oj 3.13.16 → 3.13.19

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: d07b4081b5dda88146594adba591542d11106f1c1eadcf7c56e38358d0ee0af3
-  data.tar.gz: 6bdc82c3812f45a8ae9d72ec0b2f43ee223d95ac250399fc8dae8daf34cd3b49
+  metadata.gz: bfbe0dd2157d1aba97ad11d82eee1870fbb64f4f197ae75b17a2e781c64144f7
+  data.tar.gz: 907ff02298adc4555834ca1c1c6078a3586abfde85f790b4dcba284c5be60248
 SHA512:
-  metadata.gz: 4358aa15b7956a5b577163e43fbdc0512212a218124878ceedc2932483f27acabed2d2b92eb888c8e54601f4f4aaf3dce3b959a36effbea989e7e7ea8956d1c1
-  data.tar.gz: 1e1710bce62aadfbcc1196357d2e692892b75cc18c982fb6f53c2e35782d9bd2256cb3b6ab98951691430f23e8c2d8db26987a95dcba29f87b472729613b8d2d
+  metadata.gz: c56828991af620bb1436db19b5660ec588fce4858b455bd2b6c7f11f2bccaf3a7afca8fe98b62ad868a1a18ab584d38220e1b5c4c1243d3a8c1201c3b0c51a54
+  data.tar.gz: f687f4bd9a0a64ecd75cee755722cc2f4f80fc356fe884fd34bf5cf68066d2e648050de7a49c93c121740e790afbcc168b1a8efc5f8280f64ca9bf3d23bc736b

data/CHANGELOG.md

@@ -1,5 +1,17 @@
 # CHANGELOG
 
+## 3.13.19 - 2022-07-29
+
+- TruffleRuby issues resolved.
+
+## 3.13.18 - 2022-07-25
+
+- Fixed SSE detection at run time.
+
+## 3.13.17 - 2022-07-15
+
+- Fixed Oj::Parser to detect unterminated arrays and objects.
+
 ## 3.13.16 - 2022-07-06
 
 - Added line and column as optional arguments to the Oj::Parser.saj parser.

data/ext/oj/buf.h

@@ -39,6 +39,10 @@ inline static const char *buf_str(Buf buf) {
 }
 
 inline static void buf_append_string(Buf buf, const char *s, size_t slen) {
+    if (0 == slen) {
+        return;
+    }
+
     if (buf->end <= buf->tail + slen) {
         size_t len     = buf->end - buf->head;
         size_t toff    = buf->tail - buf->head;

data/ext/oj/custom.c

@@ -484,7 +484,7 @@ static VALUE dump_common(VALUE obj, int depth, Out out) {
         const char *   s;
         int            len;
 
-        if (Yes == out->opts->trace) {
+        if (RB_UNLIKELY(Yes == out->opts->trace)) {
             oj_trace("to_json", obj, __FILE__, __LINE__, depth + 1, TraceRubyIn);
         }
         if (0 == rb_obj_method_arity(obj, oj_to_json_id)) {
@@ -492,7 +492,7 @@ static VALUE dump_common(VALUE obj, int depth, Out out) {
         } else {
             rs = rb_funcall2(obj, oj_to_json_id, out->argc, out->argv);
         }
-        if (Yes == out->opts->trace) {
+        if (RB_UNLIKELY(Yes == out->opts->trace)) {
             oj_trace("to_json", obj, __FILE__, __LINE__, depth + 1, TraceRubyOut);
         }
         s   = RSTRING_PTR(rs);
@@ -504,7 +504,7 @@ static VALUE dump_common(VALUE obj, int depth, Out out) {
     } else if (Yes == out->opts->as_json && rb_respond_to(obj, oj_as_json_id)) {
         volatile VALUE aj;
 
-        if (Yes == out->opts->trace) {
+        if (RB_UNLIKELY(Yes == out->opts->trace)) {
             oj_trace("as_json", obj, __FILE__, __LINE__, depth + 1, TraceRubyIn);
         }
         // Some classes elect to not take an options argument so check the arity
@@ -514,7 +514,7 @@ static VALUE dump_common(VALUE obj, int depth, Out out) {
         } else {
             aj = rb_funcall2(obj, oj_as_json_id, out->argc, out->argv);
         }
-        if (Yes == out->opts->trace) {
+        if (RB_UNLIKELY(Yes == out->opts->trace)) {
             oj_trace("as_json", obj, __FILE__, __LINE__, depth + 1, TraceRubyOut);
         }
         // Catch the obvious brain damaged recursive dumping.
@@ -885,7 +885,7 @@ static DumpFunc custom_funcs[] = {
 void oj_dump_custom_val(VALUE obj, int depth, Out out, bool as_ok) {
     int type = rb_type(obj);
 
