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 1.51 by root, Mon Jul 2 00:48:18 2007 UTC vs.
Revision 1.87 by root, Tue Jun 3 06:43:45 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 && (json)->incr_mode == INCR_M_JSON)
85
65typedef struct { 86typedef struct {
66 U32 flags; 87 U32 flags;
67 SV *cb_object, *cb_sk_object; 88 U32 max_depth;
89 STRLEN max_size;
90
91 SV *cb_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 unsigned char incr_nest; // {[]}-nesting level
98 unsigned char incr_mode;
68} JSON; 99} JSON;
100
101INLINE void
102json_init (JSON *json)
103{
104 Zero (json, 1, JSON);
105 json->max_depth = 512;
106}
69 107
70///////////////////////////////////////////////////////////////////////////// 108/////////////////////////////////////////////////////////////////////////////
71// utility functions 109// utility functions
72 110
73inline 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
74shrink (SV *sv) 123shrink (SV *sv)
75{ 124{
76 sv_utf8_downgrade (sv, 1); 125 sv_utf8_downgrade (sv, 1);
126
77 if (SvLEN (sv) > SvCUR (sv) + 1) 127 if (SvLEN (sv) > SvCUR (sv) + 1)
78 { 128 {
79#ifdef SvPV_shrink_to_cur 129#ifdef SvPV_shrink_to_cur
80 SvPV_shrink_to_cur (sv); 130 SvPV_shrink_to_cur (sv);
81#elif defined (SvPV_renew) 131#elif defined (SvPV_renew)
… …
87// 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
88// case of an error. 138// case of an error.
89// we special-case "safe" characters from U+80 .. U+7FF, 139// we special-case "safe" characters from U+80 .. U+7FF,
90// but use the very good perl function to parse anything else. 140// but use the very good perl function to parse anything else.
91// note that we never call this function for a ascii codepoints 141// note that we never call this function for a ascii codepoints
92inline UV 142INLINE UV
93decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) 143decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
94{ 144{
95 if (expect_false (s[0] > 0xdf || s[0] < 0xc2)) 145 if (expect_true (len >= 2
96 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); 146 && IN_RANGE_INC (char, s[0], 0xc2, 0xdf)
97 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) 147 && IN_RANGE_INC (char, s[1], 0x80, 0xbf)))
98 { 148 {
99 *clen = 2; 149 *clen = 2;
100 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 150 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
101 } 151 }
102 else 152 else
103 { 153 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
104 *clen = (STRLEN)-1; 154}
105 return (UV)-1; 155
106 } 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;
107} 178}
108 179
109///////////////////////////////////////////////////////////////////////////// 180/////////////////////////////////////////////////////////////////////////////
110// encoder 181// encoder
111 182
… …
115 char *cur; // SvPVX (sv) + current output position 186 char *cur; // SvPVX (sv) + current output position
116 char *end; // SvEND (sv) 187 char *end; // SvEND (sv)
117 SV *sv; // result scalar 188 SV *sv; // result scalar
118 JSON json; 189 JSON json;
119 U32 indent; // indentation level 190 U32 indent; // indentation level
120 U32 maxdepth; // max. indentation/recursion level 191 UV limit; // escape character values >= this value when encoding
121} enc_t; 192} enc_t;
122 193
123inline void 194INLINE void
124need (enc_t *enc, STRLEN len) 195need (enc_t *enc, STRLEN len)
125{ 196{
126 if (expect_false (enc->cur + len >= enc->end)) 197 if (expect_false (enc->cur + len >= enc->end))
127 { 198 {
128 STRLEN cur = enc->cur - SvPVX (enc->sv); 199 STRLEN cur = enc->cur - SvPVX (enc->sv);
… …
130 enc->cur = SvPVX (enc->sv) + cur; 201 enc->cur = SvPVX (enc->sv) + cur;
131 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 202 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
132 } 203 }
133} 204}
134 205
135inline void 206INLINE void
136encode_ch (enc_t *enc, char ch) 207encode_ch (enc_t *enc, char ch)
137{ 208{
138 need (enc, 1); 209 need (enc, 1);
139 *enc->cur++ = ch; 210 *enc->cur++ = ch;
140} 211}
… …
194 { 265 {
195 uch = ch; 266 uch = ch;
196 clen = 1; 267 clen = 1;
197 } 268 }
198 269
199 if (uch > 0x10FFFFUL) 270 if (uch < 0x80/*0x20*/ || uch >= enc->limit)
200 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
201
202 if (uch < 0x80 || enc->json.flags & F_ASCII || (enc->json.flags & F_LATIN1 && uch > 0xFF))
203 { 271 {
204 if (uch > 0xFFFFUL) 272 if (uch >= 0x10000UL)
205 { 273 {
274 if (uch >= 0x110000UL)
275 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
276
206 need (enc, len += 11); 277 need (enc, len += 11);
207 sprintf (enc->cur, "\\u%04x\\u%04x", 278 sprintf (enc->cur, "\\u%04x\\u%04x",
208 (int)((uch - 0x10000) / 0x400 + 0xD800), 279 (int)((uch - 0x10000) / 0x400 + 0xD800),
209 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 280 (int)((uch - 0x10000) % 0x400 + 0xDC00));
210 enc->cur += 12; 281 enc->cur += 12;
… …
238 while (--clen); 309 while (--clen);
239 } 310 }
240 else 311 else
241 { 312 {
242 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
243 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 314 enc->cur = encode_utf8 (enc->cur, uch);
244 ++str; 315 ++str;
245 } 316 }
246 } 317 }
247 } 318 }
248 } 319 }
249 320
250 --len; 321 --len;
251 } 322 }
252} 323}
253 324
254inline void 325INLINE void
255encode_indent (enc_t *enc) 326encode_indent (enc_t *enc)
256{ 327{
257 if (enc->json.flags & F_INDENT) 328 if (enc->json.flags & F_INDENT)
258 { 329 {
259 int spaces = enc->indent * INDENT_STEP; 330 int spaces = enc->indent * INDENT_STEP;
… …
262 memset (enc->cur, ' ', spaces); 333 memset (enc->cur, ' ', spaces);
263 enc->cur += spaces; 334 enc->cur += spaces;
264 } 335 }
265} 336}
266 337
267inline void 338INLINE void
268encode_space (enc_t *enc) 339encode_space (enc_t *enc)
269{ 340{
270 need (enc, 1); 341 need (enc, 1);
271 encode_ch (enc, ' '); 342 encode_ch (enc, ' ');
272} 343}
273 344
274inline void 345INLINE void
275encode_nl (enc_t *enc) 346encode_nl (enc_t *enc)
276{ 347{
277 if (enc->json.flags & F_INDENT) 348 if (enc->json.flags & F_INDENT)
278 { 349 {
279 need (enc, 1); 350 need (enc, 1);
280 encode_ch (enc, '\n'); 351 encode_ch (enc, '\n');
281 } 352 }
282} 353}
283 354
284inline void 355INLINE void
285encode_comma (enc_t *enc) 356encode_comma (enc_t *enc)
286{ 357{
287 encode_ch (enc, ','); 358 encode_ch (enc, ',');
288 359
289 if (enc->json.flags & F_INDENT) 360 if (enc->json.flags & F_INDENT)
… …
297static void 368static void
298encode_av (enc_t *enc, AV *av) 369encode_av (enc_t *enc, AV *av)
299{ 370{
300 int i, len = av_len (av); 371 int i, len = av_len (av);
301 372
302 if (enc->indent >= enc->maxdepth) 373 if (enc->indent >= enc->json.max_depth)
303 croak ("data structure too deep (hit recursion limit)"); 374 croak (ERR_NESTING_EXCEEDED);
304 375
305 encode_ch (enc, '['); encode_nl (enc); 376 encode_ch (enc, '[');
306 ++enc->indent; 377
378 if (len >= 0)
379 {
380 encode_nl (enc); ++enc->indent;
307 381
308 for (i = 0; i <= len; ++i) 382 for (i = 0; i <= len; ++i)
309 { 383 {
384 SV **svp = av_fetch (av, i, 0);
385
310 encode_indent (enc); 386 encode_indent (enc);
311 encode_sv (enc, *av_fetch (av, i, 0));
312 387
388 if (svp)
389 encode_sv (enc, *svp);
390 else
391 encode_str (enc, "null", 4, 0);
392
313 if (i < len) 393 if (i < len)
314 encode_comma (enc); 394 encode_comma (enc);
315 } 395 }
316 396
397 encode_nl (enc); --enc->indent; encode_indent (enc);
398 }
399
317 encode_nl (enc); 400 encode_ch (enc, ']');
318
319 --enc->indent;
320 encode_indent (enc); encode_ch (enc, ']');
321} 401}
322 402
323static void 403static void
324encode_he (enc_t *enc, HE *he) 404encode_hk (enc_t *enc, HE *he)
325{ 405{
326 encode_ch (enc, '"'); 406 encode_ch (enc, '"');
327 407
328 if (HeKLEN (he) == HEf_SVKEY) 408 if (HeKLEN (he) == HEf_SVKEY)
329 { 409 {
… …
342 encode_ch (enc, '"'); 422 encode_ch (enc, '"');
343 423
344 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc); 424 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
345 encode_ch (enc, ':'); 425 encode_ch (enc, ':');
346 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc); 426 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
347 encode_sv (enc, HeVAL (he));
348} 427}
349 428
350// compare hash entries, used when all keys are bytestrings 429// compare hash entries, used when all keys are bytestrings
351static int 430static int
352he_cmp_fast (const void *a_, const void *b_) 431he_cmp_fast (const void *a_, const void *b_)
… …
357 HE *b = *(HE **)b_; 436 HE *b = *(HE **)b_;
358 437
359 STRLEN la = HeKLEN (a); 438 STRLEN la = HeKLEN (a);
360 STRLEN lb = HeKLEN (b); 439 STRLEN lb = HeKLEN (b);
361 440
362 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 441 if (!(cmp = memcmp (HeKEY (b), HeKEY (a), lb < la ? lb : la)))
363 cmp = la - lb; 442 cmp = lb - la;
364 443
365 return cmp; 444 return cmp;
366} 445}
367 446
368// 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
369static int 448static int
370he_cmp_slow (const void *a, const void *b) 449he_cmp_slow (const void *a, const void *b)
371{ 450{
372 return sv_cmp (HeSVKEY_force (*(HE **)a), HeSVKEY_force (*(HE **)b)); 451 return sv_cmp (HeSVKEY_force (*(HE **)b), HeSVKEY_force (*(HE **)a));
373} 452}
374 453
375static void 454static void
376encode_hv (enc_t *enc, HV *hv) 455encode_hv (enc_t *enc, HV *hv)
377{ 456{
378 int count, i; 457 HE *he;
379 458
380 if (enc->indent >= enc->maxdepth) 459 if (enc->indent >= enc->json.max_depth)
381 croak ("data structure too deep (hit recursion limit)"); 460 croak (ERR_NESTING_EXCEEDED);
382 461
383 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 462 encode_ch (enc, '{');
384 463
385 if ((count = hv_iterinit (hv)))
386 {
387 // for canonical output we have to sort by keys first 464 // for canonical output we have to sort by keys first
388 // actually, this is mostly due to the stupid so-called 465 // actually, this is mostly due to the stupid so-called
389 // security workaround added somewhere in 5.8.x. 466 // security workaround added somewhere in 5.8.x.
