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 1.9 by root, Fri Mar 23 17:40:29 2007 UTC vs.
Revision 1.27 by root, Mon Apr 9 05:09:57 2007 UTC

… …
3#include "XSUB.h" 3#include "XSUB.h"
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#include "stdio.h"
8 9
10#if defined(__BORLANDC__) || defined(_MSC_VER)
11# define snprintf _snprintf // C compilers have this in stdio.h
12#endif
13
9#define F_ASCII 0x00000001 14#define F_ASCII 0x00000001UL
10#define F_UTF8 0x00000002 15#define F_UTF8 0x00000002UL
11#define F_INDENT 0x00000004 16#define F_INDENT 0x00000004UL
12#define F_CANONICAL 0x00000008 17#define F_CANONICAL 0x00000008UL
13#define F_SPACE_BEFORE 0x00000010 18#define F_SPACE_BEFORE 0x00000010UL
14#define F_SPACE_AFTER 0x00000020 19#define F_SPACE_AFTER 0x00000020UL
15#define F_JSON_RPC 0x00000040
16#define F_ALLOW_NONREF 0x00000080 20#define F_ALLOW_NONREF 0x00000080UL
17#define F_SHRINK 0x00000100 21#define F_SHRINK 0x00000100UL
22#define F_MAXDEPTH 0xf8000000UL
23#define S_MAXDEPTH 27
24
25#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
26
27// F_SELFCONVERT? <=> to_json/toJson
28// F_BLESSED? <=> { $__class__$ => }
18 29
19#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 30#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
20#define F_DEFAULT 0 31#define F_DEFAULT (9UL << S_MAXDEPTH)
21 32
22#define INIT_SIZE 32 // initial scalar size to be allocated 33#define INIT_SIZE 32 // initial scalar size to be allocated
34#define INDENT_STEP 3 // spaces per indentation level
35
36#define UTF8_MAX_LEN 11 // for perls UTF-X: max. number of octets per character
37#define SHORT_STRING_LEN 512 // special-case strings of up to this size
23 38
24#define SB do { 39#define SB do {
25#define SE } while (0) 40#define SE } while (0)
26 41
27static HV *json_stash; 42static HV *json_stash; // JSON::XS::
43
44/////////////////////////////////////////////////////////////////////////////
45// utility functions
46
47static UV *
48SvJSON (SV *sv)
49{
50 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash))
51 croak ("object is not of type JSON::XS");
52
53 return &SvUVX (SvRV (sv));
54}
55
56static void
57shrink (SV *sv)
58{
59 sv_utf8_downgrade (sv, 1);
60 if (SvLEN (sv) > SvCUR (sv) + 1)
61 {
62#ifdef SvPV_shrink_to_cur
63 SvPV_shrink_to_cur (sv);
64#elif defined (SvPV_renew)
65 SvPV_renew (sv, SvCUR (sv) + 1);
66#endif
67 }
68}
69
70// decode an utf-8 character and return it, or (UV)-1 in
71// case of an error.
72// we special-case "safe" characters from U+80 .. U+7FF,
73// but use the very good perl function to parse anything else.
74// note that we never call this function for a ascii codepoints
75static UV
76decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
77{
78 if (s[0] > 0xdf || s[0] < 0xc2)
79 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
80 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)
81 {
82 *clen = 2;
83 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
84 }
85 else
86 {
87 *clen = (STRLEN)-1;
88 return (UV)-1;
89 }
90}
91
92/////////////////////////////////////////////////////////////////////////////
93// encoder
28 94
29// structure used for encoding JSON 95// structure used for encoding JSON
30typedef struct 96typedef struct
31{ 97{
32 char *cur; 98 char *cur; // SvPVX (sv) + current output position
33 STRLEN len; // SvLEN (sv)
34 char *end; // SvEND (sv) 99 char *end; // SvEND (sv)
35 SV *sv; 100 SV *sv; // result scalar
36 UV flags; 101 U32 flags; // F_*
37 int max_recurse; 102 U32 indent; // indentation level
38 int indent; 103 U32 maxdepth; // max. indentation/recursion level
39} enc_t; 104} 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/////////////////////////////////////////////////////////////////////////////
69 105
70static void 106static void
71need (enc_t *enc, STRLEN len) 107need (enc_t *enc, STRLEN len)
72{ 108{
73 if (enc->cur + len >= enc->end) 109 if (enc->cur + len >= enc->end)
