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 1.11 by root, Sat Mar 24 19:42:14 2007 UTC vs.
Revision 1.24 by root, Tue Apr 3 23:59:04 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 (9UL << 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 {
82 *clen = (STRLEN)-1;
83 return (UV)-1;
84 }
85}
86
87/////////////////////////////////////////////////////////////////////////////
88// encoder
27 89
28// structure used for encoding JSON 90// structure used for encoding JSON
29typedef struct 91typedef struct
30{ 92{
31 char *cur; 93 char *cur; // SvPVX (sv) + current output position
32 STRLEN len; // SvLEN (sv)
33 char *end; // SvEND (sv) 94 char *end; // SvEND (sv)
34 SV *sv; 95 SV *sv; // result scalar
35 UV flags; 96 U32 flags; // F_*
36 int max_recurse; 97 U32 indent; // indentation level
37 int indent; 98 U32 maxdepth; // max. indentation/recursion level
38} enc_t; 99} 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/////////////////////////////////////////////////////////////////////////////
68 100
69static void 101static void
70need (enc_t *enc, STRLEN len) 102need (enc_t *enc, STRLEN len)
71{ 103{
72 if (enc->cur + len >= enc->end) 104 if (enc->cur + len >= enc->end)
… …
130 STRLEN clen; 162 STRLEN clen;
131 UV uch; 163 UV uch;
132 164
133 if (is_utf8) 165 if (is_utf8)
134 { 166 {
135 uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY); 167 //uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
168 uch = decode_utf8 (str, end - str, &clen);
136 if (clen == (STRLEN)-1) 169 if (clen == (STRLEN)-1)
137 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str); 170 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
138 } 171 }
139 else 172 else
140 { 173 {
… …
178 } 211 }
179 while (--clen); 212 while (--clen);
180 } 213 }
181 else 214 else
182 { 215 {
183 need (enc, len += 10); // never more than 11 bytes needed 216 need (enc, len += UTF8_MAX_LEN - 1); // never more than 11 bytes needed
184 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 217 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
185 ++str; 218 ++str;
186 } 219 }
187 } 220 }
188 } 221 }
… …
190 223
191 --len; 224 --len;
192 } 225 }
193} 226}
194 227
195#define INDENT SB \ 228static void
229encode_indent (enc_t *enc)
230{
196 if (enc->flags & F_INDENT) \ 231 if (enc->flags & F_INDENT)
197 { \ 232 {
198 int i_; \ 233 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
204 234
205#define SPACE SB need (enc, 1); encode_ch (enc, ' '); SE 235 need (enc, spaces);
206#define NL SB if (enc->flags & F_INDENT) { need (enc, 1); encode_ch (enc, '\n'); } SE 236 memset (enc->cur, ' ', spaces);
207#define COMMA SB \ 237 enc->cur += spaces;
238 }
239}
240
241static void
242encode_space (enc_t *enc)
243{
244 need (enc, 1);
208 encode_ch (enc, ','); \ 245 encode_ch (enc, ' ');
246}
247
248static void
249encode_nl (enc_t *enc)
250{
209 if (enc->flags & F_INDENT) \ 251 if (enc->flags & F_INDENT)
210 NL; \ 252 {
253 need (enc, 1);
254 encode_ch (enc, '\n');
255 }
256}
257
258static void
259encode_comma (enc_t *enc)
260{
261 encode_ch (enc, ',');
262
263 if (enc->flags & F_INDENT)
264 encode_nl (enc);
211 else if (enc->flags & F_SPACE_AFTER) \ 265 else if (enc->flags & F_SPACE_AFTER)
212 SPACE; \ 266 encode_space (enc);
213 SE 267}
214 268
215static void encode_sv (enc_t *enc, SV *sv); 269static void encode_sv (enc_t *enc, SV *sv);
216 270
217static void 271static void
218encode_av (enc_t *enc, AV *av) 272encode_av (enc_t *enc, AV *av)
219{ 273{
220 int i, len = av_len (av); 274 int i, len = av_len (av);
221 275
222 encode_ch (enc, '['); NL; 276 if (enc->indent >= enc->maxdepth)
277 croak ("data structure too deep (hit recursion limit)");
278
279 encode_ch (enc, '['); encode_nl (enc);
223 ++enc->indent; 280 ++enc->indent;
224 281
225 for (i = 0; i <= len; ++i) 282 for (i = 0; i <= len; ++i)
226 { 283 {
227 INDENT; 284 encode_indent (enc);
228 encode_sv (enc, *av_fetch (av, i, 0)); 285 encode_sv (enc, *av_fetch (av, i, 0));
229 286
230 if (i < len) 287 if (i < len)
231 COMMA; 288 encode_comma (enc);
232 } 289 }
233 290
234 NL; 291 encode_nl (enc);
235 292
236 --enc->indent; 293 --enc->indent;
237 INDENT; encode_ch (enc, ']'); 294 encode_indent (enc); encode_ch (enc, ']');
238} 295}
239 296
240static void 297static void
241encode_he (enc_t *enc, HE *he) 298encode_he (enc_t *enc, HE *he)
242{ 299{
… …
256 else 313 else
257 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 314 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
258 315
259 encode_ch (enc, '"'); 316 encode_ch (enc, '"');
260 317
261 if (enc->flags & F_SPACE_BEFORE) SPACE; 318 if (enc->flags & F_SPACE_BEFORE) encode_space (enc);
262 encode_ch (enc, ':'); 319 encode_ch (enc, ':');
263 if (enc->flags & F_SPACE_AFTER ) SPACE; 320 if (enc->flags & F_SPACE_AFTER ) encode_space (enc);
264 encode_sv (enc, HeVAL (he)); 321 encode_sv (enc, HeVAL (he));
265} 322}
266 323
267// compare hash entries, used when all keys are bytestrings 324// compare hash entries, used when all keys are bytestrings
268static int 325static int
… …
292static void 349static void
293encode_hv (enc_t *enc, HV *hv) 350encode_hv (enc_t *enc, HV *hv)
294{ 351{
295 int count, i; 352 int count, i;
296 353
354 if (enc->indent >= enc->maxdepth)
355 croak ("data structure too deep (hit recursion limit)");
356
297 encode_ch (enc, '{'); NL; ++enc->indent; 357 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
298 358
299 if ((count = hv_iterinit (hv))) 359 if ((count = hv_iterinit (hv)))
300 { 360 {
301 // for canonical output we have to sort by keys first 361 // for canonical output we have to sort by keys first
302 // actually, this is mostly due to the stupid so-called 362 // actually, this is mostly due to the stupid so-called
303 // security workaround added somewhere in 5.8.x. 363 // security workaround added somewhere in 5.8.x.
304 // that randomises hash orderings 364 // that randomises hash orderings
305 if (enc->flags & F_CANONICAL) 365 if (enc->flags & F_CANONICAL)
306 { 366 {
307 HE *he, *hes [count]; 367 HE *he, *hes [count]; // if your compiler dies here, you need to enable C99 mode
308 int fast = 1; 368 int fast = 1;
309 369
310 i = 0; 370 i = 0;
311 while ((he = hv_iternext (hv))) 371 while ((he = hv_iternext (hv)))
312 { 372 {
… …
337 LEAVE; 397 LEAVE;
338 } 398 }
339 399
340 for (i = 0; i < count; ++i) 400 for (i = 0; i < count; ++i)
341 { 401 {
342 INDENT; 402 encode_indent (enc);
343 encode_he (enc, hes [i]); 403 encode_he (enc, hes [i]);
344 404
345 if (i < count - 1) 405 if (i < count - 1)
346 COMMA; 406 encode_comma (enc);
347 } 407 }
348 408
349 NL; 409 encode_nl (enc);
350 } 410 }
351 else 411 else
352 { 412 {
353 SV *sv; 413 SV *sv;
354 HE *he = hv_iternext (hv); 414 HE *he = hv_iternext (hv);
355 415
356 for (;;) 416 for (;;)
357 { 417 {
358 INDENT; 418 encode_indent (enc);
359 encode_he (enc, he); 419 encode_he (enc, he);
360 420
361 if (!(he = hv_iternext (hv))) 421 if (!(he = hv_iternext (hv)))
362 break; 422 break;
363 423
364 COMMA; 424 encode_comma (enc);
365 } 425 }
366 426
367 NL; 427 encode_nl (enc);
368 } 428 }
369 } 429 }
370 430
371 --enc->indent; INDENT; encode_ch (enc, '}'); 431 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
432}
433
434// encode objects, arrays and special \0=false and \1=true values.
435static void
436encode_rv (enc_t *enc, SV *sv)
437{
438 SvGETMAGIC (sv);
439
440 svtype svt = SvTYPE (sv);
441
442 if (svt == SVt_PVHV)
443 encode_hv (enc, (HV *)sv);
444 else if (svt == SVt_PVAV)
445 encode_av (enc, (AV *)sv);
446 else if (svt < SVt_PVAV)
447 {
448 if (SvNIOK (sv) && SvIV (sv) == 0)
449 encode_str (enc, "false", 5, 0);
450 else if (SvNIOK (sv) && SvIV (sv) == 1)
451 encode_str (enc, "true", 4, 0);
452 else
453 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
454 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
455 }
456 else
457 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
458 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
372} 459}
373 460
374static void 461static void
375encode_sv (enc_t *enc, SV *sv) 462encode_sv (enc_t *enc, SV *sv)
376{ 463{
… …
397 SvIsUV(sv) 484 SvIsUV(sv)
398 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 485 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv))
399 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 486 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv));
400 } 487 }
401 else if (SvROK (sv)) 488 else if (SvROK (sv))
402 { 489 encode_rv (enc, SvRV (sv));
403 SV *rv = SvRV (sv);
404
405 if (!--enc->max_recurse)
406 croak ("data structure too deep (hit recursion limit)");
407
408 switch (SvTYPE (rv))
409 {
410 case SVt_PVAV: encode_av (enc, (AV *)rv); break;
411 case SVt_PVHV: encode_hv (enc, (HV *)rv); break;
412
413 default:
414 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
415 SvPV_nolen (sv));
416 }
417 }
418 else if (!SvOK (sv)) 490 else if (!SvOK (sv))
419 encode_str (enc, "null", 4, 0); 491 encode_str (enc, "null", 4, 0);
420 else 492 else
421 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 493 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
422 SvPV_nolen (sv), SvFLAGS (sv)); 494 SvPV_nolen (sv), SvFLAGS (sv));
423} 495}
424 496
425static SV * 497static SV *
426encode_json (SV *scalar, UV flags) 498encode_json (SV *scalar, U32 flags)
427{ 499{
428 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 500 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar))
429 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 501 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
430 502
431 enc_t enc; 503 enc_t enc;
432 enc.flags = flags; 504 enc.flags = flags;
433 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 505 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
434 enc.cur = SvPVX (enc.sv); 506 enc.cur = SvPVX (enc.sv);
435 enc.end = SvEND (enc.sv); 507 enc.end = SvEND (enc.sv);
436 enc.max_recurse = 0;
437 enc.indent = 0; 508 enc.indent = 0;
509 enc.maxdepth = DEC_DEPTH (flags);
438 510
439 SvPOK_only (enc.sv); 511 SvPOK_only (enc.sv);
440 encode_sv (&enc, scalar); 512 encode_sv (&enc, scalar);
441 513
442 if (!(flags & (F_ASCII | F_UTF8))) 514 if (!(flags & (F_ASCII | F_UTF8)))
… …
449 521
450 return enc.sv; 522 return enc.sv;
451} 523}
452 524
453///////////////////////////////////////////////////////////////////////////// 525/////////////////////////////////////////////////////////////////////////////
526// decoder
454 527
455#define WS \ 528// structure used for decoding JSON
529typedef struct
530{
531 char *cur; // current parser pointer
532 char *end; // end of input string
533 const char *err; // parse error, if != 0
534 U32 flags; // F_*
535 U32 depth; // recursion depth
536 U32 maxdepth; // recursion depth limit
537} dec_t;
538
539static void
540decode_ws (dec_t *dec)
541{
456 for (;;) \ 542 for (;;)
457 { \ 543 {
458 char ch = *dec->cur; \ 544 char ch = *dec->cur;
545
459 if (ch > 0x20 \ 546 if (ch > 0x20
460 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) \ 547 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09))
461 break; \ 548 break;
549
462 ++dec->cur; \ 550 ++dec->cur;
463 } 551 }
552}
464 553
465#define ERR(reason) SB dec->err = reason; goto fail; SE 554#define ERR(reason) SB dec->err = reason; goto fail; SE
555
466#define EXPECT_CH(ch) SB \ 556#define EXPECT_CH(ch) SB \
467 if (*dec->cur != ch) \ 557 if (*dec->cur != ch) \
468 ERR (# ch " expected"); \ 558 ERR (# ch " expected"); \
469 ++dec->cur; \ 559 ++dec->cur; \
470 SE 560 SE
471 561
562#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)")
563#define DEC_DEC_DEPTH --dec->depth
564
472static SV *decode_sv (dec_t *dec); 565static SV *decode_sv (dec_t *dec);
473 566
474static signed char decode_hexdigit[256]; 567static signed char decode_hexdigit[256];
475 568
476static UV 569static UV
477decode_4hex (dec_t *dec) 570decode_4hex (dec_t *dec)
478{ 571{
479 signed char d1, d2, d3, d4; 572 signed char d1, d2, d3, d4;
573 unsigned char *cur = (unsigned char *)dec->cur;
480 574
481 d1 = decode_hexdigit [((unsigned char *)dec->cur) [0]];
482 if (d1 < 0) ERR ("four hexadecimal digits expected"); 575 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"); 576 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"); 577 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"); 578 d4 = decode_hexdigit [cur [3]]; if (d4 < 0) ERR ("four hexadecimal digits expected");
489 579
490 dec->cur += 4; 580 dec->cur += 4;
491 581
492 return ((UV)d1) << 12 582 return ((UV)d1) << 12
493 | ((UV)d2) << 8 583 | ((UV)d2) << 8
… …
496 586
497fail: 587fail:
498 return (UV)-1; 588 return (UV)-1;
499} 589}
500 590
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 * 591static SV *
517decode_str (dec_t *dec) 592decode_str (dec_t *dec)
518{ 593{
519 SV *sv = NEWSV (0,2); 594 SV *sv = 0;
520 int utf8 = 0; 595 int utf8 = 0;
521 char *cur = SvPVX (sv);
522 char *end = SvEND (sv);
523 596
524 for (;;) 597 do
525 { 598 {
526 unsigned char ch = *(unsigned char *)dec->cur; 599 char buf [SHORT_STRING_LEN + UTF8_MAX_LEN];
600 char *cur = buf;
527 601
528 if (ch == '"') 602 do
529 break;
530 else if (ch == '\\')
531 { 603 {
532 switch (*++dec->cur) 604 unsigned char ch = *(unsigned char *)dec->cur++;
605
606 if (ch == '"')
533 { 607 {
534 case '\\': 608 --dec->cur;
535 case '/': 609 break;
536 case '"': APPEND_CH (*dec->cur++); break; 610 }
537 611 else if (ch == '\\')
538 case 'b': APPEND_CH ('\010'); ++dec->cur; break; 612 {
539 case 't': APPEND_CH ('\011'); ++dec->cur; break; 613 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 { 614 {
546 UV lo, hi; 615 case '\\':
547 ++dec->cur; 616 case '/':
617 case '"': *cur++ = *dec->cur++; break;
548 618
549 hi = decode_4hex (dec); 619 case 'b': ++dec->cur; *cur++ = '\010'; break;
550 if (hi == (UV)-1) 620 case 't': ++dec->cur; *cur++ = '\011'; break;
551 goto fail; 621 case 'n': ++dec->cur; *cur++ = '\012'; break;
622 case 'f': ++dec->cur; *cur++ = '\014'; break;
623 case 'r': ++dec->cur; *cur++ = '\015'; break;
552 624
553 // possibly a surrogate pair 625 case 'u':
554 if (hi >= 0xd800 && hi < 0xdc00)
555 { 626 {
556 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 627 UV lo, hi;
557 ERR ("missing low surrogate character in surrogate pair");
558
559 dec->cur += 2; 628 ++dec->cur;
560 629
561 lo = decode_4hex (dec); 630 hi = decode_4hex (dec);
562 if (lo == (UV)-1) 631 if (hi == (UV)-1)
563 goto fail; 632 goto fail;
564 633
634 // possibly a surrogate pair
635 if (hi >= 0xd800)
636 if (hi < 0xdc00)
