… | |
… | |
12 | #if defined(__BORLANDC__) || defined(_MSC_VER) |
12 | #if defined(__BORLANDC__) || defined(_MSC_VER) |
13 | # define snprintf _snprintf // C compilers have this in stdio.h |
13 | # define snprintf _snprintf // C compilers have this in stdio.h |
14 | #endif |
14 | #endif |
15 | |
15 | |
16 | // some old perls do not have this, try to make it work, no |
16 | // some old perls do not have this, try to make it work, no |
17 | // guarentees, though. if it breaks, you get to keep the pieces. |
17 | // guarantees, though. if it breaks, you get to keep the pieces. |
18 | #ifndef UTF8_MAXBYTES |
18 | #ifndef UTF8_MAXBYTES |
19 | # define UTF8_MAXBYTES 13 |
19 | # define UTF8_MAXBYTES 13 |
20 | #endif |
20 | #endif |
21 | |
21 | |
|
|
22 | // three extra for rounding, sign, and end of string |
22 | #define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 2) |
23 | #define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 3) |
23 | |
24 | |
24 | #define F_ASCII 0x00000001UL |
25 | #define F_ASCII 0x00000001UL |
25 | #define F_LATIN1 0x00000002UL |
26 | #define F_LATIN1 0x00000002UL |
26 | #define F_UTF8 0x00000004UL |
27 | #define F_UTF8 0x00000004UL |
27 | #define F_INDENT 0x00000008UL |
28 | #define F_INDENT 0x00000008UL |
… | |
… | |
31 | #define F_ALLOW_NONREF 0x00000100UL |
32 | #define F_ALLOW_NONREF 0x00000100UL |
32 | #define F_SHRINK 0x00000200UL |
33 | #define F_SHRINK 0x00000200UL |
33 | #define F_ALLOW_BLESSED 0x00000400UL |
34 | #define F_ALLOW_BLESSED 0x00000400UL |
34 | #define F_CONV_BLESSED 0x00000800UL |
35 | #define F_CONV_BLESSED 0x00000800UL |
35 | #define F_RELAXED 0x00001000UL |
36 | #define F_RELAXED 0x00001000UL |
36 | |
37 | #define F_ALLOW_UNKNOWN 0x00002000UL |
37 | #define F_MAXDEPTH 0xf8000000UL |
|
|
38 | #define S_MAXDEPTH 27 |
|
|
39 | #define F_MAXSIZE 0x01f00000UL |
|
|
40 | #define S_MAXSIZE 20 |
|
|
41 | #define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing |
38 | #define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing |
42 | |
39 | |
43 | #define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH)) |
|
|
44 | #define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE )) |
|
|
45 | |
|
|
46 | #define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER |
40 | #define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER |
47 | #define F_DEFAULT (9UL << S_MAXDEPTH) |
|
|
48 | |
41 | |
49 | #define INIT_SIZE 32 // initial scalar size to be allocated |
42 | #define INIT_SIZE 32 // initial scalar size to be allocated |
50 | #define INDENT_STEP 3 // spaces per indentation level |
43 | #define INDENT_STEP 3 // spaces per indentation level |
51 | |
44 | |
52 | #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 |
|
|
46 | |
|
|
47 | #define DECODE_WANTS_OCTETS(json) ((json)->flags & F_UTF8) |
53 | |
48 | |
54 | #define SB do { |
49 | #define SB do { |
55 | #define SE } while (0) |
50 | #define SE } while (0) |
56 | |
51 | |
57 | #if __GNUC__ >= 3 |
52 | #if __GNUC__ >= 3 |
… | |
… | |
67 | |
62 | |
68 | #define IN_RANGE_INC(type,val,beg,end) \ |
63 | #define IN_RANGE_INC(type,val,beg,end) \ |
69 | ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \ |
64 | ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \ |
70 | <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg))) |
65 | <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg))) |
71 | |
66 | |
|
|
67 | #define ERR_NESTING_EXCEEDED "json text or perl structure exceeds maximum nesting level (max_depth set too low?)" |
|
|
68 | |
72 | #ifdef USE_ITHREADS |
69 | #ifdef USE_ITHREADS |
73 | # define JSON_SLOW 1 |
70 | # define JSON_SLOW 1 |
74 | # define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1)) |
71 | # define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1)) |
75 | #else |
72 | #else |
76 | # define JSON_SLOW 0 |
73 | # define JSON_SLOW 0 |
… | |
… | |
82 | |
79 | |
83 | enum { |
80 | enum { |
84 | INCR_M_WS = 0, // initial whitespace skipping, must be 0 |
81 | INCR_M_WS = 0, // initial whitespace skipping, must be 0 |
85 | INCR_M_STR, // inside string |
82 | INCR_M_STR, // inside string |
86 | INCR_M_BS, // inside backslash |
83 | INCR_M_BS, // inside backslash |
|
|
84 | INCR_M_C0, // inside comment in initial whitespace sequence |
|
|
85 | INCR_M_C1, // inside comment in other places |
87 | INCR_M_JSON // outside anything, count nesting |
86 | INCR_M_JSON // outside anything, count nesting |
88 | }; |
87 | }; |
89 | |
88 | |
90 | #define INCR_DONE(json) (!(json)->incr_nest && (json)->incr_mode == INCR_M_JSON) |
89 | #define INCR_DONE(json) ((json)->incr_nest <= 0 && (json)->incr_mode == INCR_M_JSON) |
91 | |
90 | |
92 | typedef struct { |
91 | typedef struct { |
93 | U32 flags; |
92 | U32 flags; |
|
|
93 | U32 max_depth; |
|
|
94 | STRLEN max_size; |
|
|
95 | |
94 | SV *cb_object; |
96 | SV *cb_object; |
95 | HV *cb_sk_object; |
97 | HV *cb_sk_object; |
96 | |
98 | |
97 | // for the incremental parser |
99 | // for the incremental parser |
98 | SV *incr_text; // the source text so far |
100 | SV *incr_text; // the source text so far |
99 | STRLEN incr_pos; // the current offset into the text |
101 | STRLEN incr_pos; // the current offset into the text |
100 | int incr_nest; // {[]}-nesting level |
102 | int incr_nest; // {[]}-nesting level |
101 | int incr_mode; |
103 | unsigned char incr_mode; |
102 | } JSON; |
104 | } JSON; |
|
|
105 | |
|
|
106 | INLINE void |
|
|
107 | json_init (JSON *json) |
|
|
108 | { |
|
|
109 | Zero (json, 1, JSON); |
|
|
110 | json->max_depth = 512; |
|
|
111 | } |
103 | |
112 | |
104 | ///////////////////////////////////////////////////////////////////////////// |
113 | ///////////////////////////////////////////////////////////////////////////// |
105 | // utility functions |
114 | // utility functions |
106 | |
115 | |
|
|
116 | INLINE SV * |
|
|
117 | get_bool (const char *name) |
|
|
118 | { |
|
|
119 | SV *sv = get_sv (name, 1); |
|
|
120 | |
|
|
121 | SvREADONLY_on (sv); |
|
|
122 | SvREADONLY_on (SvRV (sv)); |
|
|
123 | |
|
|
124 | return sv; |
|
|
125 | } |
|
|
126 | |
107 | INLINE void |
127 | INLINE void |
108 | shrink (SV *sv) |
128 | shrink (SV *sv) |
109 | { |
129 | { |
110 | sv_utf8_downgrade (sv, 1); |
130 | sv_utf8_downgrade (sv, 1); |
|
|
131 | |
111 | if (SvLEN (sv) > SvCUR (sv) + 1) |
132 | if (SvLEN (sv) > SvCUR (sv) + 1) |
112 | { |
133 | { |
113 | #ifdef SvPV_shrink_to_cur |
134 | #ifdef SvPV_shrink_to_cur |
114 | SvPV_shrink_to_cur (sv); |
135 | SvPV_shrink_to_cur (sv); |
115 | #elif defined (SvPV_renew) |
136 | #elif defined (SvPV_renew) |
… | |
… | |
159 | *s++ = 0x80 | ( ch & 0x3f); |
180 | *s++ = 0x80 | ( ch & 0x3f); |
160 | |
181 | |
161 | return s; |
182 | return s; |
162 | } |
183 | } |
163 | |
184 | |
|
|
185 | // convert offset pointer to character index, sv must be string |
|
|
186 | static STRLEN |
|
|
187 | ptr_to_index (SV *sv, char *offset) |
|
|
188 | { |
|
|
189 | return SvUTF8 (sv) |
|
|
190 | ? utf8_distance (offset, SvPVX (sv)) |
|
|
191 | : offset - SvPVX (sv); |
|
|
192 | } |
|
|
193 | |
|
|
194 | ///////////////////////////////////////////////////////////////////////////// |
|
|
195 | // fp hell |
|
|
196 | |
|
|
197 | // scan a group of digits, and a trailing exponent |
|
|
198 | static void |
|
|
199 | json_atof_scan1 (const char *s, NV *accum, int *expo, int postdp, int maxdepth) |
|
|
200 | { |
|
|
201 | UV uaccum = 0; |
|
|
202 | int eaccum = 0; |
|
|
203 | |
|
|
204 | // if we recurse too deep, skip all remaining digits |
|
|
205 | // to avoid a stack overflow attack |