-    if (Yes == out->opts->trace) {
+    if (RB_UNLIKELY(Yes == out->opts->trace)) {
         oj_trace("dump", obj, __FILE__, __LINE__, depth, TraceIn);
     }
     if (MAX_DEPTH < depth) {
@@ -896,14 +896,14 @@ void oj_dump_custom_val(VALUE obj, int depth, Out out, bool as_ok) {
 
         if (NULL != f) {
             f(obj, depth, out, true);
-            if (Yes == out->opts->trace) {
+            if (RB_UNLIKELY(Yes == out->opts->trace)) {
                 oj_trace("dump", obj, __FILE__, __LINE__, depth, TraceOut);
             }
             return;
         }
     }
     oj_dump_nil(Qnil, depth, out, false);
-    if (Yes == out->opts->trace) {
+    if (RB_UNLIKELY(Yes == out->opts->trace)) {
         oj_trace("dump", Qnil, __FILE__, __LINE__, depth, TraceOut);
     }
 }

data/ext/oj/dump.c

@@ -736,11 +736,11 @@ void oj_dump_raw_json(VALUE obj, int depth, Out out) {
     } else {
         volatile VALUE jv;
 
-        if (Yes == out->opts->trace) {
+        if (RB_UNLIKELY(Yes == out->opts->trace)) {
             oj_trace("raw_json", obj, __FILE__, __LINE__, depth + 1, TraceRubyIn);
         }
         jv = rb_funcall(obj, oj_raw_json_id, 2, RB_INT2NUM(depth), RB_INT2NUM(out->indent));
-        if (Yes == out->opts->trace) {
+        if (RB_UNLIKELY(Yes == out->opts->trace)) {
             oj_trace("raw_json", obj, __FILE__, __LINE__, depth + 1, TraceRubyOut);
         }
         oj_dump_raw(RSTRING_PTR(jv), (size_t)RSTRING_LEN(jv), out);

data/ext/oj/dump_compat.c

@@ -109,7 +109,7 @@ dump_to_json(VALUE obj, Out out) {
     const char		*s;
     int			len;
 
-    if (Yes == out->opts->trace) {
+    if (RB_UNLIKELY(Yes == out->opts->trace)) {
 	oj_trace("to_json", obj, __FILE__, __LINE__, 0, TraceRubyIn);
     }
     if (0 == rb_obj_method_arity(obj, oj_to_json_id)) {
@@ -117,7 +117,7 @@ dump_to_json(VALUE obj, Out out) {
     } else {
 	rs = rb_funcall2(obj, oj_to_json_id, out->argc, out->argv);
     }
-    if (Yes == out->opts->trace) {
+    if (RB_UNLIKELY(Yes == out->opts->trace)) {
 	oj_trace("to_json", obj, __FILE__, __LINE__, 0, TraceRubyOut);
     }
 
@@ -893,7 +893,7 @@ void
 oj_dump_compat_val(VALUE obj, int depth, Out out, bool as_ok) {
     int	type = rb_type(obj);
 
-    if (Yes == out->opts->trace) {
+    if (RB_UNLIKELY(Yes == out->opts->trace)) {
 	oj_trace("dump", obj, __FILE__, __LINE__, depth, TraceIn);
     }
     if (out->opts->dump_opts.max_depth <= depth) {
@@ -918,14 +918,14 @@ oj_dump_compat_val(VALUE obj, int depth, Out out, bool as_ok) {
 
 	if (NULL != f) {
 	    f(obj, depth, out, as_ok);
-	    if (Yes == out->opts->trace) {
+	    if (RB_UNLIKELY(Yes == out->opts->trace)) {
 		oj_trace("dump", obj, __FILE__, __LINE__, depth, TraceOut);
 	    }
 	    return;
 	}
     }
     oj_dump_nil(Qnil, depth, out, false);
-    if (Yes == out->opts->trace) {
+    if (RB_UNLIKELY(Yes == out->opts->trace)) {
 	oj_trace("dump", Qnil, __FILE__, __LINE__, depth, TraceOut);
     }
 }

data/ext/oj/dump_object.c

@@ -682,7 +682,7 @@ static DumpFunc obj_funcs[] = {
 void oj_dump_obj_val(VALUE obj, int depth, Out out) {
     int type = rb_type(obj);
 
-    if (Yes == out->opts->trace) {
+    if (RB_UNLIKELY(Yes == out->opts->trace)) {
         oj_trace("dump", obj, __FILE__, __LINE__, depth, TraceIn);
     }
     if (MAX_DEPTH < depth) {
@@ -693,14 +693,14 @@ void oj_dump_obj_val(VALUE obj, int depth, Out out) {
 
         if (NULL != f) {
             f(obj, depth, out, false);
-            if (Yes == out->opts->trace) {
+            if (RB_UNLIKELY(Yes == out->opts->trace)) {
                 oj_trace("dump", obj, __FILE__, __LINE__, depth, TraceOut);
             }
             return;
         }
     }
     oj_dump_nil(Qnil, depth, out, false);
-    if (Yes == out->opts->trace) {
+    if (RB_UNLIKELY(Yes == out->opts->trace)) {
         oj_trace("dump", Qnil, __FILE__, __LINE__, depth, TraceOut);
     }
 }

data/ext/oj/dump_strict.c

@@ -338,7 +338,7 @@ static DumpFunc strict_funcs[] = {
 void oj_dump_strict_val(VALUE obj, int depth, Out out) {
     int type = rb_type(obj);
 
-    if (Yes == out->opts->trace) {
+    if (RB_UNLIKELY(Yes == out->opts->trace)) {
         oj_trace("dump", obj, __FILE__, __LINE__, depth, TraceIn);
     }
     if (MAX_DEPTH < depth) {
@@ -349,7 +349,7 @@ void oj_dump_strict_val(VALUE obj, int depth, Out out) {
 
         if (NULL != f) {
             f(obj, depth, out, false);
-            if (Yes == out->opts->trace) {
+            if (RB_UNLIKELY(Yes == out->opts->trace)) {
                 oj_trace("dump", obj, __FILE__, __LINE__, depth, TraceOut);
             }
             return;
@@ -386,7 +386,7 @@ static DumpFunc null_funcs[] = {
 void oj_dump_null_val(VALUE obj, int depth, Out out) {
     int type = rb_type(obj);
 
-    if (Yes == out->opts->trace) {
+    if (RB_UNLIKELY(Yes == out->opts->trace)) {
         oj_trace("dump", obj, __FILE__, __LINE__, depth, TraceOut);
     }
     if (MAX_DEPTH < depth) {
@@ -397,14 +397,14 @@ void oj_dump_null_val(VALUE obj, int depth, Out out) {
 
         if (NULL != f) {
             f(obj, depth, out, false);
-            if (Yes == out->opts->trace) {
+            if (RB_UNLIKELY(Yes == out->opts->trace)) {
                 oj_trace("dump", obj, __FILE__, __LINE__, depth, TraceOut);
             }
             return;
         }
     }
     oj_dump_nil(Qnil, depth, out, false);
-    if (Yes == out->opts->trace) {
+    if (RB_UNLIKELY(Yes == out->opts->trace)) {
         oj_trace("dump", Qnil, __FILE__, __LINE__, depth, TraceOut);
     }
 }

data/ext/oj/extconf.rb

@@ -34,9 +34,21 @@ have_func('rb_hash_bulk_insert', 'ruby.h') unless '2' == version[0] && '6' == ve
 
 dflags['OJ_DEBUG'] = true unless ENV['OJ_DEBUG'].nil?
 
-if try_cflags('-msse4.2')
+src =<<~SRC
+#include <string.h>
+#include <nmmintrin.h>
+int main(int argc, char **argv) {
+    const char *str = "hello               ";
+    const char chars[16] = "\x00\\\"";
+    const __m128i terminate = _mm_loadu_si128((const __m128i *)&chars[0]);
+    const __m128i string = _mm_loadu_si128((const __m128i *)str);
+    int r = _mm_cmpestri(terminate, 3, string, 16, _SIDD_UBYTE_OPS | _SIDD_CMP_EQUAL_ANY | _SIDD_LEAST_SIGNIFICANT);
+    return r == 16 ? 0 : 1;
+}
+SRC
+
+if try_run(src, '-msse4.2')
   $CPPFLAGS += ' -msse4.2'
-  dflags['OJ_USE_SSE4_2'] = 1
 end
 
 dflags.each do |k,v|

data/ext/oj/oj.c

@@ -2043,4 +2043,5 @@ void Init_oj(void) {
     oj_init_doc();
 
     oj_parser_init();
+    oj_scanner_init();
 }

data/ext/oj/parse.c

@@ -183,6 +183,78 @@ static void unicode_to_chars(ParseInfo pi, Buf buf, uint32_t code) {
     }
 }
 
+static inline const char *scan_string_noSIMD(const char *str, const char *end) {
+    for (; '"' != *str; str++) {
+        if (end <= str || '\0' == *str || '\\' == *str) {
+            break;
+        }
+    }
+    return str;
+}
+
+// Taken from Tensorflow:
+// https://github.com/tensorflow/tensorflow/blob/5dcfc51118817f27fad5246812d83e5dccdc5f72/tensorflow/core/lib/hash/crc32c_accelerate.cc#L21-L38
+#ifdef __SSE4_2__
+#if defined(__x86_64__) && defined(__GNUC__) && \
+    (__GNUC__ > 4 || (__GNUC__ == 4 && __GNUC_MINOR__ >= 8))
+#define USE_SSE_DETECT 1
+#elif defined(__x86_64__) && defined(__clang__)
+#if __has_builtin(__builtin_cpu_supports)
+#define USE_SSE_DETECT 1
+#endif
+#endif
+#endif /* __SSE4_2__ */
+
+// This version of Apple clang has a bug:
+// https://llvm.org/bugs/show_bug.cgi?id=25510
+#if defined(__APPLE__) && (__clang_major__ <= 8)
+#undef USE_SSE_DETECT
+#endif
+
+#if defined(TRUFFLERUBY)
+#undef USE_SSE_DETECT
+#endif
+
+#ifdef USE_SSE_DETECT
+#include <nmmintrin.h>
+
+static inline const char *scan_string_SIMD(const char *str, const char *end) {
+    static const char chars[16] = "\x00\\\"";
+    const __m128i terminate = _mm_loadu_si128((const __m128i *)&chars[0]);
+    const char *_end = (const char *)(end - 16);
+
+    for (; str <= _end; str += 16) {
+        const __m128i string = _mm_loadu_si128((const __m128i *)str);
+        const int r = _mm_cmpestri(terminate, 3, string, 16, _SIDD_UBYTE_OPS | _SIDD_CMP_EQUAL_ANY | _SIDD_LEAST_SIGNIFICANT);
+        if (r != 16) {
+            str = (char*)(str + r);
+            return str;
+        }
+    }
+
+    return scan_string_noSIMD(str, end);
+}
+#endif
+
+static bool cpu_supports_sse42(void) {
+#if USE_SSE_DETECT
+    __builtin_cpu_init();
+    return (__builtin_cpu_supports("sse4.2"));
+#else
+    return false;
+#endif
+}
+
+static const char *(*scan_func) (const char *str, const char *end) = scan_string_noSIMD;
+
+void oj_scanner_init(void) {
+    if (cpu_supports_sse42()) {
+#if USE_SSE_DETECT
+        scan_func = scan_string_SIMD;
+#endif
+    }
+}
+
 // entered at /
 static void read_escaped_str(ParseInfo pi, const char *start) {
     struct _buf buf;
@@ -192,11 +264,11 @@ static void read_escaped_str(ParseInfo pi, const char *start) {
     Val         parent = stack_peek(&pi->stack);
 
