ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/JSON-XS/XS.xs
(Generate patch)

Comparing JSON-XS/XS.xs (file contents):
Revision 1.62 by root, Sun Aug 26 22:27:32 2007 UTC vs.
Revision 1.112 by root, Thu Aug 11 17:06:53 2011 UTC

… …
4 4
5#include <assert.h> 5#include <assert.h>
6#include <string.h> 6#include <string.h>
7#include <stdlib.h> 7#include <stdlib.h>
8#include <stdio.h> 8#include <stdio.h>
9#include <limits.h>
9#include <float.h> 10#include <float.h>
10 11
11#if defined(__BORLANDC__) || defined(_MSC_VER) 12#if defined(__BORLANDC__) || defined(_MSC_VER)
12# define snprintf _snprintf // C compilers have this in stdio.h 13# define snprintf _snprintf // C compilers have this in stdio.h
13#endif 14#endif
14 15
15// some old perls do not have this, try to make it work, no 16// some old perls do not have this, try to make it work, no
16// guarentees, though. if it breaks, you get to keep the pieces. 17// guarantees, though. if it breaks, you get to keep the pieces.
17#ifndef UTF8_MAXBYTES 18#ifndef UTF8_MAXBYTES
18# define UTF8_MAXBYTES 13 19# define UTF8_MAXBYTES 13
19#endif 20#endif
21
22// three extra for rounding, sign, and end of string
23#define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 3)
20 24
21#define F_ASCII 0x00000001UL 25#define F_ASCII 0x00000001UL
22#define F_LATIN1 0x00000002UL 26#define F_LATIN1 0x00000002UL
23#define F_UTF8 0x00000004UL 27#define F_UTF8 0x00000004UL
24#define F_INDENT 0x00000008UL 28#define F_INDENT 0x00000008UL
… …
27#define F_SPACE_AFTER 0x00000040UL 31#define F_SPACE_AFTER 0x00000040UL
28#define F_ALLOW_NONREF 0x00000100UL 32#define F_ALLOW_NONREF 0x00000100UL
29#define F_SHRINK 0x00000200UL 33#define F_SHRINK 0x00000200UL
30#define F_ALLOW_BLESSED 0x00000400UL 34#define F_ALLOW_BLESSED 0x00000400UL
31#define F_CONV_BLESSED 0x00000800UL 35#define F_CONV_BLESSED 0x00000800UL
32#define F_MAXDEPTH 0xf8000000UL
33#define S_MAXDEPTH 27
34#define F_MAXSIZE 0x01f00000UL 36#define F_RELAXED 0x00001000UL
35#define S_MAXSIZE 20 37#define F_ALLOW_UNKNOWN 0x00002000UL
36#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing 38#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
37 39
38#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
39#define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE ))
40
41#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 40#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
42#define F_DEFAULT (9UL << S_MAXDEPTH)
43 41
44#define INIT_SIZE 32 // initial scalar size to be allocated 42#define INIT_SIZE 32 // initial scalar size to be allocated
45#define INDENT_STEP 3 // spaces per indentation level 43#define INDENT_STEP 3 // spaces per indentation level
46 44
47#define SHORT_STRING_LEN 16384 // special-case strings of up to this size 45#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
48 46
47#define DECODE_WANTS_OCTETS(json) ((json)->flags & F_UTF8)
48
49#define SB do { 49#define SB do {
50#define SE } while (0) 50#define SE } while (0)
51 51
52#if __GNUC__ >= 3 52#if __GNUC__ >= 3
53# define expect(expr,value) __builtin_expect ((expr),(value)) 53# define expect(expr,value) __builtin_expect ((expr), (value))
54# define inline inline 54# define INLINE static inline
55#else 55#else
56# define expect(expr,value) (expr) 56# define expect(expr,value) (expr)
57# define inline static 57# define INLINE static
58#endif 58#endif
59 59
60#define expect_false(expr) expect ((expr) != 0, 0) 60#define expect_false(expr) expect ((expr) != 0, 0)
61#define expect_true(expr) expect ((expr) != 0, 1) 61#define expect_true(expr) expect ((expr) != 0, 1)
62
63#define IN_RANGE_INC(type,val,beg,end) \
64 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \
65 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
66
67#define ERR_NESTING_EXCEEDED "json text or perl structure exceeds maximum nesting level (max_depth set too low?)"
62 68
63#ifdef USE_ITHREADS 69#ifdef USE_ITHREADS
64# define JSON_SLOW 1 70# define JSON_SLOW 1
65# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1)) 71# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1))
66#else 72#else
… …
69#endif 75#endif
70 76
71static HV *json_stash, *json_boolean_stash; // JSON::XS:: 77static HV *json_stash, *json_boolean_stash; // JSON::XS::
72static SV *json_true, *json_false; 78static SV *json_true, *json_false;
73 79
80enum {
81 INCR_M_WS = 0, // initial whitespace skipping, must be 0
82 INCR_M_STR, // inside string
83 INCR_M_BS, // inside backslash
84 INCR_M_C0, // inside comment in initial whitespace sequence
85 INCR_M_C1, // inside comment in other places
86 INCR_M_JSON // outside anything, count nesting
87};
88
89#define INCR_DONE(json) ((json)->incr_nest <= 0 && (json)->incr_mode == INCR_M_JSON)
90
74typedef struct { 91typedef struct {
75 U32 flags; 92 U32 flags;
93 U32 max_depth;
94 STRLEN max_size;
95
76 SV *cb_object; 96 SV *cb_object;
77 HV *cb_sk_object; 97 HV *cb_sk_object;
98
99 // for the incremental parser
100 SV *incr_text; // the source text so far
101 STRLEN incr_pos; // the current offset into the text
102 int incr_nest; // {[]}-nesting level
103 unsigned char incr_mode;
78} JSON; 104} JSON;
105
106INLINE void
107json_init (JSON *json)
108{
109 Zero (json, 1, JSON);
110 json->max_depth = 512;
111}
79 112
80///////////////////////////////////////////////////////////////////////////// 113/////////////////////////////////////////////////////////////////////////////
81// utility functions 114// utility functions
82 115
83inline void 116INLINE SV *
117get_bool (const char *name)
118{
119 SV *sv = get_sv (name, 1);
120
121 SvREADONLY_on (sv);
122 SvREADONLY_on (SvRV (sv));
123
124 return sv;
125}
126
127INLINE void
84shrink (SV *sv) 128shrink (SV *sv)
85{ 129{
86 sv_utf8_downgrade (sv, 1); 130 sv_utf8_downgrade (sv, 1);
131
87 if (SvLEN (sv) > SvCUR (sv) + 1) 132 if (SvLEN (sv) > SvCUR (sv) + 1)
88 { 133 {
89#ifdef SvPV_shrink_to_cur 134#ifdef SvPV_shrink_to_cur
90 SvPV_shrink_to_cur (sv); 135 SvPV_shrink_to_cur (sv);
91#elif defined (SvPV_renew) 136#elif defined (SvPV_renew)
… …
97// decode an utf-8 character and return it, or (UV)-1 in 142// decode an utf-8 character and return it, or (UV)-1 in
98// case of an error. 143// case of an error.
99// we special-case "safe" characters from U+80 .. U+7FF, 144// we special-case "safe" characters from U+80 .. U+7FF,
100// but use the very good perl function to parse anything else. 145// but use the very good perl function to parse anything else.
101// note that we never call this function for a ascii codepoints 146// note that we never call this function for a ascii codepoints
102inline UV 147INLINE UV
103decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) 148decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
104{ 149{
105 if (expect_false (s[0] > 0xdf || s[0] < 0xc2)) 150 if (expect_true (len >= 2
106 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); 151 && IN_RANGE_INC (char, s[0], 0xc2, 0xdf)
107 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) 152 && IN_RANGE_INC (char, s[1], 0x80, 0xbf)))
108 { 153 {
109 *clen = 2; 154 *clen = 2;
110 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 155 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
111 } 156 }
112 else 157 else
113 { 158 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
114 *clen = (STRLEN)-1; 159}
115 return (UV)-1; 160
161// likewise for encoding, also never called for ascii codepoints
162// this function takes advantage of this fact, although current gccs
163// seem to optimise the check for >= 0x80 away anyways
164INLINE unsigned char *
165encode_utf8 (unsigned char *s, UV ch)
166{
167 if (expect_false (ch < 0x000080))
168 *s++ = ch;
169 else if (expect_true (ch < 0x000800))
170 *s++ = 0xc0 | ( ch >> 6),
171 *s++ = 0x80 | ( ch & 0x3f);
172 else if ( ch < 0x010000)
173 *s++ = 0xe0 | ( ch >> 12),
174 *s++ = 0x80 | ((ch >> 6) & 0x3f),
175 *s++ = 0x80 | ( ch & 0x3f);
176 else if ( ch < 0x110000)
177 *s++ = 0xf0 | ( ch >> 18),
178 *s++ = 0x80 | ((ch >> 12) & 0x3f),
179 *s++ = 0x80 | ((ch >> 6) & 0x3f),
180 *s++ = 0x80 | ( ch & 0x3f);
181
182 return s;
183}
184
185// convert offset pointer to character index, sv must be string
186static STRLEN
187ptr_to_index (SV *sv, char *offset)
188{
189 return SvUTF8 (sv)
190 ? utf8_distance (offset, SvPVX (sv))
191 : offset - SvPVX (sv);
192}
193
194/////////////////////////////////////////////////////////////////////////////
195// fp hell
196
197// scan a group of digits, and a trailing exponent
198static void
199json_atof_scan1 (const char *s, NV *accum, int *expo, int postdp, int maxdepth)
200{
201 UV uaccum = 0;
202 int eaccum = 0;
203
204 // if we recurse too deep, skip all remaining digits
205 // to avoid a stack overflow attack
206 if (expect_false (--maxdepth <= 0))
207 while (((U8)*s - '0') < 10)
208 ++s;
209
210 for (;;)
116 } 211 {
117} 212 U8 dig = (U8)*s - '0';
118 213
214 if (expect_false (dig >= 10))
215 {
216 if (dig == (U8)((U8)'.' - (U8)'0'))
217 {
218 ++s;
219 json_atof_scan1 (s, accum, expo, 1, maxdepth);
220 }
221 else if ((dig | ' ') == 'e' - '0')
222 {
223 int exp2 = 0;
224 int neg = 0;
225
226 ++s;
227
228 if (*s == '-')
229 {
230 ++s;
231 neg = 1;
232 }
233 else if (*s == '+')
234 ++s;
235
236 while ((dig = (U8)*s - '0') < 10)
237 exp2 = exp2 * 10 + *s++ - '0';
238
239 *expo += neg ? -exp2 : exp2;
240 }
241
242 break;
243 }
244
245 ++s;
246
247 uaccum = uaccum * 10 + dig;
248 ++eaccum;
249
250 // if we have too many digits, then recurse for more
251 // we actually do this for rather few digits
252 if (uaccum >= (UV_MAX - 9) / 10)
253 {
254 if (postdp) *expo -= eaccum;
255 json_atof_scan1 (s, accum, expo, postdp, maxdepth);
256 if (postdp) *expo += eaccum;
257
258 break;
259 }
260 }
261
262 // this relies greatly on the quality of the pow ()
263 // implementation of the platform, but a good
264 // implementation is hard to beat.
265 // (IEEE 754 conformant ones are required to be exact)
266 if (postdp) *expo -= eaccum;
267 *accum += uaccum * Perl_pow (10., *expo);
268 *expo += eaccum;
269}
270
271static NV
272json_atof (const char *s)
273{
274 NV accum = 0.;
275 int expo = 0;
276 int neg = 0;
277
278 if (*s == '-')
279 {
280 ++s;
281 neg = 1;
282 }
283
284 // a recursion depth of ten gives us >>500 bits
285 json_atof_scan1 (s, &accum, &expo, 0, 10);
286
287 return neg ? -accum : accum;
288}
119///////////////////////////////////////////////////////////////////////////// 289/////////////////////////////////////////////////////////////////////////////
120// encoder 290// encoder
121 291
122// structure used for encoding JSON 292// structure used for encoding JSON
123typedef struct 293typedef struct
… …
125 char *cur; // SvPVX (sv) + current output position 295 char *cur; // SvPVX (sv) + current output position
126 char *end; // SvEND (sv) 296 char *end; // SvEND (sv)
127 SV *sv; // result scalar 297 SV *sv; // result scalar
128 JSON json; 298 JSON json;
129 U32 indent; // indentation level 299 U32 indent; // indentation level
130 U32 maxdepth; // max. indentation/recursion level 300 UV limit; // escape character values >= this value when encoding
131} enc_t; 301} enc_t;
132 302
133inline void 303INLINE void
134need (enc_t *enc, STRLEN len) 304need (enc_t *enc, STRLEN len)
135{ 305{
136 if (expect_false (enc->cur + len >= enc->end)) 306 if (expect_false (enc->cur + len >= enc->end))
137 { 307 {
138 STRLEN cur = enc->cur - SvPVX (enc->sv); 308 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
139 SvGROW (enc->sv, cur + len + 1); 309 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
140 enc->cur = SvPVX (enc->sv) + cur; 310 enc->cur = SvPVX (enc->sv) + cur;
141 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 311 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
142 } 312 }
143} 313}
144 314
145inline void 315INLINE void
146encode_ch (enc_t *enc, char ch) 316encode_ch (enc_t *enc, char ch)
147{ 317{
148 need (enc, 1); 318 need (enc, 1);
149 *enc->cur++ = ch; 319 *enc->cur++ = ch;
150} 320}
… …
204 { 374 {
205 uch = ch; 375 uch = ch;
206 clen = 1; 376 clen = 1;
207 } 377 }
208 378
209 if (uch > 0x10FFFFUL) 379 if (uch < 0x80/*0x20*/ || uch >= enc->limit)
210 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
211
212 if (uch < 0x80 || enc->json.flags & F_ASCII || (enc->json.flags & F_LATIN1 && uch > 0xFF))
213 { 380 {
214 if (uch > 0xFFFFUL) 381 if (uch >= 0x10000UL)
215 { 382 {
383 if (uch >= 0x110000UL)
384 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
385
216 need (enc, len += 11); 386 need (enc, len += 11);
217 sprintf (enc->cur, "\\u%04x\\u%04x", 387 sprintf (enc->cur, "\\u%04x\\u%04x",
218 (int)((uch - 0x10000) / 0x400 + 0xD800), 388 (int)((uch - 0x10000) / 0x400 + 0xD800),
219 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 389 (int)((uch - 0x10000) % 0x400 + 0xDC00));
220 enc->cur += 12; 390 enc->cur += 12;
221 } 391 }
222 else 392 else
223 { 393 {
224 static char hexdigit [16] = "0123456789abcdef";
225 need (enc, len += 5); 394 need (enc, len += 5);
226 *enc->cur++ = '\\'; 395 *enc->cur++ = '\\';
227 *enc->cur++ = 'u'; 396 *enc->cur++ = 'u';
228 *enc->cur++ = hexdigit [ uch >> 12 ]; 397 *enc->cur++ = PL_hexdigit [ uch >> 12 ];
229 *enc->cur++ = hexdigit [(uch >> 8) & 15]; 398 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
230 *enc->cur++ = hexdigit [(uch >> 4) & 15]; 399 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
231 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 400 *enc->cur++ = PL_hexdigit [(uch >> 0) & 15];
232 } 401 }
233 402
234 str += clen; 403 str += clen;
235 } 404 }
236 else if (enc->json.flags & F_LATIN1) 405 else if (enc->json.flags & F_LATIN1)
… …
248 while (--clen); 417 while (--clen);
249 } 418 }
250 else 419 else
251 { 420 {
252 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed 421 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
253 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 422 enc->cur = encode_utf8 (enc->cur, uch);
254 ++str; 423 ++str;
255 } 424 }
256 } 425 }
257 } 426 }
258 } 427 }
259 428
260 --len; 429 --len;
261 } 430 }
262} 431}
263 432
264inline void 433INLINE void
265encode_indent (enc_t *enc) 434encode_indent (enc_t *enc)
266{ 435{
267 if (enc->json.flags & F_INDENT) 436 if (enc->json.flags & F_INDENT)
268 { 437 {
269 int spaces = enc->indent * INDENT_STEP; 438 int spaces = enc->indent * INDENT_STEP;
… …
272 memset (enc->cur, ' ', spaces); 441 memset (enc->cur, ' ', spaces);
273 enc->cur += spaces; 442 enc->cur += spaces;
274 } 443 }
275} 444}
276 445
277inline void 446INLINE void
278encode_space (enc_t *enc) 447encode_space (enc_t *enc)
279{ 448{
280 need (enc, 1); 449 need (enc, 1);
281 encode_ch (enc, ' '); 450 encode_ch (enc, ' ');
282} 451}
283 452
284inline void 453INLINE void
285encode_nl (enc_t *enc) 454encode_nl (enc_t *enc)
286{ 455{
287 if (enc->json.flags & F_INDENT) 456 if (enc->json.flags & F_INDENT)
288 { 457 {
289 need (enc, 1); 458 need (enc, 1);
290 encode_ch (enc, '\n'); 459 encode_ch (enc, '\n');
291 } 460 }
292} 461}
293 462
294inline void 463INLINE void
295encode_comma (enc_t *enc) 464encode_comma (enc_t *enc)
296{ 465{
297 encode_ch (enc, ','); 466 encode_ch (enc, ',');
298 467
299 if (enc->json.flags & F_INDENT) 468 if (enc->json.flags & F_INDENT)
… …
307static void 476static void
308encode_av (enc_t *enc, AV *av) 477encode_av (enc_t *enc, AV *av)
309{ 478{
310 int i, len = av_len (av); 479 int i, len = av_len (av);
311 480
312 if (enc->indent >= enc->maxdepth) 481 if (enc->indent >= enc->json.max_depth)
313 croak ("data structure too deep (hit recursion limit)"); 482 croak (ERR_NESTING_EXCEEDED);
314 483
315 encode_ch (enc, '['); encode_nl (enc); 484 encode_ch (enc, '[');
316 ++enc->indent; 485
486 if (len >= 0)
487 {
488 encode_nl (enc); ++enc->indent;
317 489
318 for (i = 0; i <= len; ++i) 490 for (i = 0; i <= len; ++i)
319 { 491 {
320 SV **svp = av_fetch (av, i, 0); 492 SV **svp = av_fetch (av, i, 0);
321 493
322 encode_indent (enc); 494 encode_indent (enc);
323 495
324 if (svp) 496 if (svp)
325 encode_sv (enc, *svp); 497 encode_sv (enc, *svp);
326 else 498 else
327 encode_str (enc, "null", 4, 0); 499 encode_str (enc, "null", 4, 0);
328 500
329 if (i < len) 501 if (i < len)
330 encode_comma (enc); 502 encode_comma (enc);
331 } 503 }
332 504
505 encode_nl (enc); --enc->indent; encode_indent (enc);
506 }
507
333 encode_nl (enc); 508 encode_ch (enc, ']');
334
335 --enc->indent;
336 encode_indent (enc); encode_ch (enc, ']');
337} 509}
338 510
339static void 511static void
340encode_hk (enc_t *enc, HE *he) 512encode_hk (enc_t *enc, HE *he)
341{ 513{
… …
389 561
390static void 562static void
391encode_hv (enc_t *enc, HV *hv) 563encode_hv (enc_t *enc, HV *hv)
392{ 564{
393 HE *he; 565 HE *he;
394 int count;
395 566
396 if (enc->indent >= enc->maxdepth) 567 if (enc->indent >= enc->json.max_depth)
397 croak ("data structure too deep (hit recursion limit)"); 568 croak (ERR_NESTING_EXCEEDED);
398 569
399 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 570 encode_ch (enc, '{');
400 571
401 // for canonical output we have to sort by keys first 572 // for canonical output we have to sort by keys first
402 // actually, this is mostly due to the stupid so-called 573 // actually, this is mostly due to the stupid so-called
403 // security workaround added somewhere in 5.8.x. 574 // security workaround added somewhere in 5.8.x
404 // that randomises hash orderings 575 // that randomises hash orderings
405 if (enc->json.flags & F_CANONICAL) 576 if (enc->json.flags & F_CANONICAL && !SvRMAGICAL (hv))
406 { 577 {
407 int count = hv_iterinit (hv); 578 int count = hv_iterinit (hv);
408 579
409 if (SvMAGICAL (hv)) 580 if (SvMAGICAL (hv))
410 { 581 {
… …
457 628
458 FREETMPS; 629 FREETMPS;
459 LEAVE; 630 LEAVE;
460 } 631 }
461 632
633 encode_nl (enc); ++enc->indent;
634
462 while (count--) 635 while (count--)
463 { 636 {
464 encode_indent (enc); 637 encode_indent (enc);
465 he = hes [count]; 638 he = hes [count];
466 encode_hk (enc, he); 639 encode_hk (enc, he);
467 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he)); 640 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
468 641
469 if (count) 642 if (count)
470 encode_comma (enc); 643 encode_comma (enc);
471 } 644 }
645
646 encode_nl (enc); --enc->indent; encode_indent (enc);
472 } 647 }
473 } 648 }
474 else 649 else
475 { 650 {
476 if (hv_iterinit (hv) || SvMAGICAL (hv)) 651 if (hv_iterinit (hv) || SvMAGICAL (hv))
477 if ((he = hv_iternext (hv))) 652 if ((he = hv_iternext (hv)))
653 {
654 encode_nl (enc); ++enc->indent;
655
478 for (;;) 656 for (;;)
479 { 657 {
480 encode_indent (enc); 658 encode_indent (enc);
481 encode_hk (enc, he); 659 encode_hk (enc, he);
482 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he)); 660 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
483 661
484 if (!(he = hv_iternext (hv))) 662 if (!(he = hv_iternext (hv)))
485 break; 663 break;
486 664
487 encode_comma (enc); 665 encode_comma (enc);
488 } 666 }
489 }
490 667
668 encode_nl (enc); --enc->indent; encode_indent (enc);
669 }
670 }
671
491 encode_nl (enc); 672 encode_ch (enc, '}');
492
493 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
494} 673}
495 674
496// encode objects, arrays and special \0=false and \1=true values. 675// encode objects, arrays and special \0=false and \1=true values.
497static void 676static void
498encode_rv (enc_t *enc, SV *sv) 677encode_rv (enc_t *enc, SV *sv)
… …
575 754
576 if (len == 1 && *pv == '1') 755 if (len == 1 && *pv == '1')
577 encode_str (enc, "true", 4, 0); 756 encode_str (enc, "true", 4, 0);
578 else if (len == 1 && *pv == '0') 757 else if (len == 1 && *pv == '0')
579 encode_str (enc, "false", 5, 0); 758 encode_str (enc, "false", 5, 0);
759 else if (enc->json.flags & F_ALLOW_UNKNOWN)
760 encode_str (enc, "null", 4, 0);
580 else 761 else
581 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", 762 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
582 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 763 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
583 } 764 }
765 else if (enc->json.flags & F_ALLOW_UNKNOWN)
766 encode_str (enc, "null", 4, 0);
584 else 767 else
585 croak ("encountered %s, but JSON can only represent references to arrays or hashes", 768 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
586 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 769 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
587} 770}
588 771
… …
606 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 789 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
607 enc->cur += strlen (enc->cur); 790 enc->cur += strlen (enc->cur);
608 } 791 }
609 else if (SvIOKp (sv)) 792 else if (SvIOKp (sv))
610 { 793 {
611 // we assume we can always read an IV as a UV 794 // we assume we can always read an IV as a UV and vice versa
612 if (SvUV (sv) & ~(UV)0x7fff) 795 // we assume two's complement
613 { 796 // we assume no aliasing issues in the union
614 // large integer, use the (rather slow) snprintf way. 797 if (SvIsUV (sv) ? SvUVX (sv) <= 59000
615 need (enc, sizeof (UV) * 3); 798 : SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000)
616 enc->cur +=
617 SvIsUV(sv)
618 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
619 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
620 }
621 else
622 { 799 {
623 // optimise the "small number case" 800 // optimise the "small number case"
624 // code will likely be branchless and use only a single multiplication 801 // code will likely be branchless and use only a single multiplication
802 // works for numbers up to 59074
625 I32 i = SvIV (sv); 803 I32 i = SvIVX (sv);
626 U32 u; 804 U32 u;
627 char digit, nz = 0; 805 char digit, nz = 0;
628 806
629 need (enc, 6); 807 need (enc, 6);
630 808
… …
636 814
637 // now output digit by digit, each time masking out the integer part 815 // now output digit by digit, each time masking out the integer part
638 // and multiplying by 5 while moving the decimal point one to the right, 816 // and multiplying by 5 while moving the decimal point one to the right,
639 // resulting in a net multiplication by 10. 817 // resulting in a net multiplication by 10.
640 // we always write the digit to memory but conditionally increment 818 // we always write the digit to memory but conditionally increment
641 // the pointer, to ease the usage of conditional move instructions. 819 // the pointer, to enable the use of conditional move instructions.
642 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5; 820 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffffUL) * 5;
643 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5; 821 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffffUL) * 5;
644 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5; 822 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffffUL) * 5;
645 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5; 823 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffffUL) * 5;
646 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0' 824 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
647 } 825 }
826 else
827 {
828 // large integer, use the (rather slow) snprintf way.
829 need (enc, IVUV_MAXCHARS);
830 enc->cur +=
831 SvIsUV(sv)
832 ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))
833 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
834 }
648 } 835 }
649 else if (SvROK (sv)) 836 else if (SvROK (sv))
650 encode_rv (enc, SvRV (sv)); 837 encode_rv (enc, SvRV (sv));
651 else if (!SvOK (sv)) 838 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN)
652 encode_str (enc, "null", 4, 0); 839 encode_str (enc, "null", 4, 0);
653 else 840 else
654 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 841 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
655 SvPV_nolen (sv), SvFLAGS (sv)); 842 SvPV_nolen (sv), SvFLAGS (sv));
656} 843}
… …
666 enc.json = *json; 853 enc.json = *json;
667 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 854 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
668 enc.cur = SvPVX (enc.sv); 855 enc.cur = SvPVX (enc.sv);
669 enc.end = SvEND (enc.sv); 856 enc.end = SvEND (enc.sv);
670 enc.indent = 0; 857 enc.indent = 0;
671 enc.maxdepth = DEC_DEPTH (enc.json.flags); 858 enc.limit = enc.json.flags & F_ASCII ? 0x000080UL
859 : enc.json.flags & F_LATIN1 ? 0x000100UL
860 : 0x110000UL;
672 861
673 SvPOK_only (enc.sv); 862 SvPOK_only (enc.sv);
674 encode_sv (&enc, scalar); 863 encode_sv (&enc, scalar);
864 encode_nl (&enc);
675 865
676 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 866 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
677 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 867 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
678 868
679 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8))) 869 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
… …
697 JSON json; 887 JSON json;
698 U32 depth; // recursion depth 888 U32 depth; // recursion depth
699 U32 maxdepth; // recursion depth limit 889 U32 maxdepth; // recursion depth limit
700} dec_t; 890} dec_t;
701 891
702inline void 892INLINE void
893decode_comment (dec_t *dec)
894{
895 // only '#'-style comments allowed a.t.m.
896
897 while (*dec->cur && *dec->cur != 0x0a && *dec->cur != 0x0d)
898 ++dec->cur;
899}
900
901INLINE void
703decode_ws (dec_t *dec) 902decode_ws (dec_t *dec)
704{ 903{
705 for (;;) 904 for (;;)
706 { 905 {
707 char ch = *dec->cur; 906 char ch = *dec->cur;
708 907
709 if (ch > 0x20 908 if (ch > 0x20)
909 {
910 if (expect_false (ch == '#'))
911 {
912 if (dec->json.flags & F_RELAXED)
913 decode_comment (dec);
914 else
915 break;
916 }
917 else
918 break;
919 }
710 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) 920 else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)
711 break; 921 break; // parse error, but let higher level handle it, gives better error messages
712 922
713 ++dec->cur; 923 ++dec->cur;
714 } 924 }
715} 925}
716 926
… …
720 if (*dec->cur != ch) \ 930 if (*dec->cur != ch) \
721 ERR (# ch " expected"); \ 931 ERR (# ch " expected"); \
722 ++dec->cur; \ 932 ++dec->cur; \
723 SE 933 SE
724 934
725#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)") 935#define DEC_INC_DEPTH if (++dec->depth > dec->json.max_depth) ERR (ERR_NESTING_EXCEEDED)
726#define DEC_DEC_DEPTH --dec->depth 936#define DEC_DEC_DEPTH --dec->depth
727 937
728static SV *decode_sv (dec_t *dec); 938static SV *decode_sv (dec_t *dec);
729 939
730static signed char decode_hexdigit[256]; 940static signed char decode_hexdigit[256];
… …
822 1032
823 if (hi >= 0x80) 1033 if (hi >= 0x80)
824 { 1034 {
825 utf8 = 1; 1035 utf8 = 1;
826 1036
827 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0); 1037 cur = encode_utf8 (cur, hi);
828 } 1038 }
829 else 1039 else
830 *cur++ = hi; 1040 *cur++ = hi;
831 } 1041 }
832 break; 1042 break;
… …
834 default: 1044 default:
835 --dec_cur; 1045 --dec_cur;
836 ERR ("illegal backslash escape sequence in string"); 1046 ERR ("illegal backslash escape sequence in string");
837 } 1047 }
838 } 1048 }
839 else if (expect_true (ch >= 0x20 && ch <= 0x7f)) 1049 else if (expect_true (ch >= 0x20 && ch < 0x80))
840 *cur++ = ch; 1050 *cur++ = ch;
841 else if (ch >= 0x80) 1051 else if (ch >= 0x80)
842 { 1052 {
843 STRLEN clen; 1053 STRLEN clen;
844 UV uch;
845 1054
846 --dec_cur; 1055 --dec_cur;
847 1056
848 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen); 1057 decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
849 if (clen == (STRLEN)-1) 1058 if (clen == (STRLEN)-1)
850 ERR ("malformed UTF-8 character in JSON string"); 1059 ERR ("malformed UTF-8 character in JSON string");
851 1060
852 do 1061 do
853 *cur++ = *dec_cur++; 1062 *cur++ = *dec_cur++;
… …
870 { 1079 {
871 STRLEN len = cur - buf; 1080 STRLEN len = cur - buf;
872 1081
873 if (sv) 1082 if (sv)
874 { 1083 {
875 SvGROW (sv, SvCUR (sv) + len + 1); 1084 STRLEN cur = SvCUR (sv);
1085
1086 if (SvLEN (sv) <= cur + len)
1087 SvGROW (sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
1088
876 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 1089 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
877 SvCUR_set (sv, SvCUR (sv) + len); 1090 SvCUR_set (sv, SvCUR (sv) + len);
878 } 1091 }
879 else 1092 else
880 sv = newSVpvn (buf, len); 1093 sv = newSVpvn (buf, len);
… …
967 1180
968 if (!is_nv) 1181 if (!is_nv)
969 { 1182 {
970 int len = dec->cur - start; 1183 int len = dec->cur - start;
971 1184
972 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so 1185 // special case the rather common 1..5-digit-int case
973 if (*start == '-') 1186 if (*start == '-')
974 switch (len) 1187 switch (len)
975 { 1188 {
976 case 2: return newSViv (-( start [1] - '0' * 1)); 1189 case 2: return newSViv (-(IV)( start [1] - '0' * 1));
977 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); 1190 case 3: return newSViv (-(IV)( start [1] * 10 + start [2] - '0' * 11));
978 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); 1191 case 4: return newSViv (-(IV)( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
979 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); 1192 case 5: return newSViv (-(IV)( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
1193 case 6: return newSViv (-(IV)(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
980 } 1194 }
981 else 1195 else
982 switch (len) 1196 switch (len)
983 { 1197 {
984 case 1: return newSViv ( start [0] - '0' * 1); 1198 case 1: return newSViv ( start [0] - '0' * 1);
985 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); 1199 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
986 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); 1200 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
987 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); 1201 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
1202 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
988 } 1203 }
989 1204
990 { 1205 {
991 UV uv; 1206 UV uv;
992 int numtype = grok_number (start, len, &uv); 1207 int numtype = grok_number (start, len, &uv);
… …
1001 } 1216 }
1002 1217
1003 len -= *start == '-' ? 1 : 0; 1218 len -= *start == '-' ? 1 : 0;
1004 1219
1005 // does not fit into IV or UV, try NV 1220 // does not fit into IV or UV, try NV
1006 if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len) 1221 if (len <= NV_DIG)
1007 #if defined (LDBL_DIG)
1008 || (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len)
1009 #endif
1010 )
1011 // fits into NV without loss of precision 1222 // fits into NV without loss of precision
1012 return newSVnv (Atof (start)); 1223 return newSVnv (json_atof (start));
1013 1224
1014 // everything else fails, convert it to a string 1225 // everything else fails, convert it to a string
1015 return newSVpvn (start, dec->cur - start); 1226 return newSVpvn (start, dec->cur - start);
1016 } 1227 }
1017 1228
1018 // loss of precision here 1229 // loss of precision here
1019 return newSVnv (Atof (start)); 1230 return newSVnv (json_atof (start));
1020 1231
1021fail: 1232fail:
1022 return 0; 1233 return 0;
1023} 1234}
1024 1235
… …
1053 1264
1054 if (*dec->cur != ',') 1265 if (*dec->cur != ',')
1055 ERR (", or ] expected while parsing array"); 1266 ERR (", or ] expected while parsing array");
1056 1267
1057 ++dec->cur; 1268 ++dec->cur;
1269
1270 decode_ws (dec);
1271
1272 if (*dec->cur == ']' && dec->json.flags & F_RELAXED)
1273 {
1274 ++dec->cur;
1275 break;
1276 }
1058 } 1277 }
1059 1278
1060 DEC_DEC_DEPTH; 1279 DEC_DEC_DEPTH;
1061 return newRV_noinc ((SV *)av); 1280 return newRV_noinc ((SV *)av);
1062 1281
… …
1078 if (*dec->cur == '}') 1297 if (*dec->cur == '}')
1079 ++dec->cur; 1298 ++dec->cur;
1080 else 1299 else
1081 for (;;) 1300 for (;;)
1082 { 1301 {
1083 decode_ws (dec); EXPECT_CH ('"'); 1302 EXPECT_CH ('"');
1084 1303
1085 // heuristic: assume that 1304 // heuristic: assume that
1086 // a) decode_str + hv_store_ent are abysmally slow. 1305 // a) decode_str + hv_store_ent are abysmally slow.
1087 // b) most hash keys are short, simple ascii text. 1306 // b) most hash keys are short, simple ascii text.
1088 // => try to "fast-match" such strings to avoid 1307 // => try to "fast-match" such strings to avoid
… …
1092 char *p = dec->cur; 1311 char *p = dec->cur;
1093 char *e = p + 24; // only try up to 24 bytes 1312 char *e = p + 24; // only try up to 24 bytes
1094 1313
1095 for (;;) 1314 for (;;)
1096 { 1315 {
1097 // the >= 0x80 is true on most architectures 1316 // the >= 0x80 is false on most architectures
1098 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\') 1317 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1099 { 1318 {
1100 // slow path, back up and use decode_str 1319 // slow path, back up and use decode_str
1101 SV *key = decode_str (dec); 1320 SV *key = decode_str (dec);
1102 if (!key) 1321 if (!key)
1103 goto fail; 1322 goto fail;
1104 1323
1105 decode_ws (dec); EXPECT_CH (':'); 1324 decode_ws (dec); EXPECT_CH (':');
1106 1325
1326 decode_ws (dec);
1107 value = decode_sv (dec); 1327 value = decode_sv (dec);
1108 if (!value) 1328 if (!value)
1109 { 1329 {
1110 SvREFCNT_dec (key); 1330 SvREFCNT_dec (key);
1111 goto fail; 1331 goto fail;
… …
1123 int len = p - key; 1343 int len = p - key;
1124 dec->cur = p + 1; 1344 dec->cur = p + 1;
1125 1345
1126 decode_ws (dec); EXPECT_CH (':'); 1346 decode_ws (dec); EXPECT_CH (':');
1127 1347
1348 decode_ws (dec);
1128 value = decode_sv (dec); 1349 value = decode_sv (dec);
1129 if (!value) 1350 if (!value)
1130 goto fail; 1351 goto fail;
1131 1352
1132 hv_store (hv, key, len, value, 0); 1353 hv_store (hv, key, len, value, 0);
… …
1148 1369
1149 if (*dec->cur != ',') 1370 if (*dec->cur != ',')
1150 ERR (", or } expected while parsing object/hash"); 1371 ERR (", or } expected while parsing object/hash");
1151 1372
1152 ++dec->cur; 1373 ++dec->cur;
1374
1375 decode_ws (dec);
1376
1377 if (*dec->cur == '}' && dec->json.flags & F_RELAXED)
1378 {
1379 ++dec->cur;
1380 break;
1381 }
1153 } 1382 }
1154 1383
1155 DEC_DEC_DEPTH; 1384 DEC_DEC_DEPTH;
1156 sv = newRV_noinc ((SV *)hv); 1385 sv = newRV_noinc ((SV *)hv);
1157 1386
… …
1175 dSP; 1404 dSP;
1176 int count; 1405 int count;
1177 1406
1178 ENTER; SAVETMPS; PUSHMARK (SP); 1407 ENTER; SAVETMPS; PUSHMARK (SP);
1179 XPUSHs (HeVAL (he)); 1408 XPUSHs (HeVAL (he));
1409 sv_2mortal (sv);
1180 1410
1181 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN; 1411 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1182 1412
1183 if (count == 1) 1413 if (count == 1)
1184 { 1414 {
1185 sv = newSVsv (POPs); 1415 sv = newSVsv (POPs);
1186 FREETMPS; LEAVE; 1416 FREETMPS; LEAVE;
1187 return sv; 1417 return sv;
1188 } 1418 }
1189 1419
1420 SvREFCNT_inc (sv);
1190 FREETMPS; LEAVE; 1421 FREETMPS; LEAVE;
1191 } 1422 }
1192 } 1423 }
1193 1424
1194 if (dec->json.cb_object) 1425 if (dec->json.cb_object)
… …
1222} 1453}
1223 1454
1224static SV * 1455static SV *
1225decode_sv (dec_t *dec) 1456decode_sv (dec_t *dec)
1226{ 1457{
1227 decode_ws (dec);
1228
1229 // the beauty of JSON: you need exactly one character lookahead 1458 // the beauty of JSON: you need exactly one character lookahead
1230 // to parse anything. 1459 // to parse everything.
1231 switch (*dec->cur) 1460 switch (*dec->cur)
1232 { 1461 {
1233 case '"': ++dec->cur; return decode_str (dec); 1462 case '"': ++dec->cur; return decode_str (dec);
1234 case '[': ++dec->cur; return decode_av (dec); 1463 case '[': ++dec->cur; return decode_av (dec);
1235 case '{': ++dec->cur; return decode_hv (dec); 1464 case '{': ++dec->cur; return decode_hv (dec);
1236 1465
1237 case '-': 1466 case '-':
1238 case '0': case '1': case '2': case '3': case '4': 1467 case '0': case '1': case '2': case '3': case '4':
1239 case '5': case '6': case '7': case '8': case '9': 1468 case '5': case '6': case '7': case '8': case '9':
1240 return decode_num (dec); 1469 return decode_num (dec);
… …
1242 case 't': 1471 case 't':
1243 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1472 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1244 { 1473 {
1245 dec->cur += 4; 1474 dec->cur += 4;
1246#if JSON_SLOW 1475#if JSON_SLOW
1247 json_true = get_sv ("JSON::XS::true", 1); SvREADONLY_on (json_true); 1476 json_true = get_bool ("JSON::XS::true");
1248#endif 1477#endif
1249 return SvREFCNT_inc (json_true); 1478 return newSVsv (json_true);
1250 } 1479 }
1251 else 1480 else
1252 ERR ("'true' expected"); 1481 ERR ("'true' expected");
1253 1482
1254 break; 1483 break;
… …
1256 case 'f': 1485 case 'f':
1257 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1486 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1258 { 1487 {
1259 dec->cur += 5; 1488 dec->cur += 5;
1260#if JSON_SLOW 1489#if JSON_SLOW
1261 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); 1490 json_false = get_bool ("JSON::XS::false");
1262#endif 1491#endif
1263 return SvREFCNT_inc (json_false); 1492 return newSVsv (json_false);
1264 } 1493 }
1265 else 1494 else
1266 ERR ("'false' expected"); 1495 ERR ("'false' expected");
1267 1496
1268 break; 1497 break;
… …
1286fail: 1515fail:
1287 return 0; 1516 return 0;
1288} 1517}
1289 1518
1290static SV * 1519static SV *
1291decode_json (SV *string, JSON *json, UV *offset_return) 1520decode_json (SV *string, JSON *json, char **offset_return)
1292{ 1521{
1293 dec_t dec; 1522 dec_t dec;
1294 UV offset;
1295 SV *sv; 1523 SV *sv;
1296 1524
1525 /* work around bugs in 5.10 where manipulating magic values
1526 * will perl ignore the magic in subsequent accesses.
1527 * also make a copy of non-PV values, to get them into a clean
1528 * state (SvPV should do that, but it's buggy, see below).
1529 */
1297 SvGETMAGIC (string); 1530 /*SvGETMAGIC (string);*/
1531 if (SvMAGICAL (string) || !SvPOK (string))
1532 string = sv_2mortal (newSVsv (string));
1533
1298 SvUPGRADE (string, SVt_PV); 1534 SvUPGRADE (string, SVt_PV);
1299 1535
1300 if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags)) 1536 /* work around a bug in perl 5.10, which causes SvCUR to fail an
1537 * assertion with -DDEBUGGING, although SvCUR is documented to
1538 * return the xpv_cur field which certainly exists after upgrading.
1539 * according to nicholas clark, calling SvPOK fixes this.
1540 * But it doesn't fix it, so try another workaround, call SvPV_nolen
1541 * and hope for the best.
1542 * Damnit, SvPV_nolen still trips over yet another assertion. This
1543 * assertion business is seriously broken, try yet another workaround
1544 * for the broken -DDEBUGGING.
1545 */
1546 {
1547#ifdef DEBUGGING
1548 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
1549#else
1550 STRLEN offset = SvCUR (string);
1551#endif
1552
1553 if (offset > json->max_size && json->max_size)
1301 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", 1554 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1302 (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags)); 1555 (unsigned long)SvCUR (string), (unsigned long)json->max_size);
1556 }
1303 1557
1304 if (json->flags & F_UTF8) 1558 if (DECODE_WANTS_OCTETS (json))
1305 sv_utf8_downgrade (string, 0); 1559 sv_utf8_downgrade (string, 0);
1306 else 1560 else
1307 sv_utf8_upgrade (string); 1561 sv_utf8_upgrade (string);
1308 1562
1309 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1563 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
1310 1564
1311 dec.json = *json; 1565 dec.json = *json;
1312 dec.cur = SvPVX (string); 1566 dec.cur = SvPVX (string);
1313 dec.end = SvEND (string); 1567 dec.end = SvEND (string);
1314 dec.err = 0; 1568 dec.err = 0;
1315 dec.depth = 0; 1569 dec.depth = 0;
1316 dec.maxdepth = DEC_DEPTH (dec.json.flags);
1317 1570
1318 if (dec.json.cb_object || dec.json.cb_sk_object) 1571 if (dec.json.cb_object || dec.json.cb_sk_object)
1319 dec.json.flags |= F_HOOK; 1572 dec.json.flags |= F_HOOK;
1320 1573
1321 *dec.end = 0; // this should basically be a nop, too, but make sure it's there 1574 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1575
1576 decode_ws (&dec);
1322 sv = decode_sv (&dec); 1577 sv = decode_sv (&dec);
1578
1579 if (offset_return)
1580 *offset_return = dec.cur;
1323 1581
1324 if (!(offset_return || !sv)) 1582 if (!(offset_return || !sv))
1325 { 1583 {
1326 // check for trailing garbage 1584 // check for trailing garbage
1327 decode_ws (&dec); 1585 decode_ws (&dec);
… …
1330 { 1588 {
1331 dec.err = "garbage after JSON object"; 1589 dec.err = "garbage after JSON object";
1332 SvREFCNT_dec (sv); 1590 SvREFCNT_dec (sv);
1333 sv = 0; 1591 sv = 0;
1334 } 1592 }
1335 }
1336
1337 if (offset_return || !sv)
1338 {
1339 offset = dec.json.flags & F_UTF8
1340 ? dec.cur - SvPVX (string)
1341 : utf8_distance (dec.cur, SvPVX (string));
1342
1343 if (offset_return)
1344 *offset_return = offset;
1345 } 1593 }
1346 1594
1347 if (!sv) 1595 if (!sv)
1348 { 1596 {
1349 SV *uni = sv_newmortal (); 1597 SV *uni = sv_newmortal ();
… …
1355 SAVEVPTR (PL_curcop); 1603 SAVEVPTR (PL_curcop);
1356 PL_curcop = &cop; 1604 PL_curcop = &cop;
1357 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1605 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1358 LEAVE; 1606 LEAVE;
1359 1607
1360 croak ("%s, at character offset %d [\"%s\"]", 1608 croak ("%s, at character offset %d (before \"%s\")",
1361 dec.err, 1609 dec.err,
1362 (int)offset, 1610 ptr_to_index (string, dec.cur),
1363 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1611 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1364 } 1612 }
1365 1613
1366 sv = sv_2mortal (sv); 1614 sv = sv_2mortal (sv);
1367 1615
1368 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1616 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))
1369 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); 1617 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1370 1618
1371 return sv; 1619 return sv;
1620}
1621
1622/////////////////////////////////////////////////////////////////////////////
1623// incremental parser
1624
1625static void
1626incr_parse (JSON *self)
1627{
1628 const char *p = SvPVX (self->incr_text) + self->incr_pos;
1629
1630 // the state machine here is a bit convoluted and could be simplified a lot
1631 // but this would make it slower, so...
1632
1633 for (;;)
1634 {
1635 //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
1636 switch (self->incr_mode)
1637 {
1638 // only used for initial whitespace skipping
1639 case INCR_M_WS:
1640 for (;;)
1641 {
1642 if (*p > 0x20)
1643 {
1644 if (*p == '#')
1645 {
1646 self->incr_mode = INCR_M_C0;
1647 goto incr_m_c;
1648 }
1649 else
1650 {
1651 self->incr_mode = INCR_M_JSON;
1652 goto incr_m_json;
1653 }
1654 }
1655 else if (!*p)
1656 goto interrupt;
1657
1658 ++p;
1659 }
1660
1661 // skip a single char inside a string (for \\-processing)
1662 case INCR_M_BS:
1663 if (!*p)
1664 goto interrupt;
1665
1666 ++p;
1667 self->incr_mode = INCR_M_STR;
1668 goto incr_m_str;
1669
1670 // inside #-style comments
1671 case INCR_M_C0:
1672 case INCR_M_C1:
1673 incr_m_c:
1674 for (;;)
1675 {
1676 if (*p == '\n')
1677 {
1678 self->incr_mode = self->incr_mode == INCR_M_C0 ? INCR_M_WS : INCR_M_JSON;
1679 break;
1680 }
1681 else if (!*p)
1682 goto interrupt;
1683
1684 ++p;
1685 }
1686
1687 break;
1688
1689 // inside a string
1690 case INCR_M_STR:
1691 incr_m_str:
1692 for (;;)
1693 {
1694 if (*p == '"')
1695 {
1696 ++p;
1697 self->incr_mode = INCR_M_JSON;
1698
1699 if (!self->incr_nest)
1700 goto interrupt;
1701
1702 goto incr_m_json;
1703 }
1704 else if (*p == '\\')
1705 {
1706 ++p; // "virtually" consumes character after \
1707
1708 if (!*p) // if at end of string we have to switch modes
1709 {
1710 self->incr_mode = INCR_M_BS;
1711 goto interrupt;
1712 }
1713 }
1714 else if (!*p)
1715 goto interrupt;
1716
1717 ++p;
1718 }
1719
1720 // after initial ws, outside string
1721 case INCR_M_JSON:
1722 incr_m_json:
1723 for (;;)
1724 {
1725 switch (*p++)
1726 {
1727 case 0:
1728 --p;
1729 goto interrupt;
1730
1731 case 0x09:
1732 case 0x0a:
1733 case 0x0d:
1734 case 0x20:
1735 if (!self->incr_nest)
1736 {
1737 --p; // do not eat the whitespace, let the next round do it
1738 goto interrupt;
1739 }
1740 break;
1741
1742 case '"':
1743 self->incr_mode = INCR_M_STR;
1744 goto incr_m_str;
1745
1746 case '[':
1747 case '{':
1748 if (++self->incr_nest > self->max_depth)
1749 croak (ERR_NESTING_EXCEEDED);
1750 break;
1751
1752 case ']':
1753 case '}':
1754 if (--self->incr_nest <= 0)
1755 goto interrupt;
1756 break;
1757
1758 case '#':
1759 self->incr_mode = INCR_M_C1;
1760 goto incr_m_c;
1761 }
1762 }
1763 }
1764
1765 modechange:
1766 ;
1767 }
1768
1769interrupt:
1770 self->incr_pos = p - SvPVX (self->incr_text);
1771 //printf ("interrupt<%.*s>\n", self->incr_pos, SvPVX(self->incr_text));//D
1772 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D
1372} 1773}
1373 1774
1374///////////////////////////////////////////////////////////////////////////// 1775/////////////////////////////////////////////////////////////////////////////
1375// XS interface functions 1776// XS interface functions
1376 1777
… …
1388 : -1; 1789 : -1;
1389 1790
1390 json_stash = gv_stashpv ("JSON::XS" , 1); 1791 json_stash = gv_stashpv ("JSON::XS" , 1);
1391 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); 1792 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1392 1793
1393 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true ); 1794 json_true = get_bool ("JSON::XS::true");
1394 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); 1795 json_false = get_bool ("JSON::XS::false");
1796
1797 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */
1395} 1798}
1396 1799
1397PROTOTYPES: DISABLE 1800PROTOTYPES: DISABLE
1398 1801
1399void CLONE (...) 1802void CLONE (...)
… …
1404void new (char *klass) 1807void new (char *klass)
1405 PPCODE: 1808 PPCODE:
1406{ 1809{
1407 SV *pv = NEWSV (0, sizeof (JSON)); 1810 SV *pv = NEWSV (0, sizeof (JSON));
1408 SvPOK_only (pv); 1811 SvPOK_only (pv);
1409 Zero (SvPVX (pv), 1, JSON); 1812 json_init ((JSON *)SvPVX (pv));
1410 ((JSON *)SvPVX (pv))->flags = F_DEFAULT; 1813 XPUSHs (sv_2mortal (sv_bless (
1411 XPUSHs (sv_2mortal (sv_bless (newRV_noinc (pv), JSON_STASH))); 1814 newRV_noinc (pv),
1815 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1816 )));
1412} 1817}
1413 1818
1414void ascii (JSON *self, int enable = 1) 1819void ascii (JSON *self, int enable = 1)
1415 ALIAS: 1820 ALIAS:
1416 ascii = F_ASCII 1821 ascii = F_ASCII
… …
1423 pretty = F_PRETTY 1828 pretty = F_PRETTY
1424 allow_nonref = F_ALLOW_NONREF 1829 allow_nonref = F_ALLOW_NONREF
1425 shrink = F_SHRINK 1830 shrink = F_SHRINK
1426 allow_blessed = F_ALLOW_BLESSED 1831 allow_blessed = F_ALLOW_BLESSED
1427 convert_blessed = F_CONV_BLESSED 1832 convert_blessed = F_CONV_BLESSED
1833 relaxed = F_RELAXED
1834 allow_unknown = F_ALLOW_UNKNOWN
1428 PPCODE: 1835 PPCODE:
1429{ 1836{
1430 if (enable) 1837 if (enable)
1431 self->flags |= ix; 1838 self->flags |= ix;
1432 else 1839 else
1433 self->flags &= ~ix; 1840 self->flags &= ~ix;
1434 1841
1435 XPUSHs (ST (0)); 1842 XPUSHs (ST (0));
1436} 1843}
1437 1844
1438void max_depth (JSON *self, UV max_depth = 0x80000000UL) 1845void get_ascii (JSON *self)
1846 ALIAS:
1847 get_ascii = F_ASCII
1848 get_latin1 = F_LATIN1
1849 get_utf8 = F_UTF8
1850 get_indent = F_INDENT
1851 get_canonical = F_CANONICAL
1852 get_space_before = F_SPACE_BEFORE
1853 get_space_after = F_SPACE_AFTER
1854 get_allow_nonref = F_ALLOW_NONREF
1855 get_shrink = F_SHRINK
1856 get_allow_blessed = F_ALLOW_BLESSED
1857 get_convert_blessed = F_CONV_BLESSED
1858 get_relaxed = F_RELAXED
1859 get_allow_unknown = F_ALLOW_UNKNOWN
1439 PPCODE: 1860 PPCODE:
1440{ 1861 XPUSHs (boolSV (self->flags & ix));
1441 UV log2 = 0;
1442 1862
1443 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL; 1863void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
1444 1864 PPCODE:
1445 while ((1UL << log2) < max_depth) 1865 self->max_depth = max_depth;
1446 ++log2;
1447
1448 self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1449
1450 XPUSHs (ST (0)); 1866 XPUSHs (ST (0));
1451}
1452 1867
1868U32 get_max_depth (JSON *self)
1869 CODE:
1870 RETVAL = self->max_depth;
1871 OUTPUT:
1872 RETVAL
1873
1453void max_size (JSON *self, UV max_size = 0) 1874void max_size (JSON *self, U32 max_size = 0)
1454 PPCODE: 1875 PPCODE:
1455{ 1876 self->max_size = max_size;
1456 UV log2 = 0;
1457
1458 if (max_size > 0x80000000UL) max_size = 0x80000000UL;
1459 if (max_size == 1) max_size = 2;
1460
1461 while ((1UL << log2) < max_size)
1462 ++log2;
1463
1464 self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1465
1466 XPUSHs (ST (0)); 1877 XPUSHs (ST (0));
1467} 1878
1879int get_max_size (JSON *self)
1880 CODE:
1881 RETVAL = self->max_size;
1882 OUTPUT:
1883 RETVAL
1468 1884
1469void filter_json_object (JSON *self, SV *cb = &PL_sv_undef) 1885void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1470 PPCODE: 1886 PPCODE:
1471{ 1887{
1472 SvREFCNT_dec (self->cb_object); 1888 SvREFCNT_dec (self->cb_object);
… …
1506 XPUSHs (decode_json (jsonstr, self, 0)); 1922 XPUSHs (decode_json (jsonstr, self, 0));
1507 1923
1508void decode_prefix (JSON *self, SV *jsonstr) 1924void decode_prefix (JSON *self, SV *jsonstr)
1509 PPCODE: 1925 PPCODE:
1510{ 1926{
1511 UV offset; 1927 char *offset;
1512 EXTEND (SP, 2); 1928 EXTEND (SP, 2);
1513 PUSHs (decode_json (jsonstr, self, &offset)); 1929 PUSHs (decode_json (jsonstr, self, &offset));
1514 PUSHs (sv_2mortal (newSVuv (offset))); 1930 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset))));
1931}
1932
1933void incr_parse (JSON *self, SV *jsonstr = 0)
1934 PPCODE:
1935{
1936 if (!self->incr_text)
1937 self->incr_text = newSVpvn ("", 0);
1938
1939 /* if utf8-ness doesn't match the decoder, need to upgrade/downgrade */
1940 if (!DECODE_WANTS_OCTETS (self) == !SvUTF8 (self->incr_text))
1941 if (DECODE_WANTS_OCTETS (self))
1942 {
1943 if (self->incr_pos)
1944 self->incr_pos = utf8_length ((U8 *)SvPVX (self->incr_text),
1945 (U8 *)SvPVX (self->incr_text) + self->incr_pos);
1946
1947 sv_utf8_downgrade (self->incr_text, 0);
1948 }
1949 else
1950 {
1951 sv_utf8_upgrade (self->incr_text);
1952
1953 if (self->incr_pos)
1954 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
1955 - (U8 *)SvPVX (self->incr_text);
1956 }
1957
1958 // append data, if any
1959 if (jsonstr)
1960 {
1961 /* make sure both strings have same encoding */
1962 if (SvUTF8 (jsonstr) != SvUTF8 (self->incr_text))
1963 if (SvUTF8 (jsonstr))
1964 sv_utf8_downgrade (jsonstr, 0);
1965 else
1966 sv_utf8_upgrade (jsonstr);
1967
1968 /* and then just blindly append */
1969 {
1970 STRLEN len;
1971 const char *str = SvPV (jsonstr, len);
1972 STRLEN cur = SvCUR (self->incr_text);
1973
1974 if (SvLEN (self->incr_text) <= cur + len)
1975 SvGROW (self->incr_text, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
1976
1977 Move (str, SvEND (self->incr_text), len, char);
1978 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
1979 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
1980 }
1981 }
1982
1983 if (GIMME_V != G_VOID)
1984 do
1985 {
1986 char *offset;
1987
1988 if (!INCR_DONE (self))
1989 {
1990 incr_parse (self);
1991
1992 if (self->incr_pos > self->max_size && self->max_size)
1993 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1994 (unsigned long)self->incr_pos, (unsigned long)self->max_size);
1995
1996 if (!INCR_DONE (self))
1997 {
1998 // as an optimisation, do not accumulate white space in the incr buffer
1999 if (self->incr_mode == INCR_M_WS && self->incr_pos)
2000 {
2001 self->incr_pos = 0;
2002 SvCUR_set (self->incr_text, 0);
2003 }
2004
2005 break;
2006 }
2007 }
2008
2009 XPUSHs (decode_json (self->incr_text, self, &offset));
2010
2011 self->incr_pos -= offset - SvPVX (self->incr_text);
2012 self->incr_nest = 0;
2013 self->incr_mode = 0;
2014
2015 sv_chop (self->incr_text, offset);
2016 }
2017 while (GIMME_V == G_ARRAY);
2018}
2019
2020SV *incr_text (JSON *self)
2021 ATTRS: lvalue
2022 CODE:
2023{
2024 if (self->incr_pos)
2025 croak ("incr_text can not be called when the incremental parser already started parsing");
2026
2027 RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef;
2028}
2029 OUTPUT:
2030 RETVAL
2031
2032void incr_skip (JSON *self)
2033 CODE:
2034{
2035 if (self->incr_pos)
2036 {
2037 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + self->incr_pos);
2038 self->incr_pos = 0;
2039 self->incr_nest = 0;
2040 self->incr_mode = 0;
2041 }
2042}
2043
2044void incr_reset (JSON *self)
2045 CODE:
2046{
2047 SvREFCNT_dec (self->incr_text);
2048 self->incr_text = 0;
2049 self->incr_pos = 0;
2050 self->incr_nest = 0;
2051 self->incr_mode = 0;
1515} 2052}
1516 2053
1517void DESTROY (JSON *self) 2054void DESTROY (JSON *self)
1518 CODE: 2055 CODE:
1519 SvREFCNT_dec (self->cb_sk_object); 2056 SvREFCNT_dec (self->cb_sk_object);
1520 SvREFCNT_dec (self->cb_object); 2057 SvREFCNT_dec (self->cb_object);
2058 SvREFCNT_dec (self->incr_text);
1521 2059
1522PROTOTYPES: ENABLE 2060PROTOTYPES: ENABLE
1523 2061
1524void to_json (SV *scalar) 2062void encode_json (SV *scalar)
2063 ALIAS:
2064 to_json_ = 0
2065 encode_json = F_UTF8
1525 PPCODE: 2066 PPCODE:
1526{ 2067{
1527 JSON json = { F_DEFAULT | F_UTF8 }; 2068 JSON json;
2069 json_init (&json);
2070 json.flags |= ix;
1528 XPUSHs (encode_json (scalar, &json)); 2071 XPUSHs (encode_json (scalar, &json));
1529} 2072}
1530 2073
1531void from_json (SV *jsonstr) 2074void decode_json (SV *jsonstr)
2075 ALIAS:
2076 from_json_ = 0
2077 decode_json = F_UTF8
1532 PPCODE: 2078 PPCODE:
1533{ 2079{
1534 JSON json = { F_DEFAULT | F_UTF8 }; 2080 JSON json;
2081 json_init (&json);
2082 json.flags |= ix;
1535 XPUSHs (decode_json (jsonstr, &json, 0)); 2083 XPUSHs (decode_json (jsonstr, &json, 0));
1536} 2084}
1537 2085

Diff Legend

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