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 1.64 by root, Mon Aug 27 02:03:23 2007 UTC vs.
Revision 1.94 by root, Thu Nov 20 03:59:53 2008 UTC

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>
9#include <float.h> 10#include <float.h>
10 11
11#if defined(__BORLANDC__) || defined(_MSC_VER) 12#if defined(__BORLANDC__) || defined(_MSC_VER)
12# define snprintf _snprintf // C compilers have this in stdio.h 13# define snprintf _snprintf // C compilers have this in stdio.h
13#endif 14#endif
15// 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
16// guarentees, though. if it breaks, you get to keep the pieces. 17// guarentees, though. if it breaks, you get to keep the pieces.
17#ifndef UTF8_MAXBYTES 18#ifndef UTF8_MAXBYTES
18# define UTF8_MAXBYTES 13 19# define UTF8_MAXBYTES 13
19#endif 20#endif
21
22#define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 2)
20 23
21#define F_ASCII 0x00000001UL 24#define F_ASCII 0x00000001UL
22#define F_LATIN1 0x00000002UL 25#define F_LATIN1 0x00000002UL
23#define F_UTF8 0x00000004UL 26#define F_UTF8 0x00000004UL
24#define F_INDENT 0x00000008UL 27#define F_INDENT 0x00000008UL
28#define F_ALLOW_NONREF 0x00000100UL 31#define F_ALLOW_NONREF 0x00000100UL
29#define F_SHRINK 0x00000200UL 32#define F_SHRINK 0x00000200UL
30#define F_ALLOW_BLESSED 0x00000400UL 33#define F_ALLOW_BLESSED 0x00000400UL
31#define F_CONV_BLESSED 0x00000800UL 34#define F_CONV_BLESSED 0x00000800UL
32#define F_RELAXED 0x00001000UL 35#define F_RELAXED 0x00001000UL
33 36#define F_ALLOW_UNKNOWN 0x00002000UL
34#define F_MAXDEPTH 0xf8000000UL
35#define S_MAXDEPTH 27
36#define F_MAXSIZE 0x01f00000UL
37#define S_MAXSIZE 20
38#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing 37#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
39 38
40#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
41#define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE ))
42
43#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 39#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
44#define F_DEFAULT (9UL << S_MAXDEPTH)
45 40
46#define INIT_SIZE 32 // initial scalar size to be allocated 41#define INIT_SIZE 32 // initial scalar size to be allocated
47#define INDENT_STEP 3 // spaces per indentation level 42#define INDENT_STEP 3 // spaces per indentation level
48 43
49#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
50 45
51#define SB do { 46#define SB do {
52#define SE } while (0) 47#define SE } while (0)
53 48
54#if __GNUC__ >= 3 49#if __GNUC__ >= 3
55# define expect(expr,value) __builtin_expect ((expr),(value)) 50# define expect(expr,value) __builtin_expect ((expr), (value))
56# define inline inline 51# define INLINE static inline
57#else 52#else
58# define expect(expr,value) (expr) 53# define expect(expr,value) (expr)
59# define inline static 54# define INLINE static
60#endif 55#endif
61 56
62#define expect_false(expr) expect ((expr) != 0, 0) 57#define expect_false(expr) expect ((expr) != 0, 0)
63#define expect_true(expr) expect ((expr) != 0, 1) 58#define expect_true(expr) expect ((expr) != 0, 1)
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?)"
64 65
65#ifdef USE_ITHREADS 66#ifdef USE_ITHREADS
66# define JSON_SLOW 1 67# define JSON_SLOW 1
67# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1)) 68# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1))
68#else 69#else
71#endif 72#endif
72 73
73static HV *json_stash, *json_boolean_stash; // JSON::XS:: 74static HV *json_stash, *json_boolean_stash; // JSON::XS::
74static SV *json_true, *json_false; 75static SV *json_true, *json_false;
75 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
76typedef struct { 86typedef struct {
77 U32 flags; 87 U32 flags;
88 U32 max_depth;
89 STRLEN max_size;
90
78 SV *cb_object; 91 SV *cb_object;
79 HV *cb_sk_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;
80} JSON; 99} JSON;
100
101INLINE void
102json_init (JSON *json)
103{
104 Zero (json, 1, JSON);
105 json->max_depth = 512;
106}
81 107
82///////////////////////////////////////////////////////////////////////////// 108/////////////////////////////////////////////////////////////////////////////
83// utility functions 109// utility functions
84 110
85inline 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
86shrink (SV *sv) 123shrink (SV *sv)
87{ 124{
88 sv_utf8_downgrade (sv, 1); 125 sv_utf8_downgrade (sv, 1);
126
89 if (SvLEN (sv) > SvCUR (sv) + 1) 127 if (SvLEN (sv) > SvCUR (sv) + 1)
90 { 128 {
91#ifdef SvPV_shrink_to_cur 129#ifdef SvPV_shrink_to_cur
92 SvPV_shrink_to_cur (sv); 130 SvPV_shrink_to_cur (sv);
93#elif defined (SvPV_renew) 131#elif defined (SvPV_renew)
99// 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
100// case of an error. 138// case of an error.
101// we special-case "safe" characters from U+80 .. U+7FF, 139// we special-case "safe" characters from U+80 .. U+7FF,
102// but use the very good perl function to parse anything else. 140// but use the very good perl function to parse anything else.
103// note that we never call this function for a ascii codepoints 141// note that we never call this function for a ascii codepoints
104inline UV 142INLINE UV
105decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) 143decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
106{ 144{
107 if (expect_false (s[0] > 0xdf || s[0] < 0xc2)) 145 if (expect_true (len >= 2
108 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); 146 && IN_RANGE_INC (char, s[0], 0xc2, 0xdf)
109 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) 147 && IN_RANGE_INC (char, s[1], 0x80, 0xbf)))
110 { 148 {
111 *clen = 2; 149 *clen = 2;
112 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 150 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
113 } 151 }
114 else 152 else
115 { 153 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
116 *clen = (STRLEN)-1; 154}
117 return (UV)-1; 155
118 } 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;
119} 178}
120 179
121///////////////////////////////////////////////////////////////////////////// 180/////////////////////////////////////////////////////////////////////////////
122// encoder 181// encoder
123 182
127 char *cur; // SvPVX (sv) + current output position 186 char *cur; // SvPVX (sv) + current output position
128 char *end; // SvEND (sv) 187 char *end; // SvEND (sv)
129 SV *sv; // result scalar 188 SV *sv; // result scalar
130 JSON json; 189 JSON json;
131 U32 indent; // indentation level 190 U32 indent; // indentation level
132 U32 maxdepth; // max. indentation/recursion level 191 UV limit; // escape character values >= this value when encoding
133} enc_t; 192} enc_t;
134 193
135inline void 194INLINE void
136need (enc_t *enc, STRLEN len) 195need (enc_t *enc, STRLEN len)
137{ 196{
138 if (expect_false (enc->cur + len >= enc->end)) 197 if (expect_false (enc->cur + len >= enc->end))
139 { 198 {
140 STRLEN cur = enc->cur - SvPVX (enc->sv); 199 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
141 SvGROW (enc->sv, cur + len + 1); 200 SvGROW (enc->sv, cur + len + 1);
142 enc->cur = SvPVX (enc->sv) + cur; 201 enc->cur = SvPVX (enc->sv) + cur;
143 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 202 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
144 } 203 }
145} 204}
146 205
147inline void 206INLINE void
148encode_ch (enc_t *enc, char ch) 207encode_ch (enc_t *enc, char ch)
149{ 208{
150 need (enc, 1); 209 need (enc, 1);
151 *enc->cur++ = ch; 210 *enc->cur++ = ch;
152} 211}
206 { 265 {
207 uch = ch; 266 uch = ch;
208 clen = 1; 267 clen = 1;
209 } 268 }
210 269
211 if (uch > 0x10FFFFUL) 270 if (uch < 0x80/*0x20*/ || uch >= enc->limit)
212 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
213
214 if (uch < 0x80 || enc->json.flags & F_ASCII || (enc->json.flags & F_LATIN1 && uch > 0xFF))
215 { 271 {
216 if (uch > 0xFFFFUL) 272 if (uch >= 0x10000UL)
217 { 273 {
274 if (uch >= 0x110000UL)
275 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
276
218 need (enc, len += 11); 277 need (enc, len += 11);
219 sprintf (enc->cur, "\\u%04x\\u%04x", 278 sprintf (enc->cur, "\\u%04x\\u%04x",
220 (int)((uch - 0x10000) / 0x400 + 0xD800), 279 (int)((uch - 0x10000) / 0x400 + 0xD800),
221 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 280 (int)((uch - 0x10000) % 0x400 + 0xDC00));
222 enc->cur += 12; 281 enc->cur += 12;
223 } 282 }
224 else 283 else
225 { 284 {
226 static char hexdigit [16] = "0123456789abcdef";
227 need (enc, len += 5); 285 need (enc, len += 5);
228 *enc->cur++ = '\\'; 286 *enc->cur++ = '\\';
229 *enc->cur++ = 'u'; 287 *enc->cur++ = 'u';
230 *enc->cur++ = hexdigit [ uch >> 12 ]; 288 *enc->cur++ = PL_hexdigit [ uch >> 12 ];
231 *enc->cur++ = hexdigit [(uch >> 8) & 15]; 289 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
232 *enc->cur++ = hexdigit [(uch >> 4) & 15]; 290 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