74 { 110 {
75 STRLEN cur = enc->cur - SvPVX (enc->sv); 111 STRLEN cur = enc->cur - SvPVX (enc->sv);
76 SvGROW (enc->sv, cur + len + 1); 112 SvGROW (enc->sv, cur + len + 1);
77 enc->cur = SvPVX (enc->sv) + cur; 113 enc->cur = SvPVX (enc->sv) + cur;
78 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv); 114 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
79 } 115 }
80} 116}
81 117
82static void 118static void
83encode_ch (enc_t *enc, char ch) 119encode_ch (enc_t *enc, char ch)
… …
131 STRLEN clen; 167 STRLEN clen;
132 UV uch; 168 UV uch;
133 169
134 if (is_utf8) 170 if (is_utf8)
135 { 171 {
136 uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY); 172 //uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
173 uch = decode_utf8 (str, end - str, &clen);
137 if (clen == (STRLEN)-1) 174 if (clen == (STRLEN)-1)
138 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str); 175 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
139 } 176 }
140 else 177 else
141 { 178 {
… …
150 { 187 {
151 if (uch > 0xFFFFUL) 188 if (uch > 0xFFFFUL)
152 { 189 {
153 need (enc, len += 11); 190 need (enc, len += 11);
154 sprintf (enc->cur, "\\u%04x\\u%04x", 191 sprintf (enc->cur, "\\u%04x\\u%04x",
155 (uch - 0x10000) / 0x400 + 0xD800, 192 (int)((uch - 0x10000) / 0x400 + 0xD800),
156 (uch - 0x10000) % 0x400 + 0xDC00); 193 (int)((uch - 0x10000) % 0x400 + 0xDC00));
157 enc->cur += 12; 194 enc->cur += 12;
158 } 195 }
159 else 196 else
160 { 197 {
161 static char hexdigit [16] = "0123456789abcdef"; 198 static char hexdigit [16] = "0123456789abcdef";
… …
179 } 216 }
180 while (--clen); 217 while (--clen);
181 } 218 }
182 else 219 else
183 { 220 {
184 need (enc, len += 10); // never more than 11 bytes needed 221 need (enc, len += UTF8_MAX_LEN - 1); // never more than 11 bytes needed
185 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 222 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
186 ++str; 223 ++str;
187 } 224 }
188 } 225 }
189 } 226 }
… …
191 228
192 --len; 229 --len;
193 } 230 }
194} 231}
195 232
196#define INDENT SB \ 233static void
234encode_indent (enc_t *enc)
235{
197 if (enc->flags & F_INDENT) \ 236 if (enc->flags & F_INDENT)
198 { \ 237 {
199 int i_; \ 238 int spaces = enc->indent * INDENT_STEP;
200 need (enc, enc->indent); \
201 for (i_ = enc->indent * 3; i_--; )\
202 encode_ch (enc, ' '); \
203 } \
204 SE
205 239
206#define SPACE SB need (enc, 1); encode_ch (enc, ' '); SE 240 need (enc, spaces);
207#define NL SB if (enc->flags & F_INDENT) { need (enc, 1); encode_ch (enc, '\n'); } SE 241 memset (enc->cur, ' ', spaces);
208#define COMMA SB \ 242 enc->cur += spaces;
243 }
244}
245
246static void
247encode_space (enc_t *enc)
248{
249 need (enc, 1);
209 encode_ch (enc, ','); \ 250 encode_ch (enc, ' ');
251}
252
253static void
254encode_nl (enc_t *enc)
255{
210 if (enc->flags & F_INDENT) \ 256 if (enc->flags & F_INDENT)
211 NL; \ 257 {
258 need (enc, 1);
259 encode_ch (enc, '\n');
260 }
261}
262
263static void
264encode_comma (enc_t *enc)
265{
266 encode_ch (enc, ',');
267
268 if (enc->flags & F_INDENT)
269 encode_nl (enc);
212 else if (enc->flags & F_SPACE_AFTER) \ 270 else if (enc->flags & F_SPACE_AFTER)
213 SPACE; \ 271 encode_space (enc);
214 SE 272}
215 273
216static void encode_sv (enc_t *enc, SV *sv); 274static void encode_sv (enc_t *enc, SV *sv);
217 275
218static void 276static void
219encode_av (enc_t *enc, AV *av) 277encode_av (enc_t *enc, AV *av)
220{ 278{
221 int i, len = av_len (av); 279 int i, len = av_len (av);
222 280
223 encode_ch (enc, '['); NL; 281 if (enc->indent >= enc->maxdepth)
282 croak ("data structure too deep (hit recursion limit)");
283
284 encode_ch (enc, '['); encode_nl (enc);
224 ++enc->indent; 285 ++enc->indent;
225 286
226 for (i = 0; i <= len; ++i) 287 for (i = 0; i <= len; ++i)
227 { 288 {
228 INDENT; 289 encode_indent (enc);
229 encode_sv (enc, *av_fetch (av, i, 0)); 290 encode_sv (enc, *av_fetch (av, i, 0));
