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 1.62 by root, Sun Aug 26 22:27:32 2007 UTC vs.
Revision 1.84 by root, Thu Mar 27 06:37:35 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
… …
27#define F_SPACE_AFTER 0x00000040UL 30#define F_SPACE_AFTER 0x00000040UL
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
35#define F_RELAXED 0x00001000UL
36#define F_ALLOW_UNKNOWN 0x00002000UL
37
32#define F_MAXDEPTH 0xf8000000UL 38#define F_MAXDEPTH 0xf8000000UL
33#define S_MAXDEPTH 27 39#define S_MAXDEPTH 27
34#define F_MAXSIZE 0x01f00000UL 40#define F_MAXSIZE 0x01f00000UL
35#define S_MAXSIZE 20 41#define S_MAXSIZE 20
36#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing 42#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
… …
48 54
49#define SB do { 55#define SB do {
50#define SE } while (0) 56#define SE } while (0)
51 57
52#if __GNUC__ >= 3 58#if __GNUC__ >= 3
53# define expect(expr,value) __builtin_expect ((expr),(value)) 59# define expect(expr,value) __builtin_expect ((expr), (value))
54# define inline inline 60# define INLINE static inline
55#else 61#else
56# define expect(expr,value) (expr) 62# define expect(expr,value) (expr)
57# define inline static 63# define INLINE static
58#endif 64#endif
59 65
60#define expect_false(expr) expect ((expr) != 0, 0) 66#define expect_false(expr) expect ((expr) != 0, 0)
61#define expect_true(expr) expect ((expr) != 0, 1) 67#define expect_true(expr) expect ((expr) != 0, 1)
68
69#define IN_RANGE_INC(type,val,beg,end) \
70 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \
71 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
62 72
63#ifdef USE_ITHREADS 73#ifdef USE_ITHREADS
64# define JSON_SLOW 1 74# define JSON_SLOW 1
65# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1)) 75# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1))
66#else 76#else
… …
69#endif 79#endif
70 80
71static HV *json_stash, *json_boolean_stash; // JSON::XS:: 81static HV *json_stash, *json_boolean_stash; // JSON::XS::
72static SV *json_true, *json_false; 82static SV *json_true, *json_false;
73 83
84enum {
85 INCR_M_WS = 0, // initial whitespace skipping, must be 0
86 INCR_M_STR, // inside string
87 INCR_M_BS, // inside backslash
88 INCR_M_JSON // outside anything, count nesting
89};
90
91#define INCR_DONE(json) (!(json)->incr_nest && (json)->incr_mode == INCR_M_JSON)
92
74typedef struct { 93typedef struct {
75 U32 flags; 94 U32 flags;
76 SV *cb_object; 95 SV *cb_object;
77 HV *cb_sk_object; 96 HV *cb_sk_object;
97
98 // for the incremental parser
99 SV *incr_text; // the source text so far
100 STRLEN incr_pos; // the current offset into the text
101 int incr_nest; // {[]}-nesting level
102 int incr_mode;
78} JSON; 103} JSON;
79 104
80///////////////////////////////////////////////////////////////////////////// 105/////////////////////////////////////////////////////////////////////////////
81// utility functions 106// utility functions
82 107
83inline void 108INLINE SV *
109get_bool (const char *name)
110{
111 SV *sv = get_sv (name, 1);
112
113 SvREADONLY_on (sv);
114 SvREADONLY_on (SvRV (sv));
115
116 return sv;
117}
118
119INLINE void
84shrink (SV *sv) 120shrink (SV *sv)
85{ 121{
86 sv_utf8_downgrade (sv, 1); 122 sv_utf8_downgrade (sv, 1);
87 if (SvLEN (sv) > SvCUR (sv) + 1) 123 if (SvLEN (sv) > SvCUR (sv) + 1)
88 { 124 {
… …
97// decode an utf-8 character and return it, or (UV)-1 in 133// decode an utf-8 character and return it, or (UV)-1 in
98// case of an error. 134// case of an error.
99// we special-case "safe" characters from U+80 .. U+7FF, 135// we special-case "safe" characters from U+80 .. U+7FF,
100// but use the very good perl function to parse anything else. 136// but use the very good perl function to parse anything else.
