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 1.1 by root, Thu Mar 22 16:40:16 2007 UTC vs.
Revision 1.126 by root, Sun Feb 21 16:18:19 2016 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 <limits.h>
10#include <float.h>
8 11
12#if defined(__BORLANDC__) || defined(_MSC_VER)
13# define snprintf _snprintf // C compilers have this in stdio.h
14#endif
15
16// some old perls do not have this, try to make it work, no
17// guarantees, though. if it breaks, you get to keep the pieces.
18#ifndef UTF8_MAXBYTES
19# define UTF8_MAXBYTES 13
20#endif
21
22// compatibility with perl <5.18
23#ifndef HvNAMELEN_get
24# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
25#endif
26#ifndef HvNAMELEN
27# define HvNAMELEN(hv) HvNAMELEN_get (hv)
28#endif
29#ifndef HvNAMEUTF8
30# define HvNAMEUTF8(hv) 0
31#endif
32
33// three extra for rounding, sign, and end of string
34#define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 3)
35
9#define F_ASCII 0x00000001 36#define F_ASCII 0x00000001UL
37#define F_LATIN1 0x00000002UL
10#define F_UTF8 0x00000002 38#define F_UTF8 0x00000004UL
11#define F_INDENT 0x00000004 39#define F_INDENT 0x00000008UL
12#define F_CANONICAL 0x00000008 40#define F_CANONICAL 0x00000010UL
13#define F_SPACE_BEFORE 0x00000010 41#define F_SPACE_BEFORE 0x00000020UL
14#define F_SPACE_AFTER 0x00000020 42#define F_SPACE_AFTER 0x00000040UL
15#define F_JSON_RPC 0x00000040 43#define F_ALLOW_NONREF 0x00000100UL
44#define F_SHRINK 0x00000200UL
45#define F_ALLOW_BLESSED 0x00000400UL
46#define F_CONV_BLESSED 0x00000800UL
47#define F_RELAXED 0x00001000UL
48#define F_ALLOW_UNKNOWN 0x00002000UL
49#define F_ALLOW_TAGS 0x00004000UL
50#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
16 51
17#define F_DEFAULT 0 52#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
18 53
19#define INIT_SIZE 32 // initial scalar size to be allocated 54#define INIT_SIZE 32 // initial scalar size to be allocated
55#define INDENT_STEP 3 // spaces per indentation level
56
57#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
58
59#define DECODE_WANTS_OCTETS(json) ((json)->flags & F_UTF8)
20 60
21#define SB do { 61#define SB do {
22#define SE } while (0) 62#define SE } while (0)
23 63
24static HV *json_stash; 64#if __GNUC__ >= 3
65# define expect(expr,value) __builtin_expect ((expr), (value))
66# define INLINE static inline
67#else
68# define expect(expr,value) (expr)
69# define INLINE static
70#endif
71
72#define expect_false(expr) expect ((expr) != 0, 0)
73#define expect_true(expr) expect ((expr) != 0, 1)
74
75#define IN_RANGE_INC(type,val,beg,end) \
76 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \
77 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
78
79#define ERR_NESTING_EXCEEDED "json text or perl structure exceeds maximum nesting level (max_depth set too low?)"
80
81#ifdef USE_ITHREADS
82# define JSON_SLOW 1
83# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1))
84# define BOOL_STASH (bool_stash ? bool_stash : gv_stashpv ("Types::Serialiser::Boolean", 1))
85#else
86# define JSON_SLOW 0
87# define JSON_STASH json_stash
88# define BOOL_STASH bool_stash
89#endif
90
91// the amount of HEs to allocate on the stack, when sorting keys
92#define STACK_HES 64
93
94static HV *json_stash, *bool_stash; // JSON::XS::, Types::Serialiser::Boolean::
95static SV *bool_true, *bool_false, *sv_json;
96
97enum {
98 INCR_M_WS = 0, // initial whitespace skipping, must be 0
99 INCR_M_STR, // inside string
100 INCR_M_BS, // inside backslash
101 INCR_M_C0, // inside comment in initial whitespace sequence
102 INCR_M_C1, // inside comment in other places
103 INCR_M_JSON // outside anything, count nesting
104};
105
106#define INCR_DONE(json) ((json)->incr_nest <= 0 && (json)->incr_mode == INCR_M_JSON)
107
108typedef struct {
109 U32 flags;
110 U32 max_depth;
111 STRLEN max_size;
112
113 SV *cb_object;
114 HV *cb_sk_object;
115
116 // for the incremental parser
117 SV *incr_text; // the source text so far
118 STRLEN incr_pos; // the current offset into the text
119 int incr_nest; // {[]}-nesting level
120 unsigned char incr_mode;
121} JSON;
122
123INLINE void
124json_init (JSON *json)
125{
126 Zero (json, 1, JSON);
127 json->max_depth = 512;
128}
129
130/////////////////////////////////////////////////////////////////////////////
131// utility functions
132
133INLINE SV *
134get_bool (const char *name)
135{
136 SV *sv = get_sv (name, 1);
137
138 SvREADONLY_on (sv);
139 SvREADONLY_on (SvRV (sv));
140
141 return sv;
142}
143
144INLINE void
145shrink (SV *sv)
146{
147 sv_utf8_downgrade (sv, 1);
148
149 if (SvLEN (sv) > SvCUR (sv) + 1)
150 {
151#ifdef SvPV_shrink_to_cur
152 SvPV_shrink_to_cur (sv);
153#elif defined (SvPV_renew)
154 SvPV_renew (sv, SvCUR (sv) + 1);
155#endif
156 }
157}
158
159// decode an utf-8 character and return it, or (UV)-1 in
160// case of an error.
161// we special-case "safe" characters from U+80 .. U+7FF,
162// but use the very good perl function to parse anything else.
163// note that we never call this function for a ascii codepoints
164INLINE UV
165decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
166{
167 if (expect_true (len >= 2
168 && IN_RANGE_INC (char, s[0], 0xc2, 0xdf)
169 && IN_RANGE_INC (char, s[1], 0x80, 0xbf)))
170 {
171 *clen = 2;
172 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
173 }
174 else
175 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
176}
177
178// likewise for encoding, also never called for ascii codepoints
179// this function takes advantage of this fact, although current gccs
180// seem to optimise the check for >= 0x80 away anyways
181INLINE unsigned char *
182encode_utf8 (unsigned char *s, UV ch)
183{
184 if (expect_false (ch < 0x000080))
185 *s++ = ch;
186 else if (expect_true (ch < 0x000800))
187 *s++ = 0xc0 | ( ch >> 6),
188 *s++ = 0x80 | ( ch & 0x3f);
189 else if ( ch < 0x010000)
190 *s++ = 0xe0 | ( ch >> 12),
191 *s++ = 0x80 | ((ch >> 6) & 0x3f),
192 *s++ = 0x80 | ( ch & 0x3f);
193 else if ( ch < 0x110000)
194 *s++ = 0xf0 | ( ch >> 18),
195 *s++ = 0x80 | ((ch >> 12) & 0x3f),
196 *s++ = 0x80 | ((ch >> 6) & 0x3f),
197 *s++ = 0x80 | ( ch & 0x3f);
198
199 return s;
200}
201
202// convert offset pointer to character index, sv must be string
203static STRLEN
204ptr_to_index (SV *sv, char *offset)
205{
206 return SvUTF8 (sv)
207 ? utf8_distance (offset, SvPVX (sv))
208 : offset - SvPVX (sv);
209}
210
211/////////////////////////////////////////////////////////////////////////////
212// fp hell
213
214// scan a group of digits, and a trailing exponent
215static void
216json_atof_scan1 (const char *s, NV *accum, int *expo, int postdp, int maxdepth)
217{
218 UV uaccum = 0;
219 int eaccum = 0;
220
221 // if we recurse too deep, skip all remaining digits
222 // to avoid a stack overflow attack
223 if (expect_false (--maxdepth <= 0))
224 while (((U8)*s - '0') < 10)
225 ++s;
226
227 for (;;)
228 {
229 U8 dig = (U8)*s - '0';
230
231 if (expect_false (dig >= 10))
232 {
233 if (dig == (U8)((U8)'.' - (U8)'0'))
234 {
235 ++s;
236 json_atof_scan1 (s, accum, expo, 1, maxdepth);
237 }
238 else if ((dig | ' ') == 'e' - '0')
239 {
240 int exp2 = 0;
241 int neg = 0;
242
243 ++s;
244
245 if (*s == '-')
246 {
247 ++s;
248 neg = 1;
249 }
250 else if (*s == '+')
251 ++s;
252
253 while ((dig = (U8)*s - '0') < 10)
254 exp2 = exp2 * 10 + *s++ - '0';
255
256 *expo += neg ? -exp2 : exp2;
257 }
258
259 break;
260 }
261
262 ++s;
263
264 uaccum = uaccum * 10 + dig;
265 ++eaccum;
266
267 // if we have too many digits, then recurse for more
268 // we actually do this for rather few digits
269 if (uaccum >= (UV_MAX - 9) / 10)
270 {
271 if (postdp) *expo -= eaccum;
272 json_atof_scan1 (s, accum, expo, postdp, maxdepth);
273 if (postdp) *expo += eaccum;
274
275 break;
276 }
277 }
278
279 // this relies greatly on the quality of the pow ()
280 // implementation of the platform, but a good
281 // implementation is hard to beat.
282 // (IEEE 754 conformant ones are required to be exact)
283 if (postdp) *expo -= eaccum;
284 *accum += uaccum * Perl_pow (10., *expo);
285 *expo += eaccum;
286}
287
288static NV
289json_atof (const char *s)
290{
291 NV accum = 0.;
292 int expo = 0;
293 int neg = 0;
294
295 if (*s == '-')
296 {
297 ++s;
298 neg = 1;
299 }
300
301 // a recursion depth of ten gives us >>500 bits
302 json_atof_scan1 (s, &accum, &expo, 0, 10);
303
304 return neg ? -accum : accum;
305}
306
307// target of scalar reference is bool? -1 == nope, 0 == false, 1 == true
308static int
309ref_bool_type (SV *sv)
310{
311 svtype svt = SvTYPE (sv);
312
313 if (svt < SVt_PVAV)
314 {
315 STRLEN len = 0;
316 char *pv = svt ? SvPV (sv, len) : 0;
317
318 if (len == 1)
319 if (*pv == '1')
320 return 1;
321 else if (*pv == '0')
322 return 0;
323
324 }
325
326 return -1;
327}
328
329// returns whether scalar is not a reference in the sense of allow_nonref
330static int
331json_nonref (SV *scalar)
332{
333 if (!SvROK (scalar))
334 return 1;
335
336 scalar = SvRV (scalar);
337
338 if (SvSTASH (scalar) == bool_stash)
339 return 1;
340
341 if (!SvOBJECT (scalar) && ref_bool_type (scalar) >= 0)
342 return 1;
343
344 return 0;
345}
346
347/////////////////////////////////////////////////////////////////////////////
348// encoder
25 349
26// structure used for encoding JSON 350// structure used for encoding JSON
27typedef struct 351typedef struct
28{ 352{
29 char *cur; 353 char *cur; // SvPVX (sv) + current output position
30 STRLEN len; // SvLEN (sv)
31 char *end; // SvEND (sv) 354 char *end; // SvEND (sv)
32 SV *sv; 355 SV *sv; // result scalar
33 UV flags; 356 JSON json;
34 int max_recurse; 357 U32 indent; // indentation level
35 int indent; 358 UV limit; // escape character values >= this value when encoding
36} enc_t; 359} enc_t;
37 360
38// structure used for decoding JSON 361INLINE void
39typedef struct
40{
41 char *cur;
42 char *end;
43 char *err;
44 UV flags;
45} dec_t;
46
47static UV *
48SvJSON (SV *sv)
49{
50 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash))
51 croak ("object is not of type JSON::XS");
52
53 return &SvUVX (SvRV (sv));
54}
55
56/////////////////////////////////////////////////////////////////////////////
57
58static void
59need (enc_t *enc, STRLEN len) 362need (enc_t *enc, STRLEN len)
60{ 363{
61 if (enc->cur + len >= enc->end) 364 if (expect_false (enc->cur + len >= enc->end))
62 { 365 {
63 STRLEN cur = enc->cur - SvPVX (enc->sv); 366 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
64 SvGROW (enc->sv, cur + len + 1); 367 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
65 enc->cur = SvPVX (enc->sv) + cur; 368 enc->cur = SvPVX (enc->sv) + cur;
66 enc->end = SvEND (enc->sv); 369 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
67 } 370 }
68} 371}
69 372
70static void 373INLINE void
71encode_ch (enc_t *enc, char ch) 374encode_ch (enc_t *enc, char ch)
72{ 375{
73 need (enc, 1); 376 need (enc, 1);
74 *enc->cur++ = ch; 377 *enc->cur++ = ch;
75} 378}
… …
77static void 380static void
78encode_str (enc_t *enc, char *str, STRLEN len, int is_utf8) 381encode_str (enc_t *enc, char *str, STRLEN len, int is_utf8)
79{ 382{
80 char *end = str + len; 383 char *end = str + len;
81 384
385 need (enc, len);
386
82 while (str < end) 387 while (str < end)
83 { 388 {
84 unsigned char ch = *(unsigned char *)str; 389 unsigned char ch = *(unsigned char *)str;
390
85 if (ch >= 0x20 && ch < 0x80) // most common case 391 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
86 { 392 {
393 if (expect_false (ch == '"')) // but with slow exceptions
394 {
395 need (enc, len += 1);
396 *enc->cur++ = '\\';
397 *enc->cur++ = '"';
398 }
399 else if (expect_false (ch == '\\'))
400 {
401 need (enc, len += 1);
402 *enc->cur++ = '\\';
403 *enc->cur++ = '\\';
404 }
405 else
87 *enc->cur++ = ch; 406 *enc->cur++ = ch;
407
88 str++; 408 ++str;
89 } 409 }
90 else 410 else
91 { 411 {
92 STRLEN clen; 412 switch (ch)
93 UV uch;
94
95 if (is_utf8)
96 { 413 {
97 uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY); 414 case '\010': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'b'; ++str; break;
98 if (clen < 0) 415 case '\011': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 't'; ++str; break;
99 croak ("malformed UTF-8 character in string, cannot convert to JSON"); 416 case '\012': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'n'; ++str; break;
100 } 417 case '\014': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'f'; ++str; break;
101 else 418 case '\015': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'r'; ++str; break;
102 {
103 uch = ch;
104 clen = 1;
105 }
106 419
107 need (enc, len += 6); 420 default:
108
109 if (uch < 0xa0 || enc->flags & F_ASCII)
110 {
111 if (uch > 0xFFFFUL)
112 { 421 {
422 STRLEN clen;
423 UV uch;
424
425 if (is_utf8)
426 {
427 uch = decode_utf8 (str, end - str, &clen);
428 if (clen == (STRLEN)-1)
429 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
430 }
431 else
432 {
433 uch = ch;
113 len += 6; 434 clen = 1;
435 }
436
437 if (uch < 0x80/*0x20*/ || uch >= enc->limit)
438 {
439 if (uch >= 0x10000UL)
440 {
441 if (uch >= 0x110000UL)
442 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
443
114 need (enc, len += 6); 444 need (enc, len += 11);
115 sprintf (enc->cur, "\\u%04x\\u%04x", 445 sprintf (enc->cur, "\\u%04x\\u%04x",
116 (uch - 0x10000) / 0x400 + 0xD800, 446 (int)((uch - 0x10000) / 0x400 + 0xD800),
117 (uch - 0x10000) % 0x400 + 0xDC00); 447 (int)((uch - 0x10000) % 0x400 + 0xDC00));
118 enc->cur += 12; 448 enc->cur += 12;
449 }
450 else
451 {
452 need (enc, len += 5);
453 *enc->cur++ = '\\';
454 *enc->cur++ = 'u';
455 *enc->cur++ = PL_hexdigit [ uch >> 12 ];
456 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
457 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
458 *enc->cur++ = PL_hexdigit [(uch >> 0) & 15];
459 }
460
461 str += clen;
119 } 462 }
463 else if (enc->json.flags & F_LATIN1)
464 {
465 *enc->cur++ = uch;
466 str += clen;
467 }
468 else if (is_utf8)
469 {
470 need (enc, len += clen);
471 do
472 {
473 *enc->cur++ = *str++;
474 }
475 while (--clen);
476 }
120 else 477 else
121 { 478 {
122 sprintf (enc->cur, "\\u%04x", uch); 479 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
123 enc->cur += 6; 480 enc->cur = encode_utf8 (enc->cur, uch);
481 ++str;
482 }
124 } 483 }
125 } 484 }
126 else if (is_utf8)
127 {
128 memcpy (enc->cur, str, clen);
129 enc->cur += clen;
130 }
131 else
132 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
133
134 str += clen;
135 } 485 }
136 486
137 --len; 487 --len;
138 } 488 }
139} 489}
140 490
141#define INDENT SB \ 491INLINE void
492encode_indent (enc_t *enc)
493{
142 if (enc->flags & F_INDENT) \ 494 if (enc->json.flags & F_INDENT)
143 { \ 495 {
144 int i_; \ 496 int spaces = enc->indent * INDENT_STEP;
145 need (enc, enc->indent); \ 497
146 for (i_ = enc->indent * 3; i_--; )\ 498 need (enc, spaces);
499 memset (enc->cur, ' ', spaces);
500 enc->cur += spaces;
501 }
502}
503
504INLINE void
505encode_space (enc_t *enc)
506{
507 need (enc, 1);
508 encode_ch (enc, ' ');
509}
510
511INLINE void
512encode_nl (enc_t *enc)
513{
514 if (enc->json.flags & F_INDENT)
515 {
516 need (enc, 1);
147 encode_ch (enc, ' '); \ 517 encode_ch (enc, '\n');
148 } \ 518 }
149 SE 519}
150 520
151#define SPACE SB if (enc->flags & F_INDENT) { need (enc, 1); encode_ch (enc, ' '); } SE 521INLINE void
152#define NL SB if (enc->flags & F_INDENT) { need (enc, 1); encode_ch (enc, '\n'); } SE 522encode_comma (enc_t *enc)
153#define COMMA SB \ 523{
154 encode_ch (enc, ','); \ 524 encode_ch (enc, ',');
525
155 if (enc->flags & F_INDENT) \ 526 if (enc->json.flags & F_INDENT)
156 NL; \ 527 encode_nl (enc);
157 else if (enc->flags & F_SPACE_AFTER) \ 528 else if (enc->json.flags & F_SPACE_AFTER)
158 SPACE; \ 529 encode_space (enc);
159 SE 530}
160 531
161static void encode_sv (enc_t *enc, SV *sv); 532static void encode_sv (enc_t *enc, SV *sv);
162 533
163static void 534static void
164encode_av (enc_t *enc, AV *av) 535encode_av (enc_t *enc, AV *av)
165{ 536{
166 int i, len = av_len (av); 537 int i, len = av_len (av);
167 538
539 if (enc->indent >= enc->json.max_depth)
540 croak (ERR_NESTING_EXCEEDED);
541
168 encode_ch (enc, '['); NL; 542 encode_ch (enc, '[');
169 ++enc->indent;
170 543
544 if (len >= 0)
545 {
546 encode_nl (enc); ++enc->indent;
547
171 for (i = 0; i <= len; ++i) 548 for (i = 0; i <= len; ++i)
172 { 549 {
173 INDENT; 550 SV **svp = av_fetch (av, i, 0);
174 encode_sv (enc, *av_fetch (av, i, 0));
175 551
552 encode_indent (enc);
553
554 if (svp)
555 encode_sv (enc, *svp);
556 else
557 encode_str (enc, "null", 4, 0);
558
176 if (i < len) 559 if (i < len)
177 COMMA; 560 encode_comma (enc);
178 } 561 }
179 562
180 NL; 563 encode_nl (enc); --enc->indent; encode_indent (enc);
564 }
181 565
182 --enc->indent;
183 INDENT; encode_ch (enc, ']'); 566 encode_ch (enc, ']');
184} 567}
185 568
186static void 569static void
187encode_he (enc_t *enc, HE *he) 570encode_hk (enc_t *enc, HE *he)
188{ 571{
189 encode_ch (enc, '"'); 572 encode_ch (enc, '"');
190 573
191 if (HeKLEN (he) == HEf_SVKEY) 574 if (HeKLEN (he) == HEf_SVKEY)
192 { 575 {
193 SV *sv = HeSVKEY (he); 576 SV *sv = HeSVKEY (he);
194 STRLEN len; 577 STRLEN len;
578 char *str;
579
580 SvGETMAGIC (sv);
195 char *str = SvPV (sv, len); 581 str = SvPV (sv, len);
196 582
197 encode_str (enc, str, len, SvUTF8 (sv)); 583 encode_str (enc, str, len, SvUTF8 (sv));
198 } 584 }
199 else 585 else
200 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 586 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
201 587
202 encode_ch (enc, '"'); 588 encode_ch (enc, '"');
203 589
204 if (enc->flags & F_SPACE_BEFORE) SPACE; 590 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
205 encode_ch (enc, ':'); 591 encode_ch (enc, ':');
206 if (enc->flags & F_SPACE_AFTER ) SPACE; 592 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
207 encode_sv (enc, HeVAL (he));
208} 593}
209 594
210// compare hash entries, used when all keys are bytestrings 595// compare hash entries, used when all keys are bytestrings
211static int 596static int
212he_cmp_fast (const void *a_, const void *b_) 597he_cmp_fast (const void *a_, const void *b_)
… …
217 HE *b = *(HE **)b_; 602 HE *b = *(HE **)b_;
218 603
219 STRLEN la = HeKLEN (a); 604 STRLEN la = HeKLEN (a);
220 STRLEN lb = HeKLEN (b); 605 STRLEN lb = HeKLEN (b);
221 606
222 if (!(cmp == memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 607 if (!(cmp = memcmp (HeKEY (b), HeKEY (a), lb < la ? lb : la)))
223 cmp = la < lb ? -1 : la == lb ? 0 : 1; 608 cmp = lb - la;
224 609
225 return cmp; 610 return cmp;
226} 611}
227 612
228// compare hash entries, used when some keys are sv's or utf-x 613// compare hash entries, used when some keys are sv's or utf-x
229static int 614static int
230he_cmp_slow (const void *a, const void *b) 615he_cmp_slow (const void *a, const void *b)
231{ 616{
232 return sv_cmp (HeSVKEY_force (*(HE **)a), HeSVKEY_force (*(HE **)b)); 617 return sv_cmp (HeSVKEY_force (*(HE **)b), HeSVKEY_force (*(HE **)a));
233} 618}
234 619
235static void 620static void
236encode_hv (enc_t *enc, HV *hv) 621encode_hv (enc_t *enc, HV *hv)
237{ 622{
238 int count, i; 623 HE *he;
239 624
240 encode_ch (enc, '{'); NL; ++enc->indent; 625 if (enc->indent >= enc->json.max_depth)
626 croak (ERR_NESTING_EXCEEDED);
241 627
242 if ((count = hv_iterinit (hv))) 628 encode_ch (enc, '{');
243 { 629
244 // for canonical output we have to sort by keys first 630 // for canonical output we have to sort by keys first
245 // actually, this is mostly due to the stupid so-called 631 // actually, this is mostly due to the stupid so-called
246 // security workaround added somewhere in 5.8.x. 632 // security workaround added somewhere in 5.8.x
247 // that randomises hash orderings 633 // that randomises hash orderings
248 if (enc->flags & F_CANONICAL) 634 if (enc->json.flags & F_CANONICAL && !SvRMAGICAL (hv))
635 {
636 int count = hv_iterinit (hv);
637
638 if (SvMAGICAL (hv))
249 { 639 {
250 HE *he, *hes [count]; 640 // need to count by iterating. could improve by dynamically building the vector below
641 // but I don't care for the speed of this special case.
642 // note also that we will run into undefined behaviour when the two iterations
643 // do not result in the same count, something I might care for in some later release.
644
645 count = 0;
646 while (hv_iternext (hv))
647 ++count;
648
649 hv_iterinit (hv);
650 }
651
652 if (count)
653 {
251 int fast = 1; 654 int i, fast = 1;
655 HE *hes_stack [STACK_HES];
656 HE **hes = hes_stack;
657
658 // allocate larger arrays on the heap
659 if (count > STACK_HES)
660 {
661 SV *sv = sv_2mortal (NEWSV (0, count * sizeof (*hes)));
662 hes = (HE **)SvPVX (sv);
663 }
252 664
253 i = 0; 665 i = 0;
254 while ((he = hv_iternext (hv))) 666 while ((he = hv_iternext (hv)))
255 { 667 {
256 hes [i++] = he; 668 hes [i++] = he;
… …
262 674
263 if (fast) 675 if (fast)
264 qsort (hes, count, sizeof (HE *), he_cmp_fast); 676 qsort (hes, count, sizeof (HE *), he_cmp_fast);
265 else 677 else
266 { 678 {
267 // hack to disable "use bytes" 679 // hack to forcefully disable "use bytes"
268 COP *oldcop = PL_curcop, cop; 680 COP cop = *PL_curcop;
269 cop.op_private = 0; 681 cop.op_private = 0;
682
683 ENTER;
684 SAVETMPS;
685
686 SAVEVPTR (PL_curcop);
270 PL_curcop = &cop; 687 PL_curcop = &cop;
271 688
272 SAVETMPS;
273 qsort (hes, count, sizeof (HE *), he_cmp_slow); 689 qsort (hes, count, sizeof (HE *), he_cmp_slow);
690
274 FREETMPS; 691 FREETMPS;
275 692 LEAVE;
276 PL_curcop = oldcop;
277 } 693 }
278 694
279 for (i = 0; i < count; ++i) 695 encode_nl (enc); ++enc->indent;
696
697 while (count--)
280 { 698 {
281 INDENT; 699 encode_indent (enc);
700 he = hes [count];
282 encode_he (enc, hes [i]); 701 encode_hk (enc, he);
702 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
283 703
284 if (i < count - 1) 704 if (count)
285 COMMA; 705 encode_comma (enc);
286 } 706 }
287 707
288 NL; 708 encode_nl (enc); --enc->indent; encode_indent (enc);
289 } 709 }
710 }
711 else
712 {
713 if (hv_iterinit (hv) || SvMAGICAL (hv))
714 if ((he = hv_iternext (hv)))
715 {
716 encode_nl (enc); ++enc->indent;
717
718 for (;;)
719 {
720 encode_indent (enc);
721 encode_hk (enc, he);
722 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
723
724 if (!(he = hv_iternext (hv)))
725 break;
726
727 encode_comma (enc);
728 }
729
730 encode_nl (enc); --enc->indent; encode_indent (enc);
731 }
732 }
733
734 encode_ch (enc, '}');
735}
736
737// encode objects, arrays and special \0=false and \1=true values.
738static void
739encode_rv (enc_t *enc, SV *sv)
740{
741 svtype svt;
742 GV *method;
743
744 SvGETMAGIC (sv);
745 svt = SvTYPE (sv);
746
747 if (expect_false (SvOBJECT (sv)))
748 {
749 HV *stash = SvSTASH (sv);
750
751 if (stash == bool_stash)
752 {
753 if (SvIV (sv))
754 encode_str (enc, "true", 4, 0);
755 else
756 encode_str (enc, "false", 5, 0);
757 }
758 else if ((enc->json.flags & F_ALLOW_TAGS) && (method = gv_fetchmethod_autoload (stash, "FREEZE", 0)))
759 {
760 int count;
761 dSP;
762
763 ENTER; SAVETMPS;
764 SAVESTACK_POS ();
765 PUSHMARK (SP);
766 EXTEND (SP, 2);
767 // we re-bless the reference to get overload and other niceties right
768 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
769 PUSHs (sv_json);
770
771 PUTBACK;
772 count = call_sv ((SV *)GvCV (method), G_ARRAY);
773 SPAGAIN;
774
775 // catch this surprisingly common error
776 if (SvROK (TOPs) && SvRV (TOPs) == sv)
777 croak ("%s::FREEZE method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
778
779 encode_ch (enc, '(');
780 encode_ch (enc, '"');
781 encode_str (enc, HvNAME (stash), HvNAMELEN (stash), HvNAMEUTF8 (stash));
782 encode_ch (enc, '"');
783 encode_ch (enc, ')');
784 encode_ch (enc, '[');
785
786 while (count)
787 {
788 encode_sv (enc, SP[1 - count--]);
789
790 if (count)
791 encode_ch (enc, ',');
792 }
793
794 encode_ch (enc, ']');
795
796 FREETMPS; LEAVE;
797 }
798 else if ((enc->json.flags & F_CONV_BLESSED) && (method = gv_fetchmethod_autoload (stash, "TO_JSON", 0)))
799 {
800 dSP;
801
802 ENTER; SAVETMPS;
803 PUSHMARK (SP);
804 // we re-bless the reference to get overload and other niceties right
805 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
806
807 // calling with G_SCALAR ensures that we always get a 1 return value
808 PUTBACK;
809 call_sv ((SV *)GvCV (method), G_SCALAR);
810 SPAGAIN;
811
812 // catch this surprisingly common error
813 if (SvROK (TOPs) && SvRV (TOPs) == sv)
814 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
815
816 sv = POPs;
817 PUTBACK;
818
819 encode_sv (enc, sv);
820
821 FREETMPS; LEAVE;
822 }
823 else if (enc->json.flags & F_ALLOW_BLESSED)
824 encode_str (enc, "null", 4, 0);
290 else 825 else
291 { 826 croak ("encountered object '%s', but neither allow_blessed, convert_blessed nor allow_tags settings are enabled (or TO_JSON/FREEZE method missing)",
292 SV *sv; 827 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
293 HE *he = hv_iternext (hv); 828 }
294 829 else if (svt == SVt_PVHV)
295 for (;;) 830 encode_hv (enc, (HV *)sv);
296 { 831 else if (svt == SVt_PVAV)
297 INDENT; 832 encode_av (enc, (AV *)sv);
298 encode_he (enc, he); 833 else if (svt < SVt_PVAV)
299
300 if (!(he = hv_iternext (hv)))
301 break;
302
303 COMMA;
304 }
305
306 NL;
307 }
308 } 834 {
835 int bool_type = ref_bool_type (sv);
309 836
310 --enc->indent; INDENT; encode_ch (enc, '}'); 837 if (bool_type == 1)
838 encode_str (enc, "true", 4, 0);
839 else if (bool_type == 0)
840 encode_str (enc, "false", 5, 0);
841 else if (enc->json.flags & F_ALLOW_UNKNOWN)
842 encode_str (enc, "null", 4, 0);
843 else
844 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
845 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
846 }
847 else if (enc->json.flags & F_ALLOW_UNKNOWN)
848 encode_str (enc, "null", 4, 0);
849 else
850 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
851 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
311} 852}
312 853
313static void 854static void
314encode_sv (enc_t *enc, SV *sv) 855encode_sv (enc_t *enc, SV *sv)
315{ 856{
857 SvGETMAGIC (sv);
858
316 if (SvPOKp (sv)) 859 if (SvPOKp (sv))
317 { 860 {
318 STRLEN len; 861 STRLEN len;
319 char *str = SvPV (sv, len); 862 char *str = SvPV (sv, len);
320 encode_ch (enc, '"'); 863 encode_ch (enc, '"');
321 encode_str (enc, str, len, SvUTF8 (sv)); 864 encode_str (enc, str, len, SvUTF8 (sv));
322 encode_ch (enc, '"'); 865 encode_ch (enc, '"');
323 } 866 }
324 else if (SvNOKp (sv)) 867 else if (SvNOKp (sv))
325 { 868 {
869 // trust that perl will do the right thing w.r.t. JSON syntax.
326 need (enc, NV_DIG + 32); 870 need (enc, NV_DIG + 32);
327 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 871 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
328 enc->cur += strlen (enc->cur); 872 enc->cur += strlen (enc->cur);
329 } 873 }
330 else if (SvIOKp (sv)) 874 else if (SvIOKp (sv))
331 { 875 {
876 // we assume we can always read an IV as a UV and vice versa
877 // we assume two's complement
878 // we assume no aliasing issues in the union
879 if (SvIsUV (sv) ? SvUVX (sv) <= 59000
880 : SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000)
881 {
882 // optimise the "small number case"
883 // code will likely be branchless and use only a single multiplication
884 // works for numbers up to 59074
885 I32 i = SvIVX (sv);
886 U32 u;
887 char digit, nz = 0;
888
332 need (enc, 64); 889 need (enc, 6);
890
891 *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
892 u = i < 0 ? -i : i;
893
894 // convert to 4.28 fixed-point representation
895 u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
896
897 // now output digit by digit, each time masking out the integer part
898 // and multiplying by 5 while moving the decimal point one to the right,
899 // resulting in a net multiplication by 10.
900 // we always write the digit to memory but conditionally increment
901 // the pointer, to enable the use of conditional move instructions.
902 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffffUL) * 5;
903 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffffUL) * 5;
904 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffffUL) * 5;
905 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffffUL) * 5;
906 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
907 }
908 else
909 {
910 // large integer, use the (rather slow) snprintf way.
911 need (enc, IVUV_MAXCHARS);
333 enc->cur += 912 enc->cur +=
334 SvIsUV(sv) 913 SvIsUV(sv)
335 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 914 ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))
336 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 915 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
916 }
337 } 917 }
338 else if (SvROK (sv)) 918 else if (SvROK (sv))
339 { 919 encode_rv (enc, SvRV (sv));
340 if (!--enc->max_recurse) 920 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN)
341 croak ("data structure too deep (hit recursion limit)");
342
343 sv = SvRV (sv);
344
345 switch (SvTYPE (sv))
346 {
347 case SVt_PVAV: encode_av (enc, (AV *)sv); break;
348 case SVt_PVHV: encode_hv (enc, (HV *)sv); break;
349
350 default:
351 croak ("JSON can only represent references to arrays or hashes");
352 }
353 }
354 else if (!SvOK (sv))
355 encode_str (enc, "null", 4, 0); 921 encode_str (enc, "null", 4, 0);
356 else 922 else
357 croak ("encountered perl type that JSON cannot handle"); 923 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, check your input data",
924 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
358} 925}
359 926
360static SV * 927static SV *
361encode_json (SV *scalar, UV flags) 928encode_json (SV *scalar, JSON *json)
362{ 929{
363 enc_t enc; 930 enc_t enc;
364 enc.flags = flags; 931
932 if (!(json->flags & F_ALLOW_NONREF) && json_nonref (scalar))
933 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
934
935 enc.json = *json;
365 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 936 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
366 enc.cur = SvPVX (enc.sv); 937 enc.cur = SvPVX (enc.sv);
367 enc.end = SvEND (enc.sv); 938 enc.end = SvEND (enc.sv);
368 enc.max_recurse = 0;
369 enc.indent = 0; 939 enc.indent = 0;
940 enc.limit = enc.json.flags & F_ASCII ? 0x000080UL
941 : enc.json.flags & F_LATIN1 ? 0x000100UL
942 : 0x110000UL;
370 943
371 SvPOK_only (enc.sv); 944 SvPOK_only (enc.sv);
372 encode_sv (&enc, scalar); 945 encode_sv (&enc, scalar);
946 encode_nl (&enc);
373 947
948 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
949 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
950
374 if (!(flags & (F_ASCII | F_UTF8))) 951 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
375 SvUTF8_on (enc.sv); 952 SvUTF8_on (enc.sv);
376 953
377 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 954 if (enc.json.flags & F_SHRINK)
955 shrink (enc.sv);
956
378 return enc.sv; 957 return enc.sv;
379} 958}
380 959
381///////////////////////////////////////////////////////////////////////////// 960/////////////////////////////////////////////////////////////////////////////
961// decoder
382 962
383#define WS \ 963// structure used for decoding JSON
964typedef struct
965{
966 char *cur; // current parser pointer
967 char *end; // end of input string
968 const char *err; // parse error, if != 0
969 JSON json;
970 U32 depth; // recursion depth
971 U32 maxdepth; // recursion depth limit
972} dec_t;
973
974INLINE void
975decode_comment (dec_t *dec)
976{
977 // only '#'-style comments allowed a.t.m.
978
979 while (*dec->cur && *dec->cur != 0x0a && *dec->cur != 0x0d)
980 ++dec->cur;
981}
982
983INLINE void
984decode_ws (dec_t *dec)
985{
384 for (;;) \ 986 for (;;)
385 { \ 987 {
386 char ch = *dec->cur; \ 988 char ch = *dec->cur;
989
387 if (ch > 0x20 \ 990 if (ch > 0x20)
991 {
992 if (expect_false (ch == '#'))
993 {
994 if (dec->json.flags & F_RELAXED)
995 decode_comment (dec);
996 else
997 break;
998 }
999 else
1000 break;
1001 }
388 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) \ 1002 else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)
389 break; \ 1003 break; // parse error, but let higher level handle it, gives better error messages
1004
390 ++dec->cur; \ 1005 ++dec->cur;
391 } 1006 }
1007}
392 1008
393#define ERR(reason) SB dec->err = reason; goto fail; SE 1009#define ERR(reason) SB dec->err = reason; goto fail; SE
1010
394#define EXPECT_CH(ch) SB \ 1011#define EXPECT_CH(ch) SB \
395 if (*dec->cur != ch) \ 1012 if (*dec->cur != ch) \
396 ERR (# ch " expected"); \ 1013 ERR (# ch " expected"); \
397 ++dec->cur; \ 1014 ++dec->cur; \
398 SE 1015 SE
399 1016
1017#define DEC_INC_DEPTH if (++dec->depth > dec->json.max_depth) ERR (ERR_NESTING_EXCEEDED)
1018#define DEC_DEC_DEPTH --dec->depth
1019
400static SV *decode_sv (dec_t *dec); 1020static SV *decode_sv (dec_t *dec);
401
402#define APPEND_CH(ch) SB \
403 SvGROW (sv, cur + 1 + 1); \
404 SvPVX (sv)[cur++] = (ch); \
405 SE
406 1021
407static signed char decode_hexdigit[256]; 1022static signed char decode_hexdigit[256];
408 1023
409static UV 1024static UV
410decode_4hex (dec_t *dec) 1025decode_4hex (dec_t *dec)
411{ 1026{
412 signed char d1, d2, d3, d4; 1027 signed char d1, d2, d3, d4;
1028 unsigned char *cur = (unsigned char *)dec->cur;
413 1029
414 d1 = decode_hexdigit [((unsigned char *)dec->cur) [0]]; 1030 d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected");
415 if (d1 < 0) ERR ("four hexadecimal digits expected"); 1031 d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected");
416 d2 = decode_hexdigit [((unsigned char *)dec->cur) [1]]; 1032 d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected");
417 if (d2 < 0) ERR ("four hexadecimal digits expected"); 1033 d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected");
418 d3 = decode_hexdigit [((unsigned char *)dec->cur) [2]];
419 if (d3 < 0) ERR ("four hexadecimal digits expected");
420 d4 = decode_hexdigit [((unsigned char *)dec->cur) [3]];
421 if (d4 < 0) ERR ("four hexadecimal digits expected");
422 1034
423 dec->cur += 4; 1035 dec->cur += 4;
424 1036
425 return ((UV)d1) << 12 1037 return ((UV)d1) << 12
426 | ((UV)d2) << 8 1038 | ((UV)d2) << 8
… …
432} 1044}
433 1045
434static SV * 1046static SV *
435decode_str (dec_t *dec) 1047decode_str (dec_t *dec)
436{ 1048{
437 SV *sv = NEWSV (0,2); 1049 SV *sv = 0;
438 STRLEN cur = 0;
439 int utf8 = 0; 1050 int utf8 = 0;
1051 char *dec_cur = dec->cur;
440 1052
441 for (;;) 1053 do
442 { 1054 {
443 unsigned char ch = *(unsigned char *)dec->cur; 1055 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
1056 char *cur = buf;
444 1057
445 if (ch == '"') 1058 do
446 break;
447 else if (ch == '\\')
448 { 1059 {
449 switch (*++dec->cur) 1060 unsigned char ch = *(unsigned char *)dec_cur++;
1061
1062 if (expect_false (ch == '"'))
450 { 1063 {
451 case '\\': 1064 --dec_cur;
452 case '/': 1065 break;
453 case '"': APPEND_CH (*dec->cur++); break; 1066 }
454 1067 else if (expect_false (ch == '\\'))
455 case 'b': APPEND_CH ('\010'); ++dec->cur; break; 1068 {
456 case 't': APPEND_CH ('\011'); ++dec->cur; break; 1069 switch (*dec_cur)
457 case 'n': APPEND_CH ('\012'); ++dec->cur; break;
458 case 'f': APPEND_CH ('\014'); ++dec->cur; break;
459 case 'r': APPEND_CH ('\015'); ++dec->cur; break;
460
461 case 'u':
462 { 1070 {
463 UV lo, hi; 1071 case '\\':
464 ++dec->cur; 1072 case '/':
1073 case '"': *cur++ = *dec_cur++; break;
465 1074
466 hi = decode_4hex (dec); 1075 case 'b': ++dec_cur; *cur++ = '\010'; break;
467 if (hi == (UV)-1) 1076 case 't': ++dec_cur; *cur++ = '\011'; break;
468 goto fail; 1077 case 'n': ++dec_cur; *cur++ = '\012'; break;
1078 case 'f': ++dec_cur; *cur++ = '\014'; break;
1079 case 'r': ++dec_cur; *cur++ = '\015'; break;
469 1080
470 // possibly a surrogate pair 1081 case 'u':
471 if (hi >= 0xd800 && hi < 0xdc00)
472 { 1082 {
473 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 1083 UV lo, hi;
474 ERR ("illegal surrogate character"); 1084 ++dec_cur;
475 1085
476 dec->cur += 2; 1086 dec->cur = dec_cur;
477
478 lo = decode_4hex (dec); 1087 hi = decode_4hex (dec);
1088 dec_cur = dec->cur;
479 if (lo == (UV)-1) 1089 if (hi == (UV)-1)
480 goto fail; 1090 goto fail;
481 1091
1092 // possibly a surrogate pair
1093 if (hi >= 0xd800)
1094 if (hi < 0xdc00)
1095 {
1096 if (dec_cur [0] != '\\' || dec_cur [1] != 'u')
1097 ERR ("missing low surrogate character in surrogate pair");
1098
1099 dec_cur += 2;
1100
1101 dec->cur = dec_cur;
1102 lo = decode_4hex (dec);
1103 dec_cur = dec->cur;
1104 if (lo == (UV)-1)
1105 goto fail;
1106
482 if (lo < 0xdc00 || lo >= 0xe000) 1107 if (lo < 0xdc00 || lo >= 0xe000)
483 ERR ("surrogate pair expected"); 1108 ERR ("surrogate pair expected");
484 1109
485 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000; 1110 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000;
1111 }
1112 else if (hi < 0xe000)
1113 ERR ("missing high surrogate character in surrogate pair");
1114
1115 if (hi >= 0x80)
1116 {
1117 utf8 = 1;
1118
1119 cur = encode_utf8 (cur, hi);
1120 }
1121 else
1122 *cur++ = hi;
486 } 1123 }
487 else if (lo >= 0xdc00 && lo < 0xe000)
488 ERR ("illegal surrogate character");
489
490 if (hi >= 0x80)
491 { 1124 break;
492 utf8 = 1;
493 1125
494 SvGROW (sv, cur + 4 + 1); // at most 4 bytes for 21 bits
495 cur = (char *)uvuni_to_utf8_flags (SvPVX (sv) + cur, hi, 0) - SvPVX (sv);
496 }
497 else 1126 default:
498 APPEND_CH (hi); 1127 --dec_cur;
1128 ERR ("illegal backslash escape sequence in string");
499 } 1129 }
1130 }
1131 else if (expect_true (ch >= 0x20 && ch < 0x80))
1132 *cur++ = ch;
1133 else if (ch >= 0x80)
1134 {
1135 STRLEN clen;
1136
1137 --dec_cur;
1138
1139 decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
1140 if (clen == (STRLEN)-1)
1141 ERR ("malformed UTF-8 character in JSON string");
1142
1143 do
1144 *cur++ = *dec_cur++;
1145 while (--clen);
1146
1147 utf8 = 1;
1148 }
1149 else if (ch == '\t' && dec->json.flags & F_RELAXED)
1150 *cur++ = ch;
1151 else
1152 {
1153 --dec_cur;
1154
1155 if (!ch)
1156 ERR ("unexpected end of string while parsing JSON string");
500 break; 1157 else
1158 ERR ("invalid character encountered while parsing JSON string");
501 } 1159 }
502 } 1160 }
503 else if (ch >= 0x20 && ch <= 0x7f) 1161 while (cur < buf + SHORT_STRING_LEN);
504 APPEND_CH (*dec->cur++); 1162
505 else if (ch >= 0x80) 1163 {
1164 STRLEN len = cur - buf;
1165
1166 if (sv)
506 { 1167 {
507 STRLEN clen; 1168 STRLEN cur = SvCUR (sv);
508 UV uch = utf8n_to_uvuni (dec->cur, dec->end - dec->cur, &clen, UTF8_CHECK_ONLY);
509 if (clen < 0)
510 ERR ("malformed UTF-8 character in string, cannot convert to JSON");
511 1169
512 SvGROW (sv, cur + clen + 1); // at most 4 bytes for 21 bits 1170 if (SvLEN (sv) <= cur + len)
513 memcpy (SvPVX (sv) + cur, dec->cur, clen); 1171 SvGROW (sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
514 dec->cur += clen; 1172
1173 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
1174 SvCUR_set (sv, SvCUR (sv) + len);
515 } 1175 }
516 else 1176 else
517 ERR ("invalid character encountered"); 1177 sv = newSVpvn (buf, len);
518 } 1178 }
1179 }
1180 while (*dec_cur != '"');
519 1181
520 ++dec->cur; 1182 ++dec_cur;
521 1183
1184 if (sv)
1185 {
522 SvPOK_only (sv); 1186 SvPOK_only (sv);
523
524 SvCUR_set (sv, cur);
525 *SvEND (sv) = 0; 1187 *SvEND (sv) = 0;
526 1188
527 if (utf8) 1189 if (utf8)
528 SvUTF8_on (sv); 1190 SvUTF8_on (sv);
1191 }
1192 else
1193 sv = newSVpvn ("", 0);
529 1194
1195 dec->cur = dec_cur;
530 return sv; 1196 return sv;
531 1197
532fail: 1198fail:
533 SvREFCNT_dec (sv); 1199 dec->cur = dec_cur;
534 return 0; 1200 return 0;
535} 1201}
536 1202
537static SV * 1203static SV *
538decode_num (dec_t *dec) 1204decode_num (dec_t *dec)
… …
548 { 1214 {
549 ++dec->cur; 1215 ++dec->cur;
550 if (*dec->cur >= '0' && *dec->cur <= '9') 1216 if (*dec->cur >= '0' && *dec->cur <= '9')
551 ERR ("malformed number (leading zero must not be followed by another digit)"); 1217 ERR ("malformed number (leading zero must not be followed by another digit)");
552 } 1218 }
553 1219 else if (*dec->cur < '0' || *dec->cur > '9')
554 // int 1220 ERR ("malformed number (no digits after initial minus)");
1221 else
1222 do
1223 {
1224 ++dec->cur;
1225 }
555 while (*dec->cur >= '0' && *dec->cur <= '9') 1226 while (*dec->cur >= '0' && *dec->cur <= '9');
556 ++dec->cur;
557 1227
558 // [frac] 1228 // [frac]
559 if (*dec->cur == '.') 1229 if (*dec->cur == '.')
560 { 1230 {
561 is_nv = 1; 1231 ++dec->cur;
1232
1233 if (*dec->cur < '0' || *dec->cur > '9')
1234 ERR ("malformed number (no digits after decimal point)");
562 1235
563 do 1236 do
564 { 1237 {
565 ++dec->cur; 1238 ++dec->cur;
566 } 1239 }
567 while (*dec->cur >= '0' && *dec->cur <= '9'); 1240 while (*dec->cur >= '0' && *dec->cur <= '9');
1241
1242 is_nv = 1;
568 } 1243 }
569 1244
570 // [exp] 1245 // [exp]
571 if (*dec->cur == 'e' || *dec->cur == 'E') 1246 if (*dec->cur == 'e' || *dec->cur == 'E')
572 { 1247 {
573 is_nv = 1;
574
575 ++dec->cur; 1248 ++dec->cur;
1249
576 if (*dec->cur == '-' || *dec->cur == '+') 1250 if (*dec->cur == '-' || *dec->cur == '+')
577 ++dec->cur; 1251 ++dec->cur;
578 1252
1253 if (*dec->cur < '0' || *dec->cur > '9')
1254 ERR ("malformed number (no digits after exp sign)");
1255
1256 do
1257 {
1258 ++dec->cur;
1259 }
579 while (*dec->cur >= '0' && *dec->cur <= '9') 1260 while (*dec->cur >= '0' && *dec->cur <= '9');
580 ++dec->cur; 1261
1262 is_nv = 1;
581 } 1263 }
582 1264
583 if (!is_nv) 1265 if (!is_nv)
584 { 1266 {
585 UV uv; 1267 int len = dec->cur - start;
586 int numtype = grok_number (start, dec->cur - start, &uv); 1268
587 if (numtype & IS_NUMBER_IN_UV) 1269 // special case the rather common 1..5-digit-int case
588 if (numtype & IS_NUMBER_NEG) 1270 if (*start == '-')
1271 switch (len)
589 { 1272 {
590 if (uv < (UV)IV_MIN) 1273 case 2: return newSViv (-(IV)( start [1] - '0' * 1));
591 return newSViv (-(IV)uv); 1274 case 3: return newSViv (-(IV)( start [1] * 10 + start [2] - '0' * 11));
1275 case 4: return newSViv (-(IV)( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
1276 case 5: return newSViv (-(IV)( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
1277 case 6: return newSViv (-(IV)(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
592 } 1278 }
1279 else
1280 switch (len)
1281 {
1282 case 1: return newSViv ( start [0] - '0' * 1);
1283 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
1284 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
1285 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
1286 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
1287 }
1288
1289 {
1290 UV uv;
1291 int numtype = grok_number (start, len, &uv);
1292 if (numtype & IS_NUMBER_IN_UV)
1293 if (numtype & IS_NUMBER_NEG)
1294 {
1295 if (uv < (UV)IV_MIN)
1296 return newSViv (-(IV)uv);
1297 }
593 else 1298 else
594 return newSVuv (uv); 1299 return newSVuv (uv);
595 } 1300 }
596 1301
1302 len -= *start == '-' ? 1 : 0;
1303
1304 // does not fit into IV or UV, try NV
1305 if (len <= NV_DIG)
1306 // fits into NV without loss of precision
1307 return newSVnv (json_atof (start));
1308
1309 // everything else fails, convert it to a string
1310 return newSVpvn (start, dec->cur - start);
1311 }
1312
1313 // loss of precision here
597 return newSVnv (Atof (start)); 1314 return newSVnv (json_atof (start));
598 1315
599fail: 1316fail:
600 return 0; 1317 return 0;
601} 1318}
602 1319
603static SV * 1320static SV *
604decode_av (dec_t *dec) 1321decode_av (dec_t *dec)
605{ 1322{
606 AV *av = newAV (); 1323 AV *av = newAV ();
607 1324
1325 DEC_INC_DEPTH;
1326 decode_ws (dec);
1327
1328 if (*dec->cur == ']')
1329 ++dec->cur;
1330 else
608 for (;;) 1331 for (;;)
609 { 1332 {
610 SV *value; 1333 SV *value;
611 1334
612 value = decode_sv (dec); 1335 value = decode_sv (dec);
613 if (!value) 1336 if (!value)
614 goto fail; 1337 goto fail;
615 1338
616 av_push (av, value); 1339 av_push (av, value);
617 1340
618 WS; 1341 decode_ws (dec);
619 1342
620 if (*dec->cur == ']') 1343 if (*dec->cur == ']')
621 { 1344 {
622 ++dec->cur; 1345 ++dec->cur;
623 break; 1346 break;
624 } 1347 }
625 1348
626 if (*dec->cur != ',') 1349 if (*dec->cur != ',')
627 ERR (", or ] expected while parsing array"); 1350 ERR (", or ] expected while parsing array");
628 1351
629 ++dec->cur; 1352 ++dec->cur;
1353
1354 decode_ws (dec);
1355
1356 if (*dec->cur == ']' && dec->json.flags & F_RELAXED)
1357 {
1358 ++dec->cur;
1359 break;
1360 }
630 } 1361 }
631 1362
1363 DEC_DEC_DEPTH;
632 return newRV_noinc ((SV *)av); 1364 return newRV_noinc ((SV *)av);
633 1365
634fail: 1366fail:
635 SvREFCNT_dec (av); 1367 SvREFCNT_dec (av);
1368 DEC_DEC_DEPTH;
636 return 0; 1369 return 0;
637} 1370}
638 1371
639static SV * 1372static SV *
640decode_hv (dec_t *dec) 1373decode_hv (dec_t *dec)
641{ 1374{
1375 SV *sv;
642 HV *hv = newHV (); 1376 HV *hv = newHV ();
643 1377
1378 DEC_INC_DEPTH;
1379 decode_ws (dec);
1380
1381 if (*dec->cur == '}')
1382 ++dec->cur;
1383 else
644 for (;;) 1384 for (;;)
645 { 1385 {
646 SV *key, *value;
647
648 WS; EXPECT_CH ('"'); 1386 EXPECT_CH ('"');
649 1387
650 key = decode_str (dec); 1388 // heuristic: assume that
651 if (!key) 1389 // a) decode_str + hv_store_ent are abysmally slow.
652 goto fail; 1390 // b) most hash keys are short, simple ascii text.
653 1391 // => try to "fast-match" such strings to avoid
654 WS; EXPECT_CH (':'); 1392 // the overhead of decode_str + hv_store_ent.
655
656 value = decode_sv (dec);
657 if (!value)
658 { 1393 {
1394 SV *value;
1395 char *p = dec->cur;
1396 char *e = p + 24; // only try up to 24 bytes
1397
1398 for (;;)
1399 {
1400 // the >= 0x80 is false on most architectures
1401 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1402 {
1403 // slow path, back up and use decode_str
1404 SV *key = decode_str (dec);
1405 if (!key)
1406 goto fail;
1407
1408 decode_ws (dec); EXPECT_CH (':');
1409
1410 decode_ws (dec);
1411 value = decode_sv (dec);
1412 if (!value)
1413 {
1414 SvREFCNT_dec (key);
1415 goto fail;
1416 }
1417
1418 hv_store_ent (hv, key, value, 0);
659 SvREFCNT_dec (key); 1419 SvREFCNT_dec (key);
1420
1421 break;
1422 }
1423 else if (*p == '"')
1424 {
1425 // fast path, got a simple key
1426 char *key = dec->cur;
1427 int len = p - key;
1428 dec->cur = p + 1;
1429
1430 decode_ws (dec); EXPECT_CH (':');
1431
1432 decode_ws (dec);
1433 value = decode_sv (dec);
1434 if (!value)
660 goto fail; 1435 goto fail;
1436
1437 hv_store (hv, key, len, value, 0);
1438
1439 break;
1440 }
1441
1442 ++p;
1443 }
661 } 1444 }
662 1445
663 //TODO: optimise 1446 decode_ws (dec);
664 hv_store_ent (hv, key, value, 0);
665 1447
666 WS;
667
668 if (*dec->cur == '}') 1448 if (*dec->cur == '}')
1449 {
1450 ++dec->cur;
1451 break;
1452 }
1453
1454 if (*dec->cur != ',')
1455 ERR (", or } expected while parsing object/hash");
1456
1457 ++dec->cur;
1458
1459 decode_ws (dec);
1460
1461 if (*dec->cur == '}' && dec->json.flags & F_RELAXED)
1462 {
1463 ++dec->cur;
1464 break;
1465 }
1466 }
1467
1468 DEC_DEC_DEPTH;
1469 sv = newRV_noinc ((SV *)hv);
1470
1471 // check filter callbacks
1472 if (dec->json.flags & F_HOOK)
1473 {
1474 if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
669 { 1475 {
670 ++dec->cur; 1476 HE *cb, *he;
671 break; 1477
1478 hv_iterinit (hv);
1479 he = hv_iternext (hv);
1480 hv_iterinit (hv);
1481
1482 // the next line creates a mortal sv each time it's called.
1483 // might want to optimise this for common cases.
1484 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1485
1486 if (cb)
1487 {
1488 dSP;
1489 int count;
1490
1491 ENTER; SAVETMPS;
1492 SAVESTACK_POS ();
1493 PUSHMARK (SP);
1494 XPUSHs (HeVAL (he));
1495 sv_2mortal (sv);
1496
1497 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1498
1499 if (count == 1)
1500 {
1501 sv = newSVsv (POPs);
1502 FREETMPS; LEAVE;
1503 return sv;
1504 }
1505
1506 SvREFCNT_inc (sv);
1507 FREETMPS; LEAVE;
1508 }
672 } 1509 }
673 1510
674 if (*dec->cur != ',') 1511 if (dec->json.cb_object)
675 ERR (", or } expected while parsing object/hash"); 1512 {
1513 dSP;
1514 int count;
676 1515
677 ++dec->cur; 1516 ENTER; SAVETMPS;
678 } 1517 SAVESTACK_POS ();
1518 PUSHMARK (SP);
1519 XPUSHs (sv_2mortal (sv));
679 1520
680 return newRV_noinc ((SV *)hv); 1521 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1522
1523 if (count == 1)
1524 {
1525 sv = newSVsv (POPs);
1526 FREETMPS; LEAVE;
1527 return sv;
1528 }
1529
1530 SvREFCNT_inc (sv);
1531 FREETMPS; LEAVE;
1532 }
1533 }
1534
1535 return sv;
681 1536
682fail: 1537fail:
683 SvREFCNT_dec (hv); 1538 SvREFCNT_dec (hv);
1539 DEC_DEC_DEPTH;
1540 return 0;
1541}
1542
1543static SV *
1544decode_tag (dec_t *dec)
1545{
1546 SV *tag = 0;
1547 SV *val = 0;
1548
1549 if (!(dec->json.flags & F_ALLOW_TAGS))
1550 ERR ("malformed JSON string, neither array, object, number, string or atom");
1551
1552 ++dec->cur;
1553
1554 decode_ws (dec);
1555
1556 tag = decode_sv (dec);
1557 if (!tag)
1558 goto fail;
1559
1560 if (!SvPOK (tag))
1561 ERR ("malformed JSON string, (tag) must be a string");
1562
1563 decode_ws (dec);
1564
1565 if (*dec->cur != ')')
1566 ERR (") expected after tag");
1567
1568 ++dec->cur;
1569
1570 decode_ws (dec);
1571
1572 val = decode_sv (dec);
1573 if (!val)
1574 goto fail;
1575
1576 if (!SvROK (val) || SvTYPE (SvRV (val)) != SVt_PVAV)
1577 ERR ("malformed JSON string, tag value must be an array");
1578
1579 {
1580 AV *av = (AV *)SvRV (val);
1581 int i, len = av_len (av) + 1;
1582 HV *stash = gv_stashsv (tag, 0);
1583 SV *sv;
1584
1585 if (!stash)
1586 ERR ("cannot decode perl-object (package does not exist)");
1587
1588 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
1589
1590 if (!method)
1591 ERR ("cannot decode perl-object (package does not have a THAW method)");
1592
1593 dSP;
1594
1595 ENTER; SAVETMPS;
1596 PUSHMARK (SP);
1597 EXTEND (SP, len + 2);
1598 // we re-bless the reference to get overload and other niceties right
1599 PUSHs (tag);
1600 PUSHs (sv_json);
1601
1602 for (i = 0; i < len; ++i)
1603 PUSHs (*av_fetch (av, i, 1));
1604
1605 PUTBACK;
1606 call_sv ((SV *)GvCV (method), G_SCALAR);
1607 SPAGAIN;
1608
1609 SvREFCNT_dec (tag);
1610 SvREFCNT_dec (val);
1611 sv = SvREFCNT_inc (POPs);
1612
1613 PUTBACK;
1614
1615 FREETMPS; LEAVE;
1616
1617 return sv;
1618 }
1619
1620fail:
1621 SvREFCNT_dec (tag);
1622 SvREFCNT_dec (val);
684 return 0; 1623 return 0;
685} 1624}
686 1625
687static SV * 1626static SV *
688decode_sv (dec_t *dec) 1627decode_sv (dec_t *dec)
689{ 1628{
690 WS; 1629 // the beauty of JSON: you need exactly one character lookahead
1630 // to parse everything.
691 switch (*dec->cur) 1631 switch (*dec->cur)
692 { 1632 {
693 case '"': ++dec->cur; return decode_str (dec); 1633 case '"': ++dec->cur; return decode_str (dec);
694 case '[': ++dec->cur; return decode_av (dec); 1634 case '[': ++dec->cur; return decode_av (dec);
695 case '{': ++dec->cur; return decode_hv (dec); 1635 case '{': ++dec->cur; return decode_hv (dec);
1636 case '(': return decode_tag (dec);
696 1637
697 case '-': 1638 case '-':
698 case '0': case '1': case '2': case '3': case '4': 1639 case '0': case '1': case '2': case '3': case '4':
699 case '5': case '6': case '7': case '8': case '9': 1640 case '5': case '6': case '7': case '8': case '9':
700 return decode_num (dec); 1641 return decode_num (dec);
701 1642
702 case 't': 1643 case 't':
703 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1644 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
704 { 1645 {
705 dec->cur += 4; 1646 dec->cur += 4;
1647#if JSON_SLOW
1648 bool_true = get_bool ("Types::Serialiser::true");
1649#endif
706 return newSViv (1); 1650 return newSVsv (bool_true);
707 } 1651 }
708 else 1652 else
709 ERR ("'true' expected"); 1653 ERR ("'true' expected");
710 1654
711 break; 1655 break;
712 1656
713 case 'f': 1657 case 'f':
714 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1658 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
715 { 1659 {
716 dec->cur += 5; 1660 dec->cur += 5;
1661#if JSON_SLOW
1662 bool_false = get_bool ("Types::Serialiser::false");
1663#endif
717 return newSViv (0); 1664 return newSVsv (bool_false);
718 } 1665 }
719 else 1666 else
720 ERR ("'false' expected"); 1667 ERR ("'false' expected");
721 1668
722 break; 1669 break;
723 1670
724 case 'n': 1671 case 'n':
725 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4)) 1672 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4))
726 { 1673 {
727 dec->cur += 4; 1674 dec->cur += 4;
728 return newSViv (1); 1675 return newSVsv (&PL_sv_undef);
729 } 1676 }
730 else 1677 else
731 ERR ("'null' expected"); 1678 ERR ("'null' expected");
732 1679
733 break; 1680 break;
734 1681
735 default: 1682 default:
736 ERR ("malformed json string"); 1683 ERR ("malformed JSON string, neither tag, array, object, number, string or atom");
737 break; 1684 break;
738 } 1685 }
739 1686
740fail: 1687fail:
741 return 0; 1688 return 0;
742} 1689}
743 1690
744static SV * 1691static SV *
745decode_json (SV *string, UV flags) 1692decode_json (SV *string, JSON *json, char **offset_return)
746{ 1693{
1694 dec_t dec;
747 SV *sv; 1695 SV *sv;
748 1696
749 if (!(flags & F_UTF8)) 1697 /* work around bugs in 5.10 where manipulating magic values
1698 * makes perl ignore the magic in subsequent accesses.
1699 * also make a copy of non-PV values, to get them into a clean
1700 * state (SvPV should do that, but it's buggy, see below).
1701 */
1702 /*SvGETMAGIC (string);*/
1703 if (SvMAGICAL (string) || !SvPOK (string))
1704 string = sv_2mortal (newSVsv (string));
1705
1706 SvUPGRADE (string, SVt_PV);
1707
1708 /* work around a bug in perl 5.10, which causes SvCUR to fail an
1709 * assertion with -DDEBUGGING, although SvCUR is documented to
1710 * return the xpv_cur field which certainly exists after upgrading.
1711 * according to nicholas clark, calling SvPOK fixes this.
1712 * But it doesn't fix it, so try another workaround, call SvPV_nolen
1713 * and hope for the best.
1714 * Damnit, SvPV_nolen still trips over yet another assertion. This
1715 * assertion business is seriously broken, try yet another workaround
1716 * for the broken -DDEBUGGING.
1717 */
1718 {
1719#ifdef DEBUGGING
1720 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
1721#else
1722 STRLEN offset = SvCUR (string);
1723#endif
1724
1725 if (offset > json->max_size && json->max_size)
1726 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1727 (unsigned long)SvCUR (string), (unsigned long)json->max_size);
1728 }
1729
1730 if (DECODE_WANTS_OCTETS (json))
1731 sv_utf8_downgrade (string, 0);
1732 else
750 sv_utf8_upgrade (string); 1733 sv_utf8_upgrade (string);
751 1734
752 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1735 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
753 1736
754 dec_t dec; 1737 dec.json = *json;
755 dec.flags = flags;
756 dec.cur = SvPVX (string); 1738 dec.cur = SvPVX (string);
757 dec.end = SvEND (string); 1739 dec.end = SvEND (string);
758 dec.err = 0; 1740 dec.err = 0;
1741 dec.depth = 0;
759 1742
760 *dec.end = 1; // invalid anywhere 1743 if (dec.json.cb_object || dec.json.cb_sk_object)
1744 dec.json.flags |= F_HOOK;
1745
1746 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1747
1748 decode_ws (&dec);
761 sv = decode_sv (&dec); 1749 sv = decode_sv (&dec);
762 *dec.end = 0; 1750
1751 if (offset_return)
1752 *offset_return = dec.cur;
1753
1754 if (!(offset_return || !sv))
1755 {
1756 // check for trailing garbage
1757 decode_ws (&dec);
1758
1759 if (*dec.cur)
1760 {
1761 dec.err = "garbage after JSON object";
1762 SvREFCNT_dec (sv);
1763 sv = 0;
1764 }
1765 }
763 1766
764 if (!sv) 1767 if (!sv)
765 { 1768 {
766 IV offset = utf8_distance (dec.cur, SvPVX (string));
767 SV *uni = sv_newmortal (); 1769 SV *uni = sv_newmortal ();
768 1770
1771 // horrible hack to silence warning inside pv_uni_display
1772 COP cop = *PL_curcop;
1773 cop.cop_warnings = pWARN_NONE;
1774 ENTER;
1775 SAVEVPTR (PL_curcop);
1776 PL_curcop = &cop;
769 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1777 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1778 LEAVE;
1779
770 croak ("%s, at character %d (%s)", 1780 croak ("%s, at character offset %d (before \"%s\")",
771 dec.err, 1781 dec.err,
772 (int)offset, 1782 (int)ptr_to_index (string, dec.cur),
773 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1783 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
774 } 1784 }
775 1785
776 sv_dump (sv);//D
777 return sv_2mortal (sv); 1786 sv = sv_2mortal (sv);
1787
1788 if (!(dec.json.flags & F_ALLOW_NONREF) && json_nonref (sv))
1789 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1790
1791 return sv;
778} 1792}
1793
1794/////////////////////////////////////////////////////////////////////////////
1795// incremental parser
1796
1797static void
1798incr_parse (JSON *self)
1799{
1800 const char *p = SvPVX (self->incr_text) + self->incr_pos;
1801
1802 // the state machine here is a bit convoluted and could be simplified a lot
1803 // but this would make it slower, so...
1804
1805 for (;;)
1806 {
1807 //printf ("loop pod %d *p<%c><%s>, mode %d nest %d\n", p - SvPVX (self->incr_text), *p, p, self->incr_mode, self->incr_nest);//D
1808 switch (self->incr_mode)
1809 {
1810 // only used for initial whitespace skipping
1811 case INCR_M_WS:
1812 for (;;)
1813 {
1814 if (*p > 0x20)
1815 {
1816 if (*p == '#')
1817 {
1818 self->incr_mode = INCR_M_C0;
1819 goto incr_m_c;
1820 }
1821 else
1822 {
1823 self->incr_mode = INCR_M_JSON;
1824 goto incr_m_json;
1825 }
1826 }
1827 else if (!*p)
1828 goto interrupt;
1829
1830 ++p;
1831 }
1832
1833 // skip a single char inside a string (for \\-processing)
1834 case INCR_M_BS:
1835 if (!*p)
1836 goto interrupt;
1837
1838 ++p;
1839 self->incr_mode = INCR_M_STR;
1840 goto incr_m_str;
1841
1842 // inside #-style comments
1843 case INCR_M_C0:
1844 case INCR_M_C1:
1845 incr_m_c:
1846 for (;;)
1847 {
1848 if (*p == '\n')
1849 {
1850 self->incr_mode = self->incr_mode == INCR_M_C0 ? INCR_M_WS : INCR_M_JSON;
1851 break;
1852 }
1853 else if (!*p)
1854 goto interrupt;
1855
1856 ++p;
1857 }
1858
1859 break;
1860
1861 // inside a string
1862 case INCR_M_STR:
1863 incr_m_str:
1864 for (;;)
1865 {
1866 if (*p == '"')
1867 {
1868 ++p;
1869 self->incr_mode = INCR_M_JSON;
1870
1871 if (!self->incr_nest)
1872 goto interrupt;
1873
1874 goto incr_m_json;
1875 }
1876 else if (*p == '\\')
1877 {
1878 ++p; // "virtually" consumes character after \
1879
1880 if (!*p) // if at end of string we have to switch modes
1881 {
1882 self->incr_mode = INCR_M_BS;
1883 goto interrupt;
1884 }
1885 }
1886 else if (!*p)
1887 goto interrupt;
1888
1889 ++p;
1890 }
1891
1892 // after initial ws, outside string
1893 case INCR_M_JSON:
1894 incr_m_json:
1895 for (;;)
1896 {
1897 switch (*p++)
1898 {
1899 case 0:
1900 --p;
1901 goto interrupt;
1902
1903 case 0x09:
1904 case 0x0a:
1905 case 0x0d:
1906 case 0x20:
1907 if (!self->incr_nest)
1908 {
1909 --p; // do not eat the whitespace, let the next round do it
1910 goto interrupt;
1911 }
1912 break;
1913
1914 case '"':
1915 self->incr_mode = INCR_M_STR;
1916 goto incr_m_str;
1917
1918 case '[':
1919 case '{':
1920 case '(':
1921 if (++self->incr_nest > self->max_depth)
1922 croak (ERR_NESTING_EXCEEDED);
1923 break;
1924
1925 case ']':
1926 case '}':
1927 if (--self->incr_nest <= 0)
1928 goto interrupt;
1929 break;
1930
1931 case ')':
1932 --self->incr_nest;
1933 break;
1934
1935 case '#':
1936 self->incr_mode = INCR_M_C1;
1937 goto incr_m_c;
1938 }
1939 }
1940 }
1941
1942 modechange:
1943 ;
1944 }
1945
1946interrupt:
1947 self->incr_pos = p - SvPVX (self->incr_text);
1948 //printf ("interrupt<%.*s>\n", self->incr_pos, SvPVX(self->incr_text));//D
1949 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D
1950}
1951
1952/////////////////////////////////////////////////////////////////////////////
1953// XS interface functions
779 1954
780MODULE = JSON::XS PACKAGE = JSON::XS 1955MODULE = JSON::XS PACKAGE = JSON::XS
781 1956
782BOOT: 1957BOOT:
783{ 1958{
784 int i; 1959 int i;
785 1960
786 memset (decode_hexdigit, 0xff, 256);
787 for (i = 10; i--; ) 1961 for (i = 0; i < 256; ++i)
788 decode_hexdigit ['0' + i] = i; 1962 decode_hexdigit [i] =
1963 i >= '0' && i <= '9' ? i - '0'
1964 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1965 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1966 : -1;
789 1967
790 for (i = 6; --i; )
791 {
792 decode_hexdigit ['a' + i] = 10 + i;
793 decode_hexdigit ['A' + i] = 10 + i;
794 }
795
796 json_stash = gv_stashpv ("JSON::XS", 1); 1968 json_stash = gv_stashpv ("JSON::XS" , 1);
797} 1969 bool_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
1970 bool_true = get_bool ("Types::Serialiser::true");
1971 bool_false = get_bool ("Types::Serialiser::false");
798 1972
799SV *new (char *dummy) 1973 sv_json = newSVpv ("JSON", 0);
1974 SvREADONLY_on (sv_json);
1975
1976 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */
1977}
1978
1979PROTOTYPES: DISABLE
1980
1981void CLONE (...)
800 CODE: 1982 CODE:
801 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1983 json_stash = 0;
1984 bool_stash = 0;
1985
1986void new (char *klass)
1987 PPCODE:
1988{
1989 SV *pv = NEWSV (0, sizeof (JSON));
1990 SvPOK_only (pv);
1991 json_init ((JSON *)SvPVX (pv));
1992 XPUSHs (sv_2mortal (sv_bless (
1993 newRV_noinc (pv),
1994 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1995 )));
1996}
1997
1998void ascii (JSON *self, int enable = 1)
1999 ALIAS:
2000 ascii = F_ASCII
2001 latin1 = F_LATIN1
2002 utf8 = F_UTF8
2003 indent = F_INDENT
2004 canonical = F_CANONICAL
2005 space_before = F_SPACE_BEFORE
2006 space_after = F_SPACE_AFTER
2007 pretty = F_PRETTY
2008 allow_nonref = F_ALLOW_NONREF
2009 shrink = F_SHRINK
2010 allow_blessed = F_ALLOW_BLESSED
2011 convert_blessed = F_CONV_BLESSED
2012 relaxed = F_RELAXED
2013 allow_unknown = F_ALLOW_UNKNOWN
2014 allow_tags = F_ALLOW_TAGS
2015 PPCODE:
2016{
2017 if (enable)
2018 self->flags |= ix;
2019 else
2020 self->flags &= ~ix;
2021
2022 XPUSHs (ST (0));
2023}
2024
2025void get_ascii (JSON *self)
2026 ALIAS:
2027 get_ascii = F_ASCII
2028 get_latin1 = F_LATIN1
2029 get_utf8 = F_UTF8
2030 get_indent = F_INDENT
2031 get_canonical = F_CANONICAL
2032 get_space_before = F_SPACE_BEFORE
2033 get_space_after = F_SPACE_AFTER
2034 get_allow_nonref = F_ALLOW_NONREF
2035 get_shrink = F_SHRINK
2036 get_allow_blessed = F_ALLOW_BLESSED
2037 get_convert_blessed = F_CONV_BLESSED
2038 get_relaxed = F_RELAXED
2039 get_allow_unknown = F_ALLOW_UNKNOWN
2040 get_allow_tags = F_ALLOW_TAGS
2041 PPCODE:
2042 XPUSHs (boolSV (self->flags & ix));
2043
2044void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
2045 PPCODE:
2046 self->max_depth = max_depth;
2047 XPUSHs (ST (0));
2048
2049U32 get_max_depth (JSON *self)
2050 CODE:
2051 RETVAL = self->max_depth;
802 OUTPUT: 2052 OUTPUT:
803 RETVAL 2053 RETVAL
804 2054
805SV *ascii (SV *self, int enable) 2055void max_size (JSON *self, U32 max_size = 0)
806 ALIAS: 2056 PPCODE:
807 ascii = F_ASCII 2057 self->max_size = max_size;
808 utf8 = F_UTF8 2058 XPUSHs (ST (0));
809 indent = F_INDENT 2059
810 canonical = F_CANONICAL 2060int get_max_size (JSON *self)
811 space_before = F_SPACE_BEFORE
812 space_after = F_SPACE_AFTER
813 json_rpc = F_JSON_RPC
814 CODE: 2061 CODE:
815{ 2062 RETVAL = self->max_size;
816 UV *uv = SvJSON (self);
817 if (enable)
818 *uv |= ix;
819 else
820 *uv &= ~ix;
821
822 RETVAL = newSVsv (self);
823}
824 OUTPUT: 2063 OUTPUT:
825 RETVAL 2064 RETVAL
826 2065
827void encode (SV *self, SV *scalar) 2066void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
828 PPCODE: 2067 PPCODE:
829 XPUSHs (encode_json (scalar, *SvJSON (self))); 2068{
2069 SvREFCNT_dec (self->cb_object);
2070 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
830 2071
831void decode (SV *self, SV *jsondata) 2072 XPUSHs (ST (0));
2073}
2074
2075void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
832 PPCODE: 2076 PPCODE:
833 XPUSHs (decode_json (jsondata, *SvJSON (self))); 2077{
2078 if (!self->cb_sk_object)
2079 self->cb_sk_object = newHV ();
834 2080
2081 if (SvOK (cb))
2082 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
2083 else
2084 {
2085 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
2086
2087 if (!HvKEYS (self->cb_sk_object))
2088 {
2089 SvREFCNT_dec (self->cb_sk_object);
2090 self->cb_sk_object = 0;
2091 }
2092 }
2093
2094 XPUSHs (ST (0));
2095}
2096
2097void encode (JSON *self, SV *scalar)
2098 PPCODE:
2099 PUTBACK; scalar = encode_json (scalar, self); SPAGAIN;
2100 XPUSHs (scalar);
2101
2102void decode (JSON *self, SV *jsonstr)
2103 PPCODE:
2104 PUTBACK; jsonstr = decode_json (jsonstr, self, 0); SPAGAIN;
2105 XPUSHs (jsonstr);
2106
2107void decode_prefix (JSON *self, SV *jsonstr)
2108 PPCODE:
2109{
2110 SV *sv;
2111 char *offset;
2112 PUTBACK; sv = decode_json (jsonstr, self, &offset); SPAGAIN;
2113 EXTEND (SP, 2);
2114 PUSHs (sv);
2115 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset))));
2116}
2117
2118void incr_parse (JSON *self, SV *jsonstr = 0)
2119 PPCODE:
2120{
2121 if (!self->incr_text)
2122 self->incr_text = newSVpvn ("", 0);
2123
2124 /* if utf8-ness doesn't match the decoder, need to upgrade/downgrade */
2125 if (!DECODE_WANTS_OCTETS (self) == !SvUTF8 (self->incr_text))
2126 if (DECODE_WANTS_OCTETS (self))
2127 {
2128 if (self->incr_pos)
2129 self->incr_pos = utf8_length ((U8 *)SvPVX (self->incr_text),
2130 (U8 *)SvPVX (self->incr_text) + self->incr_pos);
2131
2132 sv_utf8_downgrade (self->incr_text, 0);
2133 }
2134 else
2135 {
2136 sv_utf8_upgrade (self->incr_text);
2137
2138 if (self->incr_pos)
2139 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
2140 - (U8 *)SvPVX (self->incr_text);
2141 }
2142
2143 // append data, if any
2144 if (jsonstr)
2145 {
2146 /* make sure both strings have same encoding */
2147 if (SvUTF8 (jsonstr) != SvUTF8 (self->incr_text))
2148 if (SvUTF8 (jsonstr))
2149 sv_utf8_downgrade (jsonstr, 0);
2150 else
2151 sv_utf8_upgrade (jsonstr);
2152
2153 /* and then just blindly append */
2154 {
2155 STRLEN len;
2156 const char *str = SvPV (jsonstr, len);
2157 STRLEN cur = SvCUR (self->incr_text);
2158
2159 if (SvLEN (self->incr_text) <= cur + len)
2160 SvGROW (self->incr_text, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
2161
2162 Move (str, SvEND (self->incr_text), len, char);
2163 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
2164 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
2165 }
2166 }
2167
2168 if (GIMME_V != G_VOID)
2169 do
2170 {
2171 SV *sv;
2172 char *offset;
2173
2174 if (!INCR_DONE (self))
2175 {
2176 incr_parse (self);
2177
2178 if (self->incr_pos > self->max_size && self->max_size)
2179 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
2180 (unsigned long)self->incr_pos, (unsigned long)self->max_size);
2181
2182 if (!INCR_DONE (self))
2183 {
2184 // as an optimisation, do not accumulate white space in the incr buffer
2185 if (self->incr_mode == INCR_M_WS && self->incr_pos)
2186 {
2187 self->incr_pos = 0;
2188 SvCUR_set (self->incr_text, 0);
2189 }
2190
2191 break;
2192 }
2193 }
2194
2195 PUTBACK; sv = decode_json (self->incr_text, self, &offset); SPAGAIN;
2196 XPUSHs (sv);
2197
2198 self->incr_pos -= offset - SvPVX (self->incr_text);
2199 self->incr_nest = 0;
2200 self->incr_mode = 0;
2201
2202 sv_chop (self->incr_text, offset);
2203 }
2204 while (GIMME_V == G_ARRAY);
2205}
2206
2207SV *incr_text (JSON *self)
2208 ATTRS: lvalue
2209 CODE:
2210{
2211 if (self->incr_pos)
2212 croak ("incr_text can not be called when the incremental parser already started parsing");
2213
2214 RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef;
2215}
2216 OUTPUT:
2217 RETVAL
2218
2219void incr_skip (JSON *self)
2220 CODE:
2221{
2222 if (self->incr_pos)
2223 {
2224 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + self->incr_pos);
2225 self->incr_pos = 0;
2226 self->incr_nest = 0;
2227 self->incr_mode = 0;
2228 }
2229}
2230
2231void incr_reset (JSON *self)
2232 CODE:
2233{
2234 SvREFCNT_dec (self->incr_text);
2235 self->incr_text = 0;
2236 self->incr_pos = 0;
2237 self->incr_nest = 0;
2238 self->incr_mode = 0;
2239}
2240
2241void DESTROY (JSON *self)
2242 CODE:
2243 SvREFCNT_dec (self->cb_sk_object);
2244 SvREFCNT_dec (self->cb_object);
2245 SvREFCNT_dec (self->incr_text);
2246
2247PROTOTYPES: ENABLE
2248
2249void encode_json (SV *scalar)
2250 PPCODE:
2251{
2252 JSON json;
2253 json_init (&json);
2254 json.flags |= F_UTF8;
2255 PUTBACK; scalar = encode_json (scalar, &json); SPAGAIN;
2256 XPUSHs (scalar);
2257}
2258
2259void decode_json (SV *jsonstr)
2260 PPCODE:
2261{
2262 JSON json;
2263 json_init (&json);
2264 json.flags |= F_UTF8;
2265 PUTBACK; jsonstr = decode_json (jsonstr, &json, 0); SPAGAIN;
2266 XPUSHs (jsonstr);
2267}
2268

Diff Legend

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