|
|
206 | if (expect_false (--maxdepth <= 0)) |
|
|
207 | while (((U8)*s - '0') < 10) |
|
|
208 | ++s; |
|
|
209 | |
|
|
210 | for (;;) |
|
|
211 | { |
|
|
212 | U8 dig = (U8)*s - '0'; |
|
|
213 | |
|
|
214 | if (expect_false (dig >= 10)) |
|
|
215 | { |
|
|
216 | if (dig == (U8)((U8)'.' - (U8)'0')) |
|
|
217 | { |
|
|
218 | ++s; |
|
|
219 | json_atof_scan1 (s, accum, expo, 1, maxdepth); |
|
|
220 | } |
|
|
221 | else if ((dig | ' ') == 'e' - '0') |
|
|
222 | { |
|
|
223 | int exp2 = 0; |
|
|
224 | int neg = 0; |
|
|
225 | |
|
|
226 | ++s; |
|
|
227 | |
|
|
228 | if (*s == '-') |
|
|
229 | { |
|
|
230 | ++s; |
|
|
231 | neg = 1; |
|
|
232 | } |
|
|
233 | else if (*s == '+') |
|
|
234 | ++s; |
|
|
235 | |
|
|
236 | while ((dig = (U8)*s - '0') < 10) |
|
|
237 | exp2 = exp2 * 10 + *s++ - '0'; |
|
|
238 | |
|
|
239 | *expo += neg ? -exp2 : exp2; |
|
|
240 | } |
|
|
241 | |
|
|
242 | break; |
|
|
243 | } |
|
|
244 | |
|
|
245 | ++s; |
|
|
246 | |
|
|
247 | uaccum = uaccum * 10 + dig; |
|
|
248 | ++eaccum; |
|
|
249 | |
|
|
250 | // if we have too many digits, then recurse for more |
|
|
251 | // we actually do this for rather few digits |
|
|
252 | if (uaccum >= (UV_MAX - 9) / 10) |
|
|
253 | { |
|
|
254 | if (postdp) *expo -= eaccum; |
|
|
255 | json_atof_scan1 (s, accum, expo, postdp, maxdepth); |
|
|
256 | if (postdp) *expo += eaccum; |
|
|
257 | |
|
|
258 | break; |
|
|
259 | } |
|
|
260 | } |
|
|
261 | |
|
|
262 | // this relies greatly on the quality of the pow () |
|
|
263 | // implementation of the platform, but a good |
|
|
264 | // implementation is hard to beat. |
|
|
265 | // (IEEE 754 conformant ones are required to be exact) |
|
|
266 | if (postdp) *expo -= eaccum; |
|
|
267 | *accum += uaccum * Perl_pow (10., *expo); |
|
|
268 | *expo += eaccum; |
|
|
269 | } |
|
|
270 | |
|
|
271 | static NV |
|
|
272 | json_atof (const char *s) |
|
|
273 | { |
|
|
274 | NV accum = 0.; |
|
|
275 | int expo = 0; |
|
|
276 | int neg = 0; |
|
|
277 | |
|
|
278 | if (*s == '-') |
|
|
279 | { |
|
|
280 | ++s; |
|
|
281 | neg = 1; |
|
|
282 | } |
|
|
283 | |
|
|
284 | // a recursion depth of ten gives us >>500 bits |
|
|
285 | json_atof_scan1 (s, &accum, &expo, 0, 10); |
|
|
286 | |
|
|
287 | return neg ? -accum : accum; |
|
|
288 | } |
164 | ///////////////////////////////////////////////////////////////////////////// |
289 | ///////////////////////////////////////////////////////////////////////////// |
165 | // encoder |
290 | // encoder |
166 | |
291 | |
167 | // structure used for encoding JSON |
292 | // structure used for encoding JSON |
168 | typedef struct |
293 | typedef struct |
… | |
… | |
170 | char *cur; // SvPVX (sv) + current output position |
295 | char *cur; // SvPVX (sv) + current output position |
171 | char *end; // SvEND (sv) |
296 | char *end; // SvEND (sv) |
172 | SV *sv; // result scalar |
297 | SV *sv; // result scalar |
173 | JSON json; |
298 | JSON json; |
174 | U32 indent; // indentation level |
299 | U32 indent; // indentation level |
175 | U32 maxdepth; // max. indentation/recursion level |
|
|
176 | UV limit; // escape character values >= this value when encoding |
300 | UV limit; // escape character values >= this value when encoding |
177 | } enc_t; |
301 | } enc_t; |
178 | |
302 | |
179 | INLINE void |
303 | INLINE void |
180 | need (enc_t *enc, STRLEN len) |
304 | need (enc_t *enc, STRLEN len) |
181 | { |
305 | { |
182 | if (expect_false (enc->cur + len >= enc->end)) |
306 | if (expect_false (enc->cur + len >= enc->end)) |
183 | { |
307 | { |
184 | STRLEN cur = enc->cur - SvPVX (enc->sv); |
308 | STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); |
185 | SvGROW (enc->sv, cur + len + 1); |
309 | SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); |
186 | enc