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 1.88 by root, Sun Jun 22 12:05:49 2008 UTC vs.
Revision 1.139 by root, Wed Mar 6 07:32:06 2019 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 - 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 static char hexdigit [16] = "0123456789abcdef";
286 need (enc, len += 5); 486 need (enc, len + 5);
287 *enc->cur++ = '\\'; 487 *enc->cur++ = '\\';
288 *enc->cur++ = 'u'; 488 *enc->cur++ = 'u';
289 *enc->cur++ = hexdigit [ uch >> 12 ]; 489 *enc->cur++ = PL_hexdigit [ uch >> 12 ];
290 *enc->cur++ = hexdigit [(uch >> 8) & 15]; 490 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
291 *enc->cur++ = hexdigit [(uch >> 4) & 15]; 491 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
292 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 492 *enc->cur++ = PL_hexdigit [(uch >> 0) & 15];
293 } 493 }
294 494
295 str += clen; 495 str += clen;
296 } 496 }
297 else if (enc->json.flags & F_LATIN1) 497 else if (enc->json.flags & F_LATIN1)
299 *enc->cur++ = uch; 499 *enc->cur++ = uch;
300 str += clen; 500 str += clen;
301 } 501 }
302 else if (is_utf8) 502 else if (is_utf8)
303 { 503 {
304 need (enc, len += clen); 504 need (enc, len + clen);
305 do 505 do
306 { 506 {
307 *enc->cur++ = *str++; 507 *enc->cur++ = *str++;
308 } 508 }
309 while (--clen); 509 while (--clen);
310 } 510 }
311 else 511 else
312 { 512 {
313 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
314 enc->cur = encode_utf8 (enc->cur, uch); 514 enc->cur = encode_utf8 (enc->cur, uch);
315 ++str; 515 ++str;
316 } 516 }
317 } 517 }
318 } 518 }
372 572
373 if (enc->indent >= enc->json.max_depth) 573 if (enc->indent >= enc->json.max_depth)
374 croak (ERR_NESTING_EXCEEDED); 574 croak (ERR_NESTING_EXCEEDED);
375 575
376 encode_ch (enc, '['); 576 encode_ch (enc, '[');
377 577
378 if (len >= 0) 578 if (len >= 0)
379 { 579 {
380 encode_nl (enc); ++enc->indent; 580 encode_nl (enc); ++enc->indent;
381 581
382 for (i = 0; i <= len; ++i) 582 for (i = 0; i <= len; ++i)
394 encode_comma (enc); 594 encode_comma (enc);
395 } 595 }
396 596
397 encode_nl (enc); --enc->indent; encode_indent (enc); 597 encode_nl (enc); --enc->indent; encode_indent (enc);
398 } 598 }
399 599
400 encode_ch (enc, ']'); 600 encode_ch (enc, ']');
401} 601}
402 602
403static void 603static void
404encode_hk (enc_t *enc, HE *he) 604encode_hk (enc_t *enc, HE *he)
408 if (HeKLEN (he) == HEf_SVKEY) 608 if (HeKLEN (he) == HEf_SVKEY)
409 { 609 {
410 SV *sv = HeSVKEY (he); 610 SV *sv = HeSVKEY (he);
411 STRLEN len; 611 STRLEN len;
412 char *str; 612 char *str;
413 613
414 SvGETMAGIC (sv); 614 SvGETMAGIC (sv);
415 str = SvPV (sv, len); 615 str = SvPV (sv, len);
416 616
417 encode_str (enc, str, len, SvUTF8 (sv)); 617 encode_str (enc, str, len, SvUTF8 (sv));
418 } 618 }
461 661
462 encode_ch (enc, '{'); 662 encode_ch (enc, '{');
463 663
464 // for canonical output we have to sort by keys first 664 // for canonical output we have to sort by keys first
465 // actually, this is mostly due to the stupid so-called 665 // actually, this is mostly due to the stupid so-called
466 // security workaround added somewhere in 5.8.x. 666 // security workaround added somewhere in 5.8.x
467 // that randomises hash orderings 667 // that randomises hash orderings
468 if (enc->json.flags & F_CANONICAL) 668 if (enc->json.flags & F_CANONICAL && !SvRMAGICAL (hv))
469 { 669 {
470 int count = hv_iterinit (hv); 670 int count = hv_iterinit (hv);
471 671
472 if (SvMAGICAL (hv)) 672 if (SvMAGICAL (hv))
473 { 673 {
484 } 684 }
485 685
486 if (count) 686 if (count)
487 { 687 {
488 int i, fast = 1; 688 int i, fast = 1;
489#if defined(__BORLANDC__) || defined(_MSC_VER) 689 HE *hes_stack [STACK_HES];
490 HE **hes = _alloca (count * sizeof (HE)); 690 HE **hes = hes_stack;
491#else 691
492 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode 692 // allocate larger arrays on the heap
493#endif 693 if (count > STACK_HES)
694 {
695 SV *sv = sv_2mortal (NEWSV (0, count * sizeof (*hes)));
696 hes = (HE **)SvPVX (sv);
697 }
494 698
495 i = 0; 699 i = 0;
496 while ((he = hv_iternext (hv))) 700 while ((he = hv_iternext (hv)))
497 { 701 {
498 hes [i++] = he; 702 hes [i++] = he;
567// encode objects, arrays and special \0=false and \1=true values. 771// encode objects, arrays and special \0=false and \1=true values.
568static void 772static void
569encode_rv (enc_t *enc, SV *sv) 773encode_rv (enc_t *enc, SV *sv)
570{ 774{
571 svtype svt; 775 svtype svt;
776 GV *method;
572 777
573 SvGETMAGIC (sv); 778 SvGETMAGIC (sv);
574 svt = SvTYPE (sv); 779 svt = SvTYPE (sv);
575 780
576 if (expect_false (SvOBJECT (sv))) 781 if (expect_false (SvOBJECT (sv)))
577 { 782 {
578 HV *stash = !JSON_SLOW || json_boolean_stash 783 HV *stash = SvSTASH (sv);
579 ? json_boolean_stash
580 : gv_stashpv ("JSON::XS::Boolean", 1);
581 784
582 if (SvSTASH (sv) == stash) 785 if (stash == bool_stash)
583 { 786 {
584 if (SvIV (sv))
585 encode_str (enc, "true", 4, 0); 787 if (SvIV (sv)) encode_str (enc, "true" , 4, 0);
586 else
587 encode_str (enc, "false", 5, 0); 788 else encode_str (enc, "false", 5, 0);
588 } 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 PUTBACK;
832
833 encode_ch (enc, ']');
834
835 FREETMPS; LEAVE;
836 }
837 else if ((enc->json.flags & F_CONV_BLESSED) && (method = gv_fetchmethod_autoload (stash, "TO_JSON", 0)))
838 {
839 dSP;
840
841 ENTER; SAVETMPS;
842 PUSHMARK (SP);
843 // we re-bless the reference to get overload and other niceties right
844 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
845
846 // calling with G_SCALAR ensures that we always get a 1 return value
847 PUTBACK;
848 call_sv ((SV *)GvCV (method), G_SCALAR);
849 SPAGAIN;
850
851 // catch this surprisingly common error
852 if (SvROK (TOPs) && SvRV (TOPs) == sv)
853 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
854
855 sv = POPs;
856 PUTBACK;
857
858 encode_sv (enc, sv);
859
860 FREETMPS; LEAVE;
861 }
862 else if (enc->json.flags & F_ALLOW_BLESSED)
863 encode_str (enc, "null", 4, 0);
589 else 864 else
590 {
591#if 0
592 if (0 && sv_derived_from (rv, "JSON::Literal"))
593 {
594 // not yet
595 }
596#endif
597 if (enc->json.flags & F_CONV_BLESSED)
598 {
599 // we re-bless the reference to get overload and other niceties right
600 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
601
602 if (to_json)
603 {
604 dSP;
605
606 ENTER; SAVETMPS; PUSHMARK (SP);
607 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
608
609 // calling with G_SCALAR ensures that we always get a 1 return value
610 PUTBACK;
611 call_sv ((SV *)GvCV (to_json), G_SCALAR);
612 SPAGAIN;
613
614 // catch this surprisingly common error
615 if (SvROK (TOPs) && SvRV (TOPs) == sv)
616 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
617
618 sv = POPs;
619 PUTBACK;
620
621 encode_sv (enc, sv);
622
623 FREETMPS; LEAVE;
624 }
625 else if (enc->json.flags & F_ALLOW_BLESSED)
626 encode_str (enc, "null", 4, 0);
627 else
628 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
629 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
630 }
631 else if (enc->json.flags & F_ALLOW_BLESSED)
632 encode_str (enc, "null", 4, 0);
633 else
634 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled", 865 croak ("encountered object '%s', but neither allow_blessed, convert_blessed nor allow_tags settings are enabled (or TO_JSON/FREEZE method missing)",
635 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 866 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
636 }
637 } 867 }
638 else if (svt == SVt_PVHV) 868 else if (svt == SVt_PVHV)
639 encode_hv (enc, (HV *)sv); 869 encode_hv (enc, (HV *)sv);
640 else if (svt == SVt_PVAV) 870 else if (svt == SVt_PVAV)
641 encode_av (enc, (AV *)sv); 871 encode_av (enc, (AV *)sv);
642 else if (svt < SVt_PVAV) 872 else if (svt < SVt_PVAV)
643 { 873 {
644 STRLEN len = 0; 874 int bool_type = ref_bool_type (sv);
645 char *pv = svt ? SvPV (sv, len) : 0;
646 875
647 if (len == 1 && *pv == '1') 876 if (bool_type == 1)
648 encode_str (enc, "true", 4, 0); 877 encode_str (enc, "true", 4, 0);
649 else if (len == 1 && *pv == '0') 878 else if (bool_type == 0)
650 encode_str (enc, "false", 5, 0); 879 encode_str (enc, "false", 5, 0);
651 else if (enc->json.flags & F_ALLOW_UNKNOWN) 880 else if (enc->json.flags & F_ALLOW_UNKNOWN)
652 encode_str (enc, "null", 4, 0); 881 encode_str (enc, "null", 4, 0);
653 else 882 else
654 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", 883 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
717 } 946 }
718 else 947 else
719 { 948 {
720 // large integer, use the (rather slow) snprintf way. 949 // large integer, use the (rather slow) snprintf way.
721 need (enc, IVUV_MAXCHARS); 950 need (enc, IVUV_MAXCHARS);
722 enc->cur += 951 enc->cur +=
723 SvIsUV(sv) 952 SvIsUV(sv)
724 ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv)) 953 ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))
725 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv)); 954 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
726 } 955 }
727 } 956 }
728 else if (SvROK (sv)) 957 else if (SvROK (sv))
729 encode_rv (enc, SvRV (sv)); 958 encode_rv (enc, SvRV (sv));
730 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN) 959 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN)
731 encode_str (enc, "null", 4, 0); 960 encode_str (enc, "null", 4, 0);
732 else 961 else
733 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 962 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, check your input data",
734 SvPV_nolen (sv), SvFLAGS (sv)); 963 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
735} 964}
736 965
737static SV * 966static SV *
738encode_json (SV *scalar, JSON *json) 967encode_json (SV *scalar, JSON *json)
739{ 968{
740 enc_t enc; 969 enc_t enc;
741 970
742 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar)) 971 if (!(json->flags & F_ALLOW_NONREF) && json_nonref (scalar))
743 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 972 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
744 973
745 enc.json = *json; 974 enc.json = *json;
746 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 975 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
747 enc.cur = SvPVX (enc.sv); 976 enc.cur = SvPVX (enc.sv);
751 : enc.json.flags & F_LATIN1 ? 0x000100UL 980 : enc.json.flags & F_LATIN1 ? 0x000100UL
752 : 0x110000UL; 981 : 0x110000UL;
753 982
754 SvPOK_only (enc.sv); 983 SvPOK_only (enc.sv);
755 encode_sv (&enc, scalar); 984 encode_sv (&enc, scalar);
985 encode_nl (&enc);
756 986
757 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 987 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
758 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 988 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
759 989
760 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8))) 990 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
940 else if (expect_true (ch >= 0x20 && ch < 0x80)) 1170 else if (expect_true (ch >= 0x20 && ch < 0x80))
941 *cur++ = ch; 1171 *cur++ = ch;
942 else if (ch >= 0x80) 1172 else if (ch >= 0x80)
943 { 1173 {
944 STRLEN clen; 1174 STRLEN clen;
945 UV uch;
946 1175
947 --dec_cur; 1176 --dec_cur;
948 1177
949 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen); 1178 decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
950 if (clen == (STRLEN)-1) 1179 if (clen == (STRLEN)-1)
951 ERR ("malformed UTF-8 character in JSON string"); 1180 ERR ("malformed UTF-8 character in JSON string");
952 1181
953 do 1182 do
954 *cur++ = *dec_cur++; 1183 *cur++ = *dec_cur++;
955 while (--clen); 1184 while (--clen);
956 1185
957 utf8 = 1; 1186 utf8 = 1;
958 } 1187 }
1188 else if (ch == '\t' && dec->json.flags & F_RELAXED)
1189 *cur++ = ch;
959 else 1190 else
960 { 1191 {
961 --dec_cur; 1192 --dec_cur;
962 1193
963 if (!ch) 1194 if (!ch)
971 { 1202 {
972 STRLEN len = cur - buf; 1203 STRLEN len = cur - buf;
973 1204
974 if (sv) 1205 if (sv)
975 { 1206 {
976 SvGROW (sv, SvCUR (sv) + len + 1); 1207 STRLEN cur = SvCUR (sv);
1208
1209 if (SvLEN (sv) - cur <= len)
1210 json_sv_grow (sv, cur, len);
1211
977 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 1212 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
978 SvCUR_set (sv, SvCUR (sv) + len); 1213 SvCUR_set (sv, SvCUR (sv) + len);
979 } 1214 }
980 else 1215 else
981 sv = newSVpvn (buf, len); 1216 sv = newSVpvn (buf, len);
1072 1307
1073 // special case the rather common 1..5-digit-int case 1308 // special case the rather common 1..5-digit-int case
1074 if (*start == '-') 1309 if (*start == '-')
1075 switch (len) 1310 switch (len)
1076 { 1311 {
1077 case 2: return newSViv (-( start [1] - '0' * 1)); 1312 case 2: return newSViv (-(IV)( start [1] - '0' * 1));
1078 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); 1313 case 3: return newSViv (-(IV)( start [1] * 10 + start [2] - '0' * 11));
1079 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); 1314 case 4: return newSViv (-(IV)( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
1080 case 5: return newSViv (-( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); 1315 case 5: return newSViv (-(IV)( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
1081 case 6: return newSViv (-(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111)); 1316 case 6: return newSViv (-(IV)(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
1082 } 1317 }
1083 else 1318 else
1084 switch (len) 1319 switch (len)
1085 { 1320 {
1086 case 1: return newSViv ( start [0] - '0' * 1); 1321 case 1: return newSViv ( start [0] - '0' * 1);
1087 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); 1322 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
1088 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); 1323 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
1089 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); 1324 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
1090 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111); 1325 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
1091 } 1326 }
1092 1327
1093 { 1328 {
1094 UV uv; 1329 UV uv;
1095 int numtype = grok_number (start, len, &uv); 1330 int numtype = grok_number (start, len, &uv);
1104 } 1339 }
1105 1340
1106 len -= *start == '-' ? 1 : 0; 1341 len -= *start == '-' ? 1 : 0;
1107 1342
1108 // does not fit into IV or UV, try NV 1343 // does not fit into IV or UV, try NV
1109 if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len) 1344 if (len <= NV_DIG)
1110 #if defined (LDBL_DIG)
1111 || (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len)
1112 #endif
1113 )
1114 // fits into NV without loss of precision 1345 // fits into NV without loss of precision
1115 return newSVnv (Atof (start)); 1346 return newSVnv (json_atof (start));
1116 1347
1117 // everything else fails, convert it to a string 1348 // everything else fails, convert it to a string
1118 return newSVpvn (start, dec->cur - start); 1349 return newSVpvn (start, dec->cur - start);
1119 } 1350 }
1120 1351
1121 // loss of precision here 1352 // loss of precision here
1122 return newSVnv (Atof (start)); 1353 return newSVnv (json_atof (start));
1123 1354
1124fail: 1355fail:
1125 return 0; 1356 return 0;
1126} 1357}
1127 1358
1151 if (*dec->cur == ']') 1382 if (*dec->cur == ']')
1152 { 1383 {
1153 ++dec->cur; 1384 ++dec->cur;
1154 break; 1385 break;
1155 } 1386 }
1156 1387
1157 if (*dec->cur != ',') 1388 if (*dec->cur != ',')
1158 ERR (", or ] expected while parsing array"); 1389 ERR (", or ] expected while parsing array");
1159 1390
1160 ++dec->cur; 1391 ++dec->cur;
1161 1392
1275 1506
1276 DEC_DEC_DEPTH; 1507 DEC_DEC_DEPTH;
1277 sv = newRV_noinc ((SV *)hv); 1508 sv = newRV_noinc ((SV *)hv);
1278 1509
1279 // check filter callbacks 1510 // check filter callbacks
1280 if (dec->json.flags & F_HOOK) 1511 if (expect_false (dec->json.flags & F_HOOK))
1281 { 1512 {
1282 if (dec->json.cb_sk_object && HvKEYS (hv) == 1) 1513 if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
1283 { 1514 {
1284 HE *cb, *he; 1515 HE *cb, *he;
1285 1516
1286 hv_iterinit (hv); 1517 hv_iterinit (hv);
1287 he = hv_iternext (hv); 1518 he = hv_iternext (hv);
1288 hv_iterinit (hv); 1519 hv_iterinit (hv);
1289 1520
1290 // the next line creates a mortal sv each time its called. 1521 // the next line creates a mortal sv each time it's called.
1291 // might want to optimise this for common cases. 1522 // might want to optimise this for common cases.
1292 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0); 1523 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1293 1524
1294 if (cb) 1525 if (cb)
1295 { 1526 {
1296 dSP; 1527 dSP;
1297 int count; 1528 int count;
1298 1529
1299 ENTER; SAVETMPS; PUSHMARK (SP); 1530 ENTER; SAVETMPS;
1531 PUSHMARK (SP);
1300 XPUSHs (HeVAL (he)); 1532 XPUSHs (HeVAL (he));
1533 sv_2mortal (sv);
1301 1534
1302 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN; 1535 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1303 1536
1304 if (count == 1) 1537 if (count == 1)
1305 { 1538 {
1306 sv = newSVsv (POPs); 1539 sv = newSVsv (POPs);
1540 PUTBACK;
1307 FREETMPS; LEAVE; 1541 FREETMPS; LEAVE;
1308 return sv; 1542 return sv;
1309 } 1543 }
1544 else if (count)
1545 croak ("filter_json_single_key_object callbacks must not return more than one scalar");
1546
1547 PUTBACK;
1548
1549 SvREFCNT_inc (sv);
1310 1550
1311 FREETMPS; LEAVE; 1551 FREETMPS; LEAVE;
1312 } 1552 }
1313 } 1553 }
1314 1554
1315 if (dec->json.cb_object) 1555 if (dec->json.cb_object)
1316 { 1556 {
1317 dSP; 1557 dSP;
1318 int count; 1558 int count;
1319 1559
1320 ENTER; SAVETMPS; PUSHMARK (SP); 1560 ENTER; SAVETMPS;
1561 PUSHMARK (SP);
1321 XPUSHs (sv_2mortal (sv)); 1562 XPUSHs (sv_2mortal (sv));
1322 1563
1323 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN; 1564 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1324 1565
1325 if (count == 1) 1566 if (count == 1)
1326 {
1327 sv = newSVsv (POPs); 1567 sv = newSVsv (POPs);
1328 FREETMPS; LEAVE; 1568 else if (count == 0)
1329 return sv;
1330 }
1331
1332 SvREFCNT_inc (sv); 1569 SvREFCNT_inc (sv);
1570 else
1571 croak ("filter_json_object callbacks must not return more than one scalar");
1572
1573 PUTBACK;
1574
1333 FREETMPS; LEAVE; 1575 FREETMPS; LEAVE;
1334 } 1576 }
1335 } 1577 }
1336 1578
1337 return sv; 1579 return sv;
1341 DEC_DEC_DEPTH; 1583 DEC_DEC_DEPTH;
1342 return 0; 1584 return 0;
1343} 1585}
1344 1586
1345static SV * 1587static SV *
1588decode_tag (dec_t *dec)
1589{
1590 SV *tag = 0;
1591 SV *val = 0;
1592
1593 if (!(dec->json.flags & F_ALLOW_TAGS))
1594 ERR ("malformed JSON string, neither array, object, number, string or atom");
1595
1596 ++dec->cur;
1597
1598 decode_ws (dec);
1599
1600 tag = decode_sv (dec);
1601 if (!tag)
1602 goto fail;
1603
1604 if (!SvPOK (tag))
1605 ERR ("malformed JSON string, (tag) must be a string");
1606
1607 decode_ws (dec);
1608
1609 if (*dec->cur != ')')
1610 ERR (") expected after tag");
1611
1612 ++dec->cur;
1613
1614 decode_ws (dec);
1615
1616 val = decode_sv (dec);
1617 if (!val)
1618 goto fail;
1619
1620 if (!SvROK (val) || SvTYPE (SvRV (val)) != SVt_PVAV)
1621 ERR ("malformed JSON string, tag value must be an array");
1622
1623 {
1624 AV *av = (AV *)SvRV (val);
1625 int i, len = av_len (av) + 1;
1626 HV *stash = gv_stashsv (tag, 0);
1627 SV *sv;
1628
1629 if (!stash)
1630 ERR ("cannot decode perl-object (package does not exist)");
1631
1632 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
1633
1634 if (!method)
1635 ERR ("cannot decode perl-object (package does not have a THAW method)");
1636
1637 dSP;
1638
1639 ENTER; SAVETMPS;
1640 PUSHMARK (SP);
1641 EXTEND (SP, len + 2);
1642 // we re-bless the reference to get overload and other niceties right
1643 PUSHs (tag);
1644 PUSHs (sv_json);
1645
1646 for (i = 0; i < len; ++i)
1647 PUSHs (*av_fetch (av, i, 1));
1648
1649 PUTBACK;
1650 call_sv ((SV *)GvCV (method), G_SCALAR);
1651 SPAGAIN;
1652
1653 SvREFCNT_dec (tag);
1654 SvREFCNT_dec (val);
1655 sv = SvREFCNT_inc (POPs);
1656
1657 PUTBACK;
1658
1659 FREETMPS; LEAVE;
1660
1661 return sv;
1662 }
1663
1664fail:
1665 SvREFCNT_dec (tag);
1666 SvREFCNT_dec (val);
1667 return 0;
1668}
1669
1670static SV *
1346decode_sv (dec_t *dec) 1671decode_sv (dec_t *dec)
1347{ 1672{
1348 // the beauty of JSON: you need exactly one character lookahead 1673 // the beauty of JSON: you need exactly one character lookahead
1349 // to parse everything. 1674 // to parse everything.
1350 switch (*dec->cur) 1675 switch (*dec->cur)
1351 { 1676 {
1352 case '"': ++dec->cur; return decode_str (dec); 1677 case '"': ++dec->cur; return decode_str (dec);
1353 case '[': ++dec->cur; return decode_av (dec); 1678 case '[': ++dec->cur; return decode_av (dec);
1354 case '{': ++dec->cur; return decode_hv (dec); 1679 case '{': ++dec->cur; return decode_hv (dec);
1680 case '(': return decode_tag (dec);
1355 1681
1356 case '-': 1682 case '-':
1357 case '0': case '1': case '2': case '3': case '4': 1683 case '0': case '1': case '2': case '3': case '4':
1358 case '5': case '6': case '7': case '8': case '9': 1684 case '5': case '6': case '7': case '8': case '9':
1359 return decode_num (dec); 1685 return decode_num (dec);
1360 1686
1687 case 'f':
1688 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1689 {
1690 dec->cur += 5;
1691
1692 if (expect_false (!dec->json.v_false))
1693 dec->json.v_false = GET_BOOL (false);
1694
1695 return newSVsv (dec->json.v_false);
1696 }
1697 else
1698 ERR ("'false' expected");
1699
1700 break;
1701
1361 case 't': 1702 case 't':
1362 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1703 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1363 { 1704 {
1364 dec->cur += 4; 1705 dec->cur += 4;
1365#if JSON_SLOW 1706
1366 json_true = get_bool ("JSON::XS::true"); 1707 if (expect_false (!dec->json.v_true))
1367#endif 1708 dec->json.v_true = GET_BOOL (true);
1709
1368 return newSVsv (json_true); 1710 return newSVsv (dec->json.v_true);
1369 } 1711 }
1370 else 1712 else
1371 ERR ("'true' expected"); 1713 ERR ("'true' expected");
1372
1373 break;
1374
1375 case 'f':
1376 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1377 {
1378 dec->cur += 5;
1379#if JSON_SLOW
1380 json_false = get_bool ("JSON::XS::false");
1381#endif
1382 return newSVsv (json_false);
1383 }
1384 else
1385 ERR ("'false' expected");
1386 1714
1387 break; 1715 break;
1388 1716
1389 case 'n': 1717 case 'n':
1390 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4)) 1718 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4))
1396 ERR ("'null' expected"); 1724 ERR ("'null' expected");
1397 1725
1398 break; 1726 break;
1399 1727
1400 default: 1728 default:
1401 ERR ("malformed JSON string, neither array, object, number, string or atom"); 1729 ERR ("malformed JSON string, neither tag, array, object, number, string or atom");
1402 break; 1730 break;
1403 } 1731 }
1404 1732
1405fail: 1733fail:
1406 return 0; 1734 return 0;
1408 1736
1409static SV * 1737static SV *
1410decode_json (SV *string, JSON *json, STRLEN *offset_return) 1738decode_json (SV *string, JSON *json, STRLEN *offset_return)
1411{ 1739{
1412 dec_t dec; 1740 dec_t dec;
1413 STRLEN offset;
1414 SV *sv; 1741 SV *sv;
1415 1742
1743 /* work around bugs in 5.10 where manipulating magic values
1744 * makes perl ignore the magic in subsequent accesses.
1745 * also make a copy of non-PV values, to get them into a clean
1746 * state (SvPV should do that, but it's buggy, see below).
1747 *
1748 * SvIsCOW_shared_hash works around a bug in perl (possibly 5.16),
1749 * as reported by Reini Urban.
1750 */
1416 SvGETMAGIC (string); 1751 /*SvGETMAGIC (string);*/
1752 if (SvMAGICAL (string) || !SvPOK (string) || SvIsCOW_shared_hash (string))
1753 string = sv_2mortal (newSVsv (string));
1754
1417 SvUPGRADE (string, SVt_PV); 1755 SvUPGRADE (string, SVt_PV);
1418 1756
1419 /* work around a bug in perl 5.10, which causes SvCUR to fail an 1757 /* work around a bug in perl 5.10, which causes SvCUR to fail an
1420 * assertion with -DDEBUGGING, although SvCUR is documented to 1758 * assertion with -DDEBUGGING, although SvCUR is documented to
1421 * return the xpv_cur field which certainly exists after upgrading. 1759 * return the xpv_cur field which certainly exists after upgrading.
1422 * according to nicholas clark, calling SvPOK fixes this. 1760 * according to nicholas clark, calling SvPOK fixes this.
1761 * But it doesn't fix it, so try another workaround, call SvPV_nolen
1762 * and hope for the best.
1763 * Damnit, SvPV_nolen still trips over yet another assertion. This
1764 * assertion business is seriously broken, try yet another workaround
1765 * for the broken -DDEBUGGING.
1423 */ 1766 */
1424 SvPOK (string); 1767 {
1768#ifdef DEBUGGING
1769 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
1770#else
1771 STRLEN offset = SvCUR (string);
1772#endif
1425 1773
1426 if (SvCUR (string) > json->max_size && json->max_size) 1774 if (offset > json->max_size && json->max_size)
1427 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", 1775 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1428 (unsigned long)SvCUR (string), (unsigned long)json->max_size); 1776 (unsigned long)SvCUR (string), (unsigned long)json->max_size);
1777 }
1429 1778
1430 if (json->flags & F_UTF8) 1779 if (DECODE_WANTS_OCTETS (json))
1431 sv_utf8_downgrade (string, 0); 1780 sv_utf8_downgrade (string, 0);
1432 else 1781 else
1433 sv_utf8_upgrade (string); 1782 sv_utf8_upgrade (string);
1434 1783
1435 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1784 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
1446 *dec.end = 0; // this should basically be a nop, too, but make sure it's there 1795 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1447 1796
1448 decode_ws (&dec); 1797 decode_ws (&dec);
1449 sv = decode_sv (&dec); 1798 sv = decode_sv (&dec);
1450 1799
1451 if (!(offset_return || !sv)) 1800 if (offset_return)
1801 *offset_return = dec.cur - SvPVX (string);
1802 else if (sv)
1452 { 1803 {
1453 // check for trailing garbage 1804 // check for trailing garbage
1454 decode_ws (&dec); 1805 decode_ws (&dec);
1455 1806
1456 if (*dec.cur) 1807 if (dec.cur != dec.end)
1457 { 1808 {
1458 dec.err = "garbage after JSON object"; 1809 dec.err = "garbage after JSON object";
1459 SvREFCNT_dec (sv); 1810 SvREFCNT_dec (sv);
1460 sv = 0; 1811 sv = 0;
1461 } 1812 }
1462 }
1463
1464 if (offset_return || !sv)
1465 {
1466 offset = dec.json.flags & F_UTF8
1467 ? dec.cur - SvPVX (string)
1468 : utf8_distance (dec.cur, SvPVX (string));
1469
1470 if (offset_return)
1471 *offset_return = offset;
1472 } 1813 }
1473 1814
1474 if (!sv) 1815 if (!sv)
1475 { 1816 {
1476 SV *uni = sv_newmortal (); 1817 SV *uni = sv_newmortal ();
1482 SAVEVPTR (PL_curcop); 1823 SAVEVPTR (PL_curcop);
1483 PL_curcop = &cop; 1824 PL_curcop = &cop;
1484 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1825 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1485 LEAVE; 1826 LEAVE;
1486 1827
1487 croak ("%s, at character offset %d [\"%s\"]", 1828 croak ("%s, at character offset %d (before \"%s\")",
1488 dec.err, 1829 dec.err,
1489 (int)offset, 1830 (int)ptr_to_index (string, dec.cur),
1490 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1831 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1491 } 1832 }
1492 1833
1493 sv = sv_2mortal (sv); 1834 sv = sv_2mortal (sv);
1494 1835
1495 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1836 if (!(dec.json.flags & F_ALLOW_NONREF) && json_nonref (sv))
1496 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); 1837 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1497 1838
1498 return sv; 1839 return sv;
1499} 1840}
1500 1841
1504static void 1845static void
1505incr_parse (JSON *self) 1846incr_parse (JSON *self)
1506{ 1847{
1507 const char *p = SvPVX (self->incr_text) + self->incr_pos; 1848 const char *p = SvPVX (self->incr_text) + self->incr_pos;
1508 1849
1850 // the state machine here is a bit convoluted and could be simplified a lot
1851 // but this would make it slower, so...
1852
1509 for (;;) 1853 for (;;)
1510 { 1854 {
1511 //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
1512 switch (self->incr_mode) 1855 switch (self->incr_mode)
1513 { 1856 {
1857 // reached end of a scalar, see if we are inside a nested structure or not
1858 end_of_scalar:
1859 self->incr_mode = INCR_M_JSON;
1860
1861 if (self->incr_nest) // end of a scalar inside array, object or tag
1862 goto incr_m_json;
1863 else // end of scalar outside structure, json text ends here
1864 goto interrupt;
1865
1514 // only used for intiial whitespace skipping 1866 // only used for initial whitespace skipping
1515 case INCR_M_WS: 1867 case INCR_M_WS:
1516 for (;;) 1868 for (;;)
1517 { 1869 {
1518 if (*p > 0x20) 1870 if (*p > 0x20)
1519 { 1871 {
1872 if (*p == '#')
1873 {
1874 self->incr_mode = INCR_M_C0;
1875 goto incr_m_c;
1876 }
1877 else
1878 {
1520 self->incr_mode = INCR_M_JSON; 1879 self->incr_mode = INCR_M_JSON;
1521 goto incr_m_json; 1880 goto incr_m_json;
1881 }
1522 } 1882 }
1523 else if (!*p) 1883 else if (!*p)
1524 goto interrupt; 1884 goto interrupt;
1525 1885
1526 ++p; 1886 ++p;
1532 goto interrupt; 1892 goto interrupt;
1533 1893
1534 ++p; 1894 ++p;
1535 self->incr_mode = INCR_M_STR; 1895 self->incr_mode = INCR_M_STR;
1536 goto incr_m_str; 1896 goto incr_m_str;
1897
1898 // inside #-style comments
1899 case INCR_M_C0:
1900 case INCR_M_C1:
1901 incr_m_c:
1902 for (;;)
1903 {
1904 if (*p == '\n')
1905 {
1906 self->incr_mode = self->incr_mode == INCR_M_C0 ? INCR_M_WS : INCR_M_JSON;
1907 break;
1908 }
1909 else if (!*p)
1910 goto interrupt;
1911
1912 ++p;
1913 }
1914
1915 break;
1916
1917 // inside true/false/null
1918 case INCR_M_TFN:
1919 incr_m_tfn:
1920 for (;;)
1921 switch (*p++)
1922 {
1923 case 'r': case 'u': case 'e': // tRUE, falsE, nUll
1924 case 'a': case 'l': case 's': // fALSe, nuLL
1925 // allowed
1926 break;
1927
1928 default:
1929 --p;
1930 goto end_of_scalar;
1931 }
1932
1933 // inside a number
1934 case INCR_M_NUM:
1935 incr_m_num:
1936 for (;;)
1937 switch (*p++)
1938 {
1939 case 'e': case 'E': case '.': case '+':
1940 case '-':
1941 case '0': case '1': case '2': case '3': case '4':
1942 case '5': case '6': case '7': case '8': case '9':
1943 // allowed
1944 break;
1945
1946 default:
1947 --p;
1948 goto end_of_scalar;
1949 }
1537 1950
1538 // inside a string 1951 // inside a string
1539 case INCR_M_STR: 1952 case INCR_M_STR:
1540 incr_m_str: 1953 incr_m_str:
1541 for (;;) 1954 for (;;)
1542 { 1955 {
1543 if (*p == '"') 1956 if (*p == '"')
1544 { 1957 {
1545 ++p; 1958 ++p;
1546 self->incr_mode = INCR_M_JSON;
1547
1548 if (!self->incr_nest)
1549 goto interrupt;
1550
1551 goto incr_m_json; 1959 goto end_of_scalar;
1552 } 1960 }
1553 else if (*p == '\\') 1961 else if (*p == '\\')
1554 { 1962 {
1555 ++p; // "virtually" consumes character after \ 1963 ++p; // "virtually" consumes character after \
1556 1964
1586 --p; // do not eat the whitespace, let the next round do it 1994 --p; // do not eat the whitespace, let the next round do it
1587 goto interrupt; 1995 goto interrupt;
1588 } 1996 }
1589 break; 1997 break;
1590 1998
1999 // the following three blocks handle scalars. this makes the parser
2000 // more strict than required inside arrays or objects, and could
2001 // be moved to a special case on the toplevel (except strings)
2002 case 't':
2003 case 'f':
2004 case 'n':
2005 self->incr_mode = INCR_M_TFN;
2006 goto incr_m_tfn;
2007
2008 case '-':
2009 case '0': case '1': case '2': case '3': case '4':
2010 case '5': case '6': case '7': case '8': case '9':
2011 self->incr_mode = INCR_M_NUM;
2012 goto incr_m_num;
2013
1591 case '"': 2014 case '"':
1592 self->incr_mode = INCR_M_STR; 2015 self->incr_mode = INCR_M_STR;
1593 goto incr_m_str; 2016 goto incr_m_str;
1594 2017
1595 case '[': 2018 case '[':
1596 case '{': 2019 case '{':
2020 case '(':
1597 if (++self->incr_nest > self->max_depth) 2021 if (++self->incr_nest > self->max_depth)
1598 croak (ERR_NESTING_EXCEEDED); 2022 croak (ERR_NESTING_EXCEEDED);
1599 break; 2023 break;
1600 2024
1601 case ']': 2025 case ']':
1602 case '}': 2026 case '}':
1603 if (--self->incr_nest <= 0) 2027 if (--self->incr_nest <= 0)
1604 goto interrupt; 2028 goto interrupt;
2029 break;
2030
2031 case ')':
2032 --self->incr_nest;
2033 break;
2034
2035 case '#':
2036 self->incr_mode = INCR_M_C1;
2037 goto incr_m_c;
1605 } 2038 }
1606 } 2039 }
1607 } 2040 }
1608 2041
1609 modechange: 2042 modechange:
1610 ; 2043 ;
1611 } 2044 }
1612 2045
1613interrupt: 2046interrupt:
1614 self->incr_pos = p - SvPVX (self->incr_text); 2047 self->incr_pos = p - SvPVX (self->incr_text);
2048 //printf ("interrupt<%.*s>\n", self->incr_pos, SvPVX(self->incr_text));//D
1615 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D 2049 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D
1616} 2050}
1617 2051
1618///////////////////////////////////////////////////////////////////////////// 2052/////////////////////////////////////////////////////////////////////////////
1619// XS interface functions 2053// XS interface functions
1629 i >= '0' && i <= '9' ? i - '0' 2063 i >= '0' && i <= '9' ? i - '0'
1630 : i >= 'a' && i <= 'f' ? i - 'a' + 10 2064 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1631 : i >= 'A' && i <= 'F' ? i - 'A' + 10 2065 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1632 : -1; 2066 : -1;
1633 2067
1634 json_stash = gv_stashpv ("JSON::XS" , 1); 2068 json_stash = gv_stashpv ("JSON::XS" , 1);
1635 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); 2069 bool_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
1636
1637 json_true = get_bool ("JSON::XS::true");
1638 json_false = get_bool ("JSON::XS::false"); 2070 bool_false = get_bool ("Types::Serialiser::false");
2071 bool_true = get_bool ("Types::Serialiser::true");
2072
2073 sv_json = newSVpv ("JSON", 0);
2074 SvREADONLY_on (sv_json);
2075
2076 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */
1639} 2077}
1640 2078
1641PROTOTYPES: DISABLE 2079PROTOTYPES: DISABLE
1642 2080
1643void CLONE (...) 2081void CLONE (...)
1644 CODE: 2082 CODE:
2083 // as long as these writes are atomic, the race should not matter
2084 // as existing threads either already use 0, or use the old value,
2085 // which is sitll correct for the initial thread.
1645 json_stash = 0; 2086 json_stash = 0;
1646 json_boolean_stash = 0; 2087 bool_stash = 0;
2088 bool_false = 0;
2089 bool_true = 0;
1647 2090
1648void new (char *klass) 2091void new (char *klass)
1649 PPCODE: 2092 PPCODE:
1650{ 2093{
1651 SV *pv = NEWSV (0, sizeof (JSON)); 2094 SV *pv = NEWSV (0, sizeof (JSON));
1652 SvPOK_only (pv); 2095 SvPOK_only (pv);
1653 json_init ((JSON *)SvPVX (pv)); 2096 json_init ((JSON *)SvPVX (pv));
1654 XPUSHs (sv_2mortal (sv_bless ( 2097 XPUSHs (sv_2mortal (sv_bless (
1655 newRV_noinc (pv), 2098 newRV_noinc (pv),
1656 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1) 2099 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1657 ))); 2100 )));
1658} 2101}
2102
2103void boolean_values (JSON *self, SV *v_false = 0, SV *v_true = 0)
2104 PPCODE:
2105 self->v_false = newSVsv (v_false);
2106 self->v_true = newSVsv (v_true);
2107 XPUSHs (ST (0));
2108
2109void get_boolean_values (JSON *self)
2110 PPCODE:
2111 if (self->v_false && self->v_true)
2112 {
2113 EXTEND (SP, 2);
2114 PUSHs (self->v_false);
2115 PUSHs (self->v_true);
2116 }
1659 2117
1660void ascii (JSON *self, int enable = 1) 2118void ascii (JSON *self, int enable = 1)
1661 ALIAS: 2119 ALIAS:
1662 ascii = F_ASCII 2120 ascii = F_ASCII
1663 latin1 = F_LATIN1 2121 latin1 = F_LATIN1
1671 shrink = F_SHRINK 2129 shrink = F_SHRINK
1672 allow_blessed = F_ALLOW_BLESSED 2130 allow_blessed = F_ALLOW_BLESSED
1673 convert_blessed = F_CONV_BLESSED 2131 convert_blessed = F_CONV_BLESSED
1674 relaxed = F_RELAXED 2132 relaxed = F_RELAXED
1675 allow_unknown = F_ALLOW_UNKNOWN 2133 allow_unknown = F_ALLOW_UNKNOWN
2134 allow_tags = F_ALLOW_TAGS
1676 PPCODE: 2135 PPCODE:
1677{ 2136{
1678 if (enable) 2137 if (enable)
1679 self->flags |= ix; 2138 self->flags |= ix;
1680 else 2139 else
1696 get_shrink = F_SHRINK 2155 get_shrink = F_SHRINK
1697 get_allow_blessed = F_ALLOW_BLESSED 2156 get_allow_blessed = F_ALLOW_BLESSED
1698 get_convert_blessed = F_CONV_BLESSED 2157 get_convert_blessed = F_CONV_BLESSED
1699 get_relaxed = F_RELAXED 2158 get_relaxed = F_RELAXED
1700 get_allow_unknown = F_ALLOW_UNKNOWN 2159 get_allow_unknown = F_ALLOW_UNKNOWN
2160 get_allow_tags = F_ALLOW_TAGS
1701 PPCODE: 2161 PPCODE:
1702 XPUSHs (boolSV (self->flags & ix)); 2162 XPUSHs (boolSV (self->flags & ix));
1703 2163
1704void max_depth (JSON *self, U32 max_depth = 0x80000000UL) 2164void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
1705 PPCODE: 2165 PPCODE:
1733} 2193}
1734 2194
1735void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef) 2195void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
1736 PPCODE: 2196 PPCODE:
1737{ 2197{
1738 if (!self->cb_sk_object) 2198 if (!self->cb_sk_object)
1739 self->cb_sk_object = newHV (); 2199 self->cb_sk_object = newHV ();
1740 2200
1741 if (SvOK (cb)) 2201 if (SvOK (cb))
1742 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0); 2202 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1743 else 2203 else
1754 XPUSHs (ST (0)); 2214 XPUSHs (ST (0));
1755} 2215}
1756 2216
1757void encode (JSON *self, SV *scalar) 2217void encode (JSON *self, SV *scalar)
1758 PPCODE: 2218 PPCODE:
1759 XPUSHs (encode_json (scalar, self)); 2219 PUTBACK; scalar = encode_json (scalar, self); SPAGAIN;
2220 XPUSHs (scalar);
1760 2221
1761void decode (JSON *self, SV *jsonstr) 2222void decode (JSON *self, SV *jsonstr)
1762 PPCODE: 2223 PPCODE:
1763 XPUSHs (decode_json (jsonstr, self, 0)); 2224 PUTBACK; jsonstr = decode_json (jsonstr, self, 0); SPAGAIN;
2225 XPUSHs (jsonstr);
1764 2226
1765void decode_prefix (JSON *self, SV *jsonstr) 2227void decode_prefix (JSON *self, SV *jsonstr)
1766 PPCODE: 2228 PPCODE:
1767{ 2229{
2230 SV *sv;
1768 STRLEN offset; 2231 STRLEN offset;
2232 PUTBACK; sv = decode_json (jsonstr, self, &offset); SPAGAIN;
1769 EXTEND (SP, 2); 2233 EXTEND (SP, 2);
1770 PUSHs (decode_json (jsonstr, self, &offset)); 2234 PUSHs (sv);
1771 PUSHs (sv_2mortal (newSVuv (offset))); 2235 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, SvPV_nolen (jsonstr) + offset))));
1772} 2236}
1773 2237
1774void incr_parse (JSON *self, SV *jsonstr = 0) 2238void incr_parse (JSON *self, SV *jsonstr = 0)
1775 PPCODE: 2239 PPCODE:
1776{ 2240{
1777 if (!self->incr_text) 2241 if (!self->incr_text)
1778 self->incr_text = newSVpvn ("", 0); 2242 self->incr_text = newSVpvn ("", 0);
2243
2244 /* if utf8-ness doesn't match the decoder, need to upgrade/downgrade */
2245 if (!DECODE_WANTS_OCTETS (self) == !SvUTF8 (self->incr_text))
2246 if (DECODE_WANTS_OCTETS (self))
2247 {
2248 if (self->incr_pos)
2249 self->incr_pos = utf8_length ((U8 *)SvPVX (self->incr_text),
2250 (U8 *)SvPVX (self->incr_text) + self->incr_pos);
2251
2252 sv_utf8_downgrade (self->incr_text, 0);
2253 }
2254 else
2255 {
2256 sv_utf8_upgrade (self->incr_text);
2257
2258 if (self->incr_pos)
2259 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
2260 - (U8 *)SvPVX (self->incr_text);
2261 }
1779 2262
1780 // append data, if any 2263 // append data, if any
1781 if (jsonstr) 2264 if (jsonstr)
1782 { 2265 {
2266 /* make sure both strings have same encoding */
1783 if (SvUTF8 (jsonstr) && !SvUTF8 (self->incr_text)) 2267 if (SvUTF8 (jsonstr) != SvUTF8 (self->incr_text))
2268 if (SvUTF8 (jsonstr))
2269 sv_utf8_downgrade (jsonstr, 0);
1784 { 2270 else
1785 /* utf-8-ness differs, need to upgrade */
1786 sv_utf8_upgrade (self->incr_text); 2271 sv_utf8_upgrade (jsonstr);
1787 2272
1788 if (self->incr_pos) 2273 /* and then just blindly append */
1789 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
1790 - (U8 *)SvPVX (self->incr_text);
1791 }
1792
1793 { 2274 {
1794 STRLEN len; 2275 STRLEN len;
1795 const char *str = SvPV (jsonstr, len); 2276 const char *str = SvPV (jsonstr, len);
1796 SvGROW (self->incr_text, SvCUR (self->incr_text) + len + 1); 2277 STRLEN cur = SvCUR (self->incr_text);
2278
2279 if (SvLEN (self->incr_text) - cur <= len)
2280 json_sv_grow (self->incr_text, cur, len);
2281
1797 Move (str, SvEND (self->incr_text), len, char); 2282 Move (str, SvEND (self->incr_text), len, char);
1798 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len); 2283 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
1799 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there 2284 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
1800 } 2285 }
1801 } 2286 }
1802 2287
1803 if (GIMME_V != G_VOID) 2288 if (GIMME_V != G_VOID)
1804 do 2289 do
1805 { 2290 {
2291 SV *sv;
1806 STRLEN offset; 2292 STRLEN offset;
1807 2293
1808 if (!INCR_DONE (self)) 2294 if (!INCR_DONE (self))
1809 { 2295 {
1810 incr_parse (self); 2296 incr_parse (self);
1812 if (self->incr_pos > self->max_size && self->max_size) 2298 if (self->incr_pos > self->max_size && self->max_size)
1813 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", 2299 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1814 (unsigned long)self->incr_pos, (unsigned long)self->max_size); 2300 (unsigned long)self->incr_pos, (unsigned long)self->max_size);
1815 2301
1816 if (!INCR_DONE (self)) 2302 if (!INCR_DONE (self))
2303 {
2304 // as an optimisation, do not accumulate white space in the incr buffer
2305 if (self->incr_mode == INCR_M_WS && self->incr_pos)
2306 {
2307 self->incr_pos = 0;
2308 SvCUR_set (self->incr_text, 0);
2309 }
2310
1817 break; 2311 break;
2312 }
1818 } 2313 }
1819 2314
1820 XPUSHs (decode_json (self->incr_text, self, &offset)); 2315 PUTBACK; sv = decode_json (self->incr_text, self, &offset); SPAGAIN;
2316 XPUSHs (sv);
1821 2317
1822 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + offset);
1823 self->incr_pos -= offset; 2318 self->incr_pos -= offset;
1824 self->incr_nest = 0; 2319 self->incr_nest = 0;
1825 self->incr_mode = 0; 2320 self->incr_mode = 0;
2321
2322 sv_chop (self->incr_text, SvPVX (self->incr_text) + offset);
1826 } 2323 }
1827 while (GIMME_V == G_ARRAY); 2324 while (GIMME_V == G_ARRAY);
1828} 2325}
1829 2326
1830SV *incr_text (JSON *self) 2327SV *incr_text (JSON *self)
1861 self->incr_mode = 0; 2358 self->incr_mode = 0;
1862} 2359}
1863 2360
1864void DESTROY (JSON *self) 2361void DESTROY (JSON *self)
1865 CODE: 2362 CODE:
2363 SvREFCNT_dec (self->v_false);
2364 SvREFCNT_dec (self->v_true);
1866 SvREFCNT_dec (self->cb_sk_object); 2365 SvREFCNT_dec (self->cb_sk_object);
1867 SvREFCNT_dec (self->cb_object); 2366 SvREFCNT_dec (self->cb_object);
1868 SvREFCNT_dec (self->incr_text); 2367 SvREFCNT_dec (self->incr_text);
1869 2368
1870PROTOTYPES: ENABLE 2369PROTOTYPES: ENABLE
1871 2370
1872void encode_json (SV *scalar) 2371void encode_json (SV *scalar)
1873 ALIAS:
1874 to_json_ = 0
1875 encode_json = F_UTF8
1876 PPCODE: 2372 PPCODE:
1877{ 2373{
1878 JSON json; 2374 JSON json;
1879 json_init (&json); 2375 json_init (&json);
1880 json.flags |= ix; 2376 json.flags |= F_UTF8;
1881 XPUSHs (encode_json (scalar, &json)); 2377 PUTBACK; scalar = encode_json (scalar, &json); SPAGAIN;
2378 XPUSHs (scalar);
1882} 2379}
1883 2380
1884void decode_json (SV *jsonstr) 2381void decode_json (SV *jsonstr)
1885 ALIAS:
1886 from_json_ = 0
1887 decode_json = F_UTF8
1888 PPCODE: 2382 PPCODE:
1889{ 2383{
1890 JSON json; 2384 JSON json;
1891 json_init (&json); 2385 json_init (&json);
1892 json.flags |= ix; 2386 json.flags |= F_UTF8;
1893 XPUSHs (decode_json (jsonstr, &json, 0)); 2387 PUTBACK; jsonstr = decode_json (jsonstr, &json, 0); SPAGAIN;
2388 XPUSHs (jsonstr);
1894} 2389}
1895 2390
1896

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines