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 1.49 by root, Sun Jul 1 23:40:07 2007 UTC vs.
Revision 1.109 by root, Tue Mar 8 10:18:50 2011 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
14// some old perls do not have this, try to make it work, no 16// some old perls do not have this, try to make it work, no
15// guarentees, though. if it breaks, you get to keep the pieces. 17// guarantees, though. if it breaks, you get to keep the pieces.
16#ifndef UTF8_MAXBYTES 18#ifndef UTF8_MAXBYTES
17# define UTF8_MAXBYTES 13 19# define UTF8_MAXBYTES 13
18#endif 20#endif
21
22// three extra for rounding, sign, and end of string
23#define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 3)
19 24
20#define F_ASCII 0x00000001UL 25#define F_ASCII 0x00000001UL
21#define F_LATIN1 0x00000002UL 26#define F_LATIN1 0x00000002UL
22#define F_UTF8 0x00000004UL 27#define F_UTF8 0x00000004UL
23#define F_INDENT 0x00000008UL 28#define F_INDENT 0x00000008UL
… …
26#define F_SPACE_AFTER 0x00000040UL 31#define F_SPACE_AFTER 0x00000040UL
27#define F_ALLOW_NONREF 0x00000100UL 32#define F_ALLOW_NONREF 0x00000100UL
28#define F_SHRINK 0x00000200UL 33#define F_SHRINK 0x00000200UL
29#define F_ALLOW_BLESSED 0x00000400UL 34#define F_ALLOW_BLESSED 0x00000400UL
30#define F_CONV_BLESSED 0x00000800UL 35#define F_CONV_BLESSED 0x00000800UL
31#define F_MAXDEPTH 0xf8000000UL
32#define S_MAXDEPTH 27
33#define F_MAXSIZE 0x01f00000UL 36#define F_RELAXED 0x00001000UL
34#define S_MAXSIZE 20 37#define F_ALLOW_UNKNOWN 0x00002000UL
35#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing 38#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
36 39
37#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
38#define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE ))
39
40#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 40#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
41#define F_DEFAULT (9UL << S_MAXDEPTH)
42 41
43#define INIT_SIZE 32 // initial scalar size to be allocated 42#define INIT_SIZE 32 // initial scalar size to be allocated
44#define INDENT_STEP 3 // spaces per indentation level 43#define INDENT_STEP 3 // spaces per indentation level
45 44
46#define SHORT_STRING_LEN 16384 // special-case strings of up to this size 45#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
47 46
47#define DECODE_WANTS_OCTETS(json) ((json)->flags & F_UTF8)
48
48#define SB do { 49#define SB do {
49#define SE } while (0) 50#define SE } while (0)
50 51
51#if __GNUC__ >= 3 52#if __GNUC__ >= 3
52# define expect(expr,value) __builtin_expect ((expr),(value)) 53# define expect(expr,value) __builtin_expect ((expr), (value))
53# define inline inline 54# define INLINE static inline
54#else 55#else
55# define expect(expr,value) (expr) 56# define expect(expr,value) (expr)
56# define inline static 57# define INLINE static
57#endif 58#endif
58 59
59#define expect_false(expr) expect ((expr) != 0, 0) 60#define expect_false(expr) expect ((expr) != 0, 0)
60#define expect_true(expr) expect ((expr) != 0, 1) 61#define expect_true(expr) expect ((expr) != 0, 1)
61 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
62static HV *json_stash, *json_boolean_stash; // JSON::XS:: 77static HV *json_stash, *json_boolean_stash; // JSON::XS::
63static SV *json_true, *json_false; 78static SV *json_true, *json_false;
64 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
65typedef struct { 91typedef struct {
66 U32 flags; 92 U32 flags;
67 SV *cb_object, *cb_sk_object; 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;
68} JSON; 104} JSON;
105
106INLINE void
107json_init (JSON *json)
108{
109 Zero (json, 1, JSON);
110 json->max_depth = 512;
111}
69 112
70///////////////////////////////////////////////////////////////////////////// 113/////////////////////////////////////////////////////////////////////////////
71// utility functions 114// utility functions
72 115
73inline void 116INLINE SV *
117get_bool (const char *name)
118{
119 SV *sv = get_sv (name, 1);
120
121 SvREADONLY_on (sv);
122 SvREADONLY_on (SvRV (sv));
123
124 return sv;
125}
126
127INLINE void
74shrink (SV *sv) 128shrink (SV *sv)
75{ 129{
76 sv_utf8_downgrade (sv, 1); 130 sv_utf8_downgrade (sv, 1);
131
77 if (SvLEN (sv) > SvCUR (sv) + 1) 132 if (SvLEN (sv) > SvCUR (sv) + 1)
78 { 133 {
79#ifdef SvPV_shrink_to_cur 134#ifdef SvPV_shrink_to_cur
80 SvPV_shrink_to_cur (sv); 135 SvPV_shrink_to_cur (sv);
81#elif defined (SvPV_renew) 136#elif defined (SvPV_renew)
… …
87// 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
88// case of an error. 143// case of an error.
89// we special-case "safe" characters from U+80 .. U+7FF, 144// we special-case "safe" characters from U+80 .. U+7FF,
90// but use the very good perl function to parse anything else. 145// but use the very good perl function to parse anything else.
91// note that we never call this function for a ascii codepoints 146// note that we never call this function for a ascii codepoints
92inline UV 147INLINE UV
93decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) 148decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
94{ 149{
95 if (expect_false (s[0] > 0xdf || s[0] < 0xc2)) 150 if (expect_true (len >= 2
96 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); 151 && IN_RANGE_INC (char, s[0], 0xc2, 0xdf)
97 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) 152 && IN_RANGE_INC (char, s[1], 0x80, 0xbf)))
98 { 153 {
99 *clen = 2; 154 *clen = 2;
100 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 155 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
101 } 156 }
102 else 157 else
103 { 158 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
104 *clen = (STRLEN)-1; 159}
105 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 (;;)
106 } 211 {
107} 212 U8 dig = (U8)*s - '0';
108 213
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 if (postdp) *expo -= eaccum;
266 *accum += uaccum * Perl_pow (10., *expo);
267 *expo += eaccum;
268}
269
270static NV
271json_atof (const char *s)
272{
273 NV accum = 0.;
274 int expo = 0;
275 int neg = 0;
276
277 if (*s == '-')
278 {
279 ++s;
280 neg = 1;
281 }
282
283 // a recursion depth of ten gives us >>500 bits
284 json_atof_scan1 (s, &accum, &expo, 0, 10);
285
286 return neg ? -accum : accum;
287}
109///////////////////////////////////////////////////////////////////////////// 288/////////////////////////////////////////////////////////////////////////////
110// encoder 289// encoder
111 290
112// structure used for encoding JSON 291// structure used for encoding JSON
113typedef struct 292typedef struct
… …
115 char *cur; // SvPVX (sv) + current output position 294 char *cur; // SvPVX (sv) + current output position
116 char *end; // SvEND (sv) 295 char *end; // SvEND (sv)
117 SV *sv; // result scalar 296 SV *sv; // result scalar
118 JSON json; 297 JSON json;
119 U32 indent; // indentation level 298 U32 indent; // indentation level
120 U32 maxdepth; // max. indentation/recursion level 299 UV limit; // escape character values >= this value when encoding
121} enc_t; 300} enc_t;
122 301
123inline void 302INLINE void
124need (enc_t *enc, STRLEN len) 303need (enc_t *enc, STRLEN len)
125{ 304{
126 if (expect_false (enc->cur + len >= enc->end)) 305 if (expect_false (enc->cur + len >= enc->end))
127 { 306 {
128 STRLEN cur = enc->cur - SvPVX (enc->sv); 307 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
129 SvGROW (enc->sv, cur + len + 1); 308 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
130 enc->cur = SvPVX (enc->sv) + cur; 309 enc->cur = SvPVX (enc->sv) + cur;
131 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 310 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
132 } 311 }
133} 312}
134 313
135inline void 314INLINE void
136encode_ch (enc_t *enc, char ch) 315encode_ch (enc_t *enc, char ch)
137{ 316{
138 need (enc, 1); 317 need (enc, 1);
139 *enc->cur++ = ch; 318 *enc->cur++ = ch;
140} 319}
… …
194 { 373 {
195 uch = ch; 374 uch = ch;
196 clen = 1; 375 clen = 1;
197 } 376 }
198 377
199 if (uch > 0x10FFFFUL) 378 if (uch < 0x80/*0x20*/ || uch >= enc->limit)
200 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
201
202 if (uch < 0x80 || enc->json.flags & F_ASCII || (enc->json.flags & F_LATIN1 && uch > 0xFF))
203 { 379 {
204 if (uch > 0xFFFFUL) 380 if (uch >= 0x10000UL)
205 { 381 {
382 if (uch >= 0x110000UL)
383 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
384
206 need (enc, len += 11); 385 need (enc, len += 11);
207 sprintf (enc->cur, "\\u%04x\\u%04x", 386 sprintf (enc->cur, "\\u%04x\\u%04x",
208 (int)((uch - 0x10000) / 0x400 + 0xD800), 387 (int)((uch - 0x10000) / 0x400 + 0xD800),
209 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 388 (int)((uch - 0x10000) % 0x400 + 0xDC00));
210 enc->cur += 12; 389 enc->cur += 12;
211 } 390 }
212 else 391 else
213 { 392 {
214 static char hexdigit [16] = "0123456789abcdef";
215 need (enc, len += 5); 393 need (enc, len += 5);
216 *enc->cur++ = '\\'; 394 *enc->cur++ = '\\';
217 *enc->cur++ = 'u'; 395 *enc->cur++ = 'u';
218 *enc->cur++ = hexdigit [ uch >> 12 ]; 396 *enc->cur++ = PL_hexdigit [ uch >> 12 ];
219 *enc->cur++ = hexdigit [(uch >> 8) & 15]; 397 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
220 *enc->cur++ = hexdigit [(uch >> 4) & 15]; 398 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
221 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 399 *enc->cur++ = PL_hexdigit [(uch >> 0) & 15];
222 } 400 }
223 401
224 str += clen; 402 str += clen;
225 } 403 }
226 else if (enc->json.flags & F_LATIN1) 404 else if (enc->json.flags & F_LATIN1)
… …
238 while (--clen); 416 while (--clen);
239 } 417 }
240 else 418 else
241 { 419 {
242 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed 420 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
243 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 421 enc->cur = encode_utf8 (enc->cur, uch);
244 ++str; 422 ++str;
245 } 423 }
246 } 424 }
247 } 425 }
248 } 426 }
249 427
250 --len; 428 --len;
251 } 429 }
252} 430}
253 431
254inline void 432INLINE void
255encode_indent (enc_t *enc) 433encode_indent (enc_t *enc)
256{ 434{
257 if (enc->json.flags & F_INDENT) 435 if (enc->json.flags & F_INDENT)
258 { 436 {
259 int spaces = enc->indent * INDENT_STEP; 437 int spaces = enc->indent * INDENT_STEP;
… …
262 memset (enc->cur, ' ', spaces); 440 memset (enc->cur, ' ', spaces);
263 enc->cur += spaces; 441 enc->cur += spaces;
264 } 442 }
265} 443}
266 444
267inline void 445INLINE void
268encode_space (enc_t *enc) 446encode_space (enc_t *enc)
269{ 447{
270 need (enc, 1); 448 need (enc, 1);
271 encode_ch (enc, ' '); 449 encode_ch (enc, ' ');
272} 450}
273 451
274inline void 452INLINE void
275encode_nl (enc_t *enc) 453encode_nl (enc_t *enc)
276{ 454{
277 if (enc->json.flags & F_INDENT) 455 if (enc->json.flags & F_INDENT)
278 { 456 {
279 need (enc, 1); 457 need (enc, 1);
280 encode_ch (enc, '\n'); 458 encode_ch (enc, '\n');
281 } 459 }
282} 460}
283 461
284inline void 462INLINE void
285encode_comma (enc_t *enc) 463encode_comma (enc_t *enc)
286{ 464{
287 encode_ch (enc, ','); 465 encode_ch (enc, ',');
288 466
289 if (enc->json.flags & F_INDENT) 467 if (enc->json.flags & F_INDENT)
… …
297static void 475static void
298encode_av (enc_t *enc, AV *av) 476encode_av (enc_t *enc, AV *av)
299{ 477{
300 int i, len = av_len (av); 478 int i, len = av_len (av);
301 479
302 if (enc->indent >= enc->maxdepth) 480 if (enc->indent >= enc->json.max_depth)
303 croak ("data structure too deep (hit recursion limit)"); 481 croak (ERR_NESTING_EXCEEDED);
304 482
305 encode_ch (enc, '['); encode_nl (enc); 483 encode_ch (enc, '[');
306 ++enc->indent; 484
485 if (len >= 0)
486 {
487 encode_nl (enc); ++enc->indent;
307 488
308 for (i = 0; i <= len; ++i) 489 for (i = 0; i <= len; ++i)
309 { 490 {
491 SV **svp = av_fetch (av, i, 0);
492
310 encode_indent (enc); 493 encode_indent (enc);
311 encode_sv (enc, *av_fetch (av, i, 0));
312 494
495 if (svp)
496 encode_sv (enc, *svp);
497 else
498 encode_str (enc, "null", 4, 0);
499
313 if (i < len) 500 if (i < len)
314 encode_comma (enc); 501 encode_comma (enc);
315 } 502 }
316 503
504 encode_nl (enc); --enc->indent; encode_indent (enc);
505 }
506
317 encode_nl (enc); 507 encode_ch (enc, ']');
318
319 --enc->indent;
320 encode_indent (enc); encode_ch (enc, ']');
321} 508}
322 509
323static void 510static void
324encode_he (enc_t *enc, HE *he) 511encode_hk (enc_t *enc, HE *he)
325{ 512{
326 encode_ch (enc, '"'); 513 encode_ch (enc, '"');
327 514
328 if (HeKLEN (he) == HEf_SVKEY) 515 if (HeKLEN (he) == HEf_SVKEY)
329 { 516 {
… …
342 encode_ch (enc, '"'); 529 encode_ch (enc, '"');
343 530
344 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc); 531 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
345 encode_ch (enc, ':'); 532 encode_ch (enc, ':');
346 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc); 533 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
347 encode_sv (enc, HeVAL (he));
348} 534}
349 535
350// compare hash entries, used when all keys are bytestrings 536// compare hash entries, used when all keys are bytestrings
351static int 537static int
352he_cmp_fast (const void *a_, const void *b_) 538he_cmp_fast (const void *a_, const void *b_)
… …
357 HE *b = *(HE **)b_; 543 HE *b = *(HE **)b_;
358 544
359 STRLEN la = HeKLEN (a); 545 STRLEN la = HeKLEN (a);
360 STRLEN lb = HeKLEN (b); 546 STRLEN lb = HeKLEN (b);
361 547
362 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 548 if (!(cmp = memcmp (HeKEY (b), HeKEY (a), lb < la ? lb : la)))
363 cmp = la - lb; 549 cmp = lb - la;
364 550
365 return cmp; 551 return cmp;
366} 552}
367 553
368// compare hash entries, used when some keys are sv's or utf-x 554// compare hash entries, used when some keys are sv's or utf-x
369static int 555static int
370he_cmp_slow (const void *a, const void *b) 556he_cmp_slow (const void *a, const void *b)
371{ 557{
372 return sv_cmp (HeSVKEY_force (*(HE **)a), HeSVKEY_force (*(HE **)b)); 558 return sv_cmp (HeSVKEY_force (*(HE **)b), HeSVKEY_force (*(HE **)a));
373} 559}
374 560
375static void 561static void
376encode_hv (enc_t *enc, HV *hv) 562encode_hv (enc_t *enc, HV *hv)
377{ 563{
378 int count, i; 564 HE *he;
379 565
380 if (enc->indent >= enc->maxdepth) 566 if (enc->indent >= enc->json.max_depth)
381 croak ("data structure too deep (hit recursion limit)"); 567 croak (ERR_NESTING_EXCEEDED);
382 568
383 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 569 encode_ch (enc, '{');
384 570
385 if ((count = hv_iterinit (hv)))
386 {
387 // for canonical output we have to sort by keys first 571 // for canonical output we have to sort by keys first
388 // actually, this is mostly due to the stupid so-called 572 // actually, this is mostly due to the stupid so-called
389 // security workaround added somewhere in 5.8.x. 573 // security workaround added somewhere in 5.8.x
390 // that randomises hash orderings 574 // that randomises hash orderings
391 if (enc->json.flags & F_CANONICAL) 575 if (enc->json.flags & F_CANONICAL && !SvRMAGICAL (hv))
576 {
577 int count = hv_iterinit (hv);
578
579 if (SvMAGICAL (hv))
392 { 580 {
581 // need to count by iterating. could improve by dynamically building the vector below
582 // but I don't care for the speed of this special case.
583 // note also that we will run into undefined behaviour when the two iterations
584 // do not result in the same count, something I might care for in some later release.
585
586 count = 0;
587 while (hv_iternext (hv))
588 ++count;
589
590 hv_iterinit (hv);
591 }
592
593 if (count)
594 {
393 int fast = 1; 595 int i, fast = 1;
394 HE *he;
395#if defined(__BORLANDC__) || defined(_MSC_VER) 596#if defined(__BORLANDC__) || defined(_MSC_VER)
396 HE **hes = _alloca (count * sizeof (HE)); 597 HE **hes = _alloca (count * sizeof (HE));
397#else 598#else
398 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode 599 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
399#endif 600#endif
… …
426 627
427 FREETMPS; 628 FREETMPS;
428 LEAVE; 629 LEAVE;
429 } 630 }
430 631
431 for (i = 0; i < count; ++i) 632 encode_nl (enc); ++enc->indent;
633
634 while (count--)
432 { 635 {
433 encode_indent (enc); 636 encode_indent (enc);
637 he = hes [count];
434 encode_he (enc, hes [i]); 638 encode_hk (enc, he);
639 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
435 640
436 if (i < count - 1) 641 if (count)
437 encode_comma (enc); 642 encode_comma (enc);
438 } 643 }
439 644
440 encode_nl (enc); 645 encode_nl (enc); --enc->indent; encode_indent (enc);
441 } 646 }
647 }
442 else 648 else
649 {
650 if (hv_iterinit (hv) || SvMAGICAL (hv))
651 if ((he = hv_iternext (hv)))
443 { 652 {
444 HE *he = hv_iternext (hv); 653 encode_nl (enc); ++enc->indent;
445 654
446 for (;;) 655 for (;;)
447 { 656 {
448 encode_indent (enc); 657 encode_indent (enc);
449 encode_he (enc, he); 658 encode_hk (enc, he);
659 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
450 660
451 if (!(he = hv_iternext (hv))) 661 if (!(he = hv_iternext (hv)))
452 break; 662 break;
453 663
454 encode_comma (enc); 664 encode_comma (enc);
455 } 665 }
456 666
457 encode_nl (enc); 667 encode_nl (enc); --enc->indent; encode_indent (enc);
458 } 668 }
459 } 669 }
460 670
461 --enc->indent; encode_indent (enc); encode_ch (enc, '}'); 671 encode_ch (enc, '}');
462} 672}
463 673
464// encode objects, arrays and special \0=false and \1=true values. 674// encode objects, arrays and special \0=false and \1=true values.
465static void 675static void
466encode_rv (enc_t *enc, SV *sv) 676encode_rv (enc_t *enc, SV *sv)
… …
470 SvGETMAGIC (sv); 680 SvGETMAGIC (sv);
471 svt = SvTYPE (sv); 681 svt = SvTYPE (sv);
472 682
473 if (expect_false (SvOBJECT (sv))) 683 if (expect_false (SvOBJECT (sv)))
474 { 684 {
685 HV *stash = !JSON_SLOW || json_boolean_stash
686 ? json_boolean_stash
687 : gv_stashpv ("JSON::XS::Boolean", 1);
688
475 if (SvSTASH (sv) == json_boolean_stash) 689 if (SvSTASH (sv) == stash)
476 { 690 {
477 if (SvIV (sv) == 0) 691 if (SvIV (sv))
692 encode_str (enc, "true", 4, 0);
693 else
478 encode_str (enc, "false", 5, 0); 694 encode_str (enc, "false", 5, 0);
479 else
480 encode_str (enc, "true", 4, 0);
481 } 695 }
482 else 696 else
483 { 697 {
484#if 0 698#if 0
485 if (0 && sv_derived_from (rv, "JSON::Literal")) 699 if (0 && sv_derived_from (rv, "JSON::Literal"))
… …
488 } 702 }
489#endif 703#endif
490 if (enc->json.flags & F_CONV_BLESSED) 704 if (enc->json.flags & F_CONV_BLESSED)
491 { 705 {
492 // we re-bless the reference to get overload and other niceties right 706 // we re-bless the reference to get overload and other niceties right
493 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 1); 707 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
494 708
495 if (to_json) 709 if (to_json)
496 { 710 {
711 dSP;
712
497 dSP; ENTER; SAVETMPS; PUSHMARK (SP); 713 ENTER; SAVETMPS; PUSHMARK (SP);
498 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv))); 714 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
499 715
500 // calling with G_SCALAR ensures that we always get a 1 reutrn value 716 // calling with G_SCALAR ensures that we always get a 1 return value
501 // check anyways.
502 PUTBACK; 717 PUTBACK;
503 assert (1 == call_sv ((SV *)GvCV (to_json), G_SCALAR)); 718 call_sv ((SV *)GvCV (to_json), G_SCALAR);
504 SPAGAIN; 719 SPAGAIN;
505 720
721 // catch this surprisingly common error
722 if (SvROK (TOPs) && SvRV (TOPs) == sv)
723 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
724
725 sv = POPs;
726 PUTBACK;
727
506 encode_sv (enc, POPs); 728 encode_sv (enc, sv);
507 729
508 FREETMPS; LEAVE; 730 FREETMPS; LEAVE;
509 } 731 }
510 else if (enc->json.flags & F_ALLOW_BLESSED) 732 else if (enc->json.flags & F_ALLOW_BLESSED)
511 encode_str (enc, "null", 4, 0); 733 encode_str (enc, "null", 4, 0);
… …
524 encode_hv (enc, (HV *)sv); 746 encode_hv (enc, (HV *)sv);
525 else if (svt == SVt_PVAV) 747 else if (svt == SVt_PVAV)
526 encode_av (enc, (AV *)sv); 748 encode_av (enc, (AV *)sv);
527 else if (svt < SVt_PVAV) 749 else if (svt < SVt_PVAV)
528 { 750 {
529 if (SvNIOK (sv) && SvIV (sv) == 0) 751 STRLEN len = 0;
752 char *pv = svt ? SvPV (sv, len) : 0;
753
754 if (len == 1 && *pv == '1')
755 encode_str (enc, "true", 4, 0);
756 else if (len == 1 && *pv == '0')
530 encode_str (enc, "false", 5, 0); 757 encode_str (enc, "false", 5, 0);
531 else if (SvNIOK (sv) && SvIV (sv) == 1) 758 else if (enc->json.flags & F_ALLOW_UNKNOWN)
532 encode_str (enc, "true", 4, 0); 759 encode_str (enc, "null", 4, 0);
533 else 760 else
534 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", 761 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
535 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 762 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
536 } 763 }
764 else if (enc->json.flags & F_ALLOW_UNKNOWN)
765 encode_str (enc, "null", 4, 0);
537 else 766 else
538 croak ("encountered %s, but JSON can only represent references to arrays or hashes", 767 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
539 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 768 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
540} 769}
541 770
… …
559 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 788 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
560 enc->cur += strlen (enc->cur); 789 enc->cur += strlen (enc->cur);
561 } 790 }
562 else if (SvIOKp (sv)) 791 else if (SvIOKp (sv))
563 { 792 {
564 // we assume we can always read an IV as a UV 793 // we assume we can always read an IV as a UV and vice versa
565 if (SvUV (sv) & ~(UV)0x7fff) 794 // we assume two's complement
566 { 795 // we assume no aliasing issues in the union
567 // large integer, use the (rather slow) snprintf way. 796 if (SvIsUV (sv) ? SvUVX (sv) <= 59000
568 need (enc, sizeof (UV) * 3); 797 : SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000)
569 enc->cur +=
570 SvIsUV(sv)
571 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
572 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
573 }
574 else
575 { 798 {
576 // optimise the "small number case" 799 // optimise the "small number case"
577 // code will likely be branchless and use only a single multiplication 800 // code will likely be branchless and use only a single multiplication
801 // works for numbers up to 59074
578 I32 i = SvIV (sv); 802 I32 i = SvIVX (sv);
579 U32 u; 803 U32 u;
580 char digit, nz = 0; 804 char digit, nz = 0;
581 805
582 need (enc, 6); 806 need (enc, 6);
583 807
… …
589 813
590 // now output digit by digit, each time masking out the integer part 814 // now output digit by digit, each time masking out the integer part
591 // and multiplying by 5 while moving the decimal point one to the right, 815 // and multiplying by 5 while moving the decimal point one to the right,
592 // resulting in a net multiplication by 10. 816 // resulting in a net multiplication by 10.
593 // we always write the digit to memory but conditionally increment 817 // we always write the digit to memory but conditionally increment
594 // the pointer, to ease the usage of conditional move instructions. 818 // the pointer, to enable the use of conditional move instructions.
595 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5; 819 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffffUL) * 5;
596 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5; 820 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffffUL) * 5;
597 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5; 821 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffffUL) * 5;
598 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5; 822 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffffUL) * 5;
599 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0' 823 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
600 } 824 }
825 else
826 {
827 // large integer, use the (rather slow) snprintf way.
828 need (enc, IVUV_MAXCHARS);
829 enc->cur +=
830 SvIsUV(sv)
831 ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))
832 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
833 }
601 } 834 }
602 else if (SvROK (sv)) 835 else if (SvROK (sv))
603 encode_rv (enc, SvRV (sv)); 836 encode_rv (enc, SvRV (sv));
604 else if (!SvOK (sv)) 837 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN)
605 encode_str (enc, "null", 4, 0); 838 encode_str (enc, "null", 4, 0);
606 else 839 else
607 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 840 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
608 SvPV_nolen (sv), SvFLAGS (sv)); 841 SvPV_nolen (sv), SvFLAGS (sv));
609} 842}
… …
619 enc.json = *json; 852 enc.json = *json;
620 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 853 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
621 enc.cur = SvPVX (enc.sv); 854 enc.cur = SvPVX (enc.sv);
622 enc.end = SvEND (enc.sv); 855 enc.end = SvEND (enc.sv);
623 enc.indent = 0; 856 enc.indent = 0;
624 enc.maxdepth = DEC_DEPTH (enc.json.flags); 857 enc.limit = enc.json.flags & F_ASCII ? 0x000080UL
858 : enc.json.flags & F_LATIN1 ? 0x000100UL
859 : 0x110000UL;
625 860
626 SvPOK_only (enc.sv); 861 SvPOK_only (enc.sv);
627 encode_sv (&enc, scalar); 862 encode_sv (&enc, scalar);
863 encode_nl (&enc);
628 864
629 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 865 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
630 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 866 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
631 867
632 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8))) 868 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
… …
650 JSON json; 886 JSON json;
651 U32 depth; // recursion depth 887 U32 depth; // recursion depth
652 U32 maxdepth; // recursion depth limit 888 U32 maxdepth; // recursion depth limit
653} dec_t; 889} dec_t;
654 890
655inline void 891INLINE void
892decode_comment (dec_t *dec)
893{
894 // only '#'-style comments allowed a.t.m.
895
896 while (*dec->cur && *dec->cur != 0x0a && *dec->cur != 0x0d)
897 ++dec->cur;
898}
899
900INLINE void
656decode_ws (dec_t *dec) 901decode_ws (dec_t *dec)
657{ 902{
658 for (;;) 903 for (;;)
659 { 904 {
660 char ch = *dec->cur; 905 char ch = *dec->cur;
661 906
662 if (ch > 0x20 907 if (ch > 0x20)
908 {
909 if (expect_false (ch == '#'))
910 {
911 if (dec->json.flags & F_RELAXED)
912 decode_comment (dec);
913 else
914 break;
915 }
916 else
917 break;
918 }
663 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) 919 else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)
664 break; 920 break; // parse error, but let higher level handle it, gives better error messages
665 921
666 ++dec->cur; 922 ++dec->cur;
667 } 923 }
668} 924}
669 925
… …
673 if (*dec->cur != ch) \ 929 if (*dec->cur != ch) \
674 ERR (# ch " expected"); \ 930 ERR (# ch " expected"); \
675 ++dec->cur; \ 931 ++dec->cur; \
676 SE 932 SE
677 933
678#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)") 934#define DEC_INC_DEPTH if (++dec->depth > dec->json.max_depth) ERR (ERR_NESTING_EXCEEDED)
679#define DEC_DEC_DEPTH --dec->depth 935#define DEC_DEC_DEPTH --dec->depth
680 936
681static SV *decode_sv (dec_t *dec); 937static SV *decode_sv (dec_t *dec);
682 938
683static signed char decode_hexdigit[256]; 939static signed char decode_hexdigit[256];
… …
775 1031
776 if (hi >= 0x80) 1032 if (hi >= 0x80)
777 { 1033 {
778 utf8 = 1; 1034 utf8 = 1;
779 1035
780 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0); 1036 cur = encode_utf8 (cur, hi);
781 } 1037 }
782 else 1038 else
783 *cur++ = hi; 1039 *cur++ = hi;
784 } 1040 }
785 break; 1041 break;
… …
787 default: 1043 default:
788 --dec_cur; 1044 --dec_cur;
789 ERR ("illegal backslash escape sequence in string"); 1045 ERR ("illegal backslash escape sequence in string");
790 } 1046 }
791 } 1047 }
792 else if (expect_true (ch >= 0x20 && ch <= 0x7f)) 1048 else if (expect_true (ch >= 0x20 && ch < 0x80))
793 *cur++ = ch; 1049 *cur++ = ch;
794 else if (ch >= 0x80) 1050 else if (ch >= 0x80)
795 { 1051 {
796 STRLEN clen; 1052 STRLEN clen;
797 UV uch;
798 1053
799 --dec_cur; 1054 --dec_cur;
800 1055
801 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen); 1056 decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
802 if (clen == (STRLEN)-1) 1057 if (clen == (STRLEN)-1)
803 ERR ("malformed UTF-8 character in JSON string"); 1058 ERR ("malformed UTF-8 character in JSON string");
804 1059
805 do 1060 do
806 *cur++ = *dec_cur++; 1061 *cur++ = *dec_cur++;
… …
823 { 1078 {
824 STRLEN len = cur - buf; 1079 STRLEN len = cur - buf;
825 1080
826 if (sv) 1081 if (sv)
827 { 1082 {
828 SvGROW (sv, SvCUR (sv) + len + 1); 1083 STRLEN cur = SvCUR (sv);
1084
1085 if (SvLEN (sv) <= cur + len)
1086 SvGROW (sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
1087
829 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 1088 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
830 SvCUR_set (sv, SvCUR (sv) + len); 1089 SvCUR_set (sv, SvCUR (sv) + len);
831 } 1090 }
832 else 1091 else
833 sv = newSVpvn (buf, len); 1092 sv = newSVpvn (buf, len);
… …
918 is_nv = 1; 1177 is_nv = 1;
919 } 1178 }
920 1179
921 if (!is_nv) 1180 if (!is_nv)
922 { 1181 {
1182 int len = dec->cur - start;
1183
923 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so 1184 // special case the rather common 1..5-digit-int case
924 if (*start == '-') 1185 if (*start == '-')
925 switch (dec->cur - start) 1186 switch (len)
926 { 1187 {
927 case 2: return newSViv (-( start [1] - '0' * 1)); 1188 case 2: return newSViv (-(IV)( start [1] - '0' * 1));
928 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); 1189 case 3: return newSViv (-(IV)( start [1] * 10 + start [2] - '0' * 11));
929 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); 1190 case 4: return newSViv (-(IV)( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
930 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); 1191 case 5: return newSViv (-(IV)( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
1192 case 6: return newSViv (-(IV)(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
931 } 1193 }
932 else 1194 else
933 switch (dec->cur - start) 1195 switch (len)
934 { 1196 {
935 case 1: return newSViv ( start [0] - '0' * 1); 1197 case 1: return newSViv ( start [0] - '0' * 1);
936 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); 1198 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
937 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); 1199 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
938 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); 1200 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
1201 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
939 } 1202 }
940 1203
941 { 1204 {
942 UV uv; 1205 UV uv;
943 int numtype = grok_number (start, dec->cur - start, &uv); 1206 int numtype = grok_number (start, len, &uv);
944 if (numtype & IS_NUMBER_IN_UV) 1207 if (numtype & IS_NUMBER_IN_UV)
945 if (numtype & IS_NUMBER_NEG) 1208 if (numtype & IS_NUMBER_NEG)
946 { 1209 {
947 if (uv < (UV)IV_MIN) 1210 if (uv < (UV)IV_MIN)
948 return newSViv (-(IV)uv); 1211 return newSViv (-(IV)uv);
949 } 1212 }
950 else 1213 else
951 return newSVuv (uv); 1214 return newSVuv (uv);
952
953 // here would likely be the place for bigint support
954 } 1215 }
955 }
956 1216
957 // if we ever support bigint or bigfloat, this is the place for bigfloat 1217 len -= *start == '-' ? 1 : 0;
1218
1219 // does not fit into IV or UV, try NV
1220 if (len <= NV_DIG)
1221 // fits into NV without loss of precision
1222 return newSVnv (json_atof (start));
1223
1224 // everything else fails, convert it to a string
1225 return newSVpvn (start, dec->cur - start);
1226 }
1227
1228 // loss of precision here
958 return newSVnv (Atof (start)); 1229 return newSVnv (json_atof (start));
959 1230
960fail: 1231fail:
961 return 0; 1232 return 0;
962} 1233}
963 1234
… …
992 1263
993 if (*dec->cur != ',') 1264 if (*dec->cur != ',')
994 ERR (", or ] expected while parsing array"); 1265 ERR (", or ] expected while parsing array");
995 1266
996 ++dec->cur; 1267 ++dec->cur;
1268
1269 decode_ws (dec);
1270
1271 if (*dec->cur == ']' && dec->json.flags & F_RELAXED)
1272 {
1273 ++dec->cur;
1274 break;
1275 }
997 } 1276 }
998 1277
999 DEC_DEC_DEPTH; 1278 DEC_DEC_DEPTH;
1000 return newRV_noinc ((SV *)av); 1279 return newRV_noinc ((SV *)av);
1001 1280
… …
1017 if (*dec->cur == '}') 1296 if (*dec->cur == '}')
1018 ++dec->cur; 1297 ++dec->cur;
1019 else 1298 else
1020 for (;;) 1299 for (;;)
1021 { 1300 {
1022 decode_ws (dec); EXPECT_CH ('"'); 1301 EXPECT_CH ('"');
1023 1302
1024 // heuristic: assume that 1303 // heuristic: assume that
1025 // a) decode_str + hv_store_ent are abysmally slow. 1304 // a) decode_str + hv_store_ent are abysmally slow.
1026 // b) most hash keys are short, simple ascii text. 1305 // b) most hash keys are short, simple ascii text.
1027 // => try to "fast-match" such strings to avoid 1306 // => try to "fast-match" such strings to avoid
… …
1031 char *p = dec->cur; 1310 char *p = dec->cur;
1032 char *e = p + 24; // only try up to 24 bytes 1311 char *e = p + 24; // only try up to 24 bytes
1033 1312
1034 for (;;) 1313 for (;;)
1035 { 1314 {
1036 // the >= 0x80 is true on most architectures 1315 // the >= 0x80 is false on most architectures
1037 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\') 1316 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1038 { 1317 {
1039 // slow path, back up and use decode_str 1318 // slow path, back up and use decode_str
1040 SV *key = decode_str (dec); 1319 SV *key = decode_str (dec);
1041 if (!key) 1320 if (!key)
1042 goto fail; 1321 goto fail;
1043 1322
1044 decode_ws (dec); EXPECT_CH (':'); 1323 decode_ws (dec); EXPECT_CH (':');
1045 1324
1325 decode_ws (dec);
1046 value = decode_sv (dec); 1326 value = decode_sv (dec);
1047 if (!value) 1327 if (!value)
1048 { 1328 {
1049 SvREFCNT_dec (key); 1329 SvREFCNT_dec (key);
1050 goto fail; 1330 goto fail;
… …
1062 int len = p - key; 1342 int len = p - key;
1063 dec->cur = p + 1; 1343 dec->cur = p + 1;
1064 1344
1065 decode_ws (dec); EXPECT_CH (':'); 1345 decode_ws (dec); EXPECT_CH (':');
1066 1346
1347 decode_ws (dec);
1067 value = decode_sv (dec); 1348 value = decode_sv (dec);
1068 if (!value) 1349 if (!value)
1069 goto fail; 1350 goto fail;
1070 1351
1071 hv_store (hv, key, len, value, 0); 1352 hv_store (hv, key, len, value, 0);
… …
1087 1368
1088 if (*dec->cur != ',') 1369 if (*dec->cur != ',')
1089 ERR (", or } expected while parsing object/hash"); 1370 ERR (", or } expected while parsing object/hash");
1090 1371
1091 ++dec->cur; 1372 ++dec->cur;
1373
1374 decode_ws (dec);
1375
1376 if (*dec->cur == '}' && dec->json.flags & F_RELAXED)
1377 {
1378 ++dec->cur;
1379 break;
1380 }
1092 } 1381 }
1093 1382
1094 DEC_DEC_DEPTH; 1383 DEC_DEC_DEPTH;
1095 sv = newRV_noinc ((SV *)hv); 1384 sv = newRV_noinc ((SV *)hv);
1096 1385
1097 // check filter callbacks 1386 // check filter callbacks
1098 if (dec->json.flags & F_HOOK) 1387 if (dec->json.flags & F_HOOK)
1099 { 1388 {
1100 ENTER; SAVETMPS; 1389 if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
1101
1102 if (HvKEYS (hv) == 1 && dec->json.cb_sk_object)
1103 { 1390 {
1391 HE *cb, *he;
1392
1393 hv_iterinit (hv);
1394 he = hv_iternext (hv);
1395 hv_iterinit (hv);
1396
1397 // the next line creates a mortal sv each time its called.
1398 // might want to optimise this for common cases.
1399 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1400
1401 if (cb)
1402 {
1403 dSP;
1104 int count; 1404 int count;
1105 1405
1106 dSP; PUSHMARK (SP); 1406 ENTER; SAVETMPS; PUSHMARK (SP);
1407 XPUSHs (HeVAL (he));
1107 XPUSHs (sv_2mortal (sv)); 1408 sv_2mortal (sv);
1108 1409
1109 PUTBACK; count = call_sv (dec->json.cb_sk_object, G_ARRAY); SPAGAIN; 1410 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1110 1411
1111 if (count == 1) 1412 if (count == 1)
1413 {
1112 sv = newSVsv (POPs); 1414 sv = newSVsv (POPs);
1113 else 1415 FREETMPS; LEAVE;
1416 return sv;
1417 }
1418
1114 SvREFCNT_inc (sv); 1419 SvREFCNT_inc (sv);
1420 FREETMPS; LEAVE;
1421 }
1115 } 1422 }
1116 1423
1117 if (dec->json.cb_object) 1424 if (dec->json.cb_object)
1118 { 1425 {
1426 dSP;
1119 int count; 1427 int count;
1120 1428
1121 dSP; ENTER; SAVETMPS; PUSHMARK (SP); 1429 ENTER; SAVETMPS; PUSHMARK (SP);
1122 XPUSHs (sv_2mortal (sv)); 1430 XPUSHs (sv_2mortal (sv));
1123 1431
1124 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN; 1432 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1125 1433
1126 if (count == 1) 1434 if (count == 1)
1435 {
1127 sv = newSVsv (POPs); 1436 sv = newSVsv (POPs);
1128 else 1437 FREETMPS; LEAVE;
1438 return sv;
1439 }
1440
1129 SvREFCNT_inc (sv); 1441 SvREFCNT_inc (sv);
1442 FREETMPS; LEAVE;
1130 } 1443 }
1131
1132 FREETMPS; LEAVE;
1133 } 1444 }
1134 1445
1135 return newRV_noinc ((SV *)hv); 1446 return sv;
1136 1447
1137fail: 1448fail:
1138 SvREFCNT_dec (hv); 1449 SvREFCNT_dec (hv);
1139 DEC_DEC_DEPTH; 1450 DEC_DEC_DEPTH;
1140 return 0; 1451 return 0;
1141} 1452}
1142 1453
1143static SV * 1454static SV *
1144decode_sv (dec_t *dec) 1455decode_sv (dec_t *dec)
1145{ 1456{
1146 decode_ws (dec);
1147
1148 // the beauty of JSON: you need exactly one character lookahead 1457 // the beauty of JSON: you need exactly one character lookahead
1149 // to parse anything. 1458 // to parse everything.
1150 switch (*dec->cur) 1459 switch (*dec->cur)
1151 { 1460 {
1152 case '"': ++dec->cur; return decode_str (dec); 1461 case '"': ++dec->cur; return decode_str (dec);
1153 case '[': ++dec->cur; return decode_av (dec); 1462 case '[': ++dec->cur; return decode_av (dec);
1154 case '{': ++dec->cur; return decode_hv (dec); 1463 case '{': ++dec->cur; return decode_hv (dec);
1155 1464
1156 case '-': 1465 case '-':
1157 case '0': case '1': case '2': case '3': case '4': 1466 case '0': case '1': case '2': case '3': case '4':
1158 case '5': case '6': case '7': case '8': case '9': 1467 case '5': case '6': case '7': case '8': case '9':
1159 return decode_num (dec); 1468 return decode_num (dec);
1160 1469
1161 case 't': 1470 case 't':
1162 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1471 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1163 { 1472 {
1164 dec->cur += 4; 1473 dec->cur += 4;
1474#if JSON_SLOW
1475 json_true = get_bool ("JSON::XS::true");
1476#endif
1165 return SvREFCNT_inc (json_true); 1477 return newSVsv (json_true);
1166 } 1478 }
1167 else 1479 else
1168 ERR ("'true' expected"); 1480 ERR ("'true' expected");
1169 1481
1170 break; 1482 break;
1171 1483
1172 case 'f': 1484 case 'f':
1173 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1485 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1174 { 1486 {
1175 dec->cur += 5; 1487 dec->cur += 5;
1488#if JSON_SLOW
1489 json_false = get_bool ("JSON::XS::false");
1490#endif
1176 return SvREFCNT_inc (json_false); 1491 return newSVsv (json_false);
1177 } 1492 }
1178 else 1493 else
1179 ERR ("'false' expected"); 1494 ERR ("'false' expected");
1180 1495
1181 break; 1496 break;
… …
1199fail: 1514fail:
1200 return 0; 1515 return 0;
1201} 1516}
1202 1517
1203static SV * 1518static SV *
1204decode_json (SV *string, JSON *json, UV *offset_return) 1519decode_json (SV *string, JSON *json, char **offset_return)
1205{ 1520{
1206 dec_t dec; 1521 dec_t dec;
1207 UV offset;
1208 SV *sv; 1522 SV *sv;
1209 1523
1524 /* work around bugs in 5.10 where manipulating magic values
1525 * will perl ignore the magic in subsequent accesses.
1526 * also make a copy of non-PV values, to get them into a clean
1527 * state (SvPV should do that, but it's buggy, see below).
1528 */
1210 SvGETMAGIC (string); 1529 /*SvGETMAGIC (string);*/
1530 if (SvMAGICAL (string) || !SvPOK (string))
1531 string = sv_2mortal (newSVsv (string));
1532
1211 SvUPGRADE (string, SVt_PV); 1533 SvUPGRADE (string, SVt_PV);
1212 1534
1213 if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags)) 1535 /* work around a bug in perl 5.10, which causes SvCUR to fail an
1536 * assertion with -DDEBUGGING, although SvCUR is documented to
1537 * return the xpv_cur field which certainly exists after upgrading.
1538 * according to nicholas clark, calling SvPOK fixes this.
1539 * But it doesn't fix it, so try another workaround, call SvPV_nolen
1540 * and hope for the best.
1541 * Damnit, SvPV_nolen still trips over yet another assertion. This
1542 * assertion business is seriously broken, try yet another workaround
1543 * for the broken -DDEBUGGING.
1544 */
1545 {
1546#ifdef DEBUGGING
1547 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
1548#else
1549 STRLEN offset = SvCUR (string);
1550#endif
1551
1552 if (offset > json->max_size && json->max_size)
1214 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", 1553 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1215 (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags)); 1554 (unsigned long)SvCUR (string), (unsigned long)json->max_size);
1555 }
1216 1556
1217 if (json->flags & F_UTF8) 1557 if (DECODE_WANTS_OCTETS (json))
1218 sv_utf8_downgrade (string, 0); 1558 sv_utf8_downgrade (string, 0);
1219 else 1559 else
1220 sv_utf8_upgrade (string); 1560 sv_utf8_upgrade (string);
1221 1561
1222 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1562 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
1223 1563
1224 dec.json = *json; 1564 dec.json = *json;
1225 dec.cur = SvPVX (string); 1565 dec.cur = SvPVX (string);
1226 dec.end = SvEND (string); 1566 dec.end = SvEND (string);
1227 dec.err = 0; 1567 dec.err = 0;
1228 dec.depth = 0; 1568 dec.depth = 0;
1229 dec.maxdepth = DEC_DEPTH (dec.json.flags);
1230 1569
1231 if (dec.json.cb_object || dec.json.cb_sk_object) 1570 if (dec.json.cb_object || dec.json.cb_sk_object)
1232 dec.json.flags |= F_HOOK; 1571 dec.json.flags |= F_HOOK;
1233 1572
1234 *dec.end = 0; // this should basically be a nop, too, but make sure it's there 1573 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1574
1575 decode_ws (&dec);
1235 sv = decode_sv (&dec); 1576 sv = decode_sv (&dec);
1577
1578 if (offset_return)
1579 *offset_return = dec.cur;
1236 1580
1237 if (!(offset_return || !sv)) 1581 if (!(offset_return || !sv))
1238 { 1582 {
1239 // check for trailing garbage 1583 // check for trailing garbage
1240 decode_ws (&dec); 1584 decode_ws (&dec);
… …
1243 { 1587 {
1244 dec.err = "garbage after JSON object"; 1588 dec.err = "garbage after JSON object";
1245 SvREFCNT_dec (sv); 1589 SvREFCNT_dec (sv);
1246 sv = 0; 1590 sv = 0;
1247 } 1591 }
1248 }
1249
1250 if (offset_return || !sv)
1251 {
1252 offset = dec.json.flags & F_UTF8
1253 ? dec.cur - SvPVX (string)
1254 : utf8_distance (dec.cur, SvPVX (string));
1255
1256 if (offset_return)
1257 *offset_return = offset;
1258 } 1592 }
1259 1593
1260 if (!sv) 1594 if (!sv)
1261 { 1595 {
1262 SV *uni = sv_newmortal (); 1596 SV *uni = sv_newmortal ();
… …
1268 SAVEVPTR (PL_curcop); 1602 SAVEVPTR (PL_curcop);
1269 PL_curcop = &cop; 1603 PL_curcop = &cop;
1270 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1604 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1271 LEAVE; 1605 LEAVE;
1272 1606
1273 croak ("%s, at character offset %d [\"%s\"]", 1607 croak ("%s, at character offset %d (before \"%s\")",
1274 dec.err, 1608 dec.err,
1275 (int)offset, 1609 ptr_to_index (string, dec.cur),
1276 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1610 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1277 } 1611 }
1278 1612
1279 sv = sv_2mortal (sv); 1613 sv = sv_2mortal (sv);
1280 1614
1281 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1615 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))
1282 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); 1616 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1283 1617
1284 return sv; 1618 return sv;
1619}
1620
1621/////////////////////////////////////////////////////////////////////////////
1622// incremental parser
1623
1624static void
1625incr_parse (JSON *self)
1626{
1627 const char *p = SvPVX (self->incr_text) + self->incr_pos;
1628
1629 // the state machine here is a bit convoluted and could be simplified a lot
1630 // but this would make it slower, so...
1631
1632 for (;;)
1633 {
1634 //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
1635 switch (self->incr_mode)
1636 {
1637 // only used for initial whitespace skipping
1638 case INCR_M_WS:
1639 for (;;)
1640 {
1641 if (*p > 0x20)
1642 {
1643 if (*p == '#')
1644 {
1645 self->incr_mode = INCR_M_C0;
1646 goto incr_m_c;
1647 }
1648 else
1649 {
1650 self->incr_mode = INCR_M_JSON;
1651 goto incr_m_json;
1652 }
1653 }
1654 else if (!*p)
1655 goto interrupt;
1656
1657 ++p;
1658 }
1659
1660 // skip a single char inside a string (for \\-processing)
1661 case INCR_M_BS:
1662 if (!*p)
1663 goto interrupt;
1664
1665 ++p;
1666 self->incr_mode = INCR_M_STR;
1667 goto incr_m_str;
1668
1669 // inside #-style comments
1670 case INCR_M_C0:
1671 case INCR_M_C1:
1672 incr_m_c:
1673 for (;;)
1674 {
1675 if (*p == '\n')
1676 {
1677 self->incr_mode = self->incr_mode == INCR_M_C0 ? INCR_M_WS : INCR_M_JSON;
1678 break;
1679 }
1680 else if (!*p)
1681 goto interrupt;
1682
1683 ++p;
1684 }
1685
1686 break;
1687
1688 // inside a string
1689 case INCR_M_STR:
1690 incr_m_str:
1691 for (;;)
1692 {
1693 if (*p == '"')
1694 {
1695 ++p;
1696 self->incr_mode = INCR_M_JSON;
1697
1698 if (!self->incr_nest)
1699 goto interrupt;
1700
1701 goto incr_m_json;
1702 }
1703 else if (*p == '\\')
1704 {
1705 ++p; // "virtually" consumes character after \
1706
1707 if (!*p) // if at end of string we have to switch modes
1708 {
1709 self->incr_mode = INCR_M_BS;
1710 goto interrupt;
1711 }
1712 }
1713 else if (!*p)
1714 goto interrupt;
1715
1716 ++p;
1717 }
1718
1719 // after initial ws, outside string
1720 case INCR_M_JSON:
1721 incr_m_json:
1722 for (;;)
1723 {
1724 switch (*p++)
1725 {
1726 case 0:
1727 --p;
1728 goto interrupt;
1729
1730 case 0x09:
1731 case 0x0a:
1732 case 0x0d:
1733 case 0x20:
1734 if (!self->incr_nest)
1735 {
1736 --p; // do not eat the whitespace, let the next round do it
1737 goto interrupt;
1738 }
1739 break;
1740
1741 case '"':
1742 self->incr_mode = INCR_M_STR;
1743 goto incr_m_str;
1744
1745 case '[':
1746 case '{':
1747 if (++self->incr_nest > self->max_depth)
1748 croak (ERR_NESTING_EXCEEDED);
1749 break;
1750
1751 case ']':
1752 case '}':
1753 if (--self->incr_nest <= 0)
1754 goto interrupt;
1755 break;
1756
1757 case '#':
1758 self->incr_mode = INCR_M_C1;
1759 goto incr_m_c;
1760 }
1761 }
1762 }
1763
1764 modechange:
1765 ;
1766 }
1767
1768interrupt:
1769 self->incr_pos = p - SvPVX (self->incr_text);
1770 //printf ("interrupt<%.*s>\n", self->incr_pos, SvPVX(self->incr_text));//D
1771 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D
1285} 1772}
1286 1773
1287///////////////////////////////////////////////////////////////////////////// 1774/////////////////////////////////////////////////////////////////////////////
1288// XS interface functions 1775// XS interface functions
1289 1776
… …
1301 : -1; 1788 : -1;
1302 1789
1303 json_stash = gv_stashpv ("JSON::XS" , 1); 1790 json_stash = gv_stashpv ("JSON::XS" , 1);
1304 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); 1791 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1305 1792
1306 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true ); 1793 json_true = get_bool ("JSON::XS::true");
1307 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); 1794 json_false = get_bool ("JSON::XS::false");
1795
1796 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */
1308} 1797}
1309 1798
1310PROTOTYPES: DISABLE 1799PROTOTYPES: DISABLE
1800
1801void CLONE (...)
1802 CODE:
1803 json_stash = 0;
1804 json_boolean_stash = 0;
1311 1805
1312void new (char *klass) 1806void new (char *klass)
1313 PPCODE: 1807 PPCODE:
1314{ 1808{
1315 SV *pv = NEWSV (0, sizeof (JSON)); 1809 SV *pv = NEWSV (0, sizeof (JSON));
1316 SvPOK_only (pv); 1810 SvPOK_only (pv);
1317 Zero (SvPVX (pv), 1, JSON); 1811 json_init ((JSON *)SvPVX (pv));
1318 ((JSON *)SvPVX (pv))->flags = F_DEFAULT; 1812 XPUSHs (sv_2mortal (sv_bless (
1319 XPUSHs (sv_2mortal (sv_bless (newRV_noinc (pv), json_stash))); 1813 newRV_noinc (pv),
1814 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1815 )));
1320} 1816}
1321 1817
1322void ascii (JSON *self, int enable = 1) 1818void ascii (JSON *self, int enable = 1)
1323 ALIAS: 1819 ALIAS:
1324 ascii = F_ASCII 1820 ascii = F_ASCII
… …
1331 pretty = F_PRETTY 1827 pretty = F_PRETTY
1332 allow_nonref = F_ALLOW_NONREF 1828 allow_nonref = F_ALLOW_NONREF
1333 shrink = F_SHRINK 1829 shrink = F_SHRINK
1334 allow_blessed = F_ALLOW_BLESSED 1830 allow_blessed = F_ALLOW_BLESSED
1335 convert_blessed = F_CONV_BLESSED 1831 convert_blessed = F_CONV_BLESSED
1832 relaxed = F_RELAXED
1833 allow_unknown = F_ALLOW_UNKNOWN
1336 PPCODE: 1834 PPCODE:
1337{ 1835{
1338 if (enable) 1836 if (enable)
1339 self->flags |= ix; 1837 self->flags |= ix;
1340 else 1838 else
1341 self->flags &= ~ix; 1839 self->flags &= ~ix;
1342 1840
1343 XPUSHs (ST (0)); 1841 XPUSHs (ST (0));
1344} 1842}
1345 1843
1346void max_depth (JSON *self, UV max_depth = 0x80000000UL) 1844void get_ascii (JSON *self)
1845 ALIAS:
1846 get_ascii = F_ASCII
1847 get_latin1 = F_LATIN1
1848 get_utf8 = F_UTF8
1849 get_indent = F_INDENT
1850 get_canonical = F_CANONICAL
1851 get_space_before = F_SPACE_BEFORE
1852 get_space_after = F_SPACE_AFTER
1853 get_allow_nonref = F_ALLOW_NONREF
1854 get_shrink = F_SHRINK
1855 get_allow_blessed = F_ALLOW_BLESSED
1856 get_convert_blessed = F_CONV_BLESSED
1857 get_relaxed = F_RELAXED
1858 get_allow_unknown = F_ALLOW_UNKNOWN
1347 PPCODE: 1859 PPCODE:
1348{ 1860 XPUSHs (boolSV (self->flags & ix));
1349 UV log2 = 0;
1350 1861
1351 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL; 1862void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
1352 1863 PPCODE:
1353 while ((1UL << log2) < max_depth) 1864 self->max_depth = max_depth;
1354 ++log2;
1355
1356 self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1357
1358 XPUSHs (ST (0)); 1865 XPUSHs (ST (0));
1359}
1360 1866
1867U32 get_max_depth (JSON *self)
1868 CODE:
1869 RETVAL = self->max_depth;
1870 OUTPUT:
1871 RETVAL
1872
1361void max_size (JSON *self, UV max_size = 0) 1873void max_size (JSON *self, U32 max_size = 0)
1362 PPCODE: 1874 PPCODE:
1363{ 1875 self->max_size = max_size;
1364 UV log2 = 0;
1365
1366 if (max_size > 0x80000000UL) max_size = 0x80000000UL;
1367 if (max_size == 1) max_size = 2;
1368
1369 while ((1UL << log2) < max_size)
1370 ++log2;
1371
1372 self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1373
1374 XPUSHs (ST (0)); 1876 XPUSHs (ST (0));
1375}
1376 1877
1878int get_max_size (JSON *self)
1879 CODE:
1880 RETVAL = self->max_size;
1881 OUTPUT:
1882 RETVAL
1883
1377void filter_json_objects (JSON *self, SV *cb = &PL_sv_undef) 1884void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1378 ALIAS:
1379 filter_sk_json_objects = 1
1380 PPCODE: 1885 PPCODE:
1381{ 1886{
1887 SvREFCNT_dec (self->cb_object);
1888 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
1889
1890 XPUSHs (ST (0));
1891}
1892
1893void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
1894 PPCODE:
1895{
1896 if (!self->cb_sk_object)
1897 self->cb_sk_object = newHV ();
1898
1382 if (!SvOK (cb)) 1899 if (SvOK (cb))
1383 cb = 0; 1900 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1384 1901 else
1385 switch (ix)
1386 { 1902 {
1387 case 0: self->cb_object = cb; break; 1903 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1904
1905 if (!HvKEYS (self->cb_sk_object))
1906 {
1907 SvREFCNT_dec (self->cb_sk_object);
1388 case 1: self->cb_sk_object = cb; break; 1908 self->cb_sk_object = 0;
1909 }
1389 } 1910 }
1390 1911
1391 XPUSHs (ST (0)); 1912 XPUSHs (ST (0));
1392} 1913}
1393 1914
… …
1400 XPUSHs (decode_json (jsonstr, self, 0)); 1921 XPUSHs (decode_json (jsonstr, self, 0));
1401 1922
1402void decode_prefix (JSON *self, SV *jsonstr) 1923void decode_prefix (JSON *self, SV *jsonstr)
1403 PPCODE: 1924 PPCODE:
1404{ 1925{
1405 UV offset; 1926 char *offset;
1406 EXTEND (SP, 2); 1927 EXTEND (SP, 2);
1407 PUSHs (decode_json (jsonstr, self, &offset)); 1928 PUSHs (decode_json (jsonstr, self, &offset));
1408 PUSHs (sv_2mortal (newSVuv (offset))); 1929 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset))));
1409} 1930}
1931
1932void incr_parse (JSON *self, SV *jsonstr = 0)
1933 PPCODE:
1934{
1935 if (!self->incr_text)
1936 self->incr_text = newSVpvn ("", 0);
1937
1938 /* if utf8-ness doesn't match the decoder, need to upgrade/downgrade */
1939 if (!DECODE_WANTS_OCTETS (self) == !SvUTF8 (self->incr_text))
1940 if (DECODE_WANTS_OCTETS (self))
1941 {
1942 if (self->incr_pos)
1943 self->incr_pos = utf8_length ((U8 *)SvPVX (self->incr_text),
1944 (U8 *)SvPVX (self->incr_text) + self->incr_pos);
1945
1946 sv_utf8_downgrade (self->incr_text, 0);
1947 }
1948 else
1949 {
1950 sv_utf8_upgrade (self->incr_text);
1951
1952 if (self->incr_pos)
1953 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
1954 - (U8 *)SvPVX (self->incr_text);
1955 }
1956
1957 // append data, if any
1958 if (jsonstr)
1959 {
1960 /* make sure both strings have same encoding */
1961 if (SvUTF8 (jsonstr) != SvUTF8 (self->incr_text))
1962 if (SvUTF8 (jsonstr))
1963 sv_utf8_downgrade (jsonstr, 0);
1964 else
1965 sv_utf8_upgrade (jsonstr);
1966
1967 /* and then just blindly append */
1968 {
1969 STRLEN len;
1970 const char *str = SvPV (jsonstr, len);
1971 STRLEN cur = SvCUR (self->incr_text);
1972
1973 if (SvLEN (self->incr_text) <= cur + len)
1974 SvGROW (self->incr_text, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
1975
1976 Move (str, SvEND (self->incr_text), len, char);
1977 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
1978 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
1979 }
1980 }
1981
1982 if (GIMME_V != G_VOID)
1983 do
1984 {
1985 char *offset;
1986
1987 if (!INCR_DONE (self))
1988 {
1989 incr_parse (self);
1990
1991 if (self->incr_pos > self->max_size && self->max_size)
1992 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1993 (unsigned long)self->incr_pos, (unsigned long)self->max_size);
1994
1995 if (!INCR_DONE (self))
1996 break;
1997 }
1998
1999 XPUSHs (decode_json (self->incr_text, self, &offset));
2000
2001 self->incr_pos -= offset - SvPVX (self->incr_text);
2002 self->incr_nest = 0;
2003 self->incr_mode = 0;
2004
2005 sv_chop (self->incr_text, offset);
2006 }
2007 while (GIMME_V == G_ARRAY);
2008}
2009
2010SV *incr_text (JSON *self)
2011 ATTRS: lvalue
2012 CODE:
2013{
2014 if (self->incr_pos)
2015 croak ("incr_text can not be called when the incremental parser already started parsing");
2016
2017 RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef;
2018}
2019 OUTPUT:
2020 RETVAL
2021
2022void incr_skip (JSON *self)
2023 CODE:
2024{
2025 if (self->incr_pos)
2026 {
2027 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + self->incr_pos);
2028 self->incr_pos = 0;
2029 self->incr_nest = 0;
2030 self->incr_mode = 0;
2031 }
2032}
2033
2034void incr_reset (JSON *self)
2035 CODE:
2036{
2037 SvREFCNT_dec (self->incr_text);
2038 self->incr_text = 0;
2039 self->incr_pos = 0;
2040 self->incr_nest = 0;
2041 self->incr_mode = 0;
2042}
2043
2044void DESTROY (JSON *self)
2045 CODE:
2046 SvREFCNT_dec (self->cb_sk_object);
2047 SvREFCNT_dec (self->cb_object);
2048 SvREFCNT_dec (self->incr_text);
1410 2049
1411PROTOTYPES: ENABLE 2050PROTOTYPES: ENABLE
1412 2051
1413void to_json (SV *scalar) 2052void encode_json (SV *scalar)
2053 ALIAS:
2054 to_json_ = 0
2055 encode_json = F_UTF8
1414 PPCODE: 2056 PPCODE:
1415{ 2057{
1416 JSON json = { F_DEFAULT | F_UTF8 }; 2058 JSON json;
2059 json_init (&json);
2060 json.flags |= ix;
1417 XPUSHs (encode_json (scalar, &json)); 2061 XPUSHs (encode_json (scalar, &json));
1418} 2062}
1419 2063
1420void from_json (SV *jsonstr) 2064void decode_json (SV *jsonstr)
2065 ALIAS:
2066 from_json_ = 0
2067 decode_json = F_UTF8
1421 PPCODE: 2068 PPCODE:
1422{ 2069{
1423 JSON json = { F_DEFAULT | F_UTF8 }; 2070 JSON json;
2071 json_init (&json);
2072 json.flags |= ix;
1424 XPUSHs (decode_json (jsonstr, &json, 0)); 2073 XPUSHs (decode_json (jsonstr, &json, 0));
1425} 2074}
1426 2075

Diff Legend

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