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 1.13 by root, Sat Mar 24 22:55:16 2007 UTC vs.
Revision 1.54 by root, Tue Jul 10 16:22:31 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 256 // 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", 0);
495
496 if (to_json)
497 {
498 int count;
499 dSP;
500
501 ENTER; SAVETMPS; PUSHMARK (SP);
502 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
503
504 // calling with G_SCALAR ensures that we always get a 1 return value
505 PUTBACK;
506 call_sv ((SV *)GvCV (to_json), G_SCALAR);
507 SPAGAIN;
508
509 // catch this surprisingly common error
510 if (SvROK (TOPs) && SvRV (TOPs) == sv)
511 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
512
513 sv = POPs;
514 PUTBACK;
515
516 encode_sv (enc, sv);
517
518 FREETMPS; LEAVE;
519 }
520 else if (enc->json.flags & F_ALLOW_BLESSED)
521 encode_str (enc, "null", 4, 0);
522 else
523 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
524 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
525 }
526 else if (enc->json.flags & F_ALLOW_BLESSED)
527 encode_str (enc, "null", 4, 0);
528 else
529 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",
530 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
531 }
532 }
533 else if (svt == SVt_PVHV)
534 encode_hv (enc, (HV *)sv);
535 else if (svt == SVt_PVAV)
536 encode_av (enc, (AV *)sv);
537 else if (svt < SVt_PVAV)
538 {
539 STRLEN len = 0;
540 char *pv = svt ? SvPV (sv, len) : 0;
541
542 if (len == 1 && *pv == '1')
543 encode_str (enc, "true", 4, 0);
544 else if (len == 1 && *pv == '0')
545 encode_str (enc, "false", 5, 0);
546 else
547 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
548 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
549 }
550 else
551 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
552 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
416} 553}
417 554
418static void 555static void
419encode_sv (enc_t *enc, SV *sv) 556encode_sv (enc_t *enc, SV *sv)
420{ 557{
… …
428 encode_str (enc, str, len, SvUTF8 (sv)); 565 encode_str (enc, str, len, SvUTF8 (sv));
429 encode_ch (enc, '"'); 566 encode_ch (enc, '"');
430 } 567 }
431 else if (SvNOKp (sv)) 568 else if (SvNOKp (sv))
432 { 569 {
570 // trust that perl will do the right thing w.r.t. JSON syntax.
433 need (enc, NV_DIG + 32); 571 need (enc, NV_DIG + 32);
434 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 572 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
435 enc->cur += strlen (enc->cur); 573 enc->cur += strlen (enc->cur);
436 } 574 }
437 else if (SvIOKp (sv)) 575 else if (SvIOKp (sv))
438 { 576 {
439 need (enc, 64); 577 // we assume we can always read an IV as a UV
578 if (SvUV (sv) & ~(UV)0x7fff)
579 {
580 // large integer, use the (rather slow) snprintf way.
581 need (enc, sizeof (UV) * 3);
440 enc->cur += 582 enc->cur +=
441 SvIsUV(sv) 583 SvIsUV(sv)
442 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 584 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
443 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 585 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
586 }
587 else
588 {
589 // optimise the "small number case"
590 // code will likely be branchless and use only a single multiplication
591 I32 i = SvIV (sv);
592 U32 u;
593 char digit, nz = 0;
594
595 need (enc, 6);
596
597 *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
598 u = i < 0 ? -i : i;
599
600 // convert to 4.28 fixed-point representation
601 u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
602
603 // now output digit by digit, each time masking out the integer part
604 // and multiplying by 5 while moving the decimal point one to the right,
605 // resulting in a net multiplication by 10.
606 // we always write the digit to memory but conditionally increment
607 // the pointer, to ease the usage of conditional move instructions.
608 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5;
609 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5;
610 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5;
611 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5;
612 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
613 }
444 } 614 }
445 else if (SvROK (sv)) 615 else if (SvROK (sv))
446 { 616 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)) 617 else if (!SvOK (sv))
463 encode_str (enc, "null", 4, 0); 618 encode_str (enc, "null", 4, 0);
464 else 619 else
465 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 620 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
466 SvPV_nolen (sv), SvFLAGS (sv)); 621 SvPV_nolen (sv), SvFLAGS (sv));
467} 622}
468 623
469static SV * 624static SV *
470encode_json (SV *scalar, UV flags) 625encode_json (SV *scalar, JSON *json)
471{ 626{
627 enc_t enc;
628
472 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 629 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
473 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 630 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
474 631
475 enc_t enc; 632 enc.json = *json;
476 enc.flags = flags;
477 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 633 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
478 enc.cur = SvPVX (enc.sv); 634 enc.cur = SvPVX (enc.sv);
479 enc.end = SvEND (enc.sv); 635 enc.end = SvEND (enc.sv);
480 enc.indent = 0; 636 enc.indent = 0;
481 enc.max_depth = 0x7fffffffUL; 637 enc.maxdepth = DEC_DEPTH (enc.json.flags);
482 638
483 SvPOK_only (enc.sv); 639 SvPOK_only (enc.sv);
484 encode_sv (&enc, scalar); 640 encode_sv (&enc, scalar);
485 641
642 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
643 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
644
486 if (!(flags & (F_ASCII | F_UTF8))) 645 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
487 SvUTF8_on (enc.sv); 646 SvUTF8_on (enc.sv);
488 647
489 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
490
491 if (enc.flags & F_SHRINK) 648 if (enc.json.flags & F_SHRINK)
492 shrink (enc.sv); 649 shrink (enc.sv);
493 650
494 return enc.sv; 651 return enc.sv;
495} 652}
496 653
… …
501typedef struct 658typedef struct
502{ 659{
503 char *cur; // current parser pointer 660 char *cur; // current parser pointer
504 char *end; // end of input string 661 char *end; // end of input string
505 const char *err; // parse error, if != 0 662 const char *err; // parse error, if != 0
506 UV flags; // F_* 663 JSON json;
664 U32 depth; // recursion depth
665 U32 maxdepth; // recursion depth limit
507} dec_t; 666} dec_t;
508 667
509static void 668inline void
510decode_ws (dec_t *dec) 669decode_ws (dec_t *dec)
511{ 670{
512 for (;;) 671 for (;;)
513 { 672 {
514 char ch = *dec->cur; 673 char ch = *dec->cur;
… …
520 ++dec->cur; 679 ++dec->cur;
521 } 680 }
522} 681}
523 682
524#define ERR(reason) SB dec->err = reason; goto fail; SE 683#define ERR(reason) SB dec->err = reason; goto fail; SE
684
525#define EXPECT_CH(ch) SB \ 685#define EXPECT_CH(ch) SB \
526 if (*dec->cur != ch) \ 686 if (*dec->cur != ch) \
527 ERR (# ch " expected"); \ 687 ERR (# ch " expected"); \
528 ++dec->cur; \ 688 ++dec->cur; \
529 SE 689 SE
530 690
691#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)")
692#define DEC_DEC_DEPTH --dec->depth
693
531static SV *decode_sv (dec_t *dec); 694static SV *decode_sv (dec_t *dec);
532 695
533static signed char decode_hexdigit[256]; 696static signed char decode_hexdigit[256];
534 697
535static UV 698static UV
536decode_4hex (dec_t *dec) 699decode_4hex (dec_t *dec)
537{ 700{
538 signed char d1, d2, d3, d4; 701 signed char d1, d2, d3, d4;
539 unsigned char *cur = (unsigned char *)dec->cur; 702 unsigned char *cur = (unsigned char *)dec->cur;
540 703
541 d1 = decode_hexdigit [cur [0]]; if (d1 < 0) ERR ("four hexadecimal digits expected"); 704 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"); 705 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"); 706 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"); 707 d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected");
545 708
546 dec->cur += 4; 709 dec->cur += 4;
547 710
548 return ((UV)d1) << 12 711 return ((UV)d1) << 12
549 | ((UV)d2) << 8 712 | ((UV)d2) << 8
… …
557static SV * 720static SV *
558decode_str (dec_t *dec) 721decode_str (dec_t *dec)
559{ 722{
560 SV *sv = 0; 723 SV *sv = 0;
561 int utf8 = 0; 724 int utf8 = 0;
725 char *dec_cur = dec->cur;
562 726
563 do 727 do
564 { 728 {
565 char buf [SHORT_STRING_LEN + UTF8_MAX_LEN]; 729 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
566 char *cur = buf; 730 char *cur = buf;
567 731
568 do 732 do
569 { 733 {
570 unsigned char ch = *(unsigned char *)dec->cur++; 734 unsigned char ch = *(unsigned char *)dec_cur++;
571 735
572 if (ch == '"') 736 if (expect_false (ch == '"'))
573 { 737 {
574 --dec->cur; 738 --dec_cur;
575 break; 739 break;
576 } 740 }
577 else if (ch == '\\') 741 else if (expect_false (ch == '\\'))
578 { 742 {
579 switch (*dec->cur) 743 switch (*dec_cur)
580 { 744 {
581 case '\\': 745 case '\\':
582 case '/': 746 case '/':
583 case '"': *cur++ = *dec->cur++; break; 747 case '"': *cur++ = *dec_cur++; break;
584 748
585 case 'b': ++dec->cur; *cur++ = '\010'; break; 749 case 'b': ++dec_cur; *cur++ = '\010'; break;
586 case 't': ++dec->cur; *cur++ = '\011'; break; 750 case 't': ++dec_cur; *cur++ = '\011'; break;
587 case 'n': ++dec->cur; *cur++ = '\012'; break; 751 case 'n': ++dec_cur; *cur++ = '\012'; break;
588 case 'f': ++dec->cur; *cur++ = '\014'; break; 752 case 'f': ++dec_cur; *cur++ = '\014'; break;
589 case 'r': ++dec->cur; *cur++ = '\015'; break; 753 case 'r': ++dec_cur; *cur++ = '\015'; break;
590 754
591 case 'u': 755 case 'u':
592 { 756 {
593 UV lo, hi; 757 UV lo, hi;
594 ++dec->cur; 758 ++dec_cur;
595 759
760 dec->cur = dec_cur;
596 hi = decode_4hex (dec); 761 hi = decode_4hex (dec);
762 dec_cur = dec->cur;
597 if (hi == (UV)-1) 763 if (hi == (UV)-1)
598 goto fail; 764 goto fail;
599 765
600 // possibly a surrogate pair 766 // possibly a surrogate pair
601 if (hi >= 0xd800) 767 if (hi >= 0xd800)
602 if (hi < 0xdc00) 768 if (hi < 0xdc00)
603 { 769 {
604 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 770 if (dec_cur [0] != '\\' || dec_cur [1] != 'u')
605 ERR ("missing low surrogate character in surrogate pair"); 771 ERR ("missing low surrogate character in surrogate pair");
606 772
607 dec->cur += 2; 773 dec_cur += 2;
608 774
775 dec->cur = dec_cur;
609 lo = decode_4hex (dec); 776 lo = decode_4hex (dec);
777 dec_cur = dec->cur;
610 if (lo == (UV)-1) 778 if (lo == (UV)-1)
611 goto fail; 779 goto fail;
612 780
613 if (lo < 0xdc00 || lo >= 0xe000) 781 if (lo < 0xdc00 || lo >= 0xe000)
614 ERR ("surrogate pair expected"); 782 ERR ("surrogate pair expected");
… …
628 *cur++ = hi; 796 *cur++ = hi;
629 } 797 }
630 break; 798 break;
631 799
632 default: 800 default:
633 --dec->cur; 801 --dec_cur;
634 ERR ("illegal backslash escape sequence in string"); 802 ERR ("illegal backslash escape sequence in string");
635 } 803 }
636 } 804 }
637 else if (ch >= 0x20 && ch <= 0x7f) 805 else if (expect_true (ch >= 0x20 && ch <= 0x7f))
638 *cur++ = ch; 806 *cur++ = ch;
639 else if (ch >= 0x80) 807 else if (ch >= 0x80)
640 { 808 {
641 --dec->cur;
642
643 STRLEN clen; 809 STRLEN clen;
810 UV uch;
811
812 --dec_cur;
813
644 UV uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen); 814 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
645 if (clen == (STRLEN)-1) 815 if (clen == (STRLEN)-1)
646 ERR ("malformed UTF-8 character in JSON string"); 816 ERR ("malformed UTF-8 character in JSON string");
647 817
648 do 818 do
649 {
650 *cur++ = *dec->cur++; 819 *cur++ = *dec_cur++;
651 }
652 while (--clen); 820 while (--clen);
653 821
654 utf8 = 1; 822 utf8 = 1;
655 } 823 }
656 else if (!ch)
657 ERR ("unexpected end of string while parsing json string");
658 else 824 else
825 {
826 --dec_cur;
827
828 if (!ch)
829 ERR ("unexpected end of string while parsing JSON string");
830 else
659 ERR ("invalid character encountered"); 831 ERR ("invalid character encountered while parsing JSON string");
660 832 }
661 } 833 }
662 while (cur < buf + SHORT_STRING_LEN); 834 while (cur < buf + SHORT_STRING_LEN);
663 835
836 {
664 STRLEN len = cur - buf; 837 STRLEN len = cur - buf;
665 838
666 if (sv) 839 if (sv)
667 { 840 {
668 SvGROW (sv, SvCUR (sv) + len + 1); 841 SvGROW (sv, SvCUR (sv) + len + 1);
669 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 842 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
670 SvCUR_set (sv, SvCUR (sv) + len); 843 SvCUR_set (sv, SvCUR (sv) + len);
671 } 844 }
672 else 845 else
673 sv = newSVpvn (buf, len); 846 sv = newSVpvn (buf, len);
674 } 847 }
848 }
675 while (*dec->cur != '"'); 849 while (*dec_cur != '"');
676 850
677 ++dec->cur; 851 ++dec_cur;
678 852
679 if (sv) 853 if (sv)
680 { 854 {
681 SvPOK_only (sv); 855 SvPOK_only (sv);
682 *SvEND (sv) = 0; 856 *SvEND (sv) = 0;
… …
685 SvUTF8_on (sv); 859 SvUTF8_on (sv);
686 } 860 }
687 else 861 else
688 sv = newSVpvn ("", 0); 862 sv = newSVpvn ("", 0);
689 863
864 dec->cur = dec_cur;
690 return sv; 865 return sv;
691 866
692fail: 867fail:
868 dec->cur = dec_cur;
693 return 0; 869 return 0;
694} 870}
695 871
696static SV * 872static SV *
697decode_num (dec_t *dec) 873decode_num (dec_t *dec)
… …
755 is_nv = 1; 931 is_nv = 1;
756 } 932 }
757 933
758 if (!is_nv) 934 if (!is_nv)
759 { 935 {
760 UV uv; 936 // 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); 937 if (*start == '-')
762 if (numtype & IS_NUMBER_IN_UV) 938 switch (dec->cur - start)
763 if (numtype & IS_NUMBER_NEG)
764 { 939 {
765 if (uv < (UV)IV_MIN) 940 case 2: return newSViv (-( start [1] - '0' * 1));
766 return newSViv (-(IV)uv); 941 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11));
942 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
943 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
767 } 944 }
945 else
946 switch (dec->cur - start)
947 {
948 case 1: return newSViv ( start [0] - '0' * 1);
949 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
950 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
951 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
952 }
953
954 {
955 UV uv;
956 int numtype = grok_number (start, dec->cur - start, &uv);
957 if (numtype & IS_NUMBER_IN_UV)
958 if (numtype & IS_NUMBER_NEG)
959 {
960 if (uv < (UV)IV_MIN)
961 return newSViv (-(IV)uv);
962 }
768 else 963 else
769 return newSVuv (uv); 964 return newSVuv (uv);
965
966 // here would likely be the place for bigint support
770 } 967 }
968 }
771 969
970 // if we ever support bigint or bigfloat, this is the place for bigfloat
772 return newSVnv (Atof (start)); 971 return newSVnv (Atof (start));
773 972
774fail: 973fail:
775 return 0; 974 return 0;
776} 975}
… …
778static SV * 977static SV *
779decode_av (dec_t *dec) 978decode_av (dec_t *dec)
780{ 979{
781 AV *av = newAV (); 980 AV *av = newAV ();
782 981
982 DEC_INC_DEPTH;
783 decode_ws (dec); 983 decode_ws (dec);
984
784 if (*dec->cur == ']') 985 if (*dec->cur == ']')
785 ++dec->cur; 986 ++dec->cur;
786 else 987 else
787 for (;;) 988 for (;;)
788 { 989 {
… …
806 ERR (", or ] expected while parsing array"); 1007 ERR (", or ] expected while parsing array");
807 1008
808 ++dec->cur; 1009 ++dec->cur;
809 } 1010 }
810 1011
1012 DEC_DEC_DEPTH;
811 return newRV_noinc ((SV *)av); 1013 return newRV_noinc ((SV *)av);
812 1014
813fail: 1015fail:
814 SvREFCNT_dec (av); 1016 SvREFCNT_dec (av);
1017 DEC_DEC_DEPTH;
815 return 0; 1018 return 0;
816} 1019}
817 1020
818static SV * 1021static SV *
819decode_hv (dec_t *dec) 1022decode_hv (dec_t *dec)
820{ 1023{
1024 SV *sv;
821 HV *hv = newHV (); 1025 HV *hv = newHV ();
822 1026
1027 DEC_INC_DEPTH;
823 decode_ws (dec); 1028 decode_ws (dec);
1029
824 if (*dec->cur == '}') 1030 if (*dec->cur == '}')
825 ++dec->cur; 1031 ++dec->cur;
826 else 1032 else
827 for (;;) 1033 for (;;)
828 { 1034 {
829 SV *key, *value;
830
831 decode_ws (dec); EXPECT_CH ('"'); 1035 decode_ws (dec); EXPECT_CH ('"');
832 1036
833 key = decode_str (dec); 1037 // heuristic: assume that
834 if (!key) 1038 // a) decode_str + hv_store_ent are abysmally slow.
835 goto fail; 1039 // b) most hash keys are short, simple ascii text.
1040 // => try to "fast-match" such strings to avoid
1041 // the overhead of decode_str + hv_store_ent.
1042 {
1043 SV *value;
1044 char *p = dec->cur;
1045 char *e = p + 24; // only try up to 24 bytes
836 1046
837 decode_ws (dec); EXPECT_CH (':'); 1047 for (;;)
838
839 value = decode_sv (dec);
840 if (!value)
841 { 1048 {
1049 // the >= 0x80 is true on most architectures
1050 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1051 {
1052 // slow path, back up and use decode_str
1053 SV *key = decode_str (dec);
1054 if (!key)
1055 goto fail;
1056
1057 decode_ws (dec); EXPECT_CH (':');
1058
1059 value = decode_sv (dec);
1060 if (!value)
1061 {
1062 SvREFCNT_dec (key);
1063 goto fail;
1064 }
1065
1066 hv_store_ent (hv, key, value, 0);
842 SvREFCNT_dec (key); 1067 SvREFCNT_dec (key);
1068
1069 break;
1070 }
1071 else if (*p == '"')
1072 {
1073 // fast path, got a simple key
1074 char *key = dec->cur;
1075 int len = p - key;
1076 dec->cur = p + 1;
1077
1078 decode_ws (dec); EXPECT_CH (':');
1079
1080 value = decode_sv (dec);
1081 if (!value)
843 goto fail; 1082 goto fail;
1083
1084 hv_store (hv, key, len, value, 0);
1085
1086 break;
1087 }
1088
1089 ++p;
844 } 1090 }
845 1091 }
846 //TODO: optimise
847 hv_store_ent (hv, key, value, 0);
848 1092
849 decode_ws (dec); 1093 decode_ws (dec);
850 1094
851 if (*dec->cur == '}') 1095 if (*dec->cur == '}')
852 { 1096 {
… …
858 ERR (", or } expected while parsing object/hash"); 1102 ERR (", or } expected while parsing object/hash");
859 1103
860 ++dec->cur; 1104 ++dec->cur;
861 } 1105 }
862 1106
1107 DEC_DEC_DEPTH;
863 return newRV_noinc ((SV *)hv); 1108 sv = newRV_noinc ((SV *)hv);
1109
1110 // check filter callbacks
1111 if (dec->json.flags & F_HOOK)
1112 {
1113 if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
1114 {
1115 HE *cb, *he;
1116
1117 hv_iterinit (hv);
1118 he = hv_iternext (hv);
1119 hv_iterinit (hv);
1120
1121 // the next line creates a mortal sv each time its called.
1122 // might want to optimise this for common cases.
1123 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1124
1125 if (cb)
1126 {
1127 dSP;
1128 int count;
1129
1130 ENTER; SAVETMPS; PUSHMARK (SP);
1131 XPUSHs (HeVAL (he));
1132
1133 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1134
1135 if (count == 1)
1136 {
1137 sv = newSVsv (POPs);
1138 FREETMPS; LEAVE;
1139 return sv;
1140 }
1141
1142 FREETMPS; LEAVE;
1143 }
1144 }
1145
1146 if (dec->json.cb_object)
1147 {
1148 dSP;
1149 int count;
1150
1151 ENTER; SAVETMPS; PUSHMARK (SP);
1152 XPUSHs (sv_2mortal (sv));
1153
1154 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1155
1156 if (count == 1)
1157 {
1158 sv = newSVsv (POPs);
1159 FREETMPS; LEAVE;
1160 return sv;
1161 }
1162
1163 SvREFCNT_inc (sv);
1164 FREETMPS; LEAVE;
1165 }
1166 }
1167
1168 return sv;
864 1169
865fail: 1170fail:
866 SvREFCNT_dec (hv); 1171 SvREFCNT_dec (hv);
1172 DEC_DEC_DEPTH;
867 return 0; 1173 return 0;
868} 1174}
869 1175
870static SV * 1176static SV *
871decode_sv (dec_t *dec) 1177decode_sv (dec_t *dec)
872{ 1178{
873 decode_ws (dec); 1179 decode_ws (dec);
1180
1181 // the beauty of JSON: you need exactly one character lookahead
1182 // to parse anything.
874 switch (*dec->cur) 1183 switch (*dec->cur)
875 { 1184 {
876 case '"': ++dec->cur; return decode_str (dec); 1185 case '"': ++dec->cur; return decode_str (dec);
877 case '[': ++dec->cur; return decode_av (dec); 1186 case '[': ++dec->cur; return decode_av (dec);
878 case '{': ++dec->cur; return decode_hv (dec); 1187 case '{': ++dec->cur; return decode_hv (dec);
… …
884 1193
885 case 't': 1194 case 't':
886 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1195 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
887 { 1196 {
888 dec->cur += 4; 1197 dec->cur += 4;
889 return newSViv (1); 1198 return SvREFCNT_inc (json_true);
890 } 1199 }
891 else 1200 else
892 ERR ("'true' expected"); 1201 ERR ("'true' expected");
893 1202
894 break; 1203 break;
895 1204
896 case 'f': 1205 case 'f':
897 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1206 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
898 { 1207 {
899 dec->cur += 5; 1208 dec->cur += 5;
900 return newSViv (0); 1209 return SvREFCNT_inc (json_false);
901 } 1210 }
902 else 1211 else
903 ERR ("'false' expected"); 1212 ERR ("'false' expected");
904 1213
905 break; 1214 break;
… …
914 ERR ("'null' expected"); 1223 ERR ("'null' expected");
915 1224
916 break; 1225 break;
917 1226
918 default: 1227 default:
919 ERR ("malformed json string, neither array, object, number, string or atom"); 1228 ERR ("malformed JSON string, neither array, object, number, string or atom");
920 break; 1229 break;
921 } 1230 }
922 1231
923fail: 1232fail:
924 return 0; 1233 return 0;
925} 1234}
926 1235
927static SV * 1236static SV *
928decode_json (SV *string, UV flags) 1237decode_json (SV *string, JSON *json, UV *offset_return)
929{ 1238{
1239 dec_t dec;
1240 UV offset;
930 SV *sv; 1241 SV *sv;
931 1242
1243 SvGETMAGIC (string);
1244 SvUPGRADE (string, SVt_PV);
1245
1246 if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags))
1247 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1248 (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags));
1249
932 if (flags & F_UTF8) 1250 if (json->flags & F_UTF8)
933 sv_utf8_downgrade (string, 0); 1251 sv_utf8_downgrade (string, 0);
934 else 1252 else
935 sv_utf8_upgrade (string); 1253 sv_utf8_upgrade (string);
936 1254
937 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1255 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
938 1256
939 dec_t dec; 1257 dec.json = *json;
940 dec.flags = flags;
941 dec.cur = SvPVX (string); 1258 dec.cur = SvPVX (string);
942 dec.end = SvEND (string); 1259 dec.end = SvEND (string);
943 dec.err = 0; 1260 dec.err = 0;
1261 dec.depth = 0;
1262 dec.maxdepth = DEC_DEPTH (dec.json.flags);
944 1263
1264 if (dec.json.cb_object || dec.json.cb_sk_object)
1265 dec.json.flags |= F_HOOK;
1266
1267 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
945 sv = decode_sv (&dec); 1268 sv = decode_sv (&dec);
946 1269
1270 if (!(offset_return || !sv))
1271 {
1272 // check for trailing garbage
1273 decode_ws (&dec);
1274
1275 if (*dec.cur)
1276 {
1277 dec.err = "garbage after JSON object";
1278 SvREFCNT_dec (sv);
1279 sv = 0;
1280 }
1281 }
1282
1283 if (offset_return || !sv)
1284 {
1285 offset = dec.json.flags & F_UTF8
1286 ? dec.cur - SvPVX (string)
1287 : utf8_distance (dec.cur, SvPVX (string));
1288
1289 if (offset_return)
1290 *offset_return = offset;
1291 }
1292
947 if (!sv) 1293 if (!sv)
948 { 1294 {
949 IV offset = dec.flags & F_UTF8
950 ? dec.cur - SvPVX (string)
951 : utf8_distance (dec.cur, SvPVX (string));
952 SV *uni = sv_newmortal (); 1295 SV *uni = sv_newmortal ();
953 1296
954 // horrible hack to silence warning inside pv_uni_display 1297 // horrible hack to silence warning inside pv_uni_display
955 COP cop = *PL_curcop; 1298 COP cop = *PL_curcop;
956 cop.cop_warnings = pWARN_NONE; 1299 cop.cop_warnings = pWARN_NONE;
… …
958 SAVEVPTR (PL_curcop); 1301 SAVEVPTR (PL_curcop);
959 PL_curcop = &cop; 1302 PL_curcop = &cop;
960 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1303 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
961 LEAVE; 1304 LEAVE;
962 1305
963 croak ("%s, at character offset %d (%s)", 1306 croak ("%s, at character offset %d [\"%s\"]",
964 dec.err, 1307 dec.err,
965 (int)offset, 1308 (int)offset,
966 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1309 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
967 } 1310 }
968 1311
969 sv = sv_2mortal (sv); 1312 sv = sv_2mortal (sv);
970 1313
971 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1314 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)"); 1315 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
973 1316
974 return sv; 1317 return sv;
975} 1318}
976 1319
… …
981 1324
982BOOT: 1325BOOT:
983{ 1326{
984 int i; 1327 int i;
985 1328
986 memset (decode_hexdigit, 0xff, 256);
987 for (i = 10; i--; ) 1329 for (i = 0; i < 256; ++i)
988 decode_hexdigit ['0' + i] = i; 1330 decode_hexdigit [i] =
1331 i >= '0' && i <= '9' ? i - '0'
1332 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1333 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1334 : -1;
989 1335
990 for (i = 7; i--; ) 1336 json_stash = gv_stashpv ("JSON::XS" , 1);
1337 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1338
1339 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true );
1340 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
1341}
1342
1343PROTOTYPES: DISABLE
1344
1345void new (char *klass)
1346 PPCODE:
1347{
1348 SV *pv = NEWSV (0, sizeof (JSON));
1349 SvPOK_only (pv);
1350 Zero (SvPVX (pv), 1, JSON);
1351 ((JSON *)SvPVX (pv))->flags = F_DEFAULT;
1352 XPUSHs (sv_2mortal (sv_bless (newRV_noinc (pv), json_stash)));
1353}
1354
1355void ascii (JSON *self, int enable = 1)
1356 ALIAS:
1357 ascii = F_ASCII
1358 latin1 = F_LATIN1
1359 utf8 = F_UTF8
1360 indent = F_INDENT
1361 canonical = F_CANONICAL
1362 space_before = F_SPACE_BEFORE
1363 space_after = F_SPACE_AFTER
1364 pretty = F_PRETTY
1365 allow_nonref = F_ALLOW_NONREF
1366 shrink = F_SHRINK
1367 allow_blessed = F_ALLOW_BLESSED
1368 convert_blessed = F_CONV_BLESSED
1369 PPCODE:
1370{
1371 if (enable)
1372 self->flags |= ix;
1373 else
1374 self->flags &= ~ix;
1375
1376 XPUSHs (ST (0));
1377}
1378
1379void max_depth (JSON *self, UV max_depth = 0x80000000UL)
1380 PPCODE:
1381{
1382 UV log2 = 0;
1383
1384 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1385
1386 while ((1UL << log2) < max_depth)
1387 ++log2;
1388
1389 self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1390
1391 XPUSHs (ST (0));
1392}
1393
1394void max_size (JSON *self, UV max_size = 0)
1395 PPCODE:
1396{
1397 UV log2 = 0;
1398
1399 if (max_size > 0x80000000UL) max_size = 0x80000000UL;
1400 if (max_size == 1) max_size = 2;
1401
1402 while ((1UL << log2) < max_size)
1403 ++log2;
1404
1405 self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1406
1407 XPUSHs (ST (0));
1408}
1409
1410void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1411 PPCODE:
1412{
1413 SvREFCNT_dec (self->cb_object);
1414 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
1415
1416 XPUSHs (ST (0));
1417}
1418
1419void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
1420 PPCODE:
1421{
1422 if (!self->cb_sk_object)
1423 self->cb_sk_object = newHV ();
1424
1425 if (SvOK (cb))
1426 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1427 else
991 { 1428 {
992 decode_hexdigit ['a' + i] = 10 + i; 1429 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
993 decode_hexdigit ['A' + i] = 10 + i; 1430
1431 if (!HvKEYS (self->cb_sk_object))
1432 {
1433 SvREFCNT_dec (self->cb_sk_object);
1434 self->cb_sk_object = 0;
1435 }
994 } 1436 }
995 1437
996 json_stash = gv_stashpv ("JSON::XS", 1); 1438 XPUSHs (ST (0));
997} 1439}
998 1440
999PROTOTYPES: DISABLE 1441void encode (JSON *self, SV *scalar)
1442 PPCODE:
1443 XPUSHs (encode_json (scalar, self));
1000 1444
1001SV *new (char *dummy) 1445void decode (JSON *self, SV *jsonstr)
1446 PPCODE:
1447 XPUSHs (decode_json (jsonstr, self, 0));
1448
1449void decode_prefix (JSON *self, SV *jsonstr)
1450 PPCODE:
1451{
1452 UV offset;
1453 EXTEND (SP, 2);
1454 PUSHs (decode_json (jsonstr, self, &offset));
1455 PUSHs (sv_2mortal (newSVuv (offset)));
1456}
1457
1458void DESTROY (JSON *self)
1002 CODE: 1459 CODE:
1003 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1460 SvREFCNT_dec (self->cb_sk_object);
1004 OUTPUT: 1461 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 1462
1039PROTOTYPES: ENABLE 1463PROTOTYPES: ENABLE
1040 1464
1041void to_json (SV *scalar) 1465void to_json (SV *scalar)
1042 PPCODE: 1466 PPCODE:
1467{
1468 JSON json = { F_DEFAULT | F_UTF8 };
1043 XPUSHs (encode_json (scalar, F_UTF8)); 1469 XPUSHs (encode_json (scalar, &json));
1470}
1044 1471
1045void from_json (SV *jsonstr) 1472void from_json (SV *jsonstr)
1046 PPCODE: 1473 PPCODE:
1474{
1475 JSON json = { F_DEFAULT | F_UTF8 };
1047 XPUSHs (decode_json (jsonstr, F_UTF8)); 1476 XPUSHs (decode_json (jsonstr, &json, 0));
1477}
1048 1478

Diff Legend

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