p->buf);
+    p->map = value_map;
+    oj_set_parser_safe(p, options);
+
+    return TypedData_Wrap_Struct(parser_class, &oj_parser_type, p);
+}
+
 /* Document-class: Oj::Parser
  *
  * A reusable parser that makes use of named delegates to determine the
@@ -1604,4 +1670,7 @@ void oj_parser_init(void) {
     rb_define_module_function(parser_class, "usual", parser_usual, 0);
     rb_define_module_function(parser_class, "saj", parser_saj, 0);
     rb_define_module_function(parser_class, "validate", parser_validate, 0);
+    rb_define_module_function(parser_class, "safe", parser_safe, -1);
+
+    oj_safe_init(parser_class);
 }

data/ext/oj/rails.c

@@ -920,7 +920,7 @@ static VALUE encode(VALUE obj, ROptTable ropts, Options opts, int argc, VALUE *a
     if (Yes == copts.circular) {
         oj_cache8_new(&out.circ_cache);
     }
-    // dump_rails_val(*argv, 0, &out, true);
+
     rb_protect(protect_dump, (VALUE)&oo, &line);
 
     if (0 == line) {
@@ -1219,8 +1219,7 @@ static void dump_array(VALUE a, int depth, Out out, bool as_ok) {
             return;
         }
     }
-    // if (!oj_rails_array_opt && as_ok && 0 < out->argc && rb_respond_to(a, oj_as_json_id)) {
-    if (as_ok && 0 < out->argc && rb_respond_to(a, oj_as_json_id)) {
+    if (!oj_rails_array_opt && as_ok && rb_respond_to(a, oj_as_json_id)) {
         dump_as_json(a, depth, out, false);
         return;
     }
@@ -1287,18 +1286,19 @@ static int hash_cb(VALUE key, VALUE value, VALUE ov) {
     if (out->omit_nil && Qnil == value) {
         return ST_CONTINUE;
     }
-    if (rtype != T_STRING && rtype != T_SYMBOL) {
-        key   = oj_safe_string_convert(key);
-        rtype = rb_type(key);
+    if (NULL != out->opts->dump_opts.only || NULL != out->opts->dump_opts.except) {
+        if (oj_key_skip(key, out->opts->dump_opts.only, out->opts->dump_opts.except)) {
+            return ST_CONTINUE;
+        }
     }
     if (!out->opts->dump_opts.use) {
         size = depth * out->indent + 1;
         assure_size(out, size);
         fill_indent(out, depth);
-        if (rtype == T_STRING) {
-            oj_dump_str(key, 0, out, false);
-        } else {
-            oj_dump_sym(key, 0, out, false);
+        switch (rtype) {
+        case T_STRING: oj_dump_str(key, 0, out, false); break;
+        case T_SYMBOL: oj_dump_sym(key, 0, out, false); break;
+        default: oj_dump_str(oj_safe_string_convert(key), 0, out, false); break;
         }
         *out->cur++ = ':';
     } else {
@@ -1313,10 +1313,10 @@ static int hash_cb(VALUE key, VALUE value, VALUE ov) {
                 APPEND_CHARS(out->cur, out->opts->dump_opts.indent_str, out->opts->dump_opts.indent_size);
             }
         }
-        if (rtype == T_STRING) {
-            oj_dump_str(key, 0, out, false);
-        } else {
-            oj_dump_sym(key, 0, out, false);
+        switch (rtype) {
+        case T_STRING: oj_dump_str(key, 0, out, false); break;
+        case T_SYMBOL: oj_dump_sym(key, 0, out, false); break;
+        default: oj_dump_str(oj_safe_string_convert(key), 0, out, false); break;
         }
         size = out->opts->dump_opts.before_size + out->opts->dump_opts.after_size + 2;
         assure_size(out, size);
@@ -1345,7 +1345,7 @@ static void dump_hash(VALUE obj, int depth, Out out, bool as_ok) {
             return;
         }
     }
-    if ((!oj_rails_hash_opt || 0 < out->argc) && as_ok && rb_respond_to(obj, oj_as_json_id)) {
+    if (!oj_rails_hash_opt && as_ok && rb_respond_to(obj, oj_as_json_id)) {
         dump_as_json(obj, depth, out, false);
         return;
     }

data/ext/oj/rxclass.c

@@ -96,7 +96,7 @@ int oj_rxclass_append(RxClass rc, const char *expr, VALUE clas) {
 }
 
 VALUE
-oj_rxclass_match(RxClass rc, const char *str, int len) {
+oj_rxclass_match(RxClass rc, const char *str, size_t len) {
     RxC  rxc;
     char buf[4096];
 

data/ext/oj/rxclass.h

@@ -19,7 +19,7 @@ typedef struct _rxClass {
 extern void  oj_rxclass_init(RxClass rc);
 extern void  oj_rxclass_cleanup(RxClass rc);
 extern int   oj_rxclass_append(RxClass rc, const char *expr, VALUE clas);
-extern VALUE oj_rxclass_match(RxClass rc, const char *str, int len);
+extern VALUE oj_rxclass_match(RxClass rc, const char *str, size_t len);
 extern void  oj_rxclass_copy(RxClass src, RxClass dest);
 extern void  oj_rxclass_rappend(RxClass rc, VALUE rx, VALUE clas);