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 1.40 by root, Tue Jun 12 01:27:02 2007 UTC vs.
Revision 1.75 by root, Wed Mar 19 15:37:54 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
16// some old perls do not have this, try to make it work, no
17// guarentees, though. if it breaks, you get to keep the pieces.
18#ifndef UTF8_MAXBYTES
19# define UTF8_MAXBYTES 13
20#endif
21
22#define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 2)
23
14#define F_ASCII 0x00000001UL 24#define F_ASCII 0x00000001UL
15#define F_LATIN1 0x00000002UL 25#define F_LATIN1 0x00000002UL
16#define F_UTF8 0x00000004UL 26#define F_UTF8 0x00000004UL
17#define F_INDENT 0x00000008UL 27#define F_INDENT 0x00000008UL
18#define F_CANONICAL 0x00000010UL 28#define F_CANONICAL 0x00000010UL
19#define F_SPACE_BEFORE 0x00000020UL 29#define F_SPACE_BEFORE 0x00000020UL
20#define F_SPACE_AFTER 0x00000040UL 30#define F_SPACE_AFTER 0x00000040UL
21#define F_ALLOW_NONREF 0x00000100UL 31#define F_ALLOW_NONREF 0x00000100UL
22#define F_SHRINK 0x00000200UL 32#define F_SHRINK 0x00000200UL
33#define F_ALLOW_BLESSED 0x00000400UL
34#define F_CONV_BLESSED 0x00000800UL
35#define F_RELAXED 0x00001000UL
36
23#define F_MAXDEPTH 0xf8000000UL 37#define F_MAXDEPTH 0xf8000000UL
24#define S_MAXDEPTH 27 38#define S_MAXDEPTH 27
39#define F_MAXSIZE 0x01f00000UL
40#define S_MAXSIZE 20
41#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
25 42
26#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH)) 43#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
27 44#define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE ))
28// F_SELFCONVERT? <=> to_json/toJson
29// F_BLESSED? <=> { $__class__$ => }
30 45
31#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 46#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
32#define F_DEFAULT (9UL << S_MAXDEPTH) 47#define F_DEFAULT (9UL << S_MAXDEPTH)
33 48
34#define INIT_SIZE 32 // initial scalar size to be allocated 49#define INIT_SIZE 32 // initial scalar size to be allocated
… …
38 53
39#define SB do { 54#define SB do {
40#define SE } while (0) 55#define SE } while (0)
41 56
42#if __GNUC__ >= 3 57#if __GNUC__ >= 3
43# define expect(expr,value) __builtin_expect ((expr),(value)) 58# define expect(expr,value) __builtin_expect ((expr), (value))
44# define inline inline 59# define INLINE static inline
45#else 60#else
46# define expect(expr,value) (expr) 61# define expect(expr,value) (expr)
47# define inline static 62# define INLINE static
48#endif 63#endif
49 64
50#define expect_false(expr) expect ((expr) != 0, 0) 65#define expect_false(expr) expect ((expr) != 0, 0)
51#define expect_true(expr) expect ((expr) != 0, 1) 66#define expect_true(expr) expect ((expr) != 0, 1)
52 67
68#define IN_RANGE_INC(type,val,beg,end) \
69 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \
70 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
71
72#ifdef USE_ITHREADS
73# define JSON_SLOW 1
74# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1))
75#else
76# define JSON_SLOW 0
77# define JSON_STASH json_stash
78#endif
79
53static HV *json_stash; // JSON::XS:: 80static HV *json_stash, *json_boolean_stash; // JSON::XS::
81static SV *json_true, *json_false;
82
83typedef struct {
84 U32 flags;
85 SV *cb_object;
86 HV *cb_sk_object;
87} JSON;
54 88
55///////////////////////////////////////////////////////////////////////////// 89/////////////////////////////////////////////////////////////////////////////
56// utility functions 90// utility functions
57 91
58static UV * 92INLINE void
59SvJSON (SV *sv)
60{
61 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash))
62 croak ("object is not of type JSON::XS");
63
64 return &SvUVX (SvRV (sv));
65}
66
67static void
68shrink (SV *sv) 93shrink (SV *sv)
69{ 94{
70 sv_utf8_downgrade (sv, 1); 95 sv_utf8_downgrade (sv, 1);
71 if (SvLEN (sv) > SvCUR (sv) + 1) 96 if (SvLEN (sv) > SvCUR (sv) + 1)
72 { 97 {
… …
81// decode an utf-8 character and return it, or (UV)-1 in 106// decode an utf-8 character and return it, or (UV)-1 in
82// case of an error. 107// case of an error.
83// we special-case "safe" characters from U+80 .. U+7FF, 108// we special-case "safe" characters from U+80 .. U+7FF,
84// but use the very good perl function to parse anything else. 109// but use the very good perl function to parse anything else.
