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 1.73 by root, Wed Mar 19 13:44:43 2008 UTC vs.
Revision 1.135 by root, Thu Nov 15 20:49:12 2018 UTC

… …
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>
9#include <float.h> 10#include <float.h>
11#include <inttypes.h>
10 12
11#if defined(__BORLANDC__) || defined(_MSC_VER) 13#if defined(__BORLANDC__) || defined(_MSC_VER)
12# define snprintf _snprintf // C compilers have this in stdio.h 14# define snprintf _snprintf // C compilers have this in stdio.h
13#endif 15#endif
14 16
15// some old perls do not have this, try to make it work, no 17// some old perls do not have this, try to make it work, no
16// guarentees, though. if it breaks, you get to keep the pieces. 18// guarantees, though. if it breaks, you get to keep the pieces.
17#ifndef UTF8_MAXBYTES 19#ifndef UTF8_MAXBYTES
18# define UTF8_MAXBYTES 13 20# define UTF8_MAXBYTES 13
19#endif 21#endif
22
23// compatibility with perl <5.18
24#ifndef HvNAMELEN_get
25# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
26#endif
27#ifndef HvNAMELEN
28# define HvNAMELEN(hv) HvNAMELEN_get (hv)
29#endif
30#ifndef HvNAMEUTF8
31# define HvNAMEUTF8(hv) 0
32#endif
33
34// three extra for rounding, sign, and end of string
35#define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 3)
20 36
21#define F_ASCII 0x00000001UL 37#define F_ASCII 0x00000001UL
22#define F_LATIN1 0x00000002UL 38#define F_LATIN1 0x00000002UL
23#define F_UTF8 0x00000004UL 39#define F_UTF8 0x00000004UL
24#define F_INDENT 0x00000008UL 40#define F_INDENT 0x00000008UL
… …
28#define F_ALLOW_NONREF 0x00000100UL 44#define F_ALLOW_NONREF 0x00000100UL
29#define F_SHRINK 0x00000200UL 45#define F_SHRINK 0x00000200UL
30#define F_ALLOW_BLESSED 0x00000400UL 46#define F_ALLOW_BLESSED 0x00000400UL
31#define F_CONV_BLESSED 0x00000800UL 47#define F_CONV_BLESSED 0x00000800UL
32#define F_RELAXED 0x00001000UL 48#define F_RELAXED 0x00001000UL
33 49#define F_ALLOW_UNKNOWN 0x00002000UL
34#define F_MAXDEPTH 0xf8000000UL 50#define F_ALLOW_TAGS 0x00004000UL
35#define S_MAXDEPTH 27
36#define F_MAXSIZE 0x01f00000UL
37#define S_MAXSIZE 20
38#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing 51#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
39 52
40#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
41#define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE ))
42
43#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 53#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
44#define F_DEFAULT (9UL << S_MAXDEPTH)
45 54
46#define INIT_SIZE 32 // initial scalar size to be allocated 55#define INIT_SIZE 64 // initial scalar size to be allocated
47#define INDENT_STEP 3 // spaces per indentation level 56#define INDENT_STEP 3 // spaces per indentation level
48 57
49#define SHORT_STRING_LEN 16384 // special-case strings of up to this size 58#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
59
60#define DECODE_WANTS_OCTETS(json) ((json)->flags & F_UTF8)
50 61
51#define SB do { 62#define SB do {
52#define SE } while (0) 63#define SE } while (0)
53 64
54#if __GNUC__ >= 3 65#if __GNUC__ >= 3
… …
64 75
65#define IN_RANGE_INC(type,val,beg,end) \ 76#define IN_RANGE_INC(type,val,beg,end) \
66 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \ 77 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \
67 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg))) 78 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
68 79
80#define ERR_NESTING_EXCEEDED "json text or perl structure exceeds maximum nesting level (max_depth set too low?)"
81
69#ifdef USE_ITHREADS 82#ifdef USE_ITHREADS
70# define JSON_SLOW 1
71# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1)) 83# define JSON_STASH (expect_true (json_stash) ? json_stash : gv_stashpv ("JSON::XS", 1))
84# define BOOL_STASH (expect_true (bool_stash) ? bool_stash : gv_stashpv ("Types::Serialiser::Boolean", 1))
85# define GET_BOOL(value) (expect_true (bool_ ## value) ? bool_ ## value : get_bool ("Types::Serialiser::" # value))
72#else 86#else
73# define JSON_SLOW 0
74# define JSON_STASH json_stash 87# define JSON_STASH json_stash
88# define BOOL_STASH bool_stash
89# define GET_BOOL(value) bool_ ## value
75#endif 90#endif
76 91
77static HV *json_stash, *json_boolean_stash; // JSON::XS:: 92// the amount of HEs to allocate on the stack, when sorting keys
78static SV *json_true, *json_false; 93#define STACK_HES 64
94
95static HV *json_stash, *bool_stash; // JSON::XS::, Types::Serialiser::Boolean::
96static SV *bool_false, *bool_true;
97static SV *sv_json;
98
99enum {
100 INCR_M_WS = 0, // initial whitespace skipping, must be 0
101 INCR_M_STR, // inside string
102 INCR_M_BS, // inside backslash
103 INCR_M_C0, // inside comment in initial whitespace sequence
104 INCR_M_C1, // inside comment in other places
105 INCR_M_JSON // outside anything, count nesting
106};
107
108#define INCR_DONE(json) ((json)->incr_nest <= 0 && (json)->incr_mode == INCR_M_JSON)
79 109
80typedef struct { 110typedef struct {
81 U32 flags; 111 U32 flags;
112 U32 max_depth;
113 STRLEN max_size;
114
82 SV *cb_object; 115 SV *cb_object;
83 HV *cb_sk_object; 116 HV *cb_sk_object;
117
118 // for the incremental parser
119 SV *incr_text; // the source text so far
120 STRLEN incr_pos; // the current offset into the text
121 int incr_nest; // {[]}-nesting level
122 unsigned char incr_mode;
123
124 SV *v_false, *v_true;
84} JSON; 125} JSON;
126
127INLINE void
128json_init (JSON *json)
129{
130 static const JSON init = { F_ALLOW_NONREF, 512 };
131
132 *json = init;
133}
85 134
86///////////////////////////////////////////////////////////////////////////// 135/////////////////////////////////////////////////////////////////////////////
87// utility functions 136// utility functions
88 137
138INLINE SV *
139get_bool (const char *name)
140{
141 SV *sv = get_sv (name, 1);
142
143 SvREADONLY_on (sv);
144 SvREADONLY_on (SvRV (sv));
145
146 return sv;
147}
148
89INLINE void 149INLINE void
90shrink (SV *sv) 150shrink (SV *sv)
91{ 151{
92 sv_utf8_downgrade (sv, 1); 152 sv_utf8_downgrade (sv, 1);
153
93 if (SvLEN (sv) > SvCUR (sv) + 1) 154 if (SvLEN (sv) > SvCUR (sv) + 1)
94 { 155 {
95#ifdef SvPV_shrink_to_cur 156#ifdef SvPV_shrink_to_cur
96 SvPV_shrink_to_cur (sv); 157 SvPV_shrink_to_cur (sv);
97#elif defined (SvPV_renew) 158#elif defined (SvPV_renew)
98 SvPV_renew (sv, SvCUR (sv) + 1); 159 SvPV_renew (sv, SvCUR (sv) + 1);
99#endif 160#endif
100 } 161 }
101} 162}
102 163
164/* adds two STRLENs together, slow, and with paranoia */
165STRLEN
166strlen_sum (STRLEN l1, STRLEN l2)
167{
168 size_t sum = l1 + l2;
169
170 if (sum < (size_t)l2 || sum != (size_t)(STRLEN)sum)
171 croak ("JSON::XS: string size overflow");
172
173 return sum;
174}
175
176/* similar to SvGROW, but somewhat safer and guarantees exponential realloc strategy */
177static char *
178json_sv_grow (SV *sv, size_t len1, size_t len2)
179{
180 len1 = strlen_sum (len1, len2);
181 len1 = strlen_sum (len1, len1 >> 1);
182
183 if (len1 > 4096 - 24)
184 len1 = (len1 | 4095) - 24;
185
186 return SvGROW (sv, len1);
187}
188
103// decode an utf-8 character and return it, or (UV)-1 in 189// decode a utf-8 character and return it, or (UV)-1 in
104// case of an error. 190// case of an error.
105// we special-case "safe" characters from U+80 .. U+7FF, 191// we special-case "safe" characters from U+80 .. U+7FF,
106// but use the very good perl function to parse anything else. 192// but use the very good perl function to parse anything else.
107// note that we never call this function for a ascii codepoints 193// note that we never call this function for a ascii codepoints
108INLINE UV 194INLINE UV
… …
123// this function takes advantage of this fact, although current gccs 209// this function takes advantage of this fact, although current gccs
124// seem to optimise the check for >= 0x80 away anyways 210// seem to optimise the check for >= 0x80 away anyways
125INLINE unsigned char * 211INLINE unsigned char *
126encode_utf8 (unsigned char *s, UV ch) 212encode_utf8 (unsigned char *s, UV ch)
127{ 213{
128 if (ch <= 0x7FF) 214 if (expect_false (ch < 0x000080))
129 { 215 *s++ = ch;
130 *s++ = (ch >> 6) | 0xc0; 216 else if (expect_true (ch < 0x000800))
131 *s++ = (ch & 0x3f) | 0x80; 217 *s++ = 0xc0 | ( ch >> 6),
132 } 218 *s++ = 0x80 | ( ch & 0x3f);
133 else 219 else if ( ch < 0x010000)
134 s = uvuni_to_utf8_flags (s, ch, 0); 220 *s++ = 0xe0 | ( ch >> 12),
221 *s++ = 0x80 | ((ch >> 6) & 0x3f),
222 *s++ = 0x80 | ( ch & 0x3f);
223 else if ( ch < 0x110000)
224 *s++ = 0xf0 | ( ch >> 18),
225 *s++ = 0x80 | ((ch >> 12) & 0x3f),
226 *s++ = 0x80 | ((ch >> 6) & 0x3f),
227 *s++ = 0x80 | ( ch & 0x3f);
135 228
136 return s; 229 return s;
230}
231
232// convert offset pointer to character index, sv must be string
233static STRLEN
234ptr_to_index (SV *sv, char *offset)
235{
236 return SvUTF8 (sv)
237 ? utf8_distance (offset, SvPVX (sv))
238 : offset - SvPVX (sv);
239}
240
241/////////////////////////////////////////////////////////////////////////////
242// fp hell
243
244// scan a group of digits, and a trailing exponent
245static void
246json_atof_scan1 (const char *s, NV *accum, int *expo, int postdp, int maxdepth)
247{
248 UV uaccum = 0;
249 int eaccum = 0;
250
251 // if we recurse too deep, skip all remaining digits
252 // to avoid a stack overflow attack
253 if (expect_false (--maxdepth <= 0))
254 while (((U8)*s - '0') < 10)
255 ++s;
256
257 for (;;)
258 {
259 U8 dig = (U8)*s - '0';
260
261 if (expect_false (dig >= 10))
262 {
263 if (dig == (U8)((U8)'.' - (U8)'0'))
264 {
265 ++s;
266 json_atof_scan1 (s, accum, expo, 1, maxdepth);
267 }
268 else if ((dig | ' ') == 'e' - '0')
269 {
270 int exp2 = 0;
271 int neg = 0;
272
273 ++s;
274
275 if (*s == '-')
276 {
277 ++s;
278 neg = 1;
279 }
280 else if (*s == '+')
281 ++s;
282
283 while ((dig = (U8)*s - '0') < 10)
284 exp2 = exp2 * 10 + *s++ - '0';
285
286 *expo += neg ? -exp2 : exp2;
287 }
288
289 break;
290 }
291
292 ++s;
293
294 uaccum = uaccum * 10 + dig;
295 ++eaccum;
296
297 // if we have too many digits, then recurse for more
298 // we actually do this for rather few digits
299 if (uaccum >= (UV_MAX - 9) / 10)
300 {
301 if (postdp) *expo -= eaccum;
302 json_atof_scan1 (s, accum, expo, postdp, maxdepth);
303 if (postdp) *expo += eaccum;
304
305 break;
306 }
307 }
308
309 // this relies greatly on the quality of the pow ()
310 // implementation of the platform, but a good
311 // implementation is hard to beat.
312 // (IEEE 754 conformant ones are required to be exact)
313 if (postdp) *expo -= eaccum;
314 *accum += uaccum * Perl_pow (10., *expo);
315 *expo += eaccum;
316}
317
318static NV
319json_atof (const char *s)
320{
321 NV accum = 0.;
322 int expo = 0;
323 int neg = 0;
324
325 if (*s == '-')
326 {
327 ++s;
328 neg = 1;
329 }
330
331 // a recursion depth of ten gives us >>500 bits
332 json_atof_scan1 (s, &accum, &expo, 0, 10);
333
334 return neg ? -accum : accum;
335}
336
337// target of scalar reference is bool? -1 == nope, 0 == false, 1 == true
338static int
339ref_bool_type (SV *sv)
340{
341 svtype svt = SvTYPE (sv);
342
343 if (svt < SVt_PVAV)
344 {
345 STRLEN len = 0;
346 char *pv = svt ? SvPV (sv, len) : 0;
347
348 if (len == 1)
349 if (*pv == '1')
350 return 1;
351 else if (*pv == '0')
352 return 0;
353 }
354
355 return -1;
356}
357
358// returns whether scalar is not a reference in the sense of allow_nonref
359static int
360json_nonref (SV *scalar)
361{
362 if (!SvROK (scalar))
363 return 1;
364
365 scalar = SvRV (scalar);
366
367 if (SvTYPE (scalar) >= SVt_PVMG)
368 {
369 if (SvSTASH (scalar) == bool_stash)
370 return 1;
371
372 if (!SvOBJECT (scalar) && ref_bool_type (scalar) >= 0)
373 return 1;
374 }
375
376 return 0;
137} 377}
138 378
139///////////////////////////////////////////////////////////////////////////// 379/////////////////////////////////////////////////////////////////////////////
140// encoder 380// encoder
141 381
… …
145 char *cur; // SvPVX (sv) + current output position 385 char *cur; // SvPVX (sv) + current output position
146 char *end; // SvEND (sv) 386 char *end; // SvEND (sv)
147 SV *sv; // result scalar 387 SV *sv; // result scalar
148 JSON json; 388 JSON json;
149 U32 indent; // indentation level 389 U32 indent; // indentation level
150 U32 maxdepth; // max. indentation/recursion level
151 UV limit; // escape character values >= this value when encoding 390 UV limit; // escape character values >= this value when encoding
152} enc_t; 391} enc_t;
153 392
154INLINE void 393INLINE void
155need (enc_t *enc, STRLEN len) 394need (enc_t *enc, STRLEN len)
156{ 395{
157 if (expect_false (enc->cur + len >= enc->end)) 396 if (expect_false ((uintptr_t)(enc->end - enc->cur) < len))
158 { 397 {
159 STRLEN cur = enc->cur - SvPVX (enc->sv); 398 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
160 SvGROW (enc->sv, cur + len + 1); 399 char *buf = json_sv_grow (enc->sv, cur, len);
161 enc->cur = SvPVX (enc->sv) + cur; 400 enc->cur = buf + cur;
162 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 401 enc->end = buf + SvLEN (enc->sv) - 1;
163 } 402 }
164} 403}
165 404
166INLINE void 405INLINE void
167encode_ch (enc_t *enc, char ch) 406encode_ch (enc_t *enc, char ch)
… …
183 422
184 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case 423 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
185 { 424 {
186 if (expect_false (ch == '"')) // but with slow exceptions 425 if (expect_false (ch == '"')) // but with slow exceptions
187 { 426 {
188 need (enc, len += 1); 427 need (enc, len + 1);
189 *enc->cur++ = '\\'; 428 *enc->cur++ = '\\';
190 *enc->cur++ = '"'; 429 *enc->cur++ = '"';
191 } 430 }
192 else if (expect_false (ch == '\\')) 431 else if (expect_false (ch == '\\'))
193 { 432 {
194 need (enc, len += 1); 433 need (enc, len + 1);
195 *enc->cur++ = '\\'; 434 *enc->cur++ = '\\';
196 *enc->cur++ = '\\'; 435 *enc->cur++ = '\\';
197 } 436 }
198 else 437 else
199 *enc->cur++ = ch; 438 *enc->cur++ = ch;
… …
202 } 441 }
203 else 442 else
204 { 443 {
205 switch (ch) 444 switch (ch)
206 { 445 {
207 case '\010': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'b'; ++str; break; 446 case '\010': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 'b'; ++str; break;
208 case '\011': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 't'; ++str; break; 447 case '\011': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 't'; ++str; break;
209 case '\012': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'n'; ++str; break; 448 case '\012': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 'n'; ++str; break;
210 case '\014': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'f'; ++str; break; 449 case '\014': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 'f'; ++str; break;
211 case '\015': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'r'; ++str; break; 450 case '\015': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 'r'; ++str; break;
212 451
213 default: 452 default:
214 { 453 {
215 STRLEN clen; 454 STRLEN clen;
216 UV uch; 455 UV uch;
… …
227 clen = 1; 466 clen = 1;
228 } 467 }
229 468
230 if (uch < 0x80/*0x20*/ || uch >= enc->limit) 469 if (uch < 0x80/*0x20*/ || uch >= enc->limit)
231 { 470 {
232 if (uch > 0xFFFFUL) 471 if (uch >= 0x10000UL)
233 { 472 {
234 if (uch > 0x10FFFFUL) 473 if (uch >= 0x110000UL)
235 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); 474 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
236 475
237 need (enc, len += 11); 476 need (enc, len + 11);
238 sprintf (enc->cur, "\\u%04x\\u%04x", 477 sprintf (enc->cur, "\\u%04x\\u%04x",
239 (int)((uch - 0x10000) / 0x400 + 0xD800), 478 (int)((uch - 0x10000) / 0x400 + 0xD800),
240 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 479 (int)((uch - 0x10000) % 0x400 + 0xDC00));
241 enc->cur += 12; 480 enc->cur += 12;
242 } 481 }
243 else 482 else
244 { 483 {
245 static char hexdigit [16] = "0123456789abcdef";
246 need (enc, len += 5); 484 need (enc, len + 5);
247 *enc->cur++ = '\\'; 485 *enc->cur++ = '\\';
248 *enc->cur++ = 'u'; 486 *enc->cur++ = 'u';
249 *enc->cur++ = hexdigit [ uch >> 12 ]; 487 *enc->cur++ = PL_hexdigit [ uch >> 12 ];
250 *enc->cur++ = hexdigit [(uch >> 8) & 15]; 488 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
251 *enc->cur++ = hexdigit [(uch >> 4) & 15]; 489 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
252 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 490 *enc->cur++ = PL_hexdigit [(uch >> 0) & 15];
253 } 491 }
254 492
255 str += clen; 493 str += clen;
256 } 494 }
257 else if (enc->json.flags & F_LATIN1) 495 else if (enc->json.flags & F_LATIN1)
… …
259 *enc->cur++ = uch; 497 *enc->cur++ = uch;
260 str += clen; 498 str += clen;
261 } 499 }
262 else if (is_utf8) 500 else if (is_utf8)
263 { 501 {
264 need (enc, len += clen); 502 need (enc, len + clen);
265 do 503 do
266 { 504 {
267 *enc->cur++ = *str++; 505 *enc->cur++ = *str++;
268 } 506 }
269 while (--clen); 507 while (--clen);
270 } 508 }
271 else 509 else
272 { 510 {
273 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed 511 need (enc, len + UTF8_MAXBYTES - 1); // never more than 11 bytes needed
274 enc->cur = encode_utf8 (enc->cur, uch); 512 enc->cur = encode_utf8 (enc->cur, uch);
275 ++str; 513 ++str;
276 } 514 }
277 } 515 }
278 } 516 }
… …
328static void 566static void
329encode_av (enc_t *enc, AV *av) 567encode_av (enc_t *enc, AV *av)
330{ 568{
331 int i, len = av_len (av); 569 int i, len = av_len (av);
332 570
333 if (enc->indent >= enc->maxdepth) 571 if (enc->indent >= enc->json.max_depth)
334 croak ("data structure too deep (hit recursion limit)"); 572 croak (ERR_NESTING_EXCEEDED);
335 573
336 encode_ch (enc, '['); 574 encode_ch (enc, '[');
337 575
338 if (len >= 0) 576 if (len >= 0)
339 { 577 {
340 encode_nl (enc); ++enc->indent; 578 encode_nl (enc); ++enc->indent;
341 579
342 for (i = 0; i <= len; ++i) 580 for (i = 0; i <= len; ++i)
… …
354 encode_comma (enc); 592 encode_comma (enc);
355 } 593 }
356 594
357 encode_nl (enc); --enc->indent; encode_indent (enc); 595 encode_nl (enc); --enc->indent; encode_indent (enc);
358 } 596 }
359 597
360 encode_ch (enc, ']'); 598 encode_ch (enc, ']');
361} 599}
362 600
363static void 601static void
364encode_hk (enc_t *enc, HE *he) 602encode_hk (enc_t *enc, HE *he)
… …
368 if (HeKLEN (he) == HEf_SVKEY) 606 if (HeKLEN (he) == HEf_SVKEY)
369 { 607 {
370 SV *sv = HeSVKEY (he); 608 SV *sv = HeSVKEY (he);
371 STRLEN len; 609 STRLEN len;
372 char *str; 610 char *str;
373 611
374 SvGETMAGIC (sv); 612 SvGETMAGIC (sv);
375 str = SvPV (sv, len); 613 str = SvPV (sv, len);
376 614
377 encode_str (enc, str, len, SvUTF8 (sv)); 615 encode_str (enc, str, len, SvUTF8 (sv));
378 } 616 }
… …
413 651
414static void 652static void
415encode_hv (enc_t *enc, HV *hv) 653encode_hv (enc_t *enc, HV *hv)
416{ 654{
417 HE *he; 655 HE *he;
418 int count;
419 656
420 if (enc->indent >= enc->maxdepth) 657 if (enc->indent >= enc->json.max_depth)
421 croak ("data structure too deep (hit recursion limit)"); 658 croak (ERR_NESTING_EXCEEDED);
422 659
423 encode_ch (enc, '{'); 660 encode_ch (enc, '{');
424 661
425 // for canonical output we have to sort by keys first 662 // for canonical output we have to sort by keys first
426 // actually, this is mostly due to the stupid so-called 663 // actually, this is mostly due to the stupid so-called
427 // security workaround added somewhere in 5.8.x. 664 // security workaround added somewhere in 5.8.x
428 // that randomises hash orderings 665 // that randomises hash orderings
429 if (enc->json.flags & F_CANONICAL) 666 if (enc->json.flags & F_CANONICAL && !SvRMAGICAL (hv))
430 { 667 {
431 int count = hv_iterinit (hv); 668 int count = hv_iterinit (hv);
432 669
433 if (SvMAGICAL (hv)) 670 if (SvMAGICAL (hv))
434 { 671 {
… …
445 } 682 }
446 683
447 if (count) 684 if (count)
448 { 685 {
449 int i, fast = 1; 686 int i, fast = 1;
450#if defined(__BORLANDC__) || defined(_MSC_VER) 687 HE *hes_stack [STACK_HES];
451 HE **hes = _alloca (count * sizeof (HE)); 688 HE **hes = hes_stack;
452#else 689
453 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode 690 // allocate larger arrays on the heap
454#endif 691 if (count > STACK_HES)
692 {
693 SV *sv = sv_2mortal (NEWSV (0, count * sizeof (*hes)));
694 hes = (HE **)SvPVX (sv);
695 }
455 696
456 i = 0; 697 i = 0;
457 while ((he = hv_iternext (hv))) 698 while ((he = hv_iternext (hv)))
458 { 699 {
459 hes [i++] = he; 700 hes [i++] = he;
… …
528// encode objects, arrays and special \0=false and \1=true values. 769// encode objects, arrays and special \0=false and \1=true values.
529static void 770static void
530encode_rv (enc_t *enc, SV *sv) 771encode_rv (enc_t *enc, SV *sv)
531{ 772{
532 svtype svt; 773 svtype svt;
774 GV *method;
533 775
534 SvGETMAGIC (sv); 776 SvGETMAGIC (sv);
535 svt = SvTYPE (sv); 777 svt = SvTYPE (sv);
536 778
537 if (expect_false (SvOBJECT (sv))) 779 if (expect_false (SvOBJECT (sv)))
538 { 780 {
539 HV *stash = !JSON_SLOW || json_boolean_stash 781 HV *stash = SvSTASH (sv);
540 ? json_boolean_stash
541 : gv_stashpv ("JSON::XS::Boolean", 1);
542 782
543 if (SvSTASH (sv) == stash) 783 if (stash == bool_stash)
544 { 784 {
545 if (SvIV (sv))
546 encode_str (enc, "true", 4, 0); 785 if (SvIV (sv)) encode_str (enc, "true" , 4, 0);
547 else
548 encode_str (enc, "false", 5, 0); 786 else encode_str (enc, "false", 5, 0);
549 } 787 }
788 else if ((enc->json.flags & F_ALLOW_TAGS) && (method = gv_fetchmethod_autoload (stash, "FREEZE", 0)))
789 {
790 int count;
791 dSP;
792
793 ENTER; SAVETMPS;
794 PUSHMARK (SP);
795 EXTEND (SP, 2);
796 // we re-bless the reference to get overload and other niceties right
797 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
798 PUSHs (sv_json);
799
800 PUTBACK;
801 count = call_sv ((SV *)GvCV (method), G_ARRAY);
802 SPAGAIN;
803
804 // catch this surprisingly common error
805 if (SvROK (TOPs) && SvRV (TOPs) == sv)
806 croak ("%s::FREEZE method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
807
808 encode_ch (enc, '(');
809 encode_ch (enc, '"');
810 encode_str (enc, HvNAME (stash), HvNAMELEN (stash), HvNAMEUTF8 (stash));
811 encode_ch (enc, '"');
812 encode_ch (enc, ')');
813 encode_ch (enc, '[');
814
815 if (count)
816 {
817 int i;
818
819 for (i = 0; i < count - 1; ++i)
820 {
821 encode_sv (enc, SP[i + 1 - count]);
822 encode_ch (enc, ',');
823 }
824
825 encode_sv (enc, TOPs);
826 SP -= count;
827 }
828
829 encode_ch (enc, ']');
830
831 FREETMPS; LEAVE;
832 }
833 else if ((enc->json.flags & F_CONV_BLESSED) && (method = gv_fetchmethod_autoload (stash, "TO_JSON", 0)))
834 {
835 dSP;
836
837 ENTER; SAVETMPS;
838 PUSHMARK (SP);
839 // we re-bless the reference to get overload and other niceties right
840 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
841
842 // calling with G_SCALAR ensures that we always get a 1 return value
843 PUTBACK;
844 call_sv ((SV *)GvCV (method), G_SCALAR);
845 SPAGAIN;
846
847 // catch this surprisingly common error
848 if (SvROK (TOPs) && SvRV (TOPs) == sv)
849 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
850
851 sv = POPs;
852 PUTBACK;
853
854 encode_sv (enc, sv);
855
856 FREETMPS; LEAVE;
857 }
858 else if (enc->json.flags & F_ALLOW_BLESSED)
859 encode_str (enc, "null", 4, 0);
550 else 860 else
551 {
552#if 0
553 if (0 && sv_derived_from (rv, "JSON::Literal"))
554 {
555 // not yet
556 }
557#endif
558 if (enc->json.flags & F_CONV_BLESSED)
559 {
560 // we re-bless the reference to get overload and other niceties right
561 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
562
563 if (to_json)
564 {
565 dSP;
566
567 ENTER; SAVETMPS; PUSHMARK (SP);
568 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
569
570 // calling with G_SCALAR ensures that we always get a 1 return value
571 PUTBACK;
572 call_sv ((SV *)GvCV (to_json), G_SCALAR);
573 SPAGAIN;
574
575 // catch this surprisingly common error
576 if (SvROK (TOPs) && SvRV (TOPs) == sv)
577 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
578
579 sv = POPs;
580 PUTBACK;
581
582 encode_sv (enc, sv);
583
584 FREETMPS; LEAVE;
585 }
586 else if (enc->json.flags & F_ALLOW_BLESSED)
587 encode_str (enc, "null", 4, 0);
588 else
589 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
590 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
591 }
592 else if (enc->json.flags & F_ALLOW_BLESSED)
593 encode_str (enc, "null", 4, 0);
594 else
595 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled", 861 croak ("encountered object '%s', but neither allow_blessed, convert_blessed nor allow_tags settings are enabled (or TO_JSON/FREEZE method missing)",
