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 1.3 by root, Thu Mar 22 18:10:29 2007 UTC vs.
Revision 1.41 by root, Sat Jun 23 22:53:16 2007 UTC

… …
3#include "XSUB.h" 3#include "XSUB.h"
4 4
5#include "assert.h" 5#include "assert.h"
6#include "string.h" 6#include "string.h"
7#include "stdlib.h" 7#include "stdlib.h"
8#include "stdio.h"
8 9
10#if defined(__BORLANDC__) || defined(_MSC_VER)
11# define snprintf _snprintf // C compilers have this in stdio.h
12#endif
13
14// some old perls do not have this, try to make it work, no
15// guarentees, though.
16#ifndef UTF8_MAXBYTES
17# define UTF8_MAXBYTES 13
18#endif
19
9#define F_ASCII 0x00000001 20#define F_ASCII 0x00000001UL
21#define F_LATIN1 0x00000002UL
10#define F_UTF8 0x00000002 22#define F_UTF8 0x00000004UL
11#define F_INDENT 0x00000004 23#define F_INDENT 0x00000008UL
12#define F_CANONICAL 0x00000008 24#define F_CANONICAL 0x00000010UL
13#define F_SPACE_BEFORE 0x00000010 25#define F_SPACE_BEFORE 0x00000020UL
14#define F_SPACE_AFTER 0x00000020 26#define F_SPACE_AFTER 0x00000040UL
15#define F_JSON_RPC 0x00000040
16#define F_ALLOW_NONREF 0x00000080 27#define F_ALLOW_NONREF 0x00000100UL
28#define F_SHRINK 0x00000200UL
29#define F_MAXDEPTH 0xf8000000UL
30#define S_MAXDEPTH 27
31
32#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
33
34// F_SELFCONVERT? <=> to_json/toJson
35// F_BLESSED? <=> { $__class__$ => }
17 36
18#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 37#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
19#define F_DEFAULT 0 38#define F_DEFAULT (9UL << S_MAXDEPTH)
20 39
21#define INIT_SIZE 32 // initial scalar size to be allocated 40#define INIT_SIZE 32 // initial scalar size to be allocated
41#define INDENT_STEP 3 // spaces per indentation level
42
43#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
22 44
23#define SB do { 45#define SB do {
24#define SE } while (0) 46#define SE } while (0)
25 47
48#if __GNUC__ >= 3
49# define expect(expr,value) __builtin_expect ((expr),(value))
50# define inline inline
51#else
52# define expect(expr,value) (expr)
53# define inline static
54#endif
55
56#define expect_false(expr) expect ((expr) != 0, 0)
57#define expect_true(expr) expect ((expr) != 0, 1)
58
26static HV *json_stash; 59static HV *json_stash; // JSON::XS::
60
61/////////////////////////////////////////////////////////////////////////////
62// utility functions
63
64static UV *
65SvJSON (SV *sv)
66{
67 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash))
68 croak ("object is not of type JSON::XS");
69
70 return &SvUVX (SvRV (sv));
71}
72
73static void
74shrink (SV *sv)
75{
76 sv_utf8_downgrade (sv, 1);
77 if (SvLEN (sv) > SvCUR (sv) + 1)
78 {
79#ifdef SvPV_shrink_to_cur
80 SvPV_shrink_to_cur (sv);
81#elif defined (SvPV_renew)
82 SvPV_renew (sv, SvCUR (sv) + 1);
83#endif
84 }
85}
86
87// decode an utf-8 character and return it, or (UV)-1 in
88// case of an error.
89// we special-case "safe" characters from U+80 .. U+7FF,
90// but use the very good perl function to parse anything else.
91// note that we never call this function for a ascii codepoints
92inline UV
93decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
94{
95 if (expect_false (s[0] > 0xdf || s[0] < 0xc2))
96 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
97 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)
98 {
99 *clen = 2;
100 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
101 }
102 else
103 {
104 *clen = (STRLEN)-1;
105 return (UV)-1;
106 }
107}
108
109/////////////////////////////////////////////////////////////////////////////
110// encoder
27 111
28// structure used for encoding JSON 112// structure used for encoding JSON
29typedef struct 113typedef struct
30{ 114{
31 char *cur; 115 char *cur; // SvPVX (sv) + current output position
32 STRLEN len; // SvLEN (sv)
33 char *end; // SvEND (sv) 116 char *end; // SvEND (sv)
34 SV *sv; 117 SV *sv; // result scalar
35 UV flags; 118 U32 flags; // F_*
36 int max_recurse; 119 U32 indent; // indentation level
37 int indent; 120 U32 maxdepth; // max. indentation/recursion level
38} enc_t; 121} enc_t;
39 122
40// structure used for decoding JSON 123inline void
41typedef struct
42{
43 char *cur;
44 char *end;
45 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
60static void
61need (enc_t *enc, STRLEN len) 124need (enc_t *enc, STRLEN len)
62{ 125{
63 if (enc->cur + len >= enc->end) 126 if (expect_false (enc->cur + len >= enc->end))
64 { 127 {
65 STRLEN cur = enc->cur - SvPVX (enc->sv); 128 STRLEN cur = enc->cur - SvPVX (enc->sv);
66 SvGROW (enc->sv, cur + len + 1); 129 SvGROW (enc->sv, cur + len + 1);
67 enc->cur = SvPVX (enc->sv) + cur; 130 enc->cur = SvPVX (enc->sv) + cur;
68 enc->end = SvEND (enc->sv); 131 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
69 } 132 }
70} 133}
71 134
72static void 135inline void
73encode_ch (enc_t *enc, char ch) 136encode_ch (enc_t *enc, char ch)
74{ 137{
75 need (enc, 1); 138 need (enc, 1);
76 *enc->cur++ = ch; 139 *enc->cur++ = ch;
77} 140}
… …
79static void 142static void
80encode_str (enc_t *enc, char *str, STRLEN len, int is_utf8) 143encode_str (enc_t *enc, char *str, STRLEN len, int is_utf8)
81{ 144{
82 char *end = str + len; 145 char *end = str + len;
83 146
147 need (enc, len);
148
84 while (str < end) 149 while (str < end)
85 { 150 {
86 unsigned char ch = *(unsigned char *)str; 151 unsigned char ch = *(unsigned char *)str;
152
87 if (ch >= 0x20 && ch < 0x80) // most common case 153 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
88 { 154 {
155 if (expect_false (ch == '"')) // but with slow exceptions
156 {
157 need (enc, len += 1);
158 *enc->cur++ = '\\';
159 *enc->cur++ = '"';
160 }
161 else if (expect_false (ch == '\\'))
162 {
163 need (enc, len += 1);
164 *enc->cur++ = '\\';
165 *enc->cur++ = '\\';
166 }
167 else
89 *enc->cur++ = ch; 168 *enc->cur++ = ch;
169
90 str++; 170 ++str;
91 } 171 }
92 else 172 else
93 { 173 {
94 STRLEN clen; 174 switch (ch)
95 UV uch;
96
97 if (is_utf8)
98 { 175 {
99 uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY); 176 case '\010': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'b'; ++str; break;
100 if (clen < 0) 177 case '\011': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 't'; ++str; break;
101 croak ("malformed UTF-8 character in string, cannot convert to JSON"); 178 case '\012': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'n'; ++str; break;
102 } 179 case '\014': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'f'; ++str; break;
103 else 180 case '\015': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'r'; ++str; break;
104 {
105 uch = ch;
106 clen = 1;
107 }
108 181
109 need (enc, len += 6); 182 default:
110
111 if (uch < 0xa0 || enc->flags & F_ASCII)
112 {
113 if (uch > 0xFFFFUL)
114 { 183 {
184 STRLEN clen;
185 UV uch;
186
187 if (is_utf8)
188 {
189 uch = decode_utf8 (str, end - str, &clen);
190 if (clen == (STRLEN)-1)
191 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
192 }
193 else
194 {
195 uch = ch;
115 len += 6; 196 clen = 1;
197 }
198
199 if (uch > 0x10FFFFUL)
200 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
201
202 if (uch < 0x80 || enc->flags & F_ASCII || (enc->flags & F_LATIN1 && uch > 0xFF))
203 {
204 if (uch > 0xFFFFUL)
205 {
116 need (enc, len += 6); 206 need (enc, len += 11);
117 sprintf (enc->cur, "\\u%04x\\u%04x", 207 sprintf (enc->cur, "\\u%04x\\u%04x",
118 (uch - 0x10000) / 0x400 + 0xD800, 208 (int)((uch - 0x10000) / 0x400 + 0xD800),
119 (uch - 0x10000) % 0x400 + 0xDC00); 209 (int)((uch - 0x10000) % 0x400 + 0xDC00));
120 enc->cur += 12; 210 enc->cur += 12;
211 }
212 else
213 {
214 static char hexdigit [16] = "0123456789abcdef";
215 need (enc, len += 5);
216 *enc->cur++ = '\\';
217 *enc->cur++ = 'u';
218 *enc->cur++ = hexdigit [ uch >> 12 ];
219 *enc->cur++ = hexdigit [(uch >> 8) & 15];
220 *enc->cur++ = hexdigit [(uch >> 4) & 15];
221 *enc->cur++ = hexdigit [(uch >> 0) & 15];
222 }
223
224 str += clen;
121 } 225 }
226 else if (enc->flags & F_LATIN1)
227 {
228 *enc->cur++ = uch;
229 str += clen;
230 }
231 else if (is_utf8)
232 {
233 need (enc, len += clen);
234 do
235 {
236 *enc->cur++ = *str++;
237 }
238 while (--clen);
239 }
122 else 240 else
123 { 241 {
124 sprintf (enc->cur, "\\u%04x", uch); 242 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
125 enc->cur += 6; 243 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
244 ++str;
245 }
126 } 246 }
127 } 247 }
128 else if (is_utf8)
129 {
130 memcpy (enc->cur, str, clen);
131 enc->cur += clen;
132 }
133 else
134 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
135
136 str += clen;
137 } 248 }
138 249
139 --len; 250 --len;
140 } 251 }
141} 252}
142 253
143#define INDENT SB \ 254inline void
255encode_indent (enc_t *enc)
256{
144 if (enc->flags & F_INDENT) \ 257 if (enc->flags & F_INDENT)
145 { \ 258 {
146 int i_; \ 259 int spaces = enc->indent * INDENT_STEP;
147 need (enc, enc->indent); \
148 for (i_ = enc->indent * 3; i_--; )\
149 encode_ch (enc, ' '); \
150 } \
151 SE
152 260
153#define SPACE SB need (enc, 1); encode_ch (enc, ' '); SE 261 need (enc, spaces);
154#define NL SB if (enc->flags & F_INDENT) { need (enc, 1); encode_ch (enc, '\n'); } SE 262 memset (enc->cur, ' ', spaces);
155#define COMMA SB \ 263 enc->cur += spaces;
264 }
265}
266
267inline void
268encode_space (enc_t *enc)
269{
270 need (enc, 1);
156 encode_ch (enc, ','); \ 271 encode_ch (enc, ' ');
272}
273
274inline void
275encode_nl (enc_t *enc)
276{
157 if (enc->flags & F_INDENT) \ 277 if (enc->flags & F_INDENT)
158 NL; \ 278 {
279 need (enc, 1);
280 encode_ch (enc, '\n');
281 }
282}
283
284inline void
285encode_comma (enc_t *enc)
286{
287 encode_ch (enc, ',');
288
289 if (enc->flags & F_INDENT)
290 encode_nl (enc);
159 else if (enc->flags & F_SPACE_AFTER) \ 291 else if (enc->flags & F_SPACE_AFTER)
160 SPACE; \ 292 encode_space (enc);
161 SE 293}
162 294
163static void encode_sv (enc_t *enc, SV *sv); 295static void encode_sv (enc_t *enc, SV *sv);
164 296
165static void 297static void
166encode_av (enc_t *enc, AV *av) 298encode_av (enc_t *enc, AV *av)
167{ 299{
168 int i, len = av_len (av); 300 int i, len = av_len (av);
169 301
170 encode_ch (enc, '['); NL; 302 if (enc->indent >= enc->maxdepth)
303 croak ("data structure too deep (hit recursion limit)");
304
305 encode_ch (enc, '['); encode_nl (enc);
171 ++enc->indent; 306 ++enc->indent;
172 307
173 for (i = 0; i <= len; ++i) 308 for (i = 0; i <= len; ++i)
174 { 309 {
175 INDENT; 310 encode_indent (enc);
176 encode_sv (enc, *av_fetch (av, i, 0)); 311 encode_sv (enc, *av_fetch (av, i, 0));
177 312
178 if (i < len) 313 if (i < len)
179 COMMA; 314 encode_comma (enc);
180 } 315 }
181 316
182 NL; 317 encode_nl (enc);
183 318
184 --enc->indent; 319 --enc->indent;
185 INDENT; encode_ch (enc, ']'); 320 encode_indent (enc); encode_ch (enc, ']');
186} 321}
187 322
188static void 323static void
189encode_he (enc_t *enc, HE *he) 324encode_he (enc_t *enc, HE *he)
190{ 325{
… …
192 327
193 if (HeKLEN (he) == HEf_SVKEY) 328 if (HeKLEN (he) == HEf_SVKEY)
194 { 329 {
195 SV *sv = HeSVKEY (he); 330 SV *sv = HeSVKEY (he);
196 STRLEN len; 331 STRLEN len;
332 char *str;
333
334 SvGETMAGIC (sv);
197 char *str = SvPV (sv, len); 335 str = SvPV (sv, len);
198 336
199 encode_str (enc, str, len, SvUTF8 (sv)); 337 encode_str (enc, str, len, SvUTF8 (sv));
200 } 338 }
201 else 339 else
202 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 340 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
203 341
204 encode_ch (enc, '"'); 342 encode_ch (enc, '"');
205 343
206 if (enc->flags & F_SPACE_BEFORE) SPACE; 344 if (enc->flags & F_SPACE_BEFORE) encode_space (enc);
207 encode_ch (enc, ':'); 345 encode_ch (enc, ':');
208 if (enc->flags & F_SPACE_AFTER ) SPACE; 346 if (enc->flags & F_SPACE_AFTER ) encode_space (enc);
209 encode_sv (enc, HeVAL (he)); 347 encode_sv (enc, HeVAL (he));
210} 348}
211 349
212// compare hash entries, used when all keys are bytestrings 350// compare hash entries, used when all keys are bytestrings
213static int 351static int
… …
219 HE *b = *(HE **)b_; 357 HE *b = *(HE **)b_;
220 358
221 STRLEN la = HeKLEN (a); 359 STRLEN la = HeKLEN (a);
222 STRLEN lb = HeKLEN (b); 360 STRLEN lb = HeKLEN (b);
223 361
224 if (!(cmp == memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 362 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb)))
225 cmp = la < lb ? -1 : la == lb ? 0 : 1; 363 cmp = la - lb;
226 364
227 return cmp; 365 return cmp;
228} 366}
229 367
230// compare hash entries, used when some keys are sv's or utf-x 368// compare hash entries, used when some keys are sv's or utf-x
… …
237static void 375static void
238encode_hv (enc_t *enc, HV *hv) 376encode_hv (enc_t *enc, HV *hv)
239{ 377{
240 int count, i; 378 int count, i;
241 379
380 if (enc->indent >= enc->maxdepth)
381 croak ("data structure too deep (hit recursion limit)");
382
242 encode_ch (enc, '{'); NL; ++enc->indent; 383 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
243 384
244 if ((count = hv_iterinit (hv))) 385 if ((count = hv_iterinit (hv)))
245 { 386 {
246 // for canonical output we have to sort by keys first 387 // for canonical output we have to sort by keys first
247 // actually, this is mostly due to the stupid so-called 388 // actually, this is mostly due to the stupid so-called
248 // security workaround added somewhere in 5.8.x. 389 // security workaround added somewhere in 5.8.x.
249 // that randomises hash orderings 390 // that randomises hash orderings
250 if (enc->flags & F_CANONICAL) 391 if (enc->flags & F_CANONICAL)
251 { 392 {
252 HE *he, *hes [count];
253 int fast = 1; 393 int fast = 1;
394 HE *he;
395#if defined(__BORLANDC__) || defined(_MSC_VER)
396 HE **hes = _alloca (count * sizeof (HE));
397#else
398 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
399#endif
254 400
255 i = 0; 401 i = 0;
256 while ((he = hv_iternext (hv))) 402 while ((he = hv_iternext (hv)))
257 { 403 {
258 hes [i++] = he; 404 hes [i++] = he;
… …
264 410
265 if (fast) 411 if (fast)
266 qsort (hes, count, sizeof (HE *), he_cmp_fast); 412 qsort (hes, count, sizeof (HE *), he_cmp_fast);
267 else 413 else
268 { 414 {
269 // hack to disable "use bytes" 415 // hack to forcefully disable "use bytes"
270 COP *oldcop = PL_curcop, cop; 416 COP cop = *PL_curcop;
271 cop.op_private = 0; 417 cop.op_private = 0;
418
419 ENTER;
420 SAVETMPS;
421
422 SAVEVPTR (PL_curcop);
272 PL_curcop = &cop; 423 PL_curcop = &cop;
273 424
274 SAVETMPS;
275 qsort (hes, count, sizeof (HE *), he_cmp_slow); 425 qsort (hes, count, sizeof (HE *), he_cmp_slow);
426
276 FREETMPS; 427 FREETMPS;
277 428 LEAVE;
278 PL_curcop = oldcop;
279 } 429 }
280 430
281 for (i = 0; i < count; ++i) 431 for (i = 0; i < count; ++i)
282 { 432 {
283 INDENT; 433 encode_indent (enc);
284 encode_he (enc, hes [i]); 434 encode_he (enc, hes [i]);
285 435
286 if (i < count - 1) 436 if (i < count - 1)
287 COMMA; 437 encode_comma (enc);
288 } 438 }
289 439
290 NL; 440 encode_nl (enc);
291 } 441 }
292 else 442 else
293 { 443 {
294 SV *sv;
295 HE *he = hv_iternext (hv); 444 HE *he = hv_iternext (hv);
296 445
297 for (;;) 446 for (;;)
298 { 447 {
299 INDENT; 448 encode_indent (enc);
300 encode_he (enc, he); 449 encode_he (enc, he);
301 450
302 if (!(he = hv_iternext (hv))) 451 if (!(he = hv_iternext (hv)))
303 break; 452 break;
304 453
305 COMMA; 454 encode_comma (enc);
306 } 455 }
307 456
308 NL; 457 encode_nl (enc);
309 } 458 }
310 } 459 }
311 460
312 --enc->indent; INDENT; encode_ch (enc, '}'); 461 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
462}
463
464// encode objects, arrays and special \0=false and \1=true values.
465static void
466encode_rv (enc_t *enc, SV *sv)
467{
468 svtype svt;
469
470 SvGETMAGIC (sv);
471 svt = SvTYPE (sv);
472
473 if (svt == SVt_PVHV)
474 encode_hv (enc, (HV *)sv);
475 else if (svt == SVt_PVAV)
476 encode_av (enc, (AV *)sv);
477 else if (svt < SVt_PVAV)
478 {
479 if (SvNIOK (sv) && SvIV (sv) == 0)
480 encode_str (enc, "false", 5, 0);
481 else if (SvNIOK (sv) && SvIV (sv) == 1)
482 encode_str (enc, "true", 4, 0);
483 else
484 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
485 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
486 }
487 else
488 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
489 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
313} 490}
314 491
315static void 492static void
316encode_sv (enc_t *enc, SV *sv) 493encode_sv (enc_t *enc, SV *sv)
317{ 494{
495 SvGETMAGIC (sv);
496
318 if (SvPOKp (sv)) 497 if (SvPOKp (sv))
319 { 498 {
320 STRLEN len; 499 STRLEN len;
321 char *str = SvPV (sv, len); 500 char *str = SvPV (sv, len);
322 encode_ch (enc, '"'); 501 encode_ch (enc, '"');
323 encode_str (enc, str, len, SvUTF8 (sv)); 502 encode_str (enc, str, len, SvUTF8 (sv));
324 encode_ch (enc, '"'); 503 encode_ch (enc, '"');
325 } 504 }
326 else if (SvNOKp (sv)) 505 else if (SvNOKp (sv))
327 { 506 {
507 // trust that perl will do the right thing w.r.t. JSON syntax.
328 need (enc, NV_DIG + 32); 508 need (enc, NV_DIG + 32);
329 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 509 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
330 enc->cur += strlen (enc->cur); 510 enc->cur += strlen (enc->cur);
331 } 511 }
332 else if (SvIOKp (sv)) 512 else if (SvIOKp (sv))
333 { 513 {
334 need (enc, 64); 514 // we assume we can always read an IV as a UV
515 if (SvUV (sv) & ~(UV)0x7fff)
516 {
517 // large integer, use the (rather slow) snprintf way.
518 need (enc, sizeof (UV) * 3);
335 enc->cur += 519 enc->cur +=
336 SvIsUV(sv) 520 SvIsUV(sv)
337 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 521 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
338 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 522 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
523 }
524 else
525 {
526 // optimise the "small number case"
527 // code will likely be branchless and use only a single multiplication
528 I32 i = SvIV (sv);
529 U32 u;
530 char digit, nz = 0;
531
532 need (enc, 6);
533
534 *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
535 u = i < 0 ? -i : i;
536
537 // convert to 4.28 fixed-point representation
538 u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
539
540 // now output digit by digit, each time masking out the integer part
541 // and multiplying by 5 while moving the decimal point one to the right,
542 // resulting in a net multiplication by 10.
543 // we always write the digit to memory but conditionally increment
544 // the pointer, to ease the usage of conditional move instructions.
545 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5;
546 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5;
547 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5;
548 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5;
549 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
550 }
339 } 551 }
340 else if (SvROK (sv)) 552 else if (SvROK (sv))
341 { 553 encode_rv (enc, SvRV (sv));
342 if (!--enc->max_recurse)
343 croak ("data structure too deep (hit recursion limit)");
344
345 sv = SvRV (sv);
346
347 switch (SvTYPE (sv))
348 {
349 case SVt_PVAV: encode_av (enc, (AV *)sv); break;
350 case SVt_PVHV: encode_hv (enc, (HV *)sv); break;
351
352 default:
353 croak ("JSON can only represent references to arrays or hashes");
354 }
355 }
356 else if (!SvOK (sv)) 554 else if (!SvOK (sv))
357 encode_str (enc, "null", 4, 0); 555 encode_str (enc, "null", 4, 0);
358 else 556 else
359 croak ("encountered perl type that JSON cannot handle"); 557 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
558 SvPV_nolen (sv), SvFLAGS (sv));
360} 559}
361 560
362static SV * 561static SV *
363encode_json (SV *scalar, UV flags) 562encode_json (SV *scalar, U32 flags)
364{ 563{
564 enc_t enc;
565
365 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 566 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar))
366 croak ("hash- or arraref required (not a simple scalar, use allow_nonref to allow this)"); 567 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
367 568
368 enc_t enc;
369 enc.flags = flags; 569 enc.flags = flags;
370 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 570 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
371 enc.cur = SvPVX (enc.sv); 571 enc.cur = SvPVX (enc.sv);
372 enc.end = SvEND (enc.sv); 572 enc.end = SvEND (enc.sv);
373 enc.max_recurse = 0;
374 enc.indent = 0; 573 enc.indent = 0;
574 enc.maxdepth = DEC_DEPTH (flags);
375 575
376 SvPOK_only (enc.sv); 576 SvPOK_only (enc.sv);
377 encode_sv (&enc, scalar); 577 encode_sv (&enc, scalar);
378 578
579 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
580 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
581
379 if (!(flags & (F_ASCII | F_UTF8))) 582 if (!(flags & (F_ASCII | F_LATIN1 | F_UTF8)))
380 SvUTF8_on (enc.sv); 583 SvUTF8_on (enc.sv);
381 584
382 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 585 if (enc.flags & F_SHRINK)
586 shrink (enc.sv);
587
383 return enc.sv; 588 return enc.sv;
384} 589}
385 590
386///////////////////////////////////////////////////////////////////////////// 591/////////////////////////////////////////////////////////////////////////////
592// decoder
387 593
388#define WS \ 594// structure used for decoding JSON
595typedef struct
596{
597 char *cur; // current parser pointer
598 char *end; // end of input string
599 const char *err; // parse error, if != 0
600 U32 flags; // F_*
601 U32 depth; // recursion depth
602 U32 maxdepth; // recursion depth limit
603} dec_t;
604
605inline void
606decode_ws (dec_t *dec)
607{
389 for (;;) \ 608 for (;;)
390 { \ 609 {
391 char ch = *dec->cur; \ 610 char ch = *dec->cur;
611
392 if (ch > 0x20 \ 612 if (ch > 0x20
393 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) \ 613 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09))
394 break; \ 614 break;
615
395 ++dec->cur; \ 616 ++dec->cur;
396 } 617 }
618}
397 619
398#define ERR(reason) SB dec->err = reason; goto fail; SE 620#define ERR(reason) SB dec->err = reason; goto fail; SE
621
399#define EXPECT_CH(ch) SB \ 622#define EXPECT_CH(ch) SB \
400 if (*dec->cur != ch) \ 623 if (*dec->cur != ch) \
401 ERR (# ch " expected"); \ 624 ERR (# ch " expected"); \
402 ++dec->cur; \ 625 ++dec->cur; \
403 SE 626 SE
404 627
628#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)")
629#define DEC_DEC_DEPTH --dec->depth
630
405static SV *decode_sv (dec_t *dec); 631static SV *decode_sv (dec_t *dec);
406
407#define APPEND_CH(ch) SB \
408 SvGROW (sv, cur + 1 + 1); \
409 SvPVX (sv)[cur++] = (ch); \
410 SE
411 632
412static signed char decode_hexdigit[256]; 633static signed char decode_hexdigit[256];
413 634
414static UV 635static UV
415decode_4hex (dec_t *dec) 636decode_4hex (dec_t *dec)
416{ 637{
417 signed char d1, d2, d3, d4; 638 signed char d1, d2, d3, d4;
639 unsigned char *cur = (unsigned char *)dec->cur;
418 640
419 d1 = decode_hexdigit [((unsigned char *)dec->cur) [0]]; 641 d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected");
420 if (d1 < 0) ERR ("four hexadecimal digits expected"); 642 d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected");
421 d2 = decode_hexdigit [((unsigned char *)dec->cur) [1]]; 643 d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected");
422 if (d2 < 0) ERR ("four hexadecimal digits expected"); 644 d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected");
423 d3 = decode_hexdigit [((unsigned char *)dec->cur) [2]];
424 if (d3 < 0) ERR ("four hexadecimal digits expected");
425 d4 = decode_hexdigit [((unsigned char *)dec->cur) [3]];
426 if (d4 < 0) ERR ("four hexadecimal digits expected");
427 645
428 dec->cur += 4; 646 dec->cur += 4;
429 647
430 return ((UV)d1) << 12 648 return ((UV)d1) << 12
431 | ((UV)d2) << 8 649 | ((UV)d2) << 8
… …
437} 655}
438 656
439static SV * 657static SV *
440decode_str (dec_t *dec) 658decode_str (dec_t *dec)
441{ 659{
442 SV *sv = NEWSV (0,2); 660 SV *sv = 0;
443 STRLEN cur = 0;
444 int utf8 = 0; 661 int utf8 = 0;
662 char *dec_cur = dec->cur;
445 663
446 for (;;) 664 do
447 { 665 {
448 unsigned char ch = *(unsigned char *)dec->cur; 666 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
667 char *cur = buf;
449 668
450 if (ch == '"') 669 do
451 break;
452 else if (ch == '\\')
453 { 670 {
454 switch (*++dec->cur) 671 unsigned char ch = *(unsigned char *)dec_cur++;
672
673 if (expect_false (ch == '"'))
455 { 674 {
456 case '\\': 675 --dec_cur;
457 case '/': 676 break;
458 case '"': APPEND_CH (*dec->cur++); break; 677 }
459 678 else if (expect_false (ch == '\\'))
460 case 'b': APPEND_CH ('\010'); ++dec->cur; break; 679 {
461 case 't': APPEND_CH ('\011'); ++dec->cur; break; 680 switch (*dec_cur)
462 case 'n': APPEND_CH ('\012'); ++dec->cur; break;
463 case 'f': APPEND_CH ('\014'); ++dec->cur; break;
464 case 'r': APPEND_CH ('\015'); ++dec->cur; break;
465
466 case 'u':
467 { 681 {
468 UV lo, hi; 682 case '\\':
469 ++dec->cur; 683 case '/':
684 case '"': *cur++ = *dec_cur++; break;
470 685
471 hi = decode_4hex (dec); 686 case 'b': ++dec_cur; *cur++ = '\010'; break;
472 if (hi == (UV)-1) 687 case 't': ++dec_cur; *cur++ = '\011'; break;
473 goto fail; 688 case 'n': ++dec_cur; *cur++ = '\012'; break;
689 case 'f': ++dec_cur; *cur++ = '\014'; break;
690 case 'r': ++dec_cur; *cur++ = '\015'; break;
474 691
475 // possibly a surrogate pair 692 case 'u':
476 if (hi >= 0xd800 && hi < 0xdc00)
477 { 693 {
478 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 694 UV lo, hi;
479 ERR ("illegal surrogate character"); 695 ++dec_cur;
480 696
481 dec->cur += 2; 697 dec->cur = dec_cur;
482
483 lo = decode_4hex (dec); 698 hi = decode_4hex (dec);
699 dec_cur = dec->cur;
484 if (lo == (UV)-1) 700 if (hi == (UV)-1)
485 goto fail; 701 goto fail;
486 702
703 // possibly a surrogate pair
704 if (hi >= 0xd800)
705 if (hi < 0xdc00)
706 {
707 if (dec_cur [0] != '\\' || dec_cur [1] != 'u')
708 ERR ("missing low surrogate character in surrogate pair");
709
710 dec_cur += 2;
711
712 dec->cur = dec_cur;
713 lo = decode_4hex (dec);
714 dec_cur = dec->cur;
715 if (lo == (UV)-1)
716 goto fail;
717
487 if (lo < 0xdc00 || lo >= 0xe000) 718 if (lo < 0xdc00 || lo >= 0xe000)
488 ERR ("surrogate pair expected"); 719 ERR ("surrogate pair expected");
489 720
490 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000; 721 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000;
722 }
723 else if (hi < 0xe000)
724 ERR ("missing high surrogate character in surrogate pair");
725
726 if (hi >= 0x80)
727 {
728 utf8 = 1;
729
730 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
731 }
732 else
733 *cur++ = hi;
491 } 734 }
492 else if (lo >= 0xdc00 && lo < 0xe000)
493 ERR ("illegal surrogate character");
494
495 if (hi >= 0x80)
496 { 735 break;
497 utf8 = 1;
498 736
499 SvGROW (sv, cur + 4 + 1); // at most 4 bytes for 21 bits
500 cur = (char *)uvuni_to_utf8_flags (SvPVX (sv) + cur, hi, 0) - SvPVX (sv);
501 }
502 else 737 default:
503 APPEND_CH (hi); 738 --dec_cur;
739 ERR ("illegal backslash escape sequence in string");
504 } 740 }
741 }
742 else if (expect_true (ch >= 0x20 && ch <= 0x7f))
743 *cur++ = ch;
744 else if (ch >= 0x80)
745 {
746 STRLEN clen;
747 UV uch;
748
749 --dec_cur;
750
751 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
752 if (clen == (STRLEN)-1)
753 ERR ("malformed UTF-8 character in JSON string");
754
755 do
756 *cur++ = *dec_cur++;
757 while (--clen);
758
759 utf8 = 1;
760 }
761 else
762 {
763 --dec_cur;
764
765 if (!ch)
766 ERR ("unexpected end of string while parsing JSON string");
505 break; 767 else
768 ERR ("invalid character encountered while parsing JSON string");
506 } 769 }
507 } 770 }
508 else if (ch >= 0x20 && ch <= 0x7f) 771 while (cur < buf + SHORT_STRING_LEN);
509 APPEND_CH (*dec->cur++); 772
510 else if (ch >= 0x80) 773 {
774 STRLEN len = cur - buf;
775
776 if (sv)
511 { 777 {
512 STRLEN clen; 778 SvGROW (sv, SvCUR (sv) + len + 1);
513 UV uch = utf8n_to_uvuni (dec->cur, dec->end - dec->cur, &clen, UTF8_CHECK_ONLY); 779 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
514 if (clen < 0) 780 SvCUR_set (sv, SvCUR (sv) + len);
515 ERR ("malformed UTF-8 character in string, cannot convert to JSON");
516
517 SvGROW (sv, cur + clen + 1); // at most 4 bytes for 21 bits
518 memcpy (SvPVX (sv) + cur, dec->cur, clen);
519 dec->cur += clen;
520 } 781 }
521 else 782 else
522 ERR ("invalid character encountered"); 783 sv = newSVpvn (buf, len);
523 } 784 }
785 }
786 while (*dec_cur != '"');
524 787
525 ++dec->cur; 788 ++dec_cur;
526 789
790 if (sv)
791 {
527 SvPOK_only (sv); 792 SvPOK_only (sv);
528
529 SvCUR_set (sv, cur);
530 *SvEND (sv) = 0; 793 *SvEND (sv) = 0;
531 794
532 if (utf8) 795 if (utf8)
533 SvUTF8_on (sv); 796 SvUTF8_on (sv);
797 }
798 else
799 sv = newSVpvn ("", 0);
534 800
801 dec->cur = dec_cur;
535 return sv; 802 return sv;
536 803
537fail: 804fail:
538 SvREFCNT_dec (sv); 805 dec->cur = dec_cur;
539 return 0; 806 return 0;
540} 807}
541 808
542static SV * 809static SV *
543decode_num (dec_t *dec) 810decode_num (dec_t *dec)
… …
553 { 820 {
554 ++dec->cur; 821 ++dec->cur;
555 if (*dec->cur >= '0' && *dec->cur <= '9') 822 if (*dec->cur >= '0' && *dec->cur <= '9')
556 ERR ("malformed number (leading zero must not be followed by another digit)"); 823 ERR ("malformed number (leading zero must not be followed by another digit)");
557 } 824 }
558 825 else if (*dec->cur < '0' || *dec->cur > '9')
559 // int 826 ERR ("malformed number (no digits after initial minus)");
827 else
828 do
829 {
830 ++dec->cur;
831 }
560 while (*dec->cur >= '0' && *dec->cur <= '9') 832 while (*dec->cur >= '0' && *dec->cur <= '9');
561 ++dec->cur;
562 833
563 // [frac] 834 // [frac]
564 if (*dec->cur == '.') 835 if (*dec->cur == '.')
565 { 836 {
566 is_nv = 1; 837 ++dec->cur;
838
839 if (*dec->cur < '0' || *dec->cur > '9')
840 ERR ("malformed number (no digits after decimal point)");
567 841
568 do 842 do
569 { 843 {
570 ++dec->cur; 844 ++dec->cur;
571 } 845 }
572 while (*dec->cur >= '0' && *dec->cur <= '9'); 846 while (*dec->cur >= '0' && *dec->cur <= '9');
847
848 is_nv = 1;
573 } 849 }
574 850
575 // [exp] 851 // [exp]
576 if (*dec->cur == 'e' || *dec->cur == 'E') 852 if (*dec->cur == 'e' || *dec->cur == 'E')
577 { 853 {
578 is_nv = 1;
579
580 ++dec->cur; 854 ++dec->cur;
855
581 if (*dec->cur == '-' || *dec->cur == '+') 856 if (*dec->cur == '-' || *dec->cur == '+')
582 ++dec->cur; 857 ++dec->cur;
583 858
859 if (*dec->cur < '0' || *dec->cur > '9')
860 ERR ("malformed number (no digits after exp sign)");
861
862 do
863 {
864 ++dec->cur;
865 }
584 while (*dec->cur >= '0' && *dec->cur <= '9') 866 while (*dec->cur >= '0' && *dec->cur <= '9');
585 ++dec->cur; 867
868 is_nv = 1;
586 } 869 }
587 870
588 if (!is_nv) 871 if (!is_nv)
589 { 872 {
590 UV uv; 873 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so
591 int numtype = grok_number (start, dec->cur - start, &uv); 874 if (*start == '-')
592 if (numtype & IS_NUMBER_IN_UV) 875 switch (dec->cur - start)
593 if (numtype & IS_NUMBER_NEG)
594 { 876 {
595 if (uv < (UV)IV_MIN) 877 case 2: return newSViv (-( start [1] - '0' * 1));
596 return newSViv (-(IV)uv); 878 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
879 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
880 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
597 } 881 }
882 else
883 switch (dec->cur - start)
884 {
885 case 1: return newSViv ( start [0] - '0' * 1);
886 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
887 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
888 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
889 }
890
891 {
892 UV uv;
893 int numtype = grok_number (start, dec->cur - start, &uv);
894 if (numtype & IS_NUMBER_IN_UV)
895 if (numtype & IS_NUMBER_NEG)
896 {
897 if (uv < (UV)IV_MIN)
898 return newSViv (-(IV)uv);
899 }
598 else 900 else
599 return newSVuv (uv); 901 return newSVuv (uv);
902
903 // here would likely be the place for bigint support
600 } 904 }
905 }
601 906
907 // if we ever support bigint or bigfloat, this is the place for bigfloat
602 return newSVnv (Atof (start)); 908 return newSVnv (Atof (start));
603 909
604fail: 910fail:
605 return 0; 911 return 0;
606} 912}
… …
608static SV * 914static SV *
609decode_av (dec_t *dec) 915decode_av (dec_t *dec)
610{ 916{
611 AV *av = newAV (); 917 AV *av = newAV ();
612 918
919 DEC_INC_DEPTH;
920 decode_ws (dec);
921
922 if (*dec->cur == ']')
923 ++dec->cur;
924 else
613 for (;;) 925 for (;;)
614 { 926 {
615 SV *value; 927 SV *value;
616 928
617 value = decode_sv (dec); 929 value = decode_sv (dec);
618 if (!value) 930 if (!value)
619 goto fail; 931 goto fail;
620 932
621 av_push (av, value); 933 av_push (av, value);
622 934
623 WS; 935 decode_ws (dec);
624 936
625 if (*dec->cur == ']') 937 if (*dec->cur == ']')
626 { 938 {
627 ++dec->cur; 939 ++dec->cur;
628 break; 940 break;
941 }
629 } 942
630
631 if (*dec->cur != ',') 943 if (*dec->cur != ',')
632 ERR (", or ] expected while parsing array"); 944 ERR (", or ] expected while parsing array");
633 945
634 ++dec->cur; 946 ++dec->cur;
635 } 947 }
636 948
949 DEC_DEC_DEPTH;
637 return newRV_noinc ((SV *)av); 950 return newRV_noinc ((SV *)av);
638 951
639fail: 952fail:
640 SvREFCNT_dec (av); 953 SvREFCNT_dec (av);
954 DEC_DEC_DEPTH;
641 return 0; 955 return 0;
642} 956}
643 957
644static SV * 958static SV *
645decode_hv (dec_t *dec) 959decode_hv (dec_t *dec)
646{ 960{
647 HV *hv = newHV (); 961 HV *hv = newHV ();
648 962
963 DEC_INC_DEPTH;
964 decode_ws (dec);
965
966 if (*dec->cur == '}')
967 ++dec->cur;
968 else
649 for (;;) 969 for (;;)
650 { 970 {
651 SV *key, *value; 971 SV *key, *value;
652 972
653 WS; EXPECT_CH ('"'); 973 decode_ws (dec); EXPECT_CH ('"');
654 974
655 key = decode_str (dec); 975 key = decode_str (dec);
656 if (!key) 976 if (!key)
657 goto fail;
658
659 WS; EXPECT_CH (':');
660
661 value = decode_sv (dec);
662 if (!value)
663 {
664 SvREFCNT_dec (key);
665 goto fail; 977 goto fail;
978
979 decode_ws (dec); EXPECT_CH (':');
980
981 value = decode_sv (dec);
982 if (!value)
983 {
984 SvREFCNT_dec (key);
985 goto fail;
666 } 986 }
667 987
668 //TODO: optimise
669 hv_store_ent (hv, key, value, 0); 988 hv_store_ent (hv, key, value, 0);
989 SvREFCNT_dec (key);
670 990
671 WS; 991 decode_ws (dec);
672 992
673 if (*dec->cur == '}') 993 if (*dec->cur == '}')
674 { 994 {
675 ++dec->cur; 995 ++dec->cur;
676 break; 996 break;
677 } 997 }
678 998
679 if (*dec->cur != ',') 999 if (*dec->cur != ',')
680 ERR (", or } expected while parsing object/hash"); 1000 ERR (", or } expected while parsing object/hash");
681 1001
682 ++dec->cur; 1002 ++dec->cur;
683 } 1003 }
684 1004
1005 DEC_DEC_DEPTH;
685 return newRV_noinc ((SV *)hv); 1006 return newRV_noinc ((SV *)hv);
686 1007
687fail: 1008fail:
688 SvREFCNT_dec (hv); 1009 SvREFCNT_dec (hv);
1010 DEC_DEC_DEPTH;
689 return 0; 1011 return 0;
690} 1012}
691 1013
692static SV * 1014static SV *
693decode_sv (dec_t *dec) 1015decode_sv (dec_t *dec)
694{ 1016{
695 WS; 1017 decode_ws (dec);
1018
1019 // the beauty of JSON: you need exactly one character lookahead
1020 // to parse anything.
696 switch (*dec->cur) 1021 switch (*dec->cur)
697 { 1022 {
698 case '"': ++dec->cur; return decode_str (dec); 1023 case '"': ++dec->cur; return decode_str (dec);
699 case '[': ++dec->cur; return decode_av (dec); 1024 case '[': ++dec->cur; return decode_av (dec);
700 case '{': ++dec->cur; return decode_hv (dec); 1025 case '{': ++dec->cur; return decode_hv (dec);
… …
728 1053
729 case 'n': 1054 case 'n':
730 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4)) 1055 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4))
731 { 1056 {
732 dec->cur += 4; 1057 dec->cur += 4;
733 return newSViv (1); 1058 return newSVsv (&PL_sv_undef);
734 } 1059 }
735 else 1060 else
736 ERR ("'null' expected"); 1061 ERR ("'null' expected");
737 1062
738 break; 1063 break;
739 1064
740 default: 1065 default:
741 ERR ("malformed json string"); 1066 ERR ("malformed JSON string, neither array, object, number, string or atom");
742 break; 1067 break;
743 } 1068 }
744 1069
745fail: 1070fail:
746 return 0; 1071 return 0;
747} 1072}
748 1073
749static SV * 1074static SV *
750decode_json (SV *string, UV flags) 1075decode_json (SV *string, U32 flags, UV *offset_return)
751{ 1076{
1077 dec_t dec;
1078 UV offset;
752 SV *sv; 1079 SV *sv;
753 1080
1081 SvGETMAGIC (string);
1082 SvUPGRADE (string, SVt_PV);
1083
754 if (!(flags & F_UTF8)) 1084 if (flags & F_UTF8)
1085 sv_utf8_downgrade (string, 0);
1086 else
755 sv_utf8_upgrade (string); 1087 sv_utf8_upgrade (string);
756 1088
757 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1089 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
758 1090
759 dec_t dec;
760 dec.flags = flags; 1091 dec.flags = flags;
761 dec.cur = SvPVX (string); 1092 dec.cur = SvPVX (string);
762 dec.end = SvEND (string); 1093 dec.end = SvEND (string);
763 dec.err = 0; 1094 dec.err = 0;
1095 dec.depth = 0;
1096 dec.maxdepth = DEC_DEPTH (dec.flags);
764 1097
765 *dec.end = 1; // invalid anywhere 1098 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
766 sv = decode_sv (&dec); 1099 sv = decode_sv (&dec);
767 *dec.end = 0; 1100
1101 if (!(offset_return || !sv))
1102 {
1103 // check for trailing garbage
1104 decode_ws (&dec);
1105
1106 if (*dec.cur)
1107 {
1108 dec.err = "garbage after JSON object";
1109 SvREFCNT_dec (sv);
1110 sv = 0;
1111 }
1112 }
1113
1114 if (offset_return || !sv)
1115 {
1116 offset = dec.flags & F_UTF8
1117 ? dec.cur - SvPVX (string)
1118 : utf8_distance (dec.cur, SvPVX (string));
1119
1120 if (offset_return)
1121 *offset_return = offset;
1122 }
768 1123
769 if (!sv) 1124 if (!sv)
770 { 1125 {
771 IV offset = utf8_distance (dec.cur, SvPVX (string));
772 SV *uni = sv_newmortal (); 1126 SV *uni = sv_newmortal ();
773 1127
1128 // horrible hack to silence warning inside pv_uni_display
1129 COP cop = *PL_curcop;
1130 cop.cop_warnings = pWARN_NONE;
1131 ENTER;
1132 SAVEVPTR (PL_curcop);
1133 PL_curcop = &cop;
774 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1134 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1135 LEAVE;
1136
775 croak ("%s, at character %d (%s)", 1137 croak ("%s, at character offset %d [\"%s\"]",
776 dec.err, 1138 dec.err,
777 (int)offset, 1139 (int)offset,
778 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1140 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
779 } 1141 }
780 1142
781 sv = sv_2mortal (sv); 1143 sv = sv_2mortal (sv);
782 1144
783 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1145 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv))
784 croak ("JSON object or array expected (but number, string, true, false or null found, use allow_nonref to allow this)"); 1146 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
785 1147
786 return sv; 1148 return sv;
787} 1149}
788 1150
1151/////////////////////////////////////////////////////////////////////////////
1152// XS interface functions
1153
789MODULE = JSON::XS PACKAGE = JSON::XS 1154MODULE = JSON::XS PACKAGE = JSON::XS
790 1155
791BOOT: 1156BOOT:
792{ 1157{
793 int i; 1158 int i;
794 1159
795 memset (decode_hexdigit, 0xff, 256);
796 for (i = 10; i--; ) 1160 for (i = 0; i < 256; ++i)
797 decode_hexdigit ['0' + i] = i; 1161 decode_hexdigit [i] =
798 1162 i >= '0' && i <= '9' ? i - '0'
799 for (i = 6; --i; ) 1163 : i >= 'a' && i <= 'f' ? i - 'a' + 10
800 { 1164 : i >= 'A' && i <= 'F' ? i - 'A' + 10
801 decode_hexdigit ['a' + i] = 10 + i; 1165 : -1;
802 decode_hexdigit ['A' + i] = 10 + i;
803 }
804 1166
805 json_stash = gv_stashpv ("JSON::XS", 1); 1167 json_stash = gv_stashpv ("JSON::XS", 1);
806} 1168}
1169
1170PROTOTYPES: DISABLE
807 1171
808SV *new (char *dummy) 1172SV *new (char *dummy)
809 CODE: 1173 CODE:
810 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1174 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash);
811 OUTPUT: 1175 OUTPUT:
812 RETVAL 1176 RETVAL
813 1177
814SV *ascii (SV *self, int enable) 1178SV *ascii (SV *self, int enable = 1)
815 ALIAS: 1179 ALIAS:
816 ascii = F_ASCII 1180 ascii = F_ASCII
1181 latin1 = F_LATIN1
817 utf8 = F_UTF8 1182 utf8 = F_UTF8
818 indent = F_INDENT 1183 indent = F_INDENT
819 canonical = F_CANONICAL 1184 canonical = F_CANONICAL
820 space_before = F_SPACE_BEFORE 1185 space_before = F_SPACE_BEFORE
821 space_after = F_SPACE_AFTER 1186 space_after = F_SPACE_AFTER
822 json_rpc = F_JSON_RPC
823 pretty = F_PRETTY 1187 pretty = F_PRETTY
824 allow_nonref = F_ALLOW_NONREF 1188 allow_nonref = F_ALLOW_NONREF
1189 shrink = F_SHRINK
825 CODE: 1190 CODE:
826{ 1191{
827 UV *uv = SvJSON (self); 1192 UV *uv = SvJSON (self);
828 if (enable) 1193 if (enable)
829 *uv |= ix; 1194 *uv |= ix;
… …
833 RETVAL = newSVsv (self); 1198 RETVAL = newSVsv (self);
834} 1199}
835 OUTPUT: 1200 OUTPUT:
836 RETVAL 1201 RETVAL
837 1202
1203SV *max_depth (SV *self, UV max_depth = 0x80000000UL)
1204 CODE:
1205{
1206 UV *uv = SvJSON (self);
1207 UV log2 = 0;
1208
1209 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1210
1211 while ((1UL << log2) < max_depth)
1212 ++log2;
1213
1214 *uv = *uv & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1215
1216 RETVAL = newSVsv (self);
1217}
1218 OUTPUT:
1219 RETVAL
1220
838void encode (SV *self, SV *scalar) 1221void encode (SV *self, SV *scalar)
839 PPCODE: 1222 PPCODE:
840 XPUSHs (encode_json (scalar, *SvJSON (self))); 1223 XPUSHs (encode_json (scalar, *SvJSON (self)));
841 1224
842void decode (SV *self, SV *jsonstr) 1225void decode (SV *self, SV *jsonstr)
843 PPCODE: 1226 PPCODE:
844 XPUSHs (decode_json (jsonstr, *SvJSON (self))); 1227 XPUSHs (decode_json (jsonstr, *SvJSON (self), 0));
1228
1229void decode_prefix (SV *self, SV *jsonstr)
1230 PPCODE:
1231{
1232 UV offset;
1233 EXTEND (SP, 2);
1234 PUSHs (decode_json (jsonstr, *SvJSON (self), &offset));
1235 PUSHs (sv_2mortal (newSVuv (offset)));
1236}
1237
1238PROTOTYPES: ENABLE
845 1239
846void to_json (SV *scalar) 1240void to_json (SV *scalar)
1241 ALIAS:
1242 objToJson = 0
847 PPCODE: 1243 PPCODE:
848 XPUSHs (encode_json (scalar, F_UTF8)); 1244 XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8));
849 1245
850void from_json (SV *jsonstr) 1246void from_json (SV *jsonstr)
1247 ALIAS:
1248 jsonToObj = 0
851 PPCODE: 1249 PPCODE:
852 XPUSHs (decode_json (jsonstr, F_UTF8)); 1250 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8, 0));
853 1251

Diff Legend

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