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 1.10 by root, Sat Mar 24 01:15:22 2007 UTC vs.
Revision 1.20 by root, Sun Mar 25 21:53:19 2007 UTC

… …
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 8
9#define F_ASCII 0x00000001 9#define F_ASCII 0x00000001UL
10#define F_UTF8 0x00000002 10#define F_UTF8 0x00000002UL
11#define F_INDENT 0x00000004 11#define F_INDENT 0x00000004UL
12#define F_CANONICAL 0x00000008 12#define F_CANONICAL 0x00000008UL
13#define F_SPACE_BEFORE 0x00000010 13#define F_SPACE_BEFORE 0x00000010UL
14#define F_SPACE_AFTER 0x00000020 14#define F_SPACE_AFTER 0x00000020UL
15#define F_ALLOW_NONREF 0x00000080 15#define F_ALLOW_NONREF 0x00000080UL
16#define F_SHRINK 0x00000100 16#define F_SHRINK 0x00000100UL
17#define F_MAXDEPTH 0xf8000000UL
18#define S_MAXDEPTH 27
19
20#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
21
22// F_SELFCONVERT? <=> to_json/toJson
23// F_BLESSED? <=> { $__class__$ => }
17 24
18#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 25#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
19#define F_DEFAULT 0 26#define F_DEFAULT (13UL << S_MAXDEPTH)
20 27
21#define INIT_SIZE 32 // initial scalar size to be allocated 28#define INIT_SIZE 32 // initial scalar size to be allocated
29#define INDENT_STEP 3 // spaces per indentation level
30
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
22 33
23#define SB do { 34#define SB do {
24#define SE } while (0) 35#define SE } while (0)
25 36
26static HV *json_stash; 37static HV *json_stash; // JSON::XS::
38
39/////////////////////////////////////////////////////////////////////////////
40// utility functions
41
42static UV *
43SvJSON (SV *sv)
44{
45 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash))
46 croak ("object is not of type JSON::XS");
47
48 return &SvUVX (SvRV (sv));
49}
50
51static void
52shrink (SV *sv)
53{
54 sv_utf8_downgrade (sv, 1);
55 if (SvLEN (sv) > SvCUR (sv) + 1)
56 {
57#ifdef SvPV_shrink_to_cur
58 SvPV_shrink_to_cur (sv);
59#elif defined (SvPV_renew)
60 SvPV_renew (sv, SvCUR (sv) + 1);
61#endif
62 }
63}
64
65// decode an utf-8 character and return it, or (UV)-1 in
66// case of an error.
67// we special-case "safe" characters from U+80 .. U+7FF,
68// but use the very good perl function to parse anything else.
69// note that we never call this function for a ascii codepoints
70static UV
71decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
72{
73 if (s[0] > 0xdf || s[0] < 0xc2)
74 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
75 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)
76 {
77 *clen = 2;
78 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
79 }
80 else
81 return (UV)-1;
82}
83
84/////////////////////////////////////////////////////////////////////////////
85// encoder
27 86
28// structure used for encoding JSON 87// structure used for encoding JSON
29typedef struct 88typedef struct
30{ 89{
31 char *cur; 90 char *cur; // SvPVX (sv) + current output position
32 STRLEN len; // SvLEN (sv)
33 char *end; // SvEND (sv) 91 char *end; // SvEND (sv)
34 SV *sv; 92 SV *sv; // result scalar
35 UV flags; 93 U32 flags; // F_*
36 int max_recurse; 94 U32 indent; // indentation level
37 int indent; 95 U32 maxdepth; // max. indentation/recursion level
38} enc_t; 96} enc_t;
39
40// structure used for decoding JSON
41typedef struct
42{
43 char *cur;
44 char *end;
45 const char *err;
46 UV flags;
47} dec_t;
48
49static UV *
50SvJSON (SV *sv)
51{
52 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash))
53 croak ("object is not of type JSON::XS");
54
55 return &SvUVX (SvRV (sv));
56}
57
58static void
59shrink (SV *sv)
60{
61 sv_utf8_downgrade (sv, 1);
62#ifdef SvPV_shrink_to_cur
63 SvPV_shrink_to_cur (sv);
64#endif
65}
66
67/////////////////////////////////////////////////////////////////////////////
68 97
69static void 98static void
70need (enc_t *enc, STRLEN len) 99need (enc_t *enc, STRLEN len)
71{ 100{
72 if (enc->cur + len >= enc->end) 101 if (enc->cur + len >= enc->end)
… …
130 STRLEN clen; 159 STRLEN clen;
131 UV uch; 160 UV uch;
132 161
133 if (is_utf8) 162 if (is_utf8)
134 { 163 {
135 uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY); 164 //uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
165 uch = decode_utf8 (str, end - str, &clen);
136 if (clen == (STRLEN)-1) 166 if (clen == (STRLEN)-1)
137 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str); 167 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
138 } 168 }
139 else 169 else
140 { 170 {
… …
178 } 208 }
179 while (--clen); 209 while (--clen);
180 } 210 }
181 else 211 else
182 { 212 {
183 need (enc, len += 10); // never more than 11 bytes needed 213 need (enc, len += UTF8_MAX_LEN - 1); // never more than 11 bytes needed
184 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 214 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
185 ++str; 215 ++str;
186 } 216 }
187 } 217 }
188 } 218 }
… …
190 220
191 --len; 221 --len;
192 } 222 }
193} 223}
194 224
195#define INDENT SB \ 225static void
226encode_indent (enc_t *enc)
227{
196 if (enc->flags & F_INDENT) \ 228 if (enc->flags & F_INDENT)
197 { \ 229 {
198 int i_; \ 230 int spaces = enc->indent * INDENT_STEP;
199 need (enc, enc->indent); \
200 for (i_ = enc->indent * 3; i_--; )\
201 encode_ch (enc, ' '); \
202 } \
203 SE
204 231
205#define SPACE SB need (enc, 1); encode_ch (enc, ' '); SE 232 need (enc, spaces);
206#define NL SB if (enc->flags & F_INDENT) { need (enc, 1); encode_ch (enc, '\n'); } SE 233 memset (enc->cur, ' ', spaces);
207#define COMMA SB \ 234 enc->cur += spaces;
235 }
236}
237
238static void
239encode_space (enc_t *enc)
240{
241 need (enc, 1);
208 encode_ch (enc, ','); \ 242 encode_ch (enc, ' ');
243}
244
245static void
246encode_nl (enc_t *enc)
247{
209 if (enc->flags & F_INDENT) \ 248 if (enc->flags & F_INDENT)
210 NL; \ 249 {
250 need (enc, 1);
251 encode_ch (enc, '\n');
252 }
253}
254
255static void
256encode_comma (enc_t *enc)
257{
258 encode_ch (enc, ',');
259
260 if (enc->flags & F_INDENT)
261 encode_nl (enc);
211 else if (enc->flags & F_SPACE_AFTER) \ 262 else if (enc->flags & F_SPACE_AFTER)
212 SPACE; \ 263 encode_space (enc);
213 SE 264}
214 265
215static void encode_sv (enc_t *enc, SV *sv); 266static void encode_sv (enc_t *enc, SV *sv);
216 267
217static void 268static void
218encode_av (enc_t *enc, AV *av) 269encode_av (enc_t *enc, AV *av)
219{ 270{
220 int i, len = av_len (av); 271 int i, len = av_len (av);
221 272
222 encode_ch (enc, '['); NL; 273 encode_ch (enc, '['); encode_nl (enc);
223 ++enc->indent; 274 ++enc->indent;
224 275
225 for (i = 0; i <= len; ++i) 276 for (i = 0; i <= len; ++i)
226 { 277 {
227 INDENT; 278 encode_indent (enc);
228 encode_sv (enc, *av_fetch (av, i, 0)); 279 encode_sv (enc, *av_fetch (av, i, 0));
229 280
230 if (i < len) 281 if (i < len)
231 COMMA; 282 encode_comma (enc);
232 } 283 }
