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 1.61 by root, Sun Aug 26 21:56:47 2007 UTC vs.
Revision 1.104 by root, Tue Jan 19 00:31:13 2010 UTC

… …
4 4
5#include <assert.h> 5#include <assert.h>
6#include <string.h> 6#include <string.h>
7#include <stdlib.h> 7#include <stdlib.h>
8#include <stdio.h> 8#include <stdio.h>
9#include <limits.h>
9#include <float.h> 10#include <float.h>
10 11
11#if defined(__BORLANDC__) || defined(_MSC_VER) 12#if defined(__BORLANDC__) || defined(_MSC_VER)
12# define snprintf _snprintf // C compilers have this in stdio.h 13# define snprintf _snprintf // C compilers have this in stdio.h
13#endif 14#endif
14 15
15// some old perls do not have this, try to make it work, no 16// some old perls do not have this, try to make it work, no
16// guarentees, though. if it breaks, you get to keep the pieces. 17// guarantees, though. if it breaks, you get to keep the pieces.
17#ifndef UTF8_MAXBYTES 18#ifndef UTF8_MAXBYTES
18# define UTF8_MAXBYTES 13 19# define UTF8_MAXBYTES 13
19#endif 20#endif
21
22// three extra for rounding, sign, and end of string
23#define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 3)
20 24
21#define F_ASCII 0x00000001UL 25#define F_ASCII 0x00000001UL
22#define F_LATIN1 0x00000002UL 26#define F_LATIN1 0x00000002UL
23#define F_UTF8 0x00000004UL 27#define F_UTF8 0x00000004UL
24#define F_INDENT 0x00000008UL 28#define F_INDENT 0x00000008UL
… …
27#define F_SPACE_AFTER 0x00000040UL 31#define F_SPACE_AFTER 0x00000040UL
28#define F_ALLOW_NONREF 0x00000100UL 32#define F_ALLOW_NONREF 0x00000100UL
29#define F_SHRINK 0x00000200UL 33#define F_SHRINK 0x00000200UL
30#define F_ALLOW_BLESSED 0x00000400UL 34#define F_ALLOW_BLESSED 0x00000400UL
31#define F_CONV_BLESSED 0x00000800UL 35#define F_CONV_BLESSED 0x00000800UL
32#define F_MAXDEPTH 0xf8000000UL
33#define S_MAXDEPTH 27
34#define F_MAXSIZE 0x01f00000UL 36#define F_RELAXED 0x00001000UL
35#define S_MAXSIZE 20 37#define F_ALLOW_UNKNOWN 0x00002000UL
36#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing 38#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
37 39
38#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
39#define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE ))
40
41#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 40#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
42#define F_DEFAULT (9UL << S_MAXDEPTH)
43 41
44#define INIT_SIZE 32 // initial scalar size to be allocated 42#define INIT_SIZE 32 // initial scalar size to be allocated
45#define INDENT_STEP 3 // spaces per indentation level 43#define INDENT_STEP 3 // spaces per indentation level
46 44
47#define SHORT_STRING_LEN 16384 // special-case strings of up to this size 45#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
48 46
49#define SB do { 47#define SB do {
50#define SE } while (0) 48#define SE } while (0)
51 49
52#if __GNUC__ >= 3 50#if __GNUC__ >= 3
53# define expect(expr,value) __builtin_expect ((expr),(value)) 51# define expect(expr,value) __builtin_expect ((expr), (value))
54# define inline inline 52# define INLINE static inline
55#else 53#else
56# define expect(expr,value) (expr) 54# define expect(expr,value) (expr)
57# define inline static 55# define INLINE static
58#endif 56#endif
59 57
60#define expect_false(expr) expect ((expr) != 0, 0) 58#define expect_false(expr) expect ((expr) != 0, 0)
61#define expect_true(expr) expect ((expr) != 0, 1) 59#define expect_true(expr) expect ((expr) != 0, 1)
60
61#define IN_RANGE_INC(type,val,beg,end) \
62 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \
63 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
64
65#define ERR_NESTING_EXCEEDED "json text or perl structure exceeds maximum nesting level (max_depth set too low?)"
62 66
63#ifdef USE_ITHREADS 67#ifdef USE_ITHREADS
64# define JSON_SLOW 1 68# define JSON_SLOW 1
65# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1)) 69# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1))
66#else 70#else
… …
69#endif 73#endif
70 74
71static HV *json_stash, *json_boolean_stash; // JSON::XS:: 75static HV *json_stash, *json_boolean_stash; // JSON::XS::
72static SV *json_true, *json_false; 76static SV *json_true, *json_false;
73 77
78enum {
79 INCR_M_WS = 0, // initial whitespace skipping, must be 0
80 INCR_M_STR, // inside string
81 INCR_M_BS, // inside backslash
82 INCR_M_C0, // inside comment in initial whitespace sequence
83 INCR_M_C1, // inside comment in other places
84 INCR_M_JSON // outside anything, count nesting
85};
86
87#define INCR_DONE(json) ((json)->incr_nest <= 0 && (json)->incr_mode == INCR_M_JSON)
88
74typedef struct { 89typedef struct {
75 U32 flags; 90 U32 flags;
91 U32 max_depth;
92 STRLEN max_size;
93
76 SV *cb_object; 94 SV *cb_object;
77 HV *cb_sk_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 unsigned char incr_mode;
78} JSON; 102} JSON;
103
104INLINE void
105json_init (JSON *json)
106{
107 Zero (json, 1, JSON);
108 json->max_depth = 512;
109}
79 110
80///////////////////////////////////////////////////////////////////////////// 111/////////////////////////////////////////////////////////////////////////////
81// utility functions 112// utility functions
82 113
83inline void 114INLINE SV *
115get_bool (const char *name)
116{
117 SV *sv = get_sv (name, 1);
118
119 SvREADONLY_on (sv);
120 SvREADONLY_on (SvRV (sv));
121
122 return sv;
123}
124
125INLINE void
84shrink (SV *sv) 126shrink (SV *sv)
85{ 127{
86 sv_utf8_downgrade (sv, 1); 128 sv_utf8_downgrade (sv, 1);
129
87 if (SvLEN (sv) > SvCUR (sv) + 1) 130 if (SvLEN (sv) > SvCUR (sv) + 1)
88 { 131 {
89#ifdef SvPV_shrink_to_cur 132#ifdef SvPV_shrink_to_cur
90 SvPV_shrink_to_cur (sv); 133 SvPV_shrink_to_cur (sv);
91#elif defined (SvPV_renew) 134#elif defined (SvPV_renew)
… …
97// decode an utf-8 character and return it, or (UV)-1 in 140// decode an utf-8 character and return it, or (UV)-1 in
98// case of an error. 141// case of an error.
99// we special-case "safe" characters from U+80 .. U+7FF, 142// we special-case "safe" characters from U+80 .. U+7FF,
100// but use the very good perl function to parse anything else. 143// but use the very good perl function to parse anything else.
101// note that we never call this function for a ascii codepoints 144// note that we never call this function for a ascii codepoints
102inline UV 145INLINE UV
103decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) 146decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
104{ 147{
105 if (expect_false (s[0] > 0xdf || s[0] < 0xc2)) 148 if (expect_true (len >= 2
106 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); 149 && IN_RANGE_INC (char, s[0], 0xc2, 0xdf)
107 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) 150 && IN_RANGE_INC (char, s[1], 0x80, 0xbf)))
108 { 151 {
109 *clen = 2; 152 *clen = 2;
110 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 153 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
111 } 154 }
112 else 155 else
113 { 156 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
114 *clen = (STRLEN)-1; 157}
115 return (UV)-1; 158
159// likewise for encoding, also never called for ascii codepoints
160// this function takes advantage of this fact, although current gccs
161// seem to optimise the check for >= 0x80 away anyways
162INLINE unsigned char *
163encode_utf8 (unsigned char *s, UV ch)
164{
165 if (expect_false (ch < 0x000080))
166 *s++ = ch;
167 else if (expect_true (ch < 0x000800))
168 *s++ = 0xc0 | ( ch >> 6),
169 *s++ = 0x80 | ( ch & 0x3f);
170 else if ( ch < 0x010000)
171 *s++ = 0xe0 | ( ch >> 12),
172 *s++ = 0x80 | ((ch >> 6) & 0x3f),
173 *s++ = 0x80 | ( ch & 0x3f);
174 else if ( ch < 0x110000)
175 *s++ = 0xf0 | ( ch >> 18),
176 *s++ = 0x80 | ((ch >> 12) & 0x3f),
177 *s++ = 0x80 | ((ch >> 6) & 0x3f),
178 *s++ = 0x80 | ( ch & 0x3f);
179
180 return s;
181}
182
183// convert offset pointer to character index, sv must be string
184static STRLEN
185ptr_to_index (SV *sv, char *offset)
186{
187 return SvUTF8 (sv)
188 ? utf8_distance (offset, SvPVX (sv))
189 : offset - SvPVX (sv);
190}
191
192/////////////////////////////////////////////////////////////////////////////
193// fp hell
194
195// scan a group of digits, and a trailing exponent
196static void
197json_atof_scan1 (const char *s, NV *accum, int *expo, int postdp)
198{
199 UV uaccum = 0;
200 int eaccum = 0;
201
202 for (;;)
116 } 203 {
117} 204 U8 dig = (U8)*s - '0';
118 205
206 if (expect_false (dig >= 10))
207 {
208 if (dig == (U8)((U8)'.' - (U8)'0'))
209 {
210 ++s;
211 json_atof_scan1 (s, accum, expo, 1);
212 }
213 else if ((dig | ' ') == 'e' - '0')
214 {
215 int exp2 = 0;
216 int neg = 0;
217
218 ++s;
219
220 if (*s == '-')
221 {
222 ++s;
223 neg = 1;
224 }
225 else if (*s == '+')
226 ++s;
227
228 while ((dig = (U8)*s - '0') < 10)
229 exp2 = exp2 * 10 + *s++ - '0';
230
231 *expo += neg ? -exp2 : exp2;
232 }
233
234 break;
235 }
236
237 ++s;
238
239 uaccum = uaccum * 10 + dig;
240 ++eaccum;
241
242 // if we have too many digits, then recurse for more
243 // we actually do this for rather few digits
244 if (uaccum >= (UV_MAX - 9) / 10)
245 {
246 if (postdp) *expo -= eaccum;
247 json_atof_scan1 (s, accum, expo, postdp);
248 if (postdp) *expo += eaccum;
249
250 break;
251 }
252 }
253
254 if (postdp) *expo -= eaccum;
255 *accum += uaccum * pow (10., *expo);
256 *expo += eaccum;
257}
258
259static NV
260json_atof (const char *s)
261{
262 NV accum = 0.;
263 int expo = 0;
264 int neg = 0;
265
266 if (*s == '-')
267 {
268 ++s;
269 neg = 1;
270 }
271
272 json_atof_scan1 (s, &accum, &expo, 0);
273
274 return neg ? -accum : accum;
275}
119///////////////////////////////////////////////////////////////////////////// 276/////////////////////////////////////////////////////////////////////////////
120// encoder 277// encoder
121 278
122// structure used for encoding JSON 279// structure used for encoding JSON
123typedef struct 280typedef struct
… …
125 char *cur; // SvPVX (sv) + current output position 282 char *cur; // SvPVX (sv) + current output position
126 char *end; // SvEND (sv) 283 char *end; // SvEND (sv)
127 SV *sv; // result scalar 284 SV *sv; // result scalar
128 JSON json; 285 JSON json;
129 U32 indent; // indentation level 286 U32 indent; // indentation level
130 U32 maxdepth; // max. indentation/recursion level 287 UV limit; // escape character values >= this value when encoding
131} enc_t; 288} enc_t;
132 289
133inline void 290INLINE void
134need (enc_t *enc, STRLEN len) 291need (enc_t *enc, STRLEN len)
135{ 292{
136 if (expect_false (enc->cur + len >= enc->end)) 293 if (expect_false (enc->cur + len >= enc->end))
137 { 294 {
138 STRLEN cur = enc->cur - SvPVX (enc->sv); 295 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
139 SvGROW (enc->sv, cur + len + 1); 296 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
140 enc->cur = SvPVX (enc->sv) + cur; 297 enc->cur = SvPVX (enc->sv) + cur;
141 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 298 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
142 } 299 }
143} 300}
144 301
145inline void 302INLINE void
146encode_ch (enc_t *enc, char ch) 303encode_ch (enc_t *enc, char ch)
147{ 304{
148 need (enc, 1); 305 need (enc, 1);
149 *enc->cur++ = ch; 306 *enc->cur++ = ch;
150} 307}
… …
204 { 361 {
205 uch = ch; 362 uch = ch;
206 clen = 1; 363 clen = 1;
207 } 364 }
208 365
209 if (uch > 0x10FFFFUL) 366 if (uch < 0x80/*0x20*/ || uch >= enc->limit)
210 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
211
212 if (uch < 0x80 || enc->json.flags & F_ASCII || (enc->json.flags & F_LATIN1 && uch > 0xFF))
213 { 367 {
214 if (uch > 0xFFFFUL) 368 if (uch >= 0x10000UL)
215 { 369 {
370 if (uch >= 0x110000UL)
371 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
372
216 need (enc, len += 11); 373 need (enc, len += 11);
217 sprintf (enc->cur, "\\u%04x\\u%04x", 374 sprintf (enc->cur, "\\u%04x\\u%04x",
218 (int)((uch - 0x10000) / 0x400 + 0xD800), 375 (int)((uch - 0x10000) / 0x400 + 0xD800),
219 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 376 (int)((uch - 0x10000) % 0x400 + 0xDC00));
220 enc->cur += 12; 377 enc->cur += 12;
221 } 378 }
222 else 379 else
223 { 380 {
224 static char hexdigit [16] = "0123456789abcdef";
225 need (enc, len += 5); 381 need (enc, len += 5);
226 *enc->cur++ = '\\'; 382 *enc->cur++ = '\\';
227 *enc->cur++ = 'u'; 383 *enc->cur++ = 'u';
228 *enc->cur++ = hexdigit [ uch >> 12 ]; 384 *enc->cur++ = PL_hexdigit [ uch >> 12 ];
229 *enc->cur++ = hexdigit [(uch >> 8) & 15]; 385 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
230 *enc->cur++ = hexdigit [(uch >> 4) & 15]; 386 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
231 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 387 *enc->cur++ = PL_hexdigit [(uch >> 0) & 15];
232 } 388 }
233 389
234 str += clen; 390 str += clen;
235 } 391 }
236 else if (enc->json.flags & F_LATIN1) 392 else if (enc->json.flags & F_LATIN1)
… …
248 while (--clen); 404 while (--clen);
249 } 405 }
250 else 406 else
251 { 407 {
252 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed 408 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
253 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 409 enc->cur = encode_utf8 (enc->cur, uch);
254 ++str; 410 ++str;
255 } 411 }
256 } 412 }
257 } 413 }
258 } 414 }
259 415
260 --len; 416 --len;
261 } 417 }
262} 418}
263 419
264inline void 420INLINE void
265encode_indent (enc_t *enc) 421encode_indent (enc_t *enc)
266{ 422{
267 if (enc->json.flags & F_INDENT) 423 if (enc->json.flags & F_INDENT)
268 { 424 {
269 int spaces = enc->indent * INDENT_STEP; 425 int spaces = enc->indent * INDENT_STEP;
… …
272 memset (enc->cur, ' ', spaces); 428 memset (enc->cur, ' ', spaces);
273 enc->cur += spaces; 429 enc->cur += spaces;
274 } 430 }
275} 431}
276 432
277inline void 433INLINE void
278encode_space (enc_t *enc) 434encode_space (enc_t *enc)
279{ 435{
280 need (enc, 1); 436 need (enc, 1);
281 encode_ch (enc, ' '); 437 encode_ch (enc, ' ');
282} 438}
283 439
284inline void 440INLINE void
285encode_nl (enc_t *enc) 441encode_nl (enc_t *enc)
286{ 442{
287 if (enc->json.flags & F_INDENT) 443 if (enc->json.flags & F_INDENT)
288 { 444 {
289 need (enc, 1); 445 need (enc, 1);
290 encode_ch (enc, '\n'); 446 encode_ch (enc, '\n');
291 } 447 }
292} 448}
293 449
294inline void 450INLINE void
295encode_comma (enc_t *enc) 451encode_comma (enc_t *enc)
296{ 452{
297 encode_ch (enc, ','); 453 encode_ch (enc, ',');
298 454
299 if (enc->json.flags & F_INDENT) 455 if (enc->json.flags & F_INDENT)
… …
307static void 463static void
308encode_av (enc_t *enc, AV *av) 464encode_av (enc_t *enc, AV *av)
309{ 465{
310 int i, len = av_len (av); 466 int i, len = av_len (av);
311 467
312 if (enc->indent >= enc->maxdepth) 468 if (enc->indent >= enc->json.max_depth)
313 croak ("data structure too deep (hit recursion limit)"); 469 croak (ERR_NESTING_EXCEEDED);
314 470
315 encode_ch (enc, '['); encode_nl (enc); 471 encode_ch (enc, '[');
316 ++enc->indent; 472
473 if (len >= 0)
474 {
475 encode_nl (enc); ++enc->indent;
317 476
318 for (i = 0; i <= len; ++i) 477 for (i = 0; i <= len; ++i)
319 { 478 {
320 SV **svp = av_fetch (av, i, 0); 479 SV **svp = av_fetch (av, i, 0);
321 480
322 encode_indent (enc); 481 encode_indent (enc);
323 482
324 if (svp) 483 if (svp)
325 encode_sv (enc, *svp); 484 encode_sv (enc, *svp);
326 else 485 else
327 encode_str (enc, "null", 4, 0); 486 encode_str (enc, "null", 4, 0);
328 487
329 if (i < len) 488 if (i < len)
330 encode_comma (enc); 489 encode_comma (enc);
331 } 490 }
332 491
492 encode_nl (enc); --enc->indent; encode_indent (enc);
493 }
494
333 encode_nl (enc); 495 encode_ch (enc, ']');
334
335 --enc->indent;
336 encode_indent (enc); encode_ch (enc, ']');
337} 496}
338 497
339static void 498static void
340encode_hk (enc_t *enc, HE *he) 499encode_hk (enc_t *enc, HE *he)
341{ 500{
… …
389 548
390static void 549static void
391encode_hv (enc_t *enc, HV *hv) 550encode_hv (enc_t *enc, HV *hv)
392{ 551{
393 HE *he; 552 HE *he;
394 int count;
395 553
396 if (enc->indent >= enc->maxdepth) 554 if (enc->indent >= enc->json.max_depth)
397 croak ("data structure too deep (hit recursion limit)"); 555 croak (ERR_NESTING_EXCEEDED);
398 556
399 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 557 encode_ch (enc, '{');
400 558
401 // for canonical output we have to sort by keys first 559 // for canonical output we have to sort by keys first
402 // actually, this is mostly due to the stupid so-called 560 // actually, this is mostly due to the stupid so-called
403 // security workaround added somewhere in 5.8.x. 561 // security workaround added somewhere in 5.8.x
404 // that randomises hash orderings 562 // that randomises hash orderings
405 if (enc->json.flags & F_CANONICAL) 563 if (enc->json.flags & F_CANONICAL && !SvRMAGICAL (hv))
406 { 564 {
407 int count = hv_iterinit (hv); 565 int count = hv_iterinit (hv);
408 566
409 if (SvMAGICAL (hv)) 567 if (SvMAGICAL (hv))
410 { 568 {
… …
457 615
458 FREETMPS; 616 FREETMPS;
459 LEAVE; 617 LEAVE;
460 } 618 }
461 619
620 encode_nl (enc); ++enc->indent;
621
462 while (count--) 622 while (count--)
463 { 623 {
464 encode_indent (enc); 624 encode_indent (enc);
465 he = hes [count]; 625 he = hes [count];
466 encode_hk (enc, he); 626 encode_hk (enc, he);
467 encode_sv (enc, SvMAGICAL (hv) ? hv_iterval (hv, he) : HeVAL (he)); 627 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
468 628
469 if (count) 629 if (count)
470 encode_comma (enc); 630 encode_comma (enc);
471 } 631 }
632
633 encode_nl (enc); --enc->indent; encode_indent (enc);
472 } 634 }
473 } 635 }
474 else 636 else
475 { 637 {
476 if (hv_iterinit (hv) || SvMAGICAL (hv)) 638 if (hv_iterinit (hv) || SvMAGICAL (hv))
477 if ((he = hv_iternext (hv))) 639 if ((he = hv_iternext (hv)))
640 {
641 encode_nl (enc); ++enc->indent;
642
478 for (;;) 643 for (;;)
479 { 644 {
480 encode_indent (enc); 645 encode_indent (enc);
481 encode_hk (enc, he); 646 encode_hk (enc, he);
482 encode_sv (enc, SvMAGICAL (hv) ? hv_iterval (hv, he) : HeVAL (he)); 647 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
483 648
484 if (!(he = hv_iternext (hv))) 649 if (!(he = hv_iternext (hv)))
485 break; 650 break;
486 651
487 encode_comma (enc); 652 encode_comma (enc);
488 } 653 }
489 }
490 654
655 encode_nl (enc); --enc->indent; encode_indent (enc);
656 }
657 }
658
491 encode_nl (enc); 659 encode_ch (enc, '}');
492
493 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
494} 660}
495 661
496// encode objects, arrays and special \0=false and \1=true values. 662// encode objects, arrays and special \0=false and \1=true values.
497static void 663static void
498encode_rv (enc_t *enc, SV *sv) 664encode_rv (enc_t *enc, SV *sv)
… …
575 741
576 if (len == 1 && *pv == '1') 742 if (len == 1 && *pv == '1')
577 encode_str (enc, "true", 4, 0); 743 encode_str (enc, "true", 4, 0);
578 else if (len == 1 && *pv == '0') 744 else if (len == 1 && *pv == '0')
579 encode_str (enc, "false", 5, 0); 745 encode_str (enc, "false", 5, 0);
746 else if (enc->json.flags & F_ALLOW_UNKNOWN)
747 encode_str (enc, "null", 4, 0);
580 else 748 else
581 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", 749 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
582 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 750 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
583 } 751 }
752 else if (enc->json.flags & F_ALLOW_UNKNOWN)
753 encode_str (enc, "null", 4, 0);
584 else 754 else
585 croak ("encountered %s, but JSON can only represent references to arrays or hashes", 755 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
586 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 756 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
587} 757}
588 758
… …
606 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 776 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
607 enc->cur += strlen (enc->cur); 777 enc->cur += strlen (enc->cur);
608 } 778 }
609 else if (SvIOKp (sv)) 779 else if (SvIOKp (sv))
610 { 780 {
611 // we assume we can always read an IV as a UV 781 // we assume we can always read an IV as a UV and vice versa
612 if (SvUV (sv) & ~(UV)0x7fff) 782 // we assume two's complement
613 { 783 // we assume no aliasing issues in the union
614 // large integer, use the (rather slow) snprintf way. 784 if (SvIsUV (sv) ? SvUVX (sv) <= 59000
615 need (enc, sizeof (UV) * 3); 785 : SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000)
616 enc->cur +=
617 SvIsUV(sv)
618 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
619 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
620 }
621 else
622 { 786 {
623 // optimise the "small number case" 787 // optimise the "small number case"
624 // code will likely be branchless and use only a single multiplication 788 // code will likely be branchless and use only a single multiplication
789 // works for numbers up to 59074
625 I32 i = SvIV (sv); 790 I32 i = SvIVX (sv);
626 U32 u; 791 U32 u;
627 char digit, nz = 0; 792 char digit, nz = 0;
628 793
629 need (enc, 6); 794 need (enc, 6);
630 795
… …
636 801
637 // now output digit by digit, each time masking out the integer part 802 // now output digit by digit, each time masking out the integer part
638 // and multiplying by 5 while moving the decimal point one to the right, 803 // and multiplying by 5 while moving the decimal point one to the right,
639 // resulting in a net multiplication by 10. 804 // resulting in a net multiplication by 10.
640 // we always write the digit to memory but conditionally increment 805 // we always write the digit to memory but conditionally increment
641 // the pointer, to ease the usage of conditional move instructions. 806 // the pointer, to enable the use of conditional move instructions.
642 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5; 807 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffffUL) * 5;
643 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5; 808 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffffUL) * 5;
644 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5; 809 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffffUL) * 5;
645 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5; 810 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffffUL) * 5;
646 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0' 811 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
647 } 812 }
813 else
814 {
815 // large integer, use the (rather slow) snprintf way.
816 need (enc, IVUV_MAXCHARS);
817 enc->cur +=
818 SvIsUV(sv)
819 ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))
820 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
821 }
648 } 822 }
649 else if (SvROK (sv)) 823 else if (SvROK (sv))
650 encode_rv (enc, SvRV (sv)); 824 encode_rv (enc, SvRV (sv));
651 else if (!SvOK (sv)) 825 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN)
652 encode_str (enc, "null", 4, 0); 826 encode_str (enc, "null", 4, 0);
653 else 827 else
654 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 828 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
655 SvPV_nolen (sv), SvFLAGS (sv)); 829 SvPV_nolen (sv), SvFLAGS (sv));
656} 830}
… …
666 enc.json = *json; 840 enc.json = *json;
667 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 841 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
668 enc.cur = SvPVX (enc.sv); 842 enc.cur = SvPVX (enc.sv);
669 enc.end = SvEND (enc.sv); 843 enc.end = SvEND (enc.sv);
670 enc.indent = 0; 844 enc.indent = 0;
671 enc.maxdepth = DEC_DEPTH (enc.json.flags); 845 enc.limit = enc.json.flags & F_ASCII ? 0x000080UL
846 : enc.json.flags & F_LATIN1 ? 0x000100UL
847 : 0x110000UL;
672 848
673 SvPOK_only (enc.sv); 849 SvPOK_only (enc.sv);
674 encode_sv (&enc, scalar); 850 encode_sv (&enc, scalar);
851 encode_nl (&enc);
675 852
676 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 853 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
677 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 854 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
678 855
679 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8))) 856 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
… …
697 JSON json; 874 JSON json;
698 U32 depth; // recursion depth 875 U32 depth; // recursion depth
699 U32 maxdepth; // recursion depth limit 876 U32 maxdepth; // recursion depth limit
700} dec_t; 877} dec_t;
701 878
702inline void 879INLINE void
880decode_comment (dec_t *dec)
881{
882 // only '#'-style comments allowed a.t.m.
883
884 while (*dec->cur && *dec->cur != 0x0a && *dec->cur != 0x0d)
885 ++dec->cur;
886}
887
888INLINE void
703decode_ws (dec_t *dec) 889decode_ws (dec_t *dec)
704{ 890{
705 for (;;) 891 for (;;)
706 { 892 {
707 char ch = *dec->cur; 893 char ch = *dec->cur;
708 894
709 if (ch > 0x20 895 if (ch > 0x20)
896 {
897 if (expect_false (ch == '#'))
898 {
899 if (dec->json.flags & F_RELAXED)
900 decode_comment (dec);
901 else
902 break;
903 }
904 else
905 break;
906 }
710 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) 907 else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)
711 break; 908 break; // parse error, but let higher level handle it, gives better error messages
712 909
713 ++dec->cur; 910 ++dec->cur;
714 } 911 }
715} 912}
716 913
… …
720 if (*dec->cur != ch) \ 917 if (*dec->cur != ch) \
721 ERR (# ch " expected"); \ 918 ERR (# ch " expected"); \
722 ++dec->cur; \ 919 ++dec->cur; \
723 SE 920 SE
724 921
725#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)") 922#define DEC_INC_DEPTH if (++dec->depth > dec->json.max_depth) ERR (ERR_NESTING_EXCEEDED)
726#define DEC_DEC_DEPTH --dec->depth 923#define DEC_DEC_DEPTH --dec->depth
727 924
728static SV *decode_sv (dec_t *dec); 925static SV *decode_sv (dec_t *dec);
729 926
730static signed char decode_hexdigit[256]; 927static signed char decode_hexdigit[256];
… …
822 1019
823 if (hi >= 0x80) 1020 if (hi >= 0x80)
824 { 1021 {
825 utf8 = 1; 1022 utf8 = 1;
826 1023
827 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0); 1024 cur = encode_utf8 (cur, hi);
828 } 1025 }
829 else 1026 else
830 *cur++ = hi; 1027 *cur++ = hi;
831 } 1028 }
832 break; 1029 break;
… …
834 default: 1031 default:
835 --dec_cur; 1032 --dec_cur;
836 ERR ("illegal backslash escape sequence in string"); 1033 ERR ("illegal backslash escape sequence in string");
837 } 1034 }
838 } 1035 }
839 else if (expect_true (ch >= 0x20 && ch <= 0x7f)) 1036 else if (expect_true (ch >= 0x20 && ch < 0x80))
840 *cur++ = ch; 1037 *cur++ = ch;
841 else if (ch >= 0x80) 1038 else if (ch >= 0x80)
842 { 1039 {
843 STRLEN clen; 1040 STRLEN clen;
844 UV uch;
845 1041
846 --dec_cur; 1042 --dec_cur;
847 1043
848 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen); 1044 decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
849 if (clen == (STRLEN)-1) 1045 if (clen == (STRLEN)-1)
850 ERR ("malformed UTF-8 character in JSON string"); 1046 ERR ("malformed UTF-8 character in JSON string");
851 1047
852 do 1048 do
853 *cur++ = *dec_cur++; 1049 *cur++ = *dec_cur++;
… …
870 { 1066 {
871 STRLEN len = cur - buf; 1067 STRLEN len = cur - buf;
872 1068
873 if (sv) 1069 if (sv)
874 { 1070 {
875 SvGROW (sv, SvCUR (sv) + len + 1); 1071 STRLEN cur = SvCUR (sv);
1072
1073 if (SvLEN (sv) <= cur + len)
1074 SvGROW (sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
1075
876 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 1076 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
877 SvCUR_set (sv, SvCUR (sv) + len); 1077 SvCUR_set (sv, SvCUR (sv) + len);
878 } 1078 }
879 else 1079 else
880 sv = newSVpvn (buf, len); 1080 sv = newSVpvn (buf, len);
… …
967 1167
968 if (!is_nv) 1168 if (!is_nv)
969 { 1169 {
970 int len = dec->cur - start; 1170 int len = dec->cur - start;
971 1171
972 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so 1172 // special case the rather common 1..5-digit-int case
973 if (*start == '-') 1173 if (*start == '-')
974 switch (len) 1174 switch (len)
975 { 1175 {
976 case 2: return newSViv (-( start [1] - '0' * 1)); 1176 case 2: return newSViv (-(IV)( start [1] - '0' * 1));
977 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); 1177 case 3: return newSViv (-(IV)( start [1] * 10 + start [2] - '0' * 11));
978 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); 1178 case 4: return newSViv (-(IV)( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
979 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); 1179 case 5: return newSViv (-(IV)( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
1180 case 6: return newSViv (-(IV)(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
980 } 1181 }
981 else 1182 else
982 switch (len) 1183 switch (len)
983 { 1184 {
984 case 1: return newSViv ( start [0] - '0' * 1); 1185 case 1: return newSViv ( start [0] - '0' * 1);
985 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); 1186 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
986 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); 1187 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
987 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); 1188 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
1189 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
988 } 1190 }
989 1191
990 { 1192 {
991 UV uv; 1193 UV uv;
992 int numtype = grok_number (start, len, &uv); 1194 int numtype = grok_number (start, len, &uv);
… …
1001 } 1203 }
1002 1204
1003 len -= *start == '-' ? 1 : 0; 1205 len -= *start == '-' ? 1 : 0;
1004 1206
1005 // does not fit into IV or UV, try NV 1207 // does not fit into IV or UV, try NV
1006 if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len) 1208 if (len <= NV_DIG)
1007 #if defined (LDBL_DIG)
1008 || (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len)
1009 #endif
1010 )
1011 // fits into NV without loss of precision 1209 // fits into NV without loss of precision
1012 return newSVnv (Atof (start)); 1210 return newSVnv (json_atof (start));
1013 1211
1014 // everything else fails, convert it to a string 1212 // everything else fails, convert it to a string
1015 return newSVpvn (start, dec->cur - start); 1213 return newSVpvn (start, dec->cur - start);
1016 } 1214 }
1017 1215
1018 // loss of precision here 1216 // loss of precision here
1019 return newSVnv (Atof (start)); 1217 return newSVnv (json_atof (start));
1020 1218
1021fail: 1219fail:
1022 return 0; 1220 return 0;
1023} 1221}
1024 1222
… …
1053 1251
1054 if (*dec->cur != ',') 1252 if (*dec->cur != ',')
1055 ERR (", or ] expected while parsing array"); 1253 ERR (", or ] expected while parsing array");
1056 1254
1057 ++dec->cur; 1255 ++dec->cur;
1256
1257 decode_ws (dec);
1258
1259 if (*dec->cur == ']' && dec->json.flags & F_RELAXED)
1260 {
1261 ++dec->cur;
1262 break;
1263 }
1058 } 1264 }
1059 1265
1060 DEC_DEC_DEPTH; 1266 DEC_DEC_DEPTH;
1061 return newRV_noinc ((SV *)av); 1267 return newRV_noinc ((SV *)av);
1062 1268
… …
1078 if (*dec->cur == '}') 1284 if (*dec->cur == '}')
1079 ++dec->cur; 1285 ++dec->cur;
1080 else 1286 else
1081 for (;;) 1287 for (;;)
1082 { 1288 {
1083 decode_ws (dec); EXPECT_CH ('"'); 1289 EXPECT_CH ('"');
1084 1290
1085 // heuristic: assume that 1291 // heuristic: assume that
1086 // a) decode_str + hv_store_ent are abysmally slow. 1292 // a) decode_str + hv_store_ent are abysmally slow.
1087 // b) most hash keys are short, simple ascii text. 1293 // b) most hash keys are short, simple ascii text.
1088 // => try to "fast-match" such strings to avoid 1294 // => try to "fast-match" such strings to avoid
… …
1092 char *p = dec->cur; 1298 char *p = dec->cur;
1093 char *e = p + 24; // only try up to 24 bytes 1299 char *e = p + 24; // only try up to 24 bytes
1094 1300
1095 for (;;) 1301 for (;;)
1096 { 1302 {
1097 // the >= 0x80 is true on most architectures 1303 // the >= 0x80 is false on most architectures
1098 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\') 1304 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1099 { 1305 {
1100 // slow path, back up and use decode_str 1306 // slow path, back up and use decode_str
1101 SV *key = decode_str (dec); 1307 SV *key = decode_str (dec);
1102 if (!key) 1308 if (!key)
1103 goto fail; 1309 goto fail;
1104 1310
1105 decode_ws (dec); EXPECT_CH (':'); 1311 decode_ws (dec); EXPECT_CH (':');
1106 1312
1313 decode_ws (dec);
1107 value = decode_sv (dec); 1314 value = decode_sv (dec);
1108 if (!value) 1315 if (!value)
1109 { 1316 {
1110 SvREFCNT_dec (key); 1317 SvREFCNT_dec (key);
1111 goto fail; 1318 goto fail;
… …
1123 int len = p - key; 1330 int len = p - key;
1124 dec->cur = p + 1; 1331 dec->cur = p + 1;
1125 1332
1126 decode_ws (dec); EXPECT_CH (':'); 1333 decode_ws (dec); EXPECT_CH (':');
1127 1334
1335 decode_ws (dec);
1128 value = decode_sv (dec); 1336 value = decode_sv (dec);
1129 if (!value) 1337 if (!value)
1130 goto fail; 1338 goto fail;
1131 1339
1132 hv_store (hv, key, len, value, 0); 1340 hv_store (hv, key, len, value, 0);
… …
1148 1356
1149 if (*dec->cur != ',') 1357 if (*dec->cur != ',')
1150 ERR (", or } expected while parsing object/hash"); 1358 ERR (", or } expected while parsing object/hash");
1151 1359
1152 ++dec->cur; 1360 ++dec->cur;
1361
1362 decode_ws (dec);
1363
1364 if (*dec->cur == '}' && dec->json.flags & F_RELAXED)
1365 {
1366 ++dec->cur;
1367 break;
1368 }
1153 } 1369 }
1154 1370
1155 DEC_DEC_DEPTH; 1371 DEC_DEC_DEPTH;
1156 sv = newRV_noinc ((SV *)hv); 1372 sv = newRV_noinc ((SV *)hv);
1157 1373
… …
1222} 1438}
1223 1439
1224static SV * 1440static SV *
1225decode_sv (dec_t *dec) 1441decode_sv (dec_t *dec)
1226{ 1442{
1227 decode_ws (dec);
1228
1229 // the beauty of JSON: you need exactly one character lookahead 1443 // the beauty of JSON: you need exactly one character lookahead
1230 // to parse anything. 1444 // to parse everything.
1231 switch (*dec->cur) 1445 switch (*dec->cur)
1232 { 1446 {
1233 case '"': ++dec->cur; return decode_str (dec); 1447 case '"': ++dec->cur; return decode_str (dec);
1234 case '[': ++dec->cur; return decode_av (dec); 1448 case '[': ++dec->cur; return decode_av (dec);
1235 case '{': ++dec->cur; return decode_hv (dec); 1449 case '{': ++dec->cur; return decode_hv (dec);
1236 1450
1237 case '-': 1451 case '-':
1238 case '0': case '1': case '2': case '3': case '4': 1452 case '0': case '1': case '2': case '3': case '4':
1239 case '5': case '6': case '7': case '8': case '9': 1453 case '5': case '6': case '7': case '8': case '9':
1240 return decode_num (dec); 1454 return decode_num (dec);
… …
1242 case 't': 1456 case 't':
1243 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1457 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1244 { 1458 {
1245 dec->cur += 4; 1459 dec->cur += 4;
1246#if JSON_SLOW 1460#if JSON_SLOW
1247 json_true = get_sv ("JSON::XS::true", 1); SvREADONLY_on (json_true); 1461 json_true = get_bool ("JSON::XS::true");
1248#endif 1462#endif
1249 return SvREFCNT_inc (json_true); 1463 return newSVsv (json_true);
1250 } 1464 }
1251 else 1465 else
1252 ERR ("'true' expected"); 1466 ERR ("'true' expected");
1253 1467
1254 break; 1468 break;
… …
1256 case 'f': 1470 case 'f':
1257 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1471 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1258 { 1472 {
1259 dec->cur += 5; 1473 dec->cur += 5;
1260#if JSON_SLOW 1474#if JSON_SLOW
1261 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); 1475 json_false = get_bool ("JSON::XS::false");
1262#endif 1476#endif
1263 return SvREFCNT_inc (json_false); 1477 return newSVsv (json_false);
1264 } 1478 }
1265 else 1479 else
1266 ERR ("'false' expected"); 1480 ERR ("'false' expected");
1267 1481
1268 break; 1482 break;
… …
1286fail: 1500fail:
1287 return 0; 1501 return 0;
1288} 1502}
1289 1503
1290static SV * 1504static SV *
1291decode_json (SV *string, JSON *json, UV *offset_return) 1505decode_json (SV *string, JSON *json, char **offset_return)
1292{ 1506{
1293 dec_t dec; 1507 dec_t dec;
1294 UV offset;
1295 SV *sv; 1508 SV *sv;
1296 1509
1510 /* work around bugs in 5.10 where manipulating magic values
1511 * will perl ignore the magic in subsequent accesses
1512 */
1297 SvGETMAGIC (string); 1513 /*SvGETMAGIC (string);*/
1514 if (SvMAGICAL (string))
1515 string = sv_2mortal (newSVsv (string));
1516
1298 SvUPGRADE (string, SVt_PV); 1517 SvUPGRADE (string, SVt_PV);
1299 1518
1300 if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags)) 1519 /* work around a bug in perl 5.10, which causes SvCUR to fail an
1520 * assertion with -DDEBUGGING, although SvCUR is documented to
1521 * return the xpv_cur field which certainly exists after upgrading.
1522 * according to nicholas clark, calling SvPOK fixes this.
1523 * But it doesn't fix it, so try another workaround, call SvPV_nolen
1524 * and hope for the best.
1525 * Damnit, SvPV_nolen still trips over yet another assertion. This
1526 * assertion business is seriously broken, try yet another workaround
1527 * for the broken -DDEBUGGING.
1528 */
1529 {
1530#ifdef DEBUGGING
1531 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
1532#else
1533 STRLEN offset = SvCUR (string);
1534#endif
1535
1536 if (offset > json->max_size && json->max_size)
1301 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", 1537 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1302 (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags)); 1538 (unsigned long)SvCUR (string), (unsigned long)json->max_size);
1539 }
1303 1540
1304 if (json->flags & F_UTF8) 1541 if (json->flags & F_UTF8)
1305 sv_utf8_downgrade (string, 0); 1542 sv_utf8_downgrade (string, 0);
1306 else 1543 else
1307 sv_utf8_upgrade (string); 1544 sv_utf8_upgrade (string);
1308 1545
1309 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1546 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
1310 1547
1311 dec.json = *json; 1548 dec.json = *json;
1312 dec.cur = SvPVX (string); 1549 dec.cur = SvPVX (string);
1313 dec.end = SvEND (string); 1550 dec.end = SvEND (string);
1314 dec.err = 0; 1551 dec.err = 0;
1315 dec.depth = 0; 1552 dec.depth = 0;
1316 dec.maxdepth = DEC_DEPTH (dec.json.flags);
1317 1553
1318 if (dec.json.cb_object || dec.json.cb_sk_object) 1554 if (dec.json.cb_object || dec.json.cb_sk_object)
1319 dec.json.flags |= F_HOOK; 1555 dec.json.flags |= F_HOOK;
1320 1556
1321 *dec.end = 0; // this should basically be a nop, too, but make sure it's there 1557 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1558
1559 decode_ws (&dec);
1322 sv = decode_sv (&dec); 1560 sv = decode_sv (&dec);
1561
1562 if (offset_return)
1563 *offset_return = dec.cur;
1323 1564
1324 if (!(offset_return || !sv)) 1565 if (!(offset_return || !sv))
1325 { 1566 {
1326 // check for trailing garbage 1567 // check for trailing garbage
1327 decode_ws (&dec); 1568 decode_ws (&dec);
… …
1330 { 1571 {
1331 dec.err = "garbage after JSON object"; 1572 dec.err = "garbage after JSON object";
1332 SvREFCNT_dec (sv); 1573 SvREFCNT_dec (sv);
1333 sv = 0; 1574 sv = 0;
1334 } 1575 }
1335 }
1336
1337 if (offset_return || !sv)
1338 {
1339 offset = dec.json.flags & F_UTF8
1340 ? dec.cur - SvPVX (string)
1341 : utf8_distance (dec.cur, SvPVX (string));
1342
1343 if (offset_return)
1344 *offset_return = offset;
1345 } 1576 }
1346 1577
1347 if (!sv) 1578 if (!sv)
1348 { 1579 {
1349 SV *uni = sv_newmortal (); 1580 SV *uni = sv_newmortal ();
… …
1355 SAVEVPTR (PL_curcop); 1586 SAVEVPTR (PL_curcop);
1356 PL_curcop = &cop; 1587 PL_curcop = &cop;
1357 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1588 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1358 LEAVE; 1589 LEAVE;
1359 1590
1360 croak ("%s, at character offset %d [\"%s\"]", 1591 croak ("%s, at character offset %d (before \"%s\")",
1361 dec.err, 1592 dec.err,
1362 (int)offset, 1593 ptr_to_index (string, dec.cur),
1363 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1594 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1364 } 1595 }
1365 1596
1366 sv = sv_2mortal (sv); 1597 sv = sv_2mortal (sv);
1367 1598
1368 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1599 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))
1369 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); 1600 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1370 1601
1371 return sv; 1602 return sv;
1603}
1604
1605/////////////////////////////////////////////////////////////////////////////
1606// incremental parser
1607
1608static void
1609incr_parse (JSON *self)
1610{
1611 const char *p = SvPVX (self->incr_text) + self->incr_pos;
1612
1613 // the state machine here is a bit convoluted and could be simplified a lot
1614 // but this would make it slower, so...
1615
1616 for (;;)
1617 {
1618 //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
1619 switch (self->incr_mode)
1620 {
1621 // only used for initial whitespace skipping
1622 case INCR_M_WS:
1623 for (;;)
1624 {
1625 if (*p > 0x20)
1626 {
1627 if (*p == '#')
1628 {
1629 self->incr_mode = INCR_M_C0;
1630 goto incr_m_c;
1631 }
1632 else
1633 {
1634 self->incr_mode = INCR_M_JSON;
1635 goto incr_m_json;
1636 }
1637 }
1638 else if (!*p)
1639 goto interrupt;
1640
1641 ++p;
1642 }
1643
1644 // skip a single char inside a string (for \\-processing)
1645 case INCR_M_BS:
1646 if (!*p)
1647 goto interrupt;
1648
1649 ++p;
1650 self->incr_mode = INCR_M_STR;
1651 goto incr_m_str;
1652
1653 // inside #-style comments
1654 case INCR_M_C0:
1655 case INCR_M_C1:
1656 incr_m_c:
1657 for (;;)
1658 {
1659 if (*p == '\n')
1660 {
1661 self->incr_mode = self->incr_mode == INCR_M_C0 ? INCR_M_WS : INCR_M_JSON;
1662 break;
1663 }
1664 else if (!*p)
1665 goto interrupt;
1666
1667 ++p;
1668 }
1669
1670 break;
1671
1672 // inside a string
1673 case INCR_M_STR:
1674 incr_m_str:
1675 for (;;)
1676 {
1677 if (*p == '"')
1678 {
1679 ++p;
1680 self->incr_mode = INCR_M_JSON;
1681
1682 if (!self->incr_nest)
1683 goto interrupt;
1684
1685 goto incr_m_json;
1686 }
1687 else if (*p == '\\')
1688 {
1689 ++p; // "virtually" consumes character after \
1690
1691 if (!*p) // if at end of string we have to switch modes
1692 {
1693 self->incr_mode = INCR_M_BS;
1694 goto interrupt;
1695 }
1696 }
1697 else if (!*p)
1698 goto interrupt;
1699
1700 ++p;
1701 }
1702
1703 // after initial ws, outside string
1704 case INCR_M_JSON:
1705 incr_m_json:
1706 for (;;)
1707 {
1708 switch (*p++)
1709 {
1710 case 0:
1711 --p;
1712 goto interrupt;
1713
1714 case 0x09:
1715 case 0x0a:
1716 case 0x0d:
1717 case 0x20:
1718 if (!self->incr_nest)
1719 {
1720 --p; // do not eat the whitespace, let the next round do it
1721 goto interrupt;
1722 }
1723 break;
1724
1725 case '"':
1726 self->incr_mode = INCR_M_STR;
1727 goto incr_m_str;
1728
1729 case '[':
1730 case '{':
1731 if (++self->incr_nest > self->max_depth)
1732 croak (ERR_NESTING_EXCEEDED);
1733 break;
1734
1735 case ']':
1736 case '}':
1737 if (--self->incr_nest <= 0)
1738 goto interrupt;
1739 break;
1740
1741 case '#':
1742 self->incr_mode = INCR_M_C1;
1743 goto incr_m_c;
1744 }
1745 }
1746 }
1747
1748 modechange:
1749 ;
1750 }
1751
1752interrupt:
1753 self->incr_pos = p - SvPVX (self->incr_text);
1754 //printf ("interrupt<%.*s>\n", self->incr_pos, SvPVX(self->incr_text));//D
1755 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D
1372} 1756}
1373 1757
1374///////////////////////////////////////////////////////////////////////////// 1758/////////////////////////////////////////////////////////////////////////////
1375// XS interface functions 1759// XS interface functions
1376 1760
… …
1388 : -1; 1772 : -1;
1389 1773
1390 json_stash = gv_stashpv ("JSON::XS" , 1); 1774 json_stash = gv_stashpv ("JSON::XS" , 1);
1391 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); 1775 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1392 1776
1393 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true ); 1777 json_true = get_bool ("JSON::XS::true");
1394 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); 1778 json_false = get_bool ("JSON::XS::false");
1779
1780 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */
1395} 1781}
1396 1782
1397PROTOTYPES: DISABLE 1783PROTOTYPES: DISABLE
1398 1784
1399void CLONE (...) 1785void CLONE (...)
… …
1404void new (char *klass) 1790void new (char *klass)
1405 PPCODE: 1791 PPCODE:
1406{ 1792{
1407 SV *pv = NEWSV (0, sizeof (JSON)); 1793 SV *pv = NEWSV (0, sizeof (JSON));
1408 SvPOK_only (pv); 1794 SvPOK_only (pv);
1409 Zero (SvPVX (pv), 1, JSON); 1795 json_init ((JSON *)SvPVX (pv));
1410 ((JSON *)SvPVX (pv))->flags = F_DEFAULT; 1796 XPUSHs (sv_2mortal (sv_bless (
1411 XPUSHs (sv_2mortal (sv_bless (newRV_noinc (pv), JSON_STASH))); 1797 newRV_noinc (pv),
1798 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1799 )));
1412} 1800}
1413 1801
1414void ascii (JSON *self, int enable = 1) 1802void ascii (JSON *self, int enable = 1)
1415 ALIAS: 1803 ALIAS:
1416 ascii = F_ASCII 1804 ascii = F_ASCII
… …
1423 pretty = F_PRETTY 1811 pretty = F_PRETTY
1424 allow_nonref = F_ALLOW_NONREF 1812 allow_nonref = F_ALLOW_NONREF
1425 shrink = F_SHRINK 1813 shrink = F_SHRINK
1426 allow_blessed = F_ALLOW_BLESSED 1814 allow_blessed = F_ALLOW_BLESSED
1427 convert_blessed = F_CONV_BLESSED 1815 convert_blessed = F_CONV_BLESSED
1816 relaxed = F_RELAXED
1817 allow_unknown = F_ALLOW_UNKNOWN
1428 PPCODE: 1818 PPCODE:
1429{ 1819{
1430 if (enable) 1820 if (enable)
1431 self->flags |= ix; 1821 self->flags |= ix;
1432 else 1822 else
1433 self->flags &= ~ix; 1823 self->flags &= ~ix;
1434 1824
1435 XPUSHs (ST (0)); 1825 XPUSHs (ST (0));
1436} 1826}
1437 1827
1438void max_depth (JSON *self, UV max_depth = 0x80000000UL) 1828void get_ascii (JSON *self)
1829 ALIAS:
1830 get_ascii = F_ASCII
1831 get_latin1 = F_LATIN1
1832 get_utf8 = F_UTF8
1833 get_indent = F_INDENT
1834 get_canonical = F_CANONICAL
1835 get_space_before = F_SPACE_BEFORE
1836 get_space_after = F_SPACE_AFTER
1837 get_allow_nonref = F_ALLOW_NONREF
1838 get_shrink = F_SHRINK
1839 get_allow_blessed = F_ALLOW_BLESSED
1840 get_convert_blessed = F_CONV_BLESSED
1841 get_relaxed = F_RELAXED
1842 get_allow_unknown = F_ALLOW_UNKNOWN
1439 PPCODE: 1843 PPCODE:
1440{ 1844 XPUSHs (boolSV (self->flags & ix));
1441 UV log2 = 0;
1442 1845
1443 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL; 1846void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
1444 1847 PPCODE:
1445 while ((1UL << log2) < max_depth) 1848 self->max_depth = max_depth;
1446 ++log2;
1447
1448 self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1449
1450 XPUSHs (ST (0)); 1849 XPUSHs (ST (0));
1451}
1452 1850
1851U32 get_max_depth (JSON *self)
1852 CODE:
1853 RETVAL = self->max_depth;
1854 OUTPUT:
1855 RETVAL
1856
1453void max_size (JSON *self, UV max_size = 0) 1857void max_size (JSON *self, U32 max_size = 0)
1454 PPCODE: 1858 PPCODE:
1455{ 1859 self->max_size = max_size;
1456 UV log2 = 0;
1457
1458 if (max_size > 0x80000000UL) max_size = 0x80000000UL;
1459 if (max_size == 1) max_size = 2;
1460
1461 while ((1UL << log2) < max_size)
1462 ++log2;
1463
1464 self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1465
1466 XPUSHs (ST (0)); 1860 XPUSHs (ST (0));
1467} 1861
1862int get_max_size (JSON *self)
1863 CODE:
1864 RETVAL = self->max_size;
1865 OUTPUT:
1866 RETVAL
1468 1867
1469void filter_json_object (JSON *self, SV *cb = &PL_sv_undef) 1868void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1470 PPCODE: 1869 PPCODE:
1471{ 1870{
1472 SvREFCNT_dec (self->cb_object); 1871 SvREFCNT_dec (self->cb_object);
… …
1506 XPUSHs (decode_json (jsonstr, self, 0)); 1905 XPUSHs (decode_json (jsonstr, self, 0));
1507 1906
1508void decode_prefix (JSON *self, SV *jsonstr) 1907void decode_prefix (JSON *self, SV *jsonstr)
1509 PPCODE: 1908 PPCODE:
1510{ 1909{
1511 UV offset; 1910 char *offset;
1512 EXTEND (SP, 2); 1911 EXTEND (SP, 2);
1513 PUSHs (decode_json (jsonstr, self, &offset)); 1912 PUSHs (decode_json (jsonstr, self, &offset));
1514 PUSHs (sv_2mortal (newSVuv (offset))); 1913 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset))));
1914}
1915
1916void incr_parse (JSON *self, SV *jsonstr = 0)
1917 PPCODE:
1918{
1919 if (!self->incr_text)
1920 self->incr_text = newSVpvn ("", 0);
1921
1922 // append data, if any
1923 if (jsonstr)
1924 {
1925 if (SvUTF8 (jsonstr))
1926 {
1927 if (!SvUTF8 (self->incr_text))
1928 {
1929 /* utf-8-ness differs, need to upgrade */
1930 sv_utf8_upgrade (self->incr_text);
1931
1932 if (self->incr_pos)
1933 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
1934 - (U8 *)SvPVX (self->incr_text);
1935 }
1936 }
1937 else if (SvUTF8 (self->incr_text))
1938 sv_utf8_upgrade (jsonstr);
1939
1940 {
1941 STRLEN len;
1942 const char *str = SvPV (jsonstr, len);
1943 STRLEN cur = SvCUR (self->incr_text);
1944
1945 if (SvLEN (self->incr_text) <= cur + len)
1946 SvGROW (self->incr_text, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
1947
1948 Move (str, SvEND (self->incr_text), len, char);
1949 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
1950 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
1951 }
1952 }
1953
1954 if (GIMME_V != G_VOID)
1955 do
1956 {
1957 char *offset;
1958
1959 if (!INCR_DONE (self))
1960 {
1961 incr_parse (self);
1962
1963 if (self->incr_pos > self->max_size && self->max_size)
1964 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1965 (unsigned long)self->incr_pos, (unsigned long)self->max_size);
1966
1967 if (!INCR_DONE (self))
1968 break;
1969 }
1970
1971 XPUSHs (decode_json (self->incr_text, self, &offset));
1972
1973 self->incr_pos -= offset - SvPVX (self->incr_text);
1974 self->incr_nest = 0;
1975 self->incr_mode = 0;
1976
1977 sv_chop (self->incr_text, offset);
1978 }
1979 while (GIMME_V == G_ARRAY);
1980}
1981
1982SV *incr_text (JSON *self)
1983 ATTRS: lvalue
1984 CODE:
1985{
1986 if (self->incr_pos)
1987 croak ("incr_text can not be called when the incremental parser already started parsing");
1988
1989 RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef;
1990}
1991 OUTPUT:
1992 RETVAL
1993
1994void incr_skip (JSON *self)
1995 CODE:
1996{
1997 if (self->incr_pos)
1998 {
1999 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + self->incr_pos);
2000 self->incr_pos = 0;
2001 self->incr_nest = 0;
2002 self->incr_mode = 0;
2003 }
2004}
2005
2006void incr_reset (JSON *self)
2007 CODE:
2008{
2009 SvREFCNT_dec (self->incr_text);
2010 self->incr_text = 0;
2011 self->incr_pos = 0;
2012 self->incr_nest = 0;
2013 self->incr_mode = 0;
1515} 2014}
1516 2015
1517void DESTROY (JSON *self) 2016void DESTROY (JSON *self)
1518 CODE: 2017 CODE:
1519 SvREFCNT_dec (self->cb_sk_object); 2018 SvREFCNT_dec (self->cb_sk_object);
1520 SvREFCNT_dec (self->cb_object); 2019 SvREFCNT_dec (self->cb_object);
2020 SvREFCNT_dec (self->incr_text);
1521 2021
1522PROTOTYPES: ENABLE 2022PROTOTYPES: ENABLE
1523 2023
1524void to_json (SV *scalar) 2024void encode_json (SV *scalar)
2025 ALIAS:
2026 to_json_ = 0
2027 encode_json = F_UTF8
1525 PPCODE: 2028 PPCODE:
1526{ 2029{
1527 JSON json = { F_DEFAULT | F_UTF8 }; 2030 JSON json;
2031 json_init (&json);
2032 json.flags |= ix;
1528 XPUSHs (encode_json (scalar, &json)); 2033 XPUSHs (encode_json (scalar, &json));
1529} 2034}
1530 2035
1531void from_json (SV *jsonstr) 2036void decode_json (SV *jsonstr)
2037 ALIAS:
2038 from_json_ = 0
2039 decode_json = F_UTF8
1532 PPCODE: 2040 PPCODE:
1533{ 2041{
1534 JSON json = { F_DEFAULT | F_UTF8 }; 2042 JSON json;
2043 json_init (&json);
2044 json.flags |= ix;
1535 XPUSHs (decode_json (jsonstr, &json, 0)); 2045 XPUSHs (decode_json (jsonstr, &json, 0));
1536} 2046}
1537 2047

Diff Legend

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