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 1.22 by root, Sat Mar 31 14:20:06 2007 UTC vs.
Revision 1.53 by root, Tue Jul 10 15:45:34 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 0x00000001UL 20#define F_ASCII 0x00000001UL
21#define F_LATIN1 0x00000002UL
10#define F_UTF8 0x00000002UL 22#define F_UTF8 0x00000004UL
11#define F_INDENT 0x00000004UL 23#define F_INDENT 0x00000008UL
12#define F_CANONICAL 0x00000008UL 24#define F_CANONICAL 0x00000010UL
13#define F_SPACE_BEFORE 0x00000010UL 25#define F_SPACE_BEFORE 0x00000020UL
14#define F_SPACE_AFTER 0x00000020UL 26#define F_SPACE_AFTER 0x00000040UL
15#define F_ALLOW_NONREF 0x00000080UL 27#define F_ALLOW_NONREF 0x00000100UL
16#define F_SHRINK 0x00000100UL 28#define F_SHRINK 0x00000200UL
29#define F_ALLOW_BLESSED 0x00000400UL
30#define F_CONV_BLESSED 0x00000800UL
17#define F_MAXDEPTH 0xf8000000UL 31#define F_MAXDEPTH 0xf8000000UL
18#define S_MAXDEPTH 27 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
19 36
20#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH)) 37#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
21 38#define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE ))
22// F_SELFCONVERT? <=> to_json/toJson
23// F_BLESSED? <=> { $__class__$ => }
24 39
25#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 40#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
26#define F_DEFAULT (12UL << S_MAXDEPTH) 41#define F_DEFAULT (9UL << S_MAXDEPTH)
27 42
28#define INIT_SIZE 32 // initial scalar size to be allocated 43#define INIT_SIZE 32 // initial scalar size to be allocated
29#define INDENT_STEP 3 // spaces per indentation level 44#define INDENT_STEP 3 // spaces per indentation level
30 45
31#define UTF8_MAX_LEN 11 // for perls UTF-X: max. number of octets per character
32#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
33 47
34#define SB do { 48#define SB do {
35#define SE } while (0) 49#define SE } while (0)
36 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
37static 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;
38 70
39///////////////////////////////////////////////////////////////////////////// 71/////////////////////////////////////////////////////////////////////////////
40// utility functions 72// utility functions
41 73
42static UV * 74inline void
43SvJSON (SV *sv)
44{
45 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash))
46 croak ("object is not of type JSON::XS");
47
48 return &SvUVX (SvRV (sv));
49}
50
51static void
52shrink (SV *sv) 75shrink (SV *sv)
53{ 76{
54 sv_utf8_downgrade (sv, 1); 77 sv_utf8_downgrade (sv, 1);
55 if (SvLEN (sv) > SvCUR (sv) + 1) 78 if (SvLEN (sv) > SvCUR (sv) + 1)
56 { 79 {
… …
65// 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
66// case of an error. 89// case of an error.
67// we special-case "safe" characters from U+80 .. U+7FF, 90// we special-case "safe" characters from U+80 .. U+7FF,
68// but use the very good perl function to parse anything else. 91// but use the very good perl function to parse anything else.
69// note that we never call this function for a ascii codepoints 92// note that we never call this function for a ascii codepoints
70static UV 93inline UV
71decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) 94decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
72{ 95{
73 if (s[0] > 0xdf || s[0] < 0xc2) 96 if (expect_false (s[0] > 0xdf || s[0] < 0xc2))
74 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); 97 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
75 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) 98 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)
76 { 99 {
77 *clen = 2; 100 *clen = 2;
78 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 101 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
79 } 102 }
80 else 103 else
104 {
105 *clen = (STRLEN)-1;
81 return (UV)-1; 106 return (UV)-1;
107 }
82} 108}
83 109
84///////////////////////////////////////////////////////////////////////////// 110/////////////////////////////////////////////////////////////////////////////
85// encoder 111// encoder
86 112
… …
88typedef struct 114typedef struct
89{ 115{
90 char *cur; // SvPVX (sv) + current output position 116 char *cur; // SvPVX (sv) + current output position
91 char *end; // SvEND (sv) 117 char *end; // SvEND (sv)
92 SV *sv; // result scalar 118 SV *sv; // result scalar
93 U32 flags; // F_* 119 JSON json;
94 U32 indent; // indentation level 120 U32 indent; // indentation level
95 U32 maxdepth; // max. indentation/recursion level 121 U32 maxdepth; // max. indentation/recursion level
96} enc_t; 122} enc_t;
97 123
98static void 124inline void
99need (enc_t *enc, STRLEN len) 125need (enc_t *enc, STRLEN len)
100{ 126{
101 if (enc->cur + len >= enc->end) 127 if (expect_false (enc->cur + len >= enc->end))
102 { 128 {
103 STRLEN cur = enc->cur - SvPVX (enc->sv); 129 STRLEN cur = enc->cur - SvPVX (enc->sv);
104 SvGROW (enc->sv, cur + len + 1); 130 SvGROW (enc->sv, cur + len + 1);
105 enc->cur = SvPVX (enc->sv) + cur; 131 enc->cur = SvPVX (enc->sv) + cur;
106 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv); 132 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
107 } 133 }
108} 134}
109 135
110static void 136inline void
111encode_ch (enc_t *enc, char ch) 137encode_ch (enc_t *enc, char ch)
112{ 138{
113 need (enc, 1); 139 need (enc, 1);
114 *enc->cur++ = ch; 140 *enc->cur++ = ch;
115} 141}
… …
123 149
124 while (str < end) 150 while (str < end)
125 { 151 {
126 unsigned char ch = *(unsigned char *)str; 152 unsigned char ch = *(unsigned char *)str;
127 153
128 if (ch >= 0x20 && ch < 0x80) // most common case 154 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
129 { 155 {
130 if (ch == '"') // but with slow exceptions 156 if (expect_false (ch == '"')) // but with slow exceptions
131 { 157 {
132 need (enc, len += 1); 158 need (enc, len += 1);
133 *enc->cur++ = '\\'; 159 *enc->cur++ = '\\';
134 *enc->cur++ = '"'; 160 *enc->cur++ = '"';
135 } 161 }
136 else if (ch == '\\') 162 else if (expect_false (ch == '\\'))
137 { 163 {
138 need (enc, len += 1); 164 need (enc, len += 1);
139 *enc->cur++ = '\\'; 165 *enc->cur++ = '\\';
140 *enc->cur++ = '\\'; 166 *enc->cur++ = '\\';
141 } 167 }
… …
159 STRLEN clen; 185 STRLEN clen;
160 UV uch; 186 UV uch;
161 187
162 if (is_utf8) 188 if (is_utf8)
163 { 189 {
164 //uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
165 uch = decode_utf8 (str, end - str, &clen); 190 uch = decode_utf8 (str, end - str, &clen);
166 if (clen == (STRLEN)-1) 191 if (clen == (STRLEN)-1)
167 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);
168 } 193 }
169 else 194 else
… …
173 } 198 }
174 199
175 if (uch > 0x10FFFFUL) 200 if (uch > 0x10FFFFUL)
176 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);
177 202
178 if (uch < 0x80 || enc->flags & F_ASCII) 203 if (uch < 0x80 || enc->json.flags & F_ASCII || (enc->json.flags & F_LATIN1 && uch > 0xFF))
179 { 204 {
180 if (uch > 0xFFFFUL) 205 if (uch > 0xFFFFUL)
181 { 206 {
182 need (enc, len += 11); 207 need (enc, len += 11);
183 sprintf (enc->cur, "\\u%04x\\u%04x", 208 sprintf (enc->cur, "\\u%04x\\u%04x",
… …
197 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 222 *enc->cur++ = hexdigit [(uch >> 0) & 15];
198 } 223 }
199 224
200 str += clen; 225 str += clen;
201 } 226 }
227 else if (enc->json.flags & F_LATIN1)
228 {
229 *enc->cur++ = uch;
230 str += clen;
231 }
202 else if (is_utf8) 232 else if (is_utf8)
203 { 233 {
204 need (enc, len += clen); 234 need (enc, len += clen);
205 do 235 do
206 { 236 {
… …
208 } 238 }
209 while (--clen); 239 while (--clen);
210 } 240 }
211 else 241 else
212 { 242 {
213 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
214 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 244 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
215 ++str; 245 ++str;
216 } 246 }
217 } 247 }
218 } 248 }
… …
220 250
221 --len; 251 --len;
222 } 252 }
223} 253}
224 254
225static void 255inline void
226encode_indent (enc_t *enc) 256encode_indent (enc_t *enc)
227{ 257{
228 if (enc->flags & F_INDENT) 258 if (enc->json.flags & F_INDENT)
229 { 259 {
230 int spaces = enc->indent * INDENT_STEP; 260 int spaces = enc->indent * INDENT_STEP;
231 261
232 need (enc, spaces); 262 need (enc, spaces);
233 memset (enc->cur, ' ', spaces); 263 memset (enc->cur, ' ', spaces);
234 enc->cur += spaces; 264 enc->cur += spaces;
235 } 265 }
236} 266}
237 267
238static void 268inline void
239encode_space (enc_t *enc) 269encode_space (enc_t *enc)
240{ 270{
241 need (enc, 1); 271 need (enc, 1);
242 encode_ch (enc, ' '); 272 encode_ch (enc, ' ');
243} 273}
244 274
245static void 275inline void
246encode_nl (enc_t *enc) 276encode_nl (enc_t *enc)
247{ 277{
248 if (enc->flags & F_INDENT) 278 if (enc->json.flags & F_INDENT)
249 { 279 {
250 need (enc, 1); 280 need (enc, 1);
251 encode_ch (enc, '\n'); 281 encode_ch (enc, '\n');
252 } 282 }
253} 283}
254 284
255static void 285inline void
256encode_comma (enc_t *enc) 286encode_comma (enc_t *enc)
257{ 287{
258 encode_ch (enc, ','); 288 encode_ch (enc, ',');
259 289
260 if (enc->flags & F_INDENT) 290 if (enc->json.flags & F_INDENT)
261 encode_nl (enc); 291 encode_nl (enc);
262 else if (enc->flags & F_SPACE_AFTER) 292 else if (enc->json.flags & F_SPACE_AFTER)
263 encode_space (enc); 293 encode_space (enc);
264} 294}
265 295
266static void encode_sv (enc_t *enc, SV *sv); 296static void encode_sv (enc_t *enc, SV *sv);
267 297
… …
310 else 340 else
311 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 341 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
312 342
313 encode_ch (enc, '"'); 343 encode_ch (enc, '"');
314 344
315 if (enc->flags & F_SPACE_BEFORE) encode_space (enc); 345 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
316 encode_ch (enc, ':'); 346 encode_ch (enc, ':');
317 if (enc->flags & F_SPACE_AFTER ) encode_space (enc); 347 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
318 encode_sv (enc, HeVAL (he)); 348 encode_sv (enc, HeVAL (he));
319} 349}
320 350
321// compare hash entries, used when all keys are bytestrings 351// compare hash entries, used when all keys are bytestrings
322static int 352static int
… …
357 { 387 {
358 // for canonical output we have to sort by keys first 388 // for canonical output we have to sort by keys first
359 // actually, this is mostly due to the stupid so-called 389 // actually, this is mostly due to the stupid so-called
360 // security workaround added somewhere in 5.8.x. 390 // security workaround added somewhere in 5.8.x.
361 // that randomises hash orderings 391 // that randomises hash orderings
362 if (enc->flags & F_CANONICAL) 392 if (enc->json.flags & F_CANONICAL)
363 { 393 {
364 HE *he, *hes [count]; // if your compiler dies here, you need to enable C99 mode
365 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
366 401
367 i = 0; 402 i = 0;
368 while ((he = hv_iternext (hv))) 403 while ((he = hv_iternext (hv)))
369 { 404 {
370 hes [i++] = he; 405 hes [i++] = he;
… …
405 440
406 encode_nl (enc); 441 encode_nl (enc);
407 } 442 }
408 else 443 else
409 { 444 {
410 SV *sv;
411 HE *he = hv_iternext (hv); 445 HE *he = hv_iternext (hv);
412 446
413 for (;;) 447 for (;;)
414 { 448 {
415 encode_indent (enc); 449 encode_indent (enc);
… …
430 464
431// encode objects, arrays and special \0=false and \1=true values. 465// encode objects, arrays and special \0=false and \1=true values.
432static void 466static void
433encode_rv (enc_t *enc, SV *sv) 467encode_rv (enc_t *enc, SV *sv)
434{ 468{
469 svtype svt;
470
435 SvGETMAGIC (sv); 471 SvGETMAGIC (sv);
436
437 svtype svt = SvTYPE (sv); 472 svt = SvTYPE (sv);
438 473
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 dSP;
499
500 ENTER; SAVETMPS; PUSHMARK (SP);
501 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
502
503 // calling with G_SCALAR ensures that we always get a 1 return value
504 // check anyways.
505 PUTBACK;
506 assert (1 == 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 }
439 if (svt == SVt_PVHV) 533 else if (svt == SVt_PVHV)
440 encode_hv (enc, (HV *)sv); 534 encode_hv (enc, (HV *)sv);
441 else if (svt == SVt_PVAV) 535 else if (svt == SVt_PVAV)
442 encode_av (enc, (AV *)sv); 536 encode_av (enc, (AV *)sv);
443 else if (svt < SVt_PVAV) 537 else if (svt < SVt_PVAV)
444 { 538 {
445 if (SvNIOK (sv) && SvIV (sv) == 0) 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')
446 encode_str (enc, "false", 5, 0); 545 encode_str (enc, "false", 5, 0);
447 else if (SvNIOK (sv) && SvIV (sv) == 1)
448 encode_str (enc, "true", 4, 0);
449 else 546 else
450 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", 547 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
451 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 548 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
452 } 549 }
453 else 550 else
… …
468 encode_str (enc, str, len, SvUTF8 (sv)); 565 encode_str (enc, str, len, SvUTF8 (sv));
469 encode_ch (enc, '"'); 566 encode_ch (enc, '"');
470 } 567 }
471 else if (SvNOKp (sv)) 568 else if (SvNOKp (sv))
472 { 569 {
570 // trust that perl will do the right thing w.r.t. JSON syntax.
473 need (enc, NV_DIG + 32); 571 need (enc, NV_DIG + 32);
474 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 572 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
475 enc->cur += strlen (enc->cur); 573 enc->cur += strlen (enc->cur);
476 } 574 }
477 else if (SvIOKp (sv)) 575 else if (SvIOKp (sv))
478 { 576 {
479 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);
480 enc->cur += 582 enc->cur +=
481 SvIsUV(sv) 583 SvIsUV(sv)
482 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 584 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
483 : 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 }
484 } 614 }
485 else if (SvROK (sv)) 615 else if (SvROK (sv))
486 encode_rv (enc, SvRV (sv)); 616 encode_rv (enc, SvRV (sv));
487 else if (!SvOK (sv)) 617 else if (!SvOK (sv))
488 encode_str (enc, "null", 4, 0); 618 encode_str (enc, "null", 4, 0);
… …
490 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",
491 SvPV_nolen (sv), SvFLAGS (sv)); 621 SvPV_nolen (sv), SvFLAGS (sv));
492} 622}
493 623
494static SV * 624static SV *
495encode_json (SV *scalar, U32 flags) 625encode_json (SV *scalar, JSON *json)
496{ 626{
627 enc_t enc;
628
497 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 629 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
498 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)");
499 631
500 enc_t enc; 632 enc.json = *json;
501 enc.flags = flags;
502 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 633 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
503 enc.cur = SvPVX (enc.sv); 634 enc.cur = SvPVX (enc.sv);
504 enc.end = SvEND (enc.sv); 635 enc.end = SvEND (enc.sv);
505 enc.indent = 0; 636 enc.indent = 0;
506 enc.maxdepth = DEC_DEPTH (flags); 637 enc.maxdepth = DEC_DEPTH (enc.json.flags);
507 638
508 SvPOK_only (enc.sv); 639 SvPOK_only (enc.sv);
509 encode_sv (&enc, scalar); 640 encode_sv (&enc, scalar);
510 641
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
511 if (!(flags & (F_ASCII | F_UTF8))) 645 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
512 SvUTF8_on (enc.sv); 646 SvUTF8_on (enc.sv);
513 647
514 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
515
516 if (enc.flags & F_SHRINK) 648 if (enc.json.flags & F_SHRINK)
517 shrink (enc.sv); 649 shrink (enc.sv);
518 650
519 return enc.sv; 651 return enc.sv;
520} 652}
521 653
… …
526typedef struct 658typedef struct
527{ 659{
528 char *cur; // current parser pointer 660 char *cur; // current parser pointer
529 char *end; // end of input string 661 char *end; // end of input string
530 const char *err; // parse error, if != 0 662 const char *err; // parse error, if != 0
531 U32 flags; // F_* 663 JSON json;
532 U32 depth; // recursion depth 664 U32 depth; // recursion depth
533 U32 maxdepth; // recursion depth limit 665 U32 maxdepth; // recursion depth limit
534} dec_t; 666} dec_t;
535 667
536static void 668inline void
537decode_ws (dec_t *dec) 669decode_ws (dec_t *dec)
538{ 670{
539 for (;;) 671 for (;;)
540 { 672 {
541 char ch = *dec->cur; 673 char ch = *dec->cur;
… …
567decode_4hex (dec_t *dec) 699decode_4hex (dec_t *dec)
568{ 700{
569 signed char d1, d2, d3, d4; 701 signed char d1, d2, d3, d4;
570 unsigned char *cur = (unsigned char *)dec->cur; 702 unsigned char *cur = (unsigned char *)dec->cur;
571 703
572 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");
573 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");
574 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");
575 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");
576 708
577 dec->cur += 4; 709 dec->cur += 4;
578 710
579 return ((UV)d1) << 12 711 return ((UV)d1) << 12
580 | ((UV)d2) << 8 712 | ((UV)d2) << 8
… …
588static SV * 720static SV *
589decode_str (dec_t *dec) 721decode_str (dec_t *dec)
590{ 722{
591 SV *sv = 0; 723 SV *sv = 0;
592 int utf8 = 0; 724 int utf8 = 0;
725 char *dec_cur = dec->cur;
593 726
594 do 727 do
595 { 728 {
596 char buf [SHORT_STRING_LEN + UTF8_MAX_LEN]; 729 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
597 char *cur = buf; 730 char *cur = buf;
598 731
599 do 732 do
600 { 733 {
601 unsigned char ch = *(unsigned char *)dec->cur++; 734 unsigned char ch = *(unsigned char *)dec_cur++;
602 735
603 if (ch == '"') 736 if (expect_false (ch == '"'))
604 { 737 {
605 --dec->cur; 738 --dec_cur;
606 break; 739 break;
607 } 740 }
608 else if (ch == '\\') 741 else if (expect_false (ch == '\\'))
609 { 742 {
610 switch (*dec->cur) 743 switch (*dec_cur)
611 { 744 {
612 case '\\': 745 case '\\':
613 case '/': 746 case '/':
614 case '"': *cur++ = *dec->cur++; break; 747 case '"': *cur++ = *dec_cur++; break;
615 748
616 case 'b': ++dec->cur; *cur++ = '\010'; break; 749 case 'b': ++dec_cur; *cur++ = '\010'; break;
617 case 't': ++dec->cur; *cur++ = '\011'; break; 750 case 't': ++dec_cur; *cur++ = '\011'; break;
618 case 'n': ++dec->cur; *cur++ = '\012'; break; 751 case 'n': ++dec_cur; *cur++ = '\012'; break;
619 case 'f': ++dec->cur; *cur++ = '\014'; break; 752 case 'f': ++dec_cur; *cur++ = '\014'; break;
620 case 'r': ++dec->cur; *cur++ = '\015'; break; 753 case 'r': ++dec_cur; *cur++ = '\015'; break;
621 754
622 case 'u': 755 case 'u':
623 { 756 {
624 UV lo, hi; 757 UV lo, hi;
625 ++dec->cur; 758 ++dec_cur;
626 759
760 dec->cur = dec_cur;
627 hi = decode_4hex (dec); 761 hi = decode_4hex (dec);
762 dec_cur = dec->cur;
628 if (hi == (UV)-1) 763 if (hi == (UV)-1)
629 goto fail; 764 goto fail;
630 765
631 // possibly a surrogate pair 766 // possibly a surrogate pair
632 if (hi >= 0xd800) 767 if (hi >= 0xd800)
633 if (hi < 0xdc00) 768 if (hi < 0xdc00)
634 { 769 {
635 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 770 if (dec_cur [0] != '\\' || dec_cur [1] != 'u')
636 ERR ("missing low surrogate character in surrogate pair"); 771 ERR ("missing low surrogate character in surrogate pair");
637 772
638 dec->cur += 2; 773 dec_cur += 2;
639 774
775 dec->cur = dec_cur;
640 lo = decode_4hex (dec); 776 lo = decode_4hex (dec);
777 dec_cur = dec->cur;
641 if (lo == (UV)-1) 778 if (lo == (UV)-1)
642 goto fail; 779 goto fail;
643 780
644 if (lo < 0xdc00 || lo >= 0xe000) 781 if (lo < 0xdc00 || lo >= 0xe000)
645 ERR ("surrogate pair expected"); 782 ERR ("surrogate pair expected");
… …
659 *cur++ = hi; 796 *cur++ = hi;
660 } 797 }
661 break; 798 break;
662 799
663 default: 800 default:
664 --dec->cur; 801 --dec_cur;
665 ERR ("illegal backslash escape sequence in string"); 802 ERR ("illegal backslash escape sequence in string");
666 } 803 }
667 } 804 }
668 else if (ch >= 0x20 && ch <= 0x7f) 805 else if (expect_true (ch >= 0x20 && ch <= 0x7f))
669 *cur++ = ch; 806 *cur++ = ch;
670 else if (ch >= 0x80) 807 else if (ch >= 0x80)
671 { 808 {
672 --dec->cur;
673
674 STRLEN clen; 809 STRLEN clen;
810 UV uch;
811
812 --dec_cur;
813
675 UV uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen); 814 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
676 if (clen == (STRLEN)-1) 815 if (clen == (STRLEN)-1)
677 ERR ("malformed UTF-8 character in JSON string"); 816 ERR ("malformed UTF-8 character in JSON string");
678 817
679 do 818 do
680 {
681 *cur++ = *dec->cur++; 819 *cur++ = *dec_cur++;
682 }
683 while (--clen); 820 while (--clen);
684 821
685 utf8 = 1; 822 utf8 = 1;
686 } 823 }
687 else if (!ch)
688 ERR ("unexpected end of string while parsing json string");
689 else 824 else
825 {
826 --dec_cur;
827
828 if (!ch)
829 ERR ("unexpected end of string while parsing JSON string");
830 else
690 ERR ("invalid character encountered"); 831 ERR ("invalid character encountered while parsing JSON string");
691 832 }
692 } 833 }
693 while (cur < buf + SHORT_STRING_LEN); 834 while (cur < buf + SHORT_STRING_LEN);
694 835
836 {
695 STRLEN len = cur - buf; 837 STRLEN len = cur - buf;
696 838
697 if (sv) 839 if (sv)
698 { 840 {
699 SvGROW (sv, SvCUR (sv) + len + 1); 841 SvGROW (sv, SvCUR (sv) + len + 1);
700 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 842 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
701 SvCUR_set (sv, SvCUR (sv) + len); 843 SvCUR_set (sv, SvCUR (sv) + len);
702 } 844 }
703 else 845 else
704 sv = newSVpvn (buf, len); 846 sv = newSVpvn (buf, len);
705 } 847 }
848 }
706 while (*dec->cur != '"'); 849 while (*dec_cur != '"');
707 850
708 ++dec->cur; 851 ++dec_cur;
709 852
710 if (sv) 853 if (sv)
711 { 854 {
712 SvPOK_only (sv); 855 SvPOK_only (sv);
713 *SvEND (sv) = 0; 856 *SvEND (sv) = 0;
… …
716 SvUTF8_on (sv); 859 SvUTF8_on (sv);
717 } 860 }
718 else 861 else
719 sv = newSVpvn ("", 0); 862 sv = newSVpvn ("", 0);
720 863
864 dec->cur = dec_cur;
721 return sv; 865 return sv;
722 866
723fail: 867fail:
868 dec->cur = dec_cur;
724 return 0; 869 return 0;
725} 870}
726 871
727static SV * 872static SV *
728decode_num (dec_t *dec) 873decode_num (dec_t *dec)
… …
786 is_nv = 1; 931 is_nv = 1;
787 } 932 }
788 933
789 if (!is_nv) 934 if (!is_nv)
790 { 935 {
791 UV uv; 936 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so
792 int numtype = grok_number (start, dec->cur - start, &uv); 937 if (*start == '-')
793 if (numtype & IS_NUMBER_IN_UV) 938 switch (dec->cur - start)
794 if (numtype & IS_NUMBER_NEG)
795 { 939 {
796 if (uv < (UV)IV_MIN) 940 case 2: return newSViv (-( start [1] - '0' * 1));
797 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));
798 } 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 }
799 else 963 else
800 return newSVuv (uv); 964 return newSVuv (uv);
965
966 // here would likely be the place for bigint support
801 } 967 }
968 }
802 969
970 // if we ever support bigint or bigfloat, this is the place for bigfloat
803 return newSVnv (Atof (start)); 971 return newSVnv (Atof (start));
804 972
805fail: 973fail:
806 return 0; 974 return 0;
807} 975}
… …
851} 1019}
852 1020
853static SV * 1021static SV *
854decode_hv (dec_t *dec) 1022decode_hv (dec_t *dec)
855{ 1023{
1024 SV *sv;
856 HV *hv = newHV (); 1025 HV *hv = newHV ();
857 1026
858 DEC_INC_DEPTH; 1027 DEC_INC_DEPTH;
859 decode_ws (dec); 1028 decode_ws (dec);
860 1029
861 if (*dec->cur == '}') 1030 if (*dec->cur == '}')
862 ++dec->cur; 1031 ++dec->cur;
863 else 1032 else
864 for (;;) 1033 for (;;)
865 { 1034 {
866 SV *key, *value;
867
868 decode_ws (dec); EXPECT_CH ('"'); 1035 decode_ws (dec); EXPECT_CH ('"');
869 1036
870 key = decode_str (dec); 1037 // heuristic: assume that
871 if (!key) 1038 // a) decode_str + hv_store_ent are abysmally slow.
872 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
873 1046
874 decode_ws (dec); EXPECT_CH (':'); 1047 for (;;)
875
876 value = decode_sv (dec);
877 if (!value)
878 { 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);
879 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)
880 goto fail; 1082 goto fail;
1083
1084 hv_store (hv, key, len, value, 0);
1085
1086 break;
1087 }
1088
1089 ++p;
881 } 1090 }
882 1091 }
883 hv_store_ent (hv, key, value, 0);
884 SvREFCNT_dec (key);
885 1092
886 decode_ws (dec); 1093 decode_ws (dec);
887 1094
888 if (*dec->cur == '}') 1095 if (*dec->cur == '}')
889 { 1096 {
… …
896 1103
897 ++dec->cur; 1104 ++dec->cur;
898 } 1105 }
899 1106
900 DEC_DEC_DEPTH; 1107 DEC_DEC_DEPTH;
901 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;
902 1169
903fail: 1170fail:
904 SvREFCNT_dec (hv); 1171 SvREFCNT_dec (hv);
905 DEC_DEC_DEPTH; 1172 DEC_DEC_DEPTH;
906 return 0; 1173 return 0;
… …
908 1175
909static SV * 1176static SV *
910decode_sv (dec_t *dec) 1177decode_sv (dec_t *dec)
911{ 1178{
912 decode_ws (dec); 1179 decode_ws (dec);
1180
1181 // the beauty of JSON: you need exactly one character lookahead
1182 // to parse anything.
913 switch (*dec->cur) 1183 switch (*dec->cur)
914 { 1184 {
915 case '"': ++dec->cur; return decode_str (dec); 1185 case '"': ++dec->cur; return decode_str (dec);
916 case '[': ++dec->cur; return decode_av (dec); 1186 case '[': ++dec->cur; return decode_av (dec);
917 case '{': ++dec->cur; return decode_hv (dec); 1187 case '{': ++dec->cur; return decode_hv (dec);
… …
923 1193
924 case 't': 1194 case 't':
925 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1195 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
926 { 1196 {
927 dec->cur += 4; 1197 dec->cur += 4;
928 return newSViv (1); 1198 return SvREFCNT_inc (json_true);
929 } 1199 }
930 else 1200 else
931 ERR ("'true' expected"); 1201 ERR ("'true' expected");
932 1202
933 break; 1203 break;
934 1204
935 case 'f': 1205 case 'f':
936 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1206 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
937 { 1207 {
938 dec->cur += 5; 1208 dec->cur += 5;
939 return newSViv (0); 1209 return SvREFCNT_inc (json_false);
940 } 1210 }
941 else 1211 else
942 ERR ("'false' expected"); 1212 ERR ("'false' expected");
943 1213
944 break; 1214 break;
… …
953 ERR ("'null' expected"); 1223 ERR ("'null' expected");
954 1224
955 break; 1225 break;
956 1226
957 default: 1227 default:
958 ERR ("malformed json string, neither array, object, number, string or atom"); 1228 ERR ("malformed JSON string, neither array, object, number, string or atom");
959 break; 1229 break;
960 } 1230 }
961 1231
962fail: 1232fail:
963 return 0; 1233 return 0;
964} 1234}
965 1235
966static SV * 1236static SV *
967decode_json (SV *string, U32 flags) 1237decode_json (SV *string, JSON *json, UV *offset_return)
968{ 1238{
1239 dec_t dec;
1240 UV offset;
969 SV *sv; 1241 SV *sv;
970 1242
1243 SvGETMAGIC (string);
971 SvUPGRADE (string, SVt_PV); 1244 SvUPGRADE (string, SVt_PV);
972 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
973 if (flags & F_UTF8) 1250 if (json->flags & F_UTF8)
974 sv_utf8_downgrade (string, 0); 1251 sv_utf8_downgrade (string, 0);
975 else 1252 else
976 sv_utf8_upgrade (string); 1253 sv_utf8_upgrade (string);
977 1254
978 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1255 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
979 1256
980 dec_t dec; 1257 dec.json = *json;
981 dec.flags = flags;
982 dec.cur = SvPVX (string); 1258 dec.cur = SvPVX (string);
983 dec.end = SvEND (string); 1259 dec.end = SvEND (string);
984 dec.err = 0; 1260 dec.err = 0;
985 dec.depth = 0; 1261 dec.depth = 0;
986 dec.maxdepth = DEC_DEPTH (dec.flags); 1262 dec.maxdepth = DEC_DEPTH (dec.json.flags);
987 1263
1264 if (dec.json.cb_object || dec.json.cb_sk_object)
1265 dec.json.flags |= F_HOOK;
1266
988 *dec.end = 0; // this should basically be a nop, too, but make sure its there 1267 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
989 sv = decode_sv (&dec); 1268 sv = decode_sv (&dec);
990 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
991 if (!sv) 1293 if (!sv)
992 { 1294 {
993 IV offset = dec.flags & F_UTF8
994 ? dec.cur - SvPVX (string)
995 : utf8_distance (dec.cur, SvPVX (string));
996 SV *uni = sv_newmortal (); 1295 SV *uni = sv_newmortal ();
997 1296
998 // horrible hack to silence warning inside pv_uni_display 1297 // horrible hack to silence warning inside pv_uni_display
999 COP cop = *PL_curcop; 1298 COP cop = *PL_curcop;
1000 cop.cop_warnings = pWARN_NONE; 1299 cop.cop_warnings = pWARN_NONE;
… …
1002 SAVEVPTR (PL_curcop); 1301 SAVEVPTR (PL_curcop);
1003 PL_curcop = &cop; 1302 PL_curcop = &cop;
1004 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);
1005 LEAVE; 1304 LEAVE;
1006 1305
1007 croak ("%s, at character offset %d (%s)", 1306 croak ("%s, at character offset %d [\"%s\"]",
1008 dec.err, 1307 dec.err,
1009 (int)offset, 1308 (int)offset,
1010 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1309 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1011 } 1310 }
1012 1311
1013 sv = sv_2mortal (sv); 1312 sv = sv_2mortal (sv);
1014 1313
1015 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1314 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))
1016 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)");
1017 1316
1018 return sv; 1317 return sv;
1019} 1318}
1020 1319
… …
1024MODULE = JSON::XS PACKAGE = JSON::XS 1323MODULE = JSON::XS PACKAGE = JSON::XS
1025 1324
1026BOOT: 1325BOOT:
1027{ 1326{
1028 int i; 1327 int i;
1029
1030 memset (decode_hexdigit, 0xff, 256);
1031 1328
1032 for (i = 0; i < 256; ++i) 1329 for (i = 0; i < 256; ++i)
1033 decode_hexdigit [i] = 1330 decode_hexdigit [i] =
1034 i >= '0' && i <= '9' ? i - '0' 1331 i >= '0' && i <= '9' ? i - '0'
1035 : i >= 'a' && i <= 'f' ? i - 'a' + 10 1332 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1036 : i >= 'A' && i <= 'F' ? i - 'A' + 10 1333 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1037 : -1; 1334 : -1;
1038 1335
1039 json_stash = gv_stashpv ("JSON::XS", 1); 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);
1040} 1341}
1041 1342
1042PROTOTYPES: DISABLE 1343PROTOTYPES: DISABLE
1043 1344
1044SV *new (char *dummy) 1345void new (char *klass)
1045 CODE: 1346 PPCODE:
1046 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1347{
1047 OUTPUT: 1348 SV *pv = NEWSV (0, sizeof (JSON));
1048 RETVAL 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}
1049 1354
1050SV *ascii (SV *self, int enable = 1) 1355void ascii (JSON *self, int enable = 1)
1051 ALIAS: 1356 ALIAS:
1052 ascii = F_ASCII 1357 ascii = F_ASCII
1358 latin1 = F_LATIN1
1053 utf8 = F_UTF8 1359 utf8 = F_UTF8
1054 indent = F_INDENT 1360 indent = F_INDENT
1055 canonical = F_CANONICAL 1361 canonical = F_CANONICAL
1056 space_before = F_SPACE_BEFORE 1362 space_before = F_SPACE_BEFORE
1057 space_after = F_SPACE_AFTER 1363 space_after = F_SPACE_AFTER
1058 pretty = F_PRETTY 1364 pretty = F_PRETTY
1059 allow_nonref = F_ALLOW_NONREF 1365 allow_nonref = F_ALLOW_NONREF
1060 shrink = F_SHRINK 1366 shrink = F_SHRINK
1367 allow_blessed = F_ALLOW_BLESSED
1368 convert_blessed = F_CONV_BLESSED
1061 CODE: 1369 PPCODE:
1062{ 1370{
1063 UV *uv = SvJSON (self);
1064 if (enable) 1371 if (enable)
1065 *uv |= ix; 1372 self->flags |= ix;
1066 else 1373 else
1067 *uv &= ~ix; 1374 self->flags &= ~ix;
1068 1375
1069 RETVAL = newSVsv (self); 1376 XPUSHs (ST (0));
1070} 1377}
1071 OUTPUT:
1072 RETVAL
1073 1378
1074SV *max_depth (SV *self, int max_depth = 0x80000000UL) 1379void max_depth (JSON *self, UV max_depth = 0x80000000UL)
1075 CODE: 1380 PPCODE:
1076{ 1381{
1077 UV *uv = SvJSON (self);
1078 UV log2 = 0; 1382 UV log2 = 0;
1079 1383
1080 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL; 1384 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1081 1385
1082 while ((1UL << log2) < max_depth) 1386 while ((1UL << log2) < max_depth)
1083 ++log2; 1387 ++log2;
1084 1388
1085 *uv = *uv & ~F_MAXDEPTH | (log2 << S_MAXDEPTH); 1389 self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1086 1390
1087 RETVAL = newSVsv (self); 1391 XPUSHs (ST (0));
1088} 1392}
1089 OUTPUT:
1090 RETVAL
1091 1393
1092void encode (SV *self, SV *scalar) 1394void max_size (JSON *self, UV max_size = 0)
1093 PPCODE: 1395 PPCODE:
1094 XPUSHs (encode_json (scalar, *SvJSON (self))); 1396{
1397 UV log2 = 0;
1095 1398
1096void decode (SV *self, SV *jsonstr) 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)
1097 PPCODE: 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
1428 {
1429 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1430
1431 if (!HvKEYS (self->cb_sk_object))
1432 {
1433 SvREFCNT_dec (self->cb_sk_object);
1434 self->cb_sk_object = 0;
1435 }
1436 }
1437
1438 XPUSHs (ST (0));
1439}
1440
1441void encode (JSON *self, SV *scalar)
1442 PPCODE:
1443 XPUSHs (encode_json (scalar, self));
1444
1445void decode (JSON *self, SV *jsonstr)
1446 PPCODE:
1098 XPUSHs (decode_json (jsonstr, *SvJSON (self))); 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)
1459 CODE:
1460 SvREFCNT_dec (self->cb_sk_object);
1461 SvREFCNT_dec (self->cb_object);
1099 1462
1100PROTOTYPES: ENABLE 1463PROTOTYPES: ENABLE
1101 1464
1102void to_json (SV *scalar) 1465void to_json (SV *scalar)
1103 ALIAS:
1104 objToJson = 0
1105 PPCODE: 1466 PPCODE:
1467{
1468 JSON json = { F_DEFAULT | F_UTF8 };
1106 XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8)); 1469 XPUSHs (encode_json (scalar, &json));
1470}
1107 1471
1108void from_json (SV *jsonstr) 1472void from_json (SV *jsonstr)
1109 ALIAS:
1110 jsonToObj = 0
1111 PPCODE: 1473 PPCODE:
1474{
1475 JSON json = { F_DEFAULT | F_UTF8 };
1112 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8)); 1476 XPUSHs (decode_json (jsonstr, &json, 0));
1477}
1113 1478

Diff Legend

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