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 1.9 by root, Fri Mar 23 17:40:29 2007 UTC vs.
Revision 1.18 by root, Sun Mar 25 21:19:13 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_JSON_RPC 0x00000040
16#define F_ALLOW_NONREF 0x00000080 15#define F_ALLOW_NONREF 0x00000080UL
17#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? <=> { $__class__$ => }
18 24
19#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 25#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
20#define F_DEFAULT 0 26#define F_DEFAULT (13UL << S_MAXDEPTH)
21 27
22#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
23 33
24#define SB do { 34#define SB do {
25#define SE } while (0) 35#define SE } while (0)
26 36
27static 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
28 86
29// structure used for encoding JSON 87// structure used for encoding JSON
30typedef struct 88typedef struct
31{ 89{
32 char *cur; 90 char *cur; // SvPVX (sv) + current output position
33 STRLEN len; // SvLEN (sv)
34 char *end; // SvEND (sv) 91 char *end; // SvEND (sv)
35 SV *sv; 92 SV *sv; // result scalar
36 UV flags; 93 U32 flags; // F_*
37 int max_recurse; 94 U32 indent; // indentation level
38 int indent; 95 U32 maxdepth; // max. indentation/recursion level
39} enc_t; 96} enc_t;
40
41// structure used for decoding JSON
42typedef struct
43{
44 char *cur;
45 char *end;
46 const char *err;
47 UV flags;
48} dec_t;
49
50static UV *
51SvJSON (SV *sv)
52{
53 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash))
54 croak ("object is not of type JSON::XS");
55
56 return &SvUVX (SvRV (sv));
57}
58
59static void
60shrink (SV *sv)
61{
62 sv_utf8_downgrade (sv, 1);
63#ifdef SvPV_shrink_to_cur
64 SvPV_shrink_to_cur (sv);
65#endif
66}
67
68/////////////////////////////////////////////////////////////////////////////
69 97
70static void 98static void
71need (enc_t *enc, STRLEN len) 99need (enc_t *enc, STRLEN len)
72{ 100{
73 if (enc->cur + len >= enc->end) 101 if (enc->cur + len >= enc->end)
… …
131 STRLEN clen; 159 STRLEN clen;
132 UV uch; 160 UV uch;
133 161
134 if (is_utf8) 162 if (is_utf8)
135 { 163 {
136 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);
137 if (clen == (STRLEN)-1) 166 if (clen == (STRLEN)-1)
138 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);
139 } 168 }
140 else 169 else
141 { 170 {
… …
150 { 179 {
151 if (uch > 0xFFFFUL) 180 if (uch > 0xFFFFUL)
152 { 181 {
153 need (enc, len += 11); 182 need (enc, len += 11);
154 sprintf (enc->cur, "\\u%04x\\u%04x", 183 sprintf (enc->cur, "\\u%04x\\u%04x",
155 (uch - 0x10000) / 0x400 + 0xD800, 184 (int)((uch - 0x10000) / 0x400 + 0xD800),
156 (uch - 0x10000) % 0x400 + 0xDC00); 185 (int)((uch - 0x10000) % 0x400 + 0xDC00));
157 enc->cur += 12; 186 enc->cur += 12;
158 } 187 }
159 else 188 else
160 { 189 {
161 static char hexdigit [16] = "0123456789abcdef"; 190 static char hexdigit [16] = "0123456789abcdef";
… …
179 } 208 }
180 while (--clen); 209 while (--clen);
181 } 210 }
182 else 211 else
183 { 212 {
184 need (enc, len += 10); // never more than 11 bytes needed 213 need (enc, len += UTF8_MAX_LEN - 1); // never more than 11 bytes needed
185 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 214 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
186 ++str; 215 ++str;
187 } 216 }
188 } 217 }
189 } 218 }
… …
191 220
192 --len; 221 --len;
193 } 222 }
194} 223}
195 224
196#define INDENT SB \ 225static void
226encode_indent (enc_t *enc)
227{
197 if (enc->flags & F_INDENT) \ 228 if (enc->flags & F_INDENT)
198 { \ 229 {
199 int i_; \ 230 int spaces = enc->indent * INDENT_STEP;
200 need (enc, enc->indent); \
201 for (i_ = enc->indent * 3; i_--; )\
202 encode_ch (enc, ' '); \
203 } \
204 SE
205 231
206#define SPACE SB need (enc, 1); encode_ch (enc, ' '); SE 232 need (enc, spaces);
207#define NL SB if (enc->flags & F_INDENT) { need (enc, 1); encode_ch (enc, '\n'); } SE 233 memset (enc->cur, ' ', spaces);
208#define COMMA SB \ 234 enc->cur += spaces;
235 }
236}
237
238static void
239encode_space (enc_t *enc)
240{
241 need (enc, 1);
209 encode_ch (enc, ','); \ 242 encode_ch (enc, ' ');
243}
244
245static void
246encode_nl (enc_t *enc)
247{
210 if (enc->flags & F_INDENT) \ 248 if (enc->flags & F_INDENT)
211 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);
212 else if (enc->flags & F_SPACE_AFTER) \ 262 else if (enc->flags & F_SPACE_AFTER)
213 SPACE; \ 263 encode_space (enc);
214 SE 264}
215 265
216static void encode_sv (enc_t *enc, SV *sv); 266static void encode_sv (enc_t *enc, SV *sv);
217 267
218static void 268static void
219encode_av (enc_t *enc, AV *av) 269encode_av (enc_t *enc, AV *av)
220{ 270{
221 int i, len = av_len (av); 271 int i, len = av_len (av);
222 272
223 encode_ch (enc, '['); NL; 273 encode_ch (enc, '['); encode_nl (enc);
224 ++enc->indent; 274 ++enc->indent;
225 275
226 for (i = 0; i <= len; ++i) 276 for (i = 0; i <= len; ++i)
227 { 277 {
228 INDENT; 278 encode_indent (enc);
229 encode_sv (enc, *av_fetch (av, i, 0)); 279 encode_sv (enc, *av_fetch (av, i, 0));
230 280
231 if (i < len) 281 if (i < len)
232 COMMA; 282 encode_comma (enc);
233 } 283 }
234 284
235 NL; 285 encode_nl (enc);
236 286
237 --enc->indent; 287 --enc->indent;
238 INDENT; encode_ch (enc, ']'); 288 encode_indent (enc); encode_ch (enc, ']');
239} 289}
240 290
241static void 291static void
242encode_he (enc_t *enc, HE *he) 292encode_he (enc_t *enc, HE *he)
243{ 293{
… …
257 else 307 else
258 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 308 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
259 309
260 encode_ch (enc, '"'); 310 encode_ch (enc, '"');
261 311
262 if (enc->flags & F_SPACE_BEFORE) SPACE; 312 if (enc->flags & F_SPACE_BEFORE) encode_space (enc);
263 encode_ch (enc, ':'); 313 encode_ch (enc, ':');
264 if (enc->flags & F_SPACE_AFTER ) SPACE; 314 if (enc->flags & F_SPACE_AFTER ) encode_space (enc);
265 encode_sv (enc, HeVAL (he)); 315 encode_sv (enc, HeVAL (he));
266} 316}
267 317
268// compare hash entries, used when all keys are bytestrings 318// compare hash entries, used when all keys are bytestrings
269static int 319static int
… …
275 HE *b = *(HE **)b_; 325 HE *b = *(HE **)b_;
276 326
277 STRLEN la = HeKLEN (a); 327 STRLEN la = HeKLEN (a);
278 STRLEN lb = HeKLEN (b); 328 STRLEN lb = HeKLEN (b);
279 329
280 if (!(cmp == memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 330 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb)))
281 cmp = la < lb ? -1 : la == lb ? 0 : 1; 331 cmp = la - lb;
282 332
283 return cmp; 333 return cmp;
284} 334}
285 335
286// 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
… …
293static void 343static void
294encode_hv (enc_t *enc, HV *hv) 344encode_hv (enc_t *enc, HV *hv)
295{ 345{
296 int count, i; 346 int count, i;
297 347
298 encode_ch (enc, '{'); NL; ++enc->indent; 348 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
299 349
300 if ((count = hv_iterinit (hv))) 350 if ((count = hv_iterinit (hv)))
301 { 351 {
302 // for canonical output we have to sort by keys first 352 // for canonical output we have to sort by keys first
303 // actually, this is mostly due to the stupid so-called 353 // actually, this is mostly due to the stupid so-called
304 // security workaround added somewhere in 5.8.x. 354 // security workaround added somewhere in 5.8.x.
305 // that randomises hash orderings 355 // that randomises hash orderings
306 if (enc->flags & F_CANONICAL) 356 if (enc->flags & F_CANONICAL)
307 { 357 {
308 HE *he, *hes [count]; 358 HE *he, *hes [count]; // if your compiler dies here, you need to enable C99 mode
309 int fast = 1; 359 int fast = 1;
310 360
311 i = 0; 361 i = 0;
312 while ((he = hv_iternext (hv))) 362 while ((he = hv_iternext (hv)))
313 { 363 {
… …
338 LEAVE; 388 LEAVE;
339 } 389 }
340 390
341 for (i = 0; i < count; ++i) 391 for (i = 0; i < count; ++i)
342 { 392 {
343 INDENT; 393 encode_indent (enc);
344 encode_he (enc, hes [i]); 394 encode_he (enc, hes [i]);
345 395
346 if (i < count - 1) 396 if (i < count - 1)
347 COMMA; 397 encode_comma (enc);
348 } 398 }
349 399
350 NL; 400 encode_nl (enc);
351 } 401 }
352 else 402 else
353 { 403 {
354 SV *sv; 404 SV *sv;
355 HE *he = hv_iternext (hv); 405 HE *he = hv_iternext (hv);
356 406
357 for (;;) 407 for (;;)
358 { 408 {
359 INDENT; 409 encode_indent (enc);
360 encode_he (enc, he); 410 encode_he (enc, he);
361 411
362 if (!(he = hv_iternext (hv))) 412 if (!(he = hv_iternext (hv)))
363 break; 413 break;
364 414
365 COMMA; 415 encode_comma (enc);
366 } 416 }
367 417
368 NL; 418 encode_nl (enc);
369 } 419 }
370 } 420 }
371 421
372 --enc->indent; INDENT; encode_ch (enc, '}'); 422 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
373} 423}
374 424
375static void 425static void
376encode_sv (enc_t *enc, SV *sv) 426encode_sv (enc_t *enc, SV *sv)
377{ 427{
… …
401 } 451 }
402 else if (SvROK (sv)) 452 else if (SvROK (sv))
403 { 453 {
404 SV *rv = SvRV (sv); 454 SV *rv = SvRV (sv);
405 455
406 if (!--enc->max_recurse) 456 if (enc->indent >= enc->maxdepth)
407 croak ("data structure too deep (hit recursion limit)"); 457 croak ("data structure too deep (hit recursion limit)");
408 458
409 switch (SvTYPE (rv)) 459 switch (SvTYPE (rv))
410 { 460 {
411 case SVt_PVAV: encode_av (enc, (AV *)rv); break; 461 case SVt_PVAV: encode_av (enc, (AV *)rv); break;
… …
422 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",
423 SvPV_nolen (sv), SvFLAGS (sv)); 473 SvPV_nolen (sv), SvFLAGS (sv));
424} 474}
425 475
426static SV * 476static SV *
427encode_json (SV *scalar, UV flags) 477encode_json (SV *scalar, U32 flags)
428{ 478{
429 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 479 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar))
430 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)");
431 481
432 enc_t enc; 482 enc_t enc;
433 enc.flags = flags; 483 enc.flags = flags;
434 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 484 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
435 enc.cur = SvPVX (enc.sv); 485 enc.cur = SvPVX (enc.sv);
436 enc.end = SvEND (enc.sv); 486 enc.end = SvEND (enc.sv);
437 enc.max_recurse = 0;
438 enc.indent = 0; 487 enc.indent = 0;
488 enc.maxdepth = DEC_DEPTH (flags);
439 489
440 SvPOK_only (enc.sv); 490 SvPOK_only (enc.sv);
441 encode_sv (&enc, scalar); 491 encode_sv (&enc, scalar);
442 492
443 if (!(flags & (F_ASCII | F_UTF8))) 493 if (!(flags & (F_ASCII | F_UTF8)))
… …
450 500
451 return enc.sv; 501 return enc.sv;
452} 502}
453 503
454///////////////////////////////////////////////////////////////////////////// 504/////////////////////////////////////////////////////////////////////////////
505// decoder
455 506
456#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{
457 for (;;) \ 521 for (;;)
458 { \ 522 {
459 char ch = *dec->cur; \ 523 char ch = *dec->cur;
524
460 if (ch > 0x20 \ 525 if (ch > 0x20
461 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) \ 526 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09))
462 break; \ 527 break;
528
463 ++dec->cur; \ 529 ++dec->cur;
464 } 530 }
531}
465 532
466#define ERR(reason) SB dec->err = reason; goto fail; SE 533#define ERR(reason) SB dec->err = reason; goto fail; SE
534
467#define EXPECT_CH(ch) SB \ 535#define EXPECT_CH(ch) SB \
468 if (*dec->cur != ch) \ 536 if (*dec->cur != ch) \
469 ERR (# ch " expected"); \ 537 ERR (# ch " expected"); \
470 ++dec->cur; \ 538 ++dec->cur; \
471 SE 539 SE
472 540
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
473static SV *decode_sv (dec_t *dec); 544static SV *decode_sv (dec_t *dec);
474 545
475static signed char decode_hexdigit[256]; 546static signed char decode_hexdigit[256];
476 547
477static UV 548static UV
478decode_4hex (dec_t *dec) 549decode_4hex (dec_t *dec)
479{ 550{
480 signed char d1, d2, d3, d4; 551 signed char d1, d2, d3, d4;
552 unsigned char *cur = (unsigned char *)dec->cur;
481 553
482 d1 = decode_hexdigit [((unsigned char *)dec->cur) [0]];
483 if (d1 < 0) ERR ("four hexadecimal digits expected"); 554 d1 = decode_hexdigit [cur [0]]; if (d1 < 0) ERR ("four hexadecimal digits expected");
484 d2 = decode_hexdigit [((unsigned char *)dec->cur) [1]];
485 if (d2 < 0) ERR ("four hexadecimal digits expected"); 555 d2 = decode_hexdigit [cur [1]]; if (d2 < 0) ERR ("four hexadecimal digits expected");
486 d3 = decode_hexdigit [((unsigned char *)dec->cur) [2]];
487 if (d3 < 0) ERR ("four hexadecimal digits expected"); 556 d3 = decode_hexdigit [cur [2]]; if (d3 < 0) ERR ("four hexadecimal digits expected");
488 d4 = decode_hexdigit [((unsigned char *)dec->cur) [3]];
489 if (d4 < 0) ERR ("four hexadecimal digits expected"); 557 d4 = decode_hexdigit [cur [3]]; if (d4 < 0) ERR ("four hexadecimal digits expected");
490 558
491 dec->cur += 4; 559 dec->cur += 4;
492 560
493 return ((UV)d1) << 12 561 return ((UV)d1) << 12
494 | ((UV)d2) << 8 562 | ((UV)d2) << 8
… …
497 565
498fail: 566fail:
499 return (UV)-1; 567 return (UV)-1;
500} 568}
501 569
502#define APPEND_GROW(n) SB \
503 if (cur + (n) >= end) \
504 { \
505 STRLEN ofs = cur - SvPVX (sv); \
506 SvGROW (sv, ofs + (n) + 1); \
507 cur = SvPVX (sv) + ofs; \
508 end = SvEND (sv); \
509 } \
510 SE
511
512#define APPEND_CH(ch) SB \
513 APPEND_GROW (1); \
514 *cur++ = (ch); \
515 SE
516
517static SV * 570static SV *
518decode_str (dec_t *dec) 571decode_str (dec_t *dec)
519{ 572{
520 SV *sv = NEWSV (0,2); 573 SV *sv = 0;
521 int utf8 = 0; 574 int utf8 = 0;
522 char *cur = SvPVX (sv);
523 char *end = SvEND (sv);
524 575
525 for (;;) 576 do
526 { 577 {
527 unsigned char ch = *(unsigned char *)dec->cur; 578 char buf [SHORT_STRING_LEN + UTF8_MAX_LEN];
579 char *cur = buf;
528 580
529 if (ch == '"') 581 do
530 break;
531 else if (ch == '\\')
532 { 582 {
533 switch (*++dec->cur) 583 unsigned char ch = *(unsigned char *)dec->cur++;
584
585 if (ch == '"')
534 { 586 {
535 case '\\': 587 --dec->cur;
536 case '/': 588 break;
537 case '"': APPEND_CH (*dec->cur++); break; 589 }
538 590 else if (ch == '\\')
539 case 'b': APPEND_CH ('\010'); ++dec->cur; break; 591 {
540 case 't': APPEND_CH ('\011'); ++dec->cur; break; 592 switch (*dec->cur)
541 case 'n': APPEND_CH ('\012'); ++dec->cur; break;
542 case 'f': APPEND_CH ('\014'); ++dec->cur; break;
543 case 'r': APPEND_CH ('\015'); ++dec->cur; break;
544
545 case 'u':
546 { 593 {
547 UV lo, hi; 594 case '\\':
548 ++dec->cur; 595 case '/':
596 case '"': *cur++ = *dec->cur++; break;
549 597
550 hi = decode_4hex (dec); 598 case 'b': ++dec->cur; *cur++ = '\010'; break;
551 if (hi == (UV)-1) 599 case 't': ++dec->cur; *cur++ = '\011'; break;
552 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;
553 603
554 // possibly a surrogate pair 604 case 'u':
555 if (hi >= 0xd800 && hi < 0xdc00)
556 { 605 {
557 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 606 UV lo, hi;
558 ERR ("missing low surrogate character in surrogate pair");
559
560 dec->cur += 2; 607 ++dec->cur;
561 608
562 lo = decode_4hex (dec); 609 hi = decode_4hex (dec);
563 if (lo == (UV)-1) 610 if (hi == (UV)-1)
564 goto fail; 611 goto fail;
565 612
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
566 if (lo < 0xdc00 || lo >= 0xe000) 626 if (lo < 0xdc00 || lo >= 0xe000)
567 ERR ("surrogate pair expected"); 627 ERR ("surrogate pair expected");
568 628
569 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;
570 } 642 }
571 else if (hi >= 0xdc00 && hi < 0xe000)
572 ERR ("missing high surrogate character in surrogate pair");
573
574 if (hi >= 0x80)
575 { 643 break;
576 utf8 = 1;
577 644
578 APPEND_GROW (4); // at most 4 bytes for 21 bits
579 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
580 }
581 else 645 default:
582 APPEND_CH (hi); 646 --dec->cur;
647 ERR ("illegal backslash escape sequence in string");
583 } 648 }
584 break;
585
586 default:
587 --dec->cur;
588 ERR ("illegal backslash escape sequence in string");
589 } 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
590 } 674 }
591 else if (ch >= 0x20 && ch <= 0x7f) 675 while (cur < buf + SHORT_STRING_LEN);
592 APPEND_CH (*dec->cur++); 676
593 else if (ch >= 0x80) 677 STRLEN len = cur - buf;
678
679 if (sv)
594 { 680 {
595 STRLEN clen; 681 SvGROW (sv, SvCUR (sv) + len + 1);
596 UV uch = utf8n_to_uvuni (dec->cur, dec->end - dec->cur, &clen, UTF8_CHECK_ONLY); 682 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
597 if (clen == (STRLEN)-1) 683 SvCUR_set (sv, SvCUR (sv) + len);
598 ERR ("malformed UTF-8 character in JSON string");
599
600 APPEND_GROW (clen);
601 do
602 {
603 *cur++ = *dec->cur++;
604 }
605 while (--clen);
606
607 utf8 = 1;
608 } 684 }
609 else if (dec->cur == dec->end)
610 ERR ("unexpected end of string while parsing json string");
611 else 685 else
612 ERR ("invalid character encountered"); 686 sv = newSVpvn (buf, len);
613 } 687 }
688 while (*dec->cur != '"');
614 689
615 ++dec->cur; 690 ++dec->cur;
616 691
617 SvCUR_set (sv, cur - SvPVX (sv)); 692 if (sv)
618 693 {
619 SvPOK_only (sv); 694 SvPOK_only (sv);
620 *SvEND (sv) = 0; 695 *SvEND (sv) = 0;
621 696
622 if (utf8) 697 if (utf8)
623 SvUTF8_on (sv); 698 SvUTF8_on (sv);
624 699 }
625 if (dec->flags & F_SHRINK) 700 else
626 shrink (sv); 701 sv = newSVpvn ("", 0);
627 702
628 return sv; 703 return sv;
629 704
630fail: 705fail:
631 SvREFCNT_dec (sv);
632 return 0; 706 return 0;
633} 707}
634 708
635static SV * 709static SV *
636decode_num (dec_t *dec) 710decode_num (dec_t *dec)
… …
717static SV * 791static SV *
718decode_av (dec_t *dec) 792decode_av (dec_t *dec)
719{ 793{
720 AV *av = newAV (); 794 AV *av = newAV ();
721 795
722 WS; 796 DEC_INC_DEPTH;
797 decode_ws (dec);
798
723 if (*dec->cur == ']') 799 if (*dec->cur == ']')
724 ++dec->cur; 800 ++dec->cur;
725 else 801 else
726 for (;;) 802 for (;;)
727 { 803 {
… …
731 if (!value) 807 if (!value)
732 goto fail; 808 goto fail;
733 809
734 av_push (av, value); 810 av_push (av, value);
735 811
736 WS; 812 decode_ws (dec);
737 813
738 if (*dec->cur == ']') 814 if (*dec->cur == ']')
739 { 815 {
740 ++dec->cur; 816 ++dec->cur;
741 break; 817 break;
… …
745 ERR (", or ] expected while parsing array"); 821 ERR (", or ] expected while parsing array");
746 822
747 ++dec->cur; 823 ++dec->cur;
748 } 824 }
749 825
826 DEC_DEC_DEPTH;
750 return newRV_noinc ((SV *)av); 827 return newRV_noinc ((SV *)av);
751 828
752fail: 829fail:
753 SvREFCNT_dec (av); 830 SvREFCNT_dec (av);
831 DEC_DEC_DEPTH;
754 return 0; 832 return 0;
755} 833}
756 834
757static SV * 835static SV *
758decode_hv (dec_t *dec) 836decode_hv (dec_t *dec)
759{ 837{
760 HV *hv = newHV (); 838 HV *hv = newHV ();
761 839
762 WS; 840 DEC_INC_DEPTH;
841 decode_ws (dec);
842
763 if (*dec->cur == '}') 843 if (*dec->cur == '}')
764 ++dec->cur; 844 ++dec->cur;
765 else 845 else
766 for (;;) 846 for (;;)
767 { 847 {
768 SV *key, *value; 848 SV *key, *value;
769 849
770 WS; EXPECT_CH ('"'); 850 decode_ws (dec); EXPECT_CH ('"');
771 851
772 key = decode_str (dec); 852 key = decode_str (dec);
773 if (!key) 853 if (!key)
774 goto fail; 854 goto fail;
775 855
776 WS; EXPECT_CH (':'); 856 decode_ws (dec); EXPECT_CH (':');
777 857
778 value = decode_sv (dec); 858 value = decode_sv (dec);
779 if (!value) 859 if (!value)
780 { 860 {
781 SvREFCNT_dec (key); 861 SvREFCNT_dec (key);
782 goto fail; 862 goto fail;
783 } 863 }
784 864
785 //TODO: optimise
786 hv_store_ent (hv, key, value, 0); 865 hv_store_ent (hv, key, value, 0);
866 SvREFCNT_dec (key);
787 867
788 WS; 868 decode_ws (dec);
789 869
790 if (*dec->cur == '}') 870 if (*dec->cur == '}')
791 { 871 {
792 ++dec->cur; 872 ++dec->cur;
793 break; 873 break;
… …
797 ERR (", or } expected while parsing object/hash"); 877 ERR (", or } expected while parsing object/hash");
798 878
799 ++dec->cur; 879 ++dec->cur;
800 } 880 }
801 881
882 DEC_DEC_DEPTH;
802 return newRV_noinc ((SV *)hv); 883 return newRV_noinc ((SV *)hv);
803 884
804fail: 885fail:
805 SvREFCNT_dec (hv); 886 SvREFCNT_dec (hv);
887 DEC_DEC_DEPTH;
806 return 0; 888 return 0;
807} 889}
808 890
809static SV * 891static SV *
810decode_sv (dec_t *dec) 892decode_sv (dec_t *dec)
811{ 893{
812 WS; 894 decode_ws (dec);
813 switch (*dec->cur) 895 switch (*dec->cur)
814 { 896 {
815 case '"': ++dec->cur; return decode_str (dec); 897 case '"': ++dec->cur; return decode_str (dec);
816 case '[': ++dec->cur; return decode_av (dec); 898 case '[': ++dec->cur; return decode_av (dec);
817 case '{': ++dec->cur; return decode_hv (dec); 899 case '{': ++dec->cur; return decode_hv (dec);
… …
862fail: 944fail:
863 return 0; 945 return 0;
864} 946}
865 947
866static SV * 948static SV *
867decode_json (SV *string, UV flags) 949decode_json (SV *string, U32 flags)
868{ 950{
869 SV *sv; 951 SV *sv;
870 952
871 if (flags & F_UTF8) 953 if (flags & F_UTF8)
872 sv_utf8_downgrade (string, 0); 954 sv_utf8_downgrade (string, 0);
… …
874 sv_utf8_upgrade (string); 956 sv_utf8_upgrade (string);
875 957
876 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 958 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
877 959
878 dec_t dec; 960 dec_t dec;
879 dec.flags = flags; 961 dec.flags = flags;
880 dec.cur = SvPVX (string); 962 dec.cur = SvPVX (string);
881 dec.end = SvEND (string); 963 dec.end = SvEND (string);
882 dec.err = 0; 964 dec.err = 0;
965 dec.depth = 0;
966 dec.maxdepth = DEC_DEPTH (dec.flags);
883 967
968 *SvEND (sv) = 0; // this shou[ld basically be a nop, too
884 sv = decode_sv (&dec); 969 sv = decode_sv (&dec);
885 970
886 if (!sv) 971 if (!sv)
887 { 972 {
888 IV offset = dec.flags & F_UTF8 973 IV offset = dec.flags & F_UTF8
… …
911 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)");
912 997
913 return sv; 998 return sv;
914} 999}
915 1000
1001/////////////////////////////////////////////////////////////////////////////
1002// XS interface functions
1003
916MODULE = JSON::XS PACKAGE = JSON::XS 1004MODULE = JSON::XS PACKAGE = JSON::XS
917 1005
918BOOT: 1006BOOT:
919{ 1007{
920 int i; 1008 int i;
921 1009
922 memset (decode_hexdigit, 0xff, 256); 1010 memset (decode_hexdigit, 0xff, 256);
1011
923 for (i = 10; i--; ) 1012 for (i = 0; i < 256; ++i)
924 decode_hexdigit ['0' + i] = i; 1013 decode_hexdigit [i] =
925 1014 i >= '0' && i <= '9' ? i - '0'
926 for (i = 7; i--; ) 1015 : i >= 'a' && i <= 'f' ? i - 'a' + 10
927 { 1016 : i >= 'A' && i <= 'F' ? i - 'A' + 10
928 decode_hexdigit ['a' + i] = 10 + i; 1017 : -1;
929 decode_hexdigit ['A' + i] = 10 + i;
930 }
931 1018
932 json_stash = gv_stashpv ("JSON::XS", 1); 1019 json_stash = gv_stashpv ("JSON::XS", 1);
933} 1020}
934 1021
935PROTOTYPES: DISABLE 1022PROTOTYPES: DISABLE
… …
946 utf8 = F_UTF8 1033 utf8 = F_UTF8
947 indent = F_INDENT 1034 indent = F_INDENT
948 canonical = F_CANONICAL 1035 canonical = F_CANONICAL
949 space_before = F_SPACE_BEFORE 1036 space_before = F_SPACE_BEFORE
950 space_after = F_SPACE_AFTER 1037 space_after = F_SPACE_AFTER
951 json_rpc = F_JSON_RPC
952 pretty = F_PRETTY 1038 pretty = F_PRETTY
953 allow_nonref = F_ALLOW_NONREF 1039 allow_nonref = F_ALLOW_NONREF
954 shrink = F_SHRINK 1040 shrink = F_SHRINK
955 CODE: 1041 CODE:
956{ 1042{
… …
963 RETVAL = newSVsv (self); 1049 RETVAL = newSVsv (self);
964} 1050}
965 OUTPUT: 1051 OUTPUT:
966 RETVAL 1052 RETVAL
967 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
968void encode (SV *self, SV *scalar) 1072void encode (SV *self, SV *scalar)
969 PPCODE: 1073 PPCODE:
970 XPUSHs (encode_json (scalar, *SvJSON (self))); 1074 XPUSHs (encode_json (scalar, *SvJSON (self)));
971 1075
972void decode (SV *self, SV *jsonstr) 1076void decode (SV *self, SV *jsonstr)
… …
974 XPUSHs (decode_json (jsonstr, *SvJSON (self))); 1078 XPUSHs (decode_json (jsonstr, *SvJSON (self)));
975 1079
976PROTOTYPES: ENABLE 1080PROTOTYPES: ENABLE
977 1081
978void to_json (SV *scalar) 1082void to_json (SV *scalar)
1083 ALIAS:
1084 objToJson = 0
979 PPCODE: 1085 PPCODE:
980 XPUSHs (encode_json (scalar, F_UTF8)); 1086 XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8));
981 1087
982void from_json (SV *jsonstr) 1088void from_json (SV *jsonstr)
1089 ALIAS:
1090 jsonToObj = 0
983 PPCODE: 1091 PPCODE:
984 XPUSHs (decode_json (jsonstr, F_UTF8)); 1092 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8));
985 1093

Diff Legend

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