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 1.93 by root, Mon Sep 29 03:09:27 2008 UTC vs.
Revision 1.136 by root, Thu Nov 15 22:35:35 2018 UTC

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 <limits.h>
10#include <float.h> 10#include <float.h>
11#include <inttypes.h>
11 12
12#if defined(__BORLANDC__) || defined(_MSC_VER) 13#if defined(__BORLANDC__) || defined(_MSC_VER)
13# define snprintf _snprintf // C compilers have this in stdio.h 14# define snprintf _snprintf // C compilers have this in stdio.h
14#endif 15#endif
15 16
16// some old perls do not have this, try to make it work, no 17// some old perls do not have this, try to make it work, no
17// guarentees, though. if it breaks, you get to keep the pieces. 18// guarantees, though. if it breaks, you get to keep the pieces.
18#ifndef UTF8_MAXBYTES 19#ifndef UTF8_MAXBYTES
19# define UTF8_MAXBYTES 13 20# define UTF8_MAXBYTES 13
20#endif 21#endif
21 22
23// compatibility with perl <5.18
24#ifndef HvNAMELEN_get
25# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
26#endif
27#ifndef HvNAMELEN
28# define HvNAMELEN(hv) HvNAMELEN_get (hv)
29#endif
30#ifndef HvNAMEUTF8
31# define HvNAMEUTF8(hv) 0
32#endif
33
34// three extra for rounding, sign, and end of string
22#define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 2) 35#define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 3)
23 36
24#define F_ASCII 0x00000001UL 37#define F_ASCII 0x00000001UL
25#define F_LATIN1 0x00000002UL 38#define F_LATIN1 0x00000002UL
26#define F_UTF8 0x00000004UL 39#define F_UTF8 0x00000004UL
27#define F_INDENT 0x00000008UL 40#define F_INDENT 0x00000008UL
32#define F_SHRINK 0x00000200UL 45#define F_SHRINK 0x00000200UL
33#define F_ALLOW_BLESSED 0x00000400UL 46#define F_ALLOW_BLESSED 0x00000400UL
34#define F_CONV_BLESSED 0x00000800UL 47#define F_CONV_BLESSED 0x00000800UL
35#define F_RELAXED 0x00001000UL 48#define F_RELAXED 0x00001000UL
36#define F_ALLOW_UNKNOWN 0x00002000UL 49#define F_ALLOW_UNKNOWN 0x00002000UL
50#define F_ALLOW_TAGS 0x00004000UL
37#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing 51#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
38 52
39#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 53#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
40 54
41#define INIT_SIZE 32 // initial scalar size to be allocated 55#define INIT_SIZE 64 // initial scalar size to be allocated
42#define INDENT_STEP 3 // spaces per indentation level 56#define INDENT_STEP 3 // spaces per indentation level
43 57
44#define SHORT_STRING_LEN 16384 // special-case strings of up to this size 58#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
59
60#define DECODE_WANTS_OCTETS(json) ((json)->flags & F_UTF8)
45 61
46#define SB do { 62#define SB do {
47#define SE } while (0) 63#define SE } while (0)
48 64
49#if __GNUC__ >= 3 65#if __GNUC__ >= 3
62 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg))) 78 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
63 79
64#define ERR_NESTING_EXCEEDED "json text or perl structure exceeds maximum nesting level (max_depth set too low?)" 80#define ERR_NESTING_EXCEEDED "json text or perl structure exceeds maximum nesting level (max_depth set too low?)"
65 81
66#ifdef USE_ITHREADS 82#ifdef USE_ITHREADS
67# define JSON_SLOW 1
68# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1)) 83# define JSON_STASH (expect_true (json_stash) ? json_stash : gv_stashpv ("JSON::XS", 1))
84# define BOOL_STASH (expect_true (bool_stash) ? bool_stash : gv_stashpv ("Types::Serialiser::Boolean", 1))
85# define GET_BOOL(value) (expect_true (bool_ ## value) ? bool_ ## value : get_bool ("Types::Serialiser::" # value))
69#else 86#else
70# define JSON_SLOW 0
71# define JSON_STASH json_stash 87# define JSON_STASH json_stash
88# define BOOL_STASH bool_stash
89# define GET_BOOL(value) bool_ ## value
72#endif 90#endif
73 91
74static HV *json_stash, *json_boolean_stash; // JSON::XS:: 92// the amount of HEs to allocate on the stack, when sorting keys
75static SV *json_true, *json_false; 93#define STACK_HES 64
94
95static HV *json_stash, *bool_stash; // JSON::XS::, Types::Serialiser::Boolean::
96static SV *bool_false, *bool_true;
97static SV *sv_json;
76 98
77enum { 99enum {
78 INCR_M_WS = 0, // initial whitespace skipping, must be 0 100 INCR_M_WS = 0, // initial whitespace skipping, must be 0
101 INCR_M_TFN, // inside true/false/null
102 INCR_M_NUM, // inside number
79 INCR_M_STR, // inside string 103 INCR_M_STR, // inside string
80 INCR_M_BS, // inside backslash 104 INCR_M_BS, // inside backslash
105 INCR_M_C0, // inside comment in initial whitespace sequence
106 INCR_M_C1, // inside comment in other places
81 INCR_M_JSON // outside anything, count nesting 107 INCR_M_JSON // outside anything, count nesting
82}; 108};
83 109
84#define INCR_DONE(json) ((json)->incr_nest <= 0 && (json)->incr_mode == INCR_M_JSON) 110#define INCR_DONE(json) ((json)->incr_nest <= 0 && (json)->incr_mode == INCR_M_JSON)
85 111
94 // for the incremental parser 120 // for the incremental parser
95 SV *incr_text; // the source text so far 121 SV *incr_text; // the source text so far
96 STRLEN incr_pos; // the current offset into the text 122 STRLEN incr_pos; // the current offset into the text
97 int incr_nest; // {[]}-nesting level 123 int incr_nest; // {[]}-nesting level
98 unsigned char incr_mode; 124 unsigned char incr_mode;
125
126 SV *v_false, *v_true;
99} JSON; 127} JSON;
100 128
101INLINE void 129INLINE void
102json_init (JSON *json) 130json_init (JSON *json)
103{ 131{
104 Zero (json, 1, JSON); 132 static const JSON init = { F_ALLOW_NONREF, 512 };
105 json->max_depth = 512; 133
134 *json = init;
106} 135}
107 136
108///////////////////////////////////////////////////////////////////////////// 137/////////////////////////////////////////////////////////////////////////////
109// utility functions 138// utility functions
110 139
132 SvPV_renew (sv, SvCUR (sv) + 1); 161 SvPV_renew (sv, SvCUR (sv) + 1);
133#endif 162#endif
134 } 163 }
135} 164}
136 165
166/* adds two STRLENs together, slow, and with paranoia */
167STRLEN
168strlen_sum (STRLEN l1, STRLEN l2)
169{
170 size_t sum = l1 + l2;
171
172 if (sum < (size_t)l2 || sum != (size_t)(STRLEN)sum)
173 croak ("JSON::XS: string size overflow");
174
175 return sum;
176}
177
178/* similar to SvGROW, but somewhat safer and guarantees exponential realloc strategy */
179static char *
180json_sv_grow (SV *sv, size_t len1, size_t len2)
181{
182 len1 = strlen_sum (len1, len2);
183 len1 = strlen_sum (len1, len1 >> 1);
184
185 if (len1 > 4096 - 24)
186 len1 = (len1 | 4095) - 24;
187
188 return SvGROW (sv, len1);
189}
190
137// decode an utf-8 character and return it, or (UV)-1 in 191// decode a utf-8 character and return it, or (UV)-1 in
138// case of an error. 192// case of an error.
139// we special-case "safe" characters from U+80 .. U+7FF, 193// we special-case "safe" characters from U+80 .. U+7FF,
140// but use the very good perl function to parse anything else. 194// but use the very good perl function to parse anything else.
