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 1.39 by root, Mon Jun 11 03:42:57 2007 UTC vs.
Revision 1.106 by root, Tue Jan 19 01:36:34 2010 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
15#define F_LATIN1 0x00000002UL 26#define F_LATIN1 0x00000002UL
16#define F_UTF8 0x00000004UL 27#define F_UTF8 0x00000004UL
17#define F_INDENT 0x00000008UL 28#define F_INDENT 0x00000008UL
18#define F_CANONICAL 0x00000010UL 29#define F_CANONICAL 0x00000010UL
19#define F_SPACE_BEFORE 0x00000020UL 30#define F_SPACE_BEFORE 0x00000020UL
20#define F_SPACE_AFTER 0x00000040UL 31#define F_SPACE_AFTER 0x00000040UL
21#define F_ALLOW_NONREF 0x00000100UL 32#define F_ALLOW_NONREF 0x00000100UL
22#define F_SHRINK 0x00000200UL 33#define F_SHRINK 0x00000200UL
34#define F_ALLOW_BLESSED 0x00000400UL
35#define F_CONV_BLESSED 0x00000800UL
23#define F_MAXDEPTH 0xf8000000UL 36#define F_RELAXED 0x00001000UL
24#define S_MAXDEPTH 27 37#define F_ALLOW_UNKNOWN 0x00002000UL
25 38#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
26#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
27
28// F_SELFCONVERT? <=> to_json/toJson
29// F_BLESSED? <=> { $__class__$ => }
30 39
31#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 40#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
32#define F_DEFAULT (9UL << S_MAXDEPTH)
33 41
34#define INIT_SIZE 32 // initial scalar size to be allocated 42#define INIT_SIZE 32 // initial scalar size to be allocated
35#define INDENT_STEP 3 // spaces per indentation level 43#define INDENT_STEP 3 // spaces per indentation level
36 44
37#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
38 46
39#define SB do { 47#define SB do {
40#define SE } while (0) 48#define SE } while (0)
41 49
42#if __GNUC__ >= 3 50#if __GNUC__ >= 3
43# define expect(expr,value) __builtin_expect ((expr),(value)) 51# define expect(expr,value) __builtin_expect ((expr), (value))
44# define inline inline 52# define INLINE static inline
45#else 53#else
46# define expect(expr,value) (expr) 54# define expect(expr,value) (expr)
47# define inline static 55# define INLINE static
48#endif 56#endif
49 57
50#define expect_false(expr) expect ((expr) != 0, 0) 58#define expect_false(expr) expect ((expr) != 0, 0)
51#define expect_true(expr) expect ((expr) != 0, 1) 59#define expect_true(expr) expect ((expr) != 0, 1)
52 60
61#define IN_RANGE_INC(type,val,beg,end) \
62 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \
63 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
64
65#define ERR_NESTING_EXCEEDED "json text or perl structure exceeds maximum nesting level (max_depth set too low?)"
66
67#ifdef USE_ITHREADS
68# define JSON_SLOW 1
69# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1))
70#else
71# define JSON_SLOW 0
72# define JSON_STASH json_stash
73#endif
74
53static HV *json_stash; // JSON::XS:: 75static HV *json_stash, *json_boolean_stash; // JSON::XS::
76static SV *json_true, *json_false;
77
78enum {
79 INCR_M_WS = 0, // initial whitespace skipping, must be 0
80 INCR_M_STR, // inside string
81 INCR_M_BS, // inside backslash
82 INCR_M_C0, // inside comment in initial whitespace sequence
83 INCR_M_C1, // inside comment in other places
84 INCR_M_JSON // outside anything, count nesting
85};
86
87#define INCR_DONE(json) ((json)->incr_nest <= 0 && (json)->incr_mode == INCR_M_JSON)
88
89typedef struct {
90 U32 flags;
91 U32 max_depth;
92 STRLEN max_size;
93
94 SV *cb_object;
95 HV *cb_sk_object;
96
97 // for the incremental parser
98 SV *incr_text; // the source text so far
99 STRLEN incr_pos; // the current offset into the text
100 int incr_nest; // {[]}-nesting level
101 unsigned char incr_mode;
102} JSON;
103
104INLINE void
105json_init (JSON *json)
106{
107 Zero (json, 1, JSON);
108 json->max_depth = 512;
109}
54 110
55///////////////////////////////////////////////////////////////////////////// 111/////////////////////////////////////////////////////////////////////////////
56// utility functions 112// utility functions
57 113
58static UV * 114INLINE SV *
59SvJSON (SV *sv) 115get_bool (const char *name)
60{ 116{
61 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash)) 117 SV *sv = get_sv (name, 1);
62 croak ("object is not of type JSON::XS");
63 118
64 return &SvUVX (SvRV (sv)); 119 SvREADONLY_on (sv);
65} 120 SvREADONLY_on (SvRV (sv));
66 121
67static void 122 return sv;
123}
124
125INLINE void
68shrink (SV *sv) 126shrink (SV *sv)
69{ 127{
70 sv_utf8_downgrade (sv, 1); 128 sv_utf8_downgrade (sv, 1);
129
71 if (SvLEN (sv) > SvCUR (sv) + 1) 130 if (SvLEN (sv) > SvCUR (sv) + 1)
72 { 131 {
73#ifdef SvPV_shrink_to_cur 132#ifdef SvPV_shrink_to_cur
74 SvPV_shrink_to_cur (sv); 133 SvPV_shrink_to_cur (sv);
75#elif defined (SvPV_renew) 134#elif defined (SvPV_renew)
… …
81// decode an utf-8 character and return it, or (UV)-1 in 140// decode an utf-8 character and return it, or (UV)-1 in
82// case of an error. 141// case of an error.
83// we special-case "safe" characters from U+80 .. U+7FF, 142// we special-case "safe" characters from U+80 .. U+7FF,
84// but use the very good perl function to parse anything else. 143// but use the very good perl function to parse anything else.
85// note that we never call this function for a ascii codepoints 144// note that we never call this function for a ascii codepoints
86inline UV 145INLINE UV
87decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) 146decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
88{ 147{
89 if (expect_false (s[0] > 0xdf || s[0] < 0xc2)) 148 if (expect_true (len >= 2
90 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); 149 && IN_RANGE_INC (char, s[0], 0xc2, 0xdf)
91 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) 150 && IN_RANGE_INC (char, s[1], 0x80, 0xbf)))
92 { 151 {
93 *clen = 2; 152 *clen = 2;
94 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 153 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
95 } 154 }
96 else 155 else
97 { 156 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
98 *clen = (STRLEN)-1; 157}
99 return (UV)-1; 158
159// likewise for encoding, also never called for ascii codepoints
160// this function takes advantage of this fact, although current gccs
161// seem to optimise the check for >= 0x80 away anyways
162INLINE unsigned char *
163encode_utf8 (unsigned char *s, UV ch)
164{
165 if (expect_false (ch < 0x000080))
166 *s++ = ch;
167 else if (expect_true (ch < 0x000800))
168 *s++ = 0xc0 | ( ch >> 6),
169 *s++ = 0x80 | ( ch & 0x3f);
170 else if ( ch < 0x010000)
171 *s++ = 0xe0 | ( ch >> 12),
172 *s++ = 0x80 | ((ch >> 6) & 0x3f),
173 *s++ = 0x80 | ( ch & 0x3f);
174 else if ( ch < 0x110000)
175 *s++ = 0xf0 | ( ch >> 18),
176 *s++ = 0x80 | ((ch >> 12) & 0x3f),
177 *s++ = 0x80 | ((ch >> 6) & 0x3f),
178 *s++ = 0x80 | ( ch & 0x3f);
179
180 return s;
181}
182
183// convert offset pointer to character index, sv must be string
184static STRLEN
185ptr_to_index (SV *sv, char *offset)
186{
187 return SvUTF8 (sv)
188 ? utf8_distance (offset, SvPVX (sv))
189 : offset - SvPVX (sv);
190}
191
192/////////////////////////////////////////////////////////////////////////////
193// fp hell
194
195// scan a group of digits, and a trailing exponent
196static void
197json_atof_scan1 (const char *s, NV *accum, int *expo, int postdp, int maxdepth)
198{
199 UV uaccum = 0;
200 int eaccum = 0;
201
202 // if we recurse too deep, skip all remaining digits
203 // to avoid a stack overflow attack
204 if (expect_false (--maxdepth <= 0))
205 while (((U8)*s - '0') < 10)
206 ++s;
207
208 for (;;)
100 } 209 {
101} 210 U8 dig = (U8)*s - '0';
102 211
212 if (expect_false (dig >= 10))
213 {
214 if (dig == (U8)((U8)'.' - (U8)'0'))
215 {
216 ++s;
217 json_atof_scan1 (s, accum, expo, 1, maxdepth);
218 }
219 else if ((dig | ' ') == 'e' - '0')
220 {
221 int exp2 = 0;
222 int neg = 0;
223
224 ++s;
225
226 if (*s == '-')
227 {
228 ++s;
229 neg = 1;
230 }
231 else if (*s == '+')
232 ++s;
233
234 while ((dig = (U8)*s - '0') < 10)
235 exp2 = exp2 * 10 + *s++ - '0';
236
237 *expo += neg ? -exp2 : exp2;
238 }
239
240 break;
241 }
242
243 ++s;
244
245 uaccum = uaccum * 10 + dig;
246 ++eaccum;
247
248 // if we have too many digits, then recurse for more
249 // we actually do this for rather few digits
250 if (uaccum >= (UV_MAX - 9) / 10)
251 {
252 if (postdp) *expo -= eaccum;
253 json_atof_scan1 (s, accum, expo, postdp, maxdepth);
254 if (postdp) *expo += eaccum;
255
256 break;
257 }
258 }
259
260 // this relies greatly on the quality of the pow ()
261 // implementation of the platform, but a good
262 // implementation is hard to beat.
