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 1.4 by root, Thu Mar 22 21:13:58 2007 UTC vs.
Revision 1.40 by root, Tue Jun 12 01:27:02 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
9#define F_ASCII 0x00000001 14#define F_ASCII 0x00000001UL
15#define F_LATIN1 0x00000002UL
10#define F_UTF8 0x00000002 16#define F_UTF8 0x00000004UL
11#define F_INDENT 0x00000004 17#define F_INDENT 0x00000008UL
12#define F_CANONICAL 0x00000008 18#define F_CANONICAL 0x00000010UL
13#define F_SPACE_BEFORE 0x00000010 19#define F_SPACE_BEFORE 0x00000020UL
14#define F_SPACE_AFTER 0x00000020 20#define F_SPACE_AFTER 0x00000040UL
15#define F_JSON_RPC 0x00000040
16#define F_ALLOW_NONREF 0x00000080 21#define F_ALLOW_NONREF 0x00000100UL
22#define F_SHRINK 0x00000200UL
23#define F_MAXDEPTH 0xf8000000UL
24#define S_MAXDEPTH 27
25
26#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
27
28// F_SELFCONVERT? <=> to_json/toJson
29// F_BLESSED? <=> { $__class__$ => }
17 30
18#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 31#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
19#define F_DEFAULT 0 32#define F_DEFAULT (9UL << S_MAXDEPTH)
20 33
21#define INIT_SIZE 32 // initial scalar size to be allocated 34#define INIT_SIZE 32 // initial scalar size to be allocated
35#define INDENT_STEP 3 // spaces per indentation level
36
37#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
22 38
23#define SB do { 39#define SB do {
24#define SE } while (0) 40#define SE } while (0)
25 41
42#if __GNUC__ >= 3
43# define expect(expr,value) __builtin_expect ((expr),(value))
44# define inline inline
45#else
46# define expect(expr,value) (expr)
47# define inline static
48#endif
49
50#define expect_false(expr) expect ((expr) != 0, 0)
51#define expect_true(expr) expect ((expr) != 0, 1)
52
26static HV *json_stash; 53static HV *json_stash; // JSON::XS::
54
55/////////////////////////////////////////////////////////////////////////////
56// utility functions
57
58static UV *
59SvJSON (SV *sv)
60{
61 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash))
62 croak ("object is not of type JSON::XS");
63
64 return &SvUVX (SvRV (sv));
65}
66
67static void
68shrink (SV *sv)
69{
70 sv_utf8_downgrade (sv, 1);
71 if (SvLEN (sv) > SvCUR (sv) + 1)
72 {
73#ifdef SvPV_shrink_to_cur
74 SvPV_shrink_to_cur (sv);
75#elif defined (SvPV_renew)
76 SvPV_renew (sv, SvCUR (sv) + 1);
77#endif
78 }
79}
80
81// decode an utf-8 character and return it, or (UV)-1 in
82// case of an error.
83// we special-case "safe" characters from U+80 .. U+7FF,
84// but use the very good perl function to parse anything else.
85// note that we never call this function for a ascii codepoints
86inline UV
87decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
88{
89 if (expect_false (s[0] > 0xdf || s[0] < 0xc2))
90 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
91 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)
92 {
93 *clen = 2;
94 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
95 }
96 else
97 {
98 *clen = (STRLEN)-1;
99 return (UV)-1;
100 }
101}
102
103/////////////////////////////////////////////////////////////////////////////
104// encoder
27 105
28// structure used for encoding JSON 106// structure used for encoding JSON
29typedef struct 107typedef struct
30{ 108{
31 char *cur; 109 char *cur; // SvPVX (sv) + current output position
32 STRLEN len; // SvLEN (sv)
33 char *end; // SvEND (sv) 110 char *end; // SvEND (sv)
34 SV *sv; 111 SV *sv; // result scalar
35 UV flags; 112 U32 flags; // F_*
36 int max_recurse; 113 U32 indent; // indentation level
37 int indent; 114 U32 maxdepth; // max. indentation/recursion level
38} enc_t; 115} enc_t;
39 116
40// structure used for decoding JSON 117inline void
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
60static void
61need (enc_t *enc, STRLEN len) 118need (enc_t *enc, STRLEN len)
62{ 119{
63 if (enc->cur + len >= enc->end) 120 if (expect_false (enc->cur + len >= enc->end))
64 { 121 {
65 STRLEN cur = enc->cur - SvPVX (enc->sv); 122 STRLEN cur = enc->cur - SvPVX (enc->sv);
66 SvGROW (enc->sv, cur + len + 1); 123 SvGROW (enc->sv, cur + len + 1);
67 enc->cur = SvPVX (enc->sv) + cur; 124 enc->cur = SvPVX (enc->sv) + cur;
68 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv); 125 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
69 } 126 }
70} 127}
71 128
72static void 129inline void
73encode_ch (enc_t *enc, char ch) 130encode_ch (enc_t *enc, char ch)
74{ 131{
75 need (enc, 1); 132 need (enc, 1);
76 *enc->cur++ = ch; 133 *enc->cur++ = ch;
77} 134}
… …
85 142
86 while (str < end) 143 while (str < end)
87 { 144 {
88 unsigned char ch = *(unsigned char *)str; 145 unsigned char ch = *(unsigned char *)str;
89 146
90 if (ch == '"') 147 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
91 { 148 {
149 if (expect_false (ch == '"')) // but with slow exceptions
150 {
92 need (enc, len += 1); 151 need (enc, len += 1);
93 *enc->cur++ = '\\'; 152 *enc->cur++ = '\\';
94 *enc->cur++ = '"'; 153 *enc->cur++ = '"';
95 ++str;
96 } 154 }
97 else if (ch == '\\') 155 else if (expect_false (ch == '\\'))
98 { 156 {
99 need (enc, len += 1); 157 need (enc, len += 1);
100 *enc->cur++ = '\\'; 158 *enc->cur++ = '\\';
101 *enc->cur++ = '\\'; 159 *enc->cur++ = '\\';
102 ++str;
103 } 160 }
104 else if (ch >= 0x20 && ch < 0x80) // most common case 161 else
105 {
106 *enc->cur++ = ch; 162 *enc->cur++ = ch;
107 ++str; 163
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; 164 ++str;
122 } 165 }
123 else 166 else
124 { 167 {
125 STRLEN clen; 168 switch (ch)
126 UV uch;
127
128 if (is_utf8)
129 { 169 {
130 uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY); 170 case '\010': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'b'; ++str; break;
131 if (clen < 0) 171 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"); 172 case '\012': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'n'; ++str; break;
133 } 173 case '\014': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'f'; ++str; break;
134 else 174 case '\015': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'r'; ++str; break;
135 {
136 uch = ch;
137 clen = 1;
138 }
139 175
140 if (uch < 0x80 || enc->flags & F_ASCII) 176 default:
141 {
142 if (uch > 0xFFFFUL)
143 { 177 {
178 STRLEN clen;
179 UV uch;
180
181 if (is_utf8)
182 {
183 uch = decode_utf8 (str, end - str, &clen);
184 if (clen == (STRLEN)-1)
185 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
186 }
187 else
188 {
189 uch = ch;
190 clen = 1;
191 }
192
193 if (uch > 0x10FFFFUL)
194 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
195
196 if (uch < 0x80 || enc->flags & F_ASCII || (enc->flags & F_LATIN1 && uch > 0xFF))
197 {
198 if (uch > 0xFFFFUL)
199 {
144 need (enc, len += 11); 200 need (enc, len += 11);
145 sprintf (enc->cur, "\\u%04x\\u%04x", 201 sprintf (enc->cur, "\\u%04x\\u%04x",
146 (uch - 0x10000) / 0x400 + 0xD800, 202 (int)((uch - 0x10000) / 0x400 + 0xD800),
147 (uch - 0x10000) % 0x400 + 0xDC00); 203 (int)((uch - 0x10000) % 0x400 + 0xDC00));
148 enc->cur += 12; 204 enc->cur += 12;
205 }
206 else
207 {
208 static char hexdigit [16] = "0123456789abcdef";
209 need (enc, len += 5);
210 *enc->cur++ = '\\';
211 *enc->cur++ = 'u';
212 *enc->cur++ = hexdigit [ uch >> 12 ];
213 *enc->cur++ = hexdigit [(uch >> 8) & 15];
214 *enc->cur++ = hexdigit [(uch >> 4) & 15];
215 *enc->cur++ = hexdigit [(uch >> 0) & 15];
216 }
217
218 str += clen;
219 }
220 else if (enc->flags & F_LATIN1)
221 {
222 *enc->cur++ = uch;
223 str += clen;
224 }
225 else if (is_utf8)
226 {
227 need (enc, len += clen);
228 do
229 {
230 *enc->cur++ = *str++;
231 }
232 while (--clen);
233 }
234 else
235 {
236 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
237 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
238 ++str;
239 }
149 } 240 }
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 while (clen--)
168 *enc->cur++ = *str++;
169 }
170 else
171 {
172 need (enc, 10); // never more than 11 bytes needed
173 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
174 ++str;
175 } 241 }
176 } 242 }
177 243
178 --len; 244 --len;
179 } 245 }
180} 246}
181 247
182#define INDENT SB \ 248inline void
249encode_indent (enc_t *enc)
250{
183 if (enc->flags & F_INDENT) \ 251 if (enc->flags & F_INDENT)
184 { \ 252 {
185 int i_; \ 253 int spaces = enc->indent * INDENT_STEP;
186 need (enc, enc->indent); \
187 for (i_ = enc->indent * 3; i_--; )\
188 encode_ch (enc, ' '); \
189 } \
190 SE
191 254
192#define SPACE SB need (enc, 1); encode_ch (enc, ' '); SE 255 need (enc, spaces);
193#define NL SB if (enc->flags & F_INDENT) { need (enc, 1); encode_ch (enc, '\n'); } SE 256 memset (enc->cur, ' ', spaces);
194#define COMMA SB \ 257 enc->cur += spaces;
258 }
259}
260
261inline void
262encode_space (enc_t *enc)
263{
264 need (enc, 1);
195 encode_ch (enc, ','); \ 265 encode_ch (enc, ' ');
266}
267
268inline void
269encode_nl (enc_t *enc)
270{
196 if (enc->flags & F_INDENT) \ 271 if (enc->flags & F_INDENT)
197 NL; \ 272 {
273 need (enc, 1);
274 encode_ch (enc, '\n');
275 }
276}
277
278inline void
279encode_comma (enc_t *enc)
280{
281 encode_ch (enc, ',');
282
283 if (enc->flags & F_INDENT)
284 encode_nl (enc);
198 else if (enc->flags & F_SPACE_AFTER) \ 285 else if (enc->flags & F_SPACE_AFTER)
199 SPACE; \ 286 encode_space (enc);
200 SE 287}
201 288
202static void encode_sv (enc_t *enc, SV *sv); 289static void encode_sv (enc_t *enc, SV *sv);
203 290
204static void 291static void
205encode_av (enc_t *enc, AV *av) 292encode_av (enc_t *enc, AV *av)
206{ 293{
207 int i, len = av_len (av); 294 int i, len = av_len (av);
208 295
209 encode_ch (enc, '['); NL; 296 if (enc->indent >= enc->maxdepth)
297 croak ("data structure too deep (hit recursion limit)");
298
299 encode_ch (enc, '['); encode_nl (enc);
210 ++enc->indent; 300 ++enc->indent;
211 301
212 for (i = 0; i <= len; ++i) 302 for (i = 0; i <= len; ++i)
213 { 303 {
214 INDENT; 304 encode_indent (enc);
215 encode_sv (enc, *av_fetch (av, i, 0)); 305 encode_sv (enc, *av_fetch (av, i, 0));
216 306
217 if (i < len) 307 if (i < len)
218 COMMA; 308 encode_comma (enc);
219 } 309 }
220 310
221 NL; 311 encode_nl (enc);
222 312
223 --enc->indent; 313 --enc->indent;
224 INDENT; encode_ch (enc, ']'); 314 encode_indent (enc); encode_ch (enc, ']');
225} 315}
226 316
227static void 317static void
228encode_he (enc_t *enc, HE *he) 318encode_he (enc_t *enc, HE *he)
229{ 319{
… …
243 else 333 else
244 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 334 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