101// note that we never call this function for a ascii codepoints 137// note that we never call this function for a ascii codepoints
102inline UV 138INLINE UV
103decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) 139decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
104{ 140{
105 if (expect_false (s[0] > 0xdf || s[0] < 0xc2)) 141 if (expect_true (len >= 2
106 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); 142 && IN_RANGE_INC (char, s[0], 0xc2, 0xdf)
107 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) 143 && IN_RANGE_INC (char, s[1], 0x80, 0xbf)))
108 { 144 {
109 *clen = 2; 145 *clen = 2;
110 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 146 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
111 } 147 }
112 else 148 else
113 { 149 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
114 *clen = (STRLEN)-1; 150}
115 return (UV)-1; 151
116 } 152// likewise for encoding, also never called for ascii codepoints
153// this function takes advantage of this fact, although current gccs
154// seem to optimise the check for >= 0x80 away anyways
155INLINE unsigned char *
156encode_utf8 (unsigned char *s, UV ch)
157{
158 if (expect_false (ch < 0x000080))
159 *s++ = ch;
160 else if (expect_true (ch < 0x000800))
161 *s++ = 0xc0 | ( ch >> 6),
162 *s++ = 0x80 | ( ch & 0x3f);
163 else if ( ch < 0x010000)
164 *s++ = 0xe0 | ( ch >> 12),
165 *s++ = 0x80 | ((ch >> 6) & 0x3f),
166 *s++ = 0x80 | ( ch & 0x3f);
167 else if ( ch < 0x110000)
168 *s++ = 0xf0 | ( ch >> 18),
169 *s++ = 0x80 | ((ch >> 12) & 0x3f),
170 *s++ = 0x80 | ((ch >> 6) & 0x3f),
171 *s++ = 0x80 | ( ch & 0x3f);
172
173 return s;
117} 174}
118 175
119///////////////////////////////////////////////////////////////////////////// 176/////////////////////////////////////////////////////////////////////////////
120// encoder 177// encoder
121 178
… …
126 char *end; // SvEND (sv) 183 char *end; // SvEND (sv)
127 SV *sv; // result scalar 184 SV *sv; // result scalar
128 JSON json; 185 JSON json;
129 U32 indent; // indentation level 186 U32 indent; // indentation level
130 U32 maxdepth; // max. indentation/recursion level 187 U32 maxdepth; // max. indentation/recursion level
188 UV limit; // escape character values >= this value when encoding
131} enc_t; 189} enc_t;
132 190
133inline void 191INLINE void
134need (enc_t *enc, STRLEN len) 192need (enc_t *enc, STRLEN len)
135{ 193{
136 if (expect_false (enc->cur + len >= enc->end)) 194 if (expect_false (enc->cur + len >= enc->end))
137 { 195 {
138 STRLEN cur = enc->cur - SvPVX (enc->sv); 196 STRLEN cur = enc->cur - SvPVX (enc->sv);
… …
140 enc->cur = SvPVX (enc->sv) + cur; 198 enc->cur = SvPVX (enc->sv) + cur;
141 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 199 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
142 } 200 }
143} 201}
144 202
145inline void 203INLINE void
146encode_ch (enc_t *enc, char ch) 204encode_ch (enc_t *enc, char ch)
147{ 205{
148 need (enc, 1); 206 need (enc, 1);
149 *enc->cur++ = ch; 207 *enc->cur++ = ch;
150} 208}
… …
204 { 262 {
205 uch = ch; 263 uch = ch;
206 clen = 1; 264 clen = 1;
207 } 265 }
208 266
209 if (uch > 0x10FFFFUL) 267 if (uch < 0x80/*0x20*/ || uch >= enc->limit)
210 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
211
212 if (uch < 0x80 || enc->json.flags & F_ASCII || (enc->json.flags & F_LATIN1 && uch > 0xFF))
213 { 268 {
214 if (uch > 0xFFFFUL) 269 if (uch >= 0x10000UL)
215 { 270 {
271 if (uch >= 0x110000UL)
272 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
273
216 need (enc, len += 11); 274 need (enc, len += 11);
217 sprintf (enc->cur, "\\u%04x\\u%04x", 275 sprintf (enc->cur, "\\u%04x\\u%04x",
218 (int)((uch - 0x10000) / 0x400 + 0xD800), 276 (int)((uch - 0x10000) / 0x400 + 0xD800),
219 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 277 (int)((uch - 0x10000) % 0x400 + 0xDC00));
220 enc->cur += 12; 278 enc->cur += 12;
… …
248 while (--clen); 306 while (--clen);
249 } 307 }
250 else 308 else
251 { 309 {
252 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed 310 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
253 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 311 enc->cur = encode_utf8 (enc->cur, uch);
254 ++str; 312 ++str;
255 } 313 }
256 } 314 }
257 } 315 }
258 } 316 }
259 317
260 --len; 318 --len;
261 } 319 }
262} 320}
263 321
264inline void 322INLINE void
265encode_indent (enc_t *enc) 323encode_indent (enc_t *enc)
266{ 324{
267 if (enc->json.flags & F_INDENT) 325 if (enc->json.flags & F_INDENT)
268 { 326 {
269 int spaces = enc->indent * INDENT_STEP; 327 int spaces = enc->indent * INDENT_STEP;
… …
272 memset (enc->cur, ' ', spaces); 330 memset (enc->cur, ' ', spaces);
273 enc->cur += spaces; 331 enc->cur += spaces;
274 } 332 }
275} 333}
276 334
277inline void 335INLINE void
278encode_space (enc_t *enc) 336encode_space (enc_t *enc)
279{ 337{
280 need (enc, 1); 338 need (enc, 1);
281 encode_ch (enc, ' '); 339 encode_ch (enc, ' ');
282} 340}
283 341
284inline void 342INLINE void
285encode_nl (enc_t *enc) 343encode_nl (enc_t *enc)
286{ 344{
287 if (enc->json.flags & F_INDENT) 345 if (enc->json.flags & F_INDENT)
288 { 346 {
289 need (enc, 1); 347 need (enc, 1);
290 encode_ch (enc, '\n'); 348 encode_ch (enc, '\n');
291 } 349 }
292} 350}
293 351
294inline void 352INLINE void
295encode_comma (enc_t *enc) 353encode_comma (enc_t *enc)
296{ 354{
297 encode_ch (enc, ','); 355 encode_ch (enc, ',');
298 356
299 if (enc->json.flags & F_INDENT) 357 if (enc->json.flags & F_INDENT)
… …
310 int i, len = av_len (av); 368 int i, len = av_len (av);
311 369
312 if (enc->indent >= enc->maxdepth) 370 if (enc->indent >= enc->maxdepth)
313 croak ("data structure too deep (hit recursion limit)"); 371 croak ("data structure too deep (hit recursion limit)");
314 372
315 encode_ch (enc, '['); encode_nl (enc); 373 encode_ch (enc, '[');
316 ++enc->indent; 374
375 if (len >= 0)
376 {
377 encode_nl (enc); ++enc->indent;
317 378
318 for (i = 0; i <= len; ++i) 379 for (i = 0; i <= len; ++i)
319 { 380 {
320 SV **svp = av_fetch (av, i, 0); 381 SV **svp = av_fetch (av, i, 0);
321 382
322 encode_indent (enc); 383 encode_indent (enc);
323 384
324 if (svp) 385 if (svp)
325 encode_sv (enc, *svp); 386 encode_sv (enc, *svp);
326 else 387 else
327 encode_str (enc, "null", 4, 0); 388 encode_str (enc, "null", 4, 0);
328 389
329 if (i < len) 390 if (i < len)
330 encode_comma (enc); 391 encode_comma (enc);
331 } 392 }
332 393
394 encode_nl (enc); --enc->indent; encode_indent (enc);
395 }
396
333 encode_nl (enc); 397 encode_ch (enc, ']');
334
335 --enc->indent;
336 encode_indent (enc); encode_ch (enc, ']');
337} 398}
338 399
339static void 400static void
340encode_hk (enc_t *enc, HE *he) 401encode_hk (enc_t *enc, HE *he)
341{ 402{
… …
389 450
390static void 451static void
391encode_hv (enc_t *enc, HV *hv) 452encode_hv (enc_t *enc, HV *hv)
392{ 453{
393 HE *he; 454 HE *he;
394 int count;
395 455
396 if (enc->indent >= enc->maxdepth) 456 if (enc->indent >= enc->maxdepth)
397 croak ("data structure too deep (hit recursion limit)"); 457 croak ("data structure too deep (hit recursion limit)");
398 458
399 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 459 encode_ch (enc, '{');
400 460
401 // for canonical output we have to sort by keys first 461 // for canonical output we have to sort by keys first
402 // actually, this is mostly due to the stupid so-called 462 // actually, this is mostly due to the stupid so-called
403 // security workaround added somewhere in 5.8.x. 463 // security workaround added somewhere in 5.8.x.
404 // that randomises hash orderings 464 // that randomises hash orderings
… …
457 517
458 FREETMPS; 518 FREETMPS;
459 LEAVE; 519 LEAVE;
460 } 520 }
461 521
522 encode_nl (enc); ++enc->indent;
523
462 while (count--) 524 while (count--)
463 { 525 {
464 encode_indent (enc); 526 encode_indent (enc);
465 he = hes [count]; 527 he = hes [count];
466 encode_hk (enc, he); 528 encode_hk (enc, he);
467 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he)); 529 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
468 530
469 if (count) 531 if (count)
470 encode_comma (enc); 532 encode_comma (enc);
471 } 533 }
534
535 encode_nl (enc); --enc->indent; encode_indent (enc);
472 } 536 }
473 } 537 }
474 else 538 else
475 { 539 {
476 if (hv_iterinit (hv) || SvMAGICAL (hv)) 540 if (hv_iterinit (hv) || SvMAGICAL (hv))
477 if ((he = hv_iternext (hv))) 541 if ((he = hv_iternext (hv)))
542 {
543 encode_nl (enc); ++enc->indent;
544
478 for (;;) 545 for (;;)
479 { 546 {
480 encode_indent (enc); 547 encode_indent (enc);
481 encode_hk (enc, he); 548 encode_hk (enc, he);
482 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he)); 549 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
483 550
484 if (!(he = hv_iternext (hv))) 551 if (!(he = hv_iternext (hv)))
485 break; 552 break;
486 553
487 encode_comma (enc); 554 encode_comma (enc);
488 } 555 }
489 }
490 556
557 encode_nl (enc); --enc->indent; encode_indent (enc);
558 }
559 }
560
491 encode_nl (enc); 561 encode_ch (enc, '}');
492
493 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
494} 562}
495 563
496// encode objects, arrays and special \0=false and \1=true values. 564// encode objects, arrays and special \0=false and \1=true values.
497static void 565static void
498encode_rv (enc_t *enc, SV *sv) 566encode_rv (enc_t *enc, SV *sv)
… …
575 643
576 if (len == 1 && *pv == '1') 644 if (len == 1 && *pv == '1')
577 encode_str (enc, "true", 4, 0); 645 encode_str (enc, "true", 4, 0);
578 else if (len == 1 && *pv == '0') 646 else if (len == 1 && *pv == '0')
579 encode_str (enc, "false", 5, 0); 647 encode_str (enc, "false", 5, 0);
648 else if (enc->json.flags & F_ALLOW_UNKNOWN)
649 encode_str (enc, "null", 4, 0);
580 else 650 else
581 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", 651 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
582 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 652 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
583 } 653 }
654 else if (enc->json.flags & F_ALLOW_UNKNOWN)
655 encode_str (enc, "null", 4, 0);
584 else 656 else
585 croak ("encountered %s, but JSON can only represent references to arrays or hashes", 657 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
586 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 658 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
587} 659}
588 660
… …
606 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 678 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
607 enc->cur += strlen (enc->cur); 679 enc->cur += strlen (enc->cur);
608 } 680 }
609 else if (SvIOKp (sv)) 681 else if (SvIOKp (sv))
610 { 682 {
611 // we assume we can always read an IV as a UV 683 // we assume we can always read an IV as a UV and vice versa
612 if (SvUV (sv) & ~(UV)0x7fff) 684 // we assume two's complement
613 { 685 // we assume no aliasing issues in the union
614 // large integer, use the (rather slow) snprintf way. 686 if (SvIsUV (sv) ? SvUVX (sv) <= 59000
615 need (enc, sizeof (UV) * 3); 687 : SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000)
616 enc->cur +=
617 SvIsUV(sv)
618 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
619 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
620 }
621 else
622 { 688 {
623 // optimise the "small number case" 689 // optimise the "small number case"
624 // code will likely be branchless and use only a single multiplication 690 // code will likely be branchless and use only a single multiplication
691 // works for numbers up to 59074
625 I32 i = SvIV (sv); 692 I32 i = SvIVX (sv);
626 U32 u; 693 U32 u;
627 char digit, nz = 0; 694 char digit, nz = 0;
628 695
629 need (enc, 6); 696 need (enc, 6);
630 697
… …
636 703
637 // now output digit by digit, each time masking out the integer part 704 // now output digit by digit, each time masking out the integer part
638 // and multiplying by 5 while moving the decimal point one to the right, 705 // and multiplying by 5 while moving the decimal point one to the right,
639 // resulting in a net multiplication by 10. 706 // resulting in a net multiplication by 10.
640 // we always write the digit to memory but conditionally increment 707 // we always write the digit to memory but conditionally increment
641 // the pointer, to ease the usage of conditional move instructions. 708 // the pointer, to enable the use of conditional move instructions.
642 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5; 709 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffffUL) * 5;
643 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5; 710 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffffUL) * 5;
644 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5; 711 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffffUL) * 5;
645 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5; 712 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffffUL) * 5;
646 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0' 713 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
647 } 714 }
715 else
716 {
717 // large integer, use the (rather slow) snprintf way.
718 need (enc, IVUV_MAXCHARS);
719 enc->cur +=
720 SvIsUV(sv)
721 ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))
722 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
723 }
648 } 724 }
649 else if (SvROK (sv)) 725 else if (SvROK (sv))
650 encode_rv (enc, SvRV (sv)); 726 encode_rv (enc, SvRV (sv));
651 else if (!SvOK (sv)) 727 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN)
652 encode_str (enc, "null", 4, 0); 728 encode_str (enc, "null", 4, 0);
653 else 729 else
654 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 730 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
655 SvPV_nolen (sv), SvFLAGS (sv)); 731 SvPV_nolen (sv), SvFLAGS (sv));
656} 732}
… …
667 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 743 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
668 enc.cur = SvPVX (enc.sv); 744 enc.cur = SvPVX (enc.sv);
669 enc.end = SvEND (enc.sv); 745 enc.end = SvEND (enc.sv);
670 enc.indent = 0; 746 enc.indent = 0;
671 enc.maxdepth = DEC_DEPTH (enc.json.flags); 747 enc.maxdepth = DEC_DEPTH (enc.json.flags);
748 enc.limit = enc.json.flags & F_ASCII ? 0x000080UL
749 : enc.json.flags & F_LATIN1 ? 0x000100UL
750 : 0x110000UL;
672 751
673 SvPOK_only (enc.sv); 752 SvPOK_only (enc.sv);
674 encode_sv (&enc, scalar); 753 encode_sv (&enc, scalar);
675 754
676 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 755 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
… …
697 JSON json; 776 JSON json;
698 U32 depth; // recursion depth 777 U32 depth; // recursion depth
699 U32 maxdepth; // recursion depth limit 778 U32 maxdepth; // recursion depth limit
700} dec_t; 779} dec_t;
701 780
702inline void 781INLINE void
782decode_comment (dec_t *dec)
783{
784 // only '#'-style comments allowed a.t.m.
785
786 while (*dec->cur && *dec->cur != 0x0a && *dec->cur != 0x0d)
787 ++dec->cur;
788}
789
790INLINE void
703decode_ws (dec_t *dec) 791decode_ws (dec_t *dec)
704{ 792{
705 for (;;) 793 for (;;)
706 { 794 {
707 char ch = *dec->cur; 795 char ch = *dec->cur;
708 796
709 if (ch > 0x20 797 if (ch > 0x20)
798 {
799 if (expect_false (ch == '#'))
800 {
801 if (dec->json.flags & F_RELAXED)
802 decode_comment (dec);
803 else
804 break;
805 }
806 else
807 break;
808 }
710 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) 809 else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)
711 break; 810 break; // parse error, but let higher level handle it, gives better error messages
712 811
713 ++dec->cur; 812 ++dec->cur;
714 } 813 }
715} 814}
716 815
… …
822 921
823 if (hi >= 0x80) 922 if (hi >= 0x80)
824 { 923 {
825 utf8 = 1; 924 utf8 = 1;
826 925
827 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0); 926 cur = encode_utf8 (cur, hi);
828 } 927 }
829 else 928 else
830 *cur++ = hi; 929 *cur++ = hi;
831 } 930 }
832 break; 931 break;
… …
834 default: 933 default:
835 --dec_cur; 934 --dec_cur;
836 ERR ("illegal backslash escape sequence in string"); 935 ERR ("illegal backslash escape sequence in string");
837 } 936 }
838 } 937 }
839 else if (expect_true (ch >= 0x20 && ch <= 0x7f)) 938 else if (expect_true (ch >= 0x20 && ch < 0x80))
840 *cur++ = ch; 939 *cur++ = ch;
841 else if (ch >= 0x80) 940 else if (ch >= 0x80)
842 { 941 {
843 STRLEN clen; 942 STRLEN clen;
844 UV uch; 943 UV uch;
… …
967 1066
968 if (!is_nv) 1067 if (!is_nv)
969 { 1068 {
970 int len = dec->cur - start; 1069 int len = dec->cur - start;
971 1070
972 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so 1071 // special case the rather common 1..5-digit-int case
973 if (*start == '-') 1072 if (*start == '-')
974 switch (len) 1073 switch (len)
975 { 1074 {
976 case 2: return newSViv (-( start [1] - '0' * 1)); 1075 case 2: return newSViv (-( start [1] - '0' * 1));
977 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); 1076 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
978 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); 1077 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
979 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); 1078 case 5: return newSViv (-( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
1079 case 6: return newSViv (-(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
980 } 1080 }
981 else 1081 else
982 switch (len) 1082 switch (len)
983 { 1083 {
984 case 1: return newSViv ( start [0] - '0' * 1); 1084 case 1: return newSViv ( start [0] - '0' * 1);
985 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); 1085 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
986 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); 1086 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
987 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); 1087 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
1088 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
988 } 1089 }
989 1090
990 { 1091 {
991 UV uv; 1092 UV uv;
992 int numtype = grok_number (start, len, &uv); 1093 int numtype = grok_number (start, len, &uv);
… …
1053 1154
1054 if (*dec->cur != ',') 1155 if (*dec->cur != ',')
1055 ERR (", or ] expected while parsing array"); 1156 ERR (", or ] expected while parsing array");
1056 1157
1057 ++dec->cur; 1158 ++dec->cur;
1159
1160 decode_ws (dec);
1161
1162 if (*dec->cur == ']' && dec->json.flags & F_RELAXED)
1163 {
1164 ++dec->cur;
1165 break;
1166 }
1058 } 1167 }
1059 1168
1060 DEC_DEC_DEPTH; 1169 DEC_DEC_DEPTH;
1061 return newRV_noinc ((SV *)av); 1170 return newRV_noinc ((SV *)av);
1062 1171
… …
1078 if (*dec->cur == '}') 1187 if (*dec->cur == '}')
1079 ++dec->cur; 1188 ++dec->cur;
1080 else 1189 else
1081 for (;;) 1190 for (;;)
1082 { 1191 {
1083 decode_ws (dec); EXPECT_CH ('"'); 1192 EXPECT_CH ('"');
1084 1193
1085 // heuristic: assume that 1194 // heuristic: assume that
1086 // a) decode_str + hv_store_ent are abysmally slow. 1195 // a) decode_str + hv_store_ent are abysmally slow.
1087 // b) most hash keys are short, simple ascii text. 1196 // b) most hash keys are short, simple ascii text.
1088 // => try to "fast-match" such strings to avoid 1197 // => try to "fast-match" such strings to avoid
… …
1092 char *p = dec->cur; 1201 char *p = dec->cur;
1093 char *e = p + 24; // only try up to 24 bytes 1202 char *e = p + 24; // only try up to 24 bytes
1094 1203
1095 for (;;) 1204 for (;;)
1096 { 1205 {
1097 // the >= 0x80 is true on most architectures 1206 // the >= 0x80 is false on most architectures
1098 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\') 1207 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1099 { 1208 {
1100 // slow path, back up and use decode_str 1209 // slow path, back up and use decode_str
1101 SV *key = decode_str (dec); 1210 SV *key = decode_str (dec);
1102 if (!key) 1211 if (!key)
1103 goto fail; 1212 goto fail;
1104 1213
1105 decode_ws (dec); EXPECT_CH (':'); 1214 decode_ws (dec); EXPECT_CH (':');
1106 1215
1216 decode_ws (dec);
1107 value = decode_sv (dec); 1217 value = decode_sv (dec);
1108 if (!value) 1218 if (!value)
1109 { 1219 {
1110 SvREFCNT_dec (key); 1220 SvREFCNT_dec (key);
1111 goto fail; 1221 goto fail;
… …
1123 int len = p - key; 1233 int len = p - key;
1124 dec->cur = p + 1; 1234 dec->cur = p + 1;
1125 1235
1126 decode_ws (dec); EXPECT_CH (':'); 1236 decode_ws (dec); EXPECT_CH (':');
1127 1237
1238 decode_ws (dec);
1128 value = decode_sv (dec); 1239 value = decode_sv (dec);
1129 if (!value) 1240 if (!value)
1130 goto fail; 1241 goto fail;
1131 1242
1132 hv_store (hv, key, len, value, 0); 1243 hv_store (hv, key, len, value, 0);
… …
1148 1259
1149 if (*dec->cur != ',') 1260 if (*dec->cur != ',')
1150 ERR (", or } expected while parsing object/hash"); 1261 ERR (", or } expected while parsing object/hash");
1151 1262
1152 ++dec->cur; 1263 ++dec->cur;
1264
1265 decode_ws (dec);
1266
1267 if (*dec->cur == '}' && dec->json.flags & F_RELAXED)
1268 {
1269 ++dec->cur;
1270 break;
1271 }
1153 } 1272 }
1154 1273
1155 DEC_DEC_DEPTH; 1274 DEC_DEC_DEPTH;
1156 sv = newRV_noinc ((SV *)hv); 1275 sv = newRV_noinc ((SV *)hv);
1157 1276
… …
1222} 1341}
1223 1342
1224static SV * 1343static SV *
1225decode_sv (dec_t *dec) 1344decode_sv (dec_t *dec)
1226{ 1345{
1227 decode_ws (dec);
1228
1229 // the beauty of JSON: you need exactly one character lookahead 1346 // the beauty of JSON: you need exactly one character lookahead
1230 // to parse anything. 1347 // to parse everything.
1231 switch (*dec->cur) 1348 switch (*dec->cur)
1232 { 1349 {
1233 case '"': ++dec->cur; return decode_str (dec); 1350 case '"': ++dec->cur; return decode_str (dec);
1234 case '[': ++dec->cur; return decode_av (dec); 1351 case '[': ++dec->cur; return decode_av (dec);
1235 case '{': ++dec->cur; return decode_hv (dec); 1352 case '{': ++dec->cur; return decode_hv (dec);
1236 1353
1237 case '-': 1354 case '-':
1238 case '0': case '1': case '2': case '3': case '4': 1355 case '0': case '1': case '2': case '3': case '4':
1239 case '5': case '6': case '7': case '8': case '9': 1356 case '5': case '6': case '7': case '8': case '9':
1240 return decode_num (dec); 1357 return decode_num (dec);
… …
1242 case 't': 1359 case 't':
1243 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1360 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1244 { 1361 {
1245 dec->cur += 4; 1362 dec->cur += 4;
1246#if JSON_SLOW 1363#if JSON_SLOW
1247 json_true = get_sv ("JSON::XS::true", 1); SvREADONLY_on (json_true); 1364 json_true = get_bool ("JSON::XS::true");
1248#endif 1365#endif
1249 return SvREFCNT_inc (json_true); 1366 return newSVsv (json_true);
1250 } 1367 }
1251 else 1368 else
1252 ERR ("'true' expected"); 1369 ERR ("'true' expected");
1253 1370
1254 break; 1371 break;
… …
1256 case 'f': 1373 case 'f':
1257 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1374 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1258 { 1375 {
1259 dec->cur += 5; 1376 dec->cur += 5;
1260#if JSON_SLOW 1377#if JSON_SLOW
1261 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); 1378 json_false = get_bool ("JSON::XS::false");
1262#endif 1379#endif
1263 return SvREFCNT_inc (json_false); 1380 return newSVsv (json_false);
1264 } 1381 }
1265 else 1382 else
1266 ERR ("'false' expected"); 1383 ERR ("'false' expected");
1267 1384
1268 break; 1385 break;
… …
1286fail: 1403fail:
1287 return 0; 1404 return 0;
1288} 1405}
1289 1406
1290static SV * 1407static SV *
1291decode_json (SV *string, JSON *json, UV *offset_return) 1408decode_json (SV *string, JSON *json, STRLEN *offset_return)
1292{ 1409{
1293 dec_t dec; 1410 dec_t dec;
1294 UV offset; 1411 STRLEN offset;
1295 SV *sv; 1412 SV *sv;
1296 1413
1297 SvGETMAGIC (string); 1414 SvGETMAGIC (string);
1298 SvUPGRADE (string, SVt_PV); 1415 SvUPGRADE (string, SVt_PV);
1299 1416
… …
1317 1434
1318 if (dec.json.cb_object || dec.json.cb_sk_object) 1435 if (dec.json.cb_object || dec.json.cb_sk_object)
1319 dec.json.flags |= F_HOOK; 1436 dec.json.flags |= F_HOOK;
1320 1437
1321 *dec.end = 0; // this should basically be a nop, too, but make sure it's there 1438 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1439
1440 decode_ws (&dec);
1322 sv = decode_sv (&dec); 1441 sv = decode_sv (&dec);
1323 1442
1324 if (!(offset_return || !sv)) 1443 if (!(offset_return || !sv))
1325 { 1444 {
1326 // check for trailing garbage 1445 // check for trailing garbage
… …
1370 1489
1371 return sv; 1490 return sv;
1372} 1491}
1373 1492
1374///////////////////////////////////////////////////////////////////////////// 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 ++self->incr_nest;
1590 break;
1591
1592 case ']':
1593 case '}':
1594 if (!--self->incr_nest)
1595 goto interrupt;
1596 }
1597 }
1598 }
1599
1600 modechange:
1601 ;
1602 }
1603
1604interrupt:
1605 self->incr_pos = p - SvPVX (self->incr_text);
1606 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D
1607}
1608
1609/////////////////////////////////////////////////////////////////////////////
1375// XS interface functions 1610// XS interface functions
1376 1611
1377MODULE = JSON::XS PACKAGE = JSON::XS 1612MODULE = JSON::XS PACKAGE = JSON::XS
1378 1613
1379BOOT: 1614BOOT:
… …
1388 : -1; 1623 : -1;
1389 1624
1390 json_stash = gv_stashpv ("JSON::XS" , 1); 1625 json_stash = gv_stashpv ("JSON::XS" , 1);
1391 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); 1626 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1392 1627
1393 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true ); 1628 json_true = get_bool ("JSON::XS::true");
1394 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); 1629 json_false = get_bool ("JSON::XS::false");
1395} 1630}
1396 1631
1397PROTOTYPES: DISABLE 1632PROTOTYPES: DISABLE
1398 1633
1399void CLONE (...) 1634void CLONE (...)
… …
1406{ 1641{
1407 SV *pv = NEWSV (0, sizeof (JSON)); 1642 SV *pv = NEWSV (0, sizeof (JSON));
1408 SvPOK_only (pv); 1643 SvPOK_only (pv);
1409 Zero (SvPVX (pv), 1, JSON); 1644 Zero (SvPVX (pv), 1, JSON);
1410 ((JSON *)SvPVX (pv))->flags = F_DEFAULT; 1645 ((JSON *)SvPVX (pv))->flags = F_DEFAULT;
1411 XPUSHs (sv_2mortal (sv_bless (newRV_noinc (pv), JSON_STASH))); 1646 XPUSHs (sv_2mortal (sv_bless (
1647 newRV_noinc (pv),
1648 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1649 )));
1412} 1650}
1413 1651
1414void ascii (JSON *self, int enable = 1) 1652void ascii (JSON *self, int enable = 1)
1415 ALIAS: 1653 ALIAS:
1416 ascii = F_ASCII 1654 ascii = F_ASCII
… …
1423 pretty = F_PRETTY 1661 pretty = F_PRETTY
1424 allow_nonref = F_ALLOW_NONREF 1662 allow_nonref = F_ALLOW_NONREF
1425 shrink = F_SHRINK 1663 shrink = F_SHRINK
1426 allow_blessed = F_ALLOW_BLESSED 1664 allow_blessed = F_ALLOW_BLESSED
1427 convert_blessed = F_CONV_BLESSED 1665 convert_blessed = F_CONV_BLESSED
1666 relaxed = F_RELAXED
1667 allow_unknown = F_ALLOW_UNKNOWN
1428 PPCODE: 1668 PPCODE:
1429{ 1669{
1430 if (enable) 1670 if (enable)
1431 self->flags |= ix; 1671 self->flags |= ix;
1432 else 1672 else
1433 self->flags &= ~ix; 1673 self->flags &= ~ix;
1434 1674
1435 XPUSHs (ST (0)); 1675 XPUSHs (ST (0));
1436} 1676}
1437 1677
1678void get_ascii (JSON *self)
1679 ALIAS:
1680 get_ascii = F_ASCII
1681 get_latin1 = F_LATIN1
1682 get_utf8 = F_UTF8
1683 get_indent = F_INDENT
1684 get_canonical = F_CANONICAL
1685 get_space_before = F_SPACE_BEFORE
1686 get_space_after = F_SPACE_AFTER
1687 get_allow_nonref = F_ALLOW_NONREF
1688 get_shrink = F_SHRINK
1689 get_allow_blessed = F_ALLOW_BLESSED
1690 get_convert_blessed = F_CONV_BLESSED
1691 get_relaxed = F_RELAXED
1692 get_allow_unknown = F_ALLOW_UNKNOWN
1693 PPCODE:
1694 XPUSHs (boolSV (self->flags & ix));
1695
1438void max_depth (JSON *self, UV max_depth = 0x80000000UL) 1696void max_depth (JSON *self, UV max_depth = 0x80000000UL)
1439 PPCODE: 1697 PPCODE:
1440{ 1698{
1441 UV log2 = 0; 1699 UV log2 = 0;
1442 1700
… …
1448 self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH); 1706 self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1449 1707
1450 XPUSHs (ST (0)); 1708 XPUSHs (ST (0));
1451} 1709}
1452 1710
1711U32 get_max_depth (JSON *self)
1712 CODE:
1713 RETVAL = DEC_DEPTH (self->flags);
1714 OUTPUT:
1715 RETVAL
1716
1453void max_size (JSON *self, UV max_size = 0) 1717void max_size (JSON *self, UV max_size = 0)
1454 PPCODE: 1718 PPCODE:
1455{ 1719{
1456 UV log2 = 0; 1720 UV log2 = 0;
1457 1721
… …
1463 1727
1464 self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE); 1728 self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1465 1729
1466 XPUSHs (ST (0)); 1730 XPUSHs (ST (0));
1467} 1731}
1732
1733int get_max_size (JSON *self)
1734 CODE:
1735 RETVAL = DEC_SIZE (self->flags);
1736 OUTPUT:
1737 RETVAL
1468 1738
1469void filter_json_object (JSON *self, SV *cb = &PL_sv_undef) 1739void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1470 PPCODE: 1740 PPCODE:
1471{ 1741{
1472 SvREFCNT_dec (self->cb_object); 1742 SvREFCNT_dec (self->cb_object);
… …
1506 XPUSHs (decode_json (jsonstr, self, 0)); 1776 XPUSHs (decode_json (jsonstr, self, 0));
1507 1777
1508void decode_prefix (JSON *self, SV *jsonstr) 1778void decode_prefix (JSON *self, SV *jsonstr)
1509 PPCODE: 1779 PPCODE:
1510{ 1780{
1511 UV offset; 1781 STRLEN offset;
1512 EXTEND (SP, 2); 1782 EXTEND (SP, 2);
1513 PUSHs (decode_json (jsonstr, self, &offset)); 1783 PUSHs (decode_json (jsonstr, self, &offset));
1514 PUSHs (sv_2mortal (newSVuv (offset))); 1784 PUSHs (sv_2mortal (newSVuv (offset)));
1785}
1786
1787void incr_parse (JSON *self, SV *jsonstr = 0)
1788 PPCODE:
1789{
1790 if (!self->incr_text)
1791 self->incr_text = newSVpvn ("", 0);
1792
1793 // append data, if any
1794 if (jsonstr)
1795 {
1796 if (SvUTF8 (jsonstr) && !SvUTF8 (self->incr_text))
1797 {
1798 /* utf-8-ness differs, need to upgrade */
1799 sv_utf8_upgrade (self->incr_text);
1800
1801 if (self->incr_pos)
1802 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
1803 - (U8 *)SvPVX (self->incr_text);
1804 }
1805
1806 {
1807 STRLEN len;
1808 const char *str = SvPV (jsonstr, len);
1809 SvGROW (self->incr_text, SvCUR (self->incr_text) + len + 1);
1810 Move (str, SvEND (self->incr_text), len, char);
1811 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
1812 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
1813 }
1814 }
1815
1816 if (GIMME_V != G_VOID)
1817 do
1818 {
1819 STRLEN offset;
1820
1821 if (!INCR_DONE (self))
1822 {
1823 incr_parse (self);
1824 if (!INCR_DONE (self))
1825 break;
1826 }
1827
1828 XPUSHs (decode_json (self->incr_text, self, &offset));
1829
1830 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + offset);
1831 self->incr_pos -= offset;
1832 self->incr_nest = 0;
1833 self->incr_mode = 0;
1834 }
1835 while (GIMME_V == G_ARRAY);
1836}
1837
1838SV *incr_text (JSON *self)
1839 ATTRS: lvalue
1840 CODE:
1841{
1842 if (self->incr_pos)
1843 croak ("incr_text can not be called when the incremental parser already started parsing");
1844
1845 RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef;
1846}
1847 OUTPUT:
1848 RETVAL
1849
1850void incr_skip (JSON *self)
1851 CODE:
1852{
1853 if (self->incr_pos)
1854 {
1855 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + self->incr_pos);
1856 self->incr_pos = 0;
1857 self->incr_nest = 0;
1858 self->incr_mode = 0;
1859 }
1515} 1860}
1516 1861
1517void DESTROY (JSON *self) 1862void DESTROY (JSON *self)
1518 CODE: 1863 CODE:
1519 SvREFCNT_dec (self->cb_sk_object); 1864 SvREFCNT_dec (self->cb_sk_object);
1520 SvREFCNT_dec (self->cb_object); 1865 SvREFCNT_dec (self->cb_object);
1866 SvREFCNT_dec (self->incr_text);
1521 1867
1522PROTOTYPES: ENABLE 1868PROTOTYPES: ENABLE
1523 1869
1524void to_json (SV *scalar) 1870void encode_json (SV *scalar)
1871 ALIAS:
1872 to_json_ = 0
1873 encode_json = F_UTF8
1525 PPCODE: 1874 PPCODE:
1526{ 1875{
1527 JSON json = { F_DEFAULT | F_UTF8 }; 1876 JSON json = { F_DEFAULT | ix };
1528 XPUSHs (encode_json (scalar, &json)); 1877 XPUSHs (encode_json (scalar, &json));
1529} 1878}
1530 1879
1531void from_json (SV *jsonstr) 1880void decode_json (SV *jsonstr)
1881 ALIAS:
1882 from_json_ = 0
1883 decode_json = F_UTF8
1532 PPCODE: 1884 PPCODE:
1533{ 1885{
1534 JSON json = { F_DEFAULT | F_UTF8 }; 1886 JSON json = { F_DEFAULT | ix };
1535 XPUSHs (decode_json (jsonstr, &json, 0)); 1887 XPUSHs (decode_json (jsonstr, &json, 0));
1536} 1888}
1537 1889
1890

Diff Legend

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