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 1.42 by root, Sat Jun 23 22:56:08 2007 UTC vs.
Revision 1.86 by root, Tue May 27 05:31:39 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
… …
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
19 21
22#define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 2)
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
24#define F_CANONICAL 0x00000010UL 28#define F_CANONICAL 0x00000010UL
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
33#define F_ALLOW_BLESSED 0x00000400UL
34#define F_CONV_BLESSED 0x00000800UL
29#define F_MAXDEPTH 0xf8000000UL 35#define F_RELAXED 0x00001000UL
30#define S_MAXDEPTH 27 36#define F_ALLOW_UNKNOWN 0x00002000UL
31 37#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
32#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
33
34// F_SELFCONVERT? <=> to_json/toJson
35// F_BLESSED? <=> { $__class__$ => }
36 38
37#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 39#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
38#define F_DEFAULT (9UL << S_MAXDEPTH)
39 40
40#define INIT_SIZE 32 // initial scalar size to be allocated 41#define INIT_SIZE 32 // initial scalar size to be allocated
41#define INDENT_STEP 3 // spaces per indentation level 42#define INDENT_STEP 3 // spaces per indentation level
42 43
43#define SHORT_STRING_LEN 16384 // special-case strings of up to this size 44#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
44 45
45#define SB do { 46#define SB do {
46#define SE } while (0) 47#define SE } while (0)
47 48
48#if __GNUC__ >= 3 49#if __GNUC__ >= 3
49# define expect(expr,value) __builtin_expect ((expr),(value)) 50# define expect(expr,value) __builtin_expect ((expr), (value))
50# define inline inline 51# define INLINE static inline
51#else 52#else
52# define expect(expr,value) (expr) 53# define expect(expr,value) (expr)
53# define inline static 54# define INLINE static
54#endif 55#endif
55 56
56#define expect_false(expr) expect ((expr) != 0, 0) 57#define expect_false(expr) expect ((expr) != 0, 0)
57#define expect_true(expr) expect ((expr) != 0, 1) 58#define expect_true(expr) expect ((expr) != 0, 1)
58 59
60#define IN_RANGE_INC(type,val,beg,end) \
61 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \
62 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
63
64#define ERR_NESTING_EXCEEDED "json text or perl structure exceeds maximum nesting level (max_depth set too low?)"
65
66#ifdef USE_ITHREADS
67# define JSON_SLOW 1
68# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1))
69#else
70# define JSON_SLOW 0
71# define JSON_STASH json_stash
72#endif
73
59static HV *json_stash; // JSON::XS:: 74static HV *json_stash, *json_boolean_stash; // JSON::XS::
75static SV *json_true, *json_false;
76
77enum {
78 INCR_M_WS = 0, // initial whitespace skipping, must be 0
79 INCR_M_STR, // inside string
80 INCR_M_BS, // inside backslash
81 INCR_M_JSON // outside anything, count nesting
82};
83
84#define INCR_DONE(json) (!(json)->incr_nest && (json)->incr_mode == INCR_M_JSON)
85
86typedef struct {
87 U32 flags;
88 U32 max_depth;
89 STRLEN max_size;
90
91 SV *cb_object;
92 HV *cb_sk_object;
93
94 // for the incremental parser
95 SV *incr_text; // the source text so far
96 STRLEN incr_pos; // the current offset into the text
97 unsigned char incr_nest; // {[]}-nesting level
98 unsigned char incr_mode;
99} JSON;
100
101INLINE void
102json_init (JSON *json)
103{
104 Zero (json, 1, JSON);
105 json->max_depth = 512;
106}
60 107
61///////////////////////////////////////////////////////////////////////////// 108/////////////////////////////////////////////////////////////////////////////
62// utility functions 109// utility functions
63 110
64static UV * 111INLINE SV *
65SvJSON (SV *sv) 112get_bool (const char *name)
66{ 113{
67 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash)) 114 SV *sv = get_sv (name, 1);
68 croak ("object is not of type JSON::XS");
69 115
70 return &SvUVX (SvRV (sv)); 116 SvREADONLY_on (sv);
71} 117 SvREADONLY_on (SvRV (sv));
72 118
73static void 119 return sv;
120}
121
122INLINE void
74shrink (SV *sv) 123shrink (SV *sv)
75{ 124{
76 sv_utf8_downgrade (sv, 1); 125 sv_utf8_downgrade (sv, 1);
77 if (SvLEN (sv) > SvCUR (sv) + 1) 126 if (SvLEN (sv) > SvCUR (sv) + 1)
78 { 127 {
… …
87// decode an utf-8 character and return it, or (UV)-1 in 136// decode an utf-8 character and return it, or (UV)-1 in
88// case of an error. 137// case of an error.
89// we special-case "safe" characters from U+80 .. U+7FF, 138// we special-case "safe" characters from U+80 .. U+7FF,
90// but use the very good perl function to parse anything else. 139// but use the very good perl function to parse anything else.
