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 1.47 by root, Sun Jul 1 14:08:03 2007 UTC vs.
Revision 1.84 by root, Thu Mar 27 06:37:35 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
… …
25#define F_SPACE_BEFORE 0x00000020UL 29#define F_SPACE_BEFORE 0x00000020UL
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 // NYI 34#define F_CONV_BLESSED 0x00000800UL
35#define F_RELAXED 0x00001000UL
36#define F_ALLOW_UNKNOWN 0x00002000UL
37
31#define F_MAXDEPTH 0xf8000000UL 38#define F_MAXDEPTH 0xf8000000UL
32#define S_MAXDEPTH 27 39#define S_MAXDEPTH 27
33#define F_MAXSIZE 0x01f00000UL 40#define F_MAXSIZE 0x01f00000UL
34#define S_MAXSIZE 20 41#define S_MAXSIZE 20
42#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
35 43
36#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH)) 44#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
37#define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE )) 45#define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE ))
38 46
39#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 47#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
… …
46 54
47#define SB do { 55#define SB do {
48#define SE } while (0) 56#define SE } while (0)
49 57
50#if __GNUC__ >= 3 58#if __GNUC__ >= 3
51# define expect(expr,value) __builtin_expect ((expr),(value)) 59# define expect(expr,value) __builtin_expect ((expr), (value))
52# define inline inline 60# define INLINE static inline
53#else 61#else
54# define expect(expr,value) (expr) 62# define expect(expr,value) (expr)
55# define inline static 63# define INLINE static
56#endif 64#endif
57 65
58#define expect_false(expr) expect ((expr) != 0, 0) 66#define expect_false(expr) expect ((expr) != 0, 0)
59#define expect_true(expr) expect ((expr) != 0, 1) 67#define expect_true(expr) expect ((expr) != 0, 1)
60 68
69#define IN_RANGE_INC(type,val,beg,end) \
70 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \
71 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
72
73#ifdef USE_ITHREADS
74# define JSON_SLOW 1
75# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1))
76#else
77# define JSON_SLOW 0
78# define JSON_STASH json_stash
79#endif
80
61static HV *json_stash, *json_boolean_stash; // JSON::XS:: 81static HV *json_stash, *json_boolean_stash; // JSON::XS::
62static SV *json_true, *json_false; 82static SV *json_true, *json_false;
63 83
84enum {
85 INCR_M_WS = 0, // initial whitespace skipping, must be 0
86 INCR_M_STR, // inside string
87 INCR_M_BS, // inside backslash
88 INCR_M_JSON // outside anything, count nesting
89};
90
91#define INCR_DONE(json) (!(json)->incr_nest && (json)->incr_mode == INCR_M_JSON)
92
64typedef struct json { 93typedef struct {
65 U32 flags; 94 U32 flags;
95 SV *cb_object;
96 HV *cb_sk_object;
97
98 // for the incremental parser
99 SV *incr_text; // the source text so far
100 STRLEN incr_pos; // the current offset into the text
101 int incr_nest; // {[]}-nesting level
102 int incr_mode;
66} JSON__XS; 103} JSON;
67 104
68///////////////////////////////////////////////////////////////////////////// 105/////////////////////////////////////////////////////////////////////////////
69// utility functions 106// utility functions
70 107
71static UV * 108INLINE SV *
72SvJSON (SV *sv) 109get_bool (const char *name)
73{ 110{
74 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash)) 111 SV *sv = get_sv (name, 1);
75 croak ("object is not of type JSON::XS");
76 112
77 return &SvUVX (SvRV (sv)); 113 SvREADONLY_on (sv);
78} 114 SvREADONLY_on (SvRV (sv));
79 115
80static void 116 return sv;
117}
118
119INLINE void
81shrink (SV *sv) 120shrink (SV *sv)
82{ 121{
83 sv_utf8_downgrade (sv, 1); 122 sv_utf8_downgrade (sv, 1);
84 if (SvLEN (sv) > SvCUR (sv) + 1) 123 if (SvLEN (sv) > SvCUR (sv) + 1)
85 { 124 {
… …
94// decode an utf-8 character and return it, or (UV)-1 in 133// decode an utf-8 character and return it, or (UV)-1 in
95// case of an error. 134// case of an error.
96// we special-case "safe" characters from U+80 .. U+7FF, 135// we special-case "safe" characters from U+80 .. U+7FF,
97// but use the very good perl function to parse anything else. 136// but use the very good perl function to parse anything else.
98// note that we never call this function for a ascii codepoints 137// note that we never call this function for a ascii codepoints
99inline UV 138INLINE UV
100decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) 139decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
101{ 140{
102 if (expect_false (s[0] > 0xdf || s[0] < 0xc2)) 141 if (expect_true (len >= 2
103 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); 142 && IN_RANGE_INC (char, s[0], 0xc2, 0xdf)
104 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) 143 && IN_RANGE_INC (char, s[1], 0x80, 0xbf)))
105 { 144 {
106 *clen = 2; 145 *clen = 2;
107 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 146 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
108 } 147 }
109 else 148 else
110 { 149 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
111 *clen = (STRLEN)-1; 150}
112 return (UV)-1; 151
113 } 152// likewise for encoding, also never called for ascii codepoints
153// this function takes advantage of this fact, although current gccs
154// seem to optimise the check for >= 0x80 away anyways
155INLINE unsigned char *
156encode_utf8 (unsigned char *s, UV ch)
157{
158 if (expect_false (ch < 0x000080))
159 *s++ = ch;
160 else if (expect_true (ch < 0x000800))
161 *s++ = 0xc0 | ( ch >> 6),
162 *s++ = 0x80 | ( ch & 0x3f);
163 else if ( ch < 0x010000)
164 *s++ = 0xe0 | ( ch >> 12),
165 *s++ = 0x80 | ((ch >> 6) & 0x3f),
166 *s++ = 0x80 | ( ch & 0x3f);
167 else if ( ch < 0x110000)
168 *s++ = 0xf0 | ( ch >> 18),
169 *s++ = 0x80 | ((ch >> 12) & 0x3f),
170 *s++ = 0x80 | ((ch >> 6) & 0x3f),
171 *s++ = 0x80 | ( ch & 0x3f);
172
173 return s;
114} 174}
115 175
116///////////////////////////////////////////////////////////////////////////// 176/////////////////////////////////////////////////////////////////////////////
117// encoder 177// encoder
118 178
… …
120typedef struct 180typedef struct
121{ 181{
122 char *cur; // SvPVX (sv) + current output position 182 char *cur; // SvPVX (sv) + current output position
123 char *end; // SvEND (sv) 183 char *end; // SvEND (sv)
124 SV *sv; // result scalar 184 SV *sv; // result scalar
125 struct json json; 185 JSON json;
126 U32 indent; // indentation level 186 U32 indent; // indentation level
127 U32 maxdepth; // max. indentation/recursion level 187 U32 maxdepth; // max. indentation/recursion level
188 UV limit; // escape character values >= this value when encoding
128} enc_t; 189} enc_t;
129 190
130inline void 191INLINE void
131need (enc_t *enc, STRLEN len) 192need (enc_t *enc, STRLEN len)
132{ 193{
133 if (expect_false (enc->cur + len >= enc->end)) 194 if (expect_false (enc->cur + len >= enc->end))
134 { 195 {
135 STRLEN cur = enc->cur - SvPVX (enc->sv); 196 STRLEN cur = enc->cur - SvPVX (enc->sv);
… …
137 enc->cur = SvPVX (enc->sv) + cur; 198 enc->cur = SvPVX (enc->sv) + cur;
138 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 199 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
139 } 200 }
140} 201}
141 202
142inline void 203INLINE void
143encode_ch (enc_t *enc, char ch) 204encode_ch (enc_t *enc, char ch)
144{ 205{
145 need (enc, 1); 206 need (enc, 1);
146 *enc->cur++ = ch; 207 *enc->cur++ = ch;
147} 208}
… …
201 { 262 {
202 uch = ch; 263 uch = ch;
203 clen = 1; 264 clen = 1;
204 } 265 }
205 266
206 if (uch > 0x10FFFFUL) 267 if (uch < 0x80/*0x20*/ || uch >= enc->limit)
207 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
208
209 if (uch < 0x80 || enc->json.flags & F_ASCII || (enc->json.flags & F_LATIN1 && uch > 0xFF))
210 { 268 {
211 if (uch > 0xFFFFUL) 269 if (uch >= 0x10000UL)
212 { 270 {
271 if (uch >= 0x110000UL)
272 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
273
213 need (enc, len += 11); 274 need (enc, len += 11);
214 sprintf (enc->cur, "\\u%04x\\u%04x", 275 sprintf (enc->cur, "\\u%04x\\u%04x",
215 (int)((uch - 0x10000) / 0x400 + 0xD800), 276 (int)((uch - 0x10000) / 0x400 + 0xD800),
216 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 277 (int)((uch - 0x10000) % 0x400 + 0xDC00));
217 enc->cur += 12; 278 enc->cur += 12;
… …
245 while (--clen); 306 while (--clen);
246 } 307 }
247 else 308 else
248 { 309 {
249 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed 310 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
250 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 311 enc->cur = encode_utf8 (enc->cur, uch);
251 ++str; 312 ++str;
252 } 313 }
253 } 314 }
254 } 315 }
255 } 316 }
256 317
257 --len; 318 --len;
258 } 319 }
259} 320}
260 321
261inline void 322INLINE void
262encode_indent (enc_t *enc) 323encode_indent (enc_t *enc)
263{ 324{
264 if (enc->json.flags & F_INDENT) 325 if (enc->json.flags & F_INDENT)
265 { 326 {
266 int spaces = enc->indent * INDENT_STEP; 327 int spaces = enc->indent * INDENT_STEP;
… …
269 memset (enc->cur, ' ', spaces); 330 memset (enc->cur, ' ', spaces);
270 enc->cur += spaces; 331 enc->cur += spaces;
271 } 332 }
272} 333}
273 334
274inline void 335INLINE void
275encode_space (enc_t *enc) 336encode_space (enc_t *enc)
276{ 337{
277 need (enc, 1); 338 need (enc, 1);
278 encode_ch (enc, ' '); 339 encode_ch (enc, ' ');
279} 340}
280 341
281inline void 342INLINE void
282encode_nl (enc_t *enc) 343encode_nl (enc_t *enc)
283{ 344{
284 if (enc->json.flags & F_INDENT) 345 if (enc->json.flags & F_INDENT)
285 { 346 {
286 need (enc, 1); 347 need (enc, 1);
287 encode_ch (enc, '\n'); 348 encode_ch (enc, '\n');
288 } 349 }
289} 350}
290 351
291inline void 352INLINE void
292encode_comma (enc_t *enc) 353encode_comma (enc_t *enc)
293{ 354{
294 encode_ch (enc, ','); 355 encode_ch (enc, ',');
295 356
296 if (enc->json.flags & F_INDENT) 357 if (enc->json.flags & F_INDENT)
… …
307 int i, len = av_len (av); 368 int i, len = av_len (av);
308 369
309 if (enc->indent >= enc->maxdepth) 370 if (enc->indent >= enc->maxdepth)
310 croak ("data structure too deep (hit recursion limit)"); 371 croak ("data structure too deep (hit recursion limit)");
311 372
312 encode_ch (enc, '['); encode_nl (enc); 373 encode_ch (enc, '[');
313 ++enc->indent; 374
375 if (len >= 0)
376 {
377 encode_nl (enc); ++enc->indent;
314 378
315 for (i = 0; i <= len; ++i) 379 for (i = 0; i <= len; ++i)
316 { 380 {
381 SV **svp = av_fetch (av, i, 0);
382
317 encode_indent (enc); 383 encode_indent (enc);
318 encode_sv (enc, *av_fetch (av, i, 0));
319 384
385 if (svp)
386 encode_sv (enc, *svp);
387 else
388 encode_str (enc, "null", 4, 0);
389
320 if (i < len) 390 if (i < len)
321 encode_comma (enc); 391 encode_comma (enc);
322 } 392 }
323 393
394 encode_nl (enc); --enc->indent; encode_indent (enc);
395 }
396
324 encode_nl (enc); 397 encode_ch (enc, ']');
325
326 --enc->indent;
327 encode_indent (enc); encode_ch (enc, ']');
328} 398}
329 399
330static void 400static void
331encode_he (enc_t *enc, HE *he) 401encode_hk (enc_t *enc, HE *he)
332{ 402{
333 encode_ch (enc, '"'); 403 encode_ch (enc, '"');
334 404
335 if (HeKLEN (he) == HEf_SVKEY) 405 if (HeKLEN (he) == HEf_SVKEY)
336 { 406 {
… …
349 encode_ch (enc, '"'); 419 encode_ch (enc, '"');
350 420
351 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc); 421 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
352 encode_ch (enc, ':'); 422 encode_ch (enc, ':');
353 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc); 423 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
354 encode_sv (enc, HeVAL (he));
355} 424}
356 425
357// compare hash entries, used when all keys are bytestrings 426// compare hash entries, used when all keys are bytestrings
358static int 427static int
359he_cmp_fast (const void *a_, const void *b_) 428he_cmp_fast (const void *a_, const void *b_)
… …
364 HE *b = *(HE **)b_; 433 HE *b = *(HE **)b_;
365 434
366 STRLEN la = HeKLEN (a); 435 STRLEN la = HeKLEN (a);
367 STRLEN lb = HeKLEN (b); 436 STRLEN lb = HeKLEN (b);
368 437
369 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 438 if (!(cmp = memcmp (HeKEY (b), HeKEY (a), lb < la ? lb : la)))
370 cmp = la - lb; 439 cmp = lb - la;
371 440
372 return cmp; 441 return cmp;
373} 442}
374 443
375// compare hash entries, used when some keys are sv's or utf-x 444// compare hash entries, used when some keys are sv's or utf-x
376static int 445static int
377he_cmp_slow (const void *a, const void *b) 446he_cmp_slow (const void *a, const void *b)
378{ 447{
379 return sv_cmp (HeSVKEY_force (*(HE **)a), HeSVKEY_force (*(HE **)b)); 448 return sv_cmp (HeSVKEY_force (*(HE **)b), HeSVKEY_force (*(HE **)a));
380} 449}
381 450
382static void 451static void
383encode_hv (enc_t *enc, HV *hv) 452encode_hv (enc_t *enc, HV *hv)
384{ 453{
385 int count, i; 454 HE *he;
386 455
387 if (enc->indent >= enc->maxdepth) 456 if (enc->indent >= enc->maxdepth)
388 croak ("data structure too deep (hit recursion limit)"); 457 croak ("data structure too deep (hit recursion limit)");
389 458
390 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 459 encode_ch (enc, '{');
391 460
392 if ((count = hv_iterinit (hv)))
393 {
394 // for canonical output we have to sort by keys first 461 // for canonical output we have to sort by keys first
395 // actually, this is mostly due to the stupid so-called 462 // actually, this is mostly due to the stupid so-called
396 // security workaround added somewhere in 5.8.x. 463 // security workaround added somewhere in 5.8.x.
397 // that randomises hash orderings 464 // that randomises hash orderings
398 if (enc->json.flags & F_CANONICAL) 465 if (enc->json.flags & F_CANONICAL)
466 {
467 int count = hv_iterinit (hv);
468
469 if (SvMAGICAL (hv))
399 { 470 {
471 // need to count by iterating. could improve by dynamically building the vector below
472 // but I don't care for the speed of this special case.
473 // note also that we will run into undefined behaviour when the two iterations
474 // do not result in the same count, something I might care for in some later release.
475
476 count = 0;
477 while (hv_iternext (hv))
478 ++count;
479
480 hv_iterinit (hv);
481 }
482
483 if (count)
484 {
400 int fast = 1; 485 int i, fast = 1;
401 HE *he;
402#if defined(__BORLANDC__) || defined(_MSC_VER) 486#if defined(__BORLANDC__) || defined(_MSC_VER)
403 HE **hes = _alloca (count * sizeof (HE)); 487 HE **hes = _alloca (count * sizeof (HE));
404#else 488#else
405 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode 489 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
406#endif 490#endif
… …
433 517
434 FREETMPS; 518 FREETMPS;
435 LEAVE; 519 LEAVE;
436 } 520 }
437 521
438 for (i = 0; i < count; ++i) 522 encode_nl (enc); ++enc->indent;
523
524 while (count--)
439 { 525 {
440 encode_indent (enc); 526 encode_indent (enc);
527 he = hes [count];
441 encode_he (enc, hes [i]); 528 encode_hk (enc, he);
529 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
442 530
443 if (i < count - 1) 531 if (count)
444 encode_comma (enc); 532 encode_comma (enc);
445 } 533 }
446 534
447 encode_nl (enc); 535 encode_nl (enc); --enc->indent; encode_indent (enc);
448 } 536 }
537 }
449 else 538 else
539 {
540 if (hv_iterinit (hv) || SvMAGICAL (hv))
541 if ((he = hv_iternext (hv)))
450 { 542 {
451 HE *he = hv_iternext (hv); 543 encode_nl (enc); ++enc->indent;
452 544
453 for (;;) 545 for (;;)
454 { 546 {
455 encode_indent (enc); 547 encode_indent (enc);
456 encode_he (enc, he); 548 encode_hk (enc, he);
549 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
457 550
458 if (!(he = hv_iternext (hv))) 551 if (!(he = hv_iternext (hv)))
459 break; 552 break;
460 553
461 encode_comma (enc); 554 encode_comma (enc);
462 } 555 }
463 556
464 encode_nl (enc); 557 encode_nl (enc); --enc->indent; encode_indent (enc);
465 } 558 }
466 } 559 }
467 560
468 --enc->indent; encode_indent (enc); encode_ch (enc, '}'); 561 encode_ch (enc, '}');
469} 562}
470 563
471// encode objects, arrays and special \0=false and \1=true values. 564// encode objects, arrays and special \0=false and \1=true values.
472static void 565static void
473encode_rv (enc_t *enc, SV *sv) 566encode_rv (enc_t *enc, SV *sv)
… …
477 SvGETMAGIC (sv); 570 SvGETMAGIC (sv);
478 svt = SvTYPE (sv); 571 svt = SvTYPE (sv);
479 572
480 if (expect_false (SvOBJECT (sv))) 573 if (expect_false (SvOBJECT (sv)))
481 { 574 {
575 HV *stash = !JSON_SLOW || json_boolean_stash
576 ? json_boolean_stash
577 : gv_stashpv ("JSON::XS::Boolean", 1);
578
482 if (SvSTASH (sv) == json_boolean_stash) 579 if (SvSTASH (sv) == stash)
483 { 580 {
484 if (SvIV (sv) == 0) 581 if (SvIV (sv))
582 encode_str (enc, "true", 4, 0);
583 else
485 encode_str (enc, "false", 5, 0); 584 encode_str (enc, "false", 5, 0);
486 else
487 encode_str (enc, "true", 4, 0);
488 } 585 }
489 else 586 else
490 { 587 {
491#if 0 588#if 0
492 if (0 && sv_derived_from (rv, "JSON::Literal")) 589 if (0 && sv_derived_from (rv, "JSON::Literal"))
… …
495 } 592 }
496#endif 593#endif
497 if (enc->json.flags & F_CONV_BLESSED) 594 if (enc->json.flags & F_CONV_BLESSED)
498 { 595 {
499 // we re-bless the reference to get overload and other niceties right 596 // we re-bless the reference to get overload and other niceties right
500 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 1); 597 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
501 598
502 if (to_json) 599 if (to_json)
503 { 600 {
504 dSP; 601 dSP;
505 ENTER; 602
506 SAVETMPS;
507 PUSHMARK (SP); 603 ENTER; SAVETMPS; PUSHMARK (SP);
508 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv))); 604 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
509 605
510 // calling with G_SCALAR ensures that we always get a 1 reutrn value 606 // calling with G_SCALAR ensures that we always get a 1 return value
511 // check anyways.
