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 1.14 by root, Sat Mar 24 22:55:43 2007 UTC vs.
Revision 1.58 by root, Mon Aug 13 16:06:25 2007 UTC

1#include "EXTERN.h" 1#include "EXTERN.h"
2#include "perl.h" 2#include "perl.h"
3#include "XSUB.h" 3#include "XSUB.h"
4 4
5#include "assert.h" 5#include <assert.h>
6#include "string.h" 6#include <string.h>
7#include "stdlib.h" 7#include <stdlib.h>
8#include <stdio.h>
9#include <float.h>
8 10
11#if defined(__BORLANDC__) || defined(_MSC_VER)
12# define snprintf _snprintf // C compilers have this in stdio.h
13#endif
14
15// some old perls do not have this, try to make it work, no
16// guarentees, though. if it breaks, you get to keep the pieces.
17#ifndef UTF8_MAXBYTES
18# define UTF8_MAXBYTES 13
19#endif
20
9#define F_ASCII 0x00000001 21#define F_ASCII 0x00000001UL
22#define F_LATIN1 0x00000002UL
10#define F_UTF8 0x00000002 23#define F_UTF8 0x00000004UL
11#define F_INDENT 0x00000004 24#define F_INDENT 0x00000008UL
12#define F_CANONICAL 0x00000008 25#define F_CANONICAL 0x00000010UL
13#define F_SPACE_BEFORE 0x00000010 26#define F_SPACE_BEFORE 0x00000020UL
14#define F_SPACE_AFTER 0x00000020 27#define F_SPACE_AFTER 0x00000040UL
15#define F_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 512 // 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#else
66# define JSON_SLOW 0
67#endif
68
30static HV *json_stash; // JSON::XS:: 69static HV *json_stash, *json_boolean_stash; // JSON::XS::
70static SV *json_true, *json_false;
71
72typedef struct {
73 U32 flags;
74 SV *cb_object;
75 HV *cb_sk_object;
76} JSON;
31 77
32///////////////////////////////////////////////////////////////////////////// 78/////////////////////////////////////////////////////////////////////////////
33// utility functions 79// utility functions
34 80
35static UV * 81inline 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) 82shrink (SV *sv)
46{ 83{
47 sv_utf8_downgrade (sv, 1); 84 sv_utf8_downgrade (sv, 1);
48 if (SvLEN (sv) > SvCUR (sv) + 1) 85 if (SvLEN (sv) > SvCUR (sv) + 1)
49 { 86 {
… …
58// decode an utf-8 character and return it, or (UV)-1 in 95// decode an utf-8 character and return it, or (UV)-1 in
59// case of an error. 96// case of an error.
60// we special-case "safe" characters from U+80 .. U+7FF, 97// we special-case "safe" characters from U+80 .. U+7FF,
61// but use the very good perl function to parse anything else. 98// but use the very good perl function to parse anything else.
62// note that we never call this function for a ascii codepoints 99// note that we never call this function for a ascii codepoints
63static UV 100inline UV
64decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) 101decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
65{ 102{
66 if (s[0] > 0xdf || s[0] < 0xc2) 103 if (expect_false (s[0] > 0xdf || s[0] < 0xc2))
67 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); 104 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
68 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) 105 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)
69 { 106 {
70 *clen = 2; 107 *clen = 2;
71 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 108 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
72 } 109 }
73 else 110 else
111 {
112 *clen = (STRLEN)-1;
74 return (UV)-1; 113 return (UV)-1;
114 }
75} 115}
76 116
77///////////////////////////////////////////////////////////////////////////// 117/////////////////////////////////////////////////////////////////////////////
78// encoder 118// encoder
79 119
… …
81typedef struct 121typedef struct
82{ 122{
83 char *cur; // SvPVX (sv) + current output position 123 char *cur; // SvPVX (sv) + current output position
84 char *end; // SvEND (sv) 124 char *end; // SvEND (sv)
85 SV *sv; // result scalar 125 SV *sv; // result scalar
86 UV flags; // F_* 126 JSON json;
87 int indent; // indentation level 127 U32 indent; // indentation level
88 int max_depth; // max. recursion level 128 U32 maxdepth; // max. indentation/recursion level
89} enc_t; 129} enc_t;
90 130
91static void 131inline void
92need (enc_t *enc, STRLEN len) 132need (enc_t *enc, STRLEN len)
93{ 133{
94 if (enc->cur + len >= enc->end) 134 if (expect_false (enc->cur + len >= enc->end))
95 { 135 {
96 STRLEN cur = enc->cur - SvPVX (enc->sv); 136 STRLEN cur = enc->cur - SvPVX (enc->sv);
97 SvGROW (enc->sv, cur + len + 1); 137 SvGROW (enc->sv, cur + len + 1);
98 enc->cur = SvPVX (enc->sv) + cur; 138 enc->cur = SvPVX (enc->sv) + cur;
99 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv); 139 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
100 } 140 }
101} 141}
102 142
103static void 143inline void
104encode_ch (enc_t *enc, char ch) 144encode_ch (enc_t *enc, char ch)
105{ 145{
106 need (enc, 1); 146 need (enc, 1);
107 *enc->cur++ = ch; 147 *enc->cur++ = ch;
108} 148}
… …
116 156
117 while (str < end) 157 while (str < end)
118 { 158 {
119 unsigned char ch = *(unsigned char *)str; 159 unsigned char ch = *(unsigned char *)str;
120 160
121 if (ch >= 0x20 && ch < 0x80) // most common case 161 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
122 { 162 {
123 if (ch == '"') // but with slow exceptions 163 if (expect_false (ch == '"')) // but with slow exceptions
124 { 164 {
125 need (enc, len += 1); 165 need (enc, len += 1);
126 *enc->cur++ = '\\'; 166 *enc->cur++ = '\\';
127 *enc->cur++ = '"'; 167 *enc->cur++ = '"';
128 } 168 }
129 else if (ch == '\\') 169 else if (expect_false (ch == '\\'))
130 { 170 {
131 need (enc, len += 1); 171 need (enc, len += 1);
132 *enc->cur++ = '\\'; 172 *enc->cur++ = '\\';
133 *enc->cur++ = '\\'; 173 *enc->cur++ = '\\';
134 } 174 }
… …
152 STRLEN clen; 192 STRLEN clen;
153 UV uch; 193 UV uch;
154 194
155 if (is_utf8) 195 if (is_utf8)
156 { 196 {
157 //uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
158 uch = decode_utf8 (str, end - str, &clen); 197 uch = decode_utf8 (str, end - str, &clen);
159 if (clen == (STRLEN)-1) 198 if (clen == (STRLEN)-1)
160 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str); 199 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
161 } 200 }
162 else 201 else
… …
166 } 205 }
167 206
168 if (uch > 0x10FFFFUL) 207 if (uch > 0x10FFFFUL)
169 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); 208 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
170 209
171 if (uch < 0x80 || enc->flags & F_ASCII) 210 if (uch < 0x80 || enc->json.flags & F_ASCII || (enc->json.flags & F_LATIN1 && uch > 0xFF))
172 { 211 {
173 if (uch > 0xFFFFUL) 212 if (uch > 0xFFFFUL)
174 { 213 {
175 need (enc, len += 11); 214 need (enc, len += 11);
176 sprintf (enc->cur, "\\u%04x\\u%04x", 215 sprintf (enc->cur, "\\u%04x\\u%04x",
… …
190 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 229 *enc->cur++ = hexdigit [(uch >> 0) & 15];
191 } 230 }
192 231
193 str += clen; 232 str += clen;
194 } 233 }
234 else if (enc->json.flags & F_LATIN1)
235 {
236 *enc->cur++ = uch;
237 str += clen;
238 }
195 else if (is_utf8) 239 else if (is_utf8)
196 { 240 {
197 need (enc, len += clen); 241 need (enc, len += clen);
198 do 242 do
199 { 243 {
… …
201 } 245 }
202 while (--clen); 246 while (--clen);
203 } 247 }
204 else 248 else
205 { 249 {
206 need (enc, len += UTF8_MAX_LEN - 1); // never more than 11 bytes needed 250 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
207 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 251 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
208 ++str; 252 ++str;
209 } 253 }
210 } 254 }
211 } 255 }
… …
213 257
214 --len; 258 --len;
215 } 259 }
216} 260}
217 261
218static void 262inline void
219encode_indent (enc_t *enc) 263encode_indent (enc_t *enc)
220{ 264{
221 if (enc->flags & F_INDENT) 265 if (enc->json.flags & F_INDENT)
222 { 266 {
223 int spaces = enc->indent * INDENT_STEP; 267 int spaces = enc->indent * INDENT_STEP;
224 268
225 need (enc, spaces); 269 need (enc, spaces);
226 memset (enc->cur, ' ', spaces); 270 memset (enc->cur, ' ', spaces);
227 enc->cur += spaces; 271 enc->cur += spaces;
228 } 272 }
229} 273}
230 274
231static void 275inline void
232encode_space (enc_t *enc) 276encode_space (enc_t *enc)
233{ 277{
234 need (enc, 1); 278 need (enc, 1);
235 encode_ch (enc, ' '); 279 encode_ch (enc, ' ');
236} 280}
237 281
238static void 282inline void
239encode_nl (enc_t *enc) 283encode_nl (enc_t *enc)
240{ 284{
241 if (enc->flags & F_INDENT) 285 if (enc->json.flags & F_INDENT)
242 { 286 {
243 need (enc, 1); 287 need (enc, 1);
244 encode_ch (enc, '\n'); 288 encode_ch (enc, '\n');
245 } 289 }
246} 290}
247 291
248static void 292inline void
249encode_comma (enc_t *enc) 293encode_comma (enc_t *enc)
250{ 294{
251 encode_ch (enc, ','); 295 encode_ch (enc, ',');
252 296
253 if (enc->flags & F_INDENT) 297 if (enc->json.flags & F_INDENT)
254 encode_nl (enc); 298 encode_nl (enc);
255 else if (enc->flags & F_SPACE_AFTER) 299 else if (enc->json.flags & F_SPACE_AFTER)
256 encode_space (enc); 300 encode_space (enc);
257} 301}
258 302
259static void encode_sv (enc_t *enc, SV *sv); 303static void encode_sv (enc_t *enc, SV *sv);
260 304
261static void 305static void
262encode_av (enc_t *enc, AV *av) 306encode_av (enc_t *enc, AV *av)
263{ 307{
264 int i, len = av_len (av); 308 int i, len = av_len (av);
265 309
310 if (enc->indent >= enc->maxdepth)
311 croak ("data structure too deep (hit recursion limit)");
312
266 encode_ch (enc, '['); encode_nl (enc); 313 encode_ch (enc, '['); encode_nl (enc);
267 ++enc->indent; 314 ++enc->indent;
268 315
269 for (i = 0; i <= len; ++i) 316 for (i = 0; i <= len; ++i)
270 { 317 {
318 SV **svp = av_fetch (av, i, 0);
319
271 encode_indent (enc); 320 encode_indent (enc);
272 encode_sv (enc, *av_fetch (av, i, 0)); 321
322 if (svp)
323 encode_sv (enc, *svp);
324 else
325 encode_str (enc, "null", 4, 0);
273 326
274 if (i < len) 327 if (i < len)
275 encode_comma (enc); 328 encode_comma (enc);
276 } 329 }
277 330
… …
300 else 353 else
301 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 354 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
302 355
303 encode_ch (enc, '"'); 356 encode_ch (enc, '"');
304 357
305 if (enc->flags & F_SPACE_BEFORE) encode_space (enc); 358 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
306 encode_ch (enc, ':'); 359 encode_ch (enc, ':');
307 if (enc->flags & F_SPACE_AFTER ) encode_space (enc); 360 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
308 encode_sv (enc, HeVAL (he)); 361 encode_sv (enc, HeVAL (he));
309} 362}
310 363
311// compare hash entries, used when all keys are bytestrings 364// compare hash entries, used when all keys are bytestrings
312static int 365static int
… …
335 388
336static void 389static void
337encode_hv (enc_t *enc, HV *hv) 390encode_hv (enc_t *enc, HV *hv)
338{ 391{
339 int count, i; 392 int count, i;
393
394 if (enc->indent >= enc->maxdepth)
395 croak ("data structure too deep (hit recursion limit)");
340 396
341 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 397 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
342 398
343 if ((count = hv_iterinit (hv))) 399 if ((count = hv_iterinit (hv)))
344 { 400 {
345 // for canonical output we have to sort by keys first 401 // for canonical output we have to sort by keys first
346 // actually, this is mostly due to the stupid so-called 402 // actually, this is mostly due to the stupid so-called
347 // security workaround added somewhere in 5.8.x. 403 // security workaround added somewhere in 5.8.x.
348 // that randomises hash orderings 404 // that randomises hash orderings
349 if (enc->flags & F_CANONICAL) 405 if (enc->json.flags & F_CANONICAL)
350 { 406 {
351 HE *he, *hes [count]; // if your compiler dies here, you need to enable C99 mode
352 int fast = 1; 407 int fast = 1;
408 HE *he;
409#if defined(__BORLANDC__) || defined(_MSC_VER)
410 HE **hes = _alloca (count * sizeof (HE));
411#else
412 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
413#endif
353 414
354 i = 0; 415 i = 0;
355 while ((he = hv_iternext (hv))) 416 while ((he = hv_iternext (hv)))
356 { 417 {
357 hes [i++] = he; 418 hes [i++] = he;
… …
392 453
393 encode_nl (enc); 454 encode_nl (enc);
394 } 455 }
395 else 456 else
396 { 457 {
397 SV *sv;
398 HE *he = hv_iternext (hv); 458 HE *he = hv_iternext (hv);
399 459
400 for (;;) 460 for (;;)
401 { 461 {
402 encode_indent (enc); 462 encode_indent (enc);
… …
411 encode_nl (enc); 471 encode_nl (enc);
412 } 472 }
413 } 473 }
414 474
415 --enc->indent; encode_indent (enc); encode_ch (enc, '}'); 475 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
476}
477
478// encode objects, arrays and special \0=false and \1=true values.
479static void
480encode_rv (enc_t *enc, SV *sv)
481{
482 svtype svt;
483
484 SvGETMAGIC (sv);
485 svt = SvTYPE (sv);
486
487 if (expect_false (SvOBJECT (sv)))
488 {
489 HV *stash = !JSON_SLOW || json_boolean_stash
490 ? json_boolean_stash
491 : gv_stashpv ("JSON::XS::Boolean", 1);
492
493 if (SvSTASH (sv) == stash)
494 {
495 if (SvIV (sv))
496 encode_str (enc, "true", 4, 0);
497 else
498 encode_str (enc, "false", 5, 0);
499 }
500 else
501 {
502#if 0
503 if (0 && sv_derived_from (rv, "JSON::Literal"))
504 {
505 // not yet
506 }
507#endif
508 if (enc->json.flags & F_CONV_BLESSED)
509 {
510 // we re-bless the reference to get overload and other niceties right
511 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
512
513 if (to_json)
514 {
515 dSP;
516
517 ENTER; SAVETMPS; PUSHMARK (SP);
518 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
519
520 // calling with G_SCALAR ensures that we always get a 1 return value
521 PUTBACK;
522 call_sv ((SV *)GvCV (to_json), G_SCALAR);
523 SPAGAIN;
524
525 // catch this surprisingly common error
526 if (SvROK (TOPs) && SvRV (TOPs) == sv)
527 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
528
529 sv = POPs;
530 PUTBACK;
531
532 encode_sv (enc, sv);
533
534 FREETMPS; LEAVE;
535 }
536 else if (enc->json.flags & F_ALLOW_BLESSED)
537 encode_str (enc, "null", 4, 0);
538 else
539 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
540 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
541 }
542 else if (enc->json.flags & F_ALLOW_BLESSED)
543 encode_str (enc, "null", 4, 0);
544 else
545 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",
546 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
547 }
548 }
549 else if (svt == SVt_PVHV)
550 encode_hv (enc, (HV *)sv);
551 else if (svt == SVt_PVAV)
552 encode_av (enc, (AV *)sv);
553 else if (svt < SVt_PVAV)
554 {
555 STRLEN len = 0;
556 char *pv = svt ? SvPV (sv, len) : 0;
557
558 if (len == 1 && *pv == '1')
559 encode_str (enc, "true", 4, 0);
560 else if (len == 1 && *pv == '0')
561 encode_str (enc, "false", 5, 0);
562 else
563 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
564 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
565 }
566 else
567 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
568 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
416} 569}
417 570
418static void 571static void
419encode_sv (enc_t *enc, SV *sv) 572encode_sv (enc_t *enc, SV *sv)
420{ 573{
… …
428 encode_str (enc, str, len, SvUTF8 (sv)); 581 encode_str (enc, str, len, SvUTF8 (sv));
429 encode_ch (enc, '"'); 582 encode_ch (enc, '"');
430 } 583 }
431 else if (SvNOKp (sv)) 584 else if (SvNOKp (sv))
432 { 585 {
586 // trust that perl will do the right thing w.r.t. JSON syntax.
433 need (enc, NV_DIG + 32); 587 need (enc, NV_DIG + 32);
434 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 588 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
435 enc->cur += strlen (enc->cur); 589 enc->cur += strlen (enc->cur);
436 } 590 }
437 else if (SvIOKp (sv)) 591 else if (SvIOKp (sv))
438 { 592 {
439 need (enc, 64); 593 // we assume we can always read an IV as a UV
594 if (SvUV (sv) & ~(UV)0x7fff)
595 {
596 // large integer, use the (rather slow) snprintf way.
597 need (enc, sizeof (UV) * 3);
440 enc->cur += 598 enc->cur +=
441 SvIsUV(sv) 599 SvIsUV(sv)
442 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 600 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
443 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 601 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
602 }
603 else
604 {
605 // optimise the "small number case"
606 // code will likely be branchless and use only a single multiplication
607 I32 i = SvIV (sv);
608 U32 u;
609 char digit, nz = 0;
610
611 need (enc, 6);
612
613 *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
614 u = i < 0 ? -i : i;
615
616 // convert to 4.28 fixed-point representation
617 u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
618
619 // now output digit by digit, each time masking out the integer part
620 // and multiplying by 5 while moving the decimal point one to the right,
621 // resulting in a net multiplication by 10.
622 // we always write the digit to memory but conditionally increment
623 // the pointer, to ease the usage of conditional move instructions.
624 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5;
625 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5;
626 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5;
627 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5;
628 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
629 }
444 } 630 }
445 else if (SvROK (sv)) 631 else if (SvROK (sv))
446 { 632 encode_rv (enc, SvRV (sv));
447 SV *rv = SvRV (sv);
448
449 if (enc->indent >= enc->max_depth)
450 croak ("data structure too deep (hit recursion limit)");
451
452 switch (SvTYPE (rv))
453 {
454 case SVt_PVAV: encode_av (enc, (AV *)rv); break;
455 case SVt_PVHV: encode_hv (enc, (HV *)rv); break;
456
457 default:
458 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
459 SvPV_nolen (sv));
460 }
461 }
462 else if (!SvOK (sv)) 633 else if (!SvOK (sv))
463 encode_str (enc, "null", 4, 0); 634 encode_str (enc, "null", 4, 0);
464 else 635 else
465 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 636 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
466 SvPV_nolen (sv), SvFLAGS (sv)); 637 SvPV_nolen (sv), SvFLAGS (sv));
467} 638}
468 639
469static SV * 640static SV *
470encode_json (SV *scalar, UV flags) 641encode_json (SV *scalar, JSON *json)
471{ 642{
643 enc_t enc;
644
472 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 645 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
473 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 646 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
474 647
475 enc_t enc; 648 enc.json = *json;
476 enc.flags = flags;
477 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 649 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
478 enc.cur = SvPVX (enc.sv); 650 enc.cur = SvPVX (enc.sv);
479 enc.end = SvEND (enc.sv); 651 enc.end = SvEND (enc.sv);
480 enc.indent = 0; 652 enc.indent = 0;
481 enc.max_depth = 0x7fffffffUL; 653 enc.maxdepth = DEC_DEPTH (enc.json.flags);
482 654
483 SvPOK_only (enc.sv); 655 SvPOK_only (enc.sv);
484 encode_sv (&enc, scalar); 656 encode_sv (&enc, scalar);
485 657
658 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
659 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
660
486 if (!(flags & (F_ASCII | F_UTF8))) 661 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
487 SvUTF8_on (enc.sv); 662 SvUTF8_on (enc.sv);
488 663
489 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
490
491 if (enc.flags & F_SHRINK) 664 if (enc.json.flags & F_SHRINK)
492 shrink (enc.sv); 665 shrink (enc.sv);
493 666
494 return enc.sv; 667 return enc.sv;
495} 668}
496 669
… …
501typedef struct 674typedef struct
502{ 675{
503 char *cur; // current parser pointer 676 char *cur; // current parser pointer
504 char *end; // end of input string 677 char *end; // end of input string
505 const char *err; // parse error, if != 0 678 const char *err; // parse error, if != 0
506 UV flags; // F_* 679 JSON json;
680 U32 depth; // recursion depth
681 U32 maxdepth; // recursion depth limit
507} dec_t; 682} dec_t;
508 683
509static void 684inline void
510decode_ws (dec_t *dec) 685decode_ws (dec_t *dec)
511{ 686{
512 for (;;) 687 for (;;)
513 { 688 {
514 char ch = *dec->cur; 689 char ch = *dec->cur;
… …
520 ++dec->cur; 695 ++dec->cur;
521 } 696 }
522} 697}
523 698
524#define ERR(reason) SB dec->err = reason; goto fail; SE 699#define ERR(reason) SB dec->err = reason; goto fail; SE
700
525#define EXPECT_CH(ch) SB \ 701#define EXPECT_CH(ch) SB \
526 if (*dec->cur != ch) \ 702 if (*dec->cur != ch) \
527 ERR (# ch " expected"); \ 703 ERR (# ch " expected"); \
528 ++dec->cur; \ 704 ++dec->cur; \
529 SE 705 SE
530 706
707#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)")
708#define DEC_DEC_DEPTH --dec->depth
709
531static SV *decode_sv (dec_t *dec); 710static SV *decode_sv (dec_t *dec);
532 711
533static signed char decode_hexdigit[256]; 712static signed char decode_hexdigit[256];
534 713
535static UV 714static UV
536decode_4hex (dec_t *dec) 715decode_4hex (dec_t *dec)
537{ 716{
538 signed char d1, d2, d3, d4; 717 signed char d1, d2, d3, d4;
539 unsigned char *cur = (unsigned char *)dec->cur; 718 unsigned char *cur = (unsigned char *)dec->cur;
540 719
541 d1 = decode_hexdigit [cur [0]]; if (d1 < 0) ERR ("four hexadecimal digits expected"); 720 d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected");
542 d2 = decode_hexdigit [cur [1]]; if (d2 < 0) ERR ("four hexadecimal digits expected"); 721 d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected");
543 d3 = decode_hexdigit [cur [2]]; if (d3 < 0) ERR ("four hexadecimal digits expected"); 722 d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected");
544 d4 = decode_hexdigit [cur [3]]; if (d4 < 0) ERR ("four hexadecimal digits expected"); 723 d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected");
545 724
546 dec->cur += 4; 725 dec->cur += 4;
547 726
548 return ((UV)d1) << 12 727 return ((UV)d1) << 12
549 | ((UV)d2) << 8 728 | ((UV)d2) << 8
… …
557static SV * 736static SV *
558decode_str (dec_t *dec) 737decode_str (dec_t *dec)
559{ 738{
560 SV *sv = 0; 739 SV *sv = 0;
561 int utf8 = 0; 740 int utf8 = 0;
741 char *dec_cur = dec->cur;
562 742
563 do 743 do
564 { 744 {
565 char buf [SHORT_STRING_LEN + UTF8_MAX_LEN]; 745 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
566 char *cur = buf; 746 char *cur = buf;
567 747
568 do 748 do
569 { 749 {
570 unsigned char ch = *(unsigned char *)dec->cur++; 750 unsigned char ch = *(unsigned char *)dec_cur++;
571 751
572 if (ch == '"') 752 if (expect_false (ch == '"'))
573 { 753 {
574 --dec->cur; 754 --dec_cur;
575 break; 755 break;
576 } 756 }
577 else if (ch == '\\') 757 else if (expect_false (ch == '\\'))
578 { 758 {
579 switch (*dec->cur) 759 switch (*dec_cur)
580 { 760 {
581 case '\\': 761 case '\\':
582 case '/': 762 case '/':
583 case '"': *cur++ = *dec->cur++; break; 763 case '"': *cur++ = *dec_cur++; break;
584 764
585 case 'b': ++dec->cur; *cur++ = '\010'; break; 765 case 'b': ++dec_cur; *cur++ = '\010'; break;
586 case 't': ++dec->cur; *cur++ = '\011'; break; 766 case 't': ++dec_cur; *cur++ = '\011'; break;
587 case 'n': ++dec->cur; *cur++ = '\012'; break; 767 case 'n': ++dec_cur; *cur++ = '\012'; break;
588 case 'f': ++dec->cur; *cur++ = '\014'; break; 768 case 'f': ++dec_cur; *cur++ = '\014'; break;
589 case 'r': ++dec->cur; *cur++ = '\015'; break; 769 case 'r': ++dec_cur; *cur++ = '\015'; break;
590 770
591 case 'u': 771 case 'u':
592 { 772 {
593 UV lo, hi; 773 UV lo, hi;
594 ++dec->cur; 774 ++dec_cur;
595 775
776 dec->cur = dec_cur;
596 hi = decode_4hex (dec); 777 hi = decode_4hex (dec);
778 dec_cur = dec->cur;
597 if (hi == (UV)-1) 779 if (hi == (UV)-1)
598 goto fail; 780 goto fail;
599 781
600 // possibly a surrogate pair 782 // possibly a surrogate pair
601 if (hi >= 0xd800) 783 if (hi >= 0xd800)
602 if (hi < 0xdc00) 784 if (hi < 0xdc00)
603 { 785 {
604 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 786 if (dec_cur [0] != '\\' || dec_cur [1] != 'u')
605 ERR ("missing low surrogate character in surrogate pair"); 787 ERR ("missing low surrogate character in surrogate pair");
606 788
607 dec->cur += 2; 789 dec_cur += 2;
608 790
791 dec->cur = dec_cur;
609 lo = decode_4hex (dec); 792 lo = decode_4hex (dec);
793 dec_cur = dec->cur;
610 if (lo == (UV)-1) 794 if (lo == (UV)-1)
611 goto fail; 795 goto fail;
612 796
613 if (lo < 0xdc00 || lo >= 0xe000) 797 if (lo < 0xdc00 || lo >= 0xe000)
614 ERR ("surrogate pair expected"); 798 ERR ("surrogate pair expected");
… …
628 *cur++ = hi; 812 *cur++ = hi;
629 } 813 }
630 break; 814 break;
631 815
632 default: 816 default:
633 --dec->cur; 817 --dec_cur;
634 ERR ("illegal backslash escape sequence in string"); 818 ERR ("illegal backslash escape sequence in string");
635 } 819 }
636 } 820 }
637 else if (ch >= 0x20 && ch <= 0x7f) 821 else if (expect_true (ch >= 0x20 && ch <= 0x7f))
638 *cur++ = ch; 822 *cur++ = ch;
639 else if (ch >= 0x80) 823 else if (ch >= 0x80)
640 { 824 {
641 --dec->cur;
642
643 STRLEN clen; 825 STRLEN clen;
826 UV uch;
827
828 --dec_cur;
829
644 UV uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen); 830 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
645 if (clen == (STRLEN)-1) 831 if (clen == (STRLEN)-1)
646 ERR ("malformed UTF-8 character in JSON string"); 832 ERR ("malformed UTF-8 character in JSON string");
647 833
648 do 834 do
649 {
650 *cur++ = *dec->cur++; 835 *cur++ = *dec_cur++;
651 }
652 while (--clen); 836 while (--clen);
653 837
654 utf8 = 1; 838 utf8 = 1;
655 } 839 }
656 else if (!ch)
657 ERR ("unexpected end of string while parsing json string");
658 else 840 else
841 {
842 --dec_cur;
843
844 if (!ch)
845 ERR ("unexpected end of string while parsing JSON string");
846 else
659 ERR ("invalid character encountered"); 847 ERR ("invalid character encountered while parsing JSON string");
660 848 }
661 } 849 }
662 while (cur < buf + SHORT_STRING_LEN); 850 while (cur < buf + SHORT_STRING_LEN);
663 851
852 {
664 STRLEN len = cur - buf; 853 STRLEN len = cur - buf;
665 854
666 if (sv) 855 if (sv)
667 { 856 {
668 SvGROW (sv, SvCUR (sv) + len + 1); 857 SvGROW (sv, SvCUR (sv) + len + 1);
669 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 858 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
670 SvCUR_set (sv, SvCUR (sv) + len); 859 SvCUR_set (sv, SvCUR (sv) + len);
671 } 860 }
672 else 861 else
673 sv = newSVpvn (buf, len); 862 sv = newSVpvn (buf, len);
674 } 863 }
864 }
675 while (*dec->cur != '"'); 865 while (*dec_cur != '"');
676 866
677 ++dec->cur; 867 ++dec_cur;
678 868
679 if (sv) 869 if (sv)
680 { 870 {
681 SvPOK_only (sv); 871 SvPOK_only (sv);
682 *SvEND (sv) = 0; 872 *SvEND (sv) = 0;
… …
685 SvUTF8_on (sv); 875 SvUTF8_on (sv);
686 } 876 }
687 else 877 else
688 sv = newSVpvn ("", 0); 878 sv = newSVpvn ("", 0);
689 879
880 dec->cur = dec_cur;
690 return sv; 881 return sv;
691 882
692fail: 883fail:
884 dec->cur = dec_cur;
693 return 0; 885 return 0;
694} 886}
695 887
696static SV * 888static SV *
697decode_num (dec_t *dec) 889decode_num (dec_t *dec)
… …
755 is_nv = 1; 947 is_nv = 1;
756 } 948 }
757 949
758 if (!is_nv) 950 if (!is_nv)
759 { 951 {
760 UV uv; 952 int len = dec->cur - start;
761 int numtype = grok_number (start, dec->cur - start, &uv); 953
762 if (numtype & IS_NUMBER_IN_UV) 954 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so
763 if (numtype & IS_NUMBER_NEG) 955 if (*start == '-')
956 switch (len)
764 { 957 {
765 if (uv < (UV)IV_MIN) 958 case 2: return newSViv (-( start [1] - '0' * 1));
766 return newSViv (-(IV)uv); 959 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
960 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
961 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
767 } 962 }
963 else
964 switch (len)
965 {
966 case 1: return newSViv ( start [0] - '0' * 1);
967 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
968 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
969 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
970 }
971
972 {
973 UV uv;
974 int numtype = grok_number (start, len, &uv);
975 if (numtype & IS_NUMBER_IN_UV)
976 if (numtype & IS_NUMBER_NEG)
977 {
978 if (uv < (UV)IV_MIN)
979 return newSViv (-(IV)uv);
980 }
768 else 981 else
769 return newSVuv (uv); 982 return newSVuv (uv);
770 } 983 }
771 984
985 len -= *start == '-' ? 1 : 0;
986
987 // does not fit into IV or UV, try NV
988 if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len)
989 #if defined (LDBL_DIG)
990 || (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len)
991 #endif
992 )
993 // fits into NV without loss of precision
994 return newSVnv (Atof (start));
995
996 // everything else fails, convert it to a string
997 return newSVpvn (start, dec->cur - start);
998 }
999
1000 // loss of precision here
772 return newSVnv (Atof (start)); 1001 return newSVnv (Atof (start));
773 1002
774fail: 1003fail:
775 return 0; 1004 return 0;
776} 1005}
… …
778static SV * 1007static SV *
779decode_av (dec_t *dec) 1008decode_av (dec_t *dec)
780{ 1009{
781 AV *av = newAV (); 1010 AV *av = newAV ();
782 1011
1012 DEC_INC_DEPTH;
783 decode_ws (dec); 1013 decode_ws (dec);
1014
784 if (*dec->cur == ']') 1015 if (*dec->cur == ']')
785 ++dec->cur; 1016 ++dec->cur;
786 else 1017 else
787 for (;;) 1018 for (;;)
788 { 1019 {
… …
806 ERR (", or ] expected while parsing array"); 1037 ERR (", or ] expected while parsing array");
807 1038
808 ++dec->cur; 1039 ++dec->cur;
809 } 1040 }
810 1041
1042 DEC_DEC_DEPTH;
811 return newRV_noinc ((SV *)av); 1043 return newRV_noinc ((SV *)av);
812 1044
813fail: 1045fail:
814 SvREFCNT_dec (av); 1046 SvREFCNT_dec (av);
1047 DEC_DEC_DEPTH;
815 return 0; 1048 return 0;
816} 1049}
817 1050
818static SV * 1051static SV *
819decode_hv (dec_t *dec) 1052decode_hv (dec_t *dec)
820{ 1053{
1054 SV *sv;
821 HV *hv = newHV (); 1055 HV *hv = newHV ();
822 1056
1057 DEC_INC_DEPTH;
823 decode_ws (dec); 1058 decode_ws (dec);
1059
824 if (*dec->cur == '}') 1060 if (*dec->cur == '}')
825 ++dec->cur; 1061 ++dec->cur;
826 else 1062 else
827 for (;;) 1063 for (;;)
828 { 1064 {
829 SV *key, *value;
830
831 decode_ws (dec); EXPECT_CH ('"'); 1065 decode_ws (dec); EXPECT_CH ('"');
832 1066
833 key = decode_str (dec); 1067 // heuristic: assume that
834 if (!key) 1068 // a) decode_str + hv_store_ent are abysmally slow.
835 goto fail; 1069 // b) most hash keys are short, simple ascii text.
1070 // => try to "fast-match" such strings to avoid
1071 // the overhead of decode_str + hv_store_ent.
1072 {
1073 SV *value;
1074 char *p = dec->cur;
1075 char *e = p + 24; // only try up to 24 bytes
836 1076
837 decode_ws (dec); EXPECT_CH (':'); 1077 for (;;)
838
839 value = decode_sv (dec);
840 if (!value)
841 { 1078 {
1079 // the >= 0x80 is true on most architectures
1080 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1081 {
1082 // slow path, back up and use decode_str
1083 SV *key = decode_str (dec);
1084 if (!key)
1085 goto fail;
1086
1087 decode_ws (dec); EXPECT_CH (':');
1088
1089 value = decode_sv (dec);
1090 if (!value)
1091 {
1092 SvREFCNT_dec (key);
1093 goto fail;
1094 }
1095
1096 hv_store_ent (hv, key, value, 0);
842 SvREFCNT_dec (key); 1097 SvREFCNT_dec (key);
1098
1099 break;
1100 }
1101 else if (*p == '"')
1102 {
1103 // fast path, got a simple key
1104 char *key = dec->cur;
1105 int len = p - key;
1106 dec->cur = p + 1;
1107
1108 decode_ws (dec); EXPECT_CH (':');
1109
1110 value = decode_sv (dec);
1111 if (!value)
843 goto fail; 1112 goto fail;
1113
1114 hv_store (hv, key, len, value, 0);
1115
1116 break;
1117 }
1118
1119 ++p;
844 } 1120 }
845 1121 }
846 //TODO: optimise
847 hv_store_ent (hv, key, value, 0);
848 1122
849 decode_ws (dec); 1123 decode_ws (dec);
850 1124
851 if (*dec->cur == '}') 1125 if (*dec->cur == '}')
852 { 1126 {
… …
858 ERR (", or } expected while parsing object/hash"); 1132 ERR (", or } expected while parsing object/hash");
859 1133
860 ++dec->cur; 1134 ++dec->cur;
861 } 1135 }
862 1136
1137 DEC_DEC_DEPTH;
863 return newRV_noinc ((SV *)hv); 1138 sv = newRV_noinc ((SV *)hv);
1139
1140 // check filter callbacks
1141 if (dec->json.flags & F_HOOK)
1142 {
1143 if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
1144 {
1145 HE *cb, *he;
1146
1147 hv_iterinit (hv);
1148 he = hv_iternext (hv);
1149 hv_iterinit (hv);
1150
1151 // the next line creates a mortal sv each time its called.
1152 // might want to optimise this for common cases.
1153 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1154
1155 if (cb)
1156 {
1157 dSP;
1158 int count;
1159
1160 ENTER; SAVETMPS; PUSHMARK (SP);
1161 XPUSHs (HeVAL (he));
1162
1163 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1164
1165 if (count == 1)
1166 {
1167 sv = newSVsv (POPs);
1168 FREETMPS; LEAVE;
1169 return sv;
1170 }
1171
1172 FREETMPS; LEAVE;
1173 }
1174 }
1175
1176 if (dec->json.cb_object)
1177 {
1178 dSP;
1179 int count;
1180
1181 ENTER; SAVETMPS; PUSHMARK (SP);
1182 XPUSHs (sv_2mortal (sv));
1183
1184 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1185
1186 if (count == 1)
1187 {
1188 sv = newSVsv (POPs);
1189 FREETMPS; LEAVE;
1190 return sv;
1191 }
1192
1193 SvREFCNT_inc (sv);
1194 FREETMPS; LEAVE;
1195 }
1196 }
1197
1198 return sv;
864 1199
865fail: 1200fail:
866 SvREFCNT_dec (hv); 1201 SvREFCNT_dec (hv);
1202 DEC_DEC_DEPTH;
867 return 0; 1203 return 0;
868} 1204}
869 1205
870static SV * 1206static SV *
871decode_sv (dec_t *dec) 1207decode_sv (dec_t *dec)
872{ 1208{
873 decode_ws (dec); 1209 decode_ws (dec);
1210
1211 // the beauty of JSON: you need exactly one character lookahead
1212 // to parse anything.
874 switch (*dec->cur) 1213 switch (*dec->cur)
875 { 1214 {
876 case '"': ++dec->cur; return decode_str (dec); 1215 case '"': ++dec->cur; return decode_str (dec);
877 case '[': ++dec->cur; return decode_av (dec); 1216 case '[': ++dec->cur; return decode_av (dec);
878 case '{': ++dec->cur; return decode_hv (dec); 1217 case '{': ++dec->cur; return decode_hv (dec);
… …
884 1223
885 case 't': 1224 case 't':
886 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1225 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
887 { 1226 {
888 dec->cur += 4; 1227 dec->cur += 4;
889 return newSViv (1); 1228 return SvREFCNT_inc (json_true);
890 } 1229 }
891 else 1230 else
892 ERR ("'true' expected"); 1231 ERR ("'true' expected");
893 1232
894 break; 1233 break;
895 1234
896 case 'f': 1235 case 'f':
897 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1236 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
898 { 1237 {
899 dec->cur += 5; 1238 dec->cur += 5;
900 return newSViv (0); 1239 return SvREFCNT_inc (json_false);
901 } 1240 }
902 else 1241 else
903 ERR ("'false' expected"); 1242 ERR ("'false' expected");
904 1243
905 break; 1244 break;
… …
914 ERR ("'null' expected"); 1253 ERR ("'null' expected");
915 1254
916 break; 1255 break;
917 1256
918 default: 1257 default:
919 ERR ("malformed json string, neither array, object, number, string or atom"); 1258 ERR ("malformed JSON string, neither array, object, number, string or atom");
920 break; 1259 break;
921 } 1260 }
922 1261
923fail: 1262fail:
924 return 0; 1263 return 0;
925} 1264}
926 1265
927static SV * 1266static SV *
928decode_json (SV *string, UV flags) 1267decode_json (SV *string, JSON *json, UV *offset_return)
929{ 1268{
1269 dec_t dec;
1270 UV offset;
930 SV *sv; 1271 SV *sv;
931 1272
1273 SvGETMAGIC (string);
1274 SvUPGRADE (string, SVt_PV);
1275
1276 if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags))
1277 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1278 (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags));
1279
932 if (flags & F_UTF8) 1280 if (json->flags & F_UTF8)
933 sv_utf8_downgrade (string, 0); 1281 sv_utf8_downgrade (string, 0);
934 else 1282 else
935 sv_utf8_upgrade (string); 1283 sv_utf8_upgrade (string);
936 1284
937 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1285 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
938 1286
939 dec_t dec; 1287 dec.json = *json;
940 dec.flags = flags;
941 dec.cur = SvPVX (string); 1288 dec.cur = SvPVX (string);
942 dec.end = SvEND (string); 1289 dec.end = SvEND (string);
943 dec.err = 0; 1290 dec.err = 0;
1291 dec.depth = 0;
1292 dec.maxdepth = DEC_DEPTH (dec.json.flags);
944 1293
1294 if (dec.json.cb_object || dec.json.cb_sk_object)
1295 dec.json.flags |= F_HOOK;
1296
1297 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
945 sv = decode_sv (&dec); 1298 sv = decode_sv (&dec);
946 1299
1300 if (!(offset_return || !sv))
1301 {
1302 // check for trailing garbage
1303 decode_ws (&dec);
1304
1305 if (*dec.cur)
1306 {
1307 dec.err = "garbage after JSON object";
1308 SvREFCNT_dec (sv);
1309 sv = 0;
1310 }
1311 }
1312
1313 if (offset_return || !sv)
1314 {
1315 offset = dec.json.flags & F_UTF8
1316 ? dec.cur - SvPVX (string)
1317 : utf8_distance (dec.cur, SvPVX (string));
1318
1319 if (offset_return)
1320 *offset_return = offset;
1321 }
1322
947 if (!sv) 1323 if (!sv)
948 { 1324 {
949 IV offset = dec.flags & F_UTF8
950 ? dec.cur - SvPVX (string)
951 : utf8_distance (dec.cur, SvPVX (string));
952 SV *uni = sv_newmortal (); 1325 SV *uni = sv_newmortal ();
953 1326
954 // horrible hack to silence warning inside pv_uni_display 1327 // horrible hack to silence warning inside pv_uni_display
955 COP cop = *PL_curcop; 1328 COP cop = *PL_curcop;
956 cop.cop_warnings = pWARN_NONE; 1329 cop.cop_warnings = pWARN_NONE;
… …
958 SAVEVPTR (PL_curcop); 1331 SAVEVPTR (PL_curcop);
959 PL_curcop = &cop; 1332 PL_curcop = &cop;
960 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1333 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
961 LEAVE; 1334 LEAVE;
962 1335
963 croak ("%s, at character offset %d (%s)", 1336 croak ("%s, at character offset %d [\"%s\"]",
964 dec.err, 1337 dec.err,
965 (int)offset, 1338 (int)offset,
966 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1339 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
967 } 1340 }
968 1341
969 sv = sv_2mortal (sv); 1342 sv = sv_2mortal (sv);
970 1343
971 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1344 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))
972 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); 1345 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
973 1346
974 return sv; 1347 return sv;
975} 1348}
976 1349
… …
981 1354
982BOOT: 1355BOOT:
983{ 1356{
984 int i; 1357 int i;
985 1358
986 memset (decode_hexdigit, 0xff, 256);
987 for (i = 10; i--; ) 1359 for (i = 0; i < 256; ++i)
988 decode_hexdigit ['0' + i] = i; 1360 decode_hexdigit [i] =
1361 i >= '0' && i <= '9' ? i - '0'
1362 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1363 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1364 : -1;
989 1365
990 for (i = 7; i--; ) 1366 json_stash = gv_stashpv ("JSON::XS" , 1);
1367 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1368
1369 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true );
1370 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
1371}
1372
1373PROTOTYPES: DISABLE
1374
1375void CLONE (...)
1376 CODE:
1377 json_stash = 0;
1378 json_boolean_stash = 0;
1379
1380void new (char *klass)
1381 PPCODE:
1382{
1383 HV *stash = !JSON_SLOW || json_stash
1384 ? json_stash
1385 : gv_stashpv ("JSON::XS", 1);
1386 SV *pv = NEWSV (0, sizeof (JSON));
1387 SvPOK_only (pv);
1388 Zero (SvPVX (pv), 1, JSON);
1389 ((JSON *)SvPVX (pv))->flags = F_DEFAULT;
1390 XPUSHs (sv_2mortal (sv_bless (newRV_noinc (pv), stash)));
1391}
1392
1393void ascii (JSON *self, int enable = 1)
1394 ALIAS:
1395 ascii = F_ASCII
1396 latin1 = F_LATIN1
1397 utf8 = F_UTF8
1398 indent = F_INDENT
1399 canonical = F_CANONICAL
1400 space_before = F_SPACE_BEFORE
1401 space_after = F_SPACE_AFTER
1402 pretty = F_PRETTY
1403 allow_nonref = F_ALLOW_NONREF
1404 shrink = F_SHRINK
1405 allow_blessed = F_ALLOW_BLESSED
1406 convert_blessed = F_CONV_BLESSED
1407 PPCODE:
1408{
1409 if (enable)
1410 self->flags |= ix;
1411 else
1412 self->flags &= ~ix;
1413
1414 XPUSHs (ST (0));
1415}
1416
1417void max_depth (JSON *self, UV max_depth = 0x80000000UL)
1418 PPCODE:
1419{
1420 UV log2 = 0;
1421
1422 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1423
1424 while ((1UL << log2) < max_depth)
1425 ++log2;
1426
1427 self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1428
1429 XPUSHs (ST (0));
1430}
1431
1432void max_size (JSON *self, UV max_size = 0)
1433 PPCODE:
1434{
1435 UV log2 = 0;
1436
1437 if (max_size > 0x80000000UL) max_size = 0x80000000UL;
1438 if (max_size == 1) max_size = 2;
1439
1440 while ((1UL << log2) < max_size)
1441 ++log2;
1442
1443 self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1444
1445 XPUSHs (ST (0));
1446}
1447
1448void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1449 PPCODE:
1450{
1451 SvREFCNT_dec (self->cb_object);
1452 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
1453
1454 XPUSHs (ST (0));
1455}
1456
1457void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
1458 PPCODE:
1459{
1460 if (!self->cb_sk_object)
1461 self->cb_sk_object = newHV ();
1462
1463 if (SvOK (cb))
1464 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1465 else
991 { 1466 {
992 decode_hexdigit ['a' + i] = 10 + i; 1467 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
993 decode_hexdigit ['A' + i] = 10 + i; 1468
1469 if (!HvKEYS (self->cb_sk_object))
1470 {
1471 SvREFCNT_dec (self->cb_sk_object);
1472 self->cb_sk_object = 0;
1473 }
994 } 1474 }
995 1475
996 json_stash = gv_stashpv ("JSON::XS", 1); 1476 XPUSHs (ST (0));
997} 1477}
998 1478
999PROTOTYPES: DISABLE 1479void encode (JSON *self, SV *scalar)
1480 PPCODE:
1481 XPUSHs (encode_json (scalar, self));
1000 1482
1001SV *new (char *dummy) 1483void decode (JSON *self, SV *jsonstr)
1484 PPCODE:
1485 XPUSHs (decode_json (jsonstr, self, 0));
1486
1487void decode_prefix (JSON *self, SV *jsonstr)
1488 PPCODE:
1489{
1490 UV offset;
1491 EXTEND (SP, 2);
1492 PUSHs (decode_json (jsonstr, self, &offset));
1493 PUSHs (sv_2mortal (newSVuv (offset)));
1494}
1495
1496void DESTROY (JSON *self)
1002 CODE: 1497 CODE:
1003 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1498 SvREFCNT_dec (self->cb_sk_object);
1004 OUTPUT: 1499 SvREFCNT_dec (self->cb_object);
1005 RETVAL
1006
1007SV *ascii (SV *self, int enable = 1)
1008 ALIAS:
1009 ascii = F_ASCII
1010 utf8 = F_UTF8
1011 indent = F_INDENT
1012 canonical = F_CANONICAL
1013 space_before = F_SPACE_BEFORE
1014 space_after = F_SPACE_AFTER
1015 pretty = F_PRETTY
1016 allow_nonref = F_ALLOW_NONREF
1017 shrink = F_SHRINK
1018 CODE:
1019{
1020 UV *uv = SvJSON (self);
1021 if (enable)
1022 *uv |= ix;
1023 else
1024 *uv &= ~ix;
1025
1026 RETVAL = newSVsv (self);
1027}
1028 OUTPUT:
1029 RETVAL
1030
1031void encode (SV *self, SV *scalar)
1032 PPCODE:
1033 XPUSHs (encode_json (scalar, *SvJSON (self)));
1034
1035void decode (SV *self, SV *jsonstr)
1036 PPCODE:
1037 XPUSHs (decode_json (jsonstr, *SvJSON (self)));
1038 1500
1039PROTOTYPES: ENABLE 1501PROTOTYPES: ENABLE
1040 1502
1041void to_json (SV *scalar) 1503void to_json (SV *scalar)
1042 PPCODE: 1504 PPCODE:
1505{
1506 JSON json = { F_DEFAULT | F_UTF8 };
1043 XPUSHs (encode_json (scalar, F_UTF8)); 1507 XPUSHs (encode_json (scalar, &json));
1508}
1044 1509
1045void from_json (SV *jsonstr) 1510void from_json (SV *jsonstr)
1046 PPCODE: 1511 PPCODE:
1512{
1513 JSON json = { F_DEFAULT | F_UTF8 };
1047 XPUSHs (decode_json (jsonstr, F_UTF8)); 1514 XPUSHs (decode_json (jsonstr, &json, 0));
1515}
1048 1516

Diff Legend

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