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 1.52 by root, Mon Jul 2 02:57:11 2007 UTC vs.
Revision 1.90 by root, Sun Jul 20 17:55:19 2008 UTC

1#include "EXTERN.h" 1#include "EXTERN.h"
2#include "perl.h" 2#include "perl.h"
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#include <stdio.h>
9#include <limits.h>
10#include <float.h>
9 11
10#if defined(__BORLANDC__) || defined(_MSC_VER) 12#if defined(__BORLANDC__) || defined(_MSC_VER)
11# define snprintf _snprintf // C compilers have this in stdio.h 13# define snprintf _snprintf // C compilers have this in stdio.h
12#endif 14#endif
13 15
14// some old perls do not have this, try to make it work, no 16// some old perls do not have this, try to make it work, no
15// guarentees, though. if it breaks, you get to keep the pieces. 17// guarentees, though. if it breaks, you get to keep the pieces.
16#ifndef UTF8_MAXBYTES 18#ifndef UTF8_MAXBYTES
17# define UTF8_MAXBYTES 13 19# define UTF8_MAXBYTES 13
18#endif 20#endif
21
22#define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 2)
19 23
20#define F_ASCII 0x00000001UL 24#define F_ASCII 0x00000001UL
21#define F_LATIN1 0x00000002UL 25#define F_LATIN1 0x00000002UL
22#define F_UTF8 0x00000004UL 26#define F_UTF8 0x00000004UL
23#define F_INDENT 0x00000008UL 27#define F_INDENT 0x00000008UL
… …
26#define F_SPACE_AFTER 0x00000040UL 30#define F_SPACE_AFTER 0x00000040UL
27#define F_ALLOW_NONREF 0x00000100UL 31#define F_ALLOW_NONREF 0x00000100UL
28#define F_SHRINK 0x00000200UL 32#define F_SHRINK 0x00000200UL
29#define F_ALLOW_BLESSED 0x00000400UL 33#define F_ALLOW_BLESSED 0x00000400UL
30#define F_CONV_BLESSED 0x00000800UL 34#define F_CONV_BLESSED 0x00000800UL
31#define F_MAXDEPTH 0xf8000000UL
32#define S_MAXDEPTH 27
33#define F_MAXSIZE 0x01f00000UL 35#define F_RELAXED 0x00001000UL
34#define S_MAXSIZE 20 36#define F_ALLOW_UNKNOWN 0x00002000UL
35#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing 37#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
36 38
37#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
38#define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE ))
39
40#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 39#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
41#define F_DEFAULT (9UL << S_MAXDEPTH)
42 40
43#define INIT_SIZE 32 // initial scalar size to be allocated 41#define INIT_SIZE 32 // initial scalar size to be allocated
44#define INDENT_STEP 3 // spaces per indentation level 42#define INDENT_STEP 3 // spaces per indentation level
45 43
46#define SHORT_STRING_LEN 16384 // special-case strings of up to this size 44#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
47 45
48#define SB do { 46#define SB do {
49#define SE } while (0) 47#define SE } while (0)
50 48
51#if __GNUC__ >= 3 49#if __GNUC__ >= 3
52# define expect(expr,value) __builtin_expect ((expr),(value)) 50# define expect(expr,value) __builtin_expect ((expr), (value))
53# define inline inline 51# define INLINE static inline
54#else 52#else
55# define expect(expr,value) (expr) 53# define expect(expr,value) (expr)
56# define inline static 54# define INLINE static
57#endif 55#endif
58 56
59#define expect_false(expr) expect ((expr) != 0, 0) 57#define expect_false(expr) expect ((expr) != 0, 0)
60#define expect_true(expr) expect ((expr) != 0, 1) 58#define expect_true(expr) expect ((expr) != 0, 1)
61 59
60#define IN_RANGE_INC(type,val,beg,end) \
61 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \
62 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
63
64#define ERR_NESTING_EXCEEDED "json text or perl structure exceeds maximum nesting level (max_depth set too low?)"
65
66#ifdef USE_ITHREADS
67# define JSON_SLOW 1
68# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1))
69#else
70# define JSON_SLOW 0
71# define JSON_STASH json_stash
72#endif
73
62static HV *json_stash, *json_boolean_stash; // JSON::XS:: 74static HV *json_stash, *json_boolean_stash; // JSON::XS::
63static SV *json_true, *json_false; 75static SV *json_true, *json_false;
64 76
77enum {
78 INCR_M_WS = 0, // initial whitespace skipping, must be 0
79 INCR_M_STR, // inside string
80 INCR_M_BS, // inside backslash
81 INCR_M_JSON // outside anything, count nesting
82};
83
84#define INCR_DONE(json) ((json)->incr_nest <= 0 && (json)->incr_mode == INCR_M_JSON)
85
65typedef struct { 86typedef struct {
66 U32 flags; 87 U32 flags;
88 U32 max_depth;
89 STRLEN max_size;
90
67 SV *cb_object; 91 SV *cb_object;
68 HV *cb_sk_object; 92 HV *cb_sk_object;
93
94 // for the incremental parser
95 SV *incr_text; // the source text so far
96 STRLEN incr_pos; // the current offset into the text
97 int incr_nest; // {[]}-nesting level
98 unsigned char incr_mode;
69} JSON; 99} JSON;
100
101INLINE void
102json_init (JSON *json)
103{
104 Zero (json, 1, JSON);
105 json->max_depth = 512;
106}
70 107
71///////////////////////////////////////////////////////////////////////////// 108/////////////////////////////////////////////////////////////////////////////
72// utility functions 109// utility functions
73 110
74inline void 111INLINE SV *
112get_bool (const char *name)
113{
114 SV *sv = get_sv (name, 1);
115
116 SvREADONLY_on (sv);
117 SvREADONLY_on (SvRV (sv));
118
119 return sv;
120}
121
122INLINE void
75shrink (SV *sv) 123shrink (SV *sv)
76{ 124{
77 sv_utf8_downgrade (sv, 1); 125 sv_utf8_downgrade (sv, 1);
126
78 if (SvLEN (sv) > SvCUR (sv) + 1) 127 if (SvLEN (sv) > SvCUR (sv) + 1)
79 { 128 {
80#ifdef SvPV_shrink_to_cur 129#ifdef SvPV_shrink_to_cur
81 SvPV_shrink_to_cur (sv); 130 SvPV_shrink_to_cur (sv);
82#elif defined (SvPV_renew) 131#elif defined (SvPV_renew)
… …
88// decode an utf-8 character and return it, or (UV)-1 in 137// decode an utf-8 character and return it, or (UV)-1 in
89// case of an error. 138// case of an error.
90// we special-case "safe" characters from U+80 .. U+7FF, 139// we special-case "safe" characters from U+80 .. U+7FF,
91// but use the very good perl function to parse anything else. 140// but use the very good perl function to parse anything else.