263 if (postdp) *expo -= eaccum;
264 *accum += uaccum * Perl_pow (10., *expo);
265 *expo += eaccum;
266}
267
268static NV
269json_atof (const char *s)
270{
271 NV accum = 0.;
272 int expo = 0;
273 int neg = 0;
274
275 if (*s == '-')
276 {
277 ++s;
278 neg = 1;
279 }
280
281 // a recursion depth of ten gives us >>500 bits
282 json_atof_scan1 (s, &accum, &expo, 0, 10);
283
284 return neg ? -accum : accum;
285}
103///////////////////////////////////////////////////////////////////////////// 286/////////////////////////////////////////////////////////////////////////////
104// encoder 287// encoder
105 288
106// structure used for encoding JSON 289// structure used for encoding JSON
107typedef struct 290typedef struct
108{ 291{
109 char *cur; // SvPVX (sv) + current output position 292 char *cur; // SvPVX (sv) + current output position
110 char *end; // SvEND (sv) 293 char *end; // SvEND (sv)
111 SV *sv; // result scalar 294 SV *sv; // result scalar
112 U32 flags; // F_* 295 JSON json;
113 U32 indent; // indentation level 296 U32 indent; // indentation level
114 U32 maxdepth; // max. indentation/recursion level 297 UV limit; // escape character values >= this value when encoding
115} enc_t; 298} enc_t;
116 299
117inline void 300INLINE void
118need (enc_t *enc, STRLEN len) 301need (enc_t *enc, STRLEN len)
119{ 302{
120 if (expect_false (enc->cur + len >= enc->end)) 303 if (expect_false (enc->cur + len >= enc->end))
121 { 304 {
122 STRLEN cur = enc->cur - SvPVX (enc->sv); 305 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
123 SvGROW (enc->sv, cur + len + 1); 306 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
124 enc->cur = SvPVX (enc->sv) + cur; 307 enc->cur = SvPVX (enc->sv) + cur;
125 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 308 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
126 } 309 }
127} 310}
128 311
129inline void 312INLINE void
130encode_ch (enc_t *enc, char ch) 313encode_ch (enc_t *enc, char ch)
131{ 314{
132 need (enc, 1); 315 need (enc, 1);
133 *enc->cur++ = ch; 316 *enc->cur++ = ch;
134} 317}
… …
178 STRLEN clen; 361 STRLEN clen;
179 UV uch; 362 UV uch;
180 363
181 if (is_utf8) 364 if (is_utf8)
182 { 365 {
183 //uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
184 uch = decode_utf8 (str, end - str, &clen); 366 uch = decode_utf8 (str, end - str, &clen);
185 if (clen == (STRLEN)-1) 367 if (clen == (STRLEN)-1)
186 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str); 368 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
187 } 369 }
188 else 370 else
189 { 371 {
190 uch = ch; 372 uch = ch;
191 clen = 1; 373 clen = 1;
192 } 374 }
193 375
194 if (uch > 0x10FFFFUL) 376 if (uch < 0x80/*0x20*/ || uch >= enc->limit)
195 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
196
197 if (uch < 0x80 || enc->flags & F_ASCII || (enc->flags & F_LATIN1 && uch > 0xFF))
198 { 377 {
199 if (uch > 0xFFFFUL) 378 if (uch >= 0x10000UL)
200 { 379 {
380 if (uch >= 0x110000UL)
381 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
382
201 need (enc, len += 11); 383 need (enc, len += 11);
202 sprintf (enc->cur, "\\u%04x\\u%04x", 384 sprintf (enc->cur, "\\u%04x\\u%04x",
203 (int)((uch - 0x10000) / 0x400 + 0xD800), 385 (int)((uch - 0x10000) / 0x400 + 0xD800),
204 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 386 (int)((uch - 0x10000) % 0x400 + 0xDC00));
205 enc->cur += 12; 387 enc->cur += 12;
206 } 388 }
207 else 389 else
208 { 390 {
209 static char hexdigit [16] = "0123456789abcdef";
210 need (enc, len += 5); 391 need (enc, len += 5);
211 *enc->cur++ = '\\'; 392 *enc->cur++ = '\\';
212 *enc->cur++ = 'u'; 393 *enc->cur++ = 'u';
213 *enc->cur++ = hexdigit [ uch >> 12 ]; 394 *enc->cur++ = PL_hexdigit [ uch >> 12 ];
214 *enc->cur++ = hexdigit [(uch >> 8) & 15]; 395 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
215 *enc->cur++ = hexdigit [(uch >> 4) & 15]; 396 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
216 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 397 *enc->cur++ = PL_hexdigit [(uch >> 0) & 15];
217 } 398 }
218 399
219 str += clen; 400 str += clen;
220 } 401 }
221 else if (enc->flags & F_LATIN1) 402 else if (enc->json.flags & F_LATIN1)
222 { 403 {
223 *enc->cur++ = uch; 404 *enc->cur++ = uch;
224 str += clen; 405 str += clen;
225 } 406 }
226 else if (is_utf8) 407 else if (is_utf8)
… …
233 while (--clen); 414 while (--clen);
234 } 415 }
235 else 416 else
236 { 417 {
237 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed 418 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
238 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 419 enc->cur = encode_utf8 (enc->cur, uch);
239 ++str; 420 ++str;
240 } 421 }
241 } 422 }
242 } 423 }
243 } 424 }
244 425
245 --len; 426 --len;
246 } 427 }
247} 428}
248 429
249inline void 430INLINE void
250encode_indent (enc_t *enc) 431encode_indent (enc_t *enc)
251{ 432{
252 if (enc->flags & F_INDENT) 433 if (enc->json.flags & F_INDENT)
253 { 434 {
254 int spaces = enc->indent * INDENT_STEP; 435 int spaces = enc->indent * INDENT_STEP;
255 436
256 need (enc, spaces); 437 need (enc, spaces);
257 memset (enc->cur, ' ', spaces); 438 memset (enc->cur, ' ', spaces);
258 enc->cur += spaces; 439 enc->cur += spaces;
259 } 440 }
260} 441}
261 442
262inline void 443INLINE void
263encode_space (enc_t *enc) 444encode_space (enc_t *enc)
264{ 445{
265 need (enc, 1); 446 need (enc, 1);
266 encode_ch (enc, ' '); 447 encode_ch (enc, ' ');
267} 448}
268 449
269inline void 450INLINE void
270encode_nl (enc_t *enc) 451encode_nl (enc_t *enc)
271{ 452{
272 if (enc->flags & F_INDENT) 453 if (enc->json.flags & F_INDENT)
273 { 454 {
274 need (enc, 1); 455 need (enc, 1);
275 encode_ch (enc, '\n'); 456 encode_ch (enc, '\n');
276 } 457 }
277} 458}
278 459
279inline void 460INLINE void
280encode_comma (enc_t *enc) 461encode_comma (enc_t *enc)
281{ 462{
282 encode_ch (enc, ','); 463 encode_ch (enc, ',');
283 464
284 if (enc->flags & F_INDENT) 465 if (enc->json.flags & F_INDENT)
285 encode_nl (enc); 466 encode_nl (enc);
286 else if (enc->flags & F_SPACE_AFTER) 467 else if (enc->json.flags & F_SPACE_AFTER)
287 encode_space (enc); 468 encode_space (enc);
288} 469}
289 470
290static void encode_sv (enc_t *enc, SV *sv); 471static void encode_sv (enc_t *enc, SV *sv);
291 472
292static void 473static void
293encode_av (enc_t *enc, AV *av) 474encode_av (enc_t *enc, AV *av)
294{ 475{
295 int i, len = av_len (av); 476 int i, len = av_len (av);
296 477
297 if (enc->indent >= enc->maxdepth) 478 if (enc->indent >= enc->json.max_depth)
298 croak ("data structure too deep (hit recursion limit)"); 479 croak (ERR_NESTING_EXCEEDED);
299 480
300 encode_ch (enc, '['); encode_nl (enc); 481 encode_ch (enc, '[');
301 ++enc->indent; 482
483 if (len >= 0)
484 {
485 encode_nl (enc); ++enc->indent;
302 486
303 for (i = 0; i <= len; ++i) 487 for (i = 0; i <= len; ++i)
304 { 488 {
489 SV **svp = av_fetch (av, i, 0);
490
305 encode_indent (enc); 491 encode_indent (enc);
306 encode_sv (enc, *av_fetch (av, i, 0));
307 492
493 if (svp)
494 encode_sv (enc, *svp);
495 else
496 encode_str (enc, "null", 4, 0);
497
308 if (i < len) 498 if (i < len)
309 encode_comma (enc); 499 encode_comma (enc);
310 } 500 }
311 501
502 encode_nl (enc); --enc->indent; encode_indent (enc);
503 }
504
312 encode_nl (enc); 505 encode_ch (enc, ']');
313
314 --enc->indent;
315 encode_indent (enc); encode_ch (enc, ']');
316} 506}
317 507
318static void 508static void
319encode_he (enc_t *enc, HE *he) 509encode_hk (enc_t *enc, HE *he)
320{ 510{
321 encode_ch (enc, '"'); 511 encode_ch (enc, '"');
322 512
323 if (HeKLEN (he) == HEf_SVKEY) 513 if (HeKLEN (he) == HEf_SVKEY)
324 { 514 {
… …
334 else 524 else
335 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 525 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
336 526
337 encode_ch (enc, '"'); 527 encode_ch (enc, '"');
338 528
339 if (enc->flags & F_SPACE_BEFORE) encode_space (enc); 529 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
340 encode_ch (enc, ':'); 530 encode_ch (enc, ':');
341 if (enc->flags & F_SPACE_AFTER ) encode_space (enc); 531 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
342 encode_sv (enc, HeVAL (he));
343} 532}
344 533
345// compare hash entries, used when all keys are bytestrings 534// compare hash entries, used when all keys are bytestrings
346static int 535static int
347he_cmp_fast (const void *a_, const void *b_) 536he_cmp_fast (const void *a_, const void *b_)
… …
352 HE *b = *(HE **)b_; 541 HE *b = *(HE **)b_;
353 542
354 STRLEN la = HeKLEN (a); 543 STRLEN la = HeKLEN (a);
355 STRLEN lb = HeKLEN (b); 544 STRLEN lb = HeKLEN (b);
356 545
357 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 546 if (!(cmp = memcmp (HeKEY (b), HeKEY (a), lb < la ? lb : la)))
358 cmp = la - lb; 547 cmp = lb - la;
359 548
360 return cmp; 549 return cmp;
361} 550}
362 551
363// compare hash entries, used when some keys are sv's or utf-x 552// compare hash entries, used when some keys are sv's or utf-x
364static int 553static int
365he_cmp_slow (const void *a, const void *b) 554he_cmp_slow (const void *a, const void *b)
366{ 555{
367 return sv_cmp (HeSVKEY_force (*(HE **)a), HeSVKEY_force (*(HE **)b)); 556 return sv_cmp (HeSVKEY_force (*(HE **)b), HeSVKEY_force (*(HE **)a));
368} 557}
369 558
370static void 559static void
371encode_hv (enc_t *enc, HV *hv) 560encode_hv (enc_t *enc, HV *hv)
372{ 561{
373 int count, i; 562 HE *he;
374 563
375 if (enc->indent >= enc->maxdepth) 564 if (enc->indent >= enc->json.max_depth)
376 croak ("data structure too deep (hit recursion limit)"); 565 croak (ERR_NESTING_EXCEEDED);
377 566
378 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 567 encode_ch (enc, '{');
379 568
380 if ((count = hv_iterinit (hv)))
381 {
382 // for canonical output we have to sort by keys first 569 // for canonical output we have to sort by keys first
383 // actually, this is mostly due to the stupid so-called 570 // actually, this is mostly due to the stupid so-called
384 // security workaround added somewhere in 5.8.x. 571 // security workaround added somewhere in 5.8.x
385 // that randomises hash orderings 572 // that randomises hash orderings
386 if (enc->flags & F_CANONICAL) 573 if (enc->json.flags & F_CANONICAL && !SvRMAGICAL (hv))
574 {
575 int count = hv_iterinit (hv);
576
577 if (SvMAGICAL (hv))
387 { 578 {
579 // need to count by iterating. could improve by dynamically building the vector below
580 // but I don't care for the speed of this special case.
581 // note also that we will run into undefined behaviour when the two iterations
582 // do not result in the same count, something I might care for in some later release.
583
584 count = 0;
585 while (hv_iternext (hv))
586 ++count;
587
588 hv_iterinit (hv);
589 }
590
591 if (count)
592 {
388 int fast = 1; 593 int i, fast = 1;
389 HE *he;
390#if defined(__BORLANDC__) || defined(_MSC_VER) 594#if defined(__BORLANDC__) || defined(_MSC_VER)
391 HE **hes = _alloca (count * sizeof (HE)); 595 HE **hes = _alloca (count * sizeof (HE));
392#else 596#else
393 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode 597 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
394#endif 598#endif
… …
421 625
422 FREETMPS; 626 FREETMPS;
423 LEAVE; 627 LEAVE;
424 } 628 }
425 629
426 for (i = 0; i < count; ++i) 630 encode_nl (enc); ++enc->indent;
631
632 while (count--)
427 { 633 {
428 encode_indent (enc); 634 encode_indent (enc);
635 he = hes [count];
429 encode_he (enc, hes [i]); 636 encode_hk (enc, he);
637 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
430 638
431 if (i < count - 1) 639 if (count)
432 encode_comma (enc); 640 encode_comma (enc);
433 } 641 }
434 642
643 encode_nl (enc); --enc->indent; encode_indent (enc);
644 }
645 }
646 else
647 {
648 if (hv_iterinit (hv) || SvMAGICAL (hv))
649 if ((he = hv_iternext (hv)))
650 {
651 encode_nl (enc); ++enc->indent;
652
653 for (;;)
654 {
435 encode_nl (enc); 655 encode_indent (enc);
656 encode_hk (enc, he);
657 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
658
659 if (!(he = hv_iternext (hv)))
660 break;
661
662 encode_comma (enc);
663 }
664
665 encode_nl (enc); --enc->indent; encode_indent (enc);
666 }
667 }
668
669 encode_ch (enc, '}');
670}
671
672// encode objects, arrays and special \0=false and \1=true values.
673static void
674encode_rv (enc_t *enc, SV *sv)
675{
676 svtype svt;
677
678 SvGETMAGIC (sv);
679 svt = SvTYPE (sv);
680
681 if (expect_false (SvOBJECT (sv)))
682 {
683 HV *stash = !JSON_SLOW || json_boolean_stash
684 ? json_boolean_stash
685 : gv_stashpv ("JSON::XS::Boolean", 1);
686
687 if (SvSTASH (sv) == stash)
688 {
689 if (SvIV (sv))
690 encode_str (enc, "true", 4, 0);
691 else
692 encode_str (enc, "false", 5, 0);
436 } 693 }
437 else 694 else
438 { 695 {
439 HE *he = hv_iternext (hv); 696#if 0
440 697 if (0 && sv_derived_from (rv, "JSON::Literal"))
441 for (;;)
442 { 698 {
443 encode_indent (enc); 699 // not yet
444 encode_he (enc, he);
445
446 if (!(he = hv_iternext (hv)))
447 break;
448
449 encode_comma (enc);
450 } 700 }
701#endif
702 if (enc->json.flags & F_CONV_BLESSED)
703 {
704 // we re-bless the reference to get overload and other niceties right
705 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
451 706
707 if (to_json)
708 {
709 dSP;
710
711 ENTER; SAVETMPS; PUSHMARK (SP);
712 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
713
714 // calling with G_SCALAR ensures that we always get a 1 return value
715 PUTBACK;
716 call_sv ((SV *)GvCV (to_json), G_SCALAR);
717 SPAGAIN;
718
719 // catch this surprisingly common error
720 if (SvROK (TOPs) && SvRV (TOPs) == sv)
721 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
722
723 sv = POPs;
724 PUTBACK;
725
452 encode_nl (enc); 726 encode_sv (enc, sv);
727
728 FREETMPS; LEAVE;
729 }
730 else if (enc->json.flags & F_ALLOW_BLESSED)
731 encode_str (enc, "null", 4, 0);
732 else
733 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
734 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
735 }
736 else if (enc->json.flags & F_ALLOW_BLESSED)
737 encode_str (enc, "null", 4, 0);
738 else
739 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",
740 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
453 } 741 }
454 } 742 }
455
456 --enc->indent; encode_indent (enc); encode_ch (enc, '}');
457}
458
459// encode objects, arrays and special \0=false and \1=true values.
460static void
461encode_rv (enc_t *enc, SV *sv)
462{
463 svtype svt;
464
465 SvGETMAGIC (sv);
466 svt = SvTYPE (sv);
467
468 if (svt == SVt_PVHV) 743 else if (svt == SVt_PVHV)
469 encode_hv (enc, (HV *)sv); 744 encode_hv (enc, (HV *)sv);
470 else if (svt == SVt_PVAV) 745 else if (svt == SVt_PVAV)
471 encode_av (enc, (AV *)sv); 746 encode_av (enc, (AV *)sv);
472 else if (svt < SVt_PVAV) 747 else if (svt < SVt_PVAV)
473 { 748 {
474 if (SvNIOK (sv) && SvIV (sv) == 0) 749 STRLEN len = 0;
750 char *pv = svt ? SvPV (sv, len) : 0;
751
752 if (len == 1 && *pv == '1')
753 encode_str (enc, "true", 4, 0);
754 else if (len == 1 && *pv == '0')
475 encode_str (enc, "false", 5, 0); 755 encode_str (enc, "false", 5, 0);
476 else if (SvNIOK (sv) && SvIV (sv) == 1) 756 else if (enc->json.flags & F_ALLOW_UNKNOWN)
477 encode_str (enc, "true", 4, 0); 757 encode_str (enc, "null", 4, 0);
478 else 758 else
479 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", 759 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
480 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 760 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
481 } 761 }
762 else if (enc->json.flags & F_ALLOW_UNKNOWN)
763 encode_str (enc, "null", 4, 0);
482 else 764 else
483 croak ("encountered %s, but JSON can only represent references to arrays or hashes", 765 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
484 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 766 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
485} 767}
486 768
… …
504 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 786 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
505 enc->cur += strlen (enc->cur); 787 enc->cur += strlen (enc->cur);
506 } 788 }
507 else if (SvIOKp (sv)) 789 else if (SvIOKp (sv))
508 { 790 {
509 // we assume we can always read an IV as a UV 791 // we assume we can always read an IV as a UV and vice versa
510 if (SvUV (sv) & ~(UV)0x7fff) 792 // we assume two's complement
511 { 793 // we assume no aliasing issues in the union
512 // large integer, use the (rather slow) snprintf way. 794 if (SvIsUV (sv) ? SvUVX (sv) <= 59000
513 need (enc, sizeof (UV) * 3); 795 : SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000)
514 enc->cur +=
515 SvIsUV(sv)
516 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
517 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
518 }
519 else
520 { 796 {
521 // optimise the "small number case" 797 // optimise the "small number case"
522 // code will likely be branchless and use only a single multiplication 798 // code will likely be branchless and use only a single multiplication
799 // works for numbers up to 59074
523 I32 i = SvIV (sv); 800 I32 i = SvIVX (sv);
524 U32 u; 801 U32 u;
525 char digit, nz = 0; 802 char digit, nz = 0;
526 803
527 need (enc, 6); 804 need (enc, 6);
528 805
… …
534 811
535 // now output digit by digit, each time masking out the integer part 812 // now output digit by digit, each time masking out the integer part
536 // and multiplying by 5 while moving the decimal point one to the right, 813 // and multiplying by 5 while moving the decimal point one to the right,
537 // resulting in a net multiplication by 10. 814 // resulting in a net multiplication by 10.
538 // we always write the digit to memory but conditionally increment 815 // we always write the digit to memory but conditionally increment
539 // the pointer, to ease the usage of conditional move instructions. 816 // the pointer, to enable the use of conditional move instructions.
540 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5; 817 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffffUL) * 5;
541 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5; 818 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffffUL) * 5;
542 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5; 819 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffffUL) * 5;
543 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5; 820 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffffUL) * 5;
544 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0' 821 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
545 } 822 }
823 else
824 {
825 // large integer, use the (rather slow) snprintf way.
826 need (enc, IVUV_MAXCHARS);
827 enc->cur +=
828 SvIsUV(sv)
829 ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))
830 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
831 }
546 } 832 }
547 else if (SvROK (sv)) 833 else if (SvROK (sv))
548 encode_rv (enc, SvRV (sv)); 834 encode_rv (enc, SvRV (sv));
549 else if (!SvOK (sv)) 835 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN)
550 encode_str (enc, "null", 4, 0); 836 encode_str (enc, "null", 4, 0);
551 else 837 else
552 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 838 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
553 SvPV_nolen (sv), SvFLAGS (sv)); 839 SvPV_nolen (sv), SvFLAGS (sv));
554} 840}
555 841
556static SV * 842static SV *
557encode_json (SV *scalar, U32 flags) 843encode_json (SV *scalar, JSON *json)
558{ 844{
559 enc_t enc; 845 enc_t enc;
560 846
561 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 847 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
562 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 848 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
563 849
564 enc.flags = flags; 850 enc.json = *json;
565 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 851 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
566 enc.cur = SvPVX (enc.sv); 852 enc.cur = SvPVX (enc.sv);
567 enc.end = SvEND (enc.sv); 853 enc.end = SvEND (enc.sv);
568 enc.indent = 0; 854 enc.indent = 0;
569 enc.maxdepth = DEC_DEPTH (flags); 855 enc.limit = enc.json.flags & F_ASCII ? 0x000080UL
856 : enc.json.flags & F_LATIN1 ? 0x000100UL
857 : 0x110000UL;
570 858
571 SvPOK_only (enc.sv); 859 SvPOK_only (enc.sv);
572 encode_sv (&enc, scalar); 860 encode_sv (&enc, scalar);
861 encode_nl (&enc);
573 862
574 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 863 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
575 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 864 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
576 865
577 if (!(flags & (F_ASCII | F_LATIN1 | F_UTF8))) 866 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
578 SvUTF8_on (enc.sv); 867 SvUTF8_on (enc.sv);
579 868
580 if (enc.flags & F_SHRINK) 869 if (enc.json.flags & F_SHRINK)
581 shrink (enc.sv); 870 shrink (enc.sv);
582 871
583 return enc.sv; 872 return enc.sv;
584} 873}
585 874
… …
590typedef struct 879typedef struct
591{ 880{
592 char *cur; // current parser pointer 881 char *cur; // current parser pointer
593 char *end; // end of input string 882 char *end; // end of input string
594 const char *err; // parse error, if != 0 883 const char *err; // parse error, if != 0
595 U32 flags; // F_* 884 JSON json;
596 U32 depth; // recursion depth 885 U32 depth; // recursion depth
597 U32 maxdepth; // recursion depth limit 886 U32 maxdepth; // recursion depth limit
598} dec_t; 887} dec_t;
599 888
600inline void 889INLINE void
890decode_comment (dec_t *dec)
891{
892 // only '#'-style comments allowed a.t.m.
893
894 while (*dec->cur && *dec->cur != 0x0a && *dec->cur != 0x0d)
895 ++dec->cur;
896}
897
898INLINE void
601decode_ws (dec_t *dec) 899decode_ws (dec_t *dec)
602{ 900{
603 for (;;) 901 for (;;)
604 { 902 {
605 char ch = *dec->cur; 903 char ch = *dec->cur;
606 904
607 if (ch > 0x20 905 if (ch > 0x20)
906 {
907 if (expect_false (ch == '#'))
908 {
909 if (dec->json.flags & F_RELAXED)
910 decode_comment (dec);
911 else
912 break;
913 }
914 else
915 break;
916 }
608 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) 917 else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)
609 break; 918 break; // parse error, but let higher level handle it, gives better error messages
610 919
611 ++dec->cur; 920 ++dec->cur;
612 } 921 }
613} 922}
614 923
… …
618 if (*dec->cur != ch) \ 927 if (*dec->cur != ch) \
619 ERR (# ch " expected"); \ 928 ERR (# ch " expected"); \
620 ++dec->cur; \ 929 ++dec->cur; \
621 SE 930 SE
622 931
623#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)") 932#define DEC_INC_DEPTH if (++dec->depth > dec->json.max_depth) ERR (ERR_NESTING_EXCEEDED)
624#define DEC_DEC_DEPTH --dec->depth 933#define DEC_DEC_DEPTH --dec->depth
625 934
626static SV *decode_sv (dec_t *dec); 935static SV *decode_sv (dec_t *dec);
627 936
628static signed char decode_hexdigit[256]; 937static signed char decode_hexdigit[256];
… …
631decode_4hex (dec_t *dec) 940decode_4hex (dec_t *dec)
632{ 941{
633 signed char d1, d2, d3, d4; 942 signed char d1, d2, d3, d4;
634 unsigned char *cur = (unsigned char *)dec->cur; 943 unsigned char *cur = (unsigned char *)dec->cur;
635 944
636 d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("four hexadecimal digits expected"); 945 d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected");
637 d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("four hexadecimal digits expected"); 946 d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected");
638 d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("four hexadecimal digits expected"); 947 d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected");
639 d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("four hexadecimal digits expected"); 948 d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected");
640 949
641 dec->cur += 4; 950 dec->cur += 4;
642 951
643 return ((UV)d1) << 12 952 return ((UV)d1) << 12
644 | ((UV)d2) << 8 953 | ((UV)d2) << 8
… …
720 1029
721 if (hi >= 0x80) 1030 if (hi >= 0x80)
722 { 1031 {
723 utf8 = 1; 1032 utf8 = 1;
724 1033
725 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0); 1034 cur = encode_utf8 (cur, hi);
726 } 1035 }
727 else 1036 else
728 *cur++ = hi; 1037 *cur++ = hi;
729 } 1038 }
730 break; 1039 break;
… …
732 default: 1041 default:
733 --dec_cur; 1042 --dec_cur;
734 ERR ("illegal backslash escape sequence in string"); 1043 ERR ("illegal backslash escape sequence in string");
735 } 1044 }
736 } 1045 }
737 else if (expect_true (ch >= 0x20 && ch <= 0x7f)) 1046 else if (expect_true (ch >= 0x20 && ch < 0x80))
738 *cur++ = ch; 1047 *cur++ = ch;
739 else if (ch >= 0x80) 1048 else if (ch >= 0x80)
740 { 1049 {
741 STRLEN clen; 1050 STRLEN clen;
742 UV uch;
743 1051
744 --dec_cur; 1052 --dec_cur;
745 1053
746 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen); 1054 decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
747 if (clen == (STRLEN)-1) 1055 if (clen == (STRLEN)-1)
748 ERR ("malformed UTF-8 character in JSON string"); 1056 ERR ("malformed UTF-8 character in JSON string");
749 1057
750 do 1058 do
751 *cur++ = *dec_cur++; 1059 *cur++ = *dec_cur++;
… …
768 { 1076 {
769 STRLEN len = cur - buf; 1077 STRLEN len = cur - buf;
770 1078
771 if (sv) 1079 if (sv)
772 { 1080 {
773 SvGROW (sv, SvCUR (sv) + len + 1); 1081 STRLEN cur = SvCUR (sv);
1082
1083 if (SvLEN (sv) <= cur + len)
1084 SvGROW (sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
1085
774 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 1086 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
775 SvCUR_set (sv, SvCUR (sv) + len); 1087 SvCUR_set (sv, SvCUR (sv) + len);
776 } 1088 }
777 else 1089 else
778 sv = newSVpvn (buf, len); 1090 sv = newSVpvn (buf, len);
… …
863 is_nv = 1; 1175 is_nv = 1;
864 } 1176 }
865 1177
866 if (!is_nv) 1178 if (!is_nv)
867 { 1179 {
1180 int len = dec->cur - start;
1181
868 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so 1182 // special case the rather common 1..5-digit-int case
869 if (*start == '-') 1183 if (*start == '-')
870 switch (dec->cur - start) 1184 switch (len)
871 { 1185 {
872 case 2: return newSViv (-( start [1] - '0' )); 1186 case 2: return newSViv (-(IV)( start [1] - '0' * 1));
873 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); 1187 case 3: return newSViv (-(IV)( start [1] * 10 + start [2] - '0' * 11));
874 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); 1188 case 4: return newSViv (-(IV)( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
875 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); 1189 case 5: return newSViv (-(IV)( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
1190 case 6: return newSViv (-(IV)(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
876 } 1191 }
877 else 1192 else
878 switch (dec->cur - start) 1193 switch (len)
879 { 1194 {
880 case 1: return newSViv ( start [0] - '0' ); 1195 case 1: return newSViv ( start [0] - '0' * 1);
881 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); 1196 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
882 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); 1197 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
883 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); 1198 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
1199 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
884 } 1200 }
885 1201
886 { 1202 {
887 UV uv; 1203 UV uv;
888 int numtype = grok_number (start, dec->cur - start, &uv); 1204 int numtype = grok_number (start, len, &uv);
889 if (numtype & IS_NUMBER_IN_UV) 1205 if (numtype & IS_NUMBER_IN_UV)
890 if (numtype & IS_NUMBER_NEG) 1206 if (numtype & IS_NUMBER_NEG)
891 { 1207 {
892 if (uv < (UV)IV_MIN) 1208 if (uv < (UV)IV_MIN)
893 return newSViv (-(IV)uv); 1209 return newSViv (-(IV)uv);
894 } 1210 }
895 else 1211 else
896 return newSVuv (uv); 1212 return newSVuv (uv);
897 } 1213 }
898 }
899 1214
1215 len -= *start == '-' ? 1 : 0;
1216
1217 // does not fit into IV or UV, try NV
1218 if (len <= NV_DIG)
1219 // fits into NV without loss of precision
1220 return newSVnv (json_atof (start));
1221
1222 // everything else fails, convert it to a string
1223 return newSVpvn (start, dec->cur - start);
1224 }
1225
1226 // loss of precision here
900 return newSVnv (Atof (start)); 1227 return newSVnv (json_atof (start));
901 1228
902fail: 1229fail:
903 return 0; 1230 return 0;
904} 1231}
905 1232
… …
934 1261
935 if (*dec->cur != ',') 1262 if (*dec->cur != ',')
936 ERR (", or ] expected while parsing array"); 1263 ERR (", or ] expected while parsing array");
937 1264
938 ++dec->cur; 1265 ++dec->cur;
1266
1267 decode_ws (dec);
1268
1269 if (*dec->cur == ']' && dec->json.flags & F_RELAXED)
1270 {
1271 ++dec->cur;
1272 break;
1273 }
939 } 1274 }
940 1275
941 DEC_DEC_DEPTH; 1276 DEC_DEC_DEPTH;
942 return newRV_noinc ((SV *)av); 1277 return newRV_noinc ((SV *)av);
943 1278
… …
948} 1283}
949 1284
950static SV * 1285static SV *
951decode_hv (dec_t *dec) 1286decode_hv (dec_t *dec)
952{ 1287{
1288 SV *sv;
953 HV *hv = newHV (); 1289 HV *hv = newHV ();
954 1290
955 DEC_INC_DEPTH; 1291 DEC_INC_DEPTH;
956 decode_ws (dec); 1292 decode_ws (dec);
957 1293
958 if (*dec->cur == '}') 1294 if (*dec->cur == '}')
959 ++dec->cur; 1295 ++dec->cur;
960 else 1296 else
961 for (;;) 1297 for (;;)
962 { 1298 {
1299 EXPECT_CH ('"');
1300
1301 // heuristic: assume that
1302 // a) decode_str + hv_store_ent are abysmally slow.
1303 // b) most hash keys are short, simple ascii text.
1304 // => try to "fast-match" such strings to avoid
1305 // the overhead of decode_str + hv_store_ent.
1306 {
963 SV *key, *value; 1307 SV *value;
1308 char *p = dec->cur;
1309 char *e = p + 24; // only try up to 24 bytes
964 1310
965 decode_ws (dec); EXPECT_CH ('"'); 1311 for (;;)
966
967 key = decode_str (dec);
968 if (!key)
969 goto fail;
970
971 decode_ws (dec); EXPECT_CH (':');
972
973 value = decode_sv (dec);
974 if (!value)
975 { 1312 {
1313 // the >= 0x80 is false on most architectures
1314 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1315 {
1316 // slow path, back up and use decode_str
1317 SV *key = decode_str (dec);
1318 if (!key)
1319 goto fail;
1320
1321 decode_ws (dec); EXPECT_CH (':');
1322
1323 decode_ws (dec);
1324 value = decode_sv (dec);
1325 if (!value)
1326 {
1327 SvREFCNT_dec (key);
1328 goto fail;
1329 }
1330
1331 hv_store_ent (hv, key, value, 0);
976 SvREFCNT_dec (key); 1332 SvREFCNT_dec (key);
1333
1334 break;
1335 }
1336 else if (*p == '"')
1337 {
1338 // fast path, got a simple key
1339 char *key = dec->cur;
1340 int len = p - key;
1341 dec->cur = p + 1;
1342
1343 decode_ws (dec); EXPECT_CH (':');
1344
1345 decode_ws (dec);
1346 value = decode_sv (dec);
1347 if (!value)
977 goto fail; 1348 goto fail;
1349
1350 hv_store (hv, key, len, value, 0);
1351
1352 break;
1353 }
1354
1355 ++p;
978 } 1356 }
979 1357 }
980 hv_store_ent (hv, key, value, 0);
981 SvREFCNT_dec (key);
982 1358
983 decode_ws (dec); 1359 decode_ws (dec);
984 1360
985 if (*dec->cur == '}') 1361 if (*dec->cur == '}')
986 { 1362 {
… …
990 1366
991 if (*dec->cur != ',') 1367 if (*dec->cur != ',')
992 ERR (", or } expected while parsing object/hash"); 1368 ERR (", or } expected while parsing object/hash");
993 1369
994 ++dec->cur; 1370 ++dec->cur;
1371
1372 decode_ws (dec);
1373
1374 if (*dec->cur == '}' && dec->json.flags & F_RELAXED)
1375 {
1376 ++dec->cur;
1377 break;
1378 }
995 } 1379 }
996 1380
997 DEC_DEC_DEPTH; 1381 DEC_DEC_DEPTH;
998 return newRV_noinc ((SV *)hv); 1382 sv = newRV_noinc ((SV *)hv);
1383
1384 // check filter callbacks
1385 if (dec->json.flags & F_HOOK)
1386 {
1387 if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
1388 {
1389 HE *cb, *he;
1390
1391 hv_iterinit (hv);
1392 he = hv_iternext (hv);
1393 hv_iterinit (hv);
1394
1395 // the next line creates a mortal sv each time its called.
1396 // might want to optimise this for common cases.
1397 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1398
1399 if (cb)
1400 {
1401 dSP;
1402 int count;
1403
1404 ENTER; SAVETMPS; PUSHMARK (SP);
1405 XPUSHs (HeVAL (he));
1406
1407 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1408
1409 if (count == 1)
1410 {
1411 sv = newSVsv (POPs);
1412 FREETMPS; LEAVE;
1413 return sv;
1414 }
1415
1416 FREETMPS; LEAVE;
1417 }
1418 }
1419
1420 if (dec->json.cb_object)
1421 {
1422 dSP;
1423 int count;
1424
1425 ENTER; SAVETMPS; PUSHMARK (SP);
1426 XPUSHs (sv_2mortal (sv));
1427
1428 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1429
1430 if (count == 1)
1431 {
1432 sv = newSVsv (POPs);
1433 FREETMPS; LEAVE;
1434 return sv;
1435 }
1436
1437 SvREFCNT_inc (sv);
1438 FREETMPS; LEAVE;
1439 }
1440 }
1441
1442 return sv;
999 1443
1000fail: 1444fail:
1001 SvREFCNT_dec (hv); 1445 SvREFCNT_dec (hv);
1002 DEC_DEC_DEPTH; 1446 DEC_DEC_DEPTH;
1003 return 0; 1447 return 0;
1004} 1448}
1005 1449
1006static SV * 1450static SV *
1007decode_sv (dec_t *dec) 1451decode_sv (dec_t *dec)
1008{ 1452{
1009 decode_ws (dec); 1453 // the beauty of JSON: you need exactly one character lookahead
1454 // to parse everything.
1010 switch (*dec->cur) 1455 switch (*dec->cur)
1011 { 1456 {
1012 case '"': ++dec->cur; return decode_str (dec); 1457 case '"': ++dec->cur; return decode_str (dec);
1013 case '[': ++dec->cur; return decode_av (dec); 1458 case '[': ++dec->cur; return decode_av (dec);
1014 case '{': ++dec->cur; return decode_hv (dec); 1459 case '{': ++dec->cur; return decode_hv (dec);
1015 1460
1016 case '-': 1461 case '-':
1017 case '0': case '1': case '2': case '3': case '4': 1462 case '0': case '1': case '2': case '3': case '4':
1018 case '5': case '6': case '7': case '8': case '9': 1463 case '5': case '6': case '7': case '8': case '9':
1019 return decode_num (dec); 1464 return decode_num (dec);
1020 1465
1021 case 't': 1466 case 't':
1022 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1467 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1023 { 1468 {
1024 dec->cur += 4; 1469 dec->cur += 4;
1470#if JSON_SLOW
1471 json_true = get_bool ("JSON::XS::true");
1472#endif
1025 return newSViv (1); 1473 return newSVsv (json_true);
1026 } 1474 }
1027 else 1475 else
1028 ERR ("'true' expected"); 1476 ERR ("'true' expected");
1029 1477
1030 break; 1478 break;
1031 1479
1032 case 'f': 1480 case 'f':
1033 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1481 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1034 { 1482 {
1035 dec->cur += 5; 1483 dec->cur += 5;
1484#if JSON_SLOW
1485 json_false = get_bool ("JSON::XS::false");
1486#endif
1036 return newSViv (0); 1487 return newSVsv (json_false);
1037 } 1488 }
1038 else 1489 else
1039 ERR ("'false' expected"); 1490 ERR ("'false' expected");
1040 1491
1041 break; 1492 break;
… …
1059fail: 1510fail:
1060 return 0; 1511 return 0;
1061} 1512}
1062 1513
1063static SV * 1514static SV *
1064decode_json (SV *string, U32 flags, UV *offset_return) 1515decode_json (SV *string, JSON *json, char **offset_return)
1065{ 1516{
1066 dec_t dec; 1517 dec_t dec;
1067 UV offset;
1068 SV *sv; 1518 SV *sv;
1069 1519
1520 /* work around bugs in 5.10 where manipulating magic values
1521 * will perl ignore the magic in subsequent accesses
1522 */
1070 SvGETMAGIC (string); 1523 /*SvGETMAGIC (string);*/
1524 if (SvMAGICAL (string))
1525 string = sv_2mortal (newSVsv (string));
1526
1071 SvUPGRADE (string, SVt_PV); 1527 SvUPGRADE (string, SVt_PV);
1072 1528
1529 /* work around a bug in perl 5.10, which causes SvCUR to fail an
1530 * assertion with -DDEBUGGING, although SvCUR is documented to
1531 * return the xpv_cur field which certainly exists after upgrading.
1532 * according to nicholas clark, calling SvPOK fixes this.
1533 * But it doesn't fix it, so try another workaround, call SvPV_nolen
1534 * and hope for the best.
1535 * Damnit, SvPV_nolen still trips over yet another assertion. This
1536 * assertion business is seriously broken, try yet another workaround
1537 * for the broken -DDEBUGGING.
