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 1.12 by root, Sat Mar 24 22:10:08 2007 UTC vs.
Revision 1.60 by root, Mon Aug 13 16:19:13 2007 UTC

1#include "EXTERN.h" 1#include "EXTERN.h"
2#include "perl.h" 2#include "perl.h"
3#include "XSUB.h" 3#include "XSUB.h"
4 4
5#include "assert.h" 5#include <assert.h>
6#include "string.h" 6#include <string.h>
7#include "stdlib.h" 7#include <stdlib.h>
8#include <stdio.h>
9#include <float.h>
8 10
11#if defined(__BORLANDC__) || defined(_MSC_VER)
12# define snprintf _snprintf // C compilers have this in stdio.h
13#endif
14
15// some old perls do not have this, try to make it work, no
16// guarentees, though. if it breaks, you get to keep the pieces.
17#ifndef UTF8_MAXBYTES
18# define UTF8_MAXBYTES 13
19#endif
20
9#define F_ASCII 0x00000001 21#define F_ASCII 0x00000001UL
22#define F_LATIN1 0x00000002UL
10#define F_UTF8 0x00000002 23#define F_UTF8 0x00000004UL
11#define F_INDENT 0x00000004 24#define F_INDENT 0x00000008UL
12#define F_CANONICAL 0x00000008 25#define F_CANONICAL 0x00000010UL
13#define F_SPACE_BEFORE 0x00000010 26#define F_SPACE_BEFORE 0x00000020UL
14#define F_SPACE_AFTER 0x00000020 27#define F_SPACE_AFTER 0x00000040UL
15#define F_ALLOW_NONREF 0x00000080 28#define F_ALLOW_NONREF 0x00000100UL
16#define F_SHRINK 0x00000100 29#define F_SHRINK 0x00000200UL
30#define F_ALLOW_BLESSED 0x00000400UL
31#define F_CONV_BLESSED 0x00000800UL
32#define F_MAXDEPTH 0xf8000000UL
33#define S_MAXDEPTH 27
34#define F_MAXSIZE 0x01f00000UL
35#define S_MAXSIZE 20
36#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
37
38#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
39#define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE ))
17 40
18#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 41#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
19#define F_DEFAULT 0 42#define F_DEFAULT (9UL << S_MAXDEPTH)
20 43
21#define INIT_SIZE 32 // initial scalar size to be allocated 44#define INIT_SIZE 32 // initial scalar size to be allocated
22#define INDENT_STEP 3 // spaces per indentation level 45#define INDENT_STEP 3 // spaces per indentation level
23 46
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 47#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
26 48
27#define SB do { 49#define SB do {
28#define SE } while (0) 50#define SE } while (0)
29 51
52#if __GNUC__ >= 3
53# define expect(expr,value) __builtin_expect ((expr),(value))
54# define inline inline
55#else
56# define expect(expr,value) (expr)
57# define inline static
58#endif
59
60#define expect_false(expr) expect ((expr) != 0, 0)
61#define expect_true(expr) expect ((expr) != 0, 1)
62
63#ifdef USE_ITHREADS
64# define JSON_SLOW 1
65# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1))
66#else
67# define JSON_SLOW 0
68# define JSON_STASH json_stash
69#endif
70
30static HV *json_stash; // JSON::XS:: 71static HV *json_stash, *json_boolean_stash; // JSON::XS::
72static SV *json_true, *json_false;
73
74typedef struct {
75 U32 flags;
76 SV *cb_object;
77 HV *cb_sk_object;
78} JSON;
31 79
32///////////////////////////////////////////////////////////////////////////// 80/////////////////////////////////////////////////////////////////////////////
33// utility functions 81// utility functions
34 82
35static UV * 83inline 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) 84shrink (SV *sv)
46{ 85{
47 sv_utf8_downgrade (sv, 1); 86 sv_utf8_downgrade (sv, 1);
48 if (SvLEN (sv) > SvCUR (sv) + 1) 87 if (SvLEN (sv) > SvCUR (sv) + 1)
49 { 88 {
… …
53 SvPV_renew (sv, SvCUR (sv) + 1); 92 SvPV_renew (sv, SvCUR (sv) + 1);
54#endif 93#endif
55 } 94 }
56} 95}
57 96
97// decode an utf-8 character and return it, or (UV)-1 in
98// case of an error.
99// we special-case "safe" characters from U+80 .. U+7FF,
100// but use the very good perl function to parse anything else.
101// note that we never call this function for a ascii codepoints
102inline UV
103decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
104{
105 if (expect_false (s[0] > 0xdf || s[0] < 0xc2))
106 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
107 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)
108 {
109 *clen = 2;
110 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
111 }
112 else
113 {
114 *clen = (STRLEN)-1;
115 return (UV)-1;
116 }
117}
118
58///////////////////////////////////////////////////////////////////////////// 119/////////////////////////////////////////////////////////////////////////////
59// encoder 120// encoder
60 121
61// structure used for encoding JSON 122// structure used for encoding JSON
62typedef struct 123typedef struct
63{ 124{
64 char *cur; // SvPVX (sv) + current output position 125 char *cur; // SvPVX (sv) + current output position
65 char *end; // SvEND (sv) 126 char *end; // SvEND (sv)
66 SV *sv; // result scalar 127 SV *sv; // result scalar
67 UV flags; // F_* 128 JSON json;
68 int indent; // indentation level 129 U32 indent; // indentation level
69 int max_depth; // max. recursion level 130 U32 maxdepth; // max. indentation/recursion level
70} enc_t; 131} enc_t;
71 132
72static void 133inline void
73need (enc_t *enc, STRLEN len) 134need (enc_t *enc, STRLEN len)
74{ 135{
75 if (enc->cur + len >= enc->end) 136 if (expect_false (enc->cur + len >= enc->end))
76 { 137 {
77 STRLEN cur = enc->cur - SvPVX (enc->sv); 138 STRLEN cur = enc->cur - SvPVX (enc->sv);
78 SvGROW (enc->sv, cur + len + 1); 139 SvGROW (enc->sv, cur + len + 1);
79 enc->cur = SvPVX (enc->sv) + cur; 140 enc->cur = SvPVX (enc->sv) + cur;
80 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv); 141 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
81 } 142 }
82} 143}
83 144
84static void 145inline void
85encode_ch (enc_t *enc, char ch) 146encode_ch (enc_t *enc, char ch)
86{ 147{
87 need (enc, 1); 148 need (enc, 1);
88 *enc->cur++ = ch; 149 *enc->cur++ = ch;
89} 150}
… …
97 158
98 while (str < end) 159 while (str < end)
99 { 160 {
100 unsigned char ch = *(unsigned char *)str; 161 unsigned char ch = *(unsigned char *)str;
101 162
102 if (ch >= 0x20 && ch < 0x80) // most common case 163 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
103 { 164 {
104 if (ch == '"') // but with slow exceptions 165 if (expect_false (ch == '"')) // but with slow exceptions
105 { 166 {
106 need (enc, len += 1); 167 need (enc, len += 1);
107 *enc->cur++ = '\\'; 168 *enc->cur++ = '\\';
108 *enc->cur++ = '"'; 169 *enc->cur++ = '"';
109 } 170 }
110 else if (ch == '\\') 171 else if (expect_false (ch == '\\'))
111 { 172 {
112 need (enc, len += 1); 173 need (enc, len += 1);
113 *enc->cur++ = '\\'; 174 *enc->cur++ = '\\';
114 *enc->cur++ = '\\'; 175 *enc->cur++ = '\\';
115 } 176 }
… …
133 STRLEN clen; 194 STRLEN clen;
134 UV uch; 195 UV uch;
135 196
136 if (is_utf8) 197 if (is_utf8)
137 { 198 {
138 uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY); 199 uch = decode_utf8 (str, end - str, &clen);
139 if (clen == (STRLEN)-1) 200 if (clen == (STRLEN)-1)
140 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str); 201 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
141 } 202 }
142 else 203 else
143 { 204 {
… …
146 } 207 }
147 208
148 if (uch > 0x10FFFFUL) 209 if (uch > 0x10FFFFUL)
149 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); 210 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
150 211
151 if (uch < 0x80 || enc->flags & F_ASCII) 212 if (uch < 0x80 || enc->json.flags & F_ASCII || (enc->json.flags & F_LATIN1 && uch > 0xFF))
152 { 213 {
153 if (uch > 0xFFFFUL) 214 if (uch > 0xFFFFUL)
154 { 215 {
155 need (enc, len += 11); 216 need (enc, len += 11);
156 sprintf (enc->cur, "\\u%04x\\u%04x", 217 sprintf (enc->cur, "\\u%04x\\u%04x",
… …
170 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 231 *enc->cur++ = hexdigit [(uch >> 0) & 15];
171 } 232 }
172 233
173 str += clen; 234 str += clen;
174 } 235 }
236 else if (enc->json.flags & F_LATIN1)
237 {
238 *enc->cur++ = uch;
239 str += clen;
240 }
175 else if (is_utf8) 241 else if (is_utf8)
176 { 242 {
177 need (enc, len += clen); 243 need (enc, len += clen);
178 do 244 do
179 { 245 {
… …
181 } 247 }
182 while (--clen); 248 while (--clen);
183 } 249 }
184 else 250 else
185 { 251 {
186 need (enc, len += UTF8_MAX_LEN - 1); // never more than 11 bytes needed 252 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
187 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 253 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
188 ++str; 254 ++str;
189 } 255 }
190 } 256 }
191 } 257 }
… …
193 259
194 --len; 260 --len;
195 } 261 }
196} 262}
197 263
198static void 264inline void
199encode_indent (enc_t *enc) 265encode_indent (enc_t *enc)
200{ 266{
201 if (enc->flags & F_INDENT) 267 if (enc->json.flags & F_INDENT)
202 { 268 {
203 int spaces = enc->indent * INDENT_STEP; 269 int spaces = enc->indent * INDENT_STEP;
204 270
205 need (enc, spaces); 271 need (enc, spaces);
206 memset (enc->cur, ' ', spaces); 272 memset (enc->cur, ' ', spaces);
207 enc->cur += spaces; 273 enc->cur += spaces;
208 } 274 }
209} 275}
210 276
211static void 277inline void
212encode_space (enc_t *enc) 278encode_space (enc_t *enc)
213{ 279{
214 need (enc, 1); 280 need (enc, 1);
215 encode_ch (enc, ' '); 281 encode_ch (enc, ' ');
216} 282}
217 283
218static void 284inline void
219encode_nl (enc_t *enc) 285encode_nl (enc_t *enc)
220{ 286{
221 if (enc->flags & F_INDENT) 287 if (enc->json.flags & F_INDENT)
222 { 288 {
223 need (enc, 1); 289 need (enc, 1);
224 encode_ch (enc, '\n'); 290 encode_ch (enc, '\n');
225 } 291 }
226} 292}
227 293
228static void 294inline void
229encode_comma (enc_t *enc) 295encode_comma (enc_t *enc)
230{ 296{
231 encode_ch (enc, ','); 297 encode_ch (enc, ',');
232 298
233 if (enc->flags & F_INDENT) 299 if (enc->json.flags & F_INDENT)
234 encode_nl (enc); 300 encode_nl (enc);
235 else if (enc->flags & F_SPACE_AFTER) 301 else if (enc->json.flags & F_SPACE_AFTER)
236 encode_space (enc); 302 encode_space (enc);
237} 303}
238 304
239static void encode_sv (enc_t *enc, SV *sv); 305static void encode_sv (enc_t *enc, SV *sv);
240 306
241static void 307static void
242encode_av (enc_t *enc, AV *av) 308encode_av (enc_t *enc, AV *av)
243{ 309{
244 int i, len = av_len (av); 310 int i, len = av_len (av);
245 311
312 if (enc->indent >= enc->maxdepth)
313 croak ("data structure too deep (hit recursion limit)");
314
246 encode_ch (enc, '['); encode_nl (enc); 315 encode_ch (enc, '['); encode_nl (enc);
247 ++enc->indent; 316 ++enc->indent;
248 317
249 for (i = 0; i <= len; ++i) 318 for (i = 0; i <= len; ++i)
250 { 319 {
320 SV **svp = av_fetch (av, i, 0);
321
251 encode_indent (enc); 322 encode_indent (enc);
252 encode_sv (enc, *av_fetch (av, i, 0)); 323
324 if (svp)
325 encode_sv (enc, *svp);
326 else
327 encode_str (enc, "null", 4, 0);
253 328
254 if (i < len) 329 if (i < len)
255 encode_comma (enc); 330 encode_comma (enc);
256 } 331 }
257 332
… …
280 else 355 else
281 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 356 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
282 357
283 encode_ch (enc, '"'); 358 encode_ch (enc, '"');
284 359
285 if (enc->flags & F_SPACE_BEFORE) encode_space (enc); 360 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
286 encode_ch (enc, ':'); 361 encode_ch (enc, ':');
287 if (enc->flags & F_SPACE_AFTER ) encode_space (enc); 362 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
288 encode_sv (enc, HeVAL (he)); 363 encode_sv (enc, HeVAL (he));
289} 364}
290 365
291// compare hash entries, used when all keys are bytestrings 366// compare hash entries, used when all keys are bytestrings
292static int 367static int
… …
315 390
316static void 391static void
317encode_hv (enc_t *enc, HV *hv) 392encode_hv (enc_t *enc, HV *hv)
318{ 393{
319 int count, i; 394 int count, i;
395
396 if (enc->indent >= enc->maxdepth)
397 croak ("data structure too deep (hit recursion limit)");
320 398
321 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 399 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
322 400
323 if ((count = hv_iterinit (hv))) 401 if ((count = hv_iterinit (hv)))
324 { 402 {
325 // for canonical output we have to sort by keys first 403 // for canonical output we have to sort by keys first
326 // actually, this is mostly due to the stupid so-called 404 // actually, this is mostly due to the stupid so-called
327 // security workaround added somewhere in 5.8.x. 405 // security workaround added somewhere in 5.8.x.
328 // that randomises hash orderings 406 // that randomises hash orderings
329 if (enc->flags & F_CANONICAL) 407 if (enc->json.flags & F_CANONICAL)
330 { 408 {
331 HE *he, *hes [count]; // if your compiler dies here, you need to enable C99 mode
332 int fast = 1; 409 int fast = 1;
410 HE *he;
411#if defined(__BORLANDC__) || defined(_MSC_VER)
412 HE **hes = _alloca (count * sizeof (HE));
413#else
414 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
415#endif
333 416
334 i = 0; 417 i = 0;
335 while ((he = hv_iternext (hv))) 418 while ((he = hv_iternext (hv)))
336 { 419 {
337 hes [i++] = he; 420 hes [i++] = he;
… …
372 455
373 encode_nl (enc); 456 encode_nl (enc);
374 } 457 }
375 else 458 else
376 { 459 {
377 SV *sv;
378 HE *he = hv_iternext (hv); 460 HE *he = hv_iternext (hv);
379 461
380 for (;;) 462 for (;;)
381 { 463 {
382 encode_indent (enc); 464 encode_indent (enc);
… …
391 encode_nl (enc); 473 encode_nl (enc);
392 } 474 }
393 } 475 }
394 476
395 --enc->indent; encode_indent (enc); encode_ch (enc, '}'); 477 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
478}
479
480// encode objects, arrays and special \0=false and \1=true values.
481static void
482encode_rv (enc_t *enc, SV *sv)
483{
484 svtype svt;
485
486 SvGETMAGIC (sv);
487 svt = SvTYPE (sv);
488
489 if (expect_false (SvOBJECT (sv)))
490 {
491 HV *stash = !JSON_SLOW || json_boolean_stash
492 ? json_boolean_stash
493 : gv_stashpv ("JSON::XS::Boolean", 1);
494
495 if (SvSTASH (sv) == stash)
496 {
497 if (SvIV (sv))
498 encode_str (enc, "true", 4, 0);
499 else
500 encode_str (enc, "false", 5, 0);
501 }
502 else
503 {
504#if 0
505 if (0 && sv_derived_from (rv, "JSON::Literal"))
506 {
507 // not yet
508 }
509#endif
510 if (enc->json.flags & F_CONV_BLESSED)
511 {
512 // we re-bless the reference to get overload and other niceties right
513 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
514
515 if (to_json)
516 {
517 dSP;
518
519 ENTER; SAVETMPS; PUSHMARK (SP);
520 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
521
522 // calling with G_SCALAR ensures that we always get a 1 return value
523 PUTBACK;
524 call_sv ((SV *)GvCV (to_json), G_SCALAR);
525 SPAGAIN;
526
527 // catch this surprisingly common error
528 if (SvROK (TOPs) && SvRV (TOPs) == sv)
529 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
530
531 sv = POPs;
532 PUTBACK;
533
534 encode_sv (enc, sv);
535
536 FREETMPS; LEAVE;
537 }
538 else if (enc->json.flags & F_ALLOW_BLESSED)
539 encode_str (enc, "null", 4, 0);
540 else
541 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
542 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
543 }
544 else if (enc->json.flags & F_ALLOW_BLESSED)
545 encode_str (enc, "null", 4, 0);
546 else
547 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",
548 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
549 }
550 }
551 else if (svt == SVt_PVHV)
552 encode_hv (enc, (HV *)sv);
553 else if (svt == SVt_PVAV)
554 encode_av (enc, (AV *)sv);
555 else if (svt < SVt_PVAV)
556 {
557 STRLEN len = 0;
558 char *pv = svt ? SvPV (sv, len) : 0;
559
560 if (len == 1 && *pv == '1')
561 encode_str (enc, "true", 4, 0);
562 else if (len == 1 && *pv == '0')
563 encode_str (enc, "false", 5, 0);
564 else
565 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
566 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
567 }
568 else
569 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
570 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
396} 571}
397 572
398static void 573static void
399encode_sv (enc_t *enc, SV *sv) 574encode_sv (enc_t *enc, SV *sv)
400{ 575{
… …
408 encode_str (enc, str, len, SvUTF8 (sv)); 583 encode_str (enc, str, len, SvUTF8 (sv));
409 encode_ch (enc, '"'); 584 encode_ch (enc, '"');
410 } 585 }
411 else if (SvNOKp (sv)) 586 else if (SvNOKp (sv))
412 { 587 {
588 // trust that perl will do the right thing w.r.t. JSON syntax.
413 need (enc, NV_DIG + 32); 589 need (enc, NV_DIG + 32);
414 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 590 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
415 enc->cur += strlen (enc->cur); 591 enc->cur += strlen (enc->cur);
416 } 592 }
417 else if (SvIOKp (sv)) 593 else if (SvIOKp (sv))
418 { 594 {
419 need (enc, 64); 595 // we assume we can always read an IV as a UV
596 if (SvUV (sv) & ~(UV)0x7fff)
597 {
598 // large integer, use the (rather slow) snprintf way.
599 need (enc, sizeof (UV) * 3);
420 enc->cur += 600 enc->cur +=
421 SvIsUV(sv) 601 SvIsUV(sv)
422 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 602 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
423 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 603 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
604 }
605 else
606 {
607 // optimise the "small number case"
608 // code will likely be branchless and use only a single multiplication
609 I32 i = SvIV (sv);
610 U32 u;
611 char digit, nz = 0;
612
613 need (enc, 6);
614
615 *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
616 u = i < 0 ? -i : i;
617
618 // convert to 4.28 fixed-point representation
619 u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
620
621 // now output digit by digit, each time masking out the integer part
622 // and multiplying by 5 while moving the decimal point one to the right,
623 // resulting in a net multiplication by 10.
624 // we always write the digit to memory but conditionally increment
625 // the pointer, to ease the usage of conditional move instructions.
626 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5;
627 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5;
628 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5;
629 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5;
630 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
631 }
424 } 632 }
425 else if (SvROK (sv)) 633 else if (SvROK (sv))
426 { 634 encode_rv (enc, SvRV (sv));
427 SV *rv = SvRV (sv);
428
429 if (enc->indent >= enc->max_depth)
430 croak ("data structure too deep (hit recursion limit)");
431
432 switch (SvTYPE (rv))
433 {
434 case SVt_PVAV: encode_av (enc, (AV *)rv); break;
435 case SVt_PVHV: encode_hv (enc, (HV *)rv); break;
436
437 default:
438 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
439 SvPV_nolen (sv));
440 }
441 }
442 else if (!SvOK (sv)) 635 else if (!SvOK (sv))
443 encode_str (enc, "null", 4, 0); 636 encode_str (enc, "null", 4, 0);
444 else 637 else
445 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 638 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
446 SvPV_nolen (sv), SvFLAGS (sv)); 639 SvPV_nolen (sv), SvFLAGS (sv));
447} 640}
448 641
449static SV * 642static SV *
450encode_json (SV *scalar, UV flags) 643encode_json (SV *scalar, JSON *json)
451{ 644{
645 enc_t enc;
646
452 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 647 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
453 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 648 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
454 649
455 enc_t enc; 650 enc.json = *json;
456 enc.flags = flags;
457 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 651 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
458 enc.cur = SvPVX (enc.sv); 652 enc.cur = SvPVX (enc.sv);
459 enc.end = SvEND (enc.sv); 653 enc.end = SvEND (enc.sv);
460 enc.indent = 0; 654 enc.indent = 0;
461 enc.max_depth = 0x7fffffffUL; 655 enc.maxdepth = DEC_DEPTH (enc.json.flags);
462 656
463 SvPOK_only (enc.sv); 657 SvPOK_only (enc.sv);
464 encode_sv (&enc, scalar); 658 encode_sv (&enc, scalar);
465 659
660 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
661 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
662
466 if (!(flags & (F_ASCII | F_UTF8))) 663 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
467 SvUTF8_on (enc.sv); 664 SvUTF8_on (enc.sv);
468 665
469 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
470
471 if (enc.flags & F_SHRINK) 666 if (enc.json.flags & F_SHRINK)
472 shrink (enc.sv); 667 shrink (enc.sv);
473 668
474 return enc.sv; 669 return enc.sv;
475} 670}
476 671
… …
481typedef struct 676typedef struct
482{ 677{
483 char *cur; // current parser pointer 678 char *cur; // current parser pointer
484 char *end; // end of input string 679 char *end; // end of input string
485 const char *err; // parse error, if != 0 680 const char *err; // parse error, if != 0
486 UV flags; // F_* 681 JSON json;
682 U32 depth; // recursion depth
683 U32 maxdepth; // recursion depth limit
487} dec_t; 684} dec_t;
488 685
489static void 686inline void
490decode_ws (dec_t *dec) 687decode_ws (dec_t *dec)
491{ 688{
492 for (;;) 689 for (;;)
493 { 690 {
494 char ch = *dec->cur; 691 char ch = *dec->cur;
… …
500 ++dec->cur; 697 ++dec->cur;
501 } 698 }
502} 699}
503 700
504#define ERR(reason) SB dec->err = reason; goto fail; SE 701#define ERR(reason) SB dec->err = reason; goto fail; SE
702
505#define EXPECT_CH(ch) SB \ 703#define EXPECT_CH(ch) SB \
506 if (*dec->cur != ch) \ 704 if (*dec->cur != ch) \
507 ERR (# ch " expected"); \ 705 ERR (# ch " expected"); \
508 ++dec->cur; \ 706 ++dec->cur; \
509 SE 707 SE
510 708
709#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)")
710#define DEC_DEC_DEPTH --dec->depth
711
511static SV *decode_sv (dec_t *dec); 712static SV *decode_sv (dec_t *dec);
512 713
513static signed char decode_hexdigit[256]; 714static signed char decode_hexdigit[256];
514 715
515static UV 716static UV
516decode_4hex (dec_t *dec) 717decode_4hex (dec_t *dec)
517{ 718{
518 signed char d1, d2, d3, d4; 719 signed char d1, d2, d3, d4;
519 unsigned char *cur = (unsigned char *)dec->cur; 720 unsigned char *cur = (unsigned char *)dec->cur;
520 721
521 d1 = decode_hexdigit [cur [0]]; if (d1 < 0) ERR ("four hexadecimal digits expected"); 722 d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected");
522 d2 = decode_hexdigit [cur [1]]; if (d2 < 0) ERR ("four hexadecimal digits expected"); 723 d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected");
523 d3 = decode_hexdigit [cur [2]]; if (d3 < 0) ERR ("four hexadecimal digits expected"); 724 d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected");
524 d4 = decode_hexdigit [cur [3]]; if (d4 < 0) ERR ("four hexadecimal digits expected"); 725 d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected");
525 726
526 dec->cur += 4; 727 dec->cur += 4;
527 728
528 return ((UV)d1) << 12 729 return ((UV)d1) << 12
529 | ((UV)d2) << 8 730 | ((UV)d2) << 8
… …
537static SV * 738static SV *
538decode_str (dec_t *dec) 739decode_str (dec_t *dec)
539{ 740{
540 SV *sv = 0; 741 SV *sv = 0;
541 int utf8 = 0; 742 int utf8 = 0;
743 char *dec_cur = dec->cur;
542 744
543 do 745 do
544 { 746 {
545 char buf [SHORT_STRING_LEN + UTF8_MAX_LEN]; 747 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
546 char *cur = buf; 748 char *cur = buf;
547 749
548 do 750 do
549 { 751 {
550 unsigned char ch = *(unsigned char *)dec->cur++; 752 unsigned char ch = *(unsigned char *)dec_cur++;
551 753
552 if (ch == '"') 754 if (expect_false (ch == '"'))
553 { 755 {
554 --dec->cur; 756 --dec_cur;
555 break; 757 break;
556 } 758 }
557 else if (ch == '\\') 759 else if (expect_false (ch == '\\'))
558 { 760 {
559 switch (*dec->cur) 761 switch (*dec_cur)
560 { 762 {
561 case '\\': 763 case '\\':
562 case '/': 764 case '/':
563 case '"': *cur++ = *dec->cur++; break; 765 case '"': *cur++ = *dec_cur++; break;
564 766
565 case 'b': ++dec->cur; *cur++ = '\010'; break; 767 case 'b': ++dec_cur; *cur++ = '\010'; break;
566 case 't': ++dec->cur; *cur++ = '\011'; break; 768 case 't': ++dec_cur; *cur++ = '\011'; break;
567 case 'n': ++dec->cur; *cur++ = '\012'; break; 769 case 'n': ++dec_cur; *cur++ = '\012'; break;
568 case 'f': ++dec->cur; *cur++ = '\014'; break; 770 case 'f': ++dec_cur; *cur++ = '\014'; break;
569 case 'r': ++dec->cur; *cur++ = '\015'; break; 771 case 'r': ++dec_cur; *cur++ = '\015'; break;
570 772
571 case 'u': 773 case 'u':
572 { 774 {
573 UV lo, hi; 775 UV lo, hi;
574 ++dec->cur; 776 ++dec_cur;
575 777
778 dec->cur = dec_cur;
576 hi = decode_4hex (dec); 779 hi = decode_4hex (dec);
780 dec_cur = dec->cur;
577 if (hi == (UV)-1) 781 if (hi == (UV)-1)
578 goto fail; 782 goto fail;
579 783
580 // possibly a surrogate pair 784 // possibly a surrogate pair
581 if (hi >= 0xd800) 785 if (hi >= 0xd800)
582 if (hi < 0xdc00) 786 if (hi < 0xdc00)
583 { 787 {
584 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 788 if (dec_cur [0] != '\\' || dec_cur [1] != 'u')
585 ERR ("missing low surrogate character in surrogate pair"); 789 ERR ("missing low surrogate character in surrogate pair");
586 790
587 dec->cur += 2; 791 dec_cur += 2;
588 792
793 dec->cur = dec_cur;
589 lo = decode_4hex (dec); 794 lo = decode_4hex (dec);
795 dec_cur = dec->cur;
590 if (lo == (UV)-1) 796 if (lo == (UV)-1)
591 goto fail; 797 goto fail;
592 798
593 if (lo < 0xdc00 || lo >= 0xe000) 799 if (lo < 0xdc00 || lo >= 0xe000)
594 ERR ("surrogate pair expected"); 800 ERR ("surrogate pair expected");
… …
608 *cur++ = hi; 814 *cur++ = hi;
609 } 815 }
610 break; 816 break;
611 817
612 default: 818 default:
613 --dec->cur; 819 --dec_cur;
614 ERR ("illegal backslash escape sequence in string"); 820 ERR ("illegal backslash escape sequence in string");
615 } 821 }
616 } 822 }
617 else if (ch >= 0x20 && ch <= 0x7f) 823 else if (expect_true (ch >= 0x20 && ch <= 0x7f))
618 *cur++ = ch; 824 *cur++ = ch;
619 else if (ch >= 0x80) 825 else if (ch >= 0x80)
620 { 826 {
621 --dec->cur;
622
623 STRLEN clen; 827 STRLEN clen;
624 UV uch = utf8n_to_uvuni (dec->cur, dec->end - dec->cur, &clen, UTF8_CHECK_ONLY); 828 UV uch;
829
830 --dec_cur;
831
832 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
625 if (clen == (STRLEN)-1) 833 if (clen == (STRLEN)-1)
626 ERR ("malformed UTF-8 character in JSON string"); 834 ERR ("malformed UTF-8 character in JSON string");
627 835
628 do 836 do
629 {
630 *cur++ = *dec->cur++; 837 *cur++ = *dec_cur++;
631 }
632 while (--clen); 838 while (--clen);
633 839
634 utf8 = 1; 840 utf8 = 1;
635 } 841 }
636 else if (!ch)
637 ERR ("unexpected end of string while parsing json string");
638 else 842 else
843 {
844 --dec_cur;
845
846 if (!ch)
847 ERR ("unexpected end of string while parsing JSON string");
848 else
639 ERR ("invalid character encountered"); 849 ERR ("invalid character encountered while parsing JSON string");
640 850 }
641 } 851 }
642 while (cur < buf + SHORT_STRING_LEN); 852 while (cur < buf + SHORT_STRING_LEN);
643 853
854 {
644 STRLEN len = cur - buf; 855 STRLEN len = cur - buf;
645 856
646 if (sv) 857 if (sv)
647 { 858 {
648 SvGROW (sv, SvCUR (sv) + len + 1); 859 SvGROW (sv, SvCUR (sv) + len + 1);
649 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 860 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
650 SvCUR_set (sv, SvCUR (sv) + len); 861 SvCUR_set (sv, SvCUR (sv) + len);
651 } 862 }
652 else 863 else
653 sv = newSVpvn (buf, len); 864 sv = newSVpvn (buf, len);
654 } 865 }
866 }
655 while (*dec->cur != '"'); 867 while (*dec_cur != '"');
656 868
657 ++dec->cur; 869 ++dec_cur;
658 870
659 if (sv) 871 if (sv)
660 { 872 {
661 SvPOK_only (sv); 873 SvPOK_only (sv);
662 *SvEND (sv) = 0; 874 *SvEND (sv) = 0;
… …
665 SvUTF8_on (sv); 877 SvUTF8_on (sv);
666 } 878 }
667 else 879 else
668 sv = newSVpvn ("", 0); 880 sv = newSVpvn ("", 0);
669 881
882 dec->cur = dec_cur;
670 return sv; 883 return sv;
671 884
672fail: 885fail:
886 dec->cur = dec_cur;
673 return 0; 887 return 0;
674} 888}
675 889
676static SV * 890static SV *
677decode_num (dec_t *dec) 891decode_num (dec_t *dec)
… …
735 is_nv = 1; 949 is_nv = 1;
736 } 950 }
737 951
738 if (!is_nv) 952 if (!is_nv)
739 { 953 {
740 UV uv; 954 int len = dec->cur - start;
741 int numtype = grok_number (start, dec->cur - start, &uv); 955
742 if (numtype & IS_NUMBER_IN_UV) 956 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so
743 if (numtype & IS_NUMBER_NEG) 957 if (*start == '-')
958 switch (len)
744 { 959 {
745 if (uv < (UV)IV_MIN) 960 case 2: return newSViv (-( start [1] - '0' * 1));
746 return newSViv (-(IV)uv); 961 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
962 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
963 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
747 } 964 }
965 else
966 switch (len)
967 {
968 case 1: return newSViv ( start [0] - '0' * 1);
969 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
970 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
971 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
972 }
973
974 {
975 UV uv;
976 int numtype = grok_number (start, len, &uv);
977 if (numtype & IS_NUMBER_IN_UV)
978 if (numtype & IS_NUMBER_NEG)
979 {
980 if (uv < (UV)IV_MIN)
981 return newSViv (-(IV)uv);
982 }
748 else 983 else
749 return newSVuv (uv); 984 return newSVuv (uv);
750 } 985 }
751 986
987 len -= *start == '-' ? 1 : 0;
988
989 // does not fit into IV or UV, try NV
990 if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len)
991 #if defined (LDBL_DIG)
992 || (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len)
993 #endif
994 )
995 // fits into NV without loss of precision
996 return newSVnv (Atof (start));
997
998 // everything else fails, convert it to a string
999 return newSVpvn (start, dec->cur - start);
1000 }
1001
1002 // loss of precision here
752 return newSVnv (Atof (start)); 1003 return newSVnv (Atof (start));
753 1004
754fail: 1005fail:
755 return 0; 1006 return 0;
756} 1007}
… …
758static SV * 1009static SV *
759decode_av (dec_t *dec) 1010decode_av (dec_t *dec)
760{ 1011{
761 AV *av = newAV (); 1012 AV *av = newAV ();
762 1013
1014 DEC_INC_DEPTH;
763 decode_ws (dec); 1015 decode_ws (dec);
1016
764 if (*dec->cur == ']') 1017 if (*dec->cur == ']')
765 ++dec->cur; 1018 ++dec->cur;
766 else 1019 else
767 for (;;) 1020 for (;;)
768 { 1021 {
… …
786 ERR (", or ] expected while parsing array"); 1039 ERR (", or ] expected while parsing array");
787 1040
788 ++dec->cur; 1041 ++dec->cur;
789 } 1042 }
790 1043
1044 DEC_DEC_DEPTH;
791 return newRV_noinc ((SV *)av); 1045 return newRV_noinc ((SV *)av);
792 1046
793fail: 1047fail:
794 SvREFCNT_dec (av); 1048 SvREFCNT_dec (av);
1049 DEC_DEC_DEPTH;
795 return 0; 1050 return 0;
796} 1051}
797 1052
798static SV * 1053static SV *
799decode_hv (dec_t *dec) 1054decode_hv (dec_t *dec)
800{ 1055{
1056 SV *sv;
801 HV *hv = newHV (); 1057 HV *hv = newHV ();
802 1058
1059 DEC_INC_DEPTH;
803 decode_ws (dec); 1060 decode_ws (dec);
1061
804 if (*dec->cur == '}') 1062 if (*dec->cur == '}')
805 ++dec->cur; 1063 ++dec->cur;
806 else 1064 else
807 for (;;) 1065 for (;;)
808 { 1066 {
809 SV *key, *value;
810
811 decode_ws (dec); EXPECT_CH ('"'); 1067 decode_ws (dec); EXPECT_CH ('"');
812 1068
813 key = decode_str (dec); 1069 // heuristic: assume that
814 if (!key) 1070 // a) decode_str + hv_store_ent are abysmally slow.
815 goto fail; 1071 // b) most hash keys are short, simple ascii text.
1072 // => try to "fast-match" such strings to avoid
1073 // the overhead of decode_str + hv_store_ent.
1074 {
1075 SV *value;
1076 char *p = dec->cur;
1077 char *e = p + 24; // only try up to 24 bytes
816 1078
817 decode_ws (dec); EXPECT_CH (':'); 1079 for (;;)
818
819 value = decode_sv (dec);
820 if (!value)
821 { 1080 {
1081 // the >= 0x80 is true on most architectures
1082 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1083 {
1084 // slow path, back up and use decode_str
1085 SV *key = decode_str (dec);
1086 if (!key)
1087 goto fail;
1088
1089 decode_ws (dec); EXPECT_CH (':');
1090
1091 value = decode_sv (dec);
1092 if (!value)
1093 {
1094 SvREFCNT_dec (key);
1095 goto fail;
1096 }
1097
1098 hv_store_ent (hv, key, value, 0);
822 SvREFCNT_dec (key); 1099 SvREFCNT_dec (key);
1100
1101 break;
1102 }
1103 else if (*p == '"')
1104 {
1105 // fast path, got a simple key
1106 char *key = dec->cur;
1107 int len = p - key;
1108 dec->cur = p + 1;
1109
1110 decode_ws (dec); EXPECT_CH (':');
1111
1112 value = decode_sv (dec);
1113 if (!value)
823 goto fail; 1114 goto fail;
1115
1116 hv_store (hv, key, len, value, 0);
1117
1118 break;
1119 }
1120
1121 ++p;
824 } 1122 }
825 1123 }
826 //TODO: optimise
827 hv_store_ent (hv, key, value, 0);
828 1124
829 decode_ws (dec); 1125 decode_ws (dec);
830 1126
831 if (*dec->cur == '}') 1127 if (*dec->cur == '}')
832 { 1128 {
… …
838 ERR (", or } expected while parsing object/hash"); 1134 ERR (", or } expected while parsing object/hash");
839 1135
840 ++dec->cur; 1136 ++dec->cur;
841 } 1137 }
842 1138
1139 DEC_DEC_DEPTH;
843 return newRV_noinc ((SV *)hv); 1140 sv = newRV_noinc ((SV *)hv);
1141
1142 // check filter callbacks
1143 if (dec->json.flags & F_HOOK)
1144 {
1145 if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
1146 {
1147 HE *cb, *he;
1148
1149 hv_iterinit (hv);
1150 he = hv_iternext (hv);
1151 hv_iterinit (hv);
1152
1153 // the next line creates a mortal sv each time its called.
1154 // might want to optimise this for common cases.
1155 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1156
1157 if (cb)
1158 {
1159 dSP;
1160 int count;
1161
1162 ENTER; SAVETMPS; PUSHMARK (SP);
1163 XPUSHs (HeVAL (he));
1164
1165 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1166
1167 if (count == 1)
1168 {
1169 sv = newSVsv (POPs);
1170 FREETMPS; LEAVE;
1171 return sv;
1172 }
1173
1174 FREETMPS; LEAVE;
1175 }
1176 }
1177
1178 if (dec->json.cb_object)
1179 {
1180 dSP;
1181 int count;
1182
1183 ENTER; SAVETMPS; PUSHMARK (SP);
1184 XPUSHs (sv_2mortal (sv));
1185
1186 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1187
1188 if (count == 1)
1189 {
1190 sv = newSVsv (POPs);
1191 FREETMPS; LEAVE;
1192 return sv;
1193 }
1194
1195 SvREFCNT_inc (sv);
1196 FREETMPS; LEAVE;
1197 }
1198 }
1199
1200 return sv;
844 1201
845fail: 1202fail:
846 SvREFCNT_dec (hv); 1203 SvREFCNT_dec (hv);
1204 DEC_DEC_DEPTH;
847 return 0; 1205 return 0;
848} 1206}
849 1207
850static SV * 1208static SV *
851decode_sv (dec_t *dec) 1209decode_sv (dec_t *dec)
852{ 1210{
853 decode_ws (dec); 1211 decode_ws (dec);
1212
1213 // the beauty of JSON: you need exactly one character lookahead
1214 // to parse anything.
854 switch (*dec->cur) 1215 switch (*dec->cur)
855 { 1216 {
856 case '"': ++dec->cur; return decode_str (dec); 1217 case '"': ++dec->cur; return decode_str (dec);
857 case '[': ++dec->cur; return decode_av (dec); 1218 case '[': ++dec->cur; return decode_av (dec);
858 case '{': ++dec->cur; return decode_hv (dec); 1219 case '{': ++dec->cur; return decode_hv (dec);
… …
864 1225
865 case 't': 1226 case 't':
866 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1227 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
867 { 1228 {
868 dec->cur += 4; 1229 dec->cur += 4;
869 return newSViv (1); 1230#if JSON_SLOW
1231 json_true = get_sv ("JSON::XS::true", 1); SvREADONLY_on (json_true);
1232#endif
1233 return SvREFCNT_inc (json_true);
870 } 1234 }
871 else 1235 else
872 ERR ("'true' expected"); 1236 ERR ("'true' expected");
873 1237
874 break; 1238 break;
875 1239
876 case 'f': 1240 case 'f':
877 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1241 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
878 { 1242 {
879 dec->cur += 5; 1243 dec->cur += 5;
880 return newSViv (0); 1244#if JSON_SLOW
1245 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
1246#endif
1247 return SvREFCNT_inc (json_false);
881 } 1248 }
882 else 1249 else
883 ERR ("'false' expected"); 1250 ERR ("'false' expected");
884 1251
885 break; 1252 break;
… …
894 ERR ("'null' expected"); 1261 ERR ("'null' expected");
895 1262
896 break; 1263 break;
897 1264
898 default: 1265 default:
899 ERR ("malformed json string, neither array, object, number, string or atom"); 1266 ERR ("malformed JSON string, neither array, object, number, string or atom");
900 break; 1267 break;
901 } 1268 }
902 1269
903fail: 1270fail:
904 return 0; 1271 return 0;
905} 1272}
906 1273
907static SV * 1274static SV *
908decode_json (SV *string, UV flags) 1275decode_json (SV *string, JSON *json, UV *offset_return)
909{ 1276{
1277 dec_t dec;
1278 UV offset;
910 SV *sv; 1279 SV *sv;
911 1280
1281 SvGETMAGIC (string);
1282 SvUPGRADE (string, SVt_PV);
1283
1284 if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags))
1285 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1286 (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags));
1287
912 if (flags & F_UTF8) 1288 if (json->flags & F_UTF8)
913 sv_utf8_downgrade (string, 0); 1289 sv_utf8_downgrade (string, 0);
914 else 1290 else
915 sv_utf8_upgrade (string); 1291 sv_utf8_upgrade (string);
916 1292
917 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1293 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
918 1294
919 dec_t dec; 1295 dec.json = *json;
920 dec.flags = flags;
921 dec.cur = SvPVX (string); 1296 dec.cur = SvPVX (string);
922 dec.end = SvEND (string); 1297 dec.end = SvEND (string);
923 dec.err = 0; 1298 dec.err = 0;
1299 dec.depth = 0;
1300 dec.maxdepth = DEC_DEPTH (dec.json.flags);
924 1301
1302 if (dec.json.cb_object || dec.json.cb_sk_object)
1303 dec.json.flags |= F_HOOK;
1304
1305 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
925 sv = decode_sv (&dec); 1306 sv = decode_sv (&dec);
926 1307
1308 if (!(offset_return || !sv))
1309 {
1310 // check for trailing garbage
1311 decode_ws (&dec);
1312
1313 if (*dec.cur)
1314 {
1315 dec.err = "garbage after JSON object";
1316 SvREFCNT_dec (sv);
1317 sv = 0;
1318 }
1319 }
1320
1321 if (offset_return || !sv)
1322 {
1323 offset = dec.json.flags & F_UTF8
1324 ? dec.cur - SvPVX (string)
1325 : utf8_distance (dec.cur, SvPVX (string));
1326
1327 if (offset_return)
1328 *offset_return = offset;
1329 }
1330
927 if (!sv) 1331 if (!sv)
928 { 1332 {
929 IV offset = dec.flags & F_UTF8
930 ? dec.cur - SvPVX (string)
931 : utf8_distance (dec.cur, SvPVX (string));
932 SV *uni = sv_newmortal (); 1333 SV *uni = sv_newmortal ();
933 1334
934 // horrible hack to silence warning inside pv_uni_display 1335 // horrible hack to silence warning inside pv_uni_display
935 COP cop = *PL_curcop; 1336 COP cop = *PL_curcop;
936 cop.cop_warnings = pWARN_NONE; 1337 cop.cop_warnings = pWARN_NONE;
… …
938 SAVEVPTR (PL_curcop); 1339 SAVEVPTR (PL_curcop);
939 PL_curcop = &cop; 1340 PL_curcop = &cop;
940 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1341 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
941 LEAVE; 1342 LEAVE;
942 1343
943 croak ("%s, at character offset %d (%s)", 1344 croak ("%s, at character offset %d [\"%s\"]",
944 dec.err, 1345 dec.err,
945 (int)offset, 1346 (int)offset,
946 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1347 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
947 } 1348 }
948 1349
949 sv = sv_2mortal (sv); 1350 sv = sv_2mortal (sv);
950 1351
951 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1352 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))
952 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); 1353 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
953 1354
954 return sv; 1355 return sv;
955} 1356}
956 1357
… …
961 1362
962BOOT: 1363BOOT:
963{ 1364{
964 int i; 1365 int i;
965 1366
966 memset (decode_hexdigit, 0xff, 256);
967 for (i = 10; i--; ) 1367 for (i = 0; i < 256; ++i)
968 decode_hexdigit ['0' + i] = i; 1368 decode_hexdigit [i] =
1369 i >= '0' && i <= '9' ? i - '0'
1370 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1371 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1372 : -1;
969 1373
970 for (i = 7; i--; ) 1374 json_stash = gv_stashpv ("JSON::XS" , 1);
1375 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1376
1377 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true );
1378 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
1379}
1380
1381PROTOTYPES: DISABLE
1382
1383void CLONE (...)
1384 CODE:
1385 json_stash = 0;
1386 json_boolean_stash = 0;
1387
1388void new (char *klass)
1389 PPCODE:
1390{
1391 SV *pv = NEWSV (0, sizeof (JSON));
1392 SvPOK_only (pv);
1393 Zero (SvPVX (pv), 1, JSON);
1394 ((JSON *)SvPVX (pv))->flags = F_DEFAULT;
1395 XPUSHs (sv_2mortal (sv_bless (newRV_noinc (pv), JSON_STASH)));
1396}
1397
1398void ascii (JSON *self, int enable = 1)
1399 ALIAS:
1400 ascii = F_ASCII
1401 latin1 = F_LATIN1
1402 utf8 = F_UTF8
1403 indent = F_INDENT
1404 canonical = F_CANONICAL
1405 space_before = F_SPACE_BEFORE
1406 space_after = F_SPACE_AFTER
1407 pretty = F_PRETTY
1408 allow_nonref = F_ALLOW_NONREF
1409 shrink = F_SHRINK
1410 allow_blessed = F_ALLOW_BLESSED
1411 convert_blessed = F_CONV_BLESSED
1412 PPCODE:
1413{
1414 if (enable)
1415 self->flags |= ix;
1416 else
1417 self->flags &= ~ix;
1418
1419 XPUSHs (ST (0));
1420}
1421
1422void max_depth (JSON *self, UV max_depth = 0x80000000UL)
1423 PPCODE:
1424{
1425 UV log2 = 0;
1426
1427 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1428
1429 while ((1UL << log2) < max_depth)
1430 ++log2;
1431
1432 self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1433
1434 XPUSHs (ST (0));
1435}
1436
1437void max_size (JSON *self, UV max_size = 0)
1438 PPCODE:
1439{
1440 UV log2 = 0;
1441
1442 if (max_size > 0x80000000UL) max_size = 0x80000000UL;
1443 if (max_size == 1) max_size = 2;
1444
1445 while ((1UL << log2) < max_size)
1446 ++log2;
1447
1448 self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1449
1450 XPUSHs (ST (0));
1451}
1452
1453void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1454 PPCODE:
1455{
1456 SvREFCNT_dec (self->cb_object);
1457 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
1458
1459 XPUSHs (ST (0));
1460}
1461
1462void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
1463 PPCODE:
1464{
1465 if (!self->cb_sk_object)
1466 self->cb_sk_object = newHV ();
1467
1468 if (SvOK (cb))
1469 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1470 else
971 { 1471 {
972 decode_hexdigit ['a' + i] = 10 + i; 1472 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
973 decode_hexdigit ['A' + i] = 10 + i; 1473
1474 if (!HvKEYS (self->cb_sk_object))
1475 {
1476 SvREFCNT_dec (self->cb_sk_object);
1477 self->cb_sk_object = 0;
1478 }
974 } 1479 }
975 1480
976 json_stash = gv_stashpv ("JSON::XS", 1); 1481 XPUSHs (ST (0));
977} 1482}
978 1483
979PROTOTYPES: DISABLE 1484void encode (JSON *self, SV *scalar)
1485 PPCODE:
1486 XPUSHs (encode_json (scalar, self));
980 1487
981SV *new (char *dummy) 1488void decode (JSON *self, SV *jsonstr)
1489 PPCODE:
1490 XPUSHs (decode_json (jsonstr, self, 0));
1491
1492void decode_prefix (JSON *self, SV *jsonstr)
1493 PPCODE:
1494{
1495 UV offset;
1496 EXTEND (SP, 2);
1497 PUSHs (decode_json (jsonstr, self, &offset));
1498 PUSHs (sv_2mortal (newSVuv (offset)));
1499}
1500
1501void DESTROY (JSON *self)
982 CODE: 1502 CODE:
983 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1503 SvREFCNT_dec (self->cb_sk_object);
984 OUTPUT: 1504 SvREFCNT_dec (self->cb_object);
985 RETVAL
986
987SV *ascii (SV *self, int enable = 1)
988 ALIAS:
989 ascii = F_ASCII
990 utf8 = F_UTF8
991 indent = F_INDENT
992 canonical = F_CANONICAL
993 space_before = F_SPACE_BEFORE
994 space_after = F_SPACE_AFTER
995 pretty = F_PRETTY
996 allow_nonref = F_ALLOW_NONREF
997 shrink = F_SHRINK
998 CODE:
999{
1000 UV *uv = SvJSON (self);
1001 if (enable)
1002 *uv |= ix;
1003 else
1004 *uv &= ~ix;
1005
1006 RETVAL = newSVsv (self);
1007}
1008 OUTPUT:
1009 RETVAL
1010
1011void encode (SV *self, SV *scalar)
1012 PPCODE:
1013 XPUSHs (encode_json (scalar, *SvJSON (self)));
1014
1015void decode (SV *self, SV *jsonstr)
1016 PPCODE:
1017 XPUSHs (decode_json (jsonstr, *SvJSON (self)));
1018 1505
1019PROTOTYPES: ENABLE 1506PROTOTYPES: ENABLE
1020 1507
1021void to_json (SV *scalar) 1508void to_json (SV *scalar)
1022 PPCODE: 1509 PPCODE:
1510{
1511 JSON json = { F_DEFAULT | F_UTF8 };
1023 XPUSHs (encode_json (scalar, F_UTF8)); 1512 XPUSHs (encode_json (scalar, &json));
1513}
1024 1514
1025void from_json (SV *jsonstr) 1515void from_json (SV *jsonstr)
1026 PPCODE: 1516 PPCODE:
1517{
1518 JSON json = { F_DEFAULT | F_UTF8 };
1027 XPUSHs (decode_json (jsonstr, F_UTF8)); 1519 XPUSHs (decode_json (jsonstr, &json, 0));
1520}
1028 1521

Diff Legend

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