92// note that we never call this function for a ascii codepoints 141// note that we never call this function for a ascii codepoints
93inline UV 142INLINE UV
94decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) 143decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
95{ 144{
96 if (expect_false (s[0] > 0xdf || s[0] < 0xc2)) 145 if (expect_true (len >= 2
97 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); 146 && IN_RANGE_INC (char, s[0], 0xc2, 0xdf)
98 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) 147 && IN_RANGE_INC (char, s[1], 0x80, 0xbf)))
99 { 148 {
100 *clen = 2; 149 *clen = 2;
101 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 150 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
102 } 151 }
103 else 152 else
104 { 153 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
105 *clen = (STRLEN)-1; 154}
106 return (UV)-1; 155
107 } 156// likewise for encoding, also never called for ascii codepoints
157// this function takes advantage of this fact, although current gccs
158// seem to optimise the check for >= 0x80 away anyways
159INLINE unsigned char *
160encode_utf8 (unsigned char *s, UV ch)
161{
162 if (expect_false (ch < 0x000080))
163 *s++ = ch;
164 else if (expect_true (ch < 0x000800))
165 *s++ = 0xc0 | ( ch >> 6),
166 *s++ = 0x80 | ( ch & 0x3f);
167 else if ( ch < 0x010000)
168 *s++ = 0xe0 | ( ch >> 12),
169 *s++ = 0x80 | ((ch >> 6) & 0x3f),
170 *s++ = 0x80 | ( ch & 0x3f);
171 else if ( ch < 0x110000)
172 *s++ = 0xf0 | ( ch >> 18),
173 *s++ = 0x80 | ((ch >> 12) & 0x3f),
174 *s++ = 0x80 | ((ch >> 6) & 0x3f),
175 *s++ = 0x80 | ( ch & 0x3f);
176
177 return s;
108} 178}
109 179
110///////////////////////////////////////////////////////////////////////////// 180/////////////////////////////////////////////////////////////////////////////
111// encoder 181// encoder
112 182
… …
116 char *cur; // SvPVX (sv) + current output position 186 char *cur; // SvPVX (sv) + current output position
117 char *end; // SvEND (sv) 187 char *end; // SvEND (sv)
118 SV *sv; // result scalar 188 SV *sv; // result scalar
119 JSON json; 189 JSON json;
120 U32 indent; // indentation level 190 U32 indent; // indentation level
121 U32 maxdepth; // max. indentation/recursion level 191 UV limit; // escape character values >= this value when encoding
122} enc_t; 192} enc_t;
123 193
124inline void 194INLINE void
125need (enc_t *enc, STRLEN len) 195need (enc_t *enc, STRLEN len)
126{ 196{
127 if (expect_false (enc->cur + len >= enc->end)) 197 if (expect_false (enc->cur + len >= enc->end))
128 { 198 {
129 STRLEN cur = enc->cur - SvPVX (enc->sv); 199 STRLEN cur = enc->cur - SvPVX (enc->sv);
… …
131 enc->cur = SvPVX (enc->sv) + cur; 201 enc->cur = SvPVX (enc->sv) + cur;
132 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 202 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
133 } 203 }
134} 204}
135 205
136inline void 206INLINE void
137encode_ch (enc_t *enc, char ch) 207encode_ch (enc_t *enc, char ch)
138{ 208{
139 need (enc, 1); 209 need (enc, 1);
140 *enc->cur++ = ch; 210 *enc->cur++ = ch;
141} 211}
… …
195 { 265 {
196 uch = ch; 266 uch = ch;
197 clen = 1; 267 clen = 1;
198 } 268 }
199 269
200 if (uch > 0x10FFFFUL) 270 if (uch < 0x80/*0x20*/ || uch >= enc->limit)
201 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
202
203 if (uch < 0x80 || enc->json.flags & F_ASCII || (enc->json.flags & F_LATIN1 && uch > 0xFF))
204 { 271 {
205 if (uch > 0xFFFFUL) 272 if (uch >= 0x10000UL)
206 { 273 {
274 if (uch >= 0x110000UL)
275 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
276
207 need (enc, len += 11); 277 need (enc, len += 11);
208 sprintf (enc->cur, "\\u%04x\\u%04x", 278 sprintf (enc->cur, "\\u%04x\\u%04x",
209 (int)((uch - 0x10000) / 0x400 + 0xD800), 279 (int)((uch - 0x10000) / 0x400 + 0xD800),
210 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 280 (int)((uch - 0x10000) % 0x400 + 0xDC00));
211 enc->cur += 12; 281 enc->cur += 12;
… …
239 while (--clen); 309 while (--clen);
240 } 310 }
241 else 311 else
242 { 312 {
243 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed 313 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
244 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 314 enc->cur = encode_utf8 (enc->cur, uch);
245 ++str; 315 ++str;
246 } 316 }
247 } 317 }
248 } 318 }
249 } 319 }
250 320
251 --len; 321 --len;
252 } 322 }
253} 323}
254 324
255inline void 325INLINE void
256encode_indent (enc_t *enc) 326encode_indent (enc_t *enc)
257{ 327{
258 if (enc->json.flags & F_INDENT) 328 if (enc->json.flags & F_INDENT)
259 { 329 {
260 int spaces = enc->indent * INDENT_STEP; 330 int spaces = enc->indent * INDENT_STEP;
… …
263 memset (enc->cur, ' ', spaces); 333 memset (enc->cur, ' ', spaces);
264 enc->cur += spaces; 334 enc->cur += spaces;
265 } 335 }
266} 336}
267 337
268inline void 338INLINE void
269encode_space (enc_t *enc) 339encode_space (enc_t *enc)
270{ 340{
271 need (enc, 1); 341 need (enc, 1);
272 encode_ch (enc, ' '); 342 encode_ch (enc, ' ');
273} 343}
274 344
275inline void 345INLINE void
276encode_nl (enc_t *enc) 346encode_nl (enc_t *enc)
277{ 347{
278 if (enc->json.flags & F_INDENT) 348 if (enc->json.flags & F_INDENT)
279 { 349 {
280 need (enc, 1); 350 need (enc, 1);
281 encode_ch (enc, '\n'); 351 encode_ch (enc, '\n');
282 } 352 }
283} 353}
284 354
285inline void 355INLINE void
286encode_comma (enc_t *enc) 356encode_comma (enc_t *enc)
287{ 357{
288 encode_ch (enc, ','); 358 encode_ch (enc, ',');
289 359
290 if (enc->json.flags & F_INDENT) 360 if (enc->json.flags & F_INDENT)
… …
298static void 368static void
299encode_av (enc_t *enc, AV *av) 369encode_av (enc_t *enc, AV *av)
300{ 370{
301 int i, len = av_len (av); 371 int i, len = av_len (av);
302 372
303 if (enc->indent >= enc->maxdepth) 373 if (enc->indent >= enc->json.max_depth)
304 croak ("data structure too deep (hit recursion limit)"); 374 croak (ERR_NESTING_EXCEEDED);
305 375
306 encode_ch (enc, '['); encode_nl (enc); 376 encode_ch (enc, '[');
307 ++enc->indent; 377
378 if (len >= 0)
379 {
380 encode_nl (enc); ++enc->indent;
308 381
309 for (i = 0; i <= len; ++i) 382 for (i = 0; i <= len; ++i)
310 { 383 {
384 SV **svp = av_fetch (av, i, 0);
385
311 encode_indent (enc); 386 encode_indent (enc);
312 encode_sv (enc, *av_fetch (av, i, 0));
313 387
388 if (svp)
389 encode_sv (enc, *svp);
390 else
391 encode_str (enc, "null", 4, 0);
392
314 if (i < len) 393 if (i < len)
315 encode_comma (enc); 394 encode_comma (enc);
316 } 395 }
317 396
397 encode_nl (enc); --enc->indent; encode_indent (enc);
398 }
399
318 encode_nl (enc); 400 encode_ch (enc, ']');
319
320 --enc->indent;
321 encode_indent (enc); encode_ch (enc, ']');
322} 401}