390 // that randomises hash orderings 467 // that randomises hash orderings
391 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))
392 { 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 {
393 int fast = 1; 488 int i, fast = 1;
394 HE *he;
395#if defined(__BORLANDC__) || defined(_MSC_VER) 489#if defined(__BORLANDC__) || defined(_MSC_VER)
396 HE **hes = _alloca (count * sizeof (HE)); 490 HE **hes = _alloca (count * sizeof (HE));
397#else 491#else
398 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
399#endif 493#endif
… …
426 520
427 FREETMPS; 521 FREETMPS;
428 LEAVE; 522 LEAVE;
429 } 523 }
430 524
431 for (i = 0; i < count; ++i) 525 encode_nl (enc); ++enc->indent;
526
527 while (count--)
432 { 528 {
433 encode_indent (enc); 529 encode_indent (enc);
530 he = hes [count];
434 encode_he (enc, hes [i]); 531 encode_hk (enc, he);
532 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
435 533
436 if (i < count - 1) 534 if (count)
437 encode_comma (enc); 535 encode_comma (enc);
438 } 536 }
439 537
440 encode_nl (enc); 538 encode_nl (enc); --enc->indent; encode_indent (enc);
441 } 539 }
540 }
442 else 541 else
542 {
543 if (hv_iterinit (hv) || SvMAGICAL (hv))
544 if ((he = hv_iternext (hv)))
443 { 545 {
444 HE *he = hv_iternext (hv); 546 encode_nl (enc); ++enc->indent;
445 547
446 for (;;) 548 for (;;)
447 { 549 {
448 encode_indent (enc); 550 encode_indent (enc);
449 encode_he (enc, he); 551 encode_hk (enc, he);
552 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
450 553
451 if (!(he = hv_iternext (hv))) 554 if (!(he = hv_iternext (hv)))
452 break; 555 break;
453 556
454 encode_comma (enc); 557 encode_comma (enc);
455 } 558 }
456 559
457 encode_nl (enc); 560 encode_nl (enc); --enc->indent; encode_indent (enc);
458 } 561 }
459 } 562 }
460 563
461 --enc->indent; encode_indent (enc); encode_ch (enc, '}'); 564 encode_ch (enc, '}');
462} 565}
463 566
464// encode objects, arrays and special \0=false and \1=true values. 567// encode objects, arrays and special \0=false and \1=true values.
465static void 568static void
466encode_rv (enc_t *enc, SV *sv) 569encode_rv (enc_t *enc, SV *sv)
… …
470 SvGETMAGIC (sv); 573 SvGETMAGIC (sv);
471 svt = SvTYPE (sv); 574 svt = SvTYPE (sv);
472 575
473 if (expect_false (SvOBJECT (sv))) 576 if (expect_false (SvOBJECT (sv)))
474 { 577 {
578 HV *stash = !JSON_SLOW || json_boolean_stash
579 ? json_boolean_stash
580 : gv_stashpv ("JSON::XS::Boolean", 1);
581
475 if (SvSTASH (sv) == json_boolean_stash) 582 if (SvSTASH (sv) == stash)
476 { 583 {
477 if (SvIV (sv)) 584 if (SvIV (sv))
478 encode_str (enc, "true", 4, 0); 585 encode_str (enc, "true", 4, 0);
479 else 586 else
480 encode_str (enc, "false", 5, 0); 587 encode_str (enc, "false", 5, 0);
… …
488 } 595 }
489#endif 596#endif
490 if (enc->json.flags & F_CONV_BLESSED) 597 if (enc->json.flags & F_CONV_BLESSED)
491 { 598 {
492 // 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
493 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 1); 600 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
494 601
495 if (to_json) 602 if (to_json)
496 { 603 {
604 dSP;
605
497 dSP; ENTER; SAVETMPS; PUSHMARK (SP); 606 ENTER; SAVETMPS; PUSHMARK (SP);
498 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv))); 607 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
499 608
500 // 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
501 // check anyways.
502 PUTBACK; 610 PUTBACK;
503 assert (1 == call_sv ((SV *)GvCV (to_json), G_SCALAR)); 611 call_sv ((SV *)GvCV (to_json), G_SCALAR);
504 SPAGAIN; 612 SPAGAIN;
505 613
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
506 encode_sv (enc, POPs); 621 encode_sv (enc, sv);
507 622
508 FREETMPS; LEAVE; 623 FREETMPS; LEAVE;
509 } 624 }
510 else if (enc->json.flags & F_ALLOW_BLESSED) 625 else if (enc->json.flags & F_ALLOW_BLESSED)
511 encode_str (enc, "null", 4, 0); 626 encode_str (enc, "null", 4, 0);
… …
531 646
532 if (len == 1 && *pv == '1') 647 if (len == 1 && *pv == '1')
533 encode_str (enc, "true", 4, 0); 648 encode_str (enc, "true", 4, 0);
534 else if (len == 1 && *pv == '0') 649 else if (len == 1 && *pv == '0')
535 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);
536 else 653 else
537 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",
538 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 655 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
539 } 656 }
657 else if (enc->json.flags & F_ALLOW_UNKNOWN)
658 encode_str (enc, "null", 4, 0);
540 else 659 else
541 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",
542 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 661 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
543} 662}
544 663
… …
562 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 681 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
563 enc->cur += strlen (enc->cur); 682 enc->cur += strlen (enc->cur);
564 } 683 }
565 else if (SvIOKp (sv)) 684 else if (SvIOKp (sv))
566 { 685 {
567 // 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
568 if (SvUV (sv) & ~(UV)0x7fff) 687 // we assume two's complement
569 { 688 // we assume no aliasing issues in the union
570 // large integer, use the (rather slow) snprintf way. 689 if (SvIsUV (sv) ? SvUVX (sv) <= 59000
571 need (enc, sizeof (UV) * 3); 690 : SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000)
572 enc->cur +=
573 SvIsUV(sv)
574 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
575 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
576 }
577 else
578 { 691 {
579 // optimise the "small number case" 692 // optimise the "small number case"
580 // 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
581 I32 i = SvIV (sv); 695 I32 i = SvIVX (sv);
582 U32 u; 696 U32 u;
583 char digit, nz = 0; 697 char digit, nz = 0;
584 698
585 need (enc, 6); 699 need (enc, 6);
586 700
… …
592 706
593 // 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
594 // 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,
595 // resulting in a net multiplication by 10. 709 // resulting in a net multiplication by 10.
596 // we always write the digit to memory but conditionally increment 710 // we always write the digit to memory but conditionally increment
597 // the pointer, to ease the usage of conditional move instructions. 711 // the pointer, to enable the use of conditional move instructions.
