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 1.5 by root, Thu Mar 22 23:24:18 2007 UTC vs.
Revision 1.18 by root, Sun Mar 25 21:19:13 2007 UTC

… …
4 4
5#include "assert.h" 5#include "assert.h"
6#include "string.h" 6#include "string.h"
7#include "stdlib.h" 7#include "stdlib.h"
8 8
9#define F_ASCII 0x00000001 9#define F_ASCII 0x00000001UL
10#define F_UTF8 0x00000002 10#define F_UTF8 0x00000002UL
11#define F_INDENT 0x00000004 11#define F_INDENT 0x00000004UL
12#define F_CANONICAL 0x00000008 12#define F_CANONICAL 0x00000008UL
13#define F_SPACE_BEFORE 0x00000010 13#define F_SPACE_BEFORE 0x00000010UL
14#define F_SPACE_AFTER 0x00000020 14#define F_SPACE_AFTER 0x00000020UL
15#define F_JSON_RPC 0x00000040
16#define F_ALLOW_NONREF 0x00000080 15#define F_ALLOW_NONREF 0x00000080UL
16#define F_SHRINK 0x00000100UL
17#define F_MAXDEPTH 0xf8000000UL
18#define S_MAXDEPTH 27
19
20#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
21
22// F_SELFCONVERT? <=> to_json/toJson
23// F_BLESSED? <=> { $__class__$ => }
17 24
18#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 25#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
19#define F_DEFAULT 0 26#define F_DEFAULT (13UL << S_MAXDEPTH)
20 27
21#define INIT_SIZE 32 // initial scalar size to be allocated 28#define INIT_SIZE 32 // initial scalar size to be allocated
29#define INDENT_STEP 3 // spaces per indentation level
30
31#define UTF8_MAX_LEN 11 // for perls UTF-X: max. number of octets per character
32#define SHORT_STRING_LEN 512 // special-case strings of up to this size
22 33
23#define SB do { 34#define SB do {
24#define SE } while (0) 35#define SE } while (0)
25 36
26static HV *json_stash; 37static HV *json_stash; // JSON::XS::
38
39/////////////////////////////////////////////////////////////////////////////
40// utility functions
41
42static UV *
43SvJSON (SV *sv)
44{
45 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash))
46 croak ("object is not of type JSON::XS");
47
48 return &SvUVX (SvRV (sv));
49}
50
51static void
52shrink (SV *sv)
53{
54 sv_utf8_downgrade (sv, 1);
55 if (SvLEN (sv) > SvCUR (sv) + 1)
56 {
57#ifdef SvPV_shrink_to_cur
58 SvPV_shrink_to_cur (sv);
59#elif defined (SvPV_renew)
60 SvPV_renew (sv, SvCUR (sv) + 1);
61#endif
62 }
63}
64
65// decode an utf-8 character and return it, or (UV)-1 in
66// case of an error.
67// we special-case "safe" characters from U+80 .. U+7FF,
68// but use the very good perl function to parse anything else.
69// note that we never call this function for a ascii codepoints
70static UV
71decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
72{
73 if (s[0] > 0xdf || s[0] < 0xc2)
74 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
75 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)
76 {
77 *clen = 2;
78 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
79 }
80 else
81 return (UV)-1;
82}
83
84/////////////////////////////////////////////////////////////////////////////
85// encoder
27 86
28// structure used for encoding JSON 87// structure used for encoding JSON
29typedef struct 88typedef struct
30{ 89{
31 char *cur; 90 char *cur; // SvPVX (sv) + current output position
32 STRLEN len; // SvLEN (sv)
33 char *end; // SvEND (sv) 91 char *end; // SvEND (sv)
34 SV *sv; 92 SV *sv; // result scalar
35 UV flags; 93 U32 flags; // F_*
36 int max_recurse; 94 U32 indent; // indentation level
37 int indent; 95 U32 maxdepth; // max. indentation/recursion level
38} enc_t; 96} enc_t;
39
40// structure used for decoding JSON
41typedef struct
42{
43 char *cur;
44 char *end;
45 const char *err;
46 UV flags;
47} dec_t;
48
49static UV *
50SvJSON (SV *sv)
51{
52 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash))
53 croak ("object is not of type JSON::XS");
54
55 return &SvUVX (SvRV (sv));
56}
57
58/////////////////////////////////////////////////////////////////////////////
59 97
60static void 98static void
61need (enc_t *enc, STRLEN len) 99need (enc_t *enc, STRLEN len)
62{ 100{
63 if (enc->cur + len >= enc->end) 101 if (enc->cur + len >= enc->end)
… …
85 123
86 while (str < end) 124 while (str < end)
87 { 125 {
88 unsigned char ch = *(unsigned char *)str; 126 unsigned char ch = *(unsigned char *)str;
89 127
90 if (ch == '"') 128 if (ch >= 0x20 && ch < 0x80) // most common case
91 { 129 {
130 if (ch == '"') // but with slow exceptions
131 {
92 need (enc, len += 1); 132 need (enc, len += 1);
93 *enc->cur++ = '\\'; 133 *enc->cur++ = '\\';
94 *enc->cur++ = '"'; 134 *enc->cur++ = '"';
95 ++str;
96 } 135 }
97 else if (ch == '\\') 136 else if (ch == '\\')
98 { 137 {
99 need (enc, len += 1); 138 need (enc, len += 1);
100 *enc->cur++ = '\\'; 139 *enc->cur++ = '\\';
101 *enc->cur++ = '\\'; 140 *enc->cur++ = '\\';
102 ++str;
103 } 141 }
104 else if (ch >= 0x20 && ch < 0x80) // most common case 142 else
105 {
106 *enc->cur++ = ch; 143 *enc->cur++ = ch;
107 ++str; 144
108 }
109 else if (ch == '\015')
110 {
111 need (enc, len += 1);
112 *enc->cur++ = '\\';
113 *enc->cur++ = 'r';
114 ++str;
115 }
116 else if (ch == '\012')
117 {
118 need (enc, len += 1);
119 *enc->cur++ = '\\';
120 *enc->cur++ = 'n';
121 ++str; 145 ++str;
122 } 146 }
123 else 147 else
124 { 148 {
125 STRLEN clen; 149 switch (ch)
126 UV uch;
127
128 if (is_utf8)
129 { 150 {
130 uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY); 151 case '\010': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'b'; ++str; break;
131 if (clen == (STRLEN)-1) 152 case '\011': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 't'; ++str; break;
132 croak ("malformed UTF-8 character in string, cannot convert to JSON"); 153 case '\012': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'n'; ++str; break;
133 } 154 case '\014': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'f'; ++str; break;
134 else 155 case '\015': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'r'; ++str; break;
135 {
136 uch = ch;
137 clen = 1;
138 }
139 156
140 if (uch < 0x80 || enc->flags & F_ASCII) 157 default:
141 {
142 if (uch > 0xFFFFUL)
143 { 158 {
159 STRLEN clen;
160 UV uch;
161
162 if (is_utf8)
163 {
164 //uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
165 uch = decode_utf8 (str, end - str, &clen);
166 if (clen == (STRLEN)-1)
167 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
168 }
169 else
170 {
171 uch = ch;
172 clen = 1;
173 }
174
175 if (uch > 0x10FFFFUL)
176 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
177
178 if (uch < 0x80 || enc->flags & F_ASCII)
179 {
180 if (uch > 0xFFFFUL)
181 {
144 need (enc, len += 11); 182 need (enc, len += 11);
145 sprintf (enc->cur, "\\u%04x\\u%04x", 183 sprintf (enc->cur, "\\u%04x\\u%04x",
146 (uch - 0x10000) / 0x400 + 0xD800, 184 (int)((uch - 0x10000) / 0x400 + 0xD800),
147 (uch - 0x10000) % 0x400 + 0xDC00); 185 (int)((uch - 0x10000) % 0x400 + 0xDC00));
148 enc->cur += 12; 186 enc->cur += 12;
187 }
188 else
189 {
190 static char hexdigit [16] = "0123456789abcdef";
191 need (enc, len += 5);
192 *enc->cur++ = '\\';
193 *enc->cur++ = 'u';
194 *enc->cur++ = hexdigit [ uch >> 12 ];
195 *enc->cur++ = hexdigit [(uch >> 8) & 15];
196 *enc->cur++ = hexdigit [(uch >> 4) & 15];
197 *enc->cur++ = hexdigit [(uch >> 0) & 15];
198 }
199
200 str += clen;
201 }
202 else if (is_utf8)
203 {
204 need (enc, len += clen);
205 do
206 {
207 *enc->cur++ = *str++;
208 }
209 while (--clen);
210 }
211 else
212 {
213 need (enc, len += UTF8_MAX_LEN - 1); // never more than 11 bytes needed
214 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
215 ++str;
216 }
149 } 217 }
150 else
151 {
152 static char hexdigit [16] = "0123456789abcdef";
153 need (enc, len += 5);
154 *enc->cur++ = '\\';
155 *enc->cur++ = 'u';
156 *enc->cur++ = hexdigit [ uch >> 12 ];
157 *enc->cur++ = hexdigit [(uch >> 8) & 15];
158 *enc->cur++ = hexdigit [(uch >> 4) & 15];
159 *enc->cur++ = hexdigit [(uch >> 0) & 15];
160 }
161
162 str += clen;
163 }
164 else if (is_utf8)
165 {
166 need (enc, len += clen);
167 do
168 {
169 *enc->cur++ = *str++;
170 }
171 while (--clen);
172 }
173 else
174 {
175 need (enc, 10); // never more than 11 bytes needed
176 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
177 ++str;
178 } 218 }
179 } 219 }
180 220
181 --len; 221 --len;
182 } 222 }
183} 223}
184 224
185#define INDENT SB \ 225static void
226encode_indent (enc_t *enc)
227{
186 if (enc->flags & F_INDENT) \ 228 if (enc->flags & F_INDENT)
187 { \ 229 {
188 int i_; \ 230 int spaces = enc->indent * INDENT_STEP;
189 need (enc, enc->indent); \
190 for (i_ = enc->indent * 3; i_--; )\
191 encode_ch (enc, ' '); \
192 } \
193 SE
194 231
195#define SPACE SB need (enc, 1); encode_ch (enc, ' '); SE 232 need (enc, spaces);
196#define NL SB if (enc->flags & F_INDENT) { need (enc, 1); encode_ch (enc, '\n'); } SE 233 memset (enc->cur, ' ', spaces);
197#define COMMA SB \ 234 enc->cur += spaces;
235 }
236}
237
238static void
239encode_space (enc_t *enc)
240{
241 need (enc, 1);
198 encode_ch (enc, ','); \ 242 encode_ch (enc, ' ');
243}
244
245static void
246encode_nl (enc_t *enc)
247{
199 if (enc->flags & F_INDENT) \ 248 if (enc->flags & F_INDENT)
200 NL; \ 249 {
250 need (enc, 1);
251 encode_ch (enc, '\n');
252 }
253}
254
255static void
256encode_comma (enc_t *enc)
257{
258 encode_ch (enc, ',');
259
260 if (enc->flags & F_INDENT)
261 encode_nl (enc);
201 else if (enc->flags & F_SPACE_AFTER) \ 262 else if (enc->flags & F_SPACE_AFTER)
202 SPACE; \ 263 encode_space (enc);
203 SE 264}
204 265
205static void encode_sv (enc_t *enc, SV *sv); 266static void encode_sv (enc_t *enc, SV *sv);
206 267
207static void 268static void
208encode_av (enc_t *enc, AV *av) 269encode_av (enc_t *enc, AV *av)
209{ 270{
210 int i, len = av_len (av); 271 int i, len = av_len (av);
211 272
212 encode_ch (enc, '['); NL; 273 encode_ch (enc, '['); encode_nl (enc);
213 ++enc->indent; 274 ++enc->indent;
214 275
215 for (i = 0; i <= len; ++i) 276 for (i = 0; i <= len; ++i)
216 { 277 {
217 INDENT; 278 encode_indent (enc);
218 encode_sv (enc, *av_fetch (av, i, 0)); 279 encode_sv (enc, *av_fetch (av, i, 0));
219 280
220 if (i < len) 281 if (i < len)
221 COMMA; 282 encode_comma (enc);
222 } 283 }
223 284
224 NL; 285 encode_nl (enc);
225 286
226 --enc->indent; 287 --enc->indent;
227 INDENT; encode_ch (enc, ']'); 288 encode_indent (enc); encode_ch (enc, ']');
228} 289}
229 290
230static void 291static void
231encode_he (enc_t *enc, HE *he) 292encode_he (enc_t *enc, HE *he)
232{ 293{
… …
246 else 307 else
247 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 308 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
248 309
249 encode_ch (enc, '"'); 310 encode_ch (enc, '"');
250 311
251 if (enc->flags & F_SPACE_BEFORE) SPACE; 312 if (enc->flags & F_SPACE_BEFORE) encode_space (enc);
252 encode_ch (enc, ':'); 313 encode_ch (enc, ':');
253 if (enc->flags & F_SPACE_AFTER ) SPACE; 314 if (enc->flags & F_SPACE_AFTER ) encode_space (enc);
254 encode_sv (enc, HeVAL (he)); 315 encode_sv (enc, HeVAL (he));
255} 316}
256 317
257// compare hash entries, used when all keys are bytestrings 318// compare hash entries, used when all keys are bytestrings
258static int 319static int
… …
264 HE *b = *(HE **)b_; 325 HE *b = *(HE **)b_;
265 326
266 STRLEN la = HeKLEN (a); 327 STRLEN la = HeKLEN (a);
267 STRLEN lb = HeKLEN (b); 328 STRLEN lb = HeKLEN (b);
268 329
269 if (!(cmp == memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 330 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb)))
270 cmp = la < lb ? -1 : la == lb ? 0 : 1; 331 cmp = la - lb;
271 332
272 return cmp; 333 return cmp;
273} 334}
274 335
275// compare hash entries, used when some keys are sv's or utf-x 336// compare hash entries, used when some keys are sv's or utf-x
… …
282static void 343static void
283encode_hv (enc_t *enc, HV *hv) 344encode_hv (enc_t *enc, HV *hv)
284{ 345{
285 int count, i; 346 int count, i;
286 347
287 encode_ch (enc, '{'); NL; ++enc->indent; 348 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
288 349
289 if ((count = hv_iterinit (hv))) 350 if ((count = hv_iterinit (hv)))
290 { 351 {
291 // for canonical output we have to sort by keys first 352 // for canonical output we have to sort by keys first
292 // actually, this is mostly due to the stupid so-called 353 // actually, this is mostly due to the stupid so-called
293 // security workaround added somewhere in 5.8.x. 354 // security workaround added somewhere in 5.8.x.
294 // that randomises hash orderings 355 // that randomises hash orderings
295 if (enc->flags & F_CANONICAL) 356 if (enc->flags & F_CANONICAL)
296 { 357 {
297 HE *he, *hes [count]; 358 HE *he, *hes [count]; // if your compiler dies here, you need to enable C99 mode
298 int fast = 1; 359 int fast = 1;
299 360
300 i = 0; 361 i = 0;
301 while ((he = hv_iternext (hv))) 362 while ((he = hv_iternext (hv)))
302 { 363 {
… …
309 370
310 if (fast) 371 if (fast)
311 qsort (hes, count, sizeof (HE *), he_cmp_fast); 372 qsort (hes, count, sizeof (HE *), he_cmp_fast);
312 else 373 else
313 { 374 {
314 // hack to disable "use bytes" 375 // hack to forcefully disable "use bytes"
315 COP *oldcop = PL_curcop, cop; 376 COP cop = *PL_curcop;
316 cop.op_private = 0; 377 cop.op_private = 0;
378
379 ENTER;
380 SAVETMPS;
381
382 SAVEVPTR (PL_curcop);
317 PL_curcop = &cop; 383 PL_curcop = &cop;
318 384
319 SAVETMPS;
320 qsort (hes, count, sizeof (HE *), he_cmp_slow); 385 qsort (hes, count, sizeof (HE *), he_cmp_slow);
386
321 FREETMPS; 387 FREETMPS;
322 388 LEAVE;
323 PL_curcop = oldcop;
324 } 389 }
325 390
326 for (i = 0; i < count; ++i) 391 for (i = 0; i < count; ++i)
327 { 392 {
328 INDENT; 393 encode_indent (enc);
329 encode_he (enc, hes [i]); 394 encode_he (enc, hes [i]);
330 395
331 if (i < count - 1) 396 if (i < count - 1)
332 COMMA; 397 encode_comma (enc);
333 } 398 }
334 399
335 NL; 400 encode_nl (enc);
336 } 401 }
337 else 402 else
338 { 403 {
339 SV *sv; 404 SV *sv;
340 HE *he = hv_iternext (hv); 405 HE *he = hv_iternext (hv);
341 406
342 for (;;) 407 for (;;)
343 { 408 {
344 INDENT; 409 encode_indent (enc);
345 encode_he (enc, he); 410 encode_he (enc, he);
346 411
347 if (!(he = hv_iternext (hv))) 412 if (!(he = hv_iternext (hv)))
348 break; 413 break;
349 414
350 COMMA; 415 encode_comma (enc);
351 } 416 }
352 417
353 NL; 418 encode_nl (enc);
354 } 419 }
355 } 420 }
356 421
357 --enc->indent; INDENT; encode_ch (enc, '}'); 422 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
358} 423}
359 424
360static void 425static void
361encode_sv (enc_t *enc, SV *sv) 426encode_sv (enc_t *enc, SV *sv)
362{ 427{
… …
384 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 449 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv))
385 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 450 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv));
386 } 451 }
387 else if (SvROK (sv)) 452 else if (SvROK (sv))
388 { 453 {
389 if (!--enc->max_recurse) 454 SV *rv = SvRV (sv);
455
456 if (enc->indent >= enc->maxdepth)
390 croak ("data structure too deep (hit recursion limit)"); 457 croak ("data structure too deep (hit recursion limit)");
391 458
392 sv = SvRV (sv);
393
394 switch (SvTYPE (sv)) 459 switch (SvTYPE (rv))
395 { 460 {
396 case SVt_PVAV: encode_av (enc, (AV *)sv); break; 461 case SVt_PVAV: encode_av (enc, (AV *)rv); break;
397 case SVt_PVHV: encode_hv (enc, (HV *)sv); break; 462 case SVt_PVHV: encode_hv (enc, (HV *)rv); break;
398 463
399 default: 464 default:
400 croak ("JSON can only represent references to arrays or hashes"); 465 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
466 SvPV_nolen (sv));
401 } 467 }
402 } 468 }
403 else if (!SvOK (sv)) 469 else if (!SvOK (sv))
404 encode_str (enc, "null", 4, 0); 470 encode_str (enc, "null", 4, 0);
405 else 471 else
406 croak ("encountered perl type that JSON cannot handle"); 472 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
473 SvPV_nolen (sv), SvFLAGS (sv));
407} 474}
408 475
409static SV * 476static SV *
410encode_json (SV *scalar, UV flags) 477encode_json (SV *scalar, U32 flags)
411{ 478{
412 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 479 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar))
413 croak ("hash- or arraref required (not a simple scalar, use allow_nonref to allow this)"); 480 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
414 481
415 enc_t enc; 482 enc_t enc;
416 enc.flags = flags; 483 enc.flags = flags;
417 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 484 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
418 enc.cur = SvPVX (enc.sv); 485 enc.cur = SvPVX (enc.sv);
419 enc.end = SvEND (enc.sv); 486 enc.end = SvEND (enc.sv);
420 enc.max_recurse = 0;
421 enc.indent = 0; 487 enc.indent = 0;
488 enc.maxdepth = DEC_DEPTH (flags);
422 489
423 SvPOK_only (enc.sv); 490 SvPOK_only (enc.sv);
424 encode_sv (&enc, scalar); 491 encode_sv (&enc, scalar);
425 492
426 if (!(flags & (F_ASCII | F_UTF8))) 493 if (!(flags & (F_ASCII | F_UTF8)))
427 SvUTF8_on (enc.sv); 494 SvUTF8_on (enc.sv);
428 495
429 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 496 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
497
498 if (enc.flags & F_SHRINK)
499 shrink (enc.sv);
500
430 return enc.sv; 501 return enc.sv;
431} 502}
432 503
433///////////////////////////////////////////////////////////////////////////// 504/////////////////////////////////////////////////////////////////////////////
505// decoder
434 506
435#define WS \ 507// structure used for decoding JSON
508typedef struct
509{
510 char *cur; // current parser pointer
511 char *end; // end of input string
512 const char *err; // parse error, if != 0
513 U32 flags; // F_*
514 U32 depth; // recursion depth
515 U32 maxdepth; // recursion depth limit
516} dec_t;
517
518static void
519decode_ws (dec_t *dec)
520{
436 for (;;) \ 521 for (;;)
437 { \ 522 {
438 char ch = *dec->cur; \ 523 char ch = *dec->cur;
524
439 if (ch > 0x20 \ 525 if (ch > 0x20
440 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) \ 526 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09))
441 break; \ 527 break;
528
442 ++dec->cur; \ 529 ++dec->cur;
443 } 530 }
531}
444 532
445#define ERR(reason) SB dec->err = reason; goto fail; SE 533#define ERR(reason) SB dec->err = reason; goto fail; SE
534
446#define EXPECT_CH(ch) SB \ 535#define EXPECT_CH(ch) SB \
447 if (*dec->cur != ch) \ 536 if (*dec->cur != ch) \
448 ERR (# ch " expected"); \ 537 ERR (# ch " expected"); \
449 ++dec->cur; \ 538 ++dec->cur; \
450 SE 539 SE
451 540
541#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)")
542#define DEC_DEC_DEPTH --dec->depth
543
452static SV *decode_sv (dec_t *dec); 544static SV *decode_sv (dec_t *dec);
453 545
454static signed char decode_hexdigit[256]; 546static signed char decode_hexdigit[256];
455 547
456static UV 548static UV
457decode_4hex (dec_t *dec) 549decode_4hex (dec_t *dec)
458{ 550{
459 signed char d1, d2, d3, d4; 551 signed char d1, d2, d3, d4;
552 unsigned char *cur = (unsigned char *)dec->cur;
460 553
461 d1 = decode_hexdigit [((unsigned char *)dec->cur) [0]];
462 if (d1 < 0) ERR ("four hexadecimal digits expected"); 554 d1 = decode_hexdigit [cur [0]]; if (d1 < 0) ERR ("four hexadecimal digits expected");
463 d2 = decode_hexdigit [((unsigned char *)dec->cur) [1]];
464 if (d2 < 0) ERR ("four hexadecimal digits expected"); 555 d2 = decode_hexdigit [cur [1]]; if (d2 < 0) ERR ("four hexadecimal digits expected");
465 d3 = decode_hexdigit [((unsigned char *)dec->cur) [2]];
466 if (d3 < 0) ERR ("four hexadecimal digits expected"); 556 d3 = decode_hexdigit [cur [2]]; if (d3 < 0) ERR ("four hexadecimal digits expected");
467 d4 = decode_hexdigit [((unsigned char *)dec->cur) [3]];
468 if (d4 < 0) ERR ("four hexadecimal digits expected"); 557 d4 = decode_hexdigit [cur [3]]; if (d4 < 0) ERR ("four hexadecimal digits expected");
469 558
470 dec->cur += 4; 559 dec->cur += 4;
471 560
472 return ((UV)d1) << 12 561 return ((UV)d1) << 12
473 | ((UV)d2) << 8 562 | ((UV)d2) << 8
… …
476 565
477fail: 566fail:
478 return (UV)-1; 567 return (UV)-1;
479} 568}
480 569
481#define APPEND_GROW(n) SB \
482 if (cur + (n) >= end) \
483 { \
484 STRLEN ofs = cur - SvPVX (sv); \
485 SvGROW (sv, ofs + (n) + 1); \
486 cur = SvPVX (sv) + ofs; \
487 end = SvEND (sv); \
488 } \
489 SE
490
491#define APPEND_CH(ch) SB \
492 APPEND_GROW (1); \
493 *cur++ = (ch); \
494 SE
495
496static SV * 570static SV *
497decode_str (dec_t *dec) 571decode_str (dec_t *dec)
498{ 572{
499 SV *sv = NEWSV (0,2); 573 SV *sv = 0;
500 int utf8 = 0; 574 int utf8 = 0;
501 char *cur = SvPVX (sv);
502 char *end = SvEND (sv);
503 575
504 for (;;) 576 do
505 { 577 {
506 unsigned char ch = *(unsigned char *)dec->cur; 578 char buf [SHORT_STRING_LEN + UTF8_MAX_LEN];
579 char *cur = buf;
507 580
508 if (ch == '"') 581 do
509 break;
510 else if (ch == '\\')
511 { 582 {
512 switch (*++dec->cur) 583 unsigned char ch = *(unsigned char *)dec->cur++;
584
585 if (ch == '"')
513 { 586 {
514 case '\\': 587 --dec->cur;
515 case '/': 588 break;
516 case '"': APPEND_CH (*dec->cur++); break; 589 }
517 590 else if (ch == '\\')
518 case 'b': APPEND_CH ('\010'); ++dec->cur; break; 591 {
519 case 't': APPEND_CH ('\011'); ++dec->cur; break; 592 switch (*dec->cur)
520 case 'n': APPEND_CH ('\012'); ++dec->cur; break;
521 case 'f': APPEND_CH ('\014'); ++dec->cur; break;
522 case 'r': APPEND_CH ('\015'); ++dec->cur; break;
523
524 case 'u':
525 { 593 {
526 UV lo, hi; 594 case '\\':
527 ++dec->cur; 595 case '/':
596 case '"': *cur++ = *dec->cur++; break;
528 597
529 hi = decode_4hex (dec); 598 case 'b': ++dec->cur; *cur++ = '\010'; break;
530 if (hi == (UV)-1) 599 case 't': ++dec->cur; *cur++ = '\011'; break;
531 goto fail; 600 case 'n': ++dec->cur; *cur++ = '\012'; break;
601 case 'f': ++dec->cur; *cur++ = '\014'; break;
602 case 'r': ++dec->cur; *cur++ = '\015'; break;
532 603
533 // possibly a surrogate pair 604 case 'u':
534 if (hi >= 0xd800 && hi < 0xdc00)
535 { 605 {
536 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 606 UV lo, hi;
537 ERR ("missing low surrogate character in surrogate pair");
538
539 dec->cur += 2; 607 ++dec->cur;
540 608
541 lo = decode_4hex (dec); 609 hi = decode_4hex (dec);
542 if (lo == (UV)-1) 610 if (hi == (UV)-1)
543 goto fail; 611 goto fail;
544 612
613 // possibly a surrogate pair
614 if (hi >= 0xd800)
615 if (hi < 0xdc00)
616 {
617 if (dec->cur [0] != '\\' || dec->cur [1] != 'u')
618 ERR ("missing low surrogate character in surrogate pair");
619
620 dec->cur += 2;
621
622 lo = decode_4hex (dec);
623 if (lo == (UV)-1)
624 goto fail;
625
545 if (lo < 0xdc00 || lo >= 0xe000) 626 if (lo < 0xdc00 || lo >= 0xe000)
546 ERR ("surrogate pair expected"); 627 ERR ("surrogate pair expected");
547 628
548 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000; 629 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000;
630 }
631 else if (hi < 0xe000)
632 ERR ("missing high surrogate character in surrogate pair");
633
634 if (hi >= 0x80)
635 {
636 utf8 = 1;
637
638 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
639 }
640 else
641 *cur++ = hi;
549 } 642 }
550 else if (hi >= 0xdc00 && hi < 0xe000)
551 ERR ("missing high surrogate character in surrogate pair");
552
553 if (hi >= 0x80)
554 { 643 break;
555 utf8 = 1;
556 644
557 APPEND_GROW (4); // at most 4 bytes for 21 bits
558 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
559 }
560 else 645 default:
561 APPEND_CH (hi); 646 --dec->cur;
647 ERR ("illegal backslash escape sequence in string");
562 } 648 }
563 break;
564
565 default:
566 --dec->cur;
567 ERR ("illegal backslash escape sequence in string");
568 } 649 }
650 else if (ch >= 0x20 && ch <= 0x7f)
651 *cur++ = ch;
652 else if (ch >= 0x80)
653 {
654 --dec->cur;
655
656 STRLEN clen;
657 UV uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen);
658 if (clen == (STRLEN)-1)
659 ERR ("malformed UTF-8 character in JSON string");
660
661 do
662 {
663 *cur++ = *dec->cur++;
664 }
665 while (--clen);
666
667 utf8 = 1;
668 }
669 else if (!ch)
670 ERR ("unexpected end of string while parsing json string");
671 else
672 ERR ("invalid character encountered");
673
569 } 674 }
570 else if (ch >= 0x20 && ch <= 0x7f) 675 while (cur < buf + SHORT_STRING_LEN);
571 APPEND_CH (*dec->cur++); 676
572 else if (ch >= 0x80) 677 STRLEN len = cur - buf;
678
679 if (sv)
573 { 680 {
574 STRLEN clen; 681 SvGROW (sv, SvCUR (sv) + len + 1);
575 UV uch = utf8n_to_uvuni (dec->cur, dec->end - dec->cur, &clen, UTF8_CHECK_ONLY); 682 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
576 if (clen == (STRLEN)-1) 683 SvCUR_set (sv, SvCUR (sv) + len);
577 ERR ("malformed UTF-8 character in string, cannot convert to JSON");
578
579 APPEND_GROW (clen);
580 do
581 {
582 *cur++ = *dec->cur++;
583 }
584 while (--clen);
585
586 utf8 = 1;
587 } 684 }
588 else if (dec->cur == dec->end)
589 ERR ("unexpected end of string while parsing json string");
590 else 685 else
591 ERR ("invalid character encountered"); 686 sv = newSVpvn (buf, len);
592 } 687 }
688 while (*dec->cur != '"');
593 689
594 ++dec->cur; 690 ++dec->cur;
595 691
596 SvCUR_set (sv, cur - SvPVX (sv)); 692 if (sv)
597 693 {
598 SvPOK_only (sv); 694 SvPOK_only (sv);
599 *SvEND (sv) = 0; 695 *SvEND (sv) = 0;
600 696
601 if (utf8) 697 if (utf8)
602 SvUTF8_on (sv); 698 SvUTF8_on (sv);
699 }
700 else
701 sv = newSVpvn ("", 0);
603 702
604 return sv; 703 return sv;
605 704
606fail: 705fail:
607 SvREFCNT_dec (sv);
608 return 0; 706 return 0;
609} 707}
610 708
611static SV * 709static SV *
612decode_num (dec_t *dec) 710decode_num (dec_t *dec)
… …
693static SV * 791static SV *
694decode_av (dec_t *dec) 792decode_av (dec_t *dec)
695{ 793{
696 AV *av = newAV (); 794 AV *av = newAV ();
697 795
698 WS; 796 DEC_INC_DEPTH;
797 decode_ws (dec);
798
699 if (*dec->cur == ']') 799 if (*dec->cur == ']')
700 ++dec->cur; 800 ++dec->cur;
701 else 801 else
702 for (;;) 802 for (;;)
703 { 803 {
… …
707 if (!value) 807 if (!value)
708 goto fail; 808 goto fail;
709 809
710 av_push (av, value); 810 av_push (av, value);
711 811
712 WS; 812 decode_ws (dec);
713 813
714 if (*dec->cur == ']') 814 if (*dec->cur == ']')
715 { 815 {
716 ++dec->cur; 816 ++dec->cur;
717 break; 817 break;
… …
721 ERR (", or ] expected while parsing array"); 821 ERR (", or ] expected while parsing array");
722 822
723 ++dec->cur; 823 ++dec->cur;
724 } 824 }
725 825
826 DEC_DEC_DEPTH;
726 return newRV_noinc ((SV *)av); 827 return newRV_noinc ((SV *)av);
727 828
728fail: 829fail:
729 SvREFCNT_dec (av); 830 SvREFCNT_dec (av);
831 DEC_DEC_DEPTH;
730 return 0; 832 return 0;
731} 833}
732 834
733static SV * 835static SV *
734decode_hv (dec_t *dec) 836decode_hv (dec_t *dec)
735{ 837{
736 HV *hv = newHV (); 838 HV *hv = newHV ();
737 839
738 WS; 840 DEC_INC_DEPTH;
841 decode_ws (dec);
842
739 if (*dec->cur == '}') 843 if (*dec->cur == '}')
740 ++dec->cur; 844 ++dec->cur;
741 else 845 else
742 for (;;) 846 for (;;)
743 { 847 {
744 SV *key, *value; 848 SV *key, *value;
745 849
746 WS; EXPECT_CH ('"'); 850 decode_ws (dec); EXPECT_CH ('"');
747 851
748 key = decode_str (dec); 852 key = decode_str (dec);
749 if (!key) 853 if (!key)
750 goto fail; 854 goto fail;
751 855
752 WS; EXPECT_CH (':'); 856 decode_ws (dec); EXPECT_CH (':');
753 857
754 value = decode_sv (dec); 858 value = decode_sv (dec);
755 if (!value) 859 if (!value)
756 { 860 {
757 SvREFCNT_dec (key); 861 SvREFCNT_dec (key);
758 goto fail; 862 goto fail;
759 } 863 }
760 864
761 //TODO: optimise
762 hv_store_ent (hv, key, value, 0); 865 hv_store_ent (hv, key, value, 0);
866 SvREFCNT_dec (key);
763 867
764 WS; 868 decode_ws (dec);
765 869
766 if (*dec->cur == '}') 870 if (*dec->cur == '}')
767 { 871 {
768 ++dec->cur; 872 ++dec->cur;
769 break; 873 break;
… …
773 ERR (", or } expected while parsing object/hash"); 877 ERR (", or } expected while parsing object/hash");
774 878
775 ++dec->cur; 879 ++dec->cur;
776 } 880 }
777 881
882 DEC_DEC_DEPTH;
778 return newRV_noinc ((SV *)hv); 883 return newRV_noinc ((SV *)hv);
779 884
780fail: 885fail:
781 SvREFCNT_dec (hv); 886 SvREFCNT_dec (hv);
887 DEC_DEC_DEPTH;
782 return 0; 888 return 0;
783} 889}
784 890
785static SV * 891static SV *
786decode_sv (dec_t *dec) 892decode_sv (dec_t *dec)
787{ 893{
788 WS; 894 decode_ws (dec);
789 switch (*dec->cur) 895 switch (*dec->cur)
790 { 896 {
791 case '"': ++dec->cur; return decode_str (dec); 897 case '"': ++dec->cur; return decode_str (dec);
792 case '[': ++dec->cur; return decode_av (dec); 898 case '[': ++dec->cur; return decode_av (dec);
793 case '{': ++dec->cur; return decode_hv (dec); 899 case '{': ++dec->cur; return decode_hv (dec);
… …
829 ERR ("'null' expected"); 935 ERR ("'null' expected");
830 936
831 break; 937 break;
832 938
833 default: 939 default:
834 ERR ("malformed json string"); 940 ERR ("malformed json string, neither array, object, number, string or atom");
835 break; 941 break;
836 } 942 }
837 943
838fail: 944fail:
839 return 0; 945 return 0;
840} 946}
841 947
842static SV * 948static SV *
843decode_json (SV *string, UV flags) 949decode_json (SV *string, U32 flags)
844{ 950{
845 SV *sv; 951 SV *sv;
846 952
847 if (flags & F_UTF8) 953 if (flags & F_UTF8)
848 sv_utf8_downgrade (string, 0); 954 sv_utf8_downgrade (string, 0);
… …
850 sv_utf8_upgrade (string); 956 sv_utf8_upgrade (string);
851 957
852 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 958 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
853 959
854 dec_t dec; 960 dec_t dec;
855 dec.flags = flags; 961 dec.flags = flags;
856 dec.cur = SvPVX (string); 962 dec.cur = SvPVX (string);
857 dec.end = SvEND (string); 963 dec.end = SvEND (string);
858 dec.err = 0; 964 dec.err = 0;
965 dec.depth = 0;
966 dec.maxdepth = DEC_DEPTH (dec.flags);
859 967
968 *SvEND (sv) = 0; // this shou[ld basically be a nop, too
860 sv = decode_sv (&dec); 969 sv = decode_sv (&dec);
861 970
862 if (!sv) 971 if (!sv)
863 { 972 {
973 IV offset = dec.flags & F_UTF8
974 ? dec.cur - SvPVX (string)
864 IV offset = utf8_distance (dec.cur, SvPVX (string)); 975 : utf8_distance (dec.cur, SvPVX (string));
865 SV *uni = sv_newmortal (); 976 SV *uni = sv_newmortal ();
977
866 // horrible hack to silence warning inside pv_uni_display 978 // horrible hack to silence warning inside pv_uni_display
867 COP cop; 979 COP cop = *PL_curcop;
868 memset (&cop, 0, sizeof (cop));
869 cop.cop_warnings = pWARN_NONE; 980 cop.cop_warnings = pWARN_NONE;
981 ENTER;
870 SAVEVPTR (PL_curcop); 982 SAVEVPTR (PL_curcop);
871 PL_curcop = &cop; 983 PL_curcop = &cop;
872
873 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 984 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
985 LEAVE;
986
874 croak ("%s, at character offset %d (%s)", 987 croak ("%s, at character offset %d (%s)",
875 dec.err, 988 dec.err,
876 (int)offset, 989 (int)offset,
877 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 990 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
878 } 991 }
879 992
880 sv = sv_2mortal (sv); 993 sv = sv_2mortal (sv);
881 994
882 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 995 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv))
883 croak ("JSON object or array expected (but number, string, true, false or null found, use allow_nonref to allow this)"); 996 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
884 997
885 return sv; 998 return sv;
886} 999}
887 1000
1001/////////////////////////////////////////////////////////////////////////////
1002// XS interface functions
1003
888MODULE = JSON::XS PACKAGE = JSON::XS 1004MODULE = JSON::XS PACKAGE = JSON::XS
889 1005
890BOOT: 1006BOOT:
891{ 1007{
892 int i; 1008 int i;
893 1009
894 memset (decode_hexdigit, 0xff, 256); 1010 memset (decode_hexdigit, 0xff, 256);
1011
895 for (i = 10; i--; ) 1012 for (i = 0; i < 256; ++i)
896 decode_hexdigit ['0' + i] = i; 1013 decode_hexdigit [i] =
897 1014 i >= '0' && i <= '9' ? i - '0'
898 for (i = 7; i--; ) 1015 : i >= 'a' && i <= 'f' ? i - 'a' + 10
899 { 1016 : i >= 'A' && i <= 'F' ? i - 'A' + 10
900 decode_hexdigit ['a' + i] = 10 + i; 1017 : -1;
901 decode_hexdigit ['A' + i] = 10 + i;
902 }
903 1018
904 json_stash = gv_stashpv ("JSON::XS", 1); 1019 json_stash = gv_stashpv ("JSON::XS", 1);
905} 1020}
906 1021
907PROTOTYPES: DISABLE 1022PROTOTYPES: DISABLE
… …
910 CODE: 1025 CODE:
911 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1026 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash);
912 OUTPUT: 1027 OUTPUT:
913 RETVAL 1028 RETVAL
914 1029
915SV *ascii (SV *self, int enable) 1030SV *ascii (SV *self, int enable = 1)
916 ALIAS: 1031 ALIAS:
917 ascii = F_ASCII 1032 ascii = F_ASCII
918 utf8 = F_UTF8 1033 utf8 = F_UTF8
919 indent = F_INDENT 1034 indent = F_INDENT
920 canonical = F_CANONICAL 1035 canonical = F_CANONICAL
921 space_before = F_SPACE_BEFORE 1036 space_before = F_SPACE_BEFORE
922 space_after = F_SPACE_AFTER 1037 space_after = F_SPACE_AFTER
923 json_rpc = F_JSON_RPC
924 pretty = F_PRETTY 1038 pretty = F_PRETTY
925 allow_nonref = F_ALLOW_NONREF 1039 allow_nonref = F_ALLOW_NONREF
1040 shrink = F_SHRINK
926 CODE: 1041 CODE:
927{ 1042{
928 UV *uv = SvJSON (self); 1043 UV *uv = SvJSON (self);
929 if (enable) 1044 if (enable)
930 *uv |= ix; 1045 *uv |= ix;
… …
934 RETVAL = newSVsv (self); 1049 RETVAL = newSVsv (self);
935} 1050}
936 OUTPUT: 1051 OUTPUT:
937 RETVAL 1052 RETVAL
938 1053
1054SV *max_depth (SV *self, int max_depth = 0x80000000UL)
1055 CODE:
1056{
1057 UV *uv = SvJSON (self);
1058 UV log2 = 0;
1059
1060 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1061
1062 while ((1UL << log2) < max_depth)
1063 ++log2;
1064
1065 *uv = *uv & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1066
1067 RETVAL = newSVsv (self);
1068}
1069 OUTPUT:
1070 RETVAL
1071
939void encode (SV *self, SV *scalar) 1072void encode (SV *self, SV *scalar)
940 PPCODE: 1073 PPCODE:
941 XPUSHs (encode_json (scalar, *SvJSON (self))); 1074 XPUSHs (encode_json (scalar, *SvJSON (self)));
942 1075
943void decode (SV *self, SV *jsonstr) 1076void decode (SV *self, SV *jsonstr)
… …
945 XPUSHs (decode_json (jsonstr, *SvJSON (self))); 1078 XPUSHs (decode_json (jsonstr, *SvJSON (self)));
946 1079
947PROTOTYPES: ENABLE 1080PROTOTYPES: ENABLE
948 1081
949void to_json (SV *scalar) 1082void to_json (SV *scalar)
1083 ALIAS:
1084 objToJson = 0
950 PPCODE: 1085 PPCODE:
951 XPUSHs (encode_json (scalar, F_UTF8)); 1086 XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8));
952 1087
953void from_json (SV *jsonstr) 1088void from_json (SV *jsonstr)
1089 ALIAS:
1090 jsonToObj = 0
954 PPCODE: 1091 PPCODE:
955 XPUSHs (decode_json (jsonstr, F_UTF8)); 1092 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8));
956 1093

Diff Legend

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