91// note that we never call this function for a ascii codepoints 140// note that we never call this function for a ascii codepoints
92inline UV 141INLINE UV
93decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) 142decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
94{ 143{
95 if (expect_false (s[0] > 0xdf || s[0] < 0xc2)) 144 if (expect_true (len >= 2
96 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); 145 && IN_RANGE_INC (char, s[0], 0xc2, 0xdf)
97 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) 146 && IN_RANGE_INC (char, s[1], 0x80, 0xbf)))
98 { 147 {
99 *clen = 2; 148 *clen = 2;
100 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 149 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
101 } 150 }
102 else 151 else
103 { 152 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
104 *clen = (STRLEN)-1; 153}
105 return (UV)-1; 154
106 } 155// likewise for encoding, also never called for ascii codepoints
156// this function takes advantage of this fact, although current gccs
157// seem to optimise the check for >= 0x80 away anyways
158INLINE unsigned char *
159encode_utf8 (unsigned char *s, UV ch)
160{
161 if (expect_false (ch < 0x000080))
162 *s++ = ch;
163 else if (expect_true (ch < 0x000800))
164 *s++ = 0xc0 | ( ch >> 6),
165 *s++ = 0x80 | ( ch & 0x3f);
166 else if ( ch < 0x010000)
167 *s++ = 0xe0 | ( ch >> 12),
168 *s++ = 0x80 | ((ch >> 6) & 0x3f),
169 *s++ = 0x80 | ( ch & 0x3f);
170 else if ( ch < 0x110000)
171 *s++ = 0xf0 | ( ch >> 18),
172 *s++ = 0x80 | ((ch >> 12) & 0x3f),
173 *s++ = 0x80 | ((ch >> 6) & 0x3f),
174 *s++ = 0x80 | ( ch & 0x3f);
175
176 return s;
107} 177}
108 178
109///////////////////////////////////////////////////////////////////////////// 179/////////////////////////////////////////////////////////////////////////////
110// encoder 180// encoder
111 181
… …
113typedef struct 183typedef struct
114{ 184{
115 char *cur; // SvPVX (sv) + current output position 185 char *cur; // SvPVX (sv) + current output position
116 char *end; // SvEND (sv) 186 char *end; // SvEND (sv)
117 SV *sv; // result scalar 187 SV *sv; // result scalar
118 U32 flags; // F_* 188 JSON json;
119 U32 indent; // indentation level 189 U32 indent; // indentation level
120 U32 maxdepth; // max. indentation/recursion level 190 UV limit; // escape character values >= this value when encoding
121} enc_t; 191} enc_t;
122 192
123inline void 193INLINE void
124need (enc_t *enc, STRLEN len) 194need (enc_t *enc, STRLEN len)
125{ 195{
126 if (expect_false (enc->cur + len >= enc->end)) 196 if (expect_false (enc->cur + len >= enc->end))
127 { 197 {
128 STRLEN cur = enc->cur - SvPVX (enc->sv); 198 STRLEN cur = enc->cur - SvPVX (enc->sv);
… …
130 enc->cur = SvPVX (enc->sv) + cur; 200 enc->cur = SvPVX (enc->sv) + cur;
131 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 201 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
132 } 202 }
133} 203}
134 204
135inline void 205INLINE void
136encode_ch (enc_t *enc, char ch) 206encode_ch (enc_t *enc, char ch)
137{ 207{
138 need (enc, 1); 208 need (enc, 1);
139 *enc->cur++ = ch; 209 *enc->cur++ = ch;
140} 210}
… …
194 { 264 {
195 uch = ch; 265 uch = ch;
196 clen = 1; 266 clen = 1;
197 } 267 }
198 268
199 if (uch > 0x10FFFFUL) 269 if (uch < 0x80/*0x20*/ || uch >= enc->limit)
200 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
201
202 if (uch < 0x80 || enc->flags & F_ASCII || (enc->flags & F_LATIN1 && uch > 0xFF))
203 { 270 {
204 if (uch > 0xFFFFUL) 271 if (uch >= 0x10000UL)
205 { 272 {
273 if (uch >= 0x110000UL)
274 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
275
206 need (enc, len += 11); 276 need (enc, len += 11);
207 sprintf (enc->cur, "\\u%04x\\u%04x", 277 sprintf (enc->cur, "\\u%04x\\u%04x",
208 (int)((uch - 0x10000) / 0x400 + 0xD800), 278 (int)((uch - 0x10000) / 0x400 + 0xD800),
209 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 279 (int)((uch - 0x10000) % 0x400 + 0xDC00));
210 enc->cur += 12; 280 enc->cur += 12;
… …
221 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 291 *enc->cur++ = hexdigit [(uch >> 0) & 15];
222 } 292 }
223 293
224 str += clen; 294 str += clen;
225 } 295 }
226 else if (enc->flags & F_LATIN1) 296 else if (enc->json.flags & F_LATIN1)
227 { 297 {
228 *enc->cur++ = uch; 298 *enc->cur++ = uch;
229 str += clen; 299 str += clen;
230 } 300 }
231 else if (is_utf8) 301 else if (is_utf8)
… …
238 while (--clen); 308 while (--clen);
239 } 309 }
240 else 310 else
241 { 311 {
242 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed 312 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
243 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 313 enc->cur = encode_utf8 (enc->cur, uch);
244 ++str; 314 ++str;
245 } 315 }
246 } 316 }
247 } 317 }
248 } 318 }
249 319
250 --len; 320 --len;
251 } 321 }
252} 322}
253 323
254inline void 324INLINE void
255encode_indent (enc_t *enc) 325encode_indent (enc_t *enc)
256{ 326{
257 if (enc->flags & F_INDENT) 327 if (enc->json.flags & F_INDENT)
258 { 328 {
259 int spaces = enc->indent * INDENT_STEP; 329 int spaces = enc->indent * INDENT_STEP;
260 330
261 need (enc, spaces); 331 need (enc, spaces);
262 memset (enc->cur, ' ', spaces); 332 memset (enc->cur, ' ', spaces);
263 enc->cur += spaces; 333 enc->cur += spaces;
264 } 334 }
265} 335}
266 336
267inline void 337INLINE void
268encode_space (enc_t *enc) 338encode_space (enc_t *enc)
269{ 339{
270 need (enc, 1); 340 need (enc, 1);
271 encode_ch (enc, ' '); 341 encode_ch (enc, ' ');
272} 342}
273 343
274inline void 344INLINE void
275encode_nl (enc_t *enc) 345encode_nl (enc_t *enc)
276{ 346{
277 if (enc->flags & F_INDENT) 347 if (enc->json.flags & F_INDENT)
278 { 348 {
279 need (enc, 1); 349 need (enc, 1);
280 encode_ch (enc, '\n'); 350 encode_ch (enc, '\n');
281 } 351 }
282} 352}
283 353
284inline void 354INLINE void
285encode_comma (enc_t *enc) 355encode_comma (enc_t *enc)
286{ 356{
287 encode_ch (enc, ','); 357 encode_ch (enc, ',');
288 358
289 if (enc->flags & F_INDENT) 359 if (enc->json.flags & F_INDENT)
290 encode_nl (enc); 360 encode_nl (enc);
291 else if (enc->flags & F_SPACE_AFTER) 361 else if (enc->json.flags & F_SPACE_AFTER)
292 encode_space (enc); 362 encode_space (enc);
293} 363}
294 364
295static void encode_sv (enc_t *enc, SV *sv); 365static void encode_sv (enc_t *enc, SV *sv);
296 366
297static void 367static void
298encode_av (enc_t *enc, AV *av) 368encode_av (enc_t *enc, AV *av)
299{ 369{
300 int i, len = av_len (av); 370 int i, len = av_len (av);
301 371
302 if (enc->indent >= enc->maxdepth) 372 if (enc->indent >= enc->json.max_depth)
303 croak ("data structure too deep (hit recursion limit)"); 373 croak (ERR_NESTING_EXCEEDED);
304 374
305 encode_ch (enc, '['); encode_nl (enc); 375 encode_ch (enc, '[');
306 ++enc->indent; 376
377 if (len >= 0)
378 {
379 encode_nl (enc); ++enc->indent;
307 380
308 for (i = 0; i <= len; ++i) 381 for (i = 0; i <= len; ++i)
309 { 382 {
383 SV **svp = av_fetch (av, i, 0);
384
310 encode_indent (enc); 385 encode_indent (enc);
311 encode_sv (enc, *av_fetch (av, i, 0));
312 386
387 if (svp)
388 encode_sv (enc, *svp);
389 else
390 encode_str (enc, "null", 4, 0);
391
313 if (i < len) 392 if (i < len)
