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 1.2 by root, Thu Mar 22 17:28:50 2007 UTC vs.
Revision 1.55 by root, Mon Jul 23 22:57:40 2007 UTC

… …
3#include "XSUB.h" 3#include "XSUB.h"
4 4
5#include "assert.h" 5#include "assert.h"
6#include "string.h" 6#include "string.h"
7#include "stdlib.h" 7#include "stdlib.h"
8#include "stdio.h"
8 9
10#if defined(__BORLANDC__) || defined(_MSC_VER)
11# define snprintf _snprintf // C compilers have this in stdio.h
12#endif
13
14// some old perls do not have this, try to make it work, no
15// guarentees, though. if it breaks, you get to keep the pieces.
16#ifndef UTF8_MAXBYTES
17# define UTF8_MAXBYTES 13
18#endif
19
9#define F_ASCII 0x00000001 20#define F_ASCII 0x00000001UL
21#define F_LATIN1 0x00000002UL
10#define F_UTF8 0x00000002 22#define F_UTF8 0x00000004UL
11#define F_INDENT 0x00000004 23#define F_INDENT 0x00000008UL
12#define F_CANONICAL 0x00000008 24#define F_CANONICAL 0x00000010UL
13#define F_SPACE_BEFORE 0x00000010 25#define F_SPACE_BEFORE 0x00000020UL
14#define F_SPACE_AFTER 0x00000020 26#define F_SPACE_AFTER 0x00000040UL
15#define F_JSON_RPC 0x00000040 27#define F_ALLOW_NONREF 0x00000100UL
28#define F_SHRINK 0x00000200UL
29#define F_ALLOW_BLESSED 0x00000400UL
30#define F_CONV_BLESSED 0x00000800UL
31#define F_MAXDEPTH 0xf8000000UL
32#define S_MAXDEPTH 27
33#define F_MAXSIZE 0x01f00000UL
34#define S_MAXSIZE 20
35#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
36
37#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
38#define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE ))
16 39
17#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 40#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
18#define F_DEFAULT 0 41#define F_DEFAULT (9UL << S_MAXDEPTH)
19 42
20#define INIT_SIZE 32 // initial scalar size to be allocated 43#define INIT_SIZE 32 // initial scalar size to be allocated
44#define INDENT_STEP 3 // spaces per indentation level
45
46#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
21 47
22#define SB do { 48#define SB do {
23#define SE } while (0) 49#define SE } while (0)
24 50
25static HV *json_stash; 51#if __GNUC__ >= 3
52# define expect(expr,value) __builtin_expect ((expr),(value))
53# define inline inline
54#else
55# define expect(expr,value) (expr)
56# define inline static
57#endif
58
59#define expect_false(expr) expect ((expr) != 0, 0)
60#define expect_true(expr) expect ((expr) != 0, 1)
61
62static HV *json_stash, *json_boolean_stash; // JSON::XS::
63static SV *json_true, *json_false;
64
65typedef struct {
66 U32 flags;
67 SV *cb_object;
68 HV *cb_sk_object;
69} JSON;
70
71/////////////////////////////////////////////////////////////////////////////
72// utility functions
73
74inline void
75shrink (SV *sv)
76{
77 sv_utf8_downgrade (sv, 1);
78 if (SvLEN (sv) > SvCUR (sv) + 1)
79 {
80#ifdef SvPV_shrink_to_cur
81 SvPV_shrink_to_cur (sv);
82#elif defined (SvPV_renew)
83 SvPV_renew (sv, SvCUR (sv) + 1);
84#endif
85 }
86}
87
88// decode an utf-8 character and return it, or (UV)-1 in
89// case of an error.
90// we special-case "safe" characters from U+80 .. U+7FF,
91// but use the very good perl function to parse anything else.
92// note that we never call this function for a ascii codepoints
93inline UV
94decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
95{
96 if (expect_false (s[0] > 0xdf || s[0] < 0xc2))
97 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
98 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)
99 {
100 *clen = 2;
101 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
102 }
103 else
104 {
105 *clen = (STRLEN)-1;
106 return (UV)-1;
107 }
108}
109
110/////////////////////////////////////////////////////////////////////////////
111// encoder
26 112
27// structure used for encoding JSON 113// structure used for encoding JSON
28typedef struct 114typedef struct
29{ 115{
30 char *cur; 116 char *cur; // SvPVX (sv) + current output position
31 STRLEN len; // SvLEN (sv)
32 char *end; // SvEND (sv) 117 char *end; // SvEND (sv)
33 SV *sv; 118 SV *sv; // result scalar
34 UV flags; 119 JSON json;
35 int max_recurse; 120 U32 indent; // indentation level
36 int indent; 121 U32 maxdepth; // max. indentation/recursion level
37} enc_t; 122} enc_t;
38 123
39// structure used for decoding JSON 124inline void
40typedef struct
41{
42 char *cur;
43 char *end;
44 char *err;
45 UV flags;
46} dec_t;
47
48static UV *
49SvJSON (SV *sv)
50{
51 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash))
52 croak ("object is not of type JSON::XS");
53
54 return &SvUVX (SvRV (sv));
55}
56
57/////////////////////////////////////////////////////////////////////////////
58
59static void
60need (enc_t *enc, STRLEN len) 125need (enc_t *enc, STRLEN len)
61{ 126{
62 if (enc->cur + len >= enc->end) 127 if (expect_false (enc->cur + len >= enc->end))
63 { 128 {
64 STRLEN cur = enc->cur - SvPVX (enc->sv); 129 STRLEN cur = enc->cur - SvPVX (enc->sv);
65 SvGROW (enc->sv, cur + len + 1); 130 SvGROW (enc->sv, cur + len + 1);
66 enc->cur = SvPVX (enc->sv) + cur; 131 enc->cur = SvPVX (enc->sv) + cur;
67 enc->end = SvEND (enc->sv); 132 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
68 } 133 }
69} 134}
70 135
71static void 136inline void
72encode_ch (enc_t *enc, char ch) 137encode_ch (enc_t *enc, char ch)
73{ 138{
74 need (enc, 1); 139 need (enc, 1);
75 *enc->cur++ = ch; 140 *enc->cur++ = ch;
76} 141}
… …
78static void 143static void
79encode_str (enc_t *enc, char *str, STRLEN len, int is_utf8) 144encode_str (enc_t *enc, char *str, STRLEN len, int is_utf8)
80{ 145{
81 char *end = str + len; 146 char *end = str + len;
82 147
148 need (enc, len);
149
83 while (str < end) 150 while (str < end)
84 { 151 {
85 unsigned char ch = *(unsigned char *)str; 152 unsigned char ch = *(unsigned char *)str;
153
86 if (ch >= 0x20 && ch < 0x80) // most common case 154 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
87 { 155 {
156 if (expect_false (ch == '"')) // but with slow exceptions
157 {
158 need (enc, len += 1);
159 *enc->cur++ = '\\';
160 *enc->cur++ = '"';
161 }
162 else if (expect_false (ch == '\\'))
163 {
164 need (enc, len += 1);
165 *enc->cur++ = '\\';
166 *enc->cur++ = '\\';
167 }
168 else
88 *enc->cur++ = ch; 169 *enc->cur++ = ch;
170
89 str++; 171 ++str;
90 } 172 }
91 else 173 else
92 { 174 {
93 STRLEN clen; 175 switch (ch)
94 UV uch;
95
96 if (is_utf8)
97 {
98 uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
99 if (clen < 0)
100 croak ("malformed UTF-8 character in string, cannot convert to JSON");
101 } 176 {
102 else 177 case '\010': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'b'; ++str; break;
103 { 178 case '\011': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 't'; ++str; break;
104 uch = ch; 179 case '\012': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'n'; ++str; break;
105 clen = 1; 180 case '\014': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'f'; ++str; break;
106 } 181 case '\015': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'r'; ++str; break;
107 182
108 need (enc, len += 6); 183 default:
109
110 if (uch < 0xa0 || enc->flags & F_ASCII)
111 {
112 if (uch > 0xFFFFUL)
113 { 184 {
185 STRLEN clen;
186 UV uch;
187
188 if (is_utf8)
189 {
190 uch = decode_utf8 (str, end - str, &clen);
191 if (clen == (STRLEN)-1)
192 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
193 }
194 else
195 {
196 uch = ch;
114 len += 6; 197 clen = 1;
198 }
199
200 if (uch > 0x10FFFFUL)
201 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
202
203 if (uch < 0x80 || enc->json.flags & F_ASCII || (enc->json.flags & F_LATIN1 && uch > 0xFF))
204 {
205 if (uch > 0xFFFFUL)
206 {
115 need (enc, len += 6); 207 need (enc, len += 11);
116 sprintf (enc->cur, "\\u%04x\\u%04x", 208 sprintf (enc->cur, "\\u%04x\\u%04x",
117 (uch - 0x10000) / 0x400 + 0xD800, 209 (int)((uch - 0x10000) / 0x400 + 0xD800),
118 (uch - 0x10000) % 0x400 + 0xDC00); 210 (int)((uch - 0x10000) % 0x400 + 0xDC00));
119 enc->cur += 12; 211 enc->cur += 12;
212 }
213 else
214 {
215 static char hexdigit [16] = "0123456789abcdef";
216 need (enc, len += 5);
217 *enc->cur++ = '\\';
218 *enc->cur++ = 'u';
219 *enc->cur++ = hexdigit [ uch >> 12 ];
220 *enc->cur++ = hexdigit [(uch >> 8) & 15];
221 *enc->cur++ = hexdigit [(uch >> 4) & 15];
222 *enc->cur++ = hexdigit [(uch >> 0) & 15];
223 }
224
225 str += clen;
226 }
227 else if (enc->json.flags & F_LATIN1)
228 {
229 *enc->cur++ = uch;
230 str += clen;
231 }
232 else if (is_utf8)
233 {
234 need (enc, len += clen);
235 do
236 {
237 *enc->cur++ = *str++;
238 }
239 while (--clen);
240 }
241 else
242 {
243 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
244 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
245 ++str;
246 }
120 } 247 }
121 else
122 {
123 sprintf (enc->cur, "\\u%04x", uch);
124 enc->cur += 6;
125 }
126 }
127 else if (is_utf8)
128 { 248 }
129 memcpy (enc->cur, str, clen);
130 enc->cur += clen;
131 }
132 else
133 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
134
135 str += clen;
136 } 249 }
137 250
138 --len; 251 --len;
139 } 252 }
140} 253}
141 254
142#define INDENT SB \ 255inline void
256encode_indent (enc_t *enc)
257{
143 if (enc->flags & F_INDENT) \ 258 if (enc->json.flags & F_INDENT)
144 { \ 259 {
145 int i_; \ 260 int spaces = enc->indent * INDENT_STEP;
146 need (enc, enc->indent); \ 261
147 for (i_ = enc->indent * 3; i_--; )\ 262 need (enc, spaces);
263 memset (enc->cur, ' ', spaces);
264 enc->cur += spaces;
265 }
266}
267
268inline void
269encode_space (enc_t *enc)
270{
271 need (enc, 1);
272 encode_ch (enc, ' ');
273}
274
275inline void
276encode_nl (enc_t *enc)
277{
278 if (enc->json.flags & F_INDENT)
279 {
280 need (enc, 1);
148 encode_ch (enc, ' '); \ 281 encode_ch (enc, '\n');
149 } \ 282 }
150 SE 283}
151 284
152#define SPACE SB need (enc, 1); encode_ch (enc, ' '); SE 285inline void
153#define NL SB if (enc->flags & F_INDENT) { need (enc, 1); encode_ch (enc, '\n'); } SE 286encode_comma (enc_t *enc)
154#define COMMA SB \ 287{
155 encode_ch (enc, ','); \ 288 encode_ch (enc, ',');
289
156 if (enc->flags & F_INDENT) \ 290 if (enc->json.flags & F_INDENT)
157 NL; \ 291 encode_nl (enc);
158 else if (enc->flags & F_SPACE_AFTER) \ 292 else if (enc->json.flags & F_SPACE_AFTER)
159 SPACE; \ 293 encode_space (enc);
160 SE 294}
161 295
162static void encode_sv (enc_t *enc, SV *sv); 296static void encode_sv (enc_t *enc, SV *sv);
163 297
164static void 298static void
165encode_av (enc_t *enc, AV *av) 299encode_av (enc_t *enc, AV *av)
166{ 300{
167 int i, len = av_len (av); 301 int i, len = av_len (av);
168 302
169 encode_ch (enc, '['); NL; 303 if (enc->indent >= enc->maxdepth)
304 croak ("data structure too deep (hit recursion limit)");
305
306 encode_ch (enc, '['); encode_nl (enc);
170 ++enc->indent; 307 ++enc->indent;
171 308
172 for (i = 0; i <= len; ++i) 309 for (i = 0; i <= len; ++i)
173 { 310 {
174 INDENT; 311 SV **svp = av_fetch (av, i, 0);
175 encode_sv (enc, *av_fetch (av, i, 0)); 312
313 encode_indent (enc);
314
315 if (svp)
316 encode_sv (enc, *svp);
317 else
318 encode_str (enc, "null", 4, 0);
176 319
177 if (i < len) 320 if (i < len)
178 COMMA; 321 encode_comma (enc);
179 } 322 }
180 323
181 NL; 324 encode_nl (enc);
182 325
183 --enc->indent; 326 --enc->indent;
184 INDENT; encode_ch (enc, ']'); 327 encode_indent (enc); encode_ch (enc, ']');
185} 328}
186 329
187static void 330static void
188encode_he (enc_t *enc, HE *he) 331encode_he (enc_t *enc, HE *he)
189{ 332{
… …
191 334
192 if (HeKLEN (he) == HEf_SVKEY) 335 if (HeKLEN (he) == HEf_SVKEY)
193 { 336 {
194 SV *sv = HeSVKEY (he); 337 SV *sv = HeSVKEY (he);
195 STRLEN len; 338 STRLEN len;
339 char *str;
340
341 SvGETMAGIC (sv);
196 char *str = SvPV (sv, len); 342 str = SvPV (sv, len);
197 343
198 encode_str (enc, str, len, SvUTF8 (sv)); 344 encode_str (enc, str, len, SvUTF8 (sv));
199 } 345 }
200 else 346 else
201 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 347 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
202 348
203 encode_ch (enc, '"'); 349 encode_ch (enc, '"');
204 350
205 if (enc->flags & F_SPACE_BEFORE) SPACE; 351 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
206 encode_ch (enc, ':'); 352 encode_ch (enc, ':');
207 if (enc->flags & F_SPACE_AFTER ) SPACE; 353 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
208 encode_sv (enc, HeVAL (he)); 354 encode_sv (enc, HeVAL (he));
209} 355}
210 356
211// compare hash entries, used when all keys are bytestrings 357// compare hash entries, used when all keys are bytestrings
212static int 358static int
… …
218 HE *b = *(HE **)b_; 364 HE *b = *(HE **)b_;
219 365
220 STRLEN la = HeKLEN (a); 366 STRLEN la = HeKLEN (a);
221 STRLEN lb = HeKLEN (b); 367 STRLEN lb = HeKLEN (b);
222 368
223 if (!(cmp == memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 369 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb)))
224 cmp = la < lb ? -1 : la == lb ? 0 : 1; 370 cmp = la - lb;
225 371
226 return cmp; 372 return cmp;
227} 373}
228 374
229// compare hash entries, used when some keys are sv's or utf-x 375// compare hash entries, used when some keys are sv's or utf-x
… …
236static void 382static void
237encode_hv (enc_t *enc, HV *hv) 383encode_hv (enc_t *enc, HV *hv)
238{ 384{
239 int count, i; 385 int count, i;
240 386
387 if (enc->indent >= enc->maxdepth)
388 croak ("data structure too deep (hit recursion limit)");
389
241 encode_ch (enc, '{'); NL; ++enc->indent; 390 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
242 391
243 if ((count = hv_iterinit (hv))) 392 if ((count = hv_iterinit (hv)))
244 { 393 {
245 // for canonical output we have to sort by keys first 394 // for canonical output we have to sort by keys first
246 // actually, this is mostly due to the stupid so-called 395 // actually, this is mostly due to the stupid so-called
247 // security workaround added somewhere in 5.8.x. 396 // security workaround added somewhere in 5.8.x.
248 // that randomises hash orderings 397 // that randomises hash orderings
249 if (enc->flags & F_CANONICAL) 398 if (enc->json.flags & F_CANONICAL)
250 { 399 {
251 HE *he, *hes [count];
252 int fast = 1; 400 int fast = 1;
401 HE *he;
402#if defined(__BORLANDC__) || defined(_MSC_VER)
403 HE **hes = _alloca (count * sizeof (HE));
404#else
405 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
406#endif
253 407
254 i = 0; 408 i = 0;
255 while ((he = hv_iternext (hv))) 409 while ((he = hv_iternext (hv)))
256 { 410 {
257 hes [i++] = he; 411 hes [i++] = he;
… …
263 417
264 if (fast) 418 if (fast)
265 qsort (hes, count, sizeof (HE *), he_cmp_fast); 419 qsort (hes, count, sizeof (HE *), he_cmp_fast);
266 else 420 else
267 { 421 {
268 // hack to disable "use bytes" 422 // hack to forcefully disable "use bytes"
269 COP *oldcop = PL_curcop, cop; 423 COP cop = *PL_curcop;
270 cop.op_private = 0; 424 cop.op_private = 0;
425
426 ENTER;
427 SAVETMPS;
428
429 SAVEVPTR (PL_curcop);
271 PL_curcop = &cop; 430 PL_curcop = &cop;
272 431
273 SAVETMPS;
274 qsort (hes, count, sizeof (HE *), he_cmp_slow); 432 qsort (hes, count, sizeof (HE *), he_cmp_slow);
433
275 FREETMPS; 434 FREETMPS;
276 435 LEAVE;
277 PL_curcop = oldcop;
278 } 436 }
279 437
280 for (i = 0; i < count; ++i) 438 for (i = 0; i < count; ++i)
281 { 439 {
282 INDENT; 440 encode_indent (enc);
283 encode_he (enc, hes [i]); 441 encode_he (enc, hes [i]);
284 442
285 if (i < count - 1) 443 if (i < count - 1)
286 COMMA; 444 encode_comma (enc);
287 }
288
289 NL; 445 }
446
447 encode_nl (enc);
290 } 448 }
291 else 449 else
292 { 450 {
293 SV *sv;
294 HE *he = hv_iternext (hv); 451 HE *he = hv_iternext (hv);
295 452
296 for (;;) 453 for (;;)
297 { 454 {
298 INDENT; 455 encode_indent (enc);
299 encode_he (enc, he); 456 encode_he (enc, he);
300 457
301 if (!(he = hv_iternext (hv))) 458 if (!(he = hv_iternext (hv)))
302 break; 459 break;
303 460
304 COMMA; 461 encode_comma (enc);
305 }
306
307 NL; 462 }
463
464 encode_nl (enc);
308 } 465 }
309 } 466 }
310 467
311 --enc->indent; INDENT; encode_ch (enc, '}'); 468 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
469}
470
471// encode objects, arrays and special \0=false and \1=true values.
472static void
473encode_rv (enc_t *enc, SV *sv)
474{
475 svtype svt;
476
477 SvGETMAGIC (sv);
478 svt = SvTYPE (sv);
479
480 if (expect_false (SvOBJECT (sv)))
481 {
482 if (SvSTASH (sv) == json_boolean_stash)
483 {
484 if (SvIV (sv))
485 encode_str (enc, "true", 4, 0);
486 else
487 encode_str (enc, "false", 5, 0);
488 }
489 else
490 {
491#if 0
492 if (0 && sv_derived_from (rv, "JSON::Literal"))
493 {
494 // not yet
495 }
496#endif
497 if (enc->json.flags & F_CONV_BLESSED)
498 {
499 // we re-bless the reference to get overload and other niceties right
500 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
501
502 if (to_json)
503 {
504 int count;
505 dSP;
506
507 ENTER; SAVETMPS; PUSHMARK (SP);
508 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
509
510 // calling with G_SCALAR ensures that we always get a 1 return value
511 PUTBACK;
512 call_sv ((SV *)GvCV (to_json), G_SCALAR);
513 SPAGAIN;
514
515 // catch this surprisingly common error
516 if (SvROK (TOPs) && SvRV (TOPs) == sv)
517 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
518
519 sv = POPs;
520 PUTBACK;
521
522 encode_sv (enc, sv);
523
524 FREETMPS; LEAVE;
525 }
526 else if (enc->json.flags & F_ALLOW_BLESSED)
527 encode_str (enc, "null", 4, 0);
528 else
529 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
530 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
531 }
532 else if (enc->json.flags & F_ALLOW_BLESSED)
533 encode_str (enc, "null", 4, 0);
534 else
535 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",
536 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
537 }
538 }
539 else if (svt == SVt_PVHV)
540 encode_hv (enc, (HV *)sv);
541 else if (svt == SVt_PVAV)
542 encode_av (enc, (AV *)sv);
543 else if (svt < SVt_PVAV)
544 {
545 STRLEN len = 0;
546 char *pv = svt ? SvPV (sv, len) : 0;
547
548 if (len == 1 && *pv == '1')
549 encode_str (enc, "true", 4, 0);
550 else if (len == 1 && *pv == '0')
551 encode_str (enc, "false", 5, 0);
552 else
553 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
554 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
555 }
556 else
557 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
558 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
312} 559}
313 560
314static void 561static void
315encode_sv (enc_t *enc, SV *sv) 562encode_sv (enc_t *enc, SV *sv)
316{ 563{
564 SvGETMAGIC (sv);
565
317 if (SvPOKp (sv)) 566 if (SvPOKp (sv))
318 { 567 {
319 STRLEN len; 568 STRLEN len;
320 char *str = SvPV (sv, len); 569 char *str = SvPV (sv, len);
321 encode_ch (enc, '"'); 570 encode_ch (enc, '"');
322 encode_str (enc, str, len, SvUTF8 (sv)); 571 encode_str (enc, str, len, SvUTF8 (sv));
323 encode_ch (enc, '"'); 572 encode_ch (enc, '"');
324 } 573 }
325 else if (SvNOKp (sv)) 574 else if (SvNOKp (sv))
326 { 575 {
576 // trust that perl will do the right thing w.r.t. JSON syntax.
327 need (enc, NV_DIG + 32); 577 need (enc, NV_DIG + 32);
328 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 578 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
329 enc->cur += strlen (enc->cur); 579 enc->cur += strlen (enc->cur);
330 } 580 }
331 else if (SvIOKp (sv)) 581 else if (SvIOKp (sv))
332 { 582 {
333 need (enc, 64); 583 // we assume we can always read an IV as a UV
584 if (SvUV (sv) & ~(UV)0x7fff)
585 {
586 // large integer, use the (rather slow) snprintf way.
587 need (enc, sizeof (UV) * 3);
334 enc->cur += 588 enc->cur +=
335 SvIsUV(sv) 589 SvIsUV(sv)
336 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 590 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
337 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 591 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
592 }
593 else
594 {
595 // optimise the "small number case"
596 // code will likely be branchless and use only a single multiplication
597 I32 i = SvIV (sv);
598 U32 u;
599 char digit, nz = 0;
600
601 need (enc, 6);
602
603 *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
604 u = i < 0 ? -i : i;
605
606 // convert to 4.28 fixed-point representation
607 u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
608
609 // now output digit by digit, each time masking out the integer part
610 // and multiplying by 5 while moving the decimal point one to the right,
611 // resulting in a net multiplication by 10.
612 // we always write the digit to memory but conditionally increment
613 // the pointer, to ease the usage of conditional move instructions.
614 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5;
615 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5;
616 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5;
617 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5;
618 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
619 }
338 } 620 }
339 else if (SvROK (sv)) 621 else if (SvROK (sv))
340 { 622 encode_rv (enc, SvRV (sv));
341 if (!--enc->max_recurse)
342 croak ("data structure too deep (hit recursion limit)");
343
344 sv = SvRV (sv);
345
346 switch (SvTYPE (sv))
347 {
348 case SVt_PVAV: encode_av (enc, (AV *)sv); break;
349 case SVt_PVHV: encode_hv (enc, (HV *)sv); break;
350
351 default:
352 croak ("JSON can only represent references to arrays or hashes");
353 }
354 }
355 else if (!SvOK (sv)) 623 else if (!SvOK (sv))
356 encode_str (enc, "null", 4, 0); 624 encode_str (enc, "null", 4, 0);
357 else 625 else
358 croak ("encountered perl type that JSON cannot handle"); 626 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
627 SvPV_nolen (sv), SvFLAGS (sv));
359} 628}
360 629
361static SV * 630static SV *
362encode_json (SV *scalar, UV flags) 631encode_json (SV *scalar, JSON *json)
363{ 632{
364 enc_t enc; 633 enc_t enc;
365 enc.flags = flags; 634
635 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
636 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
637
638 enc.json = *json;
366 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 639 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
367 enc.cur = SvPVX (enc.sv); 640 enc.cur = SvPVX (enc.sv);
368 enc.end = SvEND (enc.sv); 641 enc.end = SvEND (enc.sv);
369 enc.max_recurse = 0;
370 enc.indent = 0; 642 enc.indent = 0;
643 enc.maxdepth = DEC_DEPTH (enc.json.flags);
371 644
372 SvPOK_only (enc.sv); 645 SvPOK_only (enc.sv);
373 encode_sv (&enc, scalar); 646 encode_sv (&enc, scalar);
374 647
648 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
649 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
650
375 if (!(flags & (F_ASCII | F_UTF8))) 651 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
376 SvUTF8_on (enc.sv); 652 SvUTF8_on (enc.sv);
377 653
378 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 654 if (enc.json.flags & F_SHRINK)
655 shrink (enc.sv);
656
379 return enc.sv; 657 return enc.sv;
380} 658}
381 659
382///////////////////////////////////////////////////////////////////////////// 660/////////////////////////////////////////////////////////////////////////////
661// decoder
383 662
384#define WS \ 663// structure used for decoding JSON
664typedef struct
665{
666 char *cur; // current parser pointer
667 char *end; // end of input string
668 const char *err; // parse error, if != 0
669 JSON json;
670 U32 depth; // recursion depth
671 U32 maxdepth; // recursion depth limit
672} dec_t;
673
674inline void
675decode_ws (dec_t *dec)
676{
385 for (;;) \ 677 for (;;)
386 { \ 678 {
387 char ch = *dec->cur; \ 679 char ch = *dec->cur;
680
388 if (ch > 0x20 \ 681 if (ch > 0x20
389 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) \ 682 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09))
390 break; \ 683 break;
684
391 ++dec->cur; \ 685 ++dec->cur;
392 } 686 }
687}
393 688
394#define ERR(reason) SB dec->err = reason; goto fail; SE 689#define ERR(reason) SB dec->err = reason; goto fail; SE
690
395#define EXPECT_CH(ch) SB \ 691#define EXPECT_CH(ch) SB \
396 if (*dec->cur != ch) \ 692 if (*dec->cur != ch) \
397 ERR (# ch " expected"); \ 693 ERR (# ch " expected"); \
398 ++dec->cur; \ 694 ++dec->cur; \
399 SE 695 SE
400 696
697#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)")
698#define DEC_DEC_DEPTH --dec->depth
699
401static SV *decode_sv (dec_t *dec); 700static SV *decode_sv (dec_t *dec);
402
403#define APPEND_CH(ch) SB \
404 SvGROW (sv, cur + 1 + 1); \
405 SvPVX (sv)[cur++] = (ch); \
406 SE
407 701
408static signed char decode_hexdigit[256]; 702static signed char decode_hexdigit[256];
409 703
410static UV 704static UV
411decode_4hex (dec_t *dec) 705decode_4hex (dec_t *dec)
412{ 706{
413 signed char d1, d2, d3, d4; 707 signed char d1, d2, d3, d4;
708 unsigned char *cur = (unsigned char *)dec->cur;
414 709
415 d1 = decode_hexdigit [((unsigned char *)dec->cur) [0]]; 710 d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected");
416 if (d1 < 0) ERR ("four hexadecimal digits expected"); 711 d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected");
417 d2 = decode_hexdigit [((unsigned char *)dec->cur) [1]]; 712 d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected");
418 if (d2 < 0) ERR ("four hexadecimal digits expected"); 713 d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected");
419 d3 = decode_hexdigit [((unsigned char *)dec->cur) [2]];
420 if (d3 < 0) ERR ("four hexadecimal digits expected");
421 d4 = decode_hexdigit [((unsigned char *)dec->cur) [3]];
422 if (d4 < 0) ERR ("four hexadecimal digits expected");
423 714
424 dec->cur += 4; 715 dec->cur += 4;
425 716
426 return ((UV)d1) << 12 717 return ((UV)d1) << 12
427 | ((UV)d2) << 8 718 | ((UV)d2) << 8
… …
433} 724}
434 725
435static SV * 726static SV *
436decode_str (dec_t *dec) 727decode_str (dec_t *dec)
437{ 728{
438 SV *sv = NEWSV (0,2); 729 SV *sv = 0;
439 STRLEN cur = 0;
440 int utf8 = 0; 730 int utf8 = 0;
731 char *dec_cur = dec->cur;
441 732
442 for (;;) 733 do
443 { 734 {
444 unsigned char ch = *(unsigned char *)dec->cur; 735 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
736 char *cur = buf;
445 737
446 if (ch == '"') 738 do
447 break;
448 else if (ch == '\\')
449 { 739 {
450 switch (*++dec->cur) 740 unsigned char ch = *(unsigned char *)dec_cur++;
741
742 if (expect_false (ch == '"'))
743 {
744 --dec_cur;
745 break;
451 { 746 }
452 case '\\': 747 else if (expect_false (ch == '\\'))
453 case '/': 748 {
454 case '"': APPEND_CH (*dec->cur++); break; 749 switch (*dec_cur)
455
456 case 'b': APPEND_CH ('\010'); ++dec->cur; break;
457 case 't': APPEND_CH ('\011'); ++dec->cur; break;
458 case 'n': APPEND_CH ('\012'); ++dec->cur; break;
459 case 'f': APPEND_CH ('\014'); ++dec->cur; break;
460 case 'r': APPEND_CH ('\015'); ++dec->cur; break;
461
462 case 'u':
463 { 750 {
464 UV lo, hi; 751 case '\\':
465 ++dec->cur; 752 case '/':
753 case '"': *cur++ = *dec_cur++; break;
466 754
467 hi = decode_4hex (dec); 755 case 'b': ++dec_cur; *cur++ = '\010'; break;
468 if (hi == (UV)-1) 756 case 't': ++dec_cur; *cur++ = '\011'; break;
469 goto fail; 757 case 'n': ++dec_cur; *cur++ = '\012'; break;
758 case 'f': ++dec_cur; *cur++ = '\014'; break;
759 case 'r': ++dec_cur; *cur++ = '\015'; break;
470 760
471 // possibly a surrogate pair 761 case 'u':
472 if (hi >= 0xd800 && hi < 0xdc00)
473 { 762 {
474 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 763 UV lo, hi;
475 ERR ("illegal surrogate character"); 764 ++dec_cur;
476 765
477 dec->cur += 2; 766 dec->cur = dec_cur;
478
479 lo = decode_4hex (dec); 767 hi = decode_4hex (dec);
768 dec_cur = dec->cur;
480 if (lo == (UV)-1) 769 if (hi == (UV)-1)
481 goto fail; 770 goto fail;
482 771
772 // possibly a surrogate pair
773 if (hi >= 0xd800)
774 if (hi < 0xdc00)
775 {
776 if (dec_cur [0] != '\\' || dec_cur [1] != 'u')
777 ERR ("missing low surrogate character in surrogate pair");
778
779 dec_cur += 2;
780
781 dec->cur = dec_cur;
782 lo = decode_4hex (dec);
783 dec_cur = dec->cur;
784 if (lo == (UV)-1)
785 goto fail;
786
483 if (lo < 0xdc00 || lo >= 0xe000) 787 if (lo < 0xdc00 || lo >= 0xe000)
484 ERR ("surrogate pair expected"); 788 ERR ("surrogate pair expected");
485 789
486 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000; 790 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000;
791 }
792 else if (hi < 0xe000)
793 ERR ("missing high surrogate character in surrogate pair");
794
795 if (hi >= 0x80)
796 {
797 utf8 = 1;
798
799 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
800 }
801 else
802 *cur++ = hi;
487 } 803 }
488 else if (lo >= 0xdc00 && lo < 0xe000)
489 ERR ("illegal surrogate character");
490
491 if (hi >= 0x80)
492 { 804 break;
493 utf8 = 1;
494 805
495 SvGROW (sv, cur + 4 + 1); // at most 4 bytes for 21 bits
496 cur = (char *)uvuni_to_utf8_flags (SvPVX (sv) + cur, hi, 0) - SvPVX (sv);
497 }
498 else 806 default:
499 APPEND_CH (hi); 807 --dec_cur;
808 ERR ("illegal backslash escape sequence in string");
500 } 809 }
810 }
811 else if (expect_true (ch >= 0x20 && ch <= 0x7f))
812 *cur++ = ch;
813 else if (ch >= 0x80)
814 {
815 STRLEN clen;
816 UV uch;
817
818 --dec_cur;
819
820 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
821 if (clen == (STRLEN)-1)
822 ERR ("malformed UTF-8 character in JSON string");
823
824 do
825 *cur++ = *dec_cur++;
826 while (--clen);
827
828 utf8 = 1;
829 }
830 else
831 {
832 --dec_cur;
833
834 if (!ch)
835 ERR ("unexpected end of string while parsing JSON string");
501 break; 836 else
837 ERR ("invalid character encountered while parsing JSON string");
502 } 838 }
503 } 839 }
504 else if (ch >= 0x20 && ch <= 0x7f) 840 while (cur < buf + SHORT_STRING_LEN);
505 APPEND_CH (*dec->cur++); 841
506 else if (ch >= 0x80) 842 {
843 STRLEN len = cur - buf;
844
845 if (sv)
507 { 846 {
508 STRLEN clen; 847 SvGROW (sv, SvCUR (sv) + len + 1);
509 UV uch = utf8n_to_uvuni (dec->cur, dec->end - dec->cur, &clen, UTF8_CHECK_ONLY); 848 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
510 if (clen < 0) 849 SvCUR_set (sv, SvCUR (sv) + len);
511 ERR ("malformed UTF-8 character in string, cannot convert to JSON");
512
513 SvGROW (sv, cur + clen + 1); // at most 4 bytes for 21 bits
514 memcpy (SvPVX (sv) + cur, dec->cur, clen);
515 dec->cur += clen;
516 } 850 }
517 else 851 else
518 ERR ("invalid character encountered"); 852 sv = newSVpvn (buf, len);
519 } 853 }
854 }
855 while (*dec_cur != '"');
520 856
521 ++dec->cur; 857 ++dec_cur;
522 858
859 if (sv)
860 {
523 SvPOK_only (sv); 861 SvPOK_only (sv);
524
525 SvCUR_set (sv, cur);
526 *SvEND (sv) = 0; 862 *SvEND (sv) = 0;
527 863
528 if (utf8) 864 if (utf8)
529 SvUTF8_on (sv); 865 SvUTF8_on (sv);
866 }
867 else
868 sv = newSVpvn ("", 0);
530 869
870 dec->cur = dec_cur;
531 return sv; 871 return sv;
532 872
533fail: 873fail:
534 SvREFCNT_dec (sv); 874 dec->cur = dec_cur;
535 return 0; 875 return 0;
536} 876}
537 877
538static SV * 878static SV *
539decode_num (dec_t *dec) 879decode_num (dec_t *dec)
… …
549 { 889 {
550 ++dec->cur; 890 ++dec->cur;
551 if (*dec->cur >= '0' && *dec->cur <= '9') 891 if (*dec->cur >= '0' && *dec->cur <= '9')
552 ERR ("malformed number (leading zero must not be followed by another digit)"); 892 ERR ("malformed number (leading zero must not be followed by another digit)");
553 } 893 }
554 894 else if (*dec->cur < '0' || *dec->cur > '9')
555 // int 895 ERR ("malformed number (no digits after initial minus)");
896 else
897 do
898 {
899 ++dec->cur;
900 }
556 while (*dec->cur >= '0' && *dec->cur <= '9') 901 while (*dec->cur >= '0' && *dec->cur <= '9');
557 ++dec->cur;
558 902
559 // [frac] 903 // [frac]
560 if (*dec->cur == '.') 904 if (*dec->cur == '.')
561 { 905 {
562 is_nv = 1; 906 ++dec->cur;
907
908 if (*dec->cur < '0' || *dec->cur > '9')
909 ERR ("malformed number (no digits after decimal point)");
563 910
564 do 911 do
565 { 912 {
566 ++dec->cur; 913 ++dec->cur;
567 } 914 }
568 while (*dec->cur >= '0' && *dec->cur <= '9'); 915 while (*dec->cur >= '0' && *dec->cur <= '9');
916
917 is_nv = 1;
569 } 918 }
570 919
571 // [exp] 920 // [exp]
572 if (*dec->cur == 'e' || *dec->cur == 'E') 921 if (*dec->cur == 'e' || *dec->cur == 'E')
573 { 922 {
574 is_nv = 1;
575
576 ++dec->cur; 923 ++dec->cur;
924
577 if (*dec->cur == '-' || *dec->cur == '+') 925 if (*dec->cur == '-' || *dec->cur == '+')
578 ++dec->cur; 926 ++dec->cur;
579 927
928 if (*dec->cur < '0' || *dec->cur > '9')
929 ERR ("malformed number (no digits after exp sign)");
930
931 do
932 {
933 ++dec->cur;
934 }
580 while (*dec->cur >= '0' && *dec->cur <= '9') 935 while (*dec->cur >= '0' && *dec->cur <= '9');
581 ++dec->cur; 936
937 is_nv = 1;
582 } 938 }
583 939
584 if (!is_nv) 940 if (!is_nv)
585 { 941 {
586 UV uv; 942 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so
587 int numtype = grok_number (start, dec->cur - start, &uv); 943 if (*start == '-')
588 if (numtype & IS_NUMBER_IN_UV) 944 switch (dec->cur - start)
589 if (numtype & IS_NUMBER_NEG)
590 { 945 {
591 if (uv < (UV)IV_MIN) 946 case 2: return newSViv (-( start [1] - '0' * 1));
592 return newSViv (-(IV)uv); 947 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
948 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
949 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
593 } 950 }
951 else
952 switch (dec->cur - start)
953 {
954 case 1: return newSViv ( start [0] - '0' * 1);
955 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
956 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
957 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
958 }
959
960 {
961 UV uv;
962 int numtype = grok_number (start, dec->cur - start, &uv);
963 if (numtype & IS_NUMBER_IN_UV)
964 if (numtype & IS_NUMBER_NEG)
965 {
966 if (uv < (UV)IV_MIN)
967 return newSViv (-(IV)uv);
968 }
594 else 969 else
595 return newSVuv (uv); 970 return newSVuv (uv);
971
972 // here would likely be the place for bigint support
596 } 973 }
974 }
597 975
976 // if we ever support bigint or bigfloat, this is the place for bigfloat
598 return newSVnv (Atof (start)); 977 return newSVnv (Atof (start));
599 978
600fail: 979fail:
601 return 0; 980 return 0;
602} 981}
… …
604static SV * 983static SV *
605decode_av (dec_t *dec) 984decode_av (dec_t *dec)
606{ 985{
607 AV *av = newAV (); 986 AV *av = newAV ();
608 987
988 DEC_INC_DEPTH;
989 decode_ws (dec);
990
991 if (*dec->cur == ']')
992 ++dec->cur;
993 else
609 for (;;) 994 for (;;)
610 { 995 {
611 SV *value; 996 SV *value;
612 997
613 value = decode_sv (dec); 998 value = decode_sv (dec);
614 if (!value) 999 if (!value)
615 goto fail; 1000 goto fail;
616 1001
617 av_push (av, value); 1002 av_push (av, value);
618 1003
619 WS; 1004 decode_ws (dec);
620 1005
621 if (*dec->cur == ']') 1006 if (*dec->cur == ']')
622 { 1007 {
623 ++dec->cur; 1008 ++dec->cur;
624 break; 1009 break;
1010 }
625 } 1011
626
627 if (*dec->cur != ',') 1012 if (*dec->cur != ',')
628 ERR (", or ] expected while parsing array"); 1013 ERR (", or ] expected while parsing array");
629 1014
630 ++dec->cur; 1015 ++dec->cur;
631 } 1016 }
632 1017
1018 DEC_DEC_DEPTH;
633 return newRV_noinc ((SV *)av); 1019 return newRV_noinc ((SV *)av);
634 1020
635fail: 1021fail:
636 SvREFCNT_dec (av); 1022 SvREFCNT_dec (av);
1023 DEC_DEC_DEPTH;
637 return 0; 1024 return 0;
638} 1025}
639 1026
640static SV * 1027static SV *
641decode_hv (dec_t *dec) 1028decode_hv (dec_t *dec)
642{ 1029{
1030 SV *sv;
643 HV *hv = newHV (); 1031 HV *hv = newHV ();
644 1032
1033 DEC_INC_DEPTH;
1034 decode_ws (dec);
1035
1036 if (*dec->cur == '}')
1037 ++dec->cur;
1038 else
645 for (;;) 1039 for (;;)
646 { 1040 {
647 SV *key, *value; 1041 decode_ws (dec); EXPECT_CH ('"');
648 1042
649 WS; EXPECT_CH ('"'); 1043 // heuristic: assume that
650 1044 // a) decode_str + hv_store_ent are abysmally slow.
651 key = decode_str (dec); 1045 // b) most hash keys are short, simple ascii text.
652 if (!key) 1046 // => try to "fast-match" such strings to avoid
653 goto fail; 1047 // the overhead of decode_str + hv_store_ent.
654
655 WS; EXPECT_CH (':');
656
657 value = decode_sv (dec);
658 if (!value)
659 { 1048 {
1049 SV *value;
1050 char *p = dec->cur;
1051 char *e = p + 24; // only try up to 24 bytes
1052
1053 for (;;)
1054 {
1055 // the >= 0x80 is true on most architectures
1056 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1057 {
1058 // slow path, back up and use decode_str
1059 SV *key = decode_str (dec);
1060 if (!key)
1061 goto fail;
1062
1063 decode_ws (dec); EXPECT_CH (':');
1064
1065 value = decode_sv (dec);
1066 if (!value)
1067 {
1068 SvREFCNT_dec (key);
1069 goto fail;
1070 }
1071
1072 hv_store_ent (hv, key, value, 0);
660 SvREFCNT_dec (key); 1073 SvREFCNT_dec (key);
1074
1075 break;
1076 }
1077 else if (*p == '"')
1078 {
1079 // fast path, got a simple key
1080 char *key = dec->cur;
1081 int len = p - key;
1082 dec->cur = p + 1;
1083
1084 decode_ws (dec); EXPECT_CH (':');
1085
1086 value = decode_sv (dec);
1087 if (!value)
661 goto fail; 1088 goto fail;
1089
1090 hv_store (hv, key, len, value, 0);
1091
1092 break;
1093 }
1094
1095 ++p;
1096 }
662 } 1097 }
663 1098
664 //TODO: optimise 1099 decode_ws (dec);
665 hv_store_ent (hv, key, value, 0);
666 1100
667 WS;
668
669 if (*dec->cur == '}') 1101 if (*dec->cur == '}')
1102 {
1103 ++dec->cur;
1104 break;
1105 }
1106
1107 if (*dec->cur != ',')
1108 ERR (", or } expected while parsing object/hash");
1109
1110 ++dec->cur;
1111 }
1112
1113 DEC_DEC_DEPTH;
1114 sv = newRV_noinc ((SV *)hv);
1115
1116 // check filter callbacks
1117 if (dec->json.flags & F_HOOK)
1118 {
1119 if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
670 { 1120 {
671 ++dec->cur; 1121 HE *cb, *he;
672 break; 1122
1123 hv_iterinit (hv);
1124 he = hv_iternext (hv);
1125 hv_iterinit (hv);
1126
1127 // the next line creates a mortal sv each time its called.
1128 // might want to optimise this for common cases.
1129 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1130
1131 if (cb)
1132 {
1133 dSP;
1134 int count;
1135
1136 ENTER; SAVETMPS; PUSHMARK (SP);
1137 XPUSHs (HeVAL (he));
1138
1139 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1140
1141 if (count == 1)
1142 {
1143 sv = newSVsv (POPs);
1144 FREETMPS; LEAVE;
1145 return sv;
1146 }
1147
1148 FREETMPS; LEAVE;
1149 }
673 } 1150 }
674 1151
675 if (*dec->cur != ',') 1152 if (dec->json.cb_object)
676 ERR (", or } expected while parsing object/hash"); 1153 {
1154 dSP;
1155 int count;
677 1156
678 ++dec->cur; 1157 ENTER; SAVETMPS; PUSHMARK (SP);
679 } 1158 XPUSHs (sv_2mortal (sv));
680 1159
681 return newRV_noinc ((SV *)hv); 1160 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1161
1162 if (count == 1)
1163 {
1164 sv = newSVsv (POPs);
1165 FREETMPS; LEAVE;
1166 return sv;
1167 }
1168
1169 SvREFCNT_inc (sv);
1170 FREETMPS; LEAVE;
1171 }
1172 }
1173
1174 return sv;
682 1175
683fail: 1176fail:
684 SvREFCNT_dec (hv); 1177 SvREFCNT_dec (hv);
1178 DEC_DEC_DEPTH;
685 return 0; 1179 return 0;
686} 1180}
687 1181
688static SV * 1182static SV *
689decode_sv (dec_t *dec) 1183decode_sv (dec_t *dec)
690{ 1184{
691 WS; 1185 decode_ws (dec);
1186
1187 // the beauty of JSON: you need exactly one character lookahead
1188 // to parse anything.
692 switch (*dec->cur) 1189 switch (*dec->cur)
693 { 1190 {
694 case '"': ++dec->cur; return decode_str (dec); 1191 case '"': ++dec->cur; return decode_str (dec);
695 case '[': ++dec->cur; return decode_av (dec); 1192 case '[': ++dec->cur; return decode_av (dec);
696 case '{': ++dec->cur; return decode_hv (dec); 1193 case '{': ++dec->cur; return decode_hv (dec);
… …
702 1199
703 case 't': 1200 case 't':
704 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1201 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
705 { 1202 {
706 dec->cur += 4; 1203 dec->cur += 4;
707 return newSViv (1); 1204 return SvREFCNT_inc (json_true);
708 } 1205 }
709 else 1206 else
710 ERR ("'true' expected"); 1207 ERR ("'true' expected");
711 1208
712 break; 1209 break;
713 1210
714 case 'f': 1211 case 'f':
715 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1212 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
716 { 1213 {
717 dec->cur += 5; 1214 dec->cur += 5;
718 return newSViv (0); 1215 return SvREFCNT_inc (json_false);
719 } 1216 }
720 else 1217 else
721 ERR ("'false' expected"); 1218 ERR ("'false' expected");
722 1219
723 break; 1220 break;
724 1221
725 case 'n': 1222 case 'n':
726 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4)) 1223 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4))
727 { 1224 {
728 dec->cur += 4; 1225 dec->cur += 4;
729 return newSViv (1); 1226 return newSVsv (&PL_sv_undef);
730 } 1227 }
731 else 1228 else
732 ERR ("'null' expected"); 1229 ERR ("'null' expected");
733 1230
734 break; 1231 break;
735 1232
736 default: 1233 default:
737 ERR ("malformed json string"); 1234 ERR ("malformed JSON string, neither array, object, number, string or atom");
738 break; 1235 break;
739 } 1236 }
740 1237
741fail: 1238fail:
742 return 0; 1239 return 0;
743} 1240}
744 1241
745static SV * 1242static SV *
746decode_json (SV *string, UV flags) 1243decode_json (SV *string, JSON *json, UV *offset_return)
747{ 1244{
1245 dec_t dec;
1246 UV offset;
748 SV *sv; 1247 SV *sv;
749 1248
1249 SvGETMAGIC (string);
1250 SvUPGRADE (string, SVt_PV);
1251
1252 if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags))
1253 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1254 (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags));
1255
750 if (!(flags & F_UTF8)) 1256 if (json->flags & F_UTF8)
1257 sv_utf8_downgrade (string, 0);
1258 else
751 sv_utf8_upgrade (string); 1259 sv_utf8_upgrade (string);
752 1260
753 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1261 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
754 1262
755 dec_t dec; 1263 dec.json = *json;
756 dec.flags = flags;
757 dec.cur = SvPVX (string); 1264 dec.cur = SvPVX (string);
758 dec.end = SvEND (string); 1265 dec.end = SvEND (string);
759 dec.err = 0; 1266 dec.err = 0;
1267 dec.depth = 0;
1268 dec.maxdepth = DEC_DEPTH (dec.json.flags);
760 1269
761 *dec.end = 1; // invalid anywhere 1270 if (dec.json.cb_object || dec.json.cb_sk_object)
1271 dec.json.flags |= F_HOOK;
1272
1273 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
762 sv = decode_sv (&dec); 1274 sv = decode_sv (&dec);
763 *dec.end = 0; 1275
1276 if (!(offset_return || !sv))
1277 {
1278 // check for trailing garbage
1279 decode_ws (&dec);
1280
1281 if (*dec.cur)
1282 {
1283 dec.err = "garbage after JSON object";
1284 SvREFCNT_dec (sv);
1285 sv = 0;
1286 }
1287 }
1288
1289 if (offset_return || !sv)
1290 {
1291 offset = dec.json.flags & F_UTF8
1292 ? dec.cur - SvPVX (string)
1293 : utf8_distance (dec.cur, SvPVX (string));
1294
1295 if (offset_return)
1296 *offset_return = offset;
1297 }
764 1298
765 if (!sv) 1299 if (!sv)
766 { 1300 {
767 IV offset = utf8_distance (dec.cur, SvPVX (string));
768 SV *uni = sv_newmortal (); 1301 SV *uni = sv_newmortal ();
769 1302
1303 // horrible hack to silence warning inside pv_uni_display
1304 COP cop = *PL_curcop;
1305 cop.cop_warnings = pWARN_NONE;
1306 ENTER;
1307 SAVEVPTR (PL_curcop);
1308 PL_curcop = &cop;
770 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1309 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1310 LEAVE;
1311
771 croak ("%s, at character %d (%s)", 1312 croak ("%s, at character offset %d [\"%s\"]",
772 dec.err, 1313 dec.err,
773 (int)offset, 1314 (int)offset,
774 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1315 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
775 } 1316 }
776 1317
777 sv_dump (sv);//D
778 return sv_2mortal (sv); 1318 sv = sv_2mortal (sv);
1319
1320 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))
1321 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1322
1323 return sv;
779} 1324}
1325
1326/////////////////////////////////////////////////////////////////////////////
1327// XS interface functions
780 1328
781MODULE = JSON::XS PACKAGE = JSON::XS 1329MODULE = JSON::XS PACKAGE = JSON::XS
782 1330
783BOOT: 1331BOOT:
784{ 1332{
785 int i; 1333 int i;
786 1334
787 memset (decode_hexdigit, 0xff, 256);
788 for (i = 10; i--; ) 1335 for (i = 0; i < 256; ++i)
789 decode_hexdigit ['0' + i] = i; 1336 decode_hexdigit [i] =
1337 i >= '0' && i <= '9' ? i - '0'
1338 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1339 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1340 : -1;
790 1341
791 for (i = 6; --i; ) 1342 json_stash = gv_stashpv ("JSON::XS" , 1);
1343 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1344
1345 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true );
1346 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
1347}
1348
1349PROTOTYPES: DISABLE
1350
1351void new (char *klass)
1352 PPCODE:
1353{
1354 SV *pv = NEWSV (0, sizeof (JSON));
1355 SvPOK_only (pv);
1356 Zero (SvPVX (pv), 1, JSON);
1357 ((JSON *)SvPVX (pv))->flags = F_DEFAULT;
1358 XPUSHs (sv_2mortal (sv_bless (newRV_noinc (pv), json_stash)));
1359}
1360
1361void ascii (JSON *self, int enable = 1)
1362 ALIAS:
1363 ascii = F_ASCII
1364 latin1 = F_LATIN1
1365 utf8 = F_UTF8
1366 indent = F_INDENT
1367 canonical = F_CANONICAL
1368 space_before = F_SPACE_BEFORE
1369 space_after = F_SPACE_AFTER
1370 pretty = F_PRETTY
1371 allow_nonref = F_ALLOW_NONREF
1372 shrink = F_SHRINK
1373 allow_blessed = F_ALLOW_BLESSED
1374 convert_blessed = F_CONV_BLESSED
1375 PPCODE:
1376{
1377 if (enable)
1378 self->flags |= ix;
1379 else
1380 self->flags &= ~ix;
1381
1382 XPUSHs (ST (0));
1383}
1384
1385void max_depth (JSON *self, UV max_depth = 0x80000000UL)
1386 PPCODE:
1387{
1388 UV log2 = 0;
1389
1390 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1391
1392 while ((1UL << log2) < max_depth)
1393 ++log2;
1394
1395 self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1396
1397 XPUSHs (ST (0));
1398}
1399
1400void max_size (JSON *self, UV max_size = 0)
1401 PPCODE:
1402{
1403 UV log2 = 0;
1404
1405 if (max_size > 0x80000000UL) max_size = 0x80000000UL;
1406 if (max_size == 1) max_size = 2;
1407
1408 while ((1UL << log2) < max_size)
1409 ++log2;
1410
1411 self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1412
1413 XPUSHs (ST (0));
1414}
1415
1416void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1417 PPCODE:
1418{
1419 SvREFCNT_dec (self->cb_object);
1420 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
1421
1422 XPUSHs (ST (0));
1423}
1424
1425void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
1426 PPCODE:
1427{
1428 if (!self->cb_sk_object)
1429 self->cb_sk_object = newHV ();
1430
1431 if (SvOK (cb))
1432 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1433 else
792 { 1434 {
793 decode_hexdigit ['a' + i] = 10 + i; 1435 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
794 decode_hexdigit ['A' + i] = 10 + i; 1436
1437 if (!HvKEYS (self->cb_sk_object))
1438 {
1439 SvREFCNT_dec (self->cb_sk_object);
1440 self->cb_sk_object = 0;
1441 }
795 } 1442 }
796 1443
797 json_stash = gv_stashpv ("JSON::XS", 1); 1444 XPUSHs (ST (0));
798} 1445}
799 1446
800SV *new (char *dummy) 1447void encode (JSON *self, SV *scalar)
1448 PPCODE:
1449 XPUSHs (encode_json (scalar, self));
1450
1451void decode (JSON *self, SV *jsonstr)
1452 PPCODE:
1453 XPUSHs (decode_json (jsonstr, self, 0));
1454
1455void decode_prefix (JSON *self, SV *jsonstr)
1456 PPCODE:
1457{
1458 UV offset;
1459 EXTEND (SP, 2);
1460 PUSHs (decode_json (jsonstr, self, &offset));
1461 PUSHs (sv_2mortal (newSVuv (offset)));
1462}
1463
1464void DESTROY (JSON *self)
801 CODE: 1465 CODE:
802 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1466 SvREFCNT_dec (self->cb_sk_object);
803 OUTPUT: 1467 SvREFCNT_dec (self->cb_object);
804 RETVAL
805 1468
806SV *ascii (SV *self, int enable) 1469PROTOTYPES: ENABLE
807 ALIAS:
808 ascii = F_ASCII
809 utf8 = F_UTF8
810 indent = F_INDENT
811 canonical = F_CANONICAL
812 space_before = F_SPACE_BEFORE
813 space_after = F_SPACE_AFTER
814 json_rpc = F_JSON_RPC
815 pretty = F_PRETTY
816 CODE:
817{
818 UV *uv = SvJSON (self);
819 if (enable)
820 *uv |= ix;
821 else
822 *uv &= ~ix;
823
824 RETVAL = newSVsv (self);
825}
826 OUTPUT:
827 RETVAL
828
829void encode (SV *self, SV *scalar)
830 PPCODE:
831 XPUSHs (encode_json (scalar, *SvJSON (self)));
832
833void decode (SV *self, SV *jsonstr)
834 PPCODE:
835 XPUSHs (decode_json (jsonstr, *SvJSON (self)));
836 1470
837void to_json (SV *scalar) 1471void to_json (SV *scalar)
838 PPCODE: 1472 PPCODE:
1473{
1474 JSON json = { F_DEFAULT | F_UTF8 };
839 XPUSHs (encode_json (scalar, F_UTF8)); 1475 XPUSHs (encode_json (scalar, &json));
1476}
840 1477
841void from_json (SV *jsonstr) 1478void from_json (SV *jsonstr)
842 PPCODE: 1479 PPCODE:
1480{
1481 JSON json = { F_DEFAULT | F_UTF8 };
843 XPUSHs (decode_json (jsonstr, F_UTF8)); 1482 XPUSHs (decode_json (jsonstr, &json, 0));
1483}
844 1484

Diff Legend

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