596 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 862 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
597 }
598 } 863 }
599 else if (svt == SVt_PVHV) 864 else if (svt == SVt_PVHV)
600 encode_hv (enc, (HV *)sv); 865 encode_hv (enc, (HV *)sv);
601 else if (svt == SVt_PVAV) 866 else if (svt == SVt_PVAV)
602 encode_av (enc, (AV *)sv); 867 encode_av (enc, (AV *)sv);
603 else if (svt < SVt_PVAV) 868 else if (svt < SVt_PVAV)
604 { 869 {
605 STRLEN len = 0; 870 int bool_type = ref_bool_type (sv);
606 char *pv = svt ? SvPV (sv, len) : 0;
607 871
608 if (len == 1 && *pv == '1') 872 if (bool_type == 1)
609 encode_str (enc, "true", 4, 0); 873 encode_str (enc, "true", 4, 0);
610 else if (len == 1 && *pv == '0') 874 else if (bool_type == 0)
611 encode_str (enc, "false", 5, 0); 875 encode_str (enc, "false", 5, 0);
876 else if (enc->json.flags & F_ALLOW_UNKNOWN)
877 encode_str (enc, "null", 4, 0);
612 else 878 else
613 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", 879 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
614 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 880 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
615 } 881 }
882 else if (enc->json.flags & F_ALLOW_UNKNOWN)
883 encode_str (enc, "null", 4, 0);
616 else 884 else
617 croak ("encountered %s, but JSON can only represent references to arrays or hashes", 885 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
618 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 886 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
619} 887}
620 888
… …
638 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 906 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
639 enc->cur += strlen (enc->cur); 907 enc->cur += strlen (enc->cur);
640 } 908 }
641 else if (SvIOKp (sv)) 909 else if (SvIOKp (sv))
642 { 910 {
643 // we assume we can always read an IV as a UV 911 // we assume we can always read an IV as a UV and vice versa
644 if (SvUV (sv) & ~(UV)0x7fff) 912 // we assume two's complement
645 { 913 // we assume no aliasing issues in the union
646 // large integer, use the (rather slow) snprintf way. 914 if (SvIsUV (sv) ? SvUVX (sv) <= 59000
647 need (enc, sizeof (UV) * 3); 915 : SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000)
648 enc->cur +=
649 SvIsUV(sv)
650 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
651 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
652 }
653 else
654 { 916 {
655 // optimise the "small number case" 917 // optimise the "small number case"
656 // code will likely be branchless and use only a single multiplication 918 // code will likely be branchless and use only a single multiplication
919 // works for numbers up to 59074
657 I32 i = SvIV (sv); 920 I32 i = SvIVX (sv);
658 U32 u; 921 U32 u;
659 char digit, nz = 0; 922 char digit, nz = 0;
660 923
661 need (enc, 6); 924 need (enc, 6);
662 925
… …
668 931
669 // now output digit by digit, each time masking out the integer part 932 // now output digit by digit, each time masking out the integer part
670 // and multiplying by 5 while moving the decimal point one to the right, 933 // and multiplying by 5 while moving the decimal point one to the right,
671 // resulting in a net multiplication by 10. 934 // resulting in a net multiplication by 10.
672 // we always write the digit to memory but conditionally increment 935 // we always write the digit to memory but conditionally increment
673 // the pointer, to ease the usage of conditional move instructions. 936 // the pointer, to enable the use of conditional move instructions.
674 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5; 937 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffffUL) * 5;
675 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5; 938 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffffUL) * 5;
676 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5; 939 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffffUL) * 5;
677 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5; 940 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffffUL) * 5;
678 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0' 941 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
679 } 942 }
943 else
944 {
945 // large integer, use the (rather slow) snprintf way.
946 need (enc, IVUV_MAXCHARS);
947 enc->cur +=
948 SvIsUV(sv)
949 ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))
950 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
951 }
680 } 952 }
681 else if (SvROK (sv)) 953 else if (SvROK (sv))
682 encode_rv (enc, SvRV (sv)); 954 encode_rv (enc, SvRV (sv));
683 else if (!SvOK (sv)) 955 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN)
684 encode_str (enc, "null", 4, 0); 956 encode_str (enc, "null", 4, 0);
685 else 957 else
686 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 958 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, check your input data",
687 SvPV_nolen (sv), SvFLAGS (sv)); 959 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
688} 960}
689 961
690static SV * 962static SV *
691encode_json (SV *scalar, JSON *json) 963encode_json (SV *scalar, JSON *json)
692{ 964{
693 enc_t enc; 965 enc_t enc;
694 966
695 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar)) 967 if (!(json->flags & F_ALLOW_NONREF) && json_nonref (scalar))
696 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 968 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
697 969
698 enc.json = *json; 970 enc.json = *json;
699 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 971 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
700 enc.cur = SvPVX (enc.sv); 972 enc.cur = SvPVX (enc.sv);
701 enc.end = SvEND (enc.sv); 973 enc.end = SvEND (enc.sv);
702 enc.indent = 0; 974 enc.indent = 0;
703 enc.maxdepth = DEC_DEPTH (enc.json.flags);
704 enc.limit = enc.json.flags & F_ASCII ? 0x000080UL 975 enc.limit = enc.json.flags & F_ASCII ? 0x000080UL
705 : enc.json.flags & F_LATIN1 ? 0x000100UL 976 : enc.json.flags & F_LATIN1 ? 0x000100UL
706 : 0x10FFFFUL; 977 : 0x110000UL;
707 978
708 SvPOK_only (enc.sv); 979 SvPOK_only (enc.sv);
709 encode_sv (&enc, scalar); 980 encode_sv (&enc, scalar);
981 encode_nl (&enc);
710 982
711 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 983 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
712 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 984 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
713 985
714 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8))) 986 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
… …
775 if (*dec->cur != ch) \ 1047 if (*dec->cur != ch) \
776 ERR (# ch " expected"); \ 1048 ERR (# ch " expected"); \
777 ++dec->cur; \ 1049 ++dec->cur; \
778 SE 1050 SE
779 1051
780#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)") 1052#define DEC_INC_DEPTH if (++dec->depth > dec->json.max_depth) ERR (ERR_NESTING_EXCEEDED)
781#define DEC_DEC_DEPTH --dec->depth 1053#define DEC_DEC_DEPTH --dec->depth
782 1054
783static SV *decode_sv (dec_t *dec); 1055static SV *decode_sv (dec_t *dec);
784 1056
785static signed char decode_hexdigit[256]; 1057static signed char decode_hexdigit[256];
… …
889 default: 1161 default:
890 --dec_cur; 1162 --dec_cur;
891 ERR ("illegal backslash escape sequence in string"); 1163 ERR ("illegal backslash escape sequence in string");
892 } 1164 }
893 } 1165 }
894 else if (expect_true (ch >= 0x20 && ch <= 0x7f)) 1166 else if (expect_true (ch >= 0x20 && ch < 0x80))
895 *cur++ = ch; 1167 *cur++ = ch;
896 else if (ch >= 0x80) 1168 else if (ch >= 0x80)
897 { 1169 {
898 STRLEN clen; 1170 STRLEN clen;
899 UV uch;
900 1171
901 --dec_cur; 1172 --dec_cur;
902 1173
903 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen); 1174 decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
904 if (clen == (STRLEN)-1) 1175 if (clen == (STRLEN)-1)
905 ERR ("malformed UTF-8 character in JSON string"); 1176 ERR ("malformed UTF-8 character in JSON string");
906 1177
907 do 1178 do
908 *cur++ = *dec_cur++; 1179 *cur++ = *dec_cur++;
909 while (--clen); 1180 while (--clen);
910 1181
911 utf8 = 1; 1182 utf8 = 1;
912 } 1183 }
1184 else if (ch == '\t' && dec->json.flags & F_RELAXED)
1185 *cur++ = ch;
913 else 1186 else
914 { 1187 {
915 --dec_cur; 1188 --dec_cur;
916 1189
917 if (!ch) 1190 if (!ch)
… …
925 { 1198 {
926 STRLEN len = cur - buf; 1199 STRLEN len = cur - buf;
927 1200
928 if (sv) 1201 if (sv)
929 { 1202 {
930 SvGROW (sv, SvCUR (sv) + len + 1); 1203 STRLEN cur = SvCUR (sv);
1204
1205 if (SvLEN (sv) - cur <= len)
1206 json_sv_grow (sv, cur, len);
1207
931 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 1208 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
932 SvCUR_set (sv, SvCUR (sv) + len); 1209 SvCUR_set (sv, SvCUR (sv) + len);
933 } 1210 }
934 else 1211 else
935 sv = newSVpvn (buf, len); 1212 sv = newSVpvn (buf, len);
… …
1026 1303
1027 // special case the rather common 1..5-digit-int case 1304 // special case the rather common 1..5-digit-int case
1028 if (*start == '-') 1305 if (*start == '-')
1029 switch (len) 1306 switch (len)
1030 { 1307 {
1031 case 2: return newSViv (-( start [1] - '0' * 1)); 1308 case 2: return newSViv (-(IV)( start [1] - '0' * 1));
1032 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); 1309 case 3: return newSViv (-(IV)( start [1] * 10 + start [2] - '0' * 11));
1033 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); 1310 case 4: return newSViv (-(IV)( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
1034 case 5: return newSViv (-( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); 1311 case 5: return newSViv (-(IV)( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
1035 case 6: return newSViv (-(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111)); 1312 case 6: return newSViv (-(IV)(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
1036 } 1313 }
1037 else 1314 else
1038 switch (len) 1315 switch (len)
1039 { 1316 {
1040 case 1: return newSViv ( start [0] - '0' * 1); 1317 case 1: return newSViv ( start [0] - '0' * 1);
1041 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); 1318 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
1042 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); 1319 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
1043 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); 1320 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
1044 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111); 1321 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
1045 } 1322 }
1046 1323
1047 { 1324 {
1048 UV uv; 1325 UV uv;
1049 int numtype = grok_number (start, len, &uv); 1326 int numtype = grok_number (start, len, &uv);
… …
1058 } 1335 }
1059 1336
1060 len -= *start == '-' ? 1 : 0; 1337 len -= *start == '-' ? 1 : 0;
1061 1338
1062 // does not fit into IV or UV, try NV 1339 // does not fit into IV or UV, try NV
1063 if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len) 1340 if (len <= NV_DIG)
1064 #if defined (LDBL_DIG)
1065 || (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len)
1066 #endif
1067 )
1068 // fits into NV without loss of precision 1341 // fits into NV without loss of precision
1069 return newSVnv (Atof (start)); 1342 return newSVnv (json_atof (start));
1070 1343
1071 // everything else fails, convert it to a string 1344 // everything else fails, convert it to a string
1072 return newSVpvn (start, dec->cur - start); 1345 return newSVpvn (start, dec->cur - start);
1073 } 1346 }
1074 1347
1075 // loss of precision here 1348 // loss of precision here
1076 return newSVnv (Atof (start)); 1349 return newSVnv (json_atof (start));
1077 1350
1078fail: 1351fail:
1079 return 0; 1352 return 0;
1080} 1353}
1081 1354
… …
1105 if (*dec->cur == ']') 1378 if (*dec->cur == ']')
1106 { 1379 {
1107 ++dec->cur; 1380 ++dec->cur;
1108 break; 1381 break;
1109 } 1382 }
1110 1383
1111 if (*dec->cur != ',') 1384 if (*dec->cur != ',')
1112 ERR (", or ] expected while parsing array"); 1385 ERR (", or ] expected while parsing array");
1113 1386
1114 ++dec->cur; 1387 ++dec->cur;
1115 1388
… …
1157 char *p = dec->cur; 1430 char *p = dec->cur;
1158 char *e = p + 24; // only try up to 24 bytes 1431 char *e = p + 24; // only try up to 24 bytes
1159 1432
1160 for (;;) 1433 for (;;)
1161 { 1434 {
1162 // the >= 0x80 is true on most architectures 1435 // the >= 0x80 is false on most architectures
1163 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\') 1436 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1164 { 1437 {
1165 // slow path, back up and use decode_str 1438 // slow path, back up and use decode_str
1166 SV *key = decode_str (dec); 1439 SV *key = decode_str (dec);
1167 if (!key) 1440 if (!key)
… …
1239 1512
1240 hv_iterinit (hv); 1513 hv_iterinit (hv);
1241 he = hv_iternext (hv); 1514 he = hv_iternext (hv);
1242 hv_iterinit (hv); 1515 hv_iterinit (hv);
1243 1516
1244 // the next line creates a mortal sv each time its called. 1517 // the next line creates a mortal sv each time it's called.
1245 // might want to optimise this for common cases. 1518 // might want to optimise this for common cases.
1246 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0); 1519 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1247 1520
1248 if (cb) 1521 if (cb)
1249 { 1522 {
1250 dSP; 1523 dSP;
1251 int count; 1524 int count;
1252 1525
1253 ENTER; SAVETMPS; PUSHMARK (SP); 1526 ENTER; SAVETMPS;
1527 PUSHMARK (SP);
1254 XPUSHs (HeVAL (he)); 1528 XPUSHs (HeVAL (he));
1529 sv_2mortal (sv);
1255 1530
1256 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN; 1531 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1257 1532
1258 if (count == 1) 1533 if (count == 1)
1259 { 1534 {
1260 sv = newSVsv (POPs); 1535 sv = newSVsv (POPs);
1261 FREETMPS; LEAVE; 1536 FREETMPS; LEAVE;
1262 return sv; 1537 return sv;
1263 } 1538 }
1539 else if (count)
1540 croak ("filter_json_single_key_object callbacks must not return more than one scalar");
1264 1541
1542 SvREFCNT_inc (sv);
1265 FREETMPS; LEAVE; 1543 FREETMPS; LEAVE;
1266 } 1544 }
1267 } 1545 }
1268 1546
1269 if (dec->json.cb_object) 1547 if (dec->json.cb_object)
1270 { 1548 {
1271 dSP; 1549 dSP;
1272 int count; 1550 int count;
1273 1551
1274 ENTER; SAVETMPS; PUSHMARK (SP); 1552 ENTER; SAVETMPS;
1553 PUSHMARK (SP);
1275 XPUSHs (sv_2mortal (sv)); 1554 XPUSHs (sv_2mortal (sv));
1276 1555
1277 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN; 1556 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1278 1557
1279 if (count == 1) 1558 if (count == 1)
1280 { 1559 {
1281 sv = newSVsv (POPs); 1560 sv = newSVsv (POPs);
1282 FREETMPS; LEAVE; 1561 FREETMPS; LEAVE;
1283 return sv; 1562 return sv;
1284 } 1563 }
1564 else if (count)
1565 croak ("filter_json_object callbacks must not return more than one scalar");
1285 1566
1286 SvREFCNT_inc (sv); 1567 SvREFCNT_inc (sv);
1287 FREETMPS; LEAVE; 1568 FREETMPS; LEAVE;
1288 } 1569 }
1289 } 1570 }
… …
1295 DEC_DEC_DEPTH; 1576 DEC_DEC_DEPTH;
1296 return 0; 1577 return 0;
1297} 1578}
1298 1579
1299static SV * 1580static SV *
1581decode_tag (dec_t *dec)
1582{
1583 SV *tag = 0;
1584 SV *val = 0;
1585
1586 if (!(dec->json.flags & F_ALLOW_TAGS))
1587 ERR ("malformed JSON string, neither array, object, number, string or atom");
1588
1589 ++dec->cur;
1590
1591 decode_ws (dec);
1592
1593 tag = decode_sv (dec);
1594 if (!tag)
1595 goto fail;
1596
1597 if (!SvPOK (tag))
1598 ERR ("malformed JSON string, (tag) must be a string");
1599
1600 decode_ws (dec);
1601
1602 if (*dec->cur != ')')
1603 ERR (") expected after tag");
1604
1605 ++dec->cur;
1606
1607 decode_ws (dec);
1608
1609 val = decode_sv (dec);
1610 if (!val)
1611 goto fail;
1612
1613 if (!SvROK (val) || SvTYPE (SvRV (val)) != SVt_PVAV)
1614 ERR ("malformed JSON string, tag value must be an array");
1615
1616 {
1617 AV *av = (AV *)SvRV (val);
1618 int i, len = av_len (av) + 1;
1619 HV *stash = gv_stashsv (tag, 0);
1620 SV *sv;
1621
1622 if (!stash)
1623 ERR ("cannot decode perl-object (package does not exist)");
1624
1625 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
1626
1627 if (!method)
1628 ERR ("cannot decode perl-object (package does not have a THAW method)");
1629
1630 dSP;
1631
1632 ENTER; SAVETMPS;
1633 PUSHMARK (SP);
1634 EXTEND (SP, len + 2);
1635 // we re-bless the reference to get overload and other niceties right
1636 PUSHs (tag);
1637 PUSHs (sv_json);
1638
1639 for (i = 0; i < len; ++i)
1640 PUSHs (*av_fetch (av, i, 1));
1641
1642 PUTBACK;
1643 call_sv ((SV *)GvCV (method), G_SCALAR);
1644 SPAGAIN;
1645
1646 SvREFCNT_dec (tag);
1647 SvREFCNT_dec (val);
1648 sv = SvREFCNT_inc (POPs);
1649
1650 PUTBACK;
1651
1652 FREETMPS; LEAVE;
1653
1654 return sv;
1655 }
1656
1657fail:
1658 SvREFCNT_dec (tag);
1659 SvREFCNT_dec (val);
1660 return 0;
1661}
1662
1663static SV *
1300decode_sv (dec_t *dec) 1664decode_sv (dec_t *dec)
1301{ 1665{
1302 // the beauty of JSON: you need exactly one character lookahead 1666 // the beauty of JSON: you need exactly one character lookahead
1303 // to parse everything. 1667 // to parse everything.
1304 switch (*dec->cur) 1668 switch (*dec->cur)
1305 { 1669 {
1306 case '"': ++dec->cur; return decode_str (dec); 1670 case '"': ++dec->cur; return decode_str (dec);
1307 case '[': ++dec->cur; return decode_av (dec); 1671 case '[': ++dec->cur; return decode_av (dec);
1308 case '{': ++dec->cur; return decode_hv (dec); 1672 case '{': ++dec->cur; return decode_hv (dec);
1673 case '(': return decode_tag (dec);
1309 1674
1310 case '-': 1675 case '-':
1311 case '0': case '1': case '2': case '3': case '4': 1676 case '0': case '1': case '2': case '3': case '4':
1312 case '5': case '6': case '7': case '8': case '9': 1677 case '5': case '6': case '7': case '8': case '9':
1313 return decode_num (dec); 1678 return decode_num (dec);
1314 1679
1680 case 'f':
1681 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1682 {
1683 dec->cur += 5;
1684
1685 if (expect_false (!dec->json.v_false))
1686 dec->json.v_false = GET_BOOL (false);
1687
1688 return newSVsv (dec->json.v_false);
1689 }
1690 else
1691 ERR ("'false' expected");
1692
1693 break;
1694
1315 case 't': 1695 case 't':
1316 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1696 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1317 { 1697 {
1318 dec->cur += 4; 1698 dec->cur += 4;
1319#if JSON_SLOW 1699
1320 json_true = get_sv ("JSON::XS::true", 1); SvREADONLY_on (json_true); 1700 if (expect_false (!dec->json.v_true))
1321#endif 1701 dec->json.v_true = GET_BOOL (true);
1322 return SvREFCNT_inc (json_true); 1702
1703 return newSVsv (dec->json.v_true);
1323 } 1704 }
1324 else 1705 else
1325 ERR ("'true' expected"); 1706 ERR ("'true' expected");
1326
1327 break;
1328
1329 case 'f':
1330 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1331 {
1332 dec->cur += 5;
1333#if JSON_SLOW
1334 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false);
1335#endif
1336 return SvREFCNT_inc (json_false);
1337 }
1338 else
1339 ERR ("'false' expected");
1340 1707
1341 break; 1708 break;
1342 1709
1343 case 'n': 1710 case 'n':
1344 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4)) 1711 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4))
… …
1350 ERR ("'null' expected"); 1717 ERR ("'null' expected");
1351 1718
1352 break; 1719 break;
1353 1720
1354 default: 1721 default:
1355 ERR ("malformed JSON string, neither array, object, number, string or atom"); 1722 ERR ("malformed JSON string, neither tag, array, object, number, string or atom");
1356 break; 1723 break;
1357 } 1724 }
1358 1725
1359fail: 1726fail:
1360 return 0; 1727 return 0;
1361} 1728}
1362 1729
1363static SV * 1730static SV *
1364decode_json (SV *string, JSON *json, UV *offset_return) 1731decode_json (SV *string, JSON *json, STRLEN *offset_return)
1365{ 1732{
1366 dec_t dec; 1733 dec_t dec;
1367 UV offset;
1368 SV *sv; 1734 SV *sv;
1369 1735
1736 /* work around bugs in 5.10 where manipulating magic values
1737 * makes perl ignore the magic in subsequent accesses.
1738 * also make a copy of non-PV values, to get them into a clean
1739 * state (SvPV should do that, but it's buggy, see below).
1740 *
1741 * SvIsCOW_shared_hash works around a bug in perl (possibly 5.16),
1742 * as reported by Reini Urban.
1743 */
1370 SvGETMAGIC (string); 1744 /*SvGETMAGIC (string);*/
1745 if (SvMAGICAL (string) || !SvPOK (string) || SvIsCOW_shared_hash (string))
1746 string = sv_2mortal (newSVsv (string));
1747
1371 SvUPGRADE (string, SVt_PV); 1748 SvUPGRADE (string, SVt_PV);
1372 1749
1373 if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags)) 1750 /* work around a bug in perl 5.10, which causes SvCUR to fail an
1751 * assertion with -DDEBUGGING, although SvCUR is documented to
1752 * return the xpv_cur field which certainly exists after upgrading.
1753 * according to nicholas clark, calling SvPOK fixes this.
1754 * But it doesn't fix it, so try another workaround, call SvPV_nolen
1755 * and hope for the best.
1756 * Damnit, SvPV_nolen still trips over yet another assertion. This
1757 * assertion business is seriously broken, try yet another workaround
1758 * for the broken -DDEBUGGING.
1759 */
1760 {
1761#ifdef DEBUGGING
1762 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
1763#else
1764 STRLEN offset = SvCUR (string);
1765#endif
1766
1767 if (offset > json->max_size && json->max_size)
1374 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", 1768 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1375 (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags)); 1769 (unsigned long)SvCUR (string), (unsigned long)json->max_size);
1770 }
1376 1771
1377 if (json->flags & F_UTF8) 1772 if (DECODE_WANTS_OCTETS (json))
1378 sv_utf8_downgrade (string, 0); 1773 sv_utf8_downgrade (string, 0);
1379 else 1774 else
1380 sv_utf8_upgrade (string); 1775 sv_utf8_upgrade (string);
1381 1776
1382 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1777 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
1383 1778
1384 dec.json = *json; 1779 dec.json = *json;
1385 dec.cur = SvPVX (string); 1780 dec.cur = SvPVX (string);
1386 dec.end = SvEND (string); 1781 dec.end = SvEND (string);
1387 dec.err = 0; 1782 dec.err = 0;
1388 dec.depth = 0; 1783 dec.depth = 0;
1389 dec.maxdepth = DEC_DEPTH (dec.json.flags);
1390 1784
1391 if (dec.json.cb_object || dec.json.cb_sk_object) 1785 if (dec.json.cb_object || dec.json.cb_sk_object)
1392 dec.json.flags |= F_HOOK; 1786 dec.json.flags |= F_HOOK;
1393 1787
1394 *dec.end = 0; // this should basically be a nop, too, but make sure it's there 1788 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1395 1789
1396 decode_ws (&dec); 1790 decode_ws (&dec);
1397 sv = decode_sv (&dec); 1791 sv = decode_sv (&dec);
1398 1792
1399 if (!(offset_return || !sv)) 1793 if (offset_return)
1794 *offset_return = dec.cur - SvPVX (string);
1795 else if (sv)
1400 { 1796 {
1401 // check for trailing garbage 1797 // check for trailing garbage
1402 decode_ws (&dec); 1798 decode_ws (&dec);
1403 1799
1404 if (*dec.cur) 1800 if (dec.cur != dec.end)
1405 { 1801 {
1406 dec.err = "garbage after JSON object"; 1802 dec.err = "garbage after JSON object";
1407 SvREFCNT_dec (sv); 1803 SvREFCNT_dec (sv);
1408 sv = 0; 1804 sv = 0;
1409 } 1805 }
1410 }
1411
1412 if (offset_return || !sv)
1413 {
1414 offset = dec.json.flags & F_UTF8
1415 ? dec.cur - SvPVX (string)
1416 : utf8_distance (dec.cur, SvPVX (string));
1417
1418 if (offset_return)
1419 *offset_return = offset;
1420 } 1806 }
1421 1807
1422 if (!sv) 1808 if (!sv)
1423 { 1809 {
1424 SV *uni = sv_newmortal (); 1810 SV *uni = sv_newmortal ();
… …
1430 SAVEVPTR (PL_curcop); 1816 SAVEVPTR (PL_curcop);
1431 PL_curcop = &cop; 1817 PL_curcop = &cop;
1432 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1818 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1433 LEAVE; 1819 LEAVE;
1434 1820
1435 croak ("%s, at character offset %d [\"%s\"]", 1821 croak ("%s, at character offset %d (before \"%s\")",
1436 dec.err, 1822 dec.err,
1437 (int)offset, 1823 (int)ptr_to_index (string, dec.cur),
1438 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1824 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1439 } 1825 }
1440 1826
1441 sv = sv_2mortal (sv); 1827 sv = sv_2mortal (sv);
1442 1828
1443 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1829 if (!(dec.json.flags & F_ALLOW_NONREF) && json_nonref (sv))
1444 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); 1830 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1445 1831
1446 return sv; 1832 return sv;
1833}
1834
1835/////////////////////////////////////////////////////////////////////////////
1836// incremental parser
1837
1838static void
1839incr_parse (JSON *self)
1840{
1841 const char *p = SvPVX (self->incr_text) + self->incr_pos;
1842
1843 // the state machine here is a bit convoluted and could be simplified a lot
1844 // but this would make it slower, so...
1845
1846 for (;;)
1847 {
1848 //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
1849 switch (self->incr_mode)
1850 {
1851 // only used for initial whitespace skipping
1852 case INCR_M_WS:
1853 for (;;)
1854 {
1855 if (*p > 0x20)
1856 {
1857 if (*p == '#')
1858 {
1859 self->incr_mode = INCR_M_C0;
1860 goto incr_m_c;
1861 }
1862 else
1863 {
1864 self->incr_mode = INCR_M_JSON;
1865 goto incr_m_json;
1866 }
1867 }
1868 else if (!*p)
1869 goto interrupt;
1870
1871 ++p;
1872 }
1873
1874 // skip a single char inside a string (for \\-processing)
1875 case INCR_M_BS:
1876 if (!*p)
1877 goto interrupt;
1878
1879 ++p;
1880 self->incr_mode = INCR_M_STR;
1881 goto incr_m_str;
1882
1883 // inside #-style comments
1884 case INCR_M_C0:
1885 case INCR_M_C1:
1886 incr_m_c:
1887 for (;;)
1888 {
1889 if (*p == '\n')
1890 {
1891 self->incr_mode = self->incr_mode == INCR_M_C0 ? INCR_M_WS : INCR_M_JSON;
1892 break;
1893 }
1894 else if (!*p)
1895 goto interrupt;
1896
1897 ++p;
1898 }
1899
1900 break;
1901
1902 // inside a string
1903 case INCR_M_STR:
1904 incr_m_str:
1905 for (;;)
1906 {
1907 if (*p == '"')
1908 {
1909 ++p;
1910 self->incr_mode = INCR_M_JSON;
1911
1912 if (!self->incr_nest)
1913 goto interrupt;
1914
1915 goto incr_m_json;
1916 }
1917 else if (*p == '\\')
1918 {
1919 ++p; // "virtually" consumes character after \
1920
1921 if (!*p) // if at end of string we have to switch modes
1922 {
1923 self->incr_mode = INCR_M_BS;
1924 goto interrupt;
1925 }
1926 }
1927 else if (!*p)
1928 goto interrupt;
1929
1930 ++p;
1931 }
1932
1933 // after initial ws, outside string
1934 case INCR_M_JSON:
1935 incr_m_json:
1936 for (;;)
1937 {
1938 switch (*p++)
1939 {
1940 case 0:
1941 --p;
1942 goto interrupt;
1943
1944 case 0x09:
1945 case 0x0a:
1946 case 0x0d:
1947 case 0x20:
1948 if (!self->incr_nest)
1949 {
1950 --p; // do not eat the whitespace, let the next round do it
1951 goto interrupt;
1952 }
1953 break;
1954
1955 case '"':
1956 self->incr_mode = INCR_M_STR;
1957 goto incr_m_str;
1958
1959 case '[':
1960 case '{':
1961 case '(':
1962 if (++self->incr_nest > self->max_depth)
1963 croak (ERR_NESTING_EXCEEDED);
1964 break;
1965
1966 case ']':
1967 case '}':
1968 if (--self->incr_nest <= 0)
1969 goto interrupt;
1970 break;
1971
1972 case ')':
1973 --self->incr_nest;
1974 break;
1975
1976 case '#':
1977 self->incr_mode = INCR_M_C1;
1978 goto incr_m_c;
1979 }
1980 }
1981 }
1982
1983 modechange:
1984 ;
1985 }
1986
1987interrupt:
1988 self->incr_pos = p - SvPVX (self->incr_text);
1989 //printf ("interrupt<%.*s>\n", self->incr_pos, SvPVX(self->incr_text));//D
1990 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D
1447} 1991}
1448 1992
1449///////////////////////////////////////////////////////////////////////////// 1993/////////////////////////////////////////////////////////////////////////////
1450// XS interface functions 1994// XS interface functions
1451 1995
… …
1460 i >= '0' && i <= '9' ? i - '0' 2004 i >= '0' && i <= '9' ? i - '0'
1461 : i >= 'a' && i <= 'f' ? i - 'a' + 10 2005 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1462 : i >= 'A' && i <= 'F' ? i - 'A' + 10 2006 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1463 : -1; 2007 : -1;
1464 2008
1465 json_stash = gv_stashpv ("JSON::XS" , 1); 2009 json_stash = gv_stashpv ("JSON::XS" , 1);
1466 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); 2010 bool_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
2011 bool_false = get_bool ("Types::Serialiser::false");
2012 bool_true = get_bool ("Types::Serialiser::true");
1467 2013
1468 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true ); 2014 sv_json = newSVpv ("JSON", 0);
1469 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); 2015 SvREADONLY_on (sv_json);
2016
2017 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */
1470} 2018}
1471 2019
1472PROTOTYPES: DISABLE 2020PROTOTYPES: DISABLE
1473 2021
1474void CLONE (...) 2022void CLONE (...)
1475 CODE: 2023 CODE:
2024 // as long as these writes are atomic, the race should not matter
2025 // as existing threads either already use 0, or use the old value,
2026 // which is sitll correct for the initial thread.
1476 json_stash = 0; 2027 json_stash = 0;
1477 json_boolean_stash = 0; 2028 bool_stash = 0;
2029 bool_false = 0;
2030 bool_true = 0;
1478 2031
1479void new (char *klass) 2032void new (char *klass)
1480 PPCODE: 2033 PPCODE:
1481{ 2034{
1482 SV *pv = NEWSV (0, sizeof (JSON)); 2035 SV *pv = NEWSV (0, sizeof (JSON));
1483 SvPOK_only (pv); 2036 SvPOK_only (pv);
1484 Zero (SvPVX (pv), 1, JSON); 2037 json_init ((JSON *)SvPVX (pv));
1485 ((JSON *)SvPVX (pv))->flags = F_DEFAULT;
1486 XPUSHs (sv_2mortal (sv_bless ( 2038 XPUSHs (sv_2mortal (sv_bless (
1487 newRV_noinc (pv), 2039 newRV_noinc (pv),
1488 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1) 2040 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1489 ))); 2041 )));
1490} 2042}
2043
2044void boolean_values (JSON *self, SV *v_false = 0, SV *v_true = 0)
2045 PPCODE:
2046 self->v_false = newSVsv (v_false);
2047 self->v_true = newSVsv (v_true);
2048 XPUSHs (ST (0));
2049
2050void get_boolean_values (JSON *self)
2051 PPCODE:
2052 if (self->v_false && self->v_true)
2053 {
2054 EXTEND (SP, 2);
2055 PUSHs (self->v_false);
2056 PUSHs (self->v_true);
2057 }
1491 2058
1492void ascii (JSON *self, int enable = 1) 2059void ascii (JSON *self, int enable = 1)
1493 ALIAS: 2060 ALIAS:
1494 ascii = F_ASCII 2061 ascii = F_ASCII
1495 latin1 = F_LATIN1 2062 latin1 = F_LATIN1
… …
1502 allow_nonref = F_ALLOW_NONREF 2069 allow_nonref = F_ALLOW_NONREF
1503 shrink = F_SHRINK 2070 shrink = F_SHRINK
1504 allow_blessed = F_ALLOW_BLESSED 2071 allow_blessed = F_ALLOW_BLESSED
1505 convert_blessed = F_CONV_BLESSED 2072 convert_blessed = F_CONV_BLESSED
1506 relaxed = F_RELAXED 2073 relaxed = F_RELAXED
2074 allow_unknown = F_ALLOW_UNKNOWN
2075 allow_tags = F_ALLOW_TAGS
1507 PPCODE: 2076 PPCODE:
1508{ 2077{
1509 if (enable) 2078 if (enable)
1510 self->flags |= ix; 2079 self->flags |= ix;
1511 else 2080 else
… …
1526 get_allow_nonref = F_ALLOW_NONREF 2095 get_allow_nonref = F_ALLOW_NONREF
1527 get_shrink = F_SHRINK 2096 get_shrink = F_SHRINK
1528 get_allow_blessed = F_ALLOW_BLESSED 2097 get_allow_blessed = F_ALLOW_BLESSED
1529 get_convert_blessed = F_CONV_BLESSED 2098 get_convert_blessed = F_CONV_BLESSED
1530 get_relaxed = F_RELAXED 2099 get_relaxed = F_RELAXED
2100 get_allow_unknown = F_ALLOW_UNKNOWN
2101 get_allow_tags = F_ALLOW_TAGS
1531 PPCODE: 2102 PPCODE:
1532 XPUSHs (boolSV (self->flags & ix)); 2103 XPUSHs (boolSV (self->flags & ix));
1533 2104
1534void max_depth (JSON *self, UV max_depth = 0x80000000UL) 2105void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
1535 PPCODE: 2106 PPCODE:
1536{ 2107 self->max_depth = max_depth;
1537 UV log2 = 0;
1538
1539 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1540
1541 while ((1UL << log2) < max_depth)
1542 ++log2;
1543
1544 self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1545
1546 XPUSHs (ST (0)); 2108 XPUSHs (ST (0));
1547}
1548 2109
1549U32 get_max_depth (JSON *self) 2110U32 get_max_depth (JSON *self)
1550 CODE: 2111 CODE:
1551 RETVAL = DEC_DEPTH (self->flags); 2112 RETVAL = self->max_depth;
1552 OUTPUT: 2113 OUTPUT:
1553 RETVAL 2114 RETVAL
1554 2115
1555void max_size (JSON *self, UV max_size = 0) 2116void max_size (JSON *self, U32 max_size = 0)
1556 PPCODE: 2117 PPCODE:
1557{ 2118 self->max_size = max_size;
1558 UV log2 = 0;
1559
1560 if (max_size > 0x80000000UL) max_size = 0x80000000UL;
1561 if (max_size == 1) max_size = 2;
1562
1563 while ((1UL << log2) < max_size)
1564 ++log2;
1565
1566 self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1567
1568 XPUSHs (ST (0)); 2119 XPUSHs (ST (0));
1569}
1570 2120
1571int get_max_size (JSON *self) 2121int get_max_size (JSON *self)
1572 CODE: 2122 CODE:
1573 RETVAL = DEC_SIZE (self->flags); 2123 RETVAL = self->max_size;
1574 OUTPUT: 2124 OUTPUT:
1575 RETVAL 2125 RETVAL
1576 2126
1577void filter_json_object (JSON *self, SV *cb = &PL_sv_undef) 2127void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1578 PPCODE: 2128 PPCODE:
… …
1584} 2134}
1585 2135
1586void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef) 2136void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
1587 PPCODE: 2137 PPCODE:
1588{ 2138{
1589 if (!self->cb_sk_object) 2139 if (!self->cb_sk_object)
1590 self->cb_sk_object = newHV (); 2140 self->cb_sk_object = newHV ();
1591 2141
1592 if (SvOK (cb)) 2142 if (SvOK (cb))
1593 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0); 2143 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1594 else 2144 else
… …
1605 XPUSHs (ST (0)); 2155 XPUSHs (ST (0));
1606} 2156}
1607 2157
1608void encode (JSON *self, SV *scalar) 2158void encode (JSON *self, SV *scalar)
1609 PPCODE: 2159 PPCODE:
1610 XPUSHs (encode_json (scalar, self)); 2160 PUTBACK; scalar = encode_json (scalar, self); SPAGAIN;
2161 XPUSHs (scalar);
1611 2162
1612void decode (JSON *self, SV *jsonstr) 2163void decode (JSON *self, SV *jsonstr)
1613 PPCODE: 2164 PPCODE:
1614 XPUSHs (decode_json (jsonstr, self, 0)); 2165 PUTBACK; jsonstr = decode_json (jsonstr, self, 0); SPAGAIN;
2166 XPUSHs (jsonstr);
1615 2167
1616void decode_prefix (JSON *self, SV *jsonstr) 2168void decode_prefix (JSON *self, SV *jsonstr)
1617 PPCODE: 2169 PPCODE:
1618{ 2170{
1619 UV offset; 2171 SV *sv;
2172 STRLEN offset;
2173 PUTBACK; sv = decode_json (jsonstr, self, &offset); SPAGAIN;
1620 EXTEND (SP, 2); 2174 EXTEND (SP, 2);
1621 PUSHs (decode_json (jsonstr, self, &offset)); 2175 PUSHs (sv);
1622 PUSHs (sv_2mortal (newSVuv (offset))); 2176 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, SvPV_nolen (jsonstr) + offset))));
2177}
2178
2179void incr_parse (JSON *self, SV *jsonstr = 0)
2180 PPCODE:
2181{
2182 if (!self->incr_text)
2183 self->incr_text = newSVpvn ("", 0);
2184
2185 /* if utf8-ness doesn't match the decoder, need to upgrade/downgrade */
2186 if (!DECODE_WANTS_OCTETS (self) == !SvUTF8 (self->incr_text))
2187 if (DECODE_WANTS_OCTETS (self))
2188 {
2189 if (self->incr_pos)
2190 self->incr_pos = utf8_length ((U8 *)SvPVX (self->incr_text),
2191 (U8 *)SvPVX (self->incr_text) + self->incr_pos);
2192
2193 sv_utf8_downgrade (self->incr_text, 0);
2194 }
2195 else
2196 {
2197 sv_utf8_upgrade (self->incr_text);
2198
2199 if (self->incr_pos)
2200 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
2201 - (U8 *)SvPVX (self->incr_text);
2202 }
2203
2204 // append data, if any
2205 if (jsonstr)
2206 {
2207 /* make sure both strings have same encoding */
2208 if (SvUTF8 (jsonstr) != SvUTF8 (self->incr_text))
2209 if (SvUTF8 (jsonstr))
2210 sv_utf8_downgrade (jsonstr, 0);
2211 else
2212 sv_utf8_upgrade (jsonstr);
2213
2214 /* and then just blindly append */
2215 {
2216 STRLEN len;
2217 const char *str = SvPV (jsonstr, len);
2218 STRLEN cur = SvCUR (self->incr_text);
2219
2220 if (SvLEN (self->incr_text) - cur <= len)
2221 json_sv_grow (self->incr_text, cur, len);
2222
2223 Move (str, SvEND (self->incr_text), len, char);
2224 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
2225 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
2226 }
2227 }
2228
2229 if (GIMME_V != G_VOID)
2230 do
2231 {
2232 SV *sv;
2233 STRLEN offset;
2234
2235 if (!INCR_DONE (self))
2236 {
2237 incr_parse (self);
2238
2239 if (self->incr_pos > self->max_size && self->max_size)
2240 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
2241 (unsigned long)self->incr_pos, (unsigned long)self->max_size);
2242
2243 if (!INCR_DONE (self))
2244 {
2245 // as an optimisation, do not accumulate white space in the incr buffer
2246 if (self->incr_mode == INCR_M_WS && self->incr_pos)
2247 {
2248 self->incr_pos = 0;
2249 SvCUR_set (self->incr_text, 0);
2250 }
2251
2252 break;
2253 }
2254 }
2255
2256 PUTBACK; sv = decode_json (self->incr_text, self, &offset); SPAGAIN;
2257 XPUSHs (sv);
2258
2259 self->incr_pos -= offset;
2260 self->incr_nest = 0;
2261 self->incr_mode = 0;
2262
2263 sv_chop (self->incr_text, SvPVX (self->incr_text) + offset);
2264 }
2265 while (GIMME_V == G_ARRAY);
2266}
2267
2268SV *incr_text (JSON *self)
2269 ATTRS: lvalue
2270 CODE:
2271{
2272 if (self->incr_pos)
2273 croak ("incr_text can not be called when the incremental parser already started parsing");
2274
2275 RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef;
2276}
2277 OUTPUT:
2278 RETVAL
2279
2280void incr_skip (JSON *self)
2281 CODE:
2282{
2283 if (self->incr_pos)
2284 {
2285 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + self->incr_pos);
2286 self->incr_pos = 0;
2287 self->incr_nest = 0;
2288 self->incr_mode = 0;
2289 }
2290}
2291
2292void incr_reset (JSON *self)
2293 CODE:
2294{
2295 SvREFCNT_dec (self->incr_text);
2296 self->incr_text = 0;
2297 self->incr_pos = 0;
2298 self->incr_nest = 0;
2299 self->incr_mode = 0;
1623} 2300}
1624 2301
1625void DESTROY (JSON *self) 2302void DESTROY (JSON *self)
1626 CODE: 2303 CODE:
2304 SvREFCNT_dec (self->v_false);
2305 SvREFCNT_dec (self->v_true);
1627 SvREFCNT_dec (self->cb_sk_object); 2306 SvREFCNT_dec (self->cb_sk_object);
1628 SvREFCNT_dec (self->cb_object); 2307 SvREFCNT_dec (self->cb_object);
2308 SvREFCNT_dec (self->incr_text);
1629 2309
1630PROTOTYPES: ENABLE 2310PROTOTYPES: ENABLE
1631 2311
1632void encode_json (SV *scalar) 2312void encode_json (SV *scalar)
1633 PPCODE: 2313 PPCODE:
1634{ 2314{
1635 JSON json = { F_DEFAULT | F_UTF8 }; 2315 JSON json;
1636 XPUSHs (encode_json (scalar, &json)); 2316 json_init (&json);
2317 json.flags |= F_UTF8;
2318 PUTBACK; scalar = encode_json (scalar, &json); SPAGAIN;
2319 XPUSHs (scalar);
1637} 2320}
1638 2321
1639void decode_json (SV *jsonstr) 2322void decode_json (SV *jsonstr)
1640 PPCODE: 2323 PPCODE:
1641{ 2324{
1642 JSON json = { F_DEFAULT | F_UTF8 }; 2325 JSON json;
1643 XPUSHs (decode_json (jsonstr, &json, 0)); 2326 json_init (&json);
2327 json.flags |= F_UTF8;
2328 PUTBACK; jsonstr = decode_json (jsonstr, &json, 0); SPAGAIN;
2329 XPUSHs (jsonstr);
1644} 2330}
1645 2331

Diff Legend

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