314 encode_comma (enc); 393 encode_comma (enc);
315 } 394 }
316 395
396 encode_nl (enc); --enc->indent; encode_indent (enc);
397 }
398
317 encode_nl (enc); 399 encode_ch (enc, ']');
318
319 --enc->indent;
320 encode_indent (enc); encode_ch (enc, ']');
321} 400}
322 401
323static void 402static void
324encode_he (enc_t *enc, HE *he) 403encode_hk (enc_t *enc, HE *he)
325{ 404{
326 encode_ch (enc, '"'); 405 encode_ch (enc, '"');
327 406
328 if (HeKLEN (he) == HEf_SVKEY) 407 if (HeKLEN (he) == HEf_SVKEY)
329 { 408 {
… …
339 else 418 else
340 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 419 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
341 420
342 encode_ch (enc, '"'); 421 encode_ch (enc, '"');
343 422
344 if (enc->flags & F_SPACE_BEFORE) encode_space (enc); 423 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
345 encode_ch (enc, ':'); 424 encode_ch (enc, ':');
346 if (enc->flags & F_SPACE_AFTER ) encode_space (enc); 425 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
347 encode_sv (enc, HeVAL (he));
348} 426}
349 427
350// compare hash entries, used when all keys are bytestrings 428// compare hash entries, used when all keys are bytestrings
351static int 429static int
352he_cmp_fast (const void *a_, const void *b_) 430he_cmp_fast (const void *a_, const void *b_)
… …
357 HE *b = *(HE **)b_; 435 HE *b = *(HE **)b_;
358 436
359 STRLEN la = HeKLEN (a); 437 STRLEN la = HeKLEN (a);
360 STRLEN lb = HeKLEN (b); 438 STRLEN lb = HeKLEN (b);
361 439
362 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 440 if (!(cmp = memcmp (HeKEY (b), HeKEY (a), lb < la ? lb : la)))
363 cmp = la - lb; 441 cmp = lb - la;
364 442
365 return cmp; 443 return cmp;
366} 444}
367 445
368// compare hash entries, used when some keys are sv's or utf-x 446// compare hash entries, used when some keys are sv's or utf-x
369static int 447static int
370he_cmp_slow (const void *a, const void *b) 448he_cmp_slow (const void *a, const void *b)
371{ 449{
372 return sv_cmp (HeSVKEY_force (*(HE **)a), HeSVKEY_force (*(HE **)b)); 450 return sv_cmp (HeSVKEY_force (*(HE **)b), HeSVKEY_force (*(HE **)a));
373} 451}
374 452
375static void 453static void
376encode_hv (enc_t *enc, HV *hv) 454encode_hv (enc_t *enc, HV *hv)
377{ 455{
378 int count, i; 456 HE *he;
379 457
380 if (enc->indent >= enc->maxdepth) 458 if (enc->indent >= enc->json.max_depth)
381 croak ("data structure too deep (hit recursion limit)"); 459 croak (ERR_NESTING_EXCEEDED);
382 460
383 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 461 encode_ch (enc, '{');
384 462
385 if ((count = hv_iterinit (hv)))
386 {
387 // for canonical output we have to sort by keys first 463 // for canonical output we have to sort by keys first
388 // actually, this is mostly due to the stupid so-called 464 // actually, this is mostly due to the stupid so-called
389 // security workaround added somewhere in 5.8.x. 465 // security workaround added somewhere in 5.8.x.
390 // that randomises hash orderings 466 // that randomises hash orderings
391 if (enc->flags & F_CANONICAL) 467 if (enc->json.flags & F_CANONICAL)
468 {
469 int count = hv_iterinit (hv);
470
471 if (SvMAGICAL (hv))
392 { 472 {
473 // need to count by iterating. could improve by dynamically building the vector below
474 // but I don't care for the speed of this special case.
475 // note also that we will run into undefined behaviour when the two iterations
476 // do not result in the same count, something I might care for in some later release.
477
478 count = 0;
479 while (hv_iternext (hv))
480 ++count;
481
482 hv_iterinit (hv);
483 }
484
485 if (count)
486 {
393 int fast = 1; 487 int i, fast = 1;
394 HE *he;
395#if defined(__BORLANDC__) || defined(_MSC_VER) 488#if defined(__BORLANDC__) || defined(_MSC_VER)
396 HE **hes = _alloca (count * sizeof (HE)); 489 HE **hes = _alloca (count * sizeof (HE));
397#else 490#else
398 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode 491 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
399#endif 492#endif
… …
426 519
427 FREETMPS; 520 FREETMPS;
428 LEAVE; 521 LEAVE;
429 } 522 }
430 523
431 for (i = 0; i < count; ++i) 524 encode_nl (enc); ++enc->indent;
525
526 while (count--)
432 { 527 {
433 encode_indent (enc); 528 encode_indent (enc);
529 he = hes [count];
434 encode_he (enc, hes [i]); 530 encode_hk (enc, he);
531 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
435 532
436 if (i < count - 1) 533 if (count)
437 encode_comma (enc); 534 encode_comma (enc);
438 } 535 }
439 536
537 encode_nl (enc); --enc->indent; encode_indent (enc);
538 }
539 }
540 else
541 {
542 if (hv_iterinit (hv) || SvMAGICAL (hv))
543 if ((he = hv_iternext (hv)))
544 {
545 encode_nl (enc); ++enc->indent;
546
547 for (;;)
548 {
440 encode_nl (enc); 549 encode_indent (enc);
550 encode_hk (enc, he);
551 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
552
553 if (!(he = hv_iternext (hv)))
554 break;
555
556 encode_comma (enc);
557 }
558
559 encode_nl (enc); --enc->indent; encode_indent (enc);
560 }
561 }
562
563 encode_ch (enc, '}');
564}
565
566// encode objects, arrays and special \0=false and \1=true values.
567static void
568encode_rv (enc_t *enc, SV *sv)
569{
570 svtype svt;
571
572 SvGETMAGIC (sv);
573 svt = SvTYPE (sv);
574
575 if (expect_false (SvOBJECT (sv)))
576 {
577 HV *stash = !JSON_SLOW || json_boolean_stash
578 ? json_boolean_stash
579 : gv_stashpv ("JSON::XS::Boolean", 1);
580
581 if (SvSTASH (sv) == stash)
582 {
583 if (SvIV (sv))
584 encode_str (enc, "true", 4, 0);
585 else
586 encode_str (enc, "false", 5, 0);
441 } 587 }
442 else 588 else
443 { 589 {
444 HE *he = hv_iternext (hv); 590#if 0
445 591 if (0 && sv_derived_from (rv, "JSON::Literal"))
446 for (;;)
447 { 592 {
448 encode_indent (enc); 593 // not yet
449 encode_he (enc, he);
450
451 if (!(he = hv_iternext (hv)))
452 break;
453
454 encode_comma (enc);
455 } 594 }
595#endif
596 if (enc->json.flags & F_CONV_BLESSED)
597 {
598 // we re-bless the reference to get overload and other niceties right
599 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
456 600
601 if (to_json)
602 {
603 dSP;
604
605 ENTER; SAVETMPS; PUSHMARK (SP);
606 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
607
608 // calling with G_SCALAR ensures that we always get a 1 return value
609 PUTBACK;
610 call_sv ((SV *)GvCV (to_json), G_SCALAR);
611 SPAGAIN;
612
613 // catch this surprisingly common error
614 if (SvROK (TOPs) && SvRV (TOPs) == sv)
615 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
616
617 sv = POPs;
618 PUTBACK;
619
457 encode_nl (enc); 620 encode_sv (enc, sv);
621
622 FREETMPS; LEAVE;
623 }
624 else if (enc->json.flags & F_ALLOW_BLESSED)
625 encode_str (enc, "null", 4, 0);
626 else
627 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
628 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
629 }
630 else if (enc->json.flags & F_ALLOW_BLESSED)
631 encode_str (enc, "null", 4, 0);
632 else
633 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",
634 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
458 } 635 }
459 } 636 }
460
461 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
462}
463
464// encode objects, arrays and special \0=false and \1=true values.