598 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;
599 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;
600 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;
601 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;
602 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'
603 } 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 }
604 } 727 }
605 else if (SvROK (sv)) 728 else if (SvROK (sv))
606 encode_rv (enc, SvRV (sv)); 729 encode_rv (enc, SvRV (sv));
607 else if (!SvOK (sv)) 730 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN)
608 encode_str (enc, "null", 4, 0); 731 encode_str (enc, "null", 4, 0);
609 else 732 else
610 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",
611 SvPV_nolen (sv), SvFLAGS (sv)); 734 SvPV_nolen (sv), SvFLAGS (sv));
612} 735}
… …
622 enc.json = *json; 745 enc.json = *json;
623 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 746 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
624 enc.cur = SvPVX (enc.sv); 747 enc.cur = SvPVX (enc.sv);
625 enc.end = SvEND (enc.sv); 748 enc.end = SvEND (enc.sv);
626 enc.indent = 0; 749 enc.indent = 0;
627 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;
628 753
629 SvPOK_only (enc.sv); 754 SvPOK_only (enc.sv);
630 encode_sv (&enc, scalar); 755 encode_sv (&enc, scalar);
631 756
632 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 757 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
… …
653 JSON json; 778 JSON json;
654 U32 depth; // recursion depth 779 U32 depth; // recursion depth
655 U32 maxdepth; // recursion depth limit 780 U32 maxdepth; // recursion depth limit
656} dec_t; 781} dec_t;
657 782
658inline 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
659decode_ws (dec_t *dec) 793decode_ws (dec_t *dec)
660{ 794{
661 for (;;) 795 for (;;)
662 { 796 {
663 char ch = *dec->cur; 797 char ch = *dec->cur;
664 798
665 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 }
666 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) 811 else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)
667 break; 812 break; // parse error, but let higher level handle it, gives better error messages
668 813
669 ++dec->cur; 814 ++dec->cur;
670 } 815 }
671} 816}
672 817
… …
676 if (*dec->cur != ch) \ 821 if (*dec->cur != ch) \
677 ERR (# ch " expected"); \ 822 ERR (# ch " expected"); \
678 ++dec->cur; \ 823 ++dec->cur; \
679 SE 824 SE
680 825
681#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)
682#define DEC_DEC_DEPTH --dec->depth 827#define DEC_DEC_DEPTH --dec->depth
683 828
684static SV *decode_sv (dec_t *dec); 829static SV *decode_sv (dec_t *dec);
685 830
686static signed char decode_hexdigit[256]; 831static signed char decode_hexdigit[256];
… …
778 923
779 if (hi >= 0x80) 924 if (hi >= 0x80)
780 { 925 {
781 utf8 = 1; 926 utf8 = 1;
782 927
783 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0); 928 cur = encode_utf8 (cur, hi);
784 } 929 }
785 else 930 else
786 *cur++ = hi; 931 *cur++ = hi;
787 } 932 }
788 break; 933 break;
… …
790 default: 935 default:
791 --dec_cur; 936 --dec_cur;
792 ERR ("illegal backslash escape sequence in string"); 937 ERR ("illegal backslash escape sequence in string");
793 } 938 }
794 } 939 }
795 else if (expect_true (ch >= 0x20 && ch <= 0x7f)) 940 else if (expect_true (ch >= 0x20 && ch < 0x80))
796 *cur++ = ch; 941 *cur++ = ch;
797 else if (ch >= 0x80) 942 else if (ch >= 0x80)
798 { 943 {
799 STRLEN clen; 944 STRLEN clen;
800 UV uch; 945 UV uch;
… …
921 is_nv = 1; 1066 is_nv = 1;
922 } 1067 }
923 1068
924 if (!is_nv) 1069 if (!is_nv)
925 { 1070 {
1071 int len = dec->cur - start;
1072
926 // 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
927 if (*start == '-') 1074 if (*start == '-')
928 switch (dec->cur - start) 1075 switch (len)
929 { 1076 {
930 case 2: return newSViv (-( start [1] - '0' * 1)); 1077 case 2: return newSViv (-( start [1] - '0' * 1));
931 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); 1078 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
932 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));
933 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));
934 } 1082 }
935 else 1083 else
936 switch (dec->cur - start) 1084 switch (len)
937 { 1085 {
938 case 1: return newSViv ( start [0] - '0' * 1); 1086 case 1: return newSViv ( start [0] - '0' * 1);
939 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); 1087 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
940 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);
941 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);
942 } 1091 }
943 1092
944 { 1093 {
945 UV uv; 1094 UV uv;
946 int numtype = grok_number (start, dec->cur - start, &uv); 1095 int numtype = grok_number (start, len, &uv);
947 if (numtype & IS_NUMBER_IN_UV) 1096 if (numtype & IS_NUMBER_IN_UV)
948 if (numtype & IS_NUMBER_NEG) 1097 if (numtype & IS_NUMBER_NEG)
949 { 1098 {
950 if (uv < (UV)IV_MIN) 1099 if (uv < (UV)IV_MIN)
951 return newSViv (-(IV)uv); 1100 return newSViv (-(IV)uv);
952 } 1101 }
953 else 1102 else
954 return newSVuv (uv); 1103 return newSVuv (uv);
955
956 // here would likely be the place for bigint support
957 } 1104 }
958 }
959 1105
960 // 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
961 return newSVnv (Atof (start)); 1122 return newSVnv (Atof (start));
962 1123
963fail: 1124fail:
964 return 0; 1125 return 0;
965} 1126}
… …
995 1156
996 if (*dec->cur != ',') 1157 if (*dec->cur != ',')
997 ERR (", or ] expected while parsing array"); 1158 ERR (", or ] expected while parsing array");
998 1159
999 ++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 }
1000 } 1169 }
1001 1170
1002 DEC_DEC_DEPTH; 1171 DEC_DEC_DEPTH;
1003 return newRV_noinc ((SV *)av); 1172 return newRV_noinc ((SV *)av);
1004 1173
… …
1020 if (*dec->cur == '}') 1189 if (*dec->cur == '}')
1021 ++dec->cur; 1190 ++dec->cur;
1022 else 1191 else
1023 for (;;) 1192 for (;;)
1024 { 1193 {
1025 decode_ws (dec); EXPECT_CH ('"'); 1194 EXPECT_CH ('"');
1026 1195
1027 // heuristic: assume that 1196 // heuristic: assume that
1028 // a) decode_str + hv_store_ent are abysmally slow. 1197 // a) decode_str + hv_store_ent are abysmally slow.
1029 // b) most hash keys are short, simple ascii text. 1198 // b) most hash keys are short, simple ascii text.
1030 // => try to "fast-match" such strings to avoid 1199 // => try to "fast-match" such strings to avoid
… …
1034 char *p = dec->cur; 1203 char *p = dec->cur;
1035 char *e = p + 24; // only try up to 24 bytes 1204 char *e = p + 24; // only try up to 24 bytes
1036 1205
1037 for (;;) 1206 for (;;)
1038 { 1207 {
1039 // the >= 0x80 is true on most architectures 1208 // the >= 0x80 is false on most architectures
1040 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\') 1209 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1041 { 1210 {
1042 // slow path, back up and use decode_str 1211 // slow path, back up and use decode_str
1043 SV *key = decode_str (dec); 1212 SV *key = decode_str (dec);
1044 if (!key) 1213 if (!key)
1045 goto fail; 1214 goto fail;
1046 1215
1047 decode_ws (dec); EXPECT_CH (':'); 1216 decode_ws (dec); EXPECT_CH (':');
1048 1217
1218 decode_ws (dec);
1049 value = decode_sv (dec); 1219 value = decode_sv (dec);
1050 if (!value) 1220 if (!value)
1051 { 1221 {
1052 SvREFCNT_dec (key); 1222 SvREFCNT_dec (key);
1053 goto fail; 1223 goto fail;
… …
1065 int len = p - key; 1235 int len = p - key;
1066 dec->cur = p + 1; 1236 dec->cur = p + 1;
1067 1237
1068 decode_ws (dec); EXPECT_CH (':'); 1238 decode_ws (dec); EXPECT_CH (':');
1069 1239
1240 decode_ws (dec);
1070 value = decode_sv (dec); 1241 value = decode_sv (dec);
1071 if (!value) 1242 if (!value)
1072 goto fail; 1243 goto fail;
1073 1244
1074 hv_store (hv, key, len, value, 0); 1245 hv_store (hv, key, len, value, 0);
… …
1090 1261
1091 if (*dec->cur != ',') 1262 if (*dec->cur != ',')
1092 ERR (", or } expected while parsing object/hash"); 1263 ERR (", or } expected while parsing object/hash");
1093 1264
1094 ++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 }
1095 } 1274 }
1096 1275
1097 DEC_DEC_DEPTH; 1276 DEC_DEC_DEPTH;
1098 sv = newRV_noinc ((SV *)hv); 1277 sv = newRV_noinc ((SV *)hv);
1099 1278
1100 // check filter callbacks 1279 // check filter callbacks
1101 if (dec->json.flags & F_HOOK) 1280 if (dec->json.flags & F_HOOK)
1102 { 1281 {
1103 ENTER; SAVETMPS;
1104
1105 if (dec->json.cb_sk_object && HvKEYS (hv) == 1) 1282 if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
1106 { 1283 {
1284 HE *cb, *he;
1285
1286 hv_iterinit (hv);
1287 he = hv_iternext (hv);
1288 hv_iterinit (hv);
1289
1290 // the next line creates a mortal sv each time its called.
1291 // might want to optimise this for common cases.
1292 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1293
1294 if (cb)
1295 {
1296 dSP;
1297 int count;
1298
1299 ENTER; SAVETMPS; PUSHMARK (SP);
1300 XPUSHs (HeVAL (he));
1301
1302 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1303
1304 if (count == 1)
1305 {
1306 sv = newSVsv (POPs);
1307 FREETMPS; LEAVE;
1308 return sv;
1309 }
1310
1311 FREETMPS; LEAVE;
1312 }
1313 }
1314
1315 if (dec->json.cb_object)
1316 {
1317 dSP;
1107 int count; 1318 int count;
1108 1319
1109 dSP; PUSHMARK (SP); 1320 ENTER; SAVETMPS; PUSHMARK (SP);
1110 XPUSHs (sv_2mortal (sv)); 1321 XPUSHs (sv_2mortal (sv));
1111 1322
1112 PUTBACK; count = call_sv (dec->json.cb_sk_object, G_ARRAY); SPAGAIN; 1323 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1113 1324
1114 if (count == 1) 1325 if (count == 1)
1115 { 1326 {
1116 sv = newSVsv (POPs); 1327 sv = newSVsv (POPs);
1117 goto filter_ok; 1328 FREETMPS; LEAVE;
1329 return sv;
1118 } 1330 }
1119 1331
1120 SvREFCNT_inc (sv); 1332 SvREFCNT_inc (sv);
1333 FREETMPS; LEAVE;
1121 } 1334 }
1122
1123 if (dec->json.cb_object)
1124 {
1125 int count;
1126
1127 dSP; ENTER; SAVETMPS; PUSHMARK (SP);
1128 XPUSHs (sv_2mortal (sv));
1129
1130 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1131
1132 if (count == 1)
1133 {
1134 sv = newSVsv (POPs);
1135 goto filter_ok;
1136 }
1137
1138 SvREFCNT_inc (sv);
1139 }
1140
1141filter_ok:
1142 FREETMPS; LEAVE;
1143 } 1335 }
1144 1336
1145 return sv; 1337 return sv;
1146 1338
1147fail: 1339fail:
… …
1151} 1343}
1152 1344
1153static SV * 1345static SV *
1154decode_sv (dec_t *dec) 1346decode_sv (dec_t *dec)
1155{ 1347{
1156 decode_ws (dec);
1157
1158 // the beauty of JSON: you need exactly one character lookahead 1348 // the beauty of JSON: you need exactly one character lookahead
1159 // to parse anything. 1349 // to parse everything.
1160 switch (*dec->cur) 1350 switch (*dec->cur)
1161 { 1351 {
1162 case '"': ++dec->cur; return decode_str (dec); 1352 case '"': ++dec->cur; return decode_str (dec);
1163 case '[': ++dec->cur; return decode_av (dec); 1353 case '[': ++dec->cur; return decode_av (dec);
1164 case '{': ++dec->cur; return decode_hv (dec); 1354 case '{': ++dec->cur; return decode_hv (dec);
1165 1355
1166 case '-': 1356 case '-':
1167 case '0': case '1': case '2': case '3': case '4': 1357 case '0': case '1': case '2': case '3': case '4':
1168 case '5': case '6': case '7': case '8': case '9': 1358 case '5': case '6': case '7': case '8': case '9':
1169 return decode_num (dec); 1359 return decode_num (dec);
1170 1360
1171 case 't': 1361 case 't':
1172 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1362 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1173 { 1363 {
1174 dec->cur += 4; 1364 dec->cur += 4;
1365#if JSON_SLOW
1366 json_true = get_bool ("JSON::XS::true");
1367#endif
1175 return SvREFCNT_inc (json_true); 1368 return newSVsv (json_true);
1176 } 1369 }
1177 else 1370 else
1178 ERR ("'true' expected"); 1371 ERR ("'true' expected");
1179 1372
1180 break; 1373 break;
1181 1374
1182 case 'f': 1375 case 'f':
1183 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1376 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1184 { 1377 {
1185 dec->cur += 5; 1378 dec->cur += 5;
1379#if JSON_SLOW
1380 json_false = get_bool ("JSON::XS::false");
1381#endif
1186 return SvREFCNT_inc (json_false); 1382 return newSVsv (json_false);
1187 } 1383 }
1188 else 1384 else
1189 ERR ("'false' expected"); 1385 ERR ("'false' expected");
1190 1386
1191 break; 1387 break;
… …
1209fail: 1405fail:
1210 return 0; 1406 return 0;
1211} 1407}
1212 1408
1213static SV * 1409static SV *
1214decode_json (SV *string, JSON *json, UV *offset_return) 1410decode_json (SV *string, JSON *json, STRLEN *offset_return)
1215{ 1411{
1216 dec_t dec; 1412 dec_t dec;
1217 UV offset; 1413 STRLEN offset;
1218 SV *sv; 1414 SV *sv;
1219 1415
1220 SvGETMAGIC (string); 1416 SvGETMAGIC (string);
1221 SvUPGRADE (string, SVt_PV); 1417 SvUPGRADE (string, SVt_PV);
1222 1418
1223 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 */
1424 SvPOK (string);
1425
1426 if (SvCUR (string) > json->max_size && json->max_size)
1224 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", 1427 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1225 (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags)); 1428 (unsigned long)SvCUR (string), (unsigned long)json->max_size);
1226 1429
1227 if (json->flags & F_UTF8) 1430 if (json->flags & F_UTF8)
1228 sv_utf8_downgrade (string, 0); 1431 sv_utf8_downgrade (string, 0);
1229 else 1432 else
1230 sv_utf8_upgrade (string); 1433 sv_utf8_upgrade (string);
1231 1434
1232 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1435 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
1233 1436
1234 dec.json = *json; 1437 dec.json = *json;
1235 dec.cur = SvPVX (string); 1438 dec.cur = SvPVX (string);
1236 dec.end = SvEND (string); 1439 dec.end = SvEND (string);
1237 dec.err = 0; 1440 dec.err = 0;
1238 dec.depth = 0; 1441 dec.depth = 0;
1239 dec.maxdepth = DEC_DEPTH (dec.json.flags);
1240 1442
1241 if (dec.json.cb_object || dec.json.cb_sk_object) 1443 if (dec.json.cb_object || dec.json.cb_sk_object)
1242 dec.json.flags |= F_HOOK; 1444 dec.json.flags |= F_HOOK;
1243 1445
1244 *dec.end = 0; // this should basically be a nop, too, but make sure it's there 1446 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1447
1448 decode_ws (&dec);
1245 sv = decode_sv (&dec); 1449 sv = decode_sv (&dec);
1246 1450
1247 if (!(offset_return || !sv)) 1451 if (!(offset_return || !sv))
1248 { 1452 {
1249 // check for trailing garbage 1453 // check for trailing garbage
… …
1293 1497
1294 return sv; 1498 return sv;
1295} 1499}
1296 1500
1297///////////////////////////////////////////////////////////////////////////// 1501/////////////////////////////////////////////////////////////////////////////
1502// incremental parser
1503
1504static void
1505incr_parse (JSON *self)
1506{
1507 const char *p = SvPVX (self->incr_text) + self->incr_pos;
1508
1509 for (;;)
1510 {
1511 //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
1512 switch (self->incr_mode)
1513 {
1514 // only used for intiial whitespace skipping
1515 case INCR_M_WS:
1516 for (;;)
1517 {
1518 if (*p > 0x20)
1519 {
1520 self->incr_mode = INCR_M_JSON;
1521 goto incr_m_json;
1522 }
1523 else if (!*p)
1524 goto interrupt;
1525
1526 ++p;
1527 }
1528
1529 // skip a single char inside a string (for \\-processing)
1530 case INCR_M_BS:
1531 if (!*p)
1532 goto interrupt;
1533
1534 ++p;
1535 self->incr_mode = INCR_M_STR;
1536 goto incr_m_str;
1537
1538 // inside a string
1539 case INCR_M_STR:
1540 incr_m_str:
1541 for (;;)
1542 {
1543 if (*p == '"')
1544 {
1545 ++p;
1546 self->incr_mode = INCR_M_JSON;
1547
1548 if (!self->incr_nest)
1549 goto interrupt;
1550
1551 goto incr_m_json;
1552 }
1553 else if (*p == '\\')
1554 {
1555 ++p; // "virtually" consumes character after \
1556
1557 if (!*p) // if at end of string we have to switch modes
1558 {
1559 self->incr_mode = INCR_M_BS;
1560 goto interrupt;
1561 }
1562 }
1563 else if (!*p)
1564 goto interrupt;
1565
1566 ++p;
1567 }
1568
1569 // after initial ws, outside string
1570 case INCR_M_JSON:
1571 incr_m_json:
1572 for (;;)
1573 {
1574 switch (*p++)
1575 {
1576 case 0:
1577 --p;
1578 goto interrupt;
1579
1580 case 0x09:
1581 case 0x0a:
1582 case 0x0d:
1583 case 0x20:
1584 if (!self->incr_nest)
1585 {
1586 --p; // do not eat the whitespace, let the next round do it
1587 goto interrupt;
1588 }
1589 break;
1590
1591 case '"':
1592 self->incr_mode = INCR_M_STR;
1593 goto incr_m_str;
1594
1595 case '[':
1596 case '{':
1597 if (++self->incr_nest > self->max_depth)
1598 croak (ERR_NESTING_EXCEEDED);
1599 break;
1600
1601 case ']':
1602 case '}':
1603 if (!--self->incr_nest)
1604 goto interrupt;
1605 }
1606 }
1607 }
1608
1609 modechange:
1610 ;
1611 }
1612
1613interrupt:
1614 self->incr_pos = p - SvPVX (self->incr_text);
1615 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D
1616}
1617
1618/////////////////////////////////////////////////////////////////////////////
1298// XS interface functions 1619// XS interface functions
1299 1620
1300MODULE = JSON::XS PACKAGE = JSON::XS 1621MODULE = JSON::XS PACKAGE = JSON::XS
1301 1622
1302BOOT: 1623BOOT:
… …
1311 : -1; 1632 : -1;
1312 1633
1313 json_stash = gv_stashpv ("JSON::XS" , 1); 1634 json_stash = gv_stashpv ("JSON::XS" , 1);
1314 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); 1635 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1315 1636
1316 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true ); 1637 json_true = get_bool ("JSON::XS::true");
1317 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); 1638 json_false = get_bool ("JSON::XS::false");
1318} 1639}
1319 1640
1320PROTOTYPES: DISABLE 1641PROTOTYPES: DISABLE
1642
1643void CLONE (...)
1644 CODE:
1645 json_stash = 0;
1646 json_boolean_stash = 0;
1321 1647
1322void new (char *klass) 1648void new (char *klass)
1323 PPCODE: 1649 PPCODE:
1324{ 1650{
1325 SV *pv = NEWSV (0, sizeof (JSON)); 1651 SV *pv = NEWSV (0, sizeof (JSON));
1326 SvPOK_only (pv); 1652 SvPOK_only (pv);
1327 Zero (SvPVX (pv), 1, JSON); 1653 json_init ((JSON *)SvPVX (pv));
1328 ((JSON *)SvPVX (pv))->flags = F_DEFAULT; 1654 XPUSHs (sv_2mortal (sv_bless (
1329 XPUSHs (sv_2mortal (sv_bless (newRV_noinc (pv), json_stash))); 1655 newRV_noinc (pv),
1656 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1657 )));
1330} 1658}
1331 1659
1332void ascii (JSON *self, int enable = 1) 1660void ascii (JSON *self, int enable = 1)
1333 ALIAS: 1661 ALIAS:
1334 ascii = F_ASCII 1662 ascii = F_ASCII
… …
1341 pretty = F_PRETTY 1669 pretty = F_PRETTY
1342 allow_nonref = F_ALLOW_NONREF 1670 allow_nonref = F_ALLOW_NONREF
1343 shrink = F_SHRINK 1671 shrink = F_SHRINK
1344 allow_blessed = F_ALLOW_BLESSED 1672 allow_blessed = F_ALLOW_BLESSED
1345 convert_blessed = F_CONV_BLESSED 1673 convert_blessed = F_CONV_BLESSED
1674 relaxed = F_RELAXED
1675 allow_unknown = F_ALLOW_UNKNOWN
1346 PPCODE: 1676 PPCODE:
1347{ 1677{
1348 if (enable) 1678 if (enable)
1349 self->flags |= ix; 1679 self->flags |= ix;
1350 else 1680 else
1351 self->flags &= ~ix; 1681 self->flags &= ~ix;
1352 1682
1353 XPUSHs (ST (0)); 1683 XPUSHs (ST (0));
1354} 1684}
1355 1685
1356void max_depth (JSON *self, UV max_depth = 0x80000000UL) 1686void get_ascii (JSON *self)
1687 ALIAS:
1688 get_ascii = F_ASCII
1689 get_latin1 = F_LATIN1
1690 get_utf8 = F_UTF8
1691 get_indent = F_INDENT
1692 get_canonical = F_CANONICAL
1693 get_space_before = F_SPACE_BEFORE
1694 get_space_after = F_SPACE_AFTER
1695 get_allow_nonref = F_ALLOW_NONREF
1696 get_shrink = F_SHRINK
1697 get_allow_blessed = F_ALLOW_BLESSED
1698 get_convert_blessed = F_CONV_BLESSED
1699 get_relaxed = F_RELAXED
1700 get_allow_unknown = F_ALLOW_UNKNOWN
1357 PPCODE: 1701 PPCODE:
1358{ 1702 XPUSHs (boolSV (self->flags & ix));
1359 UV log2 = 0;
1360 1703
1361 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL; 1704void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
1362 1705 PPCODE:
1363 while ((1UL << log2) < max_depth) 1706 self->max_depth = max_depth;
1364 ++log2;
1365
1366 self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1367
1368 XPUSHs (ST (0)); 1707 XPUSHs (ST (0));
1369}
1370 1708
1709U32 get_max_depth (JSON *self)
1710 CODE:
1711 RETVAL = self->max_depth;
1712 OUTPUT:
1713 RETVAL
1714
1371void max_size (JSON *self, UV max_size = 0) 1715void max_size (JSON *self, U32 max_size = 0)
1372 PPCODE: 1716 PPCODE:
1373{ 1717 self->max_size = max_size;
1374 UV log2 = 0;
1375
1376 if (max_size > 0x80000000UL) max_size = 0x80000000UL;
1377 if (max_size == 1) max_size = 2;
1378
1379 while ((1UL << log2) < max_size)
1380 ++log2;
1381
1382 self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1383
1384 XPUSHs (ST (0)); 1718 XPUSHs (ST (0));
1385} 1719
1720int get_max_size (JSON *self)
1721 CODE:
1722 RETVAL = self->max_size;
1723 OUTPUT:
1724 RETVAL
1386 1725
1387void filter_json_object (JSON *self, SV *cb = &PL_sv_undef) 1726void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1388 ALIAS:
1389 filter_json_single_key_object = 1
1390 PPCODE: 1727 PPCODE:
1391{ 1728{
1392 SV **svp; 1729 SvREFCNT_dec (self->cb_object);
1730 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
1393 1731
1732 XPUSHs (ST (0));
1733}
1734
1735void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
1736 PPCODE:
1737{
1738 if (!self->cb_sk_object)
1739 self->cb_sk_object = newHV ();
1740
1394 if (!SvOK (cb)) 1741 if (SvOK (cb))
1395 cb = 0; 1742 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1396 else 1743 else
1397 cb = newSVsv (cb);
1398
1399 switch (ix)
1400 { 1744 {
1401 case 0: svp = &self->cb_object ; break; 1745 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1402 case 1: svp = &self->cb_sk_object; break; 1746
1747 if (!HvKEYS (self->cb_sk_object))
1748 {
1749 SvREFCNT_dec (self->cb_sk_object);
