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 1.41 by root, Sat Jun 23 22:53:16 2007 UTC vs.
Revision 1.90 by root, Sun Jul 20 17:55:19 2008 UTC

1#include "EXTERN.h" 1#include "EXTERN.h"
2#include "perl.h" 2#include "perl.h"
3#include "XSUB.h" 3#include "XSUB.h"
4 4
5#include "assert.h" 5#include <assert.h>
6#include "string.h" 6#include <string.h>
7#include "stdlib.h" 7#include <stdlib.h>
8#include "stdio.h" 8#include <stdio.h>
9#include <limits.h>
10#include <float.h>
9 11
10#if defined(__BORLANDC__) || defined(_MSC_VER) 12#if defined(__BORLANDC__) || defined(_MSC_VER)
11# define snprintf _snprintf // C compilers have this in stdio.h 13# define snprintf _snprintf // C compilers have this in stdio.h
12#endif 14#endif
13 15
14// some old perls do not have this, try to make it work, no 16// some old perls do not have this, try to make it work, no
15// guarentees, though. 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 <= 0 && (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 int 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);
126
77 if (SvLEN (sv) > SvCUR (sv) + 1) 127 if (SvLEN (sv) > SvCUR (sv) + 1)
78 { 128 {
79#ifdef SvPV_shrink_to_cur 129#ifdef SvPV_shrink_to_cur
80 SvPV_shrink_to_cur (sv); 130 SvPV_shrink_to_cur (sv);
81#elif defined (SvPV_renew) 131#elif defined (SvPV_renew)
… …
87// decode an utf-8 character and return it, or (UV)-1 in 137// decode an utf-8 character and return it, or (UV)-1 in
88// case of an error. 138// case of an error.
89// we special-case "safe" characters from U+80 .. U+7FF, 139// we special-case "safe" characters from U+80 .. U+7FF,
90// but use the very good perl function to parse anything else. 140// but use the very good perl function to parse anything else.
91// note that we never call this function for a ascii codepoints 141// note that we never call this function for a ascii codepoints
92inline UV 142INLINE UV
93decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) 143decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
94{ 144{
95 if (expect_false (s[0] > 0xdf || s[0] < 0xc2)) 145 if (expect_true (len >= 2
96 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); 146 && IN_RANGE_INC (char, s[0], 0xc2, 0xdf)
97 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) 147 && IN_RANGE_INC (char, s[1], 0x80, 0xbf)))
98 { 148 {
99 *clen = 2; 149 *clen = 2;
100 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 150 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
101 } 151 }
102 else 152 else
103 { 153 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
104 *clen = (STRLEN)-1; 154}
105 return (UV)-1; 155
106 } 156// likewise for encoding, also never called for ascii codepoints
157// this function takes advantage of this fact, although current gccs
158// seem to optimise the check for >= 0x80 away anyways
159INLINE unsigned char *
160encode_utf8 (unsigned char *s, UV ch)
161{
162 if (expect_false (ch < 0x000080))
163 *s++ = ch;
164 else if (expect_true (ch < 0x000800))
165 *s++ = 0xc0 | ( ch >> 6),
166 *s++ = 0x80 | ( ch & 0x3f);
167 else if ( ch < 0x010000)
168 *s++ = 0xe0 | ( ch >> 12),
169 *s++ = 0x80 | ((ch >> 6) & 0x3f),
170 *s++ = 0x80 | ( ch & 0x3f);
171 else if ( ch < 0x110000)
172 *s++ = 0xf0 | ( ch >> 18),
173 *s++ = 0x80 | ((ch >> 12) & 0x3f),
174 *s++ = 0x80 | ((ch >> 6) & 0x3f),
175 *s++ = 0x80 | ( ch & 0x3f);
176
177 return s;
107} 178}
108 179
109///////////////////////////////////////////////////////////////////////////// 180/////////////////////////////////////////////////////////////////////////////
110// encoder 181// encoder
111 182
… …
113typedef struct 184typedef struct
114{ 185{
115 char *cur; // SvPVX (sv) + current output position 186 char *cur; // SvPVX (sv) + current output position
116 char *end; // SvEND (sv) 187 char *end; // SvEND (sv)
117 SV *sv; // result scalar 188 SV *sv; // result scalar
118 U32 flags; // F_* 189 JSON json;
119 U32 indent; // indentation level 190 U32 indent; // indentation level
120 U32 maxdepth; // max. indentation/recursion level 191 UV limit; // escape character values >= this value when encoding
121} enc_t; 192} enc_t;
122 193
123inline void 194INLINE void
124need (enc_t *enc, STRLEN len) 195need (enc_t *enc, STRLEN len)
125{ 196{
126 if (expect_false (enc->cur + len >= enc->end)) 197 if (expect_false (enc->cur + len >= enc->end))
127 { 198 {
128 STRLEN cur = enc->cur - SvPVX (enc->sv); 199 STRLEN cur = enc->cur - SvPVX (enc->sv);
… …
130 enc->cur = SvPVX (enc->sv) + cur; 201 enc->cur = SvPVX (enc->sv) + cur;
131 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 202 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
132 } 203 }
133} 204}
134 205
135inline void 206INLINE void
136encode_ch (enc_t *enc, char ch) 207encode_ch (enc_t *enc, char ch)
137{ 208{
138 need (enc, 1); 209 need (enc, 1);
139 *enc->cur++ = ch; 210 *enc->cur++ = ch;
140} 211}
… …
194 { 265 {
195 uch = ch; 266 uch = ch;
196 clen = 1; 267 clen = 1;
197 } 268 }
198 269
199 if (uch > 0x10FFFFUL) 270 if (uch < 0x80/*0x20*/ || uch >= enc->limit)
200 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
201
202 if (uch < 0x80 || enc->flags & F_ASCII || (enc->flags & F_LATIN1 && uch > 0xFF))
203 { 271 {
204 if (uch > 0xFFFFUL) 272 if (uch >= 0x10000UL)
205 { 273 {
274 if (uch >= 0x110000UL)
275 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
276
206 need (enc, len += 11); 277 need (enc, len += 11);
207 sprintf (enc->cur, "\\u%04x\\u%04x", 278 sprintf (enc->cur, "\\u%04x\\u%04x",
208 (int)((uch - 0x10000) / 0x400 + 0xD800), 279 (int)((uch - 0x10000) / 0x400 + 0xD800),
209 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 280 (int)((uch - 0x10000) % 0x400 + 0xDC00));
210 enc->cur += 12; 281 enc->cur += 12;
… …
221 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 292 *enc->cur++ = hexdigit [(uch >> 0) & 15];
222 } 293 }
223 294
224 str += clen; 295 str += clen;
225 } 296 }