141// note that we never call this function for a ascii codepoints 195// note that we never call this function for a ascii codepoints
142INLINE UV 196INLINE UV
175 *s++ = 0x80 | ( ch & 0x3f); 229 *s++ = 0x80 | ( ch & 0x3f);
176 230
177 return s; 231 return s;
178} 232}
179 233
234// convert offset pointer to character index, sv must be string
235static STRLEN
236ptr_to_index (SV *sv, char *offset)
237{
238 return SvUTF8 (sv)
239 ? utf8_distance (offset, SvPVX (sv))
240 : offset - SvPVX (sv);
241}
242
243/////////////////////////////////////////////////////////////////////////////
244// fp hell
245
246// scan a group of digits, and a trailing exponent
247static void
248json_atof_scan1 (const char *s, NV *accum, int *expo, int postdp, int maxdepth)
249{
250 UV uaccum = 0;
251 int eaccum = 0;
252
253 // if we recurse too deep, skip all remaining digits
254 // to avoid a stack overflow attack
255 if (expect_false (--maxdepth <= 0))
256 while (((U8)*s - '0') < 10)
257 ++s;
258
259 for (;;)
260 {
261 U8 dig = (U8)*s - '0';
262
263 if (expect_false (dig >= 10))
264 {
265 if (dig == (U8)((U8)'.' - (U8)'0'))
266 {
267 ++s;
268 json_atof_scan1 (s, accum, expo, 1, maxdepth);
269 }
270 else if ((dig | ' ') == 'e' - '0')
271 {
272 int exp2 = 0;
273 int neg = 0;
274
275 ++s;
276
277 if (*s == '-')
278 {
279 ++s;
280 neg = 1;
281 }
282 else if (*s == '+')
283 ++s;
284
285 while ((dig = (U8)*s - '0') < 10)
286 exp2 = exp2 * 10 + *s++ - '0';
287
288 *expo += neg ? -exp2 : exp2;
289 }
290
291 break;
292 }
293
294 ++s;
295
296 uaccum = uaccum * 10 + dig;
297 ++eaccum;
298
299 // if we have too many digits, then recurse for more
300 // we actually do this for rather few digits
301 if (uaccum >= (UV_MAX - 9) / 10)
302 {
303 if (postdp) *expo -= eaccum;
304 json_atof_scan1 (s, accum, expo, postdp, maxdepth);
305 if (postdp) *expo += eaccum;
306
307 break;
308 }
309 }
310
311 // this relies greatly on the quality of the pow ()
312 // implementation of the platform, but a good
313 // implementation is hard to beat.
314 // (IEEE 754 conformant ones are required to be exact)
315 if (postdp) *expo -= eaccum;
316 *accum += uaccum * Perl_pow (10., *expo);
317 *expo += eaccum;
318}
319
320static NV
321json_atof (const char *s)
322{
323 NV accum = 0.;
324 int expo = 0;
325 int neg = 0;
326
327 if (*s == '-')
328 {
329 ++s;
330 neg = 1;
331 }
332
333 // a recursion depth of ten gives us >>500 bits
334 json_atof_scan1 (s, &accum, &expo, 0, 10);
335
336 return neg ? -accum : accum;
337}
338
339// target of scalar reference is bool? -1 == nope, 0 == false, 1 == true
340static int
341ref_bool_type (SV *sv)
342{
343 svtype svt = SvTYPE (sv);
344
345 if (svt < SVt_PVAV)
346 {
347 STRLEN len = 0;
348 char *pv = svt ? SvPV (sv, len) : 0;
349
350 if (len == 1)
351 if (*pv == '1')
352 return 1;
353 else if (*pv == '0')
354 return 0;
355 }
356
357 return -1;
358}
359
360// returns whether scalar is not a reference in the sense of allow_nonref
361static int
362json_nonref (SV *scalar)
363{
364 if (!SvROK (scalar))
365 return 1;
366
367 scalar = SvRV (scalar);
368
369 if (SvTYPE (scalar) >= SVt_PVMG)
370 {
371 if (SvSTASH (scalar) == bool_stash)
372 return 1;
373
374 if (!SvOBJECT (scalar) && ref_bool_type (scalar) >= 0)
375 return 1;
376 }
377
378 return 0;
379}
380
180///////////////////////////////////////////////////////////////////////////// 381/////////////////////////////////////////////////////////////////////////////
181// encoder 382// encoder
182 383
183// structure used for encoding JSON 384// structure used for encoding JSON
184typedef struct 385typedef struct
192} enc_t; 393} enc_t;
193 394
194INLINE void 395INLINE void
195need (enc_t *enc, STRLEN len) 396need (enc_t *enc, STRLEN len)
196{ 397{
197 if (expect_false (enc->cur + len >= enc->end)) 398 if (expect_false ((uintptr_t)(enc->end - enc->cur) < len))
198 { 399 {
199 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 400 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
200 SvGROW (enc->sv, cur + len + 1); 401 char *buf = json_sv_grow (enc->sv, cur, len);
201 enc->cur = SvPVX (enc->sv) + cur; 402 enc->cur = buf + cur;
202 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 403 enc->end = buf + SvLEN (enc->sv) - 1;
203 } 404 }
204} 405}
205 406
206INLINE void 407INLINE void
207encode_ch (enc_t *enc, char ch) 408encode_ch (enc_t *enc, char ch)
223 424
224 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case 425 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
225 { 426 {
226 if (expect_false (ch == '"')) // but with slow exceptions 427 if (expect_false (ch == '"')) // but with slow exceptions
227 { 428 {
228 need (enc, len += 1); 429 need (enc, len + 1);
229 *enc->cur++ = '\\'; 430 *enc->cur++ = '\\';
230 *enc->cur++ = '"'; 431 *enc->cur++ = '"';
231 } 432 }
232 else if (expect_false (ch == '\\')) 433 else if (expect_false (ch == '\\'))
233 { 434 {
234 need (enc, len += 1); 435 need (enc, len + 1);
235 *enc->cur++ = '\\'; 436 *enc->cur++ = '\\';
236 *enc->cur++ = '\\'; 437 *enc->cur++ = '\\';
237 } 438 }
238 else 439 else
239 *enc->cur++ = ch; 440 *enc->cur++ = ch;
242 } 443 }
243 else 444 else
244 { 445 {
245 switch (ch) 446 switch (ch)
246 { 447 {
247 case '\010': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'b'; ++str; break; 448 case '\010': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 'b'; ++str; break;
248 case '\011': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 't'; ++str; break; 449 case '\011': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 't'; ++str; break;
249 case '\012': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'n'; ++str; break; 450 case '\012': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 'n'; ++str; break;
250 case '\014': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'f'; ++str; break; 451 case '\014': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 'f'; ++str; break;
251 case '\015': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'r'; ++str; break; 452 case '\015': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 'r'; ++str; break;
252 453
253 default: 454 default:
254 { 455 {
255 STRLEN clen; 456 STRLEN clen;
256 UV uch; 457 UV uch;
272 if (uch >= 0x10000UL) 473 if (uch >= 0x10000UL)
273 { 474 {
274 if (uch >= 0x110000UL) 475 if (uch >= 0x110000UL)
275 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); 476 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
276 477
277 need (enc, len += 11); 478 need (enc, len + 11);
278 sprintf (enc->cur, "\\u%04x\\u%04x", 479 sprintf (enc->cur, "\\u%04x\\u%04x",
279 (int)((uch - 0x10000) / 0x400 + 0xD800), 480 (int)((uch - 0x10000) / 0x400 + 0xD800),
280 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 481 (int)((uch - 0x10000) % 0x400 + 0xDC00));
281 enc->cur += 12; 482 enc->cur += 12;
282 } 483 }
283 else 484 else
284 { 485 {
285 need (enc, len += 5); 486 need (enc, len + 5);
286 *enc->cur++ = '\\'; 487 *enc->cur++ = '\\';
287 *enc->cur++ = 'u'; 488 *enc->cur++ = 'u';
288 *enc->cur++ = PL_hexdigit [ uch >> 12 ]; 489 *enc->cur++ = PL_hexdigit [ uch >> 12 ];
289 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15]; 490 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
290 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15]; 491 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
298 *enc->cur++ = uch; 499 *enc->cur++ = uch;
299 str += clen; 500 str += clen;
300 } 501 }
301 else if (is_utf8) 502 else if (is_utf8)
302 { 503 {
303 need (enc, len += clen); 504 need (enc, len + clen);
304 do 505 do
305 { 506 {
306 *enc->cur++ = *str++; 507 *enc->cur++ = *str++;
307 } 508 }
308 while (--clen); 509 while (--clen);
309 } 510 }
310 else 511 else
311 { 512 {
312 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed 513 need (enc, len + UTF8_MAXBYTES - 1); // never more than 11 bytes needed
313 enc->cur = encode_utf8 (enc->cur, uch); 514 enc->cur = encode_utf8 (enc->cur, uch);
314 ++str; 515 ++str;
315 } 516 }
316 } 517 }
317 } 518 }
