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 1.48 by root, Sun Jul 1 22:20:00 2007 UTC vs.
Revision 1.95 by root, Sun Feb 22 06:55:28 2009 UTC

1#include "EXTERN.h" 1#include "EXTERN.h"
2#include "perl.h" 2#include "perl.h"
3#include "XSUB.h" 3#include "XSUB.h"
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>
10#include <float.h>
9 11
10#if defined(__BORLANDC__) || defined(_MSC_VER) 12#if defined(__BORLANDC__) || defined(_MSC_VER)
11# define snprintf _snprintf // C compilers have this in stdio.h 13# define snprintf _snprintf // C compilers have this in stdio.h
12#endif 14#endif
13 15
14// 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
15// guarentees, though. if it breaks, you get to keep the pieces. 17// guarentees, though. if it breaks, you get to keep the pieces.
16#ifndef UTF8_MAXBYTES 18#ifndef UTF8_MAXBYTES
17# define UTF8_MAXBYTES 13 19# define UTF8_MAXBYTES 13
18#endif 20#endif
21
22#define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 2)
19 23
20#define F_ASCII 0x00000001UL 24#define F_ASCII 0x00000001UL
21#define F_LATIN1 0x00000002UL 25#define F_LATIN1 0x00000002UL
22#define F_UTF8 0x00000004UL 26#define F_UTF8 0x00000004UL
23#define F_INDENT 0x00000008UL 27#define F_INDENT 0x00000008UL
25#define F_SPACE_BEFORE 0x00000020UL 29#define F_SPACE_BEFORE 0x00000020UL
26#define F_SPACE_AFTER 0x00000040UL 30#define F_SPACE_AFTER 0x00000040UL
27#define F_ALLOW_NONREF 0x00000100UL 31#define F_ALLOW_NONREF 0x00000100UL
28#define F_SHRINK 0x00000200UL 32#define F_SHRINK 0x00000200UL
29#define F_ALLOW_BLESSED 0x00000400UL 33#define F_ALLOW_BLESSED 0x00000400UL
30#define F_CONV_BLESSED 0x00000800UL // NYI 34#define F_CONV_BLESSED 0x00000800UL
31#define F_MAXDEPTH 0xf8000000UL
32#define S_MAXDEPTH 27
33#define F_MAXSIZE 0x01f00000UL 35#define F_RELAXED 0x00001000UL
34#define S_MAXSIZE 20 36#define F_ALLOW_UNKNOWN 0x00002000UL
35 37#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
36#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
37#define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE ))
38 38
39#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 39#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
40#define F_DEFAULT (9UL << S_MAXDEPTH)
41 40
42#define INIT_SIZE 32 // initial scalar size to be allocated 41#define INIT_SIZE 32 // initial scalar size to be allocated
43#define INDENT_STEP 3 // spaces per indentation level 42#define INDENT_STEP 3 // spaces per indentation level
44 43
45#define SHORT_STRING_LEN 16384 // special-case strings of up to this size 44#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
46 45
47#define SB do { 46#define SB do {
48#define SE } while (0) 47#define SE } while (0)
49 48
50#if __GNUC__ >= 3 49#if __GNUC__ >= 3
51# define expect(expr,value) __builtin_expect ((expr),(value)) 50# define expect(expr,value) __builtin_expect ((expr), (value))
52# define inline inline 51# define INLINE static inline
53#else 52#else
54# define expect(expr,value) (expr) 53# define expect(expr,value) (expr)
55# define inline static 54# define INLINE static
56#endif 55#endif
57 56
58#define expect_false(expr) expect ((expr) != 0, 0) 57#define expect_false(expr) expect ((expr) != 0, 0)
59#define expect_true(expr) expect ((expr) != 0, 1) 58#define expect_true(expr) expect ((expr) != 0, 1)
60 59
60#define IN_RANGE_INC(type,val,beg,end) \
61 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \
62 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
63
64#define ERR_NESTING_EXCEEDED "json text or perl structure exceeds maximum nesting level (max_depth set too low?)"
65
66#ifdef USE_ITHREADS
67# define JSON_SLOW 1
68# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1))
69#else
70# define JSON_SLOW 0
71# define JSON_STASH json_stash
72#endif
73
61static HV *json_stash, *json_boolean_stash; // JSON::XS:: 74static HV *json_stash, *json_boolean_stash; // JSON::XS::
62static SV *json_true, *json_false; 75static SV *json_true, *json_false;
63 76
77enum {
78 INCR_M_WS = 0, // initial whitespace skipping, must be 0
79 INCR_M_STR, // inside string
80 INCR_M_BS, // inside backslash
81 INCR_M_JSON // outside anything, count nesting
82};
83
84#define INCR_DONE(json) ((json)->incr_nest <= 0 && (json)->incr_mode == INCR_M_JSON)
85
64typedef struct { 86typedef struct {
65 U32 flags; 87 U32 flags;
88 U32 max_depth;
89 STRLEN max_size;
90
91 SV *cb_object;
92 HV *cb_sk_object;
93
94 // for the incremental parser
95 SV *incr_text; // the source text so far
96 STRLEN incr_pos; // the current offset into the text
97 int incr_nest; // {[]}-nesting level
98 unsigned char incr_mode;
66} JSON; 99} JSON;
100
101INLINE void
102json_init (JSON *json)
103{
104 Zero (json, 1, JSON);
105 json->max_depth = 512;
106}
67 107
68///////////////////////////////////////////////////////////////////////////// 108/////////////////////////////////////////////////////////////////////////////
69// utility functions 109// utility functions
70 110
71inline void 111INLINE SV *
112get_bool (const char *name)
113{
114 SV *sv = get_sv (name, 1);
115
116 SvREADONLY_on (sv);
117 SvREADONLY_on (SvRV (sv));
118
119 return sv;
120}
121
122INLINE void
72shrink (SV *sv) 123shrink (SV *sv)
73{ 124{
74 sv_utf8_downgrade (sv, 1); 125 sv_utf8_downgrade (sv, 1);
126
75 if (SvLEN (sv) > SvCUR (sv) + 1) 127 if (SvLEN (sv) > SvCUR (sv) + 1)
76 { 128 {
77#ifdef SvPV_shrink_to_cur 129#ifdef SvPV_shrink_to_cur
78 SvPV_shrink_to_cur (sv); 130 SvPV_shrink_to_cur (sv);
79#elif defined (SvPV_renew) 131#elif defined (SvPV_renew)
85// decode an utf-8 character and return it, or (UV)-1 in 137// decode an utf-8 character and return it, or (UV)-1 in
86// case of an error. 138// case of an error.
87// we special-case "safe" characters from U+80 .. U+7FF, 139// we special-case "safe" characters from U+80 .. U+7FF,
88// but use the very good perl function to parse anything else. 140// but use the very good perl function to parse anything else.
89// note that we never call this function for a ascii codepoints 141// note that we never call this function for a ascii codepoints
90inline UV 142INLINE UV
91decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) 143decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
92{ 144{
93 if (expect_false (s[0] > 0xdf || s[0] < 0xc2)) 145 if (expect_true (len >= 2
94 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); 146 && IN_RANGE_INC (char, s[0], 0xc2, 0xdf)
95 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) 147 && IN_RANGE_INC (char, s[1], 0x80, 0xbf)))
96 { 148 {
97 *clen = 2; 149 *clen = 2;
98 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 150 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
99 } 151 }
100 else 152 else
101 { 153 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
102 *clen = (STRLEN)-1; 154}
103 return (UV)-1; 155
104 } 156// likewise for encoding, also never called for ascii codepoints
157// this function takes advantage of this fact, although current gccs
158// seem to optimise the check for >= 0x80 away anyways
159INLINE unsigned char *
160encode_utf8 (unsigned char *s, UV ch)
161{
162 if (expect_false (ch < 0x000080))
163 *s++ = ch;
164 else if (expect_true (ch < 0x000800))
165 *s++ = 0xc0 | ( ch >> 6),
166 *s++ = 0x80 | ( ch & 0x3f);
167 else if ( ch < 0x010000)
168 *s++ = 0xe0 | ( ch >> 12),
169 *s++ = 0x80 | ((ch >> 6) & 0x3f),
170 *s++ = 0x80 | ( ch & 0x3f);
171 else if ( ch < 0x110000)
172 *s++ = 0xf0 | ( ch >> 18),
173 *s++ = 0x80 | ((ch >> 12) & 0x3f),
174 *s++ = 0x80 | ((ch >> 6) & 0x3f),
175 *s++ = 0x80 | ( ch & 0x3f);
176
177 return s;
105} 178}
106 179
107///////////////////////////////////////////////////////////////////////////// 180/////////////////////////////////////////////////////////////////////////////
108// encoder 181// encoder
109 182
113 char *cur; // SvPVX (sv) + current output position 186 char *cur; // SvPVX (sv) + current output position
114 char *end; // SvEND (sv) 187 char *end; // SvEND (sv)
115 SV *sv; // result scalar 188 SV *sv; // result scalar
116 JSON json; 189 JSON json;
117 U32 indent; // indentation level 190 U32 indent; // indentation level
118 U32 maxdepth; // max. indentation/recursion level 191 UV limit; // escape character values >= this value when encoding
119} enc_t; 192} enc_t;
120 193
121inline void 194INLINE void
122need (enc_t *enc, STRLEN len) 195need (enc_t *enc, STRLEN len)
123{ 196{
124 if (expect_false (enc->cur + len >= enc->end)) 197 if (expect_false (enc->cur + len >= enc->end))
125 { 198 {
126 STRLEN cur = enc->cur - SvPVX (enc->sv); 199 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
127 SvGROW (enc->sv, cur + len + 1); 200 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
128 enc->cur = SvPVX (enc->sv) + cur; 201 enc->cur = SvPVX (enc->sv) + cur;
129 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 202 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
130 } 203 }
131} 204}
132 205
133inline void 206INLINE void
134encode_ch (enc_t *enc, char ch) 207encode_ch (enc_t *enc, char ch)
135{ 208{
136 need (enc, 1); 209 need (enc, 1);
137 *enc->cur++ = ch; 210 *enc->cur++ = ch;
138} 211}
192 { 265 {
193 uch = ch; 266 uch = ch;
194 clen = 1; 267 clen = 1;
195 } 268 }
196 269
197 if (uch > 0x10FFFFUL) 270 if (uch < 0x80/*0x20*/ || uch >= enc->limit)
198 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
199
200 if (uch < 0x80 || enc->json.flags & F_ASCII || (enc->json.flags & F_LATIN1 && uch > 0xFF))
201 { 271 {
202 if (uch > 0xFFFFUL) 272 if (uch >= 0x10000UL)
203 { 273 {
274 if (uch >= 0x110000UL)
275 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
276
204 need (enc, len += 11); 277 need (enc, len += 11);
205 sprintf (enc->cur, "\\u%04x\\u%04x", 278 sprintf (enc->cur, "\\u%04x\\u%04x",
206 (int)((uch - 0x10000) / 0x400 + 0xD800), 279 (int)((uch - 0x10000) / 0x400 + 0xD800),
207 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 280 (int)((uch - 0x10000) % 0x400 + 0xDC00));
208 enc->cur += 12; 281 enc->cur += 12;
209 } 282 }
210 else 283 else
211 { 284 {
212 static char hexdigit [16] = "0123456789abcdef";
213 need (enc, len += 5); 285 need (enc, len += 5);
214 *enc->cur++ = '\\'; 286 *enc->cur++ = '\\';
215 *enc->cur++ = 'u'; 287 *enc->cur++ = 'u';
216 *enc->cur++ = hexdigit [ uch >> 12 ]; 288 *enc->cur++ = PL_hexdigit [ uch >> 12 ];
217 *enc->cur++ = hexdigit [(uch >> 8) & 15]; 289 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
218 *enc->cur++ = hexdigit [(uch >> 4) & 15]; 290 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
219 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 291 *enc->cur++ = PL_hexdigit [(uch >> 0) & 15];
220 } 292 }
221 293
222 str += clen; 294 str += clen;
223 } 295 }
224 else if (enc->json.flags & F_LATIN1) 296 else if (enc->json.flags & F_LATIN1)
236 while (--clen); 308 while (--clen);
237 } 309 }
238 else 310 else
239 { 311 {
240 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed 312 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
241 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 313 enc->cur = encode_utf8 (enc->cur, uch);
242 ++str; 314 ++str;
243 } 315 }
244 } 316 }
245 } 317 }
246 } 318 }
247 319
248 --len; 320 --len;
249 } 321 }
250} 322}
251 323
252inline void 324INLINE void
253encode_indent (enc_t *enc) 325encode_indent (enc_t *enc)
254{ 326{
255 if (enc->json.flags & F_INDENT) 327 if (enc->json.flags & F_INDENT)
256 { 328 {
257 int spaces = enc->indent * INDENT_STEP; 329 int spaces = enc->indent * INDENT_STEP;
260 memset (enc->cur, ' ', spaces); 332 memset (enc->cur, ' ', spaces);
261 enc->cur += spaces; 333 enc->cur += spaces;
262 } 334 }
263} 335}
264 336
265inline void 337INLINE void
266encode_space (enc_t *enc) 338encode_space (enc_t *enc)
267{ 339{
268 need (enc, 1); 340 need (enc, 1);
269 encode_ch (enc, ' '); 341 encode_ch (enc, ' ');
270} 342}
271 343
272inline void 344INLINE void
273encode_nl (enc_t *enc) 345encode_nl (enc_t *enc)
274{ 346{
275 if (enc->json.flags & F_INDENT) 347 if (enc->json.flags & F_INDENT)
276 { 348 {
277 need (enc, 1); 349 need (enc, 1);
278 encode_ch (enc, '\n'); 350 encode_ch (enc, '\n');
279 } 351 }
280} 352}
281 353
282inline void 354INLINE void
283encode_comma (enc_t *enc) 355encode_comma (enc_t *enc)
284{ 356{
285 encode_ch (enc, ','); 357 encode_ch (enc, ',');
286 358
287 if (enc->json.flags & F_INDENT) 359 if (enc->json.flags & F_INDENT)
295static void 367static void
296encode_av (enc_t *enc, AV *av) 368encode_av (enc_t *enc, AV *av)
297{ 369{
298 int i, len = av_len (av); 370 int i, len = av_len (av);
299 371
300 if (enc->indent >= enc->maxdepth) 372 if (enc->indent >= enc->json.max_depth)
301 croak ("data structure too deep (hit recursion limit)"); 373 croak (ERR_NESTING_EXCEEDED);
302 374
303 encode_ch (enc, '['); encode_nl (enc); 375 encode_ch (enc, '[');
304 ++enc->indent; 376
377 if (len >= 0)
378 {
379 encode_nl (enc); ++enc->indent;
305 380
306 for (i = 0; i <= len; ++i) 381 for (i = 0; i <= len; ++i)
307 { 382 {
383 SV **svp = av_fetch (av, i, 0);
384
308 encode_indent (enc); 385 encode_indent (enc);
309 encode_sv (enc, *av_fetch (av, i, 0));
310 386
387 if (svp)
388 encode_sv (enc, *svp);
389 else
390 encode_str (enc, "null", 4, 0);
391
311 if (i < len) 392 if (i < len)
312 encode_comma (enc); 393 encode_comma (enc);
313 } 394 }
314 395
396 encode_nl (enc); --enc->indent; encode_indent (enc);
397 }
398
315 encode_nl (enc); 399 encode_ch (enc, ']');
316
317 --enc->indent;
318 encode_indent (enc); encode_ch (enc, ']');
319} 400}
320 401
321static void 402static void
322encode_he (enc_t *enc, HE *he) 403encode_hk (enc_t *enc, HE *he)
323{ 404{
324 encode_ch (enc, '"'); 405 encode_ch (enc, '"');
325 406
326 if (HeKLEN (he) == HEf_SVKEY) 407 if (HeKLEN (he) == HEf_SVKEY)
327 { 408 {
340 encode_ch (enc, '"'); 421 encode_ch (enc, '"');
341 422
342 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc); 423 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
343 encode_ch (enc, ':'); 424 encode_ch (enc, ':');
344 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc); 425 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
345 encode_sv (enc, HeVAL (he));
346} 426}
347 427
348// compare hash entries, used when all keys are bytestrings 428// compare hash entries, used when all keys are bytestrings
349static int 429static int
350he_cmp_fast (const void *a_, const void *b_) 430he_cmp_fast (const void *a_, const void *b_)
355 HE *b = *(HE **)b_; 435 HE *b = *(HE **)b_;
356 436
357 STRLEN la = HeKLEN (a); 437 STRLEN la = HeKLEN (a);
358 STRLEN lb = HeKLEN (b); 438 STRLEN lb = HeKLEN (b);
359 439
360 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 440 if (!(cmp = memcmp (HeKEY (b), HeKEY (a), lb < la ? lb : la)))
361 cmp = la - lb; 441 cmp = lb - la;
362 442
363 return cmp; 443 return cmp;
364} 444}
365 445
366// compare hash entries, used when some keys are sv's or utf-x 446// compare hash entries, used when some keys are sv's or utf-x
367static int 447static int
368he_cmp_slow (const void *a, const void *b) 448he_cmp_slow (const void *a, const void *b)
369{ 449{
370 return sv_cmp (HeSVKEY_force (*(HE **)a), HeSVKEY_force (*(HE **)b)); 450 return sv_cmp (HeSVKEY_force (*(HE **)b), HeSVKEY_force (*(HE **)a));
371} 451}
372 452
373static void 453static void
374encode_hv (enc_t *enc, HV *hv) 454encode_hv (enc_t *enc, HV *hv)
375{ 455{
376 int count, i; 456 HE *he;
377 457
378 if (enc->indent >= enc->maxdepth) 458 if (enc->indent >= enc->json.max_depth)
379 croak ("data structure too deep (hit recursion limit)"); 459 croak (ERR_NESTING_EXCEEDED);
380 460
381 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 461 encode_ch (enc, '{');
382 462
383 if ((count = hv_iterinit (hv)))
384 {
385 // for canonical output we have to sort by keys first 463 // for canonical output we have to sort by keys first
386 // actually, this is mostly due to the stupid so-called 464 // actually, this is mostly due to the stupid so-called
387 // security workaround added somewhere in 5.8.x. 465 // security workaround added somewhere in 5.8.x
388 // that randomises hash orderings 466 // that randomises hash orderings
389 if (enc->json.flags & F_CANONICAL) 467 if (enc->json.flags & F_CANONICAL)
468 {
469 int count = hv_iterinit (hv);
470
471 if (SvMAGICAL (hv))
390 { 472 {
473 // need to count by iterating. could improve by dynamically building the vector below
474 // but I don't care for the speed of this special case.
475 // note also that we will run into undefined behaviour when the two iterations
476 // do not result in the same count, something I might care for in some later release.
477
478 count = 0;
479 while (hv_iternext (hv))
480 ++count;
481
482 hv_iterinit (hv);
483 }
484
485 if (count)
486 {
391 int fast = 1; 487 int i, fast = 1;
392 HE *he;
393#if defined(__BORLANDC__) || defined(_MSC_VER) 488#if defined(__BORLANDC__) || defined(_MSC_VER)
394 HE **hes = _alloca (count * sizeof (HE)); 489 HE **hes = _alloca (count * sizeof (HE));
395#else 490#else
396 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode 491 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
397#endif 492#endif
424 519
425 FREETMPS; 520 FREETMPS;
426 LEAVE; 521 LEAVE;
427 } 522 }
428 523
429 for (i = 0; i < count; ++i) 524 encode_nl (enc); ++enc->indent;
525
526 while (count--)
430 { 527 {
431 encode_indent (enc); 528 encode_indent (enc);
529 he = hes [count];
432 encode_he (enc, hes [i]); 530 encode_hk (enc, he);
531 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
433 532
434 if (i < count - 1) 533 if (count)
435 encode_comma (enc); 534 encode_comma (enc);
436 } 535 }
437 536
438 encode_nl (enc); 537 encode_nl (enc); --enc->indent; encode_indent (enc);
439 } 538 }
539 }
440 else 540 else
541 {
542 if (hv_iterinit (hv) || SvMAGICAL (hv))
543 if ((he = hv_iternext (hv)))
441 { 544 {
442 HE *he = hv_iternext (hv); 545 encode_nl (enc); ++enc->indent;
443 546
444 for (;;) 547 for (;;)
445 { 548 {
446 encode_indent (enc); 549 encode_indent (enc);
447 encode_he (enc, he); 550 encode_hk (enc, he);
551 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
448 552
449 if (!(he = hv_iternext (hv))) 553 if (!(he = hv_iternext (hv)))
450 break; 554 break;
451 555
452 encode_comma (enc); 556 encode_comma (enc);
453 } 557 }
454 558
455 encode_nl (enc); 559 encode_nl (enc); --enc->indent; encode_indent (enc);
456 } 560 }
457 } 561 }
458 562
459 --enc->indent; encode_indent (enc); encode_ch (enc, '}'); 563 encode_ch (enc, '}');
460} 564}
461 565
462// encode objects, arrays and special \0=false and \1=true values. 566// encode objects, arrays and special \0=false and \1=true values.
463static void 567static void
464encode_rv (enc_t *enc, SV *sv) 568encode_rv (enc_t *enc, SV *sv)
468 SvGETMAGIC (sv); 572 SvGETMAGIC (sv);
469 svt = SvTYPE (sv); 573 svt = SvTYPE (sv);
470 574
471 if (expect_false (SvOBJECT (sv))) 575 if (expect_false (SvOBJECT (sv)))
472 { 576 {
577 HV *stash = !JSON_SLOW || json_boolean_stash
578 ? json_boolean_stash
579 : gv_stashpv ("JSON::XS::Boolean", 1);
580
473 if (SvSTASH (sv) == json_boolean_stash) 581 if (SvSTASH (sv) == stash)
474 { 582 {
475 if (SvIV (sv) == 0) 583 if (SvIV (sv))
584 encode_str (enc, "true", 4, 0);
585 else
476 encode_str (enc, "false", 5, 0); 586 encode_str (enc, "false", 5, 0);
477 else
478 encode_str (enc, "true", 4, 0);
479 } 587 }
480 else 588 else
481 { 589 {
482#if 0 590#if 0
483 if (0 && sv_derived_from (rv, "JSON::Literal")) 591 if (0 && sv_derived_from (rv, "JSON::Literal"))
486 } 594 }
487#endif 595#endif
488 if (enc->json.flags & F_CONV_BLESSED) 596 if (enc->json.flags & F_CONV_BLESSED)
489 { 597 {
490 // we re-bless the reference to get overload and other niceties right 598 // we re-bless the reference to get overload and other niceties right
491 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 1); 599 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
492 600
493 if (to_json) 601 if (to_json)
494 { 602 {
495 dSP; 603 dSP;
496 ENTER; 604
497 SAVETMPS;
498 PUSHMARK (SP); 605 ENTER; SAVETMPS; PUSHMARK (SP);
499 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv))); 606 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
500 607
501 // calling with G_SCALAR ensures that we always get a 1 reutrn value 608 // calling with G_SCALAR ensures that we always get a 1 return value
502 // check anyways.
503 PUTBACK; 609 PUTBACK;
504 assert (1 == call_sv ((SV *)GvCV (to_json), G_SCALAR)); 610 call_sv ((SV *)GvCV (to_json), G_SCALAR);
505 SPAGAIN; 611 SPAGAIN;
506 612
613 // catch this surprisingly common error
614 if (SvROK (TOPs) && SvRV (TOPs) == sv)
615 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
616
617 sv = POPs;
618 PUTBACK;
619
507 encode_sv (enc, POPs); 620 encode_sv (enc, sv);
508 621
509 FREETMPS; 622 FREETMPS; LEAVE;
510 LEAVE;
511 } 623 }
512 else if (enc->json.flags & F_ALLOW_BLESSED) 624 else if (enc->json.flags & F_ALLOW_BLESSED)
513 encode_str (enc, "null", 4, 0); 625 encode_str (enc, "null", 4, 0);
514 else 626 else
515 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it", 627 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
526 encode_hv (enc, (HV *)sv); 638 encode_hv (enc, (HV *)sv);
527 else if (svt == SVt_PVAV) 639 else if (svt == SVt_PVAV)
528 encode_av (enc, (AV *)sv); 640 encode_av (enc, (AV *)sv);
529 else if (svt < SVt_PVAV) 641 else if (svt < SVt_PVAV)
530 { 642 {
531 if (SvNIOK (sv) && SvIV (sv) == 0) 643 STRLEN len = 0;
644 char *pv = svt ? SvPV (sv, len) : 0;
645
646 if (len == 1 && *pv == '1')
647 encode_str (enc, "true", 4, 0);
648 else if (len == 1 && *pv == '0')
532 encode_str (enc, "false", 5, 0); 649 encode_str (enc, "false", 5, 0);
533 else if (SvNIOK (sv) && SvIV (sv) == 1) 650 else if (enc->json.flags & F_ALLOW_UNKNOWN)
534 encode_str (enc, "true", 4, 0); 651 encode_str (enc, "null", 4, 0);
535 else 652 else
536 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", 653 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
537 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 654 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
538 } 655 }
656 else if (enc->json.flags & F_ALLOW_UNKNOWN)
657 encode_str (enc, "null", 4, 0);
539 else 658 else
540 croak ("encountered %s, but JSON can only represent references to arrays or hashes", 659 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
541 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 660 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
542} 661}
543 662
561 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 680 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
562 enc->cur += strlen (enc->cur); 681 enc->cur += strlen (enc->cur);
563 } 682 }
564 else if (SvIOKp (sv)) 683 else if (SvIOKp (sv))
565 { 684 {
566 // we assume we can always read an IV as a UV 685 // we assume we can always read an IV as a UV and vice versa
567 if (SvUV (sv) & ~(UV)0x7fff) 686 // we assume two's complement
568 { 687 // we assume no aliasing issues in the union
569 // large integer, use the (rather slow) snprintf way. 688 if (SvIsUV (sv) ? SvUVX (sv) <= 59000
570 need (enc, sizeof (UV) * 3); 689 : SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000)
571 enc->cur +=
572 SvIsUV(sv)
573 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
574 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
575 }
576 else
577 { 690 {
578 // optimise the "small number case" 691 // optimise the "small number case"
579 // code will likely be branchless and use only a single multiplication 692 // code will likely be branchless and use only a single multiplication
693 // works for numbers up to 59074
580 I32 i = SvIV (sv); 694 I32 i = SvIVX (sv);
581 U32 u; 695 U32 u;
582 char digit, nz = 0; 696 char digit, nz = 0;
583 697
584 need (enc, 6); 698 need (enc, 6);
585 699
591 705
592 // now output digit by digit, each time masking out the integer part 706 // now output digit by digit, each time masking out the integer part
593 // and multiplying by 5 while moving the decimal point one to the right, 707 // and multiplying by 5 while moving the decimal point one to the right,
594 // resulting in a net multiplication by 10. 708 // resulting in a net multiplication by 10.
595 // we always write the digit to memory but conditionally increment 709 // we always write the digit to memory but conditionally increment
596 // the pointer, to ease the usage of conditional move instructions. 710 // the pointer, to enable the use of conditional move instructions.
597 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5; 711 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffffUL) * 5;
598 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5; 712 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffffUL) * 5;
599 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5; 713 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffffUL) * 5;
600 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5; 714 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffffUL) * 5;
601 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0' 715 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
602 } 716 }
717 else
718 {
719 // large integer, use the (rather slow) snprintf way.
720 need (enc, IVUV_MAXCHARS);
721 enc->cur +=
722 SvIsUV(sv)
723 ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))
724 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
725 }
603 } 726 }
604 else if (SvROK (sv)) 727 else if (SvROK (sv))
605 encode_rv (enc, SvRV (sv)); 728 encode_rv (enc, SvRV (sv));
606 else if (!SvOK (sv)) 729 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN)
607 encode_str (enc, "null", 4, 0); 730 encode_str (enc, "null", 4, 0);
608 else 731 else
609 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 732 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
610 SvPV_nolen (sv), SvFLAGS (sv)); 733 SvPV_nolen (sv), SvFLAGS (sv));
611} 734}
621 enc.json = *json; 744 enc.json = *json;
622 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 745 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
623 enc.cur = SvPVX (enc.sv); 746 enc.cur = SvPVX (enc.sv);
624 enc.end = SvEND (enc.sv); 747 enc.end = SvEND (enc.sv);
625 enc.indent = 0; 748 enc.indent = 0;
626 enc.maxdepth = DEC_DEPTH (enc.json.flags); 749 enc.limit = enc.json.flags & F_ASCII ? 0x000080UL
750 : enc.json.flags & F_LATIN1 ? 0x000100UL
751 : 0x110000UL;
627 752
628 SvPOK_only (enc.sv); 753 SvPOK_only (enc.sv);
629 encode_sv (&enc, scalar); 754 encode_sv (&enc, scalar);
630 755
631 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 756 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
652 JSON json; 777 JSON json;
653 U32 depth; // recursion depth 778 U32 depth; // recursion depth
654 U32 maxdepth; // recursion depth limit 779 U32 maxdepth; // recursion depth limit
655} dec_t; 780} dec_t;
656 781
657inline void 782INLINE void
783decode_comment (dec_t *dec)
784{
785 // only '#'-style comments allowed a.t.m.
786
787 while (*dec->cur && *dec->cur != 0x0a && *dec->cur != 0x0d)
788 ++dec->cur;
789}
790
791INLINE void
658decode_ws (dec_t *dec) 792decode_ws (dec_t *dec)
659{ 793{
660 for (;;) 794 for (;;)
661 { 795 {
662 char ch = *dec->cur; 796 char ch = *dec->cur;
663 797
664 if (ch > 0x20 798 if (ch > 0x20)
799 {
800 if (expect_false (ch == '#'))
801 {
802 if (dec->json.flags & F_RELAXED)
803 decode_comment (dec);
804 else
805 break;
806 }
807 else
808 break;
809 }
665 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) 810 else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)
666 break; 811 break; // parse error, but let higher level handle it, gives better error messages
667 812
668 ++dec->cur; 813 ++dec->cur;
669 } 814 }
670} 815}
671 816
675 if (*dec->cur != ch) \ 820 if (*dec->cur != ch) \
676 ERR (# ch " expected"); \ 821 ERR (# ch " expected"); \
677 ++dec->cur; \ 822 ++dec->cur; \
678 SE 823 SE
679 824
680#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)") 825#define DEC_INC_DEPTH if (++dec->depth > dec->json.max_depth) ERR (ERR_NESTING_EXCEEDED)
681#define DEC_DEC_DEPTH --dec->depth 826#define DEC_DEC_DEPTH --dec->depth
682 827
683static SV *decode_sv (dec_t *dec); 828static SV *decode_sv (dec_t *dec);
684 829
685static signed char decode_hexdigit[256]; 830static signed char decode_hexdigit[256];
777 922
778 if (hi >= 0x80) 923 if (hi >= 0x80)
779 { 924 {
780 utf8 = 1; 925 utf8 = 1;
781 926
782 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0); 927 cur = encode_utf8 (cur, hi);
783 } 928 }
784 else 929 else
785 *cur++ = hi; 930 *cur++ = hi;
786 } 931 }
787 break; 932 break;
789 default: 934 default:
790 --dec_cur; 935 --dec_cur;
791 ERR ("illegal backslash escape sequence in string"); 936 ERR ("illegal backslash escape sequence in string");
792 } 937 }
793 } 938 }
794 else if (expect_true (ch >= 0x20 && ch <= 0x7f)) 939 else if (expect_true (ch >= 0x20 && ch < 0x80))
795 *cur++ = ch; 940 *cur++ = ch;
796 else if (ch >= 0x80) 941 else if (ch >= 0x80)
797 { 942 {
798 STRLEN clen; 943 STRLEN clen;
799 UV uch; 944 UV uch;
825 { 970 {
826 STRLEN len = cur - buf; 971 STRLEN len = cur - buf;
827 972
828 if (sv) 973 if (sv)
829 { 974 {
830 SvGROW (sv, SvCUR (sv) + len + 1); 975 STRLEN cur = SvCUR (sv);
976
977 if (SvLEN (sv) <= cur + len)
978 SvGROW (sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
979
831 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 980 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
832 SvCUR_set (sv, SvCUR (sv) + len); 981 SvCUR_set (sv, SvCUR (sv) + len);
833 } 982 }
834 else 983 else
835 sv = newSVpvn (buf, len); 984 sv = newSVpvn (buf, len);
920 is_nv = 1; 1069 is_nv = 1;
921 } 1070 }
922 1071
923 if (!is_nv) 1072 if (!is_nv)
924 { 1073 {
1074 int len = dec->cur - start;
1075
925 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so 1076 // special case the rather common 1..5-digit-int case
926 if (*start == '-') 1077 if (*start == '-')
927 switch (dec->cur - start) 1078 switch (len)
928 { 1079 {
929 case 2: return newSViv (-( start [1] - '0' * 1)); 1080 case 2: return newSViv (-( start [1] - '0' * 1));
930 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); 1081 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
931 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); 1082 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
932 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); 1083 case 5: return newSViv (-( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
1084 case 6: return newSViv (-(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
933 } 1085 }
934 else 1086 else
935 switch (dec->cur - start) 1087 switch (len)
936 { 1088 {
937 case 1: return newSViv ( start [0] - '0' * 1); 1089 case 1: return newSViv ( start [0] - '0' * 1);
938 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); 1090 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
939 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); 1091 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
940 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); 1092 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
1093 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
941 } 1094 }
942 1095
943 { 1096 {
944 UV uv; 1097 UV uv;
945 int numtype = grok_number (start, dec->cur - start, &uv); 1098 int numtype = grok_number (start, len, &uv);
946 if (numtype & IS_NUMBER_IN_UV) 1099 if (numtype & IS_NUMBER_IN_UV)
947 if (numtype & IS_NUMBER_NEG) 1100 if (numtype & IS_NUMBER_NEG)
948 { 1101 {
949 if (uv < (UV)IV_MIN) 1102 if (uv < (UV)IV_MIN)
950 return newSViv (-(IV)uv); 1103 return newSViv (-(IV)uv);
951 } 1104 }
952 else 1105 else
953 return newSVuv (uv); 1106 return newSVuv (uv);
954
955 // here would likely be the place for bigint support
956 } 1107 }
957 }
958 1108
959 // if we ever support bigint or bigfloat, this is the place for bigfloat 1109 len -= *start == '-' ? 1 : 0;
1110
1111 // does not fit into IV or UV, try NV
1112 if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len)
1113 #if defined (LDBL_DIG)
1114 || (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len)
1115 #endif
1116 )
1117 // fits into NV without loss of precision
1118 return newSVnv (Atof (start));
1119
1120 // everything else fails, convert it to a string
1121 return newSVpvn (start, dec->cur - start);
1122 }
1123
1124 // loss of precision here
960 return newSVnv (Atof (start)); 1125 return newSVnv (Atof (start));
961 1126
962fail: 1127fail:
963 return 0; 1128 return 0;
964} 1129}
994 1159
995 if (*dec->cur != ',') 1160 if (*dec->cur != ',')
996 ERR (", or ] expected while parsing array"); 1161 ERR (", or ] expected while parsing array");
997 1162
998 ++dec->cur; 1163 ++dec->cur;
1164
1165 decode_ws (dec);
1166
1167 if (*dec->cur == ']' && dec->json.flags & F_RELAXED)
1168 {
1169 ++dec->cur;
1170 break;
1171 }
999 } 1172 }
1000 1173
1001 DEC_DEC_DEPTH; 1174 DEC_DEC_DEPTH;
1002 return newRV_noinc ((SV *)av); 1175 return newRV_noinc ((SV *)av);
1003 1176
1008} 1181}
1009 1182
1010static SV * 1183static SV *
1011decode_hv (dec_t *dec) 1184decode_hv (dec_t *dec)
1012{ 1185{
1186 SV *sv;
1013 HV *hv = newHV (); 1187 HV *hv = newHV ();
1014 1188
1015 DEC_INC_DEPTH; 1189 DEC_INC_DEPTH;
1016 decode_ws (dec); 1190 decode_ws (dec);
1017 1191
1018 if (*dec->cur == '}') 1192 if (*dec->cur == '}')
1019 ++dec->cur; 1193 ++dec->cur;
1020 else 1194 else
1021 for (;;) 1195 for (;;)
1022 { 1196 {
1023 decode_ws (dec); EXPECT_CH ('"'); 1197 EXPECT_CH ('"');
1024 1198
1025 // heuristic: assume that 1199 // heuristic: assume that
1026 // a) decode_str + hv_store_ent are abysmally slow. 1200 // a) decode_str + hv_store_ent are abysmally slow.
1027 // b) most hash keys are short, simple ascii text. 1201 // b) most hash keys are short, simple ascii text.
1028 // => try to "fast-match" such strings to avoid 1202 // => try to "fast-match" such strings to avoid
1032 char *p = dec->cur; 1206 char *p = dec->cur;
1033 char *e = p + 24; // only try up to 24 bytes 1207 char *e = p + 24; // only try up to 24 bytes
1034 1208
1035 for (;;) 1209 for (;;)
1036 { 1210 {
1037 // the >= 0x80 is true on most architectures 1211 // the >= 0x80 is false on most architectures
1038 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\') 1212 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1039 { 1213 {
1040 // slow path, back up and use decode_str 1214 // slow path, back up and use decode_str
1041 SV *key = decode_str (dec); 1215 SV *key = decode_str (dec);
1042 if (!key) 1216 if (!key)
1043 goto fail; 1217 goto fail;
1044 1218
1045 decode_ws (dec); EXPECT_CH (':'); 1219 decode_ws (dec); EXPECT_CH (':');
1046 1220
1221 decode_ws (dec);
1047 value = decode_sv (dec); 1222 value = decode_sv (dec);
1048 if (!value) 1223 if (!value)
1049 { 1224 {
1050 SvREFCNT_dec (key); 1225 SvREFCNT_dec (key);
1051 goto fail; 1226 goto fail;
1063 int len = p - key; 1238 int len = p - key;
1064 dec->cur = p + 1; 1239 dec->cur = p + 1;
1065 1240
1066 decode_ws (dec); EXPECT_CH (':'); 1241 decode_ws (dec); EXPECT_CH (':');
1067 1242
1243 decode_ws (dec);
1068 value = decode_sv (dec); 1244 value = decode_sv (dec);
1069 if (!value) 1245 if (!value)
1070 goto fail; 1246 goto fail;
1071 1247
1072 hv_store (hv, key, len, value, 0); 1248 hv_store (hv, key, len, value, 0);
1088 1264
1089 if (*dec->cur != ',') 1265 if (*dec->cur != ',')
1090 ERR (", or } expected while parsing object/hash"); 1266 ERR (", or } expected while parsing object/hash");
1091 1267
1092 ++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 }
1093 } 1277 }
1094 1278
1095 DEC_DEC_DEPTH; 1279 DEC_DEC_DEPTH;
1096 return newRV_noinc ((SV *)hv); 1280 sv = newRV_noinc ((SV *)hv);
1281
1282 // check filter callbacks
1283 if (dec->json.flags & F_HOOK)
1284 {
1285 if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
1286 {
1287 HE *cb, *he;
1288
1289 hv_iterinit (hv);
1290 he = hv_iternext (hv);
1291 hv_iterinit (hv);
1292
1293 // the next line creates a mortal sv each time its called.
1294 // might want to optimise this for common cases.
1295 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1296
1297 if (cb)
1298 {
1299 dSP;
1300 int count;
1301
1302 ENTER; SAVETMPS; PUSHMARK (SP);
1303 XPUSHs (HeVAL (he));
1304
1305 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1306
1307 if (count == 1)
1308 {
1309 sv = newSVsv (POPs);
1310 FREETMPS; LEAVE;
1311 return sv;
1312 }
1313
1314 FREETMPS; LEAVE;
1315 }
1316 }
1317
1318 if (dec->json.cb_object)
1319 {
1320 dSP;
1321 int count;
1322
1323 ENTER; SAVETMPS; PUSHMARK (SP);
1324 XPUSHs (sv_2mortal (sv));
1325
1326 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1327
1328 if (count == 1)
1329 {
1330 sv = newSVsv (POPs);
1331 FREETMPS; LEAVE;
1332 return sv;
1333 }
1334
1335 SvREFCNT_inc (sv);
1336 FREETMPS; LEAVE;
1337 }
1338 }
1339
1340 return sv;
1097 1341
1098fail: 1342fail:
1099 SvREFCNT_dec (hv); 1343 SvREFCNT_dec (hv);
1100 DEC_DEC_DEPTH; 1344 DEC_DEC_DEPTH;
1101 return 0; 1345 return 0;
1102} 1346}
1103 1347
1104static SV * 1348static SV *
1105decode_sv (dec_t *dec) 1349decode_sv (dec_t *dec)
1106{ 1350{
1107 decode_ws (dec);
1108
1109 // the beauty of JSON: you need exactly one character lookahead 1351 // the beauty of JSON: you need exactly one character lookahead
1110 // to parse anything. 1352 // to parse everything.
1111 switch (*dec->cur) 1353 switch (*dec->cur)
1112 { 1354 {
1113 case '"': ++dec->cur; return decode_str (dec); 1355 case '"': ++dec->cur; return decode_str (dec);
1114 case '[': ++dec->cur; return decode_av (dec); 1356 case '[': ++dec->cur; return decode_av (dec);
1115 case '{': ++dec->cur; return decode_hv (dec); 1357 case '{': ++dec->cur; return decode_hv (dec);
1116 1358
1117 case '-': 1359 case '-':
1118 case '0': case '1': case '2': case '3': case '4': 1360 case '0': case '1': case '2': case '3': case '4':
1119 case '5': case '6': case '7': case '8': case '9': 1361 case '5': case '6': case '7': case '8': case '9':
1120 return decode_num (dec); 1362 return decode_num (dec);
1121 1363
1122 case 't': 1364 case 't':
1123 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1365 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1124 { 1366 {
1125 dec->cur += 4; 1367 dec->cur += 4;
1368#if JSON_SLOW
1369 json_true = get_bool ("JSON::XS::true");
1370#endif
1126 return SvREFCNT_inc (json_true); 1371 return newSVsv (json_true);
1127 } 1372 }
1128 else 1373 else
1129 ERR ("'true' expected"); 1374 ERR ("'true' expected");
1130 1375
1131 break; 1376 break;
1132 1377
1133 case 'f': 1378 case 'f':
1134 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1379 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1135 { 1380 {
1136 dec->cur += 5; 1381 dec->cur += 5;
1382#if JSON_SLOW
1383 json_false = get_bool ("JSON::XS::false");
1384#endif
1137 return SvREFCNT_inc (json_false); 1385 return newSVsv (json_false);
1138 } 1386 }
1139 else 1387 else
1140 ERR ("'false' expected"); 1388 ERR ("'false' expected");
1141 1389
1142 break; 1390 break;
1160fail: 1408fail:
1161 return 0; 1409 return 0;
1162} 1410}
1163 1411
1164static SV * 1412static SV *
1165decode_json (SV *string, JSON *json, UV *offset_return) 1413decode_json (SV *string, JSON *json, STRLEN *offset_return)
1166{ 1414{
1167 dec_t dec; 1415 dec_t dec;
1168 UV offset; 1416 STRLEN offset;
1169 SV *sv; 1417 SV *sv;
1170 1418
1419 /* work around bugs in 5.10 where manipulating magic values
1420 * will perl ignore the magic in subsequent accesses
1421 */
1171 SvGETMAGIC (string); 1422 /*SvGETMAGIC (string);*/
1423 if (SvMAGICAL (string))
1424 string = sv_2mortal (newSVsv (string));
1425
1172 SvUPGRADE (string, SVt_PV); 1426 SvUPGRADE (string, SVt_PV);
1173 1427
1174 if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags)) 1428 /* work around a bug in perl 5.10, which causes SvCUR to fail an
1429 * assertion with -DDEBUGGING, although SvCUR is documented to
1430 * return the xpv_cur field which certainly exists after upgrading.
1431 * according to nicholas clark, calling SvPOK fixes this.
1432 * But it doesn't fix it, so try another workaround, call SvPV_nolen
1433 * and hope for the best.
1434 * Damnit, SvPV_nolen still trips over yet another assertion. This
1435 * assertion business is seriously broken, try yet another workaround
1436 * for the broken -DDEBUGGING.
1437 */
1438#ifdef DEBUGGING
1439 offset = SvOK (string) ? sv_len (string) : 0;
1440#else
1441 offset = SvCUR (string);
1442#endif
1443
1444 if (offset > json->max_size && json->max_size)
1175 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", 1445 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1176 (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags)); 1446 (unsigned long)SvCUR (string), (unsigned long)json->max_size);
1177 1447
1178 if (json->flags & F_UTF8) 1448 if (json->flags & F_UTF8)
1179 sv_utf8_downgrade (string, 0); 1449 sv_utf8_downgrade (string, 0);
1180 else 1450 else
1181 sv_utf8_upgrade (string); 1451 sv_utf8_upgrade (string);
1182 1452
1183 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1453 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
1184 1454
1185 dec.json = *json; 1455 dec.json = *json;
1186 dec.cur = SvPVX (string); 1456 dec.cur = SvPVX (string);
1187 dec.end = SvEND (string); 1457 dec.end = SvEND (string);
1188 dec.err = 0; 1458 dec.err = 0;
1189 dec.depth = 0; 1459 dec.depth = 0;
1190 dec.maxdepth = DEC_DEPTH (dec.json.flags); 1460
1461 if (dec.json.cb_object || dec.json.cb_sk_object)
1462 dec.json.flags |= F_HOOK;
1191 1463
1192 *dec.end = 0; // this should basically be a nop, too, but make sure it's there 1464 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1465
1466 decode_ws (&dec);
1193 sv = decode_sv (&dec); 1467 sv = decode_sv (&dec);
1194 1468
1195 if (!(offset_return || !sv)) 1469 if (!(offset_return || !sv))
1196 { 1470 {
1197 // check for trailing garbage 1471 // check for trailing garbage
1241 1515
1242 return sv; 1516 return sv;
1243} 1517}
1244 1518
1245///////////////////////////////////////////////////////////////////////////// 1519/////////////////////////////////////////////////////////////////////////////
1520// incremental parser
1521
1522static void
1523incr_parse (JSON *self)
1524{
1525 const char *p = SvPVX (self->incr_text) + self->incr_pos;
1526
1527 for (;;)
1528 {
1529 //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
1530 switch (self->incr_mode)
1531 {
1532 // only used for intiial whitespace skipping
1533 case INCR_M_WS:
1534 for (;;)
1535 {
1536 if (*p > 0x20)
1537 {
1538 self->incr_mode = INCR_M_JSON;
1539 goto incr_m_json;
1540 }
1541 else if (!*p)
1542 goto interrupt;
1543
1544 ++p;
1545 }
1546
1547 // skip a single char inside a string (for \\-processing)
1548 case INCR_M_BS:
1549 if (!*p)
1550 goto interrupt;
1551
1552 ++p;
1553 self->incr_mode = INCR_M_STR;
1554 goto incr_m_str;
1555
1556 // inside a string
1557 case INCR_M_STR:
1558 incr_m_str:
1559 for (;;)
1560 {
1561 if (*p == '"')
1562 {
1563 ++p;
1564 self->incr_mode = INCR_M_JSON;
1565
1566 if (!self->incr_nest)
1567 goto interrupt;
1568
1569 goto incr_m_json;
1570 }
1571 else if (*p == '\\')
1572 {
1573 ++p; // "virtually" consumes character after \
1574
1575 if (!*p) // if at end of string we have to switch modes
1576 {
1577 self->incr_mode = INCR_M_BS;
1578 goto interrupt;
1579 }
1580 }
1581 else if (!*p)
1582 goto interrupt;
1583
1584 ++p;
1585 }
1586
1587 // after initial ws, outside string
1588 case INCR_M_JSON:
1589 incr_m_json:
1590 for (;;)
1591 {
1592 switch (*p++)
1593 {
1594 case 0:
1595 --p;
1596 goto interrupt;
1597
1598 case 0x09:
1599 case 0x0a:
1600 case 0x0d:
1601 case 0x20:
1602 if (!self->incr_nest)
1603 {
1604 --p; // do not eat the whitespace, let the next round do it
1605 goto interrupt;
1606 }
1607 break;
1608
1609 case '"':
1610 self->incr_mode = INCR_M_STR;
1611 goto incr_m_str;
1612
1613 case '[':
1614 case '{':
1615 if (++self->incr_nest > self->max_depth)
1616 croak (ERR_NESTING_EXCEEDED);
1617 break;
1618
1619 case ']':
1620 case '}':
1621 if (--self->incr_nest <= 0)
1622 goto interrupt;
1623 }
1624 }
1625 }
1626
1627 modechange:
1628 ;
1629 }
1630
1631interrupt:
1632 self->incr_pos = p - SvPVX (self->incr_text);
1633 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D
1634}
1635
1636/////////////////////////////////////////////////////////////////////////////
1246// XS interface functions 1637// XS interface functions
1247 1638
1248MODULE = JSON::XS PACKAGE = JSON::XS 1639MODULE = JSON::XS PACKAGE = JSON::XS
1249 1640
1250BOOT: 1641BOOT:
1259 : -1; 1650 : -1;
1260 1651
1261 json_stash = gv_stashpv ("JSON::XS" , 1); 1652 json_stash = gv_stashpv ("JSON::XS" , 1);
1262 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); 1653 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1263 1654
1264 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true ); 1655 json_true = get_bool ("JSON::XS::true");
1265 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); 1656 json_false = get_bool ("JSON::XS::false");
1266} 1657}
1267 1658
1268PROTOTYPES: DISABLE 1659PROTOTYPES: DISABLE
1660
1661void CLONE (...)
1662 CODE:
1663 json_stash = 0;
1664 json_boolean_stash = 0;
1269 1665
1270void new (char *klass) 1666void new (char *klass)
1271 PPCODE: 1667 PPCODE:
1272{ 1668{
1273 SV *pv = NEWSV (0, sizeof (JSON)); 1669 SV *pv = NEWSV (0, sizeof (JSON));
1274 SvPOK_only (pv); 1670 SvPOK_only (pv);
1275 Zero (SvPVX (pv), 1, sizeof (JSON)); 1671 json_init ((JSON *)SvPVX (pv));
1276 ((JSON *)SvPVX (pv))->flags = F_DEFAULT; 1672 XPUSHs (sv_2mortal (sv_bless (
1277 XPUSHs (sv_2mortal (sv_bless (newRV_noinc (pv), json_stash))); 1673 newRV_noinc (pv),
1674 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1675 )));
1278} 1676}
1279 1677
1280void ascii (JSON *self, int enable = 1) 1678void ascii (JSON *self, int enable = 1)
1281 ALIAS: 1679 ALIAS:
1282 ascii = F_ASCII 1680 ascii = F_ASCII
1289 pretty = F_PRETTY 1687 pretty = F_PRETTY
1290 allow_nonref = F_ALLOW_NONREF 1688 allow_nonref = F_ALLOW_NONREF
1291 shrink = F_SHRINK 1689 shrink = F_SHRINK
1292 allow_blessed = F_ALLOW_BLESSED 1690 allow_blessed = F_ALLOW_BLESSED
1293 convert_blessed = F_CONV_BLESSED 1691 convert_blessed = F_CONV_BLESSED
1692 relaxed = F_RELAXED
1693 allow_unknown = F_ALLOW_UNKNOWN
1294 PPCODE: 1694 PPCODE:
1295{ 1695{
1296 if (enable) 1696 if (enable)
1297 self->flags |= ix; 1697 self->flags |= ix;
1298 else 1698 else
1299 self->flags &= ~ix; 1699 self->flags &= ~ix;
1300 1700
1301 XPUSHs (ST (0)); 1701 XPUSHs (ST (0));
1302} 1702}
1303 1703
1304void max_depth (JSON *self, UV max_depth = 0x80000000UL) 1704void get_ascii (JSON *self)
1705 ALIAS:
1706 get_ascii = F_ASCII
1707 get_latin1 = F_LATIN1
1708 get_utf8 = F_UTF8
1709 get_indent = F_INDENT
1710 get_canonical = F_CANONICAL
1711 get_space_before = F_SPACE_BEFORE
1712 get_space_after = F_SPACE_AFTER
1713 get_allow_nonref = F_ALLOW_NONREF
1714 get_shrink = F_SHRINK
1715 get_allow_blessed = F_ALLOW_BLESSED
1716 get_convert_blessed = F_CONV_BLESSED
1717 get_relaxed = F_RELAXED
1718 get_allow_unknown = F_ALLOW_UNKNOWN
1305 PPCODE: 1719 PPCODE:
1306{ 1720 XPUSHs (boolSV (self->flags & ix));
1307 UV log2 = 0;
1308 1721
1309 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL; 1722void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
1310 1723 PPCODE:
1311 while ((1UL << log2) < max_depth) 1724 self->max_depth = max_depth;
1312 ++log2;
1313
1314 self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1315
1316 XPUSHs (ST (0)); 1725 XPUSHs (ST (0));
1317}
1318 1726
1727U32 get_max_depth (JSON *self)
1728 CODE:
1729 RETVAL = self->max_depth;
1730 OUTPUT:
1731 RETVAL
1732
1319void max_size (JSON *self, UV max_size = 0) 1733void max_size (JSON *self, U32 max_size = 0)
1320 PPCODE: 1734 PPCODE:
1321{ 1735 self->max_size = max_size;
1322 UV log2 = 0; 1736 XPUSHs (ST (0));
1323 1737
1324 if (max_size > 0x80000000UL) max_size = 0x80000000UL; 1738int get_max_size (JSON *self)
1325 if (max_size == 1) max_size = 2; 1739 CODE:
1740 RETVAL = self->max_size;
1741 OUTPUT:
1742 RETVAL
1326 1743
1327 while ((1UL << log2) < max_size) 1744void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1328 ++log2; 1745 PPCODE:
1746{
1747 SvREFCNT_dec (self->cb_object);
1748 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
1329 1749
1330 self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE); 1750 XPUSHs (ST (0));
1751}
1752
1753void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
1754 PPCODE:
1755{
1756 if (!self->cb_sk_object)
1757 self->cb_sk_object = newHV ();
1758
1759 if (SvOK (cb))
1760 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1761 else
1762 {
1763 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1764
1765 if (!HvKEYS (self->cb_sk_object))
1766 {
1767 SvREFCNT_dec (self->cb_sk_object);
1768 self->cb_sk_object = 0;
1769 }
1770 }
1331 1771
1332 XPUSHs (ST (0)); 1772 XPUSHs (ST (0));
1333} 1773}
1334 1774
1335void encode (JSON *self, SV *scalar) 1775void encode (JSON *self, SV *scalar)
1341 XPUSHs (decode_json (jsonstr, self, 0)); 1781 XPUSHs (decode_json (jsonstr, self, 0));
1342 1782
1343void decode_prefix (JSON *self, SV *jsonstr) 1783void decode_prefix (JSON *self, SV *jsonstr)
1344 PPCODE: 1784 PPCODE:
1345{ 1785{
1346 UV offset; 1786 STRLEN offset;
1347 EXTEND (SP, 2); 1787 EXTEND (SP, 2);
1348 PUSHs (decode_json (jsonstr, self, &offset)); 1788 PUSHs (decode_json (jsonstr, self, &offset));
1349 PUSHs (sv_2mortal (newSVuv (offset))); 1789 PUSHs (sv_2mortal (newSVuv (offset)));
1350} 1790}
1351 1791
1792void incr_parse (JSON *self, SV *jsonstr = 0)
1793 PPCODE:
1794{
1795 if (!self->incr_text)
1796 self->incr_text = newSVpvn ("", 0);
1797
1798 // append data, if any
1799 if (jsonstr)
1800 {
1801 if (SvUTF8 (jsonstr) && !SvUTF8 (self->incr_text))
1802 {
1803 /* utf-8-ness differs, need to upgrade */
1804 sv_utf8_upgrade (self->incr_text);
1805
1806 if (self->incr_pos)
1807 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
1808 - (U8 *)SvPVX (self->incr_text);
1809 }
1810
1811 {
1812 STRLEN len;
1813 const char *str = SvPV (jsonstr, len);
1814 STRLEN cur = SvCUR (self->incr_text);
1815
1816 if (SvLEN (self->incr_text) <= cur + len)
1817 SvGROW (self->incr_text, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
1818
1819 Move (str, SvEND (self->incr_text), len, char);
1820 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
1821 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
1822 }
1823 }
1824
1825 if (GIMME_V != G_VOID)
1826 do
1827 {
1828 STRLEN offset;
1829
1830 if (!INCR_DONE (self))
1831 {
1832 incr_parse (self);
1833
1834 if (self->incr_pos > self->max_size && self->max_size)
1835 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1836 (unsigned long)self->incr_pos, (unsigned long)self->max_size);
1837
1838 if (!INCR_DONE (self))
1839 break;
1840 }
1841
1842 XPUSHs (decode_json (self->incr_text, self, &offset));
1843
1844 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + offset);
1845 self->incr_pos -= offset;
1846 self->incr_nest = 0;
1847 self->incr_mode = 0;
1848 }
1849 while (GIMME_V == G_ARRAY);
1850}
1851
1852SV *incr_text (JSON *self)
1853 ATTRS: lvalue
1854 CODE:
1855{
1856 if (self->incr_pos)
1857 croak ("incr_text can not be called when the incremental parser already started parsing");
1858
1859 RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef;
1860}
1861 OUTPUT:
1862 RETVAL
1863
1864void incr_skip (JSON *self)
1865 CODE:
1866{
1867 if (self->incr_pos)
1868 {
1869 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + self->incr_pos);
1870 self->incr_pos = 0;
1871 self->incr_nest = 0;
1872 self->incr_mode = 0;
1873 }
1874}
1875
1876void incr_reset (JSON *self)
1877 CODE:
1878{
1879 SvREFCNT_dec (self->incr_text);
1880 self->incr_text = 0;
1881 self->incr_pos = 0;
1882 self->incr_nest = 0;
1883 self->incr_mode = 0;
1884}
1885
1886void DESTROY (JSON *self)
1887 CODE:
1888 SvREFCNT_dec (self->cb_sk_object);
1889 SvREFCNT_dec (self->cb_object);
1890 SvREFCNT_dec (self->incr_text);
1891
1352PROTOTYPES: ENABLE 1892PROTOTYPES: ENABLE
1353 1893
1354void to_json (SV *scalar) 1894void encode_json (SV *scalar)
1895 ALIAS:
1896 to_json_ = 0
1897 encode_json = F_UTF8
1355 PPCODE: 1898 PPCODE:
1356{ 1899{
1357 JSON json = { F_DEFAULT | F_UTF8 }; 1900 JSON json;
1901 json_init (&json);
1902 json.flags |= ix;
1358 XPUSHs (encode_json (scalar, &json)); 1903 XPUSHs (encode_json (scalar, &json));
1359} 1904}
1360 1905
1361void from_json (SV *jsonstr) 1906void decode_json (SV *jsonstr)
1907 ALIAS:
1908 from_json_ = 0
1909 decode_json = F_UTF8
1362 PPCODE: 1910 PPCODE:
1363{ 1911{
1364 JSON json = { F_DEFAULT | F_UTF8 }; 1912 JSON json;
1913 json_init (&json);
1914 json.flags |= ix;
1365 XPUSHs (decode_json (jsonstr, &json, 0)); 1915 XPUSHs (decode_json (jsonstr, &json, 0));
1366} 1916}
1367 1917
1918

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines