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 1.17 by root, Sun Mar 25 02:46:41 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// F_SKIPINVALID?
19// F_EXECCODEREF?
20// F_SELFCONVERT? <=> { &__class__ => }
21
22#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
17#define F_DEFAULT 0 23#define F_DEFAULT 0
18 24
19#define INIT_SIZE 32 // initial scalar size to be allocated 25#define INIT_SIZE 32 // initial scalar size to be allocated
26#define INDENT_STEP 3 // spaces per indentation level
27
28#define UTF8_MAX_LEN 11 // for perls UTF-X: max. number of octets per character
29#define SHORT_STRING_LEN 512 // special-case strings of up to this size
20 30
21#define SB do { 31#define SB do {
22#define SE } while (0) 32#define SE } while (0)
23 33
24static HV *json_stash; 34static HV *json_stash; // JSON::XS::
35
36/////////////////////////////////////////////////////////////////////////////
37// utility functions
38
39static UV *
40SvJSON (SV *sv)
41{
42 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash))
43 croak ("object is not of type JSON::XS");
44
45 return &SvUVX (SvRV (sv));
46}
47
48static void
49shrink (SV *sv)
50{
51 sv_utf8_downgrade (sv, 1);
52 if (SvLEN (sv) > SvCUR (sv) + 1)
53 {
54#ifdef SvPV_shrink_to_cur
55 SvPV_shrink_to_cur (sv);
56#elif defined (SvPV_renew)
57 SvPV_renew (sv, SvCUR (sv) + 1);
58#endif
59 }
60}
61
62// decode an utf-8 character and return it, or (UV)-1 in
63// case of an error.
64// we special-case "safe" characters from U+80 .. U+7FF,
65// but use the very good perl function to parse anything else.
66// note that we never call this function for a ascii codepoints
67static UV
68decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
69{
70 if (s[0] > 0xdf || s[0] < 0xc2)
71 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
72 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)
73 {
74 *clen = 2;
75 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
76 }
77 else
78 return (UV)-1;
79}
80
81/////////////////////////////////////////////////////////////////////////////
82// encoder
25 83
26// structure used for encoding JSON 84// structure used for encoding JSON
27typedef struct 85typedef struct
28{ 86{
29 char *cur; 87 char *cur; // SvPVX (sv) + current output position
30 STRLEN len; // SvLEN (sv)
31 char *end; // SvEND (sv) 88 char *end; // SvEND (sv)
32 SV *sv; 89 SV *sv; // result scalar
33 UV flags; 90 UV flags; // F_*
34 int max_recurse; 91 int indent; // indentation level
35 int indent; 92 int max_depth; // max. recursion level
36} enc_t; 93} 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 94
58static void 95static void
59need (enc_t *enc, STRLEN len) 96need (enc_t *enc, STRLEN len)
60{ 97{
61 if (enc->cur + len >= enc->end) 98 if (enc->cur + len >= enc->end)
62 { 99 {
63 STRLEN cur = enc->cur - SvPVX (enc->sv); 100 STRLEN cur = enc->cur - SvPVX (enc->sv);
64 SvGROW (enc->sv, cur + len + 1); 101 SvGROW (enc->sv, cur + len + 1);
65 enc->cur = SvPVX (enc->sv) + cur; 102 enc->cur = SvPVX (enc->sv) + cur;
66 enc->end = SvEND (enc->sv); 103 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv);
67 } 104 }
68} 105}
69 106
70static void 107static void
71encode_ch (enc_t *enc, char ch) 108encode_ch (enc_t *enc, char ch)
… …
77static void 114static void
78encode_str (enc_t *enc, char *str, STRLEN len, int is_utf8) 115encode_str (enc_t *enc, char *str, STRLEN len, int is_utf8)
79{ 116{
80 char *end = str + len; 117 char *end = str + len;
81 118
119 need (enc, len);
120
82 while (str < end) 121 while (str < end)
83 { 122 {
84 unsigned char ch = *(unsigned char *)str; 123 unsigned char ch = *(unsigned char *)str;
124
85 if (ch >= 0x20 && ch < 0x80) // most common case 125 if (ch >= 0x20 && ch < 0x80) // most common case
86 { 126 {
127 if (ch == '"') // but with slow exceptions
128 {
129 need (enc, len += 1);
130 *enc->cur++ = '\\';
131 *enc->cur++ = '"';
132 }
133 else if (ch == '\\')
134 {
135 need (enc, len += 1);
136 *enc->cur++ = '\\';
137 *enc->cur++ = '\\';
138 }
139 else
87 *enc->cur++ = ch; 140 *enc->cur++ = ch;
141
88 str++; 142 ++str;
89 } 143 }
90 else 144 else
91 { 145 {
92 STRLEN clen; 146 switch (ch)
93 UV uch;
94
95 if (is_utf8)
96 { 147 {
97 uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY); 148 case '\010': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'b'; ++str; break;
98 if (clen < 0) 149 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"); 150 case '\012': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'n'; ++str; break;
100 } 151 case '\014': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'f'; ++str; break;
101 else 152 case '\015': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'r'; ++str; break;
102 {
103 uch = ch;
104 clen = 1;
105 }
106 153
107 need (enc, len += 6); 154 default:
108
109 if (uch < 0xa0 || enc->flags & F_ASCII)
110 {
111 if (uch > 0xFFFFUL)
112 { 155 {
156 STRLEN clen;
157 UV uch;
158
159 if (is_utf8)
160 {
161 //uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
162 uch = decode_utf8 (str, end - str, &clen);
163 if (clen == (STRLEN)-1)
164 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
165 }
166 else
167 {
168 uch = ch;
113 len += 6; 169 clen = 1;
170 }
171
172 if (uch > 0x10FFFFUL)
173 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
174
175 if (uch < 0x80 || enc->flags & F_ASCII)
176 {
177 if (uch > 0xFFFFUL)
178 {
114 need (enc, len += 6); 179 need (enc, len += 11);
115 sprintf (enc->cur, "\\u%04x\\u%04x", 180 sprintf (enc->cur, "\\u%04x\\u%04x",
116 (uch - 0x10000) / 0x400 + 0xD800, 181 (int)((uch - 0x10000) / 0x400 + 0xD800),
117 (uch - 0x10000) % 0x400 + 0xDC00); 182 (int)((uch - 0x10000) % 0x400 + 0xDC00));
118 enc->cur += 12; 183 enc->cur += 12;
184 }
185 else
186 {
187 static char hexdigit [16] = "0123456789abcdef";
188 need (enc, len += 5);
189 *enc->cur++ = '\\';
190 *enc->cur++ = 'u';
191 *enc->cur++ = hexdigit [ uch >> 12 ];
192 *enc->cur++ = hexdigit [(uch >> 8) & 15];
193 *enc->cur++ = hexdigit [(uch >> 4) & 15];
194 *enc->cur++ = hexdigit [(uch >> 0) & 15];
195 }
196
197 str += clen;
119 } 198 }
199 else if (is_utf8)
200 {
201 need (enc, len += clen);
202 do
203 {
204 *enc->cur++ = *str++;
205 }
206 while (--clen);
207 }
120 else 208 else
121 { 209 {
122 sprintf (enc->cur, "\\u%04x", uch); 210 need (enc, len += UTF8_MAX_LEN - 1); // never more than 11 bytes needed
123 enc->cur += 6; 211 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
212 ++str;
213 }
124 } 214 }
125 } 215 }
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 } 216 }
136 217
137 --len; 218 --len;
138 } 219 }
139} 220}
140 221
141#define INDENT SB \ 222static void
223encode_indent (enc_t *enc)
224{
142 if (enc->flags & F_INDENT) \ 225 if (enc->flags & F_INDENT)
143 { \ 226 {
144 int i_; \ 227 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
150 228
151#define SPACE SB if (enc->flags & F_INDENT) { need (enc, 1); encode_ch (enc, ' '); } SE 229 need (enc, spaces);
152#define NL SB if (enc->flags & F_INDENT) { need (enc, 1); encode_ch (enc, '\n'); } SE 230 memset (enc->cur, ' ', spaces);
153#define COMMA SB \ 231 enc->cur += spaces;
232 }
233}
234
235static void
236encode_space (enc_t *enc)
237{
238 need (enc, 1);
154 encode_ch (enc, ','); \ 239 encode_ch (enc, ' ');
240}
241
242static void
243encode_nl (enc_t *enc)
244{
155 if (enc->flags & F_INDENT) \ 245 if (enc->flags & F_INDENT)
156 NL; \ 246 {
247 need (enc, 1);
248 encode_ch (enc, '\n');
249 }
250}
251
252static void
253encode_comma (enc_t *enc)
254{
255 encode_ch (enc, ',');
256
257 if (enc->flags & F_INDENT)
258 encode_nl (enc);
157 else if (enc->flags & F_SPACE_AFTER) \ 259 else if (enc->flags & F_SPACE_AFTER)
158 SPACE; \ 260 encode_space (enc);
159 SE 261}
160 262
161static void encode_sv (enc_t *enc, SV *sv); 263static void encode_sv (enc_t *enc, SV *sv);
162 264
163static void 265static void
164encode_av (enc_t *enc, AV *av) 266encode_av (enc_t *enc, AV *av)
165{ 267{
166 int i, len = av_len (av); 268 int i, len = av_len (av);
167 269
168 encode_ch (enc, '['); NL; 270 encode_ch (enc, '['); encode_nl (enc);
169 ++enc->indent; 271 ++enc->indent;
170 272
171 for (i = 0; i <= len; ++i) 273 for (i = 0; i <= len; ++i)
172 { 274 {
173 INDENT; 275 encode_indent (enc);
174 encode_sv (enc, *av_fetch (av, i, 0)); 276 encode_sv (enc, *av_fetch (av, i, 0));
175 277
176 if (i < len) 278 if (i < len)
177 COMMA; 279 encode_comma (enc);
178 } 280 }
179 281
180 NL; 282 encode_nl (enc);
181 283
182 --enc->indent; 284 --enc->indent;
183 INDENT; encode_ch (enc, ']'); 285 encode_indent (enc); encode_ch (enc, ']');
184} 286}
185 287
186static void 288static void
187encode_he (enc_t *enc, HE *he) 289encode_he (enc_t *enc, HE *he)
188{ 290{
… …
190 292
191 if (HeKLEN (he) == HEf_SVKEY) 293 if (HeKLEN (he) == HEf_SVKEY)
192 { 294 {
193 SV *sv = HeSVKEY (he); 295 SV *sv = HeSVKEY (he);
194 STRLEN len; 296 STRLEN len;
297 char *str;
298
299 SvGETMAGIC (sv);
195 char *str = SvPV (sv, len); 300 str = SvPV (sv, len);
196 301
197 encode_str (enc, str, len, SvUTF8 (sv)); 302 encode_str (enc, str, len, SvUTF8 (sv));
198 } 303 }
199 else 304 else
200 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 305 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
201 306
202 encode_ch (enc, '"'); 307 encode_ch (enc, '"');
203 308
204 if (enc->flags & F_SPACE_BEFORE) SPACE; 309 if (enc->flags & F_SPACE_BEFORE) encode_space (enc);
205 encode_ch (enc, ':'); 310 encode_ch (enc, ':');
206 if (enc->flags & F_SPACE_AFTER ) SPACE; 311 if (enc->flags & F_SPACE_AFTER ) encode_space (enc);
207 encode_sv (enc, HeVAL (he)); 312 encode_sv (enc, HeVAL (he));
208} 313}
209 314
210// compare hash entries, used when all keys are bytestrings 315// compare hash entries, used when all keys are bytestrings
211static int 316static int
… …
217 HE *b = *(HE **)b_; 322 HE *b = *(HE **)b_;
218 323
219 STRLEN la = HeKLEN (a); 324 STRLEN la = HeKLEN (a);
220 STRLEN lb = HeKLEN (b); 325 STRLEN lb = HeKLEN (b);
221 326
222 if (!(cmp == memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 327 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb)))
223 cmp = la < lb ? -1 : la == lb ? 0 : 1; 328 cmp = la - lb;
224 329
225 return cmp; 330 return cmp;
226} 331}
227 332
228// compare hash entries, used when some keys are sv's or utf-x 333// compare hash entries, used when some keys are sv's or utf-x
… …
235static void 340static void
236encode_hv (enc_t *enc, HV *hv) 341encode_hv (enc_t *enc, HV *hv)
237{ 342{
238 int count, i; 343 int count, i;
239 344
240 encode_ch (enc, '{'); NL; ++enc->indent; 345 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
241 346
242 if ((count = hv_iterinit (hv))) 347 if ((count = hv_iterinit (hv)))
243 { 348 {
244 // for canonical output we have to sort by keys first 349 // for canonical output we have to sort by keys first
245 // actually, this is mostly due to the stupid so-called 350 // actually, this is mostly due to the stupid so-called
246 // security workaround added somewhere in 5.8.x. 351 // security workaround added somewhere in 5.8.x.
247 // that randomises hash orderings 352 // that randomises hash orderings
248 if (enc->flags & F_CANONICAL) 353 if (enc->flags & F_CANONICAL)
249 { 354 {
250 HE *he, *hes [count]; 355 HE *he, *hes [count]; // if your compiler dies here, you need to enable C99 mode
251 int fast = 1; 356 int fast = 1;
252 357
253 i = 0; 358 i = 0;
254 while ((he = hv_iternext (hv))) 359 while ((he = hv_iternext (hv)))
255 { 360 {
… …
262 367
263 if (fast) 368 if (fast)
264 qsort (hes, count, sizeof (HE *), he_cmp_fast); 369 qsort (hes, count, sizeof (HE *), he_cmp_fast);
265 else 370 else
266 { 371 {
267 // hack to disable "use bytes" 372 // hack to forcefully disable "use bytes"
268 COP *oldcop = PL_curcop, cop; 373 COP cop = *PL_curcop;
269 cop.op_private = 0; 374 cop.op_private = 0;
375
376 ENTER;
377 SAVETMPS;
378
379 SAVEVPTR (PL_curcop);
270 PL_curcop = &cop; 380 PL_curcop = &cop;
271 381
272 SAVETMPS;
273 qsort (hes, count, sizeof (HE *), he_cmp_slow); 382 qsort (hes, count, sizeof (HE *), he_cmp_slow);
383
274 FREETMPS; 384 FREETMPS;
275 385 LEAVE;
276 PL_curcop = oldcop;
277 } 386 }
278 387
279 for (i = 0; i < count; ++i) 388 for (i = 0; i < count; ++i)
280 { 389 {
281 INDENT; 390 encode_indent (enc);
282 encode_he (enc, hes [i]); 391 encode_he (enc, hes [i]);
283 392
284 if (i < count - 1) 393 if (i < count - 1)
285 COMMA; 394 encode_comma (enc);
286 } 395 }
287 396
288 NL; 397 encode_nl (enc);
289 } 398 }
290 else 399 else
291 { 400 {
292 SV *sv; 401 SV *sv;
293 HE *he = hv_iternext (hv); 402 HE *he = hv_iternext (hv);
294 403
295 for (;;) 404 for (;;)
296 { 405 {
297 INDENT; 406 encode_indent (enc);
298 encode_he (enc, he); 407 encode_he (enc, he);
299 408
300 if (!(he = hv_iternext (hv))) 409 if (!(he = hv_iternext (hv)))
301 break; 410 break;
302 411
303 COMMA; 412 encode_comma (enc);
304 } 413 }
305 414
306 NL; 415 encode_nl (enc);
307 } 416 }
308 } 417 }
309 418
310 --enc->indent; INDENT; encode_ch (enc, '}'); 419 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
311} 420}
312 421
313static void 422static void
314encode_sv (enc_t *enc, SV *sv) 423encode_sv (enc_t *enc, SV *sv)
315{ 424{
425 SvGETMAGIC (sv);
426
316 if (SvPOKp (sv)) 427 if (SvPOKp (sv))
317 { 428 {
318 STRLEN len; 429 STRLEN len;
319 char *str = SvPV (sv, len); 430 char *str = SvPV (sv, len);
320 encode_ch (enc, '"'); 431 encode_ch (enc, '"');
… …
335 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 446 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv))
336 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 447 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv));
337 } 448 }
338 else if (SvROK (sv)) 449 else if (SvROK (sv))
339 { 450 {
340 if (!--enc->max_recurse) 451 SV *rv = SvRV (sv);
452
453 if (enc->indent >= enc->max_depth)
341 croak ("data structure too deep (hit recursion limit)"); 454 croak ("data structure too deep (hit recursion limit)");
342 455
343 sv = SvRV (sv);
344
345 switch (SvTYPE (sv)) 456 switch (SvTYPE (rv))
346 { 457 {
347 case SVt_PVAV: encode_av (enc, (AV *)sv); break; 458 case SVt_PVAV: encode_av (enc, (AV *)rv); break;
348 case SVt_PVHV: encode_hv (enc, (HV *)sv); break; 459 case SVt_PVHV: encode_hv (enc, (HV *)rv); break;
349 460
350 default: 461 default:
351 croak ("JSON can only represent references to arrays or hashes"); 462 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
463 SvPV_nolen (sv));
352 } 464 }
353 } 465 }
354 else if (!SvOK (sv)) 466 else if (!SvOK (sv))
355 encode_str (enc, "null", 4, 0); 467 encode_str (enc, "null", 4, 0);
356 else 468 else
357 croak ("encountered perl type that JSON cannot handle"); 469 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
470 SvPV_nolen (sv), SvFLAGS (sv));
358} 471}
359 472
360static SV * 473static SV *
361encode_json (SV *scalar, UV flags) 474encode_json (SV *scalar, UV flags)
362{ 475{
476 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar))
477 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
478
363 enc_t enc; 479 enc_t enc;
364 enc.flags = flags; 480 enc.flags = flags;
365 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 481 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
366 enc.cur = SvPVX (enc.sv); 482 enc.cur = SvPVX (enc.sv);
367 enc.end = SvEND (enc.sv); 483 enc.end = SvEND (enc.sv);
368 enc.max_recurse = 0;
369 enc.indent = 0; 484 enc.indent = 0;
485 enc.max_depth = 0x7fffffffUL;
370 486
371 SvPOK_only (enc.sv); 487 SvPOK_only (enc.sv);
372 encode_sv (&enc, scalar); 488 encode_sv (&enc, scalar);
373 489
374 if (!(flags & (F_ASCII | F_UTF8))) 490 if (!(flags & (F_ASCII | F_UTF8)))
375 SvUTF8_on (enc.sv); 491 SvUTF8_on (enc.sv);
376 492
377 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 493 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
494
495 if (enc.flags & F_SHRINK)
496 shrink (enc.sv);
497
378 return enc.sv; 498 return enc.sv;
379} 499}
380 500
381///////////////////////////////////////////////////////////////////////////// 501/////////////////////////////////////////////////////////////////////////////
502// decoder
382 503
383#define WS \ 504// structure used for decoding JSON
505typedef struct
506{
507 char *cur; // current parser pointer
508 char *end; // end of input string
509 const char *err; // parse error, if != 0
510 UV flags; // F_*
511} dec_t;
512
513static void
514decode_ws (dec_t *dec)
515{
384 for (;;) \ 516 for (;;)
385 { \ 517 {
386 char ch = *dec->cur; \ 518 char ch = *dec->cur;
519
387 if (ch > 0x20 \ 520 if (ch > 0x20
388 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) \ 521 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09))
389 break; \ 522 break;
523
390 ++dec->cur; \ 524 ++dec->cur;
391 } 525 }
526}
392 527
393#define ERR(reason) SB dec->err = reason; goto fail; SE 528#define ERR(reason) SB dec->err = reason; goto fail; SE
394#define EXPECT_CH(ch) SB \ 529#define EXPECT_CH(ch) SB \
395 if (*dec->cur != ch) \ 530 if (*dec->cur != ch) \
396 ERR (# ch " expected"); \ 531 ERR (# ch " expected"); \
397 ++dec->cur; \ 532 ++dec->cur; \
398 SE 533 SE
399 534
400static SV *decode_sv (dec_t *dec); 535static SV *decode_sv (dec_t *dec);
401 536
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]; 537static signed char decode_hexdigit[256];
408 538
409static UV 539static UV
410decode_4hex (dec_t *dec) 540decode_4hex (dec_t *dec)
411{ 541{
412 signed char d1, d2, d3, d4; 542 signed char d1, d2, d3, d4;
543 unsigned char *cur = (unsigned char *)dec->cur;
413 544
414 d1 = decode_hexdigit [((unsigned char *)dec->cur) [0]];
415 if (d1 < 0) ERR ("four hexadecimal digits expected"); 545 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"); 546 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"); 547 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"); 548 d4 = decode_hexdigit [cur [3]]; if (d4 < 0) ERR ("four hexadecimal digits expected");
422 549
423 dec->cur += 4; 550 dec->cur += 4;
424 551
425 return ((UV)d1) << 12 552 return ((UV)d1) << 12
426 | ((UV)d2) << 8 553 | ((UV)d2) << 8
… …
432} 559}
433 560
434static SV * 561static SV *
435decode_str (dec_t *dec) 562decode_str (dec_t *dec)
436{ 563{
437 SV *sv = NEWSV (0,2); 564 SV *sv = 0;
438 STRLEN cur = 0;
439 int utf8 = 0; 565 int utf8 = 0;
440 566
441 for (;;) 567 do
442 { 568 {
443 unsigned char ch = *(unsigned char *)dec->cur; 569 char buf [SHORT_STRING_LEN + UTF8_MAX_LEN];
570 char *cur = buf;
444 571
445 if (ch == '"') 572 do
446 break;
447 else if (ch == '\\')
448 { 573 {
449 switch (*++dec->cur) 574 unsigned char ch = *(unsigned char *)dec->cur++;
575
576 if (ch == '"')
450 { 577 {
451 case '\\': 578 --dec->cur;
452 case '/': 579 break;
453 case '"': APPEND_CH (*dec->cur++); break; 580 }
454 581 else if (ch == '\\')
455 case 'b': APPEND_CH ('\010'); ++dec->cur; break; 582 {
456 case 't': APPEND_CH ('\011'); ++dec->cur; break; 583 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 { 584 {
463 UV lo, hi; 585 case '\\':
464 ++dec->cur; 586 case '/':
587 case '"': *cur++ = *dec->cur++; break;
465 588
466 hi = decode_4hex (dec); 589 case 'b': ++dec->cur; *cur++ = '\010'; break;
467 if (hi == (UV)-1) 590 case 't': ++dec->cur; *cur++ = '\011'; break;
468 goto fail; 591 case 'n': ++dec->cur; *cur++ = '\012'; break;
592 case 'f': ++dec->cur; *cur++ = '\014'; break;
593 case 'r': ++dec->cur; *cur++ = '\015'; break;
469 594
470 // possibly a surrogate pair 595 case 'u':
471 if (hi >= 0xd800 && hi < 0xdc00)
472 { 596 {
473 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 597 UV lo, hi;
474 ERR ("illegal surrogate character");
475
476 dec->cur += 2; 598 ++dec->cur;
477 599
478 lo = decode_4hex (dec); 600 hi = decode_4hex (dec);
479 if (lo == (UV)-1) 601 if (hi == (UV)-1)
480 goto fail; 602 goto fail;
481 603
604 // possibly a surrogate pair
605 if (hi >= 0xd800)
606 if (hi < 0xdc00)
607 {
608 if (dec->cur [0] != '\\' || dec->cur [1] != 'u')
609 ERR ("missing low surrogate character in surrogate pair");
610
611 dec->cur += 2;
612
613 lo = decode_4hex (dec);
614 if (lo == (UV)-1)
615 goto fail;
616
482 if (lo < 0xdc00 || lo >= 0xe000) 617 if (lo < 0xdc00 || lo >= 0xe000)
483 ERR ("surrogate pair expected"); 618 ERR ("surrogate pair expected");
484 619
485 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000; 620 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000;
621 }
622 else if (hi < 0xe000)
623 ERR ("missing high surrogate character in surrogate pair");
624
625 if (hi >= 0x80)
626 {
627 utf8 = 1;
628
629 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
630 }
631 else
632 *cur++ = hi;
486 } 633 }
487 else if (lo >= 0xdc00 && lo < 0xe000)
488 ERR ("illegal surrogate character");
489
490 if (hi >= 0x80)
491 { 634 break;
492 utf8 = 1;
493 635
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 636 default:
498 APPEND_CH (hi); 637 --dec->cur;
638 ERR ("illegal backslash escape sequence in string");
499 } 639 }
500 break;
501 } 640 }
641 else if (ch >= 0x20 && ch <= 0x7f)
642 *cur++ = ch;
643 else if (ch >= 0x80)
644 {
645 --dec->cur;
646
647 STRLEN clen;
648 UV uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen);
649 if (clen == (STRLEN)-1)
650 ERR ("malformed UTF-8 character in JSON string");
651
652 do
653 {
654 *cur++ = *dec->cur++;
655 }
656 while (--clen);
657
658 utf8 = 1;
659 }
660 else if (!ch)
661 ERR ("unexpected end of string while parsing json string");
662 else
663 ERR ("invalid character encountered");
664
502 } 665 }
503 else if (ch >= 0x20 && ch <= 0x7f) 666 while (cur < buf + SHORT_STRING_LEN);
504 APPEND_CH (*dec->cur++); 667
505 else if (ch >= 0x80) 668 STRLEN len = cur - buf;
669
670 if (sv)
506 { 671 {
507 STRLEN clen; 672 SvGROW (sv, SvCUR (sv) + len + 1);
508 UV uch = utf8n_to_uvuni (dec->cur, dec->end - dec->cur, &clen, UTF8_CHECK_ONLY); 673 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
509 if (clen < 0) 674 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 } 675 }
516 else 676 else
517 ERR ("invalid character encountered"); 677 sv = newSVpvn (buf, len);
518 } 678 }
679 while (*dec->cur != '"');
519 680
520 ++dec->cur; 681 ++dec->cur;
521 682
683 if (sv)
684 {
522 SvPOK_only (sv); 685 SvPOK_only (sv);
523
524 SvCUR_set (sv, cur);
525 *SvEND (sv) = 0; 686 *SvEND (sv) = 0;
526 687
527 if (utf8) 688 if (utf8)
528 SvUTF8_on (sv); 689 SvUTF8_on (sv);
690 }
691 else
692 sv = newSVpvn ("", 0);
529 693
530 return sv; 694 return sv;
531 695
532fail: 696fail:
533 SvREFCNT_dec (sv);
534 return 0; 697 return 0;
535} 698}
536 699
537static SV * 700static SV *
538decode_num (dec_t *dec) 701decode_num (dec_t *dec)
… …
548 { 711 {
549 ++dec->cur; 712 ++dec->cur;
550 if (*dec->cur >= '0' && *dec->cur <= '9') 713 if (*dec->cur >= '0' && *dec->cur <= '9')
551 ERR ("malformed number (leading zero must not be followed by another digit)"); 714 ERR ("malformed number (leading zero must not be followed by another digit)");
552 } 715 }
553 716 else if (*dec->cur < '0' || *dec->cur > '9')
554 // int 717 ERR ("malformed number (no digits after initial minus)");
718 else
719 do
720 {
721 ++dec->cur;
722 }
555 while (*dec->cur >= '0' && *dec->cur <= '9') 723 while (*dec->cur >= '0' && *dec->cur <= '9');
556 ++dec->cur;
557 724
558 // [frac] 725 // [frac]
559 if (*dec->cur == '.') 726 if (*dec->cur == '.')
560 { 727 {
561 is_nv = 1; 728 ++dec->cur;
729
730 if (*dec->cur < '0' || *dec->cur > '9')
731 ERR ("malformed number (no digits after decimal point)");
562 732
563 do 733 do
564 { 734 {
565 ++dec->cur; 735 ++dec->cur;
566 } 736 }
567 while (*dec->cur >= '0' && *dec->cur <= '9'); 737 while (*dec->cur >= '0' && *dec->cur <= '9');
738
739 is_nv = 1;
568 } 740 }
569 741
570 // [exp] 742 // [exp]
571 if (*dec->cur == 'e' || *dec->cur == 'E') 743 if (*dec->cur == 'e' || *dec->cur == 'E')
572 { 744 {
573 is_nv = 1;
574
575 ++dec->cur; 745 ++dec->cur;
746
576 if (*dec->cur == '-' || *dec->cur == '+') 747 if (*dec->cur == '-' || *dec->cur == '+')
577 ++dec->cur; 748 ++dec->cur;
578 749
750 if (*dec->cur < '0' || *dec->cur > '9')
751 ERR ("malformed number (no digits after exp sign)");
752
753 do
754 {
755 ++dec->cur;
756 }
579 while (*dec->cur >= '0' && *dec->cur <= '9') 757 while (*dec->cur >= '0' && *dec->cur <= '9');
580 ++dec->cur; 758
759 is_nv = 1;
581 } 760 }
582 761
583 if (!is_nv) 762 if (!is_nv)
584 { 763 {
585 UV uv; 764 UV uv;
… …
603static SV * 782static SV *
604decode_av (dec_t *dec) 783decode_av (dec_t *dec)
605{ 784{
606 AV *av = newAV (); 785 AV *av = newAV ();
607 786
787 decode_ws (dec);
788 if (*dec->cur == ']')
789 ++dec->cur;
790 else
608 for (;;) 791 for (;;)
609 { 792 {
610 SV *value; 793 SV *value;
611 794
612 value = decode_sv (dec); 795 value = decode_sv (dec);
613 if (!value) 796 if (!value)
614 goto fail; 797 goto fail;
615 798
616 av_push (av, value); 799 av_push (av, value);
617 800
618 WS; 801 decode_ws (dec);
619 802
620 if (*dec->cur == ']') 803 if (*dec->cur == ']')
621 { 804 {
622 ++dec->cur; 805 ++dec->cur;
623 break; 806 break;
807 }
624 } 808
625
626 if (*dec->cur != ',') 809 if (*dec->cur != ',')
627 ERR (", or ] expected while parsing array"); 810 ERR (", or ] expected while parsing array");
628 811
629 ++dec->cur; 812 ++dec->cur;
630 } 813 }
631 814
632 return newRV_noinc ((SV *)av); 815 return newRV_noinc ((SV *)av);
633 816
634fail: 817fail:
635 SvREFCNT_dec (av); 818 SvREFCNT_dec (av);
… …
639static SV * 822static SV *
640decode_hv (dec_t *dec) 823decode_hv (dec_t *dec)
641{ 824{
642 HV *hv = newHV (); 825 HV *hv = newHV ();
643 826
827 decode_ws (dec);
828 if (*dec->cur == '}')
829 ++dec->cur;
830 else
644 for (;;) 831 for (;;)
645 { 832 {
646 SV *key, *value; 833 SV *key, *value;
647 834
648 WS; EXPECT_CH ('"'); 835 decode_ws (dec); EXPECT_CH ('"');
649 836
650 key = decode_str (dec); 837 key = decode_str (dec);
651 if (!key) 838 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; 839 goto fail;
840
841 decode_ws (dec); EXPECT_CH (':');
842
843 value = decode_sv (dec);
844 if (!value)
845 {
846 SvREFCNT_dec (key);
847 goto fail;
661 } 848 }
662 849
663 //TODO: optimise 850 //TODO: optimise
664 hv_store_ent (hv, key, value, 0); 851 hv_store_ent (hv, key, value, 0);
665 852
666 WS; 853 decode_ws (dec);
667 854
668 if (*dec->cur == '}') 855 if (*dec->cur == '}')
669 { 856 {
670 ++dec->cur; 857 ++dec->cur;
671 break; 858 break;
672 } 859 }
673 860
674 if (*dec->cur != ',') 861 if (*dec->cur != ',')
675 ERR (", or } expected while parsing object/hash"); 862 ERR (", or } expected while parsing object/hash");
676 863
677 ++dec->cur; 864 ++dec->cur;
678 } 865 }
679 866
680 return newRV_noinc ((SV *)hv); 867 return newRV_noinc ((SV *)hv);
681 868
682fail: 869fail:
683 SvREFCNT_dec (hv); 870 SvREFCNT_dec (hv);
… …
685} 872}
686 873
687static SV * 874static SV *
688decode_sv (dec_t *dec) 875decode_sv (dec_t *dec)
689{ 876{
690 WS; 877 decode_ws (dec);
691 switch (*dec->cur) 878 switch (*dec->cur)
692 { 879 {
693 case '"': ++dec->cur; return decode_str (dec); 880 case '"': ++dec->cur; return decode_str (dec);
694 case '[': ++dec->cur; return decode_av (dec); 881 case '[': ++dec->cur; return decode_av (dec);
695 case '{': ++dec->cur; return decode_hv (dec); 882 case '{': ++dec->cur; return decode_hv (dec);
… …
723 910
724 case 'n': 911 case 'n':
725 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4)) 912 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4))
726 { 913 {
727 dec->cur += 4; 914 dec->cur += 4;
728 return newSViv (1); 915 return newSVsv (&PL_sv_undef);
729 } 916 }
730 else 917 else
731 ERR ("'null' expected"); 918 ERR ("'null' expected");
732 919
733 break; 920 break;
734 921
735 default: 922 default:
736 ERR ("malformed json string"); 923 ERR ("malformed json string, neither array, object, number, string or atom");
737 break; 924 break;
738 } 925 }
739 926
740fail: 927fail:
741 return 0; 928 return 0;
… …
744static SV * 931static SV *
745decode_json (SV *string, UV flags) 932decode_json (SV *string, UV flags)
746{ 933{
747 SV *sv; 934 SV *sv;
748 935
749 if (!(flags & F_UTF8)) 936 if (flags & F_UTF8)
937 sv_utf8_downgrade (string, 0);
938 else
750 sv_utf8_upgrade (string); 939 sv_utf8_upgrade (string);
751 940
752 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 941 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
753 942
754 dec_t dec; 943 dec_t dec;
755 dec.flags = flags; 944 dec.flags = flags;
756 dec.cur = SvPVX (string); 945 dec.cur = SvPVX (string);
757 dec.end = SvEND (string); 946 dec.end = SvEND (string);
758 dec.err = 0; 947 dec.err = 0;
759 948
760 *dec.end = 1; // invalid anywhere
761 sv = decode_sv (&dec); 949 sv = decode_sv (&dec);
762 *dec.end = 0;
763 950
764 if (!sv) 951 if (!sv)
765 { 952 {
953 IV offset = dec.flags & F_UTF8
954 ? dec.cur - SvPVX (string)
766 IV offset = utf8_distance (dec.cur, SvPVX (string)); 955 : utf8_distance (dec.cur, SvPVX (string));
767 SV *uni = sv_newmortal (); 956 SV *uni = sv_newmortal ();
768 957
958 // horrible hack to silence warning inside pv_uni_display
959 COP cop = *PL_curcop;
960 cop.cop_warnings = pWARN_NONE;
961 ENTER;
962 SAVEVPTR (PL_curcop);
963 PL_curcop = &cop;
769 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 964 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
965 LEAVE;
966
770 croak ("%s, at character %d (%s)", 967 croak ("%s, at character offset %d (%s)",
771 dec.err, 968 dec.err,
772 (int)offset, 969 (int)offset,
773 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 970 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
774 } 971 }
775 972
776 sv_dump (sv);//D
777 return sv_2mortal (sv); 973 sv = sv_2mortal (sv);
974
975 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv))
976 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
977
978 return sv;
778} 979}
980
981/////////////////////////////////////////////////////////////////////////////
982// XS interface functions
779 983
780MODULE = JSON::XS PACKAGE = JSON::XS 984MODULE = JSON::XS PACKAGE = JSON::XS
781 985
782BOOT: 986BOOT:
783{ 987{
… …
785 989
786 memset (decode_hexdigit, 0xff, 256); 990 memset (decode_hexdigit, 0xff, 256);
787 for (i = 10; i--; ) 991 for (i = 10; i--; )
788 decode_hexdigit ['0' + i] = i; 992 decode_hexdigit ['0' + i] = i;
789 993
790 for (i = 6; --i; ) 994 for (i = 7; i--; )
791 { 995 {
792 decode_hexdigit ['a' + i] = 10 + i; 996 decode_hexdigit ['a' + i] = 10 + i;
793 decode_hexdigit ['A' + i] = 10 + i; 997 decode_hexdigit ['A' + i] = 10 + i;
794 } 998 }
795 999
796 json_stash = gv_stashpv ("JSON::XS", 1); 1000 json_stash = gv_stashpv ("JSON::XS", 1);
797} 1001}
1002
1003PROTOTYPES: DISABLE
798 1004
799SV *new (char *dummy) 1005SV *new (char *dummy)
800 CODE: 1006 CODE:
801 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1007 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash);
802 OUTPUT: 1008 OUTPUT:
803 RETVAL 1009 RETVAL
804 1010
805SV *ascii (SV *self, int enable) 1011SV *ascii (SV *self, int enable = 1)
806 ALIAS: 1012 ALIAS:
807 ascii = F_ASCII 1013 ascii = F_ASCII
808 utf8 = F_UTF8 1014 utf8 = F_UTF8
809 indent = F_INDENT 1015 indent = F_INDENT
810 canonical = F_CANONICAL 1016 canonical = F_CANONICAL
811 space_before = F_SPACE_BEFORE 1017 space_before = F_SPACE_BEFORE
812 space_after = F_SPACE_AFTER 1018 space_after = F_SPACE_AFTER
813 json_rpc = F_JSON_RPC 1019 pretty = F_PRETTY
1020 allow_nonref = F_ALLOW_NONREF
1021 shrink = F_SHRINK
814 CODE: 1022 CODE:
815{ 1023{
816 UV *uv = SvJSON (self); 1024 UV *uv = SvJSON (self);
817 if (enable) 1025 if (enable)
818 *uv |= ix; 1026 *uv |= ix;
… …
826 1034
827void encode (SV *self, SV *scalar) 1035void encode (SV *self, SV *scalar)
828 PPCODE: 1036 PPCODE:
829 XPUSHs (encode_json (scalar, *SvJSON (self))); 1037 XPUSHs (encode_json (scalar, *SvJSON (self)));
830 1038
831void decode (SV *self, SV *jsondata) 1039void decode (SV *self, SV *jsonstr)
832 PPCODE: 1040 PPCODE:
833 XPUSHs (decode_json (jsondata, *SvJSON (self))); 1041 XPUSHs (decode_json (jsonstr, *SvJSON (self)));
834 1042
1043PROTOTYPES: ENABLE
1044
1045void to_json (SV *scalar)
1046 ALIAS:
1047 objToJson = 0
1048 PPCODE:
1049 XPUSHs (encode_json (scalar, F_UTF8));
1050
1051void from_json (SV *jsonstr)
1052 ALIAS:
1053 jsonToObj = 0
1054 PPCODE:
1055 XPUSHs (decode_json (jsonstr, F_UTF8));
1056

Diff Legend

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