233 284
234 NL; 285 encode_nl (enc);
235 286
236 --enc->indent; 287 --enc->indent;
237 INDENT; encode_ch (enc, ']'); 288 encode_indent (enc); encode_ch (enc, ']');
238} 289}
239 290
240static void 291static void
241encode_he (enc_t *enc, HE *he) 292encode_he (enc_t *enc, HE *he)
242{ 293{
… …
256 else 307 else
257 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 308 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
258 309
259 encode_ch (enc, '"'); 310 encode_ch (enc, '"');
260 311
261 if (enc->flags & F_SPACE_BEFORE) SPACE; 312 if (enc->flags & F_SPACE_BEFORE) encode_space (enc);
262 encode_ch (enc, ':'); 313 encode_ch (enc, ':');
263 if (enc->flags & F_SPACE_AFTER ) SPACE; 314 if (enc->flags & F_SPACE_AFTER ) encode_space (enc);
264 encode_sv (enc, HeVAL (he)); 315 encode_sv (enc, HeVAL (he));
265} 316}
266 317
267// compare hash entries, used when all keys are bytestrings 318// compare hash entries, used when all keys are bytestrings
268static int 319static int
… …
274 HE *b = *(HE **)b_; 325 HE *b = *(HE **)b_;
275 326
276 STRLEN la = HeKLEN (a); 327 STRLEN la = HeKLEN (a);
277 STRLEN lb = HeKLEN (b); 328 STRLEN lb = HeKLEN (b);
278 329
279 if (!(cmp == memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 330 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb)))
280 cmp = la < lb ? -1 : la == lb ? 0 : 1; 331 cmp = la - lb;
281 332
282 return cmp; 333 return cmp;
283} 334}
284 335
285// compare hash entries, used when some keys are sv's or utf-x 336// compare hash entries, used when some keys are sv's or utf-x
… …
292static void 343static void
293encode_hv (enc_t *enc, HV *hv) 344encode_hv (enc_t *enc, HV *hv)
294{ 345{
295 int count, i; 346 int count, i;
296 347
297 encode_ch (enc, '{'); NL; ++enc->indent; 348 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
298 349
299 if ((count = hv_iterinit (hv))) 350 if ((count = hv_iterinit (hv)))
300 { 351 {
301 // for canonical output we have to sort by keys first 352 // for canonical output we have to sort by keys first
302 // actually, this is mostly due to the stupid so-called 353 // actually, this is mostly due to the stupid so-called
303 // security workaround added somewhere in 5.8.x. 354 // security workaround added somewhere in 5.8.x.
304 // that randomises hash orderings 355 // that randomises hash orderings
305 if (enc->flags & F_CANONICAL) 356 if (enc->flags & F_CANONICAL)
306 { 357 {
307 HE *he, *hes [count]; 358 HE *he, *hes [count]; // if your compiler dies here, you need to enable C99 mode
308 int fast = 1; 359 int fast = 1;
309 360
310 i = 0; 361 i = 0;
311 while ((he = hv_iternext (hv))) 362 while ((he = hv_iternext (hv)))
312 { 363 {
… …
337 LEAVE; 388 LEAVE;
338 } 389 }
339 390
340 for (i = 0; i < count; ++i) 391 for (i = 0; i < count; ++i)
341 { 392 {
342 INDENT; 393 encode_indent (enc);
343 encode_he (enc, hes [i]); 394 encode_he (enc, hes [i]);
344 395
345 if (i < count - 1) 396 if (i < count - 1)
346 COMMA; 397 encode_comma (enc);
347 } 398 }
348 399
349 NL; 400 encode_nl (enc);
350 } 401 }
351 else 402 else
352 { 403 {
353 SV *sv; 404 SV *sv;
354 HE *he = hv_iternext (hv); 405 HE *he = hv_iternext (hv);
355 406
356 for (;;) 407 for (;;)
357 { 408 {
358 INDENT; 409 encode_indent (enc);
359 encode_he (enc, he); 410 encode_he (enc, he);
360 411
361 if (!(he = hv_iternext (hv))) 412 if (!(he = hv_iternext (hv)))
362 break; 413 break;
363 414
364 COMMA; 415 encode_comma (enc);