230 291
231 if (i < len) 292 if (i < len)
232 COMMA; 293 encode_comma (enc);
233 } 294 }
234 295
235 NL; 296 encode_nl (enc);
236 297
237 --enc->indent; 298 --enc->indent;
238 INDENT; encode_ch (enc, ']'); 299 encode_indent (enc); encode_ch (enc, ']');
239} 300}
240 301
241static void 302static void
242encode_he (enc_t *enc, HE *he) 303encode_he (enc_t *enc, HE *he)
243{ 304{
… …
257 else 318 else
258 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 319 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
259 320
260 encode_ch (enc, '"'); 321 encode_ch (enc, '"');
261 322
262 if (enc->flags & F_SPACE_BEFORE) SPACE; 323 if (enc->flags & F_SPACE_BEFORE) encode_space (enc);
263 encode_ch (enc, ':'); 324 encode_ch (enc, ':');
264 if (enc->flags & F_SPACE_AFTER ) SPACE; 325 if (enc->flags & F_SPACE_AFTER ) encode_space (enc);
265 encode_sv (enc, HeVAL (he)); 326 encode_sv (enc, HeVAL (he));
266} 327}
267 328
268// compare hash entries, used when all keys are bytestrings 329// compare hash entries, used when all keys are bytestrings
269static int 330static int
… …
275 HE *b = *(HE **)b_; 336 HE *b = *(HE **)b_;
276 337
277 STRLEN la = HeKLEN (a); 338 STRLEN la = HeKLEN (a);
278 STRLEN lb = HeKLEN (b); 339 STRLEN lb = HeKLEN (b);
279 340
280 if (!(cmp == memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 341 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb)))
281 cmp = la < lb ? -1 : la == lb ? 0 : 1; 342 cmp = la - lb;
282 343
283 return cmp; 344 return cmp;
284} 345}
285 346
286// compare hash entries, used when some keys are sv's or utf-x 347// compare hash entries, used when some keys are sv's or utf-x
… …
293static void 354static void
294encode_hv (enc_t *enc, HV *hv) 355encode_hv (enc_t *enc, HV *hv)
295{ 356{
296 int count, i; 357 int count, i;
297 358
359 if (enc->indent >= enc->maxdepth)
360 croak ("data structure too deep (hit recursion limit)");
361
298 encode_ch (enc, '{'); NL; ++enc->indent; 362 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
299 363
300 if ((count = hv_iterinit (hv))) 364 if ((count = hv_iterinit (hv)))
301 { 365 {
302 // for canonical output we have to sort by keys first 366 // for canonical output we have to sort by keys first
303 // actually, this is mostly due to the stupid so-called 367 // actually, this is mostly due to the stupid so-called
304 // security workaround added somewhere in 5.8.x. 368 // security workaround added somewhere in 5.8.x.
305 // that randomises hash orderings 369 // that randomises hash orderings
306 if (enc->flags & F_CANONICAL) 370 if (enc->flags & F_CANONICAL)
307 { 371 {
308 HE *he, *hes [count]; 372 HE *he, *hes [count]; // if your compiler dies here, you need to enable C99 mode
309 int fast = 1; 373 int fast = 1;
310 374
311 i = 0; 375 i = 0;
312 while ((he = hv_iternext (hv))) 376 while ((he = hv_iternext (hv)))
313 { 377 {
… …
338 LEAVE; 402 LEAVE;
339 } 403 }
340 404
341 for (i = 0; i < count; ++i) 405 for (i = 0; i < count; ++i)
342 { 406 {
343 INDENT; 407 encode_indent (enc);
344 encode_he (enc, hes [i]); 408 encode_he (enc, hes [i]);
345 409
346 if (i < count - 1) 410 if (i < count - 1)
347 COMMA; 411 encode_comma (enc);
348 } 412 }
349 413
350 NL; 414 encode_nl (enc);
351 } 415 }
352 else 416 else
353 { 417 {
354 SV *sv;
355 HE *he = hv_iternext (hv); 418 HE *he = hv_iternext (hv);
356 419
357 for (;;) 420 for (;;)
358 { 421 {
359 INDENT; 422 encode_indent (enc);
360 encode_he (enc, he); 423 encode_he (enc, he);
361 424
362 if (!(he = hv_iternext (hv))) 425 if (!(he = hv_iternext (hv)))
363 break; 426 break;
364 427
365 COMMA; 428 encode_comma (enc);
366 } 429 }
367 430
368 NL; 431 encode_nl (enc);
369 } 432 }
370 } 433 }
371 434
372 --enc->indent; INDENT; encode_ch (enc, '}'); 435 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
436}
437
438// encode objects, arrays and special \0=false and \1=true values.
439static void
440encode_rv (enc_t *enc, SV *sv)
441{
442 svtype svt;
443
444 SvGETMAGIC (sv);
445 svt = SvTYPE (sv);
446
447 if (svt == SVt_PVHV)
448 encode_hv (enc, (HV *)sv);
449 else if (svt == SVt_PVAV)
450 encode_av (enc, (AV *)sv);
451 else if (svt < SVt_PVAV)
452 {
453 if (SvNIOK (sv) && SvIV (sv) == 0)
454 encode_str (enc, "false", 5, 0);
455 else if (SvNIOK (sv) && SvIV (sv) == 1)
456 encode_str (enc, "true", 4, 0);
457 else
458 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
459 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
460 }
461 else
462 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
463 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
373} 464}
374 465
375static void 466static void
376encode_sv (enc_t *enc, SV *sv) 467encode_sv (enc_t *enc, SV *sv)
377{ 468{
… …
398 SvIsUV(sv) 489 SvIsUV(sv)
399 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 490 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv))
400 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 491 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv));
401 } 492 }
402 else if (SvROK (sv)) 493 else if (SvROK (sv))
403 { 494 encode_rv (enc, SvRV (sv));
404 SV *rv = SvRV (sv);
405
406 if (!--enc->max_recurse)
407 croak ("data structure too deep (hit recursion limit)");
408
409 switch (SvTYPE (rv))
410 {
411 case SVt_PVAV: encode_av (enc, (AV *)rv); break;
412 case SVt_PVHV: encode_hv (enc, (HV *)rv); break;
413
414 default:
415 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
416 SvPV_nolen (sv));
417 }
418 }
419 else if (!SvOK (sv)) 495 else if (!SvOK (sv))
420 encode_str (enc, "null", 4, 0); 496 encode_str (enc, "null", 4, 0);
421 else 497 else
422 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 498 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
423 SvPV_nolen (sv), SvFLAGS (sv)); 499 SvPV_nolen (sv), SvFLAGS (sv));
424} 500}
425 501
426static SV * 502static SV *
427encode_json (SV *scalar, UV flags) 503encode_json (SV *scalar, U32 flags)
428{ 504{
505 enc_t enc;
506
429 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 507 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar))
430 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 508 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
431 509
432 enc_t enc;
433 enc.flags = flags; 510 enc.flags = flags;
434 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 511 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
435 enc.cur = SvPVX (enc.sv); 512 enc.cur = SvPVX (enc.sv);
436 enc.end = SvEND (enc.sv); 513 enc.end = SvEND (enc.sv);
437 enc.max_recurse = 0;
438 enc.indent = 0; 514 enc.indent = 0;
515 enc.maxdepth = DEC_DEPTH (flags);
439 516
440 SvPOK_only (enc.sv); 517 SvPOK_only (enc.sv);
441 encode_sv (&enc, scalar); 518 encode_sv (&enc, scalar);
442 519
443 if (!(flags & (F_ASCII | F_UTF8))) 520 if (!(flags & (F_ASCII | F_UTF8)))
444 SvUTF8_on (enc.sv); 521 SvUTF8_on (enc.sv);
445 522
446 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 523 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
524 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
447 525
448 if (enc.flags & F_SHRINK) 526 if (enc.flags & F_SHRINK)
449 shrink (enc.sv); 527 shrink (enc.sv);
450 528
451 return enc.sv; 529 return enc.sv;
452} 530}
453 531
454///////////////////////////////////////////////////////////////////////////// 532/////////////////////////////////////////////////////////////////////////////
533// decoder
455 534
456#define WS \ 535// structure used for decoding JSON
536typedef struct
537{
538 char *cur; // current parser pointer
539 char *end; // end of input string
540 const char *err; // parse error, if != 0
541 U32 flags; // F_*
542 U32 depth; // recursion depth
543 U32 maxdepth; // recursion depth limit
544} dec_t;
545
546static void
547decode_ws (dec_t *dec)
548{
457 for (;;) \ 549 for (;;)
458 { \ 550 {
459 char ch = *dec->cur; \ 551 char ch = *dec->cur;
552
460 if (ch > 0x20 \ 553 if (ch > 0x20
461 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) \ 554 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09))
462 break; \ 555 break;
556
463 ++dec->cur; \ 557 ++dec->cur;
464 } 558 }
559}
465 560
466#define ERR(reason) SB dec->err = reason; goto fail; SE 561#define ERR(reason) SB dec->err = reason; goto fail; SE
562
467#define EXPECT_CH(ch) SB \ 563#define EXPECT_CH(ch) SB \
468 if (*dec->cur != ch) \ 564 if (*dec->cur != ch) \
469 ERR (# ch " expected"); \ 565 ERR (# ch " expected"); \
470 ++dec->cur; \ 566 ++dec->cur; \
471 SE 567 SE
472 568
569#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)")
570#define DEC_DEC_DEPTH --dec->depth
571
473static SV *decode_sv (dec_t *dec); 572static SV *decode_sv (dec_t *dec);
474 573
475static signed char decode_hexdigit[256]; 574static signed char decode_hexdigit[256];
476 575
477static UV 576static UV
478decode_4hex (dec_t *dec) 577decode_4hex (dec_t *dec)
479{ 578{
480 signed char d1, d2, d3, d4; 579 signed char d1, d2, d3, d4;
580 unsigned char *cur = (unsigned char *)dec->cur;
481 581
482 d1 = decode_hexdigit [((unsigned char *)dec->cur) [0]];
483 if (d1 < 0) ERR ("four hexadecimal digits expected"); 582 d1 = decode_hexdigit [cur [0]]; if (d1 < 0) ERR ("four hexadecimal digits expected");
484 d2 = decode_hexdigit [((unsigned char *)dec->cur) [1]];
485 if (d2 < 0) ERR ("four hexadecimal digits expected"); 583 d2 = decode_hexdigit [cur [1]]; if (d2 < 0) ERR ("four hexadecimal digits expected");
486 d3 = decode_hexdigit [((unsigned char *)dec->cur) [2]];
487 if (d3 < 0) ERR ("four hexadecimal digits expected"); 584 d3 = decode_hexdigit [cur [2]]; if (d3 < 0) ERR ("four hexadecimal digits expected");
488 d4 = decode_hexdigit [((unsigned char *)dec->cur) [3]];
489 if (d4 < 0) ERR ("four hexadecimal digits expected"); 585 d4 = decode_hexdigit [cur [3]]; if (d4 < 0) ERR ("four hexadecimal digits expected");
490 586
491 dec->cur += 4; 587 dec->cur += 4;
492 588
493 return ((UV)d1) << 12 589 return ((UV)d1) << 12
494 | ((UV)d2) << 8 590 | ((UV)d2) << 8
… …
497 593
498fail: 594fail:
499 return (UV)-1; 595 return (UV)-1;
500} 596}
501 597
502#define APPEND_GROW(n) SB \
503 if (cur + (n) >= end) \
504 { \
505 STRLEN ofs = cur - SvPVX (sv); \
506 SvGROW (sv, ofs + (n) + 1); \
507 cur = SvPVX (sv) + ofs; \
508 end = SvEND (sv); \
509 } \
510 SE
511
512#define APPEND_CH(ch) SB \
513 APPEND_GROW (1); \
514 *cur++ = (ch); \
515 SE
516
517static SV * 598static SV *
518decode_str (dec_t *dec) 599decode_str (dec_t *dec)
519{ 600{
520 SV *sv = NEWSV (0,2); 601 SV *sv = 0;
521 int utf8 = 0; 602 int utf8 = 0;
522 char *cur = SvPVX (sv);
523 char *end = SvEND (sv);
524 603
525 for (;;) 604 do
526 { 605 {
527 unsigned char ch = *(unsigned char *)dec->cur; 606 char buf [SHORT_STRING_LEN + UTF8_MAX_LEN];
607 char *cur = buf;
528 608
529 if (ch == '"') 609 do
530 break;
531 else if (ch == '\\')
532 { 610 {
533 switch (*++dec->cur) 611 unsigned char ch = *(unsigned char *)dec->cur++;
612
613 if (ch == '"')
534 { 614 {
535 case '\\': 615 --dec->cur;
536 case '/': 616 break;
537 case '"': APPEND_CH (*dec->cur++); break; 617 }
538 618 else if (ch == '\\')
539 case 'b': APPEND_CH ('\010'); ++dec->cur; break; 619 {
540 case 't': APPEND_CH ('\011'); ++dec->cur; break; 620 switch (*dec->cur)
541 case 'n': APPEND_CH ('\012'); ++dec->cur; break;
542 case 'f': APPEND_CH ('\014'); ++dec->cur; break;
543 case 'r': APPEND_CH ('\015'); ++dec->cur; break;
544
545 case 'u':
546 { 621 {
547 UV lo, hi; 622 case '\\':
548 ++dec->cur; 623 case '/':
624 case '"': *cur++ = *dec->cur++; break;
549 625
550 hi = decode_4hex (dec); 626 case 'b': ++dec->cur; *cur++ = '\010'; break;
551 if (hi == (UV)-1) 627 case 't': ++dec->cur; *cur++ = '\011'; break;
552 goto fail; 628 case 'n': ++dec->cur; *cur++ = '\012'; break;
629 case 'f': ++dec->cur; *cur++ = '\014'; break;
630 case 'r': ++dec->cur; *cur++ = '\015'; break;
553 631
554 // possibly a surrogate pair 632 case 'u':
555 if (hi >= 0xd800 && hi < 0xdc00)
556 { 633 {
557 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 634 UV lo, hi;
558 ERR ("missing low surrogate character in surrogate pair");
559
560 dec->cur += 2; 635 ++dec->cur;
561 636
562 lo = decode_4hex (dec); 637 hi = decode_4hex (dec);
563 if (lo == (UV)-1) 638 if (hi == (UV)-1)
564 goto fail; 639 goto fail;
565 640
641 // possibly a surrogate pair
642 if (hi >= 0xd800)
643 if (hi < 0xdc00)
644 {
645 if (dec->cur [0] != '\\' || dec->cur [1] != 'u')
646 ERR ("missing low surrogate character in surrogate pair");
647
648 dec->cur += 2;
649
650 lo = decode_4hex (dec);
651 if (lo == (UV)-1)
652 goto fail;
653
566 if (lo < 0xdc00 || lo >= 0xe000) 654 if (lo < 0xdc00 || lo >= 0xe000)
567 ERR ("surrogate pair expected"); 655 ERR ("surrogate pair expected");
568 656
569 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000; 657 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000;
658 }
659 else if (hi < 0xe000)
660 ERR ("missing high surrogate character in surrogate pair");
661
662 if (hi >= 0x80)
663 {
664 utf8 = 1;
665
666 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
667 }
668 else
669 *cur++ = hi;
570 } 670 }
571 else if (hi >= 0xdc00 && hi < 0xe000)
572 ERR ("missing high surrogate character in surrogate pair");
573
574 if (hi >= 0x80)
575 { 671 break;
576 utf8 = 1;
577 672
578 APPEND_GROW (4); // at most 4 bytes for 21 bits
579 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
580 }
581 else 673 default:
582 APPEND_CH (hi); 674 --dec->cur;
675 ERR ("illegal backslash escape sequence in string");
583 } 676 }
584 break; 677 }
678 else if (ch >= 0x20 && ch <= 0x7f)
679 *cur++ = ch;
680 else if (ch >= 0x80)
681 {
682 STRLEN clen;
683 UV uch;
585 684
586 default:
587 --dec->cur; 685 --dec->cur;
588 ERR ("illegal backslash escape sequence in string"); 686
687 uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen);
688 if (clen == (STRLEN)-1)
689 ERR ("malformed UTF-8 character in JSON string");
690
691 do
692 *cur++ = *dec->cur++;
693 while (--clen);
694
695 utf8 = 1;
696 }
697 else
698 {
699 --dec->cur;
700
701 if (!ch)
702 ERR ("unexpected end of string while parsing JSON string");
703 else
704 ERR ("invalid character encountered while parsing JSON string");
589 } 705 }
590 } 706 }
591 else if (ch >= 0x20 && ch <= 0x7f) 707 while (cur < buf + SHORT_STRING_LEN);
592 APPEND_CH (*dec->cur++); 708
593 else if (ch >= 0x80)
594 { 709 {
595 STRLEN clen; 710 STRLEN len = cur - buf;
596 UV uch = utf8n_to_uvuni (dec->cur, dec->end - dec->cur, &clen, UTF8_CHECK_ONLY);
597 if (clen == (STRLEN)-1)
598 ERR ("malformed UTF-8 character in JSON string");
599 711
600 APPEND_GROW (clen); 712 if (sv)
601 do
602 { 713 {
603 *cur++ = *dec->cur++; 714 SvGROW (sv, SvCUR (sv) + len + 1);
715 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
716 SvCUR_set (sv, SvCUR (sv) + len);
604 } 717 }
605 while (--clen);
606
607 utf8 = 1;
608 }
609 else if (dec->cur == dec->end)
610 ERR ("unexpected end of string while parsing json string");
611 else 718 else
612 ERR ("invalid character encountered"); 719 sv = newSVpvn (buf, len);
613 } 720 }
721 }
722 while (*dec->cur != '"');
614 723
615 ++dec->cur; 724 ++dec->cur;
616 725
617 SvCUR_set (sv, cur - SvPVX (sv)); 726 if (sv)
618 727 {
619 SvPOK_only (sv); 728 SvPOK_only (sv);
620 *SvEND (sv) = 0; 729 *SvEND (sv) = 0;
621 730
622 if (utf8) 731 if (utf8)
623 SvUTF8_on (sv); 732 SvUTF8_on (sv);
624 733 }
625 if (dec->flags & F_SHRINK) 734 else
626 shrink (sv); 735 sv = newSVpvn ("", 0);
627 736
628 return sv; 737 return sv;
629 738
630fail: 739fail:
631 SvREFCNT_dec (sv);
632 return 0; 740 return 0;
633} 741}
634 742
635static SV * 743static SV *
636decode_num (dec_t *dec) 744decode_num (dec_t *dec)
… …
717static SV * 825static SV *
718decode_av (dec_t *dec) 826decode_av (dec_t *dec)
719{ 827{
720 AV *av = newAV (); 828 AV *av = newAV ();
721 829
722 WS; 830 DEC_INC_DEPTH;
831 decode_ws (dec);
832
723 if (*dec->cur == ']') 833 if (*dec->cur == ']')
724 ++dec->cur; 834 ++dec->cur;
725 else 835 else
726 for (;;) 836 for (;;)
727 { 837 {
… …
731 if (!value) 841 if (!value)
732 goto fail; 842 goto fail;
733 843
734 av_push (av, value); 844 av_push (av, value);
735 845
736 WS; 846 decode_ws (dec);
737 847
738 if (*dec->cur == ']') 848 if (*dec->cur == ']')
739 { 849 {
740 ++dec->cur; 850 ++dec->cur;
741 break; 851 break;
… …
745 ERR (", or ] expected while parsing array"); 855 ERR (", or ] expected while parsing array");
746 856
747 ++dec->cur; 857 ++dec->cur;
748 } 858 }
749 859
860 DEC_DEC_DEPTH;
750 return newRV_noinc ((SV *)av); 861 return newRV_noinc ((SV *)av);
751 862
752fail: 863fail:
753 SvREFCNT_dec (av); 864 SvREFCNT_dec (av);
865 DEC_DEC_DEPTH;
754 return 0; 866 return 0;
755} 867}
756 868
757static SV * 869static SV *
758decode_hv (dec_t *dec) 870decode_hv (dec_t *dec)
759{ 871{
760 HV *hv = newHV (); 872 HV *hv = newHV ();
761 873
762 WS; 874 DEC_INC_DEPTH;
875 decode_ws (dec);
876
763 if (*dec->cur == '}') 877 if (*dec->cur == '}')
764 ++dec->cur; 878 ++dec->cur;
765 else 879 else
766 for (;;) 880 for (;;)
767 { 881 {
768 SV *key, *value; 882 SV *key, *value;
769 883
770 WS; EXPECT_CH ('"'); 884 decode_ws (dec); EXPECT_CH ('"');
771 885
772 key = decode_str (dec); 886 key = decode_str (dec);
773 if (!key) 887 if (!key)
774 goto fail; 888 goto fail;
775 889
776 WS; EXPECT_CH (':'); 890 decode_ws (dec); EXPECT_CH (':');
777 891
778 value = decode_sv (dec); 892 value = decode_sv (dec);
779 if (!value) 893 if (!value)
780 { 894 {
781 SvREFCNT_dec (key); 895 SvREFCNT_dec (key);
782 goto fail; 896 goto fail;
783 } 897 }
784 898
785 //TODO: optimise
786 hv_store_ent (hv, key, value, 0); 899 hv_store_ent (hv, key, value, 0);
900 SvREFCNT_dec (key);
787 901
788 WS; 902 decode_ws (dec);
789 903
790 if (*dec->cur == '}') 904 if (*dec->cur == '}')
791 { 905 {
792 ++dec->cur; 906 ++dec->cur;
793 break; 907 break;
… …
797 ERR (", or } expected while parsing object/hash"); 911 ERR (", or } expected while parsing object/hash");
798 912
799 ++dec->cur; 913 ++dec->cur;
800 } 914 }
801 915
916 DEC_DEC_DEPTH;
802 return newRV_noinc ((SV *)hv); 917 return newRV_noinc ((SV *)hv);
803 918
804fail: 919fail:
805 SvREFCNT_dec (hv); 920 SvREFCNT_dec (hv);
921 DEC_DEC_DEPTH;
806 return 0; 922 return 0;
807} 923}
808 924
809static SV * 925static SV *
810decode_sv (dec_t *dec) 926decode_sv (dec_t *dec)
811{ 927{
812 WS; 928 decode_ws (dec);
813 switch (*dec->cur) 929 switch (*dec->cur)
814 { 930 {
815 case '"': ++dec->cur; return decode_str (dec); 931 case '"': ++dec->cur; return decode_str (dec);
816 case '[': ++dec->cur; return decode_av (dec); 932 case '[': ++dec->cur; return decode_av (dec);
817 case '{': ++dec->cur; return decode_hv (dec); 933 case '{': ++dec->cur; return decode_hv (dec);
… …
853 ERR ("'null' expected"); 969 ERR ("'null' expected");
854 970
855 break; 971 break;
856 972
857 default: 973 default:
858 ERR ("malformed json string, neither array, object, number, string or atom"); 974 ERR ("malformed JSON string, neither array, object, number, string or atom");
859 break; 975 break;
860 } 976 }
861 977
862fail: 978fail:
863 return 0; 979 return 0;
864} 980}
865 981
866static SV * 982static SV *
867decode_json (SV *string, UV flags) 983decode_json (SV *string, U32 flags)
868{ 984{
985 dec_t dec;
869 SV *sv; 986 SV *sv;
987
988 SvUPGRADE (string, SVt_PV);
870 989
871 if (flags & F_UTF8) 990 if (flags & F_UTF8)
872 sv_utf8_downgrade (string, 0); 991 sv_utf8_downgrade (string, 0);
873 else 992 else
874 sv_utf8_upgrade (string); 993 sv_utf8_upgrade (string);
875 994
876 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 995 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
877 996
878 dec_t dec;
879 dec.flags = flags; 997 dec.flags = flags;
880 dec.cur = SvPVX (string); 998 dec.cur = SvPVX (string);
881 dec.end = SvEND (string); 999 dec.end = SvEND (string);
882 dec.err = 0; 1000 dec.err = 0;
1001 dec.depth = 0;
1002 dec.maxdepth = DEC_DEPTH (dec.flags);
883 1003
1004 *dec.end = 0; // this should basically be a nop, too, but make sure its there
884 sv = decode_sv (&dec); 1005 sv = decode_sv (&dec);
885 1006
886 if (!sv) 1007 if (!sv)
887 { 1008 {
888 IV offset = dec.flags & F_UTF8 1009 IV offset = dec.flags & F_UTF8
… …
897 SAVEVPTR (PL_curcop); 1018 SAVEVPTR (PL_curcop);
898 PL_curcop = &cop; 1019 PL_curcop = &cop;
899 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1020 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
900 LEAVE; 1021 LEAVE;
901 1022
902 croak ("%s, at character offset %d (%s)", 1023 croak ("%s, at character offset %d [\"%s\"]",
903 dec.err, 1024 dec.err,
904 (int)offset, 1025 (int)offset,
905 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1026 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
906 } 1027 }
907 1028
… …
911 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); 1032 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
912 1033
913 return sv; 1034 return sv;
914} 1035}
915 1036
1037/////////////////////////////////////////////////////////////////////////////
1038// XS interface functions
1039
916MODULE = JSON::XS PACKAGE = JSON::XS 1040MODULE = JSON::XS PACKAGE = JSON::XS
917 1041
918BOOT: 1042BOOT:
919{ 1043{
920 int i; 1044 int i;
921 1045
922 memset (decode_hexdigit, 0xff, 256); 1046 memset (decode_hexdigit, 0xff, 256);
1047
923 for (i = 10; i--; ) 1048 for (i = 0; i < 256; ++i)
924 decode_hexdigit ['0' + i] = i; 1049 decode_hexdigit [i] =
925 1050 i >= '0' && i <= '9' ? i - '0'
926 for (i = 7; i--; ) 1051 : i >= 'a' && i <= 'f' ? i - 'a' + 10
927 { 1052 : i >= 'A' && i <= 'F' ? i - 'A' + 10
928 decode_hexdigit ['a' + i] = 10 + i; 1053 : -1;
929 decode_hexdigit ['A' + i] = 10 + i;
930 }
931 1054
932 json_stash = gv_stashpv ("JSON::XS", 1); 1055 json_stash = gv_stashpv ("JSON::XS", 1);
933} 1056}
934 1057
935PROTOTYPES: DISABLE 1058PROTOTYPES: DISABLE
… …
946 utf8 = F_UTF8 1069 utf8 = F_UTF8
947 indent = F_INDENT 1070 indent = F_INDENT
948 canonical = F_CANONICAL 1071 canonical = F_CANONICAL
949 space_before = F_SPACE_BEFORE 1072 space_before = F_SPACE_BEFORE
950 space_after = F_SPACE_AFTER 1073 space_after = F_SPACE_AFTER
951 json_rpc = F_JSON_RPC
952 pretty = F_PRETTY 1074 pretty = F_PRETTY
953 allow_nonref = F_ALLOW_NONREF 1075 allow_nonref = F_ALLOW_NONREF
954 shrink = F_SHRINK 1076 shrink = F_SHRINK
955 CODE: 1077 CODE:
956{ 1078{
… …
963 RETVAL = newSVsv (self); 1085 RETVAL = newSVsv (self);
964} 1086}
965 OUTPUT: 1087 OUTPUT:
966 RETVAL 1088 RETVAL
967 1089
1090SV *max_depth (SV *self, UV max_depth = 0x80000000UL)
1091 CODE:
1092{
1093 UV *uv = SvJSON (self);
1094 UV log2 = 0;
1095
1096 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1097
1098 while ((1UL << log2) < max_depth)
1099 ++log2;
1100
1101 *uv = *uv & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1102
1103 RETVAL = newSVsv (self);
1104}
1105 OUTPUT:
1106 RETVAL
1107
968void encode (SV *self, SV *scalar) 1108void encode (SV *self, SV *scalar)
969 PPCODE: 1109 PPCODE:
970 XPUSHs (encode_json (scalar, *SvJSON (self))); 1110 XPUSHs (encode_json (scalar, *SvJSON (self)));
971 1111
972void decode (SV *self, SV *jsonstr) 1112void decode (SV *self, SV *jsonstr)
… …
974 XPUSHs (decode_json (jsonstr, *SvJSON (self))); 1114 XPUSHs (decode_json (jsonstr, *SvJSON (self)));
975 1115
976PROTOTYPES: ENABLE 1116PROTOTYPES: ENABLE
977 1117
978void to_json (SV *scalar) 1118void to_json (SV *scalar)
1119 ALIAS:
1120 objToJson = 0
979 PPCODE: 1121 PPCODE:
980 XPUSHs (encode_json (scalar, F_UTF8)); 1122 XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8));
981 1123
982void from_json (SV *jsonstr) 1124void from_json (SV *jsonstr)
1125 ALIAS:
1126 jsonToObj = 0
983 PPCODE: 1127 PPCODE:
984 XPUSHs (decode_json (jsonstr, F_UTF8)); 1128 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8));
985 1129

Diff Legend

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