ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/JSON-XS/XS.xs
(Generate patch)

Comparing JSON-XS/XS.xs (file contents):
Revision 1.24 by root, Tue Apr 3 23:59:04 2007 UTC vs.
Revision 1.38 by root, Mon Jun 11 03:18:07 2007 UTC

… …
3#include "XSUB.h" 3#include "XSUB.h"
4 4
5#include "assert.h" 5#include "assert.h"
6#include "string.h" 6#include "string.h"
7#include "stdlib.h" 7#include "stdlib.h"
8#include "stdio.h"
9
10#if defined(__BORLANDC__) || defined(_MSC_VER)
11# define snprintf _snprintf // C compilers have this in stdio.h
12#endif
8 13
9#define F_ASCII 0x00000001UL 14#define F_ASCII 0x00000001UL
15#define F_LATIN1 0x00000002UL
10#define F_UTF8 0x00000002UL 16#define F_UTF8 0x00000004UL
11#define F_INDENT 0x00000004UL 17#define F_INDENT 0x00000008UL
12#define F_CANONICAL 0x00000008UL 18#define F_CANONICAL 0x00000010UL
13#define F_SPACE_BEFORE 0x00000010UL 19#define F_SPACE_BEFORE 0x00000020UL
14#define F_SPACE_AFTER 0x00000020UL 20#define F_SPACE_AFTER 0x00000040UL
15#define F_ALLOW_NONREF 0x00000080UL 21#define F_ALLOW_NONREF 0x00000100UL
16#define F_SHRINK 0x00000100UL 22#define F_SHRINK 0x00000200UL
17#define F_MAXDEPTH 0xf8000000UL 23#define F_MAXDEPTH 0xf8000000UL
18#define S_MAXDEPTH 27 24#define S_MAXDEPTH 27
19 25
20#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH)) 26#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
21 27
… …
26#define F_DEFAULT (9UL << S_MAXDEPTH) 32#define F_DEFAULT (9UL << S_MAXDEPTH)
27 33
28#define INIT_SIZE 32 // initial scalar size to be allocated 34#define INIT_SIZE 32 // initial scalar size to be allocated
29#define INDENT_STEP 3 // spaces per indentation level 35#define INDENT_STEP 3 // spaces per indentation level
30 36
31#define UTF8_MAX_LEN 11 // for perls UTF-X: max. number of octets per character
32#define SHORT_STRING_LEN 512 // special-case strings of up to this size 37#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
33 38
34#define SB do { 39#define SB do {
35#define SE } while (0) 40#define SE } while (0)
41
42#if __GNUC__ >= 3
43# define expect(expr,value) __builtin_expect ((expr),(value))
44# define inline inline
45#else
46# define expect(expr,value) (expr)
47# define inline static
48#endif
49
50#define expect_false(expr) expect ((expr) != 0, 0)
51#define expect_true(expr) expect ((expr) != 0, 1)
36 52
37static HV *json_stash; // JSON::XS:: 53static HV *json_stash; // JSON::XS::
38 54
39///////////////////////////////////////////////////////////////////////////// 55/////////////////////////////////////////////////////////////////////////////
40// utility functions 56// utility functions
… …
65// decode an utf-8 character and return it, or (UV)-1 in 81// decode an utf-8 character and return it, or (UV)-1 in
66// case of an error. 82// case of an error.
67// we special-case "safe" characters from U+80 .. U+7FF, 83// we special-case "safe" characters from U+80 .. U+7FF,
68// but use the very good perl function to parse anything else. 84// but use the very good perl function to parse anything else.
69// note that we never call this function for a ascii codepoints 85// note that we never call this function for a ascii codepoints
70static UV 86inline UV
71decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) 87decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
72{ 88{
73 if (s[0] > 0xdf || s[0] < 0xc2) 89 if (expect_false (s[0] > 0xdf || s[0] < 0xc2))
74 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); 90 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
75 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) 91 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)
76 { 92 {
77 *clen = 2; 93 *clen = 2;
78 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 94 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
… …
96 U32 flags; // F_* 112 U32 flags; // F_*
97 U32 indent; // indentation level 113 U32 indent; // indentation level
98 U32 maxdepth; // max. indentation/recursion level 114 U32 maxdepth; // max. indentation/recursion level
99} enc_t; 115} enc_t;
100 116
101static void 117inline void
102need (enc_t *enc, STRLEN len) 118need (enc_t *enc, STRLEN len)
103{ 119{
104 if (enc->cur + len >= enc->end) 120 if (expect_false (enc->cur + len >= enc->end))
105 { 121 {
106 STRLEN cur = enc->cur - SvPVX (enc->sv); 122 STRLEN cur = enc->cur - SvPVX (enc->sv);
107 SvGROW (enc->sv, cur + len + 1); 123 SvGROW (enc->sv, cur + len + 1);
108 enc->cur = SvPVX (enc->sv) + cur; 124 enc->cur = SvPVX (enc->sv) + cur;
109 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv); 125 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
110 } 126 }
111} 127}
112 128
113static void 129inline void
114encode_ch (enc_t *enc, char ch) 130encode_ch (enc_t *enc, char ch)
115{ 131{
116 need (enc, 1); 132 need (enc, 1);
117 *enc->cur++ = ch; 133 *enc->cur++ = ch;
118} 134}
… …
126 142
127 while (str < end) 143 while (str < end)
128 { 144 {
129 unsigned char ch = *(unsigned char *)str; 145 unsigned char ch = *(unsigned char *)str;
130 146
131 if (ch >= 0x20 && ch < 0x80) // most common case 147 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
132 { 148 {
133 if (ch == '"') // but with slow exceptions 149 if (expect_false (ch == '"')) // but with slow exceptions
134 { 150 {
135 need (enc, len += 1); 151 need (enc, len += 1);
136 *enc->cur++ = '\\'; 152 *enc->cur++ = '\\';
137 *enc->cur++ = '"'; 153 *enc->cur++ = '"';
138 } 154 }
139 else if (ch == '\\') 155 else if (expect_false (ch == '\\'))
140 { 156 {
141 need (enc, len += 1); 157 need (enc, len += 1);
142 *enc->cur++ = '\\'; 158 *enc->cur++ = '\\';
143 *enc->cur++ = '\\'; 159 *enc->cur++ = '\\';
144 } 160 }
… …
176 } 192 }
177 193
178 if (uch > 0x10FFFFUL) 194 if (uch > 0x10FFFFUL)
179 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); 195 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
180 196
181 if (uch < 0x80 || enc->flags & F_ASCII) 197 if (uch < 0x80 || enc->flags & F_ASCII || (enc->flags & F_LATIN1 && uch > 0xFF))
182 { 198 {
183 if (uch > 0xFFFFUL) 199 if (uch > 0xFFFFUL)
184 { 200 {
185 need (enc, len += 11); 201 need (enc, len += 11);
186 sprintf (enc->cur, "\\u%04x\\u%04x", 202 sprintf (enc->cur, "\\u%04x\\u%04x",
… …
200 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 216 *enc->cur++ = hexdigit [(uch >> 0) & 15];
201 } 217 }
202 218
203 str += clen; 219 str += clen;
204 } 220 }
221 else if (enc->flags & F_LATIN1)
222 {
223 *enc->cur++ = uch;
224 str += clen;
225 }
205 else if (is_utf8) 226 else if (is_utf8)
206 { 227 {
207 need (enc, len += clen); 228 need (enc, len += clen);
208 do 229 do
209 { 230 {
… …
211 } 232 }
212 while (--clen); 233 while (--clen);
213 } 234 }
214 else 235 else
215 { 236 {
216 need (enc, len += UTF8_MAX_LEN - 1); // never more than 11 bytes needed 237 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
217 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 238 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
218 ++str; 239 ++str;
219 } 240 }
220 } 241 }
221 } 242 }
… …
223 244
224 --len; 245 --len;
225 } 246 }
226} 247}
227 248
228static void 249inline void
229encode_indent (enc_t *enc) 250encode_indent (enc_t *enc)
230{ 251{
231 if (enc->flags & F_INDENT) 252 if (enc->flags & F_INDENT)
232 { 253 {
233 int spaces = enc->indent * INDENT_STEP; 254 int spaces = enc->indent * INDENT_STEP;
… …
236 memset (enc->cur, ' ', spaces); 257 memset (enc->cur, ' ', spaces);
237 enc->cur += spaces; 258 enc->cur += spaces;
238 } 259 }
239} 260}
240 261
241static void 262inline void
242encode_space (enc_t *enc) 263encode_space (enc_t *enc)
243{ 264{
244 need (enc, 1); 265 need (enc, 1);
245 encode_ch (enc, ' '); 266 encode_ch (enc, ' ');
246} 267}
247 268
248static void 269inline void
249encode_nl (enc_t *enc) 270encode_nl (enc_t *enc)
250{ 271{
251 if (enc->flags & F_INDENT) 272 if (enc->flags & F_INDENT)
252 { 273 {
253 need (enc, 1); 274 need (enc, 1);
254 encode_ch (enc, '\n'); 275 encode_ch (enc, '\n');
255 } 276 }
256} 277}
257 278
258static void 279inline void
259encode_comma (enc_t *enc) 280encode_comma (enc_t *enc)
260{ 281{
261 encode_ch (enc, ','); 282 encode_ch (enc, ',');
262 283
263 if (enc->flags & F_INDENT) 284 if (enc->flags & F_INDENT)
… …
362 // actually, this is mostly due to the stupid so-called 383 // actually, this is mostly due to the stupid so-called
363 // security workaround added somewhere in 5.8.x. 384 // security workaround added somewhere in 5.8.x.