365 } 416 }
366 417
367 NL; 418 encode_nl (enc);
368 } 419 }
369 } 420 }
370 421
371 --enc->indent; INDENT; encode_ch (enc, '}'); 422 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
372} 423}
373 424
374static void 425static void
375encode_sv (enc_t *enc, SV *sv) 426encode_sv (enc_t *enc, SV *sv)
376{ 427{
… …
400 } 451 }
401 else if (SvROK (sv)) 452 else if (SvROK (sv))
402 { 453 {
403 SV *rv = SvRV (sv); 454 SV *rv = SvRV (sv);
404 455
405 if (!--enc->max_recurse) 456 if (enc->indent >= enc->maxdepth)
406 croak ("data structure too deep (hit recursion limit)"); 457 croak ("data structure too deep (hit recursion limit)");
407 458
408 switch (SvTYPE (rv)) 459 switch (SvTYPE (rv))
409 { 460 {
410 case SVt_PVAV: encode_av (enc, (AV *)rv); break; 461 case SVt_PVAV: encode_av (enc, (AV *)rv); break;
… …
421 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 472 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
422 SvPV_nolen (sv), SvFLAGS (sv)); 473 SvPV_nolen (sv), SvFLAGS (sv));
423} 474}
424 475
425static SV * 476static SV *
426encode_json (SV *scalar, UV flags) 477encode_json (SV *scalar, U32 flags)
427{ 478{
428 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 479 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar))
429 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 480 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
430 481
431 enc_t enc; 482 enc_t enc;
432 enc.flags = flags; 483 enc.flags = flags;
433 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 484 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
434 enc.cur = SvPVX (enc.sv); 485 enc.cur = SvPVX (enc.sv);
435 enc.end = SvEND (enc.sv); 486 enc.end = SvEND (enc.sv);
436 enc.max_recurse = 0;
437 enc.indent = 0; 487 enc.indent = 0;
488 enc.maxdepth = DEC_DEPTH (flags);
438 489
439 SvPOK_only (enc.sv); 490 SvPOK_only (enc.sv);
440 encode_sv (&enc, scalar); 491 encode_sv (&enc, scalar);
441 492
442 if (!(flags & (F_ASCII | F_UTF8))) 493 if (!(flags & (F_ASCII | F_UTF8)))
… …
449 500
450 return enc.sv; 501 return enc.sv;
451} 502}
452 503
453///////////////////////////////////////////////////////////////////////////// 504/////////////////////////////////////////////////////////////////////////////
505// decoder
454 506
455#define WS \ 507// structure used for decoding JSON
508typedef struct
509{
510 char *cur; // current parser pointer
511 char *end; // end of input string
512 const char *err; // parse error, if != 0
513 U32 flags; // F_*
514 U32 depth; // recursion depth
515 U32 maxdepth; // recursion depth limit
516} dec_t;
517
518static void
519decode_ws (dec_t *dec)
520{
456 for (;;) \ 521 for (;;)
457 { \ 522 {
458 char ch = *dec->cur; \ 523 char ch = *dec->cur;
524
459 if (ch > 0x20 \ 525 if (ch > 0x20
460 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) \ 526 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09))
461 break; \ 527 break;
528
462 ++dec->cur; \ 529 ++dec->cur;
463 } 530 }
531}
464 532
465#define ERR(reason) SB dec->err = reason; goto fail; SE 533#define ERR(reason) SB dec->err = reason; goto fail; SE
534
466#define EXPECT_CH(ch) SB \ 535#define EXPECT_CH(ch) SB \
467 if (*dec->cur != ch) \ 536 if (*dec->cur != ch) \
468 ERR (# ch " expected"); \ 537 ERR (# ch " expected"); \
469 ++dec->cur; \ 538 ++dec->cur; \
470 SE 539 SE
471 540
541#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)")
542#define DEC_DEC_DEPTH --dec->depth
