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 1.15 by root, Sun Mar 25 02:32:40 2007 UTC vs.
Revision 1.52 by root, Mon Jul 2 02:57:11 2007 UTC

… …
3#include "XSUB.h" 3#include "XSUB.h"
4 4
5#include "assert.h" 5#include "assert.h"
6#include "string.h" 6#include "string.h"
7#include "stdlib.h" 7#include "stdlib.h"
8#include "stdio.h"
8 9
10#if defined(__BORLANDC__) || defined(_MSC_VER)
11# define snprintf _snprintf // C compilers have this in stdio.h
12#endif
13
14// some old perls do not have this, try to make it work, no
15// guarentees, though. if it breaks, you get to keep the pieces.
16#ifndef UTF8_MAXBYTES
17# define UTF8_MAXBYTES 13
18#endif
19
9#define F_ASCII 0x00000001 20#define F_ASCII 0x00000001UL
21#define F_LATIN1 0x00000002UL
10#define F_UTF8 0x00000002 22#define F_UTF8 0x00000004UL
11#define F_INDENT 0x00000004 23#define F_INDENT 0x00000008UL
12#define F_CANONICAL 0x00000008 24#define F_CANONICAL 0x00000010UL
13#define F_SPACE_BEFORE 0x00000010 25#define F_SPACE_BEFORE 0x00000020UL
14#define F_SPACE_AFTER 0x00000020 26#define F_SPACE_AFTER 0x00000040UL
15#define F_ALLOW_NONREF 0x00000080 27#define F_ALLOW_NONREF 0x00000100UL
16#define F_SHRINK 0x00000100 28#define F_SHRINK 0x00000200UL
29#define F_ALLOW_BLESSED 0x00000400UL
30#define F_CONV_BLESSED 0x00000800UL
31#define F_MAXDEPTH 0xf8000000UL
32#define S_MAXDEPTH 27
33#define F_MAXSIZE 0x01f00000UL
34#define S_MAXSIZE 20
35#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
36
37#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
38#define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE ))
17 39
18#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 40#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
19#define F_DEFAULT 0 41#define F_DEFAULT (9UL << S_MAXDEPTH)
20 42
21#define INIT_SIZE 32 // initial scalar size to be allocated 43#define INIT_SIZE 32 // initial scalar size to be allocated
22#define INDENT_STEP 3 // spaces per indentation level 44#define INDENT_STEP 3 // spaces per indentation level
23 45
24#define UTF8_MAX_LEN 11 // for perls UTF-X: max. number of octets per character
25#define SHORT_STRING_LEN 512 // special-case strings of up to this size 46#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
26 47
27#define SB do { 48#define SB do {
28#define SE } while (0) 49#define SE } while (0)
29 50
51#if __GNUC__ >= 3
52# define expect(expr,value) __builtin_expect ((expr),(value))
53# define inline inline
54#else
55# define expect(expr,value) (expr)
56# define inline static
57#endif
58
59#define expect_false(expr) expect ((expr) != 0, 0)
60#define expect_true(expr) expect ((expr) != 0, 1)
61
30static HV *json_stash; // JSON::XS:: 62static HV *json_stash, *json_boolean_stash; // JSON::XS::
63static SV *json_true, *json_false;
64
65typedef struct {
66 U32 flags;
67 SV *cb_object;
68 HV *cb_sk_object;
69} JSON;
31 70
32///////////////////////////////////////////////////////////////////////////// 71/////////////////////////////////////////////////////////////////////////////
33// utility functions 72// utility functions
34 73
35static UV * 74inline 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) 75shrink (SV *sv)
46{ 76{
47 sv_utf8_downgrade (sv, 1); 77 sv_utf8_downgrade (sv, 1);
48 if (SvLEN (sv) > SvCUR (sv) + 1) 78 if (SvLEN (sv) > SvCUR (sv) + 1)
49 { 79 {
… …
58// decode an utf-8 character and return it, or (UV)-1 in 88// decode an utf-8 character and return it, or (UV)-1 in
59// case of an error. 89// case of an error.
60// we special-case "safe" characters from U+80 .. U+7FF, 90// we special-case "safe" characters from U+80 .. U+7FF,
61// but use the very good perl function to parse anything else. 91// but use the very good perl function to parse anything else.
62// note that we never call this function for a ascii codepoints 92// note that we never call this function for a ascii codepoints
63static UV 93inline UV
64decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) 94decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
65{ 95{
66 if (s[0] > 0xdf || s[0] < 0xc2) 96 if (expect_false (s[0] > 0xdf || s[0] < 0xc2))
67 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); 97 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
68 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) 98 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)
69 { 99 {
70 *clen = 2; 100 *clen = 2;
71 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 101 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
72 } 102 }
73 else 103 else
104 {
105 *clen = (STRLEN)-1;
74 return (UV)-1; 106 return (UV)-1;
107 }
75} 108}
76 109
77///////////////////////////////////////////////////////////////////////////// 110/////////////////////////////////////////////////////////////////////////////
78// encoder 111// encoder
79 112
… …
81typedef struct 114typedef struct
82{ 115{
83 char *cur; // SvPVX (sv) + current output position 116 char *cur; // SvPVX (sv) + current output position
84 char *end; // SvEND (sv) 117 char *end; // SvEND (sv)
85 SV *sv; // result scalar 118 SV *sv; // result scalar
86 UV flags; // F_* 119 JSON json;
87 int indent; // indentation level 120 U32 indent; // indentation level
88 int max_depth; // max. recursion level 121 U32 maxdepth; // max. indentation/recursion level
89} enc_t; 122} enc_t;
90 123
91static void 124inline void
92need (enc_t *enc, STRLEN len) 125need (enc_t *enc, STRLEN len)
93{ 126{
94 if (enc->cur + len >= enc->end) 127 if (expect_false (enc->cur + len >= enc->end))
95 { 128 {
96 STRLEN cur = enc->cur - SvPVX (enc->sv); 129 STRLEN cur = enc->cur - SvPVX (enc->sv);
97 SvGROW (enc->sv, cur + len + 1); 130 SvGROW (enc->sv, cur + len + 1);
98 enc->cur = SvPVX (enc->sv) + cur; 131 enc->cur = SvPVX (enc->sv) + cur;
99 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv); 132 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
100 } 133 }
101} 134}
102 135
103static void 136inline void
104encode_ch (enc_t *enc, char ch) 137encode_ch (enc_t *enc, char ch)
105{ 138{
106 need (enc, 1); 139 need (enc, 1);
107 *enc->cur++ = ch; 140 *enc->cur++ = ch;
108} 141}
… …
116 149
117 while (str < end) 150 while (str < end)
118 { 151 {
119 unsigned char ch = *(unsigned char *)str; 152 unsigned char ch = *(unsigned char *)str;
120 153
121 if (ch >= 0x20 && ch < 0x80) // most common case 154 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
122 { 155 {
123 if (ch == '"') // but with slow exceptions 156 if (expect_false (ch == '"')) // but with slow exceptions
124 { 157 {
125 need (enc, len += 1); 158 need (enc, len += 1);
126 *enc->cur++ = '\\'; 159 *enc->cur++ = '\\';
127 *enc->cur++ = '"'; 160 *enc->cur++ = '"';
128 } 161 }
129 else if (ch == '\\') 162 else if (expect_false (ch == '\\'))
130 { 163 {
131 need (enc, len += 1); 164 need (enc, len += 1);
132 *enc->cur++ = '\\'; 165 *enc->cur++ = '\\';
133 *enc->cur++ = '\\'; 166 *enc->cur++ = '\\';
134 } 167 }
… …
152 STRLEN clen; 185 STRLEN clen;
153 UV uch; 186 UV uch;
154 187
155 if (is_utf8) 188 if (is_utf8)
156 { 189 {
157 //uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
158 uch = decode_utf8 (str, end - str, &clen); 190 uch = decode_utf8 (str, end - str, &clen);
159 if (clen == (STRLEN)-1) 191 if (clen == (STRLEN)-1)
160 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str); 192 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
161 } 193 }
162 else 194 else
… …
166 } 198 }
167 199
168 if (uch > 0x10FFFFUL) 200 if (uch > 0x10FFFFUL)
169 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); 201 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
170 202
171 if (uch < 0x80 || enc->flags & F_ASCII) 203 if (uch < 0x80 || enc->json.flags & F_ASCII || (enc->json.flags & F_LATIN1 && uch > 0xFF))
172 { 204 {
173 if (uch > 0xFFFFUL) 205 if (uch > 0xFFFFUL)
174 { 206 {
175 need (enc, len += 11); 207 need (enc, len += 11);
176 sprintf (enc->cur, "\\u%04x\\u%04x", 208 sprintf (enc->cur, "\\u%04x\\u%04x",
… …
190 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 222 *enc->cur++ = hexdigit [(uch >> 0) & 15];
191 } 223 }
192 224
193 str += clen; 225 str += clen;
194 } 226 }
227 else if (enc->json.flags & F_LATIN1)
228 {
229 *enc->cur++ = uch;
230 str += clen;
231 }
195 else if (is_utf8) 232 else if (is_utf8)
196 { 233 {
197 need (enc, len += clen); 234 need (enc, len += clen);
198 do 235 do
199 { 236 {
… …
201 } 238 }
202 while (--clen); 239 while (--clen);
203 } 240 }
204 else 241 else
205 { 242 {
206 need (enc, len += UTF8_MAX_LEN - 1); // never more than 11 bytes needed 243 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
207 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 244 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
208 ++str; 245 ++str;
209 } 246 }
210 } 247 }
211 } 248 }
… …
213 250
214 --len; 251 --len;
215 } 252 }
216} 253}
217 254
218static void 255inline void
219encode_indent (enc_t *enc) 256encode_indent (enc_t *enc)
220{ 257{
221 if (enc->flags & F_INDENT) 258 if (enc->json.flags & F_INDENT)
222 { 259 {
223 int spaces = enc->indent * INDENT_STEP; 260 int spaces = enc->indent * INDENT_STEP;
224 261
225 need (enc, spaces); 262 need (enc, spaces);
226 memset (enc->cur, ' ', spaces); 263 memset (enc->cur, ' ', spaces);
227 enc->cur += spaces; 264 enc->cur += spaces;
228 } 265 }
229} 266}
230 267
231static void 268inline void
232encode_space (enc_t *enc) 269encode_space (enc_t *enc)
233{ 270{
234 need (enc, 1); 271 need (enc, 1);
235 encode_ch (enc, ' '); 272 encode_ch (enc, ' ');
236} 273}
237 274
238static void 275inline void
239encode_nl (enc_t *enc) 276encode_nl (enc_t *enc)
240{ 277{
241 if (enc->flags & F_INDENT) 278 if (enc->json.flags & F_INDENT)
242 { 279 {
243 need (enc, 1); 280 need (enc, 1);
244 encode_ch (enc, '\n'); 281 encode_ch (enc, '\n');
245 } 282 }
246} 283}
247 284
248static void 285inline void
249encode_comma (enc_t *enc) 286encode_comma (enc_t *enc)
250{ 287{
251 encode_ch (enc, ','); 288 encode_ch (enc, ',');
252 289
253 if (enc->flags & F_INDENT) 290 if (enc->json.flags & F_INDENT)
254 encode_nl (enc); 291 encode_nl (enc);
255 else if (enc->flags & F_SPACE_AFTER) 292 else if (enc->json.flags & F_SPACE_AFTER)
256 encode_space (enc); 293 encode_space (enc);
257} 294}
258 295
259static void encode_sv (enc_t *enc, SV *sv); 296static void encode_sv (enc_t *enc, SV *sv);
260 297
261static void 298static void
262encode_av (enc_t *enc, AV *av) 299encode_av (enc_t *enc, AV *av)
263{ 300{
264 int i, len = av_len (av); 301 int i, len = av_len (av);
302
303 if (enc->indent >= enc->maxdepth)
304 croak ("data structure too deep (hit recursion limit)");
265 305
266 encode_ch (enc, '['); encode_nl (enc); 306 encode_ch (enc, '['); encode_nl (enc);
267 ++enc->indent; 307 ++enc->indent;
268 308
269 for (i = 0; i <= len; ++i) 309 for (i = 0; i <= len; ++i)
… …
300 else 340 else
301 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 341 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
302 342
303 encode_ch (enc, '"'); 343 encode_ch (enc, '"');
304 344
305 if (enc->flags & F_SPACE_BEFORE) encode_space (enc); 345 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
306 encode_ch (enc, ':'); 346 encode_ch (enc, ':');
307 if (enc->flags & F_SPACE_AFTER ) encode_space (enc); 347 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
308 encode_sv (enc, HeVAL (he)); 348 encode_sv (enc, HeVAL (he));
309} 349}
310 350
311// compare hash entries, used when all keys are bytestrings 351// compare hash entries, used when all keys are bytestrings
312static int 352static int
… …
335 375
336static void 376static void
337encode_hv (enc_t *enc, HV *hv) 377encode_hv (enc_t *enc, HV *hv)
338{ 378{
339 int count, i; 379 int count, i;
380
381 if (enc->indent >= enc->maxdepth)
382 croak ("data structure too deep (hit recursion limit)");
340 383
341 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 384 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
342 385
343 if ((count = hv_iterinit (hv))) 386 if ((count = hv_iterinit (hv)))
344 { 387 {
345 // for canonical output we have to sort by keys first 388 // for canonical output we have to sort by keys first
346 // actually, this is mostly due to the stupid so-called 389 // actually, this is mostly due to the stupid so-called
347 // security workaround added somewhere in 5.8.x. 390 // security workaround added somewhere in 5.8.x.
348 // that randomises hash orderings 391 // that randomises hash orderings
349 if (enc->flags & F_CANONICAL) 392 if (enc->json.flags & F_CANONICAL)
350 { 393 {
351 HE *he, *hes [count]; // if your compiler dies here, you need to enable C99 mode
352 int fast = 1; 394 int fast = 1;
395 HE *he;
396#if defined(__BORLANDC__) || defined(_MSC_VER)
397 HE **hes = _alloca (count * sizeof (HE));
398#else
399 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
400#endif
353 401
354 i = 0; 402 i = 0;
355 while ((he = hv_iternext (hv))) 403 while ((he = hv_iternext (hv)))
356 { 404 {
357 hes [i++] = he; 405 hes [i++] = he;
… …
392 440
393 encode_nl (enc); 441 encode_nl (enc);
394 } 442 }
395 else 443 else
396 { 444 {
397 SV *sv;
398 HE *he = hv_iternext (hv); 445 HE *he = hv_iternext (hv);
399 446
400 for (;;) 447 for (;;)
401 { 448 {
402 encode_indent (enc); 449 encode_indent (enc);
… …
411 encode_nl (enc); 458 encode_nl (enc);
412 } 459 }
413 } 460 }
414 461
415 --enc->indent; encode_indent (enc); encode_ch (enc, '}'); 462 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
463}
464
465// encode objects, arrays and special \0=false and \1=true values.
466static void
467encode_rv (enc_t *enc, SV *sv)
468{
469 svtype svt;
470
471 SvGETMAGIC (sv);
472 svt = SvTYPE (sv);
473
474 if (expect_false (SvOBJECT (sv)))
475 {
476 if (SvSTASH (sv) == json_boolean_stash)
477 {
478 if (SvIV (sv))
479 encode_str (enc, "true", 4, 0);
480 else
481 encode_str (enc, "false", 5, 0);
482 }
483 else
484 {
485#if 0
486 if (0 && sv_derived_from (rv, "JSON::Literal"))
487 {
488 // not yet
489 }
490#endif
491 if (enc->json.flags & F_CONV_BLESSED)
492 {
493 // we re-bless the reference to get overload and other niceties right
494 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 1);
495
496 if (to_json)
497 {
498 dSP; ENTER; SAVETMPS; PUSHMARK (SP);
499 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
500
501 // calling with G_SCALAR ensures that we always get a 1 reutrn value
502 // check anyways.
503 PUTBACK;
504 assert (1 == call_sv ((SV *)GvCV (to_json), G_SCALAR));
505 SPAGAIN;
506
507 encode_sv (enc, POPs);
508
509 FREETMPS; LEAVE;
510 }
511 else if (enc->json.flags & F_ALLOW_BLESSED)
512 encode_str (enc, "null", 4, 0);
513 else
514 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
515 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
516 }
517 else if (enc->json.flags & F_ALLOW_BLESSED)
518 encode_str (enc, "null", 4, 0);
519 else
520 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",
521 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
522 }
523 }
524 else if (svt == SVt_PVHV)
525 encode_hv (enc, (HV *)sv);
526 else if (svt == SVt_PVAV)
527 encode_av (enc, (AV *)sv);
528 else if (svt < SVt_PVAV)
529 {
530 STRLEN len = 0;
531 char *pv = svt ? SvPV (sv, len) : 0;
532
533 if (len == 1 && *pv == '1')
534 encode_str (enc, "true", 4, 0);
535 else if (len == 1 && *pv == '0')
536 encode_str (enc, "false", 5, 0);
537 else
538 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
539 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
540 }
541 else
542 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
543 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
416} 544}
417 545
418static void 546static void
419encode_sv (enc_t *enc, SV *sv) 547encode_sv (enc_t *enc, SV *sv)
420{ 548{
… …
428 encode_str (enc, str, len, SvUTF8 (sv)); 556 encode_str (enc, str, len, SvUTF8 (sv));
429 encode_ch (enc, '"'); 557 encode_ch (enc, '"');
430 } 558 }
431 else if (SvNOKp (sv)) 559 else if (SvNOKp (sv))
432 { 560 {
561 // trust that perl will do the right thing w.r.t. JSON syntax.
433 need (enc, NV_DIG + 32); 562 need (enc, NV_DIG + 32);
434 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 563 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
435 enc->cur += strlen (enc->cur); 564 enc->cur += strlen (enc->cur);
436 } 565 }
437 else if (SvIOKp (sv)) 566 else if (SvIOKp (sv))
438 { 567 {
439 need (enc, 64); 568 // we assume we can always read an IV as a UV
569 if (SvUV (sv) & ~(UV)0x7fff)
570 {
571 // large integer, use the (rather slow) snprintf way.
572 need (enc, sizeof (UV) * 3);
440 enc->cur += 573 enc->cur +=
441 SvIsUV(sv) 574 SvIsUV(sv)
442 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 575 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
443 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 576 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
577 }
578 else
579 {
580 // optimise the "small number case"
581 // code will likely be branchless and use only a single multiplication
582 I32 i = SvIV (sv);
583 U32 u;
584 char digit, nz = 0;
585
586 need (enc, 6);
587
588 *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
589 u = i < 0 ? -i : i;
590
591 // convert to 4.28 fixed-point representation
592 u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
593
594 // now output digit by digit, each time masking out the integer part
595 // and multiplying by 5 while moving the decimal point one to the right,
596 // resulting in a net multiplication by 10.
597 // we always write the digit to memory but conditionally increment
598 // the pointer, to ease the usage of conditional move instructions.
599 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5;
600 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5;
601 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5;
602 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5;
603 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
604 }
444 } 605 }
445 else if (SvROK (sv)) 606 else if (SvROK (sv))
446 { 607 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)) 608 else if (!SvOK (sv))
463 encode_str (enc, "null", 4, 0); 609 encode_str (enc, "null", 4, 0);
464 else 610 else
465 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 611 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
466 SvPV_nolen (sv), SvFLAGS (sv)); 612 SvPV_nolen (sv), SvFLAGS (sv));
467} 613}
468 614
469static SV * 615static SV *
470encode_json (SV *scalar, UV flags) 616encode_json (SV *scalar, JSON *json)
471{ 617{
618 enc_t enc;
619
472 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 620 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
473 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 621 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
474 622
475 enc_t enc; 623 enc.json = *json;
476 enc.flags = flags;
477 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 624 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
478 enc.cur = SvPVX (enc.sv); 625 enc.cur = SvPVX (enc.sv);
479 enc.end = SvEND (enc.sv); 626 enc.end = SvEND (enc.sv);
480 enc.indent = 0; 627 enc.indent = 0;
481 enc.max_depth = 0x7fffffffUL; 628 enc.maxdepth = DEC_DEPTH (enc.json.flags);
482 629
483 SvPOK_only (enc.sv); 630 SvPOK_only (enc.sv);
484 encode_sv (&enc, scalar); 631 encode_sv (&enc, scalar);
485 632
633 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
634 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
635
486 if (!(flags & (F_ASCII | F_UTF8))) 636 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
487 SvUTF8_on (enc.sv); 637 SvUTF8_on (enc.sv);
488 638
489 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
490
491 if (enc.flags & F_SHRINK) 639 if (enc.json.flags & F_SHRINK)
492 shrink (enc.sv); 640 shrink (enc.sv);
493 641
494 return enc.sv; 642 return enc.sv;
495} 643}
496 644
… …
501typedef struct 649typedef struct
502{ 650{
503 char *cur; // current parser pointer 651 char *cur; // current parser pointer
504 char *end; // end of input string 652 char *end; // end of input string
505 const char *err; // parse error, if != 0 653 const char *err; // parse error, if != 0
506 UV flags; // F_* 654 JSON json;
655 U32 depth; // recursion depth
656 U32 maxdepth; // recursion depth limit
507} dec_t; 657} dec_t;
508 658
509static void 659inline void
510decode_ws (dec_t *dec) 660decode_ws (dec_t *dec)
511{ 661{
512 for (;;) 662 for (;;)
513 { 663 {
514 char ch = *dec->cur; 664 char ch = *dec->cur;
… …
520 ++dec->cur; 670 ++dec->cur;
521 } 671 }
522} 672}
523 673
524#define ERR(reason) SB dec->err = reason; goto fail; SE 674#define ERR(reason) SB dec->err = reason; goto fail; SE
675
525#define EXPECT_CH(ch) SB \ 676#define EXPECT_CH(ch) SB \
526 if (*dec->cur != ch) \ 677 if (*dec->cur != ch) \
527 ERR (# ch " expected"); \ 678 ERR (# ch " expected"); \
528 ++dec->cur; \ 679 ++dec->cur; \
529 SE 680 SE
530 681
682#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)")
683#define DEC_DEC_DEPTH --dec->depth
684
531static SV *decode_sv (dec_t *dec); 685static SV *decode_sv (dec_t *dec);
532 686
533static signed char decode_hexdigit[256]; 687static signed char decode_hexdigit[256];
534 688
535static UV 689static UV
536decode_4hex (dec_t *dec) 690decode_4hex (dec_t *dec)
537{ 691{
538 signed char d1, d2, d3, d4; 692 signed char d1, d2, d3, d4;
539 unsigned char *cur = (unsigned char *)dec->cur; 693 unsigned char *cur = (unsigned char *)dec->cur;
540 694
541 d1 = decode_hexdigit [cur [0]]; if (d1 < 0) ERR ("four hexadecimal digits expected"); 695 d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected");
542 d2 = decode_hexdigit [cur [1]]; if (d2 < 0) ERR ("four hexadecimal digits expected"); 696 d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected");
543 d3 = decode_hexdigit [cur [2]]; if (d3 < 0) ERR ("four hexadecimal digits expected"); 697 d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected");
544 d4 = decode_hexdigit [cur [3]]; if (d4 < 0) ERR ("four hexadecimal digits expected"); 698 d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected");
545 699
546 dec->cur += 4; 700 dec->cur += 4;
547 701
548 return ((UV)d1) << 12 702 return ((UV)d1) << 12
549 | ((UV)d2) << 8 703 | ((UV)d2) << 8
… …
557static SV * 711static SV *
558decode_str (dec_t *dec) 712decode_str (dec_t *dec)
559{ 713{
560 SV *sv = 0; 714 SV *sv = 0;
561 int utf8 = 0; 715 int utf8 = 0;
716 char *dec_cur = dec->cur;
562 717
563 do 718 do
564 { 719 {
565 char buf [SHORT_STRING_LEN + UTF8_MAX_LEN]; 720 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
566 char *cur = buf; 721 char *cur = buf;
567 722
568 do 723 do
569 { 724 {
570 unsigned char ch = *(unsigned char *)dec->cur++; 725 unsigned char ch = *(unsigned char *)dec_cur++;
571 726
572 if (ch == '"') 727 if (expect_false (ch == '"'))
573 { 728 {
574 --dec->cur; 729 --dec_cur;
575 break; 730 break;
576 } 731 }
577 else if (ch == '\\') 732 else if (expect_false (ch == '\\'))
578 { 733 {
579 switch (*dec->cur) 734 switch (*dec_cur)
580 { 735 {
581 case '\\': 736 case '\\':
582 case '/': 737 case '/':
583 case '"': *cur++ = *dec->cur++; break; 738 case '"': *cur++ = *dec_cur++; break;
584 739
585 case 'b': ++dec->cur; *cur++ = '\010'; break; 740 case 'b': ++dec_cur; *cur++ = '\010'; break;
586 case 't': ++dec->cur; *cur++ = '\011'; break; 741 case 't': ++dec_cur; *cur++ = '\011'; break;
587 case 'n': ++dec->cur; *cur++ = '\012'; break; 742 case 'n': ++dec_cur; *cur++ = '\012'; break;
588 case 'f': ++dec->cur; *cur++ = '\014'; break; 743 case 'f': ++dec_cur; *cur++ = '\014'; break;
589 case 'r': ++dec->cur; *cur++ = '\015'; break; 744 case 'r': ++dec_cur; *cur++ = '\015'; break;
590 745
591 case 'u': 746 case 'u':
592 { 747 {
593 UV lo, hi; 748 UV lo, hi;
594 ++dec->cur; 749 ++dec_cur;
595 750
751 dec->cur = dec_cur;
596 hi = decode_4hex (dec); 752 hi = decode_4hex (dec);
753 dec_cur = dec->cur;
597 if (hi == (UV)-1) 754 if (hi == (UV)-1)
598 goto fail; 755 goto fail;
599 756
600 // possibly a surrogate pair 757 // possibly a surrogate pair
601 if (hi >= 0xd800) 758 if (hi >= 0xd800)
602 if (hi < 0xdc00) 759 if (hi < 0xdc00)
603 { 760 {
604 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 761 if (dec_cur [0] != '\\' || dec_cur [1] != 'u')
605 ERR ("missing low surrogate character in surrogate pair"); 762 ERR ("missing low surrogate character in surrogate pair");
606 763
607 dec->cur += 2; 764 dec_cur += 2;
608 765
766 dec->cur = dec_cur;
609 lo = decode_4hex (dec); 767 lo = decode_4hex (dec);
768 dec_cur = dec->cur;
610 if (lo == (UV)-1) 769 if (lo == (UV)-1)
611 goto fail; 770 goto fail;
612 771
613 if (lo < 0xdc00 || lo >= 0xe000) 772 if (lo < 0xdc00 || lo >= 0xe000)
614 ERR ("surrogate pair expected"); 773 ERR ("surrogate pair expected");
… …
628 *cur++ = hi; 787 *cur++ = hi;
629 } 788 }
630 break; 789 break;
631 790
632 default: 791 default:
633 --dec->cur; 792 --dec_cur;
634 ERR ("illegal backslash escape sequence in string"); 793 ERR ("illegal backslash escape sequence in string");
635 } 794 }
636 } 795 }
637 else if (ch >= 0x20 && ch <= 0x7f) 796 else if (expect_true (ch >= 0x20 && ch <= 0x7f))
638 *cur++ = ch; 797 *cur++ = ch;
639 else if (ch >= 0x80) 798 else if (ch >= 0x80)
640 { 799 {
641 --dec->cur;
642
643 STRLEN clen; 800 STRLEN clen;
801 UV uch;
802
803 --dec_cur;
804
644 UV uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen); 805 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
645 if (clen == (STRLEN)-1) 806 if (clen == (STRLEN)-1)
646 ERR ("malformed UTF-8 character in JSON string"); 807 ERR ("malformed UTF-8 character in JSON string");
647 808
648 do 809 do
649 {
650 *cur++ = *dec->cur++; 810 *cur++ = *dec_cur++;
651 }
652 while (--clen); 811 while (--clen);
653 812
654 utf8 = 1; 813 utf8 = 1;
655 } 814 }
656 else if (!ch)
657 ERR ("unexpected end of string while parsing json string");
658 else 815 else
816 {
817 --dec_cur;
818
819 if (!ch)
820 ERR ("unexpected end of string while parsing JSON string");
821 else
659 ERR ("invalid character encountered"); 822 ERR ("invalid character encountered while parsing JSON string");
660 823 }
661 } 824 }
662 while (cur < buf + SHORT_STRING_LEN); 825 while (cur < buf + SHORT_STRING_LEN);
663 826
827 {
664 STRLEN len = cur - buf; 828 STRLEN len = cur - buf;
665 829
666 if (sv) 830 if (sv)
667 { 831 {
668 SvGROW (sv, SvCUR (sv) + len + 1); 832 SvGROW (sv, SvCUR (sv) + len + 1);
669 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 833 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
670 SvCUR_set (sv, SvCUR (sv) + len); 834 SvCUR_set (sv, SvCUR (sv) + len);
671 } 835 }
672 else 836 else
673 sv = newSVpvn (buf, len); 837 sv = newSVpvn (buf, len);
674 } 838 }
839 }
675 while (*dec->cur != '"'); 840 while (*dec_cur != '"');
676 841
677 ++dec->cur; 842 ++dec_cur;
678 843
679 if (sv) 844 if (sv)
680 { 845 {
681 SvPOK_only (sv); 846 SvPOK_only (sv);
682 *SvEND (sv) = 0; 847 *SvEND (sv) = 0;
… …
685 SvUTF8_on (sv); 850 SvUTF8_on (sv);
686 } 851 }
687 else 852 else
688 sv = newSVpvn ("", 0); 853 sv = newSVpvn ("", 0);
689 854
855 dec->cur = dec_cur;
690 return sv; 856 return sv;
691 857
692fail: 858fail:
859 dec->cur = dec_cur;
693 return 0; 860 return 0;
694} 861}
695 862
696static SV * 863static SV *
697decode_num (dec_t *dec) 864decode_num (dec_t *dec)
… …
755 is_nv = 1; 922 is_nv = 1;
756 } 923 }
757 924
758 if (!is_nv) 925 if (!is_nv)
759 { 926 {
760 UV uv; 927 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so
761 int numtype = grok_number (start, dec->cur - start, &uv); 928 if (*start == '-')
762 if (numtype & IS_NUMBER_IN_UV) 929 switch (dec->cur - start)
763 if (numtype & IS_NUMBER_NEG)
764 { 930 {
765 if (uv < (UV)IV_MIN) 931 case 2: return newSViv (-( start [1] - '0' * 1));
766 return newSViv (-(IV)uv); 932 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
933 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
934 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
767 } 935 }
936 else
937 switch (dec->cur - start)
938 {
939 case 1: return newSViv ( start [0] - '0' * 1);
940 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
941 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
942 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
943 }
944
945 {
946 UV uv;
947 int numtype = grok_number (start, dec->cur - start, &uv);
948 if (numtype & IS_NUMBER_IN_UV)
949 if (numtype & IS_NUMBER_NEG)
950 {
951 if (uv < (UV)IV_MIN)
952 return newSViv (-(IV)uv);
953 }
768 else 954 else
769 return newSVuv (uv); 955 return newSVuv (uv);
956
957 // here would likely be the place for bigint support
770 } 958 }
959 }
771 960
961 // if we ever support bigint or bigfloat, this is the place for bigfloat
772 return newSVnv (Atof (start)); 962 return newSVnv (Atof (start));
773 963
774fail: 964fail:
775 return 0; 965 return 0;
776} 966}
… …
778static SV * 968static SV *
779decode_av (dec_t *dec) 969decode_av (dec_t *dec)
780{ 970{
781 AV *av = newAV (); 971 AV *av = newAV ();
782 972
973 DEC_INC_DEPTH;
783 decode_ws (dec); 974 decode_ws (dec);
975
784 if (*dec->cur == ']') 976 if (*dec->cur == ']')
785 ++dec->cur; 977 ++dec->cur;
786 else 978 else
787 for (;;) 979 for (;;)
788 { 980 {
… …
806 ERR (", or ] expected while parsing array"); 998 ERR (", or ] expected while parsing array");
807 999
808 ++dec->cur; 1000 ++dec->cur;
809 } 1001 }
810 1002
1003 DEC_DEC_DEPTH;
811 return newRV_noinc ((SV *)av); 1004 return newRV_noinc ((SV *)av);
812 1005
813fail: 1006fail:
814 SvREFCNT_dec (av); 1007 SvREFCNT_dec (av);
1008 DEC_DEC_DEPTH;
815 return 0; 1009 return 0;
816} 1010}
817 1011
818static SV * 1012static SV *
819decode_hv (dec_t *dec) 1013decode_hv (dec_t *dec)
820{ 1014{
1015 SV *sv;
821 HV *hv = newHV (); 1016 HV *hv = newHV ();
822 1017
1018 DEC_INC_DEPTH;
823 decode_ws (dec); 1019 decode_ws (dec);
1020
824 if (*dec->cur == '}') 1021 if (*dec->cur == '}')
825 ++dec->cur; 1022 ++dec->cur;
826 else 1023 else
827 for (;;) 1024 for (;;)
828 { 1025 {
829 SV *key, *value;
830
831 decode_ws (dec); EXPECT_CH ('"'); 1026 decode_ws (dec); EXPECT_CH ('"');
832 1027
833 key = decode_str (dec); 1028 // heuristic: assume that
834 if (!key) 1029 // a) decode_str + hv_store_ent are abysmally slow.
835 goto fail; 1030 // b) most hash keys are short, simple ascii text.
1031 // => try to "fast-match" such strings to avoid
1032 // the overhead of decode_str + hv_store_ent.
1033 {
1034 SV *value;
1035 char *p = dec->cur;
1036 char *e = p + 24; // only try up to 24 bytes
836 1037
837 decode_ws (dec); EXPECT_CH (':'); 1038 for (;;)
838
839 value = decode_sv (dec);
840 if (!value)
841 { 1039 {
1040 // the >= 0x80 is true on most architectures
1041 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1042 {
1043 // slow path, back up and use decode_str
1044 SV *key = decode_str (dec);
1045 if (!key)
1046 goto fail;
1047
1048 decode_ws (dec); EXPECT_CH (':');
1049
1050 value = decode_sv (dec);
1051 if (!value)
1052 {
1053 SvREFCNT_dec (key);
1054 goto fail;
1055 }
1056
1057 hv_store_ent (hv, key, value, 0);
842 SvREFCNT_dec (key); 1058 SvREFCNT_dec (key);
1059
1060 break;
1061 }
1062 else if (*p == '"')
1063 {
1064 // fast path, got a simple key
1065 char *key = dec->cur;
1066 int len = p - key;
1067 dec->cur = p + 1;
1068
1069 decode_ws (dec); EXPECT_CH (':');
1070
1071 value = decode_sv (dec);
1072 if (!value)
843 goto fail; 1073 goto fail;
1074
1075 hv_store (hv, key, len, value, 0);
1076
1077 break;
1078 }
1079
1080 ++p;
844 } 1081 }
845 1082 }
846 //TODO: optimise
847 hv_store_ent (hv, key, value, 0);
848 1083
849 decode_ws (dec); 1084 decode_ws (dec);
850 1085
851 if (*dec->cur == '}') 1086 if (*dec->cur == '}')
852 { 1087 {
… …
858 ERR (", or } expected while parsing object/hash"); 1093 ERR (", or } expected while parsing object/hash");
859 1094
860 ++dec->cur; 1095 ++dec->cur;
861 } 1096 }
862 1097
1098 DEC_DEC_DEPTH;
863 return newRV_noinc ((SV *)hv); 1099 sv = newRV_noinc ((SV *)hv);
1100
1101 // check filter callbacks
1102 if (dec->json.flags & F_HOOK)
1103 {
1104 if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
1105 {
1106 HE *cb, *he;
1107
1108 hv_iterinit (hv);
1109 he = hv_iternext (hv);
1110 hv_iterinit (hv);
1111
1112 // the next line creates a mortal sv each time its called.
1113 // might want to optimise this for common cases.
1114 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1115
1116 if (cb)
1117 {
1118 int count;
1119 ENTER; SAVETMPS;
1120
1121 dSP; PUSHMARK (SP);
1122 XPUSHs (HeVAL (he));
1123
1124 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1125
1126 if (count == 1)
1127 {
1128 sv = newSVsv (POPs);
1129 FREETMPS; LEAVE;
1130 return sv;
1131 }
1132
1133 FREETMPS; LEAVE;
1134 }
1135 }
1136
1137 if (dec->json.cb_object)
1138 {
1139 int count;
1140 ENTER; SAVETMPS;
1141
1142 dSP; ENTER; SAVETMPS; PUSHMARK (SP);
1143 XPUSHs (sv_2mortal (sv));
1144
1145 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1146
1147 if (count == 1)
1148 {
1149 sv = newSVsv (POPs);
1150 FREETMPS; LEAVE;
1151 return sv;
1152 }
1153
1154 SvREFCNT_inc (sv);
1155 FREETMPS; LEAVE;
1156 }
1157 }
1158
1159 return sv;
864 1160
865fail: 1161fail:
866 SvREFCNT_dec (hv); 1162 SvREFCNT_dec (hv);
1163 DEC_DEC_DEPTH;
867 return 0; 1164 return 0;
868} 1165}
869 1166
870static SV * 1167static SV *
871decode_sv (dec_t *dec) 1168decode_sv (dec_t *dec)
872{ 1169{
873 decode_ws (dec); 1170 decode_ws (dec);
1171
1172 // the beauty of JSON: you need exactly one character lookahead
1173 // to parse anything.
874 switch (*dec->cur) 1174 switch (*dec->cur)
875 { 1175 {
876 case '"': ++dec->cur; return decode_str (dec); 1176 case '"': ++dec->cur; return decode_str (dec);
877 case '[': ++dec->cur; return decode_av (dec); 1177 case '[': ++dec->cur; return decode_av (dec);
878 case '{': ++dec->cur; return decode_hv (dec); 1178 case '{': ++dec->cur; return decode_hv (dec);
… …
884 1184
885 case 't': 1185 case 't':
886 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1186 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
887 { 1187 {
888 dec->cur += 4; 1188 dec->cur += 4;
889 return newSViv (1); 1189 return SvREFCNT_inc (json_true);
890 } 1190 }
891 else 1191 else
892 ERR ("'true' expected"); 1192 ERR ("'true' expected");
893 1193
894 break; 1194 break;
895 1195
896 case 'f': 1196 case 'f':
897 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1197 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
898 { 1198 {
899 dec->cur += 5; 1199 dec->cur += 5;
900 return newSViv (0); 1200 return SvREFCNT_inc (json_false);
901 } 1201 }
902 else 1202 else
903 ERR ("'false' expected"); 1203 ERR ("'false' expected");
904 1204
905 break; 1205 break;
… …
914 ERR ("'null' expected"); 1214 ERR ("'null' expected");
915 1215
916 break; 1216 break;
917 1217
918 default: 1218 default:
919 ERR ("malformed json string, neither array, object, number, string or atom"); 1219 ERR ("malformed JSON string, neither array, object, number, string or atom");
920 break; 1220 break;
921 } 1221 }
922 1222
923fail: 1223fail:
924 return 0; 1224 return 0;
925} 1225}
926 1226
927static SV * 1227static SV *
928decode_json (SV *string, UV flags) 1228decode_json (SV *string, JSON *json, UV *offset_return)
929{ 1229{
1230 dec_t dec;
1231 UV offset;
930 SV *sv; 1232 SV *sv;
931 1233
1234 SvGETMAGIC (string);
1235 SvUPGRADE (string, SVt_PV);
1236
1237 if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags))
1238 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1239 (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags));
1240
932 if (flags & F_UTF8) 1241 if (json->flags & F_UTF8)
933 sv_utf8_downgrade (string, 0); 1242 sv_utf8_downgrade (string, 0);
934 else 1243 else
935 sv_utf8_upgrade (string); 1244 sv_utf8_upgrade (string);
936 1245
937 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1246 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
938 1247
939 dec_t dec; 1248 dec.json = *json;
940 dec.flags = flags;
941 dec.cur = SvPVX (string); 1249 dec.cur = SvPVX (string);
942 dec.end = SvEND (string); 1250 dec.end = SvEND (string);
943 dec.err = 0; 1251 dec.err = 0;
1252 dec.depth = 0;
1253 dec.maxdepth = DEC_DEPTH (dec.json.flags);
944 1254
1255 if (dec.json.cb_object || dec.json.cb_sk_object)
1256 dec.json.flags |= F_HOOK;
1257
1258 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
945 sv = decode_sv (&dec); 1259 sv = decode_sv (&dec);
946 1260
1261 if (!(offset_return || !sv))
1262 {
1263 // check for trailing garbage
1264 decode_ws (&dec);
1265
1266 if (*dec.cur)
1267 {
1268 dec.err = "garbage after JSON object";
1269 SvREFCNT_dec (sv);
1270 sv = 0;
1271 }
1272 }
1273
1274 if (offset_return || !sv)
1275 {
1276 offset = dec.json.flags & F_UTF8
1277 ? dec.cur - SvPVX (string)
1278 : utf8_distance (dec.cur, SvPVX (string));
1279
1280 if (offset_return)
1281 *offset_return = offset;
1282 }
1283
947 if (!sv) 1284 if (!sv)
948 { 1285 {
949 IV offset = dec.flags & F_UTF8
950 ? dec.cur - SvPVX (string)
951 : utf8_distance (dec.cur, SvPVX (string));
952 SV *uni = sv_newmortal (); 1286 SV *uni = sv_newmortal ();
953 1287
954 // horrible hack to silence warning inside pv_uni_display 1288 // horrible hack to silence warning inside pv_uni_display
955 COP cop = *PL_curcop; 1289 COP cop = *PL_curcop;
956 cop.cop_warnings = pWARN_NONE; 1290 cop.cop_warnings = pWARN_NONE;
… …
958 SAVEVPTR (PL_curcop); 1292 SAVEVPTR (PL_curcop);
959 PL_curcop = &cop; 1293 PL_curcop = &cop;
960 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1294 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
961 LEAVE; 1295 LEAVE;
962 1296
963 croak ("%s, at character offset %d (%s)", 1297 croak ("%s, at character offset %d [\"%s\"]",
964 dec.err, 1298 dec.err,
965 (int)offset, 1299 (int)offset,
966 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1300 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
967 } 1301 }
968 1302
969 sv = sv_2mortal (sv); 1303 sv = sv_2mortal (sv);
970 1304
971 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1305 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)"); 1306 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
973 1307
974 return sv; 1308 return sv;
975} 1309}
976 1310
… …
981 1315
982BOOT: 1316BOOT:
983{ 1317{
984 int i; 1318 int i;
985 1319
986 memset (decode_hexdigit, 0xff, 256);
987 for (i = 10; i--; ) 1320 for (i = 0; i < 256; ++i)
988 decode_hexdigit ['0' + i] = i; 1321 decode_hexdigit [i] =
1322 i >= '0' && i <= '9' ? i - '0'
1323 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1324 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1325 : -1;
989 1326
990 for (i = 7; i--; ) 1327 json_stash = gv_stashpv ("JSON::XS" , 1);
1328 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1329
1330 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true );
1331 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
1332}
1333
1334PROTOTYPES: DISABLE
1335
1336void new (char *klass)
1337 PPCODE:
1338{
1339 SV *pv = NEWSV (0, sizeof (JSON));
1340 SvPOK_only (pv);
1341 Zero (SvPVX (pv), 1, JSON);
1342 ((JSON *)SvPVX (pv))->flags = F_DEFAULT;
1343 XPUSHs (sv_2mortal (sv_bless (newRV_noinc (pv), json_stash)));
1344}
1345
1346void ascii (JSON *self, int enable = 1)
1347 ALIAS:
1348 ascii = F_ASCII
1349 latin1 = F_LATIN1
1350 utf8 = F_UTF8
1351 indent = F_INDENT
1352 canonical = F_CANONICAL
1353 space_before = F_SPACE_BEFORE
1354 space_after = F_SPACE_AFTER
1355 pretty = F_PRETTY
1356 allow_nonref = F_ALLOW_NONREF
1357 shrink = F_SHRINK
1358 allow_blessed = F_ALLOW_BLESSED
1359 convert_blessed = F_CONV_BLESSED
1360 PPCODE:
1361{
1362 if (enable)
1363 self->flags |= ix;
1364 else
1365 self->flags &= ~ix;
1366
1367 XPUSHs (ST (0));
1368}
1369
1370void max_depth (JSON *self, UV max_depth = 0x80000000UL)
1371 PPCODE:
1372{
1373 UV log2 = 0;
1374
1375 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1376
1377 while ((1UL << log2) < max_depth)
1378 ++log2;
1379
1380 self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1381
1382 XPUSHs (ST (0));
1383}
1384
1385void max_size (JSON *self, UV max_size = 0)
1386 PPCODE:
1387{
1388 UV log2 = 0;
1389
1390 if (max_size > 0x80000000UL) max_size = 0x80000000UL;
1391 if (max_size == 1) max_size = 2;
1392
1393 while ((1UL << log2) < max_size)
1394 ++log2;
1395
1396 self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1397
1398 XPUSHs (ST (0));
1399}
1400
1401void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1402 PPCODE:
1403{
1404 SvREFCNT_dec (self->cb_object);
1405 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
1406
1407 XPUSHs (ST (0));
1408}
1409
1410void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
1411 PPCODE:
1412{
1413 if (!self->cb_sk_object)
1414 self->cb_sk_object = newHV ();
1415
1416 if (SvOK (cb))
1417 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1418 else
991 { 1419 {
992 decode_hexdigit ['a' + i] = 10 + i; 1420 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
993 decode_hexdigit ['A' + i] = 10 + i; 1421
1422 if (!HvKEYS (self->cb_sk_object))
1423 {
1424 SvREFCNT_dec (self->cb_sk_object);
1425 self->cb_sk_object = 0;
1426 }
994 } 1427 }
995 1428
996 json_stash = gv_stashpv ("JSON::XS", 1); 1429 XPUSHs (ST (0));
997} 1430}
998 1431
999PROTOTYPES: DISABLE 1432void encode (JSON *self, SV *scalar)
1433 PPCODE:
1434 XPUSHs (encode_json (scalar, self));
1000 1435
1001SV *new (char *dummy) 1436void decode (JSON *self, SV *jsonstr)
1437 PPCODE:
1438 XPUSHs (decode_json (jsonstr, self, 0));
1439
1440void decode_prefix (JSON *self, SV *jsonstr)
1441 PPCODE:
1442{
1443 UV offset;
1444 EXTEND (SP, 2);
1445 PUSHs (decode_json (jsonstr, self, &offset));
1446 PUSHs (sv_2mortal (newSVuv (offset)));
1447}
1448
1449void DESTROY (JSON *self)
1002 CODE: 1450 CODE:
1003 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1451 SvREFCNT_dec (self->cb_sk_object);
1004 OUTPUT: 1452 SvREFCNT_dec (self->cb_object);
1005 RETVAL
1006 1453
1007SV *ascii (SV *self, int enable = 1) 1454PROTOTYPES: ENABLE
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 1455
1026 RETVAL = newSVsv (self); 1456void to_json (SV *scalar)
1027}
1028 OUTPUT:
1029 RETVAL
1030
1031void encode (SV *self, SV *scalar)
1032 PPCODE: 1457 PPCODE:
1458{
1459 JSON json = { F_DEFAULT | F_UTF8 };
1033 XPUSHs (encode_json (scalar, *SvJSON (self))); 1460 XPUSHs (encode_json (scalar, &json));
1461}
1034 1462
1035void decode (SV *self, SV *jsonstr) 1463void from_json (SV *jsonstr)
1036 PPCODE: 1464 PPCODE:
1037 XPUSHs (decode_json (jsonstr, *SvJSON (self))); 1465{
1038 1466 JSON json = { F_DEFAULT | F_UTF8 };
1039PROTOTYPES: ENABLE
1040
1041void to_json (SV *scalar)
1042 ALIAS:
1043 objToJson = 0
1044 PPCODE:
1045 XPUSHs (encode_json (scalar, F_UTF8));
1046
1047void from_json (SV *jsonstr)
1048 ALIAS:
1049 jsonToObj = 0
1050 PPCODE:
1051 XPUSHs (decode_json (jsonstr, F_UTF8)); 1467 XPUSHs (decode_json (jsonstr, &json, 0));
1468}
1052 1469

Diff Legend

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