85// note that we never call this function for a ascii codepoints 110// note that we never call this function for a ascii codepoints
86inline UV 111INLINE UV
87decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) 112decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
88{ 113{
89 if (expect_false (s[0] > 0xdf || s[0] < 0xc2)) 114 if (expect_true (len >= 2
90 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); 115 && IN_RANGE_INC (char, s[0], 0xc2, 0xdf)
91 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) 116 && IN_RANGE_INC (char, s[1], 0x80, 0xbf)))
92 { 117 {
93 *clen = 2; 118 *clen = 2;
94 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 119 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
95 } 120 }
96 else 121 else
97 { 122 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
98 *clen = (STRLEN)-1; 123}
99 return (UV)-1; 124
100 } 125// likewise for encoding, also never called for ascii codepoints
126// this function takes advantage of this fact, although current gccs
127// seem to optimise the check for >= 0x80 away anyways
128INLINE unsigned char *
129encode_utf8 (unsigned char *s, UV ch)
130{
131 if (expect_false (ch < 0x000080))
132 *s++ = ch;
133 else if (expect_true (ch < 0x000800))
134 *s++ = 0xc0 | ( ch >> 6),
135 *s++ = 0x80 | ( ch & 0x3f);
136 else if ( ch < 0x010000)
137 *s++ = 0xe0 | ( ch >> 12),
138 *s++ = 0x80 | ((ch >> 6) & 0x3f),
139 *s++ = 0x80 | ( ch & 0x3f);
140 else if ( ch < 0x110000)
141 *s++ = 0xf0 | ( ch >> 18),
142 *s++ = 0x80 | ((ch >> 12) & 0x3f),
143 *s++ = 0x80 | ((ch >> 6) & 0x3f),
144 *s++ = 0x80 | ( ch & 0x3f);
145
146 return s;
101} 147}
102 148
103///////////////////////////////////////////////////////////////////////////// 149/////////////////////////////////////////////////////////////////////////////
104// encoder 150// encoder
105 151
… …
107typedef struct 153typedef struct
108{ 154{
109 char *cur; // SvPVX (sv) + current output position 155 char *cur; // SvPVX (sv) + current output position
110 char *end; // SvEND (sv) 156 char *end; // SvEND (sv)
111 SV *sv; // result scalar 157 SV *sv; // result scalar
112 U32 flags; // F_* 158 JSON json;
113 U32 indent; // indentation level 159 U32 indent; // indentation level
114 U32 maxdepth; // max. indentation/recursion level 160 U32 maxdepth; // max. indentation/recursion level
161 UV limit; // escape character values >= this value when encoding
115} enc_t; 162} enc_t;
116 163
117inline void 164INLINE void
118need (enc_t *enc, STRLEN len) 165need (enc_t *enc, STRLEN len)
119{ 166{
120 if (expect_false (enc->cur + len >= enc->end)) 167 if (expect_false (enc->cur + len >= enc->end))
121 { 168 {
122 STRLEN cur = enc->cur - SvPVX (enc->sv); 169 STRLEN cur = enc->cur - SvPVX (enc->sv);
… …
124 enc->cur = SvPVX (enc->sv) + cur; 171 enc->cur = SvPVX (enc->sv) + cur;
125 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 172 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
126 } 173 }
127} 174}
128 175
129inline void 176INLINE void
130encode_ch (enc_t *enc, char ch) 177encode_ch (enc_t *enc, char ch)
131{ 178{
132 need (enc, 1); 179 need (enc, 1);
133 *enc->cur++ = ch; 180 *enc->cur++ = ch;
134} 181}
… …
188 { 235 {
189 uch = ch; 236 uch = ch;
190 clen = 1; 237 clen = 1;
191 } 238 }
192 239
193 if (uch > 0x10FFFFUL) 240 if (uch < 0x80/*0x20*/ || uch >= enc->limit)
194 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
195
196 if (uch < 0x80 || enc->flags & F_ASCII || (enc->flags & F_LATIN1 && uch > 0xFF))
197 { 241 {
198 if (uch > 0xFFFFUL) 242 if (uch >= 0x10000UL)
199 { 243 {
244 if (uch >= 0x110000UL)
245 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
246
200 need (enc, len += 11); 247 need (enc, len += 11);
201 sprintf (enc->cur, "\\u%04x\\u%04x", 248 sprintf (enc->cur, "\\u%04x\\u%04x",
202 (int)((uch - 0x10000) / 0x400 + 0xD800), 249 (int)((uch - 0x10000) / 0x400 + 0xD800),
203 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 250 (int)((uch - 0x10000) % 0x400 + 0xDC00));
204 enc->cur += 12; 251 enc->cur += 12;
… …
215 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 262 *enc->cur++ = hexdigit [(uch >> 0) & 15];
216 } 263 }
217 264
218 str += clen; 265 str += clen;
219 } 266 }
220 else if (enc->flags & F_LATIN1) 267 else if (enc->json.flags & F_LATIN1)
221 { 268 {
222 *enc->cur++ = uch; 269 *enc->cur++ = uch;
223 str += clen; 270 str += clen;
224 } 271 }
225 else if (is_utf8) 272 else if (is_utf8)
… …
232 while (--clen); 279 while (--clen);
233 } 280 }
234 else 281 else
235 { 282 {
236 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed 283 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
237 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 284 enc->cur = encode_utf8 (enc->cur, uch);
238 ++str; 285 ++str;
239 } 286 }
240 } 287 }
241 } 288 }
242 } 289 }
243 290
244 --len; 291 --len;
245 } 292 }
246} 293}
247 294
248inline void 295INLINE void
249encode_indent (enc_t *enc) 296encode_indent (enc_t *enc)
250{ 297{
251 if (enc->flags & F_INDENT) 298 if (enc->json.flags & F_INDENT)
252 { 299 {
253 int spaces = enc->indent * INDENT_STEP; 300 int spaces = enc->indent * INDENT_STEP;
254 301
255 need (enc, spaces); 302 need (enc, spaces);
256 memset (enc->cur, ' ', spaces); 303 memset (enc->cur, ' ', spaces);
257 enc->cur += spaces; 304 enc->cur += spaces;
258 } 305 }
259} 306}
260 307
261inline void 308INLINE void
262encode_space (enc_t *enc) 309encode_space (enc_t *enc)
263{ 310{
264 need (enc, 1); 311 need (enc, 1);
265 encode_ch (enc, ' '); 312 encode_ch (enc, ' ');
266} 313}
267 314
268inline void 315INLINE void
269encode_nl (enc_t *enc) 316encode_nl (enc_t *enc)
270{ 317{
271 if (enc->flags & F_INDENT) 318 if (enc->json.flags & F_INDENT)
272 { 319 {
273 need (enc, 1); 320 need (enc, 1);
274 encode_ch (enc, '\n'); 321 encode_ch (enc, '\n');
275 } 322 }
276} 323}
277 324
278inline void 325INLINE void
279encode_comma (enc_t *enc) 326encode_comma (enc_t *enc)
280{ 327{
281 encode_ch (enc, ','); 328 encode_ch (enc, ',');
282 329
283 if (enc->flags & F_INDENT) 330 if (enc->json.flags & F_INDENT)
284 encode_nl (enc); 331 encode_nl (enc);
285 else if (enc->flags & F_SPACE_AFTER) 332 else if (enc->json.flags & F_SPACE_AFTER)
286 encode_space (enc); 333 encode_space (enc);
287} 334}
288 335
289static void encode_sv (enc_t *enc, SV *sv); 336static void encode_sv (enc_t *enc, SV *sv);
290 337
… …
294 int i, len = av_len (av); 341 int i, len = av_len (av);
295 342
296 if (enc->indent >= enc->maxdepth) 343 if (enc->indent >= enc->maxdepth)
297 croak ("data structure too deep (hit recursion limit)"); 344 croak ("data structure too deep (hit recursion limit)");
298 345
299 encode_ch (enc, '['); encode_nl (enc); 346 encode_ch (enc, '[');
300 ++enc->indent; 347
348 if (len >= 0)
349 {
350 encode_nl (enc); ++enc->indent;
301 351
302 for (i = 0; i <= len; ++i) 352 for (i = 0; i <= len; ++i)
303 { 353 {
354 SV **svp = av_fetch (av, i, 0);
355
304 encode_indent (enc); 356 encode_indent (enc);
305 encode_sv (enc, *av_fetch (av, i, 0));
306 357
358 if (svp)
359 encode_sv (enc, *svp);
360 else
361 encode_str (enc, "null", 4, 0);
362
307 if (i < len) 363 if (i < len)
308 encode_comma (enc); 364 encode_comma (enc);
309 } 365 }
310 366
367 encode_nl (enc); --enc->indent; encode_indent (enc);
368 }
369
311 encode_nl (enc); 370 encode_ch (enc, ']');
312
313 --enc->indent;
314 encode_indent (enc); encode_ch (enc, ']');
315} 371}
316 372
317static void 373static void
318encode_he (enc_t *enc, HE *he) 374encode_hk (enc_t *enc, HE *he)
319{ 375{
320 encode_ch (enc, '"'); 376 encode_ch (enc, '"');
321 377
322 if (HeKLEN (he) == HEf_SVKEY) 378 if (HeKLEN (he) == HEf_SVKEY)
323 { 379 {
… …
333 else 389 else
334 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 390 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
335 391
336 encode_ch (enc, '"'); 392 encode_ch (enc, '"');
337 393
338 if (enc->flags & F_SPACE_BEFORE) encode_space (enc); 394 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
339 encode_ch (enc, ':'); 395 encode_ch (enc, ':');
340 if (enc->flags & F_SPACE_AFTER ) encode_space (enc); 396 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
341 encode_sv (enc, HeVAL (he));
342} 397}
343 398
344// compare hash entries, used when all keys are bytestrings 399// compare hash entries, used when all keys are bytestrings
345static int 400static int
346he_cmp_fast (const void *a_, const void *b_) 401he_cmp_fast (const void *a_, const void *b_)
… …
351 HE *b = *(HE **)b_; 406 HE *b = *(HE **)b_;
352 407
353 STRLEN la = HeKLEN (a); 408 STRLEN la = HeKLEN (a);
354 STRLEN lb = HeKLEN (b); 409 STRLEN lb = HeKLEN (b);
355 410
356 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 411 if (!(cmp = memcmp (HeKEY (b), HeKEY (a), lb < la ? lb : la)))
357 cmp = la - lb; 412 cmp = lb - la;
358 413
359 return cmp; 414 return cmp;
360} 415}
361 416
362// compare hash entries, used when some keys are sv's or utf-x 417// compare hash entries, used when some keys are sv's or utf-x
363static int 418static int
364he_cmp_slow (const void *a, const void *b) 419he_cmp_slow (const void *a, const void *b)
365{ 420{
366 return sv_cmp (HeSVKEY_force (*(HE **)a), HeSVKEY_force (*(HE **)b)); 421 return sv_cmp (HeSVKEY_force (*(HE **)b), HeSVKEY_force (*(HE **)a));
367} 422}
368 423
369static void 424static void
370encode_hv (enc_t *enc, HV *hv) 425encode_hv (enc_t *enc, HV *hv)
371{ 426{
427 HE *he;
372 int count, i; 428 int count;
373 429
374 if (enc->indent >= enc->maxdepth) 430 if (enc->indent >= enc->maxdepth)
375 croak ("data structure too deep (hit recursion limit)"); 431 croak ("data structure too deep (hit recursion limit)");
376 432
377 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 433 encode_ch (enc, '{');
378 434
379 if ((count = hv_iterinit (hv)))
380 {
381 // for canonical output we have to sort by keys first 435 // for canonical output we have to sort by keys first
382 // actually, this is mostly due to the stupid so-called 436 // actually, this is mostly due to the stupid so-called
383 // security workaround added somewhere in 5.8.x. 437 // security workaround added somewhere in 5.8.x.
384 // that randomises hash orderings 438 // that randomises hash orderings
385 if (enc->flags & F_CANONICAL) 439 if (enc->json.flags & F_CANONICAL)
440 {
441 int count = hv_iterinit (hv);
442
443 if (SvMAGICAL (hv))
386 { 444 {
445 // need to count by iterating. could improve by dynamically building the vector below
446 // but I don't care for the speed of this special case.
447 // note also that we will run into undefined behaviour when the two iterations
448 // do not result in the same count, something I might care for in some later release.
449
450 count = 0;
451 while (hv_iternext (hv))
452 ++count;
453
454 hv_iterinit (hv);
455 }
456
457 if (count)
458 {
387 int fast = 1; 459 int i, fast = 1;
388 HE *he;
389#if defined(__BORLANDC__) || defined(_MSC_VER) 460#if defined(__BORLANDC__) || defined(_MSC_VER)
390 HE **hes = _alloca (count * sizeof (HE)); 461 HE **hes = _alloca (count * sizeof (HE));
391#else 462#else
392 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode 463 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
393#endif 464#endif
… …
420 491
421 FREETMPS; 492 FREETMPS;
422 LEAVE; 493 LEAVE;
423 } 494 }
424 495
425 for (i = 0; i < count; ++i) 496 encode_nl (enc); ++enc->indent;
497
498 while (count--)
426 { 499 {
427 encode_indent (enc); 500 encode_indent (enc);
501 he = hes [count];
428 encode_he (enc, hes [i]); 502 encode_hk (enc, he);
503 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
429 504
430 if (i < count - 1) 505 if (count)
431 encode_comma (enc); 506 encode_comma (enc);
432 } 507 }
433 508
509 encode_nl (enc); --enc->indent; encode_indent (enc);
510 }
511 }
512 else
513 {
514 if (hv_iterinit (hv) || SvMAGICAL (hv))
515 if ((he = hv_iternext (hv)))
516 {
517 encode_nl (enc); ++enc->indent;
518
519 for (;;)
520 {
434 encode_nl (enc); 521 encode_indent (enc);
522 encode_hk (enc, he);
523 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
524
525 if (!(he = hv_iternext (hv)))
526 break;
527
528 encode_comma (enc);
529 }
530
531 encode_nl (enc); --enc->indent; encode_indent (enc);
532 }
533 }
534
535 encode_ch (enc, '}');
536}
537
538// encode objects, arrays and special \0=false and \1=true values.
539static void
540encode_rv (enc_t *enc, SV *sv)
541{
542 svtype svt;
543
544 SvGETMAGIC (sv);
545 svt = SvTYPE (sv);
546
547 if (expect_false (SvOBJECT (sv)))
548 {
549 HV *stash = !JSON_SLOW || json_boolean_stash
550 ? json_boolean_stash
551 : gv_stashpv ("JSON::XS::Boolean", 1);
552
553 if (SvSTASH (sv) == stash)
554 {
555 if (SvIV (sv))
556 encode_str (enc, "true", 4, 0);
557 else
558 encode_str (enc, "false", 5, 0);
435 } 559 }
436 else 560 else
437 { 561 {
438 HE *he = hv_iternext (hv); 562#if 0
439 563 if (0 && sv_derived_from (rv, "JSON::Literal"))
440 for (;;)
441 { 564 {
442 encode_indent (enc); 565 // not yet
443 encode_he (enc, he);
444
445 if (!(he = hv_iternext (hv)))
446 break;
447
448 encode_comma (enc);
449 } 566 }
567#endif
568 if (enc->json.flags & F_CONV_BLESSED)
569 {
570 // we re-bless the reference to get overload and other niceties right
571 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
450 572
573 if (to_json)
574 {
575 dSP;
576
577 ENTER; SAVETMPS; PUSHMARK (SP);
578 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
579
580 // calling with G_SCALAR ensures that we always get a 1 return value
581 PUTBACK;
582 call_sv ((SV *)GvCV (to_json), G_SCALAR);
583 SPAGAIN;
584
585 // catch this surprisingly common error
586 if (SvROK (TOPs) && SvRV (TOPs) == sv)
587 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
588
589 sv = POPs;
590 PUTBACK;
591
451 encode_nl (enc); 592 encode_sv (enc, sv);
593
594 FREETMPS; LEAVE;
595 }
596 else if (enc->json.flags & F_ALLOW_BLESSED)
597 encode_str (enc, "null", 4, 0);
598 else
599 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
600 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
601 }
602 else if (enc->json.flags & F_ALLOW_BLESSED)
603 encode_str (enc, "null", 4, 0);
604 else
605 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",
606 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
452 } 607 }
453 } 608 }
454
455 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
456}
457
458// encode objects, arrays and special \0=false and \1=true values.
459static void
460encode_rv (enc_t *enc, SV *sv)
461{
462 svtype svt;
463
464 SvGETMAGIC (sv);
465 svt = SvTYPE (sv);
466
467 if (svt == SVt_PVHV) 609 else if (svt == SVt_PVHV)
468 encode_hv (enc, (HV *)sv); 610 encode_hv (enc, (HV *)sv);
469 else if (svt == SVt_PVAV) 611 else if (svt == SVt_PVAV)
470 encode_av (enc, (AV *)sv); 612 encode_av (enc, (AV *)sv);
471 else if (svt < SVt_PVAV) 613 else if (svt < SVt_PVAV)
472 { 614 {
473 if (SvNIOK (sv) && SvIV (sv) == 0) 615 STRLEN len = 0;
616 char *pv = svt ? SvPV (sv, len) : 0;
617
618 if (len == 1 && *pv == '1')
619 encode_str (enc, "true", 4, 0);
620 else if (len == 1 && *pv == '0')
474 encode_str (enc, "false", 5, 0); 621 encode_str (enc, "false", 5, 0);
475 else if (SvNIOK (sv) && SvIV (sv) == 1)
476 encode_str (enc, "true", 4, 0);
477 else 622 else
478 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", 623 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
479 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 624 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
480 } 625 }
481 else 626 else
… …
503 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 648 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
504 enc->cur += strlen (enc->cur); 649 enc->cur += strlen (enc->cur);
505 } 650 }
506 else if (SvIOKp (sv)) 651 else if (SvIOKp (sv))
507 { 652 {
508 // we assume we can always read an IV as a UV 653 // we assume we can always read an IV as a UV and vice versa
509 if (SvUV (sv) & ~(UV)0x7fff) 654 // we assume two's complement
510 { 655 // we assume no aliasing issues in the union
511 // large integer, use the (rather slow) snprintf way. 656 if (SvIsUV (sv) ? SvUVX (sv) <= 59000
512 need (enc, sizeof (UV) * 3); 657 : SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000)
513 enc->cur +=
514 SvIsUV(sv)
515 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
516 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
517 }
518 else
519 { 658 {
520 // optimise the "small number case" 659 // optimise the "small number case"
521 // code will likely be branchless and use only a single multiplication 660 // code will likely be branchless and use only a single multiplication
661 // works for numbers up to 59074
522 I32 i = SvIV (sv); 662 I32 i = SvIVX (sv);
523 U32 u; 663 U32 u;
524 char digit, nz = 0; 664 char digit, nz = 0;
525 665
526 need (enc, 6); 666 need (enc, 6);
527 667
… …
533 673
534 // now output digit by digit, each time masking out the integer part 674 // now output digit by digit, each time masking out the integer part
535 // and multiplying by 5 while moving the decimal point one to the right, 675 // and multiplying by 5 while moving the decimal point one to the right,
536 // resulting in a net multiplication by 10. 676 // resulting in a net multiplication by 10.
537 // we always write the digit to memory but conditionally increment 677 // we always write the digit to memory but conditionally increment
538 // the pointer, to ease the usage of conditional move instructions. 678 // the pointer, to enable the use of conditional move instructions.
539 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5; 679 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffffUL) * 5;
540 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5; 680 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffffUL) * 5;
541 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5; 681 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffffUL) * 5;
542 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5; 682 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffffUL) * 5;
543 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0' 683 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
684 }
685 else
686 {
687 // large integer, use the (rather slow) snprintf way.
688 need (enc, IVUV_MAXCHARS);
689 enc->cur +=
690 SvIsUV(sv)
691 ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))
692 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
544 } 693 }
545 } 694 }
546 else if (SvROK (sv)) 695 else if (SvROK (sv))
547 encode_rv (enc, SvRV (sv)); 696 encode_rv (enc, SvRV (sv));
548 else if (!SvOK (sv)) 697 else if (!SvOK (sv))
… …
551 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 700 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
552 SvPV_nolen (sv), SvFLAGS (sv)); 701 SvPV_nolen (sv), SvFLAGS (sv));
553} 702}
554 703
555static SV * 704static SV *
556encode_json (SV *scalar, U32 flags) 705encode_json (SV *scalar, JSON *json)
557{ 706{
558 enc_t enc; 707 enc_t enc;
559 708
560 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 709 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
561 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 710 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
562 711
563 enc.flags = flags; 712 enc.json = *json;
564 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 713 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
565 enc.cur = SvPVX (enc.sv); 714 enc.cur = SvPVX (enc.sv);
566 enc.end = SvEND (enc.sv); 715 enc.end = SvEND (enc.sv);
567 enc.indent = 0; 716 enc.indent = 0;
568 enc.maxdepth = DEC_DEPTH (flags); 717 enc.maxdepth = DEC_DEPTH (enc.json.flags);
718 enc.limit = enc.json.flags & F_ASCII ? 0x000080UL
719 : enc.json.flags & F_LATIN1 ? 0x000100UL
720 : 0x110000UL;
569 721
570 SvPOK_only (enc.sv); 722 SvPOK_only (enc.sv);
571 encode_sv (&enc, scalar); 723 encode_sv (&enc, scalar);
572 724
573 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 725 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
574 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 726 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
575 727
576 if (!(flags & (F_ASCII | F_LATIN1 | F_UTF8))) 728 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
577 SvUTF8_on (enc.sv); 729 SvUTF8_on (enc.sv);
578 730
579 if (enc.flags & F_SHRINK) 731 if (enc.json.flags & F_SHRINK)
580 shrink (enc.sv); 732 shrink (enc.sv);
581 733
582 return enc.sv; 734 return enc.sv;
583} 735}
584 736
… …
589typedef struct 741typedef struct
590{ 742{
591 char *cur; // current parser pointer 743 char *cur; // current parser pointer
592 char *end; // end of input string 744 char *end; // end of input string
593 const char *err; // parse error, if != 0 745 const char *err; // parse error, if != 0
594 U32 flags; // F_* 746 JSON json;
595 U32 depth; // recursion depth 747 U32 depth; // recursion depth
596 U32 maxdepth; // recursion depth limit 748 U32 maxdepth; // recursion depth limit
597} dec_t; 749} dec_t;
598 750
599inline void 751INLINE void
752decode_comment (dec_t *dec)
753{
754 // only '#'-style comments allowed a.t.m.
755
756 while (*dec->cur && *dec->cur != 0x0a && *dec->cur != 0x0d)
757 ++dec->cur;
758}
759
760INLINE void
600decode_ws (dec_t *dec) 761decode_ws (dec_t *dec)
601{ 762{
602 for (;;) 763 for (;;)
603 { 764 {
604 char ch = *dec->cur; 765 char ch = *dec->cur;
605 766
606 if (ch > 0x20 767 if (ch > 0x20)
768 {
769 if (expect_false (ch == '#'))
770 {
771 if (dec->json.flags & F_RELAXED)
772 decode_comment (dec);
773 else
774 break;
775 }
776 else
777 break;
778 }
607 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) 779 else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)
608 break; 780 break; // parse error, but let higher level handle it, gives better error messages
609 781
610 ++dec->cur; 782 ++dec->cur;
611 } 783 }
612} 784}
613 785
… …
719 891
720 if (hi >= 0x80) 892 if (hi >= 0x80)
721 { 893 {
722 utf8 = 1; 894 utf8 = 1;
723 895
724 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0); 896 cur = encode_utf8 (cur, hi);
725 } 897 }
726 else 898 else
727 *cur++ = hi; 899 *cur++ = hi;
728 } 900 }
729 break; 901 break;
… …
731 default: 903 default:
732 --dec_cur; 904 --dec_cur;
733 ERR ("illegal backslash escape sequence in string"); 905 ERR ("illegal backslash escape sequence in string");
734 } 906 }
735 } 907 }
736 else if (expect_true (ch >= 0x20 && ch <= 0x7f)) 908 else if (expect_true (ch >= 0x20 && ch < 0x80))
737 *cur++ = ch; 909 *cur++ = ch;
738 else if (ch >= 0x80) 910 else if (ch >= 0x80)
739 { 911 {
740 STRLEN clen; 912 STRLEN clen;
741 UV uch; 913 UV uch;
… …
862 is_nv = 1; 1034 is_nv = 1;
863 } 1035 }
864 1036
865 if (!is_nv) 1037 if (!is_nv)
866 { 1038 {
1039 int len = dec->cur - start;
1040
867 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so 1041 // special case the rather common 1..5-digit-int case
868 if (*start == '-') 1042 if (*start == '-')
869 switch (dec->cur - start) 1043 switch (len)
870 { 1044 {
871 case 2: return newSViv (-( start [1] - '0' * 1)); 1045 case 2: return newSViv (-( start [1] - '0' * 1));
872 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); 1046 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
873 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); 1047 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
874 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); 1048 case 5: return newSViv (-( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
1049 case 6: return newSViv (-(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
875 } 1050 }
876 else 1051 else
877 switch (dec->cur - start) 1052 switch (len)
878 { 1053 {
879 case 1: return newSViv ( start [0] - '0' * 1); 1054 case 1: return newSViv ( start [0] - '0' * 1);
880 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); 1055 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
881 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); 1056 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
882 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); 1057 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
1058 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
883 } 1059 }
884 1060
885 { 1061 {
886 UV uv; 1062 UV uv;
887 int numtype = grok_number (start, dec->cur - start, &uv); 1063 int numtype = grok_number (start, len, &uv);
888 if (numtype & IS_NUMBER_IN_UV) 1064 if (numtype & IS_NUMBER_IN_UV)
889 if (numtype & IS_NUMBER_NEG) 1065 if (numtype & IS_NUMBER_NEG)
890 { 1066 {
891 if (uv < (UV)IV_MIN) 1067 if (uv < (UV)IV_MIN)
892 return newSViv (-(IV)uv); 1068 return newSViv (-(IV)uv);
893 } 1069 }
894 else 1070 else
895 return newSVuv (uv); 1071 return newSVuv (uv);
896
897 // here would likely be the place for bigint support
898 } 1072 }
899 }
900 1073
901 // if we ever support bigint or bigfloat, this is the place for bigfloat 1074 len -= *start == '-' ? 1 : 0;
1075
1076 // does not fit into IV or UV, try NV
1077 if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len)
1078 #if defined (LDBL_DIG)
1079 || (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len)
1080 #endif
1081 )
1082 // fits into NV without loss of precision
1083 return newSVnv (Atof (start));
1084
1085 // everything else fails, convert it to a string
1086 return newSVpvn (start, dec->cur - start);
1087 }
1088
1089 // loss of precision here
902 return newSVnv (Atof (start)); 1090 return newSVnv (Atof (start));
903 1091
904fail: 1092fail:
905 return 0; 1093 return 0;
906} 1094}
… …
936 1124
937 if (*dec->cur != ',') 1125 if (*dec->cur != ',')
938 ERR (", or ] expected while parsing array"); 1126 ERR (", or ] expected while parsing array");
939 1127
940 ++dec->cur; 1128 ++dec->cur;
1129
1130 decode_ws (dec);
1131
1132 if (*dec->cur == ']' && dec->json.flags & F_RELAXED)
1133 {
1134 ++dec->cur;
1135 break;
1136 }
941 } 1137 }
942 1138
943 DEC_DEC_DEPTH; 1139 DEC_DEC_DEPTH;
944 return newRV_noinc ((SV *)av); 1140 return newRV_noinc ((SV *)av);
945 1141
… …
950} 1146}
951 1147
952static SV * 1148static SV *
953decode_hv (dec_t *dec) 1149decode_hv (dec_t *dec)
954{ 1150{
1151 SV *sv;
955 HV *hv = newHV (); 1152 HV *hv = newHV ();
956 1153
957 DEC_INC_DEPTH; 1154 DEC_INC_DEPTH;
958 decode_ws (dec); 1155 decode_ws (dec);
959 1156
960 if (*dec->cur == '}') 1157 if (*dec->cur == '}')
961 ++dec->cur; 1158 ++dec->cur;
962 else 1159 else
963 for (;;) 1160 for (;;)
964 { 1161 {
1162 EXPECT_CH ('"');
1163
1164 // heuristic: assume that
1165 // a) decode_str + hv_store_ent are abysmally slow.
1166 // b) most hash keys are short, simple ascii text.
1167 // => try to "fast-match" such strings to avoid
1168 // the overhead of decode_str + hv_store_ent.
1169 {
965 SV *key, *value; 1170 SV *value;
1171 char *p = dec->cur;
1172 char *e = p + 24; // only try up to 24 bytes
966 1173
967 decode_ws (dec); EXPECT_CH ('"'); 1174 for (;;)
968
969 key = decode_str (dec);
970 if (!key)
971 goto fail;
972
973 decode_ws (dec); EXPECT_CH (':');
974
975 value = decode_sv (dec);
976 if (!value)
977 { 1175 {
1176 // the >= 0x80 is true on most architectures
1177 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1178 {
1179 // slow path, back up and use decode_str
1180 SV *key = decode_str (dec);
1181 if (!key)
1182 goto fail;
1183
1184 decode_ws (dec); EXPECT_CH (':');
1185
1186 decode_ws (dec);
1187 value = decode_sv (dec);
1188 if (!value)
1189 {
1190 SvREFCNT_dec (key);
1191 goto fail;
1192 }
1193
1194 hv_store_ent (hv, key, value, 0);
978 SvREFCNT_dec (key); 1195 SvREFCNT_dec (key);
1196
1197 break;
1198 }
1199 else if (*p == '"')
1200 {
1201 // fast path, got a simple key
1202 char *key = dec->cur;
1203 int len = p - key;
1204 dec->cur = p + 1;
1205
1206 decode_ws (dec); EXPECT_CH (':');
1207
1208 decode_ws (dec);
1209 value = decode_sv (dec);
1210 if (!value)
979 goto fail; 1211 goto fail;
1212
1213 hv_store (hv, key, len, value, 0);
1214
1215 break;
1216 }
1217
1218 ++p;
980 } 1219 }
981 1220 }
982 hv_store_ent (hv, key, value, 0);
983 SvREFCNT_dec (key);
984 1221
985 decode_ws (dec); 1222 decode_ws (dec);
986 1223
987 if (*dec->cur == '}') 1224 if (*dec->cur == '}')
988 { 1225 {
… …
992 1229
993 if (*dec->cur != ',') 1230 if (*dec->cur != ',')
994 ERR (", or } expected while parsing object/hash"); 1231 ERR (", or } expected while parsing object/hash");
995 1232
996 ++dec->cur; 1233 ++dec->cur;
1234
1235 decode_ws (dec);
1236
1237 if (*dec->cur == '}' && dec->json.flags & F_RELAXED)
1238 {
1239 ++dec->cur;
1240 break;
1241 }
997 } 1242 }
998 1243
999 DEC_DEC_DEPTH; 1244 DEC_DEC_DEPTH;
1000 return newRV_noinc ((SV *)hv); 1245 sv = newRV_noinc ((SV *)hv);
1246
1247 // check filter callbacks
1248 if (dec->json.flags & F_HOOK)
1249 {
1250 if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
1251 {
1252 HE *cb, *he;
1253
1254 hv_iterinit (hv);
1255 he = hv_iternext (hv);
1256 hv_iterinit (hv);
1257
1258 // the next line creates a mortal sv each time its called.
1259 // might want to optimise this for common cases.
1260 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1261
1262 if (cb)
1263 {
1264 dSP;
1265 int count;
1266
1267 ENTER; SAVETMPS; PUSHMARK (SP);
1268 XPUSHs (HeVAL (he));
1269
1270 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1271
1272 if (count == 1)
1273 {
1274 sv = newSVsv (POPs);
1275 FREETMPS; LEAVE;
1276 return sv;
1277 }
1278
1279 FREETMPS; LEAVE;
1280 }
1281 }
1282
1283 if (dec->json.cb_object)
1284 {
1285 dSP;
1286 int count;
1287
1288 ENTER; SAVETMPS; PUSHMARK (SP);
1289 XPUSHs (sv_2mortal (sv));
1290
1291 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1292
1293 if (count == 1)
1294 {
1295 sv = newSVsv (POPs);
1296 FREETMPS; LEAVE;
1297 return sv;
1298 }
1299
1300 SvREFCNT_inc (sv);
1301 FREETMPS; LEAVE;
1302 }
1303 }
1304
1305 return sv;
1001 1306
1002fail: 1307fail:
1003 SvREFCNT_dec (hv); 1308 SvREFCNT_dec (hv);
1004 DEC_DEC_DEPTH; 1309 DEC_DEC_DEPTH;
1005 return 0; 1310 return 0;
1006} 1311}
1007 1312
1008static SV * 1313static SV *
1009decode_sv (dec_t *dec) 1314decode_sv (dec_t *dec)
1010{ 1315{
1011 decode_ws (dec);
1012
1013 // the beauty of JSON: you need exactly one character lookahead 1316 // the beauty of JSON: you need exactly one character lookahead
1014 // to parse anything. 1317 // to parse everything.
1015 switch (*dec->cur) 1318 switch (*dec->cur)
1016 { 1319 {
1017 case '"': ++dec->cur; return decode_str (dec); 1320 case '"': ++dec->cur; return decode_str (dec);
1018 case '[': ++dec->cur; return decode_av (dec); 1321 case '[': ++dec->cur; return decode_av (dec);
1019 case '{': ++dec->cur; return decode_hv (dec); 1322 case '{': ++dec->cur; return decode_hv (dec);
1020 1323
1021 case '-': 1324 case '-':
1022 case '0': case '1': case '2': case '3': case '4': 1325 case '0': case '1': case '2': case '3': case '4':
1023 case '5': case '6': case '7': case '8': case '9': 1326 case '5': case '6': case '7': case '8': case '9':
1024 return decode_num (dec); 1327 return decode_num (dec);
1025 1328
1026 case 't': 1329 case 't':
1027 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1330 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1028 { 1331 {
1029 dec->cur += 4; 1332 dec->cur += 4;
1030 return newSViv (1); 1333#if JSON_SLOW
1334 json_true = get_sv ("JSON::XS::true", 1); SvREADONLY_on (json_true);
1335#endif
1336 return SvREFCNT_inc (json_true);
1031 } 1337 }
1032 else 1338 else
1033 ERR ("'true' expected"); 1339 ERR ("'true' expected");
1034 1340
1035 break; 1341 break;
1036 1342
1037 case 'f': 1343 case 'f':
1038 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1344 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1039 { 1345 {
1040 dec->cur += 5; 1346 dec->cur += 5;
1041 return newSViv (0); 1347#if JSON_SLOW
1348 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
1349#endif
1350 return SvREFCNT_inc (json_false);
1042 } 1351 }
1043 else 1352 else
1044 ERR ("'false' expected"); 1353 ERR ("'false' expected");
1045 1354
1046 break; 1355 break;
… …
1064fail: 1373fail:
1065 return 0; 1374 return 0;
1066} 1375}
1067 1376
1068static SV * 1377static SV *
1069decode_json (SV *string, U32 flags, UV *offset_return) 1378decode_json (SV *string, JSON *json, UV *offset_return)
1070{ 1379{
1071 dec_t dec; 1380 dec_t dec;
1072 UV offset; 1381 UV offset;
1073 SV *sv; 1382 SV *sv;
1074 1383
1075 SvGETMAGIC (string); 1384 SvGETMAGIC (string);
1076 SvUPGRADE (string, SVt_PV); 1385 SvUPGRADE (string, SVt_PV);
1077 1386
1387 if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags))
1388 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1389 (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags));
1390
1078 if (flags & F_UTF8) 1391 if (json->flags & F_UTF8)
1079 sv_utf8_downgrade (string, 0); 1392 sv_utf8_downgrade (string, 0);
1080 else 1393 else
1081 sv_utf8_upgrade (string); 1394 sv_utf8_upgrade (string);
1082 1395
1083 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1396 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
1084 1397
1085 dec.flags = flags; 1398 dec.json = *json;
1086 dec.cur = SvPVX (string); 1399 dec.cur = SvPVX (string);
1087 dec.end = SvEND (string); 1400 dec.end = SvEND (string);
1088 dec.err = 0; 1401 dec.err = 0;
1089 dec.depth = 0; 1402 dec.depth = 0;
1090 dec.maxdepth = DEC_DEPTH (dec.flags); 1403 dec.maxdepth = DEC_DEPTH (dec.json.flags);
1404
1405 if (dec.json.cb_object || dec.json.cb_sk_object)
1406 dec.json.flags |= F_HOOK;
1091 1407
1092 *dec.end = 0; // this should basically be a nop, too, but make sure it's there 1408 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1409
1410 decode_ws (&dec);
1093 sv = decode_sv (&dec); 1411 sv = decode_sv (&dec);
1094 1412
1095 if (!(offset_return || !sv)) 1413 if (!(offset_return || !sv))
1096 { 1414 {
1097 // check for trailing garbage 1415 // check for trailing garbage
… …
1105 } 1423 }
1106 } 1424 }
1107 1425
1108 if (offset_return || !sv) 1426 if (offset_return || !sv)
1109 { 1427 {
1110 offset = dec.flags & F_UTF8 1428 offset = dec.json.flags & F_UTF8
1111 ? dec.cur - SvPVX (string) 1429 ? dec.cur - SvPVX (string)
1112 : utf8_distance (dec.cur, SvPVX (string)); 1430 : utf8_distance (dec.cur, SvPVX (string));
1113 1431
1114 if (offset_return) 1432 if (offset_return)
1115 *offset_return = offset; 1433 *offset_return = offset;
… …
1134 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1452 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1135 } 1453 }
1136 1454
1137 sv = sv_2mortal (sv); 1455 sv = sv_2mortal (sv);
1138 1456
1139 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1457 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))
1140 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); 1458 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1141 1459
1142 return sv; 1460 return sv;
1143} 1461}
1144 1462
… …
1156 i >= '0' && i <= '9' ? i - '0' 1474 i >= '0' && i <= '9' ? i - '0'
1157 : i >= 'a' && i <= 'f' ? i - 'a' + 10 1475 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1158 : i >= 'A' && i <= 'F' ? i - 'A' + 10 1476 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1159 : -1; 1477 : -1;
1160 1478
1161 json_stash = gv_stashpv ("JSON::XS", 1); 1479 json_stash = gv_stashpv ("JSON::XS" , 1);
1480 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1481
1482 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true );
1483 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
1162} 1484}
1163 1485
1164PROTOTYPES: DISABLE 1486PROTOTYPES: DISABLE
1165 1487
1166SV *new (char *dummy) 1488void CLONE (...)
1167 CODE: 1489 CODE:
1168 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1490 json_stash = 0;
1491 json_boolean_stash = 0;
1492
1493void new (char *klass)
1494 PPCODE:
1495{
1496 SV *pv = NEWSV (0, sizeof (JSON));
1497 SvPOK_only (pv);
1498 Zero (SvPVX (pv), 1, JSON);
1499 ((JSON *)SvPVX (pv))->flags = F_DEFAULT;
1500 XPUSHs (sv_2mortal (sv_bless (
1501 newRV_noinc (pv),
1502 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1503 )));
1504}
1505
1506void ascii (JSON *self, int enable = 1)
1507 ALIAS:
1508 ascii = F_ASCII
1509 latin1 = F_LATIN1
1510 utf8 = F_UTF8
1511 indent = F_INDENT
1512 canonical = F_CANONICAL
1513 space_before = F_SPACE_BEFORE
1514 space_after = F_SPACE_AFTER
1515 pretty = F_PRETTY
1516 allow_nonref = F_ALLOW_NONREF
1517 shrink = F_SHRINK
1518 allow_blessed = F_ALLOW_BLESSED
1519 convert_blessed = F_CONV_BLESSED
1520 relaxed = F_RELAXED
1521 PPCODE:
1522{
1523 if (enable)
1524 self->flags |= ix;
1525 else
1526 self->flags &= ~ix;
1527
1528 XPUSHs (ST (0));
1529}
1530
1531void get_ascii (JSON *self)
1532 ALIAS:
1533 get_ascii = F_ASCII
1534 get_latin1 = F_LATIN1
1535 get_utf8 = F_UTF8
1536 get_indent = F_INDENT
1537 get_canonical = F_CANONICAL
1538 get_space_before = F_SPACE_BEFORE
1539 get_space_after = F_SPACE_AFTER
1540 get_allow_nonref = F_ALLOW_NONREF
1541 get_shrink = F_SHRINK
1542 get_allow_blessed = F_ALLOW_BLESSED
1543 get_convert_blessed = F_CONV_BLESSED
1544 get_relaxed = F_RELAXED
1545 PPCODE:
1546 XPUSHs (boolSV (self->flags & ix));
1547
1548void max_depth (JSON *self, UV max_depth = 0x80000000UL)
1549 PPCODE:
1550{
1551 UV log2 = 0;
1552
1553 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1554
1555 while ((1UL << log2) < max_depth)
1556 ++log2;
1557
1558 self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1559
1560 XPUSHs (ST (0));
1561}
1562
1563U32 get_max_depth (JSON *self)
1564 CODE:
1565 RETVAL = DEC_DEPTH (self->flags);
1169 OUTPUT: 1566 OUTPUT:
1170 RETVAL 1567 RETVAL
1171 1568
1172SV *ascii (SV *self, int enable = 1) 1569void max_size (JSON *self, UV max_size = 0)
1173 ALIAS: 1570 PPCODE:
1174 ascii = F_ASCII 1571{
1175 latin1 = F_LATIN1 1572 UV log2 = 0;
1176 utf8 = F_UTF8 1573
1177 indent = F_INDENT 1574 if (max_size > 0x80000000UL) max_size = 0x80000000UL;
1178 canonical = F_CANONICAL 1575 if (max_size == 1) max_size = 2;
1179 space_before = F_SPACE_BEFORE 1576
1180 space_after = F_SPACE_AFTER 1577 while ((1UL << log2) < max_size)
1181 pretty = F_PRETTY 1578 ++log2;
1182 allow_nonref = F_ALLOW_NONREF 1579
1183 shrink = F_SHRINK 1580 self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1581
1582 XPUSHs (ST (0));
1583}
1584
1585int get_max_size (JSON *self)
1184 CODE: 1586 CODE:
1185{ 1587 RETVAL = DEC_SIZE (self->flags);
1186 UV *uv = SvJSON (self);
1187 if (enable)
1188 *uv |= ix;
1189 else
1190 *uv &= ~ix;
1191
1192 RETVAL = newSVsv (self);
1193}
1194 OUTPUT: 1588 OUTPUT:
1195 RETVAL 1589 RETVAL
1196 1590
1197SV *max_depth (SV *self, UV max_depth = 0x80000000UL) 1591void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1198 CODE:
1199{
1200 UV *uv = SvJSON (self);
1201 UV log2 = 0;
1202
1203 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1204
1205 while ((1UL << log2) < max_depth)
1206 ++log2;
1207
1208 *uv = *uv & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1209
1210 RETVAL = newSVsv (self);
1211}
1212 OUTPUT:
1213 RETVAL
1214
1215void encode (SV *self, SV *scalar)
1216 PPCODE: 1592 PPCODE:
1217 XPUSHs (encode_json (scalar, *SvJSON (self))); 1593{
1594 SvREFCNT_dec (self->cb_object);
1595 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
1218 1596
1219void decode (SV *self, SV *jsonstr) 1597 XPUSHs (ST (0));
1598}
1599
1600void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
1220 PPCODE: 1601 PPCODE:
1602{
1603 if (!self->cb_sk_object)
1604 self->cb_sk_object = newHV ();
1605
1606 if (SvOK (cb))
1607 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1608 else
1609 {
1610 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1611
1612 if (!HvKEYS (self->cb_sk_object))
1613 {
1614 SvREFCNT_dec (self->cb_sk_object);
1615 self->cb_sk_object = 0;
1616 }
1617 }
1618
1619 XPUSHs (ST (0));
1620}
1621
1622void encode (JSON *self, SV *scalar)
1623 PPCODE:
1624 XPUSHs (encode_json (scalar, self));
1625
1626void decode (JSON *self, SV *jsonstr)
1627 PPCODE:
1221 XPUSHs (decode_json (jsonstr, *SvJSON (self), 0)); 1628 XPUSHs (decode_json (jsonstr, self, 0));
1222 1629
1223void decode_prefix (SV *self, SV *jsonstr) 1630void decode_prefix (JSON *self, SV *jsonstr)
1224 PPCODE: 1631 PPCODE:
1225{ 1632{
1226 UV offset; 1633 UV offset;
1227 EXTEND (SP, 2); 1634 EXTEND (SP, 2);
1228 PUSHs (decode_json (jsonstr, *SvJSON (self), &offset)); 1635 PUSHs (decode_json (jsonstr, self, &offset));
1229 PUSHs (sv_2mortal (newSVuv (offset))); 1636 PUSHs (sv_2mortal (newSVuv (offset)));
1230} 1637}
1231 1638
1639void DESTROY (JSON *self)
1640 CODE:
1641 SvREFCNT_dec (self->cb_sk_object);
1642 SvREFCNT_dec (self->cb_object);
1643
1232PROTOTYPES: ENABLE 1644PROTOTYPES: ENABLE
1233 1645
1234void to_json (SV *scalar) 1646void encode_json (SV *scalar)
1235 ALIAS:
1236 objToJson = 0
1237 PPCODE: 1647 PPCODE:
1648{
1649 JSON json = { F_DEFAULT | F_UTF8 };
1238 XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8)); 1650 XPUSHs (encode_json (scalar, &json));
1651}
1239 1652
1240void from_json (SV *jsonstr) 1653void decode_json (SV *jsonstr)
1241 ALIAS:
1242 jsonToObj = 0
1243 PPCODE: 1654 PPCODE:
1655{
1656 JSON json = { F_DEFAULT | F_UTF8 };
1244 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8, 0)); 1657 XPUSHs (decode_json (jsonstr, &json, 0));
1658}
1245 1659

Diff Legend

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