1538 */
1539 {
1540#ifdef DEBUGGING
1541 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
1542#else
1543 STRLEN offset = SvCUR (string);
1544#endif
1545
1546 if (offset > json->max_size && json->max_size)
1547 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1548 (unsigned long)SvCUR (string), (unsigned long)json->max_size);
1549 }
1550
1073 if (flags & F_UTF8) 1551 if (json->flags & F_UTF8)
1074 sv_utf8_downgrade (string, 0); 1552 sv_utf8_downgrade (string, 0);
1075 else 1553 else
1076 sv_utf8_upgrade (string); 1554 sv_utf8_upgrade (string);
1077 1555
1078 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1556 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
1079 1557
1080 dec.flags = flags; 1558 dec.json = *json;
1081 dec.cur = SvPVX (string); 1559 dec.cur = SvPVX (string);
1082 dec.end = SvEND (string); 1560 dec.end = SvEND (string);
1083 dec.err = 0; 1561 dec.err = 0;
1084 dec.depth = 0; 1562 dec.depth = 0;
1085 dec.maxdepth = DEC_DEPTH (dec.flags); 1563
1564 if (dec.json.cb_object || dec.json.cb_sk_object)
1565 dec.json.flags |= F_HOOK;
1086 1566
1087 *dec.end = 0; // this should basically be a nop, too, but make sure it's there 1567 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1568
1569 decode_ws (&dec);
1088 sv = decode_sv (&dec); 1570 sv = decode_sv (&dec);
1571
1572 if (offset_return)
1573 *offset_return = dec.cur;
1089 1574
1090 if (!(offset_return || !sv)) 1575 if (!(offset_return || !sv))
1091 { 1576 {
1092 // check for trailing garbage 1577 // check for trailing garbage
1093 decode_ws (&dec); 1578 decode_ws (&dec);
… …
1096 { 1581 {
1097 dec.err = "garbage after JSON object"; 1582 dec.err = "garbage after JSON object";
1098 SvREFCNT_dec (sv); 1583 SvREFCNT_dec (sv);
1099 sv = 0; 1584 sv = 0;
1100 } 1585 }
1101 }
1102
1103 if (offset_return || !sv)
1104 {
1105 offset = dec.flags & F_UTF8
1106 ? dec.cur - SvPVX (string)
1107 : utf8_distance (dec.cur, SvPVX (string));
1108
1109 if (offset_return)
1110 *offset_return = offset;
1111 } 1586 }
1112 1587
1113 if (!sv) 1588 if (!sv)
1114 { 1589 {
1115 SV *uni = sv_newmortal (); 1590 SV *uni = sv_newmortal ();
… …
1121 SAVEVPTR (PL_curcop); 1596 SAVEVPTR (PL_curcop);
1122 PL_curcop = &cop; 1597 PL_curcop = &cop;
1123 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1598 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1124 LEAVE; 1599 LEAVE;
1125 1600
1126 croak ("%s, at character offset %d [\"%s\"]", 1601 croak ("%s, at character offset %d (before \"%s\")",
1127 dec.err, 1602 dec.err,
1128 (int)offset, 1603 ptr_to_index (string, dec.cur),
1129 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1604 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1130 } 1605 }
1131 1606
1132 sv = sv_2mortal (sv); 1607 sv = sv_2mortal (sv);
1133 1608
1134 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1609 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))
1135 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); 1610 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1136 1611
1137 return sv; 1612 return sv;
1613}
1614
1615/////////////////////////////////////////////////////////////////////////////
1616// incremental parser
1617
1618static void
1619incr_parse (JSON *self)
1620{
1621 const char *p = SvPVX (self->incr_text) + self->incr_pos;
1622
1623 // the state machine here is a bit convoluted and could be simplified a lot
1624 // but this would make it slower, so...
1625
1626 for (;;)
1627 {
1628 //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
1629 switch (self->incr_mode)
1630 {
1631 // only used for initial whitespace skipping
1632 case INCR_M_WS:
1633 for (;;)
1634 {
1635 if (*p > 0x20)
1636 {
1637 if (*p == '#')
1638 {
1639 self->incr_mode = INCR_M_C0;
1640 goto incr_m_c;
1641 }
1642 else
1643 {
1644 self->incr_mode = INCR_M_JSON;
1645 goto incr_m_json;
1646 }
1647 }
1648 else if (!*p)
1649 goto interrupt;
1650
1651 ++p;
1652 }
1653
1654 // skip a single char inside a string (for \\-processing)
1655 case INCR_M_BS:
1656 if (!*p)
1657 goto interrupt;
1658
1659 ++p;
1660 self->incr_mode = INCR_M_STR;
1661 goto incr_m_str;
1662
1663 // inside #-style comments
1664 case INCR_M_C0:
1665 case INCR_M_C1:
1666 incr_m_c:
1667 for (;;)
1668 {
1669 if (*p == '\n')
1670 {
1671 self->incr_mode = self->incr_mode == INCR_M_C0 ? INCR_M_WS : INCR_M_JSON;
1672 break;
1673 }
1674 else if (!*p)
1675 goto interrupt;
1676
1677 ++p;
1678 }
1679
1680 break;
1681
1682 // inside a string
1683 case INCR_M_STR:
1684 incr_m_str:
1685 for (;;)
1686 {
1687 if (*p == '"')
1688 {
1689 ++p;
1690 self->incr_mode = INCR_M_JSON;
1691
1692 if (!self->incr_nest)
1693 goto interrupt;
1694
1695 goto incr_m_json;
1696 }
1697 else if (*p == '\\')
1698 {
1699 ++p; // "virtually" consumes character after \
1700
1701 if (!*p) // if at end of string we have to switch modes
1702 {
1703 self->incr_mode = INCR_M_BS;
1704 goto interrupt;
1705 }
1706 }
1707 else if (!*p)
1708 goto interrupt;
1709
1710 ++p;
1711 }
1712
1713 // after initial ws, outside string
1714 case INCR_M_JSON:
1715 incr_m_json:
1716 for (;;)
1717 {
1718 switch (*p++)
1719 {
1720 case 0:
1721 --p;
1722 goto interrupt;
1723
1724 case 0x09:
1725 case 0x0a:
1726 case 0x0d:
1727 case 0x20:
1728 if (!self->incr_nest)
1729 {
1730 --p; // do not eat the whitespace, let the next round do it
1731 goto interrupt;
1732 }
1733 break;
1734
1735 case '"':
1736 self->incr_mode = INCR_M_STR;
1737 goto incr_m_str;
1738
1739 case '[':
1740 case '{':
1741 if (++self->incr_nest > self->max_depth)
1742 croak (ERR_NESTING_EXCEEDED);
1743 break;
1744
1745 case ']':
1746 case '}':
1747 if (--self->incr_nest <= 0)
1748 goto interrupt;
1749 break;
1750
1751 case '#':
1752 self->incr_mode = INCR_M_C1;
1753 goto incr_m_c;
1754 }
1755 }
1756 }
1757
1758 modechange:
1759 ;
1760 }
1761
1762interrupt:
1763 self->incr_pos = p - SvPVX (self->incr_text);
1764 //printf ("interrupt<%.*s>\n", self->incr_pos, SvPVX(self->incr_text));//D
1765 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D
1138} 1766}
1139 1767
1140///////////////////////////////////////////////////////////////////////////// 1768/////////////////////////////////////////////////////////////////////////////
1141// XS interface functions 1769// XS interface functions
1142 1770
1143MODULE = JSON::XS PACKAGE = JSON::XS 1771MODULE = JSON::XS PACKAGE = JSON::XS
1144 1772
1145BOOT: 1773BOOT:
1146{ 1774{
1147 int i; 1775 int i;
1148
1149 memset (decode_hexdigit, 0xff, 256);
1150 1776
1151 for (i = 0; i < 256; ++i) 1777 for (i = 0; i < 256; ++i)
1152 decode_hexdigit [i] = 1778 decode_hexdigit [i] =
1153 i >= '0' && i <= '9' ? i - '0' 1779 i >= '0' && i <= '9' ? i - '0'
1154 : i >= 'a' && i <= 'f' ? i - 'a' + 10 1780 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1155 : i >= 'A' && i <= 'F' ? i - 'A' + 10 1781 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1156 : -1; 1782 : -1;
1157 1783
1158 json_stash = gv_stashpv ("JSON::XS", 1); 1784 json_stash = gv_stashpv ("JSON::XS" , 1);
1785 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1786
1787 json_true = get_bool ("JSON::XS::true");
1788 json_false = get_bool ("JSON::XS::false");
1789
1790 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */
1159} 1791}
1160 1792
1161PROTOTYPES: DISABLE 1793PROTOTYPES: DISABLE
1162 1794
1163SV *new (char *dummy) 1795void CLONE (...)
