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 1.49 by root, Sun Jul 1 23:40:07 2007 UTC vs.
Revision 1.94 by root, Thu Nov 20 03:59:53 2008 UTC

1#include "EXTERN.h" 1#include "EXTERN.h"
2#include "perl.h" 2#include "perl.h"
3#include "XSUB.h" 3#include "XSUB.h"
4 4
5#include "assert.h" 5#include <assert.h>
6#include "string.h" 6#include <string.h>
7#include "stdlib.h" 7#include <stdlib.h>
8#include "stdio.h" 8#include <stdio.h>
9#include <limits.h>
10#include <float.h>
9 11
10#if defined(__BORLANDC__) || defined(_MSC_VER) 12#if defined(__BORLANDC__) || defined(_MSC_VER)
11# define snprintf _snprintf // C compilers have this in stdio.h 13# define snprintf _snprintf // C compilers have this in stdio.h
12#endif 14#endif
13 15
14// some old perls do not have this, try to make it work, no 16// some old perls do not have this, try to make it work, no
15// guarentees, though. if it breaks, you get to keep the pieces. 17// guarentees, though. if it breaks, you get to keep the pieces.
16#ifndef UTF8_MAXBYTES 18#ifndef UTF8_MAXBYTES
17# define UTF8_MAXBYTES 13 19# define UTF8_MAXBYTES 13
18#endif 20#endif
21
22#define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 2)
19 23
20#define F_ASCII 0x00000001UL 24#define F_ASCII 0x00000001UL
21#define F_LATIN1 0x00000002UL 25#define F_LATIN1 0x00000002UL
22#define F_UTF8 0x00000004UL 26#define F_UTF8 0x00000004UL
23#define F_INDENT 0x00000008UL 27#define F_INDENT 0x00000008UL
… …
26#define F_SPACE_AFTER 0x00000040UL 30#define F_SPACE_AFTER 0x00000040UL
27#define F_ALLOW_NONREF 0x00000100UL 31#define F_ALLOW_NONREF 0x00000100UL
28#define F_SHRINK 0x00000200UL 32#define F_SHRINK 0x00000200UL
29#define F_ALLOW_BLESSED 0x00000400UL 33#define F_ALLOW_BLESSED 0x00000400UL
30#define F_CONV_BLESSED 0x00000800UL 34#define F_CONV_BLESSED 0x00000800UL
31#define F_MAXDEPTH 0xf8000000UL
32#define S_MAXDEPTH 27
33#define F_MAXSIZE 0x01f00000UL 35#define F_RELAXED 0x00001000UL
34#define S_MAXSIZE 20 36#define F_ALLOW_UNKNOWN 0x00002000UL
35#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing 37#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
36 38
37#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
38#define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE ))
39
40#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 39#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
41#define F_DEFAULT (9UL << S_MAXDEPTH)
42 40
43#define INIT_SIZE 32 // initial scalar size to be allocated 41#define INIT_SIZE 32 // initial scalar size to be allocated
44#define INDENT_STEP 3 // spaces per indentation level 42#define INDENT_STEP 3 // spaces per indentation level
45 43
46#define SHORT_STRING_LEN 16384 // special-case strings of up to this size 44#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
47 45
48#define SB do { 46#define SB do {
49#define SE } while (0) 47#define SE } while (0)
50 48
51#if __GNUC__ >= 3 49#if __GNUC__ >= 3
52# define expect(expr,value) __builtin_expect ((expr),(value)) 50# define expect(expr,value) __builtin_expect ((expr), (value))
53# define inline inline 51# define INLINE static inline
54#else 52#else
55# define expect(expr,value) (expr) 53# define expect(expr,value) (expr)
56# define inline static 54# define INLINE static
57#endif 55#endif
58 56
59#define expect_false(expr) expect ((expr) != 0, 0) 57#define expect_false(expr) expect ((expr) != 0, 0)
60#define expect_true(expr) expect ((expr) != 0, 1) 58#define expect_true(expr) expect ((expr) != 0, 1)
61 59
60#define IN_RANGE_INC(type,val,beg,end) \
61 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \
62 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
63
64#define ERR_NESTING_EXCEEDED "json text or perl structure exceeds maximum nesting level (max_depth set too low?)"
65
66#ifdef USE_ITHREADS
67# define JSON_SLOW 1
68# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1))
69#else
70# define JSON_SLOW 0
71# define JSON_STASH json_stash
72#endif
73
62static HV *json_stash, *json_boolean_stash; // JSON::XS:: 74static HV *json_stash, *json_boolean_stash; // JSON::XS::
63static SV *json_true, *json_false; 75static SV *json_true, *json_false;
64 76
77enum {
78 INCR_M_WS = 0, // initial whitespace skipping, must be 0
79 INCR_M_STR, // inside string
80 INCR_M_BS, // inside backslash
81 INCR_M_JSON // outside anything, count nesting
82};
83
84#define INCR_DONE(json) ((json)->incr_nest <= 0 && (json)->incr_mode == INCR_M_JSON)
85
65typedef struct { 86typedef struct {
66 U32 flags; 87 U32 flags;
67 SV *cb_object, *cb_sk_object; 88 U32 max_depth;
89 STRLEN max_size;
90
91 SV *cb_object;
92 HV *cb_sk_object;
93
94 // for the incremental parser
95 SV *incr_text; // the source text so far
96 STRLEN incr_pos; // the current offset into the text
97 int incr_nest; // {[]}-nesting level
98 unsigned char incr_mode;
68} JSON; 99} JSON;
100
101INLINE void
102json_init (JSON *json)
103{
104 Zero (json, 1, JSON);
105 json->max_depth = 512;
106}
69 107
70///////////////////////////////////////////////////////////////////////////// 108/////////////////////////////////////////////////////////////////////////////
71// utility functions 109// utility functions
72 110
73inline void 111INLINE SV *
112get_bool (const char *name)
113{
114 SV *sv = get_sv (name, 1);
115
116 SvREADONLY_on (sv);
117 SvREADONLY_on (SvRV (sv));
118
119 return sv;
120}
121
122INLINE void
74shrink (SV *sv) 123shrink (SV *sv)
75{ 124{
76 sv_utf8_downgrade (sv, 1); 125 sv_utf8_downgrade (sv, 1);
126
77 if (SvLEN (sv) > SvCUR (sv) + 1) 127 if (SvLEN (sv) > SvCUR (sv) + 1)
78 { 128 {
79#ifdef SvPV_shrink_to_cur 129#ifdef SvPV_shrink_to_cur
80 SvPV_shrink_to_cur (sv); 130 SvPV_shrink_to_cur (sv);
81#elif defined (SvPV_renew) 131#elif defined (SvPV_renew)
… …
87// decode an utf-8 character and return it, or (UV)-1 in 137// decode an utf-8 character and return it, or (UV)-1 in
88// case of an error. 138// case of an error.