226 else if (enc->flags & F_LATIN1) 297 else if (enc->json.flags & F_LATIN1)
227 { 298 {
228 *enc->cur++ = uch; 299 *enc->cur++ = uch;
229 str += clen; 300 str += clen;
230 } 301 }
231 else if (is_utf8) 302 else if (is_utf8)
… …
238 while (--clen); 309 while (--clen);
239 } 310 }
240 else 311 else
241 { 312 {
242 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed 313 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
243 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 314 enc->cur = encode_utf8 (enc->cur, uch);
244 ++str; 315 ++str;
245 } 316 }
246 } 317 }
247 } 318 }
248 } 319 }
249 320
250 --len; 321 --len;
251 } 322 }
252} 323}
253 324
254inline void 325INLINE void
255encode_indent (enc_t *enc) 326encode_indent (enc_t *enc)
256{ 327{
257 if (enc->flags & F_INDENT) 328 if (enc->json.flags & F_INDENT)
258 { 329 {
259 int spaces = enc->indent * INDENT_STEP; 330 int spaces = enc->indent * INDENT_STEP;
260 331
261 need (enc, spaces); 332 need (enc, spaces);
262 memset (enc->cur, ' ', spaces); 333 memset (enc->cur, ' ', spaces);
263 enc->cur += spaces; 334 enc->cur += spaces;
264 } 335 }
265} 336}
266 337
267inline void 338INLINE void
268encode_space (enc_t *enc) 339encode_space (enc_t *enc)
269{ 340{
270 need (enc, 1); 341 need (enc, 1);
271 encode_ch (enc, ' '); 342 encode_ch (enc, ' ');
272} 343}
273 344
274inline void 345INLINE void
275encode_nl (enc_t *enc) 346encode_nl (enc_t *enc)
276{ 347{
277 if (enc->flags & F_INDENT) 348 if (enc->json.flags & F_INDENT)
278 { 349 {
279 need (enc, 1); 350 need (enc, 1);
280 encode_ch (enc, '\n'); 351 encode_ch (enc, '\n');
281 } 352 }
282} 353}
283 354
284inline void 355INLINE void
285encode_comma (enc_t *enc) 356encode_comma (enc_t *enc)
286{ 357{
287 encode_ch (enc, ','); 358 encode_ch (enc, ',');
288 359
289 if (enc->flags & F_INDENT) 360 if (enc->json.flags & F_INDENT)
290 encode_nl (enc); 361 encode_nl (enc);
291 else if (enc->flags & F_SPACE_AFTER) 362 else if (enc->json.flags & F_SPACE_AFTER)
292 encode_space (enc); 363 encode_space (enc);
293} 364}
294 365
295static void encode_sv (enc_t *enc, SV *sv); 366static void encode_sv (enc_t *enc, SV *sv);
296 367
297static void 368static void
298encode_av (enc_t *enc, AV *av) 369encode_av (enc_t *enc, AV *av)
299{ 370{
300 int i, len = av_len (av); 371 int i, len = av_len (av);
301 372
302 if (enc->indent >= enc->maxdepth) 373 if (enc->indent >= enc->json.max_depth)
303 croak ("data structure too deep (hit recursion limit)"); 374 croak (ERR_NESTING_EXCEEDED);
304 375
305 encode_ch (enc, '['); encode_nl (enc); 376 encode_ch (enc, '[');
306 ++enc->indent; 377
378 if (len >= 0)
379 {
380 encode_nl (enc); ++enc->indent;
307 381
308 for (i = 0; i <= len; ++i) 382 for (i = 0; i <= len; ++i)
309 { 383 {
384 SV **svp = av_fetch (av, i, 0);
385
310 encode_indent (enc); 386 encode_indent (enc);
311 encode_sv (enc, *av_fetch (av, i, 0));
312 387
388 if (svp)
389 encode_sv (enc, *svp);
390 else
391 encode_str (enc, "null", 4, 0);
392
313 if (i < len) 393 if (i < len)
314 encode_comma (enc); 394 encode_comma (enc);
315 } 395 }
316 396
397 encode_nl (enc); --enc->indent; encode_indent (enc);
398 }
399
317 encode_nl (enc); 400 encode_ch (enc, ']');
318
319 --enc->indent;
320 encode_indent (enc); encode_ch (enc, ']');
321} 401}
322 402
323static void 403static void
324encode_he (enc_t *enc, HE *he) 404encode_hk (enc_t *enc, HE *he)
325{ 405{
326 encode_ch (enc, '"'); 406 encode_ch (enc, '"');
327 407
328 if (HeKLEN (he) == HEf_SVKEY) 408 if (HeKLEN (he) == HEf_SVKEY)
329 { 409 {
… …
339 else 419 else
340 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 420 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
341 421
342 encode_ch (enc, '"'); 422 encode_ch (enc, '"');
343 423
344 if (enc->flags & F_SPACE_BEFORE) encode_space (enc); 424 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
345 encode_ch (enc, ':'); 425 encode_ch (enc, ':');
346 if (enc->flags & F_SPACE_AFTER ) encode_space (enc); 426 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
347 encode_sv (enc, HeVAL (he));
348} 427}
349 428
350// compare hash entries, used when all keys are bytestrings 429// compare hash entries, used when all keys are bytestrings
351static int 430static int
352he_cmp_fast (const void *a_, const void *b_) 431he_cmp_fast (const void *a_, const void *b_)
… …
357 HE *b = *(HE **)b_; 436 HE *b = *(HE **)b_;
358 437
359 STRLEN la = HeKLEN (a); 438 STRLEN la = HeKLEN (a);
360 STRLEN lb = HeKLEN (b); 439 STRLEN lb = HeKLEN (b);
361 440
362 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 441 if (!(cmp = memcmp (HeKEY (b), HeKEY (a), lb < la ? lb : la)))
363 cmp = la - lb; 442 cmp = lb - la;
364 443
365 return cmp; 444 return cmp;
366} 445}
367 446
368// compare hash entries, used when some keys are sv's or utf-x 447// compare hash entries, used when some keys are sv's or utf-x
369static int 448static int
370he_cmp_slow (const void *a, const void *b) 449he_cmp_slow (const void *a, const void *b)
371{ 450{
372 return sv_cmp (HeSVKEY_force (*(HE **)a), HeSVKEY_force (*(HE **)b)); 451 return sv_cmp (HeSVKEY_force (*(HE **)b), HeSVKEY_force (*(HE **)a));
373} 452}
374 453
375static void 454static void
376encode_hv (enc_t *enc, HV *hv) 455encode_hv (enc_t *enc, HV *hv)
377{ 456{
378 int count, i; 457 HE *he;
379 458
380 if (enc->indent >= enc->maxdepth) 459 if (enc->indent >= enc->json.max_depth)
381 croak ("data structure too deep (hit recursion limit)"); 460 croak (ERR_NESTING_EXCEEDED);
382 461
383 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 462 encode_ch (enc, '{');
384 463
385 if ((count = hv_iterinit (hv)))
386 {
387 // for canonical output we have to sort by keys first 464 // for canonical output we have to sort by keys first
388 // actually, this is mostly due to the stupid so-called 465 // actually, this is mostly due to the stupid so-called
389 // security workaround added somewhere in 5.8.x. 466 // security workaround added somewhere in 5.8.x.
390 // that randomises hash orderings 467 // that randomises hash orderings
391 if (enc->flags & F_CANONICAL) 468 if (enc->json.flags & F_CANONICAL)
469 {
470 int count = hv_iterinit (hv);
471
472 if (SvMAGICAL (hv))
392 { 473 {
474 // need to count by iterating. could improve by dynamically building the vector below
475 // but I don't care for the speed of this special case.
476 // note also that we will run into undefined behaviour when the two iterations
477 // do not result in the same count, something I might care for in some later release.
478
479 count = 0;
480 while (hv_iternext (hv))
481 ++count;
482
483 hv_iterinit (hv);
484 }
485
486 if (count)
487 {
393 int fast = 1; 488 int i, fast = 1;
394 HE *he;
395#if defined(__BORLANDC__) || defined(_MSC_VER) 489#if defined(__BORLANDC__) || defined(_MSC_VER)
396 HE **hes = _alloca (count * sizeof (HE)); 490 HE **hes = _alloca (count * sizeof (HE));
397#else 491#else
398 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode 492 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
399#endif 493#endif
… …
426 520
427 FREETMPS; 521 FREETMPS;
428 LEAVE; 522 LEAVE;
429 } 523 }
430 524
431 for (i = 0; i < count; ++i) 525 encode_nl (enc); ++enc->indent;
526
527 while (count--)
432 { 528 {
433 encode_indent (enc); 529 encode_indent (enc);
530 he = hes [count];
434 encode_he (enc, hes [i]); 531 encode_hk (enc, he);
532 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
435 533
436 if (i < count - 1) 534 if (count)
437 encode_comma (enc); 535 encode_comma (enc);
438 } 536 }
439 537
538 encode_nl (enc); --enc->indent; encode_indent (enc);
539 }
540 }
541 else
542 {
543 if (hv_iterinit (hv) || SvMAGICAL (hv))
544 if ((he = hv_iternext (hv)))
545 {
546 encode_nl (enc); ++enc->indent;
547
548 for (;;)
549 {
440 encode_nl (enc); 550 encode_indent (enc);
551 encode_hk (enc, he);
552 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
553
554 if (!(he = hv_iternext (hv)))
555 break;
556
557 encode_comma (enc);
558 }
559
560 encode_nl (enc); --enc->indent; encode_indent (enc);
561 }
562 }
563
564 encode_ch (enc, '}');
565}
566
567// encode objects, arrays and special \0=false and \1=true values.
568static void
569encode_rv (enc_t *enc, SV *sv)
570{
571 svtype svt;
572
573 SvGETMAGIC (sv);
574 svt = SvTYPE (sv);
575
576 if (expect_false (SvOBJECT (sv)))
577 {
578 HV *stash = !JSON_SLOW || json_boolean_stash
579 ? json_boolean_stash
580 : gv_stashpv ("JSON::XS::Boolean", 1);
581
582 if (SvSTASH (sv) == stash)
583 {
584 if (SvIV (sv))
585 encode_str (enc, "true", 4, 0);
586 else
587 encode_str (enc, "false", 5, 0);
441 } 588 }
442 else 589 else
443 { 590 {
444 HE *he = hv_iternext (hv); 591#if 0
445 592 if (0 && sv_derived_from (rv, "JSON::Literal"))
446 for (;;)
447 { 593 {
448 encode_indent (enc); 594 // not yet
449 encode_he (enc, he);
450
451 if (!(he = hv_iternext (hv)))
452 break;
453
454 encode_comma (enc);
455 } 595 }
596#endif
597 if (enc->json.flags & F_CONV_BLESSED)
598 {
599 // we re-bless the reference to get overload and other niceties right
600 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
456 601
602 if (to_json)
603 {
604 dSP;
605
606 ENTER; SAVETMPS; PUSHMARK (SP);
607 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
608
609 // calling with G_SCALAR ensures that we always get a 1 return value
610 PUTBACK;
611 call_sv ((SV *)GvCV (to_json), G_SCALAR);
612 SPAGAIN;
613
614 // catch this surprisingly common error
615 if (SvROK (TOPs) && SvRV (TOPs) == sv)
616 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
617
618 sv = POPs;
619 PUTBACK;
620
457 encode_nl (enc); 621 encode_sv (enc, sv);
622
623 FREETMPS; LEAVE;
624 }
625 else if (enc->json.flags & F_ALLOW_BLESSED)
626 encode_str (enc, "null", 4, 0);
627 else
628 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
629 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
630 }
631 else if (enc->json.flags & F_ALLOW_BLESSED)
632 encode_str (enc, "null", 4, 0);
633 else
634 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",
635 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
458 } 636 }
459 } 637 }
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) 638 else if (svt == SVt_PVHV)
474 encode_hv (enc, (HV *)sv); 639 encode_hv (enc, (HV *)sv);
475 else if (svt == SVt_PVAV) 640 else if (svt == SVt_PVAV)
476 encode_av (enc, (AV *)sv); 641 encode_av (enc, (AV *)sv);
477 else if (svt < SVt_PVAV) 642 else if (svt < SVt_PVAV)
478 { 643 {
479 if (SvNIOK (sv) && SvIV (sv) == 0) 644 STRLEN len = 0;
645 char *pv = svt ? SvPV (sv, len) : 0;
646
647 if (len == 1 && *pv == '1')
648 encode_str (enc, "true", 4, 0);
649 else if (len == 1 && *pv == '0')
480 encode_str (enc, "false", 5, 0); 650 encode_str (enc, "false", 5, 0);
481 else if (SvNIOK (sv) && SvIV (sv) == 1) 651 else if (enc->json.flags & F_ALLOW_UNKNOWN)
482 encode_str (enc, "true", 4, 0); 652 encode_str (enc, "null", 4, 0);
483 else 653 else
484 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", 654 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
485 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 655 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
486 } 656 }
657 else if (enc->json.flags & F_ALLOW_UNKNOWN)
658 encode_str (enc, "null", 4, 0);
487 else 659 else
488 croak ("encountered %s, but JSON can only represent references to arrays or hashes", 660 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
489 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 661 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
490} 662}
491 663
… …
509 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 681 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
510 enc->cur += strlen (enc->cur); 682 enc->cur += strlen (enc->cur);
511 } 683 }
512 else if (SvIOKp (sv)) 684 else if (SvIOKp (sv))
513 { 685 {
514 // we assume we can always read an IV as a UV 686 // we assume we can always read an IV as a UV and vice versa
515 if (SvUV (sv) & ~(UV)0x7fff) 687 // we assume two's complement
516 { 688 // we assume no aliasing issues in the union
517 // large integer, use the (rather slow) snprintf way. 689 if (SvIsUV (sv) ? SvUVX (sv) <= 59000
518 need (enc, sizeof (UV) * 3); 690 : 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 { 691 {
526 // optimise the "small number case" 692 // optimise the "small number case"
527 // code will likely be branchless and use only a single multiplication 693 // code will likely be branchless and use only a single multiplication
694 // works for numbers up to 59074
528 I32 i = SvIV (sv); 695 I32 i = SvIVX (sv);
529 U32 u; 696 U32 u;
530 char digit, nz = 0; 697 char digit, nz = 0;
531 698
532 need (enc, 6); 699 need (enc, 6);
533 700
… …
539 706
540 // now output digit by digit, each time masking out the integer part 707 // now output digit by digit, each time masking out the integer part
541 // and multiplying by 5 while moving the decimal point one to the right, 708 // and multiplying by 5 while moving the decimal point one to the right,
542 // resulting in a net multiplication by 10. 709 // resulting in a net multiplication by 10.
543 // we always write the digit to memory but conditionally increment 710 // we always write the digit to memory but conditionally increment
544 // the pointer, to ease the usage of conditional move instructions. 711 // 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; 712 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; 713 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; 714 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; 715 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' 716 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
550 } 717 }
718 else
719 {
720 // large integer, use the (rather slow) snprintf way.
721 need (enc, IVUV_MAXCHARS);
722 enc->cur +=
723 SvIsUV(sv)
724 ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))
725 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
726 }
551 } 727 }
552 else if (SvROK (sv)) 728 else if (SvROK (sv))
553 encode_rv (enc, SvRV (sv)); 729 encode_rv (enc, SvRV (sv));
554 else if (!SvOK (sv)) 730 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN)
555 encode_str (enc, "null", 4, 0); 731 encode_str (enc, "null", 4, 0);
556 else 732 else
557 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 733 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
558 SvPV_nolen (sv), SvFLAGS (sv)); 734 SvPV_nolen (sv), SvFLAGS (sv));
559} 735}
560 736
561static SV * 737static SV *
562encode_json (SV *scalar, U32 flags) 738encode_json (SV *scalar, JSON *json)
563{ 739{
564 enc_t enc; 740 enc_t enc;
565 741
566 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 742 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
567 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 743 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
568 744
569 enc.flags = flags; 745 enc.json = *json;
570 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 746 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
571 enc.cur = SvPVX (enc.sv); 747 enc.cur = SvPVX (enc.sv);
572 enc.end = SvEND (enc.sv); 748 enc.end = SvEND (enc.sv);
573 enc.indent = 0; 749 enc.indent = 0;
574 enc.maxdepth = DEC_DEPTH (flags); 750 enc.limit = enc.json.flags & F_ASCII ? 0x000080UL
751 : enc.json.flags & F_LATIN1 ? 0x000100UL
752 : 0x110000UL;
575 753
576 SvPOK_only (enc.sv); 754 SvPOK_only (enc.sv);
577 encode_sv (&enc, scalar); 755 encode_sv (&enc, scalar);
578 756
579 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 757 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
580 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 758 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
581 759
582 if (!(flags & (F_ASCII | F_LATIN1 | F_UTF8))) 760 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
583 SvUTF8_on (enc.sv); 761 SvUTF8_on (enc.sv);
584 762
585 if (enc.flags & F_SHRINK) 763 if (enc.json.flags & F_SHRINK)
586 shrink (enc.sv); 764 shrink (enc.sv);
587 765
588 return enc.sv; 766 return enc.sv;
589} 767}
590 768
… …
595typedef struct 773typedef struct
596{ 774{
597 char *cur; // current parser pointer 775 char *cur; // current parser pointer
598 char *end; // end of input string 776 char *end; // end of input string
599 const char *err; // parse error, if != 0 777 const char *err; // parse error, if != 0
600 U32 flags; // F_* 778 JSON json;
601 U32 depth; // recursion depth 779 U32 depth; // recursion depth
602 U32 maxdepth; // recursion depth limit 780 U32 maxdepth; // recursion depth limit
603} dec_t; 781} dec_t;
604 782
605inline void 783INLINE void
784decode_comment (dec_t *dec)
785{
786 // only '#'-style comments allowed a.t.m.
787
788 while (*dec->cur && *dec->cur != 0x0a && *dec->cur != 0x0d)
789 ++dec->cur;
790}
791
792INLINE void
606decode_ws (dec_t *dec) 793decode_ws (dec_t *dec)
607{ 794{
608 for (;;) 795 for (;;)
609 { 796 {
610 char ch = *dec->cur; 797 char ch = *dec->cur;
611 798
612 if (ch > 0x20 799 if (ch > 0x20)
800 {
801 if (expect_false (ch == '#'))
802 {
803 if (dec->json.flags & F_RELAXED)
804 decode_comment (dec);
805 else
806 break;
807 }
808 else
809 break;
810 }
613 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) 811 else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)
614 break; 812 break; // parse error, but let higher level handle it, gives better error messages
615 813
616 ++dec->cur; 814 ++dec->cur;
617 } 815 }
618} 816}
619 817
… …
623 if (*dec->cur != ch) \ 821 if (*dec->cur != ch) \
624 ERR (# ch " expected"); \ 822 ERR (# ch " expected"); \
625 ++dec->cur; \ 823 ++dec->cur; \
626 SE 824 SE
627 825
628#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)") 826#define DEC_INC_DEPTH if (++dec->depth > dec->json.max_depth) ERR (ERR_NESTING_EXCEEDED)
629#define DEC_DEC_DEPTH --dec->depth 827#define DEC_DEC_DEPTH --dec->depth
630 828
631static SV *decode_sv (dec_t *dec); 829static SV *decode_sv (dec_t *dec);
632 830
633static signed char decode_hexdigit[256]; 831static signed char decode_hexdigit[256];
… …
725 923
726 if (hi >= 0x80) 924 if (hi >= 0x80)
727 { 925 {
728 utf8 = 1; 926 utf8 = 1;
729 927
730 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0); 928 cur = encode_utf8 (cur, hi);
731 } 929 }
732 else 930 else
733 *cur++ = hi; 931 *cur++ = hi;
734 } 932 }
735 break; 933 break;
… …
737 default: 935 default:
738 --dec_cur; 936 --dec_cur;
739 ERR ("illegal backslash escape sequence in string"); 937 ERR ("illegal backslash escape sequence in string");
740 } 938 }
741 } 939 }
742 else if (expect_true (ch >= 0x20 && ch <= 0x7f)) 940 else if (expect_true (ch >= 0x20 && ch < 0x80))
743 *cur++ = ch; 941 *cur++ = ch;
744 else if (ch >= 0x80) 942 else if (ch >= 0x80)
745 { 943 {
746 STRLEN clen; 944 STRLEN clen;
747 UV uch; 945 UV uch;
… …
868 is_nv = 1; 1066 is_nv = 1;
869 } 1067 }
870 1068
871 if (!is_nv) 1069 if (!is_nv)
872 { 1070 {
1071 int len = dec->cur - start;
1072