323 402
324static void 403static void
325encode_he (enc_t *enc, HE *he) 404encode_hk (enc_t *enc, HE *he)
326{ 405{
327 encode_ch (enc, '"'); 406 encode_ch (enc, '"');
328 407
329 if (HeKLEN (he) == HEf_SVKEY) 408 if (HeKLEN (he) == HEf_SVKEY)
330 { 409 {
… …
343 encode_ch (enc, '"'); 422 encode_ch (enc, '"');
344 423
345 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc); 424 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
346 encode_ch (enc, ':'); 425 encode_ch (enc, ':');
347 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc); 426 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
348 encode_sv (enc, HeVAL (he));
349} 427}
350 428
351// compare hash entries, used when all keys are bytestrings 429// compare hash entries, used when all keys are bytestrings
352static int 430static int
353he_cmp_fast (const void *a_, const void *b_) 431he_cmp_fast (const void *a_, const void *b_)
… …
358 HE *b = *(HE **)b_; 436 HE *b = *(HE **)b_;
359 437
360 STRLEN la = HeKLEN (a); 438 STRLEN la = HeKLEN (a);
361 STRLEN lb = HeKLEN (b); 439 STRLEN lb = HeKLEN (b);
362 440
363 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 441 if (!(cmp = memcmp (HeKEY (b), HeKEY (a), lb < la ? lb : la)))
364 cmp = la - lb; 442 cmp = lb - la;
365 443
366 return cmp; 444 return cmp;
367} 445}
368 446
369// compare hash entries, used when some keys are sv's or utf-x 447// compare hash entries, used when some keys are sv's or utf-x
370static int 448static int
371he_cmp_slow (const void *a, const void *b) 449he_cmp_slow (const void *a, const void *b)
372{ 450{
373 return sv_cmp (HeSVKEY_force (*(HE **)a), HeSVKEY_force (*(HE **)b)); 451 return sv_cmp (HeSVKEY_force (*(HE **)b), HeSVKEY_force (*(HE **)a));
374} 452}
375 453
376static void 454static void
377encode_hv (enc_t *enc, HV *hv) 455encode_hv (enc_t *enc, HV *hv)
378{ 456{
379 int count, i; 457 HE *he;
380 458
381 if (enc->indent >= enc->maxdepth) 459 if (enc->indent >= enc->json.max_depth)
382 croak ("data structure too deep (hit recursion limit)"); 460 croak (ERR_NESTING_EXCEEDED);
383 461
384 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 462 encode_ch (enc, '{');
385 463
386 if ((count = hv_iterinit (hv)))
387 {
388 // for canonical output we have to sort by keys first 464 // for canonical output we have to sort by keys first
389 // actually, this is mostly due to the stupid so-called 465 // actually, this is mostly due to the stupid so-called
390 // security workaround added somewhere in 5.8.x. 466 // security workaround added somewhere in 5.8.x.
391 // that randomises hash orderings 467 // that randomises hash orderings
392 if (enc->json.flags & F_CANONICAL) 468 if (enc->json.flags & F_CANONICAL)
469 {
470 int count = hv_iterinit (hv);
471
472 if (SvMAGICAL (hv))
393 { 473 {
474 // need to count by iterating. could improve by dynamically building the vector below
475 // but I don't care for the speed of this special case.
476 // note also that we will run into undefined behaviour when the two iterations
477 // do not result in the same count, something I might care for in some later release.
478
479 count = 0;
480 while (hv_iternext (hv))
481 ++count;
482
483 hv_iterinit (hv);
484 }
485
486 if (count)
487 {
394 int fast = 1; 488 int i, fast = 1;
395 HE *he;
396#if defined(__BORLANDC__) || defined(_MSC_VER) 489#if defined(__BORLANDC__) || defined(_MSC_VER)
397 HE **hes = _alloca (count * sizeof (HE)); 490 HE **hes = _alloca (count * sizeof (HE));
398#else 491#else
399 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode 492 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
400#endif 493#endif
… …
427 520
428 FREETMPS; 521 FREETMPS;
429 LEAVE; 522 LEAVE;
430 } 523 }
431 524
432 for (i = 0; i < count; ++i) 525 encode_nl (enc); ++enc->indent;
526
527 while (count--)
433 { 528 {
434 encode_indent (enc); 529 encode_indent (enc);
530 he = hes [count];
435 encode_he (enc, hes [i]); 531 encode_hk (enc, he);
532 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
436 533
437 if (i < count - 1) 534 if (count)
438 encode_comma (enc); 535 encode_comma (enc);
439 } 536 }
440 537
441 encode_nl (enc); 538 encode_nl (enc); --enc->indent; encode_indent (enc);
442 } 539 }
540 }
443 else 541 else
542 {
543 if (hv_iterinit (hv) || SvMAGICAL (hv))
544 if ((he = hv_iternext (hv)))
444 { 545 {
445 HE *he = hv_iternext (hv); 546 encode_nl (enc); ++enc->indent;
446 547
447 for (;;) 548 for (;;)
448 { 549 {
449 encode_indent (enc); 550 encode_indent (enc);
450 encode_he (enc, he); 551 encode_hk (enc, he);
552 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
451 553
452 if (!(he = hv_iternext (hv))) 554 if (!(he = hv_iternext (hv)))
453 break; 555 break;
454 556
455 encode_comma (enc); 557 encode_comma (enc);
456 } 558 }
457 559
458 encode_nl (enc); 560 encode_nl (enc); --enc->indent; encode_indent (enc);
459 } 561 }
460 } 562 }
461 563
462 --enc->indent; encode_indent (enc); encode_ch (enc, '}'); 564 encode_ch (enc, '}');
463} 565}
464 566
465// encode objects, arrays and special \0=false and \1=true values. 567// encode objects, arrays and special \0=false and \1=true values.
466static void 568static void
467encode_rv (enc_t *enc, SV *sv) 569encode_rv (enc_t *enc, SV *sv)
… …
471 SvGETMAGIC (sv); 573 SvGETMAGIC (sv);
472 svt = SvTYPE (sv); 574 svt = SvTYPE (sv);
473 575
474 if (expect_false (SvOBJECT (sv))) 576 if (expect_false (SvOBJECT (sv)))
475 { 577 {
578 HV *stash = !JSON_SLOW || json_boolean_stash
579 ? json_boolean_stash
580 : gv_stashpv ("JSON::XS::Boolean", 1);
581
476 if (SvSTASH (sv) == json_boolean_stash) 582 if (SvSTASH (sv) == stash)
477 { 583 {
478 if (SvIV (sv)) 584 if (SvIV (sv))
479 encode_str (enc, "true", 4, 0); 585 encode_str (enc, "true", 4, 0);
480 else 586 else
481 encode_str (enc, "false", 5, 0); 587 encode_str (enc, "false", 5, 0);
… …
489 } 595 }
490#endif 596#endif
491 if (enc->json.flags & F_CONV_BLESSED) 597 if (enc->json.flags & F_CONV_BLESSED)
492 { 598 {
493 // we re-bless the reference to get overload and other niceties right 599 // we re-bless the reference to get overload and other niceties right
494 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 1); 600 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
495 601
496 if (to_json) 602 if (to_json)
497 { 603 {
604 dSP;
605
498 dSP; ENTER; SAVETMPS; PUSHMARK (SP); 606 ENTER; SAVETMPS; PUSHMARK (SP);
499 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv))); 607 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
500 608
501 // calling with G_SCALAR ensures that we always get a 1 reutrn value 609 // calling with G_SCALAR ensures that we always get a 1 return value
502 // check anyways.
503 PUTBACK; 610 PUTBACK;
504 assert (1 == call_sv ((SV *)GvCV (to_json), G_SCALAR)); 611 call_sv ((SV *)GvCV (to_json), G_SCALAR);
505 SPAGAIN; 612 SPAGAIN;
506 613
614 // catch this surprisingly common error
615 if (SvROK (TOPs) && SvRV (TOPs) == sv)
616 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
617
618 sv = POPs;
619 PUTBACK;
620
507 encode_sv (enc, POPs); 621 encode_sv (enc, sv);
508 622
509 FREETMPS; LEAVE; 623 FREETMPS; LEAVE;
510 } 624 }
511 else if (enc->json.flags & F_ALLOW_BLESSED) 625 else if (enc->json.flags & F_ALLOW_BLESSED)
512 encode_str (enc, "null", 4, 0); 626 encode_str (enc, "null", 4, 0);
… …
532 646
533 if (len == 1 && *pv == '1') 647 if (len == 1 && *pv == '1')
534 encode_str (enc, "true", 4, 0); 648 encode_str (enc, "true", 4, 0);
535 else if (len == 1 && *pv == '0') 649 else if (len == 1 && *pv == '0')
536 encode_str (enc, "false", 5, 0); 650 encode_str (enc, "false", 5, 0);
651 else if (enc->json.flags & F_ALLOW_UNKNOWN)
652 encode_str (enc, "null", 4, 0);
537 else 653 else
538 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", 654 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
539 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 655 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
540 } 656 }
657 else if (enc->json.flags & F_ALLOW_UNKNOWN)
658 encode_str (enc, "null", 4, 0);
541 else 659 else
542 croak ("encountered %s, but JSON can only represent references to arrays or hashes", 660 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
543 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 661 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
544} 662}
545 663
… …
563 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 681 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
564 enc->cur += strlen (enc->cur); 682 enc->cur += strlen (enc->cur);
565 } 683 }
566 else if (SvIOKp (sv)) 684 else if (SvIOKp (sv))
567 { 685 {
568 // we assume we can always read an IV as a UV 686 // we assume we can always read an IV as a UV and vice versa
569 if (SvUV (sv) & ~(UV)0x7fff) 687 // we assume two's complement
570 { 688 // we assume no aliasing issues in the union
571 // large integer, use the (rather slow) snprintf way. 689 if (SvIsUV (sv) ? SvUVX (sv) <= 59000
572 need (enc, sizeof (UV) * 3); 690 : SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000)
573 enc->cur +=
574 SvIsUV(sv)
575 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
576 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
577 }
578 else
579 { 691 {
580 // optimise the "small number case" 692 // optimise the "small number case"
581 // code will likely be branchless and use only a single multiplication 693 // code will likely be branchless and use only a single multiplication
694 // works for numbers up to 59074
582 I32 i = SvIV (sv); 695 I32 i = SvIVX (sv);
583 U32 u; 696 U32 u;
584 char digit, nz = 0; 697 char digit, nz = 0;
585 698
586 need (enc, 6); 699 need (enc, 6);
587 700
… …
593 706
594 // now output digit by digit, each time masking out the integer part 707 // now output digit by digit, each time masking out the integer part
595 // and multiplying by 5 while moving the decimal point one to the right, 708 // and multiplying by 5 while moving the decimal point one to the right,
596 // resulting in a net multiplication by 10. 709 // resulting in a net multiplication by 10.
597 // we always write the digit to memory but conditionally increment 710 // we always write the digit to memory but conditionally increment
598 // the pointer, to ease the usage of conditional move instructions. 711 // the pointer, to enable the use of conditional move instructions.
599 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5; 712 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffffUL) * 5;
600 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5; 713 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffffUL) * 5;
601 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5; 714 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffffUL) * 5;
602 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5; 715 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffffUL) * 5;
603 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0' 716 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
604 } 717 }
718 else
719 {
720 // large integer, use the (rather slow) snprintf way.
721 need (enc, IVUV_MAXCHARS);
722 enc->cur +=
723 SvIsUV(sv)
724 ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))
725 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
726 }
605 } 727 }
606 else if (SvROK (sv)) 728 else if (SvROK (sv))
607 encode_rv (enc, SvRV (sv)); 729 encode_rv (enc, SvRV (sv));
608 else if (!SvOK (sv)) 730 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN)
609 encode_str (enc, "null", 4, 0); 731 encode_str (enc, "null", 4, 0);
610 else 732 else
611 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 733 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
612 SvPV_nolen (sv), SvFLAGS (sv)); 734 SvPV_nolen (sv), SvFLAGS (sv));
613} 735}
… …
623 enc.json = *json; 745 enc.json = *json;
624 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 746 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
625 enc.cur = SvPVX (enc.sv); 747 enc.cur = SvPVX (enc.sv);
626 enc.end = SvEND (enc.sv); 748 enc.end = SvEND (enc.sv);
627 enc.indent = 0; 749 enc.indent = 0;
628 enc.maxdepth = DEC_DEPTH (enc.json.flags); 750 enc.limit = enc.json.flags & F_ASCII ? 0x000080UL
751 : enc.json.flags & F_LATIN1 ? 0x000100UL
752 : 0x110000UL;
629 753
630 SvPOK_only (enc.sv); 754 SvPOK_only (enc.sv);
631 encode_sv (&enc, scalar); 755 encode_sv (&enc, scalar);
632 756
633 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 757 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
… …
654 JSON json; 778 JSON json;
655 U32 depth; // recursion depth 779 U32 depth; // recursion depth
656 U32 maxdepth; // recursion depth limit 780 U32 maxdepth; // recursion depth limit
657} dec_t; 781} dec_t;
658 782
659inline void 783INLINE void
784decode_comment (dec_t *dec)
785{
786 // only '#'-style comments allowed a.t.m.
787
788 while (*dec->cur && *dec->cur != 0x0a && *dec->cur != 0x0d)
789 ++dec->cur;
790}
791
792INLINE void
660decode_ws (dec_t *dec) 793decode_ws (dec_t *dec)
661{ 794{
662 for (;;) 795 for (;;)
663 { 796 {
664 char ch = *dec->cur; 797 char ch = *dec->cur;
665 798
666 if (ch > 0x20 799 if (ch > 0x20)
800 {
801 if (expect_false (ch == '#'))
802 {
803 if (dec->json.flags & F_RELAXED)
804 decode_comment (dec);
805 else
806 break;
807 }
808 else
809 break;
810 }
667 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) 811 else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)
668 break; 812 break; // parse error, but let higher level handle it, gives better error messages
669 813
670 ++dec->cur; 814 ++dec->cur;
671 } 815 }
672} 816}
673 817
… …
677 if (*dec->cur != ch) \ 821 if (*dec->cur != ch) \
678 ERR (# ch " expected"); \ 822 ERR (# ch " expected"); \
679 ++dec->cur; \ 823 ++dec->cur; \
680 SE 824 SE
681 825
682#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)") 826#define DEC_INC_DEPTH if (++dec->depth > dec->json.max_depth) ERR (ERR_NESTING_EXCEEDED)
683#define DEC_DEC_DEPTH --dec->depth 827#define DEC_DEC_DEPTH --dec->depth
684 828
685static SV *decode_sv (dec_t *dec); 829static SV *decode_sv (dec_t *dec);
686 830
687static signed char decode_hexdigit[256]; 831static signed char decode_hexdigit[256];
… …
779 923
780 if (hi >= 0x80) 924 if (hi >= 0x80)
781 { 925 {
782 utf8 = 1; 926 utf8 = 1;
783 927
784 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0); 928 cur = encode_utf8 (cur, hi);
785 } 929 }
786 else 930 else
787 *cur++ = hi; 931 *cur++ = hi;
788 } 932 }
789 break; 933 break;
… …
791 default: 935 default:
792 --dec_cur; 936 --dec_cur;
793 ERR ("illegal backslash escape sequence in string"); 937 ERR ("illegal backslash escape sequence in string");
794 } 938 }
795 } 939 }
796 else if (expect_true (ch >= 0x20 && ch <= 0x7f)) 940 else if (expect_true (ch >= 0x20 && ch < 0x80))
797 *cur++ = ch; 941 *cur++ = ch;
798 else if (ch >= 0x80) 942 else if (ch >= 0x80)
799 { 943 {
800 STRLEN clen; 944 STRLEN clen;
801 UV uch; 945 UV uch;
… …
922 is_nv = 1; 1066 is_nv = 1;
923 } 1067 }
924 1068
925 if (!is_nv) 1069 if (!is_nv)
926 { 1070 {
1071 int len = dec->cur - start;
1072
927 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so 1073 // special case the rather common 1..5-digit-int case
928 if (*start == '-') 1074 if (*start == '-')
929 switch (dec->cur - start) 1075 switch (len)
930 { 1076 {
931 case 2: return newSViv (-( start [1] - '0' * 1)); 1077 case 2: return newSViv (-( start [1] - '0' * 1));
932 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); 1078 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
933 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); 1079 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
934 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); 1080 case 5: return newSViv (-( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
1081 case 6: return newSViv (-(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
935 } 1082 }
936 else 1083 else
937 switch (dec->cur - start) 1084 switch (len)
938 { 1085 {
939 case 1: return newSViv ( start [0] - '0' * 1); 1086 case 1: return newSViv ( start [0] - '0' * 1);
940 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); 1087 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
941 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); 1088 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
942 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); 1089 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
1090 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
943 } 1091 }
944 1092
945 { 1093 {
946 UV uv; 1094 UV uv;
947 int numtype = grok_number (start, dec->cur - start, &uv); 1095 int numtype = grok_number (start, len, &uv);
948 if (numtype & IS_NUMBER_IN_UV) 1096 if (numtype & IS_NUMBER_IN_UV)
949 if (numtype & IS_NUMBER_NEG) 1097 if (numtype & IS_NUMBER_NEG)
950 { 1098 {
951 if (uv < (UV)IV_MIN) 1099 if (uv < (UV)IV_MIN)
952 return newSViv (-(IV)uv); 1100 return newSViv (-(IV)uv);
953 } 1101 }
954 else 1102 else
955 return newSVuv (uv); 1103 return newSVuv (uv);
956
957 // here would likely be the place for bigint support
958 } 1104 }
959 }
960 1105
961 // if we ever support bigint or bigfloat, this is the place for bigfloat 1106 len -= *start == '-' ? 1 : 0;
1107
1108 // does not fit into IV or UV, try NV
1109 if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len)
1110 #if defined (LDBL_DIG)
1111 || (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len)
1112 #endif
1113 )
1114 // fits into NV without loss of precision
1115 return newSVnv (Atof (start));
1116
1117 // everything else fails, convert it to a string
1118 return newSVpvn (start, dec->cur - start);
1119 }
1120
1121 // loss of precision here
962 return newSVnv (Atof (start)); 1122 return newSVnv (Atof (start));
963 1123
964fail: 1124fail:
965 return 0; 1125 return 0;
966} 1126}
… …
996 1156
997 if (*dec->cur != ',') 1157 if (*dec->cur != ',')
998 ERR (", or ] expected while parsing array"); 1158 ERR (", or ] expected while parsing array");
999 1159
1000 ++dec->cur; 1160 ++dec->cur;
1161
1162 decode_ws (dec);
1163
1164 if (*dec->cur == ']' && dec->json.flags & F_RELAXED)
1165 {
1166 ++dec->cur;
1167 break;
1168 }
1001 } 1169 }
1002 1170
1003 DEC_DEC_DEPTH; 1171 DEC_DEC_DEPTH;
1004 return newRV_noinc ((SV *)av); 1172 return newRV_noinc ((SV *)av);
1005 1173
… …
1021 if (*dec->cur == '}') 1189 if (*dec->cur == '}')
1022 ++dec->cur; 1190 ++dec->cur;
1023 else 1191 else
1024 for (;;) 1192 for (;;)
1025 { 1193 {
1026 decode_ws (dec); EXPECT_CH ('"'); 1194 EXPECT_CH ('"');
1027 1195
1028 // heuristic: assume that 1196 // heuristic: assume that
1029 // a) decode_str + hv_store_ent are abysmally slow. 1197 // a) decode_str + hv_store_ent are abysmally slow.
1030 // b) most hash keys are short, simple ascii text. 1198 // b) most hash keys are short, simple ascii text.
1031 // => try to "fast-match" such strings to avoid 1199 // => try to "fast-match" such strings to avoid
… …
1035 char *p = dec->cur; 1203 char *p = dec->cur;
1036 char *e = p + 24; // only try up to 24 bytes 1204 char *e = p + 24; // only try up to 24 bytes
1037 1205
1038 for (;;) 1206 for (;;)
1039 { 1207 {
1040 // the >= 0x80 is true on most architectures 1208 // the >= 0x80 is false on most architectures
1041 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\') 1209 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1042 { 1210 {
1043 // slow path, back up and use decode_str 1211 // slow path, back up and use decode_str
1044 SV *key = decode_str (dec); 1212 SV *key = decode_str (dec);
1045 if (!key) 1213 if (!key)
1046 goto fail; 1214 goto fail;
1047 1215
1048 decode_ws (dec); EXPECT_CH (':'); 1216 decode_ws (dec); EXPECT_CH (':');
1049 1217
1218 decode_ws (dec);
1050 value = decode_sv (dec); 1219 value = decode_sv (dec);
1051 if (!value) 1220 if (!value)
1052 { 1221 {
1053 SvREFCNT_dec (key); 1222 SvREFCNT_dec (key);
1054 goto fail; 1223 goto fail;
… …
1066 int len = p - key; 1235 int len = p - key;
1067 dec->cur = p + 1; 1236 dec->cur = p + 1;
1068 1237
1069 decode_ws (dec); EXPECT_CH (':'); 1238 decode_ws (dec); EXPECT_CH (':');
1070 1239
1240 decode_ws (dec);
1071 value = decode_sv (dec); 1241 value = decode_sv (dec);
1072 if (!value) 1242 if (!value)
1073 goto fail; 1243 goto fail;
1074 1244
1075 hv_store (hv, key, len, value, 0); 1245 hv_store (hv, key, len, value, 0);
… …
1091 1261
1092 if (*dec->cur != ',') 1262 if (*dec->cur != ',')
1093 ERR (", or } expected while parsing object/hash"); 1263 ERR (", or } expected while parsing object/hash");
1094 1264
1095 ++dec->cur; 1265 ++dec->cur;
1266
1267 decode_ws (dec);
1268
1269 if (*dec->cur == '}' && dec->json.flags & F_RELAXED)
1270 {
1271 ++dec->cur;
1272 break;
1273 }
1096 } 1274 }
1097 1275
1098 DEC_DEC_DEPTH; 1276 DEC_DEC_DEPTH;
1099 sv = newRV_noinc ((SV *)hv); 1277 sv = newRV_noinc ((SV *)hv);
1100 1278
… …
1113 // might want to optimise this for common cases. 1291 // might want to optimise this for common cases.
1114 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0); 1292 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1115 1293
1116 if (cb) 1294 if (cb)
1117 { 1295 {
1296 dSP;
1118 int count; 1297 int count;
1119 ENTER; SAVETMPS;
1120 1298
1121 dSP; PUSHMARK (SP); 1299 ENTER; SAVETMPS; PUSHMARK (SP);
1122 XPUSHs (HeVAL (he)); 1300 XPUSHs (HeVAL (he));
1123 1301
1124 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN; 1302 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1125 1303
1126 if (count == 1) 1304 if (count == 1)
… …
1134 } 1312 }
1135 } 1313 }
1136 1314
1137 if (dec->json.cb_object) 1315 if (dec->json.cb_object)
1138 { 1316 {
1317 dSP;
1139 int count; 1318 int count;
1140 ENTER; SAVETMPS;
1141 1319
1142 dSP; ENTER; SAVETMPS; PUSHMARK (SP); 1320 ENTER; SAVETMPS; PUSHMARK (SP);
1143 XPUSHs (sv_2mortal (sv)); 1321 XPUSHs (sv_2mortal (sv));
1144 1322
1145 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN; 1323 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1146 1324
1147 if (count == 1) 1325 if (count == 1)
… …
1165} 1343}
1166 1344
1167static SV * 1345static SV *
1168decode_sv (dec_t *dec) 1346decode_sv (dec_t *dec)
1169{ 1347{
1170 decode_ws (dec);
1171
1172 // the beauty of JSON: you need exactly one character lookahead 1348 // the beauty of JSON: you need exactly one character lookahead
1173 // to parse anything. 1349 // to parse everything.
1174 switch (*dec->cur) 1350 switch (*dec->cur)
1175 { 1351 {
1176 case '"': ++dec->cur; return decode_str (dec); 1352 case '"': ++dec->cur; return decode_str (dec);
1177 case '[': ++dec->cur; return decode_av (dec); 1353 case '[': ++dec->cur; return decode_av (dec);
1178 case '{': ++dec->cur; return decode_hv (dec); 1354 case '{': ++dec->cur; return decode_hv (dec);
1179 1355
1180 case '-': 1356 case '-':
1181 case '0': case '1': case '2': case '3': case '4': 1357 case '0': case '1': case '2': case '3': case '4':
1182 case '5': case '6': case '7': case '8': case '9': 1358 case '5': case '6': case '7': case '8': case '9':
1183 return decode_num (dec); 1359 return decode_num (dec);
1184 1360
1185 case 't': 1361 case 't':
1186 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1362 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1187 { 1363 {
1188 dec->cur += 4; 1364 dec->cur += 4;
1365#if JSON_SLOW
1366 json_true = get_bool ("JSON::XS::true");
1367#endif
1189 return SvREFCNT_inc (json_true); 1368 return newSVsv (json_true);
1190 } 1369 }
1191 else 1370 else
1192 ERR ("'true' expected"); 1371 ERR ("'true' expected");
1193 1372
1194 break; 1373 break;
1195 1374
1196 case 'f': 1375 case 'f':
1197 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1376 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1198 { 1377 {
1199 dec->cur += 5; 1378 dec->cur += 5;
1379#if JSON_SLOW
1380 json_false = get_bool ("JSON::XS::false");
1381#endif
1200 return SvREFCNT_inc (json_false); 1382 return newSVsv (json_false);
1201 } 1383 }
1202 else 1384 else
1203 ERR ("'false' expected"); 1385 ERR ("'false' expected");
1204 1386
1205 break; 1387 break;
… …
1223fail: 1405fail:
1224 return 0; 1406 return 0;
1225} 1407}
1226 1408
1227static SV * 1409static SV *
1228decode_json (SV *string, JSON *json, UV *offset_return) 1410decode_json (SV *string, JSON *json, STRLEN *offset_return)
1229{ 1411{
1230 dec_t dec; 1412 dec_t dec;
1231 UV offset; 1413 STRLEN offset;
1232 SV *sv; 1414 SV *sv;
1233 1415
1234 SvGETMAGIC (string); 1416 SvGETMAGIC (string);
1235 SvUPGRADE (string, SVt_PV); 1417 SvUPGRADE (string, SVt_PV);
1236 1418
1237 if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags)) 1419 /* work around a bug in perl 5.10, which causes SvCUR to fail an
1420 * assertion with -DDEBUGGING, although SvCUR is documented to
1421 * return the xpv_cur field which certainly exists after upgrading.
1422 * according to nicholas clark, calling SvPOK fixes this.
1423 * But it doesn't fix it, so try another workaround, call SvPV_nolen
1424 * and hope for the best.
1425 * Damnit, SvPV_nolen still trips over yet another assertion. This
1426 * assertion business is seriously broken, try yet another workaround
1427 * for the broken -DDEBUGGING.
1428 */
1429#ifdef DEBUGGING
1430 offset = SvOK (string) ? sv_len (string) : 0;
1431#else
1432 offset = SvCUR (string);
1433#endif
1434
1435 if (offset > json->max_size && json->max_size)
1238 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", 1436 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1239 (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags)); 1437 (unsigned long)SvCUR (string), (unsigned long)json->max_size);
1240 1438
1241 if (json->flags & F_UTF8) 1439 if (json->flags & F_UTF8)
1242 sv_utf8_downgrade (string, 0); 1440 sv_utf8_downgrade (string, 0);
1243 else 1441 else
1244 sv_utf8_upgrade (string); 1442 sv_utf8_upgrade (string);
1245 1443
1246 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1444 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
1247 1445
1248 dec.json = *json; 1446 dec.json = *json;
1249 dec.cur = SvPVX (string); 1447 dec.cur = SvPVX (string);
1250 dec.end = SvEND (string); 1448 dec.end = SvEND (string);
1251 dec.err = 0; 1449 dec.err = 0;
1252 dec.depth = 0; 1450 dec.depth = 0;
1253 dec.maxdepth = DEC_DEPTH (dec.json.flags);
1254 1451
1255 if (dec.json.cb_object || dec.json.cb_sk_object) 1452 if (dec.json.cb_object || dec.json.cb_sk_object)
1256 dec.json.flags |= F_HOOK; 1453 dec.json.flags |= F_HOOK;
1257 1454
1258 *dec.end = 0; // this should basically be a nop, too, but make sure it's there 1455 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1456
1457 decode_ws (&dec);
1259 sv = decode_sv (&dec); 1458 sv = decode_sv (&dec);
1260 1459
1261 if (!(offset_return || !sv)) 1460 if (!(offset_return || !sv))
1262 { 1461 {
1263 // check for trailing garbage 1462 // check for trailing garbage
… …
1307 1506
1308 return sv; 1507 return sv;
1309} 1508}
1310 1509
1311///////////////////////////////////////////////////////////////////////////// 1510/////////////////////////////////////////////////////////////////////////////
1511// incremental parser
1512
1513static void
1514incr_parse (JSON *self)
1515{
1516 const char *p = SvPVX (self->incr_text) + self->incr_pos;
1517
1518 for (;;)
1519 {
1520 //printf ("loop pod %d *p<%c><%s>, mode %d nest %d\n", p - SvPVX (self->incr_text), *p, p, self->incr_mode, self->incr_nest);//D
1521 switch (self->incr_mode)
1522 {
1523 // only used for intiial whitespace skipping
1524 case INCR_M_WS:
1525 for (;;)
1526 {
1527 if (*p > 0x20)
1528 {
1529 self->incr_mode = INCR_M_JSON;
1530 goto incr_m_json;
1531 }
1532 else if (!*p)
1533 goto interrupt;
1534
1535 ++p;
1536 }
1537
1538 // skip a single char inside a string (for \\-processing)
1539 case INCR_M_BS:
1540 if (!*p)
1541 goto interrupt;
1542
1543 ++p;
1544 self->incr_mode = INCR_M_STR;
1545 goto incr_m_str;
1546
1547 // inside a string
1548 case INCR_M_STR:
1549 incr_m_str:
1550 for (;;)
1551 {
1552 if (*p == '"')
1553 {
1554 ++p;
1555 self->incr_mode = INCR_M_JSON;
1556
1557 if (!self->incr_nest)
1558 goto interrupt;
1559
1560 goto incr_m_json;
1561 }
1562 else if (*p == '\\')
1563 {
1564 ++p; // "virtually" consumes character after \
1565
1566 if (!*p) // if at end of string we have to switch modes
1567 {
1568 self->incr_mode = INCR_M_BS;
1569 goto interrupt;
1570 }
1571 }
1572 else if (!*p)
1573 goto interrupt;
1574
1575 ++p;
1576 }
1577
1578 // after initial ws, outside string
1579 case INCR_M_JSON:
1580 incr_m_json:
1581 for (;;)
1582 {
1583 switch (*p++)
1584 {
1585 case 0:
1586 --p;
1587 goto interrupt;
1588
1589 case 0x09:
1590 case 0x0a:
1591 case 0x0d:
1592 case 0x20:
1593 if (!self->incr_nest)
1594 {
1595 --p; // do not eat the whitespace, let the next round do it
1596 goto interrupt;
1597 }
1598 break;
1599
1600 case '"':
1601 self->incr_mode = INCR_M_STR;
1602 goto incr_m_str;
1603
1604 case '[':
1605 case '{':
1606 if (++self->incr_nest > self->max_depth)
1607 croak (ERR_NESTING_EXCEEDED);
1608 break;
1609
1610 case ']':
1611 case '}':
1612 if (--self->incr_nest <= 0)
1613 goto interrupt;
1614 }
1615 }
1616 }
1617
1618 modechange:
1619 ;
1620 }
1621
1622interrupt:
1623 self->incr_pos = p - SvPVX (self->incr_text);
1624 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D
1625}
1626
1627/////////////////////////////////////////////////////////////////////////////
1312// XS interface functions 1628// XS interface functions
1313 1629
1314MODULE = JSON::XS PACKAGE = JSON::XS 1630MODULE = JSON::XS PACKAGE = JSON::XS
1315 1631
1316BOOT: 1632BOOT:
… …
1325 : -1; 1641 : -1;
1326 1642
1327 json_stash = gv_stashpv ("JSON::XS" , 1); 1643 json_stash = gv_stashpv ("JSON::XS" , 1);
1328 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); 1644 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1329 1645
1330 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true ); 1646 json_true = get_bool ("JSON::XS::true");
1331 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); 1647 json_false = get_bool ("JSON::XS::false");
1332} 1648}
1333 1649
1334PROTOTYPES: DISABLE 1650PROTOTYPES: DISABLE
1651
1652void CLONE (...)
1653 CODE:
1654 json_stash = 0;
1655 json_boolean_stash = 0;
1335 1656
1336void new (char *klass) 1657void new (char *klass)
1337 PPCODE: 1658 PPCODE:
1338{ 1659{
1339 SV *pv = NEWSV (0, sizeof (JSON)); 1660 SV *pv = NEWSV (0, sizeof (JSON));
1340 SvPOK_only (pv); 1661 SvPOK_only (pv);
1341 Zero (SvPVX (pv), 1, JSON); 1662 json_init ((JSON *)SvPVX (pv));
1342 ((JSON *)SvPVX (pv))->flags = F_DEFAULT; 1663 XPUSHs (sv_2mortal (sv_bless (
1343 XPUSHs (sv_2mortal (sv_bless (newRV_noinc (pv), json_stash))); 1664 newRV_noinc (pv),
1665 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1666 )));
1344} 1667}
1345 1668
1346void ascii (JSON *self, int enable = 1) 1669void ascii (JSON *self, int enable = 1)
1347 ALIAS: 1670 ALIAS:
1348 ascii = F_ASCII 1671 ascii = F_ASCII
… …
1355 pretty = F_PRETTY 1678 pretty = F_PRETTY
1356 allow_nonref = F_ALLOW_NONREF 1679 allow_nonref = F_ALLOW_NONREF
1357 shrink = F_SHRINK 1680 shrink = F_SHRINK
1358 allow_blessed = F_ALLOW_BLESSED 1681 allow_blessed = F_ALLOW_BLESSED
1359 convert_blessed = F_CONV_BLESSED 1682 convert_blessed = F_CONV_BLESSED
1683 relaxed = F_RELAXED
1684 allow_unknown = F_ALLOW_UNKNOWN
1360 PPCODE: 1685 PPCODE:
1361{ 1686{
1362 if (enable) 1687 if (enable)
1363 self->flags |= ix; 1688 self->flags |= ix;
1364 else 1689 else
1365 self->flags &= ~ix; 1690 self->flags &= ~ix;
1366 1691
1367 XPUSHs (ST (0)); 1692 XPUSHs (ST (0));
1368} 1693}
1369 1694
1370void max_depth (JSON *self, UV max_depth = 0x80000000UL) 1695void get_ascii (JSON *self)
1696 ALIAS:
1697 get_ascii = F_ASCII
1698 get_latin1 = F_LATIN1
1699 get_utf8 = F_UTF8
1700 get_indent = F_INDENT
1701 get_canonical = F_CANONICAL
1702 get_space_before = F_SPACE_BEFORE
1703 get_space_after = F_SPACE_AFTER
1704 get_allow_nonref = F_ALLOW_NONREF
1705 get_shrink = F_SHRINK
1706 get_allow_blessed = F_ALLOW_BLESSED
1707 get_convert_blessed = F_CONV_BLESSED
1708 get_relaxed = F_RELAXED
1709 get_allow_unknown = F_ALLOW_UNKNOWN
1371 PPCODE: 1710 PPCODE:
1372{ 1711 XPUSHs (boolSV (self->flags & ix));
1373 UV log2 = 0;
1374 1712
1375 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL; 1713void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
1376 1714 PPCODE:
1377 while ((1UL << log2) < max_depth) 1715 self->max_depth = max_depth;
1378 ++log2;
1379
1380 self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1381
1382 XPUSHs (ST (0)); 1716 XPUSHs (ST (0));
1383}
1384 1717
1718U32 get_max_depth (JSON *self)
1719 CODE:
1720 RETVAL = self->max_depth;
1721 OUTPUT:
1722 RETVAL
1723
1385void max_size (JSON *self, UV max_size = 0) 1724void max_size (JSON *self, U32 max_size = 0)
1386 PPCODE: 1725 PPCODE:
1387{ 1726 self->max_size = max_size;
1388 UV log2 = 0;
1389
1390 if (max_size > 0x80000000UL) max_size = 0x80000000UL;
1391 if (max_size == 1) max_size = 2;
1392
1393 while ((1UL << log2) < max_size)
1394 ++log2;
1395
1396 self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1397
1398 XPUSHs (ST (0)); 1727 XPUSHs (ST (0));
1399} 1728
1729int get_max_size (JSON *self)
1730 CODE:
1731 RETVAL = self->max_size;
1732 OUTPUT:
1733 RETVAL
1400 1734
1401void filter_json_object (JSON *self, SV *cb = &PL_sv_undef) 1735void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1402 PPCODE: 1736 PPCODE:
1403{ 1737{
1404 SvREFCNT_dec (self->cb_object); 1738 SvREFCNT_dec (self->cb_object);
… …
1438 XPUSHs (decode_json (jsonstr, self, 0)); 1772 XPUSHs (decode_json (jsonstr, self, 0));
1439 1773
1440void decode_prefix (JSON *self, SV *jsonstr) 1774void decode_prefix (JSON *self, SV *jsonstr)
1441 PPCODE: 1775 PPCODE:
1442{ 1776{
1443 UV offset; 1777 STRLEN offset;
1444 EXTEND (SP, 2); 1778 EXTEND (SP, 2);
1445 PUSHs (decode_json (jsonstr, self, &offset)); 1779 PUSHs (decode_json (jsonstr, self, &offset));
1446 PUSHs (sv_2mortal (newSVuv (offset))); 1780 PUSHs (sv_2mortal (newSVuv (offset)));
1781}
1782
1783void incr_parse (JSON *self, SV *jsonstr = 0)
1784 PPCODE:
1785{
1786 if (!self->incr_text)
1787 self->incr_text = newSVpvn ("", 0);
1788
1789 // append data, if any
1790 if (jsonstr)
1791 {
1792 if (SvUTF8 (jsonstr) && !SvUTF8 (self->incr_text))
1793 {
1794 /* utf-8-ness differs, need to upgrade */
1795 sv_utf8_upgrade (self->incr_text);
1796
1797 if (self->incr_pos)
1798 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
1799 - (U8 *)SvPVX (self->incr_text);
1800 }
1801
1802 {
1803 STRLEN len;
1804 const char *str = SvPV (jsonstr, len);
1805 SvGROW (self->incr_text, SvCUR (self->incr_text) + len + 1);
1806 Move (str, SvEND (self->incr_text), len, char);
1807 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
1808 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
1809 }
1810 }
1811
1812 if (GIMME_V != G_VOID)
1813 do
1814 {
1815 STRLEN offset;
1816
1817 if (!INCR_DONE (self))
1818 {
1819 incr_parse (self);
1820
1821 if (self->incr_pos > self->max_size && self->max_size)
1822 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1823 (unsigned long)self->incr_pos, (unsigned long)self->max_size);
1824
1825 if (!INCR_DONE (self))
1826 break;
1827 }
1828
1829 XPUSHs (decode_json (self->incr_text, self, &offset));
1830
1831 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + offset);
1832 self->incr_pos -= offset;
1833 self->incr_nest = 0;
1834 self->incr_mode = 0;
1835 }
1836 while (GIMME_V == G_ARRAY);
1837}
1838
1839SV *incr_text (JSON *self)
1840 ATTRS: lvalue
1841 CODE:
1842{
1843 if (self->incr_pos)
1844 croak ("incr_text can not be called when the incremental parser already started parsing");
1845
1846 RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef;
1847}
1848 OUTPUT:
1849 RETVAL
1850
1851void incr_skip (JSON *self)
1852 CODE:
1853{
1854 if (self->incr_pos)
1855 {
1856 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + self->incr_pos);
1857 self->incr_pos = 0;
1858 self->incr_nest = 0;
1859 self->incr_mode = 0;
1860 }
1861}
1862
1863void incr_reset (JSON *self)
1864 CODE:
1865{
1866 SvREFCNT_dec (self->incr_text);
1867 self->incr_text = 0;
1868 self->incr_pos = 0;
1869 self->incr_nest = 0;
1870 self->incr_mode = 0;
1447} 1871}
1448 1872
1449void DESTROY (JSON *self) 1873void DESTROY (JSON *self)
1450 CODE: 1874 CODE:
1451 SvREFCNT_dec (self->cb_sk_object); 1875 SvREFCNT_dec (self->cb_sk_object);
1452 SvREFCNT_dec (self->cb_object); 1876 SvREFCNT_dec (self->cb_object);
1877 SvREFCNT_dec (self->incr_text);
1453 1878
1454PROTOTYPES: ENABLE 1879PROTOTYPES: ENABLE
1455 1880
1456void to_json (SV *scalar) 1881void encode_json (SV *scalar)
1882 ALIAS:
1883 to_json_ = 0
1884 encode_json = F_UTF8
1457 PPCODE: 1885 PPCODE:
1458{ 1886{
1459 JSON json = { F_DEFAULT | F_UTF8 }; 1887 JSON json;
1888 json_init (&json);
1889 json.flags |= ix;
1460 XPUSHs (encode_json (scalar, &json)); 1890 XPUSHs (encode_json (scalar, &json));
1461} 1891}
1462 1892
1463void from_json (SV *jsonstr) 1893void decode_json (SV *jsonstr)
1894 ALIAS:
1895 from_json_ = 0
1896 decode_json = F_UTF8
1464 PPCODE: 1897 PPCODE:
1465{ 1898{
1466 JSON json = { F_DEFAULT | F_UTF8 }; 1899 JSON json;
1900 json_init (&json);
1901 json.flags |= ix;
1467 XPUSHs (decode_json (jsonstr, &json, 0)); 1902 XPUSHs (decode_json (jsonstr, &json, 0));
1468} 1903}
1469 1904
1905

Diff Legend

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