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 1.40 by root, Tue Jun 12 01:27:02 2007 UTC vs.
Revision 1.93 by root, Mon Sep 29 03:09:27 2008 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
16// some old perls do not have this, try to make it work, no
17// guarentees, though. if it breaks, you get to keep the pieces.
18#ifndef UTF8_MAXBYTES
19# define UTF8_MAXBYTES 13
20#endif
21
22#define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 2)
23
14#define F_ASCII 0x00000001UL 24#define F_ASCII 0x00000001UL
15#define F_LATIN1 0x00000002UL 25#define F_LATIN1 0x00000002UL
16#define F_UTF8 0x00000004UL 26#define F_UTF8 0x00000004UL
17#define F_INDENT 0x00000008UL 27#define F_INDENT 0x00000008UL
18#define F_CANONICAL 0x00000010UL 28#define F_CANONICAL 0x00000010UL
19#define F_SPACE_BEFORE 0x00000020UL 29#define F_SPACE_BEFORE 0x00000020UL
20#define F_SPACE_AFTER 0x00000040UL 30#define F_SPACE_AFTER 0x00000040UL
21#define F_ALLOW_NONREF 0x00000100UL 31#define F_ALLOW_NONREF 0x00000100UL
22#define F_SHRINK 0x00000200UL 32#define F_SHRINK 0x00000200UL
33#define F_ALLOW_BLESSED 0x00000400UL
34#define F_CONV_BLESSED 0x00000800UL
23#define F_MAXDEPTH 0xf8000000UL 35#define F_RELAXED 0x00001000UL
24#define S_MAXDEPTH 27 36#define F_ALLOW_UNKNOWN 0x00002000UL
25 37#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
26#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
27
28// F_SELFCONVERT? <=> to_json/toJson
29// F_BLESSED? <=> { $__class__$ => }
30 38
31#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 39#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
32#define F_DEFAULT (9UL << S_MAXDEPTH)
33 40
34#define INIT_SIZE 32 // initial scalar size to be allocated 41#define INIT_SIZE 32 // initial scalar size to be allocated
35#define INDENT_STEP 3 // spaces per indentation level 42#define INDENT_STEP 3 // spaces per indentation level
36 43
37#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
38 45
39#define SB do { 46#define SB do {
40#define SE } while (0) 47#define SE } while (0)
41 48
42#if __GNUC__ >= 3 49#if __GNUC__ >= 3
43# define expect(expr,value) __builtin_expect ((expr),(value)) 50# define expect(expr,value) __builtin_expect ((expr), (value))
44# define inline inline 51# define INLINE static inline
45#else 52#else
46# define expect(expr,value) (expr) 53# define expect(expr,value) (expr)
47# define inline static 54# define INLINE static
48#endif 55#endif
49 56
50#define expect_false(expr) expect ((expr) != 0, 0) 57#define expect_false(expr) expect ((expr) != 0, 0)
51#define expect_true(expr) expect ((expr) != 0, 1) 58#define expect_true(expr) expect ((expr) != 0, 1)
52 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
53static HV *json_stash; // JSON::XS:: 74static HV *json_stash, *json_boolean_stash; // JSON::XS::
75static SV *json_true, *json_false;
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
86typedef struct {
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;
99} JSON;
100
101INLINE void
102json_init (JSON *json)
103{
104 Zero (json, 1, JSON);
105 json->max_depth = 512;
106}
54 107
55///////////////////////////////////////////////////////////////////////////// 108/////////////////////////////////////////////////////////////////////////////
56// utility functions 109// utility functions
57 110
58static UV * 111INLINE SV *
59SvJSON (SV *sv) 112get_bool (const char *name)
60{ 113{
61 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash)) 114 SV *sv = get_sv (name, 1);
62 croak ("object is not of type JSON::XS");
63 115
64 return &SvUVX (SvRV (sv)); 116 SvREADONLY_on (sv);
65} 117 SvREADONLY_on (SvRV (sv));
66 118
67static void 119 return sv;
120}
121
122INLINE void
68shrink (SV *sv) 123shrink (SV *sv)
69{ 124{
70 sv_utf8_downgrade (sv, 1); 125 sv_utf8_downgrade (sv, 1);
126
71 if (SvLEN (sv) > SvCUR (sv) + 1) 127 if (SvLEN (sv) > SvCUR (sv) + 1)
72 { 128 {
73#ifdef SvPV_shrink_to_cur 129#ifdef SvPV_shrink_to_cur
74 SvPV_shrink_to_cur (sv); 130 SvPV_shrink_to_cur (sv);
75#elif defined (SvPV_renew) 131#elif defined (SvPV_renew)
… …
81// 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
82// case of an error. 138// case of an error.
83// we special-case "safe" characters from U+80 .. U+7FF, 139// we special-case "safe" characters from U+80 .. U+7FF,
84// but use the very good perl function to parse anything else. 140// but use the very good perl function to parse anything else.
85// note that we never call this function for a ascii codepoints 141// note that we never call this function for a ascii codepoints
86inline UV 142INLINE UV
87decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) 143decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
88{ 144{
89 if (expect_false (s[0] > 0xdf || s[0] < 0xc2)) 145 if (expect_true (len >= 2
90 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); 146 && IN_RANGE_INC (char, s[0], 0xc2, 0xdf)
91 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) 147 && IN_RANGE_INC (char, s[1], 0x80, 0xbf)))
92 { 148 {
93 *clen = 2; 149 *clen = 2;
94 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 150 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
95 } 151 }
96 else 152 else
97 { 153 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
98 *clen = (STRLEN)-1; 154}
99 return (UV)-1; 155
100 } 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;
101} 178}
102 179
103///////////////////////////////////////////////////////////////////////////// 180/////////////////////////////////////////////////////////////////////////////
104// encoder 181// encoder
105 182
… …
107typedef struct 184typedef struct
108{ 185{
109 char *cur; // SvPVX (sv) + current output position 186 char *cur; // SvPVX (sv) + current output position
110 char *end; // SvEND (sv) 187 char *end; // SvEND (sv)
111 SV *sv; // result scalar 188 SV *sv; // result scalar
112 U32 flags; // F_* 189 JSON json;
113 U32 indent; // indentation level 190 U32 indent; // indentation level
114 U32 maxdepth; // max. indentation/recursion level 191 UV limit; // escape character values >= this value when encoding
115} enc_t; 192} enc_t;
116 193
117inline void 194INLINE void
118need (enc_t *enc, STRLEN len) 195need (enc_t *enc, STRLEN len)
119{ 196{
120 if (expect_false (enc->cur + len >= enc->end)) 197 if (expect_false (enc->cur + len >= enc->end))
121 { 198 {
122 STRLEN cur = enc->cur - SvPVX (enc->sv); 199 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
123 SvGROW (enc->sv, cur + len + 1); 200 SvGROW (enc->sv, cur + len + 1);
124 enc->cur = SvPVX (enc->sv) + cur; 201 enc->cur = SvPVX (enc->sv) + cur;
125 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 202 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
126 } 203 }
127} 204}
128 205
129inline void 206INLINE void
130encode_ch (enc_t *enc, char ch) 207encode_ch (enc_t *enc, char ch)
131{ 208{
132 need (enc, 1); 209 need (enc, 1);
133 *enc->cur++ = ch; 210 *enc->cur++ = ch;
134} 211}
… …
188 { 265 {
189 uch = ch; 266 uch = ch;
190 clen = 1; 267 clen = 1;
191 } 268 }
192 269
193 if (uch > 0x10FFFFUL) 270 if (uch < 0x80/*0x20*/ || uch >= enc->limit)
194 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
195
196 if (uch < 0x80 || enc->flags & F_ASCII || (enc->flags & F_LATIN1 && uch > 0xFF))
197 { 271 {
198 if (uch > 0xFFFFUL) 272 if (uch >= 0x10000UL)
199 { 273 {
274 if (uch >= 0x110000UL)
275 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
276
200 need (enc, len += 11); 277 need (enc, len += 11);
201 sprintf (enc->cur, "\\u%04x\\u%04x", 278 sprintf (enc->cur, "\\u%04x\\u%04x",
202 (int)((uch - 0x10000) / 0x400 + 0xD800), 279 (int)((uch - 0x10000) / 0x400 + 0xD800),
203 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 280 (int)((uch - 0x10000) % 0x400 + 0xDC00));
204 enc->cur += 12; 281 enc->cur += 12;
205 } 282 }
206 else 283 else
207 { 284 {
208 static char hexdigit [16] = "0123456789abcdef";
209 need (enc, len += 5); 285 need (enc, len += 5);
210 *enc->cur++ = '\\'; 286 *enc->cur++ = '\\';
211 *enc->cur++ = 'u'; 287 *enc->cur++ = 'u';
212 *enc->cur++ = hexdigit [ uch >> 12 ]; 288 *enc->cur++ = PL_hexdigit [ uch >> 12 ];
213 *enc->cur++ = hexdigit [(uch >> 8) & 15]; 289 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
214 *enc->cur++ = hexdigit [(uch >> 4) & 15]; 290 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
215 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 291 *enc->cur++ = PL_hexdigit [(uch >> 0) & 15];
216 } 292 }
217 293
218 str += clen; 294 str += clen;
219 } 295 }
220 else if (enc->flags & F_LATIN1) 296 else if (enc->json.flags & F_LATIN1)
221 { 297 {
222 *enc->cur++ = uch; 298 *enc->cur++ = uch;
223 str += clen; 299 str += clen;
224 } 300 }
225 else if (is_utf8) 301 else if (is_utf8)
… …
232 while (--clen); 308 while (--clen);
233 } 309 }
234 else 310 else
235 { 311 {
236 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
237 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 313 enc->cur = encode_utf8 (enc->cur, uch);
238 ++str; 314 ++str;
239 } 315 }
240 } 316 }
241 } 317 }
242 } 318 }
243 319
244 --len; 320 --len;
245 } 321 }
246} 322}
247 323
248inline void 324INLINE void
249encode_indent (enc_t *enc) 325encode_indent (enc_t *enc)
250{ 326{
251 if (enc->flags & F_INDENT) 327 if (enc->json.flags & F_INDENT)
252 { 328 {
253 int spaces = enc->indent * INDENT_STEP; 329 int spaces = enc->indent * INDENT_STEP;
254 330
255 need (enc, spaces); 331 need (enc, spaces);
256 memset (enc->cur, ' ', spaces); 332 memset (enc->cur, ' ', spaces);
257 enc->cur += spaces; 333 enc->cur += spaces;
258 } 334 }
259} 335}
260 336
261inline void 337INLINE void
262encode_space (enc_t *enc) 338encode_space (enc_t *enc)
263{ 339{
264 need (enc, 1); 340 need (enc, 1);
265 encode_ch (enc, ' '); 341 encode_ch (enc, ' ');
266} 342}
267 343
268inline void 344INLINE void
269encode_nl (enc_t *enc) 345encode_nl (enc_t *enc)
270{ 346{
271 if (enc->flags & F_INDENT) 347 if (enc->json.flags & F_INDENT)
272 { 348 {
273 need (enc, 1); 349 need (enc, 1);
274 encode_ch (enc, '\n'); 350 encode_ch (enc, '\n');
275 } 351 }
276} 352}
277 353
278inline void 354INLINE void
279encode_comma (enc_t *enc) 355encode_comma (enc_t *enc)
280{ 356{
281 encode_ch (enc, ','); 357 encode_ch (enc, ',');
282 358
283 if (enc->flags & F_INDENT) 359 if (enc->json.flags & F_INDENT)
284 encode_nl (enc); 360 encode_nl (enc);
285 else if (enc->flags & F_SPACE_AFTER) 361 else if (enc->json.flags & F_SPACE_AFTER)
286 encode_space (enc); 362 encode_space (enc);
287} 363}
288 364
289static void encode_sv (enc_t *enc, SV *sv); 365static void encode_sv (enc_t *enc, SV *sv);
290 366
291static void 367static void
292encode_av (enc_t *enc, AV *av) 368encode_av (enc_t *enc, AV *av)
293{ 369{
294 int i, len = av_len (av); 370 int i, len = av_len (av);
295 371
296 if (enc->indent >= enc->maxdepth) 372 if (enc->indent >= enc->json.max_depth)
297 croak ("data structure too deep (hit recursion limit)"); 373 croak (ERR_NESTING_EXCEEDED);
298 374
299 encode_ch (enc, '['); encode_nl (enc); 375 encode_ch (enc, '[');
300 ++enc->indent; 376
377 if (len >= 0)
378 {
379 encode_nl (enc); ++enc->indent;
301 380
302 for (i = 0; i <= len; ++i) 381 for (i = 0; i <= len; ++i)
303 { 382 {
383 SV **svp = av_fetch (av, i, 0);
384
304 encode_indent (enc); 385 encode_indent (enc);
305 encode_sv (enc, *av_fetch (av, i, 0));
306 386
387 if (svp)
388 encode_sv (enc, *svp);
389 else
390 encode_str (enc, "null", 4, 0);
391
307 if (i < len) 392 if (i < len)
308 encode_comma (enc); 393 encode_comma (enc);
309 } 394 }
310 395
396 encode_nl (enc); --enc->indent; encode_indent (enc);
397 }
398
311 encode_nl (enc); 399 encode_ch (enc, ']');
312
313 --enc->indent;
314 encode_indent (enc); encode_ch (enc, ']');
315} 400}
316 401
317static void 402static void
318encode_he (enc_t *enc, HE *he) 403encode_hk (enc_t *enc, HE *he)
319{ 404{
320 encode_ch (enc, '"'); 405 encode_ch (enc, '"');
321 406
322 if (HeKLEN (he) == HEf_SVKEY) 407 if (HeKLEN (he) == HEf_SVKEY)
323 { 408 {
… …
333 else 418 else
334 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 419 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
335 420
336 encode_ch (enc, '"'); 421 encode_ch (enc, '"');
337 422
338 if (enc->flags & F_SPACE_BEFORE) encode_space (enc); 423 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
339 encode_ch (enc, ':'); 424 encode_ch (enc, ':');
340 if (enc->flags & F_SPACE_AFTER ) encode_space (enc); 425 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
341 encode_sv (enc, HeVAL (he));
342} 426}
343 427
344// compare hash entries, used when all keys are bytestrings 428// compare hash entries, used when all keys are bytestrings
345static int 429static int
346he_cmp_fast (const void *a_, const void *b_) 430he_cmp_fast (const void *a_, const void *b_)
… …
351 HE *b = *(HE **)b_; 435 HE *b = *(HE **)b_;
352 436
353 STRLEN la = HeKLEN (a); 437 STRLEN la = HeKLEN (a);
354 STRLEN lb = HeKLEN (b); 438 STRLEN lb = HeKLEN (b);
355 439
356 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 440 if (!(cmp = memcmp (HeKEY (b), HeKEY (a), lb < la ? lb : la)))
357 cmp = la - lb; 441 cmp = lb - la;
358 442
359 return cmp; 443 return cmp;
360} 444}
361 445
362// 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
363static int 447static int
364he_cmp_slow (const void *a, const void *b) 448he_cmp_slow (const void *a, const void *b)
365{ 449{
366 return sv_cmp (HeSVKEY_force (*(HE **)a), HeSVKEY_force (*(HE **)b)); 450 return sv_cmp (HeSVKEY_force (*(HE **)b), HeSVKEY_force (*(HE **)a));
367} 451}
368 452
369static void 453static void
370encode_hv (enc_t *enc, HV *hv) 454encode_hv (enc_t *enc, HV *hv)
371{ 455{
372 int count, i; 456 HE *he;
373 457
374 if (enc->indent >= enc->maxdepth) 458 if (enc->indent >= enc->json.max_depth)
375 croak ("data structure too deep (hit recursion limit)"); 459 croak (ERR_NESTING_EXCEEDED);
376 460
377 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 461 encode_ch (enc, '{');
378 462
379 if ((count = hv_iterinit (hv)))
380 {
381 // for canonical output we have to sort by keys first 463 // for canonical output we have to sort by keys first
382 // actually, this is mostly due to the stupid so-called 464 // actually, this is mostly due to the stupid so-called
383 // security workaround added somewhere in 5.8.x. 465 // security workaround added somewhere in 5.8.x
384 // that randomises hash orderings 466 // that randomises hash orderings
385 if (enc->flags & F_CANONICAL) 467 if (enc->json.flags & F_CANONICAL)
468 {
469 int count = hv_iterinit (hv);
470
471 if (SvMAGICAL (hv))
386 { 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 {
387 int fast = 1; 487 int i, fast = 1;
388 HE *he;
389#if defined(__BORLANDC__) || defined(_MSC_VER) 488#if defined(__BORLANDC__) || defined(_MSC_VER)
390 HE **hes = _alloca (count * sizeof (HE)); 489 HE **hes = _alloca (count * sizeof (HE));
391#else 490#else
392 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
393#endif 492#endif
… …
420 519
421 FREETMPS; 520 FREETMPS;
422 LEAVE; 521 LEAVE;
423 } 522 }
424 523
425 for (i = 0; i < count; ++i) 524 encode_nl (enc); ++enc->indent;
525
526 while (count--)
426 { 527 {
427 encode_indent (enc); 528 encode_indent (enc);
529 he = hes [count];
428 encode_he (enc, hes [i]); 530 encode_hk (enc, he);
531 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
429 532
430 if (i < count - 1) 533 if (count)
431 encode_comma (enc); 534 encode_comma (enc);
432 } 535 }
433 536
537 encode_nl (enc); --enc->indent; encode_indent (enc);
538 }
539 }
540 else
541 {
542 if (hv_iterinit (hv) || SvMAGICAL (hv))
543 if ((he = hv_iternext (hv)))
544 {
545 encode_nl (enc); ++enc->indent;
546
547 for (;;)
548 {
434 encode_nl (enc); 549 encode_indent (enc);
550 encode_hk (enc, he);
551 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
552
553 if (!(he = hv_iternext (hv)))
554 break;
555
556 encode_comma (enc);
557 }
558
559 encode_nl (enc); --enc->indent; encode_indent (enc);
560 }
561 }
562
563 encode_ch (enc, '}');
564}
565
566// encode objects, arrays and special \0=false and \1=true values.
567static void
568encode_rv (enc_t *enc, SV *sv)
569{
570 svtype svt;
571
572 SvGETMAGIC (sv);
573 svt = SvTYPE (sv);
574
575 if (expect_false (SvOBJECT (sv)))
576 {
577 HV *stash = !JSON_SLOW || json_boolean_stash
578 ? json_boolean_stash
579 : gv_stashpv ("JSON::XS::Boolean", 1);
580
581 if (SvSTASH (sv) == stash)
582 {
583 if (SvIV (sv))
584 encode_str (enc, "true", 4, 0);
585 else
586 encode_str (enc, "false", 5, 0);
435 } 587 }
436 else 588 else
437 { 589 {
438 HE *he = hv_iternext (hv); 590#if 0
439 591 if (0 && sv_derived_from (rv, "JSON::Literal"))
440 for (;;)
441 { 592 {
442 encode_indent (enc); 593 // not yet
443 encode_he (enc, he);
444
445 if (!(he = hv_iternext (hv)))
446 break;
447
448 encode_comma (enc);
449 } 594 }
595#endif
596 if (enc->json.flags & F_CONV_BLESSED)
597 {
598 // we re-bless the reference to get overload and other niceties right
599 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
450 600
601 if (to_json)
602 {
603 dSP;
604
605 ENTER; SAVETMPS; PUSHMARK (SP);
606 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
607
608 // calling with G_SCALAR ensures that we always get a 1 return value
609 PUTBACK;
610 call_sv ((SV *)GvCV (to_json), G_SCALAR);
611 SPAGAIN;
612
613 // catch this surprisingly common error
614 if (SvROK (TOPs) && SvRV (TOPs) == sv)
615 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
616
617 sv = POPs;
618 PUTBACK;
619
451 encode_nl (enc); 620 encode_sv (enc, sv);
621
622 FREETMPS; LEAVE;
623 }
624 else if (enc->json.flags & F_ALLOW_BLESSED)
625 encode_str (enc, "null", 4, 0);
626 else
627 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
628 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
629 }
630 else if (enc->json.flags & F_ALLOW_BLESSED)
631 encode_str (enc, "null", 4, 0);
632 else
633 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",
634 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
452 } 635 }
453 } 636 }
454
455 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
456}
457
458// encode objects, arrays and special \0=false and \1=true values.
459static void
460encode_rv (enc_t *enc, SV *sv)
461{
462 svtype svt;
463
464 SvGETMAGIC (sv);
465 svt = SvTYPE (sv);
466
467 if (svt == SVt_PVHV) 637 else if (svt == SVt_PVHV)
468 encode_hv (enc, (HV *)sv); 638 encode_hv (enc, (HV *)sv);
469 else if (svt == SVt_PVAV) 639 else if (svt == SVt_PVAV)
470 encode_av (enc, (AV *)sv); 640 encode_av (enc, (AV *)sv);
471 else if (svt < SVt_PVAV) 641 else if (svt < SVt_PVAV)
472 { 642 {
473 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')
474 encode_str (enc, "false", 5, 0); 649 encode_str (enc, "false", 5, 0);
475 else if (SvNIOK (sv) && SvIV (sv) == 1) 650 else if (enc->json.flags & F_ALLOW_UNKNOWN)
476 encode_str (enc, "true", 4, 0); 651 encode_str (enc, "null", 4, 0);
477 else 652 else
478 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",
479 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 654 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
480 } 655 }
656 else if (enc->json.flags & F_ALLOW_UNKNOWN)
657 encode_str (enc, "null", 4, 0);
481 else 658 else
482 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",
483 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 660 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
484} 661}
485 662
… …
503 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 680 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
504 enc->cur += strlen (enc->cur); 681 enc->cur += strlen (enc->cur);
505 } 682 }
506 else if (SvIOKp (sv)) 683 else if (SvIOKp (sv))
507 { 684 {
508 // 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
509 if (SvUV (sv) & ~(UV)0x7fff) 686 // we assume two's complement
510 { 687 // we assume no aliasing issues in the union
511 // large integer, use the (rather slow) snprintf way. 688 if (SvIsUV (sv) ? SvUVX (sv) <= 59000
512 need (enc, sizeof (UV) * 3); 689 : SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000)
513 enc->cur +=
514 SvIsUV(sv)
515 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
516 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
517 }
518 else
519 { 690 {
520 // optimise the "small number case" 691 // optimise the "small number case"
521 // 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
522 I32 i = SvIV (sv); 694 I32 i = SvIVX (sv);
523 U32 u; 695 U32 u;
524 char digit, nz = 0; 696 char digit, nz = 0;
525 697
526 need (enc, 6); 698 need (enc, 6);
527 699
… …
533 705
534 // 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
535 // 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,
536 // resulting in a net multiplication by 10. 708 // resulting in a net multiplication by 10.
537 // we always write the digit to memory but conditionally increment 709 // we always write the digit to memory but conditionally increment
538 // the pointer, to ease the usage of conditional move instructions. 710 // the pointer, to enable the use of conditional move instructions.
539 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;
540 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;
541 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;
542 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;
543 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'
544 } 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 }
545 } 726 }
546 else if (SvROK (sv)) 727 else if (SvROK (sv))
547 encode_rv (enc, SvRV (sv)); 728 encode_rv (enc, SvRV (sv));
548 else if (!SvOK (sv)) 729 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN)
549 encode_str (enc, "null", 4, 0); 730 encode_str (enc, "null", 4, 0);
550 else 731 else
551 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",
552 SvPV_nolen (sv), SvFLAGS (sv)); 733 SvPV_nolen (sv), SvFLAGS (sv));
553} 734}
554 735
555static SV * 736static SV *
556encode_json (SV *scalar, U32 flags) 737encode_json (SV *scalar, JSON *json)
557{ 738{
558 enc_t enc; 739 enc_t enc;
559 740
560 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 741 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
561 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 742 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
562 743
563 enc.flags = flags; 744 enc.json = *json;
564 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 745 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
565 enc.cur = SvPVX (enc.sv); 746 enc.cur = SvPVX (enc.sv);
566 enc.end = SvEND (enc.sv); 747 enc.end = SvEND (enc.sv);
567 enc.indent = 0; 748 enc.indent = 0;
568 enc.maxdepth = DEC_DEPTH (flags); 749 enc.limit = enc.json.flags & F_ASCII ? 0x000080UL
750 : enc.json.flags & F_LATIN1 ? 0x000100UL
751 : 0x110000UL;
569 752
570 SvPOK_only (enc.sv); 753 SvPOK_only (enc.sv);
571 encode_sv (&enc, scalar); 754 encode_sv (&enc, scalar);
572 755
573 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 756 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
574 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 757 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
575 758
576 if (!(flags & (F_ASCII | F_LATIN1 | F_UTF8))) 759 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
577 SvUTF8_on (enc.sv); 760 SvUTF8_on (enc.sv);
578 761
579 if (enc.flags & F_SHRINK) 762 if (enc.json.flags & F_SHRINK)
580 shrink (enc.sv); 763 shrink (enc.sv);
581 764
582 return enc.sv; 765 return enc.sv;
583} 766}
584 767
… …
589typedef struct 772typedef struct
590{ 773{
591 char *cur; // current parser pointer 774 char *cur; // current parser pointer
592 char *end; // end of input string 775 char *end; // end of input string
593 const char *err; // parse error, if != 0 776 const char *err; // parse error, if != 0
594 U32 flags; // F_* 777 JSON json;
595 U32 depth; // recursion depth 778 U32 depth; // recursion depth
596 U32 maxdepth; // recursion depth limit 779 U32 maxdepth; // recursion depth limit
597} dec_t; 780} dec_t;
598 781
599inline 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
600decode_ws (dec_t *dec) 792decode_ws (dec_t *dec)
601{ 793{
602 for (;;) 794 for (;;)
603 { 795 {
604 char ch = *dec->cur; 796 char ch = *dec->cur;
605 797
606 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 }
607 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) 810 else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)
608 break; 811 break; // parse error, but let higher level handle it, gives better error messages
609 812
610 ++dec->cur; 813 ++dec->cur;
611 } 814 }
612} 815}
613 816
… …
617 if (*dec->cur != ch) \ 820 if (*dec->cur != ch) \
618 ERR (# ch " expected"); \ 821 ERR (# ch " expected"); \
619 ++dec->cur; \ 822 ++dec->cur; \
620 SE 823 SE
621 824
622#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)
623#define DEC_DEC_DEPTH --dec->depth 826#define DEC_DEC_DEPTH --dec->depth
624 827
625static SV *decode_sv (dec_t *dec); 828static SV *decode_sv (dec_t *dec);
626 829
627static signed char decode_hexdigit[256]; 830static signed char decode_hexdigit[256];
… …
719 922
720 if (hi >= 0x80) 923 if (hi >= 0x80)
721 { 924 {
722 utf8 = 1; 925 utf8 = 1;
723 926
724 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0); 927 cur = encode_utf8 (cur, hi);
725 } 928 }
726 else 929 else
727 *cur++ = hi; 930 *cur++ = hi;
728 } 931 }
729 break; 932 break;
… …
731 default: 934 default:
732 --dec_cur; 935 --dec_cur;
733 ERR ("illegal backslash escape sequence in string"); 936 ERR ("illegal backslash escape sequence in string");
734 } 937 }
735 } 938 }
736 else if (expect_true (ch >= 0x20 && ch <= 0x7f)) 939 else if (expect_true (ch >= 0x20 && ch < 0x80))
737 *cur++ = ch; 940 *cur++ = ch;
738 else if (ch >= 0x80) 941 else if (ch >= 0x80)
739 { 942 {
740 STRLEN clen; 943 STRLEN clen;
741 UV uch; 944 UV uch;
… …
862 is_nv = 1; 1065 is_nv = 1;
863 } 1066 }
864 1067
865 if (!is_nv) 1068 if (!is_nv)
866 { 1069 {
1070 int len = dec->cur - start;
1071
867 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so 1072 // special case the rather common 1..5-digit-int case
868 if (*start == '-') 1073 if (*start == '-')
869 switch (dec->cur - start) 1074 switch (len)
870 { 1075 {
871 case 2: return newSViv (-( start [1] - '0' * 1)); 1076 case 2: return newSViv (-( start [1] - '0' * 1));
872 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); 1077 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
873 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); 1078 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
874 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); 1079 case 5: return newSViv (-( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
1080 case 6: return newSViv (-(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
875 } 1081 }
876 else 1082 else
877 switch (dec->cur - start) 1083 switch (len)
878 { 1084 {
879 case 1: return newSViv ( start [0] - '0' * 1); 1085 case 1: return newSViv ( start [0] - '0' * 1);
880 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); 1086 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
881 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); 1087 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
882 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); 1088 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
1089 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
883 } 1090 }
884 1091
885 { 1092 {
886 UV uv; 1093 UV uv;
887 int numtype = grok_number (start, dec->cur - start, &uv); 1094 int numtype = grok_number (start, len, &uv);
888 if (numtype & IS_NUMBER_IN_UV) 1095 if (numtype & IS_NUMBER_IN_UV)
889 if (numtype & IS_NUMBER_NEG) 1096 if (numtype & IS_NUMBER_NEG)
890 { 1097 {
891 if (uv < (UV)IV_MIN) 1098 if (uv < (UV)IV_MIN)
892 return newSViv (-(IV)uv); 1099 return newSViv (-(IV)uv);
893 } 1100 }
894 else 1101 else
895 return newSVuv (uv); 1102 return newSVuv (uv);
896
897 // here would likely be the place for bigint support
898 } 1103 }
899 }
900 1104
901 // if we ever support bigint or bigfloat, this is the place for bigfloat 1105 len -= *start == '-' ? 1 : 0;
1106
1107 // does not fit into IV or UV, try NV
1108 if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len)
1109 #if defined (LDBL_DIG)
1110 || (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len)
1111 #endif
1112 )
1113 // fits into NV without loss of precision
1114 return newSVnv (Atof (start));
1115
1116 // everything else fails, convert it to a string
1117 return newSVpvn (start, dec->cur - start);
1118 }
1119
1120 // loss of precision here
902 return newSVnv (Atof (start)); 1121 return newSVnv (Atof (start));
903 1122
904fail: 1123fail:
905 return 0; 1124 return 0;
906} 1125}
… …
936 1155
937 if (*dec->cur != ',') 1156 if (*dec->cur != ',')
938 ERR (", or ] expected while parsing array"); 1157 ERR (", or ] expected while parsing array");
939 1158
940 ++dec->cur; 1159 ++dec->cur;
1160
1161 decode_ws (dec);
1162
1163 if (*dec->cur == ']' && dec->json.flags & F_RELAXED)
1164 {
1165 ++dec->cur;
1166 break;
1167 }
941 } 1168 }
942 1169
943 DEC_DEC_DEPTH; 1170 DEC_DEC_DEPTH;
944 return newRV_noinc ((SV *)av); 1171 return newRV_noinc ((SV *)av);
945 1172
… …
950} 1177}
951 1178
952static SV * 1179static SV *
953decode_hv (dec_t *dec) 1180decode_hv (dec_t *dec)
954{ 1181{
1182 SV *sv;
955 HV *hv = newHV (); 1183 HV *hv = newHV ();
956 1184
957 DEC_INC_DEPTH; 1185 DEC_INC_DEPTH;
958 decode_ws (dec); 1186 decode_ws (dec);
959 1187
960 if (*dec->cur == '}') 1188 if (*dec->cur == '}')
961 ++dec->cur; 1189 ++dec->cur;
962 else 1190 else
963 for (;;) 1191 for (;;)
964 { 1192 {
1193 EXPECT_CH ('"');
1194
1195 // heuristic: assume that
1196 // a) decode_str + hv_store_ent are abysmally slow.
1197 // b) most hash keys are short, simple ascii text.
1198 // => try to "fast-match" such strings to avoid
1199 // the overhead of decode_str + hv_store_ent.
1200 {
965 SV *key, *value; 1201 SV *value;
1202 char *p = dec->cur;
1203 char *e = p + 24; // only try up to 24 bytes
966 1204
967 decode_ws (dec); EXPECT_CH ('"'); 1205 for (;;)
968
969 key = decode_str (dec);
970 if (!key)
971 goto fail;
972
973 decode_ws (dec); EXPECT_CH (':');
974
975 value = decode_sv (dec);
976 if (!value)
977 { 1206 {
1207 // the >= 0x80 is false on most architectures
1208 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1209 {
1210 // slow path, back up and use decode_str
1211 SV *key = decode_str (dec);
1212 if (!key)
1213 goto fail;
1214
1215 decode_ws (dec); EXPECT_CH (':');
1216
1217 decode_ws (dec);
1218 value = decode_sv (dec);
1219 if (!value)
1220 {
1221 SvREFCNT_dec (key);
1222 goto fail;
1223 }
1224
1225 hv_store_ent (hv, key, value, 0);
978 SvREFCNT_dec (key); 1226 SvREFCNT_dec (key);
1227
1228 break;
1229 }
1230 else if (*p == '"')
1231 {
1232 // fast path, got a simple key
1233 char *key = dec->cur;
1234 int len = p - key;
1235 dec->cur = p + 1;
1236
1237 decode_ws (dec); EXPECT_CH (':');
1238
1239 decode_ws (dec);
1240 value = decode_sv (dec);
1241 if (!value)
979 goto fail; 1242 goto fail;
1243
1244 hv_store (hv, key, len, value, 0);
1245
1246 break;
1247 }
1248
1249 ++p;
980 } 1250 }
981 1251 }
982 hv_store_ent (hv, key, value, 0);
983 SvREFCNT_dec (key);
984 1252
985 decode_ws (dec); 1253 decode_ws (dec);
986 1254
987 if (*dec->cur == '}') 1255 if (*dec->cur == '}')
988 { 1256 {
… …
992 1260
993 if (*dec->cur != ',') 1261 if (*dec->cur != ',')
994 ERR (", or } expected while parsing object/hash"); 1262 ERR (", or } expected while parsing object/hash");
995 1263
996 ++dec->cur; 1264 ++dec->cur;
1265
1266 decode_ws (dec);
1267
1268 if (*dec->cur == '}' && dec->json.flags & F_RELAXED)
1269 {
1270 ++dec->cur;
1271 break;
1272 }
997 } 1273 }
998 1274
999 DEC_DEC_DEPTH; 1275 DEC_DEC_DEPTH;
1000 return newRV_noinc ((SV *)hv); 1276 sv = newRV_noinc ((SV *)hv);
1277
1278 // check filter callbacks
1279 if (dec->json.flags & F_HOOK)
1280 {
1281 if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
1282 {
1283 HE *cb, *he;
1284
1285 hv_iterinit (hv);
1286 he = hv_iternext (hv);
1287 hv_iterinit (hv);
1288
1289 // the next line creates a mortal sv each time its called.
1290 // might want to optimise this for common cases.
1291 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1292
1293 if (cb)
1294 {
1295 dSP;
1296 int count;
1297
1298 ENTER; SAVETMPS; PUSHMARK (SP);
1299 XPUSHs (HeVAL (he));
1300
1301 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1302
1303 if (count == 1)
1304 {
1305 sv = newSVsv (POPs);
1306 FREETMPS; LEAVE;
1307 return sv;
1308 }
1309
1310 FREETMPS; LEAVE;
1311 }
1312 }
1313
1314 if (dec->json.cb_object)
1315 {
1316 dSP;
1317 int count;
1318
1319 ENTER; SAVETMPS; PUSHMARK (SP);
1320 XPUSHs (sv_2mortal (sv));
1321
1322 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1323
1324 if (count == 1)
1325 {
1326 sv = newSVsv (POPs);
1327 FREETMPS; LEAVE;
1328 return sv;
1329 }
1330
1331 SvREFCNT_inc (sv);
1332 FREETMPS; LEAVE;
1333 }
1334 }
1335
1336 return sv;
1001 1337
1002fail: 1338fail:
1003 SvREFCNT_dec (hv); 1339 SvREFCNT_dec (hv);
1004 DEC_DEC_DEPTH; 1340 DEC_DEC_DEPTH;
1005 return 0; 1341 return 0;
1006} 1342}
1007 1343
1008static SV * 1344static SV *
1009decode_sv (dec_t *dec) 1345decode_sv (dec_t *dec)
1010{ 1346{
1011 decode_ws (dec);
1012
1013 // the beauty of JSON: you need exactly one character lookahead 1347 // the beauty of JSON: you need exactly one character lookahead
1014 // to parse anything. 1348 // to parse everything.
1015 switch (*dec->cur) 1349 switch (*dec->cur)
1016 { 1350 {
1017 case '"': ++dec->cur; return decode_str (dec); 1351 case '"': ++dec->cur; return decode_str (dec);
1018 case '[': ++dec->cur; return decode_av (dec); 1352 case '[': ++dec->cur; return decode_av (dec);
1019 case '{': ++dec->cur; return decode_hv (dec); 1353 case '{': ++dec->cur; return decode_hv (dec);
1020 1354
1021 case '-': 1355 case '-':
1022 case '0': case '1': case '2': case '3': case '4': 1356 case '0': case '1': case '2': case '3': case '4':
1023 case '5': case '6': case '7': case '8': case '9': 1357 case '5': case '6': case '7': case '8': case '9':
1024 return decode_num (dec); 1358 return decode_num (dec);
1025 1359
1026 case 't': 1360 case 't':
1027 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1361 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1028 { 1362 {
1029 dec->cur += 4; 1363 dec->cur += 4;
1364#if JSON_SLOW
1365 json_true = get_bool ("JSON::XS::true");
1366#endif
1030 return newSViv (1); 1367 return newSVsv (json_true);
1031 } 1368 }
1032 else 1369 else
1033 ERR ("'true' expected"); 1370 ERR ("'true' expected");
1034 1371
1035 break; 1372 break;
1036 1373
1037 case 'f': 1374 case 'f':
1038 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1375 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1039 { 1376 {
1040 dec->cur += 5; 1377 dec->cur += 5;
1378#if JSON_SLOW
1379 json_false = get_bool ("JSON::XS::false");
1380#endif
1041 return newSViv (0); 1381 return newSVsv (json_false);
1042 } 1382 }
1043 else 1383 else
1044 ERR ("'false' expected"); 1384 ERR ("'false' expected");
1045 1385
1046 break; 1386 break;
… …
1064fail: 1404fail:
1065 return 0; 1405 return 0;
1066} 1406}
1067 1407
1068static SV * 1408static SV *
1069decode_json (SV *string, U32 flags, UV *offset_return) 1409decode_json (SV *string, JSON *json, STRLEN *offset_return)
1070{ 1410{
1071 dec_t dec; 1411 dec_t dec;
1072 UV offset; 1412 STRLEN offset;
1073 SV *sv; 1413 SV *sv;
1074 1414
1075 SvGETMAGIC (string); 1415 SvGETMAGIC (string);
1076 SvUPGRADE (string, SVt_PV); 1416 SvUPGRADE (string, SVt_PV);
1077 1417
1418 /* work around a bug in perl 5.10, which causes SvCUR to fail an
1419 * assertion with -DDEBUGGING, although SvCUR is documented to
1420 * return the xpv_cur field which certainly exists after upgrading.
1421 * according to nicholas clark, calling SvPOK fixes this.
1422 * But it doesn't fix it, so try another workaround, call SvPV_nolen
1423 * and hope for the best.
1424 * Damnit, SvPV_nolen still trips over yet another assertion. This
1425 * assertion business is seriously broken, try yet another workaround
1426 * for the broken -DDEBUGGING.
1427 */
1428#ifdef DEBUGGING
1429 offset = SvOK (string) ? sv_len (string) : 0;
1430#else
1431 offset = SvCUR (string);
1432#endif
1433
1434 if (offset > json->max_size && json->max_size)
1435 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1436 (unsigned long)SvCUR (string), (unsigned long)json->max_size);
1437
1078 if (flags & F_UTF8) 1438 if (json->flags & F_UTF8)
1079 sv_utf8_downgrade (string, 0); 1439 sv_utf8_downgrade (string, 0);
1080 else 1440 else
1081 sv_utf8_upgrade (string); 1441 sv_utf8_upgrade (string);
1082 1442
1083 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1443 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
1084 1444
1085 dec.flags = flags; 1445 dec.json = *json;
1086 dec.cur = SvPVX (string); 1446 dec.cur = SvPVX (string);
1087 dec.end = SvEND (string); 1447 dec.end = SvEND (string);
1088 dec.err = 0; 1448 dec.err = 0;
1089 dec.depth = 0; 1449 dec.depth = 0;
1090 dec.maxdepth = DEC_DEPTH (dec.flags); 1450
1451 if (dec.json.cb_object || dec.json.cb_sk_object)
1452 dec.json.flags |= F_HOOK;
1091 1453
1092 *dec.end = 0; // this should basically be a nop, too, but make sure it's there 1454 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1455
1456 decode_ws (&dec);
1093 sv = decode_sv (&dec); 1457 sv = decode_sv (&dec);
1094 1458
1095 if (!(offset_return || !sv)) 1459 if (!(offset_return || !sv))
1096 { 1460 {
1097 // check for trailing garbage 1461 // check for trailing garbage
… …
1105 } 1469 }
1106 } 1470 }
1107 1471
1108 if (offset_return || !sv) 1472 if (offset_return || !sv)
1109 { 1473 {
1110 offset = dec.flags & F_UTF8 1474 offset = dec.json.flags & F_UTF8
1111 ? dec.cur - SvPVX (string) 1475 ? dec.cur - SvPVX (string)
1112 : utf8_distance (dec.cur, SvPVX (string)); 1476 : utf8_distance (dec.cur, SvPVX (string));
1113 1477
1114 if (offset_return) 1478 if (offset_return)
1115 *offset_return = offset; 1479 *offset_return = offset;
… …
1134 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1498 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1135 } 1499 }
1136 1500
1137 sv = sv_2mortal (sv); 1501 sv = sv_2mortal (sv);
1138 1502
1139 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1503 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))
1140 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); 1504 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1141 1505
1142 return sv; 1506 return sv;
1507}
1508
1509/////////////////////////////////////////////////////////////////////////////
1510// incremental parser
1511
1512static void
1513incr_parse (JSON *self)
1514{
1515 const char *p = SvPVX (self->incr_text) + self->incr_pos;
1516
1517 for (;;)
1518 {
1519 //printf ("loop pod %d *p<%c><%s>, mode %d nest %d\n", p - SvPVX (self->incr_text), *p, p, self->incr_mode, self->incr_nest);//D
1520 switch (self->incr_mode)
1521 {
1522 // only used for intiial whitespace skipping
1523 case INCR_M_WS:
1524 for (;;)
1525 {
1526 if (*p > 0x20)
1527 {
1528 self->incr_mode = INCR_M_JSON;
1529 goto incr_m_json;
1530 }
1531 else if (!*p)
1532 goto interrupt;
1533
1534 ++p;
1535 }
1536
1537 // skip a single char inside a string (for \\-processing)
1538 case INCR_M_BS:
1539 if (!*p)
1540 goto interrupt;
1541
1542 ++p;
1543 self->incr_mode = INCR_M_STR;
1544 goto incr_m_str;
1545
1546 // inside a string
1547 case INCR_M_STR:
1548 incr_m_str:
1549 for (;;)
1550 {
1551 if (*p == '"')
1552 {
1553 ++p;
1554 self->incr_mode = INCR_M_JSON;
1555
1556 if (!self->incr_nest)
1557 goto interrupt;
1558
1559 goto incr_m_json;
1560 }
1561 else if (*p == '\\')
1562 {
1563 ++p; // "virtually" consumes character after \
1564
1565 if (!*p) // if at end of string we have to switch modes
1566 {
1567 self->incr_mode = INCR_M_BS;
1568 goto interrupt;
1569 }
1570 }
1571 else if (!*p)
1572 goto interrupt;
1573
1574 ++p;
1575 }
1576
1577 // after initial ws, outside string
1578 case INCR_M_JSON:
1579 incr_m_json:
1580 for (;;)
1581 {
1582 switch (*p++)
1583 {
1584 case 0:
1585 --p;
1586 goto interrupt;
1587
1588 case 0x09:
1589 case 0x0a:
1590 case 0x0d:
1591 case 0x20:
1592 if (!self->incr_nest)
1593 {
1594 --p; // do not eat the whitespace, let the next round do it
1595 goto interrupt;
1596 }
1597 break;
1598
1599 case '"':
1600 self->incr_mode = INCR_M_STR;
1601 goto incr_m_str;
1602
1603 case '[':
1604 case '{':
1605 if (++self->incr_nest > self->max_depth)
1606 croak (ERR_NESTING_EXCEEDED);
1607 break;
1608
1609 case ']':
1610 case '}':
1611 if (--self->incr_nest <= 0)
1612 goto interrupt;
1613 }
1614 }
1615 }
1616
1617 modechange:
1618 ;
1619 }
1620
1621interrupt:
1622 self->incr_pos = p - SvPVX (self->incr_text);
1623 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D
1143} 1624}
1144 1625
1145///////////////////////////////////////////////////////////////////////////// 1626/////////////////////////////////////////////////////////////////////////////
1146// XS interface functions 1627// XS interface functions
1147 1628
… …
1156 i >= '0' && i <= '9' ? i - '0' 1637 i >= '0' && i <= '9' ? i - '0'
1157 : i >= 'a' && i <= 'f' ? i - 'a' + 10 1638 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1158 : i >= 'A' && i <= 'F' ? i - 'A' + 10 1639 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1159 : -1; 1640 : -1;
1160 1641
1161 json_stash = gv_stashpv ("JSON::XS", 1); 1642 json_stash = gv_stashpv ("JSON::XS" , 1);
1643 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1644
1645 json_true = get_bool ("JSON::XS::true");
1646 json_false = get_bool ("JSON::XS::false");
1162} 1647}
1163 1648
1164PROTOTYPES: DISABLE 1649PROTOTYPES: DISABLE
1165 1650
1166SV *new (char *dummy) 1651void CLONE (...)
1167 CODE: 1652 CODE:
1168 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1653 json_stash = 0;
1654 json_boolean_stash = 0;
1655
1656void new (char *klass)
1657 PPCODE:
1658{
1659 SV *pv = NEWSV (0, sizeof (JSON));
1660 SvPOK_only (pv);
1661 json_init ((JSON *)SvPVX (pv));
1662 XPUSHs (sv_2mortal (sv_bless (
1663 newRV_noinc (pv),
1664 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1665 )));
1666}
1667
1668void ascii (JSON *self, int enable = 1)
1669 ALIAS:
1670 ascii = F_ASCII
1671 latin1 = F_LATIN1
1672 utf8 = F_UTF8
1673 indent = F_INDENT
1674 canonical = F_CANONICAL
1675 space_before = F_SPACE_BEFORE
1676 space_after = F_SPACE_AFTER
1677 pretty = F_PRETTY
1678 allow_nonref = F_ALLOW_NONREF
1679 shrink = F_SHRINK
1680 allow_blessed = F_ALLOW_BLESSED
1681 convert_blessed = F_CONV_BLESSED
1682 relaxed = F_RELAXED
1683 allow_unknown = F_ALLOW_UNKNOWN
1684 PPCODE:
1685{
1686 if (enable)
1687 self->flags |= ix;
1688 else
1689 self->flags &= ~ix;
1690
1691 XPUSHs (ST (0));
1692}
1693
1694void get_ascii (JSON *self)
1695 ALIAS:
1696 get_ascii = F_ASCII
1697 get_latin1 = F_LATIN1
1698 get_utf8 = F_UTF8
1699 get_indent = F_INDENT
1700 get_canonical = F_CANONICAL
1701 get_space_before = F_SPACE_BEFORE
1702 get_space_after = F_SPACE_AFTER
1703 get_allow_nonref = F_ALLOW_NONREF
1704 get_shrink = F_SHRINK
1705 get_allow_blessed = F_ALLOW_BLESSED
1706 get_convert_blessed = F_CONV_BLESSED
1707 get_relaxed = F_RELAXED
1708 get_allow_unknown = F_ALLOW_UNKNOWN
1709 PPCODE:
1710 XPUSHs (boolSV (self->flags & ix));
1711
1712void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
1713 PPCODE:
1714 self->max_depth = max_depth;
1715 XPUSHs (ST (0));
1716
1717U32 get_max_depth (JSON *self)
1718 CODE:
1719 RETVAL = self->max_depth;
1169 OUTPUT: 1720 OUTPUT:
1170 RETVAL 1721 RETVAL
1171 1722
1172SV *ascii (SV *self, int enable = 1) 1723void max_size (JSON *self, U32 max_size = 0)
1173 ALIAS: 1724 PPCODE:
1174 ascii = F_ASCII 1725 self->max_size = max_size;
1175 latin1 = F_LATIN1 1726 XPUSHs (ST (0));
1176 utf8 = F_UTF8 1727
1177 indent = F_INDENT 1728int get_max_size (JSON *self)
1178 canonical = F_CANONICAL
1179 space_before = F_SPACE_BEFORE
1180 space_after = F_SPACE_AFTER
1181 pretty = F_PRETTY
1182 allow_nonref = F_ALLOW_NONREF
1183 shrink = F_SHRINK
1184 CODE: 1729 CODE:
1185{ 1730 RETVAL = self->max_size;
1186 UV *uv = SvJSON (self);
1187 if (enable)
1188 *uv |= ix;
1189 else
1190 *uv &= ~ix;
1191
1192 RETVAL = newSVsv (self);
1193}
1194 OUTPUT: 1731 OUTPUT:
1195 RETVAL 1732 RETVAL
1196 1733
1197SV *max_depth (SV *self, UV max_depth = 0x80000000UL) 1734void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1735 PPCODE:
1736{
1737 SvREFCNT_dec (self->cb_object);
1738 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
1739
1740 XPUSHs (ST (0));
1741}
1742
1743void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
1744 PPCODE:
1745{
1746 if (!self->cb_sk_object)
1747 self->cb_sk_object = newHV ();
1748
1749 if (SvOK (cb))
1750 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1751 else
1752 {
1753 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1754
1755 if (!HvKEYS (self->cb_sk_object))
1756 {
1757 SvREFCNT_dec (self->cb_sk_object);
1758 self->cb_sk_object = 0;
1759 }
1760 }
1761
1762 XPUSHs (ST (0));
1763}
1764
1765void encode (JSON *self, SV *scalar)
1766 PPCODE:
1767 XPUSHs (encode_json (scalar, self));
1768
1769void decode (JSON *self, SV *jsonstr)
1770 PPCODE:
1771 XPUSHs (decode_json (jsonstr, self, 0));
1772
1773void decode_prefix (JSON *self, SV *jsonstr)
1774 PPCODE:
1775{
1776 STRLEN offset;
1777 EXTEND (SP, 2);
1778 PUSHs (decode_json (jsonstr, self, &offset));
1779 PUSHs (sv_2mortal (newSVuv (offset)));
1780}
1781
1782void incr_parse (JSON *self, SV *jsonstr = 0)
1783 PPCODE:
1784{
1785 if (!self->incr_text)
1786 self->incr_text = newSVpvn ("", 0);
1787
1788 // append data, if any
1789 if (jsonstr)
1790 {
1791 if (SvUTF8 (jsonstr) && !SvUTF8 (self->incr_text))
1792 {
1793 /* utf-8-ness differs, need to upgrade */
1794 sv_utf8_upgrade (self->incr_text);
1795
1796 if (self->incr_pos)
1797 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
1798 - (U8 *)SvPVX (self->incr_text);
1799 }
1800
1801 {
1802 STRLEN len;
1803 const char *str = SvPV (jsonstr, len);
1804 SvGROW (self->incr_text, SvCUR (self->incr_text) + len + 1);
1805 Move (str, SvEND (self->incr_text), len, char);
1806 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
1807 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
1808 }
1809 }
1810
1811 if (GIMME_V != G_VOID)
1812 do
1813 {
1814 STRLEN offset;
1815
1816 if (!INCR_DONE (self))
1817 {
1818 incr_parse (self);
1819
1820 if (self->incr_pos > self->max_size && self->max_size)
1821 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1822 (unsigned long)self->incr_pos, (unsigned long)self->max_size);
1823
1824 if (!INCR_DONE (self))
1825 break;
1826 }
1827
1828 XPUSHs (decode_json (self->incr_text, self, &offset));
1829
1830 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + offset);
1831 self->incr_pos -= offset;
1832 self->incr_nest = 0;
1833 self->incr_mode = 0;
1834 }
1835 while (GIMME_V == G_ARRAY);
1836}
1837
1838SV *incr_text (JSON *self)
1839 ATTRS: lvalue
1198 CODE: 1840 CODE:
1199{ 1841{
1200 UV *uv = SvJSON (self); 1842 if (self->incr_pos)
1201 UV log2 = 0; 1843 croak ("incr_text can not be called when the incremental parser already started parsing");
1202 1844
1203 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL; 1845 RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef;
1204
1205 while ((1UL << log2) < max_depth)
1206 ++log2;
1207
1208 *uv = *uv & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1209
1210 RETVAL = newSVsv (self);
1211} 1846}
1212 OUTPUT: 1847 OUTPUT:
1213 RETVAL 1848 RETVAL
1214 1849
1850void incr_skip (JSON *self)
1851 CODE:
1852{
1853 if (self->incr_pos)
1854 {
1855 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + self->incr_pos);
1856 self->incr_pos = 0;
1857 self->incr_nest = 0;
1858 self->incr_mode = 0;
1859 }
1860}
1861
1862void incr_reset (JSON *self)
1863 CODE:
1864{
1865 SvREFCNT_dec (self->incr_text);
1866 self->incr_text = 0;
1867 self->incr_pos = 0;
1868 self->incr_nest = 0;
1869 self->incr_mode = 0;
1870}
1871
1872void DESTROY (JSON *self)
1873 CODE:
1874 SvREFCNT_dec (self->cb_sk_object);
1875 SvREFCNT_dec (self->cb_object);
1876 SvREFCNT_dec (self->incr_text);
1877
1878PROTOTYPES: ENABLE
1879
1215void encode (SV *self, SV *scalar) 1880void encode_json (SV *scalar)
1881 ALIAS:
1882 to_json_ = 0
1883 encode_json = F_UTF8
1216 PPCODE: 1884 PPCODE:
1885{
1886 JSON json;
1887 json_init (&json);
1888 json.flags |= ix;
1217 XPUSHs (encode_json (scalar, *SvJSON (self))); 1889 XPUSHs (encode_json (scalar, &json));
1890}
1218 1891
1219void decode (SV *self, SV *jsonstr) 1892void decode_json (SV *jsonstr)
1893 ALIAS:
1894 from_json_ = 0
1895 decode_json = F_UTF8
1220 PPCODE: 1896 PPCODE:
1897{
1898 JSON json;
1899 json_init (&json);
1900 json.flags |= ix;
1221 XPUSHs (decode_json (jsonstr, *SvJSON (self), 0)); 1901 XPUSHs (decode_json (jsonstr, &json, 0));
1222
1223void decode_prefix (SV *self, SV *jsonstr)
1224 PPCODE:
1225{
1226 UV offset;
1227 EXTEND (SP, 2);
1228 PUSHs (decode_json (jsonstr, *SvJSON (self), &offset));
1229 PUSHs (sv_2mortal (newSVuv (offset)));
1230} 1902}
1231 1903
1232PROTOTYPES: ENABLE
1233 1904
1234void to_json (SV *scalar)
1235 ALIAS:
1236 objToJson = 0
1237 PPCODE:
1238 XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8));
1239
1240void from_json (SV *jsonstr)
1241 ALIAS:
1242 jsonToObj = 0
1243 PPCODE:
1244 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8, 0));
1245

Diff Legend

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