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 1.27 by root, Mon Apr 9 05:09:57 2007 UTC vs.
Revision 1.114 by root, Wed Aug 1 19:04:41 2012 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" 8#include <stdio.h>
9#include <limits.h>
10#include <float.h>
9 11
10#if defined(__BORLANDC__) || defined(_MSC_VER) 12#if defined(__BORLANDC__) || defined(_MSC_VER)
11# define snprintf _snprintf // C compilers have this in stdio.h 13# define snprintf _snprintf // C compilers have this in stdio.h
12#endif 14#endif
13 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// three extra for rounding, sign, and end of string
23#define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 3)
24
14#define F_ASCII 0x00000001UL 25#define F_ASCII 0x00000001UL
26#define F_LATIN1 0x00000002UL
15#define F_UTF8 0x00000002UL 27#define F_UTF8 0x00000004UL
16#define F_INDENT 0x00000004UL 28#define F_INDENT 0x00000008UL
17#define F_CANONICAL 0x00000008UL 29#define F_CANONICAL 0x00000010UL
18#define F_SPACE_BEFORE 0x00000010UL 30#define F_SPACE_BEFORE 0x00000020UL
19#define F_SPACE_AFTER 0x00000020UL 31#define F_SPACE_AFTER 0x00000040UL
20#define F_ALLOW_NONREF 0x00000080UL 32#define F_ALLOW_NONREF 0x00000100UL
21#define F_SHRINK 0x00000100UL 33#define F_SHRINK 0x00000200UL
34#define F_ALLOW_BLESSED 0x00000400UL
35#define F_CONV_BLESSED 0x00000800UL
22#define F_MAXDEPTH 0xf8000000UL 36#define F_RELAXED 0x00001000UL
23#define S_MAXDEPTH 27 37#define F_ALLOW_UNKNOWN 0x00002000UL
24 38#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
25#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
26
27// F_SELFCONVERT? <=> to_json/toJson
28// F_BLESSED? <=> { $__class__$ => }
29 39
30#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 40#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
31#define F_DEFAULT (9UL << S_MAXDEPTH)
32 41
33#define INIT_SIZE 32 // initial scalar size to be allocated 42#define INIT_SIZE 32 // initial scalar size to be allocated
34#define INDENT_STEP 3 // spaces per indentation level 43#define INDENT_STEP 3 // spaces per indentation level
35 44
36#define UTF8_MAX_LEN 11 // for perls UTF-X: max. number of octets per character
37#define SHORT_STRING_LEN 512 // special-case strings of up to this size 45#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
46
47#define DECODE_WANTS_OCTETS(json) ((json)->flags & F_UTF8)
38 48
39#define SB do { 49#define SB do {
40#define SE } while (0) 50#define SE } while (0)
41 51
52#if __GNUC__ >= 3
53# define expect(expr,value) __builtin_expect ((expr), (value))
54# define INLINE static inline
55#else
56# define expect(expr,value) (expr)
57# define INLINE static
58#endif
59
60#define expect_false(expr) expect ((expr) != 0, 0)
61#define expect_true(expr) expect ((expr) != 0, 1)
62
63#define IN_RANGE_INC(type,val,beg,end) \
64 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \
65 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
66
67#define ERR_NESTING_EXCEEDED "json text or perl structure exceeds maximum nesting level (max_depth set too low?)"
68
69#ifdef USE_ITHREADS
70# define JSON_SLOW 1
71# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1))
72#else
73# define JSON_SLOW 0
74# define JSON_STASH json_stash
75#endif
76
42static HV *json_stash; // JSON::XS:: 77static HV *json_stash, *json_boolean_stash; // JSON::XS::
78static SV *json_true, *json_false;
79
80enum {
81 INCR_M_WS = 0, // initial whitespace skipping, must be 0
82 INCR_M_STR, // inside string
83 INCR_M_BS, // inside backslash
84 INCR_M_C0, // inside comment in initial whitespace sequence
85 INCR_M_C1, // inside comment in other places
86 INCR_M_JSON // outside anything, count nesting
87};
88
89#define INCR_DONE(json) ((json)->incr_nest <= 0 && (json)->incr_mode == INCR_M_JSON)
90
91typedef struct {
92 U32 flags;
93 U32 max_depth;
94 STRLEN max_size;
95
96 SV *cb_object;
97 HV *cb_sk_object;
98
99 // for the incremental parser
100 SV *incr_text; // the source text so far
101 STRLEN incr_pos; // the current offset into the text
102 int incr_nest; // {[]}-nesting level
103 unsigned char incr_mode;
104} JSON;
105
106INLINE void
107json_init (JSON *json)
108{
109 Zero (json, 1, JSON);
110 json->max_depth = 512;
111}
43 112
44///////////////////////////////////////////////////////////////////////////// 113/////////////////////////////////////////////////////////////////////////////
45// utility functions 114// utility functions
46 115
47static UV * 116INLINE SV *
48SvJSON (SV *sv) 117get_bool (const char *name)
49{ 118{
50 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash)) 119 SV *sv = get_sv (name, 1);
51 croak ("object is not of type JSON::XS");
52 120
53 return &SvUVX (SvRV (sv)); 121 SvREADONLY_on (sv);
54} 122 SvREADONLY_on (SvRV (sv));
55 123
56static void 124 return sv;
125}
126
127INLINE void
57shrink (SV *sv) 128shrink (SV *sv)
58{ 129{
59 sv_utf8_downgrade (sv, 1); 130 sv_utf8_downgrade (sv, 1);
131
60 if (SvLEN (sv) > SvCUR (sv) + 1) 132 if (SvLEN (sv) > SvCUR (sv) + 1)
61 { 133 {
62#ifdef SvPV_shrink_to_cur 134#ifdef SvPV_shrink_to_cur
63 SvPV_shrink_to_cur (sv); 135 SvPV_shrink_to_cur (sv);
64#elif defined (SvPV_renew) 136#elif defined (SvPV_renew)
… …
70// decode an utf-8 character and return it, or (UV)-1 in 142// decode an utf-8 character and return it, or (UV)-1 in
71// case of an error. 143// case of an error.
72// we special-case "safe" characters from U+80 .. U+7FF, 144// we special-case "safe" characters from U+80 .. U+7FF,
73// but use the very good perl function to parse anything else. 145// but use the very good perl function to parse anything else.
74// note that we never call this function for a ascii codepoints 146// note that we never call this function for a ascii codepoints
75static UV 147INLINE UV
76decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) 148decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
77{ 149{
78 if (s[0] > 0xdf || s[0] < 0xc2) 150 if (expect_true (len >= 2
79 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); 151 && IN_RANGE_INC (char, s[0], 0xc2, 0xdf)
80 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) 152 && IN_RANGE_INC (char, s[1], 0x80, 0xbf)))
81 { 153 {
82 *clen = 2; 154 *clen = 2;
83 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 155 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
84 } 156 }
85 else 157 else
86 { 158 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
87 *clen = (STRLEN)-1; 159}
88 return (UV)-1; 160
161// likewise for encoding, also never called for ascii codepoints
162// this function takes advantage of this fact, although current gccs
163// seem to optimise the check for >= 0x80 away anyways
164INLINE unsigned char *
165encode_utf8 (unsigned char *s, UV ch)
166{
167 if (expect_false (ch < 0x000080))
168 *s++ = ch;
169 else if (expect_true (ch < 0x000800))
170 *s++ = 0xc0 | ( ch >> 6),
171 *s++ = 0x80 | ( ch & 0x3f);
172 else if ( ch < 0x010000)
173 *s++ = 0xe0 | ( ch >> 12),
174 *s++ = 0x80 | ((ch >> 6) & 0x3f),
175 *s++ = 0x80 | ( ch & 0x3f);
176 else if ( ch < 0x110000)
177 *s++ = 0xf0 | ( ch >> 18),
178 *s++ = 0x80 | ((ch >> 12) & 0x3f),
179 *s++ = 0x80 | ((ch >> 6) & 0x3f),
180 *s++ = 0x80 | ( ch & 0x3f);
181
182 return s;
183}
184
185// convert offset pointer to character index, sv must be string
186static STRLEN
187ptr_to_index (SV *sv, char *offset)
188{
189 return SvUTF8 (sv)
190 ? utf8_distance (offset, SvPVX (sv))
191 : offset - SvPVX (sv);
192}
193
194/////////////////////////////////////////////////////////////////////////////
195// fp hell
196
197// scan a group of digits, and a trailing exponent
198static void
199json_atof_scan1 (const char *s, NV *accum, int *expo, int postdp, int maxdepth)
200{
201 UV uaccum = 0;
202 int eaccum = 0;
203
204 // if we recurse too deep, skip all remaining digits
205 // to avoid a stack overflow attack
206 if (expect_false (--maxdepth <= 0))
207 while (((U8)*s - '0') < 10)
208 ++s;
209
210 for (;;)
89 } 211 {
90} 212 U8 dig = (U8)*s - '0';
91 213
214 if (expect_false (dig >= 10))
215 {
216 if (dig == (U8)((U8)'.' - (U8)'0'))
217 {
218 ++s;
219 json_atof_scan1 (s, accum, expo, 1, maxdepth);
220 }
221 else if ((dig | ' ') == 'e' - '0')
222 {
223 int exp2 = 0;
224 int neg = 0;
225
226 ++s;
227
228 if (*s == '-')
229 {
230 ++s;
231 neg = 1;
232 }
233 else if (*s == '+')
234 ++s;
235
236 while ((dig = (U8)*s - '0') < 10)
237 exp2 = exp2 * 10 + *s++ - '0';
238
239 *expo += neg ? -exp2 : exp2;
240 }
241
242 break;
243 }
244
245 ++s;
246
247 uaccum = uaccum * 10 + dig;
248 ++eaccum;
249
250 // if we have too many digits, then recurse for more
251 // we actually do this for rather few digits
252 if (uaccum >= (UV_MAX - 9) / 10)
253 {
254 if (postdp) *expo -= eaccum;
255 json_atof_scan1 (s, accum, expo, postdp, maxdepth);
256 if (postdp) *expo += eaccum;
257
258 break;
259 }
260 }
261
262 // this relies greatly on the quality of the pow ()
263 // implementation of the platform, but a good
264 // implementation is hard to beat.
265 // (IEEE 754 conformant ones are required to be exact)
266 if (postdp) *expo -= eaccum;
267 *accum += uaccum * Perl_pow (10., *expo);
268 *expo += eaccum;
269}
270
271static NV
272json_atof (const char *s)
273{
274 NV accum = 0.;
275 int expo = 0;
276 int neg = 0;
277
278 if (*s == '-')
279 {
280 ++s;
281 neg = 1;
282 }
283
284 // a recursion depth of ten gives us >>500 bits
285 json_atof_scan1 (s, &accum, &expo, 0, 10);
286
287 return neg ? -accum : accum;
288}
92///////////////////////////////////////////////////////////////////////////// 289/////////////////////////////////////////////////////////////////////////////
93// encoder 290// encoder
94 291
95// structure used for encoding JSON 292// structure used for encoding JSON
96typedef struct 293typedef struct
97{ 294{
98 char *cur; // SvPVX (sv) + current output position 295 char *cur; // SvPVX (sv) + current output position
99 char *end; // SvEND (sv) 296 char *end; // SvEND (sv)
100 SV *sv; // result scalar 297 SV *sv; // result scalar
101 U32 flags; // F_* 298 JSON json;
102 U32 indent; // indentation level 299 U32 indent; // indentation level
103 U32 maxdepth; // max. indentation/recursion level 300 UV limit; // escape character values >= this value when encoding
104} enc_t; 301} enc_t;
105 302
106static void 303INLINE void
107need (enc_t *enc, STRLEN len) 304need (enc_t *enc, STRLEN len)
108{ 305{
109 if (enc->cur + len >= enc->end) 306 if (expect_false (enc->cur + len >= enc->end))
110 { 307 {
111 STRLEN cur = enc->cur - SvPVX (enc->sv); 308 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
112 SvGROW (enc->sv, cur + len + 1); 309 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
113 enc->cur = SvPVX (enc->sv) + cur; 310 enc->cur = SvPVX (enc->sv) + cur;
114 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 311 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
115 } 312 }
116} 313}
117 314
118static void 315INLINE void
119encode_ch (enc_t *enc, char ch) 316encode_ch (enc_t *enc, char ch)
120{ 317{
121 need (enc, 1); 318 need (enc, 1);
122 *enc->cur++ = ch; 319 *enc->cur++ = ch;
123} 320}
… …
131 328
132 while (str < end) 329 while (str < end)
133 { 330 {
134 unsigned char ch = *(unsigned char *)str; 331 unsigned char ch = *(unsigned char *)str;
135 332
136 if (ch >= 0x20 && ch < 0x80) // most common case 333 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
137 { 334 {
138 if (ch == '"') // but with slow exceptions 335 if (expect_false (ch == '"')) // but with slow exceptions
139 { 336 {
140 need (enc, len += 1); 337 need (enc, len += 1);
141 *enc->cur++ = '\\'; 338 *enc->cur++ = '\\';
142 *enc->cur++ = '"'; 339 *enc->cur++ = '"';
143 } 340 }
144 else if (ch == '\\') 341 else if (expect_false (ch == '\\'))
145 { 342 {
146 need (enc, len += 1); 343 need (enc, len += 1);
147 *enc->cur++ = '\\'; 344 *enc->cur++ = '\\';
148 *enc->cur++ = '\\'; 345 *enc->cur++ = '\\';
149 } 346 }
… …
167 STRLEN clen; 364 STRLEN clen;
168 UV uch; 365 UV uch;
169 366
170 if (is_utf8) 367 if (is_utf8)
171 { 368 {
172 //uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
173 uch = decode_utf8 (str, end - str, &clen); 369 uch = decode_utf8 (str, end - str, &clen);
174 if (clen == (STRLEN)-1) 370 if (clen == (STRLEN)-1)
175 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str); 371 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
176 } 372 }
177 else 373 else
178 { 374 {
179 uch = ch; 375 uch = ch;
180 clen = 1; 376 clen = 1;
181 } 377 }
182 378
183 if (uch > 0x10FFFFUL) 379 if (uch < 0x80/*0x20*/ || uch >= enc->limit)
184 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
185
186 if (uch < 0x80 || enc->flags & F_ASCII)
187 { 380 {
188 if (uch > 0xFFFFUL) 381 if (uch >= 0x10000UL)
189 { 382 {
383 if (uch >= 0x110000UL)
384 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
385
190 need (enc, len += 11); 386 need (enc, len += 11);
191 sprintf (enc->cur, "\\u%04x\\u%04x", 387 sprintf (enc->cur, "\\u%04x\\u%04x",
192 (int)((uch - 0x10000) / 0x400 + 0xD800), 388 (int)((uch - 0x10000) / 0x400 + 0xD800),
193 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 389 (int)((uch - 0x10000) % 0x400 + 0xDC00));
194 enc->cur += 12; 390 enc->cur += 12;
195 } 391 }
196 else 392 else
197 { 393 {
198 static char hexdigit [16] = "0123456789abcdef";
199 need (enc, len += 5); 394 need (enc, len += 5);
200 *enc->cur++ = '\\'; 395 *enc->cur++ = '\\';
201 *enc->cur++ = 'u'; 396 *enc->cur++ = 'u';
202 *enc->cur++ = hexdigit [ uch >> 12 ]; 397 *enc->cur++ = PL_hexdigit [ uch >> 12 ];
203 *enc->cur++ = hexdigit [(uch >> 8) & 15]; 398 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
204 *enc->cur++ = hexdigit [(uch >> 4) & 15]; 399 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
205 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 400 *enc->cur++ = PL_hexdigit [(uch >> 0) & 15];
206 } 401 }
207 402
403 str += clen;
404 }
405 else if (enc->json.flags & F_LATIN1)
406 {
407 *enc->cur++ = uch;
208 str += clen; 408 str += clen;
209 } 409 }
210 else if (is_utf8) 410 else if (is_utf8)
211 { 411 {
212 need (enc, len += clen); 412 need (enc, len += clen);
… …
216 } 416 }
217 while (--clen); 417 while (--clen);
218 } 418 }
219 else 419 else
220 { 420 {
221 need (enc, len += UTF8_MAX_LEN - 1); // never more than 11 bytes needed 421 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
222 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 422 enc->cur = encode_utf8 (enc->cur, uch);
223 ++str; 423 ++str;
224 } 424 }
225 } 425 }
226 } 426 }
227 } 427 }
228 428
229 --len; 429 --len;
230 } 430 }
231} 431}
232 432
233static void 433INLINE void
234encode_indent (enc_t *enc) 434encode_indent (enc_t *enc)
235{ 435{
236 if (enc->flags & F_INDENT) 436 if (enc->json.flags & F_INDENT)
237 { 437 {
238 int spaces = enc->indent * INDENT_STEP; 438 int spaces = enc->indent * INDENT_STEP;
239 439
240 need (enc, spaces); 440 need (enc, spaces);
241 memset (enc->cur, ' ', spaces); 441 memset (enc->cur, ' ', spaces);
242 enc->cur += spaces; 442 enc->cur += spaces;
243 } 443 }
244} 444}
245 445
246static void 446INLINE void
247encode_space (enc_t *enc) 447encode_space (enc_t *enc)
248{ 448{
249 need (enc, 1); 449 need (enc, 1);
250 encode_ch (enc, ' '); 450 encode_ch (enc, ' ');
251} 451}
252 452
253static void 453INLINE void
254encode_nl (enc_t *enc) 454encode_nl (enc_t *enc)
255{ 455{
256 if (enc->flags & F_INDENT) 456 if (enc->json.flags & F_INDENT)
257 { 457 {
258 need (enc, 1); 458 need (enc, 1);
259 encode_ch (enc, '\n'); 459 encode_ch (enc, '\n');
260 } 460 }
261} 461}
262 462
263static void 463INLINE void
264encode_comma (enc_t *enc) 464encode_comma (enc_t *enc)
265{ 465{
266 encode_ch (enc, ','); 466 encode_ch (enc, ',');
267 467
268 if (enc->flags & F_INDENT) 468 if (enc->json.flags & F_INDENT)
269 encode_nl (enc); 469 encode_nl (enc);
270 else if (enc->flags & F_SPACE_AFTER) 470 else if (enc->json.flags & F_SPACE_AFTER)
271 encode_space (enc); 471 encode_space (enc);
272} 472}
273 473
274static void encode_sv (enc_t *enc, SV *sv); 474static void encode_sv (enc_t *enc, SV *sv);
275 475
276static void 476static void
277encode_av (enc_t *enc, AV *av) 477encode_av (enc_t *enc, AV *av)
278{ 478{
279 int i, len = av_len (av); 479 int i, len = av_len (av);
280 480
281 if (enc->indent >= enc->maxdepth) 481 if (enc->indent >= enc->json.max_depth)
282 croak ("data structure too deep (hit recursion limit)"); 482 croak (ERR_NESTING_EXCEEDED);
283 483
284 encode_ch (enc, '['); encode_nl (enc); 484 encode_ch (enc, '[');
285 ++enc->indent; 485
486 if (len >= 0)
487 {
488 encode_nl (enc); ++enc->indent;
286 489
287 for (i = 0; i <= len; ++i) 490 for (i = 0; i <= len; ++i)
288 { 491 {
492 SV **svp = av_fetch (av, i, 0);
493
289 encode_indent (enc); 494 encode_indent (enc);
290 encode_sv (enc, *av_fetch (av, i, 0));
291 495
496 if (svp)
497 encode_sv (enc, *svp);
498 else
499 encode_str (enc, "null", 4, 0);
500
292 if (i < len) 501 if (i < len)
293 encode_comma (enc); 502 encode_comma (enc);
294 } 503 }
295 504
505 encode_nl (enc); --enc->indent; encode_indent (enc);
506 }
507
296 encode_nl (enc); 508 encode_ch (enc, ']');
297
298 --enc->indent;
299 encode_indent (enc); encode_ch (enc, ']');
300} 509}
301 510
302static void 511static void
303encode_he (enc_t *enc, HE *he) 512encode_hk (enc_t *enc, HE *he)
304{ 513{
305 encode_ch (enc, '"'); 514 encode_ch (enc, '"');
306 515
307 if (HeKLEN (he) == HEf_SVKEY) 516 if (HeKLEN (he) == HEf_SVKEY)
308 { 517 {
… …
318 else 527 else
319 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 528 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
320 529
321 encode_ch (enc, '"'); 530 encode_ch (enc, '"');
322 531
323 if (enc->flags & F_SPACE_BEFORE) encode_space (enc); 532 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
324 encode_ch (enc, ':'); 533 encode_ch (enc, ':');
325 if (enc->flags & F_SPACE_AFTER ) encode_space (enc); 534 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
326 encode_sv (enc, HeVAL (he));
327} 535}
328 536
329// compare hash entries, used when all keys are bytestrings 537// compare hash entries, used when all keys are bytestrings
330static int 538static int
331he_cmp_fast (const void *a_, const void *b_) 539he_cmp_fast (const void *a_, const void *b_)
… …
336 HE *b = *(HE **)b_; 544 HE *b = *(HE **)b_;
337 545
338 STRLEN la = HeKLEN (a); 546 STRLEN la = HeKLEN (a);
339 STRLEN lb = HeKLEN (b); 547 STRLEN lb = HeKLEN (b);
340 548
341 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 549 if (!(cmp = memcmp (HeKEY (b), HeKEY (a), lb < la ? lb : la)))
342 cmp = la - lb; 550 cmp = lb - la;
343 551
344 return cmp; 552 return cmp;
345} 553}
346 554
347// compare hash entries, used when some keys are sv's or utf-x 555// compare hash entries, used when some keys are sv's or utf-x
348static int 556static int
349he_cmp_slow (const void *a, const void *b) 557he_cmp_slow (const void *a, const void *b)
350{ 558{
351 return sv_cmp (HeSVKEY_force (*(HE **)a), HeSVKEY_force (*(HE **)b)); 559 return sv_cmp (HeSVKEY_force (*(HE **)b), HeSVKEY_force (*(HE **)a));
352} 560}
353 561
354static void 562static void
355encode_hv (enc_t *enc, HV *hv) 563encode_hv (enc_t *enc, HV *hv)
356{ 564{
357 int count, i; 565 HE *he;
358 566
359 if (enc->indent >= enc->maxdepth) 567 if (enc->indent >= enc->json.max_depth)
360 croak ("data structure too deep (hit recursion limit)"); 568 croak (ERR_NESTING_EXCEEDED);
361 569
362 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 570 encode_ch (enc, '{');
363 571
364 if ((count = hv_iterinit (hv)))
365 {
366 // for canonical output we have to sort by keys first 572 // for canonical output we have to sort by keys first
367 // actually, this is mostly due to the stupid so-called 573 // actually, this is mostly due to the stupid so-called
368 // security workaround added somewhere in 5.8.x. 574 // security workaround added somewhere in 5.8.x
369 // that randomises hash orderings 575 // that randomises hash orderings
370 if (enc->flags & F_CANONICAL) 576 if (enc->json.flags & F_CANONICAL && !SvRMAGICAL (hv))
577 {
578 int count = hv_iterinit (hv);
579
580 if (SvMAGICAL (hv))
371 { 581 {
582 // need to count by iterating. could improve by dynamically building the vector below
583 // but I don't care for the speed of this special case.
584 // note also that we will run into undefined behaviour when the two iterations
585 // do not result in the same count, something I might care for in some later release.
586
587 count = 0;
588 while (hv_iternext (hv))
589 ++count;
590
591 hv_iterinit (hv);
592 }
593
594 if (count)
595 {
596 int i, fast = 1;
597#if defined(__BORLANDC__) || defined(_MSC_VER)
598 HE **hes = _alloca (count * sizeof (HE));
599#else
372 HE *he, *hes [count]; // if your compiler dies here, you need to enable C99 mode 600 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
373 int fast = 1; 601#endif
374 602
375 i = 0; 603 i = 0;
376 while ((he = hv_iternext (hv))) 604 while ((he = hv_iternext (hv)))
377 { 605 {
378 hes [i++] = he; 606 hes [i++] = he;
… …
400 628
401 FREETMPS; 629 FREETMPS;
402 LEAVE; 630 LEAVE;
403 } 631 }
404 632
405 for (i = 0; i < count; ++i) 633 encode_nl (enc); ++enc->indent;
634
635 while (count--)
406 { 636 {
407 encode_indent (enc); 637 encode_indent (enc);
638 he = hes [count];
408 encode_he (enc, hes [i]); 639 encode_hk (enc, he);
640 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
409 641
410 if (i < count - 1) 642 if (count)
411 encode_comma (enc); 643 encode_comma (enc);
412 } 644 }
413 645
646 encode_nl (enc); --enc->indent; encode_indent (enc);
647 }
648 }
649 else
650 {
651 if (hv_iterinit (hv) || SvMAGICAL (hv))
652 if ((he = hv_iternext (hv)))
653 {
654 encode_nl (enc); ++enc->indent;
655
656 for (;;)
657 {
414 encode_nl (enc); 658 encode_indent (enc);
659 encode_hk (enc, he);
660 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
661
662 if (!(he = hv_iternext (hv)))
663 break;
664
665 encode_comma (enc);
666 }
667
668 encode_nl (enc); --enc->indent; encode_indent (enc);
669 }
670 }
671
672 encode_ch (enc, '}');
673}
674
675// encode objects, arrays and special \0=false and \1=true values.
676static void
677encode_rv (enc_t *enc, SV *sv)
678{
679 svtype svt;
680
681 SvGETMAGIC (sv);
682 svt = SvTYPE (sv);
683
684 if (expect_false (SvOBJECT (sv)))
685 {
686 HV *stash = !JSON_SLOW || json_boolean_stash
687 ? json_boolean_stash
688 : gv_stashpv ("JSON::XS::Boolean", 1);
689
690 if (SvSTASH (sv) == stash)
691 {
692 if (SvIV (sv))
693 encode_str (enc, "true", 4, 0);
694 else
695 encode_str (enc, "false", 5, 0);
415 } 696 }
416 else 697 else
417 { 698 {
418 HE *he = hv_iternext (hv); 699#if 0
419 700 if (0 && sv_derived_from (rv, "JSON::Literal"))
420 for (;;)
421 {
422 encode_indent (enc);
423 encode_he (enc, he);
424
425 if (!(he = hv_iternext (hv)))
426 break;
427
428 encode_comma (enc);
429 } 701 {
702 // not yet
703 }
704#endif
705 if (enc->json.flags & F_CONV_BLESSED)
706 {
707 // we re-bless the reference to get overload and other niceties right
708 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
430 709
710 if (to_json)
711 {
712 dSP;
713
714 ENTER; SAVETMPS; PUSHMARK (SP);
715 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
716
717 // calling with G_SCALAR ensures that we always get a 1 return value
718 PUTBACK;
719 call_sv ((SV *)GvCV (to_json), G_SCALAR);
720 SPAGAIN;
721
722 // catch this surprisingly common error
723 if (SvROK (TOPs) && SvRV (TOPs) == sv)
724 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
725
726 sv = POPs;
727 PUTBACK;
728
431 encode_nl (enc); 729 encode_sv (enc, sv);
730
731 FREETMPS; LEAVE;
732 }
733 else if (enc->json.flags & F_ALLOW_BLESSED)
734 encode_str (enc, "null", 4, 0);
735 else
736 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
737 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
738 }
739 else if (enc->json.flags & F_ALLOW_BLESSED)
740 encode_str (enc, "null", 4, 0);
741 else
742 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",
743 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
432 } 744 }
433 } 745 }
434
435 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
436}
437
438// encode objects, arrays and special \0=false and \1=true values.
439static void
440encode_rv (enc_t *enc, SV *sv)
441{
442 svtype svt;
443
444 SvGETMAGIC (sv);
445 svt = SvTYPE (sv);
446
447 if (svt == SVt_PVHV) 746 else if (svt == SVt_PVHV)
448 encode_hv (enc, (HV *)sv); 747 encode_hv (enc, (HV *)sv);
449 else if (svt == SVt_PVAV) 748 else if (svt == SVt_PVAV)
450 encode_av (enc, (AV *)sv); 749 encode_av (enc, (AV *)sv);
451 else if (svt < SVt_PVAV) 750 else if (svt < SVt_PVAV)
452 { 751 {
453 if (SvNIOK (sv) && SvIV (sv) == 0) 752 STRLEN len = 0;
753 char *pv = svt ? SvPV (sv, len) : 0;
754
755 if (len == 1 && *pv == '1')
756 encode_str (enc, "true", 4, 0);
757 else if (len == 1 && *pv == '0')
454 encode_str (enc, "false", 5, 0); 758 encode_str (enc, "false", 5, 0);
455 else if (SvNIOK (sv) && SvIV (sv) == 1) 759 else if (enc->json.flags & F_ALLOW_UNKNOWN)
456 encode_str (enc, "true", 4, 0); 760 encode_str (enc, "null", 4, 0);
457 else 761 else
458 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", 762 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
459 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 763 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
460 } 764 }
765 else if (enc->json.flags & F_ALLOW_UNKNOWN)
766 encode_str (enc, "null", 4, 0);
461 else 767 else
462 croak ("encountered %s, but JSON can only represent references to arrays or hashes", 768 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
463 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 769 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
464} 770}
465 771
… …
476 encode_str (enc, str, len, SvUTF8 (sv)); 782 encode_str (enc, str, len, SvUTF8 (sv));
477 encode_ch (enc, '"'); 783 encode_ch (enc, '"');
478 } 784 }
479 else if (SvNOKp (sv)) 785 else if (SvNOKp (sv))
480 { 786 {
787 // trust that perl will do the right thing w.r.t. JSON syntax.
481 need (enc, NV_DIG + 32); 788 need (enc, NV_DIG + 32);
482 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 789 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
483 enc->cur += strlen (enc->cur); 790 enc->cur += strlen (enc->cur);
484 } 791 }
485 else if (SvIOKp (sv)) 792 else if (SvIOKp (sv))
486 { 793 {
794 // we assume we can always read an IV as a UV and vice versa
795 // we assume two's complement
796 // we assume no aliasing issues in the union
797 if (SvIsUV (sv) ? SvUVX (sv) <= 59000
798 : SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000)
799 {
800 // optimise the "small number case"
801 // code will likely be branchless and use only a single multiplication
802 // works for numbers up to 59074
803 I32 i = SvIVX (sv);
804 U32 u;
805 char digit, nz = 0;
806
487 need (enc, 64); 807 need (enc, 6);
808
809 *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
810 u = i < 0 ? -i : i;
811
812 // convert to 4.28 fixed-point representation
813 u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
814
815 // now output digit by digit, each time masking out the integer part
816 // and multiplying by 5 while moving the decimal point one to the right,
817 // resulting in a net multiplication by 10.
818 // we always write the digit to memory but conditionally increment
819 // the pointer, to enable the use of conditional move instructions.
820 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffffUL) * 5;
821 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffffUL) * 5;
822 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffffUL) * 5;
823 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffffUL) * 5;
824 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
825 }
826 else
827 {
828 // large integer, use the (rather slow) snprintf way.
829 need (enc, IVUV_MAXCHARS);
488 enc->cur += 830 enc->cur +=
489 SvIsUV(sv) 831 SvIsUV(sv)
490 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 832 ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))
491 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 833 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
834 }
492 } 835 }
493 else if (SvROK (sv)) 836 else if (SvROK (sv))
494 encode_rv (enc, SvRV (sv)); 837 encode_rv (enc, SvRV (sv));
495 else if (!SvOK (sv)) 838 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN)
496 encode_str (enc, "null", 4, 0); 839 encode_str (enc, "null", 4, 0);
497 else 840 else
498 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 841 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
499 SvPV_nolen (sv), SvFLAGS (sv)); 842 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
500} 843}
501 844
502static SV * 845static SV *
503encode_json (SV *scalar, U32 flags) 846encode_json (SV *scalar, JSON *json)
504{ 847{
505 enc_t enc; 848 enc_t enc;
506 849
507 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 850 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
508 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 851 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
509 852
510 enc.flags = flags; 853 enc.json = *json;
511 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 854 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
512 enc.cur = SvPVX (enc.sv); 855 enc.cur = SvPVX (enc.sv);
513 enc.end = SvEND (enc.sv); 856 enc.end = SvEND (enc.sv);
514 enc.indent = 0; 857 enc.indent = 0;
515 enc.maxdepth = DEC_DEPTH (flags); 858 enc.limit = enc.json.flags & F_ASCII ? 0x000080UL
859 : enc.json.flags & F_LATIN1 ? 0x000100UL
860 : 0x110000UL;
516 861
517 SvPOK_only (enc.sv); 862 SvPOK_only (enc.sv);
518 encode_sv (&enc, scalar); 863 encode_sv (&enc, scalar);
519 864 encode_nl (&enc);
520 if (!(flags & (F_ASCII | F_UTF8)))
521 SvUTF8_on (enc.sv);
522 865
523 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 866 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
524 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 867 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
525 868
869 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
870 SvUTF8_on (enc.sv);
871
526 if (enc.flags & F_SHRINK) 872 if (enc.json.flags & F_SHRINK)
527 shrink (enc.sv); 873 shrink (enc.sv);
528 874
529 return enc.sv; 875 return enc.sv;
530} 876}
531 877
… …
536typedef struct 882typedef struct
537{ 883{
538 char *cur; // current parser pointer 884 char *cur; // current parser pointer
539 char *end; // end of input string 885 char *end; // end of input string
540 const char *err; // parse error, if != 0 886 const char *err; // parse error, if != 0
541 U32 flags; // F_* 887 JSON json;
542 U32 depth; // recursion depth 888 U32 depth; // recursion depth
543 U32 maxdepth; // recursion depth limit 889 U32 maxdepth; // recursion depth limit
544} dec_t; 890} dec_t;
545 891
546static void 892INLINE void
893decode_comment (dec_t *dec)
894{
895 // only '#'-style comments allowed a.t.m.
896
897 while (*dec->cur && *dec->cur != 0x0a && *dec->cur != 0x0d)
898 ++dec->cur;
899}
900
901INLINE void
547decode_ws (dec_t *dec) 902decode_ws (dec_t *dec)
548{ 903{
549 for (;;) 904 for (;;)
550 { 905 {
551 char ch = *dec->cur; 906 char ch = *dec->cur;
552 907
553 if (ch > 0x20 908 if (ch > 0x20)
909 {
910 if (expect_false (ch == '#'))
911 {
912 if (dec->json.flags & F_RELAXED)
913 decode_comment (dec);
914 else
915 break;
916 }
917 else
918 break;
919 }
554 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) 920 else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)
555 break; 921 break; // parse error, but let higher level handle it, gives better error messages
556 922
557 ++dec->cur; 923 ++dec->cur;
558 } 924 }
559} 925}
560 926
… …
564 if (*dec->cur != ch) \ 930 if (*dec->cur != ch) \
565 ERR (# ch " expected"); \ 931 ERR (# ch " expected"); \
566 ++dec->cur; \ 932 ++dec->cur; \
567 SE 933 SE
568 934
569#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)") 935#define DEC_INC_DEPTH if (++dec->depth > dec->json.max_depth) ERR (ERR_NESTING_EXCEEDED)
570#define DEC_DEC_DEPTH --dec->depth 936#define DEC_DEC_DEPTH --dec->depth
571 937
572static SV *decode_sv (dec_t *dec); 938static SV *decode_sv (dec_t *dec);
573 939
574static signed char decode_hexdigit[256]; 940static signed char decode_hexdigit[256];
… …
577decode_4hex (dec_t *dec) 943decode_4hex (dec_t *dec)
578{ 944{
579 signed char d1, d2, d3, d4; 945 signed char d1, d2, d3, d4;
580 unsigned char *cur = (unsigned char *)dec->cur; 946 unsigned char *cur = (unsigned char *)dec->cur;
581 947
582 d1 = decode_hexdigit [cur [0]]; if (d1 < 0) ERR ("four hexadecimal digits expected"); 948 d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected");
583 d2 = decode_hexdigit [cur [1]]; if (d2 < 0) ERR ("four hexadecimal digits expected"); 949 d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected");
584 d3 = decode_hexdigit [cur [2]]; if (d3 < 0) ERR ("four hexadecimal digits expected"); 950 d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected");
585 d4 = decode_hexdigit [cur [3]]; if (d4 < 0) ERR ("four hexadecimal digits expected"); 951 d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected");
586 952
587 dec->cur += 4; 953 dec->cur += 4;
588 954
589 return ((UV)d1) << 12 955 return ((UV)d1) << 12
590 | ((UV)d2) << 8 956 | ((UV)d2) << 8
… …
598static SV * 964static SV *
599decode_str (dec_t *dec) 965decode_str (dec_t *dec)
600{ 966{
601 SV *sv = 0; 967 SV *sv = 0;
602 int utf8 = 0; 968 int utf8 = 0;
969 char *dec_cur = dec->cur;
603 970
604 do 971 do
605 { 972 {
606 char buf [SHORT_STRING_LEN + UTF8_MAX_LEN]; 973 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
607 char *cur = buf; 974 char *cur = buf;
608 975
609 do 976 do
610 { 977 {
611 unsigned char ch = *(unsigned char *)dec->cur++; 978 unsigned char ch = *(unsigned char *)dec_cur++;
612 979
613 if (ch == '"') 980 if (expect_false (ch == '"'))
614 { 981 {
615 --dec->cur; 982 --dec_cur;
616 break; 983 break;
617 } 984 }
618 else if (ch == '\\') 985 else if (expect_false (ch == '\\'))
619 { 986 {
620 switch (*dec->cur) 987 switch (*dec_cur)
621 { 988 {
622 case '\\': 989 case '\\':
623 case '/': 990 case '/':
624 case '"': *cur++ = *dec->cur++; break; 991 case '"': *cur++ = *dec_cur++; break;
625 992
626 case 'b': ++dec->cur; *cur++ = '\010'; break; 993 case 'b': ++dec_cur; *cur++ = '\010'; break;
627 case 't': ++dec->cur; *cur++ = '\011'; break; 994 case 't': ++dec_cur; *cur++ = '\011'; break;
628 case 'n': ++dec->cur; *cur++ = '\012'; break; 995 case 'n': ++dec_cur; *cur++ = '\012'; break;
629 case 'f': ++dec->cur; *cur++ = '\014'; break; 996 case 'f': ++dec_cur; *cur++ = '\014'; break;
630 case 'r': ++dec->cur; *cur++ = '\015'; break; 997 case 'r': ++dec_cur; *cur++ = '\015'; break;
631 998
632 case 'u': 999 case 'u':
633 { 1000 {
634 UV lo, hi; 1001 UV lo, hi;
635 ++dec->cur; 1002 ++dec_cur;
636 1003
1004 dec->cur = dec_cur;
637 hi = decode_4hex (dec); 1005 hi = decode_4hex (dec);
1006 dec_cur = dec->cur;
638 if (hi == (UV)-1) 1007 if (hi == (UV)-1)
639 goto fail; 1008 goto fail;
640 1009
641 // possibly a surrogate pair 1010 // possibly a surrogate pair
642 if (hi >= 0xd800) 1011 if (hi >= 0xd800)
643 if (hi < 0xdc00) 1012 if (hi < 0xdc00)
644 { 1013 {
645 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 1014 if (dec_cur [0] != '\\' || dec_cur [1] != 'u')
646 ERR ("missing low surrogate character in surrogate pair"); 1015 ERR ("missing low surrogate character in surrogate pair");
647 1016
648 dec->cur += 2; 1017 dec_cur += 2;
649 1018
1019 dec->cur = dec_cur;
650 lo = decode_4hex (dec); 1020 lo = decode_4hex (dec);
1021 dec_cur = dec->cur;
651 if (lo == (UV)-1) 1022 if (lo == (UV)-1)
652 goto fail; 1023 goto fail;
653 1024
654 if (lo < 0xdc00 || lo >= 0xe000) 1025 if (lo < 0xdc00 || lo >= 0xe000)
655 ERR ("surrogate pair expected"); 1026 ERR ("surrogate pair expected");
… …
661 1032
662 if (hi >= 0x80) 1033 if (hi >= 0x80)
663 { 1034 {
664 utf8 = 1; 1035 utf8 = 1;
665 1036
666 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0); 1037 cur = encode_utf8 (cur, hi);
667 } 1038 }
668 else 1039 else
669 *cur++ = hi; 1040 *cur++ = hi;
670 } 1041 }
671 break; 1042 break;
672 1043
673 default: 1044 default:
674 --dec->cur; 1045 --dec_cur;
675 ERR ("illegal backslash escape sequence in string"); 1046 ERR ("illegal backslash escape sequence in string");
676 } 1047 }
677 } 1048 }
678 else if (ch >= 0x20 && ch <= 0x7f) 1049 else if (expect_true (ch >= 0x20 && ch < 0x80))
679 *cur++ = ch; 1050 *cur++ = ch;
680 else if (ch >= 0x80) 1051 else if (ch >= 0x80)
681 { 1052 {
682 STRLEN clen; 1053 STRLEN clen;
683 UV uch;
684 1054
685 --dec->cur; 1055 --dec_cur;
686 1056
687 uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen); 1057 decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
688 if (clen == (STRLEN)-1) 1058 if (clen == (STRLEN)-1)
689 ERR ("malformed UTF-8 character in JSON string"); 1059 ERR ("malformed UTF-8 character in JSON string");
690 1060
691 do 1061 do
692 *cur++ = *dec->cur++; 1062 *cur++ = *dec_cur++;
693 while (--clen); 1063 while (--clen);
694 1064
695 utf8 = 1; 1065 utf8 = 1;
696 } 1066 }
697 else 1067 else
698 { 1068 {
699 --dec->cur; 1069 --dec_cur;
700 1070
701 if (!ch) 1071 if (!ch)
702 ERR ("unexpected end of string while parsing JSON string"); 1072 ERR ("unexpected end of string while parsing JSON string");
703 else 1073 else
704 ERR ("invalid character encountered while parsing JSON string"); 1074 ERR ("invalid character encountered while parsing JSON string");
… …
709 { 1079 {
710 STRLEN len = cur - buf; 1080 STRLEN len = cur - buf;
711 1081
712 if (sv) 1082 if (sv)
713 { 1083 {
714 SvGROW (sv, SvCUR (sv) + len + 1); 1084 STRLEN cur = SvCUR (sv);
1085
1086 if (SvLEN (sv) <= cur + len)
1087 SvGROW (sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
1088
715 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 1089 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
716 SvCUR_set (sv, SvCUR (sv) + len); 1090 SvCUR_set (sv, SvCUR (sv) + len);
717 } 1091 }
718 else 1092 else
719 sv = newSVpvn (buf, len); 1093 sv = newSVpvn (buf, len);
720 } 1094 }
721 } 1095 }
722 while (*dec->cur != '"'); 1096 while (*dec_cur != '"');
723 1097
724 ++dec->cur; 1098 ++dec_cur;
725 1099
726 if (sv) 1100 if (sv)
727 { 1101 {
728 SvPOK_only (sv); 1102 SvPOK_only (sv);
729 *SvEND (sv) = 0; 1103 *SvEND (sv) = 0;
… …
732 SvUTF8_on (sv); 1106 SvUTF8_on (sv);
733 } 1107 }
734 else 1108 else
735 sv = newSVpvn ("", 0); 1109 sv = newSVpvn ("", 0);
736 1110
1111 dec->cur = dec_cur;
737 return sv; 1112 return sv;
738 1113
739fail: 1114fail:
1115 dec->cur = dec_cur;
740 return 0; 1116 return 0;
741} 1117}
742 1118
743static SV * 1119static SV *
744decode_num (dec_t *dec) 1120decode_num (dec_t *dec)
… …
802 is_nv = 1; 1178 is_nv = 1;
803 } 1179 }
804 1180
805 if (!is_nv) 1181 if (!is_nv)
806 { 1182 {
807 UV uv; 1183 int len = dec->cur - start;
808 int numtype = grok_number (start, dec->cur - start, &uv); 1184
809 if (numtype & IS_NUMBER_IN_UV) 1185 // special case the rather common 1..5-digit-int case
810 if (numtype & IS_NUMBER_NEG) 1186 if (*start == '-')
1187 switch (len)
811 { 1188 {
812 if (uv < (UV)IV_MIN) 1189 case 2: return newSViv (-(IV)( start [1] - '0' * 1));
813 return newSViv (-(IV)uv); 1190 case 3: return newSViv (-(IV)( start [1] * 10 + start [2] - '0' * 11));
1191 case 4: return newSViv (-(IV)( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
1192 case 5: return newSViv (-(IV)( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
1193 case 6: return newSViv (-(IV)(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
814 } 1194 }
1195 else
1196 switch (len)
1197 {
1198 case 1: return newSViv ( start [0] - '0' * 1);
1199 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
1200 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
1201 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
1202 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
1203 }
1204
1205 {
1206 UV uv;
1207 int numtype = grok_number (start, len, &uv);
1208 if (numtype & IS_NUMBER_IN_UV)
1209 if (numtype & IS_NUMBER_NEG)
1210 {
1211 if (uv < (UV)IV_MIN)
1212 return newSViv (-(IV)uv);
1213 }
815 else 1214 else
816 return newSVuv (uv); 1215 return newSVuv (uv);
817 } 1216 }
818 1217
1218 len -= *start == '-' ? 1 : 0;
1219
1220 // does not fit into IV or UV, try NV
1221 if (len <= NV_DIG)
1222 // fits into NV without loss of precision
1223 return newSVnv (json_atof (start));
1224
1225 // everything else fails, convert it to a string
1226 return newSVpvn (start, dec->cur - start);
1227 }
1228
1229 // loss of precision here
819 return newSVnv (Atof (start)); 1230 return newSVnv (json_atof (start));
820 1231
821fail: 1232fail:
822 return 0; 1233 return 0;
823} 1234}
824 1235
… …
853 1264
854 if (*dec->cur != ',') 1265 if (*dec->cur != ',')
855 ERR (", or ] expected while parsing array"); 1266 ERR (", or ] expected while parsing array");
856 1267
857 ++dec->cur; 1268 ++dec->cur;
1269
1270 decode_ws (dec);
1271
1272 if (*dec->cur == ']' && dec->json.flags & F_RELAXED)
1273 {
1274 ++dec->cur;
1275 break;
1276 }
858 } 1277 }
859 1278
860 DEC_DEC_DEPTH; 1279 DEC_DEC_DEPTH;
861 return newRV_noinc ((SV *)av); 1280 return newRV_noinc ((SV *)av);
862 1281
… …
867} 1286}
868 1287
869static SV * 1288static SV *
870decode_hv (dec_t *dec) 1289decode_hv (dec_t *dec)
871{ 1290{
1291 SV *sv;
872 HV *hv = newHV (); 1292 HV *hv = newHV ();
873 1293
874 DEC_INC_DEPTH; 1294 DEC_INC_DEPTH;
875 decode_ws (dec); 1295 decode_ws (dec);
876 1296
877 if (*dec->cur == '}') 1297 if (*dec->cur == '}')
878 ++dec->cur; 1298 ++dec->cur;
879 else 1299 else
880 for (;;) 1300 for (;;)
881 { 1301 {
1302 EXPECT_CH ('"');
1303
1304 // heuristic: assume that
1305 // a) decode_str + hv_store_ent are abysmally slow.
1306 // b) most hash keys are short, simple ascii text.
1307 // => try to "fast-match" such strings to avoid
1308 // the overhead of decode_str + hv_store_ent.
1309 {
882 SV *key, *value; 1310 SV *value;
1311 char *p = dec->cur;
1312 char *e = p + 24; // only try up to 24 bytes
883 1313
884 decode_ws (dec); EXPECT_CH ('"'); 1314 for (;;)
885
886 key = decode_str (dec);
887 if (!key)
888 goto fail;
889
890 decode_ws (dec); EXPECT_CH (':');
891
892 value = decode_sv (dec);
893 if (!value)
894 { 1315 {
1316 // the >= 0x80 is false on most architectures
1317 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1318 {
1319 // slow path, back up and use decode_str
1320 SV *key = decode_str (dec);
1321 if (!key)
1322 goto fail;
1323
1324 decode_ws (dec); EXPECT_CH (':');
1325
1326 decode_ws (dec);
1327 value = decode_sv (dec);
1328 if (!value)
1329 {
1330 SvREFCNT_dec (key);
1331 goto fail;
1332 }
1333
1334 hv_store_ent (hv, key, value, 0);
895 SvREFCNT_dec (key); 1335 SvREFCNT_dec (key);
1336
1337 break;
1338 }
1339 else if (*p == '"')
1340 {
1341 // fast path, got a simple key
1342 char *key = dec->cur;
1343 int len = p - key;
1344 dec->cur = p + 1;
1345
1346 decode_ws (dec); EXPECT_CH (':');
1347
1348 decode_ws (dec);
1349 value = decode_sv (dec);
1350 if (!value)
896 goto fail; 1351 goto fail;
1352
1353 hv_store (hv, key, len, value, 0);
1354
1355 break;
1356 }
1357
1358 ++p;
897 } 1359 }
898 1360 }
899 hv_store_ent (hv, key, value, 0);
900 SvREFCNT_dec (key);
901 1361
902 decode_ws (dec); 1362 decode_ws (dec);
903 1363
904 if (*dec->cur == '}') 1364 if (*dec->cur == '}')
905 { 1365 {
… …
909 1369
910 if (*dec->cur != ',') 1370 if (*dec->cur != ',')
911 ERR (", or } expected while parsing object/hash"); 1371 ERR (", or } expected while parsing object/hash");
912 1372
913 ++dec->cur; 1373 ++dec->cur;
1374
1375 decode_ws (dec);
1376
1377 if (*dec->cur == '}' && dec->json.flags & F_RELAXED)
1378 {
1379 ++dec->cur;
1380 break;
1381 }
914 } 1382 }
915 1383
916 DEC_DEC_DEPTH; 1384 DEC_DEC_DEPTH;
917 return newRV_noinc ((SV *)hv); 1385 sv = newRV_noinc ((SV *)hv);
1386
1387 // check filter callbacks
1388 if (dec->json.flags & F_HOOK)
1389 {
1390 if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
1391 {
1392 HE *cb, *he;
1393
1394 hv_iterinit (hv);
1395 he = hv_iternext (hv);
1396 hv_iterinit (hv);
1397
1398 // the next line creates a mortal sv each time its called.
1399 // might want to optimise this for common cases.
1400 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1401
1402 if (cb)
1403 {
1404 dSP;
1405 int count;
1406
1407 ENTER; SAVETMPS; PUSHMARK (SP);
1408 XPUSHs (HeVAL (he));
1409 sv_2mortal (sv);
1410
1411 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1412
1413 if (count == 1)
1414 {
1415 sv = newSVsv (POPs);
1416 FREETMPS; LEAVE;
1417 return sv;
1418 }
1419
1420 SvREFCNT_inc (sv);
1421 FREETMPS; LEAVE;
1422 }
1423 }
1424
1425 if (dec->json.cb_object)
1426 {
1427 dSP;
1428 int count;
1429
1430 ENTER; SAVETMPS; PUSHMARK (SP);
1431 XPUSHs (sv_2mortal (sv));
1432
1433 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1434
1435 if (count == 1)
1436 {
1437 sv = newSVsv (POPs);
1438 FREETMPS; LEAVE;
1439 return sv;
1440 }
1441
1442 SvREFCNT_inc (sv);
1443 FREETMPS; LEAVE;
1444 }
1445 }
1446
1447 return sv;
918 1448
919fail: 1449fail:
920 SvREFCNT_dec (hv); 1450 SvREFCNT_dec (hv);
921 DEC_DEC_DEPTH; 1451 DEC_DEC_DEPTH;
922 return 0; 1452 return 0;
923} 1453}
924 1454
925static SV * 1455static SV *
926decode_sv (dec_t *dec) 1456decode_sv (dec_t *dec)
927{ 1457{
928 decode_ws (dec); 1458 // the beauty of JSON: you need exactly one character lookahead
1459 // to parse everything.
929 switch (*dec->cur) 1460 switch (*dec->cur)
930 { 1461 {
931 case '"': ++dec->cur; return decode_str (dec); 1462 case '"': ++dec->cur; return decode_str (dec);
932 case '[': ++dec->cur; return decode_av (dec); 1463 case '[': ++dec->cur; return decode_av (dec);
933 case '{': ++dec->cur; return decode_hv (dec); 1464 case '{': ++dec->cur; return decode_hv (dec);
934 1465
935 case '-': 1466 case '-':
936 case '0': case '1': case '2': case '3': case '4': 1467 case '0': case '1': case '2': case '3': case '4':
937 case '5': case '6': case '7': case '8': case '9': 1468 case '5': case '6': case '7': case '8': case '9':
938 return decode_num (dec); 1469 return decode_num (dec);
939 1470
940 case 't': 1471 case 't':
941 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1472 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
942 { 1473 {
943 dec->cur += 4; 1474 dec->cur += 4;
1475#if JSON_SLOW
1476 json_true = get_bool ("JSON::XS::true");
1477#endif
944 return newSViv (1); 1478 return newSVsv (json_true);
945 } 1479 }
946 else 1480 else
947 ERR ("'true' expected"); 1481 ERR ("'true' expected");
948 1482
949 break; 1483 break;
950 1484
951 case 'f': 1485 case 'f':
952 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1486 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
953 { 1487 {
954 dec->cur += 5; 1488 dec->cur += 5;
1489#if JSON_SLOW
1490 json_false = get_bool ("JSON::XS::false");
1491#endif
955 return newSViv (0); 1492 return newSVsv (json_false);
956 } 1493 }
957 else 1494 else
958 ERR ("'false' expected"); 1495 ERR ("'false' expected");
959 1496
960 break; 1497 break;
… …
978fail: 1515fail:
979 return 0; 1516 return 0;
980} 1517}
981 1518
982static SV * 1519static SV *
983decode_json (SV *string, U32 flags) 1520decode_json (SV *string, JSON *json, char **offset_return)
984{ 1521{
985 dec_t dec; 1522 dec_t dec;
986 SV *sv; 1523 SV *sv;
987 1524
1525 /* work around bugs in 5.10 where manipulating magic values
1526 * will perl ignore the magic in subsequent accesses.
1527 * also make a copy of non-PV values, to get them into a clean
1528 * state (SvPV should do that, but it's buggy, see below).
1529 */
1530 /*SvGETMAGIC (string);*/
1531 if (SvMAGICAL (string) || !SvPOK (string))
1532 string = sv_2mortal (newSVsv (string));
1533
988 SvUPGRADE (string, SVt_PV); 1534 SvUPGRADE (string, SVt_PV);
989 1535
990 if (flags & F_UTF8) 1536 /* work around a bug in perl 5.10, which causes SvCUR to fail an
1537 * assertion with -DDEBUGGING, although SvCUR is documented to
1538 * return the xpv_cur field which certainly exists after upgrading.
1539 * according to nicholas clark, calling SvPOK fixes this.
1540 * But it doesn't fix it, so try another workaround, call SvPV_nolen
1541 * and hope for the best.
1542 * Damnit, SvPV_nolen still trips over yet another assertion. This
1543 * assertion business is seriously broken, try yet another workaround
1544 * for the broken -DDEBUGGING.
1545 */
1546 {
1547#ifdef DEBUGGING
1548 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
1549#else
1550 STRLEN offset = SvCUR (string);
1551#endif
1552
1553 if (offset > json->max_size && json->max_size)
1554 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1555 (unsigned long)SvCUR (string), (unsigned long)json->max_size);
1556 }
1557
1558 if (DECODE_WANTS_OCTETS (json))
991 sv_utf8_downgrade (string, 0); 1559 sv_utf8_downgrade (string, 0);
992 else 1560 else
993 sv_utf8_upgrade (string); 1561 sv_utf8_upgrade (string);
994 1562
995 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1563 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
996 1564
997 dec.flags = flags; 1565 dec.json = *json;
998 dec.cur = SvPVX (string); 1566 dec.cur = SvPVX (string);
999 dec.end = SvEND (string); 1567 dec.end = SvEND (string);
1000 dec.err = 0; 1568 dec.err = 0;
1001 dec.depth = 0; 1569 dec.depth = 0;
1002 dec.maxdepth = DEC_DEPTH (dec.flags);
1003 1570
1571 if (dec.json.cb_object || dec.json.cb_sk_object)
1572 dec.json.flags |= F_HOOK;
1573
1004 *dec.end = 0; // this should basically be a nop, too, but make sure its there 1574 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1575
1576 decode_ws (&dec);
1005 sv = decode_sv (&dec); 1577 sv = decode_sv (&dec);
1006 1578
1579 if (offset_return)
1580 *offset_return = dec.cur;
1581
1582 if (!(offset_return || !sv))
1583 {
1584 // check for trailing garbage
1585 decode_ws (&dec);
1586
1587 if (*dec.cur)
1588 {
1589 dec.err = "garbage after JSON object";
1590 SvREFCNT_dec (sv);
1591 sv = 0;
1592 }
1593 }
1594
1007 if (!sv) 1595 if (!sv)
1008 { 1596 {
1009 IV offset = dec.flags & F_UTF8
1010 ? dec.cur - SvPVX (string)
1011 : utf8_distance (dec.cur, SvPVX (string));
1012 SV *uni = sv_newmortal (); 1597 SV *uni = sv_newmortal ();
1013 1598
1014 // horrible hack to silence warning inside pv_uni_display 1599 // horrible hack to silence warning inside pv_uni_display
1015 COP cop = *PL_curcop; 1600 COP cop = *PL_curcop;
1016 cop.cop_warnings = pWARN_NONE; 1601 cop.cop_warnings = pWARN_NONE;
… …
1018 SAVEVPTR (PL_curcop); 1603 SAVEVPTR (PL_curcop);
1019 PL_curcop = &cop; 1604 PL_curcop = &cop;
1020 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1605 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1021 LEAVE; 1606 LEAVE;
1022 1607
1023 croak ("%s, at character offset %d [\"%s\"]", 1608 croak ("%s, at character offset %d (before \"%s\")",
1024 dec.err, 1609 dec.err,
1025 (int)offset, 1610 (int)ptr_to_index (string, dec.cur),
1026 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1611 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1027 } 1612 }
1028 1613
1029 sv = sv_2mortal (sv); 1614 sv = sv_2mortal (sv);
1030 1615
1031 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1616 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))
1032 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); 1617 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1033 1618
1034 return sv; 1619 return sv;
1620}
1621
1622/////////////////////////////////////////////////////////////////////////////
1623// incremental parser
1624
1625static void
1626incr_parse (JSON *self)
1627{
1628 const char *p = SvPVX (self->incr_text) + self->incr_pos;
1629
1630 // the state machine here is a bit convoluted and could be simplified a lot
1631 // but this would make it slower, so...
1632
1633 for (;;)
1634 {
1635 //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
1636 switch (self->incr_mode)
1637 {
1638 // only used for initial whitespace skipping
1639 case INCR_M_WS:
1640 for (;;)
1641 {
1642 if (*p > 0x20)
1643 {
1644 if (*p == '#')
1645 {
1646 self->incr_mode = INCR_M_C0;
1647 goto incr_m_c;
1648 }
1649 else
1650 {
1651 self->incr_mode = INCR_M_JSON;
1652 goto incr_m_json;
1653 }
1654 }
1655 else if (!*p)
1656 goto interrupt;
1657
1658 ++p;
1659 }
1660
1661 // skip a single char inside a string (for \\-processing)
1662 case INCR_M_BS:
1663 if (!*p)
1664 goto interrupt;
1665
1666 ++p;
1667 self->incr_mode = INCR_M_STR;
1668 goto incr_m_str;
1669
1670 // inside #-style comments
1671 case INCR_M_C0:
1672 case INCR_M_C1:
1673 incr_m_c:
1674 for (;;)
1675 {
1676 if (*p == '\n')
1677 {
1678 self->incr_mode = self->incr_mode == INCR_M_C0 ? INCR_M_WS : INCR_M_JSON;
1679 break;
1680 }
1681 else if (!*p)
1682 goto interrupt;
1683
1684 ++p;
1685 }
1686
1687 break;
1688
1689 // inside a string
1690 case INCR_M_STR:
1691 incr_m_str:
1692 for (;;)
1693 {
1694 if (*p == '"')
1695 {
1696 ++p;
1697 self->incr_mode = INCR_M_JSON;
1698
1699 if (!self->incr_nest)
1700 goto interrupt;
1701
1702 goto incr_m_json;
1703 }
1704 else if (*p == '\\')
1705 {
1706 ++p; // "virtually" consumes character after \
1707
1708 if (!*p) // if at end of string we have to switch modes
1709 {
1710 self->incr_mode = INCR_M_BS;
1711 goto interrupt;
1712 }
1713 }
1714 else if (!*p)
1715 goto interrupt;
1716
1717 ++p;
1718 }
1719
1720 // after initial ws, outside string
1721 case INCR_M_JSON:
1722 incr_m_json:
1723 for (;;)
1724 {
1725 switch (*p++)
1726 {
1727 case 0:
1728 --p;
1729 goto interrupt;
1730
1731 case 0x09:
1732 case 0x0a:
1733 case 0x0d:
1734 case 0x20:
1735 if (!self->incr_nest)
1736 {
1737 --p; // do not eat the whitespace, let the next round do it
1738 goto interrupt;
1739 }
1740 break;
1741
1742 case '"':
1743 self->incr_mode = INCR_M_STR;
1744 goto incr_m_str;
1745
1746 case '[':
1747 case '{':
1748 if (++self->incr_nest > self->max_depth)
1749 croak (ERR_NESTING_EXCEEDED);
1750 break;
1751
1752 case ']':
1753 case '}':
1754 if (--self->incr_nest <= 0)
1755 goto interrupt;
1756 break;
1757
1758 case '#':
1759 self->incr_mode = INCR_M_C1;
1760 goto incr_m_c;
1761 }
1762 }
1763 }
1764
1765 modechange:
1766 ;
1767 }
1768
1769interrupt:
1770 self->incr_pos = p - SvPVX (self->incr_text);
1771 //printf ("interrupt<%.*s>\n", self->incr_pos, SvPVX(self->incr_text));//D
1772 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D
1035} 1773}
1036 1774
1037///////////////////////////////////////////////////////////////////////////// 1775/////////////////////////////////////////////////////////////////////////////
1038// XS interface functions 1776// XS interface functions
1039 1777
1040MODULE = JSON::XS PACKAGE = JSON::XS 1778MODULE = JSON::XS PACKAGE = JSON::XS
1041 1779
1042BOOT: 1780BOOT:
1043{ 1781{
1044 int i; 1782 int i;
1045
1046 memset (decode_hexdigit, 0xff, 256);
1047 1783
1048 for (i = 0; i < 256; ++i) 1784 for (i = 0; i < 256; ++i)
1049 decode_hexdigit [i] = 1785 decode_hexdigit [i] =
1050 i >= '0' && i <= '9' ? i - '0' 1786 i >= '0' && i <= '9' ? i - '0'
1051 : i >= 'a' && i <= 'f' ? i - 'a' + 10 1787 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1052 : i >= 'A' && i <= 'F' ? i - 'A' + 10 1788 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1053 : -1; 1789 : -1;
1054 1790
1055 json_stash = gv_stashpv ("JSON::XS", 1); 1791 json_stash = gv_stashpv ("JSON::XS" , 1);
1792 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1793
1794 json_true = get_bool ("JSON::XS::true");
1795 json_false = get_bool ("JSON::XS::false");
1796
1797 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */
1056} 1798}
1057 1799
1058PROTOTYPES: DISABLE 1800PROTOTYPES: DISABLE
1059 1801
1060SV *new (char *dummy) 1802void CLONE (...)
1061 CODE: 1803 CODE:
1062 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1804 json_stash = 0;
1805 json_boolean_stash = 0;
1806
1807void new (char *klass)
1808 PPCODE:
1809{
1810 SV *pv = NEWSV (0, sizeof (JSON));
1811 SvPOK_only (pv);
1812 json_init ((JSON *)SvPVX (pv));
1813 XPUSHs (sv_2mortal (sv_bless (
1814 newRV_noinc (pv),
1815 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1816 )));
1817}
1818
1819void ascii (JSON *self, int enable = 1)
1820 ALIAS:
1821 ascii = F_ASCII
1822 latin1 = F_LATIN1
1823 utf8 = F_UTF8
1824 indent = F_INDENT
1825 canonical = F_CANONICAL
1826 space_before = F_SPACE_BEFORE
1827 space_after = F_SPACE_AFTER
1828 pretty = F_PRETTY
1829 allow_nonref = F_ALLOW_NONREF
1830 shrink = F_SHRINK
1831 allow_blessed = F_ALLOW_BLESSED
1832 convert_blessed = F_CONV_BLESSED
1833 relaxed = F_RELAXED
1834 allow_unknown = F_ALLOW_UNKNOWN
1835 PPCODE:
1836{
1837 if (enable)
1838 self->flags |= ix;
1839 else
1840 self->flags &= ~ix;
1841
1842 XPUSHs (ST (0));
1843}
1844
1845void get_ascii (JSON *self)
1846 ALIAS:
1847 get_ascii = F_ASCII
1848 get_latin1 = F_LATIN1
1849 get_utf8 = F_UTF8
1850 get_indent = F_INDENT
1851 get_canonical = F_CANONICAL
1852 get_space_before = F_SPACE_BEFORE
1853 get_space_after = F_SPACE_AFTER
1854 get_allow_nonref = F_ALLOW_NONREF
1855 get_shrink = F_SHRINK
1856 get_allow_blessed = F_ALLOW_BLESSED
1857 get_convert_blessed = F_CONV_BLESSED
1858 get_relaxed = F_RELAXED
1859 get_allow_unknown = F_ALLOW_UNKNOWN
1860 PPCODE:
1861 XPUSHs (boolSV (self->flags & ix));
1862
1863void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
1864 PPCODE:
1865 self->max_depth = max_depth;
1866 XPUSHs (ST (0));
1867
1868U32 get_max_depth (JSON *self)
1869 CODE:
1870 RETVAL = self->max_depth;
1063 OUTPUT: 1871 OUTPUT:
1064 RETVAL 1872 RETVAL
1065 1873
1066SV *ascii (SV *self, int enable = 1) 1874void max_size (JSON *self, U32 max_size = 0)
1067 ALIAS: 1875 PPCODE:
1068 ascii = F_ASCII 1876 self->max_size = max_size;
1069 utf8 = F_UTF8 1877 XPUSHs (ST (0));
1070 indent = F_INDENT 1878
1071 canonical = F_CANONICAL 1879int get_max_size (JSON *self)
1072 space_before = F_SPACE_BEFORE
1073 space_after = F_SPACE_AFTER
1074 pretty = F_PRETTY
1075 allow_nonref = F_ALLOW_NONREF
1076 shrink = F_SHRINK
1077 CODE: 1880 CODE:
1078{ 1881 RETVAL = self->max_size;
1079 UV *uv = SvJSON (self);
1080 if (enable)
1081 *uv |= ix;
1082 else
1083 *uv &= ~ix;
1084
1085 RETVAL = newSVsv (self);
1086}
1087 OUTPUT: 1882 OUTPUT:
1088 RETVAL 1883 RETVAL
1089 1884
1090SV *max_depth (SV *self, UV max_depth = 0x80000000UL) 1885void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1886 PPCODE:
1887{
1888 SvREFCNT_dec (self->cb_object);
1889 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
1890
1891 XPUSHs (ST (0));
1892}
1893
1894void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
1895 PPCODE:
1896{
1897 if (!self->cb_sk_object)
1898 self->cb_sk_object = newHV ();
1899
1900 if (SvOK (cb))
1901 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1902 else
1903 {
1904 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1905
1906 if (!HvKEYS (self->cb_sk_object))
1907 {
1908 SvREFCNT_dec (self->cb_sk_object);
1909 self->cb_sk_object = 0;
1910 }
1911 }
1912
1913 XPUSHs (ST (0));
1914}
1915
1916void encode (JSON *self, SV *scalar)
1917 PPCODE:
1918 PUTBACK; scalar = encode_json (scalar, self); SPAGAIN;
1919 XPUSHs (scalar);
1920
1921void decode (JSON *self, SV *jsonstr)
1922 PPCODE:
1923 PUTBACK; jsonstr = decode_json (jsonstr, self, 0); SPAGAIN;
1924 XPUSHs (jsonstr);
1925
1926void decode_prefix (JSON *self, SV *jsonstr)
1927 PPCODE:
1928{
1929 SV *sv;
1930 char *offset;
1931 PUTBACK; sv = decode_json (jsonstr, self, &offset); SPAGAIN;
1932 EXTEND (SP, 2);
1933 PUSHs (sv);
1934 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset))));
1935}
1936
1937void incr_parse (JSON *self, SV *jsonstr = 0)
1938 PPCODE:
1939{
1940 if (!self->incr_text)
1941 self->incr_text = newSVpvn ("", 0);
1942
1943 /* if utf8-ness doesn't match the decoder, need to upgrade/downgrade */
1944 if (!DECODE_WANTS_OCTETS (self) == !SvUTF8 (self->incr_text))
1945 if (DECODE_WANTS_OCTETS (self))
1946 {
1947 if (self->incr_pos)
1948 self->incr_pos = utf8_length ((U8 *)SvPVX (self->incr_text),
1949 (U8 *)SvPVX (self->incr_text) + self->incr_pos);
1950
1951 sv_utf8_downgrade (self->incr_text, 0);
1952 }
1953 else
1954 {
1955 sv_utf8_upgrade (self->incr_text);
1956
1957 if (self->incr_pos)
1958 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
1959 - (U8 *)SvPVX (self->incr_text);
1960 }
1961
1962 // append data, if any
1963 if (jsonstr)
1964 {
1965 /* make sure both strings have same encoding */
1966 if (SvUTF8 (jsonstr) != SvUTF8 (self->incr_text))
1967 if (SvUTF8 (jsonstr))
1968 sv_utf8_downgrade (jsonstr, 0);
1969 else
1970 sv_utf8_upgrade (jsonstr);
1971
1972 /* and then just blindly append */
1973 {
1974 STRLEN len;
1975 const char *str = SvPV (jsonstr, len);
1976 STRLEN cur = SvCUR (self->incr_text);
1977
1978 if (SvLEN (self->incr_text) <= cur + len)
1979 SvGROW (self->incr_text, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
1980
1981 Move (str, SvEND (self->incr_text), len, char);
1982 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
1983 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
1984 }
1985 }
1986
1987 if (GIMME_V != G_VOID)
1988 do
1989 {
1990 SV *sv;
1991 char *offset;
1992
1993 if (!INCR_DONE (self))
1994 {
1995 incr_parse (self);
1996
1997 if (self->incr_pos > self->max_size && self->max_size)
1998 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1999 (unsigned long)self->incr_pos, (unsigned long)self->max_size);
2000
2001 if (!INCR_DONE (self))
2002 {
2003 // as an optimisation, do not accumulate white space in the incr buffer
2004 if (self->incr_mode == INCR_M_WS && self->incr_pos)
2005 {
2006 self->incr_pos = 0;
2007 SvCUR_set (self->incr_text, 0);
2008 }
2009
2010 break;
2011 }
2012 }
2013
2014 PUTBACK; sv = decode_json (self->incr_text, self, &offset); SPAGAIN;
2015 XPUSHs (sv);
2016
2017 self->incr_pos -= offset - SvPVX (self->incr_text);
2018 self->incr_nest = 0;
2019 self->incr_mode = 0;
2020
2021 sv_chop (self->incr_text, offset);
2022 }
2023 while (GIMME_V == G_ARRAY);
2024}
2025
2026SV *incr_text (JSON *self)
2027 ATTRS: lvalue
1091 CODE: 2028 CODE:
1092{ 2029{
1093 UV *uv = SvJSON (self); 2030 if (self->incr_pos)
1094 UV log2 = 0; 2031 croak ("incr_text can not be called when the incremental parser already started parsing");
1095 2032
1096 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL; 2033 RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef;
1097
1098 while ((1UL << log2) < max_depth)
1099 ++log2;
1100
1101 *uv = *uv & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1102
1103 RETVAL = newSVsv (self);
1104} 2034}
1105 OUTPUT: 2035 OUTPUT:
1106 RETVAL 2036 RETVAL
1107 2037
2038void incr_skip (JSON *self)
2039 CODE:
2040{
2041 if (self->incr_pos)
2042 {
2043 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + self->incr_pos);
2044 self->incr_pos = 0;
2045 self->incr_nest = 0;
2046 self->incr_mode = 0;
2047 }
2048}
2049
2050void incr_reset (JSON *self)
2051 CODE:
2052{
2053 SvREFCNT_dec (self->incr_text);
2054 self->incr_text = 0;
2055 self->incr_pos = 0;
2056 self->incr_nest = 0;
2057 self->incr_mode = 0;
2058}
2059
2060void DESTROY (JSON *self)
2061 CODE:
2062 SvREFCNT_dec (self->cb_sk_object);
2063 SvREFCNT_dec (self->cb_object);
2064 SvREFCNT_dec (self->incr_text);
2065
2066PROTOTYPES: ENABLE
2067
1108void encode (SV *self, SV *scalar) 2068void encode_json (SV *scalar)
2069 ALIAS:
2070 to_json_ = 0
2071 encode_json = F_UTF8
1109 PPCODE: 2072 PPCODE:
1110 XPUSHs (encode_json (scalar, *SvJSON (self))); 2073{
2074 JSON json;
2075 json_init (&json);
2076 json.flags |= ix;
2077 PUTBACK; scalar = encode_json (scalar, &json); SPAGAIN;
2078 XPUSHs (scalar);
2079}
1111 2080
1112void decode (SV *self, SV *jsonstr) 2081void decode_json (SV *jsonstr)
2082 ALIAS:
2083 from_json_ = 0
2084 decode_json = F_UTF8
1113 PPCODE: 2085 PPCODE:
1114 XPUSHs (decode_json (jsonstr, *SvJSON (self))); 2086{
2087 JSON json;
2088 json_init (&json);
2089 json.flags |= ix;
2090 PUTBACK; jsonstr = decode_json (jsonstr, &json, 0); SPAGAIN;
2091 XPUSHs (jsonstr);
2092}
1115 2093
1116PROTOTYPES: ENABLE
1117
1118void to_json (SV *scalar)
1119 ALIAS:
1120 objToJson = 0
1121 PPCODE:
1122 XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8));
1123
1124void from_json (SV *jsonstr)
1125 ALIAS:
1126 jsonToObj = 0
1127 PPCODE:
1128 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8));
1129

Diff Legend

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