364 // that randomises hash orderings 385 // that randomises hash orderings
365 if (enc->flags & F_CANONICAL) 386 if (enc->flags & F_CANONICAL)
366 { 387 {
367 HE *he, *hes [count]; // if your compiler dies here, you need to enable C99 mode
368 int fast = 1; 388 int fast = 1;
389 HE *he;
390#if defined(__BORLANDC__) || defined(_MSC_VER)
391 HE **hes = _alloca (count * sizeof (HE));
392#else
393 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
394#endif
369 395
370 i = 0; 396 i = 0;
371 while ((he = hv_iternext (hv))) 397 while ((he = hv_iternext (hv)))
372 { 398 {
373 hes [i++] = he; 399 hes [i++] = he;
… …
408 434
409 encode_nl (enc); 435 encode_nl (enc);
410 } 436 }
411 else 437 else
412 { 438 {
413 SV *sv;
414 HE *he = hv_iternext (hv); 439 HE *he = hv_iternext (hv);
415 440
416 for (;;) 441 for (;;)
417 { 442 {
418 encode_indent (enc); 443 encode_indent (enc);
… …
433 458
434// encode objects, arrays and special \0=false and \1=true values. 459// encode objects, arrays and special \0=false and \1=true values.
435static void 460static void
436encode_rv (enc_t *enc, SV *sv) 461encode_rv (enc_t *enc, SV *sv)
437{ 462{
463 svtype svt;
464
438 SvGETMAGIC (sv); 465 SvGETMAGIC (sv);
439
440 svtype svt = SvTYPE (sv); 466 svt = SvTYPE (sv);
441 467
442 if (svt == SVt_PVHV) 468 if (svt == SVt_PVHV)
443 encode_hv (enc, (HV *)sv); 469 encode_hv (enc, (HV *)sv);
444 else if (svt == SVt_PVAV) 470 else if (svt == SVt_PVAV)
445 encode_av (enc, (AV *)sv); 471 encode_av (enc, (AV *)sv);
… …
477 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 503 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
478 enc->cur += strlen (enc->cur); 504 enc->cur += strlen (enc->cur);
479 } 505 }
480 else if (SvIOKp (sv)) 506 else if (SvIOKp (sv))
481 { 507 {
482 need (enc, 64); 508 // we assume we can always read an IV as a UV
509 if (SvUV (sv) & ~(UV)0x7fff)
510 {
511 need (enc, sizeof (UV) * 3);
483 enc->cur += 512 enc->cur +=
484 SvIsUV(sv) 513 SvIsUV(sv)
485 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 514 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
486 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 515 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
516 }
517 else
518 {
519 // optimise the "small number case"
520 // code will likely be branchless and use only a single multiplication
521 I32 i = SvIV (sv);
522 U32 u;
523
524 need (enc, 6);
525
526 *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
527 u = i < 0 ? -i : i;
528
529 // convert to 4.28 fixed-point representation
530 u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
531
532 char digit, nz = 0;
533
534 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5;
535 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5;
536 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5;
537 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5;
538 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1;
539 }
487 } 540 }
488 else if (SvROK (sv)) 541 else if (SvROK (sv))
489 encode_rv (enc, SvRV (sv)); 542 encode_rv (enc, SvRV (sv));
490 else if (!SvOK (sv)) 543 else if (!SvOK (sv))
491 encode_str (enc, "null", 4, 0); 544 encode_str (enc, "null", 4, 0);
… …
495} 548}
496 549
497static SV * 550static SV *
498encode_json (SV *scalar, U32 flags) 551encode_json (SV *scalar, U32 flags)
499{ 552{
553 enc_t enc;
554
500 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 555 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar))
501 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 556 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
502 557
503 enc_t enc;
504 enc.flags = flags; 558 enc.flags = flags;
505 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 559 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
506 enc.cur = SvPVX (enc.sv); 560 enc.cur = SvPVX (enc.sv);
507 enc.end = SvEND (enc.sv); 561 enc.end = SvEND (enc.sv);
508 enc.indent = 0; 562 enc.indent = 0;
509 enc.maxdepth = DEC_DEPTH (flags); 563 enc.maxdepth = DEC_DEPTH (flags);
510 564
511 SvPOK_only (enc.sv); 565 SvPOK_only (enc.sv);
512 encode_sv (&enc, scalar); 566 encode_sv (&enc, scalar);
513 567
568 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
569 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
570
514 if (!(flags & (F_ASCII | F_UTF8))) 571 if (!(flags & (F_ASCII | F_LATIN1 | F_UTF8)))
515 SvUTF8_on (enc.sv); 572 SvUTF8_on (enc.sv);
516
517 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
518 573
519 if (enc.flags & F_SHRINK) 574 if (enc.flags & F_SHRINK)
520 shrink (enc.sv); 575 shrink (enc.sv);
521 576
522 return enc.sv; 577 return enc.sv;
… …
534 U32 flags; // F_* 589 U32 flags; // F_*
535 U32 depth; // recursion depth 590 U32 depth; // recursion depth
536 U32 maxdepth; // recursion depth limit 591 U32 maxdepth; // recursion depth limit
537} dec_t; 592} dec_t;
538 593
539static void 594inline void
540decode_ws (dec_t *dec) 595decode_ws (dec_t *dec)
541{ 596{
542 for (;;) 597 for (;;)
543 { 598 {
544 char ch = *dec->cur; 599 char ch = *dec->cur;
… …
570decode_4hex (dec_t *dec) 625decode_4hex (dec_t *dec)
571{ 626{
572 signed char d1, d2, d3, d4; 627 signed char d1, d2, d3, d4;
573 unsigned char *cur = (unsigned char *)dec->cur; 628 unsigned char *cur = (unsigned char *)dec->cur;
574 629
575 d1 = decode_hexdigit [cur [0]]; if (d1 < 0) ERR ("four hexadecimal digits expected"); 630 d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("four hexadecimal digits expected");
576 d2 = decode_hexdigit [cur [1]]; if (d2 < 0) ERR ("four hexadecimal digits expected"); 631 d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("four hexadecimal digits expected");
577 d3 = decode_hexdigit [cur [2]]; if (d3 < 0) ERR ("four hexadecimal digits expected"); 632 d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("four hexadecimal digits expected");
578 d4 = decode_hexdigit [cur [3]]; if (d4 < 0) ERR ("four hexadecimal digits expected"); 633 d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("four hexadecimal digits expected");
579 634
580 dec->cur += 4; 635 dec->cur += 4;
581 636
582 return ((UV)d1) << 12 637 return ((UV)d1) << 12
583 | ((UV)d2) << 8 638 | ((UV)d2) << 8
… …
591static SV * 646static SV *
592decode_str (dec_t *dec) 647decode_str (dec_t *dec)
593{ 648{
594 SV *sv = 0; 649 SV *sv = 0;
595 int utf8 = 0; 650 int utf8 = 0;
651 char *dec_cur = dec->cur;
596 652
597 do 653 do
598 { 654 {
599 char buf [SHORT_STRING_LEN + UTF8_MAX_LEN]; 655 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
600 char *cur = buf; 656 char *cur = buf;
601 657
602 do 658 do
603 { 659 {
604 unsigned char ch = *(unsigned char *)dec->cur++; 660 unsigned char ch = *(unsigned char *)dec_cur++;
605 661
606 if (ch == '"') 662 if (expect_false (ch == '"'))
