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 1.1 by root, Thu Mar 22 16:40:16 2007 UTC vs.
Revision 1.12 by root, Sat Mar 24 22:10:08 2007 UTC

… …
10#define F_UTF8 0x00000002 10#define F_UTF8 0x00000002
11#define F_INDENT 0x00000004 11#define F_INDENT 0x00000004
12#define F_CANONICAL 0x00000008 12#define F_CANONICAL 0x00000008
13#define F_SPACE_BEFORE 0x00000010 13#define F_SPACE_BEFORE 0x00000010
14#define F_SPACE_AFTER 0x00000020 14#define F_SPACE_AFTER 0x00000020
15#define F_ALLOW_NONREF 0x00000080
15#define F_JSON_RPC 0x00000040 16#define F_SHRINK 0x00000100
16 17
18#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
17#define F_DEFAULT 0 19#define F_DEFAULT 0
18 20
19#define INIT_SIZE 32 // initial scalar size to be allocated 21#define INIT_SIZE 32 // initial scalar size to be allocated
22#define INDENT_STEP 3 // spaces per indentation level
23
24#define UTF8_MAX_LEN 11 // for perls UTF-X: max. number of octets per character
25#define SHORT_STRING_LEN 256 // special-case strings of up to this size
20 26
21#define SB do { 27#define SB do {
22#define SE } while (0) 28#define SE } while (0)
23 29
24static HV *json_stash; 30static HV *json_stash; // JSON::XS::
31
32/////////////////////////////////////////////////////////////////////////////
33// utility functions
34
35static UV *
36SvJSON (SV *sv)
37{
38 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash))
39 croak ("object is not of type JSON::XS");
40
41 return &SvUVX (SvRV (sv));
42}
43
44static void
45shrink (SV *sv)
46{
47 sv_utf8_downgrade (sv, 1);
48 if (SvLEN (sv) > SvCUR (sv) + 1)
49 {
50#ifdef SvPV_shrink_to_cur
51 SvPV_shrink_to_cur (sv);
52#elif defined (SvPV_renew)
53 SvPV_renew (sv, SvCUR (sv) + 1);
54#endif
55 }
56}
57
58/////////////////////////////////////////////////////////////////////////////
59// encoder
25 60
26// structure used for encoding JSON 61// structure used for encoding JSON
27typedef struct 62typedef struct
28{ 63{
29 char *cur; 64 char *cur; // SvPVX (sv) + current output position
30 STRLEN len; // SvLEN (sv)
31 char *end; // SvEND (sv) 65 char *end; // SvEND (sv)
32 SV *sv; 66 SV *sv; // result scalar
33 UV flags; 67 UV flags; // F_*
34 int max_recurse; 68 int indent; // indentation level
35 int indent; 69 int max_depth; // max. recursion level
36} enc_t; 70} enc_t;
37
38// structure used for decoding JSON
39typedef struct
40{
41 char *cur;
42 char *end;
43 char *err;
44 UV flags;
45} dec_t;
46
47static UV *
48SvJSON (SV *sv)
49{
50 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash))
51 croak ("object is not of type JSON::XS");
52
53 return &SvUVX (SvRV (sv));
54}
55
56/////////////////////////////////////////////////////////////////////////////
57 71
58static void 72static void
59need (enc_t *enc, STRLEN len) 73need (enc_t *enc, STRLEN len)
60{ 74{
61 if (enc->cur + len >= enc->end) 75 if (enc->cur + len >= enc->end)
62 { 76 {
63 STRLEN cur = enc->cur - SvPVX (enc->sv); 77 STRLEN cur = enc->cur - SvPVX (enc->sv);
64 SvGROW (enc->sv, cur + len + 1); 78 SvGROW (enc->sv, cur + len + 1);
65 enc->cur = SvPVX (enc->sv) + cur; 79 enc->cur = SvPVX (enc->sv) + cur;
66 enc->end = SvEND (enc->sv); 80 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv);
67 } 81 }
68} 82}
69 83
70static void 84static void
71encode_ch (enc_t *enc, char ch) 85encode_ch (enc_t *enc, char ch)
… …
77static void 91static void
78encode_str (enc_t *enc, char *str, STRLEN len, int is_utf8) 92encode_str (enc_t *enc, char *str, STRLEN len, int is_utf8)
79{ 93{
80 char *end = str + len; 94 char *end = str + len;
81 95
96 need (enc, len);
97
82 while (str < end) 98 while (str < end)
83 { 99 {
84 unsigned char ch = *(unsigned char *)str; 100 unsigned char ch = *(unsigned char *)str;
101
85 if (ch >= 0x20 && ch < 0x80) // most common case 102 if (ch >= 0x20 && ch < 0x80) // most common case
86 { 103 {
104 if (ch == '"') // but with slow exceptions
105 {
106 need (enc, len += 1);
107 *enc->cur++ = '\\';
108 *enc->cur++ = '"';
109 }
110 else if (ch == '\\')
111 {
112 need (enc, len += 1);
113 *enc->cur++ = '\\';
114 *enc->cur++ = '\\';
115 }
116 else
