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 1.16 by root, Sun Mar 25 02:37:00 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
17
18// F_SKIPINVALID?
19// F_EXECCODEREF?
20// F_SELFCONVERT?
16 21
17#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 22#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
18#define F_DEFAULT 0 23#define F_DEFAULT 0
19 24
20#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
21 30
22#define SB do { 31#define SB do {
23#define SE } while (0) 32#define SE } while (0)
24 33
25static 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
26 83
27// structure used for encoding JSON 84// structure used for encoding JSON
28typedef struct 85typedef struct
29{ 86{
30 char *cur; 87 char *cur; // SvPVX (sv) + current output position
31 STRLEN len; // SvLEN (sv)
32 char *end; // SvEND (sv) 88 char *end; // SvEND (sv)
33 SV *sv; 89 SV *sv; // result scalar
34 UV flags; 90 UV flags; // F_*
35 int max_recurse; 91 int indent; // indentation level
36 int indent; 92 int max_depth; // max. recursion level
37} enc_t; 93} enc_t;
38
39// structure used for decoding JSON
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 94
59static void 95static void
60need (enc_t *enc, STRLEN len) 96need (enc_t *enc, STRLEN len)
61{ 97{
62 if (enc->cur + len >= enc->end) 98 if (enc->cur + len >= enc->end)
63 { 99 {
64 STRLEN cur = enc->cur - SvPVX (enc->sv); 100 STRLEN cur = enc->cur - SvPVX (enc->sv);
65 SvGROW (enc->sv, cur + len + 1); 101 SvGROW (enc->sv, cur + len + 1);
66 enc->cur = SvPVX (enc->sv) + cur; 102 enc->cur = SvPVX (enc->sv) + cur;
67 enc->end = SvEND (enc->sv); 103 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv);
68 } 104 }
69} 105}
70 106
71static void 107static void
72encode_ch (enc_t *enc, char ch) 108encode_ch (enc_t *enc, char ch)
… …
78static void 114static void
79encode_str (enc_t *enc, char *str, STRLEN len, int is_utf8) 115encode_str (enc_t *enc, char *str, STRLEN len, int is_utf8)
80{ 116{
81 char *end = str + len; 117 char *end = str + len;
82 118
119 need (enc, len);
120
83 while (str < end) 121 while (str < end)
84 { 122 {
85 unsigned char ch = *(unsigned char *)str; 123 unsigned char ch = *(unsigned char *)str;
124
86 if (ch >= 0x20 && ch < 0x80) // most common case 125 if (ch >= 0x20 && ch < 0x80) // most common case
87 { 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
88 *enc->cur++ = ch; 140 *enc->cur++ = ch;
141
89 str++; 142 ++str;
90 } 143 }
91 else 144 else
92 { 145 {
93 STRLEN clen; 146 switch (ch)
94 UV uch;
95
96 if (is_utf8)
97 { 147 {
98 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;
99 if (clen < 0) 149 case '\011': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 't'; ++str; break;
100 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;
101 } 151 case '\014': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'f'; ++str; break;
102 else 152 case '\015': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'r'; ++str; break;
103 {
104 uch = ch;
105 clen = 1;
106 }
107 153
108 need (enc, len += 6); 154 default:
109
110 if (uch < 0xa0 || enc->flags & F_ASCII)
111 {
112 if (uch > 0xFFFFUL)
113 { 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;
114 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 {
115 need (enc, len += 6); 179 need (enc, len += 11);
116 sprintf (enc->cur, "\\u%04x\\u%04x", 180 sprintf (enc->cur, "\\u%04x\\u%04x",
117 (uch - 0x10000) / 0x400 + 0xD800, 181 (int)((uch - 0x10000) / 0x400 + 0xD800),
118 (uch - 0x10000) % 0x400 + 0xDC00); 182 (int)((uch - 0x10000) % 0x400 + 0xDC00));
119 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;
120 } 198 }
199 else if (is_utf8)
200 {
201 need (enc, len += clen);
202 do
203 {
204 *enc->cur++ = *str++;
205 }
206 while (--clen);
207 }
121 else 208 else
122 { 209 {
123 sprintf (enc->cur, "\\u%04x", uch); 210 need (enc, len += UTF8_MAX_LEN - 1); // never more than 11 bytes needed
124 enc->cur += 6; 211 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
212 ++str;
213 }
125 } 214 }
126 } 215 }
127 else if (is_utf8)
128 {
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 } 216 }
137 217
138 --len; 218 --len;
139 } 219 }
140} 220}
141 221
142#define INDENT SB \ 222static void
223encode_indent (enc_t *enc)
224{
143 if (enc->flags & F_INDENT) \ 225 if (enc->flags & F_INDENT)
144 { \ 226 {
145 int i_; \ 227 int spaces = enc->indent * INDENT_STEP;
146 need (enc, enc->indent); \
147 for (i_ = enc->indent * 3; i_--; )\
148 encode_ch (enc, ' '); \
149 } \
150 SE
151 228
152#define SPACE SB need (enc, 1); encode_ch (enc, ' '); SE 229 need (enc, spaces);
153#define NL SB if (enc->flags & F_INDENT) { need (enc, 1); encode_ch (enc, '\n'); } SE 230 memset (enc->cur, ' ', spaces);
154#define COMMA SB \ 231 enc->cur += spaces;
232 }
233}
234
235static void
236encode_space (enc_t *enc)
237{
238 need (enc, 1);
155 encode_ch (enc, ','); \ 239 encode_ch (enc, ' ');
240}
241
242static void
243encode_nl (enc_t *enc)
244{
156 if (enc->flags & F_INDENT) \ 245 if (enc->flags & F_INDENT)
157 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);
158 else if (enc->flags & F_SPACE_AFTER) \ 259 else if (enc->flags & F_SPACE_AFTER)
159 SPACE; \ 260 encode_space (enc);
160 SE 261}
161 262
162static void encode_sv (enc_t *enc, SV *sv); 263static void encode_sv (enc_t *enc, SV *sv);
163 264
164static void 265static void
165encode_av (enc_t *enc, AV *av) 266encode_av (enc_t *enc, AV *av)
166{ 267{
167 int i, len = av_len (av); 268 int i, len = av_len (av);
168 269
169 encode_ch (enc, '['); NL; 270 encode_ch (enc, '['); encode_nl (enc);
170 ++enc->indent; 271 ++enc->indent;
171 272
172 for (i = 0; i <= len; ++i) 273 for (i = 0; i <= len; ++i)
173 { 274 {
174 INDENT; 275 encode_indent (enc);
175 encode_sv (enc, *av_fetch (av, i, 0)); 276 encode_sv (enc, *av_fetch (av, i, 0));
176 277
177 if (i < len) 278 if (i < len)
178 COMMA; 279 encode_comma (enc);
179 } 280 }
180 281
181 NL; 282 encode_nl (enc);
182 283
183 --enc->indent; 284 --enc->indent;
184 INDENT; encode_ch (enc, ']'); 285 encode_indent (enc); encode_ch (enc, ']');
185} 286}
186 287
187static void 288static void
188encode_he (enc_t *enc, HE *he) 289encode_he (enc_t *enc, HE *he)
189{ 290{
… …
191 292
192 if (HeKLEN (he) == HEf_SVKEY) 293 if (HeKLEN (he) == HEf_SVKEY)
193 { 294 {
194 SV *sv = HeSVKEY (he); 295 SV *sv = HeSVKEY (he);
195 STRLEN len; 296 STRLEN len;
297 char *str;
298
299 SvGETMAGIC (sv);
196 char *str = SvPV (sv, len); 300 str = SvPV (sv, len);
197 301
198 encode_str (enc, str, len, SvUTF8 (sv)); 302 encode_str (enc, str, len, SvUTF8 (sv));
199 } 303 }
200 else 304 else
201 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 305 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
202 306
203 encode_ch (enc, '"'); 307 encode_ch (enc, '"');
204 308
205 if (enc->flags & F_SPACE_BEFORE) SPACE; 309 if (enc->flags & F_SPACE_BEFORE) encode_space (enc);
206 encode_ch (enc, ':'); 310 encode_ch (enc, ':');
207 if (enc->flags & F_SPACE_AFTER ) SPACE; 311 if (enc->flags & F_SPACE_AFTER ) encode_space (enc);
208 encode_sv (enc, HeVAL (he)); 312 encode_sv (enc, HeVAL (he));
209} 313}
210 314
211// compare hash entries, used when all keys are bytestrings 315// compare hash entries, used when all keys are bytestrings
212static int 316static int
… …
218 HE *b = *(HE **)b_; 322 HE *b = *(HE **)b_;
219 323
220 STRLEN la = HeKLEN (a); 324 STRLEN la = HeKLEN (a);
221 STRLEN lb = HeKLEN (b); 325 STRLEN lb = HeKLEN (b);
222 326
223 if (!(cmp == memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 327 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb)))
224 cmp = la < lb ? -1 : la == lb ? 0 : 1; 328 cmp = la - lb;
225 329
226 return cmp; 330 return cmp;
227} 331}
228 332
229// 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
… …
236static void 340static void
237encode_hv (enc_t *enc, HV *hv) 341encode_hv (enc_t *enc, HV *hv)
238{ 342{
239 int count, i; 343 int count, i;
240 344
241 encode_ch (enc, '{'); NL; ++enc->indent; 345 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
242 346
243 if ((count = hv_iterinit (hv))) 347 if ((count = hv_iterinit (hv)))
244 { 348 {
245 // for canonical output we have to sort by keys first 349 // for canonical output we have to sort by keys first
246 // actually, this is mostly due to the stupid so-called 350 // actually, this is mostly due to the stupid so-called
247 // security workaround added somewhere in 5.8.x. 351 // security workaround added somewhere in 5.8.x.
248 // that randomises hash orderings 352 // that randomises hash orderings
249 if (enc->flags & F_CANONICAL) 353 if (enc->flags & F_CANONICAL)
250 { 354 {
251 HE *he, *hes [count]; 355 HE *he, *hes [count]; // if your compiler dies here, you need to enable C99 mode
252 int fast = 1; 356 int fast = 1;
253 357
254 i = 0; 358 i = 0;
255 while ((he = hv_iternext (hv))) 359 while ((he = hv_iternext (hv)))
256 { 360 {
… …
263 367
264 if (fast) 368 if (fast)
265 qsort (hes, count, sizeof (HE *), he_cmp_fast); 369 qsort (hes, count, sizeof (HE *), he_cmp_fast);
266 else 370 else
267 { 371 {
268 // hack to disable "use bytes" 372 // hack to forcefully disable "use bytes"
269 COP *oldcop = PL_curcop, cop; 373 COP cop = *PL_curcop;
270 cop.op_private = 0; 374 cop.op_private = 0;
375
376 ENTER;
377 SAVETMPS;
378
379 SAVEVPTR (PL_curcop);
271 PL_curcop = &cop; 380 PL_curcop = &cop;
272 381
273 SAVETMPS;
274 qsort (hes, count, sizeof (HE *), he_cmp_slow); 382 qsort (hes, count, sizeof (HE *), he_cmp_slow);
383
275 FREETMPS; 384 FREETMPS;
276 385 LEAVE;
277 PL_curcop = oldcop;
278 } 386 }
279 387
280 for (i = 0; i < count; ++i) 388 for (i = 0; i < count; ++i)
281 { 389 {
282 INDENT; 390 encode_indent (enc);
283 encode_he (enc, hes [i]); 391 encode_he (enc, hes [i]);
284 392
285 if (i < count - 1) 393 if (i < count - 1)
286 COMMA; 394 encode_comma (enc);
287 } 395 }
288 396
289 NL; 397 encode_nl (enc);
290 } 398 }
291 else 399 else
292 { 400 {
293 SV *sv; 401 SV *sv;
294 HE *he = hv_iternext (hv); 402 HE *he = hv_iternext (hv);
295 403
296 for (;;) 404 for (;;)
297 { 405 {
298 INDENT; 406 encode_indent (enc);
299 encode_he (enc, he); 407 encode_he (enc, he);
300 408
301 if (!(he = hv_iternext (hv))) 409 if (!(he = hv_iternext (hv)))
302 break; 410 break;
303 411
304 COMMA; 412 encode_comma (enc);
305 } 413 }
306 414
307 NL; 415 encode_nl (enc);
308 } 416 }
309 } 417 }
310 418
311 --enc->indent; INDENT; encode_ch (enc, '}'); 419 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
312} 420}
313 421
314static void 422static void
315encode_sv (enc_t *enc, SV *sv) 423encode_sv (enc_t *enc, SV *sv)
316{ 424{
425 SvGETMAGIC (sv);
426
317 if (SvPOKp (sv)) 427 if (SvPOKp (sv))
318 { 428 {
319 STRLEN len; 429 STRLEN len;
320 char *str = SvPV (sv, len); 430 char *str = SvPV (sv, len);
321 encode_ch (enc, '"'); 431 encode_ch (enc, '"');
… …
336 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 446 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv))
337 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 447 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv));
338 } 448 }
339 else if (SvROK (sv)) 449 else if (SvROK (sv))
340 { 450 {
341 if (!--enc->max_recurse) 451 SV *rv = SvRV (sv);
452
453 if (enc->indent >= enc->max_depth)
342 croak ("data structure too deep (hit recursion limit)"); 454 croak ("data structure too deep (hit recursion limit)");
343 455
344 sv = SvRV (sv);
345
346 switch (SvTYPE (sv)) 456 switch (SvTYPE (rv))
347 { 457 {
348 case SVt_PVAV: encode_av (enc, (AV *)sv); break; 458 case SVt_PVAV: encode_av (enc, (AV *)rv); break;
349 case SVt_PVHV: encode_hv (enc, (HV *)sv); break; 459 case SVt_PVHV: encode_hv (enc, (HV *)rv); break;
350 460
351 default: 461 default:
352 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));
353 } 464 }
354 } 465 }
355 else if (!SvOK (sv)) 466 else if (!SvOK (sv))
356 encode_str (enc, "null", 4, 0); 467 encode_str (enc, "null", 4, 0);
357 else 468 else
358 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));
359} 471}
360 472
361static SV * 473static SV *
362encode_json (SV *scalar, UV flags) 474encode_json (SV *scalar, UV flags)
363{ 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
364 enc_t enc; 479 enc_t enc;
365 enc.flags = flags; 480 enc.flags = flags;
366 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 481 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
367 enc.cur = SvPVX (enc.sv); 482 enc.cur = SvPVX (enc.sv);
368 enc.end = SvEND (enc.sv); 483 enc.end = SvEND (enc.sv);
369 enc.max_recurse = 0;
370 enc.indent = 0; 484 enc.indent = 0;
485 enc.max_depth = 0x7fffffffUL;
371 486
372 SvPOK_only (enc.sv); 487 SvPOK_only (enc.sv);
373 encode_sv (&enc, scalar); 488 encode_sv (&enc, scalar);
374 489
375 if (!(flags & (F_ASCII | F_UTF8))) 490 if (!(flags & (F_ASCII | F_UTF8)))
376 SvUTF8_on (enc.sv); 491 SvUTF8_on (enc.sv);
377 492
378 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
379 return enc.sv; 498 return enc.sv;
380} 499}
381 500
382///////////////////////////////////////////////////////////////////////////// 501/////////////////////////////////////////////////////////////////////////////
502// decoder
383 503
384#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{
385 for (;;) \ 516 for (;;)
386 { \ 517 {
387 char ch = *dec->cur; \ 518 char ch = *dec->cur;
519
388 if (ch > 0x20 \ 520 if (ch > 0x20
389 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) \ 521 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09))
390 break; \ 522 break;
523
391 ++dec->cur; \ 524 ++dec->cur;
392 } 525 }
526}
393 527
394#define ERR(reason) SB dec->err = reason; goto fail; SE 528#define ERR(reason) SB dec->err = reason; goto fail; SE
395#define EXPECT_CH(ch) SB \ 529#define EXPECT_CH(ch) SB \
396 if (*dec->cur != ch) \ 530 if (*dec->cur != ch) \
397 ERR (# ch " expected"); \ 531 ERR (# ch " expected"); \
398 ++dec->cur; \ 532 ++dec->cur; \
399 SE 533 SE
400 534
401static SV *decode_sv (dec_t *dec); 535static SV *decode_sv (dec_t *dec);
402 536
403#define APPEND_CH(ch) SB \
404 SvGROW (sv, cur + 1 + 1); \
405 SvPVX (sv)[cur++] = (ch); \
406 SE
407
408static signed char decode_hexdigit[256]; 537static signed char decode_hexdigit[256];
409 538
410static UV 539static UV
411decode_4hex (dec_t *dec) 540decode_4hex (dec_t *dec)
412{ 541{
413 signed char d1, d2, d3, d4; 542 signed char d1, d2, d3, d4;
543 unsigned char *cur = (unsigned char *)dec->cur;
414 544
415 d1 = decode_hexdigit [((unsigned char *)dec->cur) [0]];
416 if (d1 < 0) ERR ("four hexadecimal digits expected"); 545 d1 = decode_hexdigit [cur [0]]; if (d1 < 0) ERR ("four hexadecimal digits expected");
417 d2 = decode_hexdigit [((unsigned char *)dec->cur) [1]];
418 if (d2 < 0) ERR ("four hexadecimal digits expected"); 546 d2 = decode_hexdigit [cur [1]]; if (d2 < 0) ERR ("four hexadecimal digits expected");
419 d3 = decode_hexdigit [((unsigned char *)dec->cur) [2]];
420 if (d3 < 0) ERR ("four hexadecimal digits expected"); 547 d3 = decode_hexdigit [cur [2]]; 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"); 548 d4 = decode_hexdigit [cur [3]]; if (d4 < 0) ERR ("four hexadecimal digits expected");
423 549
424 dec->cur += 4; 550 dec->cur += 4;
425 551
426 return ((UV)d1) << 12 552 return ((UV)d1) << 12
427 | ((UV)d2) << 8 553 | ((UV)d2) << 8
… …
433} 559}
434 560
435static SV * 561static SV *
436decode_str (dec_t *dec) 562decode_str (dec_t *dec)
437{ 563{
438 SV *sv = NEWSV (0,2); 564 SV *sv = 0;
439 STRLEN cur = 0;
440 int utf8 = 0; 565 int utf8 = 0;
441 566
442 for (;;) 567 do
443 { 568 {
444 unsigned char ch = *(unsigned char *)dec->cur; 569 char buf [SHORT_STRING_LEN + UTF8_MAX_LEN];
570 char *cur = buf;
445 571
446 if (ch == '"') 572 do
447 break;
448 else if (ch == '\\')
449 { 573 {
450 switch (*++dec->cur) 574 unsigned char ch = *(unsigned char *)dec->cur++;
575
576 if (ch == '"')
451 { 577 {
452 case '\\': 578 --dec->cur;
453 case '/': 579 break;
454 case '"': APPEND_CH (*dec->cur++); break; 580 }
455 581 else if (ch == '\\')
456 case 'b': APPEND_CH ('\010'); ++dec->cur; break; 582 {
457 case 't': APPEND_CH ('\011'); ++dec->cur; break; 583 switch (*dec->cur)
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 { 584 {
464 UV lo, hi; 585 case '\\':
465 ++dec->cur; 586 case '/':
587 case '"': *cur++ = *dec->cur++; break;
466 588
467 hi = decode_4hex (dec); 589 case 'b': ++dec->cur; *cur++ = '\010'; break;
468 if (hi == (UV)-1) 590 case 't': ++dec->cur; *cur++ = '\011'; break;
469 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;
470 594
471 // possibly a surrogate pair 595 case 'u':
472 if (hi >= 0xd800 && hi < 0xdc00)
473 { 596 {
474 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 597 UV lo, hi;
475 ERR ("illegal surrogate character");
476
477 dec->cur += 2; 598 ++dec->cur;
478 599
479 lo = decode_4hex (dec); 600 hi = decode_4hex (dec);
480 if (lo == (UV)-1) 601 if (hi == (UV)-1)
481 goto fail; 602 goto fail;
482 603
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
483 if (lo < 0xdc00 || lo >= 0xe000) 617 if (lo < 0xdc00 || lo >= 0xe000)
484 ERR ("surrogate pair expected"); 618 ERR ("surrogate pair expected");
485 619
486 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;
487 } 633 }
488 else if (lo >= 0xdc00 && lo < 0xe000)
489 ERR ("illegal surrogate character");
490
491 if (hi >= 0x80)
492 { 634 break;
493 utf8 = 1;
494 635
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 636 default:
499 APPEND_CH (hi); 637 --dec->cur;
638 ERR ("illegal backslash escape sequence in string");
500 } 639 }
501 break;
502 } 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
503 } 665 }
504 else if (ch >= 0x20 && ch <= 0x7f) 666 while (cur < buf + SHORT_STRING_LEN);
505 APPEND_CH (*dec->cur++); 667
506 else if (ch >= 0x80) 668 STRLEN len = cur - buf;
669
670 if (sv)
507 { 671 {
508 STRLEN clen; 672 SvGROW (sv, SvCUR (sv) + len + 1);
509 UV uch = utf8n_to_uvuni (dec->cur, dec->end - dec->cur, &clen, UTF8_CHECK_ONLY); 673 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
510 if (clen < 0) 674 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 } 675 }
517 else 676 else
518 ERR ("invalid character encountered"); 677 sv = newSVpvn (buf, len);
519 } 678 }
679 while (*dec->cur != '"');
520 680
521 ++dec->cur; 681 ++dec->cur;
522 682
683 if (sv)
684 {
523 SvPOK_only (sv); 685 SvPOK_only (sv);
524
525 SvCUR_set (sv, cur);
526 *SvEND (sv) = 0; 686 *SvEND (sv) = 0;
527 687
528 if (utf8) 688 if (utf8)
529 SvUTF8_on (sv); 689 SvUTF8_on (sv);
690 }
691 else
692 sv = newSVpvn ("", 0);
530 693
531 return sv; 694 return sv;
532 695
533fail: 696fail:
534 SvREFCNT_dec (sv);
535 return 0; 697 return 0;
536} 698}
537 699
538static SV * 700static SV *
539decode_num (dec_t *dec) 701decode_num (dec_t *dec)
… …
549 { 711 {
550 ++dec->cur; 712 ++dec->cur;
551 if (*dec->cur >= '0' && *dec->cur <= '9') 713 if (*dec->cur >= '0' && *dec->cur <= '9')
552 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)");
553 } 715 }
554 716 else if (*dec->cur < '0' || *dec->cur > '9')
555 // int 717 ERR ("malformed number (no digits after initial minus)");
718 else
719 do
720 {
721 ++dec->cur;
722 }
556 while (*dec->cur >= '0' && *dec->cur <= '9') 723 while (*dec->cur >= '0' && *dec->cur <= '9');
557 ++dec->cur;
558 724
559 // [frac] 725 // [frac]
560 if (*dec->cur == '.') 726 if (*dec->cur == '.')
561 { 727 {
562 is_nv = 1; 728 ++dec->cur;
729
730 if (*dec->cur < '0' || *dec->cur > '9')
731 ERR ("malformed number (no digits after decimal point)");
563 732
564 do 733 do
565 { 734 {
566 ++dec->cur; 735 ++dec->cur;
567 } 736 }
568 while (*dec->cur >= '0' && *dec->cur <= '9'); 737 while (*dec->cur >= '0' && *dec->cur <= '9');
738
739 is_nv = 1;
569 } 740 }
570 741
571 // [exp] 742 // [exp]
572 if (*dec->cur == 'e' || *dec->cur == 'E') 743 if (*dec->cur == 'e' || *dec->cur == 'E')
573 { 744 {
574 is_nv = 1;
575
576 ++dec->cur; 745 ++dec->cur;
746
577 if (*dec->cur == '-' || *dec->cur == '+') 747 if (*dec->cur == '-' || *dec->cur == '+')
578 ++dec->cur; 748 ++dec->cur;
579 749
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 }
580 while (*dec->cur >= '0' && *dec->cur <= '9') 757 while (*dec->cur >= '0' && *dec->cur <= '9');
581 ++dec->cur; 758
759 is_nv = 1;
582 } 760 }
583 761
584 if (!is_nv) 762 if (!is_nv)
585 { 763 {
586 UV uv; 764 UV uv;
… …
604static SV * 782static SV *
605decode_av (dec_t *dec) 783decode_av (dec_t *dec)
606{ 784{
607 AV *av = newAV (); 785 AV *av = newAV ();
608 786
787 decode_ws (dec);
788 if (*dec->cur == ']')
789 ++dec->cur;
790 else
609 for (;;) 791 for (;;)
610 { 792 {
611 SV *value; 793 SV *value;
612 794
613 value = decode_sv (dec); 795 value = decode_sv (dec);
614 if (!value) 796 if (!value)
615 goto fail; 797 goto fail;
616 798
617 av_push (av, value); 799 av_push (av, value);
618 800
619 WS; 801 decode_ws (dec);
620 802
621 if (*dec->cur == ']') 803 if (*dec->cur == ']')
622 { 804 {
623 ++dec->cur; 805 ++dec->cur;
624 break; 806 break;
807 }
625 } 808
626
627 if (*dec->cur != ',') 809 if (*dec->cur != ',')
628 ERR (", or ] expected while parsing array"); 810 ERR (", or ] expected while parsing array");
629 811
630 ++dec->cur; 812 ++dec->cur;
631 } 813 }
632 814
633 return newRV_noinc ((SV *)av); 815 return newRV_noinc ((SV *)av);
634 816
635fail: 817fail:
636 SvREFCNT_dec (av); 818 SvREFCNT_dec (av);
… …
640static SV * 822static SV *
641decode_hv (dec_t *dec) 823decode_hv (dec_t *dec)
642{ 824{
643 HV *hv = newHV (); 825 HV *hv = newHV ();
644 826
827 decode_ws (dec);
828 if (*dec->cur == '}')
829 ++dec->cur;
830 else
645 for (;;) 831 for (;;)
646 { 832 {
647 SV *key, *value; 833 SV *key, *value;
648 834
649 WS; EXPECT_CH ('"'); 835 decode_ws (dec); EXPECT_CH ('"');
650 836
651 key = decode_str (dec); 837 key = decode_str (dec);
652 if (!key) 838 if (!key)
653 goto fail;
654
655 WS; EXPECT_CH (':');
656
657 value = decode_sv (dec);
658 if (!value)
659 {
660 SvREFCNT_dec (key);
661 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;
662 } 848 }
663 849
664 //TODO: optimise 850 //TODO: optimise
665 hv_store_ent (hv, key, value, 0); 851 hv_store_ent (hv, key, value, 0);
666 852
667 WS; 853 decode_ws (dec);
668 854
669 if (*dec->cur == '}') 855 if (*dec->cur == '}')
670 { 856 {
671 ++dec->cur; 857 ++dec->cur;
672 break; 858 break;
673 } 859 }
674 860
675 if (*dec->cur != ',') 861 if (*dec->cur != ',')
676 ERR (", or } expected while parsing object/hash"); 862 ERR (", or } expected while parsing object/hash");
677 863
678 ++dec->cur; 864 ++dec->cur;
679 } 865 }
680 866
681 return newRV_noinc ((SV *)hv); 867 return newRV_noinc ((SV *)hv);
682 868
683fail: 869fail:
684 SvREFCNT_dec (hv); 870 SvREFCNT_dec (hv);
… …
686} 872}
687 873
688static SV * 874static SV *
689decode_sv (dec_t *dec) 875decode_sv (dec_t *dec)
690{ 876{
691 WS; 877 decode_ws (dec);
692 switch (*dec->cur) 878 switch (*dec->cur)
693 { 879 {
694 case '"': ++dec->cur; return decode_str (dec); 880 case '"': ++dec->cur; return decode_str (dec);
695 case '[': ++dec->cur; return decode_av (dec); 881 case '[': ++dec->cur; return decode_av (dec);
696 case '{': ++dec->cur; return decode_hv (dec); 882 case '{': ++dec->cur; return decode_hv (dec);
… …
724 910
725 case 'n': 911 case 'n':
726 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4)) 912 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4))
727 { 913 {
728 dec->cur += 4; 914 dec->cur += 4;
729 return newSViv (1); 915 return newSVsv (&PL_sv_undef);
730 } 916 }
731 else 917 else
732 ERR ("'null' expected"); 918 ERR ("'null' expected");
733 919
734 break; 920 break;
735 921
736 default: 922 default:
737 ERR ("malformed json string"); 923 ERR ("malformed json string, neither array, object, number, string or atom");
738 break; 924 break;
739 } 925 }
740 926
741fail: 927fail:
742 return 0; 928 return 0;
… …
745static SV * 931static SV *
746decode_json (SV *string, UV flags) 932decode_json (SV *string, UV flags)
747{ 933{
748 SV *sv; 934 SV *sv;
749 935
750 if (!(flags & F_UTF8)) 936 if (flags & F_UTF8)
937 sv_utf8_downgrade (string, 0);
938 else
751 sv_utf8_upgrade (string); 939 sv_utf8_upgrade (string);
752 940
753 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 941 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
754 942
755 dec_t dec; 943 dec_t dec;
756 dec.flags = flags; 944 dec.flags = flags;
757 dec.cur = SvPVX (string); 945 dec.cur = SvPVX (string);
758 dec.end = SvEND (string); 946 dec.end = SvEND (string);
759 dec.err = 0; 947 dec.err = 0;
760 948
761 *dec.end = 1; // invalid anywhere
762 sv = decode_sv (&dec); 949 sv = decode_sv (&dec);
763 *dec.end = 0;
764 950
765 if (!sv) 951 if (!sv)
766 { 952 {
953 IV offset = dec.flags & F_UTF8
954 ? dec.cur - SvPVX (string)
767 IV offset = utf8_distance (dec.cur, SvPVX (string)); 955 : utf8_distance (dec.cur, SvPVX (string));
768 SV *uni = sv_newmortal (); 956 SV *uni = sv_newmortal ();
769 957
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;
770 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
771 croak ("%s, at character %d (%s)", 967 croak ("%s, at character offset %d (%s)",
772 dec.err, 968 dec.err,
773 (int)offset, 969 (int)offset,
774 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 970 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
775 } 971 }
776 972
777 sv_dump (sv);//D
778 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;
779} 979}
980
981/////////////////////////////////////////////////////////////////////////////
982// XS interface functions
780 983
781MODULE = JSON::XS PACKAGE = JSON::XS 984MODULE = JSON::XS PACKAGE = JSON::XS
782 985
783BOOT: 986BOOT:
784{ 987{
… …
786 989
787 memset (decode_hexdigit, 0xff, 256); 990 memset (decode_hexdigit, 0xff, 256);
788 for (i = 10; i--; ) 991 for (i = 10; i--; )
789 decode_hexdigit ['0' + i] = i; 992 decode_hexdigit ['0' + i] = i;
790 993
791 for (i = 6; --i; ) 994 for (i = 7; i--; )
792 { 995 {
793 decode_hexdigit ['a' + i] = 10 + i; 996 decode_hexdigit ['a' + i] = 10 + i;
794 decode_hexdigit ['A' + i] = 10 + i; 997 decode_hexdigit ['A' + i] = 10 + i;
795 } 998 }
796 999
797 json_stash = gv_stashpv ("JSON::XS", 1); 1000 json_stash = gv_stashpv ("JSON::XS", 1);
798} 1001}
1002
1003PROTOTYPES: DISABLE
799 1004
800SV *new (char *dummy) 1005SV *new (char *dummy)
801 CODE: 1006 CODE:
802 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1007 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash);
803 OUTPUT: 1008 OUTPUT:
804 RETVAL 1009 RETVAL
805 1010
806SV *ascii (SV *self, int enable) 1011SV *ascii (SV *self, int enable = 1)
807 ALIAS: 1012 ALIAS:
808 ascii = F_ASCII 1013 ascii = F_ASCII
809 utf8 = F_UTF8 1014 utf8 = F_UTF8
810 indent = F_INDENT 1015 indent = F_INDENT
811 canonical = F_CANONICAL 1016 canonical = F_CANONICAL
812 space_before = F_SPACE_BEFORE 1017 space_before = F_SPACE_BEFORE
813 space_after = F_SPACE_AFTER 1018 space_after = F_SPACE_AFTER
814 json_rpc = F_JSON_RPC
815 pretty = F_PRETTY 1019 pretty = F_PRETTY
1020 allow_nonref = F_ALLOW_NONREF
1021 shrink = F_SHRINK
816 CODE: 1022 CODE:
817{ 1023{
818 UV *uv = SvJSON (self); 1024 UV *uv = SvJSON (self);
819 if (enable) 1025 if (enable)
820 *uv |= ix; 1026 *uv |= ix;
… …
832 1038
833void decode (SV *self, SV *jsonstr) 1039void decode (SV *self, SV *jsonstr)
834 PPCODE: 1040 PPCODE:
835 XPUSHs (decode_json (jsonstr, *SvJSON (self))); 1041 XPUSHs (decode_json (jsonstr, *SvJSON (self)));
836 1042
1043PROTOTYPES: ENABLE
1044
837void to_json (SV *scalar) 1045void to_json (SV *scalar)
1046 ALIAS:
1047 objToJson = 0
838 PPCODE: 1048 PPCODE:
839 XPUSHs (encode_json (scalar, F_UTF8)); 1049 XPUSHs (encode_json (scalar, F_UTF8));
840 1050
841void from_json (SV *jsonstr) 1051void from_json (SV *jsonstr)
1052 ALIAS:
1053 jsonToObj = 0
842 PPCODE: 1054 PPCODE:
843 XPUSHs (decode_json (jsonstr, F_UTF8)); 1055 XPUSHs (decode_json (jsonstr, F_UTF8));
844 1056

Diff Legend

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