873 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so 1073 // special case the rather common 1..5-digit-int case
874 if (*start == '-') 1074 if (*start == '-')
875 switch (dec->cur - start) 1075 switch (len)
876 { 1076 {
877 case 2: return newSViv (-( start [1] - '0' * 1)); 1077 case 2: return newSViv (-( start [1] - '0' * 1));
878 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); 1078 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)); 1079 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)); 1080 case 5: return newSViv (-( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
1081 case 6: return newSViv (-(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
881 } 1082 }
882 else 1083 else
883 switch (dec->cur - start) 1084 switch (len)
884 { 1085 {
885 case 1: return newSViv ( start [0] - '0' * 1); 1086 case 1: return newSViv ( start [0] - '0' * 1);
886 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); 1087 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); 1088 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); 1089 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
1090 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
889 } 1091 }
890 1092
891 { 1093 {
892 UV uv; 1094 UV uv;
893 int numtype = grok_number (start, dec->cur - start, &uv); 1095 int numtype = grok_number (start, len, &uv);
894 if (numtype & IS_NUMBER_IN_UV) 1096 if (numtype & IS_NUMBER_IN_UV)
895 if (numtype & IS_NUMBER_NEG) 1097 if (numtype & IS_NUMBER_NEG)
896 { 1098 {
897 if (uv < (UV)IV_MIN) 1099 if (uv < (UV)IV_MIN)
898 return newSViv (-(IV)uv); 1100 return newSViv (-(IV)uv);
899 } 1101 }
900 else 1102 else
901 return newSVuv (uv); 1103 return newSVuv (uv);
902
903 // here would likely be the place for bigint support
904 } 1104 }
905 }
906 1105
907 // if we ever support bigint or bigfloat, this is the place for bigfloat 1106 len -= *start == '-' ? 1 : 0;
1107
1108 // does not fit into IV or UV, try NV
1109 if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len)
1110 #if defined (LDBL_DIG)
1111 || (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len)
1112 #endif
1113 )
1114 // fits into NV without loss of precision
1115 return newSVnv (Atof (start));
1116
1117 // everything else fails, convert it to a string
1118 return newSVpvn (start, dec->cur - start);
1119 }
1120
1121 // loss of precision here
908 return newSVnv (Atof (start)); 1122 return newSVnv (Atof (start));
909 1123
910fail: 1124fail:
911 return 0; 1125 return 0;
912} 1126}
… …
942 1156
943 if (*dec->cur != ',') 1157 if (*dec->cur != ',')
944 ERR (", or ] expected while parsing array"); 1158 ERR (", or ] expected while parsing array");
945 1159
946 ++dec->cur; 1160 ++dec->cur;
1161
1162 decode_ws (dec);
1163
1164 if (*dec->cur == ']' && dec->json.flags & F_RELAXED)
1165 {
1166 ++dec->cur;
1167 break;
1168 }
947 } 1169 }
948 1170
949 DEC_DEC_DEPTH; 1171 DEC_DEC_DEPTH;
950 return newRV_noinc ((SV *)av); 1172 return newRV_noinc ((SV *)av);
951 1173
… …
956} 1178}
957 1179
958static SV * 1180static SV *
959decode_hv (dec_t *dec) 1181decode_hv (dec_t *dec)
960{ 1182{
1183 SV *sv;
961 HV *hv = newHV (); 1184 HV *hv = newHV ();
962 1185
963 DEC_INC_DEPTH; 1186 DEC_INC_DEPTH;
964 decode_ws (dec); 1187 decode_ws (dec);
965 1188
966 if (*dec->cur == '}') 1189 if (*dec->cur == '}')
967 ++dec->cur; 1190 ++dec->cur;
968 else 1191 else
969 for (;;) 1192 for (;;)
970 { 1193 {
1194 EXPECT_CH ('"');
1195
1196 // heuristic: assume that
1197 // a) decode_str + hv_store_ent are abysmally slow.
1198 // b) most hash keys are short, simple ascii text.
1199 // => try to "fast-match" such strings to avoid
1200 // the overhead of decode_str + hv_store_ent.
1201 {
971 SV *key, *value; 1202 SV *value;
1203 char *p = dec->cur;
1204 char *e = p + 24; // only try up to 24 bytes
972 1205
973 decode_ws (dec); EXPECT_CH ('"'); 1206 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 { 1207 {
1208 // the >= 0x80 is false on most architectures
1209 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1210 {
1211 // slow path, back up and use decode_str
1212 SV *key = decode_str (dec);
1213 if (!key)
1214 goto fail;
1215
1216 decode_ws (dec); EXPECT_CH (':');
1217
1218 decode_ws (dec);
1219 value = decode_sv (dec);
1220 if (!value)
1221 {
1222 SvREFCNT_dec (key);
1223 goto fail;
1224 }
1225
1226 hv_store_ent (hv, key, value, 0);
984 SvREFCNT_dec (key); 1227 SvREFCNT_dec (key);
1228
1229 break;
1230 }
1231 else if (*p == '"')
1232 {
1233 // fast path, got a simple key
1234 char *key = dec->cur;
1235 int len = p - key;
1236 dec->cur = p + 1;
1237
1238 decode_ws (dec); EXPECT_CH (':');
1239
1240 decode_ws (dec);
1241 value = decode_sv (dec);
1242 if (!value)
985 goto fail; 1243 goto fail;
1244
1245 hv_store (hv, key, len, value, 0);
1246
1247 break;
1248 }
1249
1250 ++p;
986 } 1251 }
987 1252 }
988 hv_store_ent (hv, key, value, 0);
989 SvREFCNT_dec (key);
990 1253
991 decode_ws (dec); 1254 decode_ws (dec);
992 1255
993 if (*dec->cur == '}') 1256 if (*dec->cur == '}')
994 { 1257 {
… …
998 1261
999 if (*dec->cur != ',') 1262 if (*dec->cur != ',')
1000 ERR (", or } expected while parsing object/hash"); 1263 ERR (", or } expected while parsing object/hash");
1001 1264
1002 ++dec->cur; 1265 ++dec->cur;
1266
1267 decode_ws (dec);
1268
1269 if (*dec->cur == '}' && dec->json.flags & F_RELAXED)
1270 {
1271 ++dec->cur;
1272 break;
1273 }
1003 } 1274 }
1004 1275
1005 DEC_DEC_DEPTH; 1276 DEC_DEC_DEPTH;
1006 return newRV_noinc ((SV *)hv); 1277 sv = newRV_noinc ((SV *)hv);
1278
1279 // check filter callbacks
1280 if (dec->json.flags & F_HOOK)
1281 {
1282 if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
1283 {
1284 HE *cb, *he;
1285
1286 hv_iterinit (hv);
1287 he = hv_iternext (hv);
1288 hv_iterinit (hv);
1289
1290 // the next line creates a mortal sv each time its called.
1291 // might want to optimise this for common cases.
1292 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1293
1294 if (cb)
1295 {
1296 dSP;
1297 int count;
1298
1299 ENTER; SAVETMPS; PUSHMARK (SP);
1300 XPUSHs (HeVAL (he));
1301
1302 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1303
1304 if (count == 1)
1305 {
1306 sv = newSVsv (POPs);
1307 FREETMPS; LEAVE;
1308 return sv;
1309 }
1310
1311 FREETMPS; LEAVE;
1312 }
1313 }
1314
1315 if (dec->json.cb_object)
1316 {
1317 dSP;
1318 int count;
1319
1320 ENTER; SAVETMPS; PUSHMARK (SP);
1321 XPUSHs (sv_2mortal (sv));
1322
1323 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1324
1325 if (count == 1)
1326 {
1327 sv = newSVsv (POPs);
1328 FREETMPS; LEAVE;
1329 return sv;
1330 }
1331
1332 SvREFCNT_inc (sv);
1333 FREETMPS; LEAVE;
1334 }
1335 }
1336
1337 return sv;
1007 1338
1008fail: 1339fail:
1009 SvREFCNT_dec (hv); 1340 SvREFCNT_dec (hv);
1010 DEC_DEC_DEPTH; 1341 DEC_DEC_DEPTH;
1011 return 0; 1342 return 0;
1012} 1343}
1013 1344
1014static SV * 1345static SV *
1015decode_sv (dec_t *dec) 1346decode_sv (dec_t *dec)
1016{ 1347{
1017 decode_ws (dec);
1018
1019 // the beauty of JSON: you need exactly one character lookahead 1348 // the beauty of JSON: you need exactly one character lookahead
1020 // to parse anything. 1349 // to parse everything.
1021 switch (*dec->cur) 1350 switch (*dec->cur)
1022 { 1351 {
1023 case '"': ++dec->cur; return decode_str (dec); 1352 case '"': ++dec->cur; return decode_str (dec);
1024 case '[': ++dec->cur; return decode_av (dec); 1353 case '[': ++dec->cur; return decode_av (dec);
1025 case '{': ++dec->cur; return decode_hv (dec); 1354 case '{': ++dec->cur; return decode_hv (dec);
1026 1355
1027 case '-': 1356 case '-':
1028 case '0': case '1': case '2': case '3': case '4': 1357 case '0': case '1': case '2': case '3': case '4':
1029 case '5': case '6': case '7': case '8': case '9': 1358 case '5': case '6': case '7': case '8': case '9':
1030 return decode_num (dec); 1359 return decode_num (dec);
1031 1360
1032 case 't': 1361 case 't':
1033 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1362 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1034 { 1363 {
1035 dec->cur += 4; 1364 dec->cur += 4;
1365#if JSON_SLOW
1366 json_true = get_bool ("JSON::XS::true");
1367#endif
1036 return newSViv (1); 1368 return newSVsv (json_true);
1037 } 1369 }
1038 else 1370 else
1039 ERR ("'true' expected"); 1371 ERR ("'true' expected");
1040 1372
1041 break; 1373 break;
1042 1374
1043 case 'f': 1375 case 'f':
1044 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1376 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1045 { 1377 {
1046 dec->cur += 5; 1378 dec->cur += 5;
1379#if JSON_SLOW
1380 json_false = get_bool ("JSON::XS::false");
1381#endif
1047 return newSViv (0); 1382 return newSVsv (json_false);
1048 } 1383 }
1049 else 1384 else
1050 ERR ("'false' expected"); 1385 ERR ("'false' expected");
1051 1386
1052 break; 1387 break;
… …
1070fail: 1405fail:
1071 return 0; 1406 return 0;
1072} 1407}
1073 1408
1074static SV * 1409static SV *
1075decode_json (SV *string, U32 flags, UV *offset_return) 1410decode_json (SV *string, JSON *json, STRLEN *offset_return)
1076{ 1411{
1077 dec_t dec; 1412 dec_t dec;
1078 UV offset; 1413 STRLEN offset;
1079 SV *sv; 1414 SV *sv;
1080 1415
1081 SvGETMAGIC (string); 1416 SvGETMAGIC (string);
1082 SvUPGRADE (string, SVt_PV); 1417 SvUPGRADE (string, SVt_PV);
1083 1418
1419 /* work around a bug in perl 5.10, which causes SvCUR to fail an
1420 * assertion with -DDEBUGGING, although SvCUR is documented to
1421 * return the xpv_cur field which certainly exists after upgrading.
1422 * according to nicholas clark, calling SvPOK fixes this.
1423 * But it doesn't fix it, so try another workaround, call SvPV_nolen
1424 * and hope for the best.
1425 * Damnit, SvPV_nolen still trips over yet another assertion. This
1426 * assertion business is seriously broken, try yet another workaround
1427 * for the broken -DDEBUGGING.
1428 */
1429#ifdef DEBUGGING
1430 offset = SvOK (string) ? sv_len (string) : 0;
1431#else
1432 offset = SvCUR (string);
1433#endif
1434
1435 if (offset > json->max_size && json->max_size)
1436 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1437 (unsigned long)SvCUR (string), (unsigned long)json->max_size);
1438
1084 if (flags & F_UTF8) 1439 if (json->flags & F_UTF8)
1085 sv_utf8_downgrade (string, 0); 1440 sv_utf8_downgrade (string, 0);
1086 else 1441 else
1087 sv_utf8_upgrade (string); 1442 sv_utf8_upgrade (string);
1088 1443
1089 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1444 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
1090 1445
1091 dec.flags = flags; 1446 dec.json = *json;
1092 dec.cur = SvPVX (string); 1447 dec.cur = SvPVX (string);
1093 dec.end = SvEND (string); 1448 dec.end = SvEND (string);
1094 dec.err = 0; 1449 dec.err = 0;
1095 dec.depth = 0; 1450 dec.depth = 0;
1096 dec.maxdepth = DEC_DEPTH (dec.flags); 1451
1452 if (dec.json.cb_object || dec.json.cb_sk_object)
1453 dec.json.flags |= F_HOOK;
1097 1454
1098 *dec.end = 0; // this should basically be a nop, too, but make sure it's there 1455 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1456
1457 decode_ws (&dec);
1099 sv = decode_sv (&dec); 1458 sv = decode_sv (&dec);
1100 1459
1101 if (!(offset_return || !sv)) 1460 if (!(offset_return || !sv))
1102 { 1461 {
1103 // check for trailing garbage 1462 // check for trailing garbage
… …
1111 } 1470 }
1112 } 1471 }
1113 1472
1114 if (offset_return || !sv) 1473 if (offset_return || !sv)
1115 { 1474 {
1116 offset = dec.flags & F_UTF8 1475 offset = dec.json.flags & F_UTF8
1117 ? dec.cur - SvPVX (string) 1476 ? dec.cur - SvPVX (string)
1118 : utf8_distance (dec.cur, SvPVX (string)); 1477 : utf8_distance (dec.cur, SvPVX (string));
1119 1478
1120 if (offset_return) 1479 if (offset_return)
1121 *offset_return = offset; 1480 *offset_return = offset;
… …
1140 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1499 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1141 } 1500 }
1142 1501
1143 sv = sv_2mortal (sv); 1502 sv = sv_2mortal (sv);
1144 1503
1145 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1504 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)"); 1505 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1147 1506
1148 return sv; 1507 return sv;
1508}
1509
1510/////////////////////////////////////////////////////////////////////////////
1511// incremental parser
1512
1513static void
1514incr_parse (JSON *self)
1515{
1516 const char *p = SvPVX (self->incr_text) + self->incr_pos;
1517
1518 for (;;)
1519 {
1520 //printf ("loop pod %d *p<%c><%s>, mode %d nest %d\n", p - SvPVX (self->incr_text), *p, p, self->incr_mode, self->incr_nest);//D
1521 switch (self->incr_mode)
1522 {
1523 // only used for intiial whitespace skipping
1524 case INCR_M_WS:
1525 for (;;)
1526 {
1527 if (*p > 0x20)
1528 {
1529 self->incr_mode = INCR_M_JSON;
1530 goto incr_m_json;
1531 }
1532 else if (!*p)
1533 goto interrupt;
1534
1535 ++p;
1536 }
1537
1538 // skip a single char inside a string (for \\-processing)
1539 case INCR_M_BS:
1540 if (!*p)
1541 goto interrupt;
1542
1543 ++p;
1544 self->incr_mode = INCR_M_STR;
1545 goto incr_m_str;
1546
1547 // inside a string
1548 case INCR_M_STR:
1549 incr_m_str:
1550 for (;;)
1551 {
1552 if (*p == '"')
1553 {
1554 ++p;
1555 self->incr_mode = INCR_M_JSON;
1556
1557 if (!self->incr_nest)
1558 goto interrupt;
1559
1560 goto incr_m_json;
1561 }
1562 else if (*p == '\\')
1563 {
1564 ++p; // "virtually" consumes character after \
1565
1566 if (!*p) // if at end of string we have to switch modes
1567 {
1568 self->incr_mode = INCR_M_BS;
1569 goto interrupt;
1570 }
1571 }
1572 else if (!*p)
1573 goto interrupt;
1574
1575 ++p;
1576 }
1577
1578 // after initial ws, outside string
1579 case INCR_M_JSON:
1580 incr_m_json:
1581 for (;;)
1582 {
1583 switch (*p++)
1584 {
1585 case 0:
1586 --p;
1587 goto interrupt;
1588
1589 case 0x09:
1590 case 0x0a:
1591 case 0x0d:
1592 case 0x20:
1593 if (!self->incr_nest)
1594 {
1595 --p; // do not eat the whitespace, let the next round do it
1596 goto interrupt;
1597 }
1598 break;
1599
1600 case '"':
1601 self->incr_mode = INCR_M_STR;
1602 goto incr_m_str;
1603
1604 case '[':
1605 case '{':
1606 if (++self->incr_nest > self->max_depth)
1607 croak (ERR_NESTING_EXCEEDED);
1608 break;
1609
1610 case ']':
1611 case '}':
1612 if (--self->incr_nest <= 0)
1613 goto interrupt;
1614 }
1615 }
1616 }
1617
1618 modechange:
1619 ;
1620 }
1621
1622interrupt:
1623 self->incr_pos = p - SvPVX (self->incr_text);
1624 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D
1149} 1625}
1150 1626
1151///////////////////////////////////////////////////////////////////////////// 1627/////////////////////////////////////////////////////////////////////////////
1152// XS interface functions 1628// XS interface functions
1153 1629
… …
1162 i >= '0' && i <= '9' ? i - '0' 1638 i >= '0' && i <= '9' ? i - '0'
1163 : i >= 'a' && i <= 'f' ? i - 'a' + 10 1639 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1164 : i >= 'A' && i <= 'F' ? i - 'A' + 10 1640 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1165 : -1; 1641 : -1;
1166 1642
1167 json_stash = gv_stashpv ("JSON::XS", 1); 1643 json_stash = gv_stashpv ("JSON::XS" , 1);
1644 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1645
1646 json_true = get_bool ("JSON::XS::true");
1647 json_false = get_bool ("JSON::XS::false");
1168} 1648}
1169 1649
1170PROTOTYPES: DISABLE 1650PROTOTYPES: DISABLE
1171 1651
1172SV *new (char *dummy) 1652void CLONE (...)
1173 CODE: 1653 CODE:
1174 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1654 json_stash = 0;
1655 json_boolean_stash = 0;
1656
1657void new (char *klass)
1658 PPCODE:
1659{
1660 SV *pv = NEWSV (0, sizeof (JSON));
1661 SvPOK_only (pv);
1662 json_init ((JSON *)SvPVX (pv));
1663 XPUSHs (sv_2mortal (sv_bless (
1664 newRV_noinc (pv),
1665 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1666 )));
1667}
1668
1669void ascii (JSON *self, int enable = 1)
1670 ALIAS:
1671 ascii = F_ASCII
1672 latin1 = F_LATIN1
1673 utf8 = F_UTF8
1674 indent = F_INDENT
1675 canonical = F_CANONICAL
1676 space_before = F_SPACE_BEFORE
1677 space_after = F_SPACE_AFTER
1678 pretty = F_PRETTY
1679 allow_nonref = F_ALLOW_NONREF
1680 shrink = F_SHRINK
1681 allow_blessed = F_ALLOW_BLESSED
1682 convert_blessed = F_CONV_BLESSED
1683 relaxed = F_RELAXED
1684 allow_unknown = F_ALLOW_UNKNOWN
1685 PPCODE:
1686{
1687 if (enable)
1688 self->flags |= ix;
1689 else
1690 self->flags &= ~ix;
1691
1692 XPUSHs (ST (0));
1693}
1694
1695void get_ascii (JSON *self)
1696 ALIAS:
1697 get_ascii = F_ASCII
1698 get_latin1 = F_LATIN1
1699 get_utf8 = F_UTF8
1700 get_indent = F_INDENT
1701 get_canonical = F_CANONICAL
1702 get_space_before = F_SPACE_BEFORE
1703 get_space_after = F_SPACE_AFTER
1704 get_allow_nonref = F_ALLOW_NONREF
1705 get_shrink = F_SHRINK
1706 get_allow_blessed = F_ALLOW_BLESSED
1707 get_convert_blessed = F_CONV_BLESSED
1708 get_relaxed = F_RELAXED
1709 get_allow_unknown = F_ALLOW_UNKNOWN
1710 PPCODE:
1711 XPUSHs (boolSV (self->flags & ix));
1712
1713void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
1714 PPCODE:
1715 self->max_depth = max_depth;
1716 XPUSHs (ST (0));
1717
1718U32 get_max_depth (JSON *self)
1719 CODE:
1720 RETVAL = self->max_depth;
1175 OUTPUT: 1721 OUTPUT:
1176 RETVAL 1722 RETVAL
1177 1723
1178SV *ascii (SV *self, int enable = 1) 1724void max_size (JSON *self, U32 max_size = 0)
1179 ALIAS: 1725 PPCODE:
1180 ascii = F_ASCII 1726 self->max_size = max_size;
1181 latin1 = F_LATIN1 1727 XPUSHs (ST (0));
1182 utf8 = F_UTF8 1728
1183 indent = F_INDENT 1729int 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: 1730 CODE:
1191{ 1731 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: 1732 OUTPUT:
1201 RETVAL 1733 RETVAL
1202 1734
1203SV *max_depth (SV *self, UV max_depth = 0x80000000UL) 1735void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1736 PPCODE:
1737{
1738 SvREFCNT_dec (self->cb_object);
1739 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
1740
1741 XPUSHs (ST (0));
1742}
1743
1744void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
1745 PPCODE:
1746{
1747 if (!self->cb_sk_object)
1748 self->cb_sk_object = newHV ();
1749
1750 if (SvOK (cb))
1751 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1752 else
1753 {
1754 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1755
1756 if (!HvKEYS (self->cb_sk_object))
1757 {
1758 SvREFCNT_dec (self->cb_sk_object);
1759 self->cb_sk_object = 0;
1760 }
1761 }
1762
1763 XPUSHs (ST (0));
1764}
1765
1766void encode (JSON *self, SV *scalar)
1767 PPCODE:
1768 XPUSHs (encode_json (scalar, self));
1769
1770void decode (JSON *self, SV *jsonstr)
1771 PPCODE:
1772 XPUSHs (decode_json (jsonstr, self, 0));
1773
1774void decode_prefix (JSON *self, SV *jsonstr)
1775 PPCODE:
1776{
1777 STRLEN offset;
1778 EXTEND (SP, 2);
1779 PUSHs (decode_json (jsonstr, self, &offset));
1780 PUSHs (sv_2mortal (newSVuv (offset)));
1781}
1782
1783void incr_parse (JSON *self, SV *jsonstr = 0)
1784 PPCODE:
1785{
1786 if (!self->incr_text)
1787 self->incr_text = newSVpvn ("", 0);
1788
1789 // append data, if any
1790 if (jsonstr)
1791 {
1792 if (SvUTF8 (jsonstr) && !SvUTF8 (self->incr_text))
1793 {
1794 /* utf-8-ness differs, need to upgrade */
1795 sv_utf8_upgrade (self->incr_text);
1796
1797 if (self->incr_pos)
1798 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
1799 - (U8 *)SvPVX (self->incr_text);
1800 }
1801
1802 {
1803 STRLEN len;
1804 const char *str = SvPV (jsonstr, len);
1805 SvGROW (self->incr_text, SvCUR (self->incr_text) + len + 1);
1806 Move (str, SvEND (self->incr_text), len, char);
1807 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
1808 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
1809 }
1810 }
1811
1812 if (GIMME_V != G_VOID)
1813 do
1814 {
1815 STRLEN offset;
1816
1817 if (!INCR_DONE (self))
1818 {
1819 incr_parse (self);
1820
1821 if (self->incr_pos > self->max_size && self->max_size)
1822 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1823 (unsigned long)self->incr_pos, (unsigned long)self->max_size);
1824
1825 if (!INCR_DONE (self))
1826 break;
1827 }
1828
1829 XPUSHs (decode_json (self->incr_text, self, &offset));
1830
1831 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + offset);
1832 self->incr_pos -= offset;
1833 self->incr_nest = 0;
1834 self->incr_mode = 0;
1835 }
1836 while (GIMME_V == G_ARRAY);
1837}
1838
1839SV *incr_text (JSON *self)
1840 ATTRS: lvalue
1204 CODE: 1841 CODE:
1205{ 1842{
1206 UV *uv = SvJSON (self); 1843 if (self->incr_pos)
1207 UV log2 = 0; 1844 croak ("incr_text can not be called when the incremental parser already started parsing");
1208 1845
1209 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL; 1846 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} 1847}
1218 OUTPUT: 1848 OUTPUT:
1219 RETVAL 1849 RETVAL
1220 1850
1851void incr_skip (JSON *self)
1852 CODE:
1853{
1854 if (self->incr_pos)
1855 {
1856 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + self->incr_pos);
1857 self->incr_pos = 0;
1858 self->incr_nest = 0;
1859 self->incr_mode = 0;
1860 }
1861}
1862
1863void incr_reset (JSON *self)
1864 CODE:
1865{
1866 SvREFCNT_dec (self->incr_text);
1867 self->incr_text = 0;
1868 self->incr_pos = 0;
1869 self->incr_nest = 0;
1870 self->incr_mode = 0;
1871}
1872
1873void DESTROY (JSON *self)
1874 CODE:
1875 SvREFCNT_dec (self->cb_sk_object);
1876 SvREFCNT_dec (self->cb_object);
1877 SvREFCNT_dec (self->incr_text);
1878
1879PROTOTYPES: ENABLE
1880
1221void encode (SV *self, SV *scalar) 1881void encode_json (SV *scalar)
1882 ALIAS:
1883 to_json_ = 0
1884 encode_json = F_UTF8
1222 PPCODE: 1885 PPCODE:
1886{
1887 JSON json;
1888 json_init (&json);
1889 json.flags |= ix;
1223 XPUSHs (encode_json (scalar, *SvJSON (self))); 1890 XPUSHs (encode_json (scalar, &json));
1891}
1224 1892
1225void decode (SV *self, SV *jsonstr) 1893void decode_json (SV *jsonstr)
1894 ALIAS:
1895 from_json_ = 0
1896 decode_json = F_UTF8
1226 PPCODE: 1897 PPCODE:
1898{
1899 JSON json;
1900 json_init (&json);
1901 json.flags |= ix;
1227 XPUSHs (decode_json (jsonstr, *SvJSON (self), 0)); 1902 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} 1903}
1237 1904
1238PROTOTYPES: ENABLE
1239 1905
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