233 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 291 *enc->cur++ = PL_hexdigit [(uch >> 0) & 15];
234 } 292 }
235 293
236 str += clen; 294 str += clen;
237 } 295 }
238 else if (enc->json.flags & F_LATIN1) 296 else if (enc->json.flags & F_LATIN1)
250 while (--clen); 308 while (--clen);
251 } 309 }
252 else 310 else
253 { 311 {
254 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
255 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 313 enc->cur = encode_utf8 (enc->cur, uch);
256 ++str; 314 ++str;
257 } 315 }
258 } 316 }
259 } 317 }
260 } 318 }
261 319
262 --len; 320 --len;
263 } 321 }
264} 322}
265 323
266inline void 324INLINE void
267encode_indent (enc_t *enc) 325encode_indent (enc_t *enc)
268{ 326{
269 if (enc->json.flags & F_INDENT) 327 if (enc->json.flags & F_INDENT)
270 { 328 {
271 int spaces = enc->indent * INDENT_STEP; 329 int spaces = enc->indent * INDENT_STEP;
274 memset (enc->cur, ' ', spaces); 332 memset (enc->cur, ' ', spaces);
275 enc->cur += spaces; 333 enc->cur += spaces;
276 } 334 }
277} 335}
278 336
279inline void 337INLINE void
280encode_space (enc_t *enc) 338encode_space (enc_t *enc)
281{ 339{
282 need (enc, 1); 340 need (enc, 1);
283 encode_ch (enc, ' '); 341 encode_ch (enc, ' ');
284} 342}
285 343
286inline void 344INLINE void
287encode_nl (enc_t *enc) 345encode_nl (enc_t *enc)
288{ 346{
289 if (enc->json.flags & F_INDENT) 347 if (enc->json.flags & F_INDENT)
290 { 348 {
291 need (enc, 1); 349 need (enc, 1);
292 encode_ch (enc, '\n'); 350 encode_ch (enc, '\n');
293 } 351 }
294} 352}
295 353
296inline void 354INLINE void
297encode_comma (enc_t *enc) 355encode_comma (enc_t *enc)
298{ 356{
299 encode_ch (enc, ','); 357 encode_ch (enc, ',');
300 358
301 if (enc->json.flags & F_INDENT) 359 if (enc->json.flags & F_INDENT)
309static void 367static void
310encode_av (enc_t *enc, AV *av) 368encode_av (enc_t *enc, AV *av)
311{ 369{
312 int i, len = av_len (av); 370 int i, len = av_len (av);
313 371
314 if (enc->indent >= enc->maxdepth) 372 if (enc->indent >= enc->json.max_depth)
315 croak ("data structure too deep (hit recursion limit)"); 373 croak (ERR_NESTING_EXCEEDED);
316 374
317 encode_ch (enc, '['); encode_nl (enc); 375 encode_ch (enc, '[');
318 ++enc->indent; 376
377 if (len >= 0)
378 {
379 encode_nl (enc); ++enc->indent;
319 380
320 for (i = 0; i <= len; ++i) 381 for (i = 0; i <= len; ++i)
321 { 382 {
322 SV **svp = av_fetch (av, i, 0); 383 SV **svp = av_fetch (av, i, 0);
323 384
324 encode_indent (enc); 385 encode_indent (enc);
325 386
326 if (svp) 387 if (svp)
327 encode_sv (enc, *svp); 388 encode_sv (enc, *svp);
328 else 389 else
329 encode_str (enc, "null", 4, 0); 390 encode_str (enc, "null", 4, 0);
330 391
331 if (i < len) 392 if (i < len)
332 encode_comma (enc); 393 encode_comma (enc);
333 } 394 }
334 395
396 encode_nl (enc); --enc->indent; encode_indent (enc);
397 }
398
335 encode_nl (enc); 399 encode_ch (enc, ']');
336
337 --enc->indent;
338 encode_indent (enc); encode_ch (enc, ']');
339} 400}
340 401
341static void 402static void
342encode_hk (enc_t *enc, HE *he) 403encode_hk (enc_t *enc, HE *he)
343{ 404{
391 452
392static void 453static void
393encode_hv (enc_t *enc, HV *hv) 454encode_hv (enc_t *enc, HV *hv)
394{ 455{
395 HE *he; 456 HE *he;
396 int count;
397 457
398 if (enc->indent >= enc->maxdepth) 458 if (enc->indent >= enc->json.max_depth)
399 croak ("data structure too deep (hit recursion limit)"); 459 croak (ERR_NESTING_EXCEEDED);
400 460
401 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 461 encode_ch (enc, '{');
402 462
403 // for canonical output we have to sort by keys first 463 // for canonical output we have to sort by keys first
404 // actually, this is mostly due to the stupid so-called 464 // actually, this is mostly due to the stupid so-called
405 // security workaround added somewhere in 5.8.x. 465 // security workaround added somewhere in 5.8.x
406 // that randomises hash orderings 466 // that randomises hash orderings
407 if (enc->json.flags & F_CANONICAL) 467 if (enc->json.flags & F_CANONICAL)
408 { 468 {
409 int count = hv_iterinit (hv); 469 int count = hv_iterinit (hv);
410 470
459 519
460 FREETMPS; 520 FREETMPS;
461 LEAVE; 521 LEAVE;
462 } 522 }
463 523
524 encode_nl (enc); ++enc->indent;
525
464 while (count--) 526 while (count--)
465 { 527 {
466 encode_indent (enc); 528 encode_indent (enc);
467 he = hes [count]; 529 he = hes [count];
468 encode_hk (enc, he); 530 encode_hk (enc, he);
469 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he)); 531 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
470 532
471 if (count) 533 if (count)
472 encode_comma (enc); 534 encode_comma (enc);
473 } 535 }
536
537 encode_nl (enc); --enc->indent; encode_indent (enc);
474 } 538 }
475 } 539 }
476 else 540 else
477 { 541 {
478 if (hv_iterinit (hv) || SvMAGICAL (hv)) 542 if (hv_iterinit (hv) || SvMAGICAL (hv))
479 if ((he = hv_iternext (hv))) 543 if ((he = hv_iternext (hv)))
544 {
545 encode_nl (enc); ++enc->indent;
546
480 for (;;) 547 for (;;)
481 { 548 {
482 encode_indent (enc); 549 encode_indent (enc);
483 encode_hk (enc, he); 550 encode_hk (enc, he);
484 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he)); 551 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
485 552
486 if (!(he = hv_iternext (hv))) 553 if (!(he = hv_iternext (hv)))
487 break; 554 break;
488 555
489 encode_comma (enc); 556 encode_comma (enc);
490 } 557 }
491 }
492 558
559 encode_nl (enc); --enc->indent; encode_indent (enc);
560 }
561 }
562
493 encode_nl (enc); 563 encode_ch (enc, '}');
494
495 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
496} 564}
497 565
498// encode objects, arrays and special \0=false and \1=true values. 566// encode objects, arrays and special \0=false and \1=true values.
499static void 567static void
500encode_rv (enc_t *enc, SV *sv) 568encode_rv (enc_t *enc, SV *sv)
577 645
578 if (len == 1 && *pv == '1') 646 if (len == 1 && *pv == '1')
579 encode_str (enc, "true", 4, 0); 647 encode_str (enc, "true", 4, 0);
580 else if (len == 1 && *pv == '0') 648 else if (len == 1 && *pv == '0')
581 encode_str (enc, "false", 5, 0); 649 encode_str (enc, "false", 5, 0);
650 else if (enc->json.flags & F_ALLOW_UNKNOWN)
651 encode_str (enc, "null", 4, 0);
582 else 652 else
583 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",
584 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 654 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
585 } 655 }
656 else if (enc->json.flags & F_ALLOW_UNKNOWN)
657 encode_str (enc, "null", 4, 0);
586 else 658 else
587 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",
588 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 660 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
589} 661}
590 662
608 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 680 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
609 enc->cur += strlen (enc->cur); 681 enc->cur += strlen (enc->cur);
610 } 682 }
611 else if (SvIOKp (sv)) 683 else if (SvIOKp (sv))
612 { 684 {
613 // 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
614 if (SvUV (sv) & ~(UV)0x7fff) 686 // we assume two's complement
615 { 687 // we assume no aliasing issues in the union
616 // large integer, use the (rather slow) snprintf way. 688 if (SvIsUV (sv) ? SvUVX (sv) <= 59000
617 need (enc, sizeof (UV) * 3); 689 : SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000)
618 enc->cur +=
619 SvIsUV(sv)
620 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
621 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
622 }
623 else
624 { 690 {
625 // optimise the "small number case" 691 // optimise the "small number case"
626 // 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
627 I32 i = SvIV (sv); 694 I32 i = SvIVX (sv);
628 U32 u; 695 U32 u;
629 char digit, nz = 0; 696 char digit, nz = 0;
630 697
631 need (enc, 6); 698 need (enc, 6);
632 699
638 705
639 // 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
640 // 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,
641 // resulting in a net multiplication by 10. 708 // resulting in a net multiplication by 10.
642 // we always write the digit to memory but conditionally increment 709 // we always write the digit to memory but conditionally increment
643 // the pointer, to ease the usage of conditional move instructions. 710 // the pointer, to enable the use of conditional move instructions.
644 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;
645 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;
646 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;
647 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;
648 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'
649 } 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 }
650 } 726 }
651 else if (SvROK (sv)) 727 else if (SvROK (sv))
652 encode_rv (enc, SvRV (sv)); 728 encode_rv (enc, SvRV (sv));
653 else if (!SvOK (sv)) 729 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN)
654 encode_str (enc, "null", 4, 0); 730 encode_str (enc, "null", 4, 0);
655 else 731 else
656 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",
657 SvPV_nolen (sv), SvFLAGS (sv)); 733 SvPV_nolen (sv), SvFLAGS (sv));
658} 734}
668 enc.json = *json; 744 enc.json = *json;
669 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 745 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
670 enc.cur = SvPVX (enc.sv); 746 enc.cur = SvPVX (enc.sv);
671 enc.end = SvEND (enc.sv); 747 enc.end = SvEND (enc.sv);
672 enc.indent = 0; 748 enc.indent = 0;
673 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;
674 752
675 SvPOK_only (enc.sv); 753 SvPOK_only (enc.sv);
676 encode_sv (&enc, scalar); 754 encode_sv (&enc, scalar);
677 755
678 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 756 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
699 JSON json; 777 JSON json;
700 U32 depth; // recursion depth 778 U32 depth; // recursion depth
701 U32 maxdepth; // recursion depth limit 779 U32 maxdepth; // recursion depth limit
702} dec_t; 780} dec_t;
703 781
704inline void 782INLINE void
705decode_comment (dec_t *dec) 783decode_comment (dec_t *dec)
706{ 784{
707 // only '#'-style comments allowed a.t.m. 785 // only '#'-style comments allowed a.t.m.
708 786
709 while (*dec->cur && *dec->cur != 0x0a && *dec->cur != 0x0d) 787 while (*dec->cur && *dec->cur != 0x0a && *dec->cur != 0x0d)
710 ++dec->cur; 788 ++dec->cur;
711} 789}
712 790
713inline void 791INLINE void
714decode_ws (dec_t *dec) 792decode_ws (dec_t *dec)
715{ 793{
716 for (;;) 794 for (;;)
717 { 795 {
718 char ch = *dec->cur; 796 char ch = *dec->cur;
742 if (*dec->cur != ch) \ 820 if (*dec->cur != ch) \
743 ERR (# ch " expected"); \ 821 ERR (# ch " expected"); \
744 ++dec->cur; \ 822 ++dec->cur; \
745 SE 823 SE
746 824
747#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)
748#define DEC_DEC_DEPTH --dec->depth 826#define DEC_DEC_DEPTH --dec->depth
749 827
750static SV *decode_sv (dec_t *dec); 828static SV *decode_sv (dec_t *dec);
751 829
752static signed char decode_hexdigit[256]; 830static signed char decode_hexdigit[256];
844 922
845 if (hi >= 0x80) 923 if (hi >= 0x80)
846 { 924 {
847 utf8 = 1; 925 utf8 = 1;
848 926
849 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0); 927 cur = encode_utf8 (cur, hi);
850 } 928 }
851 else 929 else
852 *cur++ = hi; 930 *cur++ = hi;
853 } 931 }
854 break; 932 break;
856 default: 934 default:
857 --dec_cur; 935 --dec_cur;
858 ERR ("illegal backslash escape sequence in string"); 936 ERR ("illegal backslash escape sequence in string");
859 } 937 }
860 } 938 }
861 else if (expect_true (ch >= 0x20 && ch <= 0x7f)) 939 else if (expect_true (ch >= 0x20 && ch < 0x80))
862 *cur++ = ch; 940 *cur++ = ch;
863 else if (ch >= 0x80) 941 else if (ch >= 0x80)
864 { 942 {
865 STRLEN clen; 943 STRLEN clen;
866 UV uch; 944 UV uch;
989 1067
990 if (!is_nv) 1068 if (!is_nv)