89// we special-case "safe" characters from U+80 .. U+7FF, 139// we special-case "safe" characters from U+80 .. U+7FF,
90// but use the very good perl function to parse anything else. 140// but use the very good perl function to parse anything else.
91// note that we never call this function for a ascii codepoints 141// note that we never call this function for a ascii codepoints
92inline UV 142INLINE UV
93decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) 143decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
94{ 144{
95 if (expect_false (s[0] > 0xdf || s[0] < 0xc2)) 145 if (expect_true (len >= 2
96 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); 146 && IN_RANGE_INC (char, s[0], 0xc2, 0xdf)
97 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) 147 && IN_RANGE_INC (char, s[1], 0x80, 0xbf)))
98 { 148 {
99 *clen = 2; 149 *clen = 2;
100 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 150 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
101 } 151 }
102 else 152 else
103 { 153 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
104 *clen = (STRLEN)-1; 154}
105 return (UV)-1; 155
106 } 156// likewise for encoding, also never called for ascii codepoints
157// this function takes advantage of this fact, although current gccs
158// seem to optimise the check for >= 0x80 away anyways
159INLINE unsigned char *
160encode_utf8 (unsigned char *s, UV ch)
161{
162 if (expect_false (ch < 0x000080))
163 *s++ = ch;
164 else if (expect_true (ch < 0x000800))
165 *s++ = 0xc0 | ( ch >> 6),
166 *s++ = 0x80 | ( ch & 0x3f);
167 else if ( ch < 0x010000)
168 *s++ = 0xe0 | ( ch >> 12),
169 *s++ = 0x80 | ((ch >> 6) & 0x3f),
170 *s++ = 0x80 | ( ch & 0x3f);
171 else if ( ch < 0x110000)
172 *s++ = 0xf0 | ( ch >> 18),
173 *s++ = 0x80 | ((ch >> 12) & 0x3f),
174 *s++ = 0x80 | ((ch >> 6) & 0x3f),
175 *s++ = 0x80 | ( ch & 0x3f);
176
177 return s;
107} 178}
108 179
109///////////////////////////////////////////////////////////////////////////// 180/////////////////////////////////////////////////////////////////////////////
110// encoder 181// encoder
111 182
… …
115 char *cur; // SvPVX (sv) + current output position 186 char *cur; // SvPVX (sv) + current output position
116 char *end; // SvEND (sv) 187 char *end; // SvEND (sv)
117 SV *sv; // result scalar 188 SV *sv; // result scalar
118 JSON json; 189 JSON json;
119 U32 indent; // indentation level 190 U32 indent; // indentation level
120 U32 maxdepth; // max. indentation/recursion level 191 UV limit; // escape character values >= this value when encoding
121} enc_t; 192} enc_t;
122 193
123inline void 194INLINE void
124need (enc_t *enc, STRLEN len) 195need (enc_t *enc, STRLEN len)
125{ 196{
126 if (expect_false (enc->cur + len >= enc->end)) 197 if (expect_false (enc->cur + len >= enc->end))
127 { 198 {
128 STRLEN cur = enc->cur - SvPVX (enc->sv); 199 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
129 SvGROW (enc->sv, cur + len + 1); 200 SvGROW (enc->sv, cur + len + 1);
130 enc->cur = SvPVX (enc->sv) + cur; 201 enc->cur = SvPVX (enc->sv) + cur;
131 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 202 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
132 } 203 }
133} 204}
134 205
135inline void 206INLINE void
136encode_ch (enc_t *enc, char ch) 207encode_ch (enc_t *enc, char ch)
137{ 208{
138 need (enc, 1); 209 need (enc, 1);
139 *enc->cur++ = ch; 210 *enc->cur++ = ch;
140} 211}
… …
194 { 265 {
195 uch = ch; 266 uch = ch;
196 clen = 1; 267 clen = 1;
197 } 268 }
198 269
199 if (uch > 0x10FFFFUL) 270 if (uch < 0x80/*0x20*/ || uch >= enc->limit)
200 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
201
202 if (uch < 0x80 || enc->json.flags & F_ASCII || (enc->json.flags & F_LATIN1 && uch > 0xFF))
203 { 271 {
204 if (uch > 0xFFFFUL) 272 if (uch >= 0x10000UL)
205 { 273 {
274 if (uch >= 0x110000UL)
275 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
276
206 need (enc, len += 11); 277 need (enc, len += 11);
207 sprintf (enc->cur, "\\u%04x\\u%04x", 278 sprintf (enc->cur, "\\u%04x\\u%04x",
208 (int)((uch - 0x10000) / 0x400 + 0xD800), 279 (int)((uch - 0x10000) / 0x400 + 0xD800),
209 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 280 (int)((uch - 0x10000) % 0x400 + 0xDC00));
210 enc->cur += 12; 281 enc->cur += 12;
211 } 282 }
212 else 283 else
213 { 284 {
214 static char hexdigit [16] = "0123456789abcdef";
215 need (enc, len += 5); 285 need (enc, len += 5);
216 *enc->cur++ = '\\'; 286 *enc->cur++ = '\\';
217 *enc->cur++ = 'u'; 287 *enc->cur++ = 'u';
218 *enc->cur++ = hexdigit [ uch >> 12 ]; 288 *enc->cur++ = PL_hexdigit [ uch >> 12 ];
219 *enc->cur++ = hexdigit [(uch >> 8) & 15]; 289 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
220 *enc->cur++ = hexdigit [(uch >> 4) & 15]; 290 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
221 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 291 *enc->cur++ = PL_hexdigit [(uch >> 0) & 15];
222 } 292 }
223 293
224 str += clen; 294 str += clen;
225 } 295 }
226 else if (enc->json.flags & F_LATIN1) 296 else if (enc->json.flags & F_LATIN1)
… …
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->json.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;
… …
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->json.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->json.flags & F_INDENT) 359 if (enc->json.flags & F_INDENT)
… …
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 {
… …
342 encode_ch (enc, '"'); 421 encode_ch (enc, '"');
343 422
344 if (enc->json.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->json.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->json.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
440 encode_nl (enc); 537 encode_nl (enc); --enc->indent; encode_indent (enc);
441 } 538 }
539 }
442 else 540 else
541 {
542 if (hv_iterinit (hv) || SvMAGICAL (hv))
543 if ((he = hv_iternext (hv)))
443 { 544 {
444 HE *he = hv_iternext (hv); 545 encode_nl (enc); ++enc->indent;
445 546
446 for (;;) 547 for (;;)
447 { 548 {
448 encode_indent (enc); 549 encode_indent (enc);
449 encode_he (enc, he); 550 encode_hk (enc, he);
551 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
450 552
451 if (!(he = hv_iternext (hv))) 553 if (!(he = hv_iternext (hv)))
452 break; 554 break;
453 555
454 encode_comma (enc); 556 encode_comma (enc);
455 } 557 }
456 558
457 encode_nl (enc); 559 encode_nl (enc); --enc->indent; encode_indent (enc);
458 } 560 }
459 } 561 }
460 562
461 --enc->indent; encode_indent (enc); encode_ch (enc, '}'); 563 encode_ch (enc, '}');
462} 564}
463 565
464// encode objects, arrays and special \0=false and \1=true values. 566// encode objects, arrays and special \0=false and \1=true values.
465static void 567static void
466encode_rv (enc_t *enc, SV *sv) 568encode_rv (enc_t *enc, SV *sv)
… …
470 SvGETMAGIC (sv); 572 SvGETMAGIC (sv);
471 svt = SvTYPE (sv); 573 svt = SvTYPE (sv);
472 574
473 if (expect_false (SvOBJECT (sv))) 575 if (expect_false (SvOBJECT (sv)))
474 { 576 {
577 HV *stash = !JSON_SLOW || json_boolean_stash
578 ? json_boolean_stash
579 : gv_stashpv ("JSON::XS::Boolean", 1);
580
475 if (SvSTASH (sv) == json_boolean_stash) 581 if (SvSTASH (sv) == stash)
476 { 582 {
477 if (SvIV (sv) == 0) 583 if (SvIV (sv))
584 encode_str (enc, "true", 4, 0);
585 else
478 encode_str (enc, "false", 5, 0); 586 encode_str (enc, "false", 5, 0);
479 else
480 encode_str (enc, "true", 4, 0);
481 } 587 }
482 else 588 else
483 { 589 {
484#if 0 590#if 0
485 if (0 && sv_derived_from (rv, "JSON::Literal")) 591 if (0 && sv_derived_from (rv, "JSON::Literal"))
… …
488 } 594 }
489#endif 595#endif
490 if (enc->json.flags & F_CONV_BLESSED) 596 if (enc->json.flags & F_CONV_BLESSED)
491 { 597 {
492 // we re-bless the reference to get overload and other niceties right 598 // we re-bless the reference to get overload and other niceties right
493 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 1); 599 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
494 600
495 if (to_json) 601 if (to_json)
496 { 602 {
603 dSP;
604
497 dSP; ENTER; SAVETMPS; PUSHMARK (SP); 605 ENTER; SAVETMPS; PUSHMARK (SP);
498 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv))); 606 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
499 607
500 // calling with G_SCALAR ensures that we always get a 1 reutrn value 608 // calling with G_SCALAR ensures that we always get a 1 return value
501 // check anyways.
502 PUTBACK; 609 PUTBACK;
503 assert (1 == call_sv ((SV *)GvCV (to_json), G_SCALAR)); 610 call_sv ((SV *)GvCV (to_json), G_SCALAR);
504 SPAGAIN; 611 SPAGAIN;
505 612
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
506 encode_sv (enc, POPs); 620 encode_sv (enc, sv);
507 621
508 FREETMPS; LEAVE; 622 FREETMPS; LEAVE;
509 } 623 }
510 else if (enc->json.flags & F_ALLOW_BLESSED) 624 else if (enc->json.flags & F_ALLOW_BLESSED)
511 encode_str (enc, "null", 4, 0); 625 encode_str (enc, "null", 4, 0);
… …
524 encode_hv (enc, (HV *)sv); 638 encode_hv (enc, (HV *)sv);
525 else if (svt == SVt_PVAV) 639 else if (svt == SVt_PVAV)
526 encode_av (enc, (AV *)sv); 640 encode_av (enc, (AV *)sv);
527 else if (svt < SVt_PVAV) 641 else if (svt < SVt_PVAV)
528 { 642 {
529 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')
530 encode_str (enc, "false", 5, 0); 649 encode_str (enc, "false", 5, 0);
531 else if (SvNIOK (sv) && SvIV (sv) == 1) 650 else if (enc->json.flags & F_ALLOW_UNKNOWN)
532 encode_str (enc, "true", 4, 0); 651 encode_str (enc, "null", 4, 0);
533 else 652 else
534 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",
535 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 654 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
536 } 655 }
656 else if (enc->json.flags & F_ALLOW_UNKNOWN)
657 encode_str (enc, "null", 4, 0);
537 else 658 else
538 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",
539 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 660 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
540} 661}
541 662
… …
559 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 680 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
560 enc->cur += strlen (enc->cur); 681 enc->cur += strlen (enc->cur);
561 } 682 }
562 else if (SvIOKp (sv)) 683 else if (SvIOKp (sv))
563 { 684 {
564 // 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
565 if (SvUV (sv) & ~(UV)0x7fff) 686 // we assume two's complement
566 { 687 // we assume no aliasing issues in the union
567 // large integer, use the (rather slow) snprintf way. 688 if (SvIsUV (sv) ? SvUVX (sv) <= 59000
568 need (enc, sizeof (UV) * 3); 689 : SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000)
569 enc->cur +=
570 SvIsUV(sv)
571 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
572 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
573 }
574 else
575 { 690 {
576 // optimise the "small number case" 691 // optimise the "small number case"
577 // 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
578 I32 i = SvIV (sv); 694 I32 i = SvIVX (sv);
579 U32 u; 695 U32 u;
580 char digit, nz = 0; 696 char digit, nz = 0;
581 697
582 need (enc, 6); 698 need (enc, 6);
583 699
… …
589 705
590 // 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
591 // 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,
592 // resulting in a net multiplication by 10. 708 // resulting in a net multiplication by 10.
