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 1.18 by root, Sun Mar 25 21:19:13 2007 UTC vs.
Revision 1.44 by root, Mon Jun 25 04:08:17 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 // NYI
17#define F_MAXDEPTH 0xf8000000UL 31#define F_MAXDEPTH 0xf8000000UL
18#define S_MAXDEPTH 27 32#define S_MAXDEPTH 27
19 33
20#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH)) 34#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
21 35
22// F_SELFCONVERT? <=> to_json/toJson
23// F_BLESSED? <=> { $__class__$ => }
24
25#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 36#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
26#define F_DEFAULT (13UL << S_MAXDEPTH) 37#define F_DEFAULT (9UL << S_MAXDEPTH)
27 38
28#define INIT_SIZE 32 // initial scalar size to be allocated 39#define INIT_SIZE 32 // initial scalar size to be allocated
29#define INDENT_STEP 3 // spaces per indentation level 40#define INDENT_STEP 3 // spaces per indentation level
30 41
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 42#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
33 43
34#define SB do { 44#define SB do {
35#define SE } while (0) 45#define SE } while (0)
36 46
47#if __GNUC__ >= 3
48# define expect(expr,value) __builtin_expect ((expr),(value))
49# define inline inline
50#else
51# define expect(expr,value) (expr)
52# define inline static
53#endif
54
55#define expect_false(expr) expect ((expr) != 0, 0)
56#define expect_true(expr) expect ((expr) != 0, 1)
57
37static HV *json_stash; // JSON::XS:: 58static HV *json_stash, *json_boolean_stash; // JSON::XS::
59static SV *json_true, *json_false;
38 60
39///////////////////////////////////////////////////////////////////////////// 61/////////////////////////////////////////////////////////////////////////////
40// utility functions 62// utility functions
41 63
42static UV * 64static UV *
… …
65// 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
66// case of an error. 88// case of an error.
67// we special-case "safe" characters from U+80 .. U+7FF, 89// we special-case "safe" characters from U+80 .. U+7FF,
68// but use the very good perl function to parse anything else. 90// but use the very good perl function to parse anything else.
69// note that we never call this function for a ascii codepoints 91// note that we never call this function for a ascii codepoints
70static UV 92inline UV
71decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) 93decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
72{ 94{
73 if (s[0] > 0xdf || s[0] < 0xc2) 95 if (expect_false (s[0] > 0xdf || s[0] < 0xc2))
74 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); 96 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
75 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) 97 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)
76 { 98 {
77 *clen = 2; 99 *clen = 2;
78 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 100 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
79 } 101 }
80 else 102 else
103 {
104 *clen = (STRLEN)-1;
81 return (UV)-1; 105 return (UV)-1;
106 }
82} 107}
83 108
84///////////////////////////////////////////////////////////////////////////// 109/////////////////////////////////////////////////////////////////////////////
85// encoder 110// encoder
86 111
… …
93 U32 flags; // F_* 118 U32 flags; // F_*
94 U32 indent; // indentation level 119 U32 indent; // indentation level
95 U32 maxdepth; // max. indentation/recursion level 120 U32 maxdepth; // max. indentation/recursion level
96} enc_t; 121} enc_t;
97 122
98static void 123inline void
99need (enc_t *enc, STRLEN len) 124need (enc_t *enc, STRLEN len)
100{ 125{
101 if (enc->cur + len >= enc->end) 126 if (expect_false (enc->cur + len >= enc->end))
102 { 127 {
103 STRLEN cur = enc->cur - SvPVX (enc->sv); 128 STRLEN cur = enc->cur - SvPVX (enc->sv);
104 SvGROW (enc->sv, cur + len + 1); 129 SvGROW (enc->sv, cur + len + 1);
105 enc->cur = SvPVX (enc->sv) + cur; 130 enc->cur = SvPVX (enc->sv) + cur;
106 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv); 131 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
107 } 132 }
108} 133}
109 134
110static void 135inline void
111encode_ch (enc_t *enc, char ch) 136encode_ch (enc_t *enc, char ch)
112{ 137{
113 need (enc, 1); 138 need (enc, 1);
114 *enc->cur++ = ch; 139 *enc->cur++ = ch;
115} 140}
… …
123 148
124 while (str < end) 149 while (str < end)
125 { 150 {
126 unsigned char ch = *(unsigned char *)str; 151 unsigned char ch = *(unsigned char *)str;
127 152
128 if (ch >= 0x20 && ch < 0x80) // most common case 153 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
129 { 154 {
130 if (ch == '"') // but with slow exceptions 155 if (expect_false (ch == '"')) // but with slow exceptions
131 { 156 {
132 need (enc, len += 1); 157 need (enc, len += 1);
133 *enc->cur++ = '\\'; 158 *enc->cur++ = '\\';
134 *enc->cur++ = '"'; 159 *enc->cur++ = '"';
135 } 160 }
136 else if (ch == '\\') 161 else if (expect_false (ch == '\\'))
137 { 162 {
138 need (enc, len += 1); 163 need (enc, len += 1);
139 *enc->cur++ = '\\'; 164 *enc->cur++ = '\\';
140 *enc->cur++ = '\\'; 165 *enc->cur++ = '\\';
141 } 166 }
… …
159 STRLEN clen; 184 STRLEN clen;
160 UV uch; 185 UV uch;
161 186
162 if (is_utf8) 187 if (is_utf8)
163 { 188 {
164 //uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
165 uch = decode_utf8 (str, end - str, &clen); 189 uch = decode_utf8 (str, end - str, &clen);
166 if (clen == (STRLEN)-1) 190 if (clen == (STRLEN)-1)
167 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);
168 } 192 }
169 else 193 else
… …
173 } 197 }
174 198
175 if (uch > 0x10FFFFUL) 199 if (uch > 0x10FFFFUL)
176 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);
177 201
178 if (uch < 0x80 || enc->flags & F_ASCII) 202 if (uch < 0x80 || enc->flags & F_ASCII || (enc->flags & F_LATIN1 && uch > 0xFF))
179 { 203 {
180 if (uch > 0xFFFFUL) 204 if (uch > 0xFFFFUL)
181 { 205 {
182 need (enc, len += 11); 206 need (enc, len += 11);
183 sprintf (enc->cur, "\\u%04x\\u%04x", 207 sprintf (enc->cur, "\\u%04x\\u%04x",
… …
197 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 221 *enc->cur++ = hexdigit [(uch >> 0) & 15];
198 } 222 }
199 223
200 str += clen; 224 str += clen;
201 } 225 }
226 else if (enc->flags & F_LATIN1)
227 {
228 *enc->cur++ = uch;
229 str += clen;
230 }
202 else if (is_utf8) 231 else if (is_utf8)
203 { 232 {
204 need (enc, len += clen); 233 need (enc, len += clen);
205 do 234 do
206 { 235 {
… …
208 } 237 }
209 while (--clen); 238 while (--clen);
210 } 239 }
211 else 240 else
212 { 241 {
213 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
214 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 243 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
215 ++str; 244 ++str;
216 } 245 }
217 } 246 }
218 } 247 }
… …
220 249
221 --len; 250 --len;
222 } 251 }
223} 252}
224 253
225static void 254inline void
226encode_indent (enc_t *enc) 255encode_indent (enc_t *enc)
227{ 256{
228 if (enc->flags & F_INDENT) 257 if (enc->flags & F_INDENT)
229 { 258 {
230 int spaces = enc->indent * INDENT_STEP; 259 int spaces = enc->indent * INDENT_STEP;
… …
233 memset (enc->cur, ' ', spaces); 262 memset (enc->cur, ' ', spaces);
234 enc->cur += spaces; 263 enc->cur += spaces;
235 } 264 }
236} 265}
237 266
238static void 267inline void
239encode_space (enc_t *enc) 268encode_space (enc_t *enc)
240{ 269{
241 need (enc, 1); 270 need (enc, 1);
242 encode_ch (enc, ' '); 271 encode_ch (enc, ' ');
243} 272}
244 273
245static void 274inline void
246encode_nl (enc_t *enc) 275encode_nl (enc_t *enc)
247{ 276{
248 if (enc->flags & F_INDENT) 277 if (enc->flags & F_INDENT)
249 { 278 {
250 need (enc, 1); 279 need (enc, 1);
251 encode_ch (enc, '\n'); 280 encode_ch (enc, '\n');
252 } 281 }
253} 282}
254 283
255static void 284inline void
256encode_comma (enc_t *enc) 285encode_comma (enc_t *enc)
257{ 286{
258 encode_ch (enc, ','); 287 encode_ch (enc, ',');
259 288
260 if (enc->flags & F_INDENT) 289 if (enc->flags & F_INDENT)
… …
267 296
268static void 297static void
269encode_av (enc_t *enc, AV *av) 298encode_av (enc_t *enc, AV *av)
270{ 299{
271 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)");
272 304
273 encode_ch (enc, '['); encode_nl (enc); 305 encode_ch (enc, '['); encode_nl (enc);
274 ++enc->indent; 306 ++enc->indent;
275 307
276 for (i = 0; i <= len; ++i) 308 for (i = 0; i <= len; ++i)
… …
343static void 375static void
344encode_hv (enc_t *enc, HV *hv) 376encode_hv (enc_t *enc, HV *hv)
345{ 377{
346 int count, i; 378 int count, i;
347 379
380 if (enc->indent >= enc->maxdepth)
381 croak ("data structure too deep (hit recursion limit)");
382
348 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 383 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
349 384
350 if ((count = hv_iterinit (hv))) 385 if ((count = hv_iterinit (hv)))
351 { 386 {
352 // for canonical output we have to sort by keys first 387 // for canonical output we have to sort by keys first
353 // actually, this is mostly due to the stupid so-called 388 // actually, this is mostly due to the stupid so-called
354 // security workaround added somewhere in 5.8.x. 389 // security workaround added somewhere in 5.8.x.
355 // that randomises hash orderings 390 // that randomises hash orderings
356 if (enc->flags & F_CANONICAL) 391 if (enc->flags & F_CANONICAL)
357 { 392 {
358 HE *he, *hes [count]; // if your compiler dies here, you need to enable C99 mode
359 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
360 400
361 i = 0; 401 i = 0;
362 while ((he = hv_iternext (hv))) 402 while ((he = hv_iternext (hv)))
363 { 403 {
364 hes [i++] = he; 404 hes [i++] = he;
… …
399 439
400 encode_nl (enc); 440 encode_nl (enc);
401 } 441 }
402 else 442 else
403 { 443 {
404 SV *sv;
405 HE *he = hv_iternext (hv); 444 HE *he = hv_iternext (hv);
406 445
407 for (;;) 446 for (;;)
408 { 447 {
409 encode_indent (enc); 448 encode_indent (enc);
… …
418 encode_nl (enc); 457 encode_nl (enc);
419 } 458 }
420 } 459 }
421 460
422 --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->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;
498 ENTER;
499 SAVETMPS;
500 PUSHMARK (SP);
501 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
502
503 // calling with G_SCALAR ensures that we always get a 1 reutrn value
504 // check anyways.
505 PUTBACK;
506 assert (1 == call_sv ((SV *)GvCV (to_json), G_SCALAR));
507 SPAGAIN;
508
509 encode_sv (enc, POPs);
510
511 FREETMPS;
512 LEAVE;
513 }
514 else if (enc->flags & F_ALLOW_BLESSED)
515 encode_str (enc, "null", 4, 0);
516 else
517 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
518 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
519 }
520 else if (enc->flags & F_ALLOW_BLESSED)
521 encode_str (enc, "null", 4, 0);
522 else
523 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",
524 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
525 }
526 }
527 else if (svt == SVt_PVHV)
528 encode_hv (enc, (HV *)sv);
529 else if (svt == SVt_PVAV)
530 encode_av (enc, (AV *)sv);
531 else if (svt < SVt_PVAV)
532 {
533 if (SvNIOK (sv) && SvIV (sv) == 0)
534 encode_str (enc, "false", 5, 0);
535 else if (SvNIOK (sv) && SvIV (sv) == 1)
536 encode_str (enc, "true", 4, 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, U32 flags)
478{ 617{
618 enc_t enc;
619
479 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 620 if (!(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;
483 enc.flags = flags; 623 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 (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 (!(flags & (F_ASCII | F_LATIN1 | F_UTF8)))
494 SvUTF8_on (enc.sv); 637 SvUTF8_on (enc.sv);
495
496 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
497 638
498 if (enc.flags & F_SHRINK) 639 if (enc.flags & F_SHRINK)
499 shrink (enc.sv); 640 shrink (enc.sv);
500 641
501 return enc.sv; 642 return enc.sv;
… …
513 U32 flags; // F_* 654 U32 flags; // F_*
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}
… …
890 1067
891static SV * 1068static SV *
892decode_sv (dec_t *dec) 1069decode_sv (dec_t *dec)
893{ 1070{
894 decode_ws (dec); 1071 decode_ws (dec);
1072
1073 // the beauty of JSON: you need exactly one character lookahead
1074 // to parse anything.
895 switch (*dec->cur) 1075 switch (*dec->cur)
896 { 1076 {
897 case '"': ++dec->cur; return decode_str (dec); 1077 case '"': ++dec->cur; return decode_str (dec);
898 case '[': ++dec->cur; return decode_av (dec); 1078 case '[': ++dec->cur; return decode_av (dec);
899 case '{': ++dec->cur; return decode_hv (dec); 1079 case '{': ++dec->cur; return decode_hv (dec);
… …
905 1085
906 case 't': 1086 case 't':
907 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1087 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
908 { 1088 {
909 dec->cur += 4; 1089 dec->cur += 4;
910 return newSViv (1); 1090 return SvREFCNT_inc (json_true);
911 } 1091 }
912 else 1092 else
913 ERR ("'true' expected"); 1093 ERR ("'true' expected");
914 1094
915 break; 1095 break;
916 1096
917 case 'f': 1097 case 'f':
918 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1098 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
919 { 1099 {
920 dec->cur += 5; 1100 dec->cur += 5;
921 return newSViv (0); 1101 return SvREFCNT_inc (json_false);
922 } 1102 }
923 else 1103 else
924 ERR ("'false' expected"); 1104 ERR ("'false' expected");
925 1105
926 break; 1106 break;
… …
935 ERR ("'null' expected"); 1115 ERR ("'null' expected");
936 1116
937 break; 1117 break;
938 1118
939 default: 1119 default:
940 ERR ("malformed json string, neither array, object, number, string or atom"); 1120 ERR ("malformed JSON string, neither array, object, number, string or atom");
941 break; 1121 break;
942 } 1122 }
943 1123
944fail: 1124fail:
945 return 0; 1125 return 0;
946} 1126}
947 1127
948static SV * 1128static SV *
949decode_json (SV *string, U32 flags) 1129decode_json (SV *string, U32 flags, UV *offset_return)
950{ 1130{
1131 dec_t dec;
1132 UV offset;
951 SV *sv; 1133 SV *sv;
1134
1135 SvGETMAGIC (string);
1136 SvUPGRADE (string, SVt_PV);
952 1137
953 if (flags & F_UTF8) 1138 if (flags & F_UTF8)
954 sv_utf8_downgrade (string, 0); 1139 sv_utf8_downgrade (string, 0);
955 else 1140 else
956 sv_utf8_upgrade (string); 1141 sv_utf8_upgrade (string);
957 1142
958 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1143 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
959 1144
960 dec_t dec;
961 dec.flags = flags; 1145 dec.flags = flags;
962 dec.cur = SvPVX (string); 1146 dec.cur = SvPVX (string);
963 dec.end = SvEND (string); 1147 dec.end = SvEND (string);
964 dec.err = 0; 1148 dec.err = 0;
965 dec.depth = 0; 1149 dec.depth = 0;
966 dec.maxdepth = DEC_DEPTH (dec.flags); 1150 dec.maxdepth = DEC_DEPTH (dec.flags);
967 1151
968 *SvEND (sv) = 0; // this shou[ld basically be a nop, too 1152 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
969 sv = decode_sv (&dec); 1153 sv = decode_sv (&dec);
970 1154
1155 if (!(offset_return || !sv))
1156 {
1157 // check for trailing garbage
1158 decode_ws (&dec);
1159
1160 if (*dec.cur)
1161 {
1162 dec.err = "garbage after JSON object";
1163 SvREFCNT_dec (sv);
1164 sv = 0;
1165 }
1166 }
1167
1168 if (offset_return || !sv)
1169 {
1170 offset = dec.flags & F_UTF8
1171 ? dec.cur - SvPVX (string)
1172 : utf8_distance (dec.cur, SvPVX (string));
1173
1174 if (offset_return)
1175 *offset_return = offset;
1176 }
1177
971 if (!sv) 1178 if (!sv)
972 { 1179 {
973 IV offset = dec.flags & F_UTF8
974 ? dec.cur - SvPVX (string)
975 : utf8_distance (dec.cur, SvPVX (string));
976 SV *uni = sv_newmortal (); 1180 SV *uni = sv_newmortal ();
977 1181
978 // horrible hack to silence warning inside pv_uni_display 1182 // horrible hack to silence warning inside pv_uni_display
979 COP cop = *PL_curcop; 1183 COP cop = *PL_curcop;
980 cop.cop_warnings = pWARN_NONE; 1184 cop.cop_warnings = pWARN_NONE;
… …
982 SAVEVPTR (PL_curcop); 1186 SAVEVPTR (PL_curcop);
983 PL_curcop = &cop; 1187 PL_curcop = &cop;
984 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1188 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
985 LEAVE; 1189 LEAVE;
986 1190
987 croak ("%s, at character offset %d (%s)", 1191 croak ("%s, at character offset %d [\"%s\"]",
988 dec.err, 1192 dec.err,
989 (int)offset, 1193 (int)offset,
990 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1194 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
991 } 1195 }
992 1196
… …
1004MODULE = JSON::XS PACKAGE = JSON::XS 1208MODULE = JSON::XS PACKAGE = JSON::XS
1005 1209
1006BOOT: 1210BOOT:
1007{ 1211{
1008 int i; 1212 int i;
1009
1010 memset (decode_hexdigit, 0xff, 256);
1011 1213
1012 for (i = 0; i < 256; ++i) 1214 for (i = 0; i < 256; ++i)
1013 decode_hexdigit [i] = 1215 decode_hexdigit [i] =
1014 i >= '0' && i <= '9' ? i - '0' 1216 i >= '0' && i <= '9' ? i - '0'
1015 : i >= 'a' && i <= 'f' ? i - 'a' + 10 1217 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1016 : i >= 'A' && i <= 'F' ? i - 'A' + 10 1218 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1017 : -1; 1219 : -1;
1018 1220
1019 json_stash = gv_stashpv ("JSON::XS", 1); 1221 json_stash = gv_stashpv ("JSON::XS" , 1);
1222 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1223
1224 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true );
1225 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
1020} 1226}
1021 1227
1022PROTOTYPES: DISABLE 1228PROTOTYPES: DISABLE
1023 1229
1024SV *new (char *dummy) 1230SV *new (char *dummy)
… …
1027 OUTPUT: 1233 OUTPUT:
1028 RETVAL 1234 RETVAL
1029 1235
1030SV *ascii (SV *self, int enable = 1) 1236SV *ascii (SV *self, int enable = 1)
1031 ALIAS: 1237 ALIAS:
1032 ascii = F_ASCII 1238 ascii = F_ASCII
1239 latin1 = F_LATIN1
1033 utf8 = F_UTF8 1240 utf8 = F_UTF8
1034 indent = F_INDENT 1241 indent = F_INDENT
1035 canonical = F_CANONICAL 1242 canonical = F_CANONICAL
1036 space_before = F_SPACE_BEFORE 1243 space_before = F_SPACE_BEFORE
1037 space_after = F_SPACE_AFTER 1244 space_after = F_SPACE_AFTER
1038 pretty = F_PRETTY 1245 pretty = F_PRETTY
1039 allow_nonref = F_ALLOW_NONREF 1246 allow_nonref = F_ALLOW_NONREF
1040 shrink = F_SHRINK 1247 shrink = F_SHRINK
1248 allow_blessed = F_ALLOW_BLESSED
1249 convert_blessed = F_CONV_BLESSED
1041 CODE: 1250 CODE:
1042{ 1251{
1043 UV *uv = SvJSON (self); 1252 UV *uv = SvJSON (self);
1044 if (enable) 1253 if (enable)
1045 *uv |= ix; 1254 *uv |= ix;
… …
1049 RETVAL = newSVsv (self); 1258 RETVAL = newSVsv (self);
1050} 1259}
1051 OUTPUT: 1260 OUTPUT:
1052 RETVAL 1261 RETVAL
1053 1262
1054SV *max_depth (SV *self, int max_depth = 0x80000000UL) 1263SV *max_depth (SV *self, UV max_depth = 0x80000000UL)
1055 CODE: 1264 CODE:
1056{ 1265{
1057 UV *uv = SvJSON (self); 1266 UV *uv = SvJSON (self);
1058 UV log2 = 0; 1267 UV log2 = 0;
1059 1268
… …
1073 PPCODE: 1282 PPCODE:
1074 XPUSHs (encode_json (scalar, *SvJSON (self))); 1283 XPUSHs (encode_json (scalar, *SvJSON (self)));
1075 1284
1076void decode (SV *self, SV *jsonstr) 1285void decode (SV *self, SV *jsonstr)
1077 PPCODE: 1286 PPCODE:
1078 XPUSHs (decode_json (jsonstr, *SvJSON (self))); 1287 XPUSHs (decode_json (jsonstr, *SvJSON (self), 0));
1288
1289void decode_prefix (SV *self, SV *jsonstr)
1290 PPCODE:
1291{
1292 UV offset;
1293 EXTEND (SP, 2);
1294 PUSHs (decode_json (jsonstr, *SvJSON (self), &offset));
1295 PUSHs (sv_2mortal (newSVuv (offset)));
1296}
1079 1297
1080PROTOTYPES: ENABLE 1298PROTOTYPES: ENABLE
1081 1299
1082void to_json (SV *scalar) 1300void to_json (SV *scalar)
1083 ALIAS: 1301 ALIAS:
… …
1087 1305
1088void from_json (SV *jsonstr) 1306void from_json (SV *jsonstr)
1089 ALIAS: 1307 ALIAS:
1090 jsonToObj = 0 1308 jsonToObj = 0
1091 PPCODE: 1309 PPCODE:
1092 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8)); 1310 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8, 0));
1093 1311

Diff Legend

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