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 1.56 by root, Thu Jul 26 11:33:35 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
30static HV *json_stash; // JSON::XS:: 63static HV *json_stash, *json_boolean_stash; // JSON::XS::
64static SV *json_true, *json_false;
65
66typedef struct {
67 U32 flags;
68 SV *cb_object;
69 HV *cb_sk_object;
70} JSON;
31 71
32///////////////////////////////////////////////////////////////////////////// 72/////////////////////////////////////////////////////////////////////////////
33// utility functions 73// utility functions
34 74
35static UV * 75inline 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) 76shrink (SV *sv)
46{ 77{
47 sv_utf8_downgrade (sv, 1); 78 sv_utf8_downgrade (sv, 1);
48 if (SvLEN (sv) > SvCUR (sv) + 1) 79 if (SvLEN (sv) > SvCUR (sv) + 1)
49 { 80 {
… …
58// decode an utf-8 character and return it, or (UV)-1 in 89// decode an utf-8 character and return it, or (UV)-1 in
59// case of an error. 90// case of an error.
60// we special-case "safe" characters from U+80 .. U+7FF, 91// we special-case "safe" characters from U+80 .. U+7FF,
61// but use the very good perl function to parse anything else. 92// but use the very good perl function to parse anything else.
62// note that we never call this function for a ascii codepoints 93// note that we never call this function for a ascii codepoints
63static UV 94inline UV
64decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) 95decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
65{ 96{
66 if (s[0] > 0xdf || s[0] < 0xc2) 97 if (expect_false (s[0] > 0xdf || s[0] < 0xc2))
67 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); 98 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
68 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) 99 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)
69 { 100 {
70 *clen = 2; 101 *clen = 2;
71 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 102 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
72 } 103 }
73 else 104 else
105 {
106 *clen = (STRLEN)-1;
74 return (UV)-1; 107 return (UV)-1;
108 }
75} 109}
76 110
77///////////////////////////////////////////////////////////////////////////// 111/////////////////////////////////////////////////////////////////////////////
78// encoder 112// encoder
79 113
… …
81typedef struct 115typedef struct
82{ 116{
83 char *cur; // SvPVX (sv) + current output position 117 char *cur; // SvPVX (sv) + current output position
84 char *end; // SvEND (sv) 118 char *end; // SvEND (sv)
85 SV *sv; // result scalar 119 SV *sv; // result scalar
86 UV flags; // F_* 120 JSON json;
87 int indent; // indentation level 121 U32 indent; // indentation level
88 int max_depth; // max. recursion level 122 U32 maxdepth; // max. indentation/recursion level
89} enc_t; 123} enc_t;
90 124
91static void 125inline void
92need (enc_t *enc, STRLEN len) 126need (enc_t *enc, STRLEN len)
93{ 127{
94 if (enc->cur + len >= enc->end) 128 if (expect_false (enc->cur + len >= enc->end))
95 { 129 {
96 STRLEN cur = enc->cur - SvPVX (enc->sv); 130 STRLEN cur = enc->cur - SvPVX (enc->sv);
97 SvGROW (enc->sv, cur + len + 1); 131 SvGROW (enc->sv, cur + len + 1);
98 enc->cur = SvPVX (enc->sv) + cur; 132 enc->cur = SvPVX (enc->sv) + cur;
99 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv); 133 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
100 } 134 }
101} 135}
102 136
103static void 137inline void
104encode_ch (enc_t *enc, char ch) 138encode_ch (enc_t *enc, char ch)
105{ 139{
106 need (enc, 1); 140 need (enc, 1);
107 *enc->cur++ = ch; 141 *enc->cur++ = ch;
108} 142}
… …
116 150
117 while (str < end) 151 while (str < end)
118 { 152 {
119 unsigned char ch = *(unsigned char *)str; 153 unsigned char ch = *(unsigned char *)str;
120 154
121 if (ch >= 0x20 && ch < 0x80) // most common case 155 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
122 { 156 {
123 if (ch == '"') // but with slow exceptions 157 if (expect_false (ch == '"')) // but with slow exceptions
124 { 158 {
125 need (enc, len += 1); 159 need (enc, len += 1);
126 *enc->cur++ = '\\'; 160 *enc->cur++ = '\\';
127 *enc->cur++ = '"'; 161 *enc->cur++ = '"';
128 } 162 }
129 else if (ch == '\\') 163 else if (expect_false (ch == '\\'))
130 { 164 {
131 need (enc, len += 1); 165 need (enc, len += 1);
132 *enc->cur++ = '\\'; 166 *enc->cur++ = '\\';
133 *enc->cur++ = '\\'; 167 *enc->cur++ = '\\';
134 } 168 }
… …
152 STRLEN clen; 186 STRLEN clen;
153 UV uch; 187 UV uch;
154 188
155 if (is_utf8) 189 if (is_utf8)
156 { 190 {
157 //uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
158 uch = decode_utf8 (str, end - str, &clen); 191 uch = decode_utf8 (str, end - str, &clen);
159 if (clen == (STRLEN)-1) 192 if (clen == (STRLEN)-1)
160 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str); 193 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
161 } 194 }
162 else 195 else
… …
166 } 199 }
167 200
168 if (uch > 0x10FFFFUL) 201 if (uch > 0x10FFFFUL)
169 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); 202 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
170 203
171 if (uch < 0x80 || enc->flags & F_ASCII) 204 if (uch < 0x80 || enc->json.flags & F_ASCII || (enc->json.flags & F_LATIN1 && uch > 0xFF))
172 { 205 {
173 if (uch > 0xFFFFUL) 206 if (uch > 0xFFFFUL)
174 { 207 {
175 need (enc, len += 11); 208 need (enc, len += 11);
176 sprintf (enc->cur, "\\u%04x\\u%04x", 209 sprintf (enc->cur, "\\u%04x\\u%04x",
… …
190 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 223 *enc->cur++ = hexdigit [(uch >> 0) & 15];
191 } 224 }
192 225
193 str += clen; 226 str += clen;
194 } 227 }
228 else if (enc->json.flags & F_LATIN1)
229 {
230 *enc->cur++ = uch;
231 str += clen;
232 }
195 else if (is_utf8) 233 else if (is_utf8)
196 { 234 {
197 need (enc, len += clen); 235 need (enc, len += clen);
198 do 236 do
199 { 237 {
… …
201 } 239 }
202 while (--clen); 240 while (--clen);
203 } 241 }
204 else 242 else
205 { 243 {
206 need (enc, len += UTF8_MAX_LEN - 1); // never more than 11 bytes needed 244 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
207 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 245 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
208 ++str; 246 ++str;
209 } 247 }
210 } 248 }
211 } 249 }
… …
213 251
214 --len; 252 --len;
215 } 253 }
216} 254}
217 255
218static void 256inline void
219encode_indent (enc_t *enc) 257encode_indent (enc_t *enc)
220{ 258{
221 if (enc->flags & F_INDENT) 259 if (enc->json.flags & F_INDENT)
222 { 260 {
223 int spaces = enc->indent * INDENT_STEP; 261 int spaces = enc->indent * INDENT_STEP;
224 262
225 need (enc, spaces); 263 need (enc, spaces);
226 memset (enc->cur, ' ', spaces); 264 memset (enc->cur, ' ', spaces);
227 enc->cur += spaces; 265 enc->cur += spaces;
228 } 266 }
229} 267}
230 268
231static void 269inline void
232encode_space (enc_t *enc) 270encode_space (enc_t *enc)
233{ 271{
234 need (enc, 1); 272 need (enc, 1);
235 encode_ch (enc, ' '); 273 encode_ch (enc, ' ');
236} 274}
237 275
238static void 276inline void
239encode_nl (enc_t *enc) 277encode_nl (enc_t *enc)
240{ 278{
241 if (enc->flags & F_INDENT) 279 if (enc->json.flags & F_INDENT)
242 { 280 {
243 need (enc, 1); 281 need (enc, 1);
244 encode_ch (enc, '\n'); 282 encode_ch (enc, '\n');
245 } 283 }
246} 284}
247 285
248static void 286inline void
249encode_comma (enc_t *enc) 287encode_comma (enc_t *enc)
250{ 288{
251 encode_ch (enc, ','); 289 encode_ch (enc, ',');
252 290
253 if (enc->flags & F_INDENT) 291 if (enc->json.flags & F_INDENT)
254 encode_nl (enc); 292 encode_nl (enc);
255 else if (enc->flags & F_SPACE_AFTER) 293 else if (enc->json.flags & F_SPACE_AFTER)
256 encode_space (enc); 294 encode_space (enc);
257} 295}
258 296
259static void encode_sv (enc_t *enc, SV *sv); 297static void encode_sv (enc_t *enc, SV *sv);
260 298
261static void 299static void
262encode_av (enc_t *enc, AV *av) 300encode_av (enc_t *enc, AV *av)
263{ 301{
264 int i, len = av_len (av); 302 int i, len = av_len (av);
265 303
304 if (enc->indent >= enc->maxdepth)
305 croak ("data structure too deep (hit recursion limit)");
306
266 encode_ch (enc, '['); encode_nl (enc); 307 encode_ch (enc, '['); encode_nl (enc);
267 ++enc->indent; 308 ++enc->indent;
268 309
269 for (i = 0; i <= len; ++i) 310 for (i = 0; i <= len; ++i)
270 { 311 {
312 SV **svp = av_fetch (av, i, 0);
313
271 encode_indent (enc); 314 encode_indent (enc);
272 encode_sv (enc, *av_fetch (av, i, 0)); 315
316 if (svp)
317 encode_sv (enc, *svp);
318 else
319 encode_str (enc, "null", 4, 0);
273 320
274 if (i < len) 321 if (i < len)
275 encode_comma (enc); 322 encode_comma (enc);
276 } 323 }
277 324
… …
300 else 347 else
301 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 348 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
302 349
303 encode_ch (enc, '"'); 350 encode_ch (enc, '"');
304 351
305 if (enc->flags & F_SPACE_BEFORE) encode_space (enc); 352 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
306 encode_ch (enc, ':'); 353 encode_ch (enc, ':');
307 if (enc->flags & F_SPACE_AFTER ) encode_space (enc); 354 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
308 encode_sv (enc, HeVAL (he)); 355 encode_sv (enc, HeVAL (he));
309} 356}
310 357
311// compare hash entries, used when all keys are bytestrings 358// compare hash entries, used when all keys are bytestrings
312static int 359static int
… …
335 382
336static void 383static void
337encode_hv (enc_t *enc, HV *hv) 384encode_hv (enc_t *enc, HV *hv)
338{ 385{
339 int count, i; 386 int count, i;
387
388 if (enc->indent >= enc->maxdepth)
389 croak ("data structure too deep (hit recursion limit)");
340 390
341 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 391 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent;
342 392
343 if ((count = hv_iterinit (hv))) 393 if ((count = hv_iterinit (hv)))
344 { 394 {
345 // for canonical output we have to sort by keys first 395 // for canonical output we have to sort by keys first
346 // actually, this is mostly due to the stupid so-called 396 // actually, this is mostly due to the stupid so-called
347 // security workaround added somewhere in 5.8.x. 397 // security workaround added somewhere in 5.8.x.
348 // that randomises hash orderings 398 // that randomises hash orderings
349 if (enc->flags & F_CANONICAL) 399 if (enc->json.flags & F_CANONICAL)
350 { 400 {
351 HE *he, *hes [count]; // if your compiler dies here, you need to enable C99 mode
352 int fast = 1; 401 int fast = 1;
402 HE *he;
403#if defined(__BORLANDC__) || defined(_MSC_VER)
404 HE **hes = _alloca (count * sizeof (HE));
405#else
406 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
407#endif
353 408
354 i = 0; 409 i = 0;
355 while ((he = hv_iternext (hv))) 410 while ((he = hv_iternext (hv)))
356 { 411 {
357 hes [i++] = he; 412 hes [i++] = he;
… …
392 447
393 encode_nl (enc); 448 encode_nl (enc);
394 } 449 }
395 else 450 else
396 { 451 {
397 SV *sv;
398 HE *he = hv_iternext (hv); 452 HE *he = hv_iternext (hv);
399 453
400 for (;;) 454 for (;;)
401 { 455 {
402 encode_indent (enc); 456 encode_indent (enc);
… …
411 encode_nl (enc); 465 encode_nl (enc);
412 } 466 }
413 } 467 }
414 468
415 --enc->indent; encode_indent (enc); encode_ch (enc, '}'); 469 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
470}
471
472// encode objects, arrays and special \0=false and \1=true values.
473static void
474encode_rv (enc_t *enc, SV *sv)
475{
476 svtype svt;
477
478 SvGETMAGIC (sv);
479 svt = SvTYPE (sv);
480
481 if (expect_false (SvOBJECT (sv)))
482 {
483 if (SvSTASH (sv) == json_boolean_stash)
484 {
485 if (SvIV (sv))
486 encode_str (enc, "true", 4, 0);
487 else
488 encode_str (enc, "false", 5, 0);
489 }
490 else
491 {
492#if 0
493 if (0 && sv_derived_from (rv, "JSON::Literal"))
494 {
495 // not yet
496 }
497#endif
498 if (enc->json.flags & F_CONV_BLESSED)
499 {
500 // we re-bless the reference to get overload and other niceties right
501 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
502
503 if (to_json)
504 {
505 int count;
506 dSP;
507
508 ENTER; SAVETMPS; PUSHMARK (SP);
509 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
510
511 // calling with G_SCALAR ensures that we always get a 1 return value
512 PUTBACK;
513 call_sv ((SV *)GvCV (to_json), G_SCALAR);
514 SPAGAIN;
515
516 // catch this surprisingly common error
517 if (SvROK (TOPs) && SvRV (TOPs) == sv)
518 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
519
520 sv = POPs;
521 PUTBACK;
522
523 encode_sv (enc, sv);
524
525 FREETMPS; LEAVE;
526 }
527 else if (enc->json.flags & F_ALLOW_BLESSED)
528 encode_str (enc, "null", 4, 0);
529 else
530 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
531 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
532 }
533 else if (enc->json.flags & F_ALLOW_BLESSED)
534 encode_str (enc, "null", 4, 0);
535 else
536 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",
537 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
538 }
539 }
540 else if (svt == SVt_PVHV)
541 encode_hv (enc, (HV *)sv);
542 else if (svt == SVt_PVAV)
543 encode_av (enc, (AV *)sv);
544 else if (svt < SVt_PVAV)
545 {
546 STRLEN len = 0;
547 char *pv = svt ? SvPV (sv, len) : 0;
548
549 if (len == 1 && *pv == '1')
550 encode_str (enc, "true", 4, 0);
551 else if (len == 1 && *pv == '0')
552 encode_str (enc, "false", 5, 0);
553 else
554 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
555 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
556 }
557 else
558 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
559 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
416} 560}
417 561
418static void 562static void
419encode_sv (enc_t *enc, SV *sv) 563encode_sv (enc_t *enc, SV *sv)
420{ 564{
… …
428 encode_str (enc, str, len, SvUTF8 (sv)); 572 encode_str (enc, str, len, SvUTF8 (sv));
429 encode_ch (enc, '"'); 573 encode_ch (enc, '"');
430 } 574 }
431 else if (SvNOKp (sv)) 575 else if (SvNOKp (sv))
432 { 576 {
577 // trust that perl will do the right thing w.r.t. JSON syntax.
433 need (enc, NV_DIG + 32); 578 need (enc, NV_DIG + 32);
434 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 579 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
435 enc->cur += strlen (enc->cur); 580 enc->cur += strlen (enc->cur);
436 } 581 }
437 else if (SvIOKp (sv)) 582 else if (SvIOKp (sv))
438 { 583 {
439 need (enc, 64); 584 // we assume we can always read an IV as a UV
585 if (SvUV (sv) & ~(UV)0x7fff)
586 {
587 // large integer, use the (rather slow) snprintf way.
588 need (enc, sizeof (UV) * 3);
440 enc->cur += 589 enc->cur +=
441 SvIsUV(sv) 590 SvIsUV(sv)
442 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 591 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
443 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 592 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
593 }
594 else
595 {
596 // optimise the "small number case"
597 // code will likely be branchless and use only a single multiplication
598 I32 i = SvIV (sv);
599 U32 u;
600 char digit, nz = 0;
601
602 need (enc, 6);
603
604 *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
605 u = i < 0 ? -i : i;
606
607 // convert to 4.28 fixed-point representation
608 u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
609
610 // now output digit by digit, each time masking out the integer part
611 // and multiplying by 5 while moving the decimal point one to the right,
612 // resulting in a net multiplication by 10.
613 // we always write the digit to memory but conditionally increment
614 // the pointer, to ease the usage of conditional move instructions.
615 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5;
616 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5;
617 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5;
618 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5;
619 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
620 }
444 } 621 }
445 else if (SvROK (sv)) 622 else if (SvROK (sv))
446 { 623 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)) 624 else if (!SvOK (sv))
463 encode_str (enc, "null", 4, 0); 625 encode_str (enc, "null", 4, 0);
464 else 626 else
465 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 627 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
466 SvPV_nolen (sv), SvFLAGS (sv)); 628 SvPV_nolen (sv), SvFLAGS (sv));
467} 629}
468 630
469static SV * 631static SV *
470encode_json (SV *scalar, UV flags) 632encode_json (SV *scalar, JSON *json)
471{ 633{
634 enc_t enc;
635
472 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 636 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
473 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 637 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
474 638
475 enc_t enc; 639 enc.json = *json;
476 enc.flags = flags;
477 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 640 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
478 enc.cur = SvPVX (enc.sv); 641 enc.cur = SvPVX (enc.sv);
479 enc.end = SvEND (enc.sv); 642 enc.end = SvEND (enc.sv);
480 enc.indent = 0; 643 enc.indent = 0;
481 enc.max_depth = 0x7fffffffUL; 644 enc.maxdepth = DEC_DEPTH (enc.json.flags);
482 645
483 SvPOK_only (enc.sv); 646 SvPOK_only (enc.sv);
484 encode_sv (&enc, scalar); 647 encode_sv (&enc, scalar);
485 648
649 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
650 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
651
486 if (!(flags & (F_ASCII | F_UTF8))) 652 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
487 SvUTF8_on (enc.sv); 653 SvUTF8_on (enc.sv);
488 654
489 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
490
491 if (enc.flags & F_SHRINK) 655 if (enc.json.flags & F_SHRINK)
492 shrink (enc.sv); 656 shrink (enc.sv);
493 657
494 return enc.sv; 658 return enc.sv;
495} 659}
496 660
… …
501typedef struct 665typedef struct
502{ 666{
503 char *cur; // current parser pointer 667 char *cur; // current parser pointer
504 char *end; // end of input string 668 char *end; // end of input string
505 const char *err; // parse error, if != 0 669 const char *err; // parse error, if != 0
506 UV flags; // F_* 670 JSON json;
671 U32 depth; // recursion depth
672 U32 maxdepth; // recursion depth limit
507} dec_t; 673} dec_t;
508 674
509static void 675inline void
510decode_ws (dec_t *dec) 676decode_ws (dec_t *dec)
511{ 677{
512 for (;;) 678 for (;;)
513 { 679 {
514 char ch = *dec->cur; 680 char ch = *dec->cur;
… …
520 ++dec->cur; 686 ++dec->cur;
521 } 687 }
522} 688}
523 689
524#define ERR(reason) SB dec->err = reason; goto fail; SE 690#define ERR(reason) SB dec->err = reason; goto fail; SE
691
525#define EXPECT_CH(ch) SB \ 692#define EXPECT_CH(ch) SB \
526 if (*dec->cur != ch) \ 693 if (*dec->cur != ch) \
527 ERR (# ch " expected"); \ 694 ERR (# ch " expected"); \
528 ++dec->cur; \ 695 ++dec->cur; \
529 SE 696 SE
530 697
698#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)")
699#define DEC_DEC_DEPTH --dec->depth
700
531static SV *decode_sv (dec_t *dec); 701static SV *decode_sv (dec_t *dec);
532 702
533static signed char decode_hexdigit[256]; 703static signed char decode_hexdigit[256];
534 704
535static UV 705static UV
536decode_4hex (dec_t *dec) 706decode_4hex (dec_t *dec)
537{ 707{
538 signed char d1, d2, d3, d4; 708 signed char d1, d2, d3, d4;
539 unsigned char *cur = (unsigned char *)dec->cur; 709 unsigned char *cur = (unsigned char *)dec->cur;
540 710
541 d1 = decode_hexdigit [cur [0]]; if (d1 < 0) ERR ("four hexadecimal digits expected"); 711 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"); 712 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"); 713 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"); 714 d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected");
545 715
546 dec->cur += 4; 716 dec->cur += 4;
547 717
548 return ((UV)d1) << 12 718 return ((UV)d1) << 12
549 | ((UV)d2) << 8 719 | ((UV)d2) << 8
… …
557static SV * 727static SV *
558decode_str (dec_t *dec) 728decode_str (dec_t *dec)
559{ 729{
560 SV *sv = 0; 730 SV *sv = 0;
561 int utf8 = 0; 731 int utf8 = 0;
732 char *dec_cur = dec->cur;
562 733
563 do 734 do
564 { 735 {
565 char buf [SHORT_STRING_LEN + UTF8_MAX_LEN]; 736 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
566 char *cur = buf; 737 char *cur = buf;
567 738
568 do 739 do
569 { 740 {
570 unsigned char ch = *(unsigned char *)dec->cur++; 741 unsigned char ch = *(unsigned char *)dec_cur++;
571 742
572 if (ch == '"') 743 if (expect_false (ch == '"'))
573 { 744 {
574 --dec->cur; 745 --dec_cur;
575 break; 746 break;
576 } 747 }
577 else if (ch == '\\') 748 else if (expect_false (ch == '\\'))
578 { 749 {
579 switch (*dec->cur) 750 switch (*dec_cur)
580 { 751 {
581 case '\\': 752 case '\\':
582 case '/': 753 case '/':
583 case '"': *cur++ = *dec->cur++; break; 754 case '"': *cur++ = *dec_cur++; break;
584 755
585 case 'b': ++dec->cur; *cur++ = '\010'; break; 756 case 'b': ++dec_cur; *cur++ = '\010'; break;
586 case 't': ++dec->cur; *cur++ = '\011'; break; 757 case 't': ++dec_cur; *cur++ = '\011'; break;
587 case 'n': ++dec->cur; *cur++ = '\012'; break; 758 case 'n': ++dec_cur; *cur++ = '\012'; break;
588 case 'f': ++dec->cur; *cur++ = '\014'; break; 759 case 'f': ++dec_cur; *cur++ = '\014'; break;
589 case 'r': ++dec->cur; *cur++ = '\015'; break; 760 case 'r': ++dec_cur; *cur++ = '\015'; break;
590 761
591 case 'u': 762 case 'u':
592 { 763 {
593 UV lo, hi; 764 UV lo, hi;
594 ++dec->cur; 765 ++dec_cur;
595 766
767 dec->cur = dec_cur;
596 hi = decode_4hex (dec); 768 hi = decode_4hex (dec);
769 dec_cur = dec->cur;
597 if (hi == (UV)-1) 770 if (hi == (UV)-1)
598 goto fail; 771 goto fail;
599 772
600 // possibly a surrogate pair 773 // possibly a surrogate pair
601 if (hi >= 0xd800) 774 if (hi >= 0xd800)
602 if (hi < 0xdc00) 775 if (hi < 0xdc00)
603 { 776 {
604 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 777 if (dec_cur [0] != '\\' || dec_cur [1] != 'u')
605 ERR ("missing low surrogate character in surrogate pair"); 778 ERR ("missing low surrogate character in surrogate pair");
606 779
607 dec->cur += 2; 780 dec_cur += 2;
608 781
782 dec->cur = dec_cur;
609 lo = decode_4hex (dec); 783 lo = decode_4hex (dec);
784 dec_cur = dec->cur;
610 if (lo == (UV)-1) 785 if (lo == (UV)-1)
611 goto fail; 786 goto fail;
612 787
613 if (lo < 0xdc00 || lo >= 0xe000) 788 if (lo < 0xdc00 || lo >= 0xe000)
614 ERR ("surrogate pair expected"); 789 ERR ("surrogate pair expected");
… …
628 *cur++ = hi; 803 *cur++ = hi;
629 } 804 }
630 break; 805 break;
631 806
632 default: 807 default:
633 --dec->cur; 808 --dec_cur;
634 ERR ("illegal backslash escape sequence in string"); 809 ERR ("illegal backslash escape sequence in string");
635 } 810 }
636 } 811 }
637 else if (ch >= 0x20 && ch <= 0x7f) 812 else if (expect_true (ch >= 0x20 && ch <= 0x7f))
638 *cur++ = ch; 813 *cur++ = ch;
639 else if (ch >= 0x80) 814 else if (ch >= 0x80)
640 { 815 {
641 --dec->cur;
642
643 STRLEN clen; 816 STRLEN clen;
817 UV uch;
818
819 --dec_cur;
820
644 UV uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen); 821 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
645 if (clen == (STRLEN)-1) 822 if (clen == (STRLEN)-1)
646 ERR ("malformed UTF-8 character in JSON string"); 823 ERR ("malformed UTF-8 character in JSON string");
647 824
648 do 825 do
649 {
650 *cur++ = *dec->cur++; 826 *cur++ = *dec_cur++;
651 }
652 while (--clen); 827 while (--clen);
653 828
654 utf8 = 1; 829 utf8 = 1;
655 } 830 }
656 else if (!ch)
657 ERR ("unexpected end of string while parsing json string");
658 else 831 else
832 {
833 --dec_cur;
834
835 if (!ch)
836 ERR ("unexpected end of string while parsing JSON string");
837 else
659 ERR ("invalid character encountered"); 838 ERR ("invalid character encountered while parsing JSON string");
660 839 }
661 } 840 }
662 while (cur < buf + SHORT_STRING_LEN); 841 while (cur < buf + SHORT_STRING_LEN);
663 842
843 {
664 STRLEN len = cur - buf; 844 STRLEN len = cur - buf;
665 845
666 if (sv) 846 if (sv)
667 { 847 {
668 SvGROW (sv, SvCUR (sv) + len + 1); 848 SvGROW (sv, SvCUR (sv) + len + 1);
669 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 849 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
670 SvCUR_set (sv, SvCUR (sv) + len); 850 SvCUR_set (sv, SvCUR (sv) + len);
671 } 851 }
672 else 852 else
673 sv = newSVpvn (buf, len); 853 sv = newSVpvn (buf, len);
674 } 854 }
855 }
675 while (*dec->cur != '"'); 856 while (*dec_cur != '"');
676 857
677 ++dec->cur; 858 ++dec_cur;
678 859
679 if (sv) 860 if (sv)
680 { 861 {
681 SvPOK_only (sv); 862 SvPOK_only (sv);
682 *SvEND (sv) = 0; 863 *SvEND (sv) = 0;
… …
685 SvUTF8_on (sv); 866 SvUTF8_on (sv);
686 } 867 }
687 else 868 else
688 sv = newSVpvn ("", 0); 869 sv = newSVpvn ("", 0);
689 870
871 dec->cur = dec_cur;
690 return sv; 872 return sv;
691 873
692fail: 874fail:
875 dec->cur = dec_cur;
693 return 0; 876 return 0;
694} 877}
695 878
696static SV * 879static SV *
697decode_num (dec_t *dec) 880decode_num (dec_t *dec)
… …
755 is_nv = 1; 938 is_nv = 1;
756 } 939 }
757 940
758 if (!is_nv) 941 if (!is_nv)
759 { 942 {
760 UV uv; 943 int len = dec->cur - start;
761 int numtype = grok_number (start, dec->cur - start, &uv); 944
762 if (numtype & IS_NUMBER_IN_UV) 945 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so
763 if (numtype & IS_NUMBER_NEG) 946 if (*start == '-')
947 switch (len)
764 { 948 {
765 if (uv < (UV)IV_MIN) 949 case 2: return newSViv (-( start [1] - '0' * 1));
766 return newSViv (-(IV)uv); 950 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
951 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
952 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
767 } 953 }
954 else
955 switch (len)
956 {
957 case 1: return newSViv ( start [0] - '0' * 1);
958 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
959 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
960 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
961 }
962
963 {
964 UV uv;
965 int numtype = grok_number (start, len, &uv);
966 if (numtype & IS_NUMBER_IN_UV)
967 if (numtype & IS_NUMBER_NEG)
968 {
969 if (uv < (UV)IV_MIN)
970 return newSViv (-(IV)uv);
971 }
768 else 972 else
769 return newSVuv (uv); 973 return newSVuv (uv);
770 } 974 }
771 975
976 len -= *start == '-' ? 1 : 0;
977
978 // does not fit into IV or UV, try NV
979 if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len)
980 #if defined (LDBL_DIG)
981 || (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len)
982 #endif
983 )
984 // fits into NV without loss of precision
985 return newSVnv (Atof (start));
986
987 // everything else fails, convert it to a string
988 return newSVpvn (start, dec->cur - start);
989 }
990
991 // loss of precision here
772 return newSVnv (Atof (start)); 992 return newSVnv (Atof (start));
773 993
774fail: 994fail:
775 return 0; 995 return 0;
776} 996}
… …
778static SV * 998static SV *
779decode_av (dec_t *dec) 999decode_av (dec_t *dec)
780{ 1000{
781 AV *av = newAV (); 1001 AV *av = newAV ();
782 1002
1003 DEC_INC_DEPTH;
783 decode_ws (dec); 1004 decode_ws (dec);
1005
784 if (*dec->cur == ']') 1006 if (*dec->cur == ']')
785 ++dec->cur; 1007 ++dec->cur;
786 else 1008 else
787 for (;;) 1009 for (;;)
788 { 1010 {
… …
806 ERR (", or ] expected while parsing array"); 1028 ERR (", or ] expected while parsing array");
807 1029
808 ++dec->cur; 1030 ++dec->cur;
809 } 1031 }
810 1032
1033 DEC_DEC_DEPTH;
811 return newRV_noinc ((SV *)av); 1034 return newRV_noinc ((SV *)av);
812 1035
813fail: 1036fail:
814 SvREFCNT_dec (av); 1037 SvREFCNT_dec (av);
1038 DEC_DEC_DEPTH;
815 return 0; 1039 return 0;
816} 1040}
817 1041
818static SV * 1042static SV *
819decode_hv (dec_t *dec) 1043decode_hv (dec_t *dec)
820{ 1044{
1045 SV *sv;
821 HV *hv = newHV (); 1046 HV *hv = newHV ();
822 1047
1048 DEC_INC_DEPTH;
823 decode_ws (dec); 1049 decode_ws (dec);
1050
824 if (*dec->cur == '}') 1051 if (*dec->cur == '}')
825 ++dec->cur; 1052 ++dec->cur;
826 else 1053 else
827 for (;;) 1054 for (;;)
828 { 1055 {
829 SV *key, *value;
830
831 decode_ws (dec); EXPECT_CH ('"'); 1056 decode_ws (dec); EXPECT_CH ('"');
832 1057
833 key = decode_str (dec); 1058 // heuristic: assume that
834 if (!key) 1059 // a) decode_str + hv_store_ent are abysmally slow.
835 goto fail; 1060 // b) most hash keys are short, simple ascii text.
1061 // => try to "fast-match" such strings to avoid
1062 // the overhead of decode_str + hv_store_ent.
1063 {
1064 SV *value;
1065 char *p = dec->cur;
1066 char *e = p + 24; // only try up to 24 bytes
836 1067
837 decode_ws (dec); EXPECT_CH (':'); 1068 for (;;)
838
839 value = decode_sv (dec);
840 if (!value)
841 { 1069 {
1070 // the >= 0x80 is true on most architectures
1071 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1072 {
1073 // slow path, back up and use decode_str
1074 SV *key = decode_str (dec);
1075 if (!key)
1076 goto fail;
1077
1078 decode_ws (dec); EXPECT_CH (':');
1079
1080 value = decode_sv (dec);
1081 if (!value)
1082 {
1083 SvREFCNT_dec (key);
1084 goto fail;
1085 }
1086
1087 hv_store_ent (hv, key, value, 0);
842 SvREFCNT_dec (key); 1088 SvREFCNT_dec (key);
1089
1090 break;
1091 }
1092 else if (*p == '"')
1093 {
1094 // fast path, got a simple key
1095 char *key = dec->cur;
1096 int len = p - key;
1097 dec->cur = p + 1;
1098
1099 decode_ws (dec); EXPECT_CH (':');
1100
1101 value = decode_sv (dec);
1102 if (!value)
843 goto fail; 1103 goto fail;
1104
1105 hv_store (hv, key, len, value, 0);
1106
1107 break;
1108 }
1109
1110 ++p;
844 } 1111 }
845 1112 }
846 //TODO: optimise
847 hv_store_ent (hv, key, value, 0);
848 1113
849 decode_ws (dec); 1114 decode_ws (dec);
850 1115
851 if (*dec->cur == '}') 1116 if (*dec->cur == '}')
852 { 1117 {
… …
858 ERR (", or } expected while parsing object/hash"); 1123 ERR (", or } expected while parsing object/hash");
859 1124
860 ++dec->cur; 1125 ++dec->cur;
861 } 1126 }
862 1127
1128 DEC_DEC_DEPTH;
863 return newRV_noinc ((SV *)hv); 1129 sv = newRV_noinc ((SV *)hv);
1130
1131 // check filter callbacks
1132 if (dec->json.flags & F_HOOK)
1133 {
1134 if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
1135 {
1136 HE *cb, *he;
1137
1138 hv_iterinit (hv);
1139 he = hv_iternext (hv);
1140 hv_iterinit (hv);
1141
1142 // the next line creates a mortal sv each time its called.
1143 // might want to optimise this for common cases.
1144 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1145
1146 if (cb)
1147 {
1148 dSP;
1149 int count;
1150
1151 ENTER; SAVETMPS; PUSHMARK (SP);
1152 XPUSHs (HeVAL (he));
1153
1154 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1155
1156 if (count == 1)
1157 {
1158 sv = newSVsv (POPs);
1159 FREETMPS; LEAVE;
1160 return sv;
1161 }
1162
1163 FREETMPS; LEAVE;
1164 }
1165 }
1166
1167 if (dec->json.cb_object)
1168 {
1169 dSP;
1170 int count;
1171
1172 ENTER; SAVETMPS; PUSHMARK (SP);
1173 XPUSHs (sv_2mortal (sv));
1174
1175 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1176
1177 if (count == 1)
1178 {
1179 sv = newSVsv (POPs);
1180 FREETMPS; LEAVE;
1181 return sv;
1182 }
1183
1184 SvREFCNT_inc (sv);
1185 FREETMPS; LEAVE;
1186 }
1187 }
1188
1189 return sv;
864 1190
865fail: 1191fail:
866 SvREFCNT_dec (hv); 1192 SvREFCNT_dec (hv);
1193 DEC_DEC_DEPTH;
867 return 0; 1194 return 0;
868} 1195}
869 1196
870static SV * 1197static SV *
871decode_sv (dec_t *dec) 1198decode_sv (dec_t *dec)
872{ 1199{
873 decode_ws (dec); 1200 decode_ws (dec);
1201
1202 // the beauty of JSON: you need exactly one character lookahead
1203 // to parse anything.
874 switch (*dec->cur) 1204 switch (*dec->cur)
875 { 1205 {
876 case '"': ++dec->cur; return decode_str (dec); 1206 case '"': ++dec->cur; return decode_str (dec);
877 case '[': ++dec->cur; return decode_av (dec); 1207 case '[': ++dec->cur; return decode_av (dec);
878 case '{': ++dec->cur; return decode_hv (dec); 1208 case '{': ++dec->cur; return decode_hv (dec);
… …
884 1214
885 case 't': 1215 case 't':
886 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1216 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
887 { 1217 {
888 dec->cur += 4; 1218 dec->cur += 4;
889 return newSViv (1); 1219 return SvREFCNT_inc (json_true);
890 } 1220 }
891 else 1221 else
892 ERR ("'true' expected"); 1222 ERR ("'true' expected");
893 1223
894 break; 1224 break;
895 1225
896 case 'f': 1226 case 'f':
897 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1227 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
898 { 1228 {
899 dec->cur += 5; 1229 dec->cur += 5;
900 return newSViv (0); 1230 return SvREFCNT_inc (json_false);
901 } 1231 }
902 else 1232 else
903 ERR ("'false' expected"); 1233 ERR ("'false' expected");
904 1234
905 break; 1235 break;
… …
914 ERR ("'null' expected"); 1244 ERR ("'null' expected");
915 1245
916 break; 1246 break;
917 1247
918 default: 1248 default:
919 ERR ("malformed json string, neither array, object, number, string or atom"); 1249 ERR ("malformed JSON string, neither array, object, number, string or atom");
920 break; 1250 break;
921 } 1251 }
922 1252
923fail: 1253fail:
924 return 0; 1254 return 0;
925} 1255}
926 1256
927static SV * 1257static SV *
928decode_json (SV *string, UV flags) 1258decode_json (SV *string, JSON *json, UV *offset_return)
929{ 1259{
1260 dec_t dec;
1261 UV offset;
930 SV *sv; 1262 SV *sv;
931 1263
1264 SvGETMAGIC (string);
1265 SvUPGRADE (string, SVt_PV);
1266
1267 if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags))
1268 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1269 (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags));
1270
932 if (flags & F_UTF8) 1271 if (json->flags & F_UTF8)
933 sv_utf8_downgrade (string, 0); 1272 sv_utf8_downgrade (string, 0);
934 else 1273 else
935 sv_utf8_upgrade (string); 1274 sv_utf8_upgrade (string);
936 1275
937 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1276 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
938 1277
939 dec_t dec; 1278 dec.json = *json;
940 dec.flags = flags;
941 dec.cur = SvPVX (string); 1279 dec.cur = SvPVX (string);
942 dec.end = SvEND (string); 1280 dec.end = SvEND (string);
943 dec.err = 0; 1281 dec.err = 0;
1282 dec.depth = 0;
1283 dec.maxdepth = DEC_DEPTH (dec.json.flags);
944 1284
1285 if (dec.json.cb_object || dec.json.cb_sk_object)
1286 dec.json.flags |= F_HOOK;
1287
1288 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
945 sv = decode_sv (&dec); 1289 sv = decode_sv (&dec);
946 1290
1291 if (!(offset_return || !sv))
1292 {
1293 // check for trailing garbage
1294 decode_ws (&dec);
1295
1296 if (*dec.cur)
1297 {
1298 dec.err = "garbage after JSON object";
1299 SvREFCNT_dec (sv);
1300 sv = 0;
1301 }
1302 }
1303
1304 if (offset_return || !sv)
1305 {
1306 offset = dec.json.flags & F_UTF8
1307 ? dec.cur - SvPVX (string)
1308 : utf8_distance (dec.cur, SvPVX (string));
1309
1310 if (offset_return)
1311 *offset_return = offset;
1312 }
1313
947 if (!sv) 1314 if (!sv)
948 { 1315 {
949 IV offset = dec.flags & F_UTF8
950 ? dec.cur - SvPVX (string)
951 : utf8_distance (dec.cur, SvPVX (string));
952 SV *uni = sv_newmortal (); 1316 SV *uni = sv_newmortal ();
953 1317
954 // horrible hack to silence warning inside pv_uni_display 1318 // horrible hack to silence warning inside pv_uni_display
955 COP cop = *PL_curcop; 1319 COP cop = *PL_curcop;
956 cop.cop_warnings = pWARN_NONE; 1320 cop.cop_warnings = pWARN_NONE;
… …
958 SAVEVPTR (PL_curcop); 1322 SAVEVPTR (PL_curcop);
959 PL_curcop = &cop; 1323 PL_curcop = &cop;
960 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1324 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
961 LEAVE; 1325 LEAVE;
962 1326
963 croak ("%s, at character offset %d (%s)", 1327 croak ("%s, at character offset %d [\"%s\"]",
964 dec.err, 1328 dec.err,
965 (int)offset, 1329 (int)offset,
966 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1330 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
967 } 1331 }
968 1332
969 sv = sv_2mortal (sv); 1333 sv = sv_2mortal (sv);
970 1334
971 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1335 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)"); 1336 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
973 1337
974 return sv; 1338 return sv;
975} 1339}
976 1340
… …
981 1345
982BOOT: 1346BOOT:
983{ 1347{
984 int i; 1348 int i;
985 1349
986 memset (decode_hexdigit, 0xff, 256);
987 for (i = 10; i--; ) 1350 for (i = 0; i < 256; ++i)
988 decode_hexdigit ['0' + i] = i; 1351 decode_hexdigit [i] =
1352 i >= '0' && i <= '9' ? i - '0'
1353 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1354 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1355 : -1;
989 1356
990 for (i = 7; i--; ) 1357 json_stash = gv_stashpv ("JSON::XS" , 1);
1358 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1359
1360 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true );
1361 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
1362}
1363
1364PROTOTYPES: DISABLE
1365
1366void new (char *klass)
1367 PPCODE:
1368{
1369 SV *pv = NEWSV (0, sizeof (JSON));
1370 SvPOK_only (pv);
1371 Zero (SvPVX (pv), 1, JSON);
1372 ((JSON *)SvPVX (pv))->flags = F_DEFAULT;
1373 XPUSHs (sv_2mortal (sv_bless (newRV_noinc (pv), json_stash)));
1374}
1375
1376void ascii (JSON *self, int enable = 1)
1377 ALIAS:
1378 ascii = F_ASCII
1379 latin1 = F_LATIN1
1380 utf8 = F_UTF8
1381 indent = F_INDENT
1382 canonical = F_CANONICAL
1383 space_before = F_SPACE_BEFORE
1384 space_after = F_SPACE_AFTER
1385 pretty = F_PRETTY
1386 allow_nonref = F_ALLOW_NONREF
1387 shrink = F_SHRINK
1388 allow_blessed = F_ALLOW_BLESSED
1389 convert_blessed = F_CONV_BLESSED
1390 PPCODE:
1391{
1392 if (enable)
1393 self->flags |= ix;
1394 else
1395 self->flags &= ~ix;
1396
1397 XPUSHs (ST (0));
1398}
1399
1400void max_depth (JSON *self, UV max_depth = 0x80000000UL)
1401 PPCODE:
1402{
1403 UV log2 = 0;
1404
1405 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1406
1407 while ((1UL << log2) < max_depth)
1408 ++log2;
1409
1410 self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1411
1412 XPUSHs (ST (0));
1413}
1414
1415void max_size (JSON *self, UV max_size = 0)
1416 PPCODE:
1417{
1418 UV log2 = 0;
1419
1420 if (max_size > 0x80000000UL) max_size = 0x80000000UL;
1421 if (max_size == 1) max_size = 2;
1422
1423 while ((1UL << log2) < max_size)
1424 ++log2;
1425
1426 self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1427
1428 XPUSHs (ST (0));
1429}
1430
1431void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1432 PPCODE:
1433{
1434 SvREFCNT_dec (self->cb_object);
1435 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
1436
1437 XPUSHs (ST (0));
1438}
1439
1440void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
1441 PPCODE:
1442{
1443 if (!self->cb_sk_object)
1444 self->cb_sk_object = newHV ();
1445
1446 if (SvOK (cb))
1447 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1448 else
991 { 1449 {
992 decode_hexdigit ['a' + i] = 10 + i; 1450 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
993 decode_hexdigit ['A' + i] = 10 + i; 1451
1452 if (!HvKEYS (self->cb_sk_object))
1453 {
1454 SvREFCNT_dec (self->cb_sk_object);
1455 self->cb_sk_object = 0;
1456 }
994 } 1457 }
995 1458
996 json_stash = gv_stashpv ("JSON::XS", 1); 1459 XPUSHs (ST (0));
997} 1460}
998 1461
999PROTOTYPES: DISABLE 1462void encode (JSON *self, SV *scalar)
1463 PPCODE:
1464 XPUSHs (encode_json (scalar, self));
1000 1465
1001SV *new (char *dummy) 1466void decode (JSON *self, SV *jsonstr)
1467 PPCODE:
1468 XPUSHs (decode_json (jsonstr, self, 0));
1469
1470void decode_prefix (JSON *self, SV *jsonstr)
1471 PPCODE:
1472{
1473 UV offset;
1474 EXTEND (SP, 2);
1475 PUSHs (decode_json (jsonstr, self, &offset));
1476 PUSHs (sv_2mortal (newSVuv (offset)));
1477}
1478
1479void DESTROY (JSON *self)
1002 CODE: 1480 CODE:
1003 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1481 SvREFCNT_dec (self->cb_sk_object);
1004 OUTPUT: 1482 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 1483
1039PROTOTYPES: ENABLE 1484PROTOTYPES: ENABLE
1040 1485
1041void to_json (SV *scalar) 1486void to_json (SV *scalar)
1042 PPCODE: 1487 PPCODE:
1488{
1489 JSON json = { F_DEFAULT | F_UTF8 };
1043 XPUSHs (encode_json (scalar, F_UTF8)); 1490 XPUSHs (encode_json (scalar, &json));
1491}
1044 1492
1045void from_json (SV *jsonstr) 1493void from_json (SV *jsonstr)
1046 PPCODE: 1494 PPCODE:
1495{
1496 JSON json = { F_DEFAULT | F_UTF8 };
1047 XPUSHs (decode_json (jsonstr, F_UTF8)); 1497 XPUSHs (decode_json (jsonstr, &json, 0));
1498}
1048 1499

Diff Legend

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