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 1.8 by root, Fri Mar 23 16:13:59 2007 UTC vs.
Revision 1.20 by root, Sun Mar 25 21:53:19 2007 UTC

… …
4 4
5#include "assert.h" 5#include "assert.h"
6#include "string.h" 6#include "string.h"
7#include "stdlib.h" 7#include "stdlib.h"
8 8
9#define F_ASCII 0x00000001 9#define F_ASCII 0x00000001UL
10#define F_UTF8 0x00000002 10#define F_UTF8 0x00000002UL
11#define F_INDENT 0x00000004 11#define F_INDENT 0x00000004UL
12#define F_CANONICAL 0x00000008 12#define F_CANONICAL 0x00000008UL
13#define F_SPACE_BEFORE 0x00000010 13#define F_SPACE_BEFORE 0x00000010UL
14#define F_SPACE_AFTER 0x00000020 14#define F_SPACE_AFTER 0x00000020UL
15#define F_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 UTF-8 character in string, cannot convert to JSON"); 167 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
139 } 168 }
140 else 169 else
141 { 170 {
142 uch = ch; 171 uch = ch;
143 clen = 1; 172 clen = 1;
144 } 173 }
145 174
175 if (uch > 0x10FFFFUL)
176 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
177
146 if (uch < 0x80 || enc->flags & F_ASCII) 178 if (uch < 0x80 || enc->flags & F_ASCII)
147 { 179 {
148 if (uch > 0xFFFFUL) 180 if (uch > 0xFFFFUL)
149 { 181 {
150 need (enc, len += 11); 182 need (enc, len += 11);
151 sprintf (enc->cur, "\\u%04x\\u%04x", 183 sprintf (enc->cur, "\\u%04x\\u%04x",
152 (uch - 0x10000) / 0x400 + 0xD800, 184 (int)((uch - 0x10000) / 0x400 + 0xD800),
153 (uch - 0x10000) % 0x400 + 0xDC00); 185 (int)((uch - 0x10000) % 0x400 + 0xDC00));
154 enc->cur += 12; 186 enc->cur += 12;
155 } 187 }
156 else 188 else
157 { 189 {
158 static char hexdigit [16] = "0123456789abcdef"; 190 static char hexdigit [16] = "0123456789abcdef";
… …
176 } 208 }
177 while (--clen); 209 while (--clen);
178 } 210 }
179 else 211 else
180 { 212 {
181 need (enc, len += 10); // never more than 11 bytes needed 213 need (enc, len += UTF8_MAX_LEN - 1); // never more than 11 bytes needed
182 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 214 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
183 ++str; 215 ++str;
184 } 216 }
185 } 217 }
186 } 218 }
… …
188 220
189 --len; 221 --len;
190 } 222 }
191} 223}
192 224
193#define INDENT SB \ 225static void
226encode_indent (enc_t *enc)
227{
194 if (enc->flags & F_INDENT) \ 228 if (enc->flags & F_INDENT)
195 { \ 229 {
196 int i_; \ 230 int spaces = enc->indent * INDENT_STEP;
197 need (enc, enc->indent); \
198 for (i_ = enc->indent * 3; i_--; )\
199 encode_ch (enc, ' '); \
200 } \
201 SE
202 231
203#define SPACE SB need (enc, 1); encode_ch (enc, ' '); SE 232 need (enc, spaces);
204#define NL SB if (enc->flags & F_INDENT) { need (enc, 1); encode_ch (enc, '\n'); } SE 233 memset (enc->cur, ' ', spaces);
205#define COMMA SB \ 234 enc->cur += spaces;
235 }
236}
237
238static void
239encode_space (enc_t *enc)
240{
241 need (enc, 1);
206 encode_ch (enc, ','); \ 242 encode_ch (enc, ' ');
243}
244
245static void
246encode_nl (enc_t *enc)
247{
207 if (enc->flags & F_INDENT) \ 248 if (enc->flags & F_INDENT)
208 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);
209 else if (enc->flags & F_SPACE_AFTER) \ 262 else if (enc->flags & F_SPACE_AFTER)
210 SPACE; \ 263 encode_space (enc);
211 SE 264}
212 265
213static void encode_sv (enc_t *enc, SV *sv); 266static void encode_sv (enc_t *enc, SV *sv);
214 267
215static void 268static void
216encode_av (enc_t *enc, AV *av) 269encode_av (enc_t *enc, AV *av)
217{ 270{
218 int i, len = av_len (av); 271 int i, len = av_len (av);
219 272
220 encode_ch (enc, '['); NL; 273 encode_ch (enc, '['); encode_nl (enc);
221 ++enc->indent; 274 ++enc->indent;
222 275
223 for (i = 0; i <= len; ++i) 276 for (i = 0; i <= len; ++i)
224 { 277 {
225 INDENT; 278 encode_indent (enc);
226 encode_sv (enc, *av_fetch (av, i, 0)); 279 encode_sv (enc, *av_fetch (av, i, 0));
227 280
228 if (i < len) 281 if (i < len)
229 COMMA; 282 encode_comma (enc);
230 } 283 }
231 284
232 NL; 285 encode_nl (enc);
233 286
234 --enc->indent; 287 --enc->indent;
235 INDENT; encode_ch (enc, ']'); 288 encode_indent (enc); encode_ch (enc, ']');
236} 289}
237 290
238static void 291static void
239encode_he (enc_t *enc, HE *he) 292encode_he (enc_t *enc, HE *he)
240{ 293{
… …
254 else 307 else
255 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 308 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
256 309
257 encode_ch (enc, '"'); 310 encode_ch (enc, '"');
258 311
259 if (enc->flags & F_SPACE_BEFORE) SPACE; 312 if (enc->flags & F_SPACE_BEFORE) encode_space (enc);
260 encode_ch (enc, ':'); 313 encode_ch (enc, ':');
261 if (enc->flags & F_SPACE_AFTER ) SPACE; 314 if (enc->flags & F_SPACE_AFTER ) encode_space (enc);
262 encode_sv (enc, HeVAL (he)); 315 encode_sv (enc, HeVAL (he));
263} 316}
264 317
265// compare hash entries, used when all keys are bytestrings 318// compare hash entries, used when all keys are bytestrings
266static int 319static int
… …
272 HE *b = *(HE **)b_; 325 HE *b = *(HE **)b_;
273 326
274 STRLEN la = HeKLEN (a); 327 STRLEN la = HeKLEN (a);
275 STRLEN lb = HeKLEN (b); 328 STRLEN lb = HeKLEN (b);
276 329
277 if (!(cmp == memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 330 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb)))
278 cmp = la < lb ? -1 : la == lb ? 0 : 1; 331 cmp = la - lb;
279 332
280 return cmp; 333 return cmp;
281} 334}
282 335
283// 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
… …
290static void 343static void
291encode_hv (enc_t *enc, HV *hv) 344encode_hv (enc_t *enc, HV *hv)
292{ 345{
293 int count, i; 346 int count, i;
294 347
295 encode_ch (enc, '{'); NL; ++enc->indent; 348 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
296 349
297 if ((count = hv_iterinit (hv))) 350 if ((count = hv_iterinit (hv)))
298 { 351 {
299 // for canonical output we have to sort by keys first 352 // for canonical output we have to sort by keys first
300 // actually, this is mostly due to the stupid so-called 353 // actually, this is mostly due to the stupid so-called
301 // security workaround added somewhere in 5.8.x. 354 // security workaround added somewhere in 5.8.x.
