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 1.16 by root, Sun Mar 25 02:37:00 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
18// F_SKIPINVALID?
19// F_EXECCODEREF?
20// F_SELFCONVERT?
21
19#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 22#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
20#define F_DEFAULT 0 23#define F_DEFAULT 0
21 24
22#define INIT_SIZE 32 // initial scalar size to be allocated 25#define INIT_SIZE 32 // initial scalar size to be allocated
26#define INDENT_STEP 3 // spaces per indentation level
27
28#define UTF8_MAX_LEN 11 // for perls UTF-X: max. number of octets per character
29#define SHORT_STRING_LEN 512 // special-case strings of up to this size
23 30
24#define SB do { 31#define SB do {
25#define SE } while (0) 32#define SE } while (0)
26 33
27static HV *json_stash; 34static HV *json_stash; // JSON::XS::
35
36/////////////////////////////////////////////////////////////////////////////
37// utility functions
38
39static UV *
40SvJSON (SV *sv)
41{
42 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash))
43 croak ("object is not of type JSON::XS");
44
45 return &SvUVX (SvRV (sv));
46}
47
48static void
49shrink (SV *sv)
50{
51 sv_utf8_downgrade (sv, 1);
52 if (SvLEN (sv) > SvCUR (sv) + 1)
53 {
54#ifdef SvPV_shrink_to_cur
55 SvPV_shrink_to_cur (sv);
56#elif defined (SvPV_renew)
57 SvPV_renew (sv, SvCUR (sv) + 1);
58#endif
59 }
60}
61
62// decode an utf-8 character and return it, or (UV)-1 in
63// case of an error.
64// we special-case "safe" characters from U+80 .. U+7FF,
65// but use the very good perl function to parse anything else.
66// note that we never call this function for a ascii codepoints
67static UV
68decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
69{
70 if (s[0] > 0xdf || s[0] < 0xc2)
71 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
72 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)
73 {
74 *clen = 2;
75 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
76 }
77 else
78 return (UV)-1;
79}
80
81/////////////////////////////////////////////////////////////////////////////
82// encoder
28 83
29// structure used for encoding JSON 84// structure used for encoding JSON
30typedef struct 85typedef struct
31{ 86{
32 char *cur; 87 char *cur; // SvPVX (sv) + current output position
33 STRLEN len; // SvLEN (sv)
34 char *end; // SvEND (sv) 88 char *end; // SvEND (sv)
35 SV *sv; 89 SV *sv; // result scalar
36 UV flags; 90 UV flags; // F_*
37 int max_recurse; 91 int indent; // indentation level
38 int indent; 92 int max_depth; // max. recursion level
39} enc_t; 93} 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 94
61static void 95static void
62need (enc_t *enc, STRLEN len) 96need (enc_t *enc, STRLEN len)
63{ 97{
64 if (enc->cur + len >= enc->end) 98 if (enc->cur + len >= enc->end)
… …
122 STRLEN clen; 156 STRLEN clen;
123 UV uch; 157 UV uch;
124 158
125 if (is_utf8) 159 if (is_utf8)
126 { 160 {
127 uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY); 161 //uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
162 uch = decode_utf8 (str, end - str, &clen);
128 if (clen == (STRLEN)-1) 163 if (clen == (STRLEN)-1)
129 croak ("malformed UTF-8 character in string, cannot convert to JSON"); 164 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
130 } 165 }
131 else 166 else
132 { 167 {
133 uch = ch; 168 uch = ch;
134 clen = 1; 169 clen = 1;
135 } 170 }
136 171
172 if (uch > 0x10FFFFUL)
173 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
174
137 if (uch < 0x80 || enc->flags & F_ASCII) 175 if (uch < 0x80 || enc->flags & F_ASCII)
138 { 176 {
139 if (uch > 0xFFFFUL) 177 if (uch > 0xFFFFUL)
140 { 178 {
141 need (enc, len += 11); 179 need (enc, len += 11);
142 sprintf (enc->cur, "\\u%04x\\u%04x", 180 sprintf (enc->cur, "\\u%04x\\u%04x",
143 (uch - 0x10000) / 0x400 + 0xD800, 181 (int)((uch - 0x10000) / 0x400 + 0xD800),
144 (uch - 0x10000) % 0x400 + 0xDC00); 182 (int)((uch - 0x10000) % 0x400 + 0xDC00));
145 enc->cur += 12; 183 enc->cur += 12;
146 } 184 }
147 else 185 else
148 { 186 {
149 static char hexdigit [16] = "0123456789abcdef"; 187 static char hexdigit [16] = "0123456789abcdef";
… …
167 } 205 }
168 while (--clen); 206 while (--clen);
169 } 207 }
170 else 208 else
171 { 209 {
172 need (enc, 10); // never more than 11 bytes needed 210 need (enc, len += UTF8_MAX_LEN - 1); // never more than 11 bytes needed
173 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 211 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
174 ++str; 212 ++str;
175 } 213 }
176 } 214 }
177 } 215 }
… …
179 217
180 --len; 218 --len;
181 } 219 }
182} 220}
183 221
184#define INDENT SB \ 222static void
223encode_indent (enc_t *enc)
224{
185 if (enc->flags & F_INDENT) \ 225 if (enc->flags & F_INDENT)
186 { \ 226 {
187 int i_; \ 227 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
193 228
194#define SPACE SB need (enc, 1); encode_ch (enc, ' '); SE 229 need (enc, spaces);
195#define NL SB if (enc->flags & F_INDENT) { need (enc, 1); encode_ch (enc, '\n'); } SE 230 memset (enc->cur, ' ', spaces);
196#define COMMA SB \ 231 enc->cur += spaces;
232 }
233}
234
235static void
236encode_space (enc_t *enc)
237{
238 need (enc, 1);
197 encode_ch (enc, ','); \ 239 encode_ch (enc, ' ');
240}
241
242static void
243encode_nl (enc_t *enc)
244{
198 if (enc->flags & F_INDENT) \ 245 if (enc->flags & F_INDENT)
199 NL; \ 246 {
247 need (enc, 1);
248 encode_ch (enc, '\n');
249 }
250}
251
252static void
253encode_comma (enc_t *enc)
254{
255 encode_ch (enc, ',');
256
257 if (enc->flags & F_INDENT)
258 encode_nl (enc);
200 else if (enc->flags & F_SPACE_AFTER) \ 259 else if (enc->flags & F_SPACE_AFTER)
201 SPACE; \ 260 encode_space (enc);
202 SE 261}
203 262
204static void encode_sv (enc_t *enc, SV *sv); 263static void encode_sv (enc_t *enc, SV *sv);
205 264
206static void 265static void
207encode_av (enc_t *enc, AV *av) 266encode_av (enc_t *enc, AV *av)
208{ 267{
209 int i, len = av_len (av); 268 int i, len = av_len (av);
210 269
211 encode_ch (enc, '['); NL; 270 encode_ch (enc, '['); encode_nl (enc);
212 ++enc->indent; 271 ++enc->indent;
213 272
214 for (i = 0; i <= len; ++i) 273 for (i = 0; i <= len; ++i)
215 { 274 {
216 INDENT; 275 encode_indent (enc);
217 encode_sv (enc, *av_fetch (av, i, 0)); 276 encode_sv (enc, *av_fetch (av, i, 0));
218 277
219 if (i < len) 278 if (i < len)
220 COMMA; 279 encode_comma (enc);
221 } 280 }
222 281
223 NL; 282 encode_nl (enc);
224 283
225 --enc->indent; 284 --enc->indent;
226 INDENT; encode_ch (enc, ']'); 285 encode_indent (enc); encode_ch (enc, ']');
227} 286}
228 287
229static void 288static void
230encode_he (enc_t *enc, HE *he) 289encode_he (enc_t *enc, HE *he)
231{ 290{
… …
245 else 304 else
246 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 305 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
247 306
248 encode_ch (enc, '"'); 307 encode_ch (enc, '"');
249 308
250 if (enc->flags & F_SPACE_BEFORE) SPACE; 309 if (enc->flags & F_SPACE_BEFORE) encode_space (enc);
251 encode_ch (enc, ':'); 310 encode_ch (enc, ':');
252 if (enc->flags & F_SPACE_AFTER ) SPACE; 311 if (enc->flags & F_SPACE_AFTER ) encode_space (enc);
253 encode_sv (enc, HeVAL (he)); 312 encode_sv (enc, HeVAL (he));
254} 313}
255 314
256// compare hash entries, used when all keys are bytestrings 315// compare hash entries, used when all keys are bytestrings
257static int 316static int
… …
263 HE *b = *(HE **)b_; 322 HE *b = *(HE **)b_;
264 323
265 STRLEN la = HeKLEN (a); 324 STRLEN la = HeKLEN (a);
266 STRLEN lb = HeKLEN (b); 325 STRLEN lb = HeKLEN (b);
267 326
268 if (!(cmp == memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 327 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb)))
269 cmp = la < lb ? -1 : la == lb ? 0 : 1; 328 cmp = la - lb;
270 329
271 return cmp; 330 return cmp;
272} 331}
273 332
274// compare hash entries, used when some keys are sv's or utf-x 333// compare hash entries, used when some keys are sv's or utf-x
… …
281static void 340static void
282encode_hv (enc_t *enc, HV *hv) 341encode_hv (enc_t *enc, HV *hv)
283{ 342{
284 int count, i; 343 int count, i;
285 344
286 encode_ch (enc, '{'); NL; ++enc->indent; 345 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
287 346
288 if ((count = hv_iterinit (hv))) 347 if ((count = hv_iterinit (hv)))
289 { 348 {
290 // for canonical output we have to sort by keys first 349 // for canonical output we have to sort by keys first
291 // actually, this is mostly due to the stupid so-called 350 // actually, this is mostly due to the stupid so-called
292 // security workaround added somewhere in 5.8.x. 351 // security workaround added somewhere in 5.8.x.
293 // that randomises hash orderings 352 // that randomises hash orderings
294 if (enc->flags & F_CANONICAL) 353 if (enc->flags & F_CANONICAL)
295 { 354 {
296 HE *he, *hes [count]; 355 HE *he, *hes [count]; // if your compiler dies here, you need to enable C99 mode
297 int fast = 1; 356 int fast = 1;
298 357
299 i = 0; 358 i = 0;
300 while ((he = hv_iternext (hv))) 359 while ((he = hv_iternext (hv)))
301 { 360 {
… …
308 367
309 if (fast) 368 if (fast)
310 qsort (hes, count, sizeof (HE *), he_cmp_fast); 369 qsort (hes, count, sizeof (HE *), he_cmp_fast);
311 else 370 else
312 { 371 {
313 // hack to disable "use bytes" 372 // hack to forcefully disable "use bytes"
314 COP *oldcop = PL_curcop, cop; 373 COP cop = *PL_curcop;
315 cop.op_private = 0; 374 cop.op_private = 0;
375
376 ENTER;
377 SAVETMPS;
378
379 SAVEVPTR (PL_curcop);
316 PL_curcop = &cop; 380 PL_curcop = &cop;
317 381
318 SAVETMPS;
319 qsort (hes, count, sizeof (HE *), he_cmp_slow); 382 qsort (hes, count, sizeof (HE *), he_cmp_slow);
383
320 FREETMPS; 384 FREETMPS;
321 385 LEAVE;
322 PL_curcop = oldcop;
323 } 386 }
324 387
325 for (i = 0; i < count; ++i) 388 for (i = 0; i < count; ++i)
326 { 389 {
327 INDENT; 390 encode_indent (enc);
328 encode_he (enc, hes [i]); 391 encode_he (enc, hes [i]);
329 392
330 if (i < count - 1) 393 if (i < count - 1)
331 COMMA; 394 encode_comma (enc);
332 } 395 }
333 396
334 NL; 397 encode_nl (enc);
335 } 398 }
336 else 399 else
337 { 400 {
338 SV *sv; 401 SV *sv;
339 HE *he = hv_iternext (hv); 402 HE *he = hv_iternext (hv);
340 403
341 for (;;) 404 for (;;)
342 { 405 {
343 INDENT; 406 encode_indent (enc);
344 encode_he (enc, he); 407 encode_he (enc, he);
345 408
346 if (!(he = hv_iternext (hv))) 409 if (!(he = hv_iternext (hv)))
347 break; 410 break;
348 411
349 COMMA; 412 encode_comma (enc);
350 } 413 }
351 414
352 NL; 415 encode_nl (enc);
353 } 416 }
354 } 417 }
355 418
356 --enc->indent; INDENT; encode_ch (enc, '}'); 419 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
357} 420}
358 421
359static void 422static void
360encode_sv (enc_t *enc, SV *sv) 423encode_sv (enc_t *enc, SV *sv)
361{ 424{
… …
383 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 446 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv))
384 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 447 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv));
385 } 448 }
386 else if (SvROK (sv)) 449 else if (SvROK (sv))
387 { 450 {
388 if (!--enc->max_recurse) 451 SV *rv = SvRV (sv);
452
453 if (enc->indent >= enc->max_depth)
389 croak ("data structure too deep (hit recursion limit)"); 454 croak ("data structure too deep (hit recursion limit)");
390 455
391 sv = SvRV (sv);
392
393 switch (SvTYPE (sv)) 456 switch (SvTYPE (rv))
394 { 457 {
395 case SVt_PVAV: encode_av (enc, (AV *)sv); break; 458 case SVt_PVAV: encode_av (enc, (AV *)rv); break;
396 case SVt_PVHV: encode_hv (enc, (HV *)sv); break; 459 case SVt_PVHV: encode_hv (enc, (HV *)rv); break;
397 460
398 default: 461 default:
399 croak ("JSON can only represent references to arrays or hashes"); 462 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
463 SvPV_nolen (sv));
400 } 464 }
401 } 465 }
402 else if (!SvOK (sv)) 466 else if (!SvOK (sv))
403 encode_str (enc, "null", 4, 0); 467 encode_str (enc, "null", 4, 0);
404 else 468 else
405 croak ("encountered perl type that JSON cannot handle"); 469 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
470 SvPV_nolen (sv), SvFLAGS (sv));
406} 471}
407 472
408static SV * 473static SV *
409encode_json (SV *scalar, UV flags) 474encode_json (SV *scalar, UV flags)
410{ 475{
411 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 476 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar))
412 croak ("hash- or arraref required (not a simple scalar, use allow_nonref to allow this)"); 477 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
413 478
414 enc_t enc; 479 enc_t enc;
415 enc.flags = flags; 480 enc.flags = flags;
416 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 481 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
417 enc.cur = SvPVX (enc.sv); 482 enc.cur = SvPVX (enc.sv);
418 enc.end = SvEND (enc.sv); 483 enc.end = SvEND (enc.sv);
419 enc.max_recurse = 0;
420 enc.indent = 0; 484 enc.indent = 0;
485 enc.max_depth = 0x7fffffffUL;
421 486
422 SvPOK_only (enc.sv); 487 SvPOK_only (enc.sv);
423 encode_sv (&enc, scalar); 488 encode_sv (&enc, scalar);
424 489
425 if (!(flags & (F_ASCII | F_UTF8))) 490 if (!(flags & (F_ASCII | F_UTF8)))
426 SvUTF8_on (enc.sv); 491 SvUTF8_on (enc.sv);
427 492
428 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 493 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
429 494
430#ifdef SvPV_shrink_to_cur
431 if (enc.flags & F_SHRINK) 495 if (enc.flags & F_SHRINK)
432 SvPV_shrink_to_cur (enc.sv); 496 shrink (enc.sv);
433#endif 497
434 return enc.sv; 498 return enc.sv;
435} 499}
436 500
437///////////////////////////////////////////////////////////////////////////// 501/////////////////////////////////////////////////////////////////////////////
502// decoder
438 503
439#define WS \ 504// structure used for decoding JSON
505typedef struct
506{
507 char *cur; // current parser pointer
508 char *end; // end of input string
509 const char *err; // parse error, if != 0
510 UV flags; // F_*
511} dec_t;
512
513static void
514decode_ws (dec_t *dec)
515{
440 for (;;) \ 516 for (;;)
441 { \ 517 {
442 char ch = *dec->cur; \ 518 char ch = *dec->cur;
519
443 if (ch > 0x20 \ 520 if (ch > 0x20
444 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) \ 521 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09))
445 break; \ 522 break;
523
446 ++dec->cur; \ 524 ++dec->cur;
447 } 525 }
526}
448 527
449#define ERR(reason) SB dec->err = reason; goto fail; SE 528#define ERR(reason) SB dec->err = reason; goto fail; SE
450#define EXPECT_CH(ch) SB \ 529#define EXPECT_CH(ch) SB \
451 if (*dec->cur != ch) \ 530 if (*dec->cur != ch) \
452 ERR (# ch " expected"); \ 531 ERR (# ch " expected"); \
… …
459 538
460static UV 539static UV
461decode_4hex (dec_t *dec) 540decode_4hex (dec_t *dec)
462{ 541{
463 signed char d1, d2, d3, d4; 542 signed char d1, d2, d3, d4;
543 unsigned char *cur = (unsigned char *)dec->cur;
464 544
465 d1 = decode_hexdigit [((unsigned char *)dec->cur) [0]];
466 if (d1 < 0) ERR ("four hexadecimal digits expected"); 545 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"); 546 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"); 547 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"); 548 d4 = decode_hexdigit [cur [3]]; if (d4 < 0) ERR ("four hexadecimal digits expected");
473 549
474 dec->cur += 4; 550 dec->cur += 4;
475 551
476 return ((UV)d1) << 12 552 return ((UV)d1) << 12
477 | ((UV)d2) << 8 553 | ((UV)d2) << 8
… …
480 556
481fail: 557fail:
482 return (UV)-1; 558 return (UV)-1;
483} 559}
484 560
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 * 561static SV *
501decode_str (dec_t *dec) 562decode_str (dec_t *dec)
502{ 563{
503 SV *sv = NEWSV (0,2); 564 SV *sv = 0;
504 int utf8 = 0; 565 int utf8 = 0;
505 char *cur = SvPVX (sv);
506 char *end = SvEND (sv);
507 566
508 for (;;) 567 do
509 { 568 {
510 unsigned char ch = *(unsigned char *)dec->cur; 569 char buf [SHORT_STRING_LEN + UTF8_MAX_LEN];
570 char *cur = buf;
511 571
512 if (ch == '"') 572 do
513 break;
514 else if (ch == '\\')
515 { 573 {
516 switch (*++dec->cur) 574 unsigned char ch = *(unsigned char *)dec->cur++;
575
576 if (ch == '"')
517 { 577 {
518 case '\\': 578 --dec->cur;
519 case '/': 579 break;
520 case '"': APPEND_CH (*dec->cur++); break; 580 }
521 581 else if (ch == '\\')
522 case 'b': APPEND_CH ('\010'); ++dec->cur; break; 582 {
523 case 't': APPEND_CH ('\011'); ++dec->cur; break; 583 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 { 584 {
530 UV lo, hi; 585 case '\\':
531 ++dec->cur; 586 case '/':
587 case '"': *cur++ = *dec->cur++; break;
532 588
533 hi = decode_4hex (dec); 589 case 'b': ++dec->cur; *cur++ = '\010'; break;
534 if (hi == (UV)-1) 590 case 't': ++dec->cur; *cur++ = '\011'; break;
535 goto fail; 591 case 'n': ++dec->cur; *cur++ = '\012'; break;
592 case 'f': ++dec->cur; *cur++ = '\014'; break;
593 case 'r': ++dec->cur; *cur++ = '\015'; break;
536 594
537 // possibly a surrogate pair 595 case 'u':
538 if (hi >= 0xd800 && hi < 0xdc00)
539 { 596 {
540 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 597 UV lo, hi;
541 ERR ("missing low surrogate character in surrogate pair");
542
543 dec->cur += 2; 598 ++dec->cur;
544 599
545 lo = decode_4hex (dec); 600 hi = decode_4hex (dec);
546 if (lo == (UV)-1) 601 if (hi == (UV)-1)
547 goto fail; 602 goto fail;
548 603
604 // possibly a surrogate pair
605 if (hi >= 0xd800)
606 if (hi < 0xdc00)
607 {
608 if (dec->cur [0] != '\\' || dec->cur [1] != 'u')
609 ERR ("missing low surrogate character in surrogate pair");
610
611 dec->cur += 2;
612
613 lo = decode_4hex (dec);
614 if (lo == (UV)-1)
615 goto fail;
616
549 if (lo < 0xdc00 || lo >= 0xe000) 617 if (lo < 0xdc00 || lo >= 0xe000)
550 ERR ("surrogate pair expected"); 618 ERR ("surrogate pair expected");
551 619
552 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000; 620 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000;
621 }
622 else if (hi < 0xe000)
623 ERR ("missing high surrogate character in surrogate pair");
624
625 if (hi >= 0x80)
626 {
627 utf8 = 1;
628
629 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
630 }
631 else
632 *cur++ = hi;
553 } 633 }
554 else if (hi >= 0xdc00 && hi < 0xe000)
555 ERR ("missing high surrogate character in surrogate pair");
556
557 if (hi >= 0x80)
558 { 634 break;
559 utf8 = 1;
560 635
561 APPEND_GROW (4); // at most 4 bytes for 21 bits
562 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
563 }
564 else 636 default:
565 APPEND_CH (hi); 637 --dec->cur;
638 ERR ("illegal backslash escape sequence in string");
566 } 639 }
567 break;
568
569 default:
570 --dec->cur;
571 ERR ("illegal backslash escape sequence in string");
572 } 640 }
641 else if (ch >= 0x20 && ch <= 0x7f)
642 *cur++ = ch;
643 else if (ch >= 0x80)
644 {
645 --dec->cur;
646
647 STRLEN clen;
648 UV uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen);
649 if (clen == (STRLEN)-1)
650 ERR ("malformed UTF-8 character in JSON string");
651
652 do
653 {
654 *cur++ = *dec->cur++;
655 }
656 while (--clen);
657
658 utf8 = 1;
659 }
660 else if (!ch)
661 ERR ("unexpected end of string while parsing json string");
662 else
663 ERR ("invalid character encountered");
664
573 } 665 }
574 else if (ch >= 0x20 && ch <= 0x7f) 666 while (cur < buf + SHORT_STRING_LEN);
575 APPEND_CH (*dec->cur++); 667
576 else if (ch >= 0x80) 668 STRLEN len = cur - buf;
669
670 if (sv)
577 { 671 {
578 STRLEN clen; 672 SvGROW (sv, SvCUR (sv) + len + 1);
579 UV uch = utf8n_to_uvuni (dec->cur, dec->end - dec->cur, &clen, UTF8_CHECK_ONLY); 673 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
580 if (clen == (STRLEN)-1) 674 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 } 675 }
592 else if (dec->cur == dec->end)
593 ERR ("unexpected end of string while parsing json string");
594 else 676 else
595 ERR ("invalid character encountered"); 677 sv = newSVpvn (buf, len);
596 } 678 }
679 while (*dec->cur != '"');
597 680
598 ++dec->cur; 681 ++dec->cur;
599 682
600 SvCUR_set (sv, cur - SvPVX (sv)); 683 if (sv)
601 684 {
602 SvPOK_only (sv); 685 SvPOK_only (sv);
603 *SvEND (sv) = 0; 686 *SvEND (sv) = 0;
604 687
605 if (utf8) 688 if (utf8)
606 SvUTF8_on (sv); 689 SvUTF8_on (sv);
607 690 }
608#ifdef SvPV_shrink_to_cur 691 else
609 if (dec->flags & F_SHRINK) 692 sv = newSVpvn ("", 0);
610 SvPV_shrink_to_cur (sv);
611#endif
612 693
613 return sv; 694 return sv;
614 695
615fail: 696fail:
616 SvREFCNT_dec (sv);
617 return 0; 697 return 0;
618} 698}
619 699
620static SV * 700static SV *
621decode_num (dec_t *dec) 701decode_num (dec_t *dec)
… …
702static SV * 782static SV *
703decode_av (dec_t *dec) 783decode_av (dec_t *dec)
704{ 784{
705 AV *av = newAV (); 785 AV *av = newAV ();
706 786
707 WS; 787 decode_ws (dec);
708 if (*dec->cur == ']') 788 if (*dec->cur == ']')
709 ++dec->cur; 789 ++dec->cur;
710 else 790 else
711 for (;;) 791 for (;;)
712 { 792 {
… …
716 if (!value) 796 if (!value)
717 goto fail; 797 goto fail;
718 798
719 av_push (av, value); 799 av_push (av, value);
720 800
721 WS; 801 decode_ws (dec);
722 802
723 if (*dec->cur == ']') 803 if (*dec->cur == ']')
724 { 804 {
725 ++dec->cur; 805 ++dec->cur;
726 break; 806 break;
… …
742static SV * 822static SV *
743decode_hv (dec_t *dec) 823decode_hv (dec_t *dec)
744{ 824{
745 HV *hv = newHV (); 825 HV *hv = newHV ();
746 826
747 WS; 827 decode_ws (dec);
748 if (*dec->cur == '}') 828 if (*dec->cur == '}')
749 ++dec->cur; 829 ++dec->cur;
750 else 830 else
751 for (;;) 831 for (;;)
752 { 832 {
753 SV *key, *value; 833 SV *key, *value;
754 834
755 WS; EXPECT_CH ('"'); 835 decode_ws (dec); EXPECT_CH ('"');
756 836
757 key = decode_str (dec); 837 key = decode_str (dec);
758 if (!key) 838 if (!key)
759 goto fail; 839 goto fail;
760 840
761 WS; EXPECT_CH (':'); 841 decode_ws (dec); EXPECT_CH (':');
762 842
763 value = decode_sv (dec); 843 value = decode_sv (dec);
764 if (!value) 844 if (!value)
765 { 845 {
766 SvREFCNT_dec (key); 846 SvREFCNT_dec (key);
… …
768 } 848 }
769 849
770 //TODO: optimise 850 //TODO: optimise
771 hv_store_ent (hv, key, value, 0); 851 hv_store_ent (hv, key, value, 0);
772 852
773 WS; 853 decode_ws (dec);
774 854
775 if (*dec->cur == '}') 855 if (*dec->cur == '}')
776 { 856 {
777 ++dec->cur; 857 ++dec->cur;
778 break; 858 break;
… …
792} 872}
793 873
794static SV * 874static SV *
795decode_sv (dec_t *dec) 875decode_sv (dec_t *dec)
796{ 876{
797 WS; 877 decode_ws (dec);
798 switch (*dec->cur) 878 switch (*dec->cur)
799 { 879 {
800 case '"': ++dec->cur; return decode_str (dec); 880 case '"': ++dec->cur; return decode_str (dec);
801 case '[': ++dec->cur; return decode_av (dec); 881 case '[': ++dec->cur; return decode_av (dec);
802 case '{': ++dec->cur; return decode_hv (dec); 882 case '{': ++dec->cur; return decode_hv (dec);
… …
838 ERR ("'null' expected"); 918 ERR ("'null' expected");
839 919
840 break; 920 break;
841 921
842 default: 922 default:
843 ERR ("malformed json string"); 923 ERR ("malformed json string, neither array, object, number, string or atom");
844 break; 924 break;
845 } 925 }
846 926
847fail: 927fail:
848 return 0; 928 return 0;
… …
868 948
869 sv = decode_sv (&dec); 949 sv = decode_sv (&dec);
870 950
871 if (!sv) 951 if (!sv)
872 { 952 {
953 IV offset = dec.flags & F_UTF8
954 ? dec.cur - SvPVX (string)
873 IV offset = utf8_distance (dec.cur, SvPVX (string)); 955 : utf8_distance (dec.cur, SvPVX (string));
874 SV *uni = sv_newmortal (); 956 SV *uni = sv_newmortal ();
957
875 // horrible hack to silence warning inside pv_uni_display 958 // horrible hack to silence warning inside pv_uni_display
876 COP cop; 959 COP cop = *PL_curcop;
877 memset (&cop, 0, sizeof (cop));
878 cop.cop_warnings = pWARN_NONE; 960 cop.cop_warnings = pWARN_NONE;
961 ENTER;
879 SAVEVPTR (PL_curcop); 962 SAVEVPTR (PL_curcop);
880 PL_curcop = &cop; 963 PL_curcop = &cop;
881
882 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 964 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
965 LEAVE;
966
883 croak ("%s, at character offset %d (%s)", 967 croak ("%s, at character offset %d (%s)",
884 dec.err, 968 dec.err,
885 (int)offset, 969 (int)offset,
886 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 970 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
887 } 971 }
888 972
889 sv = sv_2mortal (sv); 973 sv = sv_2mortal (sv);
890 974
891 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 975 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)"); 976 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
893 977
894 return sv; 978 return sv;
895} 979}
980
981/////////////////////////////////////////////////////////////////////////////
982// XS interface functions
896 983
897MODULE = JSON::XS PACKAGE = JSON::XS 984MODULE = JSON::XS PACKAGE = JSON::XS
898 985
899BOOT: 986BOOT:
900{ 987{
… …
927 utf8 = F_UTF8 1014 utf8 = F_UTF8
928 indent = F_INDENT 1015 indent = F_INDENT
929 canonical = F_CANONICAL 1016 canonical = F_CANONICAL
930 space_before = F_SPACE_BEFORE 1017 space_before = F_SPACE_BEFORE
931 space_after = F_SPACE_AFTER 1018 space_after = F_SPACE_AFTER
932 json_rpc = F_JSON_RPC
933 pretty = F_PRETTY 1019 pretty = F_PRETTY
934 allow_nonref = F_ALLOW_NONREF 1020 allow_nonref = F_ALLOW_NONREF
935 shrink = F_SHRINK 1021 shrink = F_SHRINK
936 CODE: 1022 CODE:
937{ 1023{
… …
955 XPUSHs (decode_json (jsonstr, *SvJSON (self))); 1041 XPUSHs (decode_json (jsonstr, *SvJSON (self)));
956 1042
957PROTOTYPES: ENABLE 1043PROTOTYPES: ENABLE
958 1044
959void to_json (SV *scalar) 1045void to_json (SV *scalar)
1046 ALIAS:
1047 objToJson = 0
960 PPCODE: 1048 PPCODE:
961 XPUSHs (encode_json (scalar, F_UTF8)); 1049 XPUSHs (encode_json (scalar, F_UTF8));
962 1050
963void from_json (SV *jsonstr) 1051void from_json (SV *jsonstr)
1052 ALIAS:
1053 jsonToObj = 0
964 PPCODE: 1054 PPCODE:
965 XPUSHs (decode_json (jsonstr, F_UTF8)); 1055 XPUSHs (decode_json (jsonstr, F_UTF8));
966 1056

Diff Legend

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