593 // we always write the digit to memory but conditionally increment 709 // we always write the digit to memory but conditionally increment
594 // the pointer, to ease the usage of conditional move instructions. 710 // the pointer, to enable the use of conditional move instructions.
595 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;
596 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;
597 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;
598 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;
599 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'
600 } 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 }
601 } 726 }
602 else if (SvROK (sv)) 727 else if (SvROK (sv))
603 encode_rv (enc, SvRV (sv)); 728 encode_rv (enc, SvRV (sv));
604 else if (!SvOK (sv)) 729 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN)
605 encode_str (enc, "null", 4, 0); 730 encode_str (enc, "null", 4, 0);
606 else 731 else
607 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",
608 SvPV_nolen (sv), SvFLAGS (sv)); 733 SvPV_nolen (sv), SvFLAGS (sv));
609} 734}
… …
619 enc.json = *json; 744 enc.json = *json;
620 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 745 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
621 enc.cur = SvPVX (enc.sv); 746 enc.cur = SvPVX (enc.sv);
622 enc.end = SvEND (enc.sv); 747 enc.end = SvEND (enc.sv);
623 enc.indent = 0; 748 enc.indent = 0;
624 enc.maxdepth = DEC_DEPTH (enc.json.flags); 749 enc.limit = enc.json.flags & F_ASCII ? 0x000080UL
750 : enc.json.flags & F_LATIN1 ? 0x000100UL
751 : 0x110000UL;
625 752
626 SvPOK_only (enc.sv); 753 SvPOK_only (enc.sv);
627 encode_sv (&enc, scalar); 754 encode_sv (&enc, scalar);
628 755
629 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 756 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
… …
650 JSON json; 777 JSON json;
651 U32 depth; // recursion depth 778 U32 depth; // recursion depth
652 U32 maxdepth; // recursion depth limit 779 U32 maxdepth; // recursion depth limit
653} dec_t; 780} dec_t;
654 781
655inline 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
656decode_ws (dec_t *dec) 792decode_ws (dec_t *dec)
657{ 793{
658 for (;;) 794 for (;;)
659 { 795 {
660 char ch = *dec->cur; 796 char ch = *dec->cur;
661 797
662 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 }
663 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) 810 else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)
664 break; 811 break; // parse error, but let higher level handle it, gives better error messages
665 812
666 ++dec->cur; 813 ++dec->cur;
667 } 814 }
668} 815}
669 816
… …
673 if (*dec->cur != ch) \ 820 if (*dec->cur != ch) \
674 ERR (# ch " expected"); \ 821 ERR (# ch " expected"); \
675 ++dec->cur; \ 822 ++dec->cur; \
676 SE 823 SE
677 824
678#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)
679#define DEC_DEC_DEPTH --dec->depth 826#define DEC_DEC_DEPTH --dec->depth
680 827
681static SV *decode_sv (dec_t *dec); 828static SV *decode_sv (dec_t *dec);
682 829
683static signed char decode_hexdigit[256]; 830static signed char decode_hexdigit[256];
… …
775 922
776 if (hi >= 0x80) 923 if (hi >= 0x80)
777 { 924 {
778 utf8 = 1; 925 utf8 = 1;
779 926
780 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0); 927 cur = encode_utf8 (cur, hi);
781 } 928 }
782 else 929 else
783 *cur++ = hi; 930 *cur++ = hi;
784 } 931 }
785 break; 932 break;
… …
787 default: 934 default:
788 --dec_cur; 935 --dec_cur;
789 ERR ("illegal backslash escape sequence in string"); 936 ERR ("illegal backslash escape sequence in string");
790 } 937 }
791 } 938 }
792 else if (expect_true (ch >= 0x20 && ch <= 0x7f)) 939 else if (expect_true (ch >= 0x20 && ch < 0x80))
793 *cur++ = ch; 940 *cur++ = ch;
794 else if (ch >= 0x80) 941 else if (ch >= 0x80)
795 { 942 {
796 STRLEN clen; 943 STRLEN clen;
797 UV uch; 944 UV uch;
… …
918 is_nv = 1; 1065 is_nv = 1;
919 } 1066 }
920 1067
921 if (!is_nv) 1068 if (!is_nv)
922 { 1069 {
1070 int len = dec->cur - start;
1071
923 // 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
924 if (*start == '-') 1073 if (*start == '-')
925 switch (dec->cur - start) 1074 switch (len)
926 { 1075 {
927 case 2: return newSViv (-( start [1] - '0' * 1)); 1076 case 2: return newSViv (-( start [1] - '0' * 1));
928 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); 1077 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
929 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));
930 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));
931 } 1081 }
932 else 1082 else
933 switch (dec->cur - start) 1083 switch (len)
934 { 1084 {
935 case 1: return newSViv ( start [0] - '0' * 1); 1085 case 1: return newSViv ( start [0] - '0' * 1);
936 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); 1086 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
937 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);
938 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);
939 } 1090 }
940 1091
941 { 1092 {
942 UV uv; 1093 UV uv;
943 int numtype = grok_number (start, dec->cur - start, &uv); 1094 int numtype = grok_number (start, len, &uv);
944 if (numtype & IS_NUMBER_IN_UV) 1095 if (numtype & IS_NUMBER_IN_UV)
945 if (numtype & IS_NUMBER_NEG) 1096 if (numtype & IS_NUMBER_NEG)
946 { 1097 {
947 if (uv < (UV)IV_MIN) 1098 if (uv < (UV)IV_MIN)
948 return newSViv (-(IV)uv); 1099 return newSViv (-(IV)uv);
949 } 1100 }
950 else 1101 else
951 return newSVuv (uv); 1102 return newSVuv (uv);
952
953 // here would likely be the place for bigint support
954 } 1103 }
955 }
956 1104
957 // 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
958 return newSVnv (Atof (start)); 1121 return newSVnv (Atof (start));
959 1122
960fail: 1123fail:
961 return 0; 1124 return 0;
962} 1125}
… …
992 1155
993 if (*dec->cur != ',') 1156 if (*dec->cur != ',')
994 ERR (", or ] expected while parsing array"); 1157 ERR (", or ] expected while parsing array");
995 1158
996 ++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 }
997 } 1168 }
998 1169
999 DEC_DEC_DEPTH; 1170 DEC_DEC_DEPTH;
1000 return newRV_noinc ((SV *)av); 1171 return newRV_noinc ((SV *)av);
1001 1172
… …
1017 if (*dec->cur == '}') 1188 if (*dec->cur == '}')
1018 ++dec->cur; 1189 ++dec->cur;
1019 else 1190 else
1020 for (;;) 1191 for (;;)
1021 { 1192 {
1022 decode_ws (dec); EXPECT_CH ('"'); 1193 EXPECT_CH ('"');
1023 1194
1024 // heuristic: assume that 1195 // heuristic: assume that
1025 // a) decode_str + hv_store_ent are abysmally slow. 1196 // a) decode_str + hv_store_ent are abysmally slow.
