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 1.23 by root, Tue Apr 3 23:34:17 2007 UTC vs.
Revision 1.72 by root, Wed Mar 19 04:08:22 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>
9#include <float.h>
8 10
11#if defined(__BORLANDC__) || defined(_MSC_VER)
12# define snprintf _snprintf // C compilers have this in stdio.h
13#endif
14
15// 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#ifndef UTF8_MAXBYTES
18# define UTF8_MAXBYTES 13
19#endif
20
9#define F_ASCII 0x00000001UL 21#define F_ASCII 0x00000001UL
22#define F_LATIN1 0x00000002UL
10#define F_UTF8 0x00000002UL 23#define F_UTF8 0x00000004UL
11#define F_INDENT 0x00000004UL 24#define F_INDENT 0x00000008UL
12#define F_CANONICAL 0x00000008UL 25#define F_CANONICAL 0x00000010UL
13#define F_SPACE_BEFORE 0x00000010UL 26#define F_SPACE_BEFORE 0x00000020UL
14#define F_SPACE_AFTER 0x00000020UL 27#define F_SPACE_AFTER 0x00000040UL
15#define F_ALLOW_NONREF 0x00000080UL 28#define F_ALLOW_NONREF 0x00000100UL
16#define F_SHRINK 0x00000100UL 29#define F_SHRINK 0x00000200UL
30#define F_ALLOW_BLESSED 0x00000400UL
31#define F_CONV_BLESSED 0x00000800UL
32#define F_RELAXED 0x00001000UL
33
17#define F_MAXDEPTH 0xf8000000UL 34#define F_MAXDEPTH 0xf8000000UL
18#define S_MAXDEPTH 27 35#define S_MAXDEPTH 27
36#define F_MAXSIZE 0x01f00000UL
37#define S_MAXSIZE 20
38#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
19 39
20#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH)) 40#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
21 41#define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE ))
22// F_SELFCONVERT? <=> to_json/toJson
23// F_BLESSED? <=> { $__class__$ => }
24 42
25#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 43#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
26#define F_DEFAULT (12UL << S_MAXDEPTH) 44#define F_DEFAULT (9UL << S_MAXDEPTH)
27 45
28#define INIT_SIZE 32 // initial scalar size to be allocated 46#define INIT_SIZE 32 // initial scalar size to be allocated
29#define INDENT_STEP 3 // spaces per indentation level 47#define INDENT_STEP 3 // spaces per indentation level
30 48
31#define UTF8_MAX_LEN 11 // for perls UTF-X: max. number of octets per character
32#define SHORT_STRING_LEN 512 // special-case strings of up to this size 49#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
33 50
34#define SB do { 51#define SB do {
35#define SE } while (0) 52#define SE } while (0)
36 53
54#if __GNUC__ >= 3
55# define expect(expr,value) __builtin_expect ((expr), (value))
56# define INLINE static inline
57#else
58# define expect(expr,value) (expr)
59# define INLINE static
60#endif
61
62#define expect_false(expr) expect ((expr) != 0, 0)
63#define expect_true(expr) expect ((expr) != 0, 1)
64
65#define IN_RANGE_INC(type,val,beg,end) \
66 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \
67 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
68
69#ifdef USE_ITHREADS
70# define JSON_SLOW 1
71# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1))
72#else
73# define JSON_SLOW 0
74# define JSON_STASH json_stash
75#endif
76
37static HV *json_stash; // JSON::XS:: 77static HV *json_stash, *json_boolean_stash; // JSON::XS::
78static SV *json_true, *json_false;
79
80typedef struct {
81 U32 flags;
82 SV *cb_object;
83 HV *cb_sk_object;
84} JSON;
38 85
39///////////////////////////////////////////////////////////////////////////// 86/////////////////////////////////////////////////////////////////////////////
40// utility functions 87// utility functions
41 88
42static UV * 89INLINE void
43SvJSON (SV *sv)
44{
45 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash))
46 croak ("object is not of type JSON::XS");
47
48 return &SvUVX (SvRV (sv));
49}
50
51static void
52shrink (SV *sv) 90shrink (SV *sv)
53{ 91{
54 sv_utf8_downgrade (sv, 1); 92 sv_utf8_downgrade (sv, 1);
55 if (SvLEN (sv) > SvCUR (sv) + 1) 93 if (SvLEN (sv) > SvCUR (sv) + 1)
56 { 94 {
… …
65// decode an utf-8 character and return it, or (UV)-1 in 103// decode an utf-8 character and return it, or (UV)-1 in
66// case of an error. 104// case of an error.
67// we special-case "safe" characters from U+80 .. U+7FF, 105// we special-case "safe" characters from U+80 .. U+7FF,
68// but use the very good perl function to parse anything else. 106// but use the very good perl function to parse anything else.
69// note that we never call this function for a ascii codepoints 107// note that we never call this function for a ascii codepoints
70static UV 108INLINE UV
71decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) 109decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
72{ 110{
73 if (s[0] > 0xdf || s[0] < 0xc2) 111 if (expect_true (len >= 2
74 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); 112 && IN_RANGE_INC (char, s[0], 0xc2, 0xdf)
75 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) 113 && IN_RANGE_INC (char, s[1], 0x80, 0xbf)))
76 { 114 {
77 *clen = 2; 115 *clen = 2;
78 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 116 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
79 } 117 }
80 else 118 else
81 { 119 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
82 *clen = (STRLEN)-1; 120}
83 return (UV)-1; 121
122// likewise for encoding, also never called for ascii codepoints
123// this function takes advantage of this fact, although current gccs
124// seem to optimise the check for >= 0x80 away anyways
125INLINE unsigned char *
126encode_utf8 (unsigned char *s, UV ch)
127{
128 if (ch <= 0x7FF)
84 } 129 {
130 *s++ = (ch >> 6) | 0xc0;
131 *s++ = (ch & 0x3f) | 0x80;
132 }
133 else
134 s = uvuni_to_utf8_flags (s, ch, 0);
135
136 return s;
85} 137}
86 138
87///////////////////////////////////////////////////////////////////////////// 139/////////////////////////////////////////////////////////////////////////////
88// encoder 140// encoder
89 141
… …
91typedef struct 143typedef struct
92{ 144{
93 char *cur; // SvPVX (sv) + current output position 145 char *cur; // SvPVX (sv) + current output position
94 char *end; // SvEND (sv) 146 char *end; // SvEND (sv)
95 SV *sv; // result scalar 147 SV *sv; // result scalar
96 U32 flags; // F_* 148 JSON json;
97 U32 indent; // indentation level 149 U32 indent; // indentation level
98 U32 maxdepth; // max. indentation/recursion level 150 U32 maxdepth; // max. indentation/recursion level
151 UV limit; // escape character values >= this value when encoding
99} enc_t; 152} enc_t;
100 153
101static void 154INLINE void
102need (enc_t *enc, STRLEN len) 155need (enc_t *enc, STRLEN len)
103{ 156{
104 if (enc->cur + len >= enc->end) 157 if (expect_false (enc->cur + len >= enc->end))
105 { 158 {
106 STRLEN cur = enc->cur - SvPVX (enc->sv); 159 STRLEN cur = enc->cur - SvPVX (enc->sv);
107 SvGROW (enc->sv, cur + len + 1); 160 SvGROW (enc->sv, cur + len + 1);
108 enc->cur = SvPVX (enc->sv) + cur; 161 enc->cur = SvPVX (enc->sv) + cur;
109 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv); 162 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
110 } 163 }
111} 164}
112 165
113static void 166INLINE void
114encode_ch (enc_t *enc, char ch) 167encode_ch (enc_t *enc, char ch)
115{ 168{
116 need (enc, 1); 169 need (enc, 1);
117 *enc->cur++ = ch; 170 *enc->cur++ = ch;
118} 171}
… …
126 179
127 while (str < end) 180 while (str < end)
128 { 181 {
129 unsigned char ch = *(unsigned char *)str; 182 unsigned char ch = *(unsigned char *)str;
130 183
131 if (ch >= 0x20 && ch < 0x80) // most common case 184 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
132 { 185 {
133 if (ch == '"') // but with slow exceptions 186 if (expect_false (ch == '"')) // but with slow exceptions
134 { 187 {
135 need (enc, len += 1); 188 need (enc, len += 1);
136 *enc->cur++ = '\\'; 189 *enc->cur++ = '\\';
137 *enc->cur++ = '"'; 190 *enc->cur++ = '"';
138 } 191 }
139 else if (ch == '\\') 192 else if (expect_false (ch == '\\'))
140 { 193 {
141 need (enc, len += 1); 194 need (enc, len += 1);
142 *enc->cur++ = '\\'; 195 *enc->cur++ = '\\';
143 *enc->cur++ = '\\'; 196 *enc->cur++ = '\\';
144 } 197 }
… …
162 STRLEN clen; 215 STRLEN clen;
163 UV uch; 216 UV uch;
164 217
165 if (is_utf8) 218 if (is_utf8)
166 { 219 {
167 //uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
168 uch = decode_utf8 (str, end - str, &clen); 220 uch = decode_utf8 (str, end - str, &clen);
169 if (clen == (STRLEN)-1) 221 if (clen == (STRLEN)-1)
170 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str); 222 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
171 } 223 }
172 else 224 else
173 { 225 {
174 uch = ch; 226 uch = ch;
175 clen = 1; 227 clen = 1;
176 } 228 }
177 229
178 if (uch > 0x10FFFFUL) 230 if (uch < 0x80/*0x20*/ || uch >= enc->limit)
179 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
180
181 if (uch < 0x80 || enc->flags & F_ASCII)
182 { 231 {
183 if (uch > 0xFFFFUL) 232 if (uch > 0xFFFFUL)
184 { 233 {
234 if (uch > 0x10FFFFUL)
235 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
236
185 need (enc, len += 11); 237 need (enc, len += 11);
186 sprintf (enc->cur, "\\u%04x\\u%04x", 238 sprintf (enc->cur, "\\u%04x\\u%04x",
187 (int)((uch - 0x10000) / 0x400 + 0xD800), 239 (int)((uch - 0x10000) / 0x400 + 0xD800),
188 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 240 (int)((uch - 0x10000) % 0x400 + 0xDC00));
189 enc->cur += 12; 241 enc->cur += 12;
… …
200 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 252 *enc->cur++ = hexdigit [(uch >> 0) & 15];
201 } 253 }
202 254
203 str += clen; 255 str += clen;
204 } 256 }
257 else if (enc->json.flags & F_LATIN1)
258 {
259 *enc->cur++ = uch;
260 str += clen;
261 }
205 else if (is_utf8) 262 else if (is_utf8)
206 { 263 {
207 need (enc, len += clen); 264 need (enc, len += clen);
208 do 265 do
209 { 266 {
… …
211 } 268 }
212 while (--clen); 269 while (--clen);
213 } 270 }
214 else 271 else
215 { 272 {
216 need (enc, len += UTF8_MAX_LEN - 1); // never more than 11 bytes needed 273 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
217 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 274 enc->cur = encode_utf8 (enc->cur, uch);
218 ++str; 275 ++str;
219 } 276 }
220 } 277 }
221 } 278 }
222 } 279 }
223 280
224 --len; 281 --len;
225 } 282 }
226} 283}
227 284
228static void 285INLINE void
229encode_indent (enc_t *enc) 286encode_indent (enc_t *enc)
230{ 287{
231 if (enc->flags & F_INDENT) 288 if (enc->json.flags & F_INDENT)
232 { 289 {
233 int spaces = enc->indent * INDENT_STEP; 290 int spaces = enc->indent * INDENT_STEP;
234 291
235 need (enc, spaces); 292 need (enc, spaces);
236 memset (enc->cur, ' ', spaces); 293 memset (enc->cur, ' ', spaces);
237 enc->cur += spaces; 294 enc->cur += spaces;
238 } 295 }
239} 296}
240 297
241static void 298INLINE void
242encode_space (enc_t *enc) 299encode_space (enc_t *enc)
243{ 300{
244 need (enc, 1); 301 need (enc, 1);
245 encode_ch (enc, ' '); 302 encode_ch (enc, ' ');
246} 303}
247 304
248static void 305INLINE void
249encode_nl (enc_t *enc) 306encode_nl (enc_t *enc)
250{ 307{
251 if (enc->flags & F_INDENT) 308 if (enc->json.flags & F_INDENT)
252 { 309 {
253 need (enc, 1); 310 need (enc, 1);
254 encode_ch (enc, '\n'); 311 encode_ch (enc, '\n');
255 } 312 }
256} 313}
257 314
258static void 315INLINE void
259encode_comma (enc_t *enc) 316encode_comma (enc_t *enc)
260{ 317{
261 encode_ch (enc, ','); 318 encode_ch (enc, ',');
262 319
263 if (enc->flags & F_INDENT) 320 if (enc->json.flags & F_INDENT)
264 encode_nl (enc); 321 encode_nl (enc);
265 else if (enc->flags & F_SPACE_AFTER) 322 else if (enc->json.flags & F_SPACE_AFTER)
266 encode_space (enc); 323 encode_space (enc);
267} 324}
268 325
269static void encode_sv (enc_t *enc, SV *sv); 326static void encode_sv (enc_t *enc, SV *sv);
270 327
… …
274 int i, len = av_len (av); 331 int i, len = av_len (av);
275 332
276 if (enc->indent >= enc->maxdepth) 333 if (enc->indent >= enc->maxdepth)
277 croak ("data structure too deep (hit recursion limit)"); 334 croak ("data structure too deep (hit recursion limit)");
278 335
279 encode_ch (enc, '['); encode_nl (enc); 336 encode_ch (enc, '[');
280 ++enc->indent; 337
338 if (len >= 0)
339 {
340 encode_nl (enc); ++enc->indent;
281 341
282 for (i = 0; i <= len; ++i) 342 for (i = 0; i <= len; ++i)
283 { 343 {
344 SV **svp = av_fetch (av, i, 0);
345
284 encode_indent (enc); 346 encode_indent (enc);
285 encode_sv (enc, *av_fetch (av, i, 0));
286 347
348 if (svp)
349 encode_sv (enc, *svp);
350 else
351 encode_str (enc, "null", 4, 0);
352
287 if (i < len) 353 if (i < len)
288 encode_comma (enc); 354 encode_comma (enc);
289 } 355 }
290 356
357 encode_nl (enc); --enc->indent; encode_indent (enc);
358 }
359
291 encode_nl (enc); 360 encode_ch (enc, ']');
292
293 --enc->indent;
294 encode_indent (enc); encode_ch (enc, ']');
295} 361}
296 362
297static void 363static void
298encode_he (enc_t *enc, HE *he) 364encode_hk (enc_t *enc, HE *he)
299{ 365{
300 encode_ch (enc, '"'); 366 encode_ch (enc, '"');
301 367
302 if (HeKLEN (he) == HEf_SVKEY) 368 if (HeKLEN (he) == HEf_SVKEY)
303 { 369 {
… …
313 else 379 else
314 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 380 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
315 381
316 encode_ch (enc, '"'); 382 encode_ch (enc, '"');
317 383
318 if (enc->flags & F_SPACE_BEFORE) encode_space (enc); 384 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
319 encode_ch (enc, ':'); 385 encode_ch (enc, ':');
320 if (enc->flags & F_SPACE_AFTER ) encode_space (enc); 386 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
321 encode_sv (enc, HeVAL (he));
322} 387}
323 388
324// compare hash entries, used when all keys are bytestrings 389// compare hash entries, used when all keys are bytestrings
325static int 390static int
326he_cmp_fast (const void *a_, const void *b_) 391he_cmp_fast (const void *a_, const void *b_)
… …
331 HE *b = *(HE **)b_; 396 HE *b = *(HE **)b_;
332 397
333 STRLEN la = HeKLEN (a); 398 STRLEN la = HeKLEN (a);
334 STRLEN lb = HeKLEN (b); 399 STRLEN lb = HeKLEN (b);
335 400
336 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 401 if (!(cmp = memcmp (HeKEY (b), HeKEY (a), lb < la ? lb : la)))
337 cmp = la - lb; 402 cmp = lb - la;
338 403
339 return cmp; 404 return cmp;
340} 405}
341 406
342// compare hash entries, used when some keys are sv's or utf-x 407// compare hash entries, used when some keys are sv's or utf-x
343static int 408static int
344he_cmp_slow (const void *a, const void *b) 409he_cmp_slow (const void *a, const void *b)
345{ 410{
346 return sv_cmp (HeSVKEY_force (*(HE **)a), HeSVKEY_force (*(HE **)b)); 411 return sv_cmp (HeSVKEY_force (*(HE **)b), HeSVKEY_force (*(HE **)a));
347} 412}
348 413
349static void 414static void
350encode_hv (enc_t *enc, HV *hv) 415encode_hv (enc_t *enc, HV *hv)
351{ 416{
417 HE *he;
352 int count, i; 418 int count;
353 419
354 if (enc->indent >= enc->maxdepth) 420 if (enc->indent >= enc->maxdepth)
355 croak ("data structure too deep (hit recursion limit)"); 421 croak ("data structure too deep (hit recursion limit)");
356 422
357 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 423 encode_ch (enc, '{');
358 424
359 if ((count = hv_iterinit (hv)))
360 {
361 // for canonical output we have to sort by keys first 425 // for canonical output we have to sort by keys first
362 // actually, this is mostly due to the stupid so-called 426 // actually, this is mostly due to the stupid so-called
363 // security workaround added somewhere in 5.8.x. 427 // security workaround added somewhere in 5.8.x.
364 // that randomises hash orderings 428 // that randomises hash orderings
365 if (enc->flags & F_CANONICAL) 429 if (enc->json.flags & F_CANONICAL)
430 {
431 int count = hv_iterinit (hv);
432
433 if (SvMAGICAL (hv))
366 { 434 {
435 // need to count by iterating. could improve by dynamically building the vector below
436 // but I don't care for the speed of this special case.
437 // note also that we will run into undefined behaviour when the two iterations
438 // do not result in the same count, something I might care for in some later release.
439
440 count = 0;
441 while (hv_iternext (hv))
442 ++count;
443
444 hv_iterinit (hv);
445 }
446
447 if (count)
448 {
449 int i, fast = 1;
450#if defined(__BORLANDC__) || defined(_MSC_VER)
451 HE **hes = _alloca (count * sizeof (HE));
452#else
367 HE *he, *hes [count]; // if your compiler dies here, you need to enable C99 mode 453 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
368 int fast = 1; 454#endif
369 455
370 i = 0; 456 i = 0;
371 while ((he = hv_iternext (hv))) 457 while ((he = hv_iternext (hv)))
372 { 458 {
373 hes [i++] = he; 459 hes [i++] = he;
… …
395 481
396 FREETMPS; 482 FREETMPS;
397 LEAVE; 483 LEAVE;
398 } 484 }
399 485
400 for (i = 0; i < count; ++i) 486 encode_nl (enc); ++enc->indent;
487
488 while (count--)
401 { 489 {
402 encode_indent (enc); 490 encode_indent (enc);
491 he = hes [count];
403 encode_he (enc, hes [i]); 492 encode_hk (enc, he);
493 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
404 494
405 if (i < count - 1) 495 if (count)
406 encode_comma (enc); 496 encode_comma (enc);
407 } 497 }
408 498
499 encode_nl (enc); --enc->indent; encode_indent (enc);
500 }
501 }
502 else
503 {
504 if (hv_iterinit (hv) || SvMAGICAL (hv))
505 if ((he = hv_iternext (hv)))
506 {
507 encode_nl (enc); ++enc->indent;
508
509 for (;;)
510 {
409 encode_nl (enc); 511 encode_indent (enc);
512 encode_hk (enc, he);
513 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
514
515 if (!(he = hv_iternext (hv)))
516 break;
517
518 encode_comma (enc);
519 }
520
521 encode_nl (enc); --enc->indent; encode_indent (enc);
522 }
523 }
524
525 encode_ch (enc, '}');
526}
527
528// encode objects, arrays and special \0=false and \1=true values.
529static void
530encode_rv (enc_t *enc, SV *sv)
531{
532 svtype svt;
533
534 SvGETMAGIC (sv);
535 svt = SvTYPE (sv);
536
537 if (expect_false (SvOBJECT (sv)))
538 {
539 HV *stash = !JSON_SLOW || json_boolean_stash
540 ? json_boolean_stash
541 : gv_stashpv ("JSON::XS::Boolean", 1);
542
543 if (SvSTASH (sv) == stash)
544 {
545 if (SvIV (sv))
546 encode_str (enc, "true", 4, 0);
547 else
548 encode_str (enc, "false", 5, 0);
410 } 549 }
411 else 550 else
412 { 551 {
413 SV *sv; 552#if 0
414 HE *he = hv_iternext (hv); 553 if (0 && sv_derived_from (rv, "JSON::Literal"))
415
416 for (;;)
417 { 554 {
418 encode_indent (enc); 555 // not yet
419 encode_he (enc, he);
420
421 if (!(he = hv_iternext (hv)))
422 break;
423
424 encode_comma (enc);
425 } 556 }
557#endif
558 if (enc->json.flags & F_CONV_BLESSED)
559 {
560 // we re-bless the reference to get overload and other niceties right
561 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
426 562
563 if (to_json)
564 {
565 dSP;
566
567 ENTER; SAVETMPS; PUSHMARK (SP);
568 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
569
570 // calling with G_SCALAR ensures that we always get a 1 return value
571 PUTBACK;
572 call_sv ((SV *)GvCV (to_json), G_SCALAR);
573 SPAGAIN;
574
575 // catch this surprisingly common error
576 if (SvROK (TOPs) && SvRV (TOPs) == sv)
577 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
578
579 sv = POPs;
580 PUTBACK;
581
427 encode_nl (enc); 582 encode_sv (enc, sv);
583
584 FREETMPS; LEAVE;
585 }
586 else if (enc->json.flags & F_ALLOW_BLESSED)
587 encode_str (enc, "null", 4, 0);
588 else
589 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
590 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
591 }
592 else if (enc->json.flags & F_ALLOW_BLESSED)
593 encode_str (enc, "null", 4, 0);
594 else
595 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",
596 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
428 } 597 }
429 } 598 }
430
431 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
432}
433
434// encode objects, arrays and special \0=false and \1=true values.
435static void
436encode_rv (enc_t *enc, SV *sv)
437{
438 SvGETMAGIC (sv);
439
440 svtype svt = SvTYPE (sv);
441
442 if (svt == SVt_PVHV) 599 else if (svt == SVt_PVHV)
443 encode_hv (enc, (HV *)sv); 600 encode_hv (enc, (HV *)sv);
444 else if (svt == SVt_PVAV) 601 else if (svt == SVt_PVAV)
445 encode_av (enc, (AV *)sv); 602 encode_av (enc, (AV *)sv);
446 else if (svt < SVt_PVAV) 603 else if (svt < SVt_PVAV)
447 { 604 {
448 if (SvNIOK (sv) && SvIV (sv) == 0) 605 STRLEN len = 0;
606 char *pv = svt ? SvPV (sv, len) : 0;
607
608 if (len == 1 && *pv == '1')
609 encode_str (enc, "true", 4, 0);
610 else if (len == 1 && *pv == '0')
449 encode_str (enc, "false", 5, 0); 611 encode_str (enc, "false", 5, 0);
450 else if (SvNIOK (sv) && SvIV (sv) == 1)
451 encode_str (enc, "true", 4, 0);
452 else 612 else
453 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", 613 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
454 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 614 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
455 } 615 }
456 else 616 else
… …
471 encode_str (enc, str, len, SvUTF8 (sv)); 631 encode_str (enc, str, len, SvUTF8 (sv));
472 encode_ch (enc, '"'); 632 encode_ch (enc, '"');
473 } 633 }
474 else if (SvNOKp (sv)) 634 else if (SvNOKp (sv))
475 { 635 {
636 // trust that perl will do the right thing w.r.t. JSON syntax.
476 need (enc, NV_DIG + 32); 637 need (enc, NV_DIG + 32);
477 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 638 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
478 enc->cur += strlen (enc->cur); 639 enc->cur += strlen (enc->cur);
479 } 640 }
480 else if (SvIOKp (sv)) 641 else if (SvIOKp (sv))
481 { 642 {
482 need (enc, 64); 643 // we assume we can always read an IV as a UV
644 if (SvUV (sv) & ~(UV)0x7fff)
645 {
646 // large integer, use the (rather slow) snprintf way.
647 need (enc, sizeof (UV) * 3);
483 enc->cur += 648 enc->cur +=
484 SvIsUV(sv) 649 SvIsUV(sv)
485 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 650 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
486 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 651 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
652 }
653 else
654 {
655 // optimise the "small number case"
656 // code will likely be branchless and use only a single multiplication
657 I32 i = SvIV (sv);
658 U32 u;
659 char digit, nz = 0;
660
661 need (enc, 6);
662
663 *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
664 u = i < 0 ? -i : i;
665
666 // convert to 4.28 fixed-point representation
667 u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
668
669 // now output digit by digit, each time masking out the integer part
670 // and multiplying by 5 while moving the decimal point one to the right,
671 // resulting in a net multiplication by 10.
672 // we always write the digit to memory but conditionally increment
673 // the pointer, to ease the usage of conditional move instructions.
674 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5;
675 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5;
676 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5;
677 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5;
678 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
679 }
487 } 680 }
488 else if (SvROK (sv)) 681 else if (SvROK (sv))
489 encode_rv (enc, SvRV (sv)); 682 encode_rv (enc, SvRV (sv));
490 else if (!SvOK (sv)) 683 else if (!SvOK (sv))
491 encode_str (enc, "null", 4, 0); 684 encode_str (enc, "null", 4, 0);
… …
493 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 686 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
494 SvPV_nolen (sv), SvFLAGS (sv)); 687 SvPV_nolen (sv), SvFLAGS (sv));
495} 688}
496 689
497static SV * 690static SV *
498encode_json (SV *scalar, U32 flags) 691encode_json (SV *scalar, JSON *json)
499{ 692{
693 enc_t enc;
694
500 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 695 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
501 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 696 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
502 697
503 enc_t enc; 698 enc.json = *json;
504 enc.flags = flags;
505 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 699 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
506 enc.cur = SvPVX (enc.sv); 700 enc.cur = SvPVX (enc.sv);
507 enc.end = SvEND (enc.sv); 701 enc.end = SvEND (enc.sv);
508 enc.indent = 0; 702 enc.indent = 0;
509 enc.maxdepth = DEC_DEPTH (flags); 703 enc.maxdepth = DEC_DEPTH (enc.json.flags);
704 enc.limit = enc.json.flags & F_ASCII ? 0x000080UL
705 : enc.json.flags & F_LATIN1 ? 0x000100UL
706 : 0x10FFFFUL;
510 707
511 SvPOK_only (enc.sv); 708 SvPOK_only (enc.sv);
512 encode_sv (&enc, scalar); 709 encode_sv (&enc, scalar);
513 710
711 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
712 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
713
514 if (!(flags & (F_ASCII | F_UTF8))) 714 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
515 SvUTF8_on (enc.sv); 715 SvUTF8_on (enc.sv);
516 716
517 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
518
519 if (enc.flags & F_SHRINK) 717 if (enc.json.flags & F_SHRINK)
520 shrink (enc.sv); 718 shrink (enc.sv);
521 719
522 return enc.sv; 720 return enc.sv;
523} 721}
524 722
… …
529typedef struct 727typedef struct
530{ 728{
531 char *cur; // current parser pointer 729 char *cur; // current parser pointer
532 char *end; // end of input string 730 char *end; // end of input string
533 const char *err; // parse error, if != 0 731 const char *err; // parse error, if != 0
534 U32 flags; // F_* 732 JSON json;
535 U32 depth; // recursion depth 733 U32 depth; // recursion depth
536 U32 maxdepth; // recursion depth limit 734 U32 maxdepth; // recursion depth limit
537} dec_t; 735} dec_t;
538 736
539static void 737INLINE void
738decode_comment (dec_t *dec)
739{
740 // only '#'-style comments allowed a.t.m.
741
742 while (*dec->cur && *dec->cur != 0x0a && *dec->cur != 0x0d)
743 ++dec->cur;
744}
745
746INLINE void
540decode_ws (dec_t *dec) 747decode_ws (dec_t *dec)
541{ 748{
542 for (;;) 749 for (;;)
543 { 750 {
544 char ch = *dec->cur; 751 char ch = *dec->cur;
545 752
546 if (ch > 0x20 753 if (ch > 0x20)
754 {
755 if (expect_false (ch == '#'))
756 {
757 if (dec->json.flags & F_RELAXED)
758 decode_comment (dec);
759 else
760 break;
761 }
762 else
763 break;
764 }
547 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) 765 else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)
548 break; 766 break; // parse error, but let higher level handle it, gives better error messages
549 767
550 ++dec->cur; 768 ++dec->cur;
551 } 769 }
552} 770}
553 771
… …
570decode_4hex (dec_t *dec) 788decode_4hex (dec_t *dec)
571{ 789{
572 signed char d1, d2, d3, d4; 790 signed char d1, d2, d3, d4;
573 unsigned char *cur = (unsigned char *)dec->cur; 791 unsigned char *cur = (unsigned char *)dec->cur;
574 792
575 d1 = decode_hexdigit [cur [0]]; if (d1 < 0) ERR ("four hexadecimal digits expected"); 793 d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected");
576 d2 = decode_hexdigit [cur [1]]; if (d2 < 0) ERR ("four hexadecimal digits expected"); 794 d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected");
577 d3 = decode_hexdigit [cur [2]]; if (d3 < 0) ERR ("four hexadecimal digits expected"); 795 d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected");
578 d4 = decode_hexdigit [cur [3]]; if (d4 < 0) ERR ("four hexadecimal digits expected"); 796 d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected");
579 797
580 dec->cur += 4; 798 dec->cur += 4;
581 799
582 return ((UV)d1) << 12 800 return ((UV)d1) << 12
583 | ((UV)d2) << 8 801 | ((UV)d2) << 8
… …
591static SV * 809static SV *
592decode_str (dec_t *dec) 810decode_str (dec_t *dec)
593{ 811{
594 SV *sv = 0; 812 SV *sv = 0;
595 int utf8 = 0; 813 int utf8 = 0;
814 char *dec_cur = dec->cur;
596 815
597 do 816 do
598 { 817 {
599 char buf [SHORT_STRING_LEN + UTF8_MAX_LEN]; 818 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
600 char *cur = buf; 819 char *cur = buf;
601 820
602 do 821 do
603 { 822 {
604 unsigned char ch = *(unsigned char *)dec->cur++; 823 unsigned char ch = *(unsigned char *)dec_cur++;
605 824
606 if (ch == '"') 825 if (expect_false (ch == '"'))
607 { 826 {
608 --dec->cur; 827 --dec_cur;
609 break; 828 break;
610 } 829 }
611 else if (ch == '\\') 830 else if (expect_false (ch == '\\'))
612 { 831 {
613 switch (*dec->cur) 832 switch (*dec_cur)
614 { 833 {
615 case '\\': 834 case '\\':
616 case '/': 835 case '/':
617 case '"': *cur++ = *dec->cur++; break; 836 case '"': *cur++ = *dec_cur++; break;
618 837
619 case 'b': ++dec->cur; *cur++ = '\010'; break; 838 case 'b': ++dec_cur; *cur++ = '\010'; break;
620 case 't': ++dec->cur; *cur++ = '\011'; break; 839 case 't': ++dec_cur; *cur++ = '\011'; break;
621 case 'n': ++dec->cur; *cur++ = '\012'; break; 840 case 'n': ++dec_cur; *cur++ = '\012'; break;
622 case 'f': ++dec->cur; *cur++ = '\014'; break; 841 case 'f': ++dec_cur; *cur++ = '\014'; break;
623 case 'r': ++dec->cur; *cur++ = '\015'; break; 842 case 'r': ++dec_cur; *cur++ = '\015'; break;
624 843
625 case 'u': 844 case 'u':
626 { 845 {
627 UV lo, hi; 846 UV lo, hi;
628 ++dec->cur; 847 ++dec_cur;
629 848
849 dec->cur = dec_cur;
630 hi = decode_4hex (dec); 850 hi = decode_4hex (dec);
851 dec_cur = dec->cur;
631 if (hi == (UV)-1) 852 if (hi == (UV)-1)
632 goto fail; 853 goto fail;
633 854
634 // possibly a surrogate pair 855 // possibly a surrogate pair
635 if (hi >= 0xd800) 856 if (hi >= 0xd800)
636 if (hi < 0xdc00) 857 if (hi < 0xdc00)
637 { 858 {
638 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 859 if (dec_cur [0] != '\\' || dec_cur [1] != 'u')
639 ERR ("missing low surrogate character in surrogate pair"); 860 ERR ("missing low surrogate character in surrogate pair");
640 861
641 dec->cur += 2; 862 dec_cur += 2;
642 863
864 dec->cur = dec_cur;
643 lo = decode_4hex (dec); 865 lo = decode_4hex (dec);
866 dec_cur = dec->cur;
644 if (lo == (UV)-1) 867 if (lo == (UV)-1)
645 goto fail; 868 goto fail;
646 869
647 if (lo < 0xdc00 || lo >= 0xe000) 870 if (lo < 0xdc00 || lo >= 0xe000)
648 ERR ("surrogate pair expected"); 871 ERR ("surrogate pair expected");
… …
654 877
655 if (hi >= 0x80) 878 if (hi >= 0x80)
656 { 879 {
657 utf8 = 1; 880 utf8 = 1;
658 881
659 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0); 882 cur = encode_utf8 (cur, hi);
660 } 883 }
661 else 884 else
662 *cur++ = hi; 885 *cur++ = hi;
663 } 886 }
664 break; 887 break;
665 888
666 default: 889 default:
667 --dec->cur; 890 --dec_cur;
668 ERR ("illegal backslash escape sequence in string"); 891 ERR ("illegal backslash escape sequence in string");
669 } 892 }
670 } 893 }
671 else if (ch >= 0x20 && ch <= 0x7f) 894 else if (expect_true (ch >= 0x20 && ch <= 0x7f))
672 *cur++ = ch; 895 *cur++ = ch;
673 else if (ch >= 0x80) 896 else if (ch >= 0x80)
674 { 897 {
675 --dec->cur;
676
677 STRLEN clen; 898 STRLEN clen;
899 UV uch;
900
901 --dec_cur;
902
678 UV uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen); 903 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
679 if (clen == (STRLEN)-1) 904 if (clen == (STRLEN)-1)
680 ERR ("malformed UTF-8 character in JSON string"); 905 ERR ("malformed UTF-8 character in JSON string");
681 906
682 do 907 do
683 *cur++ = *dec->cur++; 908 *cur++ = *dec_cur++;
684 while (--clen); 909 while (--clen);
685 910
686 utf8 = 1; 911 utf8 = 1;
687 } 912 }
688 else 913 else
689 { 914 {
690 --dec->cur; 915 --dec_cur;
691 916
692 if (!ch) 917 if (!ch)
693 ERR ("unexpected end of string while parsing json string"); 918 ERR ("unexpected end of string while parsing JSON string");
694 else 919 else
695 ERR ("invalid character encountered while parsing json string"); 920 ERR ("invalid character encountered while parsing JSON string");
696 } 921 }
697 } 922 }
698 while (cur < buf + SHORT_STRING_LEN); 923 while (cur < buf + SHORT_STRING_LEN);
699 924
925 {
700 STRLEN len = cur - buf; 926 STRLEN len = cur - buf;
701 927
702 if (sv) 928 if (sv)
703 { 929 {
704 SvGROW (sv, SvCUR (sv) + len + 1); 930 SvGROW (sv, SvCUR (sv) + len + 1);
705 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 931 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
706 SvCUR_set (sv, SvCUR (sv) + len); 932 SvCUR_set (sv, SvCUR (sv) + len);
707 } 933 }
708 else 934 else
709 sv = newSVpvn (buf, len); 935 sv = newSVpvn (buf, len);
710 } 936 }
937 }
711 while (*dec->cur != '"'); 938 while (*dec_cur != '"');
712 939
713 ++dec->cur; 940 ++dec_cur;
714 941
715 if (sv) 942 if (sv)
716 { 943 {
717 SvPOK_only (sv); 944 SvPOK_only (sv);
718 *SvEND (sv) = 0; 945 *SvEND (sv) = 0;
… …
721 SvUTF8_on (sv); 948 SvUTF8_on (sv);
722 } 949 }
723 else 950 else
724 sv = newSVpvn ("", 0); 951 sv = newSVpvn ("", 0);
725 952
953 dec->cur = dec_cur;
726 return sv; 954 return sv;
727 955
728fail: 956fail:
957 dec->cur = dec_cur;
729 return 0; 958 return 0;
730} 959}
731 960
732static SV * 961static SV *
733decode_num (dec_t *dec) 962decode_num (dec_t *dec)
… …
791 is_nv = 1; 1020 is_nv = 1;
792 } 1021 }
793 1022
794 if (!is_nv) 1023 if (!is_nv)
795 { 1024 {
796 UV uv; 1025 int len = dec->cur - start;
797 int numtype = grok_number (start, dec->cur - start, &uv); 1026
798 if (numtype & IS_NUMBER_IN_UV) 1027 // special case the rather common 1..5-digit-int case
799 if (numtype & IS_NUMBER_NEG) 1028 if (*start == '-')
1029 switch (len)
800 { 1030 {
801 if (uv < (UV)IV_MIN) 1031 case 2: return newSViv (-( start [1] - '0' * 1));
802 return newSViv (-(IV)uv); 1032 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
1033 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
1034 case 5: return newSViv (-( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
1035 case 6: return newSViv (-(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
803 } 1036 }
1037 else
1038 switch (len)
1039 {
1040 case 1: return newSViv ( start [0] - '0' * 1);
1041 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
1042 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
1043 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
1044 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
1045 }
1046
1047 {
1048 UV uv;
1049 int numtype = grok_number (start, len, &uv);
1050 if (numtype & IS_NUMBER_IN_UV)
1051 if (numtype & IS_NUMBER_NEG)
1052 {
1053 if (uv < (UV)IV_MIN)
1054 return newSViv (-(IV)uv);
1055 }
804 else 1056 else
805 return newSVuv (uv); 1057 return newSVuv (uv);
806 } 1058 }
807 1059
1060 len -= *start == '-' ? 1 : 0;
1061
1062 // does not fit into IV or UV, try NV
1063 if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len)
1064 #if defined (LDBL_DIG)
1065 || (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len)
1066 #endif
1067 )
1068 // fits into NV without loss of precision
1069 return newSVnv (Atof (start));
1070
1071 // everything else fails, convert it to a string
1072 return newSVpvn (start, dec->cur - start);
1073 }
1074
1075 // loss of precision here
808 return newSVnv (Atof (start)); 1076 return newSVnv (Atof (start));
809 1077
810fail: 1078fail:
811 return 0; 1079 return 0;
812} 1080}
… …
842 1110
843 if (*dec->cur != ',') 1111 if (*dec->cur != ',')
844 ERR (", or ] expected while parsing array"); 1112 ERR (", or ] expected while parsing array");
845 1113
846 ++dec->cur; 1114 ++dec->cur;
1115
1116 decode_ws (dec);
1117
1118 if (*dec->cur == ']' && dec->json.flags & F_RELAXED)
1119 {
1120 ++dec->cur;
1121 break;
1122 }
847 } 1123 }
848 1124
849 DEC_DEC_DEPTH; 1125 DEC_DEC_DEPTH;
850 return newRV_noinc ((SV *)av); 1126 return newRV_noinc ((SV *)av);
851 1127
… …
856} 1132}
857 1133
858static SV * 1134static SV *
859decode_hv (dec_t *dec) 1135decode_hv (dec_t *dec)
860{ 1136{
1137 SV *sv;
861 HV *hv = newHV (); 1138 HV *hv = newHV ();
862 1139
863 DEC_INC_DEPTH; 1140 DEC_INC_DEPTH;
864 decode_ws (dec); 1141 decode_ws (dec);
865 1142
866 if (*dec->cur == '}') 1143 if (*dec->cur == '}')
867 ++dec->cur; 1144 ++dec->cur;
868 else 1145 else
869 for (;;) 1146 for (;;)
870 { 1147 {
1148 EXPECT_CH ('"');
1149
1150 // heuristic: assume that
1151 // a) decode_str + hv_store_ent are abysmally slow.
1152 // b) most hash keys are short, simple ascii text.
1153 // => try to "fast-match" such strings to avoid
1154 // the overhead of decode_str + hv_store_ent.
1155 {
871 SV *key, *value; 1156 SV *value;
1157 char *p = dec->cur;
1158 char *e = p + 24; // only try up to 24 bytes
872 1159
873 decode_ws (dec); EXPECT_CH ('"'); 1160 for (;;)
874
875 key = decode_str (dec);
876 if (!key)
877 goto fail;
878
879 decode_ws (dec); EXPECT_CH (':');
880
881 value = decode_sv (dec);
882 if (!value)
883 { 1161 {
1162 // the >= 0x80 is true on most architectures
1163 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1164 {
1165 // slow path, back up and use decode_str
1166 SV *key = decode_str (dec);
1167 if (!key)
1168 goto fail;
1169
1170 decode_ws (dec); EXPECT_CH (':');
1171
1172 decode_ws (dec);
1173 value = decode_sv (dec);
1174 if (!value)
1175 {
1176 SvREFCNT_dec (key);
1177 goto fail;
1178 }
1179
1180 hv_store_ent (hv, key, value, 0);
884 SvREFCNT_dec (key); 1181 SvREFCNT_dec (key);
1182
1183 break;
1184 }
1185 else if (*p == '"')
1186 {
1187 // fast path, got a simple key
1188 char *key = dec->cur;
1189 int len = p - key;
1190 dec->cur = p + 1;
1191
1192 decode_ws (dec); EXPECT_CH (':');
1193
1194 decode_ws (dec);
1195 value = decode_sv (dec);
1196 if (!value)
885 goto fail; 1197 goto fail;
1198
1199 hv_store (hv, key, len, value, 0);
1200
1201 break;
1202 }
1203
1204 ++p;
886 } 1205 }
887 1206 }
888 hv_store_ent (hv, key, value, 0);
889 SvREFCNT_dec (key);
890 1207
891 decode_ws (dec); 1208 decode_ws (dec);
892 1209
893 if (*dec->cur == '}') 1210 if (*dec->cur == '}')
894 { 1211 {
… …
898 1215
899 if (*dec->cur != ',') 1216 if (*dec->cur != ',')
900 ERR (", or } expected while parsing object/hash"); 1217 ERR (", or } expected while parsing object/hash");
901 1218
902 ++dec->cur; 1219 ++dec->cur;
1220
1221 decode_ws (dec);
1222
1223 if (*dec->cur == '}' && dec->json.flags & F_RELAXED)
1224 {
1225 ++dec->cur;
1226 break;
1227 }
903 } 1228 }
904 1229
905 DEC_DEC_DEPTH; 1230 DEC_DEC_DEPTH;
906 return newRV_noinc ((SV *)hv); 1231 sv = newRV_noinc ((SV *)hv);
1232
1233 // check filter callbacks
1234 if (dec->json.flags & F_HOOK)
1235 {
1236 if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
1237 {
1238 HE *cb, *he;
1239
1240 hv_iterinit (hv);
1241 he = hv_iternext (hv);
1242 hv_iterinit (hv);
1243
1244 // the next line creates a mortal sv each time its called.
1245 // might want to optimise this for common cases.
1246 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1247
1248 if (cb)
1249 {
1250 dSP;
1251 int count;
1252
1253 ENTER; SAVETMPS; PUSHMARK (SP);
1254 XPUSHs (HeVAL (he));
1255
1256 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1257
1258 if (count == 1)
1259 {
1260 sv = newSVsv (POPs);
1261 FREETMPS; LEAVE;
1262 return sv;
1263 }
1264
1265 FREETMPS; LEAVE;
1266 }
1267 }
1268
1269 if (dec->json.cb_object)
1270 {
1271 dSP;
1272 int count;
1273
1274 ENTER; SAVETMPS; PUSHMARK (SP);
1275 XPUSHs (sv_2mortal (sv));
1276
1277 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1278
1279 if (count == 1)
1280 {
1281 sv = newSVsv (POPs);
1282 FREETMPS; LEAVE;
1283 return sv;
1284 }
1285
1286 SvREFCNT_inc (sv);
1287 FREETMPS; LEAVE;
1288 }
1289 }
1290
1291 return sv;
907 1292
908fail: 1293fail:
909 SvREFCNT_dec (hv); 1294 SvREFCNT_dec (hv);
910 DEC_DEC_DEPTH; 1295 DEC_DEC_DEPTH;
911 return 0; 1296 return 0;
912} 1297}
913 1298
914static SV * 1299static SV *
915decode_sv (dec_t *dec) 1300decode_sv (dec_t *dec)
916{ 1301{
917 decode_ws (dec); 1302 // the beauty of JSON: you need exactly one character lookahead
1303 // to parse anything.
918 switch (*dec->cur) 1304 switch (*dec->cur)
919 { 1305 {
920 case '"': ++dec->cur; return decode_str (dec); 1306 case '"': ++dec->cur; return decode_str (dec);
921 case '[': ++dec->cur; return decode_av (dec); 1307 case '[': ++dec->cur; return decode_av (dec);
922 case '{': ++dec->cur; return decode_hv (dec); 1308 case '{': ++dec->cur; return decode_hv (dec);
… …
928 1314
929 case 't': 1315 case 't':
930 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1316 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
931 { 1317 {
932 dec->cur += 4; 1318 dec->cur += 4;
933 return newSViv (1); 1319#if JSON_SLOW
1320 json_true = get_sv ("JSON::XS::true", 1); SvREADONLY_on (json_true);
1321#endif
1322 return SvREFCNT_inc (json_true);
934 } 1323 }
935 else 1324 else
936 ERR ("'true' expected"); 1325 ERR ("'true' expected");
937 1326
938 break; 1327 break;
939 1328
940 case 'f': 1329 case 'f':
941 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1330 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
942 { 1331 {
943 dec->cur += 5; 1332 dec->cur += 5;
944 return newSViv (0); 1333#if JSON_SLOW
1334 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
1335#endif
1336 return SvREFCNT_inc (json_false);
945 } 1337 }
946 else 1338 else
947 ERR ("'false' expected"); 1339 ERR ("'false' expected");
948 1340
949 break; 1341 break;
… …
958 ERR ("'null' expected"); 1350 ERR ("'null' expected");
959 1351
960 break; 1352 break;
961 1353
962 default: 1354 default:
963 ERR ("malformed json string, neither array, object, number, string or atom"); 1355 ERR ("malformed JSON string, neither array, object, number, string or atom");
964 break; 1356 break;
965 } 1357 }
966 1358
967fail: 1359fail:
968 return 0; 1360 return 0;
969} 1361}
970 1362
971static SV * 1363static SV *
972decode_json (SV *string, U32 flags) 1364decode_json (SV *string, JSON *json, UV *offset_return)
973{ 1365{
1366 dec_t dec;
1367 UV offset;
974 SV *sv; 1368 SV *sv;
975 1369
1370 SvGETMAGIC (string);
976 SvUPGRADE (string, SVt_PV); 1371 SvUPGRADE (string, SVt_PV);
977 1372
1373 if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags))
1374 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1375 (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags));
1376
978 if (flags & F_UTF8) 1377 if (json->flags & F_UTF8)
979 sv_utf8_downgrade (string, 0); 1378 sv_utf8_downgrade (string, 0);
980 else 1379 else
981 sv_utf8_upgrade (string); 1380 sv_utf8_upgrade (string);
982 1381
983 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1382 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
984 1383
985 dec_t dec; 1384 dec.json = *json;
986 dec.flags = flags;
987 dec.cur = SvPVX (string); 1385 dec.cur = SvPVX (string);
988 dec.end = SvEND (string); 1386 dec.end = SvEND (string);
989 dec.err = 0; 1387 dec.err = 0;
990 dec.depth = 0; 1388 dec.depth = 0;
991 dec.maxdepth = DEC_DEPTH (dec.flags); 1389 dec.maxdepth = DEC_DEPTH (dec.json.flags);
992 1390
1391 if (dec.json.cb_object || dec.json.cb_sk_object)
1392 dec.json.flags |= F_HOOK;
1393
993 *dec.end = 0; // this should basically be a nop, too, but make sure its there 1394 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1395
1396 decode_ws (&dec);
994 sv = decode_sv (&dec); 1397 sv = decode_sv (&dec);
995 1398
1399 if (!(offset_return || !sv))
1400 {
1401 // check for trailing garbage
1402 decode_ws (&dec);
1403
1404 if (*dec.cur)
1405 {
1406 dec.err = "garbage after JSON object";
1407 SvREFCNT_dec (sv);
1408 sv = 0;
1409 }
1410 }
1411
1412 if (offset_return || !sv)
1413 {
1414 offset = dec.json.flags & F_UTF8
1415 ? dec.cur - SvPVX (string)
1416 : utf8_distance (dec.cur, SvPVX (string));
1417
1418 if (offset_return)
1419 *offset_return = offset;
1420 }
1421
996 if (!sv) 1422 if (!sv)
997 { 1423 {
998 IV offset = dec.flags & F_UTF8
999 ? dec.cur - SvPVX (string)
1000 : utf8_distance (dec.cur, SvPVX (string));
1001 SV *uni = sv_newmortal (); 1424 SV *uni = sv_newmortal ();
1002 1425
1003 // horrible hack to silence warning inside pv_uni_display 1426 // horrible hack to silence warning inside pv_uni_display
1004 COP cop = *PL_curcop; 1427 COP cop = *PL_curcop;
1005 cop.cop_warnings = pWARN_NONE; 1428 cop.cop_warnings = pWARN_NONE;
… …
1015 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1438 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1016 } 1439 }
1017 1440
1018 sv = sv_2mortal (sv); 1441 sv = sv_2mortal (sv);
1019 1442
1020 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1443 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))
1021 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); 1444 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1022 1445
1023 return sv; 1446 return sv;
1024} 1447}
1025 1448
… …
1029MODULE = JSON::XS PACKAGE = JSON::XS 1452MODULE = JSON::XS PACKAGE = JSON::XS
1030 1453
1031BOOT: 1454BOOT:
1032{ 1455{
1033 int i; 1456 int i;
1034
1035 memset (decode_hexdigit, 0xff, 256);
1036 1457
1037 for (i = 0; i < 256; ++i) 1458 for (i = 0; i < 256; ++i)
1038 decode_hexdigit [i] = 1459 decode_hexdigit [i] =
1039 i >= '0' && i <= '9' ? i - '0' 1460 i >= '0' && i <= '9' ? i - '0'
1040 : i >= 'a' && i <= 'f' ? i - 'a' + 10 1461 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1041 : i >= 'A' && i <= 'F' ? i - 'A' + 10 1462 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1042 : -1; 1463 : -1;
1043 1464
1044 json_stash = gv_stashpv ("JSON::XS", 1); 1465 json_stash = gv_stashpv ("JSON::XS" , 1);
1466 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1467
1468 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true );
1469 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
1045} 1470}
1046 1471
1047PROTOTYPES: DISABLE 1472PROTOTYPES: DISABLE
1048 1473
1049SV *new (char *dummy) 1474void CLONE (...)
1050 CODE: 1475 CODE:
1051 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1476 json_stash = 0;
1477 json_boolean_stash = 0;
1478
1479void new (char *klass)
1480 PPCODE:
1481{
1482 SV *pv = NEWSV (0, sizeof (JSON));
1483 SvPOK_only (pv);
1484 Zero (SvPVX (pv), 1, JSON);
1485 ((JSON *)SvPVX (pv))->flags = F_DEFAULT;
1486 XPUSHs (sv_2mortal (sv_bless (
1487 newRV_noinc (pv),
1488 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1489 )));
1490}
1491
1492void ascii (JSON *self, int enable = 1)
1493 ALIAS:
1494 ascii = F_ASCII
1495 latin1 = F_LATIN1
1496 utf8 = F_UTF8
1497 indent = F_INDENT
1498 canonical = F_CANONICAL
1499 space_before = F_SPACE_BEFORE
1500 space_after = F_SPACE_AFTER
1501 pretty = F_PRETTY
1502 allow_nonref = F_ALLOW_NONREF
1503 shrink = F_SHRINK
1504 allow_blessed = F_ALLOW_BLESSED
1505 convert_blessed = F_CONV_BLESSED
1506 relaxed = F_RELAXED
1507 PPCODE:
1508{
1509 if (enable)
1510 self->flags |= ix;
1511 else
1512 self->flags &= ~ix;
1513
1514 XPUSHs (ST (0));
1515}
1516
1517void get_ascii (JSON *self)
1518 ALIAS:
1519 get_ascii = F_ASCII
1520 get_latin1 = F_LATIN1
1521 get_utf8 = F_UTF8
1522 get_indent = F_INDENT
1523 get_canonical = F_CANONICAL
1524 get_space_before = F_SPACE_BEFORE
1525 get_space_after = F_SPACE_AFTER
1526 get_allow_nonref = F_ALLOW_NONREF
1527 get_shrink = F_SHRINK
1528 get_allow_blessed = F_ALLOW_BLESSED
1529 get_convert_blessed = F_CONV_BLESSED
1530 get_relaxed = F_RELAXED
1531 PPCODE:
1532 XPUSHs (boolSV (self->flags & ix));
1533
1534void max_depth (JSON *self, UV max_depth = 0x80000000UL)
1535 PPCODE:
1536{
1537 UV log2 = 0;
1538
1539 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1540
1541 while ((1UL << log2) < max_depth)
1542 ++log2;
1543
1544 self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1545
1546 XPUSHs (ST (0));
1547}
1548
1549U32 get_max_depth (JSON *self)
1550 CODE:
1551 RETVAL = DEC_DEPTH (self->flags);
1052 OUTPUT: 1552 OUTPUT:
1053 RETVAL 1553 RETVAL
1054 1554
1055SV *ascii (SV *self, int enable = 1) 1555void max_size (JSON *self, UV max_size = 0)
1056 ALIAS: 1556 PPCODE:
1057 ascii = F_ASCII 1557{
1058 utf8 = F_UTF8 1558 UV log2 = 0;
1059 indent = F_INDENT 1559
1060 canonical = F_CANONICAL 1560 if (max_size > 0x80000000UL) max_size = 0x80000000UL;
1061 space_before = F_SPACE_BEFORE 1561 if (max_size == 1) max_size = 2;
1062 space_after = F_SPACE_AFTER 1562
1063 pretty = F_PRETTY 1563 while ((1UL << log2) < max_size)
1064 allow_nonref = F_ALLOW_NONREF 1564 ++log2;
1065 shrink = F_SHRINK 1565
1566 self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1567
1568 XPUSHs (ST (0));
1569}
1570
1571int get_max_size (JSON *self)
1066 CODE: 1572 CODE:
1067{ 1573 RETVAL = DEC_SIZE (self->flags);
1068 UV *uv = SvJSON (self);
1069 if (enable)
1070 *uv |= ix;
1071 else
1072 *uv &= ~ix;
1073
1074 RETVAL = newSVsv (self);
1075}
1076 OUTPUT: 1574 OUTPUT:
1077 RETVAL 1575 RETVAL
1078 1576
1079SV *max_depth (SV *self, int max_depth = 0x80000000UL) 1577void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1578 PPCODE:
1579{
1580 SvREFCNT_dec (self->cb_object);
1581 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
1582
1583 XPUSHs (ST (0));
1584}
1585
1586void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
1587 PPCODE:
1588{
1589 if (!self->cb_sk_object)
1590 self->cb_sk_object = newHV ();
1591
1592 if (SvOK (cb))
1593 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1594 else
1595 {
1596 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1597
1598 if (!HvKEYS (self->cb_sk_object))
1599 {
1600 SvREFCNT_dec (self->cb_sk_object);
1601 self->cb_sk_object = 0;
1602 }
1603 }
1604
1605 XPUSHs (ST (0));
1606}
1607
1608void encode (JSON *self, SV *scalar)
1609 PPCODE:
1610 XPUSHs (encode_json (scalar, self));
1611
1612void decode (JSON *self, SV *jsonstr)
1613 PPCODE:
1614 XPUSHs (decode_json (jsonstr, self, 0));
1615
1616void decode_prefix (JSON *self, SV *jsonstr)
1617 PPCODE:
1618{
1619 UV offset;
1620 EXTEND (SP, 2);
1621 PUSHs (decode_json (jsonstr, self, &offset));
1622 PUSHs (sv_2mortal (newSVuv (offset)));
1623}
1624
1625void DESTROY (JSON *self)
1080 CODE: 1626 CODE:
1081{ 1627 SvREFCNT_dec (self->cb_sk_object);
1082 UV *uv = SvJSON (self); 1628 SvREFCNT_dec (self->cb_object);
1083 UV log2 = 0;
1084 1629
1085 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL; 1630PROTOTYPES: ENABLE
1086 1631
1087 while ((1UL << log2) < max_depth)
1088 ++log2;
1089
1090 *uv = *uv & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1091
1092 RETVAL = newSVsv (self);
1093}
1094 OUTPUT:
1095 RETVAL
1096
1097void encode (SV *self, SV *scalar) 1632void encode_json (SV *scalar)
1098 PPCODE: 1633 PPCODE:
1634{
1635 JSON json = { F_DEFAULT | F_UTF8 };
1099 XPUSHs (encode_json (scalar, *SvJSON (self))); 1636 XPUSHs (encode_json (scalar, &json));
1637}
1100 1638
1101void decode (SV *self, SV *jsonstr) 1639void decode_json (SV *jsonstr)
1102 PPCODE: 1640 PPCODE:
1641{
1642 JSON json = { F_DEFAULT | F_UTF8 };
1103 XPUSHs (decode_json (jsonstr, *SvJSON (self))); 1643 XPUSHs (decode_json (jsonstr, &json, 0));
1644}
1104 1645
1105PROTOTYPES: ENABLE
1106
1107void to_json (SV *scalar)
1108 ALIAS:
1109 objToJson = 0
1110 PPCODE:
1111 XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8));
1112
1113void from_json (SV *jsonstr)
1114 ALIAS:
1115 jsonToObj = 0
1116 PPCODE:
1117 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8));
1118

Diff Legend

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