607 { 663 {
608 --dec->cur; 664 --dec_cur;
609 break; 665 break;
610 } 666 }
611 else if (ch == '\\') 667 else if (expect_false (ch == '\\'))
612 { 668 {
613 switch (*dec->cur) 669 switch (*dec_cur)
614 { 670 {
615 case '\\': 671 case '\\':
616 case '/': 672 case '/':
617 case '"': *cur++ = *dec->cur++; break; 673 case '"': *cur++ = *dec_cur++; break;
618 674
619 case 'b': ++dec->cur; *cur++ = '\010'; break; 675 case 'b': ++dec_cur; *cur++ = '\010'; break;
620 case 't': ++dec->cur; *cur++ = '\011'; break; 676 case 't': ++dec_cur; *cur++ = '\011'; break;
621 case 'n': ++dec->cur; *cur++ = '\012'; break; 677 case 'n': ++dec_cur; *cur++ = '\012'; break;
622 case 'f': ++dec->cur; *cur++ = '\014'; break; 678 case 'f': ++dec_cur; *cur++ = '\014'; break;
623 case 'r': ++dec->cur; *cur++ = '\015'; break; 679 case 'r': ++dec_cur; *cur++ = '\015'; break;
624 680
625 case 'u': 681 case 'u':
626 { 682 {
627 UV lo, hi; 683 UV lo, hi;
628 ++dec->cur; 684 ++dec_cur;
629 685
686 dec->cur = dec_cur;
630 hi = decode_4hex (dec); 687 hi = decode_4hex (dec);
688 dec_cur = dec->cur;
631 if (hi == (UV)-1) 689 if (hi == (UV)-1)
632 goto fail; 690 goto fail;
633 691
634 // possibly a surrogate pair 692 // possibly a surrogate pair
635 if (hi >= 0xd800) 693 if (hi >= 0xd800)
636 if (hi < 0xdc00) 694 if (hi < 0xdc00)
637 { 695 {
638 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 696 if (dec_cur [0] != '\\' || dec_cur [1] != 'u')
639 ERR ("missing low surrogate character in surrogate pair"); 697 ERR ("missing low surrogate character in surrogate pair");
640 698
641 dec->cur += 2; 699 dec_cur += 2;
642 700
701 dec->cur = dec_cur;
643 lo = decode_4hex (dec); 702 lo = decode_4hex (dec);
703 dec_cur = dec->cur;
644 if (lo == (UV)-1) 704 if (lo == (UV)-1)
645 goto fail; 705 goto fail;
646 706
647 if (lo < 0xdc00 || lo >= 0xe000) 707 if (lo < 0xdc00 || lo >= 0xe000)
648 ERR ("surrogate pair expected"); 708 ERR ("surrogate pair expected");
… …
662 *cur++ = hi; 722 *cur++ = hi;
663 } 723 }
664 break; 724 break;
665 725
666 default: 726 default:
667 --dec->cur; 727 --dec_cur;
668 ERR ("illegal backslash escape sequence in string"); 728 ERR ("illegal backslash escape sequence in string");
669 } 729 }
670 } 730 }
671 else if (ch >= 0x20 && ch <= 0x7f) 731 else if (expect_true (ch >= 0x20 && ch <= 0x7f))
672 *cur++ = ch; 732 *cur++ = ch;
673 else if (ch >= 0x80) 733 else if (ch >= 0x80)
674 { 734 {
675 --dec->cur;
676
677 STRLEN clen; 735 STRLEN clen;
736 UV uch;
737
738 --dec_cur;
739
678 UV uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen); 740 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
679 if (clen == (STRLEN)-1) 741 if (clen == (STRLEN)-1)
680 ERR ("malformed UTF-8 character in JSON string"); 742 ERR ("malformed UTF-8 character in JSON string");
681 743
682 do 744 do
683 *cur++ = *dec->cur++; 745 *cur++ = *dec_cur++;
684 while (--clen); 746 while (--clen);
685 747
686 utf8 = 1; 748 utf8 = 1;
687 } 749 }
688 else 750 else
689 { 751 {
690 --dec->cur; 752 --dec_cur;
691 753
692 if (!ch) 754 if (!ch)
693 ERR ("unexpected end of string while parsing JSON string"); 755 ERR ("unexpected end of string while parsing JSON string");
694 else 756 else
695 ERR ("invalid character encountered while parsing JSON string"); 757 ERR ("invalid character encountered while parsing JSON string");
696 } 758 }
697 } 759 }
698 while (cur < buf + SHORT_STRING_LEN); 760 while (cur < buf + SHORT_STRING_LEN);
699 761
762 {
700 STRLEN len = cur - buf; 763 STRLEN len = cur - buf;
701 764
702 if (sv) 765 if (sv)
703 { 766 {
704 SvGROW (sv, SvCUR (sv) + len + 1); 767 SvGROW (sv, SvCUR (sv) + len + 1);
705 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 768 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
706 SvCUR_set (sv, SvCUR (sv) + len); 769 SvCUR_set (sv, SvCUR (sv) + len);
707 } 770 }
708 else 771 else
709 sv = newSVpvn (buf, len); 772 sv = newSVpvn (buf, len);
710 } 773 }
774 }
711 while (*dec->cur != '"'); 775 while (*dec_cur != '"');
712 776
713 ++dec->cur; 777 ++dec_cur;
714 778
715 if (sv) 779 if (sv)
716 { 780 {
717 SvPOK_only (sv); 781 SvPOK_only (sv);
718 *SvEND (sv) = 0; 782 *SvEND (sv) = 0;
… …
721 SvUTF8_on (sv); 785 SvUTF8_on (sv);
722 } 786 }
723 else 787 else
724 sv = newSVpvn ("", 0); 788 sv = newSVpvn ("", 0);
725 789
790 dec->cur = dec_cur;
726 return sv; 791 return sv;
727 792
728fail: 793fail:
794 dec->cur = dec_cur;
729 return 0; 795 return 0;
730} 796}
731 797
732static SV * 798static SV *
733decode_num (dec_t *dec) 799decode_num (dec_t *dec)
… …
791 is_nv = 1; 857 is_nv = 1;
792 } 858 }
793 859
794 if (!is_nv) 860 if (!is_nv)
795 { 861 {
796 UV uv; 862 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so
797 int numtype = grok_number (start, dec->cur - start, &uv); 863 if (*start == '-')
798 if (numtype & IS_NUMBER_IN_UV) 864 switch (dec->cur - start)
799 if (numtype & IS_NUMBER_NEG)
800 { 865 {
801 if (uv < (UV)IV_MIN) 866 case 2: return newSViv (-( start [1] - '0' ));
802 return newSViv (-(IV)uv); 867 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
868 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
869 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
803 } 870 }
871 else
872 switch (dec->cur - start)
873 {
874 case 1: return newSViv ( start [0] - '0' );
875 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
876 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
877 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
878 }
879
880 {
881 UV uv;
882 int numtype = grok_number (start, dec->cur - start, &uv);
883 if (numtype & IS_NUMBER_IN_UV)
884 if (numtype & IS_NUMBER_NEG)
885 {
886 if (uv < (UV)IV_MIN)
887 return newSViv (-(IV)uv);
888 }
804 else 889 else
805 return newSVuv (uv); 890 return newSVuv (uv);
891 }
806 } 892 }
807 893
808 return newSVnv (Atof (start)); 894 return newSVnv (Atof (start));
809 895
810fail: 896fail:
… …
967fail: 1053fail:
968 return 0; 1054 return 0;
969} 1055}
970 1056
971static SV * 1057static SV *
972decode_json (SV *string, U32 flags) 1058decode_json (SV *string, U32 flags, UV *offset_return)
973{ 1059{
1060 dec_t dec;
1061 UV offset;
974 SV *sv; 1062 SV *sv;
975 1063
1064 SvGETMAGIC (string);
976 SvUPGRADE (string, SVt_PV); 1065 SvUPGRADE (string, SVt_PV);
977 1066
978 if (flags & F_UTF8) 1067 if (flags & F_UTF8)
979 sv_utf8_downgrade (string, 0); 1068 sv_utf8_downgrade (string, 0);
980 else 1069 else
981 sv_utf8_upgrade (string); 1070 sv_utf8_upgrade (string);
982 1071
983 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1072 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
984 1073
985 dec_t dec;
986 dec.flags = flags; 1074 dec.flags = flags;
987 dec.cur = SvPVX (string); 1075 dec.cur = SvPVX (string);
988 dec.end = SvEND (string); 1076 dec.end = SvEND (string);
989 dec.err = 0; 1077 dec.err = 0;
990 dec.depth = 0; 1078 dec.depth = 0;
991 dec.maxdepth = DEC_DEPTH (dec.flags); 1079 dec.maxdepth = DEC_DEPTH (dec.flags);
992 1080
993 *dec.end = 0; // this should basically be a nop, too, but make sure its there 1081 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
994 sv = decode_sv (&dec); 1082 sv = decode_sv (&dec);
995 1083
1084 if (!(offset_return || !sv))
1085 {
1086 // check for trailing garbage
1087 decode_ws (&dec);
1088
1089 if (*dec.cur)
1090 {
1091 dec.err = "garbage after JSON object";
1092 SvREFCNT_dec (sv);
1093 sv = 0;
1094 }
1095 }
1096
1097 if (offset_return || !sv)
1098 {
1099 offset = dec.flags & F_UTF8
1100 ? dec.cur - SvPVX (string)
1101 : utf8_distance (dec.cur, SvPVX (string));
1102
1103 if (offset_return)
1104 *offset_return = offset;
1105 }
1106
996 if (!sv) 1107 if (!sv)
997 { 1108 {
998 IV offset = dec.flags & F_UTF8
999 ? dec.cur - SvPVX (string)
1000 : utf8_distance (dec.cur, SvPVX (string));
1001 SV *uni = sv_newmortal (); 1109 SV *uni = sv_newmortal ();
1002 1110
1003 // horrible hack to silence warning inside pv_uni_display 1111 // horrible hack to silence warning inside pv_uni_display
1004 COP cop = *PL_curcop; 1112 COP cop = *PL_curcop;
1005 cop.cop_warnings = pWARN_NONE; 1113 cop.cop_warnings = pWARN_NONE;
… …
1053 RETVAL 1161 RETVAL
1054 1162
1055SV *ascii (SV *self, int enable = 1) 1163SV *ascii (SV *self, int enable = 1)
1056 ALIAS: 1164 ALIAS:
1057 ascii = F_ASCII 1165 ascii = F_ASCII
1166 latin1 = F_LATIN1
1058 utf8 = F_UTF8 1167 utf8 = F_UTF8
1059 indent = F_INDENT 1168 indent = F_INDENT
1060 canonical = F_CANONICAL 1169 canonical = F_CANONICAL
1061 space_before = F_SPACE_BEFORE 1170 space_before = F_SPACE_BEFORE
1062 space_after = F_SPACE_AFTER 1171 space_after = F_SPACE_AFTER
… …
1074 RETVAL = newSVsv (self); 1183 RETVAL = newSVsv (self);
1075} 1184}
1076 OUTPUT: 1185 OUTPUT:
1077 RETVAL 1186 RETVAL
1078 1187
1079SV *max_depth (SV *self, int max_depth = 0x80000000UL) 1188SV *max_depth (SV *self, UV max_depth = 0x80000000UL)
1080 CODE: 1189 CODE:
1081{ 1190{
1082 UV *uv = SvJSON (self); 1191 UV *uv = SvJSON (self);
1083 UV log2 = 0; 1192 UV log2 = 0;
1084 1193
… …
1098 PPCODE: 1207 PPCODE:
1099 XPUSHs (encode_json (scalar, *SvJSON (self))); 1208 XPUSHs (encode_json (scalar, *SvJSON (self)));
1100 1209
1101void decode (SV *self, SV *jsonstr) 1210void decode (SV *self, SV *jsonstr)
1102 PPCODE: 1211 PPCODE:
1103 XPUSHs (decode_json (jsonstr, *SvJSON (self))); 1212 XPUSHs (decode_json (jsonstr, *SvJSON (self), 0));
1213
1214void decode_prefix (SV *self, SV *jsonstr)
1215 PPCODE:
1216{
1217 UV offset;
1218 EXTEND (SP, 2);
1219 PUSHs (decode_json (jsonstr, *SvJSON (self), &offset));
1220 PUSHs (sv_2mortal (newSVuv (offset)));
1221}
1104 1222
1105PROTOTYPES: ENABLE 1223PROTOTYPES: ENABLE
1106 1224
1107void to_json (SV *scalar) 1225void to_json (SV *scalar)
1108 ALIAS: 1226 ALIAS:
… …
1112 1230
1113void from_json (SV *jsonstr) 1231void from_json (SV *jsonstr)
1114 ALIAS: 1232 ALIAS:
1115 jsonToObj = 0 1233 jsonToObj = 0
1116 PPCODE: 1234 PPCODE:
1117 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8)); 1235 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8, 0));
1118 1236

Diff Legend

– Removed lines
+ Added lines
< Changed lines
> Changed lines