     buf_init(&buf);
-    if (0 < cnt) {
-        buf_append_string(&buf, start, cnt);
-    }
-    for (s = pi->cur; '"' != *s; s++) {
-        if (s >= pi->end) {
+    buf_append_string(&buf, start, cnt);
+
+    for (s = pi->cur; '"' != *s;) {
+        const char *scanned = scan_func(s, pi->end);
+        if (scanned >= pi->end) {
             oj_set_error_at(pi,
                             oj_parse_error_class,
                             __FILE__,
@@ -204,7 +276,12 @@ static void read_escaped_str(ParseInfo pi, const char *start) {
                             "quoted string not terminated");
             buf_cleanup(&buf);
             return;
-        } else if ('\\' == *s) {
+        }
+
+        buf_append_string(&buf, s, (size_t)(scanned - s));
+        s = scanned;
+
+        if ('\\' == *s) {
             s++;
             switch (*s) {
             case 'n': buf_append(&buf, '\n'); break;
@@ -273,8 +350,7 @@ static void read_escaped_str(ParseInfo pi, const char *start) {
                 buf_cleanup(&buf);
                 return;
             }
-        } else {
-            buf_append(&buf, *s);
+            s++;
         }
     }
     if (0 == parent) {
@@ -327,44 +403,11 @@ static void read_escaped_str(ParseInfo pi, const char *start) {
     buf_cleanup(&buf);
 }
 
-static inline void scan_string_noSIMD(ParseInfo pi) {
-    for (; '"' != *pi->cur; pi->cur++) {
-        if (pi->end <= pi->cur || '\0' == *pi->cur || '\\' == *pi->cur) {
-            return;
-        }
-    }
-}
-
-#if defined(OJ_USE_SSE4_2)
-#include <nmmintrin.h>
-
-static inline void scan_string_SIMD(ParseInfo pi) {
-    static const char chars[16] = "\x00\\\"";
-    const __m128i terminate = _mm_loadu_si128((const __m128i *)&chars[0]);
-    const char *end = (const char *)(pi->end - 16);
-
-    for (; pi->cur <= end; pi->cur += 16) {
-        const __m128i string = _mm_loadu_si128((const __m128i *)pi->cur);
-        const int r = _mm_cmpestri(terminate, 3, string, 16, _SIDD_UBYTE_OPS | _SIDD_CMP_EQUAL_ANY | _SIDD_LEAST_SIGNIFICANT);
-        if (r != 16) {
-            pi->cur = (const char*)(pi->cur + r);
-            return;
-        }
-    }
-
-    scan_string_noSIMD(pi);
-}
-#endif
-
 static void read_str(ParseInfo pi) {
     const char *str    = pi->cur;
     Val         parent = stack_peek(&pi->stack);
 
-#if defined(OJ_USE_SSE4_2)
-    scan_string_SIMD(pi);
-#else
-    scan_string_noSIMD(pi);
-#endif
+    pi->cur = scan_func(pi->cur, pi->end);
     if (RB_UNLIKELY(pi->end <= pi->cur)) {
         oj_set_error_at(pi,
                         oj_parse_error_class,
@@ -494,33 +537,31 @@ static void read_num(ParseInfo pi) {
         int  dec_cnt = 0;
         bool zero1   = false;
 
+        // Skip leading zeros.
+        for (; '0' == *pi->cur; pi->cur++) {
+            zero1 = true;
+        }
+
         for (; '0' <= *pi->cur && *pi->cur <= '9'; pi->cur++) {
-            if (0 == ni.i && '0' == *pi->cur) {
-                zero1 = true;
-            }
-            if (0 < ni.i) {
-                dec_cnt++;
-            }
-            if (!ni.big) {
-                int d = (*pi->cur - '0');
+            int d = (*pi->cur - '0');
 
-                if (0 < d) {
-                    if (zero1 && CompatMode == pi->options.mode) {
-                        oj_set_error_at(pi,
-                                        oj_parse_error_class,
-                                        __FILE__,
-                                        __LINE__,
-                                        "not a number");
-                        return;
-                    }
-                    zero1 = false;
-                }
-                ni.i = ni.i * 10 + d;
-                if (INT64_MAX <= ni.i || DEC_MAX < dec_cnt) {
-                    ni.big = 1;
-                }
+            if (RB_LIKELY(0 != ni.i)) {
+                dec_cnt++;
             }
+            ni.i = ni.i * 10 + d;
         }
+        if (RB_UNLIKELY(0 != ni.i && zero1 && CompatMode == pi->options.mode)) {
+            oj_set_error_at(pi,
+                            oj_parse_error_class,
+                            __FILE__,
+                            __LINE__,
+                            "not a number");
+            return;
+        }
+        if (INT64_MAX <= ni.i || DEC_MAX < dec_cnt) {
+            ni.big = true;
+        }
+
         if ('.' == *pi->cur) {
             pi->cur++;
             // A trailing . is not a valid decimal but if encountered allow it
@@ -540,25 +581,20 @@ static void read_num(ParseInfo pi) {
             for (; '0' <= *pi->cur && *pi->cur <= '9'; pi->cur++) {
                 int d = (*pi->cur - '0');
 
-                if (0 < ni.num || 0 < ni.i) {
+                if (RB_LIKELY(0 != ni.num || 0 != ni.i)) {
                     dec_cnt++;
                 }
-                if (INT64_MAX <= ni.div) {
-                    if (!ni.no_big) {
-                        ni.big = true;
-                    }
-                } else {
-                    ni.num = ni.num * 10 + d;
-                    ni.div *= 10;
-                    ni.di++;
-                    if (INT64_MAX <= ni.div || DEC_MAX < dec_cnt) {
-                        if (!ni.no_big) {
-                            ni.big = true;
-                        }
-                    }
-                }
+                ni.num = ni.num * 10 + d;
+                ni.div *= 10;
+                ni.di++;
             }
         }
+        if (INT64_MAX <= ni.div || DEC_MAX < dec_cnt) {
+            if (!ni.no_big) {
+                ni.big = true;
+            }
+        }
+
         if ('e' == *pi->cur || 'E' == *pi->cur) {
             int eneg = 0;
 

data/ext/oj/parse.h

@@ -97,6 +97,8 @@ static inline void parse_info_init(ParseInfo pi) {
     memset(pi, 0, sizeof(struct _parseInfo));
 }
 
+extern void oj_scanner_init(void);
+
 static inline bool empty_ok(Options options) {
     switch (options->mode) {
     case ObjectMode:

data/ext/oj/parser.c

@@ -600,7 +600,7 @@ static void parse(ojParser p, const byte *json) {
     int         i;
 
     p->line = 1;
-    p->col = -1;
+    p->col  = -1;
 #if DEBUG
     printf("*** parse - mode: %c %s\n", p->map[256], (const char *)json);
 #endif
@@ -655,7 +655,7 @@ static void parse(ojParser p, const byte *json) {
             }
             buf_append_string(&p->buf, (const char *)start, b - start);
             if ('"' == *b) {
-		p->cur = b - json;
+                p->cur = b - json;
                 p->funcs[p->stack[p->depth]].add_str(p);
                 p->map = (0 == p->depth) ? value_map : after_map;
                 break;
@@ -665,14 +665,14 @@ static void parse(ojParser p, const byte *json) {
             p->next_map = (0 == p->depth) ? value_map : after_map;
             break;
         case OPEN_OBJECT:
-	    p->cur = b - json;
+            p->cur = b - json;
             p->funcs[p->stack[p->depth]].open_object(p);
             p->depth++;
             p->stack[p->depth] = OBJECT_FUN;
             p->map             = key1_map;
             break;
         case NUM_CLOSE_OBJECT:
-	    p->cur = b - json;
+            p->cur = b - json;
             calc_num(p);
             // flow through
         case CLOSE_OBJECT:
@@ -683,18 +683,18 @@ static void parse(ojParser p, const byte *json) {
                 return;
             }
             p->depth--;
-	    p->cur = b - json;
+            p->cur = b - json;
             p->funcs[p->stack[p->depth]].close_object(p);
             break;
         case OPEN_ARRAY:
-	    p->cur = b - json;
+            p->cur = b - json;
             p->funcs[p->stack[p->depth]].open_array(p);
             p->depth++;
             p->stack[p->depth] = ARRAY_FUN;
             p->map             = value_map;
             break;
         case NUM_CLOSE_ARRAY:
-	    p->cur = b - json;
+            p->cur = b - json;
             calc_num(p);
             // flow through
         case CLOSE_ARRAY:
@@ -705,11 +705,11 @@ static void parse(ojParser p, const byte *json) {
                 return;
             }
             p->depth--;
-	    p->cur = b - json;
+            p->cur = b - json;
             p->funcs[p->stack[p->depth]].close_array(p);
             break;
         case NUM_COMMA:
-	    p->cur = b - json;
+            p->cur = b - json;
             calc_num(p);
             if (0 < p->depth && OBJECT_FUN == p->stack[p->depth]) {
                 p->map = key_map;
@@ -872,13 +872,13 @@ static void parse(ojParser p, const byte *json) {
             p->map = big_exp_map;
             break;
         case NUM_SPC:
-	    p->cur = b - json;
-	    calc_num(p);
-	    break;
+            p->cur = b - json;
+            calc_num(p);
+            break;
         case NUM_NEWLINE:
-	    p->cur = b - json;
-	    calc_num(p);
-	    b++;
+            p->cur = b - json;
+            calc_num(p);
+            b++;
 #ifdef SPACE_JUMP
             // for (uint32_t *sj = (uint32_t*)b; 0x20202020 == *sj; sj++) { b += 4; }
             for (uint16_t *sj = (uint16_t *)b; 0x2020 == *sj; sj++) {
@@ -899,7 +899,7 @@ static void parse(ojParser p, const byte *json) {
                 buf_append_string(&p->buf, (const char *)start, b - start);
             }
             if ('"' == *b) {
-		p->cur = b - json;
+                p->cur = b - json;
                 p->funcs[p->stack[p->depth]].add_str(p);
                 p->map = p->next_map;
                 break;
@@ -908,7 +908,7 @@ static void parse(ojParser p, const byte *json) {
             break;
         case STR_SLASH: p->map = esc_map; break;
         case STR_QUOTE:
-	    p->cur = b - json;
+            p->cur = b - json;
             p->funcs[p->stack[p->depth]].add_str(p);
             p->map = p->next_map;
             break;
@@ -986,7 +986,7 @@ static void parse(ojParser p, const byte *json) {
         case VAL_NULL:
             if ('u' == b[1] && 'l' == b[2] && 'l' == b[3]) {
                 b += 3;
-		p->cur = b - json;
+                p->cur = b - json;
                 p->funcs[p->stack[p->depth]].add_null(p);
                 p->map = (0 == p->depth) ? value_map : after_map;
                 break;
@@ -1012,7 +1012,7 @@ static void parse(ojParser p, const byte *json) {
         case VAL_TRUE:
             if ('r' == b[1] && 'u' == b[2] && 'e' == b[3]) {
                 b += 3;
-		p->cur = b - json;
+                p->cur = b - json;
                 p->funcs[p->stack[p->depth]].add_true(p);
                 p->map = (0 == p->depth) ? value_map : after_map;
                 break;
@@ -1038,7 +1038,7 @@ static void parse(ojParser p, const byte *json) {
         case VAL_FALSE:
             if ('a' == b[1] && 'l' == b[2] && 's' == b[3] && 'e' == b[4]) {
                 b += 4;
-		p->cur = b - json;
+                p->cur = b - json;
                 p->funcs[p->stack[p->depth]].add_false(p);
                 p->map = (0 == p->depth) ? value_map : after_map;
                 break;
@@ -1072,7 +1072,7 @@ static void parse(ojParser p, const byte *json) {
                         parse_error(p, "expected null");
                         return;
                     }
-		    p->cur = b - json;
+                    p->cur = b - json;
                     p->funcs[p->stack[p->depth]].add_null(p);
                     p->map = (0 == p->depth) ? value_map : after_map;
                 }
@@ -1084,7 +1084,7 @@ static void parse(ojParser p, const byte *json) {
                         parse_error(p, "expected false");
                         return;
                     }
-		    p->cur = b - json;
+                    p->cur = b - json;
                     p->funcs[p->stack[p->depth]].add_false(p);
                     p->map = (0 == p->depth) ? value_map : after_map;
                 }
@@ -1096,7 +1096,7 @@ static void parse(ojParser p, const byte *json) {
                         parse_error(p, "expected true");
                         return;
                     }
-		    p->cur = b - json;
+                    p->cur = b - json;
                     p->funcs[p->stack[p->depth]].add_true(p);
                     p->map = (0 == p->depth) ? value_map : after_map;
                 }
@@ -1114,6 +1114,9 @@ static void parse(ojParser p, const byte *json) {
             p->map = trail_map;
         }
     }
+    if (0 < p->depth) {
+        parse_error(p, "parse error, not closed");
+    }
     if (0 == p->depth) {
         switch (p->map[256]) {
         case '0':
@@ -1125,9 +1128,9 @@ static void parse(ojParser p, const byte *json) {
         case 'g':
         case 'B':
         case 'Y':
-	    p->cur = b - json;
-	    calc_num(p);
-	    break;
+            p->cur = b - json;
+            calc_num(p);
+            break;
         }
     }
     return;

data/ext/oj/rails.c

@@ -517,7 +517,7 @@ static void dump_as_string(VALUE obj, int depth, Out out, bool as_ok) {
 static void dump_as_json(VALUE obj, int depth, Out out, bool as_ok) {
     volatile VALUE ja;
 
-    if (Yes == out->opts->trace) {
+    if (RB_UNLIKELY(Yes == out->opts->trace)) {
         oj_trace("as_json", obj, __FILE__, __LINE__, depth + 1, TraceRubyIn);
     }
     // Some classes elect to not take an options argument so check the arity
@@ -527,7 +527,7 @@ static void dump_as_json(VALUE obj, int depth, Out out, bool as_ok) {
     } else {
         ja = rb_funcall2(obj, oj_as_json_id, out->argc, out->argv);
     }
-    if (Yes == out->opts->trace) {
+    if (RB_UNLIKELY(Yes == out->opts->trace)) {
         oj_trace("as_json", obj, __FILE__, __LINE__, depth + 1, TraceRubyOut);
     }
 
@@ -1464,7 +1464,7 @@ static DumpFunc rails_funcs[] = {
 static void dump_rails_val(VALUE obj, int depth, Out out, bool as_ok) {
     int type = rb_type(obj);
 
-    if (Yes == out->opts->trace) {
+    if (RB_UNLIKELY(Yes == out->opts->trace)) {
         oj_trace("dump", obj, __FILE__, __LINE__, depth, TraceIn);
     }
     if (MAX_DEPTH < depth) {
@@ -1475,14 +1475,14 @@ static void dump_rails_val(VALUE obj, int depth, Out out, bool as_ok) {
 
         if (NULL != f) {
             f(obj, depth, out, as_ok);
-            if (Yes == out->opts->trace) {
+            if (RB_UNLIKELY(Yes == out->opts->trace)) {
                 oj_trace("dump", obj, __FILE__, __LINE__, depth, TraceOut);
             }
             return;
         }
     }
     oj_dump_nil(Qnil, depth, out, false);
-    if (Yes == out->opts->trace) {
+    if (RB_UNLIKELY(Yes == out->opts->trace)) {
         oj_trace("dump", Qnil, __FILE__, __LINE__, depth, TraceOut);
     }
 }

data/ext/oj/saj2.c

@@ -289,7 +289,7 @@ static void add_float_key_loc(ojParser p) {
 }
 
 static void add_big(ojParser p) {
-    rb_funcall((VALUE)p->ctx,
+    rb_funcall(((Delegate)p->ctx)->handler,
                oj_add_value_id,
                2,
                rb_funcall(rb_cObject, oj_bigdecimal_id, 1, rb_str_new(buf_str(&p->buf), buf_len(&p->buf))),
@@ -297,7 +297,7 @@ static void add_big(ojParser p) {
 }
 
 static void add_big_loc(ojParser p) {
-    rb_funcall((VALUE)p->ctx,
+    rb_funcall(((Delegate)p->ctx)->handler,
                oj_add_value_id,
                4,
                rb_funcall(rb_cObject, oj_bigdecimal_id, 1, rb_str_new(buf_str(&p->buf), buf_len(&p->buf))),
@@ -307,7 +307,7 @@ static void add_big_loc(ojParser p) {
 }
 
 static void add_big_key(ojParser p) {
-    rb_funcall((VALUE)p->ctx,
+    rb_funcall(((Delegate)p->ctx)->handler,
                oj_add_value_id,
                2,
                rb_funcall(rb_cObject, oj_bigdecimal_id, 1, rb_str_new(buf_str(&p->buf), buf_len(&p->buf))),

data/ext/oj/validate.c

@@ -2,50 +2,45 @@
 
 #include "parser.h"
 
-static void
-noop(ojParser p) {
+static void noop(ojParser p) {
 }
 
-static VALUE
-option(ojParser p, const char *key, VALUE value) {
+static VALUE option(ojParser p, const char *key, VALUE value) {
     rb_raise(rb_eArgError, "%s is not an option for the validate delegate", key);
     return Qnil;
 }
 
-static VALUE
-result(ojParser p) {
+static VALUE result(ojParser p) {
     return Qnil;
 }
 
-static void
-dfree(ojParser p) {
+static void dfree(ojParser p) {
 }
 
-static void
-mark(ojParser p) {
+static void mark(ojParser p) {
 }
 
 void oj_set_parser_validator(ojParser p) {
-    p->ctx = NULL;
-    Funcs	end = p->funcs + 3;
+    Funcs end = p->funcs + 3;
     Funcs f;
+    p->ctx = NULL;
 
     for (f = p->funcs; f < end; f++) {
-	f->add_null = noop;
-	f->add_true = noop;
-	f->add_false = noop;
-	f->add_int = noop;
-	f->add_float = noop;
-	f->add_big = noop;
-	f->add_str = noop;
-	f->open_array = noop;
-	f->close_array = noop;
-	f->open_object = noop;
-	f->close_object = noop;
+        f->add_null     = noop;
+        f->add_true     = noop;
+        f->add_false    = noop;
+        f->add_int      = noop;
+        f->add_float    = noop;
+        f->add_big      = noop;
+        f->add_str      = noop;
+        f->open_array   = noop;
+        f->close_array  = noop;
+        f->open_object  = noop;
+        f->close_object = noop;
     }
     p->option = option;
     p->result = result;
-    p->free = dfree;
-    p->mark = mark;
-    p->start = noop;
+    p->free   = dfree;
+    p->mark   = mark;
+    p->start  = noop;
 }

data/ext/oj/wab.c

@@ -266,7 +266,7 @@ static DumpFunc wab_funcs[] = {
 void oj_dump_wab_val(VALUE obj, int depth, Out out) {
     int type = rb_type(obj);
 
-    if (Yes == out->opts->trace) {
+    if (RB_UNLIKELY(Yes == out->opts->trace)) {
         oj_trace("dump", obj, __FILE__, __LINE__, depth, TraceIn);
     }
     if (MAX_DEPTH < depth) {
@@ -277,7 +277,7 @@ void oj_dump_wab_val(VALUE obj, int depth, Out out) {
 
         if (NULL != f) {
             f(obj, depth, out, false);
-            if (Yes == out->opts->trace) {
+            if (RB_UNLIKELY(Yes == out->opts->trace)) {
                 oj_trace("dump", obj, __FILE__, __LINE__, depth, TraceOut);
             }
             return;

data/lib/oj/version.rb

@@ -1,5 +1,5 @@
 
 module Oj
   # Current version of the module.
-  VERSION = '3.13.16'
+  VERSION = '3.13.19'
 end

data/test/bar.rb

@@ -6,11 +6,6 @@ $: << File.join(File.dirname(__FILE__), "../ext")
 
 require 'oj'
 
-puts Oj.dump({
-  "float_test" => 0.25,
-  "nan_test" => Float::NAN,
-  "inf_test" => Float::INFINITY,
-  "minus_inf_test" => -Float::INFINITY,
-  "min_test" => Float::MIN,
-  "max_test" => Float::MAX,
-}, mode: :object) # => {"float_test":0.25,"nan_test":3.3e14159265358979323846
+p = Oj::Parser.validate
+# p = Oj::Parser.new(:debug)
+p.parse(%|{|)

data/test/helper.rb

@@ -19,10 +19,16 @@ require 'pp'
 require 'oj'
 
 
-if defined?(GC.verify_compaction_references) == 'method'
+def verify_gc_compaction
   # This method was added in Ruby 3.0.0. Calling it this way asks the GC to
   # move objects around, helping to find object movement bugs.
-  GC.verify_compaction_references(double_heap: true, toward: :empty)
+  if defined?(GC.verify_compaction_references) == 'method' && !(RbConfig::CONFIG['host_os'] =~ /(mingw|mswin)/)
+    if Gem::Version.new(RUBY_VERSION) >= Gem::Version.new("3.2.0")
+      GC.verify_compaction_references(expand_heap: true, toward: :empty)
+    else
+      GC.verify_compaction_references(double_heap: true, toward: :empty)
+    end
+  end
 end
 
 

data/test/json_gem/test_helper.rb

@@ -15,10 +15,14 @@ else
   require 'oj'
   Oj.mimic_JSON
 
+  # This method was added in Ruby 3.0.0. Calling it this way asks the GC to
+  # move objects around, helping to find object movement bugs.
   if defined?(GC.verify_compaction_references) == 'method'
-    # This method was added in Ruby 3.0.0. Calling it this way asks the GC to
-    # move objects around, helping to find object movement bugs.
-    GC.verify_compaction_references(double_heap: true, toward: :empty)
+    if Gem::Version.new(RUBY_VERSION) >= Gem::Version.new("3.2.0")
+      GC.verify_compaction_references(expand_heap: true, toward: :empty)
+    else
+      GC.verify_compaction_references(double_heap: true, toward: :empty)
+    end
   end
 end
 

data/test/test_compat.rb

@@ -504,6 +504,15 @@ class CompatJuice < Minitest::Test
     assert_equal("ぴーたー", Oj.load(json)['a'])
   end
 
+  def test_parse_large_escaped_string
+    invalid_json = %|{"a":\"aaaa\\nbbbb\\rcccc\\tddd\\feee\\bf\/\\\\\\u3074\\u30fc\\u305f\\u30fc                             }|
+    error = assert_raises() { Oj.load(invalid_json) }
+    assert(error.message.include?('quoted string not terminated'))
+
+    json = "\"aaaa\\nbbbb\\rcccc\\tddd\\feee\\bf\/\\\\\\u3074\\u30fc\\u305f\\u30fc             \""
+    assert_equal("aaaa\nbbbb\rcccc\tddd\feee\bf/\\ぴーたー             ", Oj.load(json))
+  end
+
   def dump_and_load(obj, trace=false)
     json = Oj.dump(obj)
     puts json if trace

data/test/test_object.rb

@@ -224,7 +224,7 @@ class ObjectJuice < Minitest::Test
 #=begin
     if '3.1.0' <= RUBY_VERSION && !(RbConfig::CONFIG['host_os'] =~ /(mingw|mswin)/)
       #Oj::debug_odd("teardown before GC.verify_compaction_references")
-      GC.verify_compaction_references(double_heap: true, toward: :empty)
+      verify_gc_compaction
       #Oj::debug_odd("teardown after GC.verify_compaction_references")
     end
 #=end

data/test/test_parser_saj.rb

@@ -149,6 +149,17 @@ class SajTest < Minitest::Test
     assert_equal((12345.6789e7 * 10000).to_i, (handler.calls[0][1] * 10000).to_i)
   end
 
+  def test_bignum
+    handler = AllSaj.new()
+    json = %{-11.899999999999999}
+    p = Oj::Parser.new(:saj)
+    p.handler = handler
+    p.parse(json)
+    assert_equal(1, handler.calls.size)
+    assert_equal(:add_value, handler.calls[0][0])
+    assert_equal(-118999, (handler.calls[0][1] * 10000).to_i)
+  end
+
   def test_array_empty
     handler = AllSaj.new()
     json = %{[]}

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: oj
 version: !ruby/object:Gem::Version
-  version: 3.13.16
+  version: 3.13.19
 platform: ruby
 authors:
 - Peter Ohler
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2022-07-06 00:00:00.000000000 Z
+date: 2022-07-29 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rake-compiler