302 // that randomises hash orderings 355 // that randomises hash orderings
303 if (enc->flags & F_CANONICAL) 356 if (enc->flags & F_CANONICAL)
304 { 357 {
305 HE *he, *hes [count]; 358 HE *he, *hes [count]; // if your compiler dies here, you need to enable C99 mode
306 int fast = 1; 359 int fast = 1;
307 360
308 i = 0; 361 i = 0;
309 while ((he = hv_iternext (hv))) 362 while ((he = hv_iternext (hv)))
310 { 363 {
… …
335 LEAVE; 388 LEAVE;
336 } 389 }
337 390
338 for (i = 0; i < count; ++i) 391 for (i = 0; i < count; ++i)
339 { 392 {
340 INDENT; 393 encode_indent (enc);
341 encode_he (enc, hes [i]); 394 encode_he (enc, hes [i]);
342 395
343 if (i < count - 1) 396 if (i < count - 1)
344 COMMA; 397 encode_comma (enc);
345 } 398 }
346 399
347 NL; 400 encode_nl (enc);
348 } 401 }
349 else 402 else
350 { 403 {
351 SV *sv; 404 SV *sv;
352 HE *he = hv_iternext (hv); 405 HE *he = hv_iternext (hv);
353 406
354 for (;;) 407 for (;;)
355 { 408 {
356 INDENT; 409 encode_indent (enc);
357 encode_he (enc, he); 410 encode_he (enc, he);
358 411
359 if (!(he = hv_iternext (hv))) 412 if (!(he = hv_iternext (hv)))
360 break; 413 break;
361 414
362 COMMA; 415 encode_comma (enc);
363 } 416 }
364 417
365 NL; 418 encode_nl (enc);
366 } 419 }
367 } 420 }
368 421
369 --enc->indent; INDENT; encode_ch (enc, '}'); 422 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
370} 423}
371 424
372static void 425static void
373encode_sv (enc_t *enc, SV *sv) 426encode_sv (enc_t *enc, SV *sv)
374{ 427{
… …
396 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 449 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv))
397 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 450 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv));
398 } 451 }
399 else if (SvROK (sv)) 452 else if (SvROK (sv))
400 { 453 {
401 if (!--enc->max_recurse) 454 SV *rv = SvRV (sv);
455
456 if (enc->indent >= enc->maxdepth)
402 croak ("data structure too deep (hit recursion limit)"); 457 croak ("data structure too deep (hit recursion limit)");
403 458
404 sv = SvRV (sv);
405
406 switch (SvTYPE (sv)) 459 switch (SvTYPE (rv))
407 { 460 {
408 case SVt_PVAV: encode_av (enc, (AV *)sv); break; 461 case SVt_PVAV: encode_av (enc, (AV *)rv); break;
409 case SVt_PVHV: encode_hv (enc, (HV *)sv); break; 462 case SVt_PVHV: encode_hv (enc, (HV *)rv); break;
410 463
411 default: 464 default:
412 croak ("JSON can only represent references to arrays or hashes"); 465 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
466 SvPV_nolen (sv));
413 } 467 }
414 } 468 }
415 else if (!SvOK (sv)) 469 else if (!SvOK (sv))
416 encode_str (enc, "null", 4, 0); 470 encode_str (enc, "null", 4, 0);
417 else 471 else
418 croak ("encountered perl type that JSON cannot handle"); 472 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
473 SvPV_nolen (sv), SvFLAGS (sv));
419} 474}
420 475
421static SV * 476static SV *
422encode_json (SV *scalar, UV flags) 477encode_json (SV *scalar, U32 flags)
423{ 478{
424 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 479 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar))
425 croak ("hash- or arraref required (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)");
426 481
427 enc_t enc; 482 enc_t enc;
428 enc.flags = flags; 483 enc.flags = flags;
429 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 484 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
430 enc.cur = SvPVX (enc.sv); 485 enc.cur = SvPVX (enc.sv);
431 enc.end = SvEND (enc.sv); 486 enc.end = SvEND (enc.sv);
432 enc.max_recurse = 0;
433 enc.indent = 0; 487 enc.indent = 0;
488 enc.maxdepth = DEC_DEPTH (flags);
434 489
435 SvPOK_only (enc.sv); 490 SvPOK_only (enc.sv);
436 encode_sv (&enc, scalar); 491 encode_sv (&enc, scalar);
437 492
438 if (!(flags & (F_ASCII | F_UTF8))) 493 if (!(flags & (F_ASCII | F_UTF8)))
… …
445 500
446 return enc.sv; 501 return enc.sv;
447} 502}
448 503
449///////////////////////////////////////////////////////////////////////////// 504/////////////////////////////////////////////////////////////////////////////
505// decoder
450 506
451#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{
452 for (;;) \ 521 for (;;)
453 { \ 522 {
454 char ch = *dec->cur; \ 523 char ch = *dec->cur;
524
455 if (ch > 0x20 \ 525 if (ch > 0x20
456 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) \ 526 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09))
457 break; \ 527 break;
528
458 ++dec->cur; \ 529 ++dec->cur;
459 } 530 }
531}
460 532
461#define ERR(reason) SB dec->err = reason; goto fail; SE 533#define ERR(reason) SB dec->err = reason; goto fail; SE
534
462#define EXPECT_CH(ch) SB \ 535#define EXPECT_CH(ch) SB \
463 if (*dec->cur != ch) \ 536 if (*dec->cur != ch) \
464 ERR (# ch " expected"); \ 537 ERR (# ch " expected"); \
465 ++dec->cur; \ 538 ++dec->cur; \
466 SE 539 SE
467 540
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
468static SV *decode_sv (dec_t *dec); 544static SV *decode_sv (dec_t *dec);
469 545
470static signed char decode_hexdigit[256]; 546static signed char decode_hexdigit[256];
471 547
472static UV 548static UV
473decode_4hex (dec_t *dec) 549decode_4hex (dec_t *dec)
474{ 550{
475 signed char d1, d2, d3, d4; 551 signed char d1, d2, d3, d4;
552 unsigned char *cur = (unsigned char *)dec->cur;
476 553
477 d1 = decode_hexdigit [((unsigned char *)dec->cur) [0]];
478 if (d1 < 0) ERR ("four hexadecimal digits expected"); 554 d1 = decode_hexdigit [cur [0]]; if (d1 < 0) ERR ("four hexadecimal digits expected");
479 d2 = decode_hexdigit [((unsigned char *)dec->cur) [1]];
480 if (d2 < 0) ERR ("four hexadecimal digits expected"); 555 d2 = decode_hexdigit [cur [1]]; if (d2 < 0) ERR ("four hexadecimal digits expected");
481 d3 = decode_hexdigit [((unsigned char *)dec->cur) [2]];
482 if (d3 < 0) ERR ("four hexadecimal digits expected"); 556 d3 = decode_hexdigit [cur [2]]; if (d3 < 0) ERR ("four hexadecimal digits expected");
483 d4 = decode_hexdigit [((unsigned char *)dec->cur) [3]];
484 if (d4 < 0) ERR ("four hexadecimal digits expected"); 557 d4 = decode_hexdigit [cur [3]]; if (d4 < 0) ERR ("four hexadecimal digits expected");
485 558
486 dec->cur += 4; 559 dec->cur += 4;
487 560
488 return ((UV)d1) << 12 561 return ((UV)d1) << 12
489 | ((UV)d2) << 8 562 | ((UV)d2) << 8
… …
492 565
493fail: 566fail:
494 return (UV)-1; 567 return (UV)-1;
495} 568}
496 569
497#define APPEND_GROW(n) SB \
498 if (cur + (n) >= end) \
499 { \
500 STRLEN ofs = cur - SvPVX (sv); \
501 SvGROW (sv, ofs + (n) + 1); \
502 cur = SvPVX (sv) + ofs; \
503 end = SvEND (sv); \
504 } \
505 SE
506
507#define APPEND_CH(ch) SB \
508 APPEND_GROW (1); \
509 *cur++ = (ch); \
510 SE
511
512static SV * 570static SV *
513decode_str (dec_t *dec) 571decode_str (dec_t *dec)
514{ 572{
515 SV *sv = NEWSV (0,2); 573 SV *sv = 0;
516 int utf8 = 0; 574 int utf8 = 0;
517 char *cur = SvPVX (sv);
518 char *end = SvEND (sv);
519 575
520 for (;;) 576 do
521 { 577 {
522 unsigned char ch = *(unsigned char *)dec->cur; 578 char buf [SHORT_STRING_LEN + UTF8_MAX_LEN];
579 char *cur = buf;
523 580
524 if (ch == '"') 581 do
525 break;
526 else if (ch == '\\')
527 { 582 {
528 switch (*++dec->cur) 583 unsigned char ch = *(unsigned char *)dec->cur++;
584
585 if (ch == '"')
529 { 586 {
530 case '\\': 587 --dec->cur;
531 case '/': 588 break;
532 case '"': APPEND_CH (*dec->cur++); break; 589 }
533 590 else if (ch == '\\')
534 case 'b': APPEND_CH ('\010'); ++dec->cur; break; 591 {
535 case 't': APPEND_CH ('\011'); ++dec->cur; break; 592 switch (*dec->cur)
536 case 'n': APPEND_CH ('\012'); ++dec->cur; break;
537 case 'f': APPEND_CH ('\014'); ++dec->cur; break;
538 case 'r': APPEND_CH ('\015'); ++dec->cur; break;
539
540 case 'u':
541 { 593 {
542 UV lo, hi; 594 case '\\':
543 ++dec->cur; 595 case '/':
596 case '"': *cur++ = *dec->cur++; break;
544 597
545 hi = decode_4hex (dec); 598 case 'b': ++dec->cur; *cur++ = '\010'; break;
546 if (hi == (UV)-1) 599 case 't': ++dec->cur; *cur++ = '\011'; break;
547 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;
548 603
549 // possibly a surrogate pair 604 case 'u':
550 if (hi >= 0xd800 && hi < 0xdc00)
551 { 605 {
552 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 606 UV lo, hi;
553 ERR ("missing low surrogate character in surrogate pair");
554
555 dec->cur += 2; 607 ++dec->cur;
556 608
557 lo = decode_4hex (dec); 609 hi = decode_4hex (dec);
558 if (lo == (UV)-1) 610 if (hi == (UV)-1)
559 goto fail; 611 goto fail;
560 612
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
561 if (lo < 0xdc00 || lo >= 0xe000) 626 if (lo < 0xdc00 || lo >= 0xe000)
562 ERR ("surrogate pair expected"); 627 ERR ("surrogate pair expected");
563 628
564 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;
565 } 642 }
566 else if (hi >= 0xdc00 && hi < 0xe000)
567 ERR ("missing high surrogate character in surrogate pair");
568
569 if (hi >= 0x80)
570 { 643 break;
571 utf8 = 1;
572 644
573 APPEND_GROW (4); // at most 4 bytes for 21 bits
574 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
575 }
576 else 645 default:
577 APPEND_CH (hi); 646 --dec->cur;
647 ERR ("illegal backslash escape sequence in string");
578 } 648 }
579 break;
580
581 default:
582 --dec->cur;
583 ERR ("illegal backslash escape sequence in string");
584 } 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
585 } 674 }
586 else if (ch >= 0x20 && ch <= 0x7f) 675 while (cur < buf + SHORT_STRING_LEN);
587 APPEND_CH (*dec->cur++); 676
588 else if (ch >= 0x80) 677 STRLEN len = cur - buf;
678
679 if (sv)
589 { 680 {
590 STRLEN clen; 681 SvGROW (sv, SvCUR (sv) + len + 1);
591 UV uch = utf8n_to_uvuni (dec->cur, dec->end - dec->cur, &clen, UTF8_CHECK_ONLY); 682 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
592 if (clen == (STRLEN)-1) 683 SvCUR_set (sv, SvCUR (sv) + len);
593 ERR ("malformed UTF-8 character in JSON string");
594
595 APPEND_GROW (clen);
596 do
597 {
598 *cur++ = *dec->cur++;
599 }
600 while (--clen);
601
602 utf8 = 1;
603 } 684 }
604 else if (dec->cur == dec->end)
605 ERR ("unexpected end of string while parsing json string");
606 else 685 else
607 ERR ("invalid character encountered"); 686 sv = newSVpvn (buf, len);
608 } 687 }
688 while (*dec->cur != '"');
609 689
610 ++dec->cur; 690 ++dec->cur;
611 691
612 SvCUR_set (sv, cur - SvPVX (sv)); 692 if (sv)
613 693 {
614 SvPOK_only (sv); 694 SvPOK_only (sv);
615 *SvEND (sv) = 0; 695 *SvEND (sv) = 0;
616 696
617 if (utf8) 697 if (utf8)
618 SvUTF8_on (sv); 698 SvUTF8_on (sv);
619 699 }
620 if (dec->flags & F_SHRINK) 700 else
621 shrink (sv); 701 sv = newSVpvn ("", 0);
622 702
623 return sv; 703 return sv;
624 704
625fail: 705fail:
626 SvREFCNT_dec (sv);
627 return 0; 706 return 0;
628} 707}
629 708
630static SV * 709static SV *
631decode_num (dec_t *dec) 710decode_num (dec_t *dec)
… …
712static SV * 791static SV *
713decode_av (dec_t *dec) 792decode_av (dec_t *dec)
714{ 793{
715 AV *av = newAV (); 794 AV *av = newAV ();
716 795
717 WS; 796 DEC_INC_DEPTH;
797 decode_ws (dec);
798
718 if (*dec->cur == ']') 799 if (*dec->cur == ']')
719 ++dec->cur; 800 ++dec->cur;
720 else 801 else
721 for (;;) 802 for (;;)
722 { 803 {
… …
726 if (!value) 807 if (!value)
727 goto fail; 808 goto fail;
728 809
729 av_push (av, value); 810 av_push (av, value);
730 811
731 WS; 812 decode_ws (dec);
732 813
733 if (*dec->cur == ']') 814 if (*dec->cur == ']')
734 { 815 {
735 ++dec->cur; 816 ++dec->cur;
736 break; 817 break;
… …
740 ERR (", or ] expected while parsing array"); 821 ERR (", or ] expected while parsing array");
741 822
742 ++dec->cur; 823 ++dec->cur;
743 } 824 }
744 825
826 DEC_DEC_DEPTH;
745 return newRV_noinc ((SV *)av); 827 return newRV_noinc ((SV *)av);
746 828
747fail: 829fail:
748 SvREFCNT_dec (av); 830 SvREFCNT_dec (av);
831 DEC_DEC_DEPTH;
749 return 0; 832 return 0;
750} 833}
751 834
752static SV * 835static SV *
753decode_hv (dec_t *dec) 836decode_hv (dec_t *dec)
754{ 837{
755 HV *hv = newHV (); 838 HV *hv = newHV ();
756 839
757 WS; 840 DEC_INC_DEPTH;
841 decode_ws (dec);
842
758 if (*dec->cur == '}') 843 if (*dec->cur == '}')
759 ++dec->cur; 844 ++dec->cur;
760 else 845 else
761 for (;;) 846 for (;;)
762 { 847 {
763 SV *key, *value; 848 SV *key, *value;
764 849
765 WS; EXPECT_CH ('"'); 850 decode_ws (dec); EXPECT_CH ('"');
766 851
767 key = decode_str (dec); 852 key = decode_str (dec);
768 if (!key) 853 if (!key)
769 goto fail; 854 goto fail;
770 855
771 WS; EXPECT_CH (':'); 856 decode_ws (dec); EXPECT_CH (':');
772 857
773 value = decode_sv (dec); 858 value = decode_sv (dec);
774 if (!value) 859 if (!value)
775 { 860 {
776 SvREFCNT_dec (key); 861 SvREFCNT_dec (key);
777 goto fail; 862 goto fail;
778 } 863 }
779 864
780 //TODO: optimise
781 hv_store_ent (hv, key, value, 0); 865 hv_store_ent (hv, key, value, 0);
866 SvREFCNT_dec (key);
782 867
783 WS; 868 decode_ws (dec);
784 869
785 if (*dec->cur == '}') 870 if (*dec->cur == '}')
786 { 871 {
787 ++dec->cur; 872 ++dec->cur;
788 break; 873 break;
… …
792 ERR (", or } expected while parsing object/hash"); 877 ERR (", or } expected while parsing object/hash");
793 878
794 ++dec->cur; 879 ++dec->cur;
795 } 880 }
796 881
882 DEC_DEC_DEPTH;
797 return newRV_noinc ((SV *)hv); 883 return newRV_noinc ((SV *)hv);
798 884
799fail: 885fail:
800 SvREFCNT_dec (hv); 886 SvREFCNT_dec (hv);
887 DEC_DEC_DEPTH;
801 return 0; 888 return 0;
802} 889}
803 890
804static SV * 891static SV *
805decode_sv (dec_t *dec) 892decode_sv (dec_t *dec)
806{ 893{
807 WS; 894 decode_ws (dec);
808 switch (*dec->cur) 895 switch (*dec->cur)
809 { 896 {
810 case '"': ++dec->cur; return decode_str (dec); 897 case '"': ++dec->cur; return decode_str (dec);
811 case '[': ++dec->cur; return decode_av (dec); 898 case '[': ++dec->cur; return decode_av (dec);
812 case '{': ++dec->cur; return decode_hv (dec); 899 case '{': ++dec->cur; return decode_hv (dec);
… …
857fail: 944fail:
858 return 0; 945 return 0;
859} 946}
860 947
861static SV * 948static SV *
862decode_json (SV *string, UV flags) 949decode_json (SV *string, U32 flags)
863{ 950{
864 SV *sv; 951 SV *sv;
865 952
866 if (flags & F_UTF8) 953 if (flags & F_UTF8)
867 sv_utf8_downgrade (string, 0); 954 sv_utf8_downgrade (string, 0);
… …
869 sv_utf8_upgrade (string); 956 sv_utf8_upgrade (string);
870 957
871 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 958 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
872 959
873 dec_t dec; 960 dec_t dec;
874 dec.flags = flags; 961 dec.flags = flags;
875 dec.cur = SvPVX (string); 962 dec.cur = SvPVX (string);
876 dec.end = SvEND (string); 963 dec.end = SvEND (string);
877 dec.err = 0; 964 dec.err = 0;
965 dec.depth = 0;
966 dec.maxdepth = DEC_DEPTH (dec.flags);
878 967
968 *dec.end = 0; // this should basically be a nop, too, but make sure its there
879 sv = decode_sv (&dec); 969 sv = decode_sv (&dec);
880 970
881 if (!sv) 971 if (!sv)
882 { 972 {
883 IV offset = dec.flags & F_UTF8 973 IV offset = dec.flags & F_UTF8
… …
901 } 991 }
902 992
903 sv = sv_2mortal (sv); 993 sv = sv_2mortal (sv);
904 994
905 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 995 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv))
906 croak ("JSON object or array expected (but number, string, true, false or null found, 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)");
907 997
908 return sv; 998 return sv;
909} 999}
910 1000
1001/////////////////////////////////////////////////////////////////////////////
1002// XS interface functions
1003
911MODULE = JSON::XS PACKAGE = JSON::XS 1004MODULE = JSON::XS PACKAGE = JSON::XS
912 1005
913BOOT: 1006BOOT:
914{ 1007{
915 int i; 1008 int i;
916 1009
917 memset (decode_hexdigit, 0xff, 256); 1010 memset (decode_hexdigit, 0xff, 256);
1011
918 for (i = 10; i--; ) 1012 for (i = 0; i < 256; ++i)
919 decode_hexdigit ['0' + i] = i; 1013 decode_hexdigit [i] =
920 1014 i >= '0' && i <= '9' ? i - '0'
921 for (i = 7; i--; ) 1015 : i >= 'a' && i <= 'f' ? i - 'a' + 10
922 { 1016 : i >= 'A' && i <= 'F' ? i - 'A' + 10
923 decode_hexdigit ['a' + i] = 10 + i; 1017 : -1;
924 decode_hexdigit ['A' + i] = 10 + i;
925 }
926 1018
927 json_stash = gv_stashpv ("JSON::XS", 1); 1019 json_stash = gv_stashpv ("JSON::XS", 1);
928} 1020}
929 1021
930PROTOTYPES: DISABLE 1022PROTOTYPES: DISABLE
… …
941 utf8 = F_UTF8 1033 utf8 = F_UTF8
942 indent = F_INDENT 1034 indent = F_INDENT
943 canonical = F_CANONICAL 1035 canonical = F_CANONICAL
944 space_before = F_SPACE_BEFORE 1036 space_before = F_SPACE_BEFORE
945 space_after = F_SPACE_AFTER 1037 space_after = F_SPACE_AFTER
946 json_rpc = F_JSON_RPC
947 pretty = F_PRETTY 1038 pretty = F_PRETTY
948 allow_nonref = F_ALLOW_NONREF 1039 allow_nonref = F_ALLOW_NONREF
949 shrink = F_SHRINK 1040 shrink = F_SHRINK
950 CODE: 1041 CODE:
951{ 1042{
… …
958 RETVAL = newSVsv (self); 1049 RETVAL = newSVsv (self);
959} 1050}
960 OUTPUT: 1051 OUTPUT:
961 RETVAL 1052 RETVAL
962 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
963void encode (SV *self, SV *scalar) 1072void encode (SV *self, SV *scalar)
964 PPCODE: 1073 PPCODE:
965 XPUSHs (encode_json (scalar, *SvJSON (self))); 1074 XPUSHs (encode_json (scalar, *SvJSON (self)));
966 1075
967void decode (SV *self, SV *jsonstr) 1076void decode (SV *self, SV *jsonstr)
… …
969 XPUSHs (decode_json (jsonstr, *SvJSON (self))); 1078 XPUSHs (decode_json (jsonstr, *SvJSON (self)));
970 1079
971PROTOTYPES: ENABLE 1080PROTOTYPES: ENABLE
972 1081
973void to_json (SV *scalar) 1082void to_json (SV *scalar)
1083 ALIAS:
1084 objToJson = 0
974 PPCODE: 1085 PPCODE:
975 XPUSHs (encode_json (scalar, F_UTF8)); 1086 XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8));
976 1087
977void from_json (SV *jsonstr) 1088void from_json (SV *jsonstr)
1089 ALIAS:
1090 jsonToObj = 0
978 PPCODE: 1091 PPCODE:
979 XPUSHs (decode_json (jsonstr, F_UTF8)); 1092 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8));
980 1093

Diff Legend

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