543
472static SV *decode_sv (dec_t *dec); 544static SV *decode_sv (dec_t *dec);
473 545
474static signed char decode_hexdigit[256]; 546static signed char decode_hexdigit[256];
475 547
476static UV 548static UV
477decode_4hex (dec_t *dec) 549decode_4hex (dec_t *dec)
478{ 550{
479 signed char d1, d2, d3, d4; 551 signed char d1, d2, d3, d4;
552 unsigned char *cur = (unsigned char *)dec->cur;
480 553
481 d1 = decode_hexdigit [((unsigned char *)dec->cur) [0]];
482 if (d1 < 0) ERR ("four hexadecimal digits expected"); 554 d1 = decode_hexdigit [cur [0]]; if (d1 < 0) ERR ("four hexadecimal digits expected");
483 d2 = decode_hexdigit [((unsigned char *)dec->cur) [1]];
484 if (d2 < 0) ERR ("four hexadecimal digits expected"); 555 d2 = decode_hexdigit [cur [1]]; if (d2 < 0) ERR ("four hexadecimal digits expected");
485 d3 = decode_hexdigit [((unsigned char *)dec->cur) [2]];
486 if (d3 < 0) ERR ("four hexadecimal digits expected"); 556 d3 = decode_hexdigit [cur [2]]; if (d3 < 0) ERR ("four hexadecimal digits expected");
487 d4 = decode_hexdigit [((unsigned char *)dec->cur) [3]];
488 if (d4 < 0) ERR ("four hexadecimal digits expected"); 557 d4 = decode_hexdigit [cur [3]]; if (d4 < 0) ERR ("four hexadecimal digits expected");
489 558
490 dec->cur += 4; 559 dec->cur += 4;
491 560
492 return ((UV)d1) << 12 561 return ((UV)d1) << 12
493 | ((UV)d2) << 8 562 | ((UV)d2) << 8
… …
496 565
497fail: 566fail:
498 return (UV)-1; 567 return (UV)-1;
499} 568}
500 569
501#define APPEND_GROW(n) SB \
502 if (cur + (n) >= end) \
503 { \
504 STRLEN ofs = cur - SvPVX (sv); \
505 SvGROW (sv, ofs + (n) + 1); \
506 cur = SvPVX (sv) + ofs; \
507 end = SvEND (sv); \
508 } \
509 SE
510
511#define APPEND_CH(ch) SB \
512 APPEND_GROW (1); \
513 *cur++ = (ch); \
514 SE
515
516static SV * 570static SV *
517decode_str (dec_t *dec) 571decode_str (dec_t *dec)
518{ 572{
519 SV *sv = NEWSV (0,2); 573 SV *sv = 0;
520 int utf8 = 0; 574 int utf8 = 0;
521 char *cur = SvPVX (sv);
522 char *end = SvEND (sv);
523 575
524 for (;;) 576 do
525 { 577 {
526 unsigned char ch = *(unsigned char *)dec->cur; 578 char buf [SHORT_STRING_LEN + UTF8_MAX_LEN];
579 char *cur = buf;
527 580
528 if (ch == '"') 581 do
529 break;
530 else if (ch == '\\')
531 { 582 {
532 switch (*++dec->cur) 583 unsigned char ch = *(unsigned char *)dec->cur++;
584
585 if (ch == '"')
533 { 586 {
534 case '\\': 587 --dec->cur;
535 case '/': 588 break;
536 case '"': APPEND_CH (*dec->cur++); break; 589 }
537 590 else if (ch == '\\')
538 case 'b': APPEND_CH ('\010'); ++dec->cur; break; 591 {
539 case 't': APPEND_CH ('\011'); ++dec->cur; break; 592 switch (*dec->cur)
540 case 'n': APPEND_CH ('\012'); ++dec->cur; break;
541 case 'f': APPEND_CH ('\014'); ++dec->cur; break;
542 case 'r': APPEND_CH ('\015'); ++dec->cur; break;
543
544 case 'u':
545 { 593 {
546 UV lo, hi; 594 case '\\':
547 ++dec->cur; 595 case '/':
596 case '"': *cur++ = *dec->cur++; break;
548 597
549 hi = decode_4hex (dec); 598 case 'b': ++dec->cur; *cur++ = '\010'; break;
550 if (hi == (UV)-1) 599 case 't': ++dec->cur; *cur++ = '\011'; break;
551 goto fail; 600 case 'n': ++dec->cur; *cur++ = '\012'; break;
601 case 'f': ++dec->cur; *cur++ = '\014'; break;
602 case 'r': ++dec->cur; *cur++ = '\015'; break;
552 603
553 // possibly a surrogate pair 604 case 'u':
554 if (hi >= 0xd800 && hi < 0xdc00)
555 { 605 {
556 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 606 UV lo, hi;