245 335
246 encode_ch (enc, '"'); 336 encode_ch (enc, '"');
247 337
248 if (enc->flags & F_SPACE_BEFORE) SPACE; 338 if (enc->flags & F_SPACE_BEFORE) encode_space (enc);
249 encode_ch (enc, ':'); 339 encode_ch (enc, ':');
250 if (enc->flags & F_SPACE_AFTER ) SPACE; 340 if (enc->flags & F_SPACE_AFTER ) encode_space (enc);
251 encode_sv (enc, HeVAL (he)); 341 encode_sv (enc, HeVAL (he));
252} 342}
253 343
254// compare hash entries, used when all keys are bytestrings 344// compare hash entries, used when all keys are bytestrings
255static int 345static int
… …
261 HE *b = *(HE **)b_; 351 HE *b = *(HE **)b_;
262 352
263 STRLEN la = HeKLEN (a); 353 STRLEN la = HeKLEN (a);
264 STRLEN lb = HeKLEN (b); 354 STRLEN lb = HeKLEN (b);
265 355
266 if (!(cmp == memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 356 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb)))
267 cmp = la < lb ? -1 : la == lb ? 0 : 1; 357 cmp = la - lb;
268 358
269 return cmp; 359 return cmp;
270} 360}
271 361
272// compare hash entries, used when some keys are sv's or utf-x 362// compare hash entries, used when some keys are sv's or utf-x
… …
279static void 369static void
280encode_hv (enc_t *enc, HV *hv) 370encode_hv (enc_t *enc, HV *hv)
281{ 371{
282 int count, i; 372 int count, i;
283 373
374 if (enc->indent >= enc->maxdepth)
375 croak ("data structure too deep (hit recursion limit)");
376
284 encode_ch (enc, '{'); NL; ++enc->indent; 377 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
285 378
286 if ((count = hv_iterinit (hv))) 379 if ((count = hv_iterinit (hv)))
287 { 380 {
288 // for canonical output we have to sort by keys first 381 // for canonical output we have to sort by keys first
289 // actually, this is mostly due to the stupid so-called 382 // actually, this is mostly due to the stupid so-called
290 // security workaround added somewhere in 5.8.x. 383 // security workaround added somewhere in 5.8.x.
291 // that randomises hash orderings 384 // that randomises hash orderings
292 if (enc->flags & F_CANONICAL) 385 if (enc->flags & F_CANONICAL)
293 { 386 {
294 HE *he, *hes [count];
295 int fast = 1; 387 int fast = 1;
388 HE *he;
389#if defined(__BORLANDC__) || defined(_MSC_VER)
390 HE **hes = _alloca (count * sizeof (HE));
391#else
392 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
393#endif
296 394
297 i = 0; 395 i = 0;
298 while ((he = hv_iternext (hv))) 396 while ((he = hv_iternext (hv)))
299 { 397 {
300 hes [i++] = he; 398 hes [i++] = he;
… …
306 404
307 if (fast) 405 if (fast)
308 qsort (hes, count, sizeof (HE *), he_cmp_fast); 406 qsort (hes, count, sizeof (HE *), he_cmp_fast);
309 else 407 else
310 { 408 {
311 // hack to disable "use bytes" 409 // hack to forcefully disable "use bytes"
312 COP *oldcop = PL_curcop, cop; 410 COP cop = *PL_curcop;
313 cop.op_private = 0; 411 cop.op_private = 0;
412
413 ENTER;
414 SAVETMPS;
415
416 SAVEVPTR (PL_curcop);
314 PL_curcop = &cop; 417 PL_curcop = &cop;
315 418
316 SAVETMPS;
317 qsort (hes, count, sizeof (HE *), he_cmp_slow); 419 qsort (hes, count, sizeof (HE *), he_cmp_slow);
420
318 FREETMPS; 421 FREETMPS;
319 422 LEAVE;
320 PL_curcop = oldcop;
321 } 423 }
322 424
323 for (i = 0; i < count; ++i) 425 for (i = 0; i < count; ++i)
324 { 426 {
325 INDENT; 427 encode_indent (enc);
326 encode_he (enc, hes [i]); 428 encode_he (enc, hes [i]);
327 429
328 if (i < count - 1) 430 if (i < count - 1)
329 COMMA; 431 encode_comma (enc);
330 } 432 }
331 433
332 NL; 434 encode_nl (enc);
333 } 435 }
334 else 436 else
335 { 437 {
336 SV *sv;
337 HE *he = hv_iternext (hv); 438 HE *he = hv_iternext (hv);
338 439
339 for (;;) 440 for (;;)
340 { 441 {
341 INDENT; 442 encode_indent (enc);
342 encode_he (enc, he); 443 encode_he (enc, he);
343 444
344 if (!(he = hv_iternext (hv))) 445 if (!(he = hv_iternext (hv)))
345 break; 446 break;
346 447
347 COMMA; 448 encode_comma (enc);
348 } 449 }
349 450
350 NL; 451 encode_nl (enc);
351 } 452 }
352 } 453 }
353 454
354 --enc->indent; INDENT; encode_ch (enc, '}'); 455 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
456}
457
458// encode objects, arrays and special \0=false and \1=true values.
459static void
460encode_rv (enc_t *enc, SV *sv)
461{
462 svtype svt;
463
464 SvGETMAGIC (sv);
465 svt = SvTYPE (sv);
466
467 if (svt == SVt_PVHV)
468 encode_hv (enc, (HV *)sv);
469 else if (svt == SVt_PVAV)
470 encode_av (enc, (AV *)sv);
471 else if (svt < SVt_PVAV)
472 {
473 if (SvNIOK (sv) && SvIV (sv) == 0)
474 encode_str (enc, "false", 5, 0);
475 else if (SvNIOK (sv) && SvIV (sv) == 1)
476 encode_str (enc, "true", 4, 0);
477 else
478 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
479 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
480 }
481 else
482 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
483 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
355} 484}
356 485
357static void 486static void
358encode_sv (enc_t *enc, SV *sv) 487encode_sv (enc_t *enc, SV *sv)
359{ 488{
… …
367 encode_str (enc, str, len, SvUTF8 (sv)); 496 encode_str (enc, str, len, SvUTF8 (sv));
368 encode_ch (enc, '"'); 497 encode_ch (enc, '"');
369 } 498 }
370 else if (SvNOKp (sv)) 499 else if (SvNOKp (sv))
371 { 500 {
501 // trust that perl will do the right thing w.r.t. JSON syntax.
372 need (enc, NV_DIG + 32); 502 need (enc, NV_DIG + 32);
373 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 503 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
374 enc->cur += strlen (enc->cur); 504 enc->cur += strlen (enc->cur);
375 } 505 }
376 else if (SvIOKp (sv)) 506 else if (SvIOKp (sv))
377 { 507 {
378 need (enc, 64); 508 // we assume we can always read an IV as a UV
509 if (SvUV (sv) & ~(UV)0x7fff)
510 {
511 // large integer, use the (rather slow) snprintf way.
512 need (enc, sizeof (UV) * 3);
379 enc->cur += 513 enc->cur +=
380 SvIsUV(sv) 514 SvIsUV(sv)
381 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 515 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
382 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 516 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
517 }
518 else
519 {
520 // optimise the "small number case"
521 // code will likely be branchless and use only a single multiplication
522 I32 i = SvIV (sv);
523 U32 u;
524 char digit, nz = 0;
525
526 need (enc, 6);
527
528 *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
529 u = i < 0 ? -i : i;
530
531 // convert to 4.28 fixed-point representation
532 u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
533
534 // now output digit by digit, each time masking out the integer part
535 // and multiplying by 5 while moving the decimal point one to the right,
536 // resulting in a net multiplication by 10.
537 // we always write the digit to memory but conditionally increment
538 // the pointer, to ease the usage of conditional move instructions.
539 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5;
540 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5;
541 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5;
542 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5;
543 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
544 }
383 } 545 }
384 else if (SvROK (sv)) 546 else if (SvROK (sv))
385 { 547 encode_rv (enc, SvRV (sv));
386 if (!--enc->max_recurse)
387 croak ("data structure too deep (hit recursion limit)");
388
389 sv = SvRV (sv);
390
391 switch (SvTYPE (sv))
392 {
393 case SVt_PVAV: encode_av (enc, (AV *)sv); break;
394 case SVt_PVHV: encode_hv (enc, (HV *)sv); break;
395
396 default:
397 croak ("JSON can only represent references to arrays or hashes");
398 }
399 }
400 else if (!SvOK (sv)) 548 else if (!SvOK (sv))
401 encode_str (enc, "null", 4, 0); 549 encode_str (enc, "null", 4, 0);
402 else 550 else
403 croak ("encountered perl type that JSON cannot handle"); 551 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
552 SvPV_nolen (sv), SvFLAGS (sv));
404} 553}
405 554
406static SV * 555static SV *
407encode_json (SV *scalar, UV flags) 556encode_json (SV *scalar, U32 flags)
408{ 557{
558 enc_t enc;
559
409 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 560 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar))
410 croak ("hash- or arraref required (not a simple scalar, use allow_nonref to allow this)"); 561 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
411 562
412 enc_t enc;
413 enc.flags = flags; 563 enc.flags = flags;
414 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 564 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
415 enc.cur = SvPVX (enc.sv); 565 enc.cur = SvPVX (enc.sv);
416 enc.end = SvEND (enc.sv); 566 enc.end = SvEND (enc.sv);
417 enc.max_recurse = 0;
418 enc.indent = 0; 567 enc.indent = 0;
568 enc.maxdepth = DEC_DEPTH (flags);
419 569
420 SvPOK_only (enc.sv); 570 SvPOK_only (enc.sv);
421 encode_sv (&enc, scalar); 571 encode_sv (&enc, scalar);
422 572
573 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
574 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
575
423 if (!(flags & (F_ASCII | F_UTF8))) 576 if (!(flags & (F_ASCII | F_LATIN1 | F_UTF8)))
424 SvUTF8_on (enc.sv); 577 SvUTF8_on (enc.sv);
425 578
426 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 579 if (enc.flags & F_SHRINK)
580 shrink (enc.sv);
581
427 return enc.sv; 582 return enc.sv;
428} 583}
429 584
430///////////////////////////////////////////////////////////////////////////// 585/////////////////////////////////////////////////////////////////////////////
586// decoder
431 587
432#define WS \ 588// structure used for decoding JSON
589typedef struct
590{
591 char *cur; // current parser pointer
592 char *end; // end of input string
593 const char *err; // parse error, if != 0
594 U32 flags; // F_*
595 U32 depth; // recursion depth
596 U32 maxdepth; // recursion depth limit
597} dec_t;
598
599inline void
600decode_ws (dec_t *dec)
601{
433 for (;;) \ 602 for (;;)
434 { \ 603 {
435 char ch = *dec->cur; \ 604 char ch = *dec->cur;
605
436 if (ch > 0x20 \ 606 if (ch > 0x20
437 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) \ 607 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09))
438 break; \ 608 break;
609
439 ++dec->cur; \ 610 ++dec->cur;
440 } 611 }
612}
441 613
442#define ERR(reason) SB dec->err = reason; goto fail; SE 614#define ERR(reason) SB dec->err = reason; goto fail; SE
615
443#define EXPECT_CH(ch) SB \ 616#define EXPECT_CH(ch) SB \
444 if (*dec->cur != ch) \ 617 if (*dec->cur != ch) \
445 ERR (# ch " expected"); \ 618 ERR (# ch " expected"); \
446 ++dec->cur; \ 619 ++dec->cur; \
447 SE 620 SE
448 621
622#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)")
623#define DEC_DEC_DEPTH --dec->depth
624
449static SV *decode_sv (dec_t *dec); 625static SV *decode_sv (dec_t *dec);
450 626
451static signed char decode_hexdigit[256]; 627static signed char decode_hexdigit[256];
452 628
453static UV 629static UV
454decode_4hex (dec_t *dec) 630decode_4hex (dec_t *dec)
455{ 631{
456 signed char d1, d2, d3, d4; 632 signed char d1, d2, d3, d4;
633 unsigned char *cur = (unsigned char *)dec->cur;
457 634
458 d1 = decode_hexdigit [((unsigned char *)dec->cur) [0]]; 635 d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected");
459 if (d1 < 0) ERR ("four hexadecimal digits expected"); 636 d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected");
460 d2 = decode_hexdigit [((unsigned char *)dec->cur) [1]]; 637 d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected");
461 if (d2 < 0) ERR ("four hexadecimal digits expected"); 638 d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected");
462 d3 = decode_hexdigit [((unsigned char *)dec->cur) [2]];
463 if (d3 < 0) ERR ("four hexadecimal digits expected");
464 d4 = decode_hexdigit [((unsigned char *)dec->cur) [3]];
465 if (d4 < 0) ERR ("four hexadecimal digits expected");
466 639
467 dec->cur += 4; 640 dec->cur += 4;
468 641
469 return ((UV)d1) << 12 642 return ((UV)d1) << 12
470 | ((UV)d2) << 8 643 | ((UV)d2) << 8
… …
473 646
474fail: 647fail:
475 return (UV)-1; 648 return (UV)-1;
476} 649}
477 650
478#define APPEND_GROW(n) SB \
479 if (cur + (n) >= end) \
480 { \
481 STRLEN ofs = cur - SvPVX (sv); \
482 SvGROW (sv, ofs + (n) + 1); \
483 cur = SvPVX (sv) + ofs; \
484 end = SvEND (sv); \
485 } \
486 SE
487
488#define APPEND_CH(ch) SB \
489 APPEND_GROW (1); \
490 *cur++ = (ch); \
491 SE
492
493static SV * 651static SV *
494decode_str (dec_t *dec) 652decode_str (dec_t *dec)
495{ 653{
496 SV *sv = NEWSV (0,2); 654 SV *sv = 0;
497 int utf8 = 0; 655 int utf8 = 0;
498 char *cur = SvPVX (sv); 656 char *dec_cur = dec->cur;
499 char *end = SvEND (sv);
500 657
501 for (;;) 658 do
502 { 659 {
503 unsigned char ch = *(unsigned char *)dec->cur; 660 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
661 char *cur = buf;
504 662
505 if (ch == '"') 663 do
506 break;
507 else if (ch == '\\')
508 { 664 {
509 switch (*++dec->cur) 665 unsigned char ch = *(unsigned char *)dec_cur++;
666
667 if (expect_false (ch == '"'))
510 { 668 {
511 case '\\': 669 --dec_cur;
512 case '/': 670 break;
513 case '"': APPEND_CH (*dec->cur++); break; 671 }
514 672 else if (expect_false (ch == '\\'))
515 case 'b': APPEND_CH ('\010'); ++dec->cur; break; 673 {
516 case 't': APPEND_CH ('\011'); ++dec->cur; break; 674 switch (*dec_cur)
517 case 'n': APPEND_CH ('\012'); ++dec->cur; break;
518 case 'f': APPEND_CH ('\014'); ++dec->cur; break;
519 case 'r': APPEND_CH ('\015'); ++dec->cur; break;
520
521 case 'u':
522 { 675 {
523 UV lo, hi; 676 case '\\':
524 ++dec->cur; 677 case '/':
678 case '"': *cur++ = *dec_cur++; break;
525 679
526 hi = decode_4hex (dec); 680 case 'b': ++dec_cur; *cur++ = '\010'; break;
527 if (hi == (UV)-1) 681 case 't': ++dec_cur; *cur++ = '\011'; break;
528 goto fail; 682 case 'n': ++dec_cur; *cur++ = '\012'; break;
683 case 'f': ++dec_cur; *cur++ = '\014'; break;
684 case 'r': ++dec_cur; *cur++ = '\015'; break;
529 685
530 // possibly a surrogate pair 686 case 'u':
531 if (hi >= 0xd800 && hi < 0xdc00)
532 { 687 {
533 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 688 UV lo, hi;
534 ERR ("illegal surrogate character"); 689 ++dec_cur;
535 690
536 dec->cur += 2; 691 dec->cur = dec_cur;
537
538 lo = decode_4hex (dec); 692 hi = decode_4hex (dec);
693 dec_cur = dec->cur;
539 if (lo == (UV)-1) 694 if (hi == (UV)-1)
540 goto fail; 695 goto fail;
541 696
697 // possibly a surrogate pair
698 if (hi >= 0xd800)
699 if (hi < 0xdc00)
700 {
701 if (dec_cur [0] != '\\' || dec_cur [1] != 'u')
702 ERR ("missing low surrogate character in surrogate pair");
703
704 dec_cur += 2;
705
706 dec->cur = dec_cur;
707 lo = decode_4hex (dec);
708 dec_cur = dec->cur;
709 if (lo == (UV)-1)
710 goto fail;
711
542 if (lo < 0xdc00 || lo >= 0xe000) 712 if (lo < 0xdc00 || lo >= 0xe000)
543 ERR ("surrogate pair expected"); 713 ERR ("surrogate pair expected");
544 714
545 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000; 715 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000;
716 }
717 else if (hi < 0xe000)
718 ERR ("missing high surrogate character in surrogate pair");
719
720 if (hi >= 0x80)
721 {
722 utf8 = 1;
723
724 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
725 }
726 else
727 *cur++ = hi;
546 } 728 }
547 else if (lo >= 0xdc00 && lo < 0xe000)
548 ERR ("illegal surrogate character");
549
550 if (hi >= 0x80)
551 { 729 break;
552 utf8 = 1;
553 730
554 APPEND_GROW (4); // at most 4 bytes for 21 bits
555 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
556 }
557 else 731 default:
558 APPEND_CH (hi); 732 --dec_cur;
733 ERR ("illegal backslash escape sequence in string");
559 } 734 }
735 }
736 else if (expect_true (ch >= 0x20 && ch <= 0x7f))
737 *cur++ = ch;
738 else if (ch >= 0x80)
739 {
740 STRLEN clen;
741 UV uch;
742
743 --dec_cur;
744
745 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
746 if (clen == (STRLEN)-1)
747 ERR ("malformed UTF-8 character in JSON string");
748
749 do
750 *cur++ = *dec_cur++;
751 while (--clen);
752
753 utf8 = 1;
754 }
755 else
756 {
757 --dec_cur;
758
759 if (!ch)
760 ERR ("unexpected end of string while parsing JSON string");
560 break; 761 else
762 ERR ("invalid character encountered while parsing JSON string");
561 } 763 }
562 } 764 }
563 else if (ch >= 0x20 && ch <= 0x7f) 765 while (cur < buf + SHORT_STRING_LEN);
564 APPEND_CH (*dec->cur++); 766
565 else if (ch >= 0x80) 767 {
768 STRLEN len = cur - buf;
769
770 if (sv)
566 { 771 {
567 STRLEN clen; 772 SvGROW (sv, SvCUR (sv) + len + 1);
568 UV uch = utf8n_to_uvuni (dec->cur, dec->end - dec->cur, &clen, UTF8_CHECK_ONLY); 773 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
569 if (clen < 0) 774 SvCUR_set (sv, SvCUR (sv) + len);
570 ERR ("malformed UTF-8 character in string, cannot convert to JSON");
571
572 APPEND_GROW (clen);
573 memcpy (cur, dec->cur, clen);
574 cur += clen;
575 dec->cur += clen;
576 } 775 }
577 else 776 else
578 ERR ("invalid character encountered"); 777 sv = newSVpvn (buf, len);
579 } 778 }
779 }
780 while (*dec_cur != '"');
580 781
581 ++dec->cur; 782 ++dec_cur;
582 783
583 SvCUR_set (sv, cur - SvPVX (sv)); 784 if (sv)
584 785 {
585 SvPOK_only (sv); 786 SvPOK_only (sv);
586 *SvEND (sv) = 0; 787 *SvEND (sv) = 0;
587 788
588 if (utf8) 789 if (utf8)
589 SvUTF8_on (sv); 790 SvUTF8_on (sv);
791 }
792 else
793 sv = newSVpvn ("", 0);
590 794
795 dec->cur = dec_cur;
591 return sv; 796 return sv;
592 797
593fail: 798fail:
594 SvREFCNT_dec (sv); 799 dec->cur = dec_cur;
595 return 0; 800 return 0;
596} 801}
597 802
598static SV * 803static SV *
599decode_num (dec_t *dec) 804decode_num (dec_t *dec)
… …
609 { 814 {
610 ++dec->cur; 815 ++dec->cur;
611 if (*dec->cur >= '0' && *dec->cur <= '9') 816 if (*dec->cur >= '0' && *dec->cur <= '9')
612 ERR ("malformed number (leading zero must not be followed by another digit)"); 817 ERR ("malformed number (leading zero must not be followed by another digit)");
613 } 818 }
614 819 else if (*dec->cur < '0' || *dec->cur > '9')
615 // int 820 ERR ("malformed number (no digits after initial minus)");
821 else
822 do
823 {
824 ++dec->cur;
825 }
616 while (*dec->cur >= '0' && *dec->cur <= '9') 826 while (*dec->cur >= '0' && *dec->cur <= '9');
617 ++dec->cur;
618 827
619 // [frac] 828 // [frac]
620 if (*dec->cur == '.') 829 if (*dec->cur == '.')
621 { 830 {
622 is_nv = 1; 831 ++dec->cur;
832
833 if (*dec->cur < '0' || *dec->cur > '9')
834 ERR ("malformed number (no digits after decimal point)");
623 835
624 do 836 do
625 { 837 {
626 ++dec->cur; 838 ++dec->cur;
627 } 839 }
628 while (*dec->cur >= '0' && *dec->cur <= '9'); 840 while (*dec->cur >= '0' && *dec->cur <= '9');
841
842 is_nv = 1;
629 } 843 }
630 844
631 // [exp] 845 // [exp]
632 if (*dec->cur == 'e' || *dec->cur == 'E') 846 if (*dec->cur == 'e' || *dec->cur == 'E')
633 { 847 {
634 is_nv = 1;
635
636 ++dec->cur; 848 ++dec->cur;
849
637 if (*dec->cur == '-' || *dec->cur == '+') 850 if (*dec->cur == '-' || *dec->cur == '+')
638 ++dec->cur; 851 ++dec->cur;
639 852
853 if (*dec->cur < '0' || *dec->cur > '9')
854 ERR ("malformed number (no digits after exp sign)");
855
856 do
857 {
858 ++dec->cur;
859 }
640 while (*dec->cur >= '0' && *dec->cur <= '9') 860 while (*dec->cur >= '0' && *dec->cur <= '9');
641 ++dec->cur; 861
862 is_nv = 1;
642 } 863 }
643 864
644 if (!is_nv) 865 if (!is_nv)
645 { 866 {
646 UV uv; 867 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so
647 int numtype = grok_number (start, dec->cur - start, &uv); 868 if (*start == '-')
648 if (numtype & IS_NUMBER_IN_UV) 869 switch (dec->cur - start)
649 if (numtype & IS_NUMBER_NEG)
650 { 870 {
651 if (uv < (UV)IV_MIN) 871 case 2: return newSViv (-( start [1] - '0' * 1));
652 return newSViv (-(IV)uv); 872 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
873 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
874 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
653 } 875 }
876 else
877 switch (dec->cur - start)
878 {
879 case 1: return newSViv ( start [0] - '0' * 1);
880 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
881 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
882 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
883 }
884
885 {
886 UV uv;
887 int numtype = grok_number (start, dec->cur - start, &uv);
888 if (numtype & IS_NUMBER_IN_UV)
889 if (numtype & IS_NUMBER_NEG)
890 {
891 if (uv < (UV)IV_MIN)
892 return newSViv (-(IV)uv);
893 }
654 else 894 else
655 return newSVuv (uv); 895 return newSVuv (uv);
896
897 // here would likely be the place for bigint support
656 } 898 }
899 }
657 900
901 // if we ever support bigint or bigfloat, this is the place for bigfloat
658 return newSVnv (Atof (start)); 902 return newSVnv (Atof (start));
659 903
660fail: 904fail:
661 return 0; 905 return 0;
662} 906}
… …
664static SV * 908static SV *
665decode_av (dec_t *dec) 909decode_av (dec_t *dec)
666{ 910{
667 AV *av = newAV (); 911 AV *av = newAV ();
668 912
913 DEC_INC_DEPTH;
914 decode_ws (dec);
915
916 if (*dec->cur == ']')
917 ++dec->cur;
918 else
669 for (;;) 919 for (;;)
670 { 920 {
671 SV *value; 921 SV *value;
672 922
673 value = decode_sv (dec); 923 value = decode_sv (dec);
674 if (!value) 924 if (!value)
675 goto fail; 925 goto fail;
676 926
677 av_push (av, value); 927 av_push (av, value);
678 928
679 WS; 929 decode_ws (dec);
680 930
681 if (*dec->cur == ']') 931 if (*dec->cur == ']')
682 { 932 {
683 ++dec->cur; 933 ++dec->cur;
684 break; 934 break;
935 }
685 } 936
686
687 if (*dec->cur != ',') 937 if (*dec->cur != ',')
688 ERR (", or ] expected while parsing array"); 938 ERR (", or ] expected while parsing array");
689 939
690 ++dec->cur; 940 ++dec->cur;
691 } 941 }
692 942
943 DEC_DEC_DEPTH;
693 return newRV_noinc ((SV *)av); 944 return newRV_noinc ((SV *)av);
694 945
695fail: 946fail:
696 SvREFCNT_dec (av); 947 SvREFCNT_dec (av);
948 DEC_DEC_DEPTH;
697 return 0; 949 return 0;
698} 950}
699 951
700static SV * 952static SV *
701decode_hv (dec_t *dec) 953decode_hv (dec_t *dec)
702{ 954{
703 HV *hv = newHV (); 955 HV *hv = newHV ();
704 956
957 DEC_INC_DEPTH;
958 decode_ws (dec);
959
960 if (*dec->cur == '}')
961 ++dec->cur;
962 else
705 for (;;) 963 for (;;)
706 { 964 {
707 SV *key, *value; 965 SV *key, *value;
708 966
709 WS; EXPECT_CH ('"'); 967 decode_ws (dec); EXPECT_CH ('"');
710 968
711 key = decode_str (dec); 969 key = decode_str (dec);
712 if (!key) 970 if (!key)
713 goto fail;
714
715 WS; EXPECT_CH (':');
716
717 value = decode_sv (dec);
718 if (!value)
719 {
720 SvREFCNT_dec (key);
721 goto fail; 971 goto fail;
972
973 decode_ws (dec); EXPECT_CH (':');
974
975 value = decode_sv (dec);
976 if (!value)
977 {
978 SvREFCNT_dec (key);
979 goto fail;
722 } 980 }
723 981
724 //TODO: optimise
725 hv_store_ent (hv, key, value, 0); 982 hv_store_ent (hv, key, value, 0);
983 SvREFCNT_dec (key);
726 984
727 WS; 985 decode_ws (dec);
728 986
729 if (*dec->cur == '}') 987 if (*dec->cur == '}')
730 { 988 {
731 ++dec->cur; 989 ++dec->cur;
732 break; 990 break;
733 } 991 }
734 992
735 if (*dec->cur != ',') 993 if (*dec->cur != ',')
736 ERR (", or } expected while parsing object/hash"); 994 ERR (", or } expected while parsing object/hash");
737 995
738 ++dec->cur; 996 ++dec->cur;
739 } 997 }
740 998
999 DEC_DEC_DEPTH;
741 return newRV_noinc ((SV *)hv); 1000 return newRV_noinc ((SV *)hv);
742 1001
743fail: 1002fail:
744 SvREFCNT_dec (hv); 1003 SvREFCNT_dec (hv);
1004 DEC_DEC_DEPTH;
745 return 0; 1005 return 0;
746} 1006}
747 1007
748static SV * 1008static SV *
749decode_sv (dec_t *dec) 1009decode_sv (dec_t *dec)
750{ 1010{
751 WS; 1011 decode_ws (dec);
1012
1013 // the beauty of JSON: you need exactly one character lookahead
1014 // to parse anything.
752 switch (*dec->cur) 1015 switch (*dec->cur)
753 { 1016 {
754 case '"': ++dec->cur; return decode_str (dec); 1017 case '"': ++dec->cur; return decode_str (dec);
755 case '[': ++dec->cur; return decode_av (dec); 1018 case '[': ++dec->cur; return decode_av (dec);
756 case '{': ++dec->cur; return decode_hv (dec); 1019 case '{': ++dec->cur; return decode_hv (dec);
… …
784 1047
785 case 'n': 1048 case 'n':
786 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4)) 1049 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4))
787 { 1050 {
788 dec->cur += 4; 1051 dec->cur += 4;
789 return newSViv (1); 1052 return newSVsv (&PL_sv_undef);
790 } 1053 }
791 else 1054 else
792 ERR ("'null' expected"); 1055 ERR ("'null' expected");
793 1056
794 break; 1057 break;
795 1058
796 default: 1059 default:
797 ERR ("malformed json string"); 1060 ERR ("malformed JSON string, neither array, object, number, string or atom");
798 break; 1061 break;
799 } 1062 }
800 1063
801fail: 1064fail:
802 return 0; 1065 return 0;
803} 1066}
804 1067
805static SV * 1068static SV *
806decode_json (SV *string, UV flags) 1069decode_json (SV *string, U32 flags, UV *offset_return)
807{ 1070{
1071 dec_t dec;
1072 UV offset;
808 SV *sv; 1073 SV *sv;
809 1074
1075 SvGETMAGIC (string);
1076 SvUPGRADE (string, SVt_PV);
1077
810 if (!(flags & F_UTF8)) 1078 if (flags & F_UTF8)
1079 sv_utf8_downgrade (string, 0);
1080 else
811 sv_utf8_upgrade (string); 1081 sv_utf8_upgrade (string);
812 1082
813 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1083 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
814 1084
815 dec_t dec;
816 dec.flags = flags; 1085 dec.flags = flags;
817 dec.cur = SvPVX (string); 1086 dec.cur = SvPVX (string);
818 dec.end = SvEND (string); 1087 dec.end = SvEND (string);
819 dec.err = 0; 1088 dec.err = 0;
1089 dec.depth = 0;
1090 dec.maxdepth = DEC_DEPTH (dec.flags);
820 1091
821 *dec.end = 1; // invalid anywhere 1092 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
822 sv = decode_sv (&dec); 1093 sv = decode_sv (&dec);
823 *dec.end = 0; 1094
1095 if (!(offset_return || !sv))
1096 {
1097 // check for trailing garbage
1098 decode_ws (&dec);
1099
1100 if (*dec.cur)
1101 {
1102 dec.err = "garbage after JSON object";
1103 SvREFCNT_dec (sv);
1104 sv = 0;
1105 }
1106 }
1107
1108 if (offset_return || !sv)
1109 {
1110 offset = dec.flags & F_UTF8
1111 ? dec.cur - SvPVX (string)
1112 : utf8_distance (dec.cur, SvPVX (string));
1113
1114 if (offset_return)
1115 *offset_return = offset;
1116 }
824 1117
825 if (!sv) 1118 if (!sv)
826 { 1119 {
827 IV offset = utf8_distance (dec.cur, SvPVX (string));
828 SV *uni = sv_newmortal (); 1120 SV *uni = sv_newmortal ();
829 1121
1122 // horrible hack to silence warning inside pv_uni_display
1123 COP cop = *PL_curcop;
1124 cop.cop_warnings = pWARN_NONE;
1125 ENTER;
1126 SAVEVPTR (PL_curcop);
1127 PL_curcop = &cop;
830 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1128 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1129 LEAVE;
1130
831 croak ("%s, at character %d (%s)", 1131 croak ("%s, at character offset %d [\"%s\"]",
832 dec.err, 1132 dec.err,
833 (int)offset, 1133 (int)offset,
834 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1134 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
835 } 1135 }
836 1136
837 sv = sv_2mortal (sv); 1137 sv = sv_2mortal (sv);
838 1138
839 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1139 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv))
840 croak ("JSON object or array expected (but number, string, true, false or null found, use allow_nonref to allow this)"); 1140 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
841 1141
842 return sv; 1142 return sv;
843} 1143}
844 1144
1145/////////////////////////////////////////////////////////////////////////////
1146// XS interface functions
1147
845MODULE = JSON::XS PACKAGE = JSON::XS 1148MODULE = JSON::XS PACKAGE = JSON::XS
846 1149
847BOOT: 1150BOOT:
848{ 1151{
849 int i; 1152 int i;
850 1153
851 memset (decode_hexdigit, 0xff, 256);
852 for (i = 10; i--; ) 1154 for (i = 0; i < 256; ++i)
853 decode_hexdigit ['0' + i] = i; 1155 decode_hexdigit [i] =
854 1156 i >= '0' && i <= '9' ? i - '0'
855 for (i = 7; i--; ) 1157 : i >= 'a' && i <= 'f' ? i - 'a' + 10
856 { 1158 : i >= 'A' && i <= 'F' ? i - 'A' + 10
857 decode_hexdigit ['a' + i] = 10 + i; 1159 : -1;
858 decode_hexdigit ['A' + i] = 10 + i;
859 }
860 1160
861 json_stash = gv_stashpv ("JSON::XS", 1); 1161 json_stash = gv_stashpv ("JSON::XS", 1);
862} 1162}
863 1163
864PROTOTYPES: DISABLE 1164PROTOTYPES: DISABLE
… …
867 CODE: 1167 CODE:
868 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1168 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash);
869 OUTPUT: 1169 OUTPUT:
870 RETVAL 1170 RETVAL
871 1171
872SV *ascii (SV *self, int enable) 1172SV *ascii (SV *self, int enable = 1)
873 ALIAS: 1173 ALIAS:
874 ascii = F_ASCII 1174 ascii = F_ASCII
1175 latin1 = F_LATIN1
875 utf8 = F_UTF8 1176 utf8 = F_UTF8
876 indent = F_INDENT 1177 indent = F_INDENT
877 canonical = F_CANONICAL 1178 canonical = F_CANONICAL
878 space_before = F_SPACE_BEFORE 1179 space_before = F_SPACE_BEFORE
879 space_after = F_SPACE_AFTER 1180 space_after = F_SPACE_AFTER
880 json_rpc = F_JSON_RPC
881 pretty = F_PRETTY 1181 pretty = F_PRETTY
882 allow_nonref = F_ALLOW_NONREF 1182 allow_nonref = F_ALLOW_NONREF
1183 shrink = F_SHRINK
883 CODE: 1184 CODE:
884{ 1185{
885 UV *uv = SvJSON (self); 1186 UV *uv = SvJSON (self);
886 if (enable) 1187 if (enable)
887 *uv |= ix; 1188 *uv |= ix;
… …
891 RETVAL = newSVsv (self); 1192 RETVAL = newSVsv (self);
892} 1193}
893 OUTPUT: 1194 OUTPUT:
894 RETVAL 1195 RETVAL
895 1196
1197SV *max_depth (SV *self, UV max_depth = 0x80000000UL)
1198 CODE:
1199{
1200 UV *uv = SvJSON (self);
1201 UV log2 = 0;
1202
1203 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1204
1205 while ((1UL << log2) < max_depth)
1206 ++log2;
1207
1208 *uv = *uv & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1209
1210 RETVAL = newSVsv (self);
1211}
1212 OUTPUT:
1213 RETVAL
1214
896void encode (SV *self, SV *scalar) 1215void encode (SV *self, SV *scalar)
897 PPCODE: 1216 PPCODE:
898 XPUSHs (encode_json (scalar, *SvJSON (self))); 1217 XPUSHs (encode_json (scalar, *SvJSON (self)));
899 1218
900void decode (SV *self, SV *jsonstr) 1219void decode (SV *self, SV *jsonstr)
901 PPCODE: 1220 PPCODE:
902 XPUSHs (decode_json (jsonstr, *SvJSON (self))); 1221 XPUSHs (decode_json (jsonstr, *SvJSON (self), 0));
1222
1223void decode_prefix (SV *self, SV *jsonstr)
1224 PPCODE:
1225{
1226 UV offset;
1227 EXTEND (SP, 2);
1228 PUSHs (decode_json (jsonstr, *SvJSON (self), &offset));
1229 PUSHs (sv_2mortal (newSVuv (offset)));
1230}
903 1231
904PROTOTYPES: ENABLE 1232PROTOTYPES: ENABLE
905 1233
906void to_json (SV *scalar) 1234void to_json (SV *scalar)
1235 ALIAS:
1236 objToJson = 0
907 PPCODE: 1237 PPCODE:
908 XPUSHs (encode_json (scalar, F_UTF8)); 1238 XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8));
909 1239
910void from_json (SV *jsonstr) 1240void from_json (SV *jsonstr)
1241 ALIAS:
1242 jsonToObj = 0
911 PPCODE: 1243 PPCODE:
912 XPUSHs (decode_json (jsonstr, F_UTF8)); 1244 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8, 0));
913 1245

Diff Legend

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