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 1.7 by root, Fri Mar 23 15:57:18 2007 UTC vs.
Revision 1.23 by root, Tue Apr 3 23:34:17 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 (12UL << 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
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/////////////////////////////////////////////////////////////////////////////
69 100
70static void 101static void
71need (enc_t *enc, STRLEN len) 102need (enc_t *enc, STRLEN len)
72{ 103{
73 if (enc->cur + len >= enc->end) 104 if (enc->cur + len >= enc->end)
… …
131 STRLEN clen; 162 STRLEN clen;
132 UV uch; 163 UV uch;
133 164
134 if (is_utf8) 165 if (is_utf8)
135 { 166 {
136 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);
137 if (clen == (STRLEN)-1) 169 if (clen == (STRLEN)-1)
138 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);
139 } 171 }
140 else 172 else
141 { 173 {
142 uch = ch; 174 uch = ch;
143 clen = 1; 175 clen = 1;
144 } 176 }
145 177
178 if (uch > 0x10FFFFUL)
179 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
180
146 if (uch < 0x80 || enc->flags & F_ASCII) 181 if (uch < 0x80 || enc->flags & F_ASCII)
147 { 182 {
148 if (uch > 0xFFFFUL) 183 if (uch > 0xFFFFUL)
149 { 184 {
150 need (enc, len += 11); 185 need (enc, len += 11);
151 sprintf (enc->cur, "\\u%04x\\u%04x", 186 sprintf (enc->cur, "\\u%04x\\u%04x",
152 (uch - 0x10000) / 0x400 + 0xD800, 187 (int)((uch - 0x10000) / 0x400 + 0xD800),
153 (uch - 0x10000) % 0x400 + 0xDC00); 188 (int)((uch - 0x10000) % 0x400 + 0xDC00));
154 enc->cur += 12; 189 enc->cur += 12;
155 } 190 }
156 else 191 else
157 { 192 {
158 static char hexdigit [16] = "0123456789abcdef"; 193 static char hexdigit [16] = "0123456789abcdef";
… …
176 } 211 }
177 while (--clen); 212 while (--clen);
178 } 213 }
179 else 214 else
180 { 215 {
181 need (enc, len += 10); // never more than 11 bytes needed 216 need (enc, len += UTF8_MAX_LEN - 1); // never more than 11 bytes needed
182 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 217 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
183 ++str; 218 ++str;
184 } 219 }
185 } 220 }
186 } 221 }
… …
188 223
189 --len; 224 --len;
190 } 225 }
191} 226}
192 227
193#define INDENT SB \ 228static void
229encode_indent (enc_t *enc)
230{
194 if (enc->flags & F_INDENT) \ 231 if (enc->flags & F_INDENT)
195 { \ 232 {
196 int i_; \ 233 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
202 234
203#define SPACE SB need (enc, 1); encode_ch (enc, ' '); SE 235 need (enc, spaces);
204#define NL SB if (enc->flags & F_INDENT) { need (enc, 1); encode_ch (enc, '\n'); } SE 236 memset (enc->cur, ' ', spaces);
205#define COMMA SB \ 237 enc->cur += spaces;
238 }
239}
240
241static void
242encode_space (enc_t *enc)
243{
244 need (enc, 1);
206 encode_ch (enc, ','); \ 245 encode_ch (enc, ' ');
246}
247
248static void
249encode_nl (enc_t *enc)
250{
207 if (enc->flags & F_INDENT) \ 251 if (enc->flags & F_INDENT)
208 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);
209 else if (enc->flags & F_SPACE_AFTER) \ 265 else if (enc->flags & F_SPACE_AFTER)
210 SPACE; \ 266 encode_space (enc);
211 SE 267}
212 268
213static void encode_sv (enc_t *enc, SV *sv); 269static void encode_sv (enc_t *enc, SV *sv);
214 270
215static void 271static void
216encode_av (enc_t *enc, AV *av) 272encode_av (enc_t *enc, AV *av)
217{ 273{
218 int i, len = av_len (av); 274 int i, len = av_len (av);
219 275
220 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);
221 ++enc->indent; 280 ++enc->indent;
222 281
223 for (i = 0; i <= len; ++i) 282 for (i = 0; i <= len; ++i)
224 { 283 {
225 INDENT; 284 encode_indent (enc);
226 encode_sv (enc, *av_fetch (av, i, 0)); 285 encode_sv (enc, *av_fetch (av, i, 0));
227 286
228 if (i < len) 287 if (i < len)
229 COMMA; 288 encode_comma (enc);
230 } 289 }
231 290
232 NL; 291 encode_nl (enc);
233 292
234 --enc->indent; 293 --enc->indent;
235 INDENT; encode_ch (enc, ']'); 294 encode_indent (enc); encode_ch (enc, ']');
236} 295}
237 296
238static void 297static void
239encode_he (enc_t *enc, HE *he) 298encode_he (enc_t *enc, HE *he)
240{ 299{
… …
254 else 313 else
255 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 314 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
256 315
257 encode_ch (enc, '"'); 316 encode_ch (enc, '"');
258 317
259 if (enc->flags & F_SPACE_BEFORE) SPACE; 318 if (enc->flags & F_SPACE_BEFORE) encode_space (enc);
260 encode_ch (enc, ':'); 319 encode_ch (enc, ':');
261 if (enc->flags & F_SPACE_AFTER ) SPACE; 320 if (enc->flags & F_SPACE_AFTER ) encode_space (enc);
262 encode_sv (enc, HeVAL (he)); 321 encode_sv (enc, HeVAL (he));
263} 322}
264 323
265// compare hash entries, used when all keys are bytestrings 324// compare hash entries, used when all keys are bytestrings
266static int 325static int
… …
272 HE *b = *(HE **)b_; 331 HE *b = *(HE **)b_;
273 332
274 STRLEN la = HeKLEN (a); 333 STRLEN la = HeKLEN (a);
275 STRLEN lb = HeKLEN (b); 334 STRLEN lb = HeKLEN (b);
276 335
277 if (!(cmp == memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 336 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb)))
278 cmp = la < lb ? -1 : la == lb ? 0 : 1; 337 cmp = la - lb;
279 338
280 return cmp; 339 return cmp;
281} 340}
282 341
283// 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
… …
290static void 349static void
291encode_hv (enc_t *enc, HV *hv) 350encode_hv (enc_t *enc, HV *hv)
292{ 351{
293 int count, i; 352 int count, i;
294 353
354 if (enc->indent >= enc->maxdepth)
355 croak ("data structure too deep (hit recursion limit)");
356
295 encode_ch (enc, '{'); NL; ++enc->indent; 357 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
296 358
297 if ((count = hv_iterinit (hv))) 359 if ((count = hv_iterinit (hv)))
298 { 360 {
299 // for canonical output we have to sort by keys first 361 // for canonical output we have to sort by keys first
300 // actually, this is mostly due to the stupid so-called 362 // actually, this is mostly due to the stupid so-called
301 // security workaround added somewhere in 5.8.x. 363 // security workaround added somewhere in 5.8.x.
302 // that randomises hash orderings 364 // that randomises hash orderings
303 if (enc->flags & F_CANONICAL) 365 if (enc->flags & F_CANONICAL)
304 { 366 {
305 HE *he, *hes [count]; 367 HE *he, *hes [count]; // if your compiler dies here, you need to enable C99 mode
306 int fast = 1; 368 int fast = 1;
307 369
308 i = 0; 370 i = 0;
309 while ((he = hv_iternext (hv))) 371 while ((he = hv_iternext (hv)))
310 { 372 {
… …
317 379
318 if (fast) 380 if (fast)
319 qsort (hes, count, sizeof (HE *), he_cmp_fast); 381 qsort (hes, count, sizeof (HE *), he_cmp_fast);
320 else 382 else
321 { 383 {
322 // hack to disable "use bytes" 384 // hack to forcefully disable "use bytes"
323 COP *oldcop = PL_curcop, cop; 385 COP cop = *PL_curcop;
324 cop.op_private = 0; 386 cop.op_private = 0;
387
388 ENTER;
389 SAVETMPS;
390
391 SAVEVPTR (PL_curcop);
325 PL_curcop = &cop; 392 PL_curcop = &cop;
326 393
327 SAVETMPS;
328 qsort (hes, count, sizeof (HE *), he_cmp_slow); 394 qsort (hes, count, sizeof (HE *), he_cmp_slow);
395
329 FREETMPS; 396 FREETMPS;
330 397 LEAVE;
331 PL_curcop = oldcop;
332 } 398 }
333 399
334 for (i = 0; i < count; ++i) 400 for (i = 0; i < count; ++i)
335 { 401 {
336 INDENT; 402 encode_indent (enc);
337 encode_he (enc, hes [i]); 403 encode_he (enc, hes [i]);
338 404
339 if (i < count - 1) 405 if (i < count - 1)
340 COMMA; 406 encode_comma (enc);
341 } 407 }
342 408
343 NL; 409 encode_nl (enc);
344 } 410 }
345 else 411 else
346 { 412 {
347 SV *sv; 413 SV *sv;
348 HE *he = hv_iternext (hv); 414 HE *he = hv_iternext (hv);
349 415
350 for (;;) 416 for (;;)
351 { 417 {
352 INDENT; 418 encode_indent (enc);
353 encode_he (enc, he); 419 encode_he (enc, he);
354 420
355 if (!(he = hv_iternext (hv))) 421 if (!(he = hv_iternext (hv)))
356 break; 422 break;
357 423
358 COMMA; 424 encode_comma (enc);
359 } 425 }
360 426
361 NL; 427 encode_nl (enc);
362 } 428 }
363 } 429 }
364 430
365 --enc->indent; INDENT; encode_ch (enc, '}'); 431 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
432}
433
434// encode objects, arrays and special \0=false and \1=true values.
435static void
436encode_rv (enc_t *enc, SV *sv)
437{
438 SvGETMAGIC (sv);
439
440 svtype 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))));
366} 459}
367 460
368static void 461static void
369encode_sv (enc_t *enc, SV *sv) 462encode_sv (enc_t *enc, SV *sv)
370{ 463{
… …
391 SvIsUV(sv) 484 SvIsUV(sv)
392 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 485 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv))
393 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 486 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv));
394 } 487 }
395 else if (SvROK (sv)) 488 else if (SvROK (sv))
396 { 489 encode_rv (enc, SvRV (sv));
397 if (!--enc->max_recurse)
398 croak ("data structure too deep (hit recursion limit)");
399
400 sv = SvRV (sv);
401
402 switch (SvTYPE (sv))
403 {
404 case SVt_PVAV: encode_av (enc, (AV *)sv); break;
405 case SVt_PVHV: encode_hv (enc, (HV *)sv); break;
406
407 default:
408 croak ("JSON can only represent references to arrays or hashes");
409 }
410 }
411 else if (!SvOK (sv)) 490 else if (!SvOK (sv))
412 encode_str (enc, "null", 4, 0); 491 encode_str (enc, "null", 4, 0);
413 else 492 else
414 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));
415} 495}
416 496
417static SV * 497static SV *
418encode_json (SV *scalar, UV flags) 498encode_json (SV *scalar, U32 flags)
419{ 499{
420 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 500 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar))
421 croak ("hash- or arraref required (not a simple scalar, use allow_nonref to allow this)"); 501 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
422 502
423 enc_t enc; 503 enc_t enc;
424 enc.flags = flags; 504 enc.flags = flags;
425 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 505 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
426 enc.cur = SvPVX (enc.sv); 506 enc.cur = SvPVX (enc.sv);
427 enc.end = SvEND (enc.sv); 507 enc.end = SvEND (enc.sv);
428 enc.max_recurse = 0;
429 enc.indent = 0; 508 enc.indent = 0;
509 enc.maxdepth = DEC_DEPTH (flags);
430 510
431 SvPOK_only (enc.sv); 511 SvPOK_only (enc.sv);
432 encode_sv (&enc, scalar); 512 encode_sv (&enc, scalar);
433 513
434 if (!(flags & (F_ASCII | F_UTF8))) 514 if (!(flags & (F_ASCII | F_UTF8)))
… …
441 521
442 return enc.sv; 522 return enc.sv;
443} 523}
444 524
445///////////////////////////////////////////////////////////////////////////// 525/////////////////////////////////////////////////////////////////////////////
526// decoder
446 527
447#define WS \ 528// structure used for decoding JSON
529typedef struct
530{
531 char *cur; // current parser pointer
532 char *end; // end of input string
533 const char *err; // parse error, if != 0
534 U32 flags; // F_*
535 U32 depth; // recursion depth
536 U32 maxdepth; // recursion depth limit
537} dec_t;
538
539static void
540decode_ws (dec_t *dec)
541{
448 for (;;) \ 542 for (;;)
449 { \ 543 {
450 char ch = *dec->cur; \ 544 char ch = *dec->cur;
545
451 if (ch > 0x20 \ 546 if (ch > 0x20
452 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) \ 547 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09))
453 break; \ 548 break;
549
454 ++dec->cur; \ 550 ++dec->cur;
455 } 551 }
552}
456 553
457#define ERR(reason) SB dec->err = reason; goto fail; SE 554#define ERR(reason) SB dec->err = reason; goto fail; SE
555
458#define EXPECT_CH(ch) SB \ 556#define EXPECT_CH(ch) SB \
459 if (*dec->cur != ch) \ 557 if (*dec->cur != ch) \
460 ERR (# ch " expected"); \ 558 ERR (# ch " expected"); \
461 ++dec->cur; \ 559 ++dec->cur; \
462 SE 560 SE
463 561
562#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)")
563#define DEC_DEC_DEPTH --dec->depth
564
464static SV *decode_sv (dec_t *dec); 565static SV *decode_sv (dec_t *dec);
465 566
466static signed char decode_hexdigit[256]; 567static signed char decode_hexdigit[256];
467 568
468static UV 569static UV
469decode_4hex (dec_t *dec) 570decode_4hex (dec_t *dec)
470{ 571{
471 signed char d1, d2, d3, d4; 572 signed char d1, d2, d3, d4;
573 unsigned char *cur = (unsigned char *)dec->cur;
472 574
473 d1 = decode_hexdigit [((unsigned char *)dec->cur) [0]];
474 if (d1 < 0) ERR ("four hexadecimal digits expected"); 575 d1 = decode_hexdigit [cur [0]]; if (d1 < 0) ERR ("four hexadecimal digits expected");
475 d2 = decode_hexdigit [((unsigned char *)dec->cur) [1]];
476 if (d2 < 0) ERR ("four hexadecimal digits expected"); 576 d2 = decode_hexdigit [cur [1]]; if (d2 < 0) ERR ("four hexadecimal digits expected");
477 d3 = decode_hexdigit [((unsigned char *)dec->cur) [2]];
478 if (d3 < 0) ERR ("four hexadecimal digits expected"); 577 d3 = decode_hexdigit [cur [2]]; if (d3 < 0) ERR ("four hexadecimal digits expected");
479 d4 = decode_hexdigit [((unsigned char *)dec->cur) [3]];
480 if (d4 < 0) ERR ("four hexadecimal digits expected"); 578 d4 = decode_hexdigit [cur [3]]; if (d4 < 0) ERR ("four hexadecimal digits expected");
481 579
482 dec->cur += 4; 580 dec->cur += 4;
483 581
484 return ((UV)d1) << 12 582 return ((UV)d1) << 12
485 | ((UV)d2) << 8 583 | ((UV)d2) << 8
… …
488 586
489fail: 587fail:
490 return (UV)-1; 588 return (UV)-1;
491} 589}
492 590
493#define APPEND_GROW(n) SB \
494 if (cur + (n) >= end) \
495 { \
496 STRLEN ofs = cur - SvPVX (sv); \
497 SvGROW (sv, ofs + (n) + 1); \
498 cur = SvPVX (sv) + ofs; \
499 end = SvEND (sv); \
500 } \
501 SE
502
503#define APPEND_CH(ch) SB \
504 APPEND_GROW (1); \
505 *cur++ = (ch); \
506 SE
507
508static SV * 591static SV *
509decode_str (dec_t *dec) 592decode_str (dec_t *dec)
510{ 593{
511 SV *sv = NEWSV (0,2); 594 SV *sv = 0;
512 int utf8 = 0; 595 int utf8 = 0;
513 char *cur = SvPVX (sv);
514 char *end = SvEND (sv);
515 596
516 for (;;) 597 do
517 { 598 {
518 unsigned char ch = *(unsigned char *)dec->cur; 599 char buf [SHORT_STRING_LEN + UTF8_MAX_LEN];
600 char *cur = buf;
519 601
520 if (ch == '"') 602 do
521 break;
522 else if (ch == '\\')
523 { 603 {
524 switch (*++dec->cur) 604 unsigned char ch = *(unsigned char *)dec->cur++;
605
606 if (ch == '"')
525 { 607 {
526 case '\\': 608 --dec->cur;
527 case '/': 609 break;
528 case '"': APPEND_CH (*dec->cur++); break; 610 }
529 611 else if (ch == '\\')
530 case 'b': APPEND_CH ('\010'); ++dec->cur; break; 612 {
531 case 't': APPEND_CH ('\011'); ++dec->cur; break; 613 switch (*dec->cur)
532 case 'n': APPEND_CH ('\012'); ++dec->cur; break;
533 case 'f': APPEND_CH ('\014'); ++dec->cur; break;
534 case 'r': APPEND_CH ('\015'); ++dec->cur; break;
535
536 case 'u':
537 { 614 {
538 UV lo, hi; 615 case '\\':
539 ++dec->cur; 616 case '/':
617 case '"': *cur++ = *dec->cur++; break;
540 618
541 hi = decode_4hex (dec); 619 case 'b': ++dec->cur; *cur++ = '\010'; break;
542 if (hi == (UV)-1) 620 case 't': ++dec->cur; *cur++ = '\011'; break;
543 goto fail; 621 case 'n': ++dec->cur; *cur++ = '\012'; break;
622 case 'f': ++dec->cur; *cur++ = '\014'; break;
623 case 'r': ++dec->cur; *cur++ = '\015'; break;
544 624
545 // possibly a surrogate pair 625 case 'u':
546 if (hi >= 0xd800 && hi < 0xdc00)
547 { 626 {
548 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 627 UV lo, hi;
549 ERR ("missing low surrogate character in surrogate pair");
550
551 dec->cur += 2; 628 ++dec->cur;
552 629
553 lo = decode_4hex (dec); 630 hi = decode_4hex (dec);
554 if (lo == (UV)-1) 631 if (hi == (UV)-1)
555 goto fail; 632 goto fail;
556 633
634 // possibly a surrogate pair
635 if (hi >= 0xd800)
636 if (hi < 0xdc00)
637 {
638 if (dec->cur [0] != '\\' || dec->cur [1] != 'u')
639 ERR ("missing low surrogate character in surrogate pair");
640
641 dec->cur += 2;
642
643 lo = decode_4hex (dec);
644 if (lo == (UV)-1)
645 goto fail;
646
557 if (lo < 0xdc00 || lo >= 0xe000) 647 if (lo < 0xdc00 || lo >= 0xe000)
558 ERR ("surrogate pair expected"); 648 ERR ("surrogate pair expected");
559 649
560 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000; 650 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000;
651 }
652 else if (hi < 0xe000)
653 ERR ("missing high surrogate character in surrogate pair");
654
655 if (hi >= 0x80)
656 {
657 utf8 = 1;
658
659 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
660 }
661 else
662 *cur++ = hi;
561 } 663 }
562 else if (hi >= 0xdc00 && hi < 0xe000)
563 ERR ("missing high surrogate character in surrogate pair");
564
565 if (hi >= 0x80)
566 { 664 break;
567 utf8 = 1;
568 665
569 APPEND_GROW (4); // at most 4 bytes for 21 bits
570 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
571 }
572 else 666 default:
573 APPEND_CH (hi); 667 --dec->cur;
668 ERR ("illegal backslash escape sequence in string");
574 } 669 }
575 break; 670 }
576 671 else if (ch >= 0x20 && ch <= 0x7f)
577 default: 672 *cur++ = ch;
673 else if (ch >= 0x80)
674 {
578 --dec->cur; 675 --dec->cur;
579 ERR ("illegal backslash escape sequence in string"); 676
677 STRLEN clen;
678 UV uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen);
679 if (clen == (STRLEN)-1)
680 ERR ("malformed UTF-8 character in JSON string");
681
682 do
683 *cur++ = *dec->cur++;
684 while (--clen);
685
686 utf8 = 1;
687 }
688 else
689 {
690 --dec->cur;
691
692 if (!ch)
693 ERR ("unexpected end of string while parsing json string");
694 else
695 ERR ("invalid character encountered while parsing json string");
580 } 696 }
581 } 697 }
582 else if (ch >= 0x20 && ch <= 0x7f) 698 while (cur < buf + SHORT_STRING_LEN);
583 APPEND_CH (*dec->cur++); 699
584 else if (ch >= 0x80) 700 STRLEN len = cur - buf;
701
702 if (sv)
585 { 703 {
586 STRLEN clen; 704 SvGROW (sv, SvCUR (sv) + len + 1);
587 UV uch = utf8n_to_uvuni (dec->cur, dec->end - dec->cur, &clen, UTF8_CHECK_ONLY); 705 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
588 if (clen == (STRLEN)-1) 706 SvCUR_set (sv, SvCUR (sv) + len);
589 ERR ("malformed UTF-8 character in JSON string");
590
591 APPEND_GROW (clen);
592 do
593 {
594 *cur++ = *dec->cur++;
595 }
596 while (--clen);
597
598 utf8 = 1;
599 } 707 }
600 else if (dec->cur == dec->end)
601 ERR ("unexpected end of string while parsing json string");
602 else 708 else
603 ERR ("invalid character encountered"); 709 sv = newSVpvn (buf, len);
604 } 710 }
711 while (*dec->cur != '"');
605 712
606 ++dec->cur; 713 ++dec->cur;
607 714
608 SvCUR_set (sv, cur - SvPVX (sv)); 715 if (sv)
609 716 {
610 SvPOK_only (sv); 717 SvPOK_only (sv);
611 *SvEND (sv) = 0; 718 *SvEND (sv) = 0;
612 719
613 if (utf8) 720 if (utf8)
614 SvUTF8_on (sv); 721 SvUTF8_on (sv);
615 722 }
616 if (dec->flags & F_SHRINK) 723 else
617 shrink (sv); 724 sv = newSVpvn ("", 0);
618 725
619 return sv; 726 return sv;
620 727
621fail: 728fail:
622 SvREFCNT_dec (sv);
623 return 0; 729 return 0;
624} 730}
625 731
626static SV * 732static SV *
627decode_num (dec_t *dec) 733decode_num (dec_t *dec)
… …
708static SV * 814static SV *
709decode_av (dec_t *dec) 815decode_av (dec_t *dec)
710{ 816{
711 AV *av = newAV (); 817 AV *av = newAV ();
712 818
713 WS; 819 DEC_INC_DEPTH;
820 decode_ws (dec);
821
714 if (*dec->cur == ']') 822 if (*dec->cur == ']')
715 ++dec->cur; 823 ++dec->cur;
716 else 824 else
717 for (;;) 825 for (;;)
718 { 826 {
… …
722 if (!value) 830 if (!value)
723 goto fail; 831 goto fail;
724 832
725 av_push (av, value); 833 av_push (av, value);
726 834
727 WS; 835 decode_ws (dec);
728 836
729 if (*dec->cur == ']') 837 if (*dec->cur == ']')
730 { 838 {
731 ++dec->cur; 839 ++dec->cur;
732 break; 840 break;
… …
736 ERR (", or ] expected while parsing array"); 844 ERR (", or ] expected while parsing array");
737 845
738 ++dec->cur; 846 ++dec->cur;
739 } 847 }
740 848
849 DEC_DEC_DEPTH;
741 return newRV_noinc ((SV *)av); 850 return newRV_noinc ((SV *)av);
742 851
743fail: 852fail:
744 SvREFCNT_dec (av); 853 SvREFCNT_dec (av);
854 DEC_DEC_DEPTH;
745 return 0; 855 return 0;
746} 856}
747 857
748static SV * 858static SV *
749decode_hv (dec_t *dec) 859decode_hv (dec_t *dec)
750{ 860{
751 HV *hv = newHV (); 861 HV *hv = newHV ();
752 862
753 WS; 863 DEC_INC_DEPTH;
864 decode_ws (dec);
865
754 if (*dec->cur == '}') 866 if (*dec->cur == '}')
755 ++dec->cur; 867 ++dec->cur;
756 else 868 else
757 for (;;) 869 for (;;)
758 { 870 {
759 SV *key, *value; 871 SV *key, *value;
760 872
761 WS; EXPECT_CH ('"'); 873 decode_ws (dec); EXPECT_CH ('"');
762 874
763 key = decode_str (dec); 875 key = decode_str (dec);
764 if (!key) 876 if (!key)
765 goto fail; 877 goto fail;
766 878
767 WS; EXPECT_CH (':'); 879 decode_ws (dec); EXPECT_CH (':');
768 880
769 value = decode_sv (dec); 881 value = decode_sv (dec);
770 if (!value) 882 if (!value)
771 { 883 {
772 SvREFCNT_dec (key); 884 SvREFCNT_dec (key);
773 goto fail; 885 goto fail;
774 } 886 }
775 887
776 //TODO: optimise
777 hv_store_ent (hv, key, value, 0); 888 hv_store_ent (hv, key, value, 0);
889 SvREFCNT_dec (key);
778 890
779 WS; 891 decode_ws (dec);
780 892
781 if (*dec->cur == '}') 893 if (*dec->cur == '}')
782 { 894 {
783 ++dec->cur; 895 ++dec->cur;
784 break; 896 break;
… …
788 ERR (", or } expected while parsing object/hash"); 900 ERR (", or } expected while parsing object/hash");
789 901
790 ++dec->cur; 902 ++dec->cur;
791 } 903 }
792 904
905 DEC_DEC_DEPTH;
793 return newRV_noinc ((SV *)hv); 906 return newRV_noinc ((SV *)hv);
794 907
795fail: 908fail:
796 SvREFCNT_dec (hv); 909 SvREFCNT_dec (hv);
910 DEC_DEC_DEPTH;
797 return 0; 911 return 0;
798} 912}
799 913
800static SV * 914static SV *
801decode_sv (dec_t *dec) 915decode_sv (dec_t *dec)
802{ 916{
803 WS; 917 decode_ws (dec);
804 switch (*dec->cur) 918 switch (*dec->cur)
805 { 919 {
806 case '"': ++dec->cur; return decode_str (dec); 920 case '"': ++dec->cur; return decode_str (dec);
807 case '[': ++dec->cur; return decode_av (dec); 921 case '[': ++dec->cur; return decode_av (dec);
808 case '{': ++dec->cur; return decode_hv (dec); 922 case '{': ++dec->cur; return decode_hv (dec);
… …
853fail: 967fail:
854 return 0; 968 return 0;
855} 969}
856 970
857static SV * 971static SV *
858decode_json (SV *string, UV flags) 972decode_json (SV *string, U32 flags)
859{ 973{
860 SV *sv; 974 SV *sv;
975
976 SvUPGRADE (string, SVt_PV);
861 977
862 if (flags & F_UTF8) 978 if (flags & F_UTF8)
863 sv_utf8_downgrade (string, 0); 979 sv_utf8_downgrade (string, 0);
864 else 980 else
865 sv_utf8_upgrade (string); 981 sv_utf8_upgrade (string);
866 982
867 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 983 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
868 984
869 dec_t dec; 985 dec_t dec;
870 dec.flags = flags; 986 dec.flags = flags;
871 dec.cur = SvPVX (string); 987 dec.cur = SvPVX (string);
872 dec.end = SvEND (string); 988 dec.end = SvEND (string);
873 dec.err = 0; 989 dec.err = 0;
990 dec.depth = 0;
991 dec.maxdepth = DEC_DEPTH (dec.flags);
874 992
993 *dec.end = 0; // this should basically be a nop, too, but make sure its there
875 sv = decode_sv (&dec); 994 sv = decode_sv (&dec);
876 995
877 if (!sv) 996 if (!sv)
878 { 997 {
879 IV offset = dec.flags & F_UTF8 998 IV offset = dec.flags & F_UTF8
880 ? dec.cur - SvPVX (string) 999 ? dec.cur - SvPVX (string)
881 : utf8_distance (dec.cur, SvPVX (string)); 1000 : utf8_distance (dec.cur, SvPVX (string));
882 SV *uni = sv_newmortal (); 1001 SV *uni = sv_newmortal ();
1002
883 // horrible hack to silence warning inside pv_uni_display 1003 // horrible hack to silence warning inside pv_uni_display
884 COP cop; 1004 COP cop = *PL_curcop;
885 memset (&cop, 0, sizeof (cop));
886 cop.cop_warnings = pWARN_NONE; 1005 cop.cop_warnings = pWARN_NONE;
1006 ENTER;
887 SAVEVPTR (PL_curcop); 1007 SAVEVPTR (PL_curcop);
888 PL_curcop = &cop; 1008 PL_curcop = &cop;
889
890 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1009 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1010 LEAVE;
1011
891 croak ("%s, at character offset %d (%s)", 1012 croak ("%s, at character offset %d [\"%s\"]",
892 dec.err, 1013 dec.err,
893 (int)offset, 1014 (int)offset,
894 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1015 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
895 } 1016 }
896 1017
897 sv = sv_2mortal (sv); 1018 sv = sv_2mortal (sv);
898 1019
899 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1020 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv))
900 croak ("JSON object or array expected (but number, string, true, false or null found, use allow_nonref to allow this)"); 1021 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
901 1022
902 return sv; 1023 return sv;
903} 1024}
904 1025
1026/////////////////////////////////////////////////////////////////////////////
1027// XS interface functions
1028
905MODULE = JSON::XS PACKAGE = JSON::XS 1029MODULE = JSON::XS PACKAGE = JSON::XS
906 1030
907BOOT: 1031BOOT:
908{ 1032{
909 int i; 1033 int i;
910 1034
911 memset (decode_hexdigit, 0xff, 256); 1035 memset (decode_hexdigit, 0xff, 256);
1036
912 for (i = 10; i--; ) 1037 for (i = 0; i < 256; ++i)
913 decode_hexdigit ['0' + i] = i; 1038 decode_hexdigit [i] =
914 1039 i >= '0' && i <= '9' ? i - '0'
915 for (i = 7; i--; ) 1040 : i >= 'a' && i <= 'f' ? i - 'a' + 10
916 { 1041 : i >= 'A' && i <= 'F' ? i - 'A' + 10
917 decode_hexdigit ['a' + i] = 10 + i; 1042 : -1;
918 decode_hexdigit ['A' + i] = 10 + i;
919 }
920 1043
921 json_stash = gv_stashpv ("JSON::XS", 1); 1044 json_stash = gv_stashpv ("JSON::XS", 1);
922} 1045}
923 1046
924PROTOTYPES: DISABLE 1047PROTOTYPES: DISABLE
… …
935 utf8 = F_UTF8 1058 utf8 = F_UTF8
936 indent = F_INDENT 1059 indent = F_INDENT
937 canonical = F_CANONICAL 1060 canonical = F_CANONICAL
938 space_before = F_SPACE_BEFORE 1061 space_before = F_SPACE_BEFORE
939 space_after = F_SPACE_AFTER 1062 space_after = F_SPACE_AFTER
940 json_rpc = F_JSON_RPC
941 pretty = F_PRETTY 1063 pretty = F_PRETTY
942 allow_nonref = F_ALLOW_NONREF 1064 allow_nonref = F_ALLOW_NONREF
943 shrink = F_SHRINK 1065 shrink = F_SHRINK
944 CODE: 1066 CODE:
945{ 1067{
… …
952 RETVAL = newSVsv (self); 1074 RETVAL = newSVsv (self);
953} 1075}
954 OUTPUT: 1076 OUTPUT:
955 RETVAL 1077 RETVAL
956 1078
1079SV *max_depth (SV *self, int max_depth = 0x80000000UL)
1080 CODE:
1081{
1082 UV *uv = SvJSON (self);
1083 UV log2 = 0;
1084
1085 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1086
1087 while ((1UL << log2) < max_depth)
1088 ++log2;
1089
1090 *uv = *uv & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1091
1092 RETVAL = newSVsv (self);
1093}
1094 OUTPUT:
1095 RETVAL
1096
957void encode (SV *self, SV *scalar) 1097void encode (SV *self, SV *scalar)
958 PPCODE: 1098 PPCODE:
959 XPUSHs (encode_json (scalar, *SvJSON (self))); 1099 XPUSHs (encode_json (scalar, *SvJSON (self)));
960 1100
961void decode (SV *self, SV *jsonstr) 1101void decode (SV *self, SV *jsonstr)
… …
963 XPUSHs (decode_json (jsonstr, *SvJSON (self))); 1103 XPUSHs (decode_json (jsonstr, *SvJSON (self)));
964 1104
965PROTOTYPES: ENABLE 1105PROTOTYPES: ENABLE
966 1106
967void to_json (SV *scalar) 1107void to_json (SV *scalar)
1108 ALIAS:
1109 objToJson = 0
968 PPCODE: 1110 PPCODE:
969 XPUSHs (encode_json (scalar, F_UTF8)); 1111 XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8));
970 1112
971void from_json (SV *jsonstr) 1113void from_json (SV *jsonstr)
1114 ALIAS:
1115 jsonToObj = 0
972 PPCODE: 1116 PPCODE:
973 XPUSHs (decode_json (jsonstr, F_UTF8)); 1117 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8));
974 1118

Diff Legend

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