557 ERR ("missing low surrogate character in surrogate pair");
558
559 dec->cur += 2; 607 ++dec->cur;
560 608
561 lo = decode_4hex (dec); 609 hi = decode_4hex (dec);
562 if (lo == (UV)-1) 610 if (hi == (UV)-1)
563 goto fail; 611 goto fail;
564 612
613 // possibly a surrogate pair
614 if (hi >= 0xd800)
615 if (hi < 0xdc00)
616 {
617 if (dec->cur [0] != '\\' || dec->cur [1] != 'u')
618 ERR ("missing low surrogate character in surrogate pair");
619
620 dec->cur += 2;
621
622 lo = decode_4hex (dec);
623 if (lo == (UV)-1)
624 goto fail;
625
565 if (lo < 0xdc00 || lo >= 0xe000) 626 if (lo < 0xdc00 || lo >= 0xe000)
566 ERR ("surrogate pair expected"); 627 ERR ("surrogate pair expected");
567 628
568 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000; 629 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000;
630 }
631 else if (hi < 0xe000)
632 ERR ("missing high surrogate character in surrogate pair");
633
634 if (hi >= 0x80)
635 {
636 utf8 = 1;
637
638 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
639 }
640 else
641 *cur++ = hi;
569 } 642 }
570 else if (hi >= 0xdc00 && hi < 0xe000)
571 ERR ("missing high surrogate character in surrogate pair");
572
573 if (hi >= 0x80)
574 { 643 break;
575 utf8 = 1;
576 644
577 APPEND_GROW (4); // at most 4 bytes for 21 bits
578 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
579 }
580 else 645 default:
581 APPEND_CH (hi); 646 --dec->cur;
647 ERR ("illegal backslash escape sequence in string");
582 } 648 }
583 break;
584
585 default:
586 --dec->cur;
587 ERR ("illegal backslash escape sequence in string");
588 } 649 }
650 else if (ch >= 0x20 && ch <= 0x7f)
651 *cur++ = ch;
652 else if (ch >= 0x80)
653 {
654 --dec->cur;
655
656 STRLEN clen;
657 UV uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen);
658 if (clen == (STRLEN)-1)
659 ERR ("malformed UTF-8 character in JSON string");
660
661 do
662 {
663 *cur++ = *dec->cur++;
664 }
665 while (--clen);
666
667 utf8 = 1;
668 }
669 else if (!ch)
670 ERR ("unexpected end of string while parsing json string");
671 else
672 ERR ("invalid character encountered");
673
589 } 674 }
590 else if (ch >= 0x20 && ch <= 0x7f) 675 while (cur < buf + SHORT_STRING_LEN);
591 APPEND_CH (*dec->cur++); 676
592 else if (ch >= 0x80) 677 STRLEN len = cur - buf;
678
679 if (sv)
593 { 680 {
594 STRLEN clen; 681 SvGROW (sv, SvCUR (sv) + len + 1);
595 UV uch = utf8n_to_uvuni (dec->cur, dec->end - dec->cur, &clen, UTF8_CHECK_ONLY); 682 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
596 if (clen == (STRLEN)-1) 683 SvCUR_set (sv, SvCUR (sv) + len);
597 ERR ("malformed UTF-8 character in JSON string");
598
599 APPEND_GROW (clen);
600 do
601 {
602 *cur++ = *dec->cur++;
603 }
604 while (--clen);
605
606 utf8 = 1;
607 } 684 }
608 else if (dec->cur == dec->end)
609 ERR ("unexpected end of string while parsing json string");
610 else 685 else
611 ERR ("invalid character encountered"); 686 sv = newSVpvn (buf, len);
612 } 687 }
688 while (*dec->cur != '"');
613 689
614 ++dec->cur; 690 ++dec->cur;
615 691
616 SvCUR_set (sv, cur - SvPVX (sv)); 692 if (sv)
617 693 {
618 SvPOK_only (sv); 694 SvPOK_only (sv);
619 *SvEND (sv) = 0; 695 *SvEND (sv) = 0;
620 696
621 if (utf8) 697 if (utf8)
622 SvUTF8_on (sv); 698 SvUTF8_on (sv);
623 699 }
624 if (dec->flags & F_SHRINK) 700 else
625 shrink (sv); 701 sv = newSVpvn ("", 0);
626 702
627 return sv; 703 return sv;
628 704
629fail: 705fail:
630 SvREFCNT_dec (sv);
631 return 0; 706 return 0;
632} 707}
633 708
634static SV * 709static SV *
635decode_num (dec_t *dec) 710decode_num (dec_t *dec)
… …
716static SV * 791static SV *
717decode_av (dec_t *dec) 792decode_av (dec_t *dec)
718{ 793{
719 AV *av = newAV (); 794 AV *av = newAV ();
720 795
721 WS; 796 DEC_INC_DEPTH;
797 decode_ws (dec);
798
722 if (*dec->cur == ']') 799 if (*dec->cur == ']')
723 ++dec->cur; 800 ++dec->cur;
724 else 801 else
725 for (;;) 802 for (;;)
726 { 803 {
… …
730 if (!value) 807 if (!value)
731 goto fail; 808 goto fail;
732 809
733 av_push (av, value); 810 av_push (av, value);
734 811
735 WS; 812 decode_ws (dec);
736 813
737 if (*dec->cur == ']') 814 if (*dec->cur == ']')
738 { 815 {
739 ++dec->cur; 816 ++dec->cur;
740 break; 817 break;
… …
744 ERR (", or ] expected while parsing array"); 821 ERR (", or ] expected while parsing array");
745 822
746 ++dec->cur; 823 ++dec->cur;
747 } 824 }
748 825
826 DEC_DEC_DEPTH;
749 return newRV_noinc ((SV *)av); 827 return newRV_noinc ((SV *)av);
750 828
751fail: 829fail:
752 SvREFCNT_dec (av); 830 SvREFCNT_dec (av);
831 DEC_DEC_DEPTH;
753 return 0; 832 return 0;
754} 833}
755 834
756static SV * 835static SV *
757decode_hv (dec_t *dec) 836decode_hv (dec_t *dec)
758{ 837{
759 HV *hv = newHV (); 838 HV *hv = newHV ();
760 839
761 WS; 840 DEC_INC_DEPTH;
841 decode_ws (dec);
842
762 if (*dec->cur == '}') 843 if (*dec->cur == '}')
763 ++dec->cur; 844 ++dec->cur;
764 else 845 else
765 for (;;) 846 for (;;)
766 { 847 {
767 SV *key, *value; 848 SV *key, *value;
768 849
769 WS; EXPECT_CH ('"'); 850 decode_ws (dec); EXPECT_CH ('"');
770 851
771 key = decode_str (dec); 852 key = decode_str (dec);
772 if (!key) 853 if (!key)
773 goto fail; 854 goto fail;
774 855
775 WS; EXPECT_CH (':'); 856 decode_ws (dec); EXPECT_CH (':');
776 857
777 value = decode_sv (dec); 858 value = decode_sv (dec);
778 if (!value) 859 if (!value)
779 { 860 {
780 SvREFCNT_dec (key); 861 SvREFCNT_dec (key);
781 goto fail; 862 goto fail;
782 } 863 }
783 864
784 //TODO: optimise
785 hv_store_ent (hv, key, value, 0); 865 hv_store_ent (hv, key, value, 0);
866 SvREFCNT_dec (key);
786 867
787 WS; 868 decode_ws (dec);
788 869
789 if (*dec->cur == '}') 870 if (*dec->cur == '}')
790 { 871 {
791 ++dec->cur; 872 ++dec->cur;
792 break; 873 break;
… …
796 ERR (", or } expected while parsing object/hash"); 877 ERR (", or } expected while parsing object/hash");
797 878
798 ++dec->cur; 879 ++dec->cur;
799 } 880 }
800 881
882 DEC_DEC_DEPTH;
801 return newRV_noinc ((SV *)hv); 883 return newRV_noinc ((SV *)hv);
802 884
803fail: 885fail:
804 SvREFCNT_dec (hv); 886 SvREFCNT_dec (hv);
887 DEC_DEC_DEPTH;
805 return 0; 888 return 0;
806} 889}
807 890
808static SV * 891static SV *
809decode_sv (dec_t *dec) 892decode_sv (dec_t *dec)
810{ 893{
811 WS; 894 decode_ws (dec);
812 switch (*dec->cur) 895 switch (*dec->cur)
813 { 896 {
814 case '"': ++dec->cur; return decode_str (dec); 897 case '"': ++dec->cur; return decode_str (dec);
815 case '[': ++dec->cur; return decode_av (dec); 898 case '[': ++dec->cur; return decode_av (dec);
816 case '{': ++dec->cur; return decode_hv (dec); 899 case '{': ++dec->cur; return decode_hv (dec);
… …
861fail: 944fail:
862 return 0; 945 return 0;
863} 946}
864 947
865static SV * 948static SV *
866decode_json (SV *string, UV flags) 949decode_json (SV *string, U32 flags)
867{ 950{
868 SV *sv; 951 SV *sv;
869 952
870 if (flags & F_UTF8) 953 if (flags & F_UTF8)
871 sv_utf8_downgrade (string, 0); 954 sv_utf8_downgrade (string, 0);
… …
873 sv_utf8_upgrade (string); 956 sv_utf8_upgrade (string);
874 957
875 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 958 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
876 959
877 dec_t dec; 960 dec_t dec;
878 dec.flags = flags; 961 dec.flags = flags;
879 dec.cur = SvPVX (string); 962 dec.cur = SvPVX (string);
880 dec.end = SvEND (string); 963 dec.end = SvEND (string);
881 dec.err = 0; 964 dec.err = 0;
965 dec.depth = 0;
966 dec.maxdepth = DEC_DEPTH (dec.flags);
882 967
968 *dec.end = 0; // this should basically be a nop, too, but make sure its there
883 sv = decode_sv (&dec); 969 sv = decode_sv (&dec);
884 970
885 if (!sv) 971 if (!sv)
886 { 972 {
887 IV offset = dec.flags & F_UTF8 973 IV offset = dec.flags & F_UTF8
… …
910 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); 996 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
911 997
912 return sv; 998 return sv;
913} 999}
914 1000
1001/////////////////////////////////////////////////////////////////////////////
1002// XS interface functions
1003
915MODULE = JSON::XS PACKAGE = JSON::XS 1004MODULE = JSON::XS PACKAGE = JSON::XS
916 1005
917BOOT: 1006BOOT:
918{ 1007{
919 int i; 1008 int i;
920 1009
921 memset (decode_hexdigit, 0xff, 256); 1010 memset (decode_hexdigit, 0xff, 256);
1011
922 for (i = 10; i--; ) 1012 for (i = 0; i < 256; ++i)
923 decode_hexdigit ['0' + i] = i; 1013 decode_hexdigit [i] =
924 1014 i >= '0' && i <= '9' ? i - '0'
925 for (i = 7; i--; ) 1015 : i >= 'a' && i <= 'f' ? i - 'a' + 10
926 { 1016 : i >= 'A' && i <= 'F' ? i - 'A' + 10
927 decode_hexdigit ['a' + i] = 10 + i; 1017 : -1;
928 decode_hexdigit ['A' + i] = 10 + i;
929 }
930 1018
931 json_stash = gv_stashpv ("JSON::XS", 1); 1019 json_stash = gv_stashpv ("JSON::XS", 1);
932} 1020}
933 1021
934PROTOTYPES: DISABLE 1022PROTOTYPES: DISABLE
… …
961 RETVAL = newSVsv (self); 1049 RETVAL = newSVsv (self);
962} 1050}
963 OUTPUT: 1051 OUTPUT:
964 RETVAL 1052 RETVAL
965 1053
1054SV *max_depth (SV *self, int max_depth = 0x80000000UL)
1055 CODE:
1056{
1057 UV *uv = SvJSON (self);
1058 UV log2 = 0;
1059
1060 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1061
1062 while ((1UL << log2) < max_depth)
1063 ++log2;
1064
1065 *uv = *uv & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1066
1067 RETVAL = newSVsv (self);
1068}
1069 OUTPUT:
1070 RETVAL
1071
966void encode (SV *self, SV *scalar) 1072void encode (SV *self, SV *scalar)
967 PPCODE: 1073 PPCODE:
968 XPUSHs (encode_json (scalar, *SvJSON (self))); 1074 XPUSHs (encode_json (scalar, *SvJSON (self)));
969 1075
970void decode (SV *self, SV *jsonstr) 1076void decode (SV *self, SV *jsonstr)
… …
972 XPUSHs (decode_json (jsonstr, *SvJSON (self))); 1078 XPUSHs (decode_json (jsonstr, *SvJSON (self)));
973 1079
974PROTOTYPES: ENABLE 1080PROTOTYPES: ENABLE
975 1081
976void to_json (SV *scalar) 1082void to_json (SV *scalar)
1083 ALIAS:
1084 objToJson = 0
977 PPCODE: 1085 PPCODE:
978 XPUSHs (encode_json (scalar, F_UTF8)); 1086 XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8));
979 1087
980void from_json (SV *jsonstr) 1088void from_json (SV *jsonstr)
1089 ALIAS:
1090 jsonToObj = 0
981 PPCODE: 1091 PPCODE:
982 XPUSHs (decode_json (jsonstr, F_UTF8)); 1092 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8));
983 1093

Diff Legend

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