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 1.25 by root, Fri Apr 6 20:37:22 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 (9UL << 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 {
82 *clen = (STRLEN)-1;
83 return (UV)-1;
84 }
85}
86
87/////////////////////////////////////////////////////////////////////////////
88// encoder
28 89
29// structure used for encoding JSON 90// structure used for encoding JSON
30typedef struct 91typedef struct
31{ 92{
32 char *cur; 93 char *cur; // SvPVX (sv) + current output position
33 STRLEN len; // SvLEN (sv)
34 char *end; // SvEND (sv) 94 char *end; // SvEND (sv)
35 SV *sv; 95 SV *sv; // result scalar
36 UV flags; 96 U32 flags; // F_*
37 int max_recurse; 97 U32 indent; // indentation level
38 int indent; 98 U32 maxdepth; // max. indentation/recursion level
39} enc_t; 99} 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/////////////////////////////////////////////////////////////////////////////
60 100
61static void 101static void
62need (enc_t *enc, STRLEN len) 102need (enc_t *enc, STRLEN len)
63{ 103{
64 if (enc->cur + len >= enc->end) 104 if (enc->cur + len >= enc->end)
… …
122 STRLEN clen; 162 STRLEN clen;
123 UV uch; 163 UV uch;
124 164
125 if (is_utf8) 165 if (is_utf8)
126 { 166 {
127 uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY); 167 //uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
168 uch = decode_utf8 (str, end - str, &clen);
128 if (clen == (STRLEN)-1) 169 if (clen == (STRLEN)-1)
129 croak ("malformed UTF-8 character in string, cannot convert to JSON"); 170 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
130 } 171 }
131 else 172 else
132 { 173 {
133 uch = ch; 174 uch = ch;
134 clen = 1; 175 clen = 1;
135 } 176 }
136 177
178 if (uch > 0x10FFFFUL)
179 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
180
137 if (uch < 0x80 || enc->flags & F_ASCII) 181 if (uch < 0x80 || enc->flags & F_ASCII)
138 { 182 {
139 if (uch > 0xFFFFUL) 183 if (uch > 0xFFFFUL)
140 { 184 {
141 need (enc, len += 11); 185 need (enc, len += 11);
142 sprintf (enc->cur, "\\u%04x\\u%04x", 186 sprintf (enc->cur, "\\u%04x\\u%04x",
143 (uch - 0x10000) / 0x400 + 0xD800, 187 (int)((uch - 0x10000) / 0x400 + 0xD800),
144 (uch - 0x10000) % 0x400 + 0xDC00); 188 (int)((uch - 0x10000) % 0x400 + 0xDC00));
145 enc->cur += 12; 189 enc->cur += 12;
146 } 190 }
147 else 191 else
148 { 192 {
149 static char hexdigit [16] = "0123456789abcdef"; 193 static char hexdigit [16] = "0123456789abcdef";
… …
167 } 211 }
168 while (--clen); 212 while (--clen);
169 } 213 }
170 else 214 else
171 { 215 {
172 need (enc, 10); // never more than 11 bytes needed 216 need (enc, len += UTF8_MAX_LEN - 1); // never more than 11 bytes needed
173 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 217 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
174 ++str; 218 ++str;
175 } 219 }
176 } 220 }
177 } 221 }
… …
179 223
180 --len; 224 --len;
181 } 225 }
182} 226}
183 227
184#define INDENT SB \ 228static void
229encode_indent (enc_t *enc)
230{
185 if (enc->flags & F_INDENT) \ 231 if (enc->flags & F_INDENT)
186 { \ 232 {
187 int i_; \ 233 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
193 234
194#define SPACE SB need (enc, 1); encode_ch (enc, ' '); SE 235 need (enc, spaces);
195#define NL SB if (enc->flags & F_INDENT) { need (enc, 1); encode_ch (enc, '\n'); } SE 236 memset (enc->cur, ' ', spaces);
196#define COMMA SB \ 237 enc->cur += spaces;
238 }
239}
240
241static void
242encode_space (enc_t *enc)
243{
244 need (enc, 1);
197 encode_ch (enc, ','); \ 245 encode_ch (enc, ' ');
246}
247
248static void
249encode_nl (enc_t *enc)
250{
198 if (enc->flags & F_INDENT) \ 251 if (enc->flags & F_INDENT)
199 NL; \ 252 {
253 need (enc, 1);
254 encode_ch (enc, '\n');
255 }
256}
257
258static void
259encode_comma (enc_t *enc)
260{
261 encode_ch (enc, ',');
262
263 if (enc->flags & F_INDENT)
264 encode_nl (enc);
200 else if (enc->flags & F_SPACE_AFTER) \ 265 else if (enc->flags & F_SPACE_AFTER)
201 SPACE; \ 266 encode_space (enc);
202 SE 267}
203 268
204static void encode_sv (enc_t *enc, SV *sv); 269static void encode_sv (enc_t *enc, SV *sv);
205 270
206static void 271static void
207encode_av (enc_t *enc, AV *av) 272encode_av (enc_t *enc, AV *av)
208{ 273{
209 int i, len = av_len (av); 274 int i, len = av_len (av);
210 275
211 encode_ch (enc, '['); NL; 276 if (enc->indent >= enc->maxdepth)
277 croak ("data structure too deep (hit recursion limit)");
278
279 encode_ch (enc, '['); encode_nl (enc);
212 ++enc->indent; 280 ++enc->indent;
213 281
214 for (i = 0; i <= len; ++i) 282 for (i = 0; i <= len; ++i)
215 { 283 {
216 INDENT; 284 encode_indent (enc);
217 encode_sv (enc, *av_fetch (av, i, 0)); 285 encode_sv (enc, *av_fetch (av, i, 0));
218 286
219 if (i < len) 287 if (i < len)
220 COMMA; 288 encode_comma (enc);
221 } 289 }
222 290
223 NL; 291 encode_nl (enc);
224 292
225 --enc->indent; 293 --enc->indent;
226 INDENT; encode_ch (enc, ']'); 294 encode_indent (enc); encode_ch (enc, ']');
227} 295}
228 296
229static void 297static void
230encode_he (enc_t *enc, HE *he) 298encode_he (enc_t *enc, HE *he)
231{ 299{
… …
245 else 313 else
246 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 314 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
247 315
248 encode_ch (enc, '"'); 316 encode_ch (enc, '"');
249 317
250 if (enc->flags & F_SPACE_BEFORE) SPACE; 318 if (enc->flags & F_SPACE_BEFORE) encode_space (enc);
251 encode_ch (enc, ':'); 319 encode_ch (enc, ':');
252 if (enc->flags & F_SPACE_AFTER ) SPACE; 320 if (enc->flags & F_SPACE_AFTER ) encode_space (enc);
253 encode_sv (enc, HeVAL (he)); 321 encode_sv (enc, HeVAL (he));
254} 322}
255 323
256// compare hash entries, used when all keys are bytestrings 324// compare hash entries, used when all keys are bytestrings
257static int 325static int
… …
263 HE *b = *(HE **)b_; 331 HE *b = *(HE **)b_;
264 332
265 STRLEN la = HeKLEN (a); 333 STRLEN la = HeKLEN (a);
266 STRLEN lb = HeKLEN (b); 334 STRLEN lb = HeKLEN (b);
267 335
268 if (!(cmp == memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 336 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb)))
269 cmp = la < lb ? -1 : la == lb ? 0 : 1; 337 cmp = la - lb;
270 338
271 return cmp; 339 return cmp;
272} 340}
273 341
274// compare hash entries, used when some keys are sv's or utf-x 342// compare hash entries, used when some keys are sv's or utf-x
… …
281static void 349static void
282encode_hv (enc_t *enc, HV *hv) 350encode_hv (enc_t *enc, HV *hv)
283{ 351{
284 int count, i; 352 int count, i;
285 353
354 if (enc->indent >= enc->maxdepth)
355 croak ("data structure too deep (hit recursion limit)");
356
286 encode_ch (enc, '{'); NL; ++enc->indent; 357 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
287 358
288 if ((count = hv_iterinit (hv))) 359 if ((count = hv_iterinit (hv)))
289 { 360 {
290 // for canonical output we have to sort by keys first 361 // for canonical output we have to sort by keys first
291 // actually, this is mostly due to the stupid so-called 362 // actually, this is mostly due to the stupid so-called
292 // security workaround added somewhere in 5.8.x. 363 // security workaround added somewhere in 5.8.x.
293 // that randomises hash orderings 364 // that randomises hash orderings
294 if (enc->flags & F_CANONICAL) 365 if (enc->flags & F_CANONICAL)
295 { 366 {
296 HE *he, *hes [count]; 367 HE *he, *hes [count]; // if your compiler dies here, you need to enable C99 mode
297 int fast = 1; 368 int fast = 1;
298 369
299 i = 0; 370 i = 0;
300 while ((he = hv_iternext (hv))) 371 while ((he = hv_iternext (hv)))
301 { 372 {
… …
308 379
309 if (fast) 380 if (fast)
310 qsort (hes, count, sizeof (HE *), he_cmp_fast); 381 qsort (hes, count, sizeof (HE *), he_cmp_fast);
311 else 382 else
312 { 383 {
313 // hack to disable "use bytes" 384 // hack to forcefully disable "use bytes"
314 COP *oldcop = PL_curcop, cop; 385 COP cop = *PL_curcop;
315 cop.op_private = 0; 386 cop.op_private = 0;
387
388 ENTER;
389 SAVETMPS;
390
391 SAVEVPTR (PL_curcop);
316 PL_curcop = &cop; 392 PL_curcop = &cop;
317 393
318 SAVETMPS;
319 qsort (hes, count, sizeof (HE *), he_cmp_slow); 394 qsort (hes, count, sizeof (HE *), he_cmp_slow);
395
320 FREETMPS; 396 FREETMPS;
321 397 LEAVE;
322 PL_curcop = oldcop;
323 } 398 }
324 399
325 for (i = 0; i < count; ++i) 400 for (i = 0; i < count; ++i)
326 { 401 {
327 INDENT; 402 encode_indent (enc);
328 encode_he (enc, hes [i]); 403 encode_he (enc, hes [i]);
329 404
330 if (i < count - 1) 405 if (i < count - 1)
331 COMMA; 406 encode_comma (enc);
332 } 407 }
333 408
334 NL; 409 encode_nl (enc);
335 } 410 }
336 else 411 else
337 { 412 {
338 SV *sv;
339 HE *he = hv_iternext (hv); 413 HE *he = hv_iternext (hv);
340 414
341 for (;;) 415 for (;;)
342 { 416 {
343 INDENT; 417 encode_indent (enc);
344 encode_he (enc, he); 418 encode_he (enc, he);
345 419
346 if (!(he = hv_iternext (hv))) 420 if (!(he = hv_iternext (hv)))
347 break; 421 break;
348 422
349 COMMA; 423 encode_comma (enc);
350 } 424 }
351 425
352 NL; 426 encode_nl (enc);
353 } 427 }
354 } 428 }
355 429
356 --enc->indent; INDENT; encode_ch (enc, '}'); 430 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
431}
432
433// encode objects, arrays and special \0=false and \1=true values.
434static void
435encode_rv (enc_t *enc, SV *sv)
436{
437 svtype svt;
438
439 SvGETMAGIC (sv);
440 svt = SvTYPE (sv);
441
442 if (svt == SVt_PVHV)
443 encode_hv (enc, (HV *)sv);
444 else if (svt == SVt_PVAV)
445 encode_av (enc, (AV *)sv);
446 else if (svt < SVt_PVAV)
447 {
448 if (SvNIOK (sv) && SvIV (sv) == 0)
449 encode_str (enc, "false", 5, 0);
450 else if (SvNIOK (sv) && SvIV (sv) == 1)
451 encode_str (enc, "true", 4, 0);
452 else
453 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
454 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
455 }
456 else
457 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
458 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
357} 459}
358 460
359static void 461static void
360encode_sv (enc_t *enc, SV *sv) 462encode_sv (enc_t *enc, SV *sv)
361{ 463{
… …
382 SvIsUV(sv) 484 SvIsUV(sv)
383 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 485 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv))
384 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 486 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv));
385 } 487 }
386 else if (SvROK (sv)) 488 else if (SvROK (sv))
387 { 489 encode_rv (enc, SvRV (sv));
388 if (!--enc->max_recurse)
389 croak ("data structure too deep (hit recursion limit)");
390
391 sv = SvRV (sv);
392
393 switch (SvTYPE (sv))
394 {
395 case SVt_PVAV: encode_av (enc, (AV *)sv); break;
396 case SVt_PVHV: encode_hv (enc, (HV *)sv); break;
397
398 default:
399 croak ("JSON can only represent references to arrays or hashes");
400 }
401 }
402 else if (!SvOK (sv)) 490 else if (!SvOK (sv))
403 encode_str (enc, "null", 4, 0); 491 encode_str (enc, "null", 4, 0);
404 else 492 else
405 croak ("encountered perl type that JSON cannot handle"); 493 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
494 SvPV_nolen (sv), SvFLAGS (sv));
406} 495}
407 496
408static SV * 497static SV *
409encode_json (SV *scalar, UV flags) 498encode_json (SV *scalar, U32 flags)
410{ 499{
500 enc_t enc;
501
411 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 502 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar))
412 croak ("hash- or arraref required (not a simple scalar, use allow_nonref to allow this)"); 503 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
413 504
414 enc_t enc;
415 enc.flags = flags; 505 enc.flags = flags;
416 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 506 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
417 enc.cur = SvPVX (enc.sv); 507 enc.cur = SvPVX (enc.sv);
418 enc.end = SvEND (enc.sv); 508 enc.end = SvEND (enc.sv);
419 enc.max_recurse = 0;
420 enc.indent = 0; 509 enc.indent = 0;
510 enc.maxdepth = DEC_DEPTH (flags);
421 511
422 SvPOK_only (enc.sv); 512 SvPOK_only (enc.sv);
423 encode_sv (&enc, scalar); 513 encode_sv (&enc, scalar);
424 514
425 if (!(flags & (F_ASCII | F_UTF8))) 515 if (!(flags & (F_ASCII | F_UTF8)))
426 SvUTF8_on (enc.sv); 516 SvUTF8_on (enc.sv);
427 517
428 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 518 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
429 519
430#ifdef SvPV_shrink_to_cur
431 if (enc.flags & F_SHRINK) 520 if (enc.flags & F_SHRINK)
432 SvPV_shrink_to_cur (enc.sv); 521 shrink (enc.sv);
433#endif 522
434 return enc.sv; 523 return enc.sv;
435} 524}
436 525
437///////////////////////////////////////////////////////////////////////////// 526/////////////////////////////////////////////////////////////////////////////
527// decoder
438 528
439#define WS \ 529// structure used for decoding JSON
530typedef struct
531{
532 char *cur; // current parser pointer
533 char *end; // end of input string
534 const char *err; // parse error, if != 0
535 U32 flags; // F_*
536 U32 depth; // recursion depth
537 U32 maxdepth; // recursion depth limit
538} dec_t;
539
540static void
541decode_ws (dec_t *dec)
542{
440 for (;;) \ 543 for (;;)
441 { \ 544 {
442 char ch = *dec->cur; \ 545 char ch = *dec->cur;
546
443 if (ch > 0x20 \ 547 if (ch > 0x20
444 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) \ 548 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09))
445 break; \ 549 break;
550
446 ++dec->cur; \ 551 ++dec->cur;
447 } 552 }
553}
448 554
449#define ERR(reason) SB dec->err = reason; goto fail; SE 555#define ERR(reason) SB dec->err = reason; goto fail; SE
556
450#define EXPECT_CH(ch) SB \ 557#define EXPECT_CH(ch) SB \
451 if (*dec->cur != ch) \ 558 if (*dec->cur != ch) \
452 ERR (# ch " expected"); \ 559 ERR (# ch " expected"); \
453 ++dec->cur; \ 560 ++dec->cur; \
454 SE 561 SE
455 562
563#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)")
564#define DEC_DEC_DEPTH --dec->depth
565
456static SV *decode_sv (dec_t *dec); 566static SV *decode_sv (dec_t *dec);
457 567
458static signed char decode_hexdigit[256]; 568static signed char decode_hexdigit[256];
459 569
460static UV 570static UV
461decode_4hex (dec_t *dec) 571decode_4hex (dec_t *dec)
462{ 572{
463 signed char d1, d2, d3, d4; 573 signed char d1, d2, d3, d4;
574 unsigned char *cur = (unsigned char *)dec->cur;
464 575
465 d1 = decode_hexdigit [((unsigned char *)dec->cur) [0]];
466 if (d1 < 0) ERR ("four hexadecimal digits expected"); 576 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"); 577 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"); 578 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"); 579 d4 = decode_hexdigit [cur [3]]; if (d4 < 0) ERR ("four hexadecimal digits expected");
473 580
474 dec->cur += 4; 581 dec->cur += 4;
475 582
476 return ((UV)d1) << 12 583 return ((UV)d1) << 12
477 | ((UV)d2) << 8 584 | ((UV)d2) << 8
… …
480 587
481fail: 588fail:
482 return (UV)-1; 589 return (UV)-1;
483} 590}
484 591
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 * 592static SV *
501decode_str (dec_t *dec) 593decode_str (dec_t *dec)
502{ 594{
503 SV *sv = NEWSV (0,2); 595 SV *sv = 0;
504 int utf8 = 0; 596 int utf8 = 0;
505 char *cur = SvPVX (sv);
506 char *end = SvEND (sv);
507 597
508 for (;;) 598 do
509 { 599 {
510 unsigned char ch = *(unsigned char *)dec->cur; 600 char buf [SHORT_STRING_LEN + UTF8_MAX_LEN];
601 char *cur = buf;
511 602
512 if (ch == '"') 603 do
513 break;
514 else if (ch == '\\')
515 { 604 {
516 switch (*++dec->cur) 605 unsigned char ch = *(unsigned char *)dec->cur++;
606
607 if (ch == '"')
517 { 608 {
518 case '\\': 609 --dec->cur;
519 case '/': 610 break;
520 case '"': APPEND_CH (*dec->cur++); break; 611 }
521 612 else if (ch == '\\')
522 case 'b': APPEND_CH ('\010'); ++dec->cur; break; 613 {
523 case 't': APPEND_CH ('\011'); ++dec->cur; break; 614 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 { 615 {
530 UV lo, hi; 616 case '\\':
531 ++dec->cur; 617 case '/':
618 case '"': *cur++ = *dec->cur++; break;
532 619
533 hi = decode_4hex (dec); 620 case 'b': ++dec->cur; *cur++ = '\010'; break;
534 if (hi == (UV)-1) 621 case 't': ++dec->cur; *cur++ = '\011'; break;
535 goto fail; 622 case 'n': ++dec->cur; *cur++ = '\012'; break;
623 case 'f': ++dec->cur; *cur++ = '\014'; break;
624 case 'r': ++dec->cur; *cur++ = '\015'; break;
536 625
537 // possibly a surrogate pair 626 case 'u':
538 if (hi >= 0xd800 && hi < 0xdc00)
539 { 627 {
540 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 628 UV lo, hi;
541 ERR ("missing low surrogate character in surrogate pair");
542
543 dec->cur += 2; 629 ++dec->cur;
544 630
545 lo = decode_4hex (dec); 631 hi = decode_4hex (dec);
546 if (lo == (UV)-1) 632 if (hi == (UV)-1)
547 goto fail; 633 goto fail;
548 634
635 // possibly a surrogate pair
636 if (hi >= 0xd800)
637 if (hi < 0xdc00)
638 {
639 if (dec->cur [0] != '\\' || dec->cur [1] != 'u')
640 ERR ("missing low surrogate character in surrogate pair");
641
642 dec->cur += 2;
643
644 lo = decode_4hex (dec);
645 if (lo == (UV)-1)
646 goto fail;
647
549 if (lo < 0xdc00 || lo >= 0xe000) 648 if (lo < 0xdc00 || lo >= 0xe000)
550 ERR ("surrogate pair expected"); 649 ERR ("surrogate pair expected");
551 650
552 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000; 651 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000;
652 }
653 else if (hi < 0xe000)
654 ERR ("missing high surrogate character in surrogate pair");
655
656 if (hi >= 0x80)
657 {
658 utf8 = 1;
659
660 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
661 }
662 else
663 *cur++ = hi;
553 } 664 }
554 else if (hi >= 0xdc00 && hi < 0xe000)
555 ERR ("missing high surrogate character in surrogate pair");
556
557 if (hi >= 0x80)
558 { 665 break;
559 utf8 = 1;
560 666
561 APPEND_GROW (4); // at most 4 bytes for 21 bits
562 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
563 }
564 else 667 default:
565 APPEND_CH (hi); 668 --dec->cur;
669 ERR ("illegal backslash escape sequence in string");
566 } 670 }
567 break; 671 }
672 else if (ch >= 0x20 && ch <= 0x7f)
673 *cur++ = ch;
674 else if (ch >= 0x80)
675 {
676 STRLEN clen;
677 UV uch;
568 678
569 default:
570 --dec->cur; 679 --dec->cur;
571 ERR ("illegal backslash escape sequence in string"); 680
681 uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen);
682 if (clen == (STRLEN)-1)
683 ERR ("malformed UTF-8 character in JSON string");
684
685 do
686 *cur++ = *dec->cur++;
687 while (--clen);
688
689 utf8 = 1;
690 }
691 else
692 {
693 --dec->cur;
694
695 if (!ch)
696 ERR ("unexpected end of string while parsing JSON string");
697 else
698 ERR ("invalid character encountered while parsing JSON string");
572 } 699 }
573 } 700 }
574 else if (ch >= 0x20 && ch <= 0x7f) 701 while (cur < buf + SHORT_STRING_LEN);
575 APPEND_CH (*dec->cur++); 702
576 else if (ch >= 0x80)
577 { 703 {
578 STRLEN clen; 704 STRLEN len = cur - buf;
579 UV uch = utf8n_to_uvuni (dec->cur, dec->end - dec->cur, &clen, UTF8_CHECK_ONLY);
580 if (clen == (STRLEN)-1)
581 ERR ("malformed UTF-8 character in string, cannot convert to JSON");
582 705
583 APPEND_GROW (clen); 706 if (sv)
584 do
585 { 707 {
586 *cur++ = *dec->cur++; 708 SvGROW (sv, SvCUR (sv) + len + 1);
709 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
710 SvCUR_set (sv, SvCUR (sv) + len);
587 } 711 }
588 while (--clen);
589
590 utf8 = 1;
591 }
592 else if (dec->cur == dec->end)
593 ERR ("unexpected end of string while parsing json string");
594 else 712 else
595 ERR ("invalid character encountered"); 713 sv = newSVpvn (buf, len);
596 } 714 }
715 }
716 while (*dec->cur != '"');
597 717
598 ++dec->cur; 718 ++dec->cur;
599 719
600 SvCUR_set (sv, cur - SvPVX (sv)); 720 if (sv)
601 721 {
602 SvPOK_only (sv); 722 SvPOK_only (sv);
603 *SvEND (sv) = 0; 723 *SvEND (sv) = 0;
604 724
605 if (utf8) 725 if (utf8)
606 SvUTF8_on (sv); 726 SvUTF8_on (sv);
607 727 }
608#ifdef SvPV_shrink_to_cur 728 else
609 if (dec->flags & F_SHRINK) 729 sv = newSVpvn ("", 0);
610 SvPV_shrink_to_cur (sv);
611#endif
612 730
613 return sv; 731 return sv;
614 732
615fail: 733fail:
616 SvREFCNT_dec (sv);
617 return 0; 734 return 0;
618} 735}
619 736
620static SV * 737static SV *
621decode_num (dec_t *dec) 738decode_num (dec_t *dec)
… …
702static SV * 819static SV *
703decode_av (dec_t *dec) 820decode_av (dec_t *dec)
704{ 821{
705 AV *av = newAV (); 822 AV *av = newAV ();
706 823
707 WS; 824 DEC_INC_DEPTH;
825 decode_ws (dec);
826
708 if (*dec->cur == ']') 827 if (*dec->cur == ']')
709 ++dec->cur; 828 ++dec->cur;
710 else 829 else
711 for (;;) 830 for (;;)
712 { 831 {
… …
716 if (!value) 835 if (!value)
717 goto fail; 836 goto fail;
718 837
719 av_push (av, value); 838 av_push (av, value);
720 839
721 WS; 840 decode_ws (dec);
722 841
723 if (*dec->cur == ']') 842 if (*dec->cur == ']')
724 { 843 {
725 ++dec->cur; 844 ++dec->cur;
726 break; 845 break;
… …
730 ERR (", or ] expected while parsing array"); 849 ERR (", or ] expected while parsing array");
731 850
732 ++dec->cur; 851 ++dec->cur;
733 } 852 }
734 853
854 DEC_DEC_DEPTH;
735 return newRV_noinc ((SV *)av); 855 return newRV_noinc ((SV *)av);
736 856
737fail: 857fail:
738 SvREFCNT_dec (av); 858 SvREFCNT_dec (av);
859 DEC_DEC_DEPTH;
739 return 0; 860 return 0;
740} 861}
741 862
742static SV * 863static SV *
743decode_hv (dec_t *dec) 864decode_hv (dec_t *dec)
744{ 865{
745 HV *hv = newHV (); 866 HV *hv = newHV ();
746 867
747 WS; 868 DEC_INC_DEPTH;
869 decode_ws (dec);
870
748 if (*dec->cur == '}') 871 if (*dec->cur == '}')
749 ++dec->cur; 872 ++dec->cur;
750 else 873 else
751 for (;;) 874 for (;;)
752 { 875 {
753 SV *key, *value; 876 SV *key, *value;
754 877
755 WS; EXPECT_CH ('"'); 878 decode_ws (dec); EXPECT_CH ('"');
756 879
757 key = decode_str (dec); 880 key = decode_str (dec);
758 if (!key) 881 if (!key)
759 goto fail; 882 goto fail;
760 883
761 WS; EXPECT_CH (':'); 884 decode_ws (dec); EXPECT_CH (':');
762 885
763 value = decode_sv (dec); 886 value = decode_sv (dec);
764 if (!value) 887 if (!value)
765 { 888 {
766 SvREFCNT_dec (key); 889 SvREFCNT_dec (key);
767 goto fail; 890 goto fail;
768 } 891 }
769 892
770 //TODO: optimise
771 hv_store_ent (hv, key, value, 0); 893 hv_store_ent (hv, key, value, 0);
894 SvREFCNT_dec (key);
772 895
773 WS; 896 decode_ws (dec);
774 897
775 if (*dec->cur == '}') 898 if (*dec->cur == '}')
776 { 899 {
777 ++dec->cur; 900 ++dec->cur;
778 break; 901 break;
… …
782 ERR (", or } expected while parsing object/hash"); 905 ERR (", or } expected while parsing object/hash");
783 906
784 ++dec->cur; 907 ++dec->cur;
785 } 908 }
786 909
910 DEC_DEC_DEPTH;
787 return newRV_noinc ((SV *)hv); 911 return newRV_noinc ((SV *)hv);
788 912
789fail: 913fail:
790 SvREFCNT_dec (hv); 914 SvREFCNT_dec (hv);
915 DEC_DEC_DEPTH;
791 return 0; 916 return 0;
792} 917}
793 918
794static SV * 919static SV *
795decode_sv (dec_t *dec) 920decode_sv (dec_t *dec)
796{ 921{
797 WS; 922 decode_ws (dec);
798 switch (*dec->cur) 923 switch (*dec->cur)
799 { 924 {
800 case '"': ++dec->cur; return decode_str (dec); 925 case '"': ++dec->cur; return decode_str (dec);
801 case '[': ++dec->cur; return decode_av (dec); 926 case '[': ++dec->cur; return decode_av (dec);
802 case '{': ++dec->cur; return decode_hv (dec); 927 case '{': ++dec->cur; return decode_hv (dec);
… …
838 ERR ("'null' expected"); 963 ERR ("'null' expected");
839 964
840 break; 965 break;
841 966
842 default: 967 default:
843 ERR ("malformed json string"); 968 ERR ("malformed JSON string, neither array, object, number, string or atom");
844 break; 969 break;
845 } 970 }
846 971
847fail: 972fail:
848 return 0; 973 return 0;
849} 974}
850 975
851static SV * 976static SV *
852decode_json (SV *string, UV flags) 977decode_json (SV *string, U32 flags)
853{ 978{
979 dec_t dec;
854 SV *sv; 980 SV *sv;
981
982 SvUPGRADE (string, SVt_PV);
855 983
856 if (flags & F_UTF8) 984 if (flags & F_UTF8)
857 sv_utf8_downgrade (string, 0); 985 sv_utf8_downgrade (string, 0);
858 else 986 else
859 sv_utf8_upgrade (string); 987 sv_utf8_upgrade (string);
860 988
861 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 989 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
862 990
863 dec_t dec;
864 dec.flags = flags; 991 dec.flags = flags;
865 dec.cur = SvPVX (string); 992 dec.cur = SvPVX (string);
866 dec.end = SvEND (string); 993 dec.end = SvEND (string);
867 dec.err = 0; 994 dec.err = 0;
995 dec.depth = 0;
996 dec.maxdepth = DEC_DEPTH (dec.flags);
868 997
998 *dec.end = 0; // this should basically be a nop, too, but make sure its there
869 sv = decode_sv (&dec); 999 sv = decode_sv (&dec);
870 1000
871 if (!sv) 1001 if (!sv)
872 { 1002 {
1003 IV offset = dec.flags & F_UTF8
1004 ? dec.cur - SvPVX (string)
873 IV offset = utf8_distance (dec.cur, SvPVX (string)); 1005 : utf8_distance (dec.cur, SvPVX (string));
874 SV *uni = sv_newmortal (); 1006 SV *uni = sv_newmortal ();
1007
875 // horrible hack to silence warning inside pv_uni_display 1008 // horrible hack to silence warning inside pv_uni_display
876 COP cop; 1009 COP cop = *PL_curcop;
877 memset (&cop, 0, sizeof (cop));
878 cop.cop_warnings = pWARN_NONE; 1010 cop.cop_warnings = pWARN_NONE;
1011 ENTER;
879 SAVEVPTR (PL_curcop); 1012 SAVEVPTR (PL_curcop);
880 PL_curcop = &cop; 1013 PL_curcop = &cop;
881
882 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1014 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1015 LEAVE;
1016
883 croak ("%s, at character offset %d (%s)", 1017 croak ("%s, at character offset %d [\"%s\"]",
884 dec.err, 1018 dec.err,
885 (int)offset, 1019 (int)offset,
886 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1020 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
887 } 1021 }
888 1022
889 sv = sv_2mortal (sv); 1023 sv = sv_2mortal (sv);
890 1024
891 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1025 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)"); 1026 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
893 1027
894 return sv; 1028 return sv;
895} 1029}
896 1030
1031/////////////////////////////////////////////////////////////////////////////
1032// XS interface functions
1033
897MODULE = JSON::XS PACKAGE = JSON::XS 1034MODULE = JSON::XS PACKAGE = JSON::XS
898 1035
899BOOT: 1036BOOT:
900{ 1037{
901 int i; 1038 int i;
902 1039
903 memset (decode_hexdigit, 0xff, 256); 1040 memset (decode_hexdigit, 0xff, 256);
1041
904 for (i = 10; i--; ) 1042 for (i = 0; i < 256; ++i)
905 decode_hexdigit ['0' + i] = i; 1043 decode_hexdigit [i] =
906 1044 i >= '0' && i <= '9' ? i - '0'
907 for (i = 7; i--; ) 1045 : i >= 'a' && i <= 'f' ? i - 'a' + 10
908 { 1046 : i >= 'A' && i <= 'F' ? i - 'A' + 10
909 decode_hexdigit ['a' + i] = 10 + i; 1047 : -1;
910 decode_hexdigit ['A' + i] = 10 + i;
911 }
912 1048
913 json_stash = gv_stashpv ("JSON::XS", 1); 1049 json_stash = gv_stashpv ("JSON::XS", 1);
914} 1050}
915 1051
916PROTOTYPES: DISABLE 1052PROTOTYPES: DISABLE
… …
927 utf8 = F_UTF8 1063 utf8 = F_UTF8
928 indent = F_INDENT 1064 indent = F_INDENT
929 canonical = F_CANONICAL 1065 canonical = F_CANONICAL
930 space_before = F_SPACE_BEFORE 1066 space_before = F_SPACE_BEFORE
931 space_after = F_SPACE_AFTER 1067 space_after = F_SPACE_AFTER
932 json_rpc = F_JSON_RPC
933 pretty = F_PRETTY 1068 pretty = F_PRETTY
934 allow_nonref = F_ALLOW_NONREF 1069 allow_nonref = F_ALLOW_NONREF
935 shrink = F_SHRINK 1070 shrink = F_SHRINK
936 CODE: 1071 CODE:
937{ 1072{
… …
944 RETVAL = newSVsv (self); 1079 RETVAL = newSVsv (self);
945} 1080}
946 OUTPUT: 1081 OUTPUT:
947 RETVAL 1082 RETVAL
948 1083
1084SV *max_depth (SV *self, UV max_depth = 0x80000000UL)
1085 CODE:
1086{
1087 UV *uv = SvJSON (self);
1088 UV log2 = 0;
1089
1090 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1091
1092 while ((1UL << log2) < max_depth)
1093 ++log2;
1094
1095 *uv = *uv & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1096
1097 RETVAL = newSVsv (self);
1098}
1099 OUTPUT:
1100 RETVAL
1101
949void encode (SV *self, SV *scalar) 1102void encode (SV *self, SV *scalar)
950 PPCODE: 1103 PPCODE:
951 XPUSHs (encode_json (scalar, *SvJSON (self))); 1104 XPUSHs (encode_json (scalar, *SvJSON (self)));
952 1105
953void decode (SV *self, SV *jsonstr) 1106void decode (SV *self, SV *jsonstr)
… …
955 XPUSHs (decode_json (jsonstr, *SvJSON (self))); 1108 XPUSHs (decode_json (jsonstr, *SvJSON (self)));
956 1109
957PROTOTYPES: ENABLE 1110PROTOTYPES: ENABLE
958 1111
959void to_json (SV *scalar) 1112void to_json (SV *scalar)
1113 ALIAS:
1114 objToJson = 0
960 PPCODE: 1115 PPCODE:
961 XPUSHs (encode_json (scalar, F_UTF8)); 1116 XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8));
962 1117
963void from_json (SV *jsonstr) 1118void from_json (SV *jsonstr)
1119 ALIAS:
1120 jsonToObj = 0
964 PPCODE: 1121 PPCODE:
965 XPUSHs (decode_json (jsonstr, F_UTF8)); 1122 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8));
966 1123

Diff Legend

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