512 PUTBACK; 607 PUTBACK;
513 assert (1 == call_sv ((SV *)GvCV (to_json), G_SCALAR)); 608 call_sv ((SV *)GvCV (to_json), G_SCALAR);
514 SPAGAIN; 609 SPAGAIN;
515 610
611 // catch this surprisingly common error
612 if (SvROK (TOPs) && SvRV (TOPs) == sv)
613 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
614
615 sv = POPs;
616 PUTBACK;
617
516 encode_sv (enc, POPs); 618 encode_sv (enc, sv);
517 619
518 FREETMPS; 620 FREETMPS; LEAVE;
519 LEAVE;
520 } 621 }
521 else if (enc->json.flags & F_ALLOW_BLESSED) 622 else if (enc->json.flags & F_ALLOW_BLESSED)
522 encode_str (enc, "null", 4, 0); 623 encode_str (enc, "null", 4, 0);
523 else 624 else
524 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it", 625 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
… …
535 encode_hv (enc, (HV *)sv); 636 encode_hv (enc, (HV *)sv);
536 else if (svt == SVt_PVAV) 637 else if (svt == SVt_PVAV)
537 encode_av (enc, (AV *)sv); 638 encode_av (enc, (AV *)sv);
538 else if (svt < SVt_PVAV) 639 else if (svt < SVt_PVAV)
539 { 640 {
540 if (SvNIOK (sv) && SvIV (sv) == 0) 641 STRLEN len = 0;
642 char *pv = svt ? SvPV (sv, len) : 0;
643
644 if (len == 1 && *pv == '1')
645 encode_str (enc, "true", 4, 0);
646 else if (len == 1 && *pv == '0')
541 encode_str (enc, "false", 5, 0); 647 encode_str (enc, "false", 5, 0);
542 else if (SvNIOK (sv) && SvIV (sv) == 1) 648 else if (enc->json.flags & F_ALLOW_UNKNOWN)
543 encode_str (enc, "true", 4, 0); 649 encode_str (enc, "null", 4, 0);
544 else 650 else
545 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", 651 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
546 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 652 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
547 } 653 }
654 else if (enc->json.flags & F_ALLOW_UNKNOWN)
655 encode_str (enc, "null", 4, 0);
548 else 656 else
549 croak ("encountered %s, but JSON can only represent references to arrays or hashes", 657 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
550 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 658 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
551} 659}
552 660
… …
570 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 678 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
571 enc->cur += strlen (enc->cur); 679 enc->cur += strlen (enc->cur);
572 } 680 }
573 else if (SvIOKp (sv)) 681 else if (SvIOKp (sv))
574 { 682 {
575 // we assume we can always read an IV as a UV 683 // we assume we can always read an IV as a UV and vice versa
576 if (SvUV (sv) & ~(UV)0x7fff) 684 // we assume two's complement
577 { 685 // we assume no aliasing issues in the union
578 // large integer, use the (rather slow) snprintf way. 686 if (SvIsUV (sv) ? SvUVX (sv) <= 59000
579 need (enc, sizeof (UV) * 3); 687 : SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000)
580 enc->cur +=
581 SvIsUV(sv)
582 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
583 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
584 }
585 else
586 { 688 {
587 // optimise the "small number case" 689 // optimise the "small number case"
588 // code will likely be branchless and use only a single multiplication 690 // code will likely be branchless and use only a single multiplication
691 // works for numbers up to 59074
589 I32 i = SvIV (sv); 692 I32 i = SvIVX (sv);
590 U32 u; 693 U32 u;
591 char digit, nz = 0; 694 char digit, nz = 0;
592 695
593 need (enc, 6); 696 need (enc, 6);
594 697
… …
600 703
601 // now output digit by digit, each time masking out the integer part 704 // now output digit by digit, each time masking out the integer part
602 // and multiplying by 5 while moving the decimal point one to the right, 705 // and multiplying by 5 while moving the decimal point one to the right,
603 // resulting in a net multiplication by 10. 706 // resulting in a net multiplication by 10.
604 // we always write the digit to memory but conditionally increment 707 // we always write the digit to memory but conditionally increment
605 // the pointer, to ease the usage of conditional move instructions. 708 // the pointer, to enable the use of conditional move instructions.
606 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5; 709 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffffUL) * 5;
607 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5; 710 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffffUL) * 5;
608 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5; 711 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffffUL) * 5;
609 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5; 712 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffffUL) * 5;
610 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0' 713 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
611 } 714 }
715 else
716 {
717 // large integer, use the (rather slow) snprintf way.
718 need (enc, IVUV_MAXCHARS);
719 enc->cur +=
720 SvIsUV(sv)
721 ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))
722 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
723 }
612 } 724 }
613 else if (SvROK (sv)) 725 else if (SvROK (sv))
614 encode_rv (enc, SvRV (sv)); 726 encode_rv (enc, SvRV (sv));
615 else if (!SvOK (sv)) 727 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN)
616 encode_str (enc, "null", 4, 0); 728 encode_str (enc, "null", 4, 0);
617 else 729 else
618 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 730 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
619 SvPV_nolen (sv), SvFLAGS (sv)); 731 SvPV_nolen (sv), SvFLAGS (sv));
620} 732}
621 733
622static SV * 734static SV *
623encode_json (SV *scalar, struct json *json) 735encode_json (SV *scalar, JSON *json)
624{ 736{
625 enc_t enc; 737 enc_t enc;
626 738
627 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar)) 739 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
628 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 740 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
… …
631 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 743 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
632 enc.cur = SvPVX (enc.sv); 744 enc.cur = SvPVX (enc.sv);
633 enc.end = SvEND (enc.sv); 745 enc.end = SvEND (enc.sv);
634 enc.indent = 0; 746 enc.indent = 0;
635 enc.maxdepth = DEC_DEPTH (enc.json.flags); 747 enc.maxdepth = DEC_DEPTH (enc.json.flags);
748 enc.limit = enc.json.flags & F_ASCII ? 0x000080UL
749 : enc.json.flags & F_LATIN1 ? 0x000100UL
750 : 0x110000UL;
636 751
637 SvPOK_only (enc.sv); 752 SvPOK_only (enc.sv);
638 encode_sv (&enc, scalar); 753 encode_sv (&enc, scalar);
639 754
640 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 755 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
… …
656typedef struct 771typedef struct
657{ 772{
658 char *cur; // current parser pointer 773 char *cur; // current parser pointer
659 char *end; // end of input string 774 char *end; // end of input string
660 const char *err; // parse error, if != 0 775 const char *err; // parse error, if != 0
661 struct json json; 776 JSON json;
662 U32 depth; // recursion depth 777 U32 depth; // recursion depth
663 U32 maxdepth; // recursion depth limit 778 U32 maxdepth; // recursion depth limit
664} dec_t; 779} dec_t;
665 780
666inline void 781INLINE void
782decode_comment (dec_t *dec)
783{
784 // only '#'-style comments allowed a.t.m.
785
786 while (*dec->cur && *dec->cur != 0x0a && *dec->cur != 0x0d)
787 ++dec->cur;
788}
789
790INLINE void
667decode_ws (dec_t *dec) 791decode_ws (dec_t *dec)
668{ 792{
669 for (;;) 793 for (;;)
670 { 794 {
671 char ch = *dec->cur; 795 char ch = *dec->cur;
672 796
673 if (ch > 0x20 797 if (ch > 0x20)
798 {
799 if (expect_false (ch == '#'))
800 {
801 if (dec->json.flags & F_RELAXED)
802 decode_comment (dec);
803 else
804 break;
805 }
806 else
807 break;
808 }
674 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) 809 else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)
675 break; 810 break; // parse error, but let higher level handle it, gives better error messages
676 811
677 ++dec->cur; 812 ++dec->cur;
678 } 813 }
679} 814}
680 815
… …
786 921
787 if (hi >= 0x80) 922 if (hi >= 0x80)
788 { 923 {
789 utf8 = 1; 924 utf8 = 1;
790 925
791 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0); 926 cur = encode_utf8 (cur, hi);
792 } 927 }
793 else 928 else
794 *cur++ = hi; 929 *cur++ = hi;
795 } 930 }
796 break; 931 break;
… …
798 default: 933 default:
799 --dec_cur; 934 --dec_cur;
800 ERR ("illegal backslash escape sequence in string"); 935 ERR ("illegal backslash escape sequence in string");
801 } 936 }
802 } 937 }
803 else if (expect_true (ch >= 0x20 && ch <= 0x7f)) 938 else if (expect_true (ch >= 0x20 && ch < 0x80))
804 *cur++ = ch; 939 *cur++ = ch;
805 else if (ch >= 0x80) 940 else if (ch >= 0x80)
806 { 941 {
807 STRLEN clen; 942 STRLEN clen;
808 UV uch; 943 UV uch;
… …
929 is_nv = 1; 1064 is_nv = 1;
930 } 1065 }
931 1066
932 if (!is_nv) 1067 if (!is_nv)
933 { 1068 {
1069 int len = dec->cur - start;
1070
934 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so 1071 // special case the rather common 1..5-digit-int case
935 if (*start == '-') 1072 if (*start == '-')
936 switch (dec->cur - start) 1073 switch (len)
937 { 1074 {
938 case 2: return newSViv (-( start [1] - '0' * 1)); 1075 case 2: return newSViv (-( start [1] - '0' * 1));
939 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); 1076 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
940 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); 1077 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
941 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); 1078 case 5: return newSViv (-( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
1079 case 6: return newSViv (-(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
942 } 1080 }
943 else 1081 else
944 switch (dec->cur - start) 1082 switch (len)
945 { 1083 {
946 case 1: return newSViv ( start [0] - '0' * 1); 1084 case 1: return newSViv ( start [0] - '0' * 1);
947 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); 1085 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
948 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); 1086 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
949 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); 1087 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
1088 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
950 } 1089 }
951 1090
952 { 1091 {
953 UV uv; 1092 UV uv;
954 int numtype = grok_number (start, dec->cur - start, &uv); 1093 int numtype = grok_number (start, len, &uv);
955 if (numtype & IS_NUMBER_IN_UV) 1094 if (numtype & IS_NUMBER_IN_UV)
956 if (numtype & IS_NUMBER_NEG) 1095 if (numtype & IS_NUMBER_NEG)
957 { 1096 {
958 if (uv < (UV)IV_MIN) 1097 if (uv < (UV)IV_MIN)
959 return newSViv (-(IV)uv); 1098 return newSViv (-(IV)uv);
960 } 1099 }
961 else 1100 else
962 return newSVuv (uv); 1101 return newSVuv (uv);
963
964 // here would likely be the place for bigint support
965 } 1102 }
966 }
967 1103
968 // if we ever support bigint or bigfloat, this is the place for bigfloat 1104 len -= *start == '-' ? 1 : 0;
1105
1106 // does not fit into IV or UV, try NV
1107 if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len)
1108 #if defined (LDBL_DIG)
1109 || (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len)
1110 #endif
1111 )
1112 // fits into NV without loss of precision
1113 return newSVnv (Atof (start));
1114
1115 // everything else fails, convert it to a string
1116 return newSVpvn (start, dec->cur - start);
1117 }
1118
1119 // loss of precision here
969 return newSVnv (Atof (start)); 1120 return newSVnv (Atof (start));
970 1121
971fail: 1122fail:
972 return 0; 1123 return 0;
973} 1124}
… …
1003 1154
1004 if (*dec->cur != ',') 1155 if (*dec->cur != ',')
1005 ERR (", or ] expected while parsing array"); 1156 ERR (", or ] expected while parsing array");
1006 1157
1007 ++dec->cur; 1158 ++dec->cur;
1159
1160 decode_ws (dec);
1161
1162 if (*dec->cur == ']' && dec->json.flags & F_RELAXED)
1163 {
1164 ++dec->cur;
1165 break;
1166 }
1008 } 1167 }
1009 1168
1010 DEC_DEC_DEPTH; 1169 DEC_DEC_DEPTH;
1011 return newRV_noinc ((SV *)av); 1170 return newRV_noinc ((SV *)av);
1012 1171
… …
1017} 1176}
1018 1177
1019static SV * 1178static SV *
1020decode_hv (dec_t *dec) 1179decode_hv (dec_t *dec)
1021{ 1180{
1181 SV *sv;
1022 HV *hv = newHV (); 1182 HV *hv = newHV ();
1023 1183
1024 DEC_INC_DEPTH; 1184 DEC_INC_DEPTH;
1025 decode_ws (dec); 1185 decode_ws (dec);
1026 1186
1027 if (*dec->cur == '}') 1187 if (*dec->cur == '}')
1028 ++dec->cur; 1188 ++dec->cur;
1029 else 1189 else
1030 for (;;) 1190 for (;;)
1031 { 1191 {
1032 decode_ws (dec); EXPECT_CH ('"'); 1192 EXPECT_CH ('"');
1033 1193
1034 // heuristic: assume that 1194 // heuristic: assume that
1035 // a) decode_str + hv_store_ent are abysmally slow. 1195 // a) decode_str + hv_store_ent are abysmally slow.
1036 // b) most hash keys are short, simple ascii text. 1196 // b) most hash keys are short, simple ascii text.
1037 // => try to "fast-match" such strings to avoid 1197 // => try to "fast-match" such strings to avoid
… …
1041 char *p = dec->cur; 1201 char *p = dec->cur;
1042 char *e = p + 24; // only try up to 24 bytes 1202 char *e = p + 24; // only try up to 24 bytes
1043 1203
1044 for (;;) 1204 for (;;)
1045 { 1205 {
1046 // the >= 0x80 is true on most architectures 1206 // the >= 0x80 is false on most architectures
1047 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\') 1207 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1048 { 1208 {
1049 // slow path, back up and use decode_str 1209 // slow path, back up and use decode_str
1050 SV *key = decode_str (dec); 1210 SV *key = decode_str (dec);
1051 if (!key) 1211 if (!key)
1052 goto fail; 1212 goto fail;
1053 1213
1054 decode_ws (dec); EXPECT_CH (':'); 1214 decode_ws (dec); EXPECT_CH (':');
1055 1215
1216 decode_ws (dec);
1056 value = decode_sv (dec); 1217 value = decode_sv (dec);
1057 if (!value) 1218 if (!value)
1058 { 1219 {
1059 SvREFCNT_dec (key); 1220 SvREFCNT_dec (key);
1060 goto fail; 1221 goto fail;
… …
1072 int len = p - key; 1233 int len = p - key;
1073 dec->cur = p + 1; 1234 dec->cur = p + 1;
1074 1235
1075 decode_ws (dec); EXPECT_CH (':'); 1236 decode_ws (dec); EXPECT_CH (':');
1076 1237
1238 decode_ws (dec);
1077 value = decode_sv (dec); 1239 value = decode_sv (dec);
1078 if (!value) 1240 if (!value)
1079 goto fail; 1241 goto fail;
1080 1242
1081 hv_store (hv, key, len, value, 0); 1243 hv_store (hv, key, len, value, 0);
… …
1097 1259
1098 if (*dec->cur != ',') 1260 if (*dec->cur != ',')
1099 ERR (", or } expected while parsing object/hash"); 1261 ERR (", or } expected while parsing object/hash");
1100 1262
1101 ++dec->cur; 1263 ++dec->cur;
1264
1265 decode_ws (dec);
1266
1267 if (*dec->cur == '}' && dec->json.flags & F_RELAXED)
1268 {
1269 ++dec->cur;
1270 break;
1271 }
1102 } 1272 }
1103 1273
1104 DEC_DEC_DEPTH; 1274 DEC_DEC_DEPTH;
1105 return newRV_noinc ((SV *)hv); 1275 sv = newRV_noinc ((SV *)hv);
1276
1277 // check filter callbacks
1278 if (dec->json.flags & F_HOOK)
1279 {
1280 if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
1281 {
1282 HE *cb, *he;
1283
1284 hv_iterinit (hv);
1285 he = hv_iternext (hv);
1286 hv_iterinit (hv);
1287
1288 // the next line creates a mortal sv each time its called.
1289 // might want to optimise this for common cases.
1290 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1291
1292 if (cb)
1293 {
1294 dSP;
1295 int count;
1296
1297 ENTER; SAVETMPS; PUSHMARK (SP);
1298 XPUSHs (HeVAL (he));
1299
1300 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1301
1302 if (count == 1)
1303 {
1304 sv = newSVsv (POPs);
1305 FREETMPS; LEAVE;
1306 return sv;
1307 }
1308
1309 FREETMPS; LEAVE;
1310 }
1311 }
1312
1313 if (dec->json.cb_object)
1314 {
1315 dSP;
1316 int count;
1317
1318 ENTER; SAVETMPS; PUSHMARK (SP);
1319 XPUSHs (sv_2mortal (sv));
1320
1321 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1322
1323 if (count == 1)
1324 {
1325 sv = newSVsv (POPs);
1326 FREETMPS; LEAVE;
1327 return sv;
1328 }
1329
1330 SvREFCNT_inc (sv);
1331 FREETMPS; LEAVE;
1332 }
1333 }
1334
1335 return sv;
1106 1336
1107fail: 1337fail:
1108 SvREFCNT_dec (hv); 1338 SvREFCNT_dec (hv);
1109 DEC_DEC_DEPTH; 1339 DEC_DEC_DEPTH;
1110 return 0; 1340 return 0;
1111} 1341}
1112 1342
1113static SV * 1343static SV *
1114decode_sv (dec_t *dec) 1344decode_sv (dec_t *dec)
1115{ 1345{
1116 decode_ws (dec);
1117
1118 // the beauty of JSON: you need exactly one character lookahead 1346 // the beauty of JSON: you need exactly one character lookahead
1119 // to parse anything. 1347 // to parse everything.
1120 switch (*dec->cur) 1348 switch (*dec->cur)
1121 { 1349 {
1122 case '"': ++dec->cur; return decode_str (dec); 1350 case '"': ++dec->cur; return decode_str (dec);
1123 case '[': ++dec->cur; return decode_av (dec); 1351 case '[': ++dec->cur; return decode_av (dec);
1124 case '{': ++dec->cur; return decode_hv (dec); 1352 case '{': ++dec->cur; return decode_hv (dec);
1125 1353
1126 case '-': 1354 case '-':
1127 case '0': case '1': case '2': case '3': case '4': 1355 case '0': case '1': case '2': case '3': case '4':
1128 case '5': case '6': case '7': case '8': case '9': 1356 case '5': case '6': case '7': case '8': case '9':
1129 return decode_num (dec); 1357 return decode_num (dec);
1130 1358
1131 case 't': 1359 case 't':
1132 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1360 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1133 { 1361 {
1134 dec->cur += 4; 1362 dec->cur += 4;
1363#if JSON_SLOW
1364 json_true = get_bool ("JSON::XS::true");
1365#endif
1135 return SvREFCNT_inc (json_true); 1366 return newSVsv (json_true);
1136 } 1367 }
1137 else 1368 else
1138 ERR ("'true' expected"); 1369 ERR ("'true' expected");
1139 1370
1140 break; 1371 break;
1141 1372
1142 case 'f': 1373 case 'f':
1143 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1374 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1144 { 1375 {
1145 dec->cur += 5; 1376 dec->cur += 5;
1377#if JSON_SLOW
1378 json_false = get_bool ("JSON::XS::false");
1379#endif
1146 return SvREFCNT_inc (json_false); 1380 return newSVsv (json_false);
1147 } 1381 }
1148 else 1382 else
1149 ERR ("'false' expected"); 1383 ERR ("'false' expected");
1150 1384
1151 break; 1385 break;
… …
1169fail: 1403fail:
1170 return 0; 1404 return 0;
1171} 1405}
1172 1406
1173static SV * 1407static SV *
1174decode_json (SV *string, struct json *json, UV *offset_return) 1408decode_json (SV *string, JSON *json, STRLEN *offset_return)
1175{ 1409{
1176 dec_t dec; 1410 dec_t dec;
1177 UV offset; 1411 STRLEN offset;
1178 SV *sv; 1412 SV *sv;
1179 1413
1180 SvGETMAGIC (string); 1414 SvGETMAGIC (string);
1181 SvUPGRADE (string, SVt_PV); 1415 SvUPGRADE (string, SVt_PV);
1182 1416
… …
1196 dec.end = SvEND (string); 1430 dec.end = SvEND (string);
1197 dec.err = 0; 1431 dec.err = 0;
1198 dec.depth = 0; 1432 dec.depth = 0;
1199 dec.maxdepth = DEC_DEPTH (dec.json.flags); 1433 dec.maxdepth = DEC_DEPTH (dec.json.flags);
1200 1434
1435 if (dec.json.cb_object || dec.json.cb_sk_object)
1436 dec.json.flags |= F_HOOK;
1437
1201 *dec.end = 0; // this should basically be a nop, too, but make sure it's there 1438 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1439
1440 decode_ws (&dec);
1202 sv = decode_sv (&dec); 1441 sv = decode_sv (&dec);
1203 1442
1204 if (!(offset_return || !sv)) 1443 if (!(offset_return || !sv))
1205 { 1444 {
1206 // check for trailing garbage 1445 // check for trailing garbage
… …
1250 1489
1251 return sv; 1490 return sv;
1252} 1491}
1253 1492
1254///////////////////////////////////////////////////////////////////////////// 1493/////////////////////////////////////////////////////////////////////////////
1494// incremental parser
1495
1496static void
1497incr_parse (JSON *self)
1498{
1499 const char *p = SvPVX (self->incr_text) + self->incr_pos;
1500
1501 for (;;)
1502 {
1503 //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
1504 switch (self->incr_mode)
1505 {
1506 // only used for intiial whitespace skipping
1507 case INCR_M_WS:
1508 for (;;)
1509 {
1510 if (*p > 0x20)
1511 {
1512 self->incr_mode = INCR_M_JSON;
1513 goto incr_m_json;
1514 }
1515 else if (!*p)
1516 goto interrupt;
1517
1518 ++p;
1519 }
1520
1521 // skip a single char inside a string (for \\-processing)
1522 case INCR_M_BS:
1523 if (!*p)
1524 goto interrupt;
1525
1526 ++p;
1527 self->incr_mode = INCR_M_STR;
1528 goto incr_m_str;
1529
1530 // inside a string
1531 case INCR_M_STR:
1532 incr_m_str:
1533 for (;;)
1534 {
1535 if (*p == '"')
1536 {
1537 ++p;
1538 self->incr_mode = INCR_M_JSON;
1539
1540 if (!self->incr_nest)
1541 goto interrupt;
1542
1543 goto incr_m_json;
1544 }
1545 else if (*p == '\\')
1546 {
1547 ++p; // "virtually" consumes character after \
1548
1549 if (!*p) // if at end of string we have to switch modes
1550 {
1551 self->incr_mode = INCR_M_BS;
1552 goto interrupt;
1553 }
1554 }
1555 else if (!*p)
1556 goto interrupt;
1557
1558 ++p;
1559 }
1560
1561 // after initial ws, outside string
1562 case INCR_M_JSON:
1563 incr_m_json:
1564 for (;;)
1565 {
1566 switch (*p++)
1567 {
1568 case 0:
1569 --p;
1570 goto interrupt;
1571
1572 case 0x09:
1573 case 0x0a:
1574 case 0x0d:
1575 case 0x20:
1576 if (!self->incr_nest)
1577 {
1578 --p; // do not eat the whitespace, let the next round do it
1579 goto interrupt;
1580 }
1581 break;
1582
1583 case '"':
1584 self->incr_mode = INCR_M_STR;
1585 goto incr_m_str;
1586
1587 case '[':
1588 case '{':
1589 ++self->incr_nest;
1590 break;
1591
1592 case ']':
1593 case '}':
1594 if (!--self->incr_nest)
1595 goto interrupt;
1596 }
1597 }
1598 }
1599
1600 modechange:
1601 ;
1602 }
1603
1604interrupt:
1605 self->incr_pos = p - SvPVX (self->incr_text);
1606 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D
1607}
1608
1609/////////////////////////////////////////////////////////////////////////////
1255// XS interface functions 1610// XS interface functions
1256 1611
1257MODULE = JSON::XS PACKAGE = JSON::XS 1612MODULE = JSON::XS PACKAGE = JSON::XS
1258 1613
1259BOOT: 1614BOOT:
… …
1268 : -1; 1623 : -1;
1269 1624
1270 json_stash = gv_stashpv ("JSON::XS" , 1); 1625 json_stash = gv_stashpv ("JSON::XS" , 1);
1271 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); 1626 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1272 1627
1273 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true ); 1628 json_true = get_bool ("JSON::XS::true");
1274 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); 1629 json_false = get_bool ("JSON::XS::false");
1275} 1630}
1276 1631
1277PROTOTYPES: DISABLE 1632PROTOTYPES: DISABLE
1278 1633
1279SV *new (char *dummy) 1634void CLONE (...)
1280 CODE: 1635 CODE:
1281 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1636 json_stash = 0;
1282 OUTPUT: 1637 json_boolean_stash = 0;
1283 RETVAL
1284 1638
1639void new (char *klass)
1640 PPCODE:
1641{
1642 SV *pv = NEWSV (0, sizeof (JSON));
1643 SvPOK_only (pv);
1644 Zero (SvPVX (pv), 1, JSON);
1645 ((JSON *)SvPVX (pv))->flags = F_DEFAULT;
1646 XPUSHs (sv_2mortal (sv_bless (
1647 newRV_noinc (pv),
1648 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1649 )));
1650}
1651
1285SV *ascii (SV *self, int enable = 1) 1652void ascii (JSON *self, int enable = 1)
1286 ALIAS: 1653 ALIAS:
1287 ascii = F_ASCII 1654 ascii = F_ASCII
1288 latin1 = F_LATIN1 1655 latin1 = F_LATIN1
1289 utf8 = F_UTF8 1656 utf8 = F_UTF8
1290 indent = F_INDENT 1657 indent = F_INDENT
… …
1294 pretty = F_PRETTY 1661 pretty = F_PRETTY
1295 allow_nonref = F_ALLOW_NONREF 1662 allow_nonref = F_ALLOW_NONREF
1296 shrink = F_SHRINK 1663 shrink = F_SHRINK
1297 allow_blessed = F_ALLOW_BLESSED 1664 allow_blessed = F_ALLOW_BLESSED
1298 convert_blessed = F_CONV_BLESSED 1665 convert_blessed = F_CONV_BLESSED
1666 relaxed = F_RELAXED
1667 allow_unknown = F_ALLOW_UNKNOWN
1668 PPCODE:
1669{
1670 if (enable)
1671 self->flags |= ix;
1672 else
1673 self->flags &= ~ix;
1674
1675 XPUSHs (ST (0));
1676}
1677
1678void get_ascii (JSON *self)
1679 ALIAS:
1680 get_ascii = F_ASCII
1681 get_latin1 = F_LATIN1
1682 get_utf8 = F_UTF8
1683 get_indent = F_INDENT
1684 get_canonical = F_CANONICAL
1685 get_space_before = F_SPACE_BEFORE
1686 get_space_after = F_SPACE_AFTER
1687 get_allow_nonref = F_ALLOW_NONREF
1688 get_shrink = F_SHRINK
1689 get_allow_blessed = F_ALLOW_BLESSED
1690 get_convert_blessed = F_CONV_BLESSED
1691 get_relaxed = F_RELAXED
1692 get_allow_unknown = F_ALLOW_UNKNOWN
1693 PPCODE:
1694 XPUSHs (boolSV (self->flags & ix));
1695
1696void max_depth (JSON *self, UV max_depth = 0x80000000UL)
1697 PPCODE:
1698{
1699 UV log2 = 0;
1700
1701 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1702
1703 while ((1UL << log2) < max_depth)
1704 ++log2;
1705
1706 self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1707
1708 XPUSHs (ST (0));
1709}
1710
1711U32 get_max_depth (JSON *self)
1299 CODE: 1712 CODE:
1300{ 1713 RETVAL = DEC_DEPTH (self->flags);
1301 UV *uv = SvJSON (self);
1302 if (enable)
1303 *uv |= ix;
1304 else
1305 *uv &= ~ix;
1306
1307 RETVAL = newSVsv (self);
1308}
1309 OUTPUT: 1714 OUTPUT:
1310 RETVAL 1715 RETVAL
1311 1716
1312SV *max_depth (SV *self, UV max_depth = 0x80000000UL) 1717void max_size (JSON *self, UV max_size = 0)
1718 PPCODE:
1719{
1720 UV log2 = 0;
1721
1722 if (max_size > 0x80000000UL) max_size = 0x80000000UL;
1723 if (max_size == 1) max_size = 2;
1724
1725 while ((1UL << log2) < max_size)
1726 ++log2;
1727
1728 self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1729
1730 XPUSHs (ST (0));
1731}
1732
1733int get_max_size (JSON *self)
1313 CODE: 1734 CODE:
1314{ 1735 RETVAL = DEC_SIZE (self->flags);
1315 UV *uv = SvJSON (self);
1316 UV log2 = 0;
1317
1318 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1319
1320 while ((1UL << log2) < max_depth)
1321 ++log2;
1322
1323 *uv = *uv & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1324
1325 RETVAL = newSVsv (self);
1326}
1327 OUTPUT: 1736 OUTPUT:
1328 RETVAL 1737 RETVAL
1329 1738
1330SV *max_size (SV *self, UV max_size = 0) 1739void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1740 PPCODE:
1741{
1742 SvREFCNT_dec (self->cb_object);
1743 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
1744
1745 XPUSHs (ST (0));
1746}
1747
1748void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
1749 PPCODE:
1750{
1751 if (!self->cb_sk_object)
1752 self->cb_sk_object = newHV ();
1753
1754 if (SvOK (cb))
1755 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1756 else
1757 {
1758 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1759
1760 if (!HvKEYS (self->cb_sk_object))
1761 {
1762 SvREFCNT_dec (self->cb_sk_object);
1763 self->cb_sk_object = 0;
1764 }
1765 }
1766
1767 XPUSHs (ST (0));
1768}
1769
1770void encode (JSON *self, SV *scalar)
1771 PPCODE:
1772 XPUSHs (encode_json (scalar, self));
1773
1774void decode (JSON *self, SV *jsonstr)
1775 PPCODE:
1776 XPUSHs (decode_json (jsonstr, self, 0));
1777
1778void decode_prefix (JSON *self, SV *jsonstr)
1779 PPCODE:
1780{
1781 STRLEN offset;
1782 EXTEND (SP, 2);
1783 PUSHs (decode_json (jsonstr, self, &offset));
1784 PUSHs (sv_2mortal (newSVuv (offset)));
1785}
1786
1787void incr_parse (JSON *self, SV *jsonstr = 0)
1788 PPCODE:
1789{
1790 if (!self->incr_text)
1791 self->incr_text = newSVpvn ("", 0);
1792
1793 // append data, if any
1794 if (jsonstr)
1795 {
1796 if (SvUTF8 (jsonstr) && !SvUTF8 (self->incr_text))
1797 {
1798 /* utf-8-ness differs, need to upgrade */
1799 sv_utf8_upgrade (self->incr_text);
1800
1801 if (self->incr_pos)
1802 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
1803 - (U8 *)SvPVX (self->incr_text);
1804 }
1805
1806 {
1807 STRLEN len;
1808 const char *str = SvPV (jsonstr, len);
1809 SvGROW (self->incr_text, SvCUR (self->incr_text) + len + 1);
1810 Move (str, SvEND (self->incr_text), len, char);
1811 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
1812 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
1813 }
1814 }
1815
1816 if (GIMME_V != G_VOID)
1817 do
1818 {
1819 STRLEN offset;
1820
1821 if (!INCR_DONE (self))
1822 {
1823 incr_parse (self);
1824 if (!INCR_DONE (self))
1825 break;
1826 }
1827
1828 XPUSHs (decode_json (self->incr_text, self, &offset));
1829
1830 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + offset);
1831 self->incr_pos -= offset;
1832 self->incr_nest = 0;
1833 self->incr_mode = 0;
1834 }
1835 while (GIMME_V == G_ARRAY);
1836}
1837
1838SV *incr_text (JSON *self)
1839 ATTRS: lvalue
1331 CODE: 1840 CODE:
1332{ 1841{
1333 UV *uv = SvJSON (self); 1842 if (self->incr_pos)
1334 UV log2 = 0; 1843 croak ("incr_text can not be called when the incremental parser already started parsing");
1335 1844
1336 if (max_size > 0x80000000UL) max_size = 0x80000000UL; 1845 RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef;
1337 if (max_size == 1) max_size = 2;
1338
1339 while ((1UL << log2) < max_size)
1340 ++log2;
1341
1342 *uv = *uv & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1343
1344 RETVAL = newSVsv (self);
1345} 1846}
1346 OUTPUT: 1847 OUTPUT:
1347 RETVAL 1848 RETVAL
1348 1849
1850void incr_skip (JSON *self)
1851 CODE:
1852{
1853 if (self->incr_pos)
1854 {
1855 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + self->incr_pos);
1856 self->incr_pos = 0;
1857 self->incr_nest = 0;
1858 self->incr_mode = 0;
1859 }
1860}
1861
1862void DESTROY (JSON *self)
1863 CODE:
1864 SvREFCNT_dec (self->cb_sk_object);
1865 SvREFCNT_dec (self->cb_object);
1866 SvREFCNT_dec (self->incr_text);
1867
1868PROTOTYPES: ENABLE
1869
1349void encode (SV *self, SV *scalar) 1870void encode_json (SV *scalar)
1871 ALIAS:
1872 to_json_ = 0
1873 encode_json = F_UTF8
1350 PPCODE: 1874 PPCODE:
1351{ 1875{
1352 struct json json = { *SvJSON (self) }; 1876 JSON json = { F_DEFAULT | ix };
1353 XPUSHs (encode_json (scalar, &json)); 1877 XPUSHs (encode_json (scalar, &json));
1354} 1878}
1355 1879
1356void decode (SV *self, SV *jsonstr) 1880void decode_json (SV *jsonstr)
1881 ALIAS:
1882 from_json_ = 0
1883 decode_json = F_UTF8
1357 PPCODE: 1884 PPCODE:
1358{ 1885{
1359 struct json json = { *SvJSON (self) }; 1886 JSON json = { F_DEFAULT | ix };
1360 XPUSHs (decode_json (jsonstr, &json, 0)); 1887 XPUSHs (decode_json (jsonstr, &json, 0));
1361} 1888}
1362 1889
1363void decode_prefix (SV *self, SV *jsonstr)
1364 PPCODE:
1365{
1366 UV offset;
1367 struct json json = { *SvJSON (self) };
1368 EXTEND (SP, 2);
1369 PUSHs (decode_json (jsonstr, &json, &offset));
1370 PUSHs (sv_2mortal (newSVuv (offset)));
1371}
1372 1890
1373PROTOTYPES: ENABLE
1374
1375void to_json (SV *scalar)
1376 PPCODE:
1377{
1378 struct json json = { F_DEFAULT | F_UTF8 };
1379 XPUSHs (encode_json (scalar, &json));
1380}
1381
1382void from_json (SV *jsonstr)
1383 PPCODE:
1384{
1385 struct json json = { F_DEFAULT | F_UTF8 };
1386 XPUSHs (decode_json (jsonstr, &json, 0));
1387}
1388

Diff Legend

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