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 1.13 by root, Sat Mar 24 22:55:16 2007 UTC vs.
Revision 1.49 by root, Sun Jul 1 23:40:07 2007 UTC

… …
3#include "XSUB.h" 3#include "XSUB.h"
4 4
5#include "assert.h" 5#include "assert.h"
6#include "string.h" 6#include "string.h"
7#include "stdlib.h" 7#include "stdlib.h"
8#include "stdio.h"
8 9
10#if defined(__BORLANDC__) || defined(_MSC_VER)
11# define snprintf _snprintf // C compilers have this in stdio.h
12#endif
13
14// some old perls do not have this, try to make it work, no
15// guarentees, though. if it breaks, you get to keep the pieces.
16#ifndef UTF8_MAXBYTES
17# define UTF8_MAXBYTES 13
18#endif
19
9#define F_ASCII 0x00000001 20#define F_ASCII 0x00000001UL
21#define F_LATIN1 0x00000002UL
10#define F_UTF8 0x00000002 22#define F_UTF8 0x00000004UL
11#define F_INDENT 0x00000004 23#define F_INDENT 0x00000008UL
12#define F_CANONICAL 0x00000008 24#define F_CANONICAL 0x00000010UL
13#define F_SPACE_BEFORE 0x00000010 25#define F_SPACE_BEFORE 0x00000020UL
14#define F_SPACE_AFTER 0x00000020 26#define F_SPACE_AFTER 0x00000040UL
15#define F_ALLOW_NONREF 0x00000080 27#define F_ALLOW_NONREF 0x00000100UL
16#define F_SHRINK 0x00000100 28#define F_SHRINK 0x00000200UL
29#define F_ALLOW_BLESSED 0x00000400UL
30#define F_CONV_BLESSED 0x00000800UL
31#define F_MAXDEPTH 0xf8000000UL
32#define S_MAXDEPTH 27
33#define F_MAXSIZE 0x01f00000UL
34#define S_MAXSIZE 20
35#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
36
37#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
38#define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE ))
17 39
18#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 40#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
19#define F_DEFAULT 0 41#define F_DEFAULT (9UL << S_MAXDEPTH)
20 42
21#define INIT_SIZE 32 // initial scalar size to be allocated 43#define INIT_SIZE 32 // initial scalar size to be allocated
22#define INDENT_STEP 3 // spaces per indentation level 44#define INDENT_STEP 3 // spaces per indentation level
23 45
24#define UTF8_MAX_LEN 11 // for perls UTF-X: max. number of octets per character
25#define SHORT_STRING_LEN 256 // special-case strings of up to this size 46#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
26 47
27#define SB do { 48#define SB do {
28#define SE } while (0) 49#define SE } while (0)
29 50
51#if __GNUC__ >= 3
52# define expect(expr,value) __builtin_expect ((expr),(value))
53# define inline inline
54#else
55# define expect(expr,value) (expr)
56# define inline static
57#endif
58
59#define expect_false(expr) expect ((expr) != 0, 0)
60#define expect_true(expr) expect ((expr) != 0, 1)
61
30static HV *json_stash; // JSON::XS:: 62static HV *json_stash, *json_boolean_stash; // JSON::XS::
63static SV *json_true, *json_false;
64
65typedef struct {
66 U32 flags;
67 SV *cb_object, *cb_sk_object;
68} JSON;
31 69
32///////////////////////////////////////////////////////////////////////////// 70/////////////////////////////////////////////////////////////////////////////
33// utility functions 71// utility functions
34 72
35static UV * 73inline void
36SvJSON (SV *sv)
37{
38 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash))
39 croak ("object is not of type JSON::XS");
40
41 return &SvUVX (SvRV (sv));
42}
43
44static void
45shrink (SV *sv) 74shrink (SV *sv)
46{ 75{
47 sv_utf8_downgrade (sv, 1); 76 sv_utf8_downgrade (sv, 1);
48 if (SvLEN (sv) > SvCUR (sv) + 1) 77 if (SvLEN (sv) > SvCUR (sv) + 1)
49 { 78 {
… …
58// decode an utf-8 character and return it, or (UV)-1 in 87// decode an utf-8 character and return it, or (UV)-1 in
59// case of an error. 88// case of an error.
60// we special-case "safe" characters from U+80 .. U+7FF, 89// we special-case "safe" characters from U+80 .. U+7FF,
61// but use the very good perl function to parse anything else. 90// but use the very good perl function to parse anything else.
62// note that we never call this function for a ascii codepoints 91// note that we never call this function for a ascii codepoints
63static UV 92inline UV
64decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) 93decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
65{ 94{
66 if (s[0] > 0xdf || s[0] < 0xc2) 95 if (expect_false (s[0] > 0xdf || s[0] < 0xc2))
67 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); 96 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
68 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) 97 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)
69 { 98 {
70 *clen = 2; 99 *clen = 2;
71 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 100 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
72 } 101 }
73 else 102 else
103 {
104 *clen = (STRLEN)-1;
74 return (UV)-1; 105 return (UV)-1;
106 }
75} 107}
76 108
77///////////////////////////////////////////////////////////////////////////// 109/////////////////////////////////////////////////////////////////////////////
78// encoder 110// encoder
79 111
… …
81typedef struct 113typedef struct
82{ 114{
83 char *cur; // SvPVX (sv) + current output position 115 char *cur; // SvPVX (sv) + current output position
84 char *end; // SvEND (sv) 116 char *end; // SvEND (sv)
85 SV *sv; // result scalar 117 SV *sv; // result scalar
86 UV flags; // F_* 118 JSON json;
87 int indent; // indentation level 119 U32 indent; // indentation level
88 int max_depth; // max. recursion level 120 U32 maxdepth; // max. indentation/recursion level
89} enc_t; 121} enc_t;
90 122
91static void 123inline void
92need (enc_t *enc, STRLEN len) 124need (enc_t *enc, STRLEN len)
93{ 125{
94 if (enc->cur + len >= enc->end) 126 if (expect_false (enc->cur + len >= enc->end))
95 { 127 {
96 STRLEN cur = enc->cur - SvPVX (enc->sv); 128 STRLEN cur = enc->cur - SvPVX (enc->sv);
97 SvGROW (enc->sv, cur + len + 1); 129 SvGROW (enc->sv, cur + len + 1);
98 enc->cur = SvPVX (enc->sv) + cur; 130 enc->cur = SvPVX (enc->sv) + cur;
99 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv); 131 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
100 } 132 }
101} 133}
102 134
103static void 135inline void
104encode_ch (enc_t *enc, char ch) 136encode_ch (enc_t *enc, char ch)
105{ 137{
106 need (enc, 1); 138 need (enc, 1);
107 *enc->cur++ = ch; 139 *enc->cur++ = ch;
108} 140}
… …
116 148
117 while (str < end) 149 while (str < end)
118 { 150 {
119 unsigned char ch = *(unsigned char *)str; 151 unsigned char ch = *(unsigned char *)str;
120 152
121 if (ch >= 0x20 && ch < 0x80) // most common case 153 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
122 { 154 {
123 if (ch == '"') // but with slow exceptions 155 if (expect_false (ch == '"')) // but with slow exceptions
124 { 156 {
125 need (enc, len += 1); 157 need (enc, len += 1);
126 *enc->cur++ = '\\'; 158 *enc->cur++ = '\\';
127 *enc->cur++ = '"'; 159 *enc->cur++ = '"';
128 } 160 }
129 else if (ch == '\\') 161 else if (expect_false (ch == '\\'))
130 { 162 {
131 need (enc, len += 1); 163 need (enc, len += 1);
132 *enc->cur++ = '\\'; 164 *enc->cur++ = '\\';
133 *enc->cur++ = '\\'; 165 *enc->cur++ = '\\';
134 } 166 }
… …
152 STRLEN clen; 184 STRLEN clen;
153 UV uch; 185 UV uch;
154 186
155 if (is_utf8) 187 if (is_utf8)
156 { 188 {
157 //uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
158 uch = decode_utf8 (str, end - str, &clen); 189 uch = decode_utf8 (str, end - str, &clen);
159 if (clen == (STRLEN)-1) 190 if (clen == (STRLEN)-1)
160 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str); 191 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
161 } 192 }
162 else 193 else
… …
166 } 197 }
167 198
168 if (uch > 0x10FFFFUL) 199 if (uch > 0x10FFFFUL)
169 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); 200 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
170 201
171 if (uch < 0x80 || enc->flags & F_ASCII) 202 if (uch < 0x80 || enc->json.flags & F_ASCII || (enc->json.flags & F_LATIN1 && uch > 0xFF))
172 { 203 {
173 if (uch > 0xFFFFUL) 204 if (uch > 0xFFFFUL)
174 { 205 {
175 need (enc, len += 11); 206 need (enc, len += 11);
176 sprintf (enc->cur, "\\u%04x\\u%04x", 207 sprintf (enc->cur, "\\u%04x\\u%04x",
… …
190 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 221 *enc->cur++ = hexdigit [(uch >> 0) & 15];
191 } 222 }
192 223
193 str += clen; 224 str += clen;
194 } 225 }
226 else if (enc->json.flags & F_LATIN1)
227 {
228 *enc->cur++ = uch;
229 str += clen;
230 }
195 else if (is_utf8) 231 else if (is_utf8)
196 { 232 {
197 need (enc, len += clen); 233 need (enc, len += clen);
198 do 234 do
199 { 235 {
… …
201 } 237 }
202 while (--clen); 238 while (--clen);
203 } 239 }
204 else 240 else
205 { 241 {
206 need (enc, len += UTF8_MAX_LEN - 1); // never more than 11 bytes needed 242 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
207 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 243 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
208 ++str; 244 ++str;
209 } 245 }
210 } 246 }
211 } 247 }
… …
213 249
214 --len; 250 --len;
215 } 251 }
216} 252}
217 253
218static void 254inline void
219encode_indent (enc_t *enc) 255encode_indent (enc_t *enc)
220{ 256{
221 if (enc->flags & F_INDENT) 257 if (enc->json.flags & F_INDENT)
222 { 258 {
223 int spaces = enc->indent * INDENT_STEP; 259 int spaces = enc->indent * INDENT_STEP;
224 260
225 need (enc, spaces); 261 need (enc, spaces);
226 memset (enc->cur, ' ', spaces); 262 memset (enc->cur, ' ', spaces);
227 enc->cur += spaces; 263 enc->cur += spaces;
228 } 264 }
229} 265}
230 266
231static void 267inline void
232encode_space (enc_t *enc) 268encode_space (enc_t *enc)
233{ 269{
234 need (enc, 1); 270 need (enc, 1);
235 encode_ch (enc, ' '); 271 encode_ch (enc, ' ');
236} 272}
237 273
238static void 274inline void
239encode_nl (enc_t *enc) 275encode_nl (enc_t *enc)
240{ 276{
241 if (enc->flags & F_INDENT) 277 if (enc->json.flags & F_INDENT)
242 { 278 {
243 need (enc, 1); 279 need (enc, 1);
244 encode_ch (enc, '\n'); 280 encode_ch (enc, '\n');
245 } 281 }
246} 282}
247 283
248static void 284inline void
249encode_comma (enc_t *enc) 285encode_comma (enc_t *enc)
250{ 286{
251 encode_ch (enc, ','); 287 encode_ch (enc, ',');
252 288
253 if (enc->flags & F_INDENT) 289 if (enc->json.flags & F_INDENT)
254 encode_nl (enc); 290 encode_nl (enc);
255 else if (enc->flags & F_SPACE_AFTER) 291 else if (enc->json.flags & F_SPACE_AFTER)
256 encode_space (enc); 292 encode_space (enc);
257} 293}
258 294
259static void encode_sv (enc_t *enc, SV *sv); 295static void encode_sv (enc_t *enc, SV *sv);
260 296
261static void 297static void
262encode_av (enc_t *enc, AV *av) 298encode_av (enc_t *enc, AV *av)
263{ 299{
264 int i, len = av_len (av); 300 int i, len = av_len (av);
301
302 if (enc->indent >= enc->maxdepth)
303 croak ("data structure too deep (hit recursion limit)");
265 304
266 encode_ch (enc, '['); encode_nl (enc); 305 encode_ch (enc, '['); encode_nl (enc);
267 ++enc->indent; 306 ++enc->indent;
268 307
269 for (i = 0; i <= len; ++i) 308 for (i = 0; i <= len; ++i)
… …
300 else 339 else
301 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 340 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
302 341
303 encode_ch (enc, '"'); 342 encode_ch (enc, '"');
304 343
305 if (enc->flags & F_SPACE_BEFORE) encode_space (enc); 344 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
306 encode_ch (enc, ':'); 345 encode_ch (enc, ':');
307 if (enc->flags & F_SPACE_AFTER ) encode_space (enc); 346 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
308 encode_sv (enc, HeVAL (he)); 347 encode_sv (enc, HeVAL (he));
309} 348}
310 349
311// compare hash entries, used when all keys are bytestrings 350// compare hash entries, used when all keys are bytestrings
312static int 351static int
… …
335 374
336static void 375static void
337encode_hv (enc_t *enc, HV *hv) 376encode_hv (enc_t *enc, HV *hv)
338{ 377{
339 int count, i; 378 int count, i;
379
380 if (enc->indent >= enc->maxdepth)
381 croak ("data structure too deep (hit recursion limit)");
340 382
341 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 383 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
342 384
343 if ((count = hv_iterinit (hv))) 385 if ((count = hv_iterinit (hv)))
344 { 386 {
345 // for canonical output we have to sort by keys first 387 // for canonical output we have to sort by keys first
346 // actually, this is mostly due to the stupid so-called 388 // actually, this is mostly due to the stupid so-called
347 // security workaround added somewhere in 5.8.x. 389 // security workaround added somewhere in 5.8.x.
348 // that randomises hash orderings 390 // that randomises hash orderings
349 if (enc->flags & F_CANONICAL) 391 if (enc->json.flags & F_CANONICAL)
350 { 392 {
351 HE *he, *hes [count]; // if your compiler dies here, you need to enable C99 mode
352 int fast = 1; 393 int fast = 1;
394 HE *he;
395#if defined(__BORLANDC__) || defined(_MSC_VER)
396 HE **hes = _alloca (count * sizeof (HE));
397#else
398 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
399#endif
353 400
354 i = 0; 401 i = 0;
355 while ((he = hv_iternext (hv))) 402 while ((he = hv_iternext (hv)))
356 { 403 {
357 hes [i++] = he; 404 hes [i++] = he;
… …
392 439
393 encode_nl (enc); 440 encode_nl (enc);
394 } 441 }
395 else 442 else
396 { 443 {
397 SV *sv;
398 HE *he = hv_iternext (hv); 444 HE *he = hv_iternext (hv);
399 445
400 for (;;) 446 for (;;)
401 { 447 {
402 encode_indent (enc); 448 encode_indent (enc);
… …
411 encode_nl (enc); 457 encode_nl (enc);
412 } 458 }
413 } 459 }
414 460
415 --enc->indent; encode_indent (enc); encode_ch (enc, '}'); 461 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
462}
463
464// encode objects, arrays and special \0=false and \1=true values.
465static void
466encode_rv (enc_t *enc, SV *sv)
467{
468 svtype svt;
469
470 SvGETMAGIC (sv);
471 svt = SvTYPE (sv);
472
473 if (expect_false (SvOBJECT (sv)))
474 {
475 if (SvSTASH (sv) == json_boolean_stash)
476 {
477 if (SvIV (sv) == 0)
478 encode_str (enc, "false", 5, 0);
479 else
480 encode_str (enc, "true", 4, 0);
481 }
482 else
483 {
484#if 0
485 if (0 && sv_derived_from (rv, "JSON::Literal"))
486 {
487 // not yet
488 }
489#endif
490 if (enc->json.flags & F_CONV_BLESSED)
491 {
492 // we re-bless the reference to get overload and other niceties right
493 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 1);
494
495 if (to_json)
496 {
497 dSP; ENTER; SAVETMPS; PUSHMARK (SP);
498 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
499
500 // calling with G_SCALAR ensures that we always get a 1 reutrn value
501 // check anyways.
502 PUTBACK;
503 assert (1 == call_sv ((SV *)GvCV (to_json), G_SCALAR));
504 SPAGAIN;
505
506 encode_sv (enc, POPs);
507
508 FREETMPS; LEAVE;
509 }
510 else if (enc->json.flags & F_ALLOW_BLESSED)
511 encode_str (enc, "null", 4, 0);
512 else
513 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
514 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
515 }
516 else if (enc->json.flags & F_ALLOW_BLESSED)
517 encode_str (enc, "null", 4, 0);
518 else
519 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",
520 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
521 }
522 }
523 else if (svt == SVt_PVHV)
524 encode_hv (enc, (HV *)sv);
525 else if (svt == SVt_PVAV)
526 encode_av (enc, (AV *)sv);
527 else if (svt < SVt_PVAV)
528 {
529 if (SvNIOK (sv) && SvIV (sv) == 0)
530 encode_str (enc, "false", 5, 0);
531 else if (SvNIOK (sv) && SvIV (sv) == 1)
532 encode_str (enc, "true", 4, 0);
533 else
534 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
535 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
536 }
537 else
538 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
539 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
416} 540}
417 541
418static void 542static void
419encode_sv (enc_t *enc, SV *sv) 543encode_sv (enc_t *enc, SV *sv)
420{ 544{
… …
428 encode_str (enc, str, len, SvUTF8 (sv)); 552 encode_str (enc, str, len, SvUTF8 (sv));
429 encode_ch (enc, '"'); 553 encode_ch (enc, '"');
430 } 554 }
431 else if (SvNOKp (sv)) 555 else if (SvNOKp (sv))
432 { 556 {
557 // trust that perl will do the right thing w.r.t. JSON syntax.
433 need (enc, NV_DIG + 32); 558 need (enc, NV_DIG + 32);
434 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 559 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
435 enc->cur += strlen (enc->cur); 560 enc->cur += strlen (enc->cur);
436 } 561 }
437 else if (SvIOKp (sv)) 562 else if (SvIOKp (sv))
438 { 563 {
439 need (enc, 64); 564 // we assume we can always read an IV as a UV
565 if (SvUV (sv) & ~(UV)0x7fff)
566 {
567 // large integer, use the (rather slow) snprintf way.
568 need (enc, sizeof (UV) * 3);
440 enc->cur += 569 enc->cur +=
441 SvIsUV(sv) 570 SvIsUV(sv)
442 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 571 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
443 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 572 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
573 }
574 else
575 {
576 // optimise the "small number case"
577 // code will likely be branchless and use only a single multiplication
578 I32 i = SvIV (sv);
579 U32 u;
580 char digit, nz = 0;
581
582 need (enc, 6);
583
584 *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
585 u = i < 0 ? -i : i;
586
587 // convert to 4.28 fixed-point representation
588 u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
589
590 // now output digit by digit, each time masking out the integer part
591 // and multiplying by 5 while moving the decimal point one to the right,
592 // resulting in a net multiplication by 10.
593 // we always write the digit to memory but conditionally increment
594 // the pointer, to ease the usage of conditional move instructions.
595 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5;
596 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5;
597 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5;
598 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5;
599 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
600 }
444 } 601 }
445 else if (SvROK (sv)) 602 else if (SvROK (sv))
446 { 603 encode_rv (enc, SvRV (sv));
447 SV *rv = SvRV (sv);
448
449 if (enc->indent >= enc->max_depth)
450 croak ("data structure too deep (hit recursion limit)");
451
452 switch (SvTYPE (rv))
453 {
454 case SVt_PVAV: encode_av (enc, (AV *)rv); break;
455 case SVt_PVHV: encode_hv (enc, (HV *)rv); break;
456
457 default:
458 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
459 SvPV_nolen (sv));
460 }
461 }
462 else if (!SvOK (sv)) 604 else if (!SvOK (sv))
463 encode_str (enc, "null", 4, 0); 605 encode_str (enc, "null", 4, 0);
464 else 606 else
465 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 607 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
466 SvPV_nolen (sv), SvFLAGS (sv)); 608 SvPV_nolen (sv), SvFLAGS (sv));
467} 609}
468 610
469static SV * 611static SV *
470encode_json (SV *scalar, UV flags) 612encode_json (SV *scalar, JSON *json)
471{ 613{
614 enc_t enc;
615
472 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 616 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
473 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 617 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
474 618
475 enc_t enc; 619 enc.json = *json;
476 enc.flags = flags;
477 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 620 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
478 enc.cur = SvPVX (enc.sv); 621 enc.cur = SvPVX (enc.sv);
479 enc.end = SvEND (enc.sv); 622 enc.end = SvEND (enc.sv);
480 enc.indent = 0; 623 enc.indent = 0;
481 enc.max_depth = 0x7fffffffUL; 624 enc.maxdepth = DEC_DEPTH (enc.json.flags);
482 625
483 SvPOK_only (enc.sv); 626 SvPOK_only (enc.sv);
484 encode_sv (&enc, scalar); 627 encode_sv (&enc, scalar);
485 628
629 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
630 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
631
486 if (!(flags & (F_ASCII | F_UTF8))) 632 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
487 SvUTF8_on (enc.sv); 633 SvUTF8_on (enc.sv);
488 634
489 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
490
491 if (enc.flags & F_SHRINK) 635 if (enc.json.flags & F_SHRINK)
492 shrink (enc.sv); 636 shrink (enc.sv);
493 637
494 return enc.sv; 638 return enc.sv;
495} 639}
496 640
… …
501typedef struct 645typedef struct
502{ 646{
503 char *cur; // current parser pointer 647 char *cur; // current parser pointer
504 char *end; // end of input string 648 char *end; // end of input string
505 const char *err; // parse error, if != 0 649 const char *err; // parse error, if != 0
506 UV flags; // F_* 650 JSON json;
651 U32 depth; // recursion depth
652 U32 maxdepth; // recursion depth limit
507} dec_t; 653} dec_t;
508 654
509static void 655inline void
510decode_ws (dec_t *dec) 656decode_ws (dec_t *dec)
511{ 657{
512 for (;;) 658 for (;;)
513 { 659 {
514 char ch = *dec->cur; 660 char ch = *dec->cur;
… …
520 ++dec->cur; 666 ++dec->cur;
521 } 667 }
522} 668}
523 669
524#define ERR(reason) SB dec->err = reason; goto fail; SE 670#define ERR(reason) SB dec->err = reason; goto fail; SE
671
525#define EXPECT_CH(ch) SB \ 672#define EXPECT_CH(ch) SB \
526 if (*dec->cur != ch) \ 673 if (*dec->cur != ch) \
527 ERR (# ch " expected"); \ 674 ERR (# ch " expected"); \
528 ++dec->cur; \ 675 ++dec->cur; \
529 SE 676 SE
530 677
678#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)")
679#define DEC_DEC_DEPTH --dec->depth
680
531static SV *decode_sv (dec_t *dec); 681static SV *decode_sv (dec_t *dec);
532 682
533static signed char decode_hexdigit[256]; 683static signed char decode_hexdigit[256];
534 684
535static UV 685static UV
536decode_4hex (dec_t *dec) 686decode_4hex (dec_t *dec)
537{ 687{
538 signed char d1, d2, d3, d4; 688 signed char d1, d2, d3, d4;
539 unsigned char *cur = (unsigned char *)dec->cur; 689 unsigned char *cur = (unsigned char *)dec->cur;
540 690
541 d1 = decode_hexdigit [cur [0]]; if (d1 < 0) ERR ("four hexadecimal digits expected"); 691 d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected");
542 d2 = decode_hexdigit [cur [1]]; if (d2 < 0) ERR ("four hexadecimal digits expected"); 692 d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected");
543 d3 = decode_hexdigit [cur [2]]; if (d3 < 0) ERR ("four hexadecimal digits expected"); 693 d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected");
544 d4 = decode_hexdigit [cur [3]]; if (d4 < 0) ERR ("four hexadecimal digits expected"); 694 d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected");
545 695
546 dec->cur += 4; 696 dec->cur += 4;
547 697
548 return ((UV)d1) << 12 698 return ((UV)d1) << 12
549 | ((UV)d2) << 8 699 | ((UV)d2) << 8
… …
557static SV * 707static SV *
558decode_str (dec_t *dec) 708decode_str (dec_t *dec)
559{ 709{
560 SV *sv = 0; 710 SV *sv = 0;
561 int utf8 = 0; 711 int utf8 = 0;
712 char *dec_cur = dec->cur;
562 713
563 do 714 do
564 { 715 {
565 char buf [SHORT_STRING_LEN + UTF8_MAX_LEN]; 716 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
566 char *cur = buf; 717 char *cur = buf;
567 718
568 do 719 do
569 { 720 {
570 unsigned char ch = *(unsigned char *)dec->cur++; 721 unsigned char ch = *(unsigned char *)dec_cur++;
571 722
572 if (ch == '"') 723 if (expect_false (ch == '"'))
573 { 724 {
574 --dec->cur; 725 --dec_cur;
575 break; 726 break;
576 } 727 }
577 else if (ch == '\\') 728 else if (expect_false (ch == '\\'))
578 { 729 {
579 switch (*dec->cur) 730 switch (*dec_cur)
580 { 731 {
581 case '\\': 732 case '\\':
582 case '/': 733 case '/':
583 case '"': *cur++ = *dec->cur++; break; 734 case '"': *cur++ = *dec_cur++; break;
584 735
585 case 'b': ++dec->cur; *cur++ = '\010'; break; 736 case 'b': ++dec_cur; *cur++ = '\010'; break;
586 case 't': ++dec->cur; *cur++ = '\011'; break; 737 case 't': ++dec_cur; *cur++ = '\011'; break;
587 case 'n': ++dec->cur; *cur++ = '\012'; break; 738 case 'n': ++dec_cur; *cur++ = '\012'; break;
588 case 'f': ++dec->cur; *cur++ = '\014'; break; 739 case 'f': ++dec_cur; *cur++ = '\014'; break;
589 case 'r': ++dec->cur; *cur++ = '\015'; break; 740 case 'r': ++dec_cur; *cur++ = '\015'; break;
590 741
591 case 'u': 742 case 'u':
592 { 743 {
593 UV lo, hi; 744 UV lo, hi;
594 ++dec->cur; 745 ++dec_cur;
595 746
747 dec->cur = dec_cur;
596 hi = decode_4hex (dec); 748 hi = decode_4hex (dec);
749 dec_cur = dec->cur;
597 if (hi == (UV)-1) 750 if (hi == (UV)-1)
598 goto fail; 751 goto fail;
599 752
600 // possibly a surrogate pair 753 // possibly a surrogate pair
601 if (hi >= 0xd800) 754 if (hi >= 0xd800)
602 if (hi < 0xdc00) 755 if (hi < 0xdc00)
603 { 756 {
604 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 757 if (dec_cur [0] != '\\' || dec_cur [1] != 'u')
605 ERR ("missing low surrogate character in surrogate pair"); 758 ERR ("missing low surrogate character in surrogate pair");
606 759
607 dec->cur += 2; 760 dec_cur += 2;
608 761
762 dec->cur = dec_cur;
609 lo = decode_4hex (dec); 763 lo = decode_4hex (dec);
764 dec_cur = dec->cur;
610 if (lo == (UV)-1) 765 if (lo == (UV)-1)
611 goto fail; 766 goto fail;
612 767
613 if (lo < 0xdc00 || lo >= 0xe000) 768 if (lo < 0xdc00 || lo >= 0xe000)
614 ERR ("surrogate pair expected"); 769 ERR ("surrogate pair expected");
… …
628 *cur++ = hi; 783 *cur++ = hi;
629 } 784 }
630 break; 785 break;
631 786
632 default: 787 default:
633 --dec->cur; 788 --dec_cur;
634 ERR ("illegal backslash escape sequence in string"); 789 ERR ("illegal backslash escape sequence in string");
635 } 790 }
636 } 791 }
637 else if (ch >= 0x20 && ch <= 0x7f) 792 else if (expect_true (ch >= 0x20 && ch <= 0x7f))
638 *cur++ = ch; 793 *cur++ = ch;
639 else if (ch >= 0x80) 794 else if (ch >= 0x80)
640 { 795 {
641 --dec->cur;
642
643 STRLEN clen; 796 STRLEN clen;
797 UV uch;
798
799 --dec_cur;
800
644 UV uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen); 801 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
645 if (clen == (STRLEN)-1) 802 if (clen == (STRLEN)-1)
646 ERR ("malformed UTF-8 character in JSON string"); 803 ERR ("malformed UTF-8 character in JSON string");
647 804
648 do 805 do
649 {
650 *cur++ = *dec->cur++; 806 *cur++ = *dec_cur++;
651 }
652 while (--clen); 807 while (--clen);
653 808
654 utf8 = 1; 809 utf8 = 1;
655 } 810 }
656 else if (!ch)
657 ERR ("unexpected end of string while parsing json string");
658 else 811 else
812 {
813 --dec_cur;
814
815 if (!ch)
816 ERR ("unexpected end of string while parsing JSON string");
817 else
659 ERR ("invalid character encountered"); 818 ERR ("invalid character encountered while parsing JSON string");
660 819 }
661 } 820 }
662 while (cur < buf + SHORT_STRING_LEN); 821 while (cur < buf + SHORT_STRING_LEN);
663 822
823 {
664 STRLEN len = cur - buf; 824 STRLEN len = cur - buf;
665 825
666 if (sv) 826 if (sv)
667 { 827 {
668 SvGROW (sv, SvCUR (sv) + len + 1); 828 SvGROW (sv, SvCUR (sv) + len + 1);
669 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 829 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
670 SvCUR_set (sv, SvCUR (sv) + len); 830 SvCUR_set (sv, SvCUR (sv) + len);
671 } 831 }
672 else 832 else
673 sv = newSVpvn (buf, len); 833 sv = newSVpvn (buf, len);
674 } 834 }
835 }
675 while (*dec->cur != '"'); 836 while (*dec_cur != '"');
676 837
677 ++dec->cur; 838 ++dec_cur;
678 839
679 if (sv) 840 if (sv)
680 { 841 {
681 SvPOK_only (sv); 842 SvPOK_only (sv);
682 *SvEND (sv) = 0; 843 *SvEND (sv) = 0;
… …
685 SvUTF8_on (sv); 846 SvUTF8_on (sv);
686 } 847 }
687 else 848 else
688 sv = newSVpvn ("", 0); 849 sv = newSVpvn ("", 0);
689 850
851 dec->cur = dec_cur;
690 return sv; 852 return sv;
691 853
692fail: 854fail:
855 dec->cur = dec_cur;
693 return 0; 856 return 0;
694} 857}
695 858
696static SV * 859static SV *
697decode_num (dec_t *dec) 860decode_num (dec_t *dec)
… …
755 is_nv = 1; 918 is_nv = 1;
756 } 919 }
757 920
758 if (!is_nv) 921 if (!is_nv)
759 { 922 {
760 UV uv; 923 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so
761 int numtype = grok_number (start, dec->cur - start, &uv); 924 if (*start == '-')
762 if (numtype & IS_NUMBER_IN_UV) 925 switch (dec->cur - start)
763 if (numtype & IS_NUMBER_NEG)
764 { 926 {
765 if (uv < (UV)IV_MIN) 927 case 2: return newSViv (-( start [1] - '0' * 1));
766 return newSViv (-(IV)uv); 928 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
929 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
930 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
767 } 931 }
932 else
933 switch (dec->cur - start)
934 {
935 case 1: return newSViv ( start [0] - '0' * 1);
936 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
937 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
938 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
939 }
940
941 {
942 UV uv;
943 int numtype = grok_number (start, dec->cur - start, &uv);
944 if (numtype & IS_NUMBER_IN_UV)
945 if (numtype & IS_NUMBER_NEG)
946 {
947 if (uv < (UV)IV_MIN)
948 return newSViv (-(IV)uv);
949 }
768 else 950 else
769 return newSVuv (uv); 951 return newSVuv (uv);
952
953 // here would likely be the place for bigint support
770 } 954 }
955 }
771 956
957 // if we ever support bigint or bigfloat, this is the place for bigfloat
772 return newSVnv (Atof (start)); 958 return newSVnv (Atof (start));
773 959
774fail: 960fail:
775 return 0; 961 return 0;
776} 962}
… …
778static SV * 964static SV *
779decode_av (dec_t *dec) 965decode_av (dec_t *dec)
780{ 966{
781 AV *av = newAV (); 967 AV *av = newAV ();
782 968
969 DEC_INC_DEPTH;
783 decode_ws (dec); 970 decode_ws (dec);
971
784 if (*dec->cur == ']') 972 if (*dec->cur == ']')
785 ++dec->cur; 973 ++dec->cur;
786 else 974 else
787 for (;;) 975 for (;;)
788 { 976 {
… …
806 ERR (", or ] expected while parsing array"); 994 ERR (", or ] expected while parsing array");
807 995
808 ++dec->cur; 996 ++dec->cur;
809 } 997 }
810 998
999 DEC_DEC_DEPTH;
811 return newRV_noinc ((SV *)av); 1000 return newRV_noinc ((SV *)av);
812 1001
813fail: 1002fail:
814 SvREFCNT_dec (av); 1003 SvREFCNT_dec (av);
1004 DEC_DEC_DEPTH;
815 return 0; 1005 return 0;
816} 1006}
817 1007
818static SV * 1008static SV *
819decode_hv (dec_t *dec) 1009decode_hv (dec_t *dec)
820{ 1010{
1011 SV *sv;
821 HV *hv = newHV (); 1012 HV *hv = newHV ();
822 1013
1014 DEC_INC_DEPTH;
823 decode_ws (dec); 1015 decode_ws (dec);
1016
824 if (*dec->cur == '}') 1017 if (*dec->cur == '}')
825 ++dec->cur; 1018 ++dec->cur;
826 else 1019 else
827 for (;;) 1020 for (;;)
828 { 1021 {
829 SV *key, *value;
830
831 decode_ws (dec); EXPECT_CH ('"'); 1022 decode_ws (dec); EXPECT_CH ('"');
832 1023
833 key = decode_str (dec); 1024 // heuristic: assume that
834 if (!key) 1025 // a) decode_str + hv_store_ent are abysmally slow.
835 goto fail; 1026 // b) most hash keys are short, simple ascii text.
1027 // => try to "fast-match" such strings to avoid
1028 // the overhead of decode_str + hv_store_ent.
1029 {
1030 SV *value;
1031 char *p = dec->cur;
1032 char *e = p + 24; // only try up to 24 bytes
836 1033
837 decode_ws (dec); EXPECT_CH (':'); 1034 for (;;)
838
839 value = decode_sv (dec);
840 if (!value)
841 { 1035 {
1036 // the >= 0x80 is true on most architectures
1037 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1038 {
1039 // slow path, back up and use decode_str
1040 SV *key = decode_str (dec);
1041 if (!key)
1042 goto fail;
1043
1044 decode_ws (dec); EXPECT_CH (':');
1045
1046 value = decode_sv (dec);
1047 if (!value)
1048 {
1049 SvREFCNT_dec (key);
1050 goto fail;
1051 }
1052
1053 hv_store_ent (hv, key, value, 0);
842 SvREFCNT_dec (key); 1054 SvREFCNT_dec (key);
1055
1056 break;
1057 }
1058 else if (*p == '"')
1059 {
1060 // fast path, got a simple key
1061 char *key = dec->cur;
1062 int len = p - key;
1063 dec->cur = p + 1;
1064
1065 decode_ws (dec); EXPECT_CH (':');
1066
1067 value = decode_sv (dec);
1068 if (!value)
843 goto fail; 1069 goto fail;
1070
1071 hv_store (hv, key, len, value, 0);
1072
1073 break;
1074 }
1075
1076 ++p;
844 } 1077 }
845 1078 }
846 //TODO: optimise
847 hv_store_ent (hv, key, value, 0);
848 1079
849 decode_ws (dec); 1080 decode_ws (dec);
850 1081
851 if (*dec->cur == '}') 1082 if (*dec->cur == '}')
852 { 1083 {
… …
858 ERR (", or } expected while parsing object/hash"); 1089 ERR (", or } expected while parsing object/hash");
859 1090
860 ++dec->cur; 1091 ++dec->cur;
861 } 1092 }
862 1093
1094 DEC_DEC_DEPTH;
1095 sv = newRV_noinc ((SV *)hv);
1096
1097 // check filter callbacks
1098 if (dec->json.flags & F_HOOK)
1099 {
1100 ENTER; SAVETMPS;
1101
1102 if (HvKEYS (hv) == 1 && dec->json.cb_sk_object)
1103 {
1104 int count;
1105
1106 dSP; PUSHMARK (SP);
1107 XPUSHs (sv_2mortal (sv));
1108
1109 PUTBACK; count = call_sv (dec->json.cb_sk_object, G_ARRAY); SPAGAIN;
1110
1111 if (count == 1)
1112 sv = newSVsv (POPs);
1113 else
1114 SvREFCNT_inc (sv);
1115 }
1116
1117 if (dec->json.cb_object)
1118 {
1119 int count;
1120
1121 dSP; ENTER; SAVETMPS; PUSHMARK (SP);
1122 XPUSHs (sv_2mortal (sv));
1123
1124 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1125
1126 if (count == 1)
1127 sv = newSVsv (POPs);
1128 else
1129 SvREFCNT_inc (sv);
1130 }
1131
1132 FREETMPS; LEAVE;
1133 }
1134
863 return newRV_noinc ((SV *)hv); 1135 return newRV_noinc ((SV *)hv);
864 1136
865fail: 1137fail:
866 SvREFCNT_dec (hv); 1138 SvREFCNT_dec (hv);
1139 DEC_DEC_DEPTH;
867 return 0; 1140 return 0;
868} 1141}
869 1142
870static SV * 1143static SV *
871decode_sv (dec_t *dec) 1144decode_sv (dec_t *dec)
872{ 1145{
873 decode_ws (dec); 1146 decode_ws (dec);
1147
1148 // the beauty of JSON: you need exactly one character lookahead
1149 // to parse anything.
874 switch (*dec->cur) 1150 switch (*dec->cur)
875 { 1151 {
876 case '"': ++dec->cur; return decode_str (dec); 1152 case '"': ++dec->cur; return decode_str (dec);
877 case '[': ++dec->cur; return decode_av (dec); 1153 case '[': ++dec->cur; return decode_av (dec);
878 case '{': ++dec->cur; return decode_hv (dec); 1154 case '{': ++dec->cur; return decode_hv (dec);
… …
884 1160
885 case 't': 1161 case 't':
886 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1162 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
887 { 1163 {
888 dec->cur += 4; 1164 dec->cur += 4;
889 return newSViv (1); 1165 return SvREFCNT_inc (json_true);
890 } 1166 }
891 else 1167 else
892 ERR ("'true' expected"); 1168 ERR ("'true' expected");
893 1169
894 break; 1170 break;
895 1171
896 case 'f': 1172 case 'f':
897 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1173 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
898 { 1174 {
899 dec->cur += 5; 1175 dec->cur += 5;
900 return newSViv (0); 1176 return SvREFCNT_inc (json_false);
901 } 1177 }
902 else 1178 else
903 ERR ("'false' expected"); 1179 ERR ("'false' expected");
904 1180
905 break; 1181 break;
… …
914 ERR ("'null' expected"); 1190 ERR ("'null' expected");
915 1191
916 break; 1192 break;
917 1193
918 default: 1194 default:
919 ERR ("malformed json string, neither array, object, number, string or atom"); 1195 ERR ("malformed JSON string, neither array, object, number, string or atom");
920 break; 1196 break;
921 } 1197 }
922 1198
923fail: 1199fail:
924 return 0; 1200 return 0;
925} 1201}
926 1202
927static SV * 1203static SV *
928decode_json (SV *string, UV flags) 1204decode_json (SV *string, JSON *json, UV *offset_return)
929{ 1205{
1206 dec_t dec;
1207 UV offset;
930 SV *sv; 1208 SV *sv;
931 1209
1210 SvGETMAGIC (string);
1211 SvUPGRADE (string, SVt_PV);
1212
1213 if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags))
1214 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1215 (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags));
1216
932 if (flags & F_UTF8) 1217 if (json->flags & F_UTF8)
933 sv_utf8_downgrade (string, 0); 1218 sv_utf8_downgrade (string, 0);
934 else 1219 else
935 sv_utf8_upgrade (string); 1220 sv_utf8_upgrade (string);
936 1221
937 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1222 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
938 1223
939 dec_t dec; 1224 dec.json = *json;
940 dec.flags = flags;
941 dec.cur = SvPVX (string); 1225 dec.cur = SvPVX (string);
942 dec.end = SvEND (string); 1226 dec.end = SvEND (string);
943 dec.err = 0; 1227 dec.err = 0;
1228 dec.depth = 0;
1229 dec.maxdepth = DEC_DEPTH (dec.json.flags);
944 1230
1231 if (dec.json.cb_object || dec.json.cb_sk_object)
1232 dec.json.flags |= F_HOOK;
1233
1234 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
945 sv = decode_sv (&dec); 1235 sv = decode_sv (&dec);
946 1236
1237 if (!(offset_return || !sv))
1238 {
1239 // check for trailing garbage
1240 decode_ws (&dec);
1241
1242 if (*dec.cur)
1243 {
1244 dec.err = "garbage after JSON object";
1245 SvREFCNT_dec (sv);
1246 sv = 0;
1247 }
1248 }
1249
1250 if (offset_return || !sv)
1251 {
1252 offset = dec.json.flags & F_UTF8
1253 ? dec.cur - SvPVX (string)
1254 : utf8_distance (dec.cur, SvPVX (string));
1255
1256 if (offset_return)
1257 *offset_return = offset;
1258 }
1259
947 if (!sv) 1260 if (!sv)
948 { 1261 {
949 IV offset = dec.flags & F_UTF8
950 ? dec.cur - SvPVX (string)
951 : utf8_distance (dec.cur, SvPVX (string));
952 SV *uni = sv_newmortal (); 1262 SV *uni = sv_newmortal ();
953 1263
954 // horrible hack to silence warning inside pv_uni_display 1264 // horrible hack to silence warning inside pv_uni_display
955 COP cop = *PL_curcop; 1265 COP cop = *PL_curcop;
956 cop.cop_warnings = pWARN_NONE; 1266 cop.cop_warnings = pWARN_NONE;
… …
958 SAVEVPTR (PL_curcop); 1268 SAVEVPTR (PL_curcop);
959 PL_curcop = &cop; 1269 PL_curcop = &cop;
960 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1270 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
961 LEAVE; 1271 LEAVE;
962 1272
963 croak ("%s, at character offset %d (%s)", 1273 croak ("%s, at character offset %d [\"%s\"]",
964 dec.err, 1274 dec.err,
965 (int)offset, 1275 (int)offset,
966 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1276 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
967 } 1277 }
968 1278
969 sv = sv_2mortal (sv); 1279 sv = sv_2mortal (sv);
970 1280
971 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1281 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))
972 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); 1282 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
973 1283
974 return sv; 1284 return sv;
975} 1285}
976 1286
… …
981 1291
982BOOT: 1292BOOT:
983{ 1293{
984 int i; 1294 int i;
985 1295
986 memset (decode_hexdigit, 0xff, 256);
987 for (i = 10; i--; ) 1296 for (i = 0; i < 256; ++i)
988 decode_hexdigit ['0' + i] = i; 1297 decode_hexdigit [i] =
1298 i >= '0' && i <= '9' ? i - '0'
1299 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1300 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1301 : -1;
989 1302
990 for (i = 7; i--; ) 1303 json_stash = gv_stashpv ("JSON::XS" , 1);
1304 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1305
1306 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true );
1307 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
1308}
1309
1310PROTOTYPES: DISABLE
1311
1312void new (char *klass)
1313 PPCODE:
1314{
1315 SV *pv = NEWSV (0, sizeof (JSON));
1316 SvPOK_only (pv);
1317 Zero (SvPVX (pv), 1, JSON);
1318 ((JSON *)SvPVX (pv))->flags = F_DEFAULT;
1319 XPUSHs (sv_2mortal (sv_bless (newRV_noinc (pv), json_stash)));
1320}
1321
1322void ascii (JSON *self, int enable = 1)
1323 ALIAS:
1324 ascii = F_ASCII
1325 latin1 = F_LATIN1
1326 utf8 = F_UTF8
1327 indent = F_INDENT
1328 canonical = F_CANONICAL
1329 space_before = F_SPACE_BEFORE
1330 space_after = F_SPACE_AFTER
1331 pretty = F_PRETTY
1332 allow_nonref = F_ALLOW_NONREF
1333 shrink = F_SHRINK
1334 allow_blessed = F_ALLOW_BLESSED
1335 convert_blessed = F_CONV_BLESSED
1336 PPCODE:
1337{
1338 if (enable)
1339 self->flags |= ix;
1340 else
1341 self->flags &= ~ix;
1342
1343 XPUSHs (ST (0));
1344}
1345
1346void max_depth (JSON *self, UV max_depth = 0x80000000UL)
1347 PPCODE:
1348{
1349 UV log2 = 0;
1350
1351 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1352
1353 while ((1UL << log2) < max_depth)
1354 ++log2;
1355
1356 self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1357
1358 XPUSHs (ST (0));
1359}
1360
1361void max_size (JSON *self, UV max_size = 0)
1362 PPCODE:
1363{
1364 UV log2 = 0;
1365
1366 if (max_size > 0x80000000UL) max_size = 0x80000000UL;
1367 if (max_size == 1) max_size = 2;
1368
1369 while ((1UL << log2) < max_size)
1370 ++log2;
1371
1372 self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1373
1374 XPUSHs (ST (0));
1375}
1376
1377void filter_json_objects (JSON *self, SV *cb = &PL_sv_undef)
1378 ALIAS:
1379 filter_sk_json_objects = 1
1380 PPCODE:
1381{
1382 if (!SvOK (cb))
1383 cb = 0;
1384
1385 switch (ix)
991 { 1386 {
992 decode_hexdigit ['a' + i] = 10 + i; 1387 case 0: self->cb_object = cb; break;
993 decode_hexdigit ['A' + i] = 10 + i; 1388 case 1: self->cb_sk_object = cb; break;
994 } 1389 }
995 1390
996 json_stash = gv_stashpv ("JSON::XS", 1); 1391 XPUSHs (ST (0));
997} 1392}
998 1393
999PROTOTYPES: DISABLE
1000
1001SV *new (char *dummy)
1002 CODE:
1003 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash);
1004 OUTPUT:
1005 RETVAL
1006
1007SV *ascii (SV *self, int enable = 1)
1008 ALIAS:
1009 ascii = F_ASCII
1010 utf8 = F_UTF8
1011 indent = F_INDENT
1012 canonical = F_CANONICAL
1013 space_before = F_SPACE_BEFORE
1014 space_after = F_SPACE_AFTER
1015 pretty = F_PRETTY
1016 allow_nonref = F_ALLOW_NONREF
1017 shrink = F_SHRINK
1018 CODE:
1019{
1020 UV *uv = SvJSON (self);
1021 if (enable)
1022 *uv |= ix;
1023 else
1024 *uv &= ~ix;
1025
1026 RETVAL = newSVsv (self);
1027}
1028 OUTPUT:
1029 RETVAL
1030
1031void encode (SV *self, SV *scalar) 1394void encode (JSON *self, SV *scalar)
1032 PPCODE: 1395 PPCODE:
1033 XPUSHs (encode_json (scalar, *SvJSON (self))); 1396 XPUSHs (encode_json (scalar, self));
1034 1397
1035void decode (SV *self, SV *jsonstr) 1398void decode (JSON *self, SV *jsonstr)
1036 PPCODE: 1399 PPCODE:
1037 XPUSHs (decode_json (jsonstr, *SvJSON (self))); 1400 XPUSHs (decode_json (jsonstr, self, 0));
1401
1402void decode_prefix (JSON *self, SV *jsonstr)
1403 PPCODE:
1404{
1405 UV offset;
1406 EXTEND (SP, 2);
1407 PUSHs (decode_json (jsonstr, self, &offset));
1408 PUSHs (sv_2mortal (newSVuv (offset)));
1409}
1038 1410
1039PROTOTYPES: ENABLE 1411PROTOTYPES: ENABLE
1040 1412
1041void to_json (SV *scalar) 1413void to_json (SV *scalar)
1042 PPCODE: 1414 PPCODE:
1415{
1416 JSON json = { F_DEFAULT | F_UTF8 };
1043 XPUSHs (encode_json (scalar, F_UTF8)); 1417 XPUSHs (encode_json (scalar, &json));
1418}
1044 1419
1045void from_json (SV *jsonstr) 1420void from_json (SV *jsonstr)
1046 PPCODE: 1421 PPCODE:
1422{
1423 JSON json = { F_DEFAULT | F_UTF8 };
1047 XPUSHs (decode_json (jsonstr, F_UTF8)); 1424 XPUSHs (decode_json (jsonstr, &json, 0));
1425}
1048 1426

Diff Legend

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