1164 CODE: 1796 CODE:
1165 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1797 json_stash = 0;
1798 json_boolean_stash = 0;
1799
1800void new (char *klass)
1801 PPCODE:
1802{
1803 SV *pv = NEWSV (0, sizeof (JSON));
1804 SvPOK_only (pv);
1805 json_init ((JSON *)SvPVX (pv));
1806 XPUSHs (sv_2mortal (sv_bless (
1807 newRV_noinc (pv),
1808 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1809 )));
1810}
1811
1812void ascii (JSON *self, int enable = 1)
1813 ALIAS:
1814 ascii = F_ASCII
1815 latin1 = F_LATIN1
1816 utf8 = F_UTF8
1817 indent = F_INDENT
1818 canonical = F_CANONICAL
1819 space_before = F_SPACE_BEFORE
1820 space_after = F_SPACE_AFTER
1821 pretty = F_PRETTY
1822 allow_nonref = F_ALLOW_NONREF
1823 shrink = F_SHRINK
1824 allow_blessed = F_ALLOW_BLESSED
1825 convert_blessed = F_CONV_BLESSED
1826 relaxed = F_RELAXED
1827 allow_unknown = F_ALLOW_UNKNOWN
1828 PPCODE:
1829{
1830 if (enable)
1831 self->flags |= ix;
1832 else
1833 self->flags &= ~ix;
1834
1835 XPUSHs (ST (0));
1836}
1837
1838void get_ascii (JSON *self)
1839 ALIAS:
1840 get_ascii = F_ASCII
1841 get_latin1 = F_LATIN1
1842 get_utf8 = F_UTF8
1843 get_indent = F_INDENT
1844 get_canonical = F_CANONICAL
1845 get_space_before = F_SPACE_BEFORE
1846 get_space_after = F_SPACE_AFTER
1847 get_allow_nonref = F_ALLOW_NONREF
1848 get_shrink = F_SHRINK
1849 get_allow_blessed = F_ALLOW_BLESSED
1850 get_convert_blessed = F_CONV_BLESSED
1851 get_relaxed = F_RELAXED
1852 get_allow_unknown = F_ALLOW_UNKNOWN
1853 PPCODE:
1854 XPUSHs (boolSV (self->flags & ix));
1855
1856void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
1857 PPCODE:
1858 self->max_depth = max_depth;
1859 XPUSHs (ST (0));
1860
1861U32 get_max_depth (JSON *self)
1862 CODE:
1863 RETVAL = self->max_depth;
1166 OUTPUT: 1864 OUTPUT:
1167 RETVAL 1865 RETVAL
1168 1866
1169SV *ascii (SV *self, int enable = 1) 1867void max_size (JSON *self, U32 max_size = 0)
1170 ALIAS: 1868 PPCODE:
1171 ascii = F_ASCII 1869 self->max_size = max_size;
1172 latin1 = F_LATIN1 1870 XPUSHs (ST (0));
1173 utf8 = F_UTF8 1871
1174 indent = F_INDENT 1872int get_max_size (JSON *self)
1175 canonical = F_CANONICAL
1176 space_before = F_SPACE_BEFORE
1177 space_after = F_SPACE_AFTER
1178 pretty = F_PRETTY
1179 allow_nonref = F_ALLOW_NONREF
1180 shrink = F_SHRINK
1181 CODE: 1873 CODE:
1182{ 1874 RETVAL = self->max_size;
1183 UV *uv = SvJSON (self);
1184 if (enable)
1185 *uv |= ix;
1186 else
1187 *uv &= ~ix;
1188
1189 RETVAL = newSVsv (self);
1190}
1191 OUTPUT: 1875 OUTPUT:
1192 RETVAL 1876 RETVAL
1193 1877
1194SV *max_depth (SV *self, UV max_depth = 0x80000000UL) 1878void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1879 PPCODE:
1880{
1881 SvREFCNT_dec (self->cb_object);
1882 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
1883
1884 XPUSHs (ST (0));
1885}
1886
1887void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
1888 PPCODE:
1889{
1890 if (!self->cb_sk_object)
1891 self->cb_sk_object = newHV ();
1892
1893 if (SvOK (cb))
1894 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1895 else
1896 {
1897 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1898
1899 if (!HvKEYS (self->cb_sk_object))
1900 {
1901 SvREFCNT_dec (self->cb_sk_object);
1902 self->cb_sk_object = 0;
1903 }
1904 }
1905
1906 XPUSHs (ST (0));
1907}
1908
1909void encode (JSON *self, SV *scalar)
1910 PPCODE:
1911 XPUSHs (encode_json (scalar, self));
1912
1913void decode (JSON *self, SV *jsonstr)
1914 PPCODE:
1915 XPUSHs (decode_json (jsonstr, self, 0));
1916
1917void decode_prefix (JSON *self, SV *jsonstr)
1918 PPCODE:
1919{
1920 char *offset;
1921 EXTEND (SP, 2);
1922 PUSHs (decode_json (jsonstr, self, &offset));
1923 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset))));
1924}
1925
1926void incr_parse (JSON *self, SV *jsonstr = 0)
1927 PPCODE:
1928{
1929 if (!self->incr_text)
1930 self->incr_text = newSVpvn ("", 0);
1931
1932 // append data, if any
1933 if (jsonstr)
1934 {
1935 if (SvUTF8 (jsonstr))
1936 {
1937 if (!SvUTF8 (self->incr_text))
1938 {
1939 /* utf-8-ness differs, need to upgrade */
1940 sv_utf8_upgrade (self->incr_text);
1941
1942 if (self->incr_pos)
1943 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
1944 - (U8 *)SvPVX (self->incr_text);
1945 }
1946 }
1947 else if (SvUTF8 (self->incr_text))
1948 sv_utf8_upgrade (jsonstr);
1949
1950 {
1951 STRLEN len;
1952 const char *str = SvPV (jsonstr, len);
1953 STRLEN cur = SvCUR (self->incr_text);
1954
1955 if (SvLEN (self->incr_text) <= cur + len)
1956 SvGROW (self->incr_text, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
1957
1958 Move (str, SvEND (self->incr_text), len, char);
1959 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
1960 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
1961 }
1962 }
1963
1964 if (GIMME_V != G_VOID)
1965 do
1966 {
1967 char *offset;
1968
1969 if (!INCR_DONE (self))
1970 {
1971 incr_parse (self);
1972
1973 if (self->incr_pos > self->max_size && self->max_size)
1974 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1975 (unsigned long)self->incr_pos, (unsigned long)self->max_size);
1976
1977 if (!INCR_DONE (self))
1978 break;
1979 }
1980
1981 XPUSHs (decode_json (self->incr_text, self, &offset));
1982
1983 self->incr_pos -= offset - SvPVX (self->incr_text);
1984 self->incr_nest = 0;
1985 self->incr_mode = 0;
1986
1987 sv_chop (self->incr_text, offset);
1988 }
1989 while (GIMME_V == G_ARRAY);
1990}
1991
1992SV *incr_text (JSON *self)
1993 ATTRS: lvalue
1195 CODE: 1994 CODE:
1196{ 1995{
1197 UV *uv = SvJSON (self); 1996 if (self->incr_pos)
1198 UV log2 = 0; 1997 croak ("incr_text can not be called when the incremental parser already started parsing");
1199 1998
1200 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL; 1999 RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef;
1201
1202 while ((1UL << log2) < max_depth)
1203 ++log2;
1204
1205 *uv = *uv & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1206
1207 RETVAL = newSVsv (self);
1208} 2000}
1209 OUTPUT: 2001 OUTPUT:
1210 RETVAL 2002 RETVAL
1211 2003
2004void incr_skip (JSON *self)
2005 CODE:
2006{
2007 if (self->incr_pos)
2008 {
2009 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + self->incr_pos);
2010 self->incr_pos = 0;
2011 self->incr_nest = 0;
2012 self->incr_mode = 0;
2013 }
2014}
2015
2016void incr_reset (JSON *self)
2017 CODE:
2018{
2019 SvREFCNT_dec (self->incr_text);
2020 self->incr_text = 0;
2021 self->incr_pos = 0;
2022 self->incr_nest = 0;
2023 self->incr_mode = 0;
2024}
2025
2026void DESTROY (JSON *self)
2027 CODE:
2028 SvREFCNT_dec (self->cb_sk_object);
2029 SvREFCNT_dec (self->cb_object);
2030 SvREFCNT_dec (self->incr_text);
2031
2032PROTOTYPES: ENABLE
2033
1212void encode (SV *self, SV *scalar) 2034void encode_json (SV *scalar)
2035 ALIAS:
2036 to_json_ = 0
2037 encode_json = F_UTF8
1213 PPCODE: 2038 PPCODE:
2039{
2040 JSON json;
2041 json_init (&json);
2042 json.flags |= ix;
1214 XPUSHs (encode_json (scalar, *SvJSON (self))); 2043 XPUSHs (encode_json (scalar, &json));
2044}
1215 2045
1216void decode (SV *self, SV *jsonstr) 2046void decode_json (SV *jsonstr)
2047 ALIAS:
2048 from_json_ = 0
2049 decode_json = F_UTF8
1217 PPCODE: 2050 PPCODE:
2051{
2052 JSON json;
2053 json_init (&json);
2054 json.flags |= ix;
1218 XPUSHs (decode_json (jsonstr, *SvJSON (self), 0)); 2055 XPUSHs (decode_json (jsonstr, &json, 0));
1219
1220void decode_prefix (SV *self, SV *jsonstr)
1221 PPCODE:
1222{
1223 UV offset;
1224 EXTEND (SP, 2);
1225 PUSHs (decode_json (jsonstr, *SvJSON (self), &offset));
1226 PUSHs (sv_2mortal (newSVuv (offset)));
1227} 2056}
1228 2057
1229PROTOTYPES: ENABLE
1230
1231void to_json (SV *scalar)
1232 ALIAS:
1233 objToJson = 0
1234 PPCODE:
1235 XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8));
1236
1237void from_json (SV *jsonstr)
1238 ALIAS:
1239 jsonToObj = 0
1240 PPCODE:
1241 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8, 0));
1242

Diff Legend

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