87 *enc->cur++ = ch; 117 *enc->cur++ = ch;
118
88 str++; 119 ++str;
89 } 120 }
90 else 121 else
91 { 122 {
92 STRLEN clen; 123 switch (ch)
93 UV uch;
94
95 if (is_utf8)
96 { 124 {
97 uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY); 125 case '\010': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'b'; ++str; break;
98 if (clen < 0) 126 case '\011': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 't'; ++str; break;
99 croak ("malformed UTF-8 character in string, cannot convert to JSON"); 127 case '\012': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'n'; ++str; break;
100 } 128 case '\014': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'f'; ++str; break;
101 else 129 case '\015': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'r'; ++str; break;
102 {
103 uch = ch;
104 clen = 1;
105 }
106 130
107 need (enc, len += 6); 131 default:
108
109 if (uch < 0xa0 || enc->flags & F_ASCII)
110 {
111 if (uch > 0xFFFFUL)
112 { 132 {
133 STRLEN clen;
134 UV uch;
135
136 if (is_utf8)
137 {
138 uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
139 if (clen == (STRLEN)-1)
140 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
141 }
142 else
143 {
144 uch = ch;
113 len += 6; 145 clen = 1;
146 }
147
148 if (uch > 0x10FFFFUL)
149 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
150
151 if (uch < 0x80 || enc->flags & F_ASCII)
152 {
153 if (uch > 0xFFFFUL)
154 {
114 need (enc, len += 6); 155 need (enc, len += 11);
115 sprintf (enc->cur, "\\u%04x\\u%04x", 156 sprintf (enc->cur, "\\u%04x\\u%04x",
116 (uch - 0x10000) / 0x400 + 0xD800, 157 (int)((uch - 0x10000) / 0x400 + 0xD800),
117 (uch - 0x10000) % 0x400 + 0xDC00); 158 (int)((uch - 0x10000) % 0x400 + 0xDC00));
118 enc->cur += 12; 159 enc->cur += 12;
160 }
161 else
162 {
163 static char hexdigit [16] = "0123456789abcdef";
164 need (enc, len += 5);
165 *enc->cur++ = '\\';
166 *enc->cur++ = 'u';
167 *enc->cur++ = hexdigit [ uch >> 12 ];
168 *enc->cur++ = hexdigit [(uch >> 8) & 15];
169 *enc->cur++ = hexdigit [(uch >> 4) & 15];
170 *enc->cur++ = hexdigit [(uch >> 0) & 15];
171 }
172
173 str += clen;
119 } 174 }
175 else if (is_utf8)
176 {
177 need (enc, len += clen);
178 do
179 {
180 *enc->cur++ = *str++;
181 }
182 while (--clen);
183 }
120 else 184 else
121 { 185 {
122 sprintf (enc->cur, "\\u%04x", uch); 186 need (enc, len += UTF8_MAX_LEN - 1); // never more than 11 bytes needed
123 enc->cur += 6; 187 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
188 ++str;
189 }
124 } 190 }
125 } 191 }
126 else if (is_utf8)
127 {
128 memcpy (enc->cur, str, clen);
129 enc->cur += clen;
130 }
131 else
132 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
133
134 str += clen;
135 } 192 }
136 193
137 --len; 194 --len;
138 } 195 }
139} 196}
140 197
141#define INDENT SB \ 198static void
199encode_indent (enc_t *enc)
200{
142 if (enc->flags & F_INDENT) \ 201 if (enc->flags & F_INDENT)
143 { \ 202 {
144 int i_; \ 203 int spaces = enc->indent * INDENT_STEP;
145 need (enc, enc->indent); \
146 for (i_ = enc->indent * 3; i_--; )\
147 encode_ch (enc, ' '); \
148 } \
149 SE
150 204
151#define SPACE SB if (enc->flags & F_INDENT) { need (enc, 1); encode_ch (enc, ' '); } SE 205 need (enc, spaces);
152#define NL SB if (enc->flags & F_INDENT) { need (enc, 1); encode_ch (enc, '\n'); } SE 206 memset (enc->cur, ' ', spaces);
153#define COMMA SB \ 207 enc->cur += spaces;
208 }
209}
210
211static void
212encode_space (enc_t *enc)
213{
214 need (enc, 1);
154 encode_ch (enc, ','); \ 215 encode_ch (enc, ' ');
216}
217
218static void
219encode_nl (enc_t *enc)
220{
155 if (enc->flags & F_INDENT) \ 221 if (enc->flags & F_INDENT)
156 NL; \ 222 {
223 need (enc, 1);
224 encode_ch (enc, '\n');
225 }
226}
227
228static void
229encode_comma (enc_t *enc)
230{
231 encode_ch (enc, ',');
232
233 if (enc->flags & F_INDENT)
234 encode_nl (enc);
157 else if (enc->flags & F_SPACE_AFTER) \ 235 else if (enc->flags & F_SPACE_AFTER)
158 SPACE; \ 236 encode_space (enc);
159 SE 237}
160 238
161static void encode_sv (enc_t *enc, SV *sv); 239static void encode_sv (enc_t *enc, SV *sv);
162 240
163static void 241static void
164encode_av (enc_t *enc, AV *av) 242encode_av (enc_t *enc, AV *av)
165{ 243{
166 int i, len = av_len (av); 244 int i, len = av_len (av);
167 245
168 encode_ch (enc, '['); NL; 246 encode_ch (enc, '['); encode_nl (enc);
169 ++enc->indent; 247 ++enc->indent;
170 248
171 for (i = 0; i <= len; ++i) 249 for (i = 0; i <= len; ++i)
172 { 250 {
173 INDENT; 251 encode_indent (enc);
174 encode_sv (enc, *av_fetch (av, i, 0)); 252 encode_sv (enc, *av_fetch (av, i, 0));
175 253
176 if (i < len) 254 if (i < len)
177 COMMA; 255 encode_comma (enc);
178 } 256 }
179 257
180 NL; 258 encode_nl (enc);
181 259
182 --enc->indent; 260 --enc->indent;
183 INDENT; encode_ch (enc, ']'); 261 encode_indent (enc); encode_ch (enc, ']');
184} 262}
185 263
186static void 264static void
187encode_he (enc_t *enc, HE *he) 265encode_he (enc_t *enc, HE *he)
188{ 266{
… …
190 268
191 if (HeKLEN (he) == HEf_SVKEY) 269 if (HeKLEN (he) == HEf_SVKEY)
192 { 270 {
193 SV *sv = HeSVKEY (he); 271 SV *sv = HeSVKEY (he);
194 STRLEN len; 272 STRLEN len;
273 char *str;
274
275 SvGETMAGIC (sv);
195 char *str = SvPV (sv, len); 276 str = SvPV (sv, len);
196 277
197 encode_str (enc, str, len, SvUTF8 (sv)); 278 encode_str (enc, str, len, SvUTF8 (sv));
198 } 279 }
199 else 280 else
200 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 281 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
201 282
202 encode_ch (enc, '"'); 283 encode_ch (enc, '"');
203 284
204 if (enc->flags & F_SPACE_BEFORE) SPACE; 285 if (enc->flags & F_SPACE_BEFORE) encode_space (enc);
205 encode_ch (enc, ':'); 286 encode_ch (enc, ':');
206 if (enc->flags & F_SPACE_AFTER ) SPACE; 287 if (enc->flags & F_SPACE_AFTER ) encode_space (enc);
207 encode_sv (enc, HeVAL (he)); 288 encode_sv (enc, HeVAL (he));
208} 289}
209 290
210// compare hash entries, used when all keys are bytestrings 291// compare hash entries, used when all keys are bytestrings
211static int 292static int
… …
217 HE *b = *(HE **)b_; 298 HE *b = *(HE **)b_;
218 299
219 STRLEN la = HeKLEN (a); 300 STRLEN la = HeKLEN (a);
220 STRLEN lb = HeKLEN (b); 301 STRLEN lb = HeKLEN (b);
221 302
222 if (!(cmp == memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 303 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb)))
223 cmp = la < lb ? -1 : la == lb ? 0 : 1; 304 cmp = la - lb;
224 305
225 return cmp; 306 return cmp;
226} 307}
227 308
228// compare hash entries, used when some keys are sv's or utf-x 309// compare hash entries, used when some keys are sv's or utf-x
… …
235static void 316static void
236encode_hv (enc_t *enc, HV *hv) 317encode_hv (enc_t *enc, HV *hv)
237{ 318{
238 int count, i; 319 int count, i;
239 320
240 encode_ch (enc, '{'); NL; ++enc->indent; 321 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
241 322
242 if ((count = hv_iterinit (hv))) 323 if ((count = hv_iterinit (hv)))
243 { 324 {
244 // for canonical output we have to sort by keys first 325 // for canonical output we have to sort by keys first
245 // actually, this is mostly due to the stupid so-called 326 // actually, this is mostly due to the stupid so-called
246 // security workaround added somewhere in 5.8.x. 327 // security workaround added somewhere in 5.8.x.
247 // that randomises hash orderings 328 // that randomises hash orderings
248 if (enc->flags & F_CANONICAL) 329 if (enc->flags & F_CANONICAL)
249 { 330 {
250 HE *he, *hes [count]; 331 HE *he, *hes [count]; // if your compiler dies here, you need to enable C99 mode
251 int fast = 1; 332 int fast = 1;
252 333
253 i = 0; 334 i = 0;
254 while ((he = hv_iternext (hv))) 335 while ((he = hv_iternext (hv)))
255 { 336 {
… …
262 343
263 if (fast) 344 if (fast)
264 qsort (hes, count, sizeof (HE *), he_cmp_fast); 345 qsort (hes, count, sizeof (HE *), he_cmp_fast);
265 else 346 else
266 { 347 {
267 // hack to disable "use bytes" 348 // hack to forcefully disable "use bytes"
268 COP *oldcop = PL_curcop, cop; 349 COP cop = *PL_curcop;
269 cop.op_private = 0; 350 cop.op_private = 0;
351
352 ENTER;
353 SAVETMPS;
354
355 SAVEVPTR (PL_curcop);
270 PL_curcop = &cop; 356 PL_curcop = &cop;
271 357
272 SAVETMPS;
273 qsort (hes, count, sizeof (HE *), he_cmp_slow); 358 qsort (hes, count, sizeof (HE *), he_cmp_slow);
359
274 FREETMPS; 360 FREETMPS;
275 361 LEAVE;
276 PL_curcop = oldcop;
277 } 362 }
278 363
279 for (i = 0; i < count; ++i) 364 for (i = 0; i < count; ++i)
280 { 365 {
281 INDENT; 366 encode_indent (enc);
282 encode_he (enc, hes [i]); 367 encode_he (enc, hes [i]);
283 368
284 if (i < count - 1) 369 if (i < count - 1)
285 COMMA; 370 encode_comma (enc);
286 } 371 }
287 372
288 NL; 373 encode_nl (enc);
289 } 374 }
290 else 375 else
291 { 376 {
292 SV *sv; 377 SV *sv;
293 HE *he = hv_iternext (hv); 378 HE *he = hv_iternext (hv);
294 379
295 for (;;) 380 for (;;)
296 { 381 {
297 INDENT; 382 encode_indent (enc);
298 encode_he (enc, he); 383 encode_he (enc, he);
299 384
300 if (!(he = hv_iternext (hv))) 385 if (!(he = hv_iternext (hv)))
301 break; 386 break;
302 387
303 COMMA; 388 encode_comma (enc);
304 } 389 }
305 390
306 NL; 391 encode_nl (enc);
307 } 392 }
308 } 393 }
309 394
310 --enc->indent; INDENT; encode_ch (enc, '}'); 395 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
311} 396}
312 397
313static void 398static void
314encode_sv (enc_t *enc, SV *sv) 399encode_sv (enc_t *enc, SV *sv)
315{ 400{
401 SvGETMAGIC (sv);
402
316 if (SvPOKp (sv)) 403 if (SvPOKp (sv))
317 { 404 {
318 STRLEN len; 405 STRLEN len;
319 char *str = SvPV (sv, len); 406 char *str = SvPV (sv, len);
320 encode_ch (enc, '"'); 407 encode_ch (enc, '"');
… …
335 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 422 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv))
336 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 423 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv));
337 } 424 }
338 else if (SvROK (sv)) 425 else if (SvROK (sv))
339 { 426 {
340 if (!--enc->max_recurse) 427 SV *rv = SvRV (sv);
428
429 if (enc->indent >= enc->max_depth)
341 croak ("data structure too deep (hit recursion limit)"); 430 croak ("data structure too deep (hit recursion limit)");
342 431
343 sv = SvRV (sv);
344
345 switch (SvTYPE (sv)) 432 switch (SvTYPE (rv))
346 { 433 {
347 case SVt_PVAV: encode_av (enc, (AV *)sv); break; 434 case SVt_PVAV: encode_av (enc, (AV *)rv); break;
348 case SVt_PVHV: encode_hv (enc, (HV *)sv); break; 435 case SVt_PVHV: encode_hv (enc, (HV *)rv); break;
349 436
350 default: 437 default:
351 croak ("JSON can only represent references to arrays or hashes"); 438 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
439 SvPV_nolen (sv));
352 } 440 }
353 } 441 }
354 else if (!SvOK (sv)) 442 else if (!SvOK (sv))
355 encode_str (enc, "null", 4, 0); 443 encode_str (enc, "null", 4, 0);
356 else 444 else
357 croak ("encountered perl type that JSON cannot handle"); 445 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
446 SvPV_nolen (sv), SvFLAGS (sv));
358} 447}
359 448
360static SV * 449static SV *
361encode_json (SV *scalar, UV flags) 450encode_json (SV *scalar, UV flags)
362{ 451{
452 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar))
453 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
454
363 enc_t enc; 455 enc_t enc;
364 enc.flags = flags; 456 enc.flags = flags;
365 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 457 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
366 enc.cur = SvPVX (enc.sv); 458 enc.cur = SvPVX (enc.sv);
367 enc.end = SvEND (enc.sv); 459 enc.end = SvEND (enc.sv);
368 enc.max_recurse = 0;
369 enc.indent = 0; 460 enc.indent = 0;
461 enc.max_depth = 0x7fffffffUL;
370 462
371 SvPOK_only (enc.sv); 463 SvPOK_only (enc.sv);
372 encode_sv (&enc, scalar); 464 encode_sv (&enc, scalar);
373 465
374 if (!(flags & (F_ASCII | F_UTF8))) 466 if (!(flags & (F_ASCII | F_UTF8)))
375 SvUTF8_on (enc.sv); 467 SvUTF8_on (enc.sv);
376 468
377 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 469 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
470
471 if (enc.flags & F_SHRINK)
472 shrink (enc.sv);
473
378 return enc.sv; 474 return enc.sv;
379} 475}
380 476
381///////////////////////////////////////////////////////////////////////////// 477/////////////////////////////////////////////////////////////////////////////
478// decoder
382 479
383#define WS \ 480// structure used for decoding JSON
481typedef struct
482{
483 char *cur; // current parser pointer
484 char *end; // end of input string
485 const char *err; // parse error, if != 0
486 UV flags; // F_*
487} dec_t;
488
489static void
490decode_ws (dec_t *dec)
491{
384 for (;;) \ 492 for (;;)
385 { \ 493 {
386 char ch = *dec->cur; \ 494 char ch = *dec->cur;
495
387 if (ch > 0x20 \ 496 if (ch > 0x20
388 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) \ 497 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09))
389 break; \ 498 break;
499
390 ++dec->cur; \ 500 ++dec->cur;
391 } 501 }
502}
392 503
393#define ERR(reason) SB dec->err = reason; goto fail; SE 504#define ERR(reason) SB dec->err = reason; goto fail; SE
394#define EXPECT_CH(ch) SB \ 505#define EXPECT_CH(ch) SB \
395 if (*dec->cur != ch) \ 506 if (*dec->cur != ch) \
396 ERR (# ch " expected"); \ 507 ERR (# ch " expected"); \
397 ++dec->cur; \ 508 ++dec->cur; \
398 SE 509 SE
399 510
400static SV *decode_sv (dec_t *dec); 511static SV *decode_sv (dec_t *dec);
401 512
402#define APPEND_CH(ch) SB \
403 SvGROW (sv, cur + 1 + 1); \
404 SvPVX (sv)[cur++] = (ch); \
405 SE
406
407static signed char decode_hexdigit[256]; 513static signed char decode_hexdigit[256];
408 514
409static UV 515static UV
410decode_4hex (dec_t *dec) 516decode_4hex (dec_t *dec)
411{ 517{
412 signed char d1, d2, d3, d4; 518 signed char d1, d2, d3, d4;
519 unsigned char *cur = (unsigned char *)dec->cur;
413 520
414 d1 = decode_hexdigit [((unsigned char *)dec->cur) [0]];
415 if (d1 < 0) ERR ("four hexadecimal digits expected"); 521 d1 = decode_hexdigit [cur [0]]; if (d1 < 0) ERR ("four hexadecimal digits expected");
416 d2 = decode_hexdigit [((unsigned char *)dec->cur) [1]];
417 if (d2 < 0) ERR ("four hexadecimal digits expected"); 522 d2 = decode_hexdigit [cur [1]]; if (d2 < 0) ERR ("four hexadecimal digits expected");
418 d3 = decode_hexdigit [((unsigned char *)dec->cur) [2]];
419 if (d3 < 0) ERR ("four hexadecimal digits expected"); 523 d3 = decode_hexdigit [cur [2]]; if (d3 < 0) ERR ("four hexadecimal digits expected");
420 d4 = decode_hexdigit [((unsigned char *)dec->cur) [3]];
421 if (d4 < 0) ERR ("four hexadecimal digits expected"); 524 d4 = decode_hexdigit [cur [3]]; if (d4 < 0) ERR ("four hexadecimal digits expected");
422 525
423 dec->cur += 4; 526 dec->cur += 4;
424 527
425 return ((UV)d1) << 12 528 return ((UV)d1) << 12
426 | ((UV)d2) << 8 529 | ((UV)d2) << 8
… …
432} 535}
433 536
434static SV * 537static SV *
435decode_str (dec_t *dec) 538decode_str (dec_t *dec)
436{ 539{
437 SV *sv = NEWSV (0,2); 540 SV *sv = 0;
438 STRLEN cur = 0;
439 int utf8 = 0; 541 int utf8 = 0;
440 542
441 for (;;) 543 do
442 { 544 {
443 unsigned char ch = *(unsigned char *)dec->cur; 545 char buf [SHORT_STRING_LEN + UTF8_MAX_LEN];
546 char *cur = buf;
444 547
445 if (ch == '"') 548 do
446 break;
447 else if (ch == '\\')
448 { 549 {
449 switch (*++dec->cur) 550 unsigned char ch = *(unsigned char *)dec->cur++;
551
552 if (ch == '"')
450 { 553 {
451 case '\\': 554 --dec->cur;
452 case '/': 555 break;
453 case '"': APPEND_CH (*dec->cur++); break; 556 }
454 557 else if (ch == '\\')
455 case 'b': APPEND_CH ('\010'); ++dec->cur; break; 558 {
456 case 't': APPEND_CH ('\011'); ++dec->cur; break; 559 switch (*dec->cur)
457 case 'n': APPEND_CH ('\012'); ++dec->cur; break;
458 case 'f': APPEND_CH ('\014'); ++dec->cur; break;
459 case 'r': APPEND_CH ('\015'); ++dec->cur; break;
460
461 case 'u':
462 { 560 {
463 UV lo, hi; 561 case '\\':
464 ++dec->cur; 562 case '/':
563 case '"': *cur++ = *dec->cur++; break;
465 564
466 hi = decode_4hex (dec); 565 case 'b': ++dec->cur; *cur++ = '\010'; break;
467 if (hi == (UV)-1) 566 case 't': ++dec->cur; *cur++ = '\011'; break;
468 goto fail; 567 case 'n': ++dec->cur; *cur++ = '\012'; break;
568 case 'f': ++dec->cur; *cur++ = '\014'; break;
569 case 'r': ++dec->cur; *cur++ = '\015'; break;
469 570
470 // possibly a surrogate pair 571 case 'u':
471 if (hi >= 0xd800 && hi < 0xdc00)
472 { 572 {
473 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 573 UV lo, hi;
474 ERR ("illegal surrogate character");
475
476 dec->cur += 2; 574 ++dec->cur;
477 575
478 lo = decode_4hex (dec); 576 hi = decode_4hex (dec);
479 if (lo == (UV)-1) 577 if (hi == (UV)-1)
480 goto fail; 578 goto fail;
481 579
580 // possibly a surrogate pair
581 if (hi >= 0xd800)
582 if (hi < 0xdc00)
583 {
584 if (dec->cur [0] != '\\' || dec->cur [1] != 'u')
585 ERR ("missing low surrogate character in surrogate pair");
586
587 dec->cur += 2;
588
589 lo = decode_4hex (dec);
590 if (lo == (UV)-1)
591 goto fail;
592
482 if (lo < 0xdc00 || lo >= 0xe000) 593 if (lo < 0xdc00 || lo >= 0xe000)
483 ERR ("surrogate pair expected"); 594 ERR ("surrogate pair expected");
484 595
485 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000; 596 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000;
597 }
598 else if (hi < 0xe000)
599 ERR ("missing high surrogate character in surrogate pair");
600
601 if (hi >= 0x80)
602 {
603 utf8 = 1;
604
605 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
606 }
607 else
608 *cur++ = hi;
486 } 609 }
487 else if (lo >= 0xdc00 && lo < 0xe000)
488 ERR ("illegal surrogate character");
489
490 if (hi >= 0x80)
491 { 610 break;
492 utf8 = 1;
493 611
494 SvGROW (sv, cur + 4 + 1); // at most 4 bytes for 21 bits
495 cur = (char *)uvuni_to_utf8_flags (SvPVX (sv) + cur, hi, 0) - SvPVX (sv);
496 }
497 else 612 default:
498 APPEND_CH (hi); 613 --dec->cur;
614 ERR ("illegal backslash escape sequence in string");
499 } 615 }
500 break;
501 } 616 }
617 else if (ch >= 0x20 && ch <= 0x7f)
618 *cur++ = ch;
619 else if (ch >= 0x80)
620 {
621 --dec->cur;
622
623 STRLEN clen;
624 UV uch = utf8n_to_uvuni (dec->cur, dec->end - dec->cur, &clen, UTF8_CHECK_ONLY);
625 if (clen == (STRLEN)-1)
626 ERR ("malformed UTF-8 character in JSON string");
627
628 do
629 {
630 *cur++ = *dec->cur++;
631 }
632 while (--clen);
633
634 utf8 = 1;
635 }
636 else if (!ch)
637 ERR ("unexpected end of string while parsing json string");
638 else
639 ERR ("invalid character encountered");
640
502 } 641 }
503 else if (ch >= 0x20 && ch <= 0x7f) 642 while (cur < buf + SHORT_STRING_LEN);
504 APPEND_CH (*dec->cur++); 643
505 else if (ch >= 0x80) 644 STRLEN len = cur - buf;
645
646 if (sv)
506 { 647 {
507 STRLEN clen; 648 SvGROW (sv, SvCUR (sv) + len + 1);
508 UV uch = utf8n_to_uvuni (dec->cur, dec->end - dec->cur, &clen, UTF8_CHECK_ONLY); 649 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
509 if (clen < 0) 650 SvCUR_set (sv, SvCUR (sv) + len);
510 ERR ("malformed UTF-8 character in string, cannot convert to JSON");
511
512 SvGROW (sv, cur + clen + 1); // at most 4 bytes for 21 bits
513 memcpy (SvPVX (sv) + cur, dec->cur, clen);
514 dec->cur += clen;
515 } 651 }
516 else 652 else
517 ERR ("invalid character encountered"); 653 sv = newSVpvn (buf, len);
518 } 654 }
655 while (*dec->cur != '"');
519 656
520 ++dec->cur; 657 ++dec->cur;
521 658
659 if (sv)
660 {
522 SvPOK_only (sv); 661 SvPOK_only (sv);
523
524 SvCUR_set (sv, cur);
525 *SvEND (sv) = 0; 662 *SvEND (sv) = 0;
526 663
527 if (utf8) 664 if (utf8)
528 SvUTF8_on (sv); 665 SvUTF8_on (sv);
666 }
667 else
668 sv = newSVpvn ("", 0);
529 669
530 return sv; 670 return sv;
531 671
532fail: 672fail:
533 SvREFCNT_dec (sv);
534 return 0; 673 return 0;
535} 674}
536 675
537static SV * 676static SV *
538decode_num (dec_t *dec) 677decode_num (dec_t *dec)
… …
548 { 687 {
549 ++dec->cur; 688 ++dec->cur;
550 if (*dec->cur >= '0' && *dec->cur <= '9') 689 if (*dec->cur >= '0' && *dec->cur <= '9')
551 ERR ("malformed number (leading zero must not be followed by another digit)"); 690 ERR ("malformed number (leading zero must not be followed by another digit)");
552 } 691 }
553 692 else if (*dec->cur < '0' || *dec->cur > '9')
554 // int 693 ERR ("malformed number (no digits after initial minus)");
694 else
695 do
696 {
697 ++dec->cur;
698 }
555 while (*dec->cur >= '0' && *dec->cur <= '9') 699 while (*dec->cur >= '0' && *dec->cur <= '9');
556 ++dec->cur;
557 700
558 // [frac] 701 // [frac]
559 if (*dec->cur == '.') 702 if (*dec->cur == '.')
560 { 703 {
561 is_nv = 1; 704 ++dec->cur;
705
706 if (*dec->cur < '0' || *dec->cur > '9')
707 ERR ("malformed number (no digits after decimal point)");
562 708
563 do 709 do
564 { 710 {
565 ++dec->cur; 711 ++dec->cur;
566 } 712 }
567 while (*dec->cur >= '0' && *dec->cur <= '9'); 713 while (*dec->cur >= '0' && *dec->cur <= '9');
714
715 is_nv = 1;
568 } 716 }
569 717
570 // [exp] 718 // [exp]
571 if (*dec->cur == 'e' || *dec->cur == 'E') 719 if (*dec->cur == 'e' || *dec->cur == 'E')
572 { 720 {
573 is_nv = 1;
574
575 ++dec->cur; 721 ++dec->cur;
722
576 if (*dec->cur == '-' || *dec->cur == '+') 723 if (*dec->cur == '-' || *dec->cur == '+')
577 ++dec->cur; 724 ++dec->cur;
578 725
726 if (*dec->cur < '0' || *dec->cur > '9')
727 ERR ("malformed number (no digits after exp sign)");
728
729 do
730 {
731 ++dec->cur;
732 }
579 while (*dec->cur >= '0' && *dec->cur <= '9') 733 while (*dec->cur >= '0' && *dec->cur <= '9');
580 ++dec->cur; 734
735 is_nv = 1;
581 } 736 }
582 737
583 if (!is_nv) 738 if (!is_nv)
584 { 739 {
585 UV uv; 740 UV uv;
… …
603static SV * 758static SV *
604decode_av (dec_t *dec) 759decode_av (dec_t *dec)
605{ 760{
606 AV *av = newAV (); 761 AV *av = newAV ();
607 762
763 decode_ws (dec);
764 if (*dec->cur == ']')
765 ++dec->cur;
766 else
608 for (;;) 767 for (;;)
609 { 768 {
610 SV *value; 769 SV *value;
611 770
612 value = decode_sv (dec); 771 value = decode_sv (dec);
613 if (!value) 772 if (!value)
614 goto fail; 773 goto fail;
615 774
616 av_push (av, value); 775 av_push (av, value);
617 776
618 WS; 777 decode_ws (dec);
619 778
620 if (*dec->cur == ']') 779 if (*dec->cur == ']')
621 { 780 {
622 ++dec->cur; 781 ++dec->cur;
623 break; 782 break;
783 }
624 } 784
625
626 if (*dec->cur != ',') 785 if (*dec->cur != ',')
627 ERR (", or ] expected while parsing array"); 786 ERR (", or ] expected while parsing array");
628 787
629 ++dec->cur; 788 ++dec->cur;
630 } 789 }
631 790
632 return newRV_noinc ((SV *)av); 791 return newRV_noinc ((SV *)av);
633 792
634fail: 793fail:
635 SvREFCNT_dec (av); 794 SvREFCNT_dec (av);
… …
639static SV * 798static SV *
640decode_hv (dec_t *dec) 799decode_hv (dec_t *dec)
641{ 800{
642 HV *hv = newHV (); 801 HV *hv = newHV ();
643 802
803 decode_ws (dec);
804 if (*dec->cur == '}')
805 ++dec->cur;
806 else
644 for (;;) 807 for (;;)
645 { 808 {
646 SV *key, *value; 809 SV *key, *value;
647 810
648 WS; EXPECT_CH ('"'); 811 decode_ws (dec); EXPECT_CH ('"');
649 812
650 key = decode_str (dec); 813 key = decode_str (dec);
651 if (!key) 814 if (!key)
652 goto fail;
653
654 WS; EXPECT_CH (':');
655
656 value = decode_sv (dec);
657 if (!value)
658 {
659 SvREFCNT_dec (key);
660 goto fail; 815 goto fail;
816
817 decode_ws (dec); EXPECT_CH (':');
818
819 value = decode_sv (dec);
820 if (!value)
821 {
822 SvREFCNT_dec (key);
823 goto fail;
661 } 824 }
662 825
663 //TODO: optimise 826 //TODO: optimise
664 hv_store_ent (hv, key, value, 0); 827 hv_store_ent (hv, key, value, 0);
665 828
666 WS; 829 decode_ws (dec);
667 830
668 if (*dec->cur == '}') 831 if (*dec->cur == '}')
669 { 832 {
670 ++dec->cur; 833 ++dec->cur;
671 break; 834 break;
672 } 835 }
673 836
674 if (*dec->cur != ',') 837 if (*dec->cur != ',')
675 ERR (", or } expected while parsing object/hash"); 838 ERR (", or } expected while parsing object/hash");
676 839
677 ++dec->cur; 840 ++dec->cur;
678 } 841 }
679 842
680 return newRV_noinc ((SV *)hv); 843 return newRV_noinc ((SV *)hv);
681 844
682fail: 845fail:
683 SvREFCNT_dec (hv); 846 SvREFCNT_dec (hv);
… …
685} 848}
686 849
687static SV * 850static SV *
688decode_sv (dec_t *dec) 851decode_sv (dec_t *dec)
689{ 852{
690 WS; 853 decode_ws (dec);
691 switch (*dec->cur) 854 switch (*dec->cur)
692 { 855 {
693 case '"': ++dec->cur; return decode_str (dec); 856 case '"': ++dec->cur; return decode_str (dec);
694 case '[': ++dec->cur; return decode_av (dec); 857 case '[': ++dec->cur; return decode_av (dec);
695 case '{': ++dec->cur; return decode_hv (dec); 858 case '{': ++dec->cur; return decode_hv (dec);
… …
723 886
724 case 'n': 887 case 'n':
725 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4)) 888 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4))
726 { 889 {
727 dec->cur += 4; 890 dec->cur += 4;
728 return newSViv (1); 891 return newSVsv (&PL_sv_undef);
729 } 892 }
730 else 893 else
731 ERR ("'null' expected"); 894 ERR ("'null' expected");
732 895
733 break; 896 break;
734 897
735 default: 898 default:
736 ERR ("malformed json string"); 899 ERR ("malformed json string, neither array, object, number, string or atom");
737 break; 900 break;
738 } 901 }
739 902
740fail: 903fail:
741 return 0; 904 return 0;
… …
744static SV * 907static SV *
745decode_json (SV *string, UV flags) 908decode_json (SV *string, UV flags)
746{ 909{
747 SV *sv; 910 SV *sv;
748 911
749 if (!(flags & F_UTF8)) 912 if (flags & F_UTF8)
913 sv_utf8_downgrade (string, 0);
914 else
750 sv_utf8_upgrade (string); 915 sv_utf8_upgrade (string);
751 916
752 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 917 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
753 918
754 dec_t dec; 919 dec_t dec;
755 dec.flags = flags; 920 dec.flags = flags;
756 dec.cur = SvPVX (string); 921 dec.cur = SvPVX (string);
757 dec.end = SvEND (string); 922 dec.end = SvEND (string);
758 dec.err = 0; 923 dec.err = 0;
759 924
760 *dec.end = 1; // invalid anywhere
761 sv = decode_sv (&dec); 925 sv = decode_sv (&dec);
762 *dec.end = 0;
763 926
764 if (!sv) 927 if (!sv)
765 { 928 {
929 IV offset = dec.flags & F_UTF8
930 ? dec.cur - SvPVX (string)
766 IV offset = utf8_distance (dec.cur, SvPVX (string)); 931 : utf8_distance (dec.cur, SvPVX (string));
767 SV *uni = sv_newmortal (); 932 SV *uni = sv_newmortal ();
768 933
934 // horrible hack to silence warning inside pv_uni_display
935 COP cop = *PL_curcop;
936 cop.cop_warnings = pWARN_NONE;
937 ENTER;
938 SAVEVPTR (PL_curcop);
939 PL_curcop = &cop;
769 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 940 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
941 LEAVE;
942
770 croak ("%s, at character %d (%s)", 943 croak ("%s, at character offset %d (%s)",
771 dec.err, 944 dec.err,
772 (int)offset, 945 (int)offset,
773 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 946 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
774 } 947 }
775 948
776 sv_dump (sv);//D
777 return sv_2mortal (sv); 949 sv = sv_2mortal (sv);
950
951 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv))
952 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
953
954 return sv;
778} 955}
956
957/////////////////////////////////////////////////////////////////////////////
958// XS interface functions
779 959
780MODULE = JSON::XS PACKAGE = JSON::XS 960MODULE = JSON::XS PACKAGE = JSON::XS
781 961
782BOOT: 962BOOT:
783{ 963{
… …
785 965
786 memset (decode_hexdigit, 0xff, 256); 966 memset (decode_hexdigit, 0xff, 256);
787 for (i = 10; i--; ) 967 for (i = 10; i--; )
788 decode_hexdigit ['0' + i] = i; 968 decode_hexdigit ['0' + i] = i;
789 969
790 for (i = 6; --i; ) 970 for (i = 7; i--; )
791 { 971 {
792 decode_hexdigit ['a' + i] = 10 + i; 972 decode_hexdigit ['a' + i] = 10 + i;
793 decode_hexdigit ['A' + i] = 10 + i; 973 decode_hexdigit ['A' + i] = 10 + i;
794 } 974 }
795 975
796 json_stash = gv_stashpv ("JSON::XS", 1); 976 json_stash = gv_stashpv ("JSON::XS", 1);
797} 977}
978
979PROTOTYPES: DISABLE
798 980
799SV *new (char *dummy) 981SV *new (char *dummy)
800 CODE: 982 CODE:
801 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 983 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash);
802 OUTPUT: 984 OUTPUT:
803 RETVAL 985 RETVAL
804 986
805SV *ascii (SV *self, int enable) 987SV *ascii (SV *self, int enable = 1)
806 ALIAS: 988 ALIAS:
807 ascii = F_ASCII 989 ascii = F_ASCII
808 utf8 = F_UTF8 990 utf8 = F_UTF8
809 indent = F_INDENT 991 indent = F_INDENT
810 canonical = F_CANONICAL 992 canonical = F_CANONICAL
811 space_before = F_SPACE_BEFORE 993 space_before = F_SPACE_BEFORE
812 space_after = F_SPACE_AFTER 994 space_after = F_SPACE_AFTER
813 json_rpc = F_JSON_RPC 995 pretty = F_PRETTY
996 allow_nonref = F_ALLOW_NONREF
997 shrink = F_SHRINK
814 CODE: 998 CODE:
815{ 999{
816 UV *uv = SvJSON (self); 1000 UV *uv = SvJSON (self);
817 if (enable) 1001 if (enable)
818 *uv |= ix; 1002 *uv |= ix;
… …
826 1010
827void encode (SV *self, SV *scalar) 1011void encode (SV *self, SV *scalar)
828 PPCODE: 1012 PPCODE:
829 XPUSHs (encode_json (scalar, *SvJSON (self))); 1013 XPUSHs (encode_json (scalar, *SvJSON (self)));
830 1014
831void decode (SV *self, SV *jsondata) 1015void decode (SV *self, SV *jsonstr)
832 PPCODE: 1016 PPCODE:
833 XPUSHs (decode_json (jsondata, *SvJSON (self))); 1017 XPUSHs (decode_json (jsonstr, *SvJSON (self)));
834 1018
1019PROTOTYPES: ENABLE
1020
1021void to_json (SV *scalar)
1022 PPCODE:
1023 XPUSHs (encode_json (scalar, F_UTF8));
1024
1025void from_json (SV *jsonstr)
1026 PPCODE:
1027 XPUSHs (decode_json (jsonstr, F_UTF8));
1028

Diff Legend

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