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 1.39 by root, Mon Jun 11 03:42:57 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 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 = SvEND (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}
… …
79static void 136static void
80encode_str (enc_t *enc, char *str, STRLEN len, int is_utf8) 137encode_str (enc_t *enc, char *str, STRLEN len, int is_utf8)
81{ 138{
82 char *end = str + len; 139 char *end = str + len;
83 140
141 need (enc, len);
142
84 while (str < end) 143 while (str < end)
85 { 144 {
86 unsigned char ch = *(unsigned char *)str; 145 unsigned char ch = *(unsigned char *)str;
146
87 if (ch >= 0x20 && ch < 0x80) // most common case 147 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
88 { 148 {
149 if (expect_false (ch == '"')) // but with slow exceptions
150 {
151 need (enc, len += 1);
152 *enc->cur++ = '\\';
153 *enc->cur++ = '"';
154 }
155 else if (expect_false (ch == '\\'))
156 {
157 need (enc, len += 1);
158 *enc->cur++ = '\\';
159 *enc->cur++ = '\\';
160 }
161 else
89 *enc->cur++ = ch; 162 *enc->cur++ = ch;
163
90 str++; 164 ++str;
91 } 165 }
92 else 166 else
93 { 167 {
94 STRLEN clen; 168 switch (ch)
95 UV uch;
96
97 if (is_utf8)
98 { 169 {
99 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;
100 if (clen < 0) 171 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"); 172 case '\012': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'n'; ++str; break;
102 } 173 case '\014': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'f'; ++str; break;
103 else 174 case '\015': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'r'; ++str; break;
104 {
105 uch = ch;
106 clen = 1;
107 }
108 175
109 need (enc, len += 6); 176 default:
110
111 if (uch < 0xa0 || enc->flags & F_ASCII)
112 {
113 if (uch > 0xFFFFUL)
114 { 177 {
178 STRLEN clen;
179 UV uch;
180
181 if (is_utf8)
182 {
183 //uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
184 uch = decode_utf8 (str, end - str, &clen);
185 if (clen == (STRLEN)-1)
186 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
187 }
188 else
189 {
190 uch = ch;
115 len += 6; 191 clen = 1;
192 }
193
194 if (uch > 0x10FFFFUL)
195 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
196
197 if (uch < 0x80 || enc->flags & F_ASCII || (enc->flags & F_LATIN1 && uch > 0xFF))
198 {
199 if (uch > 0xFFFFUL)
200 {
116 need (enc, len += 6); 201 need (enc, len += 11);
117 sprintf (enc->cur, "\\u%04x\\u%04x", 202 sprintf (enc->cur, "\\u%04x\\u%04x",
118 (uch - 0x10000) / 0x400 + 0xD800, 203 (int)((uch - 0x10000) / 0x400 + 0xD800),
119 (uch - 0x10000) % 0x400 + 0xDC00); 204 (int)((uch - 0x10000) % 0x400 + 0xDC00));
120 enc->cur += 12; 205 enc->cur += 12;
206 }
207 else
208 {
209 static char hexdigit [16] = "0123456789abcdef";
210 need (enc, len += 5);
211 *enc->cur++ = '\\';
212 *enc->cur++ = 'u';
213 *enc->cur++ = hexdigit [ uch >> 12 ];
214 *enc->cur++ = hexdigit [(uch >> 8) & 15];
215 *enc->cur++ = hexdigit [(uch >> 4) & 15];
216 *enc->cur++ = hexdigit [(uch >> 0) & 15];
217 }
218
219 str += clen;
121 } 220 }
221 else if (enc->flags & F_LATIN1)
222 {
223 *enc->cur++ = uch;
224 str += clen;
225 }
226 else if (is_utf8)
227 {
228 need (enc, len += clen);
229 do
230 {
231 *enc->cur++ = *str++;
232 }
233 while (--clen);
234 }
122 else 235 else
123 { 236 {
124 sprintf (enc->cur, "\\u%04x", uch); 237 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
125 enc->cur += 6; 238 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
239 ++str;
240 }
126 } 241 }
127 } 242 }
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 } 243 }
138 244
139 --len; 245 --len;
140 } 246 }
141} 247}
142 248
143#define INDENT SB \ 249inline void
250encode_indent (enc_t *enc)
251{
144 if (enc->flags & F_INDENT) \ 252 if (enc->flags & F_INDENT)
145 { \ 253 {
146 int i_; \ 254 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
152 255
153#define SPACE SB need (enc, 1); encode_ch (enc, ' '); SE 256 need (enc, spaces);
154#define NL SB if (enc->flags & F_INDENT) { need (enc, 1); encode_ch (enc, '\n'); } SE 257 memset (enc->cur, ' ', spaces);
155#define COMMA SB \ 258 enc->cur += spaces;
259 }
260}
261
262inline void
263encode_space (enc_t *enc)
264{
265 need (enc, 1);
156 encode_ch (enc, ','); \ 266 encode_ch (enc, ' ');
267}
268
269inline void
270encode_nl (enc_t *enc)
271{
157 if (enc->flags & F_INDENT) \ 272 if (enc->flags & F_INDENT)
158 NL; \ 273 {
274 need (enc, 1);
275 encode_ch (enc, '\n');
276 }
277}
278
279inline void
280encode_comma (enc_t *enc)
281{
282 encode_ch (enc, ',');
283
284 if (enc->flags & F_INDENT)
285 encode_nl (enc);
159 else if (enc->flags & F_SPACE_AFTER) \ 286 else if (enc->flags & F_SPACE_AFTER)
160 SPACE; \ 287 encode_space (enc);
161 SE 288}
162 289
163static void encode_sv (enc_t *enc, SV *sv); 290static void encode_sv (enc_t *enc, SV *sv);
164 291
165static void 292static void
166encode_av (enc_t *enc, AV *av) 293encode_av (enc_t *enc, AV *av)
167{ 294{
168 int i, len = av_len (av); 295 int i, len = av_len (av);
169 296
170 encode_ch (enc, '['); NL; 297 if (enc->indent >= enc->maxdepth)
298 croak ("data structure too deep (hit recursion limit)");
299
300 encode_ch (enc, '['); encode_nl (enc);
171 ++enc->indent; 301 ++enc->indent;
172 302
173 for (i = 0; i <= len; ++i) 303 for (i = 0; i <= len; ++i)
174 { 304 {
175 INDENT; 305 encode_indent (enc);
176 encode_sv (enc, *av_fetch (av, i, 0)); 306 encode_sv (enc, *av_fetch (av, i, 0));
177 307
178 if (i < len) 308 if (i < len)
179 COMMA; 309 encode_comma (enc);
180 } 310 }
181 311
182 NL; 312 encode_nl (enc);
183 313
184 --enc->indent; 314 --enc->indent;
185 INDENT; encode_ch (enc, ']'); 315 encode_indent (enc); encode_ch (enc, ']');
186} 316}
187 317
188static void 318static void
189encode_he (enc_t *enc, HE *he) 319encode_he (enc_t *enc, HE *he)
190{ 320{
… …
192 322
193 if (HeKLEN (he) == HEf_SVKEY) 323 if (HeKLEN (he) == HEf_SVKEY)
194 { 324 {
195 SV *sv = HeSVKEY (he); 325 SV *sv = HeSVKEY (he);
196 STRLEN len; 326 STRLEN len;
327 char *str;
328
329 SvGETMAGIC (sv);
197 char *str = SvPV (sv, len); 330 str = SvPV (sv, len);
198 331
199 encode_str (enc, str, len, SvUTF8 (sv)); 332 encode_str (enc, str, len, SvUTF8 (sv));
200 } 333 }
201 else 334 else
202 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 335 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
203 336
204 encode_ch (enc, '"'); 337 encode_ch (enc, '"');
205 338
206 if (enc->flags & F_SPACE_BEFORE) SPACE; 339 if (enc->flags & F_SPACE_BEFORE) encode_space (enc);
207 encode_ch (enc, ':'); 340 encode_ch (enc, ':');
208 if (enc->flags & F_SPACE_AFTER ) SPACE; 341 if (enc->flags & F_SPACE_AFTER ) encode_space (enc);
209 encode_sv (enc, HeVAL (he)); 342 encode_sv (enc, HeVAL (he));
210} 343}
211 344
212// compare hash entries, used when all keys are bytestrings 345// compare hash entries, used when all keys are bytestrings
213static int 346static int
… …
219 HE *b = *(HE **)b_; 352 HE *b = *(HE **)b_;
220 353
221 STRLEN la = HeKLEN (a); 354 STRLEN la = HeKLEN (a);
222 STRLEN lb = HeKLEN (b); 355 STRLEN lb = HeKLEN (b);
223 356
224 if (!(cmp == memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 357 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb)))
225 cmp = la < lb ? -1 : la == lb ? 0 : 1; 358 cmp = la - lb;
226 359
227 return cmp; 360 return cmp;
228} 361}
229 362
230// compare hash entries, used when some keys are sv's or utf-x 363// compare hash entries, used when some keys are sv's or utf-x
… …
237static void 370static void
238encode_hv (enc_t *enc, HV *hv) 371encode_hv (enc_t *enc, HV *hv)
239{ 372{
240 int count, i; 373 int count, i;
241 374
375 if (enc->indent >= enc->maxdepth)
376 croak ("data structure too deep (hit recursion limit)");
377
242 encode_ch (enc, '{'); NL; ++enc->indent; 378 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
243 379
244 if ((count = hv_iterinit (hv))) 380 if ((count = hv_iterinit (hv)))
245 { 381 {
246 // for canonical output we have to sort by keys first 382 // for canonical output we have to sort by keys first
247 // actually, this is mostly due to the stupid so-called 383 // actually, this is mostly due to the stupid so-called
248 // security workaround added somewhere in 5.8.x. 384 // security workaround added somewhere in 5.8.x.
249 // that randomises hash orderings 385 // that randomises hash orderings
250 if (enc->flags & F_CANONICAL) 386 if (enc->flags & F_CANONICAL)
251 { 387 {
252 HE *he, *hes [count];
253 int fast = 1; 388 int fast = 1;
389 HE *he;
390#if defined(__BORLANDC__) || defined(_MSC_VER)
391 HE **hes = _alloca (count * sizeof (HE));
392#else
393 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
394#endif
254 395
255 i = 0; 396 i = 0;
256 while ((he = hv_iternext (hv))) 397 while ((he = hv_iternext (hv)))
257 { 398 {
258 hes [i++] = he; 399 hes [i++] = he;
… …
264 405
265 if (fast) 406 if (fast)
266 qsort (hes, count, sizeof (HE *), he_cmp_fast); 407 qsort (hes, count, sizeof (HE *), he_cmp_fast);
267 else 408 else
268 { 409 {
269 // hack to disable "use bytes" 410 // hack to forcefully disable "use bytes"
270 COP *oldcop = PL_curcop, cop; 411 COP cop = *PL_curcop;
271 cop.op_private = 0; 412 cop.op_private = 0;
413
414 ENTER;
415 SAVETMPS;
416
417 SAVEVPTR (PL_curcop);
272 PL_curcop = &cop; 418 PL_curcop = &cop;
273 419
274 SAVETMPS;
275 qsort (hes, count, sizeof (HE *), he_cmp_slow); 420 qsort (hes, count, sizeof (HE *), he_cmp_slow);
421
276 FREETMPS; 422 FREETMPS;
277 423 LEAVE;
278 PL_curcop = oldcop;
279 } 424 }
280 425
281 for (i = 0; i < count; ++i) 426 for (i = 0; i < count; ++i)
282 { 427 {
283 INDENT; 428 encode_indent (enc);
284 encode_he (enc, hes [i]); 429 encode_he (enc, hes [i]);
285 430
286 if (i < count - 1) 431 if (i < count - 1)
287 COMMA; 432 encode_comma (enc);
288 } 433 }
289 434
290 NL; 435 encode_nl (enc);
291 } 436 }
292 else 437 else
293 { 438 {
294 SV *sv;
295 HE *he = hv_iternext (hv); 439 HE *he = hv_iternext (hv);
296 440
297 for (;;) 441 for (;;)
298 { 442 {
299 INDENT; 443 encode_indent (enc);
300 encode_he (enc, he); 444 encode_he (enc, he);
301 445
302 if (!(he = hv_iternext (hv))) 446 if (!(he = hv_iternext (hv)))
303 break; 447 break;
304 448
305 COMMA; 449 encode_comma (enc);
306 } 450 }
307 451
308 NL; 452 encode_nl (enc);
309 } 453 }
310 } 454 }
311 455
312 --enc->indent; INDENT; encode_ch (enc, '}'); 456 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
457}
458
459// encode objects, arrays and special \0=false and \1=true values.
460static void
461encode_rv (enc_t *enc, SV *sv)
462{
463 svtype svt;
464
465 SvGETMAGIC (sv);
466 svt = SvTYPE (sv);
467
468 if (svt == SVt_PVHV)
469 encode_hv (enc, (HV *)sv);
470 else if (svt == SVt_PVAV)
471 encode_av (enc, (AV *)sv);
472 else if (svt < SVt_PVAV)
473 {
474 if (SvNIOK (sv) && SvIV (sv) == 0)
475 encode_str (enc, "false", 5, 0);
476 else if (SvNIOK (sv) && SvIV (sv) == 1)
477 encode_str (enc, "true", 4, 0);
478 else
479 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
480 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
481 }
482 else
483 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
484 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
313} 485}
314 486
315static void 487static void
316encode_sv (enc_t *enc, SV *sv) 488encode_sv (enc_t *enc, SV *sv)
317{ 489{
490 SvGETMAGIC (sv);
491
318 if (SvPOKp (sv)) 492 if (SvPOKp (sv))
319 { 493 {
320 STRLEN len; 494 STRLEN len;
321 char *str = SvPV (sv, len); 495 char *str = SvPV (sv, len);
322 encode_ch (enc, '"'); 496 encode_ch (enc, '"');
323 encode_str (enc, str, len, SvUTF8 (sv)); 497 encode_str (enc, str, len, SvUTF8 (sv));
324 encode_ch (enc, '"'); 498 encode_ch (enc, '"');
325 } 499 }
326 else if (SvNOKp (sv)) 500 else if (SvNOKp (sv))
327 { 501 {
502 // trust that perl will do the right thing w.r.t. JSON syntax.
328 need (enc, NV_DIG + 32); 503 need (enc, NV_DIG + 32);
329 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 504 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
330 enc->cur += strlen (enc->cur); 505 enc->cur += strlen (enc->cur);
331 } 506 }
332 else if (SvIOKp (sv)) 507 else if (SvIOKp (sv))
333 { 508 {
334 need (enc, 64); 509 // we assume we can always read an IV as a UV
510 if (SvUV (sv) & ~(UV)0x7fff)
511 {
512 // large integer, use the (rather slow) snprintf way.
513 need (enc, sizeof (UV) * 3);
335 enc->cur += 514 enc->cur +=
336 SvIsUV(sv) 515 SvIsUV(sv)
337 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 516 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
338 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 517 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
518 }
519 else
520 {
521 // optimise the "small number case"
522 // code will likely be branchless and use only a single multiplication
523 I32 i = SvIV (sv);
524 U32 u;
525 char digit, nz = 0;
526
527 need (enc, 6);
528
529 *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
530 u = i < 0 ? -i : i;
531
532 // convert to 4.28 fixed-point representation
533 u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
534
535 // now output digit by digit, each time masking out the integer part
536 // and multiplying by 5 while moving the decimal point one to the right,
537 // resulting in a net multiplication by 10.
538 // we always write the digit to memory but conditionally increment
539 // the pointer, to ease the usage of conditional move instructions.
540 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5;
541 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5;
542 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5;
543 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5;
544 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
545 }
339 } 546 }
340 else if (SvROK (sv)) 547 else if (SvROK (sv))
341 { 548 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)) 549 else if (!SvOK (sv))
357 encode_str (enc, "null", 4, 0); 550 encode_str (enc, "null", 4, 0);
358 else 551 else
359 croak ("encountered perl type that JSON cannot handle"); 552 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
553 SvPV_nolen (sv), SvFLAGS (sv));
360} 554}
361 555
362static SV * 556static SV *
363encode_json (SV *scalar, UV flags) 557encode_json (SV *scalar, U32 flags)
364{ 558{
559 enc_t enc;
560
365 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 561 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar))
366 croak ("hash- or arraref required (not a simple scalar, use allow_nonref to allow this)"); 562 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
367 563
368 enc_t enc;
369 enc.flags = flags; 564 enc.flags = flags;
370 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 565 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
371 enc.cur = SvPVX (enc.sv); 566 enc.cur = SvPVX (enc.sv);
372 enc.end = SvEND (enc.sv); 567 enc.end = SvEND (enc.sv);
373 enc.max_recurse = 0;
374 enc.indent = 0; 568 enc.indent = 0;
569 enc.maxdepth = DEC_DEPTH (flags);
375 570
376 SvPOK_only (enc.sv); 571 SvPOK_only (enc.sv);
377 encode_sv (&enc, scalar); 572 encode_sv (&enc, scalar);
378 573
574 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
575 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
576
379 if (!(flags & (F_ASCII | F_UTF8))) 577 if (!(flags & (F_ASCII | F_LATIN1 | F_UTF8)))
380 SvUTF8_on (enc.sv); 578 SvUTF8_on (enc.sv);
381 579
382 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 580 if (enc.flags & F_SHRINK)
581 shrink (enc.sv);
582
383 return enc.sv; 583 return enc.sv;
384} 584}
385 585
386///////////////////////////////////////////////////////////////////////////// 586/////////////////////////////////////////////////////////////////////////////
587// decoder
387 588
388#define WS \ 589// structure used for decoding JSON
590typedef struct
591{
592 char *cur; // current parser pointer
593 char *end; // end of input string
594 const char *err; // parse error, if != 0
595 U32 flags; // F_*
596 U32 depth; // recursion depth
597 U32 maxdepth; // recursion depth limit
598} dec_t;
599
600inline void
601decode_ws (dec_t *dec)
602{
389 for (;;) \ 603 for (;;)
390 { \ 604 {
391 char ch = *dec->cur; \ 605 char ch = *dec->cur;
606
392 if (ch > 0x20 \ 607 if (ch > 0x20
393 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) \ 608 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09))
394 break; \ 609 break;
610
395 ++dec->cur; \ 611 ++dec->cur;
396 } 612 }
613}
397 614
398#define ERR(reason) SB dec->err = reason; goto fail; SE 615#define ERR(reason) SB dec->err = reason; goto fail; SE
616
399#define EXPECT_CH(ch) SB \ 617#define EXPECT_CH(ch) SB \
400 if (*dec->cur != ch) \ 618 if (*dec->cur != ch) \
401 ERR (# ch " expected"); \ 619 ERR (# ch " expected"); \
402 ++dec->cur; \ 620 ++dec->cur; \
403 SE 621 SE
404 622
623#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)")
624#define DEC_DEC_DEPTH --dec->depth
625
405static SV *decode_sv (dec_t *dec); 626static 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
411 627
412static signed char decode_hexdigit[256]; 628static signed char decode_hexdigit[256];
413 629
414static UV 630static UV
415decode_4hex (dec_t *dec) 631decode_4hex (dec_t *dec)
416{ 632{
417 signed char d1, d2, d3, d4; 633 signed char d1, d2, d3, d4;
634 unsigned char *cur = (unsigned char *)dec->cur;
418 635
419 d1 = decode_hexdigit [((unsigned char *)dec->cur) [0]]; 636 d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("four hexadecimal digits expected");
420 if (d1 < 0) ERR ("four hexadecimal digits expected"); 637 d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("four hexadecimal digits expected");
421 d2 = decode_hexdigit [((unsigned char *)dec->cur) [1]]; 638 d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("four hexadecimal digits expected");
422 if (d2 < 0) ERR ("four hexadecimal digits expected"); 639 d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("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");
427 640
428 dec->cur += 4; 641 dec->cur += 4;
429 642
430 return ((UV)d1) << 12 643 return ((UV)d1) << 12
431 | ((UV)d2) << 8 644 | ((UV)d2) << 8
… …
437} 650}
438 651
439static SV * 652static SV *
440decode_str (dec_t *dec) 653decode_str (dec_t *dec)
441{ 654{
442 SV *sv = NEWSV (0,2); 655 SV *sv = 0;
443 STRLEN cur = 0;
444 int utf8 = 0; 656 int utf8 = 0;
657 char *dec_cur = dec->cur;
445 658
446 for (;;) 659 do
447 { 660 {
448 unsigned char ch = *(unsigned char *)dec->cur; 661 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
662 char *cur = buf;
449 663
450 if (ch == '"') 664 do
451 break;
452 else if (ch == '\\')
453 { 665 {
454 switch (*++dec->cur) 666 unsigned char ch = *(unsigned char *)dec_cur++;
667
668 if (expect_false (ch == '"'))
455 { 669 {
456 case '\\': 670 --dec_cur;
457 case '/': 671 break;
458 case '"': APPEND_CH (*dec->cur++); break; 672 }
459 673 else if (expect_false (ch == '\\'))
460 case 'b': APPEND_CH ('\010'); ++dec->cur; break; 674 {
461 case 't': APPEND_CH ('\011'); ++dec->cur; break; 675 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 { 676 {
468 UV lo, hi; 677 case '\\':
469 ++dec->cur; 678 case '/':
679 case '"': *cur++ = *dec_cur++; break;
470 680
471 hi = decode_4hex (dec); 681 case 'b': ++dec_cur; *cur++ = '\010'; break;
472 if (hi == (UV)-1) 682 case 't': ++dec_cur; *cur++ = '\011'; break;
473 goto fail; 683 case 'n': ++dec_cur; *cur++ = '\012'; break;
684 case 'f': ++dec_cur; *cur++ = '\014'; break;
685 case 'r': ++dec_cur; *cur++ = '\015'; break;
474 686
475 // possibly a surrogate pair 687 case 'u':
476 if (hi >= 0xd800 && hi < 0xdc00)
477 { 688 {
478 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 689 UV lo, hi;
479 ERR ("illegal surrogate character"); 690 ++dec_cur;
480 691
481 dec->cur += 2; 692 dec->cur = dec_cur;
482
483 lo = decode_4hex (dec); 693 hi = decode_4hex (dec);
694 dec_cur = dec->cur;
484 if (lo == (UV)-1) 695 if (hi == (UV)-1)
485 goto fail; 696 goto fail;
486 697
698 // possibly a surrogate pair
699 if (hi >= 0xd800)
700 if (hi < 0xdc00)
701 {
702 if (dec_cur [0] != '\\' || dec_cur [1] != 'u')
703 ERR ("missing low surrogate character in surrogate pair");
704
705 dec_cur += 2;
706
707 dec->cur = dec_cur;
708 lo = decode_4hex (dec);
709 dec_cur = dec->cur;
710 if (lo == (UV)-1)
711 goto fail;
712
487 if (lo < 0xdc00 || lo >= 0xe000) 713 if (lo < 0xdc00 || lo >= 0xe000)
488 ERR ("surrogate pair expected"); 714 ERR ("surrogate pair expected");
489 715
490 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000; 716 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000;
717 }
718 else if (hi < 0xe000)
719 ERR ("missing high surrogate character in surrogate pair");
720
721 if (hi >= 0x80)
722 {
723 utf8 = 1;
724
725 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
726 }
727 else
728 *cur++ = hi;
491 } 729 }
492 else if (lo >= 0xdc00 && lo < 0xe000)
493 ERR ("illegal surrogate character");
494
495 if (hi >= 0x80)
496 { 730 break;
497 utf8 = 1;
498 731
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 732 default:
503 APPEND_CH (hi); 733 --dec_cur;
734 ERR ("illegal backslash escape sequence in string");
504 } 735 }
736 }
737 else if (expect_true (ch >= 0x20 && ch <= 0x7f))
738 *cur++ = ch;
739 else if (ch >= 0x80)
740 {
741 STRLEN clen;
742 UV uch;
743
744 --dec_cur;
745
746 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
747 if (clen == (STRLEN)-1)
748 ERR ("malformed UTF-8 character in JSON string");
749
750 do
751 *cur++ = *dec_cur++;
752 while (--clen);
753
754 utf8 = 1;
755 }
756 else
757 {
758 --dec_cur;
759
760 if (!ch)
761 ERR ("unexpected end of string while parsing JSON string");
505 break; 762 else
763 ERR ("invalid character encountered while parsing JSON string");
506 } 764 }
507 } 765 }
508 else if (ch >= 0x20 && ch <= 0x7f) 766 while (cur < buf + SHORT_STRING_LEN);
509 APPEND_CH (*dec->cur++); 767
510 else if (ch >= 0x80) 768 {
769 STRLEN len = cur - buf;
770
771 if (sv)
511 { 772 {
512 STRLEN clen; 773 SvGROW (sv, SvCUR (sv) + len + 1);
513 UV uch = utf8n_to_uvuni (dec->cur, dec->end - dec->cur, &clen, UTF8_CHECK_ONLY); 774 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
514 if (clen < 0) 775 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 } 776 }
521 else 777 else
522 ERR ("invalid character encountered"); 778 sv = newSVpvn (buf, len);
523 } 779 }
780 }
781 while (*dec_cur != '"');
524 782
525 ++dec->cur; 783 ++dec_cur;
526 784
785 if (sv)
786 {
527 SvPOK_only (sv); 787 SvPOK_only (sv);
528
529 SvCUR_set (sv, cur);
530 *SvEND (sv) = 0; 788 *SvEND (sv) = 0;
531 789
532 if (utf8) 790 if (utf8)
533 SvUTF8_on (sv); 791 SvUTF8_on (sv);
792 }
793 else
794 sv = newSVpvn ("", 0);
534 795
796 dec->cur = dec_cur;
535 return sv; 797 return sv;
536 798
537fail: 799fail:
538 SvREFCNT_dec (sv); 800 dec->cur = dec_cur;
539 return 0; 801 return 0;
540} 802}
541 803
542static SV * 804static SV *
543decode_num (dec_t *dec) 805decode_num (dec_t *dec)
… …
553 { 815 {
554 ++dec->cur; 816 ++dec->cur;
555 if (*dec->cur >= '0' && *dec->cur <= '9') 817 if (*dec->cur >= '0' && *dec->cur <= '9')
556 ERR ("malformed number (leading zero must not be followed by another digit)"); 818 ERR ("malformed number (leading zero must not be followed by another digit)");
557 } 819 }
558 820 else if (*dec->cur < '0' || *dec->cur > '9')
559 // int 821 ERR ("malformed number (no digits after initial minus)");
822 else
823 do
824 {
825 ++dec->cur;
826 }
560 while (*dec->cur >= '0' && *dec->cur <= '9') 827 while (*dec->cur >= '0' && *dec->cur <= '9');
561 ++dec->cur;
562 828
563 // [frac] 829 // [frac]
564 if (*dec->cur == '.') 830 if (*dec->cur == '.')
565 { 831 {
566 is_nv = 1; 832 ++dec->cur;
833
834 if (*dec->cur < '0' || *dec->cur > '9')
835 ERR ("malformed number (no digits after decimal point)");
567 836
568 do 837 do
569 { 838 {
570 ++dec->cur; 839 ++dec->cur;
571 } 840 }
572 while (*dec->cur >= '0' && *dec->cur <= '9'); 841 while (*dec->cur >= '0' && *dec->cur <= '9');
842
843 is_nv = 1;
573 } 844 }
574 845
575 // [exp] 846 // [exp]
576 if (*dec->cur == 'e' || *dec->cur == 'E') 847 if (*dec->cur == 'e' || *dec->cur == 'E')
577 { 848 {
578 is_nv = 1;
579
580 ++dec->cur; 849 ++dec->cur;
850
581 if (*dec->cur == '-' || *dec->cur == '+') 851 if (*dec->cur == '-' || *dec->cur == '+')
582 ++dec->cur; 852 ++dec->cur;
583 853
854 if (*dec->cur < '0' || *dec->cur > '9')
855 ERR ("malformed number (no digits after exp sign)");
856
857 do
858 {
859 ++dec->cur;
860 }
584 while (*dec->cur >= '0' && *dec->cur <= '9') 861 while (*dec->cur >= '0' && *dec->cur <= '9');
585 ++dec->cur; 862
863 is_nv = 1;
586 } 864 }
587 865
588 if (!is_nv) 866 if (!is_nv)
589 { 867 {
590 UV uv; 868 // 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); 869 if (*start == '-')
592 if (numtype & IS_NUMBER_IN_UV) 870 switch (dec->cur - start)
593 if (numtype & IS_NUMBER_NEG)
594 { 871 {
595 if (uv < (UV)IV_MIN) 872 case 2: return newSViv (-( start [1] - '0' ));
596 return newSViv (-(IV)uv); 873 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
874 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
875 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
597 } 876 }
877 else
878 switch (dec->cur - start)
879 {
880 case 1: return newSViv ( start [0] - '0' );
881 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
882 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
883 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
884 }
885
886 {
887 UV uv;
888 int numtype = grok_number (start, dec->cur - start, &uv);
889 if (numtype & IS_NUMBER_IN_UV)
890 if (numtype & IS_NUMBER_NEG)
891 {
892 if (uv < (UV)IV_MIN)
893 return newSViv (-(IV)uv);
894 }
598 else 895 else
599 return newSVuv (uv); 896 return newSVuv (uv);
897 }
600 } 898 }
601 899
602 return newSVnv (Atof (start)); 900 return newSVnv (Atof (start));
603 901
604fail: 902fail:
… …
608static SV * 906static SV *
609decode_av (dec_t *dec) 907decode_av (dec_t *dec)
610{ 908{
611 AV *av = newAV (); 909 AV *av = newAV ();
612 910
911 DEC_INC_DEPTH;
912 decode_ws (dec);
913
914 if (*dec->cur == ']')
915 ++dec->cur;
916 else
613 for (;;) 917 for (;;)
614 { 918 {
615 SV *value; 919 SV *value;
616 920
617 value = decode_sv (dec); 921 value = decode_sv (dec);
618 if (!value) 922 if (!value)
619 goto fail; 923 goto fail;
620 924
621 av_push (av, value); 925 av_push (av, value);
622 926
623 WS; 927 decode_ws (dec);
624 928
625 if (*dec->cur == ']') 929 if (*dec->cur == ']')
626 { 930 {
627 ++dec->cur; 931 ++dec->cur;
628 break; 932 break;
933 }
629 } 934
630
631 if (*dec->cur != ',') 935 if (*dec->cur != ',')
632 ERR (", or ] expected while parsing array"); 936 ERR (", or ] expected while parsing array");
633 937
634 ++dec->cur; 938 ++dec->cur;
635 } 939 }
636 940
941 DEC_DEC_DEPTH;
637 return newRV_noinc ((SV *)av); 942 return newRV_noinc ((SV *)av);
638 943
639fail: 944fail:
640 SvREFCNT_dec (av); 945 SvREFCNT_dec (av);
946 DEC_DEC_DEPTH;
641 return 0; 947 return 0;
642} 948}
643 949
644static SV * 950static SV *
645decode_hv (dec_t *dec) 951decode_hv (dec_t *dec)
646{ 952{
647 HV *hv = newHV (); 953 HV *hv = newHV ();
648 954
955 DEC_INC_DEPTH;
956 decode_ws (dec);
957
958 if (*dec->cur == '}')
959 ++dec->cur;
960 else
649 for (;;) 961 for (;;)
650 { 962 {
651 SV *key, *value; 963 SV *key, *value;
652 964
653 WS; EXPECT_CH ('"'); 965 decode_ws (dec); EXPECT_CH ('"');
654 966
655 key = decode_str (dec); 967 key = decode_str (dec);
656 if (!key) 968 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; 969 goto fail;
970
971 decode_ws (dec); EXPECT_CH (':');
972
973 value = decode_sv (dec);
974 if (!value)
975 {
976 SvREFCNT_dec (key);
977 goto fail;
666 } 978 }
667 979
668 //TODO: optimise
669 hv_store_ent (hv, key, value, 0); 980 hv_store_ent (hv, key, value, 0);
981 SvREFCNT_dec (key);
670 982
671 WS; 983 decode_ws (dec);
672 984
673 if (*dec->cur == '}') 985 if (*dec->cur == '}')
674 { 986 {
675 ++dec->cur; 987 ++dec->cur;
676 break; 988 break;
677 } 989 }
678 990
679 if (*dec->cur != ',') 991 if (*dec->cur != ',')
680 ERR (", or } expected while parsing object/hash"); 992 ERR (", or } expected while parsing object/hash");
681 993
682 ++dec->cur; 994 ++dec->cur;
683 } 995 }
684 996
997 DEC_DEC_DEPTH;
685 return newRV_noinc ((SV *)hv); 998 return newRV_noinc ((SV *)hv);
686 999
687fail: 1000fail:
688 SvREFCNT_dec (hv); 1001 SvREFCNT_dec (hv);
1002 DEC_DEC_DEPTH;
689 return 0; 1003 return 0;
690} 1004}
691 1005
692static SV * 1006static SV *
693decode_sv (dec_t *dec) 1007decode_sv (dec_t *dec)
694{ 1008{
695 WS; 1009 decode_ws (dec);
696 switch (*dec->cur) 1010 switch (*dec->cur)
697 { 1011 {
698 case '"': ++dec->cur; return decode_str (dec); 1012 case '"': ++dec->cur; return decode_str (dec);
699 case '[': ++dec->cur; return decode_av (dec); 1013 case '[': ++dec->cur; return decode_av (dec);
700 case '{': ++dec->cur; return decode_hv (dec); 1014 case '{': ++dec->cur; return decode_hv (dec);
… …
728 1042
729 case 'n': 1043 case 'n':
730 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4)) 1044 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4))
731 { 1045 {
732 dec->cur += 4; 1046 dec->cur += 4;
733 return newSViv (1); 1047 return newSVsv (&PL_sv_undef);
734 } 1048 }
735 else 1049 else
736 ERR ("'null' expected"); 1050 ERR ("'null' expected");
737 1051
738 break; 1052 break;
739 1053
740 default: 1054 default:
741 ERR ("malformed json string"); 1055 ERR ("malformed JSON string, neither array, object, number, string or atom");
742 break; 1056 break;
743 } 1057 }
744 1058
745fail: 1059fail:
746 return 0; 1060 return 0;
747} 1061}
748 1062
749static SV * 1063static SV *
750decode_json (SV *string, UV flags) 1064decode_json (SV *string, U32 flags, UV *offset_return)
751{ 1065{
1066 dec_t dec;
1067 UV offset;
752 SV *sv; 1068 SV *sv;
753 1069
1070 SvGETMAGIC (string);
1071 SvUPGRADE (string, SVt_PV);
1072
754 if (!(flags & F_UTF8)) 1073 if (flags & F_UTF8)
1074 sv_utf8_downgrade (string, 0);
1075 else
755 sv_utf8_upgrade (string); 1076 sv_utf8_upgrade (string);
756 1077
757 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1078 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
758 1079
759 dec_t dec;
760 dec.flags = flags; 1080 dec.flags = flags;
761 dec.cur = SvPVX (string); 1081 dec.cur = SvPVX (string);
762 dec.end = SvEND (string); 1082 dec.end = SvEND (string);
763 dec.err = 0; 1083 dec.err = 0;
1084 dec.depth = 0;
1085 dec.maxdepth = DEC_DEPTH (dec.flags);
764 1086
765 *dec.end = 1; // invalid anywhere 1087 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
766 sv = decode_sv (&dec); 1088 sv = decode_sv (&dec);
767 *dec.end = 0; 1089
1090 if (!(offset_return || !sv))
1091 {
1092 // check for trailing garbage
1093 decode_ws (&dec);
1094
1095 if (*dec.cur)
1096 {
1097 dec.err = "garbage after JSON object";
1098 SvREFCNT_dec (sv);
1099 sv = 0;
1100 }
1101 }
1102
1103 if (offset_return || !sv)
1104 {
1105 offset = dec.flags & F_UTF8
1106 ? dec.cur - SvPVX (string)
1107 : utf8_distance (dec.cur, SvPVX (string));
1108
1109 if (offset_return)
1110 *offset_return = offset;
1111 }
768 1112
769 if (!sv) 1113 if (!sv)
770 { 1114 {
771 IV offset = utf8_distance (dec.cur, SvPVX (string));
772 SV *uni = sv_newmortal (); 1115 SV *uni = sv_newmortal ();
773 1116
1117 // horrible hack to silence warning inside pv_uni_display
1118 COP cop = *PL_curcop;
1119 cop.cop_warnings = pWARN_NONE;
1120 ENTER;
1121 SAVEVPTR (PL_curcop);
1122 PL_curcop = &cop;
774 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1123 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1124 LEAVE;
1125
775 croak ("%s, at character %d (%s)", 1126 croak ("%s, at character offset %d [\"%s\"]",
776 dec.err, 1127 dec.err,
777 (int)offset, 1128 (int)offset,
778 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1129 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
779 } 1130 }
780 1131
781 sv = sv_2mortal (sv); 1132 sv = sv_2mortal (sv);
782 1133
783 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1134 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)"); 1135 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
785 1136
786 return sv; 1137 return sv;
787} 1138}
788 1139
1140/////////////////////////////////////////////////////////////////////////////
1141// XS interface functions
1142
789MODULE = JSON::XS PACKAGE = JSON::XS 1143MODULE = JSON::XS PACKAGE = JSON::XS
790 1144
791BOOT: 1145BOOT:
792{ 1146{
793 int i; 1147 int i;
794 1148
795 memset (decode_hexdigit, 0xff, 256); 1149 memset (decode_hexdigit, 0xff, 256);
1150
796 for (i = 10; i--; ) 1151 for (i = 0; i < 256; ++i)
797 decode_hexdigit ['0' + i] = i; 1152 decode_hexdigit [i] =
798 1153 i >= '0' && i <= '9' ? i - '0'
799 for (i = 6; --i; ) 1154 : i >= 'a' && i <= 'f' ? i - 'a' + 10
800 { 1155 : i >= 'A' && i <= 'F' ? i - 'A' + 10
801 decode_hexdigit ['a' + i] = 10 + i; 1156 : -1;
802 decode_hexdigit ['A' + i] = 10 + i;
803 }
804 1157
805 json_stash = gv_stashpv ("JSON::XS", 1); 1158 json_stash = gv_stashpv ("JSON::XS", 1);
806} 1159}
1160
1161PROTOTYPES: DISABLE
807 1162
808SV *new (char *dummy) 1163SV *new (char *dummy)
809 CODE: 1164 CODE:
810 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1165 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash);
811 OUTPUT: 1166 OUTPUT:
812 RETVAL 1167 RETVAL
813 1168
814SV *ascii (SV *self, int enable) 1169SV *ascii (SV *self, int enable = 1)
815 ALIAS: 1170 ALIAS:
816 ascii = F_ASCII 1171 ascii = F_ASCII
1172 latin1 = F_LATIN1
817 utf8 = F_UTF8 1173 utf8 = F_UTF8
818 indent = F_INDENT 1174 indent = F_INDENT
819 canonical = F_CANONICAL 1175 canonical = F_CANONICAL
820 space_before = F_SPACE_BEFORE 1176 space_before = F_SPACE_BEFORE
821 space_after = F_SPACE_AFTER 1177 space_after = F_SPACE_AFTER
822 json_rpc = F_JSON_RPC
823 pretty = F_PRETTY 1178 pretty = F_PRETTY
824 allow_nonref = F_ALLOW_NONREF 1179 allow_nonref = F_ALLOW_NONREF
1180 shrink = F_SHRINK
825 CODE: 1181 CODE:
826{ 1182{
827 UV *uv = SvJSON (self); 1183 UV *uv = SvJSON (self);
828 if (enable) 1184 if (enable)
829 *uv |= ix; 1185 *uv |= ix;
… …
833 RETVAL = newSVsv (self); 1189 RETVAL = newSVsv (self);
834} 1190}
835 OUTPUT: 1191 OUTPUT:
836 RETVAL 1192 RETVAL
837 1193
1194SV *max_depth (SV *self, UV max_depth = 0x80000000UL)
1195 CODE:
1196{
1197 UV *uv = SvJSON (self);
1198 UV log2 = 0;
1199
1200 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1201
1202 while ((1UL << log2) < max_depth)
1203 ++log2;
1204
1205 *uv = *uv & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1206
1207 RETVAL = newSVsv (self);
1208}
1209 OUTPUT:
1210 RETVAL
1211
838void encode (SV *self, SV *scalar) 1212void encode (SV *self, SV *scalar)
839 PPCODE: 1213 PPCODE:
840 XPUSHs (encode_json (scalar, *SvJSON (self))); 1214 XPUSHs (encode_json (scalar, *SvJSON (self)));
841 1215
842void decode (SV *self, SV *jsonstr) 1216void decode (SV *self, SV *jsonstr)
843 PPCODE: 1217 PPCODE:
844 XPUSHs (decode_json (jsonstr, *SvJSON (self))); 1218 XPUSHs (decode_json (jsonstr, *SvJSON (self), 0));
1219
1220void decode_prefix (SV *self, SV *jsonstr)
1221 PPCODE:
1222{
1223 UV offset;
1224 EXTEND (SP, 2);
1225 PUSHs (decode_json (jsonstr, *SvJSON (self), &offset));
1226 PUSHs (sv_2mortal (newSVuv (offset)));
1227}
1228
1229PROTOTYPES: ENABLE
845 1230
846void to_json (SV *scalar) 1231void to_json (SV *scalar)
1232 ALIAS:
1233 objToJson = 0
847 PPCODE: 1234 PPCODE:
848 XPUSHs (encode_json (scalar, F_UTF8)); 1235 XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8));
849 1236
850void from_json (SV *jsonstr) 1237void from_json (SV *jsonstr)
1238 ALIAS:
1239 jsonToObj = 0
851 PPCODE: 1240 PPCODE:
852 XPUSHs (decode_json (jsonstr, F_UTF8)); 1241 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8, 0));
853 1242

Diff Legend

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