oj 3.16.3 → 3.17.4
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- checksums.yaml +4 -4
- data/CHANGELOG.md +99 -0
- data/README.md +0 -16
- data/ext/oj/compat.c +3 -3
- data/ext/oj/custom.c +17 -11
- data/ext/oj/dump.c +619 -88
- data/ext/oj/dump.h +9 -2
- data/ext/oj/dump_compat.c +16 -9
- data/ext/oj/dump_leaf.c +1 -1
- data/ext/oj/dump_object.c +52 -27
- data/ext/oj/dump_strict.c +20 -14
- data/ext/oj/extconf.rb +11 -9
- data/ext/oj/fast.c +90 -65
- data/ext/oj/intern.c +1 -1
- data/ext/oj/mimic_json.c +14 -8
- data/ext/oj/object.c +18 -12
- data/ext/oj/oj.c +426 -11
- data/ext/oj/oj.h +60 -52
- data/ext/oj/parse.c +252 -48
- data/ext/oj/parse.h +0 -1
- data/ext/oj/parser.c +179 -47
- data/ext/oj/parser.h +4 -2
- data/ext/oj/rails.c +57 -36
- data/ext/oj/reader.c +17 -5
- data/ext/oj/rxclass.c +17 -1
- data/ext/oj/rxclass.h +2 -1
- data/ext/oj/safe.c +230 -0
- data/ext/oj/safe.h +79 -0
- data/ext/oj/saj.c +55 -20
- data/ext/oj/scp.c +3 -6
- data/ext/oj/simd.h +219 -0
- data/ext/oj/sparse.c +15 -3
- data/ext/oj/stream_writer.c +19 -14
- data/ext/oj/strict.c +2 -4
- data/ext/oj/string_writer.c +26 -14
- data/ext/oj/usual.c +37 -35
- data/ext/oj/wab.c +5 -4
- data/lib/oj/mimic.rb +1 -5
- data/lib/oj/schandler.rb +5 -4
- data/lib/oj/version.rb +1 -1
- data/pages/Encoding.md +1 -1
- metadata +22 -100
- data/test/_test_active.rb +0 -75
- data/test/_test_active_mimic.rb +0 -95
- data/test/_test_mimic_rails.rb +0 -123
- data/test/activerecord/result_test.rb +0 -31
- data/test/activesupport6/abstract_unit.rb +0 -44
- data/test/activesupport6/decoding_test.rb +0 -133
- data/test/activesupport6/encoding_test.rb +0 -507
- data/test/activesupport6/encoding_test_cases.rb +0 -98
- data/test/activesupport6/test_common.rb +0 -17
- data/test/activesupport6/test_helper.rb +0 -163
- data/test/activesupport6/time_zone_test_helpers.rb +0 -39
- data/test/activesupport7/abstract_unit.rb +0 -49
- data/test/activesupport7/decoding_test.rb +0 -125
- data/test/activesupport7/encoding_test.rb +0 -486
- data/test/activesupport7/encoding_test_cases.rb +0 -104
- data/test/activesupport7/time_zone_test_helpers.rb +0 -47
- data/test/files.rb +0 -29
- data/test/foo.rb +0 -14
- data/test/helper.rb +0 -39
- data/test/isolated/shared.rb +0 -309
- data/test/isolated/test_mimic_after.rb +0 -13
- data/test/isolated/test_mimic_alone.rb +0 -12
- data/test/isolated/test_mimic_as_json.rb +0 -45
- data/test/isolated/test_mimic_before.rb +0 -13
- data/test/isolated/test_mimic_define.rb +0 -28
- data/test/isolated/test_mimic_rails_after.rb +0 -22
- data/test/isolated/test_mimic_rails_before.rb +0 -21
- data/test/isolated/test_mimic_redefine.rb +0 -15
- data/test/json_gem/json_addition_test.rb +0 -216
- data/test/json_gem/json_common_interface_test.rb +0 -155
- data/test/json_gem/json_encoding_test.rb +0 -107
- data/test/json_gem/json_ext_parser_test.rb +0 -21
- data/test/json_gem/json_fixtures_test.rb +0 -36
- data/test/json_gem/json_generator_test.rb +0 -413
- data/test/json_gem/json_generic_object_test.rb +0 -90
- data/test/json_gem/json_parser_test.rb +0 -477
- data/test/json_gem/json_string_matching_test.rb +0 -42
- data/test/json_gem/test_helper.rb +0 -30
- data/test/mem.rb +0 -34
- data/test/perf.rb +0 -102
- data/test/perf_compat.rb +0 -128
- data/test/perf_dump.rb +0 -50
- data/test/perf_fast.rb +0 -162
- data/test/perf_file.rb +0 -62
- data/test/perf_object.rb +0 -134
- data/test/perf_once.rb +0 -59
- data/test/perf_parser.rb +0 -183
- data/test/perf_saj.rb +0 -101
- data/test/perf_scp.rb +0 -140
- data/test/perf_simple.rb +0 -289
- data/test/perf_strict.rb +0 -137
- data/test/perf_wab.rb +0 -129
- data/test/prec.rb +0 -23
- data/test/sample/change.rb +0 -13
- data/test/sample/dir.rb +0 -18
- data/test/sample/doc.rb +0 -35
- data/test/sample/file.rb +0 -47
- data/test/sample/group.rb +0 -15
- data/test/sample/hasprops.rb +0 -15
- data/test/sample/layer.rb +0 -11
- data/test/sample/line.rb +0 -20
- data/test/sample/oval.rb +0 -10
- data/test/sample/rect.rb +0 -9
- data/test/sample/shape.rb +0 -34
- data/test/sample/text.rb +0 -19
- data/test/sample.rb +0 -54
- data/test/sample_json.rb +0 -37
- data/test/test_compat.rb +0 -567
- data/test/test_custom.rb +0 -555
- data/test/test_debian.rb +0 -50
- data/test/test_fast.rb +0 -526
- data/test/test_file.rb +0 -250
- data/test/test_gc.rb +0 -60
- data/test/test_generate.rb +0 -21
- data/test/test_hash.rb +0 -39
- data/test/test_integer_range.rb +0 -72
- data/test/test_null.rb +0 -376
- data/test/test_object.rb +0 -1030
- data/test/test_parser.rb +0 -11
- data/test/test_parser_debug.rb +0 -27
- data/test/test_parser_saj.rb +0 -337
- data/test/test_parser_usual.rb +0 -227
- data/test/test_rails.rb +0 -35
- data/test/test_saj.rb +0 -188
- data/test/test_scp.rb +0 -431
- data/test/test_strict.rb +0 -441
- data/test/test_various.rb +0 -801
- data/test/test_wab.rb +0 -311
- data/test/test_writer.rb +0 -380
- data/test/tests.rb +0 -33
- data/test/tests_mimic.rb +0 -23
- data/test/tests_mimic_addition.rb +0 -16
data/ext/oj/saj.c
CHANGED
|
@@ -82,7 +82,14 @@ inline static void next_non_white(ParseInfo pi) {
|
|
|
82
82
|
case '\f':
|
|
83
83
|
case '\n':
|
|
84
84
|
case '\r': break;
|
|
85
|
-
case '/':
|
|
85
|
+
case '/':
|
|
86
|
+
skip_comment(pi);
|
|
87
|
+
/* A comment that is not terminated by a newline ends on the null
|
|
88
|
+
* terminator. Stop here so the loop does not step past it. */
|
|
89
|
+
if ('\0' == *pi->s) {
|
|
90
|
+
return;
|
|
91
|
+
}
|
|
92
|
+
break;
|
|
86
93
|
default: return;
|
|
87
94
|
}
|
|
88
95
|
}
|
|
@@ -94,8 +101,7 @@ inline static void call_add_value(VALUE handler, VALUE value, const char *key) {
|
|
|
94
101
|
if (0 == key) {
|
|
95
102
|
k = Qnil;
|
|
96
103
|
} else {
|
|
97
|
-
k =
|
|
98
|
-
k = oj_encode(k);
|
|
104
|
+
k = rb_utf8_str_new_cstr(key);
|
|
99
105
|
}
|
|
100
106
|
rb_funcall(handler, oj_add_value_id, 2, value, k);
|
|
101
107
|
}
|
|
@@ -106,8 +112,7 @@ inline static void call_no_value(VALUE handler, ID method, const char *key) {
|
|
|
106
112
|
if (0 == key) {
|
|
107
113
|
k = Qnil;
|
|
108
114
|
} else {
|
|
109
|
-
k =
|
|
110
|
-
k = oj_encode(k);
|
|
115
|
+
k = rb_utf8_str_new_cstr(key);
|
|
111
116
|
}
|
|
112
117
|
rb_funcall(handler, method, 1, k);
|
|
113
118
|
}
|
|
@@ -120,14 +125,15 @@ static void skip_comment(ParseInfo pi) {
|
|
|
120
125
|
if ('*' == *pi->s && '/' == *(pi->s + 1)) {
|
|
121
126
|
pi->s++;
|
|
122
127
|
return;
|
|
123
|
-
} else if ('\0' == *pi->s) {
|
|
124
|
-
if (pi->has_error) {
|
|
125
|
-
call_error("comment not terminated", pi, __FILE__, __LINE__);
|
|
126
|
-
} else {
|
|
127
|
-
raise_error("comment not terminated", pi->str, pi->s);
|
|
128
|
-
}
|
|
129
128
|
}
|
|
130
129
|
}
|
|
130
|
+
/* The loop only ends on the null terminator so the comment was never
|
|
131
|
+
* closed. */
|
|
132
|
+
if (pi->has_error) {
|
|
133
|
+
call_error("comment not terminated", pi, __FILE__, __LINE__);
|
|
134
|
+
} else {
|
|
135
|
+
raise_error("comment not terminated", pi->str, pi->s);
|
|
136
|
+
}
|
|
131
137
|
} else if ('/' == *pi->s) {
|
|
132
138
|
for (; 1; pi->s++) {
|
|
133
139
|
switch (*pi->s) {
|
|
@@ -257,9 +263,8 @@ static void read_str(ParseInfo pi, const char *key) {
|
|
|
257
263
|
|
|
258
264
|
text = read_quoted_value(pi);
|
|
259
265
|
if (pi->has_add_value) {
|
|
260
|
-
VALUE s =
|
|
266
|
+
VALUE s = rb_utf8_str_new_cstr(text);
|
|
261
267
|
|
|
262
|
-
s = oj_encode(s);
|
|
263
268
|
call_add_value(pi->handler, s, key);
|
|
264
269
|
}
|
|
265
270
|
}
|
|
@@ -578,7 +583,7 @@ static void saj_parse(VALUE handler, char *json) {
|
|
|
578
583
|
/* initialize parse info */
|
|
579
584
|
pi.str = json;
|
|
580
585
|
pi.s = json;
|
|
581
|
-
#if IS_WINDOWS
|
|
586
|
+
#if IS_WINDOWS || !defined(HAVE_GETRLIMIT)
|
|
582
587
|
pi.stack_min = (void *)((char *)&obj - (512L * 1024L)); /* assume a 1M stack and give half to ruby */
|
|
583
588
|
#else
|
|
584
589
|
{
|
|
@@ -587,7 +592,7 @@ static void saj_parse(VALUE handler, char *json) {
|
|
|
587
592
|
if (0 == getrlimit(RLIMIT_STACK, &lim) && RLIM_INFINITY != lim.rlim_cur) {
|
|
588
593
|
pi.stack_min = (void *)((char *)&obj - (lim.rlim_cur / 4 * 3)); /* let 3/4ths of the stack be used only */
|
|
589
594
|
} else {
|
|
590
|
-
pi.stack_min = 0;
|
|
595
|
+
pi.stack_min = 0; /* indicates not to check stack limit */
|
|
591
596
|
}
|
|
592
597
|
}
|
|
593
598
|
#endif
|
|
@@ -609,6 +614,19 @@ static void saj_parse(VALUE handler, char *json) {
|
|
|
609
614
|
}
|
|
610
615
|
}
|
|
611
616
|
|
|
617
|
+
struct _sajArgs {
|
|
618
|
+
VALUE handler;
|
|
619
|
+
char *json;
|
|
620
|
+
};
|
|
621
|
+
|
|
622
|
+
static VALUE protect_saj_parse(VALUE x) {
|
|
623
|
+
struct _sajArgs *args = (struct _sajArgs *)x;
|
|
624
|
+
|
|
625
|
+
saj_parse(args->handler, args->json);
|
|
626
|
+
|
|
627
|
+
return Qnil;
|
|
628
|
+
}
|
|
629
|
+
|
|
612
630
|
/* call-seq: saj_parse(handler, io)
|
|
613
631
|
*
|
|
614
632
|
* Parses an IO stream or file containing an JSON document. Raises an exception
|
|
@@ -651,8 +669,16 @@ oj_saj_parse(int argc, VALUE *argv, VALUE self) {
|
|
|
651
669
|
len = lseek(fd, 0, SEEK_END);
|
|
652
670
|
lseek(fd, 0, SEEK_SET);
|
|
653
671
|
json = OJ_R_ALLOC_N(char, len + 1);
|
|
654
|
-
|
|
655
|
-
|
|
672
|
+
{
|
|
673
|
+
size_t total = 0;
|
|
674
|
+
|
|
675
|
+
while (total < len) {
|
|
676
|
+
cnt = read(fd, json + total, len - total);
|
|
677
|
+
if (cnt <= 0) {
|
|
678
|
+
rb_raise(rb_eIOError, "failed to read from IO Object.");
|
|
679
|
+
}
|
|
680
|
+
total += cnt;
|
|
681
|
+
}
|
|
656
682
|
}
|
|
657
683
|
json[len] = '\0';
|
|
658
684
|
#endif
|
|
@@ -665,8 +691,17 @@ oj_saj_parse(int argc, VALUE *argv, VALUE self) {
|
|
|
665
691
|
rb_raise(rb_eArgError, "saj_parse() expected a String or IO Object.");
|
|
666
692
|
}
|
|
667
693
|
}
|
|
668
|
-
|
|
669
|
-
|
|
670
|
-
|
|
694
|
+
{
|
|
695
|
+
// saj_parse() raises on a malformed document so the json buffer has to
|
|
696
|
+
// be freed even when the parse does not return normally.
|
|
697
|
+
struct _sajArgs args = {*argv, json};
|
|
698
|
+
int ex = 0;
|
|
699
|
+
|
|
700
|
+
rb_protect(protect_saj_parse, (VALUE)&args, &ex);
|
|
701
|
+
OJ_R_FREE(json);
|
|
702
|
+
if (0 != ex) {
|
|
703
|
+
rb_jump_tag(ex);
|
|
704
|
+
}
|
|
705
|
+
}
|
|
671
706
|
return Qnil;
|
|
672
707
|
}
|
data/ext/oj/scp.c
CHANGED
|
@@ -56,9 +56,8 @@ static void add_value(ParseInfo pi, VALUE val) {
|
|
|
56
56
|
}
|
|
57
57
|
|
|
58
58
|
static void add_cstr(ParseInfo pi, const char *str, size_t len, const char *orig) {
|
|
59
|
-
volatile VALUE rstr =
|
|
59
|
+
volatile VALUE rstr = rb_utf8_str_new(str, len);
|
|
60
60
|
|
|
61
|
-
rstr = oj_encode(rstr);
|
|
62
61
|
rb_funcall(pi->handler, oj_add_value_id, 1, rstr);
|
|
63
62
|
}
|
|
64
63
|
|
|
@@ -87,9 +86,8 @@ static VALUE hash_key(ParseInfo pi, const char *key, size_t klen) {
|
|
|
87
86
|
}
|
|
88
87
|
|
|
89
88
|
static void hash_set_cstr(ParseInfo pi, Val kval, const char *str, size_t len, const char *orig) {
|
|
90
|
-
volatile VALUE rstr =
|
|
89
|
+
volatile VALUE rstr = rb_utf8_str_new(str, len);
|
|
91
90
|
|
|
92
|
-
rstr = oj_encode(rstr);
|
|
93
91
|
rb_funcall(pi->handler, oj_hash_set_id, 3, stack_peek(&pi->stack)->val, oj_calc_hash_key(pi, kval), rstr);
|
|
94
92
|
}
|
|
95
93
|
|
|
@@ -107,9 +105,8 @@ static void hash_set_value(ParseInfo pi, Val kval, VALUE value) {
|
|
|
107
105
|
}
|
|
108
106
|
|
|
109
107
|
static void array_append_cstr(ParseInfo pi, const char *str, size_t len, const char *orig) {
|
|
110
|
-
volatile VALUE rstr =
|
|
108
|
+
volatile VALUE rstr = rb_utf8_str_new(str, len);
|
|
111
109
|
|
|
112
|
-
rstr = oj_encode(rstr);
|
|
113
110
|
rb_funcall(pi->handler, oj_array_append_id, 2, stack_peek(&pi->stack)->val, rstr);
|
|
114
111
|
}
|
|
115
112
|
|
data/ext/oj/simd.h
ADDED
|
@@ -0,0 +1,219 @@
|
|
|
1
|
+
#ifndef OJ_SIMD_H
|
|
2
|
+
#define OJ_SIMD_H
|
|
3
|
+
|
|
4
|
+
// SIMD architecture detection and configuration
|
|
5
|
+
// This header provides unified SIMD support across different CPU architectures
|
|
6
|
+
// with cross-platform runtime detection (Windows/Linux/Mac)
|
|
7
|
+
|
|
8
|
+
// SIMD implementation enum - used for runtime selection
|
|
9
|
+
typedef enum _simd_implementation { SIMD_NONE, SIMD_NEON, SIMD_SSE2, SIMD_SSE42 } SIMD_Implementation;
|
|
10
|
+
|
|
11
|
+
// Define in oj.c.
|
|
12
|
+
extern SIMD_Implementation SIMD_Impl;
|
|
13
|
+
|
|
14
|
+
// Runtime CPU detection function (implemented in oj.c)
|
|
15
|
+
SIMD_Implementation oj_get_simd_implementation(void);
|
|
16
|
+
|
|
17
|
+
// =============================================================================
|
|
18
|
+
// Compiler compatibility macros
|
|
19
|
+
// =============================================================================
|
|
20
|
+
|
|
21
|
+
// Branch prediction hints
|
|
22
|
+
#if defined(__GNUC__) || defined(__clang__)
|
|
23
|
+
#define OJ_LIKELY(x) __builtin_expect(!!(x), 1)
|
|
24
|
+
#define OJ_UNLIKELY(x) __builtin_expect(!!(x), 0)
|
|
25
|
+
#else
|
|
26
|
+
#define OJ_LIKELY(x) (x)
|
|
27
|
+
#define OJ_UNLIKELY(x) (x)
|
|
28
|
+
#endif
|
|
29
|
+
|
|
30
|
+
// Prefetch hints
|
|
31
|
+
#if defined(__GNUC__) || defined(__clang__)
|
|
32
|
+
#define OJ_PREFETCH(addr) __builtin_prefetch(addr, 0, 0)
|
|
33
|
+
#elif defined(_MSC_VER)
|
|
34
|
+
#include <intrin.h>
|
|
35
|
+
#define OJ_PREFETCH(addr) _mm_prefetch((const char *)(addr), _MM_HINT_T0)
|
|
36
|
+
#else
|
|
37
|
+
#define OJ_PREFETCH(addr) ((void)0)
|
|
38
|
+
#endif
|
|
39
|
+
|
|
40
|
+
// Count trailing zeros (for SSE2 mask scanning)
|
|
41
|
+
#if defined(__GNUC__) || defined(__clang__)
|
|
42
|
+
#define OJ_CTZ(x) __builtin_ctz(x)
|
|
43
|
+
#define OJ_CTZ64(x) __builtin_ctzll(x)
|
|
44
|
+
#elif defined(_MSC_VER)
|
|
45
|
+
#include <intrin.h>
|
|
46
|
+
static __inline int oj_ctz_msvc(unsigned int x) {
|
|
47
|
+
unsigned long index;
|
|
48
|
+
if (0 == x) {
|
|
49
|
+
return 32;
|
|
50
|
+
}
|
|
51
|
+
_BitScanForward(&index, x);
|
|
52
|
+
return (int)index;
|
|
53
|
+
}
|
|
54
|
+
static __inline int oj_ctz64_msvc(uint64_t x) {
|
|
55
|
+
unsigned long index;
|
|
56
|
+
if (_BitScanForward64(&index, x)) {
|
|
57
|
+
return (int)index;
|
|
58
|
+
}
|
|
59
|
+
return 64;
|
|
60
|
+
}
|
|
61
|
+
#define OJ_CTZ(x) oj_ctz_msvc(x)
|
|
62
|
+
#define OJ_CTZ64(x) oj_ctz64_msvc(x)
|
|
63
|
+
#else
|
|
64
|
+
// Fallback: naive implementation
|
|
65
|
+
static inline int oj_ctz_fallback(unsigned int x) {
|
|
66
|
+
int count = 0;
|
|
67
|
+
while ((x & 1) == 0 && count < 32) {
|
|
68
|
+
x >>= 1;
|
|
69
|
+
count++;
|
|
70
|
+
}
|
|
71
|
+
return count;
|
|
72
|
+
}
|
|
73
|
+
|
|
74
|
+
static inline int oj_ctz64_fallback(uint64_t x) {
|
|
75
|
+
int count = 0;
|
|
76
|
+
while ((x & 1) == 0 && count < 64) {
|
|
77
|
+
x >>= 1;
|
|
78
|
+
count++;
|
|
79
|
+
}
|
|
80
|
+
return count;
|
|
81
|
+
}
|
|
82
|
+
#define OJ_CTZ(x) oj_ctz_fallback(x)
|
|
83
|
+
#define OJ_CTZ64(x) oj_ctz64_fallback(x)
|
|
84
|
+
#endif
|
|
85
|
+
|
|
86
|
+
// =============================================================================
|
|
87
|
+
// x86/x86_64 SIMD detection
|
|
88
|
+
// =============================================================================
|
|
89
|
+
#if defined(__x86_64__) || defined(__i386__) || defined(_M_IX86) || defined(_M_X64)
|
|
90
|
+
#define HAVE_SIMD_X86 1
|
|
91
|
+
|
|
92
|
+
// Include appropriate SIMD headers
|
|
93
|
+
#if defined(_MSC_VER)
|
|
94
|
+
// MSVC: use intrin.h for all intrinsics
|
|
95
|
+
#include <intrin.h>
|
|
96
|
+
#define HAVE_SIMD_SSE4_2 1
|
|
97
|
+
#define HAVE_SIMD_SSE2 1
|
|
98
|
+
#elif defined(__GNUC__) || defined(__clang__)
|
|
99
|
+
// GCC/Clang: check for header availability and include them
|
|
100
|
+
// We include headers but use target attributes to enable instructions per-function
|
|
101
|
+
// Include cpuid.h for __get_cpuid fallback when __builtin_cpu_supports is unavailable
|
|
102
|
+
#if __has_include(<cpuid.h>)
|
|
103
|
+
#include <cpuid.h>
|
|
104
|
+
#endif
|
|
105
|
+
#if defined(__SSE4_2__) || defined(__SSE2__)
|
|
106
|
+
// If any SSE is enabled globally, x86intrin.h should be available
|
|
107
|
+
#include <x86intrin.h>
|
|
108
|
+
#define HAVE_SIMD_SSE4_2 1
|
|
109
|
+
#define HAVE_SIMD_SSE2 1
|
|
110
|
+
#else
|
|
111
|
+
// Try to include headers anyway for target attribute functions
|
|
112
|
+
#if __has_include(<x86intrin.h>)
|
|
113
|
+
#include <x86intrin.h>
|
|
114
|
+
#define HAVE_SIMD_SSE4_2 1
|
|
115
|
+
#define HAVE_SIMD_SSE2 1
|
|
116
|
+
#elif __has_include(<nmmintrin.h>)
|
|
117
|
+
#include <nmmintrin.h>
|
|
118
|
+
#define HAVE_SIMD_SSE4_2 1
|
|
119
|
+
#define HAVE_SIMD_SSE2 1
|
|
120
|
+
#elif __has_include(<emmintrin.h>)
|
|
121
|
+
#include <emmintrin.h>
|
|
122
|
+
#define HAVE_SIMD_SSE2 1
|
|
123
|
+
#endif
|
|
124
|
+
#endif
|
|
125
|
+
#endif
|
|
126
|
+
|
|
127
|
+
// Target attribute macros for function-level SIMD enabling
|
|
128
|
+
#if defined(__clang__) || defined(__GNUC__)
|
|
129
|
+
#define OJ_TARGET_SSE42 __attribute__((target("sse4.2")))
|
|
130
|
+
#define OJ_TARGET_SSE2 __attribute__((target("sse2")))
|
|
131
|
+
#else
|
|
132
|
+
// MSVC doesn't need target attributes - intrinsics are always available
|
|
133
|
+
#define OJ_TARGET_SSE42
|
|
134
|
+
#define OJ_TARGET_SSE2
|
|
135
|
+
#endif
|
|
136
|
+
|
|
137
|
+
#endif // x86/x86_64
|
|
138
|
+
|
|
139
|
+
// =============================================================================
|
|
140
|
+
// ARM NEON detection
|
|
141
|
+
// =============================================================================
|
|
142
|
+
#if defined(__ARM_NEON) || defined(__ARM_NEON__) || defined(__aarch64__) || defined(_M_ARM64)
|
|
143
|
+
#define HAVE_SIMD_NEON 1
|
|
144
|
+
#define SIMD_MINIMUM_THRESHOLD 6
|
|
145
|
+
#include <arm_neon.h>
|
|
146
|
+
#endif
|
|
147
|
+
|
|
148
|
+
// =============================================================================
|
|
149
|
+
// SIMD type string for debugging/logging
|
|
150
|
+
// =============================================================================
|
|
151
|
+
#if defined(HAVE_SIMD_SSE4_2) || defined(HAVE_SIMD_SSE2)
|
|
152
|
+
#define HAVE_SIMD_STRING_SCAN 1
|
|
153
|
+
#define SIMD_TYPE "x86 (runtime detected)"
|
|
154
|
+
#elif defined(HAVE_SIMD_NEON)
|
|
155
|
+
#define HAVE_SIMD_STRING_SCAN 1
|
|
156
|
+
#define SIMD_TYPE "NEON"
|
|
157
|
+
#else
|
|
158
|
+
#define SIMD_TYPE "none"
|
|
159
|
+
#endif
|
|
160
|
+
|
|
161
|
+
#if defined(HAVE_SIMD_SSE4_2)
|
|
162
|
+
|
|
163
|
+
#define SIMD_MINIMUM_THRESHOLD 6
|
|
164
|
+
|
|
165
|
+
extern void initialize_sse42(void);
|
|
166
|
+
|
|
167
|
+
static inline OJ_TARGET_SSE42 __m128i vector_lookup_sse42(__m128i input, __m128i *lookup_table, int tab_size) {
|
|
168
|
+
// Extract high 4 bits to determine which 16-byte chunk (0-15)
|
|
169
|
+
__m128i hi_index = _mm_and_si128(_mm_srli_epi32(input, 4), _mm_set1_epi8(0x0F));
|
|
170
|
+
|
|
171
|
+
// Extract low 4 bits for index within the chunk (0-15)
|
|
172
|
+
__m128i low_index = _mm_and_si128(input, _mm_set1_epi8(0x0F));
|
|
173
|
+
|
|
174
|
+
// Perform lookups in all 16 tables
|
|
175
|
+
__m128i results[16];
|
|
176
|
+
for (int i = 0; i < tab_size; i++) {
|
|
177
|
+
results[i] = _mm_shuffle_epi8(lookup_table[i], low_index);
|
|
178
|
+
}
|
|
179
|
+
|
|
180
|
+
// Create masks for each chunk and blend results
|
|
181
|
+
__m128i final_result = _mm_setzero_si128();
|
|
182
|
+
|
|
183
|
+
for (int i = 0; i < tab_size; i++) {
|
|
184
|
+
__m128i mask = _mm_cmpeq_epi8(hi_index, _mm_set1_epi8(i));
|
|
185
|
+
__m128i masked_result = _mm_and_si128(mask, results[i]);
|
|
186
|
+
final_result = _mm_or_si128(final_result, masked_result);
|
|
187
|
+
}
|
|
188
|
+
|
|
189
|
+
return final_result;
|
|
190
|
+
}
|
|
191
|
+
|
|
192
|
+
#endif
|
|
193
|
+
|
|
194
|
+
#ifndef __has_builtin
|
|
195
|
+
#define __has_builtin(x) 0
|
|
196
|
+
#endif
|
|
197
|
+
|
|
198
|
+
#if __has_builtin(__builtin_memcpy)
|
|
199
|
+
#define HAVE_FAST_MEMCPY 1
|
|
200
|
+
|
|
201
|
+
inline static void fast_memcpy16(void *dest, const void *src, size_t n) {
|
|
202
|
+
char *d = (char *)dest;
|
|
203
|
+
char *s = (char *)src;
|
|
204
|
+
if (n >= 8) {
|
|
205
|
+
__builtin_memcpy(d, s, 8);
|
|
206
|
+
__builtin_memcpy(d + n - 8, s + n - 8, 8);
|
|
207
|
+
} else if (n >= 4) {
|
|
208
|
+
__builtin_memcpy(d, s, 4);
|
|
209
|
+
__builtin_memcpy(d + n - 4, s + n - 4, 4);
|
|
210
|
+
} else if (n >= 2) {
|
|
211
|
+
__builtin_memcpy(d, s, 2);
|
|
212
|
+
__builtin_memcpy(d + n - 2, s + n - 2, 2);
|
|
213
|
+
} else if (n >= 1) {
|
|
214
|
+
*d = *s;
|
|
215
|
+
}
|
|
216
|
+
}
|
|
217
|
+
#endif
|
|
218
|
+
|
|
219
|
+
#endif /* OJ_SIMD_H */
|
data/ext/oj/sparse.c
CHANGED
|
@@ -399,6 +399,7 @@ static void read_num(ParseInfo pi) {
|
|
|
399
399
|
char c;
|
|
400
400
|
|
|
401
401
|
reader_protect(&pi->rd);
|
|
402
|
+
ni.pi = pi;
|
|
402
403
|
ni.i = 0;
|
|
403
404
|
ni.num = 0;
|
|
404
405
|
ni.div = 1;
|
|
@@ -415,7 +416,7 @@ static void read_num(ParseInfo pi) {
|
|
|
415
416
|
ni.bigdec_load = pi->options.compat_bigdec;
|
|
416
417
|
} else {
|
|
417
418
|
ni.no_big = (FloatDec == pi->options.bigdec_load || FastDec == pi->options.bigdec_load ||
|
|
418
|
-
|
|
419
|
+
RubyDec == pi->options.bigdec_load);
|
|
419
420
|
ni.bigdec_load = pi->options.bigdec_load;
|
|
420
421
|
}
|
|
421
422
|
|
|
@@ -549,6 +550,7 @@ static void read_nan(ParseInfo pi) {
|
|
|
549
550
|
struct _numInfo ni;
|
|
550
551
|
char c;
|
|
551
552
|
|
|
553
|
+
ni.pi = pi;
|
|
552
554
|
ni.str = pi->rd.str;
|
|
553
555
|
ni.i = 0;
|
|
554
556
|
ni.num = 0;
|
|
@@ -565,7 +567,7 @@ static void read_nan(ParseInfo pi) {
|
|
|
565
567
|
ni.bigdec_load = pi->options.compat_bigdec;
|
|
566
568
|
} else {
|
|
567
569
|
ni.no_big = (FloatDec == pi->options.bigdec_load || FastDec == pi->options.bigdec_load ||
|
|
568
|
-
|
|
570
|
+
RubyDec == pi->options.bigdec_load);
|
|
569
571
|
ni.bigdec_load = pi->options.bigdec_load;
|
|
570
572
|
}
|
|
571
573
|
|
|
@@ -745,6 +747,7 @@ void oj_sparse2(ParseInfo pi) {
|
|
|
745
747
|
oj_set_error_at(pi, oj_parse_error_class, __FILE__, __LINE__, "expected NaN");
|
|
746
748
|
return;
|
|
747
749
|
}
|
|
750
|
+
ni.pi = pi;
|
|
748
751
|
ni.str = pi->rd.str;
|
|
749
752
|
ni.i = 0;
|
|
750
753
|
ni.num = 0;
|
|
@@ -761,7 +764,7 @@ void oj_sparse2(ParseInfo pi) {
|
|
|
761
764
|
ni.bigdec_load = pi->options.compat_bigdec;
|
|
762
765
|
} else {
|
|
763
766
|
ni.no_big = (FloatDec == pi->options.bigdec_load || FastDec == pi->options.bigdec_load ||
|
|
764
|
-
|
|
767
|
+
RubyDec == pi->options.bigdec_load);
|
|
765
768
|
ni.bigdec_load = pi->options.bigdec_load;
|
|
766
769
|
}
|
|
767
770
|
add_num_value(pi, &ni);
|
|
@@ -904,6 +907,15 @@ CLEANUP:
|
|
|
904
907
|
oj_circ_array_free(pi->circ_array);
|
|
905
908
|
}
|
|
906
909
|
stack_cleanup(&pi->stack);
|
|
910
|
+
// A :match_string option builds a chain of regexps for this call that has to
|
|
911
|
+
// be freed. Without the option the member aliases the chain owned by the
|
|
912
|
+
// defaults and must be left alone.
|
|
913
|
+
if (pi->options.str_rx.head != oj_default_options.str_rx.head) {
|
|
914
|
+
oj_rxclass_cleanup(&pi->options.str_rx);
|
|
915
|
+
pi->options.str_rx.head = NULL;
|
|
916
|
+
pi->options.str_rx.tail = NULL;
|
|
917
|
+
}
|
|
918
|
+
oj_free_call_options(&pi->options);
|
|
907
919
|
if (0 != fd) {
|
|
908
920
|
#ifdef _WIN32
|
|
909
921
|
rb_w32_close(fd);
|
data/ext/oj/stream_writer.c
CHANGED
|
@@ -16,15 +16,29 @@ static void stream_writer_free(void *ptr) {
|
|
|
16
16
|
return;
|
|
17
17
|
}
|
|
18
18
|
sw = (StreamWriter)ptr;
|
|
19
|
+
oj_options_release(&sw->sw.opts);
|
|
19
20
|
OJ_R_FREE(sw->sw.out.buf);
|
|
20
21
|
OJ_R_FREE(sw->sw.types);
|
|
21
22
|
OJ_R_FREE(ptr);
|
|
22
23
|
}
|
|
23
24
|
|
|
25
|
+
static void stream_writer_mark(void *ptr) {
|
|
26
|
+
if (NULL != ptr) {
|
|
27
|
+
StreamWriter sw = (StreamWriter)ptr;
|
|
28
|
+
|
|
29
|
+
oj_options_mark(&sw->sw.opts);
|
|
30
|
+
// The writer can be the only reference to the stream as in
|
|
31
|
+
// Oj::StreamWriter.new(StringIO.new) so it must be marked.
|
|
32
|
+
if (Qnil != sw->stream) {
|
|
33
|
+
rb_gc_mark(sw->stream);
|
|
34
|
+
}
|
|
35
|
+
}
|
|
36
|
+
}
|
|
37
|
+
|
|
24
38
|
static const rb_data_type_t oj_stream_writer_type = {
|
|
25
39
|
"Oj/stream_writer",
|
|
26
40
|
{
|
|
27
|
-
|
|
41
|
+
stream_writer_mark,
|
|
28
42
|
stream_writer_free,
|
|
29
43
|
NULL,
|
|
30
44
|
},
|
|
@@ -42,22 +56,12 @@ static void stream_writer_write(StreamWriter sw) {
|
|
|
42
56
|
|
|
43
57
|
switch (sw->type) {
|
|
44
58
|
case STRING_IO:
|
|
45
|
-
case STREAM_IO:
|
|
46
|
-
|
|
47
|
-
|
|
48
|
-
// Oddly enough, when pushing ASCII characters with UTF-8 encoding or
|
|
49
|
-
// even ASCII-8BIT does not change the output encoding. Pushing any
|
|
50
|
-
// non-ASCII no matter what the encoding changes the output encoding
|
|
51
|
-
// to ASCII-8BIT if it the string is not forced to UTF-8 here.
|
|
52
|
-
rs = oj_encode(rs);
|
|
59
|
+
case STREAM_IO:
|
|
60
|
+
case FILE_IO: {
|
|
61
|
+
volatile VALUE rs = rb_utf8_str_new(sw->sw.out.buf, size);
|
|
53
62
|
rb_funcall(sw->stream, oj_write_id, 1, rs);
|
|
54
63
|
break;
|
|
55
64
|
}
|
|
56
|
-
case FILE_IO:
|
|
57
|
-
if (size != write(sw->fd, sw->sw.out.buf, size)) {
|
|
58
|
-
rb_raise(rb_eIOError, "Write failed. [_%d_:%s]\n", errno, strerror(errno));
|
|
59
|
-
}
|
|
60
|
-
break;
|
|
61
65
|
default: rb_raise(rb_eArgError, "expected an IO Object.");
|
|
62
66
|
}
|
|
63
67
|
stream_writer_reset_buf(sw);
|
|
@@ -126,6 +130,7 @@ static VALUE stream_writer_new(int argc, VALUE *argv, VALUE self) {
|
|
|
126
130
|
oj_str_writer_init(&sw->sw, 4096);
|
|
127
131
|
sw->flush_limit = 0;
|
|
128
132
|
}
|
|
133
|
+
oj_options_take_ownership(&sw->sw.opts);
|
|
129
134
|
sw->sw.out.indent = sw->sw.opts.indent;
|
|
130
135
|
sw->stream = stream;
|
|
131
136
|
sw->type = type;
|
data/ext/oj/strict.c
CHANGED
|
@@ -19,8 +19,7 @@ VALUE oj_cstr_to_value(const char *str, size_t len, size_t cache_str) {
|
|
|
19
19
|
if (len < cache_str) {
|
|
20
20
|
rstr = oj_str_intern(str, len);
|
|
21
21
|
} else {
|
|
22
|
-
rstr =
|
|
23
|
-
rstr = oj_encode(rstr);
|
|
22
|
+
rstr = rb_utf8_str_new(str, len);
|
|
24
23
|
}
|
|
25
24
|
return rstr;
|
|
26
25
|
}
|
|
@@ -35,8 +34,7 @@ VALUE oj_calc_hash_key(ParseInfo pi, Val parent) {
|
|
|
35
34
|
if (Yes == pi->options.sym_key) {
|
|
36
35
|
rkey = ID2SYM(rb_intern3(parent->key, parent->klen, oj_utf8_encoding));
|
|
37
36
|
} else {
|
|
38
|
-
rkey =
|
|
39
|
-
rkey = oj_encode(rkey);
|
|
37
|
+
rkey = rb_utf8_str_new(parent->key, parent->klen);
|
|
40
38
|
OBJ_FREEZE(rkey); // frozen when used as a Hash key anyway
|
|
41
39
|
}
|
|
42
40
|
return rkey;
|
data/ext/oj/string_writer.c
CHANGED
|
@@ -42,12 +42,18 @@ static void maybe_comma(StrWriter sw) {
|
|
|
42
42
|
|
|
43
43
|
// Used by stream writer also.
|
|
44
44
|
void oj_str_writer_init(StrWriter sw, int buf_size) {
|
|
45
|
-
sw->opts
|
|
46
|
-
|
|
47
|
-
|
|
48
|
-
|
|
49
|
-
|
|
50
|
-
sw->
|
|
45
|
+
sw->opts = oj_default_options;
|
|
46
|
+
// Detach from the match_string regexps owned by the defaults before the
|
|
47
|
+
// options are parsed. A :match_string option then builds a chain of its own
|
|
48
|
+
// that is freed with the writer instead of being appended to the chain the
|
|
49
|
+
// defaults are using.
|
|
50
|
+
sw->opts.str_rx.head = NULL;
|
|
51
|
+
sw->opts.str_rx.tail = NULL;
|
|
52
|
+
sw->depth = 0;
|
|
53
|
+
sw->types = OJ_R_ALLOC_N(char, 256);
|
|
54
|
+
sw->types_end = sw->types + 256;
|
|
55
|
+
*sw->types = '\0';
|
|
56
|
+
sw->keyWritten = 0;
|
|
51
57
|
|
|
52
58
|
if (0 == buf_size) {
|
|
53
59
|
buf_size = 4096;
|
|
@@ -234,15 +240,22 @@ static void string_writer_free(void *ptr) {
|
|
|
234
240
|
sw = (StrWriter)ptr;
|
|
235
241
|
|
|
236
242
|
oj_out_free(&sw->out);
|
|
243
|
+
oj_options_release(&sw->opts);
|
|
237
244
|
|
|
238
245
|
OJ_R_FREE(sw->types);
|
|
239
246
|
OJ_R_FREE(ptr);
|
|
240
247
|
}
|
|
241
248
|
|
|
249
|
+
static void string_writer_mark(void *ptr) {
|
|
250
|
+
if (NULL != ptr) {
|
|
251
|
+
oj_options_mark(&((StrWriter)ptr)->opts);
|
|
252
|
+
}
|
|
253
|
+
}
|
|
254
|
+
|
|
242
255
|
static const rb_data_type_t oj_string_writer_type = {
|
|
243
256
|
"Oj/string_writer",
|
|
244
257
|
{
|
|
245
|
-
|
|
258
|
+
string_writer_mark,
|
|
246
259
|
string_writer_free,
|
|
247
260
|
NULL,
|
|
248
261
|
},
|
|
@@ -279,6 +292,7 @@ static VALUE str_writer_new(int argc, VALUE *argv, VALUE self) {
|
|
|
279
292
|
if (1 == argc) {
|
|
280
293
|
oj_parse_options(argv[0], &sw->opts);
|
|
281
294
|
}
|
|
295
|
+
oj_options_take_ownership(&sw->opts);
|
|
282
296
|
sw->out.argc = argc - 1;
|
|
283
297
|
sw->out.argv = argv + 1;
|
|
284
298
|
sw->out.indent = sw->opts.indent;
|
|
@@ -469,22 +483,20 @@ static VALUE str_writer_reset(VALUE self) {
|
|
|
469
483
|
static VALUE str_writer_to_s(VALUE self) {
|
|
470
484
|
StrWriter sw;
|
|
471
485
|
TypedData_Get_Struct(self, struct _strWriter, &oj_string_writer_type, sw);
|
|
472
|
-
|
|
473
|
-
|
|
474
|
-
return oj_encode(rstr);
|
|
486
|
+
return rb_utf8_str_new(sw->out.buf, sw->out.cur - sw->out.buf);
|
|
475
487
|
}
|
|
476
488
|
|
|
477
489
|
/* Document-method: as_json
|
|
478
|
-
* call-seq: as_json()
|
|
490
|
+
* call-seq: as_json(*)
|
|
479
491
|
*
|
|
480
492
|
* Returns the contents of the writer as a JSON element. If called from inside
|
|
481
493
|
* an array or hash by Oj the raw buffer will be used othersize a more
|
|
482
494
|
* inefficient parse of the contents and a return of the result is
|
|
483
|
-
* completed. The parse uses the strict mode.
|
|
495
|
+
* completed. The parse uses the strict mode. Optional arguments are ignored.
|
|
484
496
|
*
|
|
485
497
|
* *return* [_Hash_|_Array_|_String_|_Integer_|_Float_|_True_|_False_|_nil|)
|
|
486
498
|
*/
|
|
487
|
-
static VALUE str_writer_as_json(VALUE self) {
|
|
499
|
+
static VALUE str_writer_as_json(int argc, VALUE *argv, VALUE self) {
|
|
488
500
|
if (string_writer_optimized) {
|
|
489
501
|
return self;
|
|
490
502
|
}
|
|
@@ -515,5 +527,5 @@ void oj_string_writer_init(void) {
|
|
|
515
527
|
rb_define_method(oj_string_writer_class, "reset", str_writer_reset, 0);
|
|
516
528
|
rb_define_method(oj_string_writer_class, "to_s", str_writer_to_s, 0);
|
|
517
529
|
rb_define_method(oj_string_writer_class, "raw_json", str_writer_to_s, 0);
|
|
518
|
-
rb_define_method(oj_string_writer_class, "as_json", str_writer_as_json,
|
|
530
|
+
rb_define_method(oj_string_writer_class, "as_json", str_writer_as_json, -1);
|
|
519
531
|
}
|