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 1.63 by root, Mon Aug 27 01:49:01 2007 UTC vs.
Revision 1.85 by root, Sat Apr 5 18:15:46 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 && (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 unsigned char 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);
89 if (SvLEN (sv) > SvCUR (sv) + 1) 126 if (SvLEN (sv) > SvCUR (sv) + 1)
90 { 127 {
99// decode an utf-8 character and return it, or (UV)-1 in 136// decode an utf-8 character and return it, or (UV)-1 in
100// case of an error. 137// case of an error.
101// we special-case "safe" characters from U+80 .. U+7FF, 138// we special-case "safe" characters from U+80 .. U+7FF,
102// but use the very good perl function to parse anything else. 139// but use the very good perl function to parse anything else.
103// note that we never call this function for a ascii codepoints 140// note that we never call this function for a ascii codepoints
104inline UV 141INLINE UV
105decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) 142decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
106{ 143{
107 if (expect_false (s[0] > 0xdf || s[0] < 0xc2)) 144 if (expect_true (len >= 2
108 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); 145 && IN_RANGE_INC (char, s[0], 0xc2, 0xdf)
109 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) 146 && IN_RANGE_INC (char, s[1], 0x80, 0xbf)))
110 { 147 {
111 *clen = 2; 148 *clen = 2;
112 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 149 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
113 } 150 }
114 else 151 else
115 { 152 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
116 *clen = (STRLEN)-1; 153}
117 return (UV)-1; 154
118 } 155// likewise for encoding, also never called for ascii codepoints
156// this function takes advantage of this fact, although current gccs
157// seem to optimise the check for >= 0x80 away anyways
158INLINE unsigned char *
159encode_utf8 (unsigned char *s, UV ch)
160{
161 if (expect_false (ch < 0x000080))
162 *s++ = ch;
163 else if (expect_true (ch < 0x000800))
164 *s++ = 0xc0 | ( ch >> 6),
165 *s++ = 0x80 | ( ch & 0x3f);
166 else if ( ch < 0x010000)
167 *s++ = 0xe0 | ( ch >> 12),
168 *s++ = 0x80 | ((ch >> 6) & 0x3f),
169 *s++ = 0x80 | ( ch & 0x3f);
170 else if ( ch < 0x110000)
171 *s++ = 0xf0 | ( ch >> 18),
172 *s++ = 0x80 | ((ch >> 12) & 0x3f),
173 *s++ = 0x80 | ((ch >> 6) & 0x3f),
174 *s++ = 0x80 | ( ch & 0x3f);
175
176 return s;
119} 177}
120 178
121///////////////////////////////////////////////////////////////////////////// 179/////////////////////////////////////////////////////////////////////////////
122// encoder 180// encoder
123 181
127 char *cur; // SvPVX (sv) + current output position 185 char *cur; // SvPVX (sv) + current output position
128 char *end; // SvEND (sv) 186 char *end; // SvEND (sv)
129 SV *sv; // result scalar 187 SV *sv; // result scalar
130 JSON json; 188 JSON json;
131 U32 indent; // indentation level 189 U32 indent; // indentation level
132 U32 maxdepth; // max. indentation/recursion level 190 UV limit; // escape character values >= this value when encoding
133} enc_t; 191} enc_t;
134 192
135inline void 193INLINE void
136need (enc_t *enc, STRLEN len) 194need (enc_t *enc, STRLEN len)
137{ 195{
138 if (expect_false (enc->cur + len >= enc->end)) 196 if (expect_false (enc->cur + len >= enc->end))
139 { 197 {
140 STRLEN cur = enc->cur - SvPVX (enc->sv); 198 STRLEN cur = enc->cur - SvPVX (enc->sv);
142 enc->cur = SvPVX (enc->sv) + cur; 200 enc->cur = SvPVX (enc->sv) + cur;
143 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 201 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
144 } 202 }
145} 203}
146 204
147inline void 205INLINE void
148encode_ch (enc_t *enc, char ch) 206encode_ch (enc_t *enc, char ch)
149{ 207{
150 need (enc, 1); 208 need (enc, 1);
151 *enc->cur++ = ch; 209 *enc->cur++ = ch;
152} 210}
206 { 264 {
207 uch = ch; 265 uch = ch;
208 clen = 1; 266 clen = 1;
209 } 267 }
210 268
211 if (uch > 0x10FFFFUL) 269 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 { 270 {
216 if (uch > 0xFFFFUL) 271 if (uch >= 0x10000UL)
217 { 272 {
273 if (uch >= 0x110000UL)
274 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
275
218 need (enc, len += 11); 276 need (enc, len += 11);
219 sprintf (enc->cur, "\\u%04x\\u%04x", 277 sprintf (enc->cur, "\\u%04x\\u%04x",
220 (int)((uch - 0x10000) / 0x400 + 0xD800), 278 (int)((uch - 0x10000) / 0x400 + 0xD800),
221 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 279 (int)((uch - 0x10000) % 0x400 + 0xDC00));
222 enc->cur += 12; 280 enc->cur += 12;
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
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
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
705decode_ws (dec_t *dec) 792decode_ws (dec_t *dec)
706{ 793{
707 for (;;) 794 for (;;)
708 { 795 {
709 char ch = *dec->cur; 796 char ch = *dec->cur;
710 797
711 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 }
712 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) 810 else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)
713 break; 811 break; // parse error, but let higher level handle it, gives better error messages
714
715 if (ch == '#' && dec->json.flags & F_RELAXED)
716 ++dec->cur;
717 812
718 ++dec->cur; 813 ++dec->cur;
719 } 814 }
720} 815}
721 816
725 if (*dec->cur != ch) \ 820 if (*dec->cur != ch) \
726 ERR (# ch " expected"); \ 821 ERR (# ch " expected"); \
727 ++dec->cur; \ 822 ++dec->cur; \
728 SE 823 SE
729 824
730#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)
731#define DEC_DEC_DEPTH --dec->depth 826#define DEC_DEC_DEPTH --dec->depth
732 827
733static SV *decode_sv (dec_t *dec); 828static SV *decode_sv (dec_t *dec);
734 829
735static signed char decode_hexdigit[256]; 830static signed char decode_hexdigit[256];
827 922
828 if (hi >= 0x80) 923 if (hi >= 0x80)
829 { 924 {
830 utf8 = 1; 925 utf8 = 1;
831 926
832 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0); 927 cur = encode_utf8 (cur, hi);
833 } 928 }
834 else 929 else
835 *cur++ = hi; 930 *cur++ = hi;
836 } 931 }
837 break; 932 break;
839 default: 934 default:
840 --dec_cur; 935 --dec_cur;
841 ERR ("illegal backslash escape sequence in string"); 936 ERR ("illegal backslash escape sequence in string");
842 } 937 }
843 } 938 }
844 else if (expect_true (ch >= 0x20 && ch <= 0x7f)) 939 else if (expect_true (ch >= 0x20 && ch < 0x80))
845 *cur++ = ch; 940 *cur++ = ch;
846 else if (ch >= 0x80) 941 else if (ch >= 0x80)
847 { 942 {
848 STRLEN clen; 943 STRLEN clen;
849 UV uch; 944 UV uch;
972 1067
973 if (!is_nv) 1068 if (!is_nv)
974 { 1069 {
975 int len = dec->cur - start; 1070 int len = dec->cur - start;
976 1071
977 // 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
978 if (*start == '-') 1073 if (*start == '-')
979 switch (len) 1074 switch (len)
980 { 1075 {
981 case 2: return newSViv (-( start [1] - '0' * 1)); 1076 case 2: return newSViv (-( start [1] - '0' * 1));
982 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); 1077 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
983 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));
984 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));
985 } 1081 }
986 else 1082 else
987 switch (len) 1083 switch (len)
988 { 1084 {
989 case 1: return newSViv ( start [0] - '0' * 1); 1085 case 1: return newSViv ( start [0] - '0' * 1);
990 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); 1086 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
991 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);
992 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);
993 } 1090 }
994 1091
995 { 1092 {
996 UV uv; 1093 UV uv;
997 int numtype = grok_number (start, len, &uv); 1094 int numtype = grok_number (start, len, &uv);
1105 char *p = dec->cur; 1202 char *p = dec->cur;
1106 char *e = p + 24; // only try up to 24 bytes 1203 char *e = p + 24; // only try up to 24 bytes
1107 1204
1108 for (;;) 1205 for (;;)
1109 { 1206 {
1110 // the >= 0x80 is true on most architectures 1207 // the >= 0x80 is false on most architectures
1111 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\') 1208 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1112 { 1209 {
1113 // slow path, back up and use decode_str 1210 // slow path, back up and use decode_str
1114 SV *key = decode_str (dec); 1211 SV *key = decode_str (dec);
1115 if (!key) 1212 if (!key)
1246 1343
1247static SV * 1344static SV *
1248decode_sv (dec_t *dec) 1345decode_sv (dec_t *dec)
1249{ 1346{
1250 // the beauty of JSON: you need exactly one character lookahead 1347 // the beauty of JSON: you need exactly one character lookahead
1251 // to parse anything. 1348 // to parse everything.
1252 switch (*dec->cur) 1349 switch (*dec->cur)
1253 { 1350 {
1254 case '"': ++dec->cur; return decode_str (dec); 1351 case '"': ++dec->cur; return decode_str (dec);
1255 case '[': ++dec->cur; return decode_av (dec); 1352 case '[': ++dec->cur; return decode_av (dec);
1256 case '{': ++dec->cur; return decode_hv (dec); 1353 case '{': ++dec->cur; return decode_hv (dec);
1257 1354
1258 case '-': 1355 case '-':
1259 case '0': case '1': case '2': case '3': case '4': 1356 case '0': case '1': case '2': case '3': case '4':
1260 case '5': case '6': case '7': case '8': case '9': 1357 case '5': case '6': case '7': case '8': case '9':
1261 return decode_num (dec); 1358 return decode_num (dec);
1263 case 't': 1360 case 't':
1264 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1361 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1265 { 1362 {
1266 dec->cur += 4; 1363 dec->cur += 4;
1267#if JSON_SLOW 1364#if JSON_SLOW
1268 json_true = get_sv ("JSON::XS::true", 1); SvREADONLY_on (json_true); 1365 json_true = get_bool ("JSON::XS::true");
1269#endif 1366#endif
1270 return SvREFCNT_inc (json_true); 1367 return newSVsv (json_true);
1271 } 1368 }
1272 else 1369 else
1273 ERR ("'true' expected"); 1370 ERR ("'true' expected");
1274 1371
1275 break; 1372 break;
1277 case 'f': 1374 case 'f':
1278 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1375 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1279 { 1376 {
1280 dec->cur += 5; 1377 dec->cur += 5;
1281#if JSON_SLOW 1378#if JSON_SLOW
1282 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); 1379 json_false = get_bool ("JSON::XS::false");
1283#endif 1380#endif
1284 return SvREFCNT_inc (json_false); 1381 return newSVsv (json_false);
1285 } 1382 }
1286 else 1383 else
1287 ERR ("'false' expected"); 1384 ERR ("'false' expected");
1288 1385
1289 break; 1386 break;
1307fail: 1404fail:
1308 return 0; 1405 return 0;
1309} 1406}
1310 1407
1311static SV * 1408static SV *
1312decode_json (SV *string, JSON *json, UV *offset_return) 1409decode_json (SV *string, JSON *json, STRLEN *offset_return)
1313{ 1410{
1314 dec_t dec; 1411 dec_t dec;
1315 UV offset; 1412 STRLEN offset;
1316 SV *sv; 1413 SV *sv;
1317 1414
1318 SvGETMAGIC (string); 1415 SvGETMAGIC (string);
1319 SvUPGRADE (string, SVt_PV); 1416 SvUPGRADE (string, SVt_PV);
1320 1417
1321 if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags)) 1418 if (SvCUR (string) > json->max_size && json->max_size)
1322 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", 1419 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1323 (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags)); 1420 (unsigned long)SvCUR (string), (unsigned long)json->max_size);
1324 1421
1325 if (json->flags & F_UTF8) 1422 if (json->flags & F_UTF8)
1326 sv_utf8_downgrade (string, 0); 1423 sv_utf8_downgrade (string, 0);
1327 else 1424 else
1328 sv_utf8_upgrade (string); 1425 sv_utf8_upgrade (string);
1329 1426
1330 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1427 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
1331 1428
1332 dec.json = *json; 1429 dec.json = *json;
1333 dec.cur = SvPVX (string); 1430 dec.cur = SvPVX (string);
1334 dec.end = SvEND (string); 1431 dec.end = SvEND (string);
1335 dec.err = 0; 1432 dec.err = 0;
1336 dec.depth = 0; 1433 dec.depth = 0;
1337 dec.maxdepth = DEC_DEPTH (dec.json.flags);
1338 1434
1339 if (dec.json.cb_object || dec.json.cb_sk_object) 1435 if (dec.json.cb_object || dec.json.cb_sk_object)
1340 dec.json.flags |= F_HOOK; 1436 dec.json.flags |= F_HOOK;
1341 1437
1342 *dec.end = 0; // this should basically be a nop, too, but make sure it's there 1438 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1393 1489
1394 return sv; 1490 return sv;
1395} 1491}
1396 1492
1397///////////////////////////////////////////////////////////////////////////// 1493/////////////////////////////////////////////////////////////////////////////
1494// incremental parser
1495
1496static void
1497incr_parse (JSON *self)
1498{
1499 const char *p = SvPVX (self->incr_text) + self->incr_pos;
1500
1501 for (;;)
1502 {
1503 //printf ("loop pod %d *p<%c><%s>, mode %d nest %d\n", p - SvPVX (self->incr_text), *p, p, self->incr_mode, self->incr_nest);//D
1504 switch (self->incr_mode)
1505 {
1506 // only used for intiial whitespace skipping
1507 case INCR_M_WS:
1508 for (;;)
1509 {
1510 if (*p > 0x20)
1511 {
1512 self->incr_mode = INCR_M_JSON;
1513 goto incr_m_json;
1514 }
1515 else if (!*p)
1516 goto interrupt;
1517
1518 ++p;
1519 }
1520
1521 // skip a single char inside a string (for \\-processing)
1522 case INCR_M_BS:
1523 if (!*p)
1524 goto interrupt;
1525
1526 ++p;
1527 self->incr_mode = INCR_M_STR;
1528 goto incr_m_str;
1529
1530 // inside a string
1531 case INCR_M_STR:
1532 incr_m_str:
1533 for (;;)
1534 {
1535 if (*p == '"')
1536 {
1537 ++p;
1538 self->incr_mode = INCR_M_JSON;
1539
1540 if (!self->incr_nest)
1541 goto interrupt;
1542
1543 goto incr_m_json;
1544 }
1545 else if (*p == '\\')
1546 {
1547 ++p; // "virtually" consumes character after \
1548
1549 if (!*p) // if at end of string we have to switch modes
1550 {
1551 self->incr_mode = INCR_M_BS;
1552 goto interrupt;
1553 }
1554 }
1555 else if (!*p)
1556 goto interrupt;
1557
1558 ++p;
1559 }
1560
1561 // after initial ws, outside string
1562 case INCR_M_JSON:
1563 incr_m_json:
1564 for (;;)
1565 {
1566 switch (*p++)
1567 {
1568 case 0:
1569 --p;
1570 goto interrupt;
1571
1572 case 0x09:
1573 case 0x0a:
1574 case 0x0d:
1575 case 0x20:
1576 if (!self->incr_nest)
1577 {
1578 --p; // do not eat the whitespace, let the next round do it
1579 goto interrupt;
1580 }
1581 break;
1582
1583 case '"':
1584 self->incr_mode = INCR_M_STR;
1585 goto incr_m_str;
1586
1587 case '[':
1588 case '{':
1589 if (++self->incr_nest > self->max_depth)
1590 croak (ERR_NESTING_EXCEEDED);
1591 break;
1592
1593 case ']':
1594 case '}':
1595 if (!--self->incr_nest)
1596 goto interrupt;
1597 }
1598 }
1599 }
1600
1601 modechange:
1602 ;
1603 }
1604
1605interrupt:
1606 self->incr_pos = p - SvPVX (self->incr_text);
1607 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D
1608}
1609
1610/////////////////////////////////////////////////////////////////////////////
1398// XS interface functions 1611// XS interface functions
1399 1612
1400MODULE = JSON::XS PACKAGE = JSON::XS 1613MODULE = JSON::XS PACKAGE = JSON::XS
1401 1614
1402BOOT: 1615BOOT:
1411 : -1; 1624 : -1;
1412 1625
1413 json_stash = gv_stashpv ("JSON::XS" , 1); 1626 json_stash = gv_stashpv ("JSON::XS" , 1);
1414 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); 1627 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1415 1628
1416 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true ); 1629 json_true = get_bool ("JSON::XS::true");
1417 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); 1630 json_false = get_bool ("JSON::XS::false");
1418} 1631}
1419 1632
1420PROTOTYPES: DISABLE 1633PROTOTYPES: DISABLE
1421 1634
1422void CLONE (...) 1635void CLONE (...)
1427void new (char *klass) 1640void new (char *klass)
1428 PPCODE: 1641 PPCODE:
1429{ 1642{
1430 SV *pv = NEWSV (0, sizeof (JSON)); 1643 SV *pv = NEWSV (0, sizeof (JSON));
1431 SvPOK_only (pv); 1644 SvPOK_only (pv);
1432 Zero (SvPVX (pv), 1, JSON); 1645 json_init ((JSON *)SvPVX (pv));
1433 ((JSON *)SvPVX (pv))->flags = F_DEFAULT; 1646 XPUSHs (sv_2mortal (sv_bless (
1434 XPUSHs (sv_2mortal (sv_bless (newRV_noinc (pv), JSON_STASH))); 1647 newRV_noinc (pv),
1648 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1649 )));
1435} 1650}
1436 1651
1437void ascii (JSON *self, int enable = 1) 1652void ascii (JSON *self, int enable = 1)
1438 ALIAS: 1653 ALIAS:
1439 ascii = F_ASCII 1654 ascii = F_ASCII
1447 allow_nonref = F_ALLOW_NONREF 1662 allow_nonref = F_ALLOW_NONREF
1448 shrink = F_SHRINK 1663 shrink = F_SHRINK
1449 allow_blessed = F_ALLOW_BLESSED 1664 allow_blessed = F_ALLOW_BLESSED
1450 convert_blessed = F_CONV_BLESSED 1665 convert_blessed = F_CONV_BLESSED
1451 relaxed = F_RELAXED 1666 relaxed = F_RELAXED
1667 allow_unknown = F_ALLOW_UNKNOWN
1452 PPCODE: 1668 PPCODE:
1453{ 1669{
1454 if (enable) 1670 if (enable)
1455 self->flags |= ix; 1671 self->flags |= ix;
1456 else 1672 else
1457 self->flags &= ~ix; 1673 self->flags &= ~ix;
1458 1674
1459 XPUSHs (ST (0)); 1675 XPUSHs (ST (0));
1460} 1676}
1461 1677
1462void max_depth (JSON *self, UV max_depth = 0x80000000UL) 1678void get_ascii (JSON *self)
1679 ALIAS:
1680 get_ascii = F_ASCII
1681 get_latin1 = F_LATIN1
1682 get_utf8 = F_UTF8
1683 get_indent = F_INDENT
1684 get_canonical = F_CANONICAL
1685 get_space_before = F_SPACE_BEFORE
1686 get_space_after = F_SPACE_AFTER
1687 get_allow_nonref = F_ALLOW_NONREF
1688 get_shrink = F_SHRINK
1689 get_allow_blessed = F_ALLOW_BLESSED
1690 get_convert_blessed = F_CONV_BLESSED
1691 get_relaxed = F_RELAXED
1692 get_allow_unknown = F_ALLOW_UNKNOWN
1463 PPCODE: 1693 PPCODE:
1464{ 1694 XPUSHs (boolSV (self->flags & ix));
1465 UV log2 = 0;
1466 1695
1467 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL; 1696void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
1468 1697 PPCODE:
1469 while ((1UL << log2) < max_depth) 1698 self->max_depth = max_depth;
1470 ++log2;
1471
1472 self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1473
1474 XPUSHs (ST (0)); 1699 XPUSHs (ST (0));
1475}
1476 1700
1701U32 get_max_depth (JSON *self)
1702 CODE:
1703 RETVAL = self->max_depth;
1704 OUTPUT:
1705 RETVAL
1706
1477void max_size (JSON *self, UV max_size = 0) 1707void max_size (JSON *self, U32 max_size = 0)
1478 PPCODE: 1708 PPCODE:
1479{ 1709 self->max_size = max_size;
1480 UV log2 = 0;
1481
1482 if (max_size > 0x80000000UL) max_size = 0x80000000UL;
1483 if (max_size == 1) max_size = 2;
1484
1485 while ((1UL << log2) < max_size)
1486 ++log2;
1487
1488 self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1489
1490 XPUSHs (ST (0)); 1710 XPUSHs (ST (0));
1491} 1711
1712int get_max_size (JSON *self)
1713 CODE:
1714 RETVAL = self->max_size;
1715 OUTPUT:
1716 RETVAL
1492 1717
1493void filter_json_object (JSON *self, SV *cb = &PL_sv_undef) 1718void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1494 PPCODE: 1719 PPCODE:
1495{ 1720{
1496 SvREFCNT_dec (self->cb_object); 1721 SvREFCNT_dec (self->cb_object);
1530 XPUSHs (decode_json (jsonstr, self, 0)); 1755 XPUSHs (decode_json (jsonstr, self, 0));
1531 1756
1532void decode_prefix (JSON *self, SV *jsonstr) 1757void decode_prefix (JSON *self, SV *jsonstr)
1533 PPCODE: 1758 PPCODE:
1534{ 1759{
1535 UV offset; 1760 STRLEN offset;
1536 EXTEND (SP, 2); 1761 EXTEND (SP, 2);
1537 PUSHs (decode_json (jsonstr, self, &offset)); 1762 PUSHs (decode_json (jsonstr, self, &offset));
1538 PUSHs (sv_2mortal (newSVuv (offset))); 1763 PUSHs (sv_2mortal (newSVuv (offset)));
1764}
1765
1766void incr_parse (JSON *self, SV *jsonstr = 0)
1767 PPCODE:
1768{
1769 if (!self->incr_text)
1770 self->incr_text = newSVpvn ("", 0);
1771
1772 // append data, if any
1773 if (jsonstr)
1774 {
1775 if (SvUTF8 (jsonstr) && !SvUTF8 (self->incr_text))
1776 {
1777 /* utf-8-ness differs, need to upgrade */
1778 sv_utf8_upgrade (self->incr_text);
1779
1780 if (self->incr_pos)
1781 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
1782 - (U8 *)SvPVX (self->incr_text);
1783 }
1784
1785 {
1786 STRLEN len;
1787 const char *str = SvPV (jsonstr, len);
1788 SvGROW (self->incr_text, SvCUR (self->incr_text) + len + 1);
1789 Move (str, SvEND (self->incr_text), len, char);
1790 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
1791 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
1792 }
1793 }
1794
1795 if (GIMME_V != G_VOID)
1796 do
1797 {
1798 STRLEN offset;
1799
1800 if (!INCR_DONE (self))
1801 {
1802 incr_parse (self);
1803
1804 if (self->incr_pos > self->max_size && self->max_size)
1805 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1806 (unsigned long)self->incr_pos, (unsigned long)self->max_size);
1807
1808 if (!INCR_DONE (self))
1809 break;
1810 }
1811
1812 XPUSHs (decode_json (self->incr_text, self, &offset));
1813
1814 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + offset);
1815 self->incr_pos -= offset;
1816 self->incr_nest = 0;
1817 self->incr_mode = 0;
1818 }
1819 while (GIMME_V == G_ARRAY);
1820}
1821
1822SV *incr_text (JSON *self)
1823 ATTRS: lvalue
1824 CODE:
1825{
1826 if (self->incr_pos)
1827 croak ("incr_text can not be called when the incremental parser already started parsing");
1828
1829 RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef;
1830}
1831 OUTPUT:
1832 RETVAL
1833
1834void incr_skip (JSON *self)
1835 CODE:
1836{
1837 if (self->incr_pos)
1838 {
1839 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + self->incr_pos);
1840 self->incr_pos = 0;
1841 self->incr_nest = 0;
1842 self->incr_mode = 0;
1843 }
1539} 1844}
1540 1845
1541void DESTROY (JSON *self) 1846void DESTROY (JSON *self)
1542 CODE: 1847 CODE:
1543 SvREFCNT_dec (self->cb_sk_object); 1848 SvREFCNT_dec (self->cb_sk_object);
1544 SvREFCNT_dec (self->cb_object); 1849 SvREFCNT_dec (self->cb_object);
1850 SvREFCNT_dec (self->incr_text);
1545 1851
1546PROTOTYPES: ENABLE 1852PROTOTYPES: ENABLE
1547 1853
1548void to_json (SV *scalar) 1854void encode_json (SV *scalar)
1855 ALIAS:
1856 to_json_ = 0
1857 encode_json = F_UTF8
1549 PPCODE: 1858 PPCODE:
1550{ 1859{
1551 JSON json = { F_DEFAULT | F_UTF8 }; 1860 JSON json;
1861 json_init (&json);
1862 json.flags |= ix;
1552 XPUSHs (encode_json (scalar, &json)); 1863 XPUSHs (encode_json (scalar, &json));
1553} 1864}
1554 1865
1555void from_json (SV *jsonstr) 1866void decode_json (SV *jsonstr)
1867 ALIAS:
1868 from_json_ = 0
1869 decode_json = F_UTF8
1556 PPCODE: 1870 PPCODE:
1557{ 1871{
1558 JSON json = { F_DEFAULT | F_UTF8 }; 1872 JSON json;
1873 json_init (&json);
1874 json.flags |= ix;
1559 XPUSHs (decode_json (jsonstr, &json, 0)); 1875 XPUSHs (decode_json (jsonstr, &json, 0));
1560} 1876}
1561 1877
1878

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines