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 1.19 by root, Sun Mar 25 21:52:47 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
59/////////////////////////////////////////////////////////////////////////////
60 97
61static void 98static void
62need (enc_t *enc, STRLEN len) 99need (enc_t *enc, STRLEN len)
63{ 100{
64 if (enc->cur + len >= enc->end) 101 if (enc->cur + len >= enc->end)
… …
122 STRLEN clen; 159 STRLEN clen;
123 UV uch; 160 UV uch;
124 161
125 if (is_utf8) 162 if (is_utf8)
126 { 163 {
127 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);
128 if (clen == (STRLEN)-1) 166 if (clen == (STRLEN)-1)
129 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);
130 } 168 }
131 else 169 else
132 { 170 {
133 uch = ch; 171 uch = ch;
134 clen = 1; 172 clen = 1;
135 } 173 }
136 174
175 if (uch > 0x10FFFFUL)
176 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
177
137 if (uch < 0x80 || enc->flags & F_ASCII) 178 if (uch < 0x80 || enc->flags & F_ASCII)
138 { 179 {
139 if (uch > 0xFFFFUL) 180 if (uch > 0xFFFFUL)
140 { 181 {
141 need (enc, len += 11); 182 need (enc, len += 11);
142 sprintf (enc->cur, "\\u%04x\\u%04x", 183 sprintf (enc->cur, "\\u%04x\\u%04x",
143 (uch - 0x10000) / 0x400 + 0xD800, 184 (int)((uch - 0x10000) / 0x400 + 0xD800),
144 (uch - 0x10000) % 0x400 + 0xDC00); 185 (int)((uch - 0x10000) % 0x400 + 0xDC00));
145 enc->cur += 12; 186 enc->cur += 12;
146 } 187 }
147 else 188 else
148 { 189 {
149 static char hexdigit [16] = "0123456789abcdef"; 190 static char hexdigit [16] = "0123456789abcdef";
… …
167 } 208 }
168 while (--clen); 209 while (--clen);
169 } 210 }
170 else 211 else
171 { 212 {
172 need (enc, 10); // never more than 11 bytes needed 213 need (enc, len += UTF8_MAX_LEN - 1); // never more than 11 bytes needed
173 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 214 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
174 ++str; 215 ++str;
175 } 216 }
176 } 217 }
177 } 218 }
… …
179 220
180 --len; 221 --len;
181 } 222 }
182} 223}
183 224
184#define INDENT SB \ 225static void
226encode_indent (enc_t *enc)
227{
185 if (enc->flags & F_INDENT) \ 228 if (enc->flags & F_INDENT)
186 { \ 229 {
187 int i_; \ 230 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
193 231
194#define SPACE SB need (enc, 1); encode_ch (enc, ' '); SE 232 need (enc, spaces);
195#define NL SB if (enc->flags & F_INDENT) { need (enc, 1); encode_ch (enc, '\n'); } SE 233 memset (enc->cur, ' ', spaces);
196#define COMMA SB \ 234 enc->cur += spaces;
235 }
236}
237
238static void
239encode_space (enc_t *enc)
240{
241 need (enc, 1);
197 encode_ch (enc, ','); \ 242 encode_ch (enc, ' ');
243}
244
245static void
246encode_nl (enc_t *enc)
247{
198 if (enc->flags & F_INDENT) \ 248 if (enc->flags & F_INDENT)
199 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);
200 else if (enc->flags & F_SPACE_AFTER) \ 262 else if (enc->flags & F_SPACE_AFTER)
201 SPACE; \ 263 encode_space (enc);
202 SE 264}
203 265
204static void encode_sv (enc_t *enc, SV *sv); 266static void encode_sv (enc_t *enc, SV *sv);
205 267
206static void 268static void
207encode_av (enc_t *enc, AV *av) 269encode_av (enc_t *enc, AV *av)
208{ 270{
209 int i, len = av_len (av); 271 int i, len = av_len (av);
210 272
211 encode_ch (enc, '['); NL; 273 encode_ch (enc, '['); encode_nl (enc);
212 ++enc->indent; 274 ++enc->indent;
213 275
214 for (i = 0; i <= len; ++i) 276 for (i = 0; i <= len; ++i)
215 { 277 {
216 INDENT; 278 encode_indent (enc);
217 encode_sv (enc, *av_fetch (av, i, 0)); 279 encode_sv (enc, *av_fetch (av, i, 0));
218 280
219 if (i < len) 281 if (i < len)
220 COMMA; 282 encode_comma (enc);
221 } 283 }
222 284
223 NL; 285 encode_nl (enc);
224 286
225 --enc->indent; 287 --enc->indent;
226 INDENT; encode_ch (enc, ']'); 288 encode_indent (enc); encode_ch (enc, ']');
227} 289}
228 290
229static void 291static void
230encode_he (enc_t *enc, HE *he) 292encode_he (enc_t *enc, HE *he)
231{ 293{
… …
245 else 307 else
246 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 308 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
247 309
248 encode_ch (enc, '"'); 310 encode_ch (enc, '"');
249 311
250 if (enc->flags & F_SPACE_BEFORE) SPACE; 312 if (enc->flags & F_SPACE_BEFORE) encode_space (enc);
251 encode_ch (enc, ':'); 313 encode_ch (enc, ':');
252 if (enc->flags & F_SPACE_AFTER ) SPACE; 314 if (enc->flags & F_SPACE_AFTER ) encode_space (enc);
253 encode_sv (enc, HeVAL (he)); 315 encode_sv (enc, HeVAL (he));
254} 316}
255 317
256// compare hash entries, used when all keys are bytestrings 318// compare hash entries, used when all keys are bytestrings
257static int 319static int
… …
263 HE *b = *(HE **)b_; 325 HE *b = *(HE **)b_;
264 326
265 STRLEN la = HeKLEN (a); 327 STRLEN la = HeKLEN (a);
266 STRLEN lb = HeKLEN (b); 328 STRLEN lb = HeKLEN (b);
267 329
268 if (!(cmp == memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 330 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb)))
269 cmp = la < lb ? -1 : la == lb ? 0 : 1; 331 cmp = la - lb;
270 332
271 return cmp; 333 return cmp;
272} 334}
273 335
274// 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
… …
281static void 343static void
282encode_hv (enc_t *enc, HV *hv) 344encode_hv (enc_t *enc, HV *hv)
283{ 345{
284 int count, i; 346 int count, i;
285 347
286 encode_ch (enc, '{'); NL; ++enc->indent; 348 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
287 349
288 if ((count = hv_iterinit (hv))) 350 if ((count = hv_iterinit (hv)))
289 { 351 {
290 // for canonical output we have to sort by keys first 352 // for canonical output we have to sort by keys first
291 // actually, this is mostly due to the stupid so-called 353 // actually, this is mostly due to the stupid so-called
292 // security workaround added somewhere in 5.8.x. 354 // security workaround added somewhere in 5.8.x.
293 // that randomises hash orderings 355 // that randomises hash orderings
294 if (enc->flags & F_CANONICAL) 356 if (enc->flags & F_CANONICAL)
295 { 357 {
296 HE *he, *hes [count]; 358 HE *he, *hes [count]; // if your compiler dies here, you need to enable C99 mode
297 int fast = 1; 359 int fast = 1;
298 360
299 i = 0; 361 i = 0;
300 while ((he = hv_iternext (hv))) 362 while ((he = hv_iternext (hv)))
301 { 363 {
… …
308 370
309 if (fast) 371 if (fast)
310 qsort (hes, count, sizeof (HE *), he_cmp_fast); 372 qsort (hes, count, sizeof (HE *), he_cmp_fast);
311 else 373 else
312 { 374 {
313 // hack to disable "use bytes" 375 // hack to forcefully disable "use bytes"
314 COP *oldcop = PL_curcop, cop; 376 COP cop = *PL_curcop;
315 cop.op_private = 0; 377 cop.op_private = 0;
378
379 ENTER;
380 SAVETMPS;
381
382 SAVEVPTR (PL_curcop);
316 PL_curcop = &cop; 383 PL_curcop = &cop;
317 384
318 SAVETMPS;
319 qsort (hes, count, sizeof (HE *), he_cmp_slow); 385 qsort (hes, count, sizeof (HE *), he_cmp_slow);
386
320 FREETMPS; 387 FREETMPS;
321 388 LEAVE;
322 PL_curcop = oldcop;
323 } 389 }
324 390
325 for (i = 0; i < count; ++i) 391 for (i = 0; i < count; ++i)
326 { 392 {
327 INDENT; 393 encode_indent (enc);
328 encode_he (enc, hes [i]); 394 encode_he (enc, hes [i]);
329 395
330 if (i < count - 1) 396 if (i < count - 1)
331 COMMA; 397 encode_comma (enc);
332 } 398 }
333 399
334 NL; 400 encode_nl (enc);
335 } 401 }
336 else 402 else
337 { 403 {
338 SV *sv; 404 SV *sv;
339 HE *he = hv_iternext (hv); 405 HE *he = hv_iternext (hv);
340 406
341 for (;;) 407 for (;;)
342 { 408 {
343 INDENT; 409 encode_indent (enc);
344 encode_he (enc, he); 410 encode_he (enc, he);
345 411
346 if (!(he = hv_iternext (hv))) 412 if (!(he = hv_iternext (hv)))
347 break; 413 break;
348 414
349 COMMA; 415 encode_comma (enc);
350 } 416 }
351 417
352 NL; 418 encode_nl (enc);
353 } 419 }
354 } 420 }
355 421
356 --enc->indent; INDENT; encode_ch (enc, '}'); 422 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
357} 423}
358 424
359static void 425static void
360encode_sv (enc_t *enc, SV *sv) 426encode_sv (enc_t *enc, SV *sv)
361{ 427{
… …
383 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 449 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv))
384 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 450 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv));
385 } 451 }
386 else if (SvROK (sv)) 452 else if (SvROK (sv))
387 { 453 {
388 if (!--enc->max_recurse) 454 SV *rv = SvRV (sv);
455
456 if (enc->indent >= enc->maxdepth)
389 croak ("data structure too deep (hit recursion limit)"); 457 croak ("data structure too deep (hit recursion limit)");
390 458
391 sv = SvRV (sv);
392
393 switch (SvTYPE (sv)) 459 switch (SvTYPE (rv))
394 { 460 {
395 case SVt_PVAV: encode_av (enc, (AV *)sv); break; 461 case SVt_PVAV: encode_av (enc, (AV *)rv); break;
396 case SVt_PVHV: encode_hv (enc, (HV *)sv); break; 462 case SVt_PVHV: encode_hv (enc, (HV *)rv); break;
397 463
398 default: 464 default:
399 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));
400 } 467 }
401 } 468 }
402 else if (!SvOK (sv)) 469 else if (!SvOK (sv))
403 encode_str (enc, "null", 4, 0); 470 encode_str (enc, "null", 4, 0);
404 else 471 else
405 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));
406} 474}
407 475
408static SV * 476static SV *
409encode_json (SV *scalar, UV flags) 477encode_json (SV *scalar, U32 flags)
410{ 478{
411 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 479 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar))
412 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)");
413 481
414 enc_t enc; 482 enc_t enc;
415 enc.flags = flags; 483 enc.flags = flags;
416 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 484 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
417 enc.cur = SvPVX (enc.sv); 485 enc.cur = SvPVX (enc.sv);
418 enc.end = SvEND (enc.sv); 486 enc.end = SvEND (enc.sv);
419 enc.max_recurse = 0;
420 enc.indent = 0; 487 enc.indent = 0;
488 enc.maxdepth = DEC_DEPTH (flags);
421 489
422 SvPOK_only (enc.sv); 490 SvPOK_only (enc.sv);
423 encode_sv (&enc, scalar); 491 encode_sv (&enc, scalar);
424 492
425 if (!(flags & (F_ASCII | F_UTF8))) 493 if (!(flags & (F_ASCII | F_UTF8)))
426 SvUTF8_on (enc.sv); 494 SvUTF8_on (enc.sv);
427 495
428 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 496 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
429 497
430#ifdef SvPV_shrink_to_cur
431 if (enc.flags & F_SHRINK) 498 if (enc.flags & F_SHRINK)
432 SvPV_shrink_to_cur (enc.sv); 499 shrink (enc.sv);
433#endif 500
434 return enc.sv; 501 return enc.sv;
435} 502}
436 503
437///////////////////////////////////////////////////////////////////////////// 504/////////////////////////////////////////////////////////////////////////////
505// decoder
438 506
439#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{
440 for (;;) \ 521 for (;;)
441 { \ 522 {
442 char ch = *dec->cur; \ 523 char ch = *dec->cur;
524
443 if (ch > 0x20 \ 525 if (ch > 0x20
444 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) \ 526 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09))
445 break; \ 527 break;
528
446 ++dec->cur; \ 529 ++dec->cur;
447 } 530 }
531}
448 532
449#define ERR(reason) SB dec->err = reason; goto fail; SE 533#define ERR(reason) SB dec->err = reason; goto fail; SE
534
450#define EXPECT_CH(ch) SB \ 535#define EXPECT_CH(ch) SB \
451 if (*dec->cur != ch) \ 536 if (*dec->cur != ch) \
452 ERR (# ch " expected"); \ 537 ERR (# ch " expected"); \
453 ++dec->cur; \ 538 ++dec->cur; \
454 SE 539 SE
455 540
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
456static SV *decode_sv (dec_t *dec); 544static SV *decode_sv (dec_t *dec);
457 545
458static signed char decode_hexdigit[256]; 546static signed char decode_hexdigit[256];
459 547
460static UV 548static UV
461decode_4hex (dec_t *dec) 549decode_4hex (dec_t *dec)
462{ 550{
463 signed char d1, d2, d3, d4; 551 signed char d1, d2, d3, d4;
552 unsigned char *cur = (unsigned char *)dec->cur;
464 553
465 d1 = decode_hexdigit [((unsigned char *)dec->cur) [0]];
466 if (d1 < 0) ERR ("four hexadecimal digits expected"); 554 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"); 555 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"); 556 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"); 557 d4 = decode_hexdigit [cur [3]]; if (d4 < 0) ERR ("four hexadecimal digits expected");
473 558
474 dec->cur += 4; 559 dec->cur += 4;
475 560
476 return ((UV)d1) << 12 561 return ((UV)d1) << 12
477 | ((UV)d2) << 8 562 | ((UV)d2) << 8
… …
480 565
481fail: 566fail:
482 return (UV)-1; 567 return (UV)-1;
483} 568}
484 569
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 * 570static SV *
501decode_str (dec_t *dec) 571decode_str (dec_t *dec)
502{ 572{
503 SV *sv = NEWSV (0,2); 573 SV *sv = 0;
504 int utf8 = 0; 574 int utf8 = 0;
505 char *cur = SvPVX (sv);
506 char *end = SvEND (sv);
507 575
508 for (;;) 576 do
509 { 577 {
510 unsigned char ch = *(unsigned char *)dec->cur; 578 char buf [SHORT_STRING_LEN + UTF8_MAX_LEN];
579 char *cur = buf;
511 580
512 if (ch == '"') 581 do
513 break;
514 else if (ch == '\\')
515 { 582 {
516 switch (*++dec->cur) 583 unsigned char ch = *(unsigned char *)dec->cur++;
584
585 if (ch == '"')
517 { 586 {
518 case '\\': 587 --dec->cur;
519 case '/': 588 break;
520 case '"': APPEND_CH (*dec->cur++); break; 589 }
521 590 else if (ch == '\\')
522 case 'b': APPEND_CH ('\010'); ++dec->cur; break; 591 {
523 case 't': APPEND_CH ('\011'); ++dec->cur; break; 592 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 { 593 {
530 UV lo, hi; 594 case '\\':
531 ++dec->cur; 595 case '/':
596 case '"': *cur++ = *dec->cur++; break;
532 597
533 hi = decode_4hex (dec); 598 case 'b': ++dec->cur; *cur++ = '\010'; break;
534 if (hi == (UV)-1) 599 case 't': ++dec->cur; *cur++ = '\011'; break;
535 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;
536 603
537 // possibly a surrogate pair 604 case 'u':
538 if (hi >= 0xd800 && hi < 0xdc00)
539 { 605 {
540 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 606 UV lo, hi;
541 ERR ("missing low surrogate character in surrogate pair");
542
543 dec->cur += 2; 607 ++dec->cur;
544 608
545 lo = decode_4hex (dec); 609 hi = decode_4hex (dec);
546 if (lo == (UV)-1) 610 if (hi == (UV)-1)
547 goto fail; 611 goto fail;
548 612
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
549 if (lo < 0xdc00 || lo >= 0xe000) 626 if (lo < 0xdc00 || lo >= 0xe000)
550 ERR ("surrogate pair expected"); 627 ERR ("surrogate pair expected");
551 628
552 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;
553 } 642 }
554 else if (hi >= 0xdc00 && hi < 0xe000)
555 ERR ("missing high surrogate character in surrogate pair");
556
557 if (hi >= 0x80)
558 { 643 break;
559 utf8 = 1;
560 644
561 APPEND_GROW (4); // at most 4 bytes for 21 bits
562 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
563 }
564 else 645 default:
565 APPEND_CH (hi); 646 --dec->cur;
647 ERR ("illegal backslash escape sequence in string");
566 } 648 }
567 break;
568
569 default:
570 --dec->cur;
571 ERR ("illegal backslash escape sequence in string");
572 } 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
573 } 674 }
574 else if (ch >= 0x20 && ch <= 0x7f) 675 while (cur < buf + SHORT_STRING_LEN);
575 APPEND_CH (*dec->cur++); 676
576 else if (ch >= 0x80) 677 STRLEN len = cur - buf;
678
679 if (sv)
577 { 680 {
578 STRLEN clen; 681 SvGROW (sv, SvCUR (sv) + len + 1);
579 UV uch = utf8n_to_uvuni (dec->cur, dec->end - dec->cur, &clen, UTF8_CHECK_ONLY); 682 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
580 if (clen == (STRLEN)-1) 683 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 } 684 }
592 else if (dec->cur == dec->end)
593 ERR ("unexpected end of string while parsing json string");
594 else 685 else
595 ERR ("invalid character encountered"); 686 sv = newSVpvn (buf, len);
596 } 687 }
688 while (*dec->cur != '"');
597 689
598 ++dec->cur; 690 ++dec->cur;
599 691
600 SvCUR_set (sv, cur - SvPVX (sv)); 692 if (sv)
601 693 {
602 SvPOK_only (sv); 694 SvPOK_only (sv);
603 *SvEND (sv) = 0; 695 *SvEND (sv) = 0;
604 696
605 if (utf8) 697 if (utf8)
606 SvUTF8_on (sv); 698 SvUTF8_on (sv);
607 699 }
608#ifdef SvPV_shrink_to_cur 700 else
609 if (dec->flags & F_SHRINK) 701 sv = newSVpvn ("", 0);
610 SvPV_shrink_to_cur (sv);
611#endif
612 702
613 return sv; 703 return sv;
614 704
615fail: 705fail:
616 SvREFCNT_dec (sv);
617 return 0; 706 return 0;
618} 707}
619 708
620static SV * 709static SV *
621decode_num (dec_t *dec) 710decode_num (dec_t *dec)
… …
702static SV * 791static SV *
703decode_av (dec_t *dec) 792decode_av (dec_t *dec)
704{ 793{
705 AV *av = newAV (); 794 AV *av = newAV ();
706 795
707 WS; 796 DEC_INC_DEPTH;
797 decode_ws (dec);
798
708 if (*dec->cur == ']') 799 if (*dec->cur == ']')
709 ++dec->cur; 800 ++dec->cur;
710 else 801 else
711 for (;;) 802 for (;;)
712 { 803 {
… …
716 if (!value) 807 if (!value)
717 goto fail; 808 goto fail;
718 809
719 av_push (av, value); 810 av_push (av, value);
720 811
721 WS; 812 decode_ws (dec);
722 813
723 if (*dec->cur == ']') 814 if (*dec->cur == ']')
724 { 815 {
725 ++dec->cur; 816 ++dec->cur;
726 break; 817 break;
… …
730 ERR (", or ] expected while parsing array"); 821 ERR (", or ] expected while parsing array");
731 822
732 ++dec->cur; 823 ++dec->cur;
733 } 824 }
734 825
826 DEC_DEC_DEPTH;
735 return newRV_noinc ((SV *)av); 827 return newRV_noinc ((SV *)av);
736 828
737fail: 829fail:
738 SvREFCNT_dec (av); 830 SvREFCNT_dec (av);
831 DEC_DEC_DEPTH;
739 return 0; 832 return 0;
740} 833}
741 834
742static SV * 835static SV *
743decode_hv (dec_t *dec) 836decode_hv (dec_t *dec)
744{ 837{
745 HV *hv = newHV (); 838 HV *hv = newHV ();
746 839
747 WS; 840 DEC_INC_DEPTH;
841 decode_ws (dec);
842
748 if (*dec->cur == '}') 843 if (*dec->cur == '}')
749 ++dec->cur; 844 ++dec->cur;
750 else 845 else
751 for (;;) 846 for (;;)
752 { 847 {
753 SV *key, *value; 848 SV *key, *value;
754 849
755 WS; EXPECT_CH ('"'); 850 decode_ws (dec); EXPECT_CH ('"');
756 851
757 key = decode_str (dec); 852 key = decode_str (dec);
758 if (!key) 853 if (!key)
759 goto fail; 854 goto fail;
760 855
761 WS; EXPECT_CH (':'); 856 decode_ws (dec); EXPECT_CH (':');
762 857
763 value = decode_sv (dec); 858 value = decode_sv (dec);
764 if (!value) 859 if (!value)
765 { 860 {
766 SvREFCNT_dec (key); 861 SvREFCNT_dec (key);
767 goto fail; 862 goto fail;
768 } 863 }
769 864
770 //TODO: optimise
771 hv_store_ent (hv, key, value, 0); 865 hv_store_ent (hv, key, value, 0);
866 SvREFCNT_dec (key);
772 867
773 WS; 868 decode_ws (dec);
774 869
775 if (*dec->cur == '}') 870 if (*dec->cur == '}')
776 { 871 {
777 ++dec->cur; 872 ++dec->cur;
778 break; 873 break;
… …
782 ERR (", or } expected while parsing object/hash"); 877 ERR (", or } expected while parsing object/hash");
783 878
784 ++dec->cur; 879 ++dec->cur;
785 } 880 }
786 881
882 DEC_DEC_DEPTH;
787 return newRV_noinc ((SV *)hv); 883 return newRV_noinc ((SV *)hv);
788 884
789fail: 885fail:
790 SvREFCNT_dec (hv); 886 SvREFCNT_dec (hv);
887 DEC_DEC_DEPTH;
791 return 0; 888 return 0;
792} 889}
793 890
794static SV * 891static SV *
795decode_sv (dec_t *dec) 892decode_sv (dec_t *dec)
796{ 893{
797 WS; 894 decode_ws (dec);
798 switch (*dec->cur) 895 switch (*dec->cur)
799 { 896 {
800 case '"': ++dec->cur; return decode_str (dec); 897 case '"': ++dec->cur; return decode_str (dec);
801 case '[': ++dec->cur; return decode_av (dec); 898 case '[': ++dec->cur; return decode_av (dec);
802 case '{': ++dec->cur; return decode_hv (dec); 899 case '{': ++dec->cur; return decode_hv (dec);
… …
838 ERR ("'null' expected"); 935 ERR ("'null' expected");
839 936
840 break; 937 break;
841 938
842 default: 939 default:
843 ERR ("malformed json string"); 940 ERR ("malformed json string, neither array, object, number, string or atom");
844 break; 941 break;
845 } 942 }
846 943
847fail: 944fail:
848 return 0; 945 return 0;
849} 946}
850 947
851static SV * 948static SV *
852decode_json (SV *string, UV flags) 949decode_json (SV *string, U32 flags)
853{ 950{
854 SV *sv; 951 SV *sv;
855 952
856 if (flags & F_UTF8) 953 if (flags & F_UTF8)
857 sv_utf8_downgrade (string, 0); 954 sv_utf8_downgrade (string, 0);
… …
859 sv_utf8_upgrade (string); 956 sv_utf8_upgrade (string);
860 957
861 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 958 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
862 959
863 dec_t dec; 960 dec_t dec;
864 dec.flags = flags; 961 dec.flags = flags;
865 dec.cur = SvPVX (string); 962 dec.cur = SvPVX (string);
866 dec.end = SvEND (string); 963 dec.end = SvEND (string);
867 dec.err = 0; 964 dec.err = 0;
965 dec.depth = 0;
966 dec.maxdepth = DEC_DEPTH (dec.flags);
868 967
968 *SvEND (string) = 0; // this should basically be a nop, too
869 sv = decode_sv (&dec); 969 sv = decode_sv (&dec);
870 970
871 if (!sv) 971 if (!sv)
872 { 972 {
973 IV offset = dec.flags & F_UTF8
974 ? dec.cur - SvPVX (string)
873 IV offset = utf8_distance (dec.cur, SvPVX (string)); 975 : utf8_distance (dec.cur, SvPVX (string));
874 SV *uni = sv_newmortal (); 976 SV *uni = sv_newmortal ();
977
875 // horrible hack to silence warning inside pv_uni_display 978 // horrible hack to silence warning inside pv_uni_display
876 COP cop; 979 COP cop = *PL_curcop;
877 memset (&cop, 0, sizeof (cop));
878 cop.cop_warnings = pWARN_NONE; 980 cop.cop_warnings = pWARN_NONE;
981 ENTER;
879 SAVEVPTR (PL_curcop); 982 SAVEVPTR (PL_curcop);
880 PL_curcop = &cop; 983 PL_curcop = &cop;
881
882 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 984 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
985 LEAVE;
986
883 croak ("%s, at character offset %d (%s)", 987 croak ("%s, at character offset %d (%s)",
884 dec.err, 988 dec.err,
885 (int)offset, 989 (int)offset,
886 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 990 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
887 } 991 }
888 992
889 sv = sv_2mortal (sv); 993 sv = sv_2mortal (sv);
890 994
891 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 995 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)"); 996 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
893 997
894 return sv; 998 return sv;
895} 999}
896 1000
1001/////////////////////////////////////////////////////////////////////////////
1002// XS interface functions
1003
897MODULE = JSON::XS PACKAGE = JSON::XS 1004MODULE = JSON::XS PACKAGE = JSON::XS
898 1005
899BOOT: 1006BOOT:
900{ 1007{
901 int i; 1008 int i;
902 1009
903 memset (decode_hexdigit, 0xff, 256); 1010 memset (decode_hexdigit, 0xff, 256);
1011
904 for (i = 10; i--; ) 1012 for (i = 0; i < 256; ++i)
905 decode_hexdigit ['0' + i] = i; 1013 decode_hexdigit [i] =
906 1014 i >= '0' && i <= '9' ? i - '0'
907 for (i = 7; i--; ) 1015 : i >= 'a' && i <= 'f' ? i - 'a' + 10
908 { 1016 : i >= 'A' && i <= 'F' ? i - 'A' + 10
909 decode_hexdigit ['a' + i] = 10 + i; 1017 : -1;
910 decode_hexdigit ['A' + i] = 10 + i;
911 }
912 1018
913 json_stash = gv_stashpv ("JSON::XS", 1); 1019 json_stash = gv_stashpv ("JSON::XS", 1);
914} 1020}
915 1021
916PROTOTYPES: DISABLE 1022PROTOTYPES: DISABLE
… …
927 utf8 = F_UTF8 1033 utf8 = F_UTF8
928 indent = F_INDENT 1034 indent = F_INDENT
929 canonical = F_CANONICAL 1035 canonical = F_CANONICAL
930 space_before = F_SPACE_BEFORE 1036 space_before = F_SPACE_BEFORE
931 space_after = F_SPACE_AFTER 1037 space_after = F_SPACE_AFTER
932 json_rpc = F_JSON_RPC
933 pretty = F_PRETTY 1038 pretty = F_PRETTY
934 allow_nonref = F_ALLOW_NONREF 1039 allow_nonref = F_ALLOW_NONREF
935 shrink = F_SHRINK 1040 shrink = F_SHRINK
936 CODE: 1041 CODE:
937{ 1042{
… …
944 RETVAL = newSVsv (self); 1049 RETVAL = newSVsv (self);
945} 1050}
946 OUTPUT: 1051 OUTPUT:
947 RETVAL 1052 RETVAL
948 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
949void encode (SV *self, SV *scalar) 1072void encode (SV *self, SV *scalar)
950 PPCODE: 1073 PPCODE:
951 XPUSHs (encode_json (scalar, *SvJSON (self))); 1074 XPUSHs (encode_json (scalar, *SvJSON (self)));
952 1075
953void decode (SV *self, SV *jsonstr) 1076void decode (SV *self, SV *jsonstr)
… …
955 XPUSHs (decode_json (jsonstr, *SvJSON (self))); 1078 XPUSHs (decode_json (jsonstr, *SvJSON (self)));
956 1079
957PROTOTYPES: ENABLE 1080PROTOTYPES: ENABLE
958 1081
959void to_json (SV *scalar) 1082void to_json (SV *scalar)
1083 ALIAS:
1084 objToJson = 0
960 PPCODE: 1085 PPCODE:
961 XPUSHs (encode_json (scalar, F_UTF8)); 1086 XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8));
962 1087
963void from_json (SV *jsonstr) 1088void from_json (SV *jsonstr)
1089 ALIAS:
1090 jsonToObj = 0
964 PPCODE: 1091 PPCODE:
965 XPUSHs (decode_json (jsonstr, F_UTF8)); 1092 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8));
966 1093

Diff Legend

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