1026 // b) most hash keys are short, simple ascii text. 1197 // b) most hash keys are short, simple ascii text.
1027 // => try to "fast-match" such strings to avoid 1198 // => try to "fast-match" such strings to avoid
… …
1031 char *p = dec->cur; 1202 char *p = dec->cur;
1032 char *e = p + 24; // only try up to 24 bytes 1203 char *e = p + 24; // only try up to 24 bytes
1033 1204
1034 for (;;) 1205 for (;;)
1035 { 1206 {
1036 // the >= 0x80 is true on most architectures 1207 // the >= 0x80 is false on most architectures
1037 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\') 1208 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1038 { 1209 {
1039 // slow path, back up and use decode_str 1210 // slow path, back up and use decode_str
1040 SV *key = decode_str (dec); 1211 SV *key = decode_str (dec);
1041 if (!key) 1212 if (!key)
1042 goto fail; 1213 goto fail;
1043 1214
1044 decode_ws (dec); EXPECT_CH (':'); 1215 decode_ws (dec); EXPECT_CH (':');
1045 1216
1217 decode_ws (dec);
1046 value = decode_sv (dec); 1218 value = decode_sv (dec);
1047 if (!value) 1219 if (!value)
1048 { 1220 {
1049 SvREFCNT_dec (key); 1221 SvREFCNT_dec (key);
1050 goto fail; 1222 goto fail;
… …
1062 int len = p - key; 1234 int len = p - key;
1063 dec->cur = p + 1; 1235 dec->cur = p + 1;
1064 1236
1065 decode_ws (dec); EXPECT_CH (':'); 1237 decode_ws (dec); EXPECT_CH (':');
1066 1238
1239 decode_ws (dec);
1067 value = decode_sv (dec); 1240 value = decode_sv (dec);
1068 if (!value) 1241 if (!value)
1069 goto fail; 1242 goto fail;
1070 1243
1071 hv_store (hv, key, len, value, 0); 1244 hv_store (hv, key, len, value, 0);
… …
1087 1260
1088 if (*dec->cur != ',') 1261 if (*dec->cur != ',')
1089 ERR (", or } expected while parsing object/hash"); 1262 ERR (", or } expected while parsing object/hash");
1090 1263
1091 ++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 }
1092 } 1273 }
1093 1274
1094 DEC_DEC_DEPTH; 1275 DEC_DEC_DEPTH;
1095 sv = newRV_noinc ((SV *)hv); 1276 sv = newRV_noinc ((SV *)hv);
1096 1277
1097 // check filter callbacks 1278 // check filter callbacks
1098 if (dec->json.flags & F_HOOK) 1279 if (dec->json.flags & F_HOOK)
1099 { 1280 {
1100 ENTER; SAVETMPS; 1281 if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
1101
1102 if (HvKEYS (hv) == 1 && dec->json.cb_sk_object)
1103 { 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;
1104 int count; 1296 int count;
1105 1297
1106 dSP; PUSHMARK (SP); 1298 ENTER; SAVETMPS; PUSHMARK (SP);
1107 XPUSHs (sv_2mortal (sv)); 1299 XPUSHs (HeVAL (he));
1108 1300
1109 PUTBACK; count = call_sv (dec->json.cb_sk_object, G_ARRAY); SPAGAIN; 1301 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1110 1302
1111 if (count == 1) 1303 if (count == 1)
1304 {
1112 sv = newSVsv (POPs); 1305 sv = newSVsv (POPs);
1113 else 1306 FREETMPS; LEAVE;
1114 SvREFCNT_inc (sv); 1307 return sv;
1308 }
1309
1310 FREETMPS; LEAVE;
1311 }
1115 } 1312 }
1116 1313
1117 if (dec->json.cb_object) 1314 if (dec->json.cb_object)
1118 { 1315 {
1316 dSP;
1119 int count; 1317 int count;
1120 1318
1121 dSP; ENTER; SAVETMPS; PUSHMARK (SP); 1319 ENTER; SAVETMPS; PUSHMARK (SP);
1122 XPUSHs (sv_2mortal (sv)); 1320 XPUSHs (sv_2mortal (sv));
1123 1321
1124 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN; 1322 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1125 1323
1126 if (count == 1) 1324 if (count == 1)
1325 {
1127 sv = newSVsv (POPs); 1326 sv = newSVsv (POPs);
1128 else 1327 FREETMPS; LEAVE;
1328 return sv;
1329 }
1330
1129 SvREFCNT_inc (sv); 1331 SvREFCNT_inc (sv);
1332 FREETMPS; LEAVE;
1130 } 1333 }
1131
1132 FREETMPS; LEAVE;
1133 } 1334 }
1134 1335
1135 return newRV_noinc ((SV *)hv); 1336 return sv;
1136 1337
1137fail: 1338fail:
1138 SvREFCNT_dec (hv); 1339 SvREFCNT_dec (hv);
1139 DEC_DEC_DEPTH; 1340 DEC_DEC_DEPTH;
1140 return 0; 1341 return 0;
1141} 1342}
1142 1343
1143static SV * 1344static SV *
1144decode_sv (dec_t *dec) 1345decode_sv (dec_t *dec)
1145{ 1346{
1146 decode_ws (dec);
1147
1148 // the beauty of JSON: you need exactly one character lookahead 1347 // the beauty of JSON: you need exactly one character lookahead
1149 // to parse anything. 1348 // to parse everything.
