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 1.6 by root, Fri Mar 23 15:10:55 2007 UTC vs.
Revision 1.12 by root, Sat Mar 24 22:10:08 2007 UTC

… …
10#define F_UTF8 0x00000002 10#define F_UTF8 0x00000002
11#define F_INDENT 0x00000004 11#define F_INDENT 0x00000004
12#define F_CANONICAL 0x00000008 12#define F_CANONICAL 0x00000008
13#define F_SPACE_BEFORE 0x00000010 13#define F_SPACE_BEFORE 0x00000010
14#define F_SPACE_AFTER 0x00000020 14#define F_SPACE_AFTER 0x00000020
15#define F_JSON_RPC 0x00000040
16#define F_ALLOW_NONREF 0x00000080 15#define F_ALLOW_NONREF 0x00000080
17#define F_SHRINK 0x00000100 16#define F_SHRINK 0x00000100
18 17
19#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 18#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
20#define F_DEFAULT 0 19#define F_DEFAULT 0
21 20
22#define INIT_SIZE 32 // initial scalar size to be allocated 21#define INIT_SIZE 32 // initial scalar size to be allocated
22#define INDENT_STEP 3 // spaces per indentation level
23
24#define UTF8_MAX_LEN 11 // for perls UTF-X: max. number of octets per character
25#define SHORT_STRING_LEN 256 // special-case strings of up to this size
23 26
24#define SB do { 27#define SB do {
25#define SE } while (0) 28#define SE } while (0)
26 29
27static HV *json_stash; 30static HV *json_stash; // JSON::XS::
31
32/////////////////////////////////////////////////////////////////////////////
33// utility functions
34
35static UV *
36SvJSON (SV *sv)
37{
38 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash))
39 croak ("object is not of type JSON::XS");
40
41 return &SvUVX (SvRV (sv));
42}
43
44static void
45shrink (SV *sv)
46{
47 sv_utf8_downgrade (sv, 1);
48 if (SvLEN (sv) > SvCUR (sv) + 1)
49 {
50#ifdef SvPV_shrink_to_cur
51 SvPV_shrink_to_cur (sv);
52#elif defined (SvPV_renew)
53 SvPV_renew (sv, SvCUR (sv) + 1);
54#endif
55 }
56}
57
58/////////////////////////////////////////////////////////////////////////////
59// encoder
28 60
29// structure used for encoding JSON 61// structure used for encoding JSON
30typedef struct 62typedef struct
31{ 63{
32 char *cur; 64 char *cur; // SvPVX (sv) + current output position
33 STRLEN len; // SvLEN (sv)
34 char *end; // SvEND (sv) 65 char *end; // SvEND (sv)
35 SV *sv; 66 SV *sv; // result scalar
36 UV flags; 67 UV flags; // F_*
37 int max_recurse; 68 int indent; // indentation level
38 int indent; 69 int max_depth; // max. recursion level
39} enc_t; 70} 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
59/////////////////////////////////////////////////////////////////////////////
60 71
61static void 72static void
62need (enc_t *enc, STRLEN len) 73need (enc_t *enc, STRLEN len)
63{ 74{
64 if (enc->cur + len >= enc->end) 75 if (enc->cur + len >= enc->end)
… …
124 135
125 if (is_utf8) 136 if (is_utf8)
126 { 137 {
127 uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY); 138 uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
128 if (clen == (STRLEN)-1) 139 if (clen == (STRLEN)-1)
129 croak ("malformed UTF-8 character in string, cannot convert to JSON"); 140 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
130 } 141 }
131 else 142 else
132 { 143 {
133 uch = ch; 144 uch = ch;
134 clen = 1; 145 clen = 1;
135 } 146 }
136 147
148 if (uch > 0x10FFFFUL)
149 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
150
137 if (uch < 0x80 || enc->flags & F_ASCII) 151 if (uch < 0x80 || enc->flags & F_ASCII)
138 { 152 {
139 if (uch > 0xFFFFUL) 153 if (uch > 0xFFFFUL)
140 { 154 {
141 need (enc, len += 11); 155 need (enc, len += 11);
142 sprintf (enc->cur, "\\u%04x\\u%04x", 156 sprintf (enc->cur, "\\u%04x\\u%04x",
143 (uch - 0x10000) / 0x400 + 0xD800, 157 (int)((uch - 0x10000) / 0x400 + 0xD800),
144 (uch - 0x10000) % 0x400 + 0xDC00); 158 (int)((uch - 0x10000) % 0x400 + 0xDC00));
145 enc->cur += 12; 159 enc->cur += 12;
146 } 160 }
147 else 161 else
148 { 162 {
149 static char hexdigit [16] = "0123456789abcdef"; 163 static char hexdigit [16] = "0123456789abcdef";
… …
167 } 181 }
168 while (--clen); 182 while (--clen);
169 } 183 }
170 else 184 else
171 { 185 {
172 need (enc, 10); // never more than 11 bytes needed 186 need (enc, len += UTF8_MAX_LEN - 1); // never more than 11 bytes needed
173 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 187 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
174 ++str; 188 ++str;
175 } 189 }
176 } 190 }
177 } 191 }
… …
179 193
180 --len; 194 --len;
181 } 195 }
182} 196}
183 197
184#define INDENT SB \ 198static void
199encode_indent (enc_t *enc)
200{
185 if (enc->flags & F_INDENT) \ 201 if (enc->flags & F_INDENT)
186 { \ 202 {
187 int i_; \ 203 int spaces = enc->indent * INDENT_STEP;
188 need (enc, enc->indent); \
189 for (i_ = enc->indent * 3; i_--; )\
190 encode_ch (enc, ' '); \
191 } \
192 SE
193 204
194#define SPACE SB need (enc, 1); encode_ch (enc, ' '); SE 205 need (enc, spaces);
195#define NL SB if (enc->flags & F_INDENT) { need (enc, 1); encode_ch (enc, '\n'); } SE 206 memset (enc->cur, ' ', spaces);
196#define COMMA SB \ 207 enc->cur += spaces;
208 }
209}
210
211static void
212encode_space (enc_t *enc)
213{
214 need (enc, 1);
197 encode_ch (enc, ','); \ 215 encode_ch (enc, ' ');
216}
217
218static void
219encode_nl (enc_t *enc)
220{
198 if (enc->flags & F_INDENT) \ 221 if (enc->flags & F_INDENT)
199 NL; \ 222 {
223 need (enc, 1);
224 encode_ch (enc, '\n');
225 }
226}
227
228static void
229encode_comma (enc_t *enc)
230{
231 encode_ch (enc, ',');
232
233 if (enc->flags & F_INDENT)
234 encode_nl (enc);
200 else if (enc->flags & F_SPACE_AFTER) \ 235 else if (enc->flags & F_SPACE_AFTER)
201 SPACE; \ 236 encode_space (enc);
202 SE 237}
203 238
204static void encode_sv (enc_t *enc, SV *sv); 239static void encode_sv (enc_t *enc, SV *sv);
205 240
206static void 241static void
207encode_av (enc_t *enc, AV *av) 242encode_av (enc_t *enc, AV *av)
208{ 243{
209 int i, len = av_len (av); 244 int i, len = av_len (av);
210 245
211 encode_ch (enc, '['); NL; 246 encode_ch (enc, '['); encode_nl (enc);
212 ++enc->indent; 247 ++enc->indent;
213 248
214 for (i = 0; i <= len; ++i) 249 for (i = 0; i <= len; ++i)
215 { 250 {
216 INDENT; 251 encode_indent (enc);
217 encode_sv (enc, *av_fetch (av, i, 0)); 252 encode_sv (enc, *av_fetch (av, i, 0));
218 253
219 if (i < len) 254 if (i < len)
220 COMMA; 255 encode_comma (enc);
221 } 256 }
222 257
223 NL; 258 encode_nl (enc);
224 259
225 --enc->indent; 260 --enc->indent;
226 INDENT; encode_ch (enc, ']'); 261 encode_indent (enc); encode_ch (enc, ']');
227} 262}
228 263
229static void 264static void
230encode_he (enc_t *enc, HE *he) 265encode_he (enc_t *enc, HE *he)
231{ 266{
… …
245 else 280 else
246 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 281 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
247 282
248 encode_ch (enc, '"'); 283 encode_ch (enc, '"');
249 284
250 if (enc->flags & F_SPACE_BEFORE) SPACE; 285 if (enc->flags & F_SPACE_BEFORE) encode_space (enc);
251 encode_ch (enc, ':'); 286 encode_ch (enc, ':');
252 if (enc->flags & F_SPACE_AFTER ) SPACE; 287 if (enc->flags & F_SPACE_AFTER ) encode_space (enc);
253 encode_sv (enc, HeVAL (he)); 288 encode_sv (enc, HeVAL (he));
254} 289}
255 290
256// compare hash entries, used when all keys are bytestrings 291// compare hash entries, used when all keys are bytestrings
257static int 292static int
… …
263 HE *b = *(HE **)b_; 298 HE *b = *(HE **)b_;
264 299
265 STRLEN la = HeKLEN (a); 300 STRLEN la = HeKLEN (a);
266 STRLEN lb = HeKLEN (b); 301 STRLEN lb = HeKLEN (b);
267 302
268 if (!(cmp == memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 303 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb)))
269 cmp = la < lb ? -1 : la == lb ? 0 : 1; 304 cmp = la - lb;
270 305
271 return cmp; 306 return cmp;
272} 307}
273 308
274// compare hash entries, used when some keys are sv's or utf-x 309// compare hash entries, used when some keys are sv's or utf-x
… …
281static void 316static void
282encode_hv (enc_t *enc, HV *hv) 317encode_hv (enc_t *enc, HV *hv)
283{ 318{
284 int count, i; 319 int count, i;
285 320
286 encode_ch (enc, '{'); NL; ++enc->indent; 321 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
287 322
288 if ((count = hv_iterinit (hv))) 323 if ((count = hv_iterinit (hv)))
289 { 324 {
290 // for canonical output we have to sort by keys first 325 // for canonical output we have to sort by keys first
291 // actually, this is mostly due to the stupid so-called 326 // actually, this is mostly due to the stupid so-called
292 // security workaround added somewhere in 5.8.x. 327 // security workaround added somewhere in 5.8.x.
293 // that randomises hash orderings 328 // that randomises hash orderings
294 if (enc->flags & F_CANONICAL) 329 if (enc->flags & F_CANONICAL)
295 { 330 {
296 HE *he, *hes [count]; 331 HE *he, *hes [count]; // if your compiler dies here, you need to enable C99 mode
297 int fast = 1; 332 int fast = 1;
298 333
299 i = 0; 334 i = 0;
300 while ((he = hv_iternext (hv))) 335 while ((he = hv_iternext (hv)))
301 { 336 {
… …
308 343
309 if (fast) 344 if (fast)
310 qsort (hes, count, sizeof (HE *), he_cmp_fast); 345 qsort (hes, count, sizeof (HE *), he_cmp_fast);
311 else 346 else
312 { 347 {
313 // hack to disable "use bytes" 348 // hack to forcefully disable "use bytes"
314 COP *oldcop = PL_curcop, cop; 349 COP cop = *PL_curcop;
315 cop.op_private = 0; 350 cop.op_private = 0;
351
352 ENTER;
353 SAVETMPS;
354
355 SAVEVPTR (PL_curcop);
316 PL_curcop = &cop; 356 PL_curcop = &cop;
317 357
318 SAVETMPS;
319 qsort (hes, count, sizeof (HE *), he_cmp_slow); 358 qsort (hes, count, sizeof (HE *), he_cmp_slow);
359
320 FREETMPS; 360 FREETMPS;
321 361 LEAVE;
322 PL_curcop = oldcop;
323 } 362 }
324 363
325 for (i = 0; i < count; ++i) 364 for (i = 0; i < count; ++i)
326 { 365 {
327 INDENT; 366 encode_indent (enc);
328 encode_he (enc, hes [i]); 367 encode_he (enc, hes [i]);
329 368
330 if (i < count - 1) 369 if (i < count - 1)
331 COMMA; 370 encode_comma (enc);
332 } 371 }
333 372
334 NL; 373 encode_nl (enc);
335 } 374 }
336 else 375 else
337 { 376 {
338 SV *sv; 377 SV *sv;
339 HE *he = hv_iternext (hv); 378 HE *he = hv_iternext (hv);
340 379
341 for (;;) 380 for (;;)
342 { 381 {
343 INDENT; 382 encode_indent (enc);
344 encode_he (enc, he); 383 encode_he (enc, he);
345 384
346 if (!(he = hv_iternext (hv))) 385 if (!(he = hv_iternext (hv)))
347 break; 386 break;
348 387
349 COMMA; 388 encode_comma (enc);
350 } 389 }
351 390
352 NL; 391 encode_nl (enc);
353 } 392 }
354 } 393 }
355 394
356 --enc->indent; INDENT; encode_ch (enc, '}'); 395 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
357} 396}
358 397
359static void 398static void
360encode_sv (enc_t *enc, SV *sv) 399encode_sv (enc_t *enc, SV *sv)
361{ 400{
… …
383 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 422 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv))
384 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 423 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv));
385 } 424 }
386 else if (SvROK (sv)) 425 else if (SvROK (sv))
387 { 426 {
388 if (!--enc->max_recurse) 427 SV *rv = SvRV (sv);
428
429 if (enc->indent >= enc->max_depth)
389 croak ("data structure too deep (hit recursion limit)"); 430 croak ("data structure too deep (hit recursion limit)");
390 431
391 sv = SvRV (sv);
392
393 switch (SvTYPE (sv)) 432 switch (SvTYPE (rv))
394 { 433 {
395 case SVt_PVAV: encode_av (enc, (AV *)sv); break; 434 case SVt_PVAV: encode_av (enc, (AV *)rv); break;
396 case SVt_PVHV: encode_hv (enc, (HV *)sv); break; 435 case SVt_PVHV: encode_hv (enc, (HV *)rv); break;
397 436
398 default: 437 default:
399 croak ("JSON can only represent references to arrays or hashes"); 438 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
439 SvPV_nolen (sv));
400 } 440 }
401 } 441 }
402 else if (!SvOK (sv)) 442 else if (!SvOK (sv))
403 encode_str (enc, "null", 4, 0); 443 encode_str (enc, "null", 4, 0);
404 else 444 else
405 croak ("encountered perl type that JSON cannot handle"); 445 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
446 SvPV_nolen (sv), SvFLAGS (sv));
406} 447}
407 448
408static SV * 449static SV *
409encode_json (SV *scalar, UV flags) 450encode_json (SV *scalar, UV flags)
410{ 451{
411 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 452 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar))
412 croak ("hash- or arraref required (not a simple scalar, use allow_nonref to allow this)"); 453 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
413 454
414 enc_t enc; 455 enc_t enc;
415 enc.flags = flags; 456 enc.flags = flags;
416 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 457 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
417 enc.cur = SvPVX (enc.sv); 458 enc.cur = SvPVX (enc.sv);
418 enc.end = SvEND (enc.sv); 459 enc.end = SvEND (enc.sv);
419 enc.max_recurse = 0;
420 enc.indent = 0; 460 enc.indent = 0;
461 enc.max_depth = 0x7fffffffUL;
421 462
422 SvPOK_only (enc.sv); 463 SvPOK_only (enc.sv);
423 encode_sv (&enc, scalar); 464 encode_sv (&enc, scalar);
424 465
425 if (!(flags & (F_ASCII | F_UTF8))) 466 if (!(flags & (F_ASCII | F_UTF8)))
426 SvUTF8_on (enc.sv); 467 SvUTF8_on (enc.sv);
427 468
428 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 469 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
429 470
430#ifdef SvPV_shrink_to_cur
431 if (enc.flags & F_SHRINK) 471 if (enc.flags & F_SHRINK)
432 SvPV_shrink_to_cur (enc.sv); 472 shrink (enc.sv);
433#endif 473
434 return enc.sv; 474 return enc.sv;
435} 475}
436 476
437///////////////////////////////////////////////////////////////////////////// 477/////////////////////////////////////////////////////////////////////////////
478// decoder
438 479
439#define WS \ 480// structure used for decoding JSON
481typedef struct
482{
483 char *cur; // current parser pointer
484 char *end; // end of input string
485 const char *err; // parse error, if != 0
486 UV flags; // F_*
487} dec_t;
488
489static void
490decode_ws (dec_t *dec)
491{
440 for (;;) \ 492 for (;;)
441 { \ 493 {
442 char ch = *dec->cur; \ 494 char ch = *dec->cur;
495
443 if (ch > 0x20 \ 496 if (ch > 0x20
444 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) \ 497 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09))
445 break; \ 498 break;
499
446 ++dec->cur; \ 500 ++dec->cur;
447 } 501 }
502}
448 503
449#define ERR(reason) SB dec->err = reason; goto fail; SE 504#define ERR(reason) SB dec->err = reason; goto fail; SE
450#define EXPECT_CH(ch) SB \ 505#define EXPECT_CH(ch) SB \
451 if (*dec->cur != ch) \ 506 if (*dec->cur != ch) \
452 ERR (# ch " expected"); \ 507 ERR (# ch " expected"); \
… …
459 514
460static UV 515static UV
461decode_4hex (dec_t *dec) 516decode_4hex (dec_t *dec)
462{ 517{
463 signed char d1, d2, d3, d4; 518 signed char d1, d2, d3, d4;
519 unsigned char *cur = (unsigned char *)dec->cur;
464 520
465 d1 = decode_hexdigit [((unsigned char *)dec->cur) [0]];
466 if (d1 < 0) ERR ("four hexadecimal digits expected"); 521 d1 = decode_hexdigit [cur [0]]; if (d1 < 0) ERR ("four hexadecimal digits expected");
467 d2 = decode_hexdigit [((unsigned char *)dec->cur) [1]];
468 if (d2 < 0) ERR ("four hexadecimal digits expected"); 522 d2 = decode_hexdigit [cur [1]]; if (d2 < 0) ERR ("four hexadecimal digits expected");
469 d3 = decode_hexdigit [((unsigned char *)dec->cur) [2]];
470 if (d3 < 0) ERR ("four hexadecimal digits expected"); 523 d3 = decode_hexdigit [cur [2]]; if (d3 < 0) ERR ("four hexadecimal digits expected");
471 d4 = decode_hexdigit [((unsigned char *)dec->cur) [3]];
472 if (d4 < 0) ERR ("four hexadecimal digits expected"); 524 d4 = decode_hexdigit [cur [3]]; if (d4 < 0) ERR ("four hexadecimal digits expected");
473 525
474 dec->cur += 4; 526 dec->cur += 4;
475 527
476 return ((UV)d1) << 12 528 return ((UV)d1) << 12
477 | ((UV)d2) << 8 529 | ((UV)d2) << 8
… …
480 532
481fail: 533fail:
482 return (UV)-1; 534 return (UV)-1;
483} 535}
484 536
485#define APPEND_GROW(n) SB \
486 if (cur + (n) >= end) \
487 { \
488 STRLEN ofs = cur - SvPVX (sv); \
489 SvGROW (sv, ofs + (n) + 1); \
490 cur = SvPVX (sv) + ofs; \
491 end = SvEND (sv); \
492 } \
493 SE
494
495#define APPEND_CH(ch) SB \
496 APPEND_GROW (1); \
497 *cur++ = (ch); \
498 SE
499
500static SV * 537static SV *
501decode_str (dec_t *dec) 538decode_str (dec_t *dec)
502{ 539{
503 SV *sv = NEWSV (0,2); 540 SV *sv = 0;
504 int utf8 = 0; 541 int utf8 = 0;
505 char *cur = SvPVX (sv);
506 char *end = SvEND (sv);
507 542
508 for (;;) 543 do
509 { 544 {
510 unsigned char ch = *(unsigned char *)dec->cur; 545 char buf [SHORT_STRING_LEN + UTF8_MAX_LEN];
546 char *cur = buf;
511 547
512 if (ch == '"') 548 do
513 break;
514 else if (ch == '\\')
515 { 549 {
516 switch (*++dec->cur) 550 unsigned char ch = *(unsigned char *)dec->cur++;
551
552 if (ch == '"')
517 { 553 {
518 case '\\': 554 --dec->cur;
519 case '/': 555 break;
520 case '"': APPEND_CH (*dec->cur++); break; 556 }
521 557 else if (ch == '\\')
522 case 'b': APPEND_CH ('\010'); ++dec->cur; break; 558 {
523 case 't': APPEND_CH ('\011'); ++dec->cur; break; 559 switch (*dec->cur)
524 case 'n': APPEND_CH ('\012'); ++dec->cur; break;
525 case 'f': APPEND_CH ('\014'); ++dec->cur; break;
526 case 'r': APPEND_CH ('\015'); ++dec->cur; break;
527
528 case 'u':
529 { 560 {
530 UV lo, hi; 561 case '\\':
531 ++dec->cur; 562 case '/':
563 case '"': *cur++ = *dec->cur++; break;
532 564
533 hi = decode_4hex (dec); 565 case 'b': ++dec->cur; *cur++ = '\010'; break;
534 if (hi == (UV)-1) 566 case 't': ++dec->cur; *cur++ = '\011'; break;
535 goto fail; 567 case 'n': ++dec->cur; *cur++ = '\012'; break;
568 case 'f': ++dec->cur; *cur++ = '\014'; break;
569 case 'r': ++dec->cur; *cur++ = '\015'; break;
536 570
537 // possibly a surrogate pair 571 case 'u':
538 if (hi >= 0xd800 && hi < 0xdc00)
539 { 572 {
540 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 573 UV lo, hi;
541 ERR ("missing low surrogate character in surrogate pair");
542
543 dec->cur += 2; 574 ++dec->cur;
544 575
545 lo = decode_4hex (dec); 576 hi = decode_4hex (dec);
546 if (lo == (UV)-1) 577 if (hi == (UV)-1)
547 goto fail; 578 goto fail;
548 579
580 // possibly a surrogate pair
581 if (hi >= 0xd800)
582 if (hi < 0xdc00)
583 {
584 if (dec->cur [0] != '\\' || dec->cur [1] != 'u')
585 ERR ("missing low surrogate character in surrogate pair");
586
587 dec->cur += 2;
588
589 lo = decode_4hex (dec);
590 if (lo == (UV)-1)
591 goto fail;
592
549 if (lo < 0xdc00 || lo >= 0xe000) 593 if (lo < 0xdc00 || lo >= 0xe000)
550 ERR ("surrogate pair expected"); 594 ERR ("surrogate pair expected");
551 595
552 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000; 596 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000;
597 }
598 else if (hi < 0xe000)
599 ERR ("missing high surrogate character in surrogate pair");
600
601 if (hi >= 0x80)
602 {
603 utf8 = 1;
604
605 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
606 }
607 else
608 *cur++ = hi;
553 } 609 }
554 else if (hi >= 0xdc00 && hi < 0xe000)
555 ERR ("missing high surrogate character in surrogate pair");
556
557 if (hi >= 0x80)
558 { 610 break;
559 utf8 = 1;
560 611
561 APPEND_GROW (4); // at most 4 bytes for 21 bits
562 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
563 }
564 else 612 default:
565 APPEND_CH (hi); 613 --dec->cur;
614 ERR ("illegal backslash escape sequence in string");
566 } 615 }
567 break;
568
569 default:
570 --dec->cur;
571 ERR ("illegal backslash escape sequence in string");
572 } 616 }
617 else if (ch >= 0x20 && ch <= 0x7f)
618 *cur++ = ch;
619 else if (ch >= 0x80)
620 {
621 --dec->cur;
622
623 STRLEN clen;
624 UV uch = utf8n_to_uvuni (dec->cur, dec->end - dec->cur, &clen, UTF8_CHECK_ONLY);
625 if (clen == (STRLEN)-1)
626 ERR ("malformed UTF-8 character in JSON string");
627
628 do
629 {
630 *cur++ = *dec->cur++;
631 }
632 while (--clen);
633
634 utf8 = 1;
635 }
636 else if (!ch)
637 ERR ("unexpected end of string while parsing json string");
638 else
639 ERR ("invalid character encountered");
640
573 } 641 }
574 else if (ch >= 0x20 && ch <= 0x7f) 642 while (cur < buf + SHORT_STRING_LEN);
575 APPEND_CH (*dec->cur++); 643
576 else if (ch >= 0x80) 644 STRLEN len = cur - buf;
645
646 if (sv)
577 { 647 {
578 STRLEN clen; 648 SvGROW (sv, SvCUR (sv) + len + 1);
579 UV uch = utf8n_to_uvuni (dec->cur, dec->end - dec->cur, &clen, UTF8_CHECK_ONLY); 649 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
580 if (clen == (STRLEN)-1) 650 SvCUR_set (sv, SvCUR (sv) + len);
581 ERR ("malformed UTF-8 character in string, cannot convert to JSON");
582
583 APPEND_GROW (clen);
584 do
585 {
586 *cur++ = *dec->cur++;
587 }
588 while (--clen);
589
590 utf8 = 1;
591 } 651 }
592 else if (dec->cur == dec->end)
593 ERR ("unexpected end of string while parsing json string");
594 else 652 else
595 ERR ("invalid character encountered"); 653 sv = newSVpvn (buf, len);
596 } 654 }
655 while (*dec->cur != '"');
597 656
598 ++dec->cur; 657 ++dec->cur;
599 658
600 SvCUR_set (sv, cur - SvPVX (sv)); 659 if (sv)
601 660 {
602 SvPOK_only (sv); 661 SvPOK_only (sv);
603 *SvEND (sv) = 0; 662 *SvEND (sv) = 0;
604 663
605 if (utf8) 664 if (utf8)
606 SvUTF8_on (sv); 665 SvUTF8_on (sv);
607 666 }
608#ifdef SvPV_shrink_to_cur 667 else
609 if (dec->flags & F_SHRINK) 668 sv = newSVpvn ("", 0);
610 SvPV_shrink_to_cur (sv);
611#endif
612 669
613 return sv; 670 return sv;
614 671
615fail: 672fail:
616 SvREFCNT_dec (sv);
617 return 0; 673 return 0;
618} 674}
619 675
620static SV * 676static SV *
621decode_num (dec_t *dec) 677decode_num (dec_t *dec)
… …
702static SV * 758static SV *
703decode_av (dec_t *dec) 759decode_av (dec_t *dec)
704{ 760{
705 AV *av = newAV (); 761 AV *av = newAV ();
706 762
707 WS; 763 decode_ws (dec);
708 if (*dec->cur == ']') 764 if (*dec->cur == ']')
709 ++dec->cur; 765 ++dec->cur;
710 else 766 else
711 for (;;) 767 for (;;)
712 { 768 {
… …
716 if (!value) 772 if (!value)
717 goto fail; 773 goto fail;
718 774
719 av_push (av, value); 775 av_push (av, value);
720 776
721 WS; 777 decode_ws (dec);
722 778
723 if (*dec->cur == ']') 779 if (*dec->cur == ']')
724 { 780 {
725 ++dec->cur; 781 ++dec->cur;
726 break; 782 break;
… …
742static SV * 798static SV *
743decode_hv (dec_t *dec) 799decode_hv (dec_t *dec)
744{ 800{
745 HV *hv = newHV (); 801 HV *hv = newHV ();
746 802
747 WS; 803 decode_ws (dec);
748 if (*dec->cur == '}') 804 if (*dec->cur == '}')
749 ++dec->cur; 805 ++dec->cur;
750 else 806 else
751 for (;;) 807 for (;;)
752 { 808 {
753 SV *key, *value; 809 SV *key, *value;
754 810
755 WS; EXPECT_CH ('"'); 811 decode_ws (dec); EXPECT_CH ('"');
756 812
757 key = decode_str (dec); 813 key = decode_str (dec);
758 if (!key) 814 if (!key)
759 goto fail; 815 goto fail;
760 816
761 WS; EXPECT_CH (':'); 817 decode_ws (dec); EXPECT_CH (':');
762 818
763 value = decode_sv (dec); 819 value = decode_sv (dec);
764 if (!value) 820 if (!value)
765 { 821 {
766 SvREFCNT_dec (key); 822 SvREFCNT_dec (key);
… …
768 } 824 }
769 825
770 //TODO: optimise 826 //TODO: optimise
771 hv_store_ent (hv, key, value, 0); 827 hv_store_ent (hv, key, value, 0);
772 828
773 WS; 829 decode_ws (dec);
774 830
775 if (*dec->cur == '}') 831 if (*dec->cur == '}')
776 { 832 {
777 ++dec->cur; 833 ++dec->cur;
778 break; 834 break;
… …
792} 848}
793 849
794static SV * 850static SV *
795decode_sv (dec_t *dec) 851decode_sv (dec_t *dec)
796{ 852{
797 WS; 853 decode_ws (dec);
798 switch (*dec->cur) 854 switch (*dec->cur)
799 { 855 {
800 case '"': ++dec->cur; return decode_str (dec); 856 case '"': ++dec->cur; return decode_str (dec);
801 case '[': ++dec->cur; return decode_av (dec); 857 case '[': ++dec->cur; return decode_av (dec);
802 case '{': ++dec->cur; return decode_hv (dec); 858 case '{': ++dec->cur; return decode_hv (dec);
… …
838 ERR ("'null' expected"); 894 ERR ("'null' expected");
839 895
840 break; 896 break;
841 897
842 default: 898 default:
843 ERR ("malformed json string"); 899 ERR ("malformed json string, neither array, object, number, string or atom");
844 break; 900 break;
845 } 901 }
846 902
847fail: 903fail:
848 return 0; 904 return 0;
… …
868 924
869 sv = decode_sv (&dec); 925 sv = decode_sv (&dec);
870 926
871 if (!sv) 927 if (!sv)
872 { 928 {
929 IV offset = dec.flags & F_UTF8
930 ? dec.cur - SvPVX (string)
873 IV offset = utf8_distance (dec.cur, SvPVX (string)); 931 : utf8_distance (dec.cur, SvPVX (string));
874 SV *uni = sv_newmortal (); 932 SV *uni = sv_newmortal ();
933
875 // horrible hack to silence warning inside pv_uni_display 934 // horrible hack to silence warning inside pv_uni_display
876 COP cop; 935 COP cop = *PL_curcop;
877 memset (&cop, 0, sizeof (cop));
878 cop.cop_warnings = pWARN_NONE; 936 cop.cop_warnings = pWARN_NONE;
937 ENTER;
879 SAVEVPTR (PL_curcop); 938 SAVEVPTR (PL_curcop);
880 PL_curcop = &cop; 939 PL_curcop = &cop;
881
882 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 940 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
941 LEAVE;
942
883 croak ("%s, at character offset %d (%s)", 943 croak ("%s, at character offset %d (%s)",
884 dec.err, 944 dec.err,
885 (int)offset, 945 (int)offset,
886 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 946 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
887 } 947 }
888 948
889 sv = sv_2mortal (sv); 949 sv = sv_2mortal (sv);
890 950
891 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 951 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv))
892 croak ("JSON object or array expected (but number, string, true, false or null found, use allow_nonref to allow this)"); 952 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
893 953
894 return sv; 954 return sv;
895} 955}
956
957/////////////////////////////////////////////////////////////////////////////
958// XS interface functions
896 959
897MODULE = JSON::XS PACKAGE = JSON::XS 960MODULE = JSON::XS PACKAGE = JSON::XS
898 961
899BOOT: 962BOOT:
900{ 963{
… …
927 utf8 = F_UTF8 990 utf8 = F_UTF8
928 indent = F_INDENT 991 indent = F_INDENT
929 canonical = F_CANONICAL 992 canonical = F_CANONICAL
930 space_before = F_SPACE_BEFORE 993 space_before = F_SPACE_BEFORE
931 space_after = F_SPACE_AFTER 994 space_after = F_SPACE_AFTER
932 json_rpc = F_JSON_RPC
933 pretty = F_PRETTY 995 pretty = F_PRETTY
934 allow_nonref = F_ALLOW_NONREF 996 allow_nonref = F_ALLOW_NONREF
935 shrink = F_SHRINK 997 shrink = F_SHRINK
936 CODE: 998 CODE:
937{ 999{

Diff Legend

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