371 572
372 if (enc->indent >= enc->json.max_depth) 573 if (enc->indent >= enc->json.max_depth)
373 croak (ERR_NESTING_EXCEEDED); 574 croak (ERR_NESTING_EXCEEDED);
374 575
375 encode_ch (enc, '['); 576 encode_ch (enc, '[');
376 577
377 if (len >= 0) 578 if (len >= 0)
378 { 579 {
379 encode_nl (enc); ++enc->indent; 580 encode_nl (enc); ++enc->indent;
380 581
381 for (i = 0; i <= len; ++i) 582 for (i = 0; i <= len; ++i)
393 encode_comma (enc); 594 encode_comma (enc);
394 } 595 }
395 596
396 encode_nl (enc); --enc->indent; encode_indent (enc); 597 encode_nl (enc); --enc->indent; encode_indent (enc);
397 } 598 }
398 599
399 encode_ch (enc, ']'); 600 encode_ch (enc, ']');
400} 601}
401 602
402static void 603static void
403encode_hk (enc_t *enc, HE *he) 604encode_hk (enc_t *enc, HE *he)
407 if (HeKLEN (he) == HEf_SVKEY) 608 if (HeKLEN (he) == HEf_SVKEY)
408 { 609 {
409 SV *sv = HeSVKEY (he); 610 SV *sv = HeSVKEY (he);
410 STRLEN len; 611 STRLEN len;
411 char *str; 612 char *str;
412 613
413 SvGETMAGIC (sv); 614 SvGETMAGIC (sv);
414 str = SvPV (sv, len); 615 str = SvPV (sv, len);
415 616
416 encode_str (enc, str, len, SvUTF8 (sv)); 617 encode_str (enc, str, len, SvUTF8 (sv));
417 } 618 }
462 663
463 // for canonical output we have to sort by keys first 664 // for canonical output we have to sort by keys first
464 // actually, this is mostly due to the stupid so-called 665 // actually, this is mostly due to the stupid so-called
465 // security workaround added somewhere in 5.8.x 666 // security workaround added somewhere in 5.8.x
466 // that randomises hash orderings 667 // that randomises hash orderings
467 if (enc->json.flags & F_CANONICAL) 668 if (enc->json.flags & F_CANONICAL && !SvRMAGICAL (hv))
468 { 669 {
469 int count = hv_iterinit (hv); 670 int count = hv_iterinit (hv);
470 671
471 if (SvMAGICAL (hv)) 672 if (SvMAGICAL (hv))
472 { 673 {
483 } 684 }
484 685
485 if (count) 686 if (count)
486 { 687 {
487 int i, fast = 1; 688 int i, fast = 1;
488#if defined(__BORLANDC__) || defined(_MSC_VER) 689 HE *hes_stack [STACK_HES];
489 HE **hes = _alloca (count * sizeof (HE)); 690 HE **hes = hes_stack;
490#else 691
491 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode 692 // allocate larger arrays on the heap
492#endif 693 if (count > STACK_HES)
694 {
695 SV *sv = sv_2mortal (NEWSV (0, count * sizeof (*hes)));
696 hes = (HE **)SvPVX (sv);
697 }
493 698
494 i = 0; 699 i = 0;
495 while ((he = hv_iternext (hv))) 700 while ((he = hv_iternext (hv)))
496 { 701 {
497 hes [i++] = he; 702 hes [i++] = he;
566// encode objects, arrays and special \0=false and \1=true values. 771// encode objects, arrays and special \0=false and \1=true values.
567static void 772static void
568encode_rv (enc_t *enc, SV *sv) 773encode_rv (enc_t *enc, SV *sv)
569{ 774{
570 svtype svt; 775 svtype svt;
776 GV *method;
571 777
572 SvGETMAGIC (sv); 778 SvGETMAGIC (sv);
573 svt = SvTYPE (sv); 779 svt = SvTYPE (sv);
574 780
575 if (expect_false (SvOBJECT (sv))) 781 if (expect_false (SvOBJECT (sv)))
576 { 782 {
577 HV *stash = !JSON_SLOW || json_boolean_stash 783 HV *stash = SvSTASH (sv);
578 ? json_boolean_stash
579 : gv_stashpv ("JSON::XS::Boolean", 1);
580 784
581 if (SvSTASH (sv) == stash) 785 if (stash == bool_stash)
582 { 786 {
583 if (SvIV (sv))
584 encode_str (enc, "true", 4, 0); 787 if (SvIV (sv)) encode_str (enc, "true" , 4, 0);
585 else
586 encode_str (enc, "false", 5, 0); 788 else encode_str (enc, "false", 5, 0);
587 } 789 }
790 else if ((enc->json.flags & F_ALLOW_TAGS) && (method = gv_fetchmethod_autoload (stash, "FREEZE", 0)))
791 {
792 int count;
793 dSP;
794
795 ENTER; SAVETMPS;
796 PUSHMARK (SP);
797 EXTEND (SP, 2);
798 // we re-bless the reference to get overload and other niceties right
799 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
800 PUSHs (sv_json);
801
802 PUTBACK;
803 count = call_sv ((SV *)GvCV (method), G_ARRAY);
804 SPAGAIN;
805
806 // catch this surprisingly common error
807 if (SvROK (TOPs) && SvRV (TOPs) == sv)
808 croak ("%s::FREEZE method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
809
810 encode_ch (enc, '(');
811 encode_ch (enc, '"');
812 encode_str (enc, HvNAME (stash), HvNAMELEN (stash), HvNAMEUTF8 (stash));
813 encode_ch (enc, '"');
814 encode_ch (enc, ')');
815 encode_ch (enc, '[');
816
817 if (count)
818 {
819 int i;
820
821 for (i = 0; i < count - 1; ++i)
822 {
823 encode_sv (enc, SP[i + 1 - count]);
824 encode_ch (enc, ',');
825 }
826
827 encode_sv (enc, TOPs);
828 SP -= count;
829 }
830
831 encode_ch (enc, ']');
832
833 FREETMPS; LEAVE;
834 }
835 else if ((enc->json.flags & F_CONV_BLESSED) && (method = gv_fetchmethod_autoload (stash, "TO_JSON", 0)))
836 {
837 dSP;
838
839 ENTER; SAVETMPS;
840 PUSHMARK (SP);
841 // we re-bless the reference to get overload and other niceties right
842 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
843
844 // calling with G_SCALAR ensures that we always get a 1 return value
845 PUTBACK;
846 call_sv ((SV *)GvCV (method), G_SCALAR);
847 SPAGAIN;
848
849 // catch this surprisingly common error
850 if (SvROK (TOPs) && SvRV (TOPs) == sv)
851 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
852
853 sv = POPs;
854 PUTBACK;
855
856 encode_sv (enc, sv);
857
858 FREETMPS; LEAVE;
859 }
860 else if (enc->json.flags & F_ALLOW_BLESSED)
861 encode_str (enc, "null", 4, 0);
588 else 862 else
589 {
590#if 0
591 if (0 && sv_derived_from (rv, "JSON::Literal"))
592 {
593 // not yet
594 }
595#endif
596 if (enc->json.flags & F_CONV_BLESSED)
597 {
598 // we re-bless the reference to get overload and other niceties right
599 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
600
601 if (to_json)
602 {
603 dSP;
604
605 ENTER; SAVETMPS; PUSHMARK (SP);
606 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
607
608 // calling with G_SCALAR ensures that we always get a 1 return value
609 PUTBACK;
610 call_sv ((SV *)GvCV (to_json), G_SCALAR);
611 SPAGAIN;
612
613 // catch this surprisingly common error
614 if (SvROK (TOPs) && SvRV (TOPs) == sv)
615 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
616
617 sv = POPs;
618 PUTBACK;
619
620 encode_sv (enc, sv);
621
622 FREETMPS; LEAVE;
623 }
624 else if (enc->json.flags & F_ALLOW_BLESSED)
625 encode_str (enc, "null", 4, 0);
626 else
627 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
628 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
629 }
630 else if (enc->json.flags & F_ALLOW_BLESSED)
631 encode_str (enc, "null", 4, 0);
632 else
633 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled", 863 croak ("encountered object '%s', but neither allow_blessed, convert_blessed nor allow_tags settings are enabled (or TO_JSON/FREEZE method missing)",
634 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 864 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
635 }
636 } 865 }
637 else if (svt == SVt_PVHV) 866 else if (svt == SVt_PVHV)
638 encode_hv (enc, (HV *)sv); 867 encode_hv (enc, (HV *)sv);
639 else if (svt == SVt_PVAV) 868 else if (svt == SVt_PVAV)
640 encode_av (enc, (AV *)sv); 869 encode_av (enc, (AV *)sv);
641 else if (svt < SVt_PVAV) 870 else if (svt < SVt_PVAV)
642 { 871 {
643 STRLEN len = 0; 872 int bool_type = ref_bool_type (sv);
644 char *pv = svt ? SvPV (sv, len) : 0;
645 873
646 if (len == 1 && *pv == '1') 874 if (bool_type == 1)
647 encode_str (enc, "true", 4, 0); 875 encode_str (enc, "true", 4, 0);
648 else if (len == 1 && *pv == '0') 876 else if (bool_type == 0)
649 encode_str (enc, "false", 5, 0); 877 encode_str (enc, "false", 5, 0);
650 else if (enc->json.flags & F_ALLOW_UNKNOWN) 878 else if (enc->json.flags & F_ALLOW_UNKNOWN)
651 encode_str (enc, "null", 4, 0); 879 encode_str (enc, "null", 4, 0);
652 else 880 else
653 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", 881 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
716 } 944 }
717 else 945 else
718 { 946 {
719 // large integer, use the (rather slow) snprintf way. 947 // large integer, use the (rather slow) snprintf way.
720 need (enc, IVUV_MAXCHARS); 948 need (enc, IVUV_MAXCHARS);
721 enc->cur += 949 enc->cur +=
722 SvIsUV(sv) 950 SvIsUV(sv)
723 ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv)) 951 ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))
724 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv)); 952 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
725 } 953 }
726 } 954 }
727 else if (SvROK (sv)) 955 else if (SvROK (sv))
728 encode_rv (enc, SvRV (sv)); 956 encode_rv (enc, SvRV (sv));
729 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN) 957 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN)
730 encode_str (enc, "null", 4, 0); 958 encode_str (enc, "null", 4, 0);
731 else 959 else
732 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 960 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, check your input data",
733 SvPV_nolen (sv), SvFLAGS (sv)); 961 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
734} 962}
735 963
736static SV * 964static SV *
737encode_json (SV *scalar, JSON *json) 965encode_json (SV *scalar, JSON *json)
738{ 966{
739 enc_t enc; 967 enc_t enc;
740 968
741 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar)) 969 if (!(json->flags & F_ALLOW_NONREF) && json_nonref (scalar))
742 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 970 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
743 971
744 enc.json = *json; 972 enc.json = *json;
745 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 973 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
746 enc.cur = SvPVX (enc.sv); 974 enc.cur = SvPVX (enc.sv);
750 : enc.json.flags & F_LATIN1 ? 0x000100UL 978 : enc.json.flags & F_LATIN1 ? 0x000100UL
751 : 0x110000UL; 979 : 0x110000UL;
752 980
753 SvPOK_only (enc.sv); 981 SvPOK_only (enc.sv);
754 encode_sv (&enc, scalar); 982 encode_sv (&enc, scalar);
983 encode_nl (&enc);
755 984
756 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 985 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
757 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 986 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
758 987
759 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8))) 988 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
939 else if (expect_true (ch >= 0x20 && ch < 0x80)) 1168 else if (expect_true (ch >= 0x20 && ch < 0x80))
940 *cur++ = ch; 1169 *cur++ = ch;
941 else if (ch >= 0x80) 1170 else if (ch >= 0x80)
942 { 1171 {
943 STRLEN clen; 1172 STRLEN clen;
944 UV uch;
945 1173
946 --dec_cur; 1174 --dec_cur;
947 1175
948 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen); 1176 decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
949 if (clen == (STRLEN)-1) 1177 if (clen == (STRLEN)-1)
950 ERR ("malformed UTF-8 character in JSON string"); 1178 ERR ("malformed UTF-8 character in JSON string");
951 1179
952 do 1180 do
953 *cur++ = *dec_cur++; 1181 *cur++ = *dec_cur++;
954 while (--clen); 1182 while (--clen);
955 1183
956 utf8 = 1; 1184 utf8 = 1;
957 } 1185 }
1186 else if (ch == '\t' && dec->json.flags & F_RELAXED)
1187 *cur++ = ch;
958 else 1188 else
959 { 1189 {
960 --dec_cur; 1190 --dec_cur;
961 1191
962 if (!ch) 1192 if (!ch)
970 { 1200 {
971 STRLEN len = cur - buf; 1201 STRLEN len = cur - buf;
972 1202
973 if (sv) 1203 if (sv)
974 { 1204 {
975 SvGROW (sv, SvCUR (sv) + len + 1); 1205 STRLEN cur = SvCUR (sv);
1206
1207 if (SvLEN (sv) - cur <= len)
1208 json_sv_grow (sv, cur, len);
1209
976 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 1210 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
977 SvCUR_set (sv, SvCUR (sv) + len); 1211 SvCUR_set (sv, SvCUR (sv) + len);
978 } 1212 }
979 else 1213 else
980 sv = newSVpvn (buf, len); 1214 sv = newSVpvn (buf, len);
1071 1305
1072 // special case the rather common 1..5-digit-int case 1306 // special case the rather common 1..5-digit-int case
1073 if (*start == '-') 1307 if (*start == '-')
1074 switch (len) 1308 switch (len)
1075 { 1309 {
1076 case 2: return newSViv (-( start [1] - '0' * 1)); 1310 case 2: return newSViv (-(IV)( start [1] - '0' * 1));
1077 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); 1311 case 3: return newSViv (-(IV)( start [1] * 10 + start [2] - '0' * 11));
1078 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); 1312 case 4: return newSViv (-(IV)( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
1079 case 5: return newSViv (-( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); 1313 case 5: return newSViv (-(IV)( 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)); 1314 case 6: return newSViv (-(IV)(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
1081 } 1315 }
1082 else 1316 else
1083 switch (len) 1317 switch (len)
1084 { 1318 {
1085 case 1: return newSViv ( start [0] - '0' * 1); 1319 case 1: return newSViv ( start [0] - '0' * 1);
1086 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); 1320 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
1087 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); 1321 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
1088 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); 1322 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); 1323 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
1090 } 1324 }
1091 1325
1092 { 1326 {
1093 UV uv; 1327 UV uv;
1094 int numtype = grok_number (start, len, &uv); 1328 int numtype = grok_number (start, len, &uv);
1103 } 1337 }
1104 1338
1105 len -= *start == '-' ? 1 : 0; 1339 len -= *start == '-' ? 1 : 0;
1106 1340
1107 // does not fit into IV or UV, try NV 1341 // does not fit into IV or UV, try NV
1108 if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len) 1342 if (len <= NV_DIG)
1109 #if defined (LDBL_DIG)
1110 || (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len)
1111 #endif
1112 )
1113 // fits into NV without loss of precision 1343 // fits into NV without loss of precision
1114 return newSVnv (Atof (start)); 1344 return newSVnv (json_atof (start));
1115 1345
1116 // everything else fails, convert it to a string 1346 // everything else fails, convert it to a string
1117 return newSVpvn (start, dec->cur - start); 1347 return newSVpvn (start, dec->cur - start);
1118 } 1348 }
1119 1349
1120 // loss of precision here 1350 // loss of precision here
1121 return newSVnv (Atof (start)); 1351 return newSVnv (json_atof (start));
1122 1352
1123fail: 1353fail:
1124 return 0; 1354 return 0;
1125} 1355}
1126 1356
1150 if (*dec->cur == ']') 1380 if (*dec->cur == ']')
1151 { 1381 {
1152 ++dec->cur; 1382 ++dec->cur;
1153 break; 1383 break;
1154 } 1384 }
1155 1385
1156 if (*dec->cur != ',') 1386 if (*dec->cur != ',')
1157 ERR (", or ] expected while parsing array"); 1387 ERR (", or ] expected while parsing array");
1158 1388
1159 ++dec->cur; 1389 ++dec->cur;
1160 1390
1284 1514
1285 hv_iterinit (hv); 1515 hv_iterinit (hv);
1286 he = hv_iternext (hv); 1516 he = hv_iternext (hv);
1287 hv_iterinit (hv); 1517 hv_iterinit (hv);
1288 1518
1289 // the next line creates a mortal sv each time its called. 1519 // the next line creates a mortal sv each time it's called.
1290 // might want to optimise this for common cases. 1520 // might want to optimise this for common cases.
1291 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0); 1521 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1292 1522
1293 if (cb) 1523 if (cb)
1294 { 1524 {
1295 dSP; 1525 dSP;
1296 int count; 1526 int count;
1297 1527
1298 ENTER; SAVETMPS; PUSHMARK (SP); 1528 ENTER; SAVETMPS;
1529 PUSHMARK (SP);
1299 XPUSHs (HeVAL (he)); 1530 XPUSHs (HeVAL (he));
1531 sv_2mortal (sv);
1300 1532
1301 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN; 1533 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1302 1534
1303 if (count == 1) 1535 if (count == 1)
1304 { 1536 {
1305 sv = newSVsv (POPs); 1537 sv = newSVsv (POPs);
1306 FREETMPS; LEAVE; 1538 FREETMPS; LEAVE;
1307 return sv; 1539 return sv;
1308 } 1540 }
1541 else if (count)
1542 croak ("filter_json_single_key_object callbacks must not return more than one scalar");
1309 1543
1544 SvREFCNT_inc (sv);
1310 FREETMPS; LEAVE; 1545 FREETMPS; LEAVE;
1311 } 1546 }
1312 } 1547 }
1313 1548
1314 if (dec->json.cb_object) 1549 if (dec->json.cb_object)
1315 { 1550 {
1316 dSP; 1551 dSP;
1317 int count; 1552 int count;
1318 1553
1319 ENTER; SAVETMPS; PUSHMARK (SP); 1554 ENTER; SAVETMPS;
1555 PUSHMARK (SP);
1320 XPUSHs (sv_2mortal (sv)); 1556 XPUSHs (sv_2mortal (sv));
1321 1557
1322 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN; 1558 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1323 1559
1324 if (count == 1) 1560 if (count == 1)
1325 { 1561 {
1326 sv = newSVsv (POPs); 1562 sv = newSVsv (POPs);
1327 FREETMPS; LEAVE; 1563 FREETMPS; LEAVE;
1328 return sv; 1564 return sv;
1329 } 1565 }
1566 else if (count)
1567 croak ("filter_json_object callbacks must not return more than one scalar");
1330 1568
1331 SvREFCNT_inc (sv); 1569 SvREFCNT_inc (sv);
1332 FREETMPS; LEAVE; 1570 FREETMPS; LEAVE;
1333 } 1571 }
1334 } 1572 }
1340 DEC_DEC_DEPTH; 1578 DEC_DEC_DEPTH;
1341 return 0; 1579 return 0;
1342} 1580}
1343 1581
1344static SV * 1582static SV *
1583decode_tag (dec_t *dec)
1584{
1585 SV *tag = 0;
1586 SV *val = 0;
1587
1588 if (!(dec->json.flags & F_ALLOW_TAGS))
1589 ERR ("malformed JSON string, neither array, object, number, string or atom");
1590
1591 ++dec->cur;
1592
1593 decode_ws (dec);
1594
1595 tag = decode_sv (dec);
1596 if (!tag)
1597 goto fail;
1598
1599 if (!SvPOK (tag))
1600 ERR ("malformed JSON string, (tag) must be a string");
1601
1602 decode_ws (dec);
1603
1604 if (*dec->cur != ')')
1605 ERR (") expected after tag");
1606
1607 ++dec->cur;
1608
1609 decode_ws (dec);
1610
1611 val = decode_sv (dec);
1612 if (!val)
1613 goto fail;
1614
1615 if (!SvROK (val) || SvTYPE (SvRV (val)) != SVt_PVAV)
1616 ERR ("malformed JSON string, tag value must be an array");
1617
1618 {
1619 AV *av = (AV *)SvRV (val);
1620 int i, len = av_len (av) + 1;
1621 HV *stash = gv_stashsv (tag, 0);
1622 SV *sv;
1623
1624 if (!stash)
1625 ERR ("cannot decode perl-object (package does not exist)");
1626
1627 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
1628
1629 if (!method)
1630 ERR ("cannot decode perl-object (package does not have a THAW method)");
1631
1632 dSP;
1633
1634 ENTER; SAVETMPS;
1635 PUSHMARK (SP);
1636 EXTEND (SP, len + 2);
1637 // we re-bless the reference to get overload and other niceties right
1638 PUSHs (tag);
1639 PUSHs (sv_json);
1640
1641 for (i = 0; i < len; ++i)
1642 PUSHs (*av_fetch (av, i, 1));
1643
1644 PUTBACK;
1645 call_sv ((SV *)GvCV (method), G_SCALAR);
1646 SPAGAIN;
1647
1648 SvREFCNT_dec (tag);
1649 SvREFCNT_dec (val);
1650 sv = SvREFCNT_inc (POPs);
1651
1652 PUTBACK;
1653
1654 FREETMPS; LEAVE;
1655
1656 return sv;
1657 }
1658
1659fail:
1660 SvREFCNT_dec (tag);
1661 SvREFCNT_dec (val);
1662 return 0;
1663}
1664
1665static SV *
1345decode_sv (dec_t *dec) 1666decode_sv (dec_t *dec)
1346{ 1667{
1347 // the beauty of JSON: you need exactly one character lookahead 1668 // the beauty of JSON: you need exactly one character lookahead
1348 // to parse everything. 1669 // to parse everything.
1349 switch (*dec->cur) 1670 switch (*dec->cur)
1350 { 1671 {
1351 case '"': ++dec->cur; return decode_str (dec); 1672 case '"': ++dec->cur; return decode_str (dec);
1352 case '[': ++dec->cur; return decode_av (dec); 1673 case '[': ++dec->cur; return decode_av (dec);
1353 case '{': ++dec->cur; return decode_hv (dec); 1674 case '{': ++dec->cur; return decode_hv (dec);
1675 case '(': return decode_tag (dec);
1354 1676
1355 case '-': 1677 case '-':
1356 case '0': case '1': case '2': case '3': case '4': 1678 case '0': case '1': case '2': case '3': case '4':
1357 case '5': case '6': case '7': case '8': case '9': 1679 case '5': case '6': case '7': case '8': case '9':
1358 return decode_num (dec); 1680 return decode_num (dec);
1359 1681
1682 case 'f':
1683 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1684 {
1685 dec->cur += 5;
1686
1687 if (expect_false (!dec->json.v_false))
1688 dec->json.v_false = GET_BOOL (false);
1689
1690 return newSVsv (dec->json.v_false);
1691 }
1692 else
1693 ERR ("'false' expected");
1694
1695 break;
1696
1360 case 't': 1697 case 't':
1361 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1698 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1362 { 1699 {
1363 dec->cur += 4; 1700 dec->cur += 4;
1364#if JSON_SLOW 1701
1365 json_true = get_bool ("JSON::XS::true"); 1702 if (expect_false (!dec->json.v_true))
1366#endif 1703 dec->json.v_true = GET_BOOL (true);
1704
1367 return newSVsv (json_true); 1705 return newSVsv (dec->json.v_true);
1368 } 1706 }
1369 else 1707 else
1370 ERR ("'true' expected"); 1708 ERR ("'true' expected");
1371
1372 break;
1373
1374 case 'f':
1375 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1376 {
1377 dec->cur += 5;
1378#if JSON_SLOW
1379 json_false = get_bool ("JSON::XS::false");
1380#endif
1381 return newSVsv (json_false);
1382 }
1383 else
1384 ERR ("'false' expected");
1385 1709
1386 break; 1710 break;
1387 1711
1388 case 'n': 1712 case 'n':
1389 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4)) 1713 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4))
1395 ERR ("'null' expected"); 1719 ERR ("'null' expected");
1396 1720
1397 break; 1721 break;
1398 1722
1399 default: 1723 default:
1400 ERR ("malformed JSON string, neither array, object, number, string or atom"); 1724 ERR ("malformed JSON string, neither tag, array, object, number, string or atom");
1401 break; 1725 break;
1402 } 1726 }
1403 1727
1404fail: 1728fail:
1405 return 0; 1729 return 0;
1407 1731
1408static SV * 1732static SV *
1409decode_json (SV *string, JSON *json, STRLEN *offset_return) 1733decode_json (SV *string, JSON *json, STRLEN *offset_return)
1410{ 1734{
1411 dec_t dec; 1735 dec_t dec;
1412 STRLEN offset;
1413 SV *sv; 1736 SV *sv;
1414 1737
1738 /* work around bugs in 5.10 where manipulating magic values
1739 * makes perl ignore the magic in subsequent accesses.
1740 * also make a copy of non-PV values, to get them into a clean
1741 * state (SvPV should do that, but it's buggy, see below).
1742 *
1743 * SvIsCOW_shared_hash works around a bug in perl (possibly 5.16),
1744 * as reported by Reini Urban.
1745 */
1415 SvGETMAGIC (string); 1746 /*SvGETMAGIC (string);*/
1747 if (SvMAGICAL (string) || !SvPOK (string) || SvIsCOW_shared_hash (string))
1748 string = sv_2mortal (newSVsv (string));
1749
1416 SvUPGRADE (string, SVt_PV); 1750 SvUPGRADE (string, SVt_PV);
1417 1751
1418 /* work around a bug in perl 5.10, which causes SvCUR to fail an 1752 /* work around a bug in perl 5.10, which causes SvCUR to fail an
1419 * assertion with -DDEBUGGING, although SvCUR is documented to 1753 * assertion with -DDEBUGGING, although SvCUR is documented to
1420 * return the xpv_cur field which certainly exists after upgrading. 1754 * return the xpv_cur field which certainly exists after upgrading.
1423 * and hope for the best. 1757 * and hope for the best.
1424 * Damnit, SvPV_nolen still trips over yet another assertion. This 1758 * Damnit, SvPV_nolen still trips over yet another assertion. This
1425 * assertion business is seriously broken, try yet another workaround 1759 * assertion business is seriously broken, try yet another workaround
1426 * for the broken -DDEBUGGING. 1760 * for the broken -DDEBUGGING.
1427 */ 1761 */
1762 {
1428#ifdef DEBUGGING 1763#ifdef DEBUGGING
1429 offset = SvOK (string) ? sv_len (string) : 0; 1764 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
1430#else 1765#else
1431 offset = SvCUR (string); 1766 STRLEN offset = SvCUR (string);
1432#endif 1767#endif
1433 1768
1434 if (offset > json->max_size && json->max_size) 1769 if (offset > json->max_size && json->max_size)
1435 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", 1770 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1436 (unsigned long)SvCUR (string), (unsigned long)json->max_size); 1771 (unsigned long)SvCUR (string), (unsigned long)json->max_size);
1772 }
1437 1773
1438 if (json->flags & F_UTF8) 1774 if (DECODE_WANTS_OCTETS (json))
1439 sv_utf8_downgrade (string, 0); 1775 sv_utf8_downgrade (string, 0);
1440 else 1776 else
1441 sv_utf8_upgrade (string); 1777 sv_utf8_upgrade (string);
1442 1778
1443 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1779 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
1454 *dec.end = 0; // this should basically be a nop, too, but make sure it's there 1790 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1455 1791
1456 decode_ws (&dec); 1792 decode_ws (&dec);
1457 sv = decode_sv (&dec); 1793 sv = decode_sv (&dec);
1458 1794
1459 if (!(offset_return || !sv)) 1795 if (offset_return)
1796 *offset_return = dec.cur - SvPVX (string);
1797 else if (sv)
1460 { 1798 {
1461 // check for trailing garbage 1799 // check for trailing garbage
1462 decode_ws (&dec); 1800 decode_ws (&dec);
1463 1801
1464 if (*dec.cur) 1802 if (dec.cur != dec.end)
1465 { 1803 {
1466 dec.err = "garbage after JSON object"; 1804 dec.err = "garbage after JSON object";
1467 SvREFCNT_dec (sv); 1805 SvREFCNT_dec (sv);
1468 sv = 0; 1806 sv = 0;
1469 } 1807 }
1470 }
1471
1472 if (offset_return || !sv)
1473 {
1474 offset = dec.json.flags & F_UTF8
1475 ? dec.cur - SvPVX (string)
1476 : utf8_distance (dec.cur, SvPVX (string));
1477
1478 if (offset_return)
1479 *offset_return = offset;
1480 } 1808 }
1481 1809
1482 if (!sv) 1810 if (!sv)
1483 { 1811 {
1484 SV *uni = sv_newmortal (); 1812 SV *uni = sv_newmortal ();
1490 SAVEVPTR (PL_curcop); 1818 SAVEVPTR (PL_curcop);
1491 PL_curcop = &cop; 1819 PL_curcop = &cop;
1492 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1820 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1493 LEAVE; 1821 LEAVE;
1494 1822
1495 croak ("%s, at character offset %d [\"%s\"]", 1823 croak ("%s, at character offset %d (before \"%s\")",
1496 dec.err, 1824 dec.err,
1497 (int)offset, 1825 (int)ptr_to_index (string, dec.cur),
1498 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1826 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1499 } 1827 }
1500 1828
1501 sv = sv_2mortal (sv); 1829 sv = sv_2mortal (sv);
1502 1830
1503 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1831 if (!(dec.json.flags & F_ALLOW_NONREF) && json_nonref (sv))
1504 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); 1832 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1505 1833
1506 return sv; 1834 return sv;
1507} 1835}
1508 1836
1512static void 1840static void
1513incr_parse (JSON *self) 1841incr_parse (JSON *self)
1514{ 1842{
1515 const char *p = SvPVX (self->incr_text) + self->incr_pos; 1843 const char *p = SvPVX (self->incr_text) + self->incr_pos;
1516 1844
1845 // the state machine here is a bit convoluted and could be simplified a lot
1846 // but this would make it slower, so...
1847
1517 for (;;) 1848 for (;;)
1518 { 1849 {
1519 //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
1520 switch (self->incr_mode) 1850 switch (self->incr_mode)
1521 { 1851 {
1852 // reached end of a scalar, see if we are inside a nested structure or not
1853 end_of_scalar:
1854 self->incr_mode = INCR_M_JSON;
1855
1856 if (self->incr_nest) // end of a scalar inside array, object or tag
1857 goto incr_m_json;
1858 else // end of scalar outside structure, json text ends here
1859 goto interrupt;
1860
1522 // only used for intiial whitespace skipping 1861 // only used for initial whitespace skipping
1523 case INCR_M_WS: 1862 case INCR_M_WS:
1524 for (;;) 1863 for (;;)
1525 { 1864 {
1526 if (*p > 0x20) 1865 if (*p > 0x20)
1527 { 1866 {
1867 if (*p == '#')
1868 {
1869 self->incr_mode = INCR_M_C0;
1870 goto incr_m_c;
1871 }
1872 else
1873 {
1528 self->incr_mode = INCR_M_JSON; 1874 self->incr_mode = INCR_M_JSON;
1529 goto incr_m_json; 1875 goto incr_m_json;
1876 }
1530 } 1877 }
1531 else if (!*p) 1878 else if (!*p)
1532 goto interrupt; 1879 goto interrupt;
1533 1880
1534 ++p; 1881 ++p;
1540 goto interrupt; 1887 goto interrupt;
1541 1888
1542 ++p; 1889 ++p;
1543 self->incr_mode = INCR_M_STR; 1890 self->incr_mode = INCR_M_STR;
1544 goto incr_m_str; 1891 goto incr_m_str;
1892
1893 // inside #-style comments
1894 case INCR_M_C0:
1895 case INCR_M_C1:
1896 incr_m_c:
1897 for (;;)
1898 {
1899 if (*p == '\n')
1900 {
1901 self->incr_mode = self->incr_mode == INCR_M_C0 ? INCR_M_WS : INCR_M_JSON;
1902 break;
1903 }
1904 else if (!*p)
1905 goto interrupt;
1906
1907 ++p;
1908 }
1909
1910 break;
1911
1912 // inside true/false/null
1913 case INCR_M_TFN:
1914 incr_m_tfn:
1915 for (;;)
1916 switch (*p++)
1917 {
1918 case 'r': case 'u': case 'e': // tRUE, falsE, nUll
1919 case 'a': case 'l': case 's': // fALSe, nuLL
1920 // allowed
1921 break;
1922
1923 default:
1924 --p;
1925 goto end_of_scalar;
1926 }
1927
1928 // inside a number
1929 case INCR_M_NUM:
1930 incr_m_num:
1931 for (;;)
1932 switch (*p++)
1933 {
1934 case 'e': case 'E': case '.': case '+':
1935 case '-':
1936 case '0': case '1': case '2': case '3': case '4':
1937 case '5': case '6': case '7': case '8': case '9':
1938 // allowed
1939 break;
1940
1941 default:
1942 --p;
1943 goto end_of_scalar;
1944 }
1545 1945
1546 // inside a string 1946 // inside a string
1547 case INCR_M_STR: 1947 case INCR_M_STR:
1548 incr_m_str: 1948 incr_m_str:
1549 for (;;) 1949 for (;;)
1550 { 1950 {
1551 if (*p == '"') 1951 if (*p == '"')
1552 { 1952 {
1553 ++p; 1953 ++p;
1554 self->incr_mode = INCR_M_JSON;
1555
1556 if (!self->incr_nest)
1557 goto interrupt;
1558
1559 goto incr_m_json; 1954 goto end_of_scalar;
1560 } 1955 }
1561 else if (*p == '\\') 1956 else if (*p == '\\')
1562 { 1957 {
1563 ++p; // "virtually" consumes character after \ 1958 ++p; // "virtually" consumes character after \
1564 1959
1594 --p; // do not eat the whitespace, let the next round do it 1989 --p; // do not eat the whitespace, let the next round do it
1595 goto interrupt; 1990 goto interrupt;
1596 } 1991 }
1597 break; 1992 break;
1598 1993
1994 // the following three blocks handle scalars. this makes the parser
1995 // more strict than required inside arrays or objects, and could
1996 // be moved to a special case on the toplevel (except strings)
1997 case 't':
1998 case 'f':
1999 case 'n':
2000 self->incr_mode = INCR_M_TFN;
2001 goto incr_m_tfn;
2002
2003 case '-':
2004 case '0': case '1': case '2': case '3': case '4':
2005 case '5': case '6': case '7': case '8': case '9':
2006 self->incr_mode = INCR_M_NUM;
2007 goto incr_m_num;
2008
1599 case '"': 2009 case '"':
1600 self->incr_mode = INCR_M_STR; 2010 self->incr_mode = INCR_M_STR;
1601 goto incr_m_str; 2011 goto incr_m_str;
1602 2012
1603 case '[': 2013 case '[':
1604 case '{': 2014 case '{':
2015 case '(':
1605 if (++self->incr_nest > self->max_depth) 2016 if (++self->incr_nest > self->max_depth)
1606 croak (ERR_NESTING_EXCEEDED); 2017 croak (ERR_NESTING_EXCEEDED);
1607 break; 2018 break;
1608 2019
1609 case ']': 2020 case ']':
1610 case '}': 2021 case '}':
1611 if (--self->incr_nest <= 0) 2022 if (--self->incr_nest <= 0)
1612 goto interrupt; 2023 goto interrupt;
2024 break;
2025
2026 case ')':
2027 --self->incr_nest;
2028 break;
2029
2030 case '#':
2031 self->incr_mode = INCR_M_C1;
2032 goto incr_m_c;
1613 } 2033 }
1614 } 2034 }
1615 } 2035 }
1616 2036
1617 modechange: 2037 modechange:
1618 ; 2038 ;
1619 } 2039 }
1620 2040
1621interrupt: 2041interrupt:
1622 self->incr_pos = p - SvPVX (self->incr_text); 2042 self->incr_pos = p - SvPVX (self->incr_text);
2043 //printf ("interrupt<%.*s>\n", self->incr_pos, SvPVX(self->incr_text));//D
1623 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D 2044 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D
1624} 2045}
1625 2046
1626///////////////////////////////////////////////////////////////////////////// 2047/////////////////////////////////////////////////////////////////////////////
1627// XS interface functions 2048// XS interface functions
1637 i >= '0' && i <= '9' ? i - '0' 2058 i >= '0' && i <= '9' ? i - '0'
1638 : i >= 'a' && i <= 'f' ? i - 'a' + 10 2059 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1639 : i >= 'A' && i <= 'F' ? i - 'A' + 10 2060 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1640 : -1; 2061 : -1;
1641 2062
1642 json_stash = gv_stashpv ("JSON::XS" , 1); 2063 json_stash = gv_stashpv ("JSON::XS" , 1);
1643 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); 2064 bool_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
1644
1645 json_true = get_bool ("JSON::XS::true");
1646 json_false = get_bool ("JSON::XS::false"); 2065 bool_false = get_bool ("Types::Serialiser::false");
2066 bool_true = get_bool ("Types::Serialiser::true");
2067
2068 sv_json = newSVpv ("JSON", 0);
2069 SvREADONLY_on (sv_json);
2070
2071 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */
1647} 2072}
1648 2073
1649PROTOTYPES: DISABLE 2074PROTOTYPES: DISABLE
1650 2075
1651void CLONE (...) 2076void CLONE (...)
1652 CODE: 2077 CODE:
2078 // as long as these writes are atomic, the race should not matter
2079 // as existing threads either already use 0, or use the old value,
2080 // which is sitll correct for the initial thread.
1653 json_stash = 0; 2081 json_stash = 0;
1654 json_boolean_stash = 0; 2082 bool_stash = 0;
2083 bool_false = 0;
2084 bool_true = 0;
1655 2085
1656void new (char *klass) 2086void new (char *klass)
1657 PPCODE: 2087 PPCODE:
1658{ 2088{
1659 SV *pv = NEWSV (0, sizeof (JSON)); 2089 SV *pv = NEWSV (0, sizeof (JSON));
1660 SvPOK_only (pv); 2090 SvPOK_only (pv);
1661 json_init ((JSON *)SvPVX (pv)); 2091 json_init ((JSON *)SvPVX (pv));
1662 XPUSHs (sv_2mortal (sv_bless ( 2092 XPUSHs (sv_2mortal (sv_bless (
1663 newRV_noinc (pv), 2093 newRV_noinc (pv),
1664 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1) 2094 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1665 ))); 2095 )));
1666} 2096}
2097
2098void boolean_values (JSON *self, SV *v_false = 0, SV *v_true = 0)
2099 PPCODE:
2100 self->v_false = newSVsv (v_false);
2101 self->v_true = newSVsv (v_true);
2102 XPUSHs (ST (0));
2103
2104void get_boolean_values (JSON *self)
2105 PPCODE:
2106 if (self->v_false && self->v_true)
2107 {
2108 EXTEND (SP, 2);
2109 PUSHs (self->v_false);
2110 PUSHs (self->v_true);
2111 }
1667 2112
1668void ascii (JSON *self, int enable = 1) 2113void ascii (JSON *self, int enable = 1)
1669 ALIAS: 2114 ALIAS:
1670 ascii = F_ASCII 2115 ascii = F_ASCII
1671 latin1 = F_LATIN1 2116 latin1 = F_LATIN1
1679 shrink = F_SHRINK 2124 shrink = F_SHRINK
1680 allow_blessed = F_ALLOW_BLESSED 2125 allow_blessed = F_ALLOW_BLESSED
1681 convert_blessed = F_CONV_BLESSED 2126 convert_blessed = F_CONV_BLESSED
1682 relaxed = F_RELAXED 2127 relaxed = F_RELAXED
1683 allow_unknown = F_ALLOW_UNKNOWN 2128 allow_unknown = F_ALLOW_UNKNOWN
2129 allow_tags = F_ALLOW_TAGS
1684 PPCODE: 2130 PPCODE:
1685{ 2131{
1686 if (enable) 2132 if (enable)
1687 self->flags |= ix; 2133 self->flags |= ix;
1688 else 2134 else
1704 get_shrink = F_SHRINK 2150 get_shrink = F_SHRINK
1705 get_allow_blessed = F_ALLOW_BLESSED 2151 get_allow_blessed = F_ALLOW_BLESSED
1706 get_convert_blessed = F_CONV_BLESSED 2152 get_convert_blessed = F_CONV_BLESSED
1707 get_relaxed = F_RELAXED 2153 get_relaxed = F_RELAXED
1708 get_allow_unknown = F_ALLOW_UNKNOWN 2154 get_allow_unknown = F_ALLOW_UNKNOWN
2155 get_allow_tags = F_ALLOW_TAGS
1709 PPCODE: 2156 PPCODE:
1710 XPUSHs (boolSV (self->flags & ix)); 2157 XPUSHs (boolSV (self->flags & ix));
1711 2158
1712void max_depth (JSON *self, U32 max_depth = 0x80000000UL) 2159void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
1713 PPCODE: 2160 PPCODE:
1741} 2188}
1742 2189
1743void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef) 2190void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
1744 PPCODE: 2191 PPCODE:
1745{ 2192{
1746 if (!self->cb_sk_object) 2193 if (!self->cb_sk_object)
1747 self->cb_sk_object = newHV (); 2194 self->cb_sk_object = newHV ();
1748 2195
1749 if (SvOK (cb)) 2196 if (SvOK (cb))
1750 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0); 2197 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1751 else 2198 else
1762 XPUSHs (ST (0)); 2209 XPUSHs (ST (0));
1763} 2210}
1764 2211
1765void encode (JSON *self, SV *scalar) 2212void encode (JSON *self, SV *scalar)
1766 PPCODE: 2213 PPCODE:
1767 XPUSHs (encode_json (scalar, self)); 2214 PUTBACK; scalar = encode_json (scalar, self); SPAGAIN;
2215 XPUSHs (scalar);
1768 2216
1769void decode (JSON *self, SV *jsonstr) 2217void decode (JSON *self, SV *jsonstr)
1770 PPCODE: 2218 PPCODE:
1771 XPUSHs (decode_json (jsonstr, self, 0)); 2219 PUTBACK; jsonstr = decode_json (jsonstr, self, 0); SPAGAIN;
2220 XPUSHs (jsonstr);
1772 2221
1773void decode_prefix (JSON *self, SV *jsonstr) 2222void decode_prefix (JSON *self, SV *jsonstr)
1774 PPCODE: 2223 PPCODE:
1775{ 2224{
2225 SV *sv;
1776 STRLEN offset; 2226 STRLEN offset;
2227 PUTBACK; sv = decode_json (jsonstr, self, &offset); SPAGAIN;
1777 EXTEND (SP, 2); 2228 EXTEND (SP, 2);
1778 PUSHs (decode_json (jsonstr, self, &offset)); 2229 PUSHs (sv);
1779 PUSHs (sv_2mortal (newSVuv (offset))); 2230 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, SvPV_nolen (jsonstr) + offset))));
1780} 2231}
1781 2232
1782void incr_parse (JSON *self, SV *jsonstr = 0) 2233void incr_parse (JSON *self, SV *jsonstr = 0)
1783 PPCODE: 2234 PPCODE:
1784{ 2235{
1785 if (!self->incr_text) 2236 if (!self->incr_text)
1786 self->incr_text = newSVpvn ("", 0); 2237 self->incr_text = newSVpvn ("", 0);
2238
2239 /* if utf8-ness doesn't match the decoder, need to upgrade/downgrade */
2240 if (!DECODE_WANTS_OCTETS (self) == !SvUTF8 (self->incr_text))
2241 if (DECODE_WANTS_OCTETS (self))
2242 {
2243 if (self->incr_pos)
2244 self->incr_pos = utf8_length ((U8 *)SvPVX (self->incr_text),
2245 (U8 *)SvPVX (self->incr_text) + self->incr_pos);
2246
2247 sv_utf8_downgrade (self->incr_text, 0);
2248 }
2249 else
2250 {
2251 sv_utf8_upgrade (self->incr_text);
2252
2253 if (self->incr_pos)
2254 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
2255 - (U8 *)SvPVX (self->incr_text);
2256 }
1787 2257
1788 // append data, if any 2258 // append data, if any
1789 if (jsonstr) 2259 if (jsonstr)
1790 { 2260 {
2261 /* make sure both strings have same encoding */
1791 if (SvUTF8 (jsonstr) && !SvUTF8 (self->incr_text)) 2262 if (SvUTF8 (jsonstr) != SvUTF8 (self->incr_text))
2263 if (SvUTF8 (jsonstr))
2264 sv_utf8_downgrade (jsonstr, 0);
1792 { 2265 else
1793 /* utf-8-ness differs, need to upgrade */
1794 sv_utf8_upgrade (self->incr_text); 2266 sv_utf8_upgrade (jsonstr);
1795 2267
1796 if (self->incr_pos) 2268 /* and then just blindly append */
1797 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
1798 - (U8 *)SvPVX (self->incr_text);
1799 }
1800
1801 { 2269 {
1802 STRLEN len; 2270 STRLEN len;
1803 const char *str = SvPV (jsonstr, len); 2271 const char *str = SvPV (jsonstr, len);
1804 SvGROW (self->incr_text, SvCUR (self->incr_text) + len + 1); 2272 STRLEN cur = SvCUR (self->incr_text);
2273
2274 if (SvLEN (self->incr_text) - cur <= len)
2275 json_sv_grow (self->incr_text, cur, len);
2276
1805 Move (str, SvEND (self->incr_text), len, char); 2277 Move (str, SvEND (self->incr_text), len, char);
1806 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len); 2278 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
1807 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there 2279 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
1808 } 2280 }
1809 } 2281 }
1810 2282
1811 if (GIMME_V != G_VOID) 2283 if (GIMME_V != G_VOID)
1812 do 2284 do
1813 { 2285 {
2286 SV *sv;
1814 STRLEN offset; 2287 STRLEN offset;
1815 2288
1816 if (!INCR_DONE (self)) 2289 if (!INCR_DONE (self))
1817 { 2290 {
1818 incr_parse (self); 2291 incr_parse (self);
1820 if (self->incr_pos > self->max_size && self->max_size) 2293 if (self->incr_pos > self->max_size && self->max_size)
1821 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", 2294 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1822 (unsigned long)self->incr_pos, (unsigned long)self->max_size); 2295 (unsigned long)self->incr_pos, (unsigned long)self->max_size);
1823 2296
1824 if (!INCR_DONE (self)) 2297 if (!INCR_DONE (self))
2298 {
2299 // as an optimisation, do not accumulate white space in the incr buffer
2300 if (self->incr_mode == INCR_M_WS && self->incr_pos)
2301 {
2302 self->incr_pos = 0;
2303 SvCUR_set (self->incr_text, 0);
2304 }
2305
1825 break; 2306 break;
2307 }
1826 } 2308 }
1827 2309
1828 XPUSHs (decode_json (self->incr_text, self, &offset)); 2310 PUTBACK; sv = decode_json (self->incr_text, self, &offset); SPAGAIN;
2311 XPUSHs (sv);
1829 2312
1830 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + offset);
1831 self->incr_pos -= offset; 2313 self->incr_pos -= offset;
1832 self->incr_nest = 0; 2314 self->incr_nest = 0;
1833 self->incr_mode = 0; 2315 self->incr_mode = 0;
2316
2317 sv_chop (self->incr_text, SvPVX (self->incr_text) + offset);
1834 } 2318 }
1835 while (GIMME_V == G_ARRAY); 2319 while (GIMME_V == G_ARRAY);
1836} 2320}
1837 2321
1838SV *incr_text (JSON *self) 2322SV *incr_text (JSON *self)
1869 self->incr_mode = 0; 2353 self->incr_mode = 0;
1870} 2354}
1871 2355
1872void DESTROY (JSON *self) 2356void DESTROY (JSON *self)
1873 CODE: 2357 CODE:
2358 SvREFCNT_dec (self->v_false);
2359 SvREFCNT_dec (self->v_true);
1874 SvREFCNT_dec (self->cb_sk_object); 2360 SvREFCNT_dec (self->cb_sk_object);
1875 SvREFCNT_dec (self->cb_object); 2361 SvREFCNT_dec (self->cb_object);
1876 SvREFCNT_dec (self->incr_text); 2362 SvREFCNT_dec (self->incr_text);
1877 2363
1878PROTOTYPES: ENABLE 2364PROTOTYPES: ENABLE
1879 2365
1880void encode_json (SV *scalar) 2366void encode_json (SV *scalar)
1881 ALIAS:
1882 to_json_ = 0
1883 encode_json = F_UTF8
1884 PPCODE: 2367 PPCODE:
1885{ 2368{
1886 JSON json; 2369 JSON json;
1887 json_init (&json); 2370 json_init (&json);
1888 json.flags |= ix; 2371 json.flags |= F_UTF8;
1889 XPUSHs (encode_json (scalar, &json)); 2372 PUTBACK; scalar = encode_json (scalar, &json); SPAGAIN;
2373 XPUSHs (scalar);
1890} 2374}
1891 2375
1892void decode_json (SV *jsonstr) 2376void decode_json (SV *jsonstr)
1893 ALIAS:
1894 from_json_ = 0
1895 decode_json = F_UTF8
1896 PPCODE: 2377 PPCODE:
1897{ 2378{
1898 JSON json; 2379 JSON json;
1899 json_init (&json); 2380 json_init (&json);
1900 json.flags |= ix; 2381 json.flags |= F_UTF8;
1901 XPUSHs (decode_json (jsonstr, &json, 0)); 2382 PUTBACK; jsonstr = decode_json (jsonstr, &json, 0); SPAGAIN;
2383 XPUSHs (jsonstr);
1902} 2384}
1903 2385
1904

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines