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 1.36 by root, Wed Jun 6 17:27:33 2007 UTC vs.
Revision 1.80 by root, Wed Mar 26 01:32:53 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
35#define F_RELAXED 0x00001000UL
36
23#define F_MAXDEPTH 0xf8000000UL 37#define F_MAXDEPTH 0xf8000000UL
24#define S_MAXDEPTH 27 38#define S_MAXDEPTH 27
39#define F_MAXSIZE 0x01f00000UL
40#define S_MAXSIZE 20
41#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
25 42
26#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH)) 43#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
27 44#define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE ))
28// F_SELFCONVERT? <=> to_json/toJson
29// F_BLESSED? <=> { $__class__$ => }
30 45
31#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 46#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
32#define F_DEFAULT (9UL << S_MAXDEPTH) 47#define F_DEFAULT (9UL << S_MAXDEPTH)
33 48
34#define INIT_SIZE 32 // initial scalar size to be allocated 49#define INIT_SIZE 32 // initial scalar size to be allocated
35#define INDENT_STEP 3 // spaces per indentation level 50#define INDENT_STEP 3 // spaces per indentation level
36 51
37#define SHORT_STRING_LEN 512 // special-case strings of up to this size 52#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
38 53
39#define SB do { 54#define SB do {
40#define SE } while (0) 55#define SE } while (0)
41 56
42#if __GNUC__ >= 3 57#if __GNUC__ >= 3
43# define expect(expr,value) __builtin_expect ((expr),(value)) 58# define expect(expr,value) __builtin_expect ((expr), (value))
44# define inline inline 59# define INLINE static inline
45#else 60#else
46# define expect(expr,value) (expr) 61# define expect(expr,value) (expr)
47# define inline static 62# define INLINE static
48#endif 63#endif
49 64
50#define expect_false(expr) expect ((expr) != 0, 0) 65#define expect_false(expr) expect ((expr) != 0, 0)
51#define expect_true(expr) expect ((expr) != 0, 1) 66#define expect_true(expr) expect ((expr) != 0, 1)
52 67
68#define IN_RANGE_INC(type,val,beg,end) \
69 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \
70 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
71
72#ifdef USE_ITHREADS
73# define JSON_SLOW 1
74# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1))
75#else
76# define JSON_SLOW 0
77# define JSON_STASH json_stash
78#endif
79
53static HV *json_stash; // JSON::XS:: 80static HV *json_stash, *json_boolean_stash; // JSON::XS::
81static SV *json_true, *json_false;
82
83enum {
84 INCR_M_WS = 0, // initial whitespace skipping, must be 0
85 INCR_M_STR, // inside string
86 INCR_M_BS, // inside backslash
87 INCR_M_JSON // outside anything, count nesting
88};
89
90#define INCR_DONE(json) (!(json)->incr_nest && (json)->incr_mode == INCR_M_JSON)
91
92typedef struct {
93 U32 flags;
94 SV *cb_object;
95 HV *cb_sk_object;
96
97 // for the incremental parser
98 SV *incr_text; // the source text so far
99 STRLEN incr_pos; // the current offset into the text
100 int incr_nest; // {[]}-nesting level
101 int incr_mode;
102} JSON;
54 103
55///////////////////////////////////////////////////////////////////////////// 104/////////////////////////////////////////////////////////////////////////////
56// utility functions 105// utility functions
57 106
58static UV * 107INLINE void
59SvJSON (SV *sv)
60{
61 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash))
62 croak ("object is not of type JSON::XS");
63
64 return &SvUVX (SvRV (sv));
65}
66
67static void
68shrink (SV *sv) 108shrink (SV *sv)
69{ 109{
70 sv_utf8_downgrade (sv, 1); 110 sv_utf8_downgrade (sv, 1);
71 if (SvLEN (sv) > SvCUR (sv) + 1) 111 if (SvLEN (sv) > SvCUR (sv) + 1)
72 { 112 {
… …
81// decode an utf-8 character and return it, or (UV)-1 in 121// decode an utf-8 character and return it, or (UV)-1 in
82// case of an error. 122// case of an error.
83// we special-case "safe" characters from U+80 .. U+7FF, 123// we special-case "safe" characters from U+80 .. U+7FF,
84// but use the very good perl function to parse anything else. 124// but use the very good perl function to parse anything else.
85// note that we never call this function for a ascii codepoints 125// note that we never call this function for a ascii codepoints
86inline UV 126INLINE UV
87decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) 127decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
88{ 128{
89 if (expect_false (s[0] > 0xdf || s[0] < 0xc2)) 129 if (expect_true (len >= 2
90 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); 130 && IN_RANGE_INC (char, s[0], 0xc2, 0xdf)
91 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) 131 && IN_RANGE_INC (char, s[1], 0x80, 0xbf)))
92 { 132 {
93 *clen = 2; 133 *clen = 2;
94 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 134 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
95 } 135 }
96 else 136 else
97 { 137 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
98 *clen = (STRLEN)-1; 138}
99 return (UV)-1; 139
100 } 140// likewise for encoding, also never called for ascii codepoints
141// this function takes advantage of this fact, although current gccs
142// seem to optimise the check for >= 0x80 away anyways
143INLINE unsigned char *
144encode_utf8 (unsigned char *s, UV ch)
145{
146 if (expect_false (ch < 0x000080))
147 *s++ = ch;
148 else if (expect_true (ch < 0x000800))
149 *s++ = 0xc0 | ( ch >> 6),
150 *s++ = 0x80 | ( ch & 0x3f);
151 else if ( ch < 0x010000)
152 *s++ = 0xe0 | ( ch >> 12),
153 *s++ = 0x80 | ((ch >> 6) & 0x3f),
154 *s++ = 0x80 | ( ch & 0x3f);
155 else if ( ch < 0x110000)
156 *s++ = 0xf0 | ( ch >> 18),
157 *s++ = 0x80 | ((ch >> 12) & 0x3f),
158 *s++ = 0x80 | ((ch >> 6) & 0x3f),
159 *s++ = 0x80 | ( ch & 0x3f);
160
161 return s;
101} 162}
102 163
103///////////////////////////////////////////////////////////////////////////// 164/////////////////////////////////////////////////////////////////////////////
104// encoder 165// encoder
105 166
… …
107typedef struct 168typedef struct
108{ 169{
109 char *cur; // SvPVX (sv) + current output position 170 char *cur; // SvPVX (sv) + current output position
110 char *end; // SvEND (sv) 171 char *end; // SvEND (sv)
111 SV *sv; // result scalar 172 SV *sv; // result scalar
112 U32 flags; // F_* 173 JSON json;
113 U32 indent; // indentation level 174 U32 indent; // indentation level
114 U32 maxdepth; // max. indentation/recursion level 175 U32 maxdepth; // max. indentation/recursion level
176 UV limit; // escape character values >= this value when encoding
115} enc_t; 177} enc_t;
116 178
117inline void 179INLINE void
118need (enc_t *enc, STRLEN len) 180need (enc_t *enc, STRLEN len)
119{ 181{
120 if (expect_false (enc->cur + len >= enc->end)) 182 if (expect_false (enc->cur + len >= enc->end))
121 { 183 {
122 STRLEN cur = enc->cur - SvPVX (enc->sv); 184 STRLEN cur = enc->cur - SvPVX (enc->sv);
… …
124 enc->cur = SvPVX (enc->sv) + cur; 186 enc->cur = SvPVX (enc->sv) + cur;
125 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 187 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
126 } 188 }
127} 189}
128 190
129inline void 191INLINE void
130encode_ch (enc_t *enc, char ch) 192encode_ch (enc_t *enc, char ch)
131{ 193{
132 need (enc, 1); 194 need (enc, 1);
133 *enc->cur++ = ch; 195 *enc->cur++ = ch;
134} 196}
… …
178 STRLEN clen; 240 STRLEN clen;
179 UV uch; 241 UV uch;
180 242
181 if (is_utf8) 243 if (is_utf8)
182 { 244 {
183 //uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
184 uch = decode_utf8 (str, end - str, &clen); 245 uch = decode_utf8 (str, end - str, &clen);
185 if (clen == (STRLEN)-1) 246 if (clen == (STRLEN)-1)
186 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str); 247 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
187 } 248 }
188 else 249 else
189 { 250 {
190 uch = ch; 251 uch = ch;
191 clen = 1; 252 clen = 1;
192 } 253 }
193 254
194 if (uch > 0x10FFFFUL) 255 if (uch < 0x80/*0x20*/ || uch >= enc->limit)
195 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
196
197 if (uch < 0x80 || enc->flags & F_ASCII || (enc->flags & F_LATIN1 && uch > 0xFF))
198 { 256 {
199 if (uch > 0xFFFFUL) 257 if (uch >= 0x10000UL)
200 { 258 {
259 if (uch >= 0x110000UL)
260 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
261
201 need (enc, len += 11); 262 need (enc, len += 11);
202 sprintf (enc->cur, "\\u%04x\\u%04x", 263 sprintf (enc->cur, "\\u%04x\\u%04x",
203 (int)((uch - 0x10000) / 0x400 + 0xD800), 264 (int)((uch - 0x10000) / 0x400 + 0xD800),
204 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 265 (int)((uch - 0x10000) % 0x400 + 0xDC00));
205 enc->cur += 12; 266 enc->cur += 12;
… …
216 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 277 *enc->cur++ = hexdigit [(uch >> 0) & 15];
217 } 278 }
218 279
219 str += clen; 280 str += clen;
220 } 281 }
221 else if (enc->flags & F_LATIN1) 282 else if (enc->json.flags & F_LATIN1)
222 { 283 {
223 *enc->cur++ = uch; 284 *enc->cur++ = uch;
224 str += clen; 285 str += clen;
225 } 286 }
226 else if (is_utf8) 287 else if (is_utf8)
… …
233 while (--clen); 294 while (--clen);
234 } 295 }
235 else 296 else
236 { 297 {
237 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed 298 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
238 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 299 enc->cur = encode_utf8 (enc->cur, uch);
239 ++str; 300 ++str;
240 } 301 }
241 } 302 }
242 } 303 }
243 } 304 }
244 305
245 --len; 306 --len;
246 } 307 }
247} 308}
248 309
249inline void 310INLINE void
250encode_indent (enc_t *enc) 311encode_indent (enc_t *enc)
251{ 312{
252 if (enc->flags & F_INDENT) 313 if (enc->json.flags & F_INDENT)
253 { 314 {
254 int spaces = enc->indent * INDENT_STEP; 315 int spaces = enc->indent * INDENT_STEP;
255 316
256 need (enc, spaces); 317 need (enc, spaces);
257 memset (enc->cur, ' ', spaces); 318 memset (enc->cur, ' ', spaces);
258 enc->cur += spaces; 319 enc->cur += spaces;
259 } 320 }
260} 321}
261 322
262inline void 323INLINE void
263encode_space (enc_t *enc) 324encode_space (enc_t *enc)
264{ 325{
265 need (enc, 1); 326 need (enc, 1);
266 encode_ch (enc, ' '); 327 encode_ch (enc, ' ');
267} 328}
268 329
269inline void 330INLINE void
270encode_nl (enc_t *enc) 331encode_nl (enc_t *enc)
271{ 332{
272 if (enc->flags & F_INDENT) 333 if (enc->json.flags & F_INDENT)
273 { 334 {
274 need (enc, 1); 335 need (enc, 1);
275 encode_ch (enc, '\n'); 336 encode_ch (enc, '\n');
276 } 337 }
277} 338}
278 339
279inline void 340INLINE void
280encode_comma (enc_t *enc) 341encode_comma (enc_t *enc)
281{ 342{
282 encode_ch (enc, ','); 343 encode_ch (enc, ',');
283 344
284 if (enc->flags & F_INDENT) 345 if (enc->json.flags & F_INDENT)
285 encode_nl (enc); 346 encode_nl (enc);
286 else if (enc->flags & F_SPACE_AFTER) 347 else if (enc->json.flags & F_SPACE_AFTER)
287 encode_space (enc); 348 encode_space (enc);
288} 349}
289 350
290static void encode_sv (enc_t *enc, SV *sv); 351static void encode_sv (enc_t *enc, SV *sv);
291 352
… …
295 int i, len = av_len (av); 356 int i, len = av_len (av);
296 357
297 if (enc->indent >= enc->maxdepth) 358 if (enc->indent >= enc->maxdepth)
298 croak ("data structure too deep (hit recursion limit)"); 359 croak ("data structure too deep (hit recursion limit)");
299 360
300 encode_ch (enc, '['); encode_nl (enc); 361 encode_ch (enc, '[');
301 ++enc->indent; 362
363 if (len >= 0)
364 {
365 encode_nl (enc); ++enc->indent;
302 366
303 for (i = 0; i <= len; ++i) 367 for (i = 0; i <= len; ++i)
304 { 368 {
369 SV **svp = av_fetch (av, i, 0);
370
305 encode_indent (enc); 371 encode_indent (enc);
306 encode_sv (enc, *av_fetch (av, i, 0));
307 372
373 if (svp)
374 encode_sv (enc, *svp);
375 else
376 encode_str (enc, "null", 4, 0);
377
308 if (i < len) 378 if (i < len)
309 encode_comma (enc); 379 encode_comma (enc);
310 } 380 }
311 381
382 encode_nl (enc); --enc->indent; encode_indent (enc);
383 }
384
312 encode_nl (enc); 385 encode_ch (enc, ']');
313
314 --enc->indent;
315 encode_indent (enc); encode_ch (enc, ']');
316} 386}
317 387
318static void 388static void
319encode_he (enc_t *enc, HE *he) 389encode_hk (enc_t *enc, HE *he)
320{ 390{
321 encode_ch (enc, '"'); 391 encode_ch (enc, '"');
322 392
323 if (HeKLEN (he) == HEf_SVKEY) 393 if (HeKLEN (he) == HEf_SVKEY)
324 { 394 {
… …
334 else 404 else
335 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 405 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
336 406
337 encode_ch (enc, '"'); 407 encode_ch (enc, '"');
338 408
339 if (enc->flags & F_SPACE_BEFORE) encode_space (enc); 409 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
340 encode_ch (enc, ':'); 410 encode_ch (enc, ':');
341 if (enc->flags & F_SPACE_AFTER ) encode_space (enc); 411 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
342 encode_sv (enc, HeVAL (he));
343} 412}
344 413
345// compare hash entries, used when all keys are bytestrings 414// compare hash entries, used when all keys are bytestrings
346static int 415static int
347he_cmp_fast (const void *a_, const void *b_) 416he_cmp_fast (const void *a_, const void *b_)
… …
352 HE *b = *(HE **)b_; 421 HE *b = *(HE **)b_;
353 422
354 STRLEN la = HeKLEN (a); 423 STRLEN la = HeKLEN (a);
355 STRLEN lb = HeKLEN (b); 424 STRLEN lb = HeKLEN (b);
356 425
357 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 426 if (!(cmp = memcmp (HeKEY (b), HeKEY (a), lb < la ? lb : la)))
358 cmp = la - lb; 427 cmp = lb - la;
359 428
360 return cmp; 429 return cmp;
361} 430}
362 431
363// compare hash entries, used when some keys are sv's or utf-x 432// compare hash entries, used when some keys are sv's or utf-x
364static int 433static int
365he_cmp_slow (const void *a, const void *b) 434he_cmp_slow (const void *a, const void *b)
366{ 435{
367 return sv_cmp (HeSVKEY_force (*(HE **)a), HeSVKEY_force (*(HE **)b)); 436 return sv_cmp (HeSVKEY_force (*(HE **)b), HeSVKEY_force (*(HE **)a));
368} 437}
369 438
370static void 439static void
371encode_hv (enc_t *enc, HV *hv) 440encode_hv (enc_t *enc, HV *hv)
372{ 441{
373 int count, i; 442 HE *he;
374 443
375 if (enc->indent >= enc->maxdepth) 444 if (enc->indent >= enc->maxdepth)
376 croak ("data structure too deep (hit recursion limit)"); 445 croak ("data structure too deep (hit recursion limit)");
377 446
378 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 447 encode_ch (enc, '{');
379 448
380 if ((count = hv_iterinit (hv)))
381 {
382 // for canonical output we have to sort by keys first 449 // for canonical output we have to sort by keys first
383 // actually, this is mostly due to the stupid so-called 450 // actually, this is mostly due to the stupid so-called
384 // security workaround added somewhere in 5.8.x. 451 // security workaround added somewhere in 5.8.x.
385 // that randomises hash orderings 452 // that randomises hash orderings
386 if (enc->flags & F_CANONICAL) 453 if (enc->json.flags & F_CANONICAL)
454 {
455 int count = hv_iterinit (hv);
456
457 if (SvMAGICAL (hv))
387 { 458 {
459 // need to count by iterating. could improve by dynamically building the vector below
460 // but I don't care for the speed of this special case.
461 // note also that we will run into undefined behaviour when the two iterations
462 // do not result in the same count, something I might care for in some later release.
463
464 count = 0;
465 while (hv_iternext (hv))
466 ++count;
467
468 hv_iterinit (hv);
469 }
470
471 if (count)
472 {
388 int fast = 1; 473 int i, fast = 1;
389 HE *he;
390#if defined(__BORLANDC__) || defined(_MSC_VER) 474#if defined(__BORLANDC__) || defined(_MSC_VER)
391 HE **hes = _alloca (count * sizeof (HE)); 475 HE **hes = _alloca (count * sizeof (HE));
392#else 476#else
393 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode 477 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
394#endif 478#endif
… …
421 505
422 FREETMPS; 506 FREETMPS;
423 LEAVE; 507 LEAVE;
424 } 508 }
425 509
426 for (i = 0; i < count; ++i) 510 encode_nl (enc); ++enc->indent;
511
512 while (count--)
427 { 513 {
428 encode_indent (enc); 514 encode_indent (enc);
515 he = hes [count];
429 encode_he (enc, hes [i]); 516 encode_hk (enc, he);
517 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
430 518
431 if (i < count - 1) 519 if (count)
432 encode_comma (enc); 520 encode_comma (enc);
433 } 521 }
434 522
523 encode_nl (enc); --enc->indent; encode_indent (enc);
524 }
525 }
526 else
527 {
528 if (hv_iterinit (hv) || SvMAGICAL (hv))
529 if ((he = hv_iternext (hv)))
530 {
531 encode_nl (enc); ++enc->indent;
532
533 for (;;)
534 {
435 encode_nl (enc); 535 encode_indent (enc);
536 encode_hk (enc, he);
537 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
538
539 if (!(he = hv_iternext (hv)))
540 break;
541
542 encode_comma (enc);
543 }
544
545 encode_nl (enc); --enc->indent; encode_indent (enc);
546 }
547 }
548
549 encode_ch (enc, '}');
550}
551
552// encode objects, arrays and special \0=false and \1=true values.
553static void
554encode_rv (enc_t *enc, SV *sv)
555{
556 svtype svt;
557
558 SvGETMAGIC (sv);
559 svt = SvTYPE (sv);
560
561 if (expect_false (SvOBJECT (sv)))
562 {
563 HV *stash = !JSON_SLOW || json_boolean_stash
564 ? json_boolean_stash
565 : gv_stashpv ("JSON::XS::Boolean", 1);
566
567 if (SvSTASH (sv) == stash)
568 {
569 if (SvIV (sv))
570 encode_str (enc, "true", 4, 0);
571 else
572 encode_str (enc, "false", 5, 0);
436 } 573 }
437 else 574 else
438 { 575 {
439 HE *he = hv_iternext (hv); 576#if 0
440 577 if (0 && sv_derived_from (rv, "JSON::Literal"))
441 for (;;)
442 { 578 {
443 encode_indent (enc); 579 // not yet
444 encode_he (enc, he);
445
446 if (!(he = hv_iternext (hv)))
447 break;
448
449 encode_comma (enc);
450 } 580 }
581#endif
582 if (enc->json.flags & F_CONV_BLESSED)
583 {
584 // we re-bless the reference to get overload and other niceties right
585 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
451 586
587 if (to_json)
588 {
589 dSP;
590
591 ENTER; SAVETMPS; PUSHMARK (SP);
592 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
593
594 // calling with G_SCALAR ensures that we always get a 1 return value
595 PUTBACK;
596 call_sv ((SV *)GvCV (to_json), G_SCALAR);
597 SPAGAIN;
598
599 // catch this surprisingly common error
600 if (SvROK (TOPs) && SvRV (TOPs) == sv)
601 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
602
603 sv = POPs;
604 PUTBACK;
605
452 encode_nl (enc); 606 encode_sv (enc, sv);
607
608 FREETMPS; LEAVE;
609 }
610 else if (enc->json.flags & F_ALLOW_BLESSED)
611 encode_str (enc, "null", 4, 0);
612 else
613 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
614 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
615 }
616 else if (enc->json.flags & F_ALLOW_BLESSED)
617 encode_str (enc, "null", 4, 0);
618 else
619 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",
620 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
453 } 621 }
454 } 622 }
455
456 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
457}
458
459// encode objects, arrays and special \0=false and \1=true values.
460static void
461encode_rv (enc_t *enc, SV *sv)
462{
463 svtype svt;
464
465 SvGETMAGIC (sv);
466 svt = SvTYPE (sv);
467
468 if (svt == SVt_PVHV) 623 else if (svt == SVt_PVHV)
469 encode_hv (enc, (HV *)sv); 624 encode_hv (enc, (HV *)sv);
470 else if (svt == SVt_PVAV) 625 else if (svt == SVt_PVAV)
471 encode_av (enc, (AV *)sv); 626 encode_av (enc, (AV *)sv);
472 else if (svt < SVt_PVAV) 627 else if (svt < SVt_PVAV)
473 { 628 {
474 if (SvNIOK (sv) && SvIV (sv) == 0) 629 STRLEN len = 0;
630 char *pv = svt ? SvPV (sv, len) : 0;
631
632 if (len == 1 && *pv == '1')
633 encode_str (enc, "true", 4, 0);
634 else if (len == 1 && *pv == '0')
475 encode_str (enc, "false", 5, 0); 635 encode_str (enc, "false", 5, 0);
476 else if (SvNIOK (sv) && SvIV (sv) == 1)
477 encode_str (enc, "true", 4, 0);
478 else 636 else
479 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", 637 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
480 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 638 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
481 } 639 }
482 else 640 else
… …
497 encode_str (enc, str, len, SvUTF8 (sv)); 655 encode_str (enc, str, len, SvUTF8 (sv));
498 encode_ch (enc, '"'); 656 encode_ch (enc, '"');
499 } 657 }
500 else if (SvNOKp (sv)) 658 else if (SvNOKp (sv))
501 { 659 {
660 // trust that perl will do the right thing w.r.t. JSON syntax.
502 need (enc, NV_DIG + 32); 661 need (enc, NV_DIG + 32);
503 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 662 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
504 enc->cur += strlen (enc->cur); 663 enc->cur += strlen (enc->cur);
505 } 664 }
506 else if (SvIOKp (sv)) 665 else if (SvIOKp (sv))
507 { 666 {
508 // we assume we can always read an IV as a UV 667 // we assume we can always read an IV as a UV and vice versa
509 if (SvUV (sv) & ~(UV)0x7fff) 668 // we assume two's complement
669 // we assume no aliasing issues in the union
670 if (SvIsUV (sv) ? SvUVX (sv) <= 59000
671 : SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000)
510 { 672 {
673 // optimise the "small number case"
674 // code will likely be branchless and use only a single multiplication
675 // works for numbers up to 59074
676 I32 i = SvIVX (sv);
677 U32 u;
678 char digit, nz = 0;
679
511 need (enc, 64); 680 need (enc, 6);
512 enc->cur += 681
513 SvIsUV(sv) 682 *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
514 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 683 u = i < 0 ? -i : i;
515 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 684
685 // convert to 4.28 fixed-point representation
686 u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
687
688 // now output digit by digit, each time masking out the integer part
689 // and multiplying by 5 while moving the decimal point one to the right,
690 // resulting in a net multiplication by 10.
691 // we always write the digit to memory but conditionally increment
692 // the pointer, to enable the use of conditional move instructions.
693 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffffUL) * 5;
694 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffffUL) * 5;
695 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffffUL) * 5;
696 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffffUL) * 5;
697 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
516 } 698 }
517 else 699 else
518 { 700 {
519 // optimise the "small number case" 701 // large integer, use the (rather slow) snprintf way.
520 // code will likely be branchless and use only a single multiplication
521 I32 i = SvIV (sv);
522 U32 u;
523
524 need (enc, 6); 702 need (enc, IVUV_MAXCHARS);
525 703 enc->cur +=
526 *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0; 704 SvIsUV(sv)
527 u = i < 0 ? -i : i; 705 ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))
528 706 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
529 // convert to 4.28 fixed-point representation
530 u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
531
532 char digit, nz = 0;
533
534 digit = u >> 28; *enc->cur = digit + '0'; nz |= digit; enc->cur += nz ? 1 : 0; u = (u & 0xfffffff) * 5;
535 digit = u >> 27; *enc->cur = digit + '0'; nz |= digit; enc->cur += nz ? 1 : 0; u = (u & 0x7ffffff) * 5;
536 digit = u >> 26; *enc->cur = digit + '0'; nz |= digit; enc->cur += nz ? 1 : 0; u = (u & 0x3ffffff) * 5;
537 digit = u >> 25; *enc->cur = digit + '0'; nz |= digit; enc->cur += nz ? 1 : 0; u = (u & 0x1ffffff) * 5;
538 digit = u >> 24; *enc->cur = digit + '0'; nz |= digit; enc->cur += 1;
539 } 707 }
540 } 708 }
541 else if (SvROK (sv)) 709 else if (SvROK (sv))
542 encode_rv (enc, SvRV (sv)); 710 encode_rv (enc, SvRV (sv));
543 else if (!SvOK (sv)) 711 else if (!SvOK (sv))
… …
546 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 714 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
547 SvPV_nolen (sv), SvFLAGS (sv)); 715 SvPV_nolen (sv), SvFLAGS (sv));
548} 716}
549 717
550static SV * 718static SV *
551encode_json (SV *scalar, U32 flags) 719encode_json (SV *scalar, JSON *json)
552{ 720{
553 enc_t enc; 721 enc_t enc;
554 722
555 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 723 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
556 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 724 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
557 725
558 enc.flags = flags; 726 enc.json = *json;
559 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 727 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
560 enc.cur = SvPVX (enc.sv); 728 enc.cur = SvPVX (enc.sv);
561 enc.end = SvEND (enc.sv); 729 enc.end = SvEND (enc.sv);
562 enc.indent = 0; 730 enc.indent = 0;
563 enc.maxdepth = DEC_DEPTH (flags); 731 enc.maxdepth = DEC_DEPTH (enc.json.flags);
732 enc.limit = enc.json.flags & F_ASCII ? 0x000080UL
733 : enc.json.flags & F_LATIN1 ? 0x000100UL
734 : 0x110000UL;
564 735
565 SvPOK_only (enc.sv); 736 SvPOK_only (enc.sv);
566 encode_sv (&enc, scalar); 737 encode_sv (&enc, scalar);
567 738
568 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 739 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
569 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 740 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
570 741
571 if (!(flags & (F_ASCII | F_LATIN1 | F_UTF8))) 742 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
572 SvUTF8_on (enc.sv); 743 SvUTF8_on (enc.sv);
573 744
574 if (enc.flags & F_SHRINK) 745 if (enc.json.flags & F_SHRINK)
575 shrink (enc.sv); 746 shrink (enc.sv);
576 747
577 return enc.sv; 748 return enc.sv;
578} 749}
579 750
… …
584typedef struct 755typedef struct
585{ 756{
586 char *cur; // current parser pointer 757 char *cur; // current parser pointer
587 char *end; // end of input string 758 char *end; // end of input string
588 const char *err; // parse error, if != 0 759 const char *err; // parse error, if != 0
589 U32 flags; // F_* 760 JSON json;
590 U32 depth; // recursion depth 761 U32 depth; // recursion depth
591 U32 maxdepth; // recursion depth limit 762 U32 maxdepth; // recursion depth limit
592} dec_t; 763} dec_t;
593 764
594inline void 765INLINE void
766decode_comment (dec_t *dec)
767{
768 // only '#'-style comments allowed a.t.m.
769
770 while (*dec->cur && *dec->cur != 0x0a && *dec->cur != 0x0d)
771 ++dec->cur;
772}
773
774INLINE void
595decode_ws (dec_t *dec) 775decode_ws (dec_t *dec)
596{ 776{
597 for (;;) 777 for (;;)
598 { 778 {
599 char ch = *dec->cur; 779 char ch = *dec->cur;
600 780
601 if (ch > 0x20 781 if (ch > 0x20)
782 {
783 if (expect_false (ch == '#'))
784 {
785 if (dec->json.flags & F_RELAXED)
786 decode_comment (dec);
787 else
788 break;
789 }
790 else
791 break;
792 }
602 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) 793 else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)
603 break; 794 break; // parse error, but let higher level handle it, gives better error messages
604 795
605 ++dec->cur; 796 ++dec->cur;
606 } 797 }
607} 798}
608 799
… …
625decode_4hex (dec_t *dec) 816decode_4hex (dec_t *dec)
626{ 817{
627 signed char d1, d2, d3, d4; 818 signed char d1, d2, d3, d4;
628 unsigned char *cur = (unsigned char *)dec->cur; 819 unsigned char *cur = (unsigned char *)dec->cur;
629 820
630 d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("four hexadecimal digits expected"); 821 d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected");
631 d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("four hexadecimal digits expected"); 822 d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected");
632 d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("four hexadecimal digits expected"); 823 d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected");
633 d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("four hexadecimal digits expected"); 824 d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected");
634 825
635 dec->cur += 4; 826 dec->cur += 4;
636 827
637 return ((UV)d1) << 12 828 return ((UV)d1) << 12
638 | ((UV)d2) << 8 829 | ((UV)d2) << 8
… …
646static SV * 837static SV *
647decode_str (dec_t *dec) 838decode_str (dec_t *dec)
648{ 839{
649 SV *sv = 0; 840 SV *sv = 0;
650 int utf8 = 0; 841 int utf8 = 0;
842 char *dec_cur = dec->cur;
651 843
652 do 844 do
653 { 845 {
654 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES]; 846 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
655 char *cur = buf; 847 char *cur = buf;
656 848
657 do 849 do
658 { 850 {
659 unsigned char ch = *(unsigned char *)dec->cur++; 851 unsigned char ch = *(unsigned char *)dec_cur++;
660 852
661 if (expect_false (ch == '"')) 853 if (expect_false (ch == '"'))
662 { 854 {
663 --dec->cur; 855 --dec_cur;
664 break; 856 break;
665 } 857 }
666 else if (expect_false (ch == '\\')) 858 else if (expect_false (ch == '\\'))
667 { 859 {
668 switch (*dec->cur) 860 switch (*dec_cur)
669 { 861 {
670 case '\\': 862 case '\\':
671 case '/': 863 case '/':
672 case '"': *cur++ = *dec->cur++; break; 864 case '"': *cur++ = *dec_cur++; break;
673 865
674 case 'b': ++dec->cur; *cur++ = '\010'; break; 866 case 'b': ++dec_cur; *cur++ = '\010'; break;
675 case 't': ++dec->cur; *cur++ = '\011'; break; 867 case 't': ++dec_cur; *cur++ = '\011'; break;
676 case 'n': ++dec->cur; *cur++ = '\012'; break; 868 case 'n': ++dec_cur; *cur++ = '\012'; break;
677 case 'f': ++dec->cur; *cur++ = '\014'; break; 869 case 'f': ++dec_cur; *cur++ = '\014'; break;
678 case 'r': ++dec->cur; *cur++ = '\015'; break; 870 case 'r': ++dec_cur; *cur++ = '\015'; break;
679 871
680 case 'u': 872 case 'u':
681 { 873 {
682 UV lo, hi; 874 UV lo, hi;
683 ++dec->cur; 875 ++dec_cur;
684 876
877 dec->cur = dec_cur;
685 hi = decode_4hex (dec); 878 hi = decode_4hex (dec);
879 dec_cur = dec->cur;
686 if (hi == (UV)-1) 880 if (hi == (UV)-1)
687 goto fail; 881 goto fail;
688 882
689 // possibly a surrogate pair 883 // possibly a surrogate pair
690 if (hi >= 0xd800) 884 if (hi >= 0xd800)
691 if (hi < 0xdc00) 885 if (hi < 0xdc00)
692 { 886 {
693 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 887 if (dec_cur [0] != '\\' || dec_cur [1] != 'u')
694 ERR ("missing low surrogate character in surrogate pair"); 888 ERR ("missing low surrogate character in surrogate pair");
695 889
696 dec->cur += 2; 890 dec_cur += 2;
697 891
892 dec->cur = dec_cur;
698 lo = decode_4hex (dec); 893 lo = decode_4hex (dec);
894 dec_cur = dec->cur;
699 if (lo == (UV)-1) 895 if (lo == (UV)-1)
700 goto fail; 896 goto fail;
701 897
702 if (lo < 0xdc00 || lo >= 0xe000) 898 if (lo < 0xdc00 || lo >= 0xe000)
703 ERR ("surrogate pair expected"); 899 ERR ("surrogate pair expected");
… …
709 905
710 if (hi >= 0x80) 906 if (hi >= 0x80)
711 { 907 {
712 utf8 = 1; 908 utf8 = 1;
713 909
714 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0); 910 cur = encode_utf8 (cur, hi);
715 } 911 }
716 else 912 else
717 *cur++ = hi; 913 *cur++ = hi;
718 } 914 }
719 break; 915 break;
720 916
721 default: 917 default:
722 --dec->cur; 918 --dec_cur;
723 ERR ("illegal backslash escape sequence in string"); 919 ERR ("illegal backslash escape sequence in string");
724 } 920 }
725 } 921 }
726 else if (expect_true (ch >= 0x20 && ch <= 0x7f)) 922 else if (expect_true (ch >= 0x20 && ch < 0x80))
727 *cur++ = ch; 923 *cur++ = ch;
728 else if (ch >= 0x80) 924 else if (ch >= 0x80)
729 { 925 {
730 STRLEN clen; 926 STRLEN clen;
731 UV uch; 927 UV uch;
732 928
733 --dec->cur; 929 --dec_cur;
734 930
735 uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen); 931 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
736 if (clen == (STRLEN)-1) 932 if (clen == (STRLEN)-1)
737 ERR ("malformed UTF-8 character in JSON string"); 933 ERR ("malformed UTF-8 character in JSON string");
738 934
739 do 935 do
740 *cur++ = *dec->cur++; 936 *cur++ = *dec_cur++;
741 while (--clen); 937 while (--clen);
742 938
743 utf8 = 1; 939 utf8 = 1;
744 } 940 }
745 else 941 else
746 { 942 {
747 --dec->cur; 943 --dec_cur;
748 944
749 if (!ch) 945 if (!ch)
750 ERR ("unexpected end of string while parsing JSON string"); 946 ERR ("unexpected end of string while parsing JSON string");
751 else 947 else
752 ERR ("invalid character encountered while parsing JSON string"); 948 ERR ("invalid character encountered while parsing JSON string");
… …
765 } 961 }
766 else 962 else
767 sv = newSVpvn (buf, len); 963 sv = newSVpvn (buf, len);
768 } 964 }
769 } 965 }
770 while (*dec->cur != '"'); 966 while (*dec_cur != '"');
771 967
772 ++dec->cur; 968 ++dec_cur;
773 969
774 if (sv) 970 if (sv)
775 { 971 {
776 SvPOK_only (sv); 972 SvPOK_only (sv);
777 *SvEND (sv) = 0; 973 *SvEND (sv) = 0;
… …
780 SvUTF8_on (sv); 976 SvUTF8_on (sv);
781 } 977 }
782 else 978 else
783 sv = newSVpvn ("", 0); 979 sv = newSVpvn ("", 0);
784 980
981 dec->cur = dec_cur;
785 return sv; 982 return sv;
786 983
787fail: 984fail:
985 dec->cur = dec_cur;
788 return 0; 986 return 0;
789} 987}
790 988
791static SV * 989static SV *
792decode_num (dec_t *dec) 990decode_num (dec_t *dec)
… …
850 is_nv = 1; 1048 is_nv = 1;
851 } 1049 }
852 1050
853 if (!is_nv) 1051 if (!is_nv)
854 { 1052 {
1053 int len = dec->cur - start;
1054
855 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so 1055 // special case the rather common 1..5-digit-int case
856 if (*start == '-') 1056 if (*start == '-')
857 switch (dec->cur - start) 1057 switch (len)
858 { 1058 {
859 case 2: return newSViv (-( start [1] - '0' )); 1059 case 2: return newSViv (-( start [1] - '0' * 1));
860 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); 1060 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
861 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); 1061 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
862 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); 1062 case 5: return newSViv (-( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
1063 case 6: return newSViv (-(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
863 } 1064 }
864 else 1065 else
865 switch (dec->cur - start) 1066 switch (len)
866 { 1067 {
867 case 1: return newSViv ( start [0] - '0' ); 1068 case 1: return newSViv ( start [0] - '0' * 1);
868 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); 1069 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
869 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); 1070 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
870 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); 1071 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
1072 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
871 } 1073 }
872 1074
873 { 1075 {
874 UV uv; 1076 UV uv;
875 int numtype = grok_number (start, dec->cur - start, &uv); 1077 int numtype = grok_number (start, len, &uv);
876 if (numtype & IS_NUMBER_IN_UV) 1078 if (numtype & IS_NUMBER_IN_UV)
877 if (numtype & IS_NUMBER_NEG) 1079 if (numtype & IS_NUMBER_NEG)
878 { 1080 {
879 if (uv < (UV)IV_MIN) 1081 if (uv < (UV)IV_MIN)
880 return newSViv (-(IV)uv); 1082 return newSViv (-(IV)uv);
881 } 1083 }
882 else 1084 else
883 return newSVuv (uv); 1085 return newSVuv (uv);
884 } 1086 }
885 }
886 1087
1088 len -= *start == '-' ? 1 : 0;
1089
1090 // does not fit into IV or UV, try NV
1091 if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len)
1092 #if defined (LDBL_DIG)
1093 || (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len)
1094 #endif
1095 )
1096 // fits into NV without loss of precision
1097 return newSVnv (Atof (start));
1098
1099 // everything else fails, convert it to a string
1100 return newSVpvn (start, dec->cur - start);
1101 }
1102
1103 // loss of precision here
887 return newSVnv (Atof (start)); 1104 return newSVnv (Atof (start));
888 1105
889fail: 1106fail:
890 return 0; 1107 return 0;
891} 1108}
… …
921 1138
922 if (*dec->cur != ',') 1139 if (*dec->cur != ',')
923 ERR (", or ] expected while parsing array"); 1140 ERR (", or ] expected while parsing array");
924 1141
925 ++dec->cur; 1142 ++dec->cur;
1143
1144 decode_ws (dec);
1145
1146 if (*dec->cur == ']' && dec->json.flags & F_RELAXED)
1147 {
1148 ++dec->cur;
1149 break;
1150 }
926 } 1151 }
927 1152
928 DEC_DEC_DEPTH; 1153 DEC_DEC_DEPTH;
929 return newRV_noinc ((SV *)av); 1154 return newRV_noinc ((SV *)av);
930 1155
… …
935} 1160}
936 1161
937static SV * 1162static SV *
938decode_hv (dec_t *dec) 1163decode_hv (dec_t *dec)
939{ 1164{
1165 SV *sv;
940 HV *hv = newHV (); 1166 HV *hv = newHV ();
941 1167
942 DEC_INC_DEPTH; 1168 DEC_INC_DEPTH;
943 decode_ws (dec); 1169 decode_ws (dec);
944 1170
945 if (*dec->cur == '}') 1171 if (*dec->cur == '}')
946 ++dec->cur; 1172 ++dec->cur;
947 else 1173 else
948 for (;;) 1174 for (;;)
949 { 1175 {
1176 EXPECT_CH ('"');
1177
1178 // heuristic: assume that
1179 // a) decode_str + hv_store_ent are abysmally slow.
1180 // b) most hash keys are short, simple ascii text.
1181 // => try to "fast-match" such strings to avoid
1182 // the overhead of decode_str + hv_store_ent.
1183 {
950 SV *key, *value; 1184 SV *value;
1185 char *p = dec->cur;
1186 char *e = p + 24; // only try up to 24 bytes
951 1187
952 decode_ws (dec); EXPECT_CH ('"'); 1188 for (;;)
953
954 key = decode_str (dec);
955 if (!key)
956 goto fail;
957
958 decode_ws (dec); EXPECT_CH (':');
959
960 value = decode_sv (dec);
961 if (!value)
962 { 1189 {
1190 // the >= 0x80 is false on most architectures
1191 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1192 {
1193 // slow path, back up and use decode_str
1194 SV *key = decode_str (dec);
1195 if (!key)
1196 goto fail;
1197
1198 decode_ws (dec); EXPECT_CH (':');
1199
1200 decode_ws (dec);
1201 value = decode_sv (dec);
1202 if (!value)
1203 {
1204 SvREFCNT_dec (key);
1205 goto fail;
1206 }
1207
1208 hv_store_ent (hv, key, value, 0);
963 SvREFCNT_dec (key); 1209 SvREFCNT_dec (key);
1210
1211 break;
1212 }
1213 else if (*p == '"')
1214 {
1215 // fast path, got a simple key
1216 char *key = dec->cur;
1217 int len = p - key;
1218 dec->cur = p + 1;
1219
1220 decode_ws (dec); EXPECT_CH (':');
1221
1222 decode_ws (dec);
1223 value = decode_sv (dec);
1224 if (!value)
964 goto fail; 1225 goto fail;
1226
1227 hv_store (hv, key, len, value, 0);
1228
1229 break;
1230 }
1231
1232 ++p;
965 } 1233 }
966 1234 }
967 hv_store_ent (hv, key, value, 0);
968 SvREFCNT_dec (key);
969 1235
970 decode_ws (dec); 1236 decode_ws (dec);
971 1237
972 if (*dec->cur == '}') 1238 if (*dec->cur == '}')
973 { 1239 {
… …
977 1243
978 if (*dec->cur != ',') 1244 if (*dec->cur != ',')
979 ERR (", or } expected while parsing object/hash"); 1245 ERR (", or } expected while parsing object/hash");
980 1246
981 ++dec->cur; 1247 ++dec->cur;
1248
1249 decode_ws (dec);
1250
1251 if (*dec->cur == '}' && dec->json.flags & F_RELAXED)
1252 {
1253 ++dec->cur;
1254 break;
1255 }
982 } 1256 }
983 1257
984 DEC_DEC_DEPTH; 1258 DEC_DEC_DEPTH;
985 return newRV_noinc ((SV *)hv); 1259 sv = newRV_noinc ((SV *)hv);
1260
1261 // check filter callbacks
1262 if (dec->json.flags & F_HOOK)
1263 {
1264 if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
1265 {
1266 HE *cb, *he;
1267
1268 hv_iterinit (hv);
1269 he = hv_iternext (hv);
1270 hv_iterinit (hv);
1271
1272 // the next line creates a mortal sv each time its called.
1273 // might want to optimise this for common cases.
1274 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1275
1276 if (cb)
1277 {
1278 dSP;
1279 int count;
1280
1281 ENTER; SAVETMPS; PUSHMARK (SP);
1282 XPUSHs (HeVAL (he));
1283
1284 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1285
1286 if (count == 1)
1287 {
1288 sv = newSVsv (POPs);
1289 FREETMPS; LEAVE;
1290 return sv;
1291 }
1292
1293 FREETMPS; LEAVE;
1294 }
1295 }
1296
1297 if (dec->json.cb_object)
1298 {
1299 dSP;
1300 int count;
1301
1302 ENTER; SAVETMPS; PUSHMARK (SP);
1303 XPUSHs (sv_2mortal (sv));
1304
1305 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1306
1307 if (count == 1)
1308 {
1309 sv = newSVsv (POPs);
1310 FREETMPS; LEAVE;
1311 return sv;
1312 }
1313
1314 SvREFCNT_inc (sv);
1315 FREETMPS; LEAVE;
1316 }
1317 }
1318
1319 return sv;
986 1320
987fail: 1321fail:
988 SvREFCNT_dec (hv); 1322 SvREFCNT_dec (hv);
989 DEC_DEC_DEPTH; 1323 DEC_DEC_DEPTH;
990 return 0; 1324 return 0;
991} 1325}
992 1326
993static SV * 1327static SV *
994decode_sv (dec_t *dec) 1328decode_sv (dec_t *dec)
995{ 1329{
996 decode_ws (dec); 1330 // the beauty of JSON: you need exactly one character lookahead
1331 // to parse everything.
997 switch (*dec->cur) 1332 switch (*dec->cur)
998 { 1333 {
999 case '"': ++dec->cur; return decode_str (dec); 1334 case '"': ++dec->cur; return decode_str (dec);
1000 case '[': ++dec->cur; return decode_av (dec); 1335 case '[': ++dec->cur; return decode_av (dec);
1001 case '{': ++dec->cur; return decode_hv (dec); 1336 case '{': ++dec->cur; return decode_hv (dec);
1002 1337
1003 case '-': 1338 case '-':
1004 case '0': case '1': case '2': case '3': case '4': 1339 case '0': case '1': case '2': case '3': case '4':
1005 case '5': case '6': case '7': case '8': case '9': 1340 case '5': case '6': case '7': case '8': case '9':
1006 return decode_num (dec); 1341 return decode_num (dec);
1007 1342
1008 case 't': 1343 case 't':
1009 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1344 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1010 { 1345 {
1011 dec->cur += 4; 1346 dec->cur += 4;
1012 return newSViv (1); 1347#if JSON_SLOW
1348 json_true = get_sv ("JSON::XS::true", 1); SvREADONLY_on (json_true);
1349#endif
1350 return SvREFCNT_inc (json_true);
1013 } 1351 }
1014 else 1352 else
1015 ERR ("'true' expected"); 1353 ERR ("'true' expected");
1016 1354
1017 break; 1355 break;
1018 1356
1019 case 'f': 1357 case 'f':
1020 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1358 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1021 { 1359 {
1022 dec->cur += 5; 1360 dec->cur += 5;
1023 return newSViv (0); 1361#if JSON_SLOW
1362 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
1363#endif
1364 return SvREFCNT_inc (json_false);
1024 } 1365 }
1025 else 1366 else
1026 ERR ("'false' expected"); 1367 ERR ("'false' expected");
1027 1368
1028 break; 1369 break;
… …
1046fail: 1387fail:
1047 return 0; 1388 return 0;
1048} 1389}
1049 1390
1050static SV * 1391static SV *
1051decode_json (SV *string, U32 flags, UV *offset_return) 1392decode_json (SV *string, JSON *json, STRLEN *offset_return)
1052{ 1393{
1053 dec_t dec; 1394 dec_t dec;
1054 UV offset; 1395 STRLEN offset;
1055 SV *sv; 1396 SV *sv;
1056 1397
1057 SvGETMAGIC (string); 1398 SvGETMAGIC (string);
1058 SvUPGRADE (string, SVt_PV); 1399 SvUPGRADE (string, SVt_PV);
1059 1400
1401 if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags))
1402 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1403 (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags));
1404
1060 if (flags & F_UTF8) 1405 if (json->flags & F_UTF8)
1061 sv_utf8_downgrade (string, 0); 1406 sv_utf8_downgrade (string, 0);
1062 else 1407 else
1063 sv_utf8_upgrade (string); 1408 sv_utf8_upgrade (string);
1064 1409
1065 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1410 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
1066 1411
1067 dec.flags = flags; 1412 dec.json = *json;
1068 dec.cur = SvPVX (string); 1413 dec.cur = SvPVX (string);
1069 dec.end = SvEND (string); 1414 dec.end = SvEND (string);
1070 dec.err = 0; 1415 dec.err = 0;
1071 dec.depth = 0; 1416 dec.depth = 0;
1072 dec.maxdepth = DEC_DEPTH (dec.flags); 1417 dec.maxdepth = DEC_DEPTH (dec.json.flags);
1418
1419 if (dec.json.cb_object || dec.json.cb_sk_object)
1420 dec.json.flags |= F_HOOK;
1073 1421
1074 *dec.end = 0; // this should basically be a nop, too, but make sure it's there 1422 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1423
1424 decode_ws (&dec);
1075 sv = decode_sv (&dec); 1425 sv = decode_sv (&dec);
1076 1426
1077 if (!(offset_return || !sv)) 1427 if (!(offset_return || !sv))
1078 { 1428 {
1079 // check for trailing garbage 1429 // check for trailing garbage
… …
1087 } 1437 }
1088 } 1438 }
1089 1439
1090 if (offset_return || !sv) 1440 if (offset_return || !sv)
1091 { 1441 {
1092 offset = dec.flags & F_UTF8 1442 offset = dec.json.flags & F_UTF8
1093 ? dec.cur - SvPVX (string) 1443 ? dec.cur - SvPVX (string)
1094 : utf8_distance (dec.cur, SvPVX (string)); 1444 : utf8_distance (dec.cur, SvPVX (string));
1095 1445
1096 if (offset_return) 1446 if (offset_return)
1097 *offset_return = offset; 1447 *offset_return = offset;
… …
1116 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1466 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1117 } 1467 }
1118 1468
1119 sv = sv_2mortal (sv); 1469 sv = sv_2mortal (sv);
1120 1470
1121 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1471 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))
1122 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); 1472 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1123 1473
1124 return sv; 1474 return sv;
1475}
1476
1477/////////////////////////////////////////////////////////////////////////////
1478// incremental parser
1479
1480static void
1481incr_parse (JSON *self)
1482{
1483 const char *p = SvPVX (self->incr_text) + self->incr_pos;
1484
1485 for (;;)
1486 {
1487 //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
1488 switch (self->incr_mode)
1489 {
1490 // only used for intiial whitespace skipping
1491 case INCR_M_WS:
1492 for (;;)
1493 {
1494 if (*p > 0x20)
1495 {
1496 self->incr_mode = INCR_M_JSON;
1497 goto incr_m_json;
1498 }
1499 else if (!*p)
1500 goto interrupt;
1501
1502 ++p;
1503 }
1504
1505 // skip a single char inside a string (for \\-processing)
1506 case INCR_M_BS:
1507 if (!*p)
1508 goto interrupt;
1509
1510 ++p;
1511 self->incr_mode = INCR_M_STR;
1512 goto incr_m_str;
1513
1514 // inside a string
1515 case INCR_M_STR:
1516 incr_m_str:
1517 for (;;)
1518 {
1519 if (*p == '"')
1520 {
1521 ++p;
1522 self->incr_mode = INCR_M_JSON;
1523
1524 if (!self->incr_nest)
1525 goto interrupt;
1526
1527 goto incr_m_json;
1528 }
1529 else if (*p == '\\')
1530 {
1531 ++p; // "virtually" consumes character after \
1532
1533 if (!*p) // if at end of string we have to switch modes
1534 {
1535 self->incr_mode = INCR_M_BS;
1536 goto interrupt;
1537 }
1538 }
1539 else if (!*p)
1540 goto interrupt;
1541
1542 ++p;
1543 }
1544
1545 // after initial ws, outside string
1546 case INCR_M_JSON:
1547 incr_m_json:
1548 for (;;)
1549 {
1550 switch (*p++)
1551 {
1552 case 0:
1553 --p;
1554 goto interrupt;
1555
1556 case 0x09:
1557 case 0x0a:
1558 case 0x0d:
1559 case 0x20:
1560 if (!self->incr_nest)
1561 {
1562 --p; // do not eat the whitespace, let the next round do it
1563 goto interrupt;
1564 }
1565 break;
1566
1567 case '"':
1568 self->incr_mode = INCR_M_STR;
1569 goto incr_m_str;
1570
1571 case '[':
1572 case '{':
1573 ++self->incr_nest;
1574 break;
1575
1576 case ']':
1577 case '}':
1578 if (!--self->incr_nest)
1579 goto interrupt;
1580 }
1581 }
1582 }
1583
1584 modechange:
1585 ;
1586 }
1587
1588interrupt:
1589 self->incr_pos = p - SvPVX (self->incr_text);
1590 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D
1125} 1591}
1126 1592
1127///////////////////////////////////////////////////////////////////////////// 1593/////////////////////////////////////////////////////////////////////////////
1128// XS interface functions 1594// XS interface functions
1129 1595
1130MODULE = JSON::XS PACKAGE = JSON::XS 1596MODULE = JSON::XS PACKAGE = JSON::XS
1131 1597
1132BOOT: 1598BOOT:
1133{ 1599{
1134 int i; 1600 int i;
1135
1136 memset (decode_hexdigit, 0xff, 256);
1137 1601
1138 for (i = 0; i < 256; ++i) 1602 for (i = 0; i < 256; ++i)
1139 decode_hexdigit [i] = 1603 decode_hexdigit [i] =
1140 i >= '0' && i <= '9' ? i - '0' 1604 i >= '0' && i <= '9' ? i - '0'
1141 : i >= 'a' && i <= 'f' ? i - 'a' + 10 1605 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1142 : i >= 'A' && i <= 'F' ? i - 'A' + 10 1606 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1143 : -1; 1607 : -1;
1144 1608
1145 json_stash = gv_stashpv ("JSON::XS", 1); 1609 json_stash = gv_stashpv ("JSON::XS" , 1);
1610 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1611
1612 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true );
1613 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
1146} 1614}
1147 1615
1148PROTOTYPES: DISABLE 1616PROTOTYPES: DISABLE
1149 1617
1150SV *new (char *dummy) 1618void CLONE (...)
1151 CODE: 1619 CODE:
1152 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1620 json_stash = 0;
1621 json_boolean_stash = 0;
1622
1623void new (char *klass)
1624 PPCODE:
1625{
1626 SV *pv = NEWSV (0, sizeof (JSON));
1627 SvPOK_only (pv);
1628 Zero (SvPVX (pv), 1, JSON);
1629 ((JSON *)SvPVX (pv))->flags = F_DEFAULT;
1630 XPUSHs (sv_2mortal (sv_bless (
1631 newRV_noinc (pv),
1632 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1633 )));
1634}
1635
1636void ascii (JSON *self, int enable = 1)
1637 ALIAS:
1638 ascii = F_ASCII
1639 latin1 = F_LATIN1
1640 utf8 = F_UTF8
1641 indent = F_INDENT
1642 canonical = F_CANONICAL
1643 space_before = F_SPACE_BEFORE
1644 space_after = F_SPACE_AFTER
1645 pretty = F_PRETTY
1646 allow_nonref = F_ALLOW_NONREF
1647 shrink = F_SHRINK
1648 allow_blessed = F_ALLOW_BLESSED
1649 convert_blessed = F_CONV_BLESSED
1650 relaxed = F_RELAXED
1651 PPCODE:
1652{
1653 if (enable)
1654 self->flags |= ix;
1655 else
1656 self->flags &= ~ix;
1657
1658 XPUSHs (ST (0));
1659}
1660
1661void get_ascii (JSON *self)
1662 ALIAS:
1663 get_ascii = F_ASCII
1664 get_latin1 = F_LATIN1
1665 get_utf8 = F_UTF8
1666 get_indent = F_INDENT
1667 get_canonical = F_CANONICAL
1668 get_space_before = F_SPACE_BEFORE
1669 get_space_after = F_SPACE_AFTER
1670 get_allow_nonref = F_ALLOW_NONREF
1671 get_shrink = F_SHRINK
1672 get_allow_blessed = F_ALLOW_BLESSED
1673 get_convert_blessed = F_CONV_BLESSED
1674 get_relaxed = F_RELAXED
1675 PPCODE:
1676 XPUSHs (boolSV (self->flags & ix));
1677
1678void max_depth (JSON *self, UV max_depth = 0x80000000UL)
1679 PPCODE:
1680{
1681 UV log2 = 0;
1682
1683 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1684
1685 while ((1UL << log2) < max_depth)
1686 ++log2;
1687
1688 self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1689
1690 XPUSHs (ST (0));
1691}
1692
1693U32 get_max_depth (JSON *self)
1694 CODE:
1695 RETVAL = DEC_DEPTH (self->flags);
1153 OUTPUT: 1696 OUTPUT:
1154 RETVAL 1697 RETVAL
1155 1698
1156SV *ascii (SV *self, int enable = 1) 1699void max_size (JSON *self, UV max_size = 0)
1157 ALIAS: 1700 PPCODE:
1158 ascii = F_ASCII 1701{
1159 latin1 = F_LATIN1 1702 UV log2 = 0;
1160 utf8 = F_UTF8 1703
1161 indent = F_INDENT 1704 if (max_size > 0x80000000UL) max_size = 0x80000000UL;
1162 canonical = F_CANONICAL 1705 if (max_size == 1) max_size = 2;
1163 space_before = F_SPACE_BEFORE 1706
1164 space_after = F_SPACE_AFTER 1707 while ((1UL << log2) < max_size)
1165 pretty = F_PRETTY 1708 ++log2;
1166 allow_nonref = F_ALLOW_NONREF 1709
1167 shrink = F_SHRINK 1710 self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1711
1712 XPUSHs (ST (0));
1713}
1714
1715int get_max_size (JSON *self)
1168 CODE: 1716 CODE:
1169{ 1717 RETVAL = DEC_SIZE (self->flags);
1170 UV *uv = SvJSON (self);
1171 if (enable)
1172 *uv |= ix;
1173 else
1174 *uv &= ~ix;
1175
1176 RETVAL = newSVsv (self);
1177}
1178 OUTPUT: 1718 OUTPUT:
1179 RETVAL 1719 RETVAL
1180 1720
1181SV *max_depth (SV *self, UV max_depth = 0x80000000UL) 1721void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1722 PPCODE:
1723{
1724 SvREFCNT_dec (self->cb_object);
1725 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
1726
1727 XPUSHs (ST (0));
1728}
1729
1730void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
1731 PPCODE:
1732{
1733 if (!self->cb_sk_object)
1734 self->cb_sk_object = newHV ();
1735
1736 if (SvOK (cb))
1737 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1738 else
1739 {
1740 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1741
1742 if (!HvKEYS (self->cb_sk_object))
1743 {
1744 SvREFCNT_dec (self->cb_sk_object);
1745 self->cb_sk_object = 0;
1746 }
1747 }
1748
1749 XPUSHs (ST (0));
1750}
1751
1752void encode (JSON *self, SV *scalar)
1753 PPCODE:
1754 XPUSHs (encode_json (scalar, self));
1755
1756void decode (JSON *self, SV *jsonstr)
1757 PPCODE:
1758 XPUSHs (decode_json (jsonstr, self, 0));
1759
1760void decode_prefix (JSON *self, SV *jsonstr)
1761 PPCODE:
1762{
1763 STRLEN offset;
1764 EXTEND (SP, 2);
1765 PUSHs (decode_json (jsonstr, self, &offset));
1766 PUSHs (sv_2mortal (newSVuv (offset)));
1767}
1768
1769void incr_parse (JSON *self, SV *jsonstr = 0)
1770 PPCODE:
1771{
1772 if (!self->incr_text)
1773 self->incr_text = newSVpvn ("", 0);
1774
1775 // append data, if any
1776 if (jsonstr)
1777 {
1778 if (SvUTF8 (jsonstr) && !SvUTF8 (self->incr_text))
1779 {
1780 /* utf-8-ness differs, need to upgrade */
1781 sv_utf8_upgrade (self->incr_text);
1782
1783 if (self->incr_pos)
1784 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
1785 - (U8 *)SvPVX (self->incr_text);
1786 }
1787
1788 {
1789 STRLEN len;
1790 const char *str = SvPV (jsonstr, len);
1791 SvGROW (self->incr_text, SvCUR (self->incr_text) + len + 1);
1792 Move (str, SvEND (self->incr_text), len, char);
1793 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
1794 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
1795 }
1796 }
1797
1798 if (GIMME_V != G_VOID)
1799 do
1800 {
1801 STRLEN offset;
1802
1803 incr_parse (self);
1804
1805 if (!INCR_DONE (self))
1806 break;
1807
1808 XPUSHs (decode_json (self->incr_text, self, &offset));
1809
1810 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + offset);
1811 self->incr_pos -= offset;
1812 self->incr_nest = 0;
1813 self->incr_mode = 0;
1814 }
1815 while (GIMME_V == G_ARRAY);
1816}
1817
1818SV *incr_text (JSON *self)
1819 ATTRS: lvalue
1182 CODE: 1820 CODE:
1183{ 1821{
1184 UV *uv = SvJSON (self); 1822 if (self->incr_pos)
1185 UV log2 = 0; 1823 croak ("incr_text can only be called after a successful incr_parse call in scalar context");
1186 1824
1187 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL; 1825 RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef;
1188
1189 while ((1UL << log2) < max_depth)
1190 ++log2;
1191
1192 *uv = *uv & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1193
1194 RETVAL = newSVsv (self);
1195} 1826}
1196 OUTPUT: 1827 OUTPUT:
1197 RETVAL 1828 RETVAL
1198 1829
1830void incr_skip (JSON *self)
1831 CODE:
1832{
1833 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + self->incr_pos);
1834 self->incr_pos = 0;
1835 self->incr_nest = 0;
1836 self->incr_mode = 0;
1837}
1838
1839void DESTROY (JSON *self)
1840 CODE:
1841 SvREFCNT_dec (self->cb_sk_object);
1842 SvREFCNT_dec (self->cb_object);
1843 SvREFCNT_dec (self->incr_text);
1844
1845PROTOTYPES: ENABLE
1846
1199void encode (SV *self, SV *scalar) 1847void encode_json (SV *scalar)
1848 ALIAS:
1849 to_json_ = 0
1850 encode_json = F_UTF8
1200 PPCODE: 1851 PPCODE:
1852{
1853 JSON json = { F_DEFAULT | ix };
1201 XPUSHs (encode_json (scalar, *SvJSON (self))); 1854 XPUSHs (encode_json (scalar, &json));
1855}
1202 1856
1203void decode (SV *self, SV *jsonstr) 1857void decode_json (SV *jsonstr)
1858 ALIAS:
1859 from_json_ = 0
1860 decode_json = F_UTF8
1204 PPCODE: 1861 PPCODE:
1862{
1863 JSON json = { F_DEFAULT | ix };
1205 XPUSHs (decode_json (jsonstr, *SvJSON (self), 0)); 1864 XPUSHs (decode_json (jsonstr, &json, 0));
1206
1207void decode_prefix (SV *self, SV *jsonstr)
1208 PPCODE:
1209{
1210 UV offset;
1211 EXTEND (SP, 2);
1212 PUSHs (decode_json (jsonstr, *SvJSON (self), &offset));
1213 PUSHs (sv_2mortal (newSVuv (offset)));
1214} 1865}
1215 1866
1216PROTOTYPES: ENABLE
1217 1867
1218void to_json (SV *scalar)
1219 ALIAS:
1220 objToJson = 0
1221 PPCODE:
1222 XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8));
1223
1224void from_json (SV *jsonstr)
1225 ALIAS:
1226 jsonToObj = 0
1227 PPCODE:
1228 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8, 0));
1229

Diff Legend

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