465static void
466encode_rv (enc_t *enc, SV *sv)
467{
468 svtype svt;
469
470 SvGETMAGIC (sv);
471 svt = SvTYPE (sv);
472
473 if (svt == SVt_PVHV) 637 else if (svt == SVt_PVHV)
474 encode_hv (enc, (HV *)sv); 638 encode_hv (enc, (HV *)sv);
475 else if (svt == SVt_PVAV) 639 else if (svt == SVt_PVAV)
476 encode_av (enc, (AV *)sv); 640 encode_av (enc, (AV *)sv);
477 else if (svt < SVt_PVAV) 641 else if (svt < SVt_PVAV)
478 { 642 {
479 if (SvNIOK (sv) && SvIV (sv) == 0) 643 STRLEN len = 0;
644 char *pv = svt ? SvPV (sv, len) : 0;
645
646 if (len == 1 && *pv == '1')
647 encode_str (enc, "true", 4, 0);
648 else if (len == 1 && *pv == '0')
480 encode_str (enc, "false", 5, 0); 649 encode_str (enc, "false", 5, 0);
481 else if (SvNIOK (sv) && SvIV (sv) == 1) 650 else if (enc->json.flags & F_ALLOW_UNKNOWN)
482 encode_str (enc, "true", 4, 0); 651 encode_str (enc, "null", 4, 0);
483 else 652 else
484 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", 653 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
485 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 654 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
486 } 655 }
656 else if (enc->json.flags & F_ALLOW_UNKNOWN)
657 encode_str (enc, "null", 4, 0);
487 else 658 else
488 croak ("encountered %s, but JSON can only represent references to arrays or hashes", 659 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
489 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 660 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
490} 661}
491 662
… …
509 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 680 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
510 enc->cur += strlen (enc->cur); 681 enc->cur += strlen (enc->cur);
511 } 682 }
512 else if (SvIOKp (sv)) 683 else if (SvIOKp (sv))
513 { 684 {
514 // we assume we can always read an IV as a UV 685 // we assume we can always read an IV as a UV and vice versa
515 if (SvUV (sv) & ~(UV)0x7fff) 686 // we assume two's complement
516 { 687 // we assume no aliasing issues in the union
517 // large integer, use the (rather slow) snprintf way. 688 if (SvIsUV (sv) ? SvUVX (sv) <= 59000
518 need (enc, sizeof (UV) * 3); 689 : SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000)
519 enc->cur +=
520 SvIsUV(sv)
521 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
522 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
523 }
524 else
525 { 690 {
526 // optimise the "small number case" 691 // optimise the "small number case"
527 // code will likely be branchless and use only a single multiplication 692 // code will likely be branchless and use only a single multiplication
693 // works for numbers up to 59074
528 I32 i = SvIV (sv); 694 I32 i = SvIVX (sv);
529 U32 u; 695 U32 u;
530 char digit, nz = 0; 696 char digit, nz = 0;
531 697
532 need (enc, 6); 698 need (enc, 6);
533 699
… …
539 705
540 // now output digit by digit, each time masking out the integer part 706 // now output digit by digit, each time masking out the integer part
541 // and multiplying by 5 while moving the decimal point one to the right, 707 // and multiplying by 5 while moving the decimal point one to the right,
542 // resulting in a net multiplication by 10. 708 // resulting in a net multiplication by 10.
543 // we always write the digit to memory but conditionally increment 709 // we always write the digit to memory but conditionally increment
544 // the pointer, to ease the usage of conditional move instructions. 710 // the pointer, to enable the use of conditional move instructions.
545 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5; 711 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffffUL) * 5;
546 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5; 712 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffffUL) * 5;
547 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5; 713 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffffUL) * 5;
548 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5; 714 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffffUL) * 5;
549 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0' 715 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
550 } 716 }
717 else
718 {
719 // large integer, use the (rather slow) snprintf way.
720 need (enc, IVUV_MAXCHARS);
721 enc->cur +=
722 SvIsUV(sv)
723 ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))
724 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
725 }
551 } 726 }
552 else if (SvROK (sv)) 727 else if (SvROK (sv))
553 encode_rv (enc, SvRV (sv)); 728 encode_rv (enc, SvRV (sv));
554 else if (!SvOK (sv)) 729 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN)
555 encode_str (enc, "null", 4, 0); 730 encode_str (enc, "null", 4, 0);
556 else 731 else
557 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 732 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
558 SvPV_nolen (sv), SvFLAGS (sv)); 733 SvPV_nolen (sv), SvFLAGS (sv));
559} 734}
560 735
561static SV * 736static SV *
562encode_json (SV *scalar, U32 flags) 737encode_json (SV *scalar, JSON *json)
563{ 738{
564 enc_t enc; 739 enc_t enc;
565 740
566 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 741 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
567 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 742 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
568 743
569 enc.flags = flags; 744 enc.json = *json;
570 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 745 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
571 enc.cur = SvPVX (enc.sv); 746 enc.cur = SvPVX (enc.sv);
572 enc.end = SvEND (enc.sv); 747 enc.end = SvEND (enc.sv);
573 enc.indent = 0; 748 enc.indent = 0;
574 enc.maxdepth = DEC_DEPTH (flags); 749 enc.limit = enc.json.flags & F_ASCII ? 0x000080UL
750 : enc.json.flags & F_LATIN1 ? 0x000100UL
751 : 0x110000UL;
575 752
576 SvPOK_only (enc.sv); 753 SvPOK_only (enc.sv);
577 encode_sv (&enc, scalar); 754 encode_sv (&enc, scalar);
578 755
579 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 756 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
580 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 757 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
581 758
582 if (!(flags & (F_ASCII | F_LATIN1 | F_UTF8))) 759 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
583 SvUTF8_on (enc.sv); 760 SvUTF8_on (enc.sv);
584 761
585 if (enc.flags & F_SHRINK) 762 if (enc.json.flags & F_SHRINK)
586 shrink (enc.sv); 763 shrink (enc.sv);
587 764
588 return enc.sv; 765 return enc.sv;
589} 766}
590 767
… …
595typedef struct 772typedef struct
596{ 773{
597 char *cur; // current parser pointer 774 char *cur; // current parser pointer
598 char *end; // end of input string 775 char *end; // end of input string
599 const char *err; // parse error, if != 0 776 const char *err; // parse error, if != 0
600 U32 flags; // F_* 777 JSON json;
601 U32 depth; // recursion depth 778 U32 depth; // recursion depth
602 U32 maxdepth; // recursion depth limit 779 U32 maxdepth; // recursion depth limit
603} dec_t; 780} dec_t;
604 781
605inline void 782INLINE void
783decode_comment (dec_t *dec)
784{
785 // only '#'-style comments allowed a.t.m.
786
787 while (*dec->cur && *dec->cur != 0x0a && *dec->cur != 0x0d)
788 ++dec->cur;
789}
790
791INLINE void
606decode_ws (dec_t *dec) 792decode_ws (dec_t *dec)
607{ 793{
608 for (;;) 794 for (;;)
609 { 795 {
610 char ch = *dec->cur; 796 char ch = *dec->cur;
611 797
612 if (ch > 0x20 798 if (ch > 0x20)
799 {
800 if (expect_false (ch == '#'))
801 {
802 if (dec->json.flags & F_RELAXED)
803 decode_comment (dec);
804 else
805 break;
806 }
807 else
808 break;
809 }
613 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) 810 else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)
614 break; 811 break; // parse error, but let higher level handle it, gives better error messages
615 812
616 ++dec->cur; 813 ++dec->cur;
617 } 814 }
618} 815}
619 816
… …
623 if (*dec->cur != ch) \ 820 if (*dec->cur != ch) \
624 ERR (# ch " expected"); \ 821 ERR (# ch " expected"); \
625 ++dec->cur; \ 822 ++dec->cur; \
626 SE 823 SE
627 824
628#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)") 825#define DEC_INC_DEPTH if (++dec->depth > dec->json.max_depth) ERR (ERR_NESTING_EXCEEDED)
629#define DEC_DEC_DEPTH --dec->depth 826#define DEC_DEC_DEPTH --dec->depth
630 827
631static SV *decode_sv (dec_t *dec); 828static SV *decode_sv (dec_t *dec);
632 829
633static signed char decode_hexdigit[256]; 830static signed char decode_hexdigit[256];
… …
725 922
726 if (hi >= 0x80) 923 if (hi >= 0x80)
727 { 924 {
728 utf8 = 1; 925 utf8 = 1;
729 926
730 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0); 927 cur = encode_utf8 (cur, hi);
731 } 928 }
732 else 929 else
733 *cur++ = hi; 930 *cur++ = hi;
734 } 931 }
735 break; 932 break;
… …
737 default: 934 default:
738 --dec_cur; 935 --dec_cur;
739 ERR ("illegal backslash escape sequence in string"); 936 ERR ("illegal backslash escape sequence in string");
740 } 937 }
741 } 938 }
742 else if (expect_true (ch >= 0x20 && ch <= 0x7f)) 939 else if (expect_true (ch >= 0x20 && ch < 0x80))
743 *cur++ = ch; 940 *cur++ = ch;
744 else if (ch >= 0x80) 941 else if (ch >= 0x80)
745 { 942 {
746 STRLEN clen; 943 STRLEN clen;
747 UV uch; 944 UV uch;
… …
868 is_nv = 1; 1065 is_nv = 1;
869 } 1066 }
870 1067
871 if (!is_nv) 1068 if (!is_nv)
872 { 1069 {
1070 int len = dec->cur - start;
1071
873 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so 1072 // special case the rather common 1..5-digit-int case
874 if (*start == '-') 1073 if (*start == '-')
875 switch (dec->cur - start) 1074 switch (len)
876 { 1075 {
877 case 2: return newSViv (-( start [1] - '0' * 1)); 1076 case 2: return newSViv (-( start [1] - '0' * 1));
878 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); 1077 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
879 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); 1078 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
880 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); 1079 case 5: return newSViv (-( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
1080 case 6: return newSViv (-(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
881 } 1081 }
882 else 1082 else
883 switch (dec->cur - start) 1083 switch (len)
884 { 1084 {
885 case 1: return newSViv ( start [0] - '0' * 1); 1085 case 1: return newSViv ( start [0] - '0' * 1);
886 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); 1086 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
887 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); 1087 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
888 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); 1088 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
1089 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
889 } 1090 }
890 1091
891 { 1092 {
892 UV uv; 1093 UV uv;
893 int numtype = grok_number (start, dec->cur - start, &uv); 1094 int numtype = grok_number (start, len, &uv);
894 if (numtype & IS_NUMBER_IN_UV) 1095 if (numtype & IS_NUMBER_IN_UV)
895 if (numtype & IS_NUMBER_NEG) 1096 if (numtype & IS_NUMBER_NEG)
896 { 1097 {
897 if (uv < (UV)IV_MIN) 1098 if (uv < (UV)IV_MIN)
898 return newSViv (-(IV)uv); 1099 return newSViv (-(IV)uv);
899 } 1100 }
900 else 1101 else
901 return newSVuv (uv); 1102 return newSVuv (uv);
902
903 // here would likely be the place for bigint support
904 } 1103 }
905 }
906 1104
907 // if we ever support bigint or bigfloat, this is the place for bigfloat 1105 len -= *start == '-' ? 1 : 0;
1106
1107 // does not fit into IV or UV, try NV
1108 if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len)
1109 #if defined (LDBL_DIG)
1110 || (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len)
1111 #endif
1112 )
1113 // fits into NV without loss of precision
1114 return newSVnv (Atof (start));
1115
1116 // everything else fails, convert it to a string
1117 return newSVpvn (start, dec->cur - start);
1118 }
1119
1120 // loss of precision here
908 return newSVnv (Atof (start)); 1121 return newSVnv (Atof (start));
909 1122
910fail: 1123fail:
911 return 0; 1124 return 0;
912} 1125}
… …
942 1155
943 if (*dec->cur != ',') 1156 if (*dec->cur != ',')
944 ERR (", or ] expected while parsing array"); 1157 ERR (", or ] expected while parsing array");
945 1158
946 ++dec->cur; 1159 ++dec->cur;
1160
1161 decode_ws (dec);
1162
1163 if (*dec->cur == ']' && dec->json.flags & F_RELAXED)
1164 {
1165 ++dec->cur;
1166 break;
1167 }
947 } 1168 }
948 1169
949 DEC_DEC_DEPTH; 1170 DEC_DEC_DEPTH;
950 return newRV_noinc ((SV *)av); 1171 return newRV_noinc ((SV *)av);
951 1172
… …
956} 1177}
957 1178
958static SV * 1179static SV *
959decode_hv (dec_t *dec) 1180decode_hv (dec_t *dec)
960{ 1181{
1182 SV *sv;
961 HV *hv = newHV (); 1183 HV *hv = newHV ();
962 1184
963 DEC_INC_DEPTH; 1185 DEC_INC_DEPTH;
964 decode_ws (dec); 1186 decode_ws (dec);
965 1187
966 if (*dec->cur == '}') 1188 if (*dec->cur == '}')
967 ++dec->cur; 1189 ++dec->cur;
968 else 1190 else
969 for (;;) 1191 for (;;)
970 { 1192 {
1193 EXPECT_CH ('"');
1194
1195 // heuristic: assume that
1196 // a) decode_str + hv_store_ent are abysmally slow.
1197 // b) most hash keys are short, simple ascii text.
1198 // => try to "fast-match" such strings to avoid
1199 // the overhead of decode_str + hv_store_ent.
1200 {
971 SV *key, *value; 1201 SV *value;
1202 char *p = dec->cur;
1203 char *e = p + 24; // only try up to 24 bytes
972 1204
973 decode_ws (dec); EXPECT_CH ('"'); 1205 for (;;)
974
975 key = decode_str (dec);
976 if (!key)
977 goto fail;
978
979 decode_ws (dec); EXPECT_CH (':');
980
981 value = decode_sv (dec);
982 if (!value)
983 { 1206 {
1207 // the >= 0x80 is false on most architectures
1208 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1209 {
1210 // slow path, back up and use decode_str
1211 SV *key = decode_str (dec);
1212 if (!key)
1213 goto fail;
1214
1215 decode_ws (dec); EXPECT_CH (':');
1216
1217 decode_ws (dec);
1218 value = decode_sv (dec);
1219 if (!value)
1220 {
1221 SvREFCNT_dec (key);
1222 goto fail;
1223 }
1224
1225 hv_store_ent (hv, key, value, 0);
984 SvREFCNT_dec (key); 1226 SvREFCNT_dec (key);
1227
1228 break;
1229 }
1230 else if (*p == '"')
1231 {
1232 // fast path, got a simple key
1233 char *key = dec->cur;
1234 int len = p - key;
1235 dec->cur = p + 1;
1236
1237 decode_ws (dec); EXPECT_CH (':');
1238
1239 decode_ws (dec);
1240 value = decode_sv (dec);
1241 if (!value)
985 goto fail; 1242 goto fail;
1243
1244 hv_store (hv, key, len, value, 0);
1245
1246 break;
1247 }
1248
1249 ++p;
986 } 1250 }
987 1251 }
988 hv_store_ent (hv, key, value, 0);
989 SvREFCNT_dec (key);
990 1252
991 decode_ws (dec); 1253 decode_ws (dec);
992 1254
993 if (*dec->cur == '}') 1255 if (*dec->cur == '}')
994 { 1256 {
… …
998 1260
999 if (*dec->cur != ',') 1261 if (*dec->cur != ',')
1000 ERR (", or } expected while parsing object/hash"); 1262 ERR (", or } expected while parsing object/hash");
1001 1263
1002 ++dec->cur; 1264 ++dec->cur;
1265
1266 decode_ws (dec);
1267
1268 if (*dec->cur == '}' && dec->json.flags & F_RELAXED)
1269 {
1270 ++dec->cur;
1271 break;
1272 }
1003 } 1273 }
1004 1274
1005 DEC_DEC_DEPTH; 1275 DEC_DEC_DEPTH;
1006 return newRV_noinc ((SV *)hv); 1276 sv = newRV_noinc ((SV *)hv);
1277
1278 // check filter callbacks
1279 if (dec->json.flags & F_HOOK)
1280 {
1281 if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
1282 {
1283 HE *cb, *he;
1284
1285 hv_iterinit (hv);
1286 he = hv_iternext (hv);
1287 hv_iterinit (hv);
1288
1289 // the next line creates a mortal sv each time its called.
1290 // might want to optimise this for common cases.
1291 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1292
1293 if (cb)
1294 {
1295 dSP;
1296 int count;
1297
1298 ENTER; SAVETMPS; PUSHMARK (SP);
1299 XPUSHs (HeVAL (he));
1300
1301 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1302
1303 if (count == 1)
1304 {
1305 sv = newSVsv (POPs);
1306 FREETMPS; LEAVE;
1307 return sv;
1308 }
1309
1310 FREETMPS; LEAVE;
1311 }
1312 }
1313
1314 if (dec->json.cb_object)
1315 {
1316 dSP;
1317 int count;
1318
1319 ENTER; SAVETMPS; PUSHMARK (SP);
1320 XPUSHs (sv_2mortal (sv));
1321
1322 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1323
1324 if (count == 1)
1325 {
1326 sv = newSVsv (POPs);
1327 FREETMPS; LEAVE;
1328 return sv;
1329 }
1330
1331 SvREFCNT_inc (sv);
1332 FREETMPS; LEAVE;
1333 }
1334 }
1335
1336 return sv;
1007 1337
1008fail: 1338fail:
1009 SvREFCNT_dec (hv); 1339 SvREFCNT_dec (hv);
1010 DEC_DEC_DEPTH; 1340 DEC_DEC_DEPTH;
1011 return 0; 1341 return 0;
1012} 1342}
1013 1343
1014static SV * 1344static SV *
1015decode_sv (dec_t *dec) 1345decode_sv (dec_t *dec)
1016{ 1346{
1017 decode_ws (dec);
1018
1019 // the beauty of JSON: you need exactly one character lookahead 1347 // the beauty of JSON: you need exactly one character lookahead
1020 // to parse anything. 1348 // to parse everything.
1021 switch (*dec->cur) 1349 switch (*dec->cur)
1022 { 1350 {
1023 case '"': ++dec->cur; return decode_str (dec); 1351 case '"': ++dec->cur; return decode_str (dec);
1024 case '[': ++dec->cur; return decode_av (dec); 1352 case '[': ++dec->cur; return decode_av (dec);
1025 case '{': ++dec->cur; return decode_hv (dec); 1353 case '{': ++dec->cur; return decode_hv (dec);
1026 1354
1027 case '-': 1355 case '-':
1028 case '0': case '1': case '2': case '3': case '4': 1356 case '0': case '1': case '2': case '3': case '4':
1029 case '5': case '6': case '7': case '8': case '9': 1357 case '5': case '6': case '7': case '8': case '9':
1030 return decode_num (dec); 1358 return decode_num (dec);
1031 1359
1032 case 't': 1360 case 't':
1033 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1361 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1034 { 1362 {
1035 dec->cur += 4; 1363 dec->cur += 4;
1364#if JSON_SLOW
1365 json_true = get_bool ("JSON::XS::true");
1366#endif
1036 return newSViv (1); 1367 return newSVsv (json_true);
1037 } 1368 }
1038 else 1369 else
1039 ERR ("'true' expected"); 1370 ERR ("'true' expected");
1040 1371
1041 break; 1372 break;
1042 1373
1043 case 'f': 1374 case 'f':
1044 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1375 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1045 { 1376 {
1046 dec->cur += 5; 1377 dec->cur += 5;
1378#if JSON_SLOW
1379 json_false = get_bool ("JSON::XS::false");
1380#endif
1047 return newSViv (0); 1381 return newSVsv (json_false);
1048 } 1382 }
1049 else 1383 else
1050 ERR ("'false' expected"); 1384 ERR ("'false' expected");
1051 1385
1052 break; 1386 break;
… …
1070fail: 1404fail:
1071 return 0; 1405 return 0;
1072} 1406}
1073 1407
1074static SV * 1408static SV *
1075decode_json (SV *string, U32 flags, UV *offset_return) 1409decode_json (SV *string, JSON *json, STRLEN *offset_return)
1076{ 1410{
1077 dec_t dec; 1411 dec_t dec;
1078 UV offset; 1412 STRLEN offset;
1079 SV *sv; 1413 SV *sv;
1080 1414
1081 SvGETMAGIC (string); 1415 SvGETMAGIC (string);
1082 SvUPGRADE (string, SVt_PV); 1416 SvUPGRADE (string, SVt_PV);
1083 1417
1418 if (SvCUR (string) > json->max_size && json->max_size)
1419 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1420 (unsigned long)SvCUR (string), (unsigned long)json->max_size);
1421
1084 if (flags & F_UTF8) 1422 if (json->flags & F_UTF8)
1085 sv_utf8_downgrade (string, 0); 1423 sv_utf8_downgrade (string, 0);
1086 else 1424 else
1087 sv_utf8_upgrade (string); 1425 sv_utf8_upgrade (string);
1088 1426
1089 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1427 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
1090 1428
1091 dec.flags = flags; 1429 dec.json = *json;
1092 dec.cur = SvPVX (string); 1430 dec.cur = SvPVX (string);
1093 dec.end = SvEND (string); 1431 dec.end = SvEND (string);
1094 dec.err = 0; 1432 dec.err = 0;
1095 dec.depth = 0; 1433 dec.depth = 0;
1096 dec.maxdepth = DEC_DEPTH (dec.flags); 1434
1435 if (dec.json.cb_object || dec.json.cb_sk_object)
1436 dec.json.flags |= F_HOOK;
1097 1437
1098 *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);
1099 sv = decode_sv (&dec); 1441 sv = decode_sv (&dec);
1100 1442
1101 if (!(offset_return || !sv)) 1443 if (!(offset_return || !sv))
1102 { 1444 {
1103 // check for trailing garbage 1445 // check for trailing garbage
… …
1111 } 1453 }
1112 } 1454 }
1113 1455
1114 if (offset_return || !sv) 1456 if (offset_return || !sv)
1115 { 1457 {
1116 offset = dec.flags & F_UTF8 1458 offset = dec.json.flags & F_UTF8
1117 ? dec.cur - SvPVX (string) 1459 ? dec.cur - SvPVX (string)
1118 : utf8_distance (dec.cur, SvPVX (string)); 1460 : utf8_distance (dec.cur, SvPVX (string));
1119 1461
1120 if (offset_return) 1462 if (offset_return)
1121 *offset_return = offset; 1463 *offset_return = offset;
… …
1140 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1482 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1141 } 1483 }
1142 1484
1143 sv = sv_2mortal (sv); 1485 sv = sv_2mortal (sv);
1144 1486
1145 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1487 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))
1146 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); 1488 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1147 1489
1148 return sv; 1490 return sv;
1491}
1492
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 if (++self->incr_nest > self->max_depth)
1590 croak (ERR_NESTING_EXCEEDED);
1591 break;
1592
1593 case ']':
1594 case '}':
1595 if (!--self->incr_nest)
1596 goto interrupt;
1597 }
1598 }
1599 }
1600
1601 modechange:
1602 ;
1603 }
1604
1605interrupt:
1606 self->incr_pos = p - SvPVX (self->incr_text);
1607 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D
1149} 1608}
1150 1609
1151///////////////////////////////////////////////////////////////////////////// 1610/////////////////////////////////////////////////////////////////////////////
1152// XS interface functions 1611// XS interface functions
1153 1612
… …
1162 i >= '0' && i <= '9' ? i - '0' 1621 i >= '0' && i <= '9' ? i - '0'
1163 : i >= 'a' && i <= 'f' ? i - 'a' + 10 1622 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1164 : i >= 'A' && i <= 'F' ? i - 'A' + 10 1623 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1165 : -1; 1624 : -1;
1166 1625
1167 json_stash = gv_stashpv ("JSON::XS", 1); 1626 json_stash = gv_stashpv ("JSON::XS" , 1);
1627 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1628
1629 json_true = get_bool ("JSON::XS::true");
1630 json_false = get_bool ("JSON::XS::false");
1168} 1631}
1169 1632
1170PROTOTYPES: DISABLE 1633PROTOTYPES: DISABLE
1171 1634
1172SV *new (char *dummy) 1635void CLONE (...)
1173 CODE: 1636 CODE:
1174 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1637 json_stash = 0;
1638 json_boolean_stash = 0;
1639
1640void new (char *klass)
1641 PPCODE:
1642{
1643 SV *pv = NEWSV (0, sizeof (JSON));
1644 SvPOK_only (pv);
1645 json_init ((JSON *)SvPVX (pv));
1646 XPUSHs (sv_2mortal (sv_bless (
1647 newRV_noinc (pv),
1648 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1649 )));
1650}
1651
1652void ascii (JSON *self, int enable = 1)
1653 ALIAS:
1654 ascii = F_ASCII
1655 latin1 = F_LATIN1
1656 utf8 = F_UTF8
1657 indent = F_INDENT
1658 canonical = F_CANONICAL
1659 space_before = F_SPACE_BEFORE
1660 space_after = F_SPACE_AFTER
1661 pretty = F_PRETTY
1662 allow_nonref = F_ALLOW_NONREF
1663 shrink = F_SHRINK
1664 allow_blessed = F_ALLOW_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, U32 max_depth = 0x80000000UL)
1697 PPCODE:
1698 self->max_depth = max_depth;
1699 XPUSHs (ST (0));
1700
1701U32 get_max_depth (JSON *self)
1702 CODE:
1703 RETVAL = self->max_depth;
1175 OUTPUT: 1704 OUTPUT:
1176 RETVAL 1705 RETVAL
1177 1706
1178SV *ascii (SV *self, int enable = 1) 1707void max_size (JSON *self, U32 max_size = 0)
1179 ALIAS: 1708 PPCODE:
1180 ascii = F_ASCII 1709 self->max_size = max_size;
1181 latin1 = F_LATIN1 1710 XPUSHs (ST (0));
1182 utf8 = F_UTF8 1711
1183 indent = F_INDENT 1712int get_max_size (JSON *self)
1184 canonical = F_CANONICAL
1185 space_before = F_SPACE_BEFORE
1186 space_after = F_SPACE_AFTER
1187 pretty = F_PRETTY
1188 allow_nonref = F_ALLOW_NONREF
1189 shrink = F_SHRINK
1190 CODE: 1713 CODE:
1191{ 1714 RETVAL = self->max_size;
1192 UV *uv = SvJSON (self);
1193 if (enable)
1194 *uv |= ix;
1195 else
1196 *uv &= ~ix;
1197
1198 RETVAL = newSVsv (self);
1199}
1200 OUTPUT: 1715 OUTPUT:
1201 RETVAL 1716 RETVAL
1202 1717
1203SV *max_depth (SV *self, UV max_depth = 0x80000000UL) 1718void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1719 PPCODE:
1720{
1721 SvREFCNT_dec (self->cb_object);
1722 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
1723
1724 XPUSHs (ST (0));
1725}
1726
1727void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
1728 PPCODE:
1729{
1730 if (!self->cb_sk_object)
1731 self->cb_sk_object = newHV ();
1732
1733 if (SvOK (cb))
1734 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1735 else
1736 {
1737 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1738
1739 if (!HvKEYS (self->cb_sk_object))
1740 {
1741 SvREFCNT_dec (self->cb_sk_object);
1742 self->cb_sk_object = 0;
1743 }
1744 }
1745
1746 XPUSHs (ST (0));
1747}
1748
1749void encode (JSON *self, SV *scalar)
1750 PPCODE:
1751 XPUSHs (encode_json (scalar, self));
1752
1753void decode (JSON *self, SV *jsonstr)
1754 PPCODE:
1755 XPUSHs (decode_json (jsonstr, self, 0));
1756
1757void decode_prefix (JSON *self, SV *jsonstr)
1758 PPCODE:
1759{
1760 STRLEN offset;
1761 EXTEND (SP, 2);
1762 PUSHs (decode_json (jsonstr, self, &offset));
1763 PUSHs (sv_2mortal (newSVuv (offset)));
1764}
1765
1766void incr_parse (JSON *self, SV *jsonstr = 0)
1767 PPCODE:
1768{
1769 if (!self->incr_text)
1770 self->incr_text = newSVpvn ("", 0);
1771
1772 // append data, if any
1773 if (jsonstr)
1774 {
1775 if (SvUTF8 (jsonstr) && !SvUTF8 (self->incr_text))
1776 {
1777 /* utf-8-ness differs, need to upgrade */
1778 sv_utf8_upgrade (self->incr_text);
1779
1780 if (self->incr_pos)
1781 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
1782 - (U8 *)SvPVX (self->incr_text);
1783 }
1784
1785 {
1786 STRLEN len;
1787 const char *str = SvPV (jsonstr, len);
1788 SvGROW (self->incr_text, SvCUR (self->incr_text) + len + 1);
1789 Move (str, SvEND (self->incr_text), len, char);
1790 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
1791 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
1792 }
1793 }
1794
1795 if (GIMME_V != G_VOID)
1796 do
1797 {
1798 STRLEN offset;
1799
1800 if (!INCR_DONE (self))
1801 {
1802 incr_parse (self);
1803
1804 if (self->incr_pos > self->max_size && self->max_size)
1805 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1806 (unsigned long)self->incr_pos, (unsigned long)self->max_size);
1807
1808 if (!INCR_DONE (self))
1809 break;
1810 }
1811
1812 XPUSHs (decode_json (self->incr_text, self, &offset));
1813
1814 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + offset);
1815 self->incr_pos -= offset;
1816 self->incr_nest = 0;
1817 self->incr_mode = 0;
1818 }
1819 while (GIMME_V == G_ARRAY);
1820}
1821
1822SV *incr_text (JSON *self)
1823 ATTRS: lvalue
1204 CODE: 1824 CODE:
1205{ 1825{
1206 UV *uv = SvJSON (self); 1826 if (self->incr_pos)
1207 UV log2 = 0; 1827 croak ("incr_text can not be called when the incremental parser already started parsing");
1208 1828
1209 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL; 1829 RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef;
1210
1211 while ((1UL << log2) < max_depth)
1212 ++log2;
1213
1214 *uv = *uv & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1215
1216 RETVAL = newSVsv (self);
1217} 1830}
1218 OUTPUT: 1831 OUTPUT:
1219 RETVAL 1832 RETVAL
1220 1833
1834void incr_skip (JSON *self)
1835 CODE:
1836{
1837 if (self->incr_pos)
1838 {
1839 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + self->incr_pos);
1840 self->incr_pos = 0;
1841 self->incr_nest = 0;
1842 self->incr_mode = 0;
1843 }
1844}
1845
1846void incr_reset (JSON *self)
1847 CODE:
1848{
1849 SvREFCNT_dec (self->incr_text);
1850 self->incr_text = 0;
1851 self->incr_pos = 0;
1852 self->incr_nest = 0;
1853 self->incr_mode = 0;
1854}
1855
1856void DESTROY (JSON *self)
1857 CODE:
1858 SvREFCNT_dec (self->cb_sk_object);
1859 SvREFCNT_dec (self->cb_object);
1860 SvREFCNT_dec (self->incr_text);
1861
1862PROTOTYPES: ENABLE
1863
1221void encode (SV *self, SV *scalar) 1864void encode_json (SV *scalar)
1865 ALIAS:
1866 to_json_ = 0
1867 encode_json = F_UTF8
1222 PPCODE: 1868 PPCODE:
1869{
1870 JSON json;
1871 json_init (&json);
1872 json.flags |= ix;
1223 XPUSHs (encode_json (scalar, *SvJSON (self))); 1873 XPUSHs (encode_json (scalar, &json));
1874}
1224 1875
1225void decode (SV *self, SV *jsonstr) 1876void decode_json (SV *jsonstr)
1877 ALIAS:
1878 from_json_ = 0
1879 decode_json = F_UTF8
1226 PPCODE: 1880 PPCODE:
1881{
1882 JSON json;
1883 json_init (&json);
1884 json.flags |= ix;
1227 XPUSHs (decode_json (jsonstr, *SvJSON (self), 0)); 1885 XPUSHs (decode_json (jsonstr, &json, 0));
1228
1229void decode_prefix (SV *self, SV *jsonstr)
1230 PPCODE:
1231{
1232 UV offset;
1233 EXTEND (SP, 2);
1234 PUSHs (decode_json (jsonstr, *SvJSON (self), &offset));
1235 PUSHs (sv_2mortal (newSVuv (offset)));
1236} 1886}
1237 1887
1238PROTOTYPES: ENABLE
1239 1888
1240void to_json (SV *scalar)
1241 ALIAS:
1242 objToJson = 0
1243 PPCODE:
1244 XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8));
1245
1246void from_json (SV *jsonstr)
1247 ALIAS:
1248 jsonToObj = 0
1249 PPCODE:
1250 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8, 0));
1251

Diff Legend

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