637 {
638 if (dec->cur [0] != '\\' || dec->cur [1] != 'u')
639 ERR ("missing low surrogate character in surrogate pair");
640
641 dec->cur += 2;
642
643 lo = decode_4hex (dec);
644 if (lo == (UV)-1)
645 goto fail;
646
565 if (lo < 0xdc00 || lo >= 0xe000) 647 if (lo < 0xdc00 || lo >= 0xe000)
566 ERR ("surrogate pair expected"); 648 ERR ("surrogate pair expected");
567 649
568 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000; 650 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000;
651 }
652 else if (hi < 0xe000)
653 ERR ("missing high surrogate character in surrogate pair");
654
655 if (hi >= 0x80)
656 {
657 utf8 = 1;
658
659 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
660 }
661 else
662 *cur++ = hi;
569 } 663 }
570 else if (hi >= 0xdc00 && hi < 0xe000)
571 ERR ("missing high surrogate character in surrogate pair");
572
573 if (hi >= 0x80)
574 { 664 break;
575 utf8 = 1;
576 665
577 APPEND_GROW (4); // at most 4 bytes for 21 bits
578 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
579 }
580 else 666 default:
581 APPEND_CH (hi); 667 --dec->cur;
668 ERR ("illegal backslash escape sequence in string");
582 } 669 }
583 break; 670 }
584 671 else if (ch >= 0x20 && ch <= 0x7f)
585 default: 672 *cur++ = ch;
673 else if (ch >= 0x80)
674 {
586 --dec->cur; 675 --dec->cur;
587 ERR ("illegal backslash escape sequence in string"); 676
677 STRLEN clen;
678 UV uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen);
679 if (clen == (STRLEN)-1)
680 ERR ("malformed UTF-8 character in JSON string");
681
682 do
683 *cur++ = *dec->cur++;
684 while (--clen);
685
686 utf8 = 1;
687 }
688 else
689 {
690 --dec->cur;
691
692 if (!ch)
693 ERR ("unexpected end of string while parsing JSON string");
694 else
695 ERR ("invalid character encountered while parsing JSON string");
588 } 696 }
589 } 697 }
590 else if (ch >= 0x20 && ch <= 0x7f) 698 while (cur < buf + SHORT_STRING_LEN);
591 APPEND_CH (*dec->cur++); 699
592 else if (ch >= 0x80) 700 STRLEN len = cur - buf;
701
702 if (sv)
593 { 703 {
594 STRLEN clen; 704 SvGROW (sv, SvCUR (sv) + len + 1);
595 UV uch = utf8n_to_uvuni (dec->cur, dec->end - dec->cur, &clen, UTF8_CHECK_ONLY); 705 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
596 if (clen == (STRLEN)-1) 706 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 } 707 }
608 else if (dec->cur == dec->end)
609 ERR ("unexpected end of string while parsing json string");
610 else 708 else
611 ERR ("invalid character encountered"); 709 sv = newSVpvn (buf, len);
612 } 710 }
711 while (*dec->cur != '"');
613 712
614 ++dec->cur; 713 ++dec->cur;
615 714
616 SvCUR_set (sv, cur - SvPVX (sv)); 715 if (sv)
617 716 {
618 SvPOK_only (sv); 717 SvPOK_only (sv);
619 *SvEND (sv) = 0; 718 *SvEND (sv) = 0;
620 719
621 if (utf8) 720 if (utf8)
622 SvUTF8_on (sv); 721 SvUTF8_on (sv);
623 722 }
624 if (dec->flags & F_SHRINK) 723 else
625 shrink (sv); 724 sv = newSVpvn ("", 0);
626 725
627 return sv; 726 return sv;
628 727
629fail: 728fail:
630 SvREFCNT_dec (sv);
631 return 0; 729 return 0;
632} 730}
633 731
634static SV * 732static SV *
635decode_num (dec_t *dec) 733decode_num (dec_t *dec)
… …
716static SV * 814static SV *
717decode_av (dec_t *dec) 815decode_av (dec_t *dec)
718{ 816{
719 AV *av = newAV (); 817 AV *av = newAV ();
720 818
721 WS; 819 DEC_INC_DEPTH;
820 decode_ws (dec);
821
722 if (*dec->cur == ']') 822 if (*dec->cur == ']')
723 ++dec->cur; 823 ++dec->cur;
724 else 824 else
725 for (;;) 825 for (;;)
726 { 826 {
… …
730 if (!value) 830 if (!value)
731 goto fail; 831 goto fail;
732 832
733 av_push (av, value); 833 av_push (av, value);