991 { 1069 {
992 int len = dec->cur - start; 1070 int len = dec->cur - start;
993 1071
994 // 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
995 if (*start == '-') 1073 if (*start == '-')
996 switch (len) 1074 switch (len)
997 { 1075 {
998 case 2: return newSViv (-( start [1] - '0' * 1)); 1076 case 2: return newSViv (-( start [1] - '0' * 1));
999 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); 1077 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
1000 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));
1001 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));
1002 } 1081 }
1003 else 1082 else
1004 switch (len) 1083 switch (len)
1005 { 1084 {
1006 case 1: return newSViv ( start [0] - '0' * 1); 1085 case 1: return newSViv ( start [0] - '0' * 1);
1007 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); 1086 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
1008 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);
1009 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);
1010 } 1090 }
1011 1091
1012 { 1092 {
1013 UV uv; 1093 UV uv;
1014 int numtype = grok_number (start, len, &uv); 1094 int numtype = grok_number (start, len, &uv);
1122 char *p = dec->cur; 1202 char *p = dec->cur;
1123 char *e = p + 24; // only try up to 24 bytes 1203 char *e = p + 24; // only try up to 24 bytes
1124 1204
1125 for (;;) 1205 for (;;)
1126 { 1206 {
1127 // the >= 0x80 is true on most architectures 1207 // the >= 0x80 is false on most architectures
1128 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\') 1208 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1129 { 1209 {
1130 // slow path, back up and use decode_str 1210 // slow path, back up and use decode_str
1131 SV *key = decode_str (dec); 1211 SV *key = decode_str (dec);
1132 if (!key) 1212 if (!key)
1263 1343
1264static SV * 1344static SV *
1265decode_sv (dec_t *dec) 1345decode_sv (dec_t *dec)
1266{ 1346{
1267 // the beauty of JSON: you need exactly one character lookahead 1347 // the beauty of JSON: you need exactly one character lookahead
1268 // to parse anything. 1348 // to parse everything.
1269 switch (*dec->cur) 1349 switch (*dec->cur)
1270 { 1350 {
1271 case '"': ++dec->cur; return decode_str (dec); 1351 case '"': ++dec->cur; return decode_str (dec);
1272 case '[': ++dec->cur; return decode_av (dec); 1352 case '[': ++dec->cur; return decode_av (dec);
1273 case '{': ++dec->cur; return decode_hv (dec); 1353 case '{': ++dec->cur; return decode_hv (dec);
1274 1354
1275 case '-': 1355 case '-':
1276 case '0': case '1': case '2': case '3': case '4': 1356 case '0': case '1': case '2': case '3': case '4':
1277 case '5': case '6': case '7': case '8': case '9': 1357 case '5': case '6': case '7': case '8': case '9':
1278 return decode_num (dec); 1358 return decode_num (dec);
1280 case 't': 1360 case 't':
1281 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1361 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1282 { 1362 {
1283 dec->cur += 4; 1363 dec->cur += 4;
1284#if JSON_SLOW 1364#if JSON_SLOW
1285 json_true = get_sv ("JSON::XS::true", 1); SvREADONLY_on (json_true); 1365 json_true = get_bool ("JSON::XS::true");
1286#endif 1366#endif
1287 return SvREFCNT_inc (json_true); 1367 return newSVsv (json_true);
1288 } 1368 }
1289 else 1369 else
1290 ERR ("'true' expected"); 1370 ERR ("'true' expected");
1291 1371
1292 break; 1372 break;
1294 case 'f': 1374 case 'f':
1295 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1375 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1296 { 1376 {
1297 dec->cur += 5; 1377 dec->cur += 5;
1298#if JSON_SLOW 1378#if JSON_SLOW
1299 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); 1379 json_false = get_bool ("JSON::XS::false");
1300#endif 1380#endif
1301 return SvREFCNT_inc (json_false); 1381 return newSVsv (json_false);
1302 } 1382 }
1303 else 1383 else
1304 ERR ("'false' expected"); 1384 ERR ("'false' expected");
1305 1385
1306 break; 1386 break;
1324fail: 1404fail:
1325 return 0; 1405 return 0;
1326} 1406}
1327 1407
1328static SV * 1408static SV *
1329decode_json (SV *string, JSON *json, UV *offset_return) 1409decode_json (SV *string, JSON *json, STRLEN *offset_return)
1330{ 1410{
1331 dec_t dec; 1411 dec_t dec;
1332 UV offset; 1412 STRLEN offset;
1333 SV *sv; 1413 SV *sv;
1334 1414
1415 /* work around bugs in 5.10 where manipulating magic values
1416 * will perl ignore the magic in subsequent accesses
1417 */
1335 SvGETMAGIC (string); 1418 /*SvGETMAGIC (string);*/
1419 if (SvMAGICAL (string))
1420 string = sv_2mortal (newSVsv (string));
1421
1336 SvUPGRADE (string, SVt_PV); 1422 SvUPGRADE (string, SVt_PV);
1337 1423
1338 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)
1339 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",
1340 (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags)); 1442 (unsigned long)SvCUR (string), (unsigned long)json->max_size);
1341 1443
1342 if (json->flags & F_UTF8) 1444 if (json->flags & F_UTF8)
1343 sv_utf8_downgrade (string, 0); 1445 sv_utf8_downgrade (string, 0);
1344 else 1446 else
1345 sv_utf8_upgrade (string); 1447 sv_utf8_upgrade (string);
1346 1448
1347 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1449 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
1348 1450
1349 dec.json = *json; 1451 dec.json = *json;
1350 dec.cur = SvPVX (string); 1452 dec.cur = SvPVX (string);
1351 dec.end = SvEND (string); 1453 dec.end = SvEND (string);
1352 dec.err = 0; 1454 dec.err = 0;
1353 dec.depth = 0; 1455 dec.depth = 0;
1354 dec.maxdepth = DEC_DEPTH (dec.json.flags);
1355 1456
1356 if (dec.json.cb_object || dec.json.cb_sk_object) 1457 if (dec.json.cb_object || dec.json.cb_sk_object)
1357 dec.json.flags |= F_HOOK; 1458 dec.json.flags |= F_HOOK;
1358 1459
1359 *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
1410 1511
1411 return sv; 1512 return sv;
1412} 1513}
1413 1514
1414///////////////////////////////////////////////////////////////////////////// 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/////////////////////////////////////////////////////////////////////////////
1415// XS interface functions 1633// XS interface functions
1416 1634
1417MODULE = JSON::XS PACKAGE = JSON::XS 1635MODULE = JSON::XS PACKAGE = JSON::XS
1418 1636
1419BOOT: 1637BOOT:
1428 : -1; 1646 : -1;
1429 1647
1430 json_stash = gv_stashpv ("JSON::XS" , 1); 1648 json_stash = gv_stashpv ("JSON::XS" , 1);
1431 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); 1649 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1432 1650
1433 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true ); 1651 json_true = get_bool ("JSON::XS::true");
1434 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); 1652 json_false = get_bool ("JSON::XS::false");
1435} 1653}
1436 1654
1437PROTOTYPES: DISABLE 1655PROTOTYPES: DISABLE
1438 1656
1439void CLONE (...) 1657void CLONE (...)
1444void new (char *klass) 1662void new (char *klass)
1445 PPCODE: 1663 PPCODE:
1446{ 1664{
1447 SV *pv = NEWSV (0, sizeof (JSON)); 1665 SV *pv = NEWSV (0, sizeof (JSON));
1448 SvPOK_only (pv); 1666 SvPOK_only (pv);
1449 Zero (SvPVX (pv), 1, JSON); 1667 json_init ((JSON *)SvPVX (pv));
1450 ((JSON *)SvPVX (pv))->flags = F_DEFAULT; 1668 XPUSHs (sv_2mortal (sv_bless (
1451 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 )));
1452} 1672}
1453 1673
1454void ascii (JSON *self, int enable = 1) 1674void ascii (JSON *self, int enable = 1)
1455 ALIAS: 1675 ALIAS:
1456 ascii = F_ASCII 1676 ascii = F_ASCII
1464 allow_nonref = F_ALLOW_NONREF 1684 allow_nonref = F_ALLOW_NONREF
1465 shrink = F_SHRINK 1685 shrink = F_SHRINK
1466 allow_blessed = F_ALLOW_BLESSED 1686 allow_blessed = F_ALLOW_BLESSED
1467 convert_blessed = F_CONV_BLESSED 1687 convert_blessed = F_CONV_BLESSED
1468 relaxed = F_RELAXED 1688 relaxed = F_RELAXED
1689 allow_unknown = F_ALLOW_UNKNOWN
1469 PPCODE: 1690 PPCODE:
1470{ 1691{
1471 if (enable) 1692 if (enable)
1472 self->flags |= ix; 1693 self->flags |= ix;
1473 else 1694 else
1474 self->flags &= ~ix; 1695 self->flags &= ~ix;
1475 1696
1476 XPUSHs (ST (0)); 1697 XPUSHs (ST (0));
1477} 1698}
1478 1699
1479void 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
1480 PPCODE: 1715 PPCODE:
1481{ 1716 XPUSHs (boolSV (self->flags & ix));
1482 UV log2 = 0;
1483 1717
1484 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL; 1718void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
1485 1719 PPCODE:
1486 while ((1UL << log2) < max_depth) 1720 self->max_depth = max_depth;
1487 ++log2;
1488
1489 self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1490
1491 XPUSHs (ST (0)); 1721 XPUSHs (ST (0));
1492}
1493 1722
1723U32 get_max_depth (JSON *self)
1724 CODE:
1725 RETVAL = self->max_depth;
1726 OUTPUT:
1727 RETVAL
1728
1494void max_size (JSON *self, UV max_size = 0) 1729void max_size (JSON *self, U32 max_size = 0)
1495 PPCODE: 1730 PPCODE:
1496{ 1731 self->max_size = max_size;
1497 UV log2 = 0;
1498
1499 if (max_size > 0x80000000UL) max_size = 0x80000000UL;
1500 if (max_size == 1) max_size = 2;
1501
1502 while ((1UL << log2) < max_size)
1503 ++log2;
1504
1505 self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1506
1507 XPUSHs (ST (0)); 1732 XPUSHs (ST (0));
1508} 1733
1734int get_max_size (JSON *self)
1735 CODE:
1736 RETVAL = self->max_size;
1737 OUTPUT:
1738 RETVAL
1509 1739
1510void filter_json_object (JSON *self, SV *cb = &PL_sv_undef) 1740void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1511 PPCODE: 1741 PPCODE:
1512{ 1742{
1513 SvREFCNT_dec (self->cb_object); 1743 SvREFCNT_dec (self->cb_object);
1547 XPUSHs (decode_json (jsonstr, self, 0)); 1777 XPUSHs (decode_json (jsonstr, self, 0));
1548 1778
1549void decode_prefix (JSON *self, SV *jsonstr) 1779void decode_prefix (JSON *self, SV *jsonstr)
1550 PPCODE: 1780 PPCODE:
1551{ 1781{
1552 UV offset; 1782 STRLEN offset;
1553 EXTEND (SP, 2); 1783 EXTEND (SP, 2);
1554 PUSHs (decode_json (jsonstr, self, &offset)); 1784 PUSHs (decode_json (jsonstr, self, &offset));
1555 PUSHs (sv_2mortal (newSVuv (offset))); 1785 PUSHs (sv_2mortal (newSVuv (offset)));
1786}
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;
1556} 1876}
1557 1877
1558void DESTROY (JSON *self) 1878void DESTROY (JSON *self)
1559 CODE: 1879 CODE:
1560 SvREFCNT_dec (self->cb_sk_object); 1880 SvREFCNT_dec (self->cb_sk_object);
1561 SvREFCNT_dec (self->cb_object); 1881 SvREFCNT_dec (self->cb_object);
1882 SvREFCNT_dec (self->incr_text);
1562 1883
1563PROTOTYPES: ENABLE 1884PROTOTYPES: ENABLE
1564 1885
1565void to_json (SV *scalar) 1886void encode_json (SV *scalar)
1887 ALIAS:
1888 to_json_ = 0
1889 encode_json = F_UTF8
1566 PPCODE: 1890 PPCODE:
1567{ 1891{
1568 JSON json = { F_DEFAULT | F_UTF8 }; 1892 JSON json;
1893 json_init (&json);
1894 json.flags |= ix;
1569 XPUSHs (encode_json (scalar, &json)); 1895 XPUSHs (encode_json (scalar, &json));
1570} 1896}
1571 1897
1572void from_json (SV *jsonstr) 1898void decode_json (SV *jsonstr)
1899 ALIAS:
1900 from_json_ = 0
1901 decode_json = F_UTF8
1573 PPCODE: 1902 PPCODE:
1574{ 1903{
1575 JSON json = { F_DEFAULT | F_UTF8 }; 1904 JSON json;
1905 json_init (&json);
1906 json.flags |= ix;
1576 XPUSHs (decode_json (jsonstr, &json, 0)); 1907 XPUSHs (decode_json (jsonstr, &json, 0));
1577} 1908}
1578 1909
1910

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines