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 1.58 by root, Mon Aug 13 16:06:25 2007 UTC

1#include "EXTERN.h" 1#include "EXTERN.h"
2#include "perl.h" 2#include "perl.h"
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>
9#include <float.h>
8 10
11#if defined(__BORLANDC__) || defined(_MSC_VER)
12# define snprintf _snprintf // C compilers have this in stdio.h
13#endif
14
15// some old perls do not have this, try to make it work, no
16// guarentees, though. if it breaks, you get to keep the pieces.
17#ifndef UTF8_MAXBYTES
18# define UTF8_MAXBYTES 13
19#endif
20
9#define F_ASCII 0x00000001 21#define F_ASCII 0x00000001UL
22#define F_LATIN1 0x00000002UL
10#define F_UTF8 0x00000002 23#define F_UTF8 0x00000004UL
11#define F_INDENT 0x00000004 24#define F_INDENT 0x00000008UL
12#define F_CANONICAL 0x00000008 25#define F_CANONICAL 0x00000010UL
13#define F_SPACE_BEFORE 0x00000010 26#define F_SPACE_BEFORE 0x00000020UL
14#define F_SPACE_AFTER 0x00000020 27#define F_SPACE_AFTER 0x00000040UL
15#define F_JSON_RPC 0x00000040
16#define F_ALLOW_NONREF 0x00000080 28#define F_ALLOW_NONREF 0x00000100UL
29#define F_SHRINK 0x00000200UL
30#define F_ALLOW_BLESSED 0x00000400UL
31#define F_CONV_BLESSED 0x00000800UL
32#define F_MAXDEPTH 0xf8000000UL
33#define S_MAXDEPTH 27
34#define F_MAXSIZE 0x01f00000UL
35#define S_MAXSIZE 20
36#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
37
38#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
39#define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE ))
17 40
18#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 41#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
19#define F_DEFAULT 0 42#define F_DEFAULT (9UL << S_MAXDEPTH)
20 43
21#define INIT_SIZE 32 // initial scalar size to be allocated 44#define INIT_SIZE 32 // initial scalar size to be allocated
45#define INDENT_STEP 3 // spaces per indentation level
46
47#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
22 48
23#define SB do { 49#define SB do {
24#define SE } while (0) 50#define SE } while (0)
25 51
26static HV *json_stash; 52#if __GNUC__ >= 3
53# define expect(expr,value) __builtin_expect ((expr),(value))
54# define inline inline
55#else
56# define expect(expr,value) (expr)
57# define inline static
58#endif
59
60#define expect_false(expr) expect ((expr) != 0, 0)
61#define expect_true(expr) expect ((expr) != 0, 1)
62
63#ifdef USE_ITHREADS
64# define JSON_SLOW 1
65#else
66# define JSON_SLOW 0
67#endif
68
69static HV *json_stash, *json_boolean_stash; // JSON::XS::
70static SV *json_true, *json_false;
71
72typedef struct {
73 U32 flags;
74 SV *cb_object;
75 HV *cb_sk_object;
76} JSON;
77
78/////////////////////////////////////////////////////////////////////////////
79// utility functions
80
81inline void
82shrink (SV *sv)
83{
84 sv_utf8_downgrade (sv, 1);
85 if (SvLEN (sv) > SvCUR (sv) + 1)
86 {
87#ifdef SvPV_shrink_to_cur
88 SvPV_shrink_to_cur (sv);
89#elif defined (SvPV_renew)
90 SvPV_renew (sv, SvCUR (sv) + 1);
91#endif
92 }
93}
94
95// decode an utf-8 character and return it, or (UV)-1 in
96// case of an error.
97// we special-case "safe" characters from U+80 .. U+7FF,
98// but use the very good perl function to parse anything else.
99// note that we never call this function for a ascii codepoints
100inline UV
101decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
102{
103 if (expect_false (s[0] > 0xdf || s[0] < 0xc2))
104 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
105 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)
106 {
107 *clen = 2;
108 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
109 }
110 else
111 {
112 *clen = (STRLEN)-1;
113 return (UV)-1;
114 }
115}
116
117/////////////////////////////////////////////////////////////////////////////
118// encoder
27 119
28// structure used for encoding JSON 120// structure used for encoding JSON
29typedef struct 121typedef struct
30{ 122{
31 char *cur; 123 char *cur; // SvPVX (sv) + current output position
32 STRLEN len; // SvLEN (sv)
33 char *end; // SvEND (sv) 124 char *end; // SvEND (sv)
34 SV *sv; 125 SV *sv; // result scalar
35 UV flags; 126 JSON json;
36 int max_recurse; 127 U32 indent; // indentation level
37 int indent; 128 U32 maxdepth; // max. indentation/recursion level
38} enc_t; 129} enc_t;
39 130
40// structure used for decoding JSON 131inline 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) 132need (enc_t *enc, STRLEN len)
62{ 133{
63 if (enc->cur + len >= enc->end) 134 if (expect_false (enc->cur + len >= enc->end))
64 { 135 {
65 STRLEN cur = enc->cur - SvPVX (enc->sv); 136 STRLEN cur = enc->cur - SvPVX (enc->sv);
66 SvGROW (enc->sv, cur + len + 1); 137 SvGROW (enc->sv, cur + len + 1);
67 enc->cur = SvPVX (enc->sv) + cur; 138 enc->cur = SvPVX (enc->sv) + cur;
68 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv); 139 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
69 } 140 }
70} 141}
71 142
72static void 143inline void
73encode_ch (enc_t *enc, char ch) 144encode_ch (enc_t *enc, char ch)
74{ 145{
75 need (enc, 1); 146 need (enc, 1);
76 *enc->cur++ = ch; 147 *enc->cur++ = ch;
77} 148}
… …
85 156
86 while (str < end) 157 while (str < end)
87 { 158 {
88 unsigned char ch = *(unsigned char *)str; 159 unsigned char ch = *(unsigned char *)str;
89 160
90 if (ch == '"') 161 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
91 { 162 {
163 if (expect_false (ch == '"')) // but with slow exceptions
164 {
92 need (enc, len += 1); 165 need (enc, len += 1);
93 *enc->cur++ = '\\'; 166 *enc->cur++ = '\\';
94 *enc->cur++ = '"'; 167 *enc->cur++ = '"';
95 ++str;
96 } 168 }
97 else if (ch == '\\') 169 else if (expect_false (ch == '\\'))
98 { 170 {
99 need (enc, len += 1); 171 need (enc, len += 1);
100 *enc->cur++ = '\\'; 172 *enc->cur++ = '\\';
101 *enc->cur++ = '\\'; 173 *enc->cur++ = '\\';
102 ++str;
103 } 174 }
104 else if (ch >= 0x20 && ch < 0x80) // most common case 175 else
105 {
106 *enc->cur++ = ch; 176 *enc->cur++ = ch;
107 ++str; 177
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; 178 ++str;
122 } 179 }
123 else 180 else
124 { 181 {
125 STRLEN clen; 182 switch (ch)
126 UV uch;
127
128 if (is_utf8)
129 {
130 uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
131 if (clen < 0)
132 croak ("malformed UTF-8 character in string, cannot convert to JSON");
133 } 183 {
134 else 184 case '\010': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'b'; ++str; break;
135 { 185 case '\011': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 't'; ++str; break;
136 uch = ch; 186 case '\012': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'n'; ++str; break;
137 clen = 1; 187 case '\014': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'f'; ++str; break;
138 } 188 case '\015': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'r'; ++str; break;
139 189
140 if (uch < 0x80 || enc->flags & F_ASCII) 190 default:
141 {
142 if (uch > 0xFFFFUL)
143 { 191 {
192 STRLEN clen;
193 UV uch;
194
195 if (is_utf8)
196 {
197 uch = decode_utf8 (str, end - str, &clen);
198 if (clen == (STRLEN)-1)
199 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
200 }
201 else
202 {
203 uch = ch;
204 clen = 1;
205 }
206
207 if (uch > 0x10FFFFUL)
208 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
209
210 if (uch < 0x80 || enc->json.flags & F_ASCII || (enc->json.flags & F_LATIN1 && uch > 0xFF))
211 {
212 if (uch > 0xFFFFUL)
213 {
144 need (enc, len += 11); 214 need (enc, len += 11);
145 sprintf (enc->cur, "\\u%04x\\u%04x", 215 sprintf (enc->cur, "\\u%04x\\u%04x",
146 (uch - 0x10000) / 0x400 + 0xD800, 216 (int)((uch - 0x10000) / 0x400 + 0xD800),
147 (uch - 0x10000) % 0x400 + 0xDC00); 217 (int)((uch - 0x10000) % 0x400 + 0xDC00));
148 enc->cur += 12; 218 enc->cur += 12;
219 }
220 else
221 {
222 static char hexdigit [16] = "0123456789abcdef";
223 need (enc, len += 5);
224 *enc->cur++ = '\\';
225 *enc->cur++ = 'u';
226 *enc->cur++ = hexdigit [ uch >> 12 ];
227 *enc->cur++ = hexdigit [(uch >> 8) & 15];
228 *enc->cur++ = hexdigit [(uch >> 4) & 15];
229 *enc->cur++ = hexdigit [(uch >> 0) & 15];
230 }
231
232 str += clen;
233 }
234 else if (enc->json.flags & F_LATIN1)
235 {
236 *enc->cur++ = uch;
237 str += clen;
238 }
239 else if (is_utf8)
240 {
241 need (enc, len += clen);
242 do
243 {
244 *enc->cur++ = *str++;
245 }
246 while (--clen);
247 }
248 else
249 {
250 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
251 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
252 ++str;
253 }
149 } 254 }
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 } 255 }
176 } 256 }
177 257
178 --len; 258 --len;
179 } 259 }
180} 260}
181 261
182#define INDENT SB \ 262inline void
263encode_indent (enc_t *enc)
264{
183 if (enc->flags & F_INDENT) \ 265 if (enc->json.flags & F_INDENT)
184 { \ 266 {
185 int i_; \ 267 int spaces = enc->indent * INDENT_STEP;
186 need (enc, enc->indent); \ 268
187 for (i_ = enc->indent * 3; i_--; )\ 269 need (enc, spaces);
270 memset (enc->cur, ' ', spaces);
271 enc->cur += spaces;
272 }
273}
274
275inline void
276encode_space (enc_t *enc)
277{
278 need (enc, 1);
279 encode_ch (enc, ' ');
280}
281
282inline void
283encode_nl (enc_t *enc)
284{
285 if (enc->json.flags & F_INDENT)
286 {
287 need (enc, 1);
188 encode_ch (enc, ' '); \ 288 encode_ch (enc, '\n');
189 } \ 289 }
190 SE 290}
191 291
192#define SPACE SB need (enc, 1); encode_ch (enc, ' '); SE 292inline void
193#define NL SB if (enc->flags & F_INDENT) { need (enc, 1); encode_ch (enc, '\n'); } SE 293encode_comma (enc_t *enc)
194#define COMMA SB \ 294{
195 encode_ch (enc, ','); \ 295 encode_ch (enc, ',');
296
196 if (enc->flags & F_INDENT) \ 297 if (enc->json.flags & F_INDENT)
197 NL; \ 298 encode_nl (enc);
198 else if (enc->flags & F_SPACE_AFTER) \ 299 else if (enc->json.flags & F_SPACE_AFTER)
199 SPACE; \ 300 encode_space (enc);
200 SE 301}
201 302
202static void encode_sv (enc_t *enc, SV *sv); 303static void encode_sv (enc_t *enc, SV *sv);
203 304
204static void 305static void
205encode_av (enc_t *enc, AV *av) 306encode_av (enc_t *enc, AV *av)
206{ 307{
207 int i, len = av_len (av); 308 int i, len = av_len (av);
208 309
209 encode_ch (enc, '['); NL; 310 if (enc->indent >= enc->maxdepth)
311 croak ("data structure too deep (hit recursion limit)");
312
313 encode_ch (enc, '['); encode_nl (enc);
210 ++enc->indent; 314 ++enc->indent;
211 315
212 for (i = 0; i <= len; ++i) 316 for (i = 0; i <= len; ++i)
213 { 317 {
214 INDENT; 318 SV **svp = av_fetch (av, i, 0);
215 encode_sv (enc, *av_fetch (av, i, 0)); 319
320 encode_indent (enc);
321
322 if (svp)
323 encode_sv (enc, *svp);
324 else
325 encode_str (enc, "null", 4, 0);
216 326
217 if (i < len) 327 if (i < len)
218 COMMA; 328 encode_comma (enc);
219 } 329 }
220 330
221 NL; 331 encode_nl (enc);
222 332
223 --enc->indent; 333 --enc->indent;
224 INDENT; encode_ch (enc, ']'); 334 encode_indent (enc); encode_ch (enc, ']');
225} 335}
226 336
227static void 337static void
228encode_he (enc_t *enc, HE *he) 338encode_he (enc_t *enc, HE *he)
229{ 339{
… …
243 else 353 else
244 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 354 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
245 355
246 encode_ch (enc, '"'); 356 encode_ch (enc, '"');
247 357
248 if (enc->flags & F_SPACE_BEFORE) SPACE; 358 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
249 encode_ch (enc, ':'); 359 encode_ch (enc, ':');
250 if (enc->flags & F_SPACE_AFTER ) SPACE; 360 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
251 encode_sv (enc, HeVAL (he)); 361 encode_sv (enc, HeVAL (he));
252} 362}
253 363
254// compare hash entries, used when all keys are bytestrings 364// compare hash entries, used when all keys are bytestrings
255static int 365static int
… …
261 HE *b = *(HE **)b_; 371 HE *b = *(HE **)b_;
262 372
263 STRLEN la = HeKLEN (a); 373 STRLEN la = HeKLEN (a);
264 STRLEN lb = HeKLEN (b); 374 STRLEN lb = HeKLEN (b);
265 375
266 if (!(cmp == memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 376 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb)))
267 cmp = la < lb ? -1 : la == lb ? 0 : 1; 377 cmp = la - lb;
268 378
269 return cmp; 379 return cmp;
270} 380}
271 381
272// compare hash entries, used when some keys are sv's or utf-x 382// compare hash entries, used when some keys are sv's or utf-x
… …
279static void 389static void
280encode_hv (enc_t *enc, HV *hv) 390encode_hv (enc_t *enc, HV *hv)
281{ 391{
282 int count, i; 392 int count, i;
283 393
394 if (enc->indent >= enc->maxdepth)
395 croak ("data structure too deep (hit recursion limit)");
396
284 encode_ch (enc, '{'); NL; ++enc->indent; 397 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
285 398
286 if ((count = hv_iterinit (hv))) 399 if ((count = hv_iterinit (hv)))
287 { 400 {
288 // for canonical output we have to sort by keys first 401 // for canonical output we have to sort by keys first
289 // actually, this is mostly due to the stupid so-called 402 // actually, this is mostly due to the stupid so-called
290 // security workaround added somewhere in 5.8.x. 403 // security workaround added somewhere in 5.8.x.
291 // that randomises hash orderings 404 // that randomises hash orderings
292 if (enc->flags & F_CANONICAL) 405 if (enc->json.flags & F_CANONICAL)
293 { 406 {
294 HE *he, *hes [count];
295 int fast = 1; 407 int fast = 1;
408 HE *he;
409#if defined(__BORLANDC__) || defined(_MSC_VER)
410 HE **hes = _alloca (count * sizeof (HE));
411#else
412 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
413#endif
296 414
297 i = 0; 415 i = 0;
298 while ((he = hv_iternext (hv))) 416 while ((he = hv_iternext (hv)))
299 { 417 {
300 hes [i++] = he; 418 hes [i++] = he;
… …
306 424
307 if (fast) 425 if (fast)
308 qsort (hes, count, sizeof (HE *), he_cmp_fast); 426 qsort (hes, count, sizeof (HE *), he_cmp_fast);
309 else 427 else
310 { 428 {
311 // hack to disable "use bytes" 429 // hack to forcefully disable "use bytes"
312 COP *oldcop = PL_curcop, cop; 430 COP cop = *PL_curcop;
313 cop.op_private = 0; 431 cop.op_private = 0;
432
433 ENTER;
434 SAVETMPS;
435
436 SAVEVPTR (PL_curcop);
314 PL_curcop = &cop; 437 PL_curcop = &cop;
315 438
316 SAVETMPS;
317 qsort (hes, count, sizeof (HE *), he_cmp_slow); 439 qsort (hes, count, sizeof (HE *), he_cmp_slow);
440
318 FREETMPS; 441 FREETMPS;
319 442 LEAVE;
320 PL_curcop = oldcop;
321 } 443 }
322 444
323 for (i = 0; i < count; ++i) 445 for (i = 0; i < count; ++i)
324 { 446 {
325 INDENT; 447 encode_indent (enc);
326 encode_he (enc, hes [i]); 448 encode_he (enc, hes [i]);
327 449
328 if (i < count - 1) 450 if (i < count - 1)
329 COMMA; 451 encode_comma (enc);
330 }
331
332 NL; 452 }
453
454 encode_nl (enc);
333 } 455 }
334 else 456 else
335 { 457 {
336 SV *sv;
337 HE *he = hv_iternext (hv); 458 HE *he = hv_iternext (hv);
338 459
339 for (;;) 460 for (;;)
340 { 461 {
341 INDENT; 462 encode_indent (enc);
342 encode_he (enc, he); 463 encode_he (enc, he);
343 464
344 if (!(he = hv_iternext (hv))) 465 if (!(he = hv_iternext (hv)))
345 break; 466 break;
346 467
347 COMMA; 468 encode_comma (enc);
348 }
349
350 NL; 469 }
470
471 encode_nl (enc);
351 } 472 }
352 } 473 }
353 474
354 --enc->indent; INDENT; encode_ch (enc, '}'); 475 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
476}
477
478// encode objects, arrays and special \0=false and \1=true values.
479static void
480encode_rv (enc_t *enc, SV *sv)
481{
482 svtype svt;
483
484 SvGETMAGIC (sv);
485 svt = SvTYPE (sv);
486
487 if (expect_false (SvOBJECT (sv)))
488 {
489 HV *stash = !JSON_SLOW || json_boolean_stash
490 ? json_boolean_stash
491 : gv_stashpv ("JSON::XS::Boolean", 1);
492
493 if (SvSTASH (sv) == stash)
494 {
495 if (SvIV (sv))
496 encode_str (enc, "true", 4, 0);
497 else
498 encode_str (enc, "false", 5, 0);
499 }
500 else
501 {
502#if 0
503 if (0 && sv_derived_from (rv, "JSON::Literal"))
504 {
505 // not yet
506 }
507#endif
508 if (enc->json.flags & F_CONV_BLESSED)
509 {
510 // we re-bless the reference to get overload and other niceties right
511 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
512
513 if (to_json)
514 {
515 dSP;
516
517 ENTER; SAVETMPS; PUSHMARK (SP);
518 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
519
520 // calling with G_SCALAR ensures that we always get a 1 return value
521 PUTBACK;
522 call_sv ((SV *)GvCV (to_json), G_SCALAR);
523 SPAGAIN;
524
525 // catch this surprisingly common error
526 if (SvROK (TOPs) && SvRV (TOPs) == sv)
527 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
528
529 sv = POPs;
530 PUTBACK;
531
532 encode_sv (enc, sv);
533
534 FREETMPS; LEAVE;
535 }
536 else if (enc->json.flags & F_ALLOW_BLESSED)
537 encode_str (enc, "null", 4, 0);
538 else
539 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
540 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
541 }
542 else if (enc->json.flags & F_ALLOW_BLESSED)
543 encode_str (enc, "null", 4, 0);
544 else
545 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",
546 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
547 }
548 }
549 else if (svt == SVt_PVHV)
550 encode_hv (enc, (HV *)sv);
551 else if (svt == SVt_PVAV)
552 encode_av (enc, (AV *)sv);
553 else if (svt < SVt_PVAV)
554 {
555 STRLEN len = 0;
556 char *pv = svt ? SvPV (sv, len) : 0;
557
558 if (len == 1 && *pv == '1')
559 encode_str (enc, "true", 4, 0);
560 else if (len == 1 && *pv == '0')
561 encode_str (enc, "false", 5, 0);
562 else
563 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
564 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
565 }
566 else
567 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
568 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
355} 569}
356 570
357static void 571static void
358encode_sv (enc_t *enc, SV *sv) 572encode_sv (enc_t *enc, SV *sv)
359{ 573{
… …
367 encode_str (enc, str, len, SvUTF8 (sv)); 581 encode_str (enc, str, len, SvUTF8 (sv));
368 encode_ch (enc, '"'); 582 encode_ch (enc, '"');
369 } 583 }
370 else if (SvNOKp (sv)) 584 else if (SvNOKp (sv))
371 { 585 {
586 // trust that perl will do the right thing w.r.t. JSON syntax.
372 need (enc, NV_DIG + 32); 587 need (enc, NV_DIG + 32);
373 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 588 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
374 enc->cur += strlen (enc->cur); 589 enc->cur += strlen (enc->cur);
375 } 590 }
376 else if (SvIOKp (sv)) 591 else if (SvIOKp (sv))
377 { 592 {
378 need (enc, 64); 593 // we assume we can always read an IV as a UV
594 if (SvUV (sv) & ~(UV)0x7fff)
595 {
596 // large integer, use the (rather slow) snprintf way.
597 need (enc, sizeof (UV) * 3);
379 enc->cur += 598 enc->cur +=
380 SvIsUV(sv) 599 SvIsUV(sv)
381 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 600 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
382 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 601 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
602 }
603 else
604 {
605 // optimise the "small number case"
606 // code will likely be branchless and use only a single multiplication
607 I32 i = SvIV (sv);
608 U32 u;
609 char digit, nz = 0;
610
611 need (enc, 6);
612
613 *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
614 u = i < 0 ? -i : i;
615
616 // convert to 4.28 fixed-point representation
617 u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
618
619 // now output digit by digit, each time masking out the integer part
620 // and multiplying by 5 while moving the decimal point one to the right,
621 // resulting in a net multiplication by 10.
622 // we always write the digit to memory but conditionally increment
623 // the pointer, to ease the usage of conditional move instructions.
624 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5;
625 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5;
626 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5;
627 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5;
628 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
629 }
383 } 630 }
384 else if (SvROK (sv)) 631 else if (SvROK (sv))
385 { 632 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)) 633 else if (!SvOK (sv))
401 encode_str (enc, "null", 4, 0); 634 encode_str (enc, "null", 4, 0);
402 else 635 else
403 croak ("encountered perl type that JSON cannot handle"); 636 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
637 SvPV_nolen (sv), SvFLAGS (sv));
404} 638}
405 639
406static SV * 640static SV *
407encode_json (SV *scalar, UV flags) 641encode_json (SV *scalar, JSON *json)
408{ 642{
409 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar))
410 croak ("hash- or arraref required (not a simple scalar, use allow_nonref to allow this)");
411
412 enc_t enc; 643 enc_t enc;
413 enc.flags = flags; 644
645 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
646 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
647
648 enc.json = *json;
414 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 649 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
415 enc.cur = SvPVX (enc.sv); 650 enc.cur = SvPVX (enc.sv);
416 enc.end = SvEND (enc.sv); 651 enc.end = SvEND (enc.sv);
417 enc.max_recurse = 0;
418 enc.indent = 0; 652 enc.indent = 0;
653 enc.maxdepth = DEC_DEPTH (enc.json.flags);
419 654
420 SvPOK_only (enc.sv); 655 SvPOK_only (enc.sv);
421 encode_sv (&enc, scalar); 656 encode_sv (&enc, scalar);
422 657
658 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
659 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
660
423 if (!(flags & (F_ASCII | F_UTF8))) 661 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
424 SvUTF8_on (enc.sv); 662 SvUTF8_on (enc.sv);
425 663
426 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 664 if (enc.json.flags & F_SHRINK)
665 shrink (enc.sv);
666
427 return enc.sv; 667 return enc.sv;
428} 668}
429 669
430///////////////////////////////////////////////////////////////////////////// 670/////////////////////////////////////////////////////////////////////////////
671// decoder
431 672
432#define WS \ 673// structure used for decoding JSON
674typedef struct
675{
676 char *cur; // current parser pointer
677 char *end; // end of input string
678 const char *err; // parse error, if != 0
679 JSON json;
680 U32 depth; // recursion depth
681 U32 maxdepth; // recursion depth limit
682} dec_t;
683
684inline void
685decode_ws (dec_t *dec)
686{
433 for (;;) \ 687 for (;;)
434 { \ 688 {
435 char ch = *dec->cur; \ 689 char ch = *dec->cur;
690
436 if (ch > 0x20 \ 691 if (ch > 0x20
437 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) \ 692 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09))
438 break; \ 693 break;
694
439 ++dec->cur; \ 695 ++dec->cur;
440 } 696 }
697}
441 698
442#define ERR(reason) SB dec->err = reason; goto fail; SE 699#define ERR(reason) SB dec->err = reason; goto fail; SE
700
443#define EXPECT_CH(ch) SB \ 701#define EXPECT_CH(ch) SB \
444 if (*dec->cur != ch) \ 702 if (*dec->cur != ch) \
445 ERR (# ch " expected"); \ 703 ERR (# ch " expected"); \
446 ++dec->cur; \ 704 ++dec->cur; \
447 SE 705 SE
448 706
707#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)")
708#define DEC_DEC_DEPTH --dec->depth
709
449static SV *decode_sv (dec_t *dec); 710static SV *decode_sv (dec_t *dec);
450 711
451static signed char decode_hexdigit[256]; 712static signed char decode_hexdigit[256];
452 713
453static UV 714static UV
454decode_4hex (dec_t *dec) 715decode_4hex (dec_t *dec)
455{ 716{
456 signed char d1, d2, d3, d4; 717 signed char d1, d2, d3, d4;
718 unsigned char *cur = (unsigned char *)dec->cur;
457 719
458 d1 = decode_hexdigit [((unsigned char *)dec->cur) [0]]; 720 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"); 721 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]]; 722 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"); 723 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");
466 724
467 dec->cur += 4; 725 dec->cur += 4;
468 726
469 return ((UV)d1) << 12 727 return ((UV)d1) << 12
470 | ((UV)d2) << 8 728 | ((UV)d2) << 8
… …
473 731
474fail: 732fail:
475 return (UV)-1; 733 return (UV)-1;
476} 734}
477 735
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 * 736static SV *
494decode_str (dec_t *dec) 737decode_str (dec_t *dec)
495{ 738{
496 SV *sv = NEWSV (0,2); 739 SV *sv = 0;
497 int utf8 = 0; 740 int utf8 = 0;
498 char *cur = SvPVX (sv); 741 char *dec_cur = dec->cur;
499 char *end = SvEND (sv);
500 742
501 for (;;) 743 do
502 { 744 {
503 unsigned char ch = *(unsigned char *)dec->cur; 745 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
746 char *cur = buf;
504 747
505 if (ch == '"') 748 do
506 break;
507 else if (ch == '\\')
508 { 749 {
509 switch (*++dec->cur) 750 unsigned char ch = *(unsigned char *)dec_cur++;
751
752 if (expect_false (ch == '"'))
753 {
754 --dec_cur;
755 break;
510 { 756 }
511 case '\\': 757 else if (expect_false (ch == '\\'))
512 case '/': 758 {
513 case '"': APPEND_CH (*dec->cur++); break; 759 switch (*dec_cur)
514
515 case 'b': APPEND_CH ('\010'); ++dec->cur; break;
516 case 't': APPEND_CH ('\011'); ++dec->cur; break;
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 { 760 {
523 UV lo, hi; 761 case '\\':
524 ++dec->cur; 762 case '/':
763 case '"': *cur++ = *dec_cur++; break;
525 764
526 hi = decode_4hex (dec); 765 case 'b': ++dec_cur; *cur++ = '\010'; break;
527 if (hi == (UV)-1) 766 case 't': ++dec_cur; *cur++ = '\011'; break;
528 goto fail; 767 case 'n': ++dec_cur; *cur++ = '\012'; break;
768 case 'f': ++dec_cur; *cur++ = '\014'; break;
769 case 'r': ++dec_cur; *cur++ = '\015'; break;
529 770
530 // possibly a surrogate pair 771 case 'u':
531 if (hi >= 0xd800 && hi < 0xdc00)
532 { 772 {
533 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 773 UV lo, hi;
534 ERR ("illegal surrogate character"); 774 ++dec_cur;
535 775
536 dec->cur += 2; 776 dec->cur = dec_cur;
537
538 lo = decode_4hex (dec); 777 hi = decode_4hex (dec);
778 dec_cur = dec->cur;
539 if (lo == (UV)-1) 779 if (hi == (UV)-1)
540 goto fail; 780 goto fail;
541 781
782 // possibly a surrogate pair
783 if (hi >= 0xd800)
784 if (hi < 0xdc00)
785 {
786 if (dec_cur [0] != '\\' || dec_cur [1] != 'u')
787 ERR ("missing low surrogate character in surrogate pair");
788
789 dec_cur += 2;
790
791 dec->cur = dec_cur;
792 lo = decode_4hex (dec);
793 dec_cur = dec->cur;
794 if (lo == (UV)-1)
795 goto fail;
796
542 if (lo < 0xdc00 || lo >= 0xe000) 797 if (lo < 0xdc00 || lo >= 0xe000)
543 ERR ("surrogate pair expected"); 798 ERR ("surrogate pair expected");
544 799
545 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000; 800 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000;
801 }
802 else if (hi < 0xe000)
803 ERR ("missing high surrogate character in surrogate pair");
804
805 if (hi >= 0x80)
806 {
807 utf8 = 1;
808
809 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
810 }
811 else
812 *cur++ = hi;
546 } 813 }
547 else if (lo >= 0xdc00 && lo < 0xe000)
548 ERR ("illegal surrogate character");
549
550 if (hi >= 0x80)
551 { 814 break;
552 utf8 = 1;
553 815
554 APPEND_GROW (4); // at most 4 bytes for 21 bits
555 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
556 }
557 else 816 default:
558 APPEND_CH (hi); 817 --dec_cur;
818 ERR ("illegal backslash escape sequence in string");
559 } 819 }
820 }
821 else if (expect_true (ch >= 0x20 && ch <= 0x7f))
822 *cur++ = ch;
823 else if (ch >= 0x80)
824 {
825 STRLEN clen;
826 UV uch;
827
828 --dec_cur;
829
830 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
831 if (clen == (STRLEN)-1)
832 ERR ("malformed UTF-8 character in JSON string");
833
834 do
835 *cur++ = *dec_cur++;
836 while (--clen);
837
838 utf8 = 1;
839 }
840 else
841 {
842 --dec_cur;
843
844 if (!ch)
845 ERR ("unexpected end of string while parsing JSON string");
560 break; 846 else
847 ERR ("invalid character encountered while parsing JSON string");
561 } 848 }
562 } 849 }
563 else if (ch >= 0x20 && ch <= 0x7f) 850 while (cur < buf + SHORT_STRING_LEN);
564 APPEND_CH (*dec->cur++); 851
565 else if (ch >= 0x80) 852 {
853 STRLEN len = cur - buf;
854
855 if (sv)
566 { 856 {
567 STRLEN clen; 857 SvGROW (sv, SvCUR (sv) + len + 1);
568 UV uch = utf8n_to_uvuni (dec->cur, dec->end - dec->cur, &clen, UTF8_CHECK_ONLY); 858 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
569 if (clen < 0) 859 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 } 860 }
577 else 861 else
578 ERR ("invalid character encountered"); 862 sv = newSVpvn (buf, len);
579 } 863 }
864 }
865 while (*dec_cur != '"');
580 866
581 ++dec->cur; 867 ++dec_cur;
582 868
583 SvCUR_set (sv, cur - SvPVX (sv)); 869 if (sv)
584 870 {
585 SvPOK_only (sv); 871 SvPOK_only (sv);
586 *SvEND (sv) = 0; 872 *SvEND (sv) = 0;
587 873
588 if (utf8) 874 if (utf8)
589 SvUTF8_on (sv); 875 SvUTF8_on (sv);
876 }
877 else
878 sv = newSVpvn ("", 0);
590 879
880 dec->cur = dec_cur;
591 return sv; 881 return sv;
592 882
593fail: 883fail:
594 SvREFCNT_dec (sv); 884 dec->cur = dec_cur;
595 return 0; 885 return 0;
596} 886}
597 887
598static SV * 888static SV *
599decode_num (dec_t *dec) 889decode_num (dec_t *dec)
… …
609 { 899 {
610 ++dec->cur; 900 ++dec->cur;
611 if (*dec->cur >= '0' && *dec->cur <= '9') 901 if (*dec->cur >= '0' && *dec->cur <= '9')
612 ERR ("malformed number (leading zero must not be followed by another digit)"); 902 ERR ("malformed number (leading zero must not be followed by another digit)");
613 } 903 }
614 904 else if (*dec->cur < '0' || *dec->cur > '9')
615 // int 905 ERR ("malformed number (no digits after initial minus)");
906 else
907 do
908 {
909 ++dec->cur;
910 }
616 while (*dec->cur >= '0' && *dec->cur <= '9') 911 while (*dec->cur >= '0' && *dec->cur <= '9');
617 ++dec->cur;
618 912
619 // [frac] 913 // [frac]
620 if (*dec->cur == '.') 914 if (*dec->cur == '.')
621 { 915 {
622 is_nv = 1; 916 ++dec->cur;
917
918 if (*dec->cur < '0' || *dec->cur > '9')
919 ERR ("malformed number (no digits after decimal point)");
623 920
624 do 921 do
625 { 922 {
626 ++dec->cur; 923 ++dec->cur;
627 } 924 }
628 while (*dec->cur >= '0' && *dec->cur <= '9'); 925 while (*dec->cur >= '0' && *dec->cur <= '9');
926
927 is_nv = 1;
629 } 928 }
630 929
631 // [exp] 930 // [exp]
632 if (*dec->cur == 'e' || *dec->cur == 'E') 931 if (*dec->cur == 'e' || *dec->cur == 'E')
633 { 932 {
634 is_nv = 1;
635
636 ++dec->cur; 933 ++dec->cur;
934
637 if (*dec->cur == '-' || *dec->cur == '+') 935 if (*dec->cur == '-' || *dec->cur == '+')
638 ++dec->cur; 936 ++dec->cur;
639 937
938 if (*dec->cur < '0' || *dec->cur > '9')
939 ERR ("malformed number (no digits after exp sign)");
940
941 do
942 {
943 ++dec->cur;
944 }
640 while (*dec->cur >= '0' && *dec->cur <= '9') 945 while (*dec->cur >= '0' && *dec->cur <= '9');
641 ++dec->cur; 946
947 is_nv = 1;
642 } 948 }
643 949
644 if (!is_nv) 950 if (!is_nv)
645 { 951 {
646 UV uv; 952 int len = dec->cur - start;
647 int numtype = grok_number (start, dec->cur - start, &uv); 953
648 if (numtype & IS_NUMBER_IN_UV) 954 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so
649 if (numtype & IS_NUMBER_NEG) 955 if (*start == '-')
956 switch (len)
650 { 957 {
651 if (uv < (UV)IV_MIN) 958 case 2: return newSViv (-( start [1] - '0' * 1));
652 return newSViv (-(IV)uv); 959 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
960 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
961 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
653 } 962 }
963 else
964 switch (len)
965 {
966 case 1: return newSViv ( start [0] - '0' * 1);
967 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
968 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
969 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
970 }
971
972 {
973 UV uv;
974 int numtype = grok_number (start, len, &uv);
975 if (numtype & IS_NUMBER_IN_UV)
976 if (numtype & IS_NUMBER_NEG)
977 {
978 if (uv < (UV)IV_MIN)
979 return newSViv (-(IV)uv);
980 }
654 else 981 else
655 return newSVuv (uv); 982 return newSVuv (uv);
656 } 983 }
657 984
985 len -= *start == '-' ? 1 : 0;
986
987 // does not fit into IV or UV, try NV
988 if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len)
989 #if defined (LDBL_DIG)
990 || (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len)
991 #endif
992 )
993 // fits into NV without loss of precision
994 return newSVnv (Atof (start));
995
996 // everything else fails, convert it to a string
997 return newSVpvn (start, dec->cur - start);
998 }
999
1000 // loss of precision here
658 return newSVnv (Atof (start)); 1001 return newSVnv (Atof (start));
659 1002
660fail: 1003fail:
661 return 0; 1004 return 0;
662} 1005}
… …
664static SV * 1007static SV *
665decode_av (dec_t *dec) 1008decode_av (dec_t *dec)
666{ 1009{
667 AV *av = newAV (); 1010 AV *av = newAV ();
668 1011
1012 DEC_INC_DEPTH;
1013 decode_ws (dec);
1014
1015 if (*dec->cur == ']')
1016 ++dec->cur;
1017 else
669 for (;;) 1018 for (;;)
670 { 1019 {
671 SV *value; 1020 SV *value;
672 1021
673 value = decode_sv (dec); 1022 value = decode_sv (dec);
674 if (!value) 1023 if (!value)
675 goto fail; 1024 goto fail;
676 1025
677 av_push (av, value); 1026 av_push (av, value);
678 1027
679 WS; 1028 decode_ws (dec);
680 1029
681 if (*dec->cur == ']') 1030 if (*dec->cur == ']')
682 { 1031 {
683 ++dec->cur; 1032 ++dec->cur;
684 break; 1033 break;
1034 }
685 } 1035
686
687 if (*dec->cur != ',') 1036 if (*dec->cur != ',')
688 ERR (", or ] expected while parsing array"); 1037 ERR (", or ] expected while parsing array");
689 1038
690 ++dec->cur; 1039 ++dec->cur;
691 } 1040 }
692 1041
1042 DEC_DEC_DEPTH;
693 return newRV_noinc ((SV *)av); 1043 return newRV_noinc ((SV *)av);
694 1044
695fail: 1045fail:
696 SvREFCNT_dec (av); 1046 SvREFCNT_dec (av);
1047 DEC_DEC_DEPTH;
697 return 0; 1048 return 0;
698} 1049}
699 1050
700static SV * 1051static SV *
701decode_hv (dec_t *dec) 1052decode_hv (dec_t *dec)
702{ 1053{
1054 SV *sv;
703 HV *hv = newHV (); 1055 HV *hv = newHV ();
704 1056
1057 DEC_INC_DEPTH;
1058 decode_ws (dec);
1059
1060 if (*dec->cur == '}')
1061 ++dec->cur;
1062 else
705 for (;;) 1063 for (;;)
706 { 1064 {
707 SV *key, *value; 1065 decode_ws (dec); EXPECT_CH ('"');
708 1066
709 WS; EXPECT_CH ('"'); 1067 // heuristic: assume that
710 1068 // a) decode_str + hv_store_ent are abysmally slow.
711 key = decode_str (dec); 1069 // b) most hash keys are short, simple ascii text.
712 if (!key) 1070 // => try to "fast-match" such strings to avoid
713 goto fail; 1071 // the overhead of decode_str + hv_store_ent.
714
715 WS; EXPECT_CH (':');
716
717 value = decode_sv (dec);
718 if (!value)
719 { 1072 {
1073 SV *value;
1074 char *p = dec->cur;
1075 char *e = p + 24; // only try up to 24 bytes
1076
1077 for (;;)
1078 {
1079 // the >= 0x80 is true on most architectures
1080 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1081 {
1082 // slow path, back up and use decode_str
1083 SV *key = decode_str (dec);
1084 if (!key)
1085 goto fail;
1086
1087 decode_ws (dec); EXPECT_CH (':');
1088
1089 value = decode_sv (dec);
1090 if (!value)
1091 {
1092 SvREFCNT_dec (key);
1093 goto fail;
1094 }
1095
1096 hv_store_ent (hv, key, value, 0);
720 SvREFCNT_dec (key); 1097 SvREFCNT_dec (key);
1098
1099 break;
1100 }
1101 else if (*p == '"')
1102 {
1103 // fast path, got a simple key
1104 char *key = dec->cur;
1105 int len = p - key;
1106 dec->cur = p + 1;
1107
1108 decode_ws (dec); EXPECT_CH (':');
1109
1110 value = decode_sv (dec);
1111 if (!value)
721 goto fail; 1112 goto fail;
1113
1114 hv_store (hv, key, len, value, 0);
1115
1116 break;
1117 }
1118
1119 ++p;
1120 }
722 } 1121 }
723 1122
724 //TODO: optimise 1123 decode_ws (dec);
725 hv_store_ent (hv, key, value, 0);
726 1124
727 WS;
728
729 if (*dec->cur == '}') 1125 if (*dec->cur == '}')
1126 {
1127 ++dec->cur;
1128 break;
1129 }
1130
1131 if (*dec->cur != ',')
1132 ERR (", or } expected while parsing object/hash");
1133
1134 ++dec->cur;
1135 }
1136
1137 DEC_DEC_DEPTH;
1138 sv = newRV_noinc ((SV *)hv);
1139
1140 // check filter callbacks
1141 if (dec->json.flags & F_HOOK)
1142 {
1143 if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
730 { 1144 {
731 ++dec->cur; 1145 HE *cb, *he;
732 break; 1146
1147 hv_iterinit (hv);
1148 he = hv_iternext (hv);
1149 hv_iterinit (hv);
1150
1151 // the next line creates a mortal sv each time its called.
1152 // might want to optimise this for common cases.
1153 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1154
1155 if (cb)
1156 {
1157 dSP;
1158 int count;
1159
1160 ENTER; SAVETMPS; PUSHMARK (SP);
1161 XPUSHs (HeVAL (he));
1162
1163 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1164
1165 if (count == 1)
1166 {
1167 sv = newSVsv (POPs);
1168 FREETMPS; LEAVE;
1169 return sv;
1170 }
1171
1172 FREETMPS; LEAVE;
1173 }
733 } 1174 }
734 1175
735 if (*dec->cur != ',') 1176 if (dec->json.cb_object)
736 ERR (", or } expected while parsing object/hash"); 1177 {
1178 dSP;
1179 int count;
737 1180
738 ++dec->cur; 1181 ENTER; SAVETMPS; PUSHMARK (SP);
739 } 1182 XPUSHs (sv_2mortal (sv));
740 1183
741 return newRV_noinc ((SV *)hv); 1184 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1185
1186 if (count == 1)
1187 {
1188 sv = newSVsv (POPs);
1189 FREETMPS; LEAVE;
1190 return sv;
1191 }
1192
1193 SvREFCNT_inc (sv);
1194 FREETMPS; LEAVE;
1195 }
1196 }
1197
1198 return sv;
742 1199
743fail: 1200fail:
744 SvREFCNT_dec (hv); 1201 SvREFCNT_dec (hv);
1202 DEC_DEC_DEPTH;
745 return 0; 1203 return 0;
746} 1204}
747 1205
748static SV * 1206static SV *
749decode_sv (dec_t *dec) 1207decode_sv (dec_t *dec)
750{ 1208{
751 WS; 1209 decode_ws (dec);
1210
1211 // the beauty of JSON: you need exactly one character lookahead
1212 // to parse anything.
752 switch (*dec->cur) 1213 switch (*dec->cur)
753 { 1214 {
754 case '"': ++dec->cur; return decode_str (dec); 1215 case '"': ++dec->cur; return decode_str (dec);
755 case '[': ++dec->cur; return decode_av (dec); 1216 case '[': ++dec->cur; return decode_av (dec);
756 case '{': ++dec->cur; return decode_hv (dec); 1217 case '{': ++dec->cur; return decode_hv (dec);
… …
762 1223
763 case 't': 1224 case 't':
764 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1225 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
765 { 1226 {
766 dec->cur += 4; 1227 dec->cur += 4;
767 return newSViv (1); 1228 return SvREFCNT_inc (json_true);
768 } 1229 }
769 else 1230 else
770 ERR ("'true' expected"); 1231 ERR ("'true' expected");
771 1232
772 break; 1233 break;
773 1234
774 case 'f': 1235 case 'f':
775 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1236 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
776 { 1237 {
777 dec->cur += 5; 1238 dec->cur += 5;
778 return newSViv (0); 1239 return SvREFCNT_inc (json_false);
779 } 1240 }
780 else 1241 else
781 ERR ("'false' expected"); 1242 ERR ("'false' expected");
782 1243
783 break; 1244 break;
784 1245
785 case 'n': 1246 case 'n':
786 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4)) 1247 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4))
787 { 1248 {
788 dec->cur += 4; 1249 dec->cur += 4;
789 return newSViv (1); 1250 return newSVsv (&PL_sv_undef);
790 } 1251 }
791 else 1252 else
792 ERR ("'null' expected"); 1253 ERR ("'null' expected");
793 1254
794 break; 1255 break;
795 1256
796 default: 1257 default:
797 ERR ("malformed json string"); 1258 ERR ("malformed JSON string, neither array, object, number, string or atom");
798 break; 1259 break;
799 } 1260 }
800 1261
801fail: 1262fail:
802 return 0; 1263 return 0;
803} 1264}
804 1265
805static SV * 1266static SV *
806decode_json (SV *string, UV flags) 1267decode_json (SV *string, JSON *json, UV *offset_return)
807{ 1268{
1269 dec_t dec;
1270 UV offset;
808 SV *sv; 1271 SV *sv;
809 1272
1273 SvGETMAGIC (string);
1274 SvUPGRADE (string, SVt_PV);
1275
1276 if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags))
1277 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1278 (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags));
1279
810 if (!(flags & F_UTF8)) 1280 if (json->flags & F_UTF8)
1281 sv_utf8_downgrade (string, 0);
1282 else
811 sv_utf8_upgrade (string); 1283 sv_utf8_upgrade (string);
812 1284
813 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1285 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
814 1286
815 dec_t dec; 1287 dec.json = *json;
816 dec.flags = flags;
817 dec.cur = SvPVX (string); 1288 dec.cur = SvPVX (string);
818 dec.end = SvEND (string); 1289 dec.end = SvEND (string);
819 dec.err = 0; 1290 dec.err = 0;
1291 dec.depth = 0;
1292 dec.maxdepth = DEC_DEPTH (dec.json.flags);
820 1293
821 *dec.end = 1; // invalid anywhere 1294 if (dec.json.cb_object || dec.json.cb_sk_object)
1295 dec.json.flags |= F_HOOK;
1296
1297 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
822 sv = decode_sv (&dec); 1298 sv = decode_sv (&dec);
823 *dec.end = 0; 1299
1300 if (!(offset_return || !sv))
1301 {
1302 // check for trailing garbage
1303 decode_ws (&dec);
1304
1305 if (*dec.cur)
1306 {
1307 dec.err = "garbage after JSON object";
1308 SvREFCNT_dec (sv);
1309 sv = 0;
1310 }
1311 }
1312
1313 if (offset_return || !sv)
1314 {
1315 offset = dec.json.flags & F_UTF8
1316 ? dec.cur - SvPVX (string)
1317 : utf8_distance (dec.cur, SvPVX (string));
1318
1319 if (offset_return)
1320 *offset_return = offset;
1321 }
824 1322
825 if (!sv) 1323 if (!sv)
826 { 1324 {
827 IV offset = utf8_distance (dec.cur, SvPVX (string));
828 SV *uni = sv_newmortal (); 1325 SV *uni = sv_newmortal ();
829 1326
1327 // horrible hack to silence warning inside pv_uni_display
1328 COP cop = *PL_curcop;
1329 cop.cop_warnings = pWARN_NONE;
1330 ENTER;
1331 SAVEVPTR (PL_curcop);
1332 PL_curcop = &cop;
830 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1333 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1334 LEAVE;
1335
831 croak ("%s, at character %d (%s)", 1336 croak ("%s, at character offset %d [\"%s\"]",
832 dec.err, 1337 dec.err,
833 (int)offset, 1338 (int)offset,
834 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1339 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
835 } 1340 }
836 1341
837 sv = sv_2mortal (sv); 1342 sv = sv_2mortal (sv);
838 1343
839 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1344 if (!(dec.json.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)"); 1345 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
841 1346
842 return sv; 1347 return sv;
843} 1348}
844 1349
1350/////////////////////////////////////////////////////////////////////////////
1351// XS interface functions
1352
845MODULE = JSON::XS PACKAGE = JSON::XS 1353MODULE = JSON::XS PACKAGE = JSON::XS
846 1354
847BOOT: 1355BOOT:
848{ 1356{
849 int i; 1357 int i;
850 1358
851 memset (decode_hexdigit, 0xff, 256);
852 for (i = 10; i--; ) 1359 for (i = 0; i < 256; ++i)
853 decode_hexdigit ['0' + i] = i; 1360 decode_hexdigit [i] =
1361 i >= '0' && i <= '9' ? i - '0'
1362 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1363 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1364 : -1;
854 1365
855 for (i = 7; i--; ) 1366 json_stash = gv_stashpv ("JSON::XS" , 1);
1367 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1368
1369 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true );
1370 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
1371}
1372
1373PROTOTYPES: DISABLE
1374
1375void CLONE (...)
1376 CODE:
1377 json_stash = 0;
1378 json_boolean_stash = 0;
1379
1380void new (char *klass)
1381 PPCODE:
1382{
1383 HV *stash = !JSON_SLOW || json_stash
1384 ? json_stash
1385 : gv_stashpv ("JSON::XS", 1);
1386 SV *pv = NEWSV (0, sizeof (JSON));
1387 SvPOK_only (pv);
1388 Zero (SvPVX (pv), 1, JSON);
1389 ((JSON *)SvPVX (pv))->flags = F_DEFAULT;
1390 XPUSHs (sv_2mortal (sv_bless (newRV_noinc (pv), stash)));
1391}
1392
1393void ascii (JSON *self, int enable = 1)
1394 ALIAS:
1395 ascii = F_ASCII
1396 latin1 = F_LATIN1
1397 utf8 = F_UTF8
1398 indent = F_INDENT
1399 canonical = F_CANONICAL
1400 space_before = F_SPACE_BEFORE
1401 space_after = F_SPACE_AFTER
1402 pretty = F_PRETTY
1403 allow_nonref = F_ALLOW_NONREF
1404 shrink = F_SHRINK
1405 allow_blessed = F_ALLOW_BLESSED
1406 convert_blessed = F_CONV_BLESSED
1407 PPCODE:
1408{
1409 if (enable)
1410 self->flags |= ix;
1411 else
1412 self->flags &= ~ix;
1413
1414 XPUSHs (ST (0));
1415}
1416
1417void max_depth (JSON *self, UV max_depth = 0x80000000UL)
1418 PPCODE:
1419{
1420 UV log2 = 0;
1421
1422 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1423
1424 while ((1UL << log2) < max_depth)
1425 ++log2;
1426
1427 self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1428
1429 XPUSHs (ST (0));
1430}
1431
1432void max_size (JSON *self, UV max_size = 0)
1433 PPCODE:
1434{
1435 UV log2 = 0;
1436
1437 if (max_size > 0x80000000UL) max_size = 0x80000000UL;
1438 if (max_size == 1) max_size = 2;
1439
1440 while ((1UL << log2) < max_size)
1441 ++log2;
1442
1443 self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1444
1445 XPUSHs (ST (0));
1446}
1447
1448void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1449 PPCODE:
1450{
1451 SvREFCNT_dec (self->cb_object);
1452 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
1453
1454 XPUSHs (ST (0));
1455}
1456
1457void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
1458 PPCODE:
1459{
1460 if (!self->cb_sk_object)
1461 self->cb_sk_object = newHV ();
1462
1463 if (SvOK (cb))
1464 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1465 else
856 { 1466 {
857 decode_hexdigit ['a' + i] = 10 + i; 1467 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
858 decode_hexdigit ['A' + i] = 10 + i; 1468
1469 if (!HvKEYS (self->cb_sk_object))
1470 {
1471 SvREFCNT_dec (self->cb_sk_object);
1472 self->cb_sk_object = 0;
1473 }
859 } 1474 }
860 1475
861 json_stash = gv_stashpv ("JSON::XS", 1); 1476 XPUSHs (ST (0));
862} 1477}
863 1478
864PROTOTYPES: DISABLE 1479void encode (JSON *self, SV *scalar)
1480 PPCODE:
1481 XPUSHs (encode_json (scalar, self));
865 1482
866SV *new (char *dummy) 1483void decode (JSON *self, SV *jsonstr)
1484 PPCODE:
1485 XPUSHs (decode_json (jsonstr, self, 0));
1486
1487void decode_prefix (JSON *self, SV *jsonstr)
1488 PPCODE:
1489{
1490 UV offset;
1491 EXTEND (SP, 2);
1492 PUSHs (decode_json (jsonstr, self, &offset));
1493 PUSHs (sv_2mortal (newSVuv (offset)));
1494}
1495
1496void DESTROY (JSON *self)
867 CODE: 1497 CODE:
868 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1498 SvREFCNT_dec (self->cb_sk_object);
869 OUTPUT: 1499 SvREFCNT_dec (self->cb_object);
870 RETVAL
871
872SV *ascii (SV *self, int enable)
873 ALIAS:
874 ascii = F_ASCII
875 utf8 = F_UTF8
876 indent = F_INDENT
877 canonical = F_CANONICAL
878 space_before = F_SPACE_BEFORE
879 space_after = F_SPACE_AFTER
880 json_rpc = F_JSON_RPC
881 pretty = F_PRETTY
882 allow_nonref = F_ALLOW_NONREF
883 CODE:
884{
885 UV *uv = SvJSON (self);
886 if (enable)
887 *uv |= ix;
888 else
889 *uv &= ~ix;
890
891 RETVAL = newSVsv (self);
892}
893 OUTPUT:
894 RETVAL
895
896void encode (SV *self, SV *scalar)
897 PPCODE:
898 XPUSHs (encode_json (scalar, *SvJSON (self)));
899
900void decode (SV *self, SV *jsonstr)
901 PPCODE:
902 XPUSHs (decode_json (jsonstr, *SvJSON (self)));
903 1500
904PROTOTYPES: ENABLE 1501PROTOTYPES: ENABLE
905 1502
906void to_json (SV *scalar) 1503void to_json (SV *scalar)
907 PPCODE: 1504 PPCODE:
1505{
1506 JSON json = { F_DEFAULT | F_UTF8 };
908 XPUSHs (encode_json (scalar, F_UTF8)); 1507 XPUSHs (encode_json (scalar, &json));
1508}
909 1509
910void from_json (SV *jsonstr) 1510void from_json (SV *jsonstr)
911 PPCODE: 1511 PPCODE:
1512{
1513 JSON json = { F_DEFAULT | F_UTF8 };
912 XPUSHs (decode_json (jsonstr, F_UTF8)); 1514 XPUSHs (decode_json (jsonstr, &json, 0));
1515}
913 1516

Diff Legend

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