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 1.19 by root, Sun Mar 25 21:52:47 2007 UTC vs.
Revision 1.52 by root, Mon Jul 2 02:57:11 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 0x00000001UL 20#define F_ASCII 0x00000001UL
21#define F_LATIN1 0x00000002UL
10#define F_UTF8 0x00000002UL 22#define F_UTF8 0x00000004UL
11#define F_INDENT 0x00000004UL 23#define F_INDENT 0x00000008UL
12#define F_CANONICAL 0x00000008UL 24#define F_CANONICAL 0x00000010UL
13#define F_SPACE_BEFORE 0x00000010UL 25#define F_SPACE_BEFORE 0x00000020UL
14#define F_SPACE_AFTER 0x00000020UL 26#define F_SPACE_AFTER 0x00000040UL
15#define F_ALLOW_NONREF 0x00000080UL 27#define F_ALLOW_NONREF 0x00000100UL
16#define F_SHRINK 0x00000100UL 28#define F_SHRINK 0x00000200UL
29#define F_ALLOW_BLESSED 0x00000400UL
30#define F_CONV_BLESSED 0x00000800UL
17#define F_MAXDEPTH 0xf8000000UL 31#define F_MAXDEPTH 0xf8000000UL
18#define S_MAXDEPTH 27 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
19 36
20#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH)) 37#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
21 38#define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE ))
22// F_SELFCONVERT? <=> to_json/toJson
23// F_BLESSED? <=> { $__class__$ => }
24 39
25#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 40#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
26#define F_DEFAULT (13UL << S_MAXDEPTH) 41#define F_DEFAULT (9UL << S_MAXDEPTH)
27 42
28#define INIT_SIZE 32 // initial scalar size to be allocated 43#define INIT_SIZE 32 // initial scalar size to be allocated
29#define INDENT_STEP 3 // spaces per indentation level 44#define INDENT_STEP 3 // spaces per indentation level
30 45
31#define UTF8_MAX_LEN 11 // for perls UTF-X: max. number of octets per character
32#define SHORT_STRING_LEN 512 // special-case strings of up to this size 46#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
33 47
34#define SB do { 48#define SB do {
35#define SE } while (0) 49#define SE } while (0)
36 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
37static 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;
68 HV *cb_sk_object;
69} JSON;
38 70
39///////////////////////////////////////////////////////////////////////////// 71/////////////////////////////////////////////////////////////////////////////
40// utility functions 72// utility functions
41 73
42static UV * 74inline void
43SvJSON (SV *sv)
44{
45 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash))
46 croak ("object is not of type JSON::XS");
47
48 return &SvUVX (SvRV (sv));
49}
50
51static void
52shrink (SV *sv) 75shrink (SV *sv)
53{ 76{
54 sv_utf8_downgrade (sv, 1); 77 sv_utf8_downgrade (sv, 1);
55 if (SvLEN (sv) > SvCUR (sv) + 1) 78 if (SvLEN (sv) > SvCUR (sv) + 1)
56 { 79 {
… …
65// decode an utf-8 character and return it, or (UV)-1 in 88// decode an utf-8 character and return it, or (UV)-1 in
66// case of an error. 89// case of an error.
67// we special-case "safe" characters from U+80 .. U+7FF, 90// we special-case "safe" characters from U+80 .. U+7FF,
68// but use the very good perl function to parse anything else. 91// but use the very good perl function to parse anything else.
69// note that we never call this function for a ascii codepoints 92// note that we never call this function for a ascii codepoints
70static UV 93inline UV
71decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) 94decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
72{ 95{
73 if (s[0] > 0xdf || s[0] < 0xc2) 96 if (expect_false (s[0] > 0xdf || s[0] < 0xc2))
74 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); 97 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
75 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) 98 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)
76 { 99 {
77 *clen = 2; 100 *clen = 2;
78 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 101 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
79 } 102 }
80 else 103 else
104 {
105 *clen = (STRLEN)-1;
81 return (UV)-1; 106 return (UV)-1;
107 }
82} 108}
83 109
84///////////////////////////////////////////////////////////////////////////// 110/////////////////////////////////////////////////////////////////////////////
85// encoder 111// encoder
86 112
… …
88typedef struct 114typedef struct
89{ 115{
90 char *cur; // SvPVX (sv) + current output position 116 char *cur; // SvPVX (sv) + current output position
91 char *end; // SvEND (sv) 117 char *end; // SvEND (sv)
92 SV *sv; // result scalar 118 SV *sv; // result scalar
93 U32 flags; // F_* 119 JSON json;
94 U32 indent; // indentation level 120 U32 indent; // indentation level
95 U32 maxdepth; // max. indentation/recursion level 121 U32 maxdepth; // max. indentation/recursion level
96} enc_t; 122} enc_t;
97 123
98static void 124inline void
99need (enc_t *enc, STRLEN len) 125need (enc_t *enc, STRLEN len)
100{ 126{
101 if (enc->cur + len >= enc->end) 127 if (expect_false (enc->cur + len >= enc->end))
102 { 128 {
103 STRLEN cur = enc->cur - SvPVX (enc->sv); 129 STRLEN cur = enc->cur - SvPVX (enc->sv);
104 SvGROW (enc->sv, cur + len + 1); 130 SvGROW (enc->sv, cur + len + 1);
105 enc->cur = SvPVX (enc->sv) + cur; 131 enc->cur = SvPVX (enc->sv) + cur;
106 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv); 132 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
107 } 133 }
108} 134}
109 135
110static void 136inline void
111encode_ch (enc_t *enc, char ch) 137encode_ch (enc_t *enc, char ch)
112{ 138{
113 need (enc, 1); 139 need (enc, 1);
114 *enc->cur++ = ch; 140 *enc->cur++ = ch;
115} 141}
… …
123 149
124 while (str < end) 150 while (str < end)
125 { 151 {
126 unsigned char ch = *(unsigned char *)str; 152 unsigned char ch = *(unsigned char *)str;
127 153
128 if (ch >= 0x20 && ch < 0x80) // most common case 154 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
129 { 155 {
130 if (ch == '"') // but with slow exceptions 156 if (expect_false (ch == '"')) // but with slow exceptions
131 { 157 {
132 need (enc, len += 1); 158 need (enc, len += 1);
133 *enc->cur++ = '\\'; 159 *enc->cur++ = '\\';
134 *enc->cur++ = '"'; 160 *enc->cur++ = '"';
135 } 161 }
136 else if (ch == '\\') 162 else if (expect_false (ch == '\\'))
137 { 163 {
138 need (enc, len += 1); 164 need (enc, len += 1);
139 *enc->cur++ = '\\'; 165 *enc->cur++ = '\\';
140 *enc->cur++ = '\\'; 166 *enc->cur++ = '\\';
141 } 167 }
… …
159 STRLEN clen; 185 STRLEN clen;
160 UV uch; 186 UV uch;
161 187
162 if (is_utf8) 188 if (is_utf8)
163 { 189 {
164 //uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
165 uch = decode_utf8 (str, end - str, &clen); 190 uch = decode_utf8 (str, end - str, &clen);
166 if (clen == (STRLEN)-1) 191 if (clen == (STRLEN)-1)
167 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str); 192 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
168 } 193 }
169 else 194 else
… …
173 } 198 }
174 199
175 if (uch > 0x10FFFFUL) 200 if (uch > 0x10FFFFUL)
176 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); 201 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
177 202
178 if (uch < 0x80 || enc->flags & F_ASCII) 203 if (uch < 0x80 || enc->json.flags & F_ASCII || (enc->json.flags & F_LATIN1 && uch > 0xFF))
179 { 204 {
180 if (uch > 0xFFFFUL) 205 if (uch > 0xFFFFUL)
181 { 206 {
182 need (enc, len += 11); 207 need (enc, len += 11);
183 sprintf (enc->cur, "\\u%04x\\u%04x", 208 sprintf (enc->cur, "\\u%04x\\u%04x",
… …
197 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 222 *enc->cur++ = hexdigit [(uch >> 0) & 15];
198 } 223 }
199 224
200 str += clen; 225 str += clen;
201 } 226 }
227 else if (enc->json.flags & F_LATIN1)
228 {
229 *enc->cur++ = uch;
230 str += clen;
231 }
202 else if (is_utf8) 232 else if (is_utf8)
203 { 233 {
204 need (enc, len += clen); 234 need (enc, len += clen);
205 do 235 do
206 { 236 {
… …
208 } 238 }
209 while (--clen); 239 while (--clen);
210 } 240 }
211 else 241 else
212 { 242 {
213 need (enc, len += UTF8_MAX_LEN - 1); // never more than 11 bytes needed 243 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
214 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 244 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
215 ++str; 245 ++str;
216 } 246 }
217 } 247 }
218 } 248 }
… …
220 250
221 --len; 251 --len;
222 } 252 }
223} 253}
224 254
225static void 255inline void
226encode_indent (enc_t *enc) 256encode_indent (enc_t *enc)
227{ 257{
228 if (enc->flags & F_INDENT) 258 if (enc->json.flags & F_INDENT)
229 { 259 {
230 int spaces = enc->indent * INDENT_STEP; 260 int spaces = enc->indent * INDENT_STEP;
231 261
232 need (enc, spaces); 262 need (enc, spaces);
233 memset (enc->cur, ' ', spaces); 263 memset (enc->cur, ' ', spaces);
234 enc->cur += spaces; 264 enc->cur += spaces;
235 } 265 }
236} 266}
237 267
238static void 268inline void
239encode_space (enc_t *enc) 269encode_space (enc_t *enc)
240{ 270{
241 need (enc, 1); 271 need (enc, 1);
242 encode_ch (enc, ' '); 272 encode_ch (enc, ' ');
243} 273}
244 274
245static void 275inline void
246encode_nl (enc_t *enc) 276encode_nl (enc_t *enc)
247{ 277{
248 if (enc->flags & F_INDENT) 278 if (enc->json.flags & F_INDENT)
249 { 279 {
250 need (enc, 1); 280 need (enc, 1);
251 encode_ch (enc, '\n'); 281 encode_ch (enc, '\n');
252 } 282 }
253} 283}
254 284
255static void 285inline void
256encode_comma (enc_t *enc) 286encode_comma (enc_t *enc)
257{ 287{
258 encode_ch (enc, ','); 288 encode_ch (enc, ',');
259 289
260 if (enc->flags & F_INDENT) 290 if (enc->json.flags & F_INDENT)
261 encode_nl (enc); 291 encode_nl (enc);
262 else if (enc->flags & F_SPACE_AFTER) 292 else if (enc->json.flags & F_SPACE_AFTER)
263 encode_space (enc); 293 encode_space (enc);
264} 294}
265 295
266static void encode_sv (enc_t *enc, SV *sv); 296static void encode_sv (enc_t *enc, SV *sv);
267 297
268static void 298static void
269encode_av (enc_t *enc, AV *av) 299encode_av (enc_t *enc, AV *av)
270{ 300{
271 int i, len = av_len (av); 301 int i, len = av_len (av);
302
303 if (enc->indent >= enc->maxdepth)
304 croak ("data structure too deep (hit recursion limit)");
272 305
273 encode_ch (enc, '['); encode_nl (enc); 306 encode_ch (enc, '['); encode_nl (enc);
274 ++enc->indent; 307 ++enc->indent;
275 308
276 for (i = 0; i <= len; ++i) 309 for (i = 0; i <= len; ++i)
… …
307 else 340 else
308 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 341 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
309 342
310 encode_ch (enc, '"'); 343 encode_ch (enc, '"');
311 344
312 if (enc->flags & F_SPACE_BEFORE) encode_space (enc); 345 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
313 encode_ch (enc, ':'); 346 encode_ch (enc, ':');
314 if (enc->flags & F_SPACE_AFTER ) encode_space (enc); 347 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
315 encode_sv (enc, HeVAL (he)); 348 encode_sv (enc, HeVAL (he));
316} 349}
317 350
318// compare hash entries, used when all keys are bytestrings 351// compare hash entries, used when all keys are bytestrings
319static int 352static int
… …
342 375
343static void 376static void
344encode_hv (enc_t *enc, HV *hv) 377encode_hv (enc_t *enc, HV *hv)
345{ 378{
346 int count, i; 379 int count, i;
380
381 if (enc->indent >= enc->maxdepth)
382 croak ("data structure too deep (hit recursion limit)");
347 383
348 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 384 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
349 385
350 if ((count = hv_iterinit (hv))) 386 if ((count = hv_iterinit (hv)))
351 { 387 {
352 // for canonical output we have to sort by keys first 388 // for canonical output we have to sort by keys first
353 // actually, this is mostly due to the stupid so-called 389 // actually, this is mostly due to the stupid so-called
354 // security workaround added somewhere in 5.8.x. 390 // security workaround added somewhere in 5.8.x.
355 // that randomises hash orderings 391 // that randomises hash orderings
356 if (enc->flags & F_CANONICAL) 392 if (enc->json.flags & F_CANONICAL)
357 { 393 {
358 HE *he, *hes [count]; // if your compiler dies here, you need to enable C99 mode
359 int fast = 1; 394 int fast = 1;
395 HE *he;
396#if defined(__BORLANDC__) || defined(_MSC_VER)
397 HE **hes = _alloca (count * sizeof (HE));
398#else
399 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
400#endif
360 401
361 i = 0; 402 i = 0;
362 while ((he = hv_iternext (hv))) 403 while ((he = hv_iternext (hv)))
363 { 404 {
364 hes [i++] = he; 405 hes [i++] = he;
… …
399 440
400 encode_nl (enc); 441 encode_nl (enc);
401 } 442 }
402 else 443 else
403 { 444 {
404 SV *sv;
405 HE *he = hv_iternext (hv); 445 HE *he = hv_iternext (hv);
406 446
407 for (;;) 447 for (;;)
408 { 448 {
409 encode_indent (enc); 449 encode_indent (enc);
… …
418 encode_nl (enc); 458 encode_nl (enc);
419 } 459 }
420 } 460 }
421 461
422 --enc->indent; encode_indent (enc); encode_ch (enc, '}'); 462 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
463}
464
465// encode objects, arrays and special \0=false and \1=true values.
466static void
467encode_rv (enc_t *enc, SV *sv)
468{
469 svtype svt;
470
471 SvGETMAGIC (sv);
472 svt = SvTYPE (sv);
473
474 if (expect_false (SvOBJECT (sv)))
475 {
476 if (SvSTASH (sv) == json_boolean_stash)
477 {
478 if (SvIV (sv))
479 encode_str (enc, "true", 4, 0);
480 else
481 encode_str (enc, "false", 5, 0);
482 }
483 else
484 {
485#if 0
486 if (0 && sv_derived_from (rv, "JSON::Literal"))
487 {
488 // not yet
489 }
490#endif
491 if (enc->json.flags & F_CONV_BLESSED)
492 {
493 // we re-bless the reference to get overload and other niceties right
494 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 1);
495
496 if (to_json)
497 {
498 dSP; ENTER; SAVETMPS; PUSHMARK (SP);
499 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
500
501 // calling with G_SCALAR ensures that we always get a 1 reutrn value
502 // check anyways.
503 PUTBACK;
504 assert (1 == call_sv ((SV *)GvCV (to_json), G_SCALAR));
505 SPAGAIN;
506
507 encode_sv (enc, POPs);
508
509 FREETMPS; LEAVE;
510 }
511 else if (enc->json.flags & F_ALLOW_BLESSED)
512 encode_str (enc, "null", 4, 0);
513 else
514 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
515 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
516 }
517 else if (enc->json.flags & F_ALLOW_BLESSED)
518 encode_str (enc, "null", 4, 0);
519 else
520 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",
521 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
522 }
523 }
524 else if (svt == SVt_PVHV)
525 encode_hv (enc, (HV *)sv);
526 else if (svt == SVt_PVAV)
527 encode_av (enc, (AV *)sv);
528 else if (svt < SVt_PVAV)
529 {
530 STRLEN len = 0;
531 char *pv = svt ? SvPV (sv, len) : 0;
532
533 if (len == 1 && *pv == '1')
534 encode_str (enc, "true", 4, 0);
535 else if (len == 1 && *pv == '0')
536 encode_str (enc, "false", 5, 0);
537 else
538 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
539 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
540 }
541 else
542 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
543 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
423} 544}
424 545
425static void 546static void
426encode_sv (enc_t *enc, SV *sv) 547encode_sv (enc_t *enc, SV *sv)
427{ 548{
… …
435 encode_str (enc, str, len, SvUTF8 (sv)); 556 encode_str (enc, str, len, SvUTF8 (sv));
436 encode_ch (enc, '"'); 557 encode_ch (enc, '"');
437 } 558 }
438 else if (SvNOKp (sv)) 559 else if (SvNOKp (sv))
439 { 560 {
561 // trust that perl will do the right thing w.r.t. JSON syntax.
440 need (enc, NV_DIG + 32); 562 need (enc, NV_DIG + 32);
441 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 563 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
442 enc->cur += strlen (enc->cur); 564 enc->cur += strlen (enc->cur);
443 } 565 }
444 else if (SvIOKp (sv)) 566 else if (SvIOKp (sv))
445 { 567 {
446 need (enc, 64); 568 // we assume we can always read an IV as a UV
569 if (SvUV (sv) & ~(UV)0x7fff)
570 {
571 // large integer, use the (rather slow) snprintf way.
572 need (enc, sizeof (UV) * 3);
447 enc->cur += 573 enc->cur +=
448 SvIsUV(sv) 574 SvIsUV(sv)
449 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 575 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
450 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 576 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
577 }
578 else
579 {
580 // optimise the "small number case"
581 // code will likely be branchless and use only a single multiplication
582 I32 i = SvIV (sv);
583 U32 u;
584 char digit, nz = 0;
585
586 need (enc, 6);
587
588 *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
589 u = i < 0 ? -i : i;
590
591 // convert to 4.28 fixed-point representation
592 u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
593
594 // now output digit by digit, each time masking out the integer part
595 // and multiplying by 5 while moving the decimal point one to the right,
596 // resulting in a net multiplication by 10.
597 // we always write the digit to memory but conditionally increment
598 // the pointer, to ease the usage of conditional move instructions.
599 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5;
600 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5;
601 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5;
602 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5;
603 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
604 }
451 } 605 }
452 else if (SvROK (sv)) 606 else if (SvROK (sv))
453 { 607 encode_rv (enc, SvRV (sv));
454 SV *rv = SvRV (sv);
455
456 if (enc->indent >= enc->maxdepth)
457 croak ("data structure too deep (hit recursion limit)");
458
459 switch (SvTYPE (rv))
460 {
461 case SVt_PVAV: encode_av (enc, (AV *)rv); break;
462 case SVt_PVHV: encode_hv (enc, (HV *)rv); break;
463
464 default:
465 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
466 SvPV_nolen (sv));
467 }
468 }
469 else if (!SvOK (sv)) 608 else if (!SvOK (sv))
470 encode_str (enc, "null", 4, 0); 609 encode_str (enc, "null", 4, 0);
471 else 610 else
472 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 611 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
473 SvPV_nolen (sv), SvFLAGS (sv)); 612 SvPV_nolen (sv), SvFLAGS (sv));
474} 613}
475 614
476static SV * 615static SV *
477encode_json (SV *scalar, U32 flags) 616encode_json (SV *scalar, JSON *json)
478{ 617{
618 enc_t enc;
619
479 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 620 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
480 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 621 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
481 622
482 enc_t enc; 623 enc.json = *json;
483 enc.flags = flags;
484 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 624 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
485 enc.cur = SvPVX (enc.sv); 625 enc.cur = SvPVX (enc.sv);
486 enc.end = SvEND (enc.sv); 626 enc.end = SvEND (enc.sv);
487 enc.indent = 0; 627 enc.indent = 0;
488 enc.maxdepth = DEC_DEPTH (flags); 628 enc.maxdepth = DEC_DEPTH (enc.json.flags);
489 629
490 SvPOK_only (enc.sv); 630 SvPOK_only (enc.sv);
491 encode_sv (&enc, scalar); 631 encode_sv (&enc, scalar);
492 632
633 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
634 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
635
493 if (!(flags & (F_ASCII | F_UTF8))) 636 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
494 SvUTF8_on (enc.sv); 637 SvUTF8_on (enc.sv);
495 638
496 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
497
498 if (enc.flags & F_SHRINK) 639 if (enc.json.flags & F_SHRINK)
499 shrink (enc.sv); 640 shrink (enc.sv);
500 641
501 return enc.sv; 642 return enc.sv;
502} 643}
503 644
… …
508typedef struct 649typedef struct
509{ 650{
510 char *cur; // current parser pointer 651 char *cur; // current parser pointer
511 char *end; // end of input string 652 char *end; // end of input string
512 const char *err; // parse error, if != 0 653 const char *err; // parse error, if != 0
513 U32 flags; // F_* 654 JSON json;
514 U32 depth; // recursion depth 655 U32 depth; // recursion depth
515 U32 maxdepth; // recursion depth limit 656 U32 maxdepth; // recursion depth limit
516} dec_t; 657} dec_t;
517 658
518static void 659inline void
519decode_ws (dec_t *dec) 660decode_ws (dec_t *dec)
520{ 661{
521 for (;;) 662 for (;;)
522 { 663 {
523 char ch = *dec->cur; 664 char ch = *dec->cur;
… …
549decode_4hex (dec_t *dec) 690decode_4hex (dec_t *dec)
550{ 691{
551 signed char d1, d2, d3, d4; 692 signed char d1, d2, d3, d4;
552 unsigned char *cur = (unsigned char *)dec->cur; 693 unsigned char *cur = (unsigned char *)dec->cur;
553 694
554 d1 = decode_hexdigit [cur [0]]; if (d1 < 0) ERR ("four hexadecimal digits expected"); 695 d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected");
555 d2 = decode_hexdigit [cur [1]]; if (d2 < 0) ERR ("four hexadecimal digits expected"); 696 d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected");
556 d3 = decode_hexdigit [cur [2]]; if (d3 < 0) ERR ("four hexadecimal digits expected"); 697 d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected");
557 d4 = decode_hexdigit [cur [3]]; if (d4 < 0) ERR ("four hexadecimal digits expected"); 698 d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected");
558 699
559 dec->cur += 4; 700 dec->cur += 4;
560 701
561 return ((UV)d1) << 12 702 return ((UV)d1) << 12
562 | ((UV)d2) << 8 703 | ((UV)d2) << 8
… …
570static SV * 711static SV *
571decode_str (dec_t *dec) 712decode_str (dec_t *dec)
572{ 713{
573 SV *sv = 0; 714 SV *sv = 0;
574 int utf8 = 0; 715 int utf8 = 0;
716 char *dec_cur = dec->cur;
575 717
576 do 718 do
577 { 719 {
578 char buf [SHORT_STRING_LEN + UTF8_MAX_LEN]; 720 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
579 char *cur = buf; 721 char *cur = buf;
580 722
581 do 723 do
582 { 724 {
583 unsigned char ch = *(unsigned char *)dec->cur++; 725 unsigned char ch = *(unsigned char *)dec_cur++;
584 726
585 if (ch == '"') 727 if (expect_false (ch == '"'))
586 { 728 {
587 --dec->cur; 729 --dec_cur;
588 break; 730 break;
589 } 731 }
590 else if (ch == '\\') 732 else if (expect_false (ch == '\\'))
591 { 733 {
592 switch (*dec->cur) 734 switch (*dec_cur)
593 { 735 {
594 case '\\': 736 case '\\':
595 case '/': 737 case '/':
596 case '"': *cur++ = *dec->cur++; break; 738 case '"': *cur++ = *dec_cur++; break;
597 739
598 case 'b': ++dec->cur; *cur++ = '\010'; break; 740 case 'b': ++dec_cur; *cur++ = '\010'; break;
599 case 't': ++dec->cur; *cur++ = '\011'; break; 741 case 't': ++dec_cur; *cur++ = '\011'; break;
600 case 'n': ++dec->cur; *cur++ = '\012'; break; 742 case 'n': ++dec_cur; *cur++ = '\012'; break;
601 case 'f': ++dec->cur; *cur++ = '\014'; break; 743 case 'f': ++dec_cur; *cur++ = '\014'; break;
602 case 'r': ++dec->cur; *cur++ = '\015'; break; 744 case 'r': ++dec_cur; *cur++ = '\015'; break;
603 745
604 case 'u': 746 case 'u':
605 { 747 {
606 UV lo, hi; 748 UV lo, hi;
607 ++dec->cur; 749 ++dec_cur;
608 750
751 dec->cur = dec_cur;
609 hi = decode_4hex (dec); 752 hi = decode_4hex (dec);
753 dec_cur = dec->cur;
610 if (hi == (UV)-1) 754 if (hi == (UV)-1)
611 goto fail; 755 goto fail;
612 756
613 // possibly a surrogate pair 757 // possibly a surrogate pair
614 if (hi >= 0xd800) 758 if (hi >= 0xd800)
615 if (hi < 0xdc00) 759 if (hi < 0xdc00)
616 { 760 {
617 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 761 if (dec_cur [0] != '\\' || dec_cur [1] != 'u')
618 ERR ("missing low surrogate character in surrogate pair"); 762 ERR ("missing low surrogate character in surrogate pair");
619 763
620 dec->cur += 2; 764 dec_cur += 2;
621 765
766 dec->cur = dec_cur;
622 lo = decode_4hex (dec); 767 lo = decode_4hex (dec);
768 dec_cur = dec->cur;
623 if (lo == (UV)-1) 769 if (lo == (UV)-1)
624 goto fail; 770 goto fail;
625 771
626 if (lo < 0xdc00 || lo >= 0xe000) 772 if (lo < 0xdc00 || lo >= 0xe000)
627 ERR ("surrogate pair expected"); 773 ERR ("surrogate pair expected");
… …
641 *cur++ = hi; 787 *cur++ = hi;
642 } 788 }
643 break; 789 break;
644 790
645 default: 791 default:
646 --dec->cur; 792 --dec_cur;
647 ERR ("illegal backslash escape sequence in string"); 793 ERR ("illegal backslash escape sequence in string");
648 } 794 }
649 } 795 }
650 else if (ch >= 0x20 && ch <= 0x7f) 796 else if (expect_true (ch >= 0x20 && ch <= 0x7f))
651 *cur++ = ch; 797 *cur++ = ch;
652 else if (ch >= 0x80) 798 else if (ch >= 0x80)
653 { 799 {
654 --dec->cur;
655
656 STRLEN clen; 800 STRLEN clen;
801 UV uch;
802
803 --dec_cur;
804
657 UV uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen); 805 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
658 if (clen == (STRLEN)-1) 806 if (clen == (STRLEN)-1)
659 ERR ("malformed UTF-8 character in JSON string"); 807 ERR ("malformed UTF-8 character in JSON string");
660 808
661 do 809 do
662 {
663 *cur++ = *dec->cur++; 810 *cur++ = *dec_cur++;
664 }
665 while (--clen); 811 while (--clen);
666 812
667 utf8 = 1; 813 utf8 = 1;
668 } 814 }
669 else if (!ch)
670 ERR ("unexpected end of string while parsing json string");
671 else 815 else
816 {
817 --dec_cur;
818
819 if (!ch)
820 ERR ("unexpected end of string while parsing JSON string");
821 else
672 ERR ("invalid character encountered"); 822 ERR ("invalid character encountered while parsing JSON string");
673 823 }
674 } 824 }
675 while (cur < buf + SHORT_STRING_LEN); 825 while (cur < buf + SHORT_STRING_LEN);
676 826
827 {
677 STRLEN len = cur - buf; 828 STRLEN len = cur - buf;
678 829
679 if (sv) 830 if (sv)
680 { 831 {
681 SvGROW (sv, SvCUR (sv) + len + 1); 832 SvGROW (sv, SvCUR (sv) + len + 1);
682 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 833 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
683 SvCUR_set (sv, SvCUR (sv) + len); 834 SvCUR_set (sv, SvCUR (sv) + len);
684 } 835 }
685 else 836 else
686 sv = newSVpvn (buf, len); 837 sv = newSVpvn (buf, len);
687 } 838 }
839 }
688 while (*dec->cur != '"'); 840 while (*dec_cur != '"');
689 841
690 ++dec->cur; 842 ++dec_cur;
691 843
692 if (sv) 844 if (sv)
693 { 845 {
694 SvPOK_only (sv); 846 SvPOK_only (sv);
695 *SvEND (sv) = 0; 847 *SvEND (sv) = 0;
… …
698 SvUTF8_on (sv); 850 SvUTF8_on (sv);
699 } 851 }
700 else 852 else
701 sv = newSVpvn ("", 0); 853 sv = newSVpvn ("", 0);
702 854
855 dec->cur = dec_cur;
703 return sv; 856 return sv;
704 857
705fail: 858fail:
859 dec->cur = dec_cur;
706 return 0; 860 return 0;
707} 861}
708 862
709static SV * 863static SV *
710decode_num (dec_t *dec) 864decode_num (dec_t *dec)
… …
768 is_nv = 1; 922 is_nv = 1;
769 } 923 }
770 924
771 if (!is_nv) 925 if (!is_nv)
772 { 926 {
773 UV uv; 927 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so
774 int numtype = grok_number (start, dec->cur - start, &uv); 928 if (*start == '-')
775 if (numtype & IS_NUMBER_IN_UV) 929 switch (dec->cur - start)
776 if (numtype & IS_NUMBER_NEG)
777 { 930 {
778 if (uv < (UV)IV_MIN) 931 case 2: return newSViv (-( start [1] - '0' * 1));
779 return newSViv (-(IV)uv); 932 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
933 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
934 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
780 } 935 }
936 else
937 switch (dec->cur - start)
938 {
939 case 1: return newSViv ( start [0] - '0' * 1);
940 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
941 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
942 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
943 }
944
945 {
946 UV uv;
947 int numtype = grok_number (start, dec->cur - start, &uv);
948 if (numtype & IS_NUMBER_IN_UV)
949 if (numtype & IS_NUMBER_NEG)
950 {
951 if (uv < (UV)IV_MIN)
952 return newSViv (-(IV)uv);
953 }
781 else 954 else
782 return newSVuv (uv); 955 return newSVuv (uv);
956
957 // here would likely be the place for bigint support
783 } 958 }
959 }
784 960
961 // if we ever support bigint or bigfloat, this is the place for bigfloat
785 return newSVnv (Atof (start)); 962 return newSVnv (Atof (start));
786 963
787fail: 964fail:
788 return 0; 965 return 0;
789} 966}
… …
833} 1010}
834 1011
835static SV * 1012static SV *
836decode_hv (dec_t *dec) 1013decode_hv (dec_t *dec)
837{ 1014{
1015 SV *sv;
838 HV *hv = newHV (); 1016 HV *hv = newHV ();
839 1017
840 DEC_INC_DEPTH; 1018 DEC_INC_DEPTH;
841 decode_ws (dec); 1019 decode_ws (dec);
842 1020
843 if (*dec->cur == '}') 1021 if (*dec->cur == '}')
844 ++dec->cur; 1022 ++dec->cur;
845 else 1023 else
846 for (;;) 1024 for (;;)
847 { 1025 {
848 SV *key, *value;
849
850 decode_ws (dec); EXPECT_CH ('"'); 1026 decode_ws (dec); EXPECT_CH ('"');
851 1027
852 key = decode_str (dec); 1028 // heuristic: assume that
853 if (!key) 1029 // a) decode_str + hv_store_ent are abysmally slow.
854 goto fail; 1030 // b) most hash keys are short, simple ascii text.
1031 // => try to "fast-match" such strings to avoid
1032 // the overhead of decode_str + hv_store_ent.
1033 {
1034 SV *value;
1035 char *p = dec->cur;
1036 char *e = p + 24; // only try up to 24 bytes
855 1037
856 decode_ws (dec); EXPECT_CH (':'); 1038 for (;;)
857
858 value = decode_sv (dec);
859 if (!value)
860 { 1039 {
1040 // the >= 0x80 is true on most architectures
1041 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1042 {
1043 // slow path, back up and use decode_str
1044 SV *key = decode_str (dec);
1045 if (!key)
1046 goto fail;
1047
1048 decode_ws (dec); EXPECT_CH (':');
1049
1050 value = decode_sv (dec);
1051 if (!value)
1052 {
1053 SvREFCNT_dec (key);
1054 goto fail;
1055 }
1056
1057 hv_store_ent (hv, key, value, 0);
861 SvREFCNT_dec (key); 1058 SvREFCNT_dec (key);
1059
1060 break;
1061 }
1062 else if (*p == '"')
1063 {
1064 // fast path, got a simple key
1065 char *key = dec->cur;
1066 int len = p - key;
1067 dec->cur = p + 1;
1068
1069 decode_ws (dec); EXPECT_CH (':');
1070
1071 value = decode_sv (dec);
1072 if (!value)
862 goto fail; 1073 goto fail;
1074
1075 hv_store (hv, key, len, value, 0);
1076
1077 break;
1078 }
1079
1080 ++p;
863 } 1081 }
864 1082 }
865 hv_store_ent (hv, key, value, 0);
866 SvREFCNT_dec (key);
867 1083
868 decode_ws (dec); 1084 decode_ws (dec);
869 1085
870 if (*dec->cur == '}') 1086 if (*dec->cur == '}')
871 { 1087 {
… …
878 1094
879 ++dec->cur; 1095 ++dec->cur;
880 } 1096 }
881 1097
882 DEC_DEC_DEPTH; 1098 DEC_DEC_DEPTH;
883 return newRV_noinc ((SV *)hv); 1099 sv = newRV_noinc ((SV *)hv);
1100
1101 // check filter callbacks
1102 if (dec->json.flags & F_HOOK)
1103 {
1104 if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
1105 {
1106 HE *cb, *he;
1107
1108 hv_iterinit (hv);
1109 he = hv_iternext (hv);
1110 hv_iterinit (hv);
1111
1112 // the next line creates a mortal sv each time its called.
1113 // might want to optimise this for common cases.
1114 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1115
1116 if (cb)
1117 {
1118 int count;
1119 ENTER; SAVETMPS;
1120
1121 dSP; PUSHMARK (SP);
1122 XPUSHs (HeVAL (he));
1123
1124 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1125
1126 if (count == 1)
1127 {
1128 sv = newSVsv (POPs);
1129 FREETMPS; LEAVE;
1130 return sv;
1131 }
1132
1133 FREETMPS; LEAVE;
1134 }
1135 }
1136
1137 if (dec->json.cb_object)
1138 {
1139 int count;
1140 ENTER; SAVETMPS;
1141
1142 dSP; ENTER; SAVETMPS; PUSHMARK (SP);
1143 XPUSHs (sv_2mortal (sv));
1144
1145 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1146
1147 if (count == 1)
1148 {
1149 sv = newSVsv (POPs);
1150 FREETMPS; LEAVE;
1151 return sv;
1152 }
1153
1154 SvREFCNT_inc (sv);
1155 FREETMPS; LEAVE;
1156 }
1157 }
1158
1159 return sv;
884 1160
885fail: 1161fail:
886 SvREFCNT_dec (hv); 1162 SvREFCNT_dec (hv);
887 DEC_DEC_DEPTH; 1163 DEC_DEC_DEPTH;
888 return 0; 1164 return 0;
… …
890 1166
891static SV * 1167static SV *
892decode_sv (dec_t *dec) 1168decode_sv (dec_t *dec)
893{ 1169{
894 decode_ws (dec); 1170 decode_ws (dec);
1171
1172 // the beauty of JSON: you need exactly one character lookahead
1173 // to parse anything.
895 switch (*dec->cur) 1174 switch (*dec->cur)
896 { 1175 {
897 case '"': ++dec->cur; return decode_str (dec); 1176 case '"': ++dec->cur; return decode_str (dec);
898 case '[': ++dec->cur; return decode_av (dec); 1177 case '[': ++dec->cur; return decode_av (dec);
899 case '{': ++dec->cur; return decode_hv (dec); 1178 case '{': ++dec->cur; return decode_hv (dec);
… …
905 1184
906 case 't': 1185 case 't':
907 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1186 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
908 { 1187 {
909 dec->cur += 4; 1188 dec->cur += 4;
910 return newSViv (1); 1189 return SvREFCNT_inc (json_true);
911 } 1190 }
912 else 1191 else
913 ERR ("'true' expected"); 1192 ERR ("'true' expected");
914 1193
915 break; 1194 break;
916 1195
917 case 'f': 1196 case 'f':
918 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1197 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
919 { 1198 {
920 dec->cur += 5; 1199 dec->cur += 5;
921 return newSViv (0); 1200 return SvREFCNT_inc (json_false);
922 } 1201 }
923 else 1202 else
924 ERR ("'false' expected"); 1203 ERR ("'false' expected");
925 1204
926 break; 1205 break;
… …
935 ERR ("'null' expected"); 1214 ERR ("'null' expected");
936 1215
937 break; 1216 break;
938 1217
939 default: 1218 default:
940 ERR ("malformed json string, neither array, object, number, string or atom"); 1219 ERR ("malformed JSON string, neither array, object, number, string or atom");
941 break; 1220 break;
942 } 1221 }
943 1222
944fail: 1223fail:
945 return 0; 1224 return 0;
946} 1225}
947 1226
948static SV * 1227static SV *
949decode_json (SV *string, U32 flags) 1228decode_json (SV *string, JSON *json, UV *offset_return)
950{ 1229{
1230 dec_t dec;
1231 UV offset;
951 SV *sv; 1232 SV *sv;
952 1233
1234 SvGETMAGIC (string);
1235 SvUPGRADE (string, SVt_PV);
1236
1237 if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags))
1238 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1239 (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags));
1240
953 if (flags & F_UTF8) 1241 if (json->flags & F_UTF8)
954 sv_utf8_downgrade (string, 0); 1242 sv_utf8_downgrade (string, 0);
955 else 1243 else
956 sv_utf8_upgrade (string); 1244 sv_utf8_upgrade (string);
957 1245
958 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1246 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
959 1247
960 dec_t dec; 1248 dec.json = *json;
961 dec.flags = flags;
962 dec.cur = SvPVX (string); 1249 dec.cur = SvPVX (string);
963 dec.end = SvEND (string); 1250 dec.end = SvEND (string);
964 dec.err = 0; 1251 dec.err = 0;
965 dec.depth = 0; 1252 dec.depth = 0;
966 dec.maxdepth = DEC_DEPTH (dec.flags); 1253 dec.maxdepth = DEC_DEPTH (dec.json.flags);
967 1254
968 *SvEND (string) = 0; // this should basically be a nop, too 1255 if (dec.json.cb_object || dec.json.cb_sk_object)
1256 dec.json.flags |= F_HOOK;
1257
1258 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
969 sv = decode_sv (&dec); 1259 sv = decode_sv (&dec);
970 1260
1261 if (!(offset_return || !sv))
1262 {
1263 // check for trailing garbage
1264 decode_ws (&dec);
1265
1266 if (*dec.cur)
1267 {
1268 dec.err = "garbage after JSON object";
1269 SvREFCNT_dec (sv);
1270 sv = 0;
1271 }
1272 }
1273
1274 if (offset_return || !sv)
1275 {
1276 offset = dec.json.flags & F_UTF8
1277 ? dec.cur - SvPVX (string)
1278 : utf8_distance (dec.cur, SvPVX (string));
1279
1280 if (offset_return)
1281 *offset_return = offset;
1282 }
1283
971 if (!sv) 1284 if (!sv)
972 { 1285 {
973 IV offset = dec.flags & F_UTF8
974 ? dec.cur - SvPVX (string)
975 : utf8_distance (dec.cur, SvPVX (string));
976 SV *uni = sv_newmortal (); 1286 SV *uni = sv_newmortal ();
977 1287
978 // horrible hack to silence warning inside pv_uni_display 1288 // horrible hack to silence warning inside pv_uni_display
979 COP cop = *PL_curcop; 1289 COP cop = *PL_curcop;
980 cop.cop_warnings = pWARN_NONE; 1290 cop.cop_warnings = pWARN_NONE;
… …
982 SAVEVPTR (PL_curcop); 1292 SAVEVPTR (PL_curcop);
983 PL_curcop = &cop; 1293 PL_curcop = &cop;
984 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1294 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
985 LEAVE; 1295 LEAVE;
986 1296
987 croak ("%s, at character offset %d (%s)", 1297 croak ("%s, at character offset %d [\"%s\"]",
988 dec.err, 1298 dec.err,
989 (int)offset, 1299 (int)offset,
990 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1300 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
991 } 1301 }
992 1302
993 sv = sv_2mortal (sv); 1303 sv = sv_2mortal (sv);
994 1304
995 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1305 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))
996 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); 1306 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
997 1307
998 return sv; 1308 return sv;
999} 1309}
1000 1310
… …
1004MODULE = JSON::XS PACKAGE = JSON::XS 1314MODULE = JSON::XS PACKAGE = JSON::XS
1005 1315
1006BOOT: 1316BOOT:
1007{ 1317{
1008 int i; 1318 int i;
1009
1010 memset (decode_hexdigit, 0xff, 256);
1011 1319
1012 for (i = 0; i < 256; ++i) 1320 for (i = 0; i < 256; ++i)
1013 decode_hexdigit [i] = 1321 decode_hexdigit [i] =
1014 i >= '0' && i <= '9' ? i - '0' 1322 i >= '0' && i <= '9' ? i - '0'
1015 : i >= 'a' && i <= 'f' ? i - 'a' + 10 1323 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1016 : i >= 'A' && i <= 'F' ? i - 'A' + 10 1324 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1017 : -1; 1325 : -1;
1018 1326
1019 json_stash = gv_stashpv ("JSON::XS", 1); 1327 json_stash = gv_stashpv ("JSON::XS" , 1);
1328 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1329
1330 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true );
1331 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
1020} 1332}
1021 1333
1022PROTOTYPES: DISABLE 1334PROTOTYPES: DISABLE
1023 1335
1024SV *new (char *dummy) 1336void new (char *klass)
1025 CODE: 1337 PPCODE:
1026 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1338{
1027 OUTPUT: 1339 SV *pv = NEWSV (0, sizeof (JSON));
1028 RETVAL 1340 SvPOK_only (pv);
1341 Zero (SvPVX (pv), 1, JSON);
1342 ((JSON *)SvPVX (pv))->flags = F_DEFAULT;
1343 XPUSHs (sv_2mortal (sv_bless (newRV_noinc (pv), json_stash)));
1344}
1029 1345
1030SV *ascii (SV *self, int enable = 1) 1346void ascii (JSON *self, int enable = 1)
1031 ALIAS: 1347 ALIAS:
1032 ascii = F_ASCII 1348 ascii = F_ASCII
1349 latin1 = F_LATIN1
1033 utf8 = F_UTF8 1350 utf8 = F_UTF8
1034 indent = F_INDENT 1351 indent = F_INDENT
1035 canonical = F_CANONICAL 1352 canonical = F_CANONICAL
1036 space_before = F_SPACE_BEFORE 1353 space_before = F_SPACE_BEFORE
1037 space_after = F_SPACE_AFTER 1354 space_after = F_SPACE_AFTER
1038 pretty = F_PRETTY 1355 pretty = F_PRETTY
1039 allow_nonref = F_ALLOW_NONREF 1356 allow_nonref = F_ALLOW_NONREF
1040 shrink = F_SHRINK 1357 shrink = F_SHRINK
1358 allow_blessed = F_ALLOW_BLESSED
1359 convert_blessed = F_CONV_BLESSED
1041 CODE: 1360 PPCODE:
1042{ 1361{
1043 UV *uv = SvJSON (self);
1044 if (enable) 1362 if (enable)
1045 *uv |= ix; 1363 self->flags |= ix;
1046 else 1364 else
1047 *uv &= ~ix; 1365 self->flags &= ~ix;
1048 1366
1049 RETVAL = newSVsv (self); 1367 XPUSHs (ST (0));
1050} 1368}
1051 OUTPUT:
1052 RETVAL
1053 1369
1054SV *max_depth (SV *self, int max_depth = 0x80000000UL) 1370void max_depth (JSON *self, UV max_depth = 0x80000000UL)
1055 CODE: 1371 PPCODE:
1056{ 1372{
1057 UV *uv = SvJSON (self);
1058 UV log2 = 0; 1373 UV log2 = 0;
1059 1374
1060 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL; 1375 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1061 1376
1062 while ((1UL << log2) < max_depth) 1377 while ((1UL << log2) < max_depth)
1063 ++log2; 1378 ++log2;
1064 1379
1065 *uv = *uv & ~F_MAXDEPTH | (log2 << S_MAXDEPTH); 1380 self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1066 1381
1067 RETVAL = newSVsv (self); 1382 XPUSHs (ST (0));
1068} 1383}
1069 OUTPUT:
1070 RETVAL
1071 1384
1072void encode (SV *self, SV *scalar) 1385void max_size (JSON *self, UV max_size = 0)
1073 PPCODE: 1386 PPCODE:
1074 XPUSHs (encode_json (scalar, *SvJSON (self))); 1387{
1388 UV log2 = 0;
1075 1389
1076void decode (SV *self, SV *jsonstr) 1390 if (max_size > 0x80000000UL) max_size = 0x80000000UL;
1391 if (max_size == 1) max_size = 2;
1392
1393 while ((1UL << log2) < max_size)
1394 ++log2;
1395
1396 self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1397
1398 XPUSHs (ST (0));
1399}
1400
1401void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1077 PPCODE: 1402 PPCODE:
1403{
1404 SvREFCNT_dec (self->cb_object);
1405 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
1406
1407 XPUSHs (ST (0));
1408}
1409
1410void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
1411 PPCODE:
1412{
1413 if (!self->cb_sk_object)
1414 self->cb_sk_object = newHV ();
1415
1416 if (SvOK (cb))
1417 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1418 else
1419 {
1420 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1421
1422 if (!HvKEYS (self->cb_sk_object))
1423 {
1424 SvREFCNT_dec (self->cb_sk_object);
1425 self->cb_sk_object = 0;
1426 }
1427 }
1428
1429 XPUSHs (ST (0));
1430}
1431
1432void encode (JSON *self, SV *scalar)
1433 PPCODE:
1434 XPUSHs (encode_json (scalar, self));
1435
1436void decode (JSON *self, SV *jsonstr)
1437 PPCODE:
1078 XPUSHs (decode_json (jsonstr, *SvJSON (self))); 1438 XPUSHs (decode_json (jsonstr, self, 0));
1439
1440void decode_prefix (JSON *self, SV *jsonstr)
1441 PPCODE:
1442{
1443 UV offset;
1444 EXTEND (SP, 2);
1445 PUSHs (decode_json (jsonstr, self, &offset));
1446 PUSHs (sv_2mortal (newSVuv (offset)));
1447}
1448
1449void DESTROY (JSON *self)
1450 CODE:
1451 SvREFCNT_dec (self->cb_sk_object);
1452 SvREFCNT_dec (self->cb_object);
1079 1453
1080PROTOTYPES: ENABLE 1454PROTOTYPES: ENABLE
1081 1455
1082void to_json (SV *scalar) 1456void to_json (SV *scalar)
1083 ALIAS:
1084 objToJson = 0
1085 PPCODE: 1457 PPCODE:
1458{
1459 JSON json = { F_DEFAULT | F_UTF8 };
1086 XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8)); 1460 XPUSHs (encode_json (scalar, &json));
1461}
1087 1462
1088void from_json (SV *jsonstr) 1463void from_json (SV *jsonstr)
1089 ALIAS:
1090 jsonToObj = 0
1091 PPCODE: 1464 PPCODE:
1465{
1466 JSON json = { F_DEFAULT | F_UTF8 };
1092 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8)); 1467 XPUSHs (decode_json (jsonstr, &json, 0));
1468}
1093 1469

Diff Legend

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