734 834
735 WS; 835 decode_ws (dec);
736 836
737 if (*dec->cur == ']') 837 if (*dec->cur == ']')
738 { 838 {
739 ++dec->cur; 839 ++dec->cur;
740 break; 840 break;
… …
744 ERR (", or ] expected while parsing array"); 844 ERR (", or ] expected while parsing array");
745 845
746 ++dec->cur; 846 ++dec->cur;
747 } 847 }
748 848
849 DEC_DEC_DEPTH;
749 return newRV_noinc ((SV *)av); 850 return newRV_noinc ((SV *)av);
750 851
751fail: 852fail:
752 SvREFCNT_dec (av); 853 SvREFCNT_dec (av);
854 DEC_DEC_DEPTH;
753 return 0; 855 return 0;
754} 856}
755 857
756static SV * 858static SV *
757decode_hv (dec_t *dec) 859decode_hv (dec_t *dec)
758{ 860{
759 HV *hv = newHV (); 861 HV *hv = newHV ();
760 862
761 WS; 863 DEC_INC_DEPTH;
864 decode_ws (dec);
865
762 if (*dec->cur == '}') 866 if (*dec->cur == '}')
763 ++dec->cur; 867 ++dec->cur;
764 else 868 else
765 for (;;) 869 for (;;)
766 { 870 {
767 SV *key, *value; 871 SV *key, *value;
768 872
769 WS; EXPECT_CH ('"'); 873 decode_ws (dec); EXPECT_CH ('"');
770 874
771 key = decode_str (dec); 875 key = decode_str (dec);
772 if (!key) 876 if (!key)
773 goto fail; 877 goto fail;
774 878
775 WS; EXPECT_CH (':'); 879 decode_ws (dec); EXPECT_CH (':');
776 880
777 value = decode_sv (dec); 881 value = decode_sv (dec);
778 if (!value) 882 if (!value)
779 { 883 {
780 SvREFCNT_dec (key); 884 SvREFCNT_dec (key);
781 goto fail; 885 goto fail;
782 } 886 }
783 887
784 //TODO: optimise
785 hv_store_ent (hv, key, value, 0); 888 hv_store_ent (hv, key, value, 0);
889 SvREFCNT_dec (key);
786 890
787 WS; 891 decode_ws (dec);
788 892
789 if (*dec->cur == '}') 893 if (*dec->cur == '}')
790 { 894 {
791 ++dec->cur; 895 ++dec->cur;
792 break; 896 break;
… …
796 ERR (", or } expected while parsing object/hash"); 900 ERR (", or } expected while parsing object/hash");
797 901
798 ++dec->cur; 902 ++dec->cur;
799 } 903 }
800 904
905 DEC_DEC_DEPTH;
801 return newRV_noinc ((SV *)hv); 906 return newRV_noinc ((SV *)hv);
802 907
803fail: 908fail:
804 SvREFCNT_dec (hv); 909 SvREFCNT_dec (hv);
910 DEC_DEC_DEPTH;
805 return 0; 911 return 0;
806} 912}
807 913
808static SV * 914static SV *
809decode_sv (dec_t *dec) 915decode_sv (dec_t *dec)
810{ 916{
811 WS; 917 decode_ws (dec);
812 switch (*dec->cur) 918 switch (*dec->cur)
813 { 919 {
814 case '"': ++dec->cur; return decode_str (dec); 920 case '"': ++dec->cur; return decode_str (dec);
815 case '[': ++dec->cur; return decode_av (dec); 921 case '[': ++dec->cur; return decode_av (dec);
816 case '{': ++dec->cur; return decode_hv (dec); 922 case '{': ++dec->cur; return decode_hv (dec);
… …
852 ERR ("'null' expected"); 958 ERR ("'null' expected");
853 959
854 break; 960 break;
855 961
856 default: 962 default:
857 ERR ("malformed json string, neither array, object, number, string or atom"); 963 ERR ("malformed JSON string, neither array, object, number, string or atom");
858 break; 964 break;
859 } 965 }
860 966
861fail: 967fail:
862 return 0; 968 return 0;
863} 969}
864 970
865static SV * 971static SV *
866decode_json (SV *string, UV flags) 972decode_json (SV *string, U32 flags)
867{ 973{
868 SV *sv; 974 SV *sv;
975
976 SvUPGRADE (string, SVt_PV);
869 977
870 if (flags & F_UTF8) 978 if (flags & F_UTF8)
871 sv_utf8_downgrade (string, 0); 979 sv_utf8_downgrade (string, 0);
872 else 980 else
873 sv_utf8_upgrade (string); 981 sv_utf8_upgrade (string);
874 982
875 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 983 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
876 984
877 dec_t dec; 985 dec_t dec;
878 dec.flags = flags; 986 dec.flags = flags;
879 dec.cur = SvPVX (string); 987 dec.cur = SvPVX (string);
880 dec.end = SvEND (string); 988 dec.end = SvEND (string);
881 dec.err = 0; 989 dec.err = 0;
990 dec.depth = 0;
991 dec.maxdepth = DEC_DEPTH (dec.flags);
882 992
993 *dec.end = 0; // this should basically be a nop, too, but make sure its there
883 sv = decode_sv (&dec); 994 sv = decode_sv (&dec);
884 995
885 if (!sv) 996 if (!sv)
886 { 997 {
887 IV offset = dec.flags & F_UTF8 998 IV offset = dec.flags & F_UTF8
… …
896 SAVEVPTR (PL_curcop); 1007 SAVEVPTR (PL_curcop);
897 PL_curcop = &cop; 1008 PL_curcop = &cop;
898 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1009 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
899 LEAVE; 1010 LEAVE;
900 1011
901 croak ("%s, at character offset %d (%s)", 1012 croak ("%s, at character offset %d [\"%s\"]",
902 dec.err, 1013 dec.err,
903 (int)offset, 1014 (int)offset,
904 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1015 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
905 } 1016 }
906 1017
… …
910 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); 1021 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
911 1022
912 return sv; 1023 return sv;
913} 1024}
914 1025
1026/////////////////////////////////////////////////////////////////////////////
1027// XS interface functions
1028
915MODULE = JSON::XS PACKAGE = JSON::XS 1029MODULE = JSON::XS PACKAGE = JSON::XS
916 1030
917BOOT: 1031BOOT:
918{ 1032{
919 int i; 1033 int i;
920 1034
921 memset (decode_hexdigit, 0xff, 256); 1035 memset (decode_hexdigit, 0xff, 256);
1036
922 for (i = 10; i--; ) 1037 for (i = 0; i < 256; ++i)
923 decode_hexdigit ['0' + i] = i; 1038 decode_hexdigit [i] =
924 1039 i >= '0' && i <= '9' ? i - '0'
925 for (i = 7; i--; ) 1040 : i >= 'a' && i <= 'f' ? i - 'a' + 10
926 { 1041 : i >= 'A' && i <= 'F' ? i - 'A' + 10
927 decode_hexdigit ['a' + i] = 10 + i; 1042 : -1;
928 decode_hexdigit ['A' + i] = 10 + i;
929 }
930 1043
931 json_stash = gv_stashpv ("JSON::XS", 1); 1044 json_stash = gv_stashpv ("JSON::XS", 1);
932} 1045}
933 1046
934PROTOTYPES: DISABLE 1047PROTOTYPES: DISABLE
… …
961 RETVAL = newSVsv (self); 1074 RETVAL = newSVsv (self);
962} 1075}
963 OUTPUT: 1076 OUTPUT:
964 RETVAL 1077 RETVAL
965 1078
1079SV *max_depth (SV *self, int max_depth = 0x80000000UL)
1080 CODE:
1081{
1082 UV *uv = SvJSON (self);
1083 UV log2 = 0;
1084
1085 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1086
1087 while ((1UL << log2) < max_depth)
1088 ++log2;
1089
1090 *uv = *uv & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1091
1092 RETVAL = newSVsv (self);
1093}
1094 OUTPUT:
1095 RETVAL
1096
966void encode (SV *self, SV *scalar) 1097void encode (SV *self, SV *scalar)
967 PPCODE: 1098 PPCODE:
968 XPUSHs (encode_json (scalar, *SvJSON (self))); 1099 XPUSHs (encode_json (scalar, *SvJSON (self)));
969 1100
970void decode (SV *self, SV *jsonstr) 1101void decode (SV *self, SV *jsonstr)
… …
972 XPUSHs (decode_json (jsonstr, *SvJSON (self))); 1103 XPUSHs (decode_json (jsonstr, *SvJSON (self)));
973 1104
974PROTOTYPES: ENABLE 1105PROTOTYPES: ENABLE
975 1106
976void to_json (SV *scalar) 1107void to_json (SV *scalar)
1108 ALIAS:
1109 objToJson = 0
977 PPCODE: 1110 PPCODE:
978 XPUSHs (encode_json (scalar, F_UTF8)); 1111 XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8));
979 1112
980void from_json (SV *jsonstr) 1113void from_json (SV *jsonstr)
1114 ALIAS:
1115 jsonToObj = 0
981 PPCODE: 1116 PPCODE:
982 XPUSHs (decode_json (jsonstr, F_UTF8)); 1117 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8));
983 1118

Diff Legend

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