1750 self->cb_sk_object = 0;
1751 }
1403 } 1752 }
1404
1405 if (*svp)
1406 SvREFCNT_dec (*svp);
1407
1408 *svp = cb;
1409 1753
1410 XPUSHs (ST (0)); 1754 XPUSHs (ST (0));
1411} 1755}
1412 1756
1413void encode (JSON *self, SV *scalar) 1757void encode (JSON *self, SV *scalar)
… …
1419 XPUSHs (decode_json (jsonstr, self, 0)); 1763 XPUSHs (decode_json (jsonstr, self, 0));
1420 1764
1421void decode_prefix (JSON *self, SV *jsonstr) 1765void decode_prefix (JSON *self, SV *jsonstr)
1422 PPCODE: 1766 PPCODE:
1423{ 1767{
1424 UV offset; 1768 STRLEN offset;
1425 EXTEND (SP, 2); 1769 EXTEND (SP, 2);
1426 PUSHs (decode_json (jsonstr, self, &offset)); 1770 PUSHs (decode_json (jsonstr, self, &offset));
1427 PUSHs (sv_2mortal (newSVuv (offset))); 1771 PUSHs (sv_2mortal (newSVuv (offset)));
1428} 1772}
1429 1773
1774void incr_parse (JSON *self, SV *jsonstr = 0)
1775 PPCODE:
1776{
1777 if (!self->incr_text)
1778 self->incr_text = newSVpvn ("", 0);
1779
1780 // append data, if any
1781 if (jsonstr)
1782 {
1783 if (SvUTF8 (jsonstr) && !SvUTF8 (self->incr_text))
1784 {
1785 /* utf-8-ness differs, need to upgrade */
1786 sv_utf8_upgrade (self->incr_text);
1787
1788 if (self->incr_pos)
1789 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
1790 - (U8 *)SvPVX (self->incr_text);
1791 }
1792
1793 {
1794 STRLEN len;
1795 const char *str = SvPV (jsonstr, len);
1796 SvGROW (self->incr_text, SvCUR (self->incr_text) + len + 1);
1797 Move (str, SvEND (self->incr_text), len, char);
1798 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
1799 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
1800 }
1801 }
1802
1803 if (GIMME_V != G_VOID)
1804 do
1805 {
1806 STRLEN offset;
1807
1808 if (!INCR_DONE (self))
1809 {
1810 incr_parse (self);
1811
1812 if (self->incr_pos > self->max_size && self->max_size)
1813 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1814 (unsigned long)self->incr_pos, (unsigned long)self->max_size);
1815
1816 if (!INCR_DONE (self))
1817 break;
1818 }
1819
1820 XPUSHs (decode_json (self->incr_text, self, &offset));
1821
1822 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + offset);
1823 self->incr_pos -= offset;
1824 self->incr_nest = 0;
1825 self->incr_mode = 0;
1826 }
1827 while (GIMME_V == G_ARRAY);
1828}
1829
1830SV *incr_text (JSON *self)
1831 ATTRS: lvalue
1832 CODE:
1833{
1834 if (self->incr_pos)
1835 croak ("incr_text can not be called when the incremental parser already started parsing");
1836
1837 RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef;
1838}
1839 OUTPUT:
1840 RETVAL
1841
1842void incr_skip (JSON *self)
1843 CODE:
1844{
1845 if (self->incr_pos)
1846 {
1847 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + self->incr_pos);
1848 self->incr_pos = 0;
1849 self->incr_nest = 0;
1850 self->incr_mode = 0;
1851 }
1852}
1853
1854void incr_reset (JSON *self)
1855 CODE:
1856{
1857 SvREFCNT_dec (self->incr_text);
1858 self->incr_text = 0;
1859 self->incr_pos = 0;
1860 self->incr_nest = 0;
1861 self->incr_mode = 0;
1862}
1863
1864void DESTROY (JSON *self)
1865 CODE:
1866 SvREFCNT_dec (self->cb_sk_object);
1867 SvREFCNT_dec (self->cb_object);
1868 SvREFCNT_dec (self->incr_text);
1869
1430PROTOTYPES: ENABLE 1870PROTOTYPES: ENABLE
1431 1871
1432void to_json (SV *scalar) 1872void encode_json (SV *scalar)
1873 ALIAS:
1874 to_json_ = 0
1875 encode_json = F_UTF8
1433 PPCODE: 1876 PPCODE:
1434{ 1877{
1435 JSON json = { F_DEFAULT | F_UTF8 }; 1878 JSON json;
1879 json_init (&json);
1880 json.flags |= ix;
1436 XPUSHs (encode_json (scalar, &json)); 1881 XPUSHs (encode_json (scalar, &json));
1437} 1882}
1438 1883
1439void from_json (SV *jsonstr) 1884void decode_json (SV *jsonstr)
1885 ALIAS:
1886 from_json_ = 0
1887 decode_json = F_UTF8
1440 PPCODE: 1888 PPCODE:
1441{ 1889{
1442 JSON json = { F_DEFAULT | F_UTF8 }; 1890 JSON json;
1891 json_init (&json);
1892 json.flags |= ix;
1443 XPUSHs (decode_json (jsonstr, &json, 0)); 1893 XPUSHs (decode_json (jsonstr, &json, 0));
1444} 1894}
1445 1895
1896

Diff Legend

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