1150 switch (*dec->cur) 1349 switch (*dec->cur)
1151 { 1350 {
1152 case '"': ++dec->cur; return decode_str (dec); 1351 case '"': ++dec->cur; return decode_str (dec);
1153 case '[': ++dec->cur; return decode_av (dec); 1352 case '[': ++dec->cur; return decode_av (dec);
1154 case '{': ++dec->cur; return decode_hv (dec); 1353 case '{': ++dec->cur; return decode_hv (dec);
1155 1354
1156 case '-': 1355 case '-':
1157 case '0': case '1': case '2': case '3': case '4': 1356 case '0': case '1': case '2': case '3': case '4':
1158 case '5': case '6': case '7': case '8': case '9': 1357 case '5': case '6': case '7': case '8': case '9':
1159 return decode_num (dec); 1358 return decode_num (dec);
1160 1359
1161 case 't': 1360 case 't':
1162 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1361 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1163 { 1362 {
1164 dec->cur += 4; 1363 dec->cur += 4;
1364#if JSON_SLOW
1365 json_true = get_bool ("JSON::XS::true");
1366#endif
1165 return SvREFCNT_inc (json_true); 1367 return newSVsv (json_true);
1166 } 1368 }
1167 else 1369 else
1168 ERR ("'true' expected"); 1370 ERR ("'true' expected");
1169 1371
1170 break; 1372 break;
1171 1373
1172 case 'f': 1374 case 'f':
1173 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1375 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1174 { 1376 {
1175 dec->cur += 5; 1377 dec->cur += 5;
1378#if JSON_SLOW
1379 json_false = get_bool ("JSON::XS::false");
1380#endif
1176 return SvREFCNT_inc (json_false); 1381 return newSVsv (json_false);
1177 } 1382 }
1178 else 1383 else
1179 ERR ("'false' expected"); 1384 ERR ("'false' expected");
1180 1385
1181 break; 1386 break;
… …
1199fail: 1404fail:
1200 return 0; 1405 return 0;
1201} 1406}
1202 1407
1203static SV * 1408static SV *
1204decode_json (SV *string, JSON *json, UV *offset_return) 1409decode_json (SV *string, JSON *json, STRLEN *offset_return)
1205{ 1410{
1206 dec_t dec; 1411 dec_t dec;
1207 UV offset; 1412 STRLEN offset;
1208 SV *sv; 1413 SV *sv;
1209 1414
1415 /* work around bugs in 5.10 where manipulating magic values
1416 * will perl ignore the magic in subsequent accesses
1417 */
1210 SvGETMAGIC (string); 1418 /*SvGETMAGIC (string);*/
1419 if (SvMAGICAL (string))
1420 string = sv_2mortal (newSVsv (string));
1421
1211 SvUPGRADE (string, SVt_PV); 1422 SvUPGRADE (string, SVt_PV);
1212 1423
1213 if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags)) 1424 /* work around a bug in perl 5.10, which causes SvCUR to fail an
1425 * assertion with -DDEBUGGING, although SvCUR is documented to
1426 * return the xpv_cur field which certainly exists after upgrading.
1427 * according to nicholas clark, calling SvPOK fixes this.
1428 * But it doesn't fix it, so try another workaround, call SvPV_nolen
1429 * and hope for the best.
1430 * Damnit, SvPV_nolen still trips over yet another assertion. This
1431 * assertion business is seriously broken, try yet another workaround
1432 * for the broken -DDEBUGGING.
1433 */
1434#ifdef DEBUGGING
1435 offset = SvOK (string) ? sv_len (string) : 0;
1436#else
1437 offset = SvCUR (string);
1438#endif
1439
1440 if (offset > json->max_size && json->max_size)
1214 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", 1441 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1215 (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags)); 1442 (unsigned long)SvCUR (string), (unsigned long)json->max_size);
1216 1443
1217 if (json->flags & F_UTF8) 1444 if (json->flags & F_UTF8)
1218 sv_utf8_downgrade (string, 0); 1445 sv_utf8_downgrade (string, 0);
1219 else 1446 else
1220 sv_utf8_upgrade (string); 1447 sv_utf8_upgrade (string);
1221 1448
1222 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1449 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
1223 1450
1224 dec.json = *json; 1451 dec.json = *json;
1225 dec.cur = SvPVX (string); 1452 dec.cur = SvPVX (string);
1226 dec.end = SvEND (string); 1453 dec.end = SvEND (string);
1227 dec.err = 0; 1454 dec.err = 0;
1228 dec.depth = 0; 1455 dec.depth = 0;
1229 dec.maxdepth = DEC_DEPTH (dec.json.flags);
1230 1456
1231 if (dec.json.cb_object || dec.json.cb_sk_object) 1457 if (dec.json.cb_object || dec.json.cb_sk_object)
1232 dec.json.flags |= F_HOOK; 1458 dec.json.flags |= F_HOOK;
1233 1459
1234 *dec.end = 0; // this should basically be a nop, too, but make sure it's there 1460 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1461
1462 decode_ws (&dec);
1235 sv = decode_sv (&dec); 1463 sv = decode_sv (&dec);
1236 1464
1237 if (!(offset_return || !sv)) 1465 if (!(offset_return || !sv))
1238 { 1466 {
1239 // check for trailing garbage 1467 // check for trailing garbage
… …
1283 1511
1284 return sv; 1512 return sv;
1285} 1513}
1286 1514
1287///////////////////////////////////////////////////////////////////////////// 1515/////////////////////////////////////////////////////////////////////////////
1516// incremental parser
1517
1518static void
1519incr_parse (JSON *self)
1520{
1521 const char *p = SvPVX (self->incr_text) + self->incr_pos;
1522
1523 for (;;)
1524 {
1525 //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
1526 switch (self->incr_mode)
1527 {
1528 // only used for intiial whitespace skipping
1529 case INCR_M_WS:
1530 for (;;)
1531 {
1532 if (*p > 0x20)
1533 {
1534 self->incr_mode = INCR_M_JSON;
1535 goto incr_m_json;
1536 }
1537 else if (!*p)
1538 goto interrupt;
1539
1540 ++p;
1541 }
1542
1543 // skip a single char inside a string (for \\-processing)
1544 case INCR_M_BS:
1545 if (!*p)
1546 goto interrupt;
1547
1548 ++p;
1549 self->incr_mode = INCR_M_STR;
1550 goto incr_m_str;
1551
1552 // inside a string
1553 case INCR_M_STR:
1554 incr_m_str:
1555 for (;;)
1556 {
1557 if (*p == '"')
1558 {
1559 ++p;
1560 self->incr_mode = INCR_M_JSON;
1561
1562 if (!self->incr_nest)
1563 goto interrupt;
1564
1565 goto incr_m_json;
1566 }
1567 else if (*p == '\\')
1568 {
1569 ++p; // "virtually" consumes character after \
1570
1571 if (!*p) // if at end of string we have to switch modes
1572 {
1573 self->incr_mode = INCR_M_BS;
1574 goto interrupt;
1575 }
1576 }
1577 else if (!*p)
1578 goto interrupt;
1579
1580 ++p;
1581 }
1582
1583 // after initial ws, outside string
1584 case INCR_M_JSON:
1585 incr_m_json:
1586 for (;;)
1587 {
1588 switch (*p++)
1589 {
1590 case 0:
1591 --p;
1592 goto interrupt;
1593
1594 case 0x09:
1595 case 0x0a:
1596 case 0x0d:
1597 case 0x20:
1598 if (!self->incr_nest)
1599 {
1600 --p; // do not eat the whitespace, let the next round do it
1601 goto interrupt;
1602 }
1603 break;
1604
1605 case '"':
1606 self->incr_mode = INCR_M_STR;
1607 goto incr_m_str;
1608
1609 case '[':
1610 case '{':
1611 if (++self->incr_nest > self->max_depth)
1612 croak (ERR_NESTING_EXCEEDED);
1613 break;
1614
1615 case ']':
1616 case '}':
1617 if (--self->incr_nest <= 0)
1618 goto interrupt;
1619 }
1620 }
1621 }
1622
1623 modechange:
1624 ;
1625 }
1626
1627interrupt:
1628 self->incr_pos = p - SvPVX (self->incr_text);
1629 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D
1630}
1631
1632/////////////////////////////////////////////////////////////////////////////
1288// XS interface functions 1633// XS interface functions
1289 1634
1290MODULE = JSON::XS PACKAGE = JSON::XS 1635MODULE = JSON::XS PACKAGE = JSON::XS
1291 1636
1292BOOT: 1637BOOT:
… …
1301 : -1; 1646 : -1;
1302 1647
1303 json_stash = gv_stashpv ("JSON::XS" , 1); 1648 json_stash = gv_stashpv ("JSON::XS" , 1);
1304 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); 1649 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1305 1650
1306 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true ); 1651 json_true = get_bool ("JSON::XS::true");
1307 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); 1652 json_false = get_bool ("JSON::XS::false");
1308} 1653}
1309 1654
1310PROTOTYPES: DISABLE 1655PROTOTYPES: DISABLE
1656
1657void CLONE (...)
1658 CODE:
1659 json_stash = 0;
1660 json_boolean_stash = 0;
1311 1661
1312void new (char *klass) 1662void new (char *klass)
1313 PPCODE: 1663 PPCODE:
1314{ 1664{
1315 SV *pv = NEWSV (0, sizeof (JSON)); 1665 SV *pv = NEWSV (0, sizeof (JSON));
1316 SvPOK_only (pv); 1666 SvPOK_only (pv);
1317 Zero (SvPVX (pv), 1, JSON); 1667 json_init ((JSON *)SvPVX (pv));
1318 ((JSON *)SvPVX (pv))->flags = F_DEFAULT; 1668 XPUSHs (sv_2mortal (sv_bless (
1319 XPUSHs (sv_2mortal (sv_bless (newRV_noinc (pv), json_stash))); 1669 newRV_noinc (pv),
1670 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1671 )));
1320} 1672}
1321 1673
1322void ascii (JSON *self, int enable = 1) 1674void ascii (JSON *self, int enable = 1)
1323 ALIAS: 1675 ALIAS:
1324 ascii = F_ASCII 1676 ascii = F_ASCII
… …
1331 pretty = F_PRETTY 1683 pretty = F_PRETTY
1332 allow_nonref = F_ALLOW_NONREF 1684 allow_nonref = F_ALLOW_NONREF
1333 shrink = F_SHRINK 1685 shrink = F_SHRINK
1334 allow_blessed = F_ALLOW_BLESSED 1686 allow_blessed = F_ALLOW_BLESSED
1335 convert_blessed = F_CONV_BLESSED 1687 convert_blessed = F_CONV_BLESSED
1688 relaxed = F_RELAXED
1689 allow_unknown = F_ALLOW_UNKNOWN
1336 PPCODE: 1690 PPCODE:
1337{ 1691{
1338 if (enable) 1692 if (enable)
1339 self->flags |= ix; 1693 self->flags |= ix;
1340 else 1694 else
1341 self->flags &= ~ix; 1695 self->flags &= ~ix;
1342 1696
1343 XPUSHs (ST (0)); 1697 XPUSHs (ST (0));
1344} 1698}
1345 1699
1346void max_depth (JSON *self, UV max_depth = 0x80000000UL) 1700void get_ascii (JSON *self)
1701 ALIAS:
1702 get_ascii = F_ASCII
1703 get_latin1 = F_LATIN1
1704 get_utf8 = F_UTF8
1705 get_indent = F_INDENT
1706 get_canonical = F_CANONICAL
1707 get_space_before = F_SPACE_BEFORE
1708 get_space_after = F_SPACE_AFTER
1709 get_allow_nonref = F_ALLOW_NONREF
1710 get_shrink = F_SHRINK
1711 get_allow_blessed = F_ALLOW_BLESSED
1712 get_convert_blessed = F_CONV_BLESSED
1713 get_relaxed = F_RELAXED
1714 get_allow_unknown = F_ALLOW_UNKNOWN
1347 PPCODE: 1715 PPCODE:
1348{ 1716 XPUSHs (boolSV (self->flags & ix));
1349 UV log2 = 0;
1350 1717
1351 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL; 1718void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
1352 1719 PPCODE:
1353 while ((1UL << log2) < max_depth) 1720 self->max_depth = max_depth;
1354 ++log2;
1355
1356 self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1357
1358 XPUSHs (ST (0)); 1721 XPUSHs (ST (0));
1359}
1360 1722
1723U32 get_max_depth (JSON *self)
1724 CODE:
1725 RETVAL = self->max_depth;
1726 OUTPUT:
1727 RETVAL
1728
1361void max_size (JSON *self, UV max_size = 0) 1729void max_size (JSON *self, U32 max_size = 0)
1362 PPCODE: 1730 PPCODE:
1363{ 1731 self->max_size = max_size;
1364 UV log2 = 0;
1365
1366 if (max_size > 0x80000000UL) max_size = 0x80000000UL;
1367 if (max_size == 1) max_size = 2;
1368
1369 while ((1UL << log2) < max_size)
1370 ++log2;
1371
1372 self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1373
1374 XPUSHs (ST (0)); 1732 XPUSHs (ST (0));
1375}
1376 1733
1734int get_max_size (JSON *self)
1735 CODE:
1736 RETVAL = self->max_size;
1737 OUTPUT:
1738 RETVAL
1739
1377void filter_json_objects (JSON *self, SV *cb = &PL_sv_undef) 1740void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1378 ALIAS:
1379 filter_sk_json_objects = 1
1380 PPCODE: 1741 PPCODE:
1381{ 1742{
1743 SvREFCNT_dec (self->cb_object);
1744 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
1745
1746 XPUSHs (ST (0));
1747}
1748
1749void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
1750 PPCODE:
1751{
1752 if (!self->cb_sk_object)
1753 self->cb_sk_object = newHV ();
1754
1382 if (!SvOK (cb)) 1755 if (SvOK (cb))
1383 cb = 0; 1756 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1384 1757 else
1385 switch (ix)
1386 { 1758 {
1387 case 0: self->cb_object = cb; break; 1759 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1760
1761 if (!HvKEYS (self->cb_sk_object))
1762 {
1763 SvREFCNT_dec (self->cb_sk_object);
1388 case 1: self->cb_sk_object = cb; break; 1764 self->cb_sk_object = 0;
1765 }
1389 } 1766 }
1390 1767
1391 XPUSHs (ST (0)); 1768 XPUSHs (ST (0));
1392} 1769}
1393 1770
… …
1400 XPUSHs (decode_json (jsonstr, self, 0)); 1777 XPUSHs (decode_json (jsonstr, self, 0));
1401 1778
1402void decode_prefix (JSON *self, SV *jsonstr) 1779void decode_prefix (JSON *self, SV *jsonstr)
1403 PPCODE: 1780 PPCODE:
1404{ 1781{
1405 UV offset; 1782 STRLEN offset;
1406 EXTEND (SP, 2); 1783 EXTEND (SP, 2);
1407 PUSHs (decode_json (jsonstr, self, &offset)); 1784 PUSHs (decode_json (jsonstr, self, &offset));
1408 PUSHs (sv_2mortal (newSVuv (offset))); 1785 PUSHs (sv_2mortal (newSVuv (offset)));
1409} 1786}
1410 1787
1788void incr_parse (JSON *self, SV *jsonstr = 0)
1789 PPCODE:
1790{
1791 if (!self->incr_text)
1792 self->incr_text = newSVpvn ("", 0);
1793
1794 // append data, if any
1795 if (jsonstr)
1796 {
1797 if (SvUTF8 (jsonstr) && !SvUTF8 (self->incr_text))
1798 {
1799 /* utf-8-ness differs, need to upgrade */
1800 sv_utf8_upgrade (self->incr_text);
1801
1802 if (self->incr_pos)
1803 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
1804 - (U8 *)SvPVX (self->incr_text);
1805 }
1806
1807 {
1808 STRLEN len;
1809 const char *str = SvPV (jsonstr, len);
1810 SvGROW (self->incr_text, SvCUR (self->incr_text) + len + 1);
1811 Move (str, SvEND (self->incr_text), len, char);
1812 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
1813 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
1814 }
1815 }
1816
1817 if (GIMME_V != G_VOID)
1818 do
1819 {
1820 STRLEN offset;
1821
1822 if (!INCR_DONE (self))
1823 {
1824 incr_parse (self);
1825
1826 if (self->incr_pos > self->max_size && self->max_size)
1827 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1828 (unsigned long)self->incr_pos, (unsigned long)self->max_size);
1829
1830 if (!INCR_DONE (self))
1831 break;
1832 }
1833
1834 XPUSHs (decode_json (self->incr_text, self, &offset));
1835
1836 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + offset);
1837 self->incr_pos -= offset;
1838 self->incr_nest = 0;
1839 self->incr_mode = 0;
1840 }
1841 while (GIMME_V == G_ARRAY);
1842}
1843
1844SV *incr_text (JSON *self)
1845 ATTRS: lvalue
1846 CODE:
1847{
1848 if (self->incr_pos)
1849 croak ("incr_text can not be called when the incremental parser already started parsing");
1850
1851 RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef;
1852}
1853 OUTPUT:
1854 RETVAL
1855
1856void incr_skip (JSON *self)
1857 CODE:
1858{
1859 if (self->incr_pos)
1860 {
1861 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + self->incr_pos);
1862 self->incr_pos = 0;
1863 self->incr_nest = 0;
1864 self->incr_mode = 0;
1865 }
1866}
1867
1868void incr_reset (JSON *self)
1869 CODE:
1870{
1871 SvREFCNT_dec (self->incr_text);
1872 self->incr_text = 0;
1873 self->incr_pos = 0;
1874 self->incr_nest = 0;
1875 self->incr_mode = 0;
1876}
1877
1878void DESTROY (JSON *self)
1879 CODE:
1880 SvREFCNT_dec (self->cb_sk_object);
1881 SvREFCNT_dec (self->cb_object);
1882 SvREFCNT_dec (self->incr_text);
1883
1411PROTOTYPES: ENABLE 1884PROTOTYPES: ENABLE
1412 1885
1413void to_json (SV *scalar) 1886void encode_json (SV *scalar)
1887 ALIAS:
1888 to_json_ = 0
1889 encode_json = F_UTF8
1414 PPCODE: 1890 PPCODE:
1415{ 1891{
1416 JSON json = { F_DEFAULT | F_UTF8 }; 1892 JSON json;
1893 json_init (&json);
1894 json.flags |= ix;
1417 XPUSHs (encode_json (scalar, &json)); 1895 XPUSHs (encode_json (scalar, &json));
1418} 1896}
1419 1897
1420void from_json (SV *jsonstr) 1898void decode_json (SV *jsonstr)
1899 ALIAS:
1900 from_json_ = 0
1901 decode_json = F_UTF8
1421 PPCODE: 1902 PPCODE:
1422{ 1903{
1423 JSON json = { F_DEFAULT | F_UTF8 }; 1904 JSON json;
1905 json_init (&json);
1906 json.flags |= ix;
1424 XPUSHs (decode_json (jsonstr, &json, 0)); 1907 XPUSHs (decode_json (jsonstr, &json, 0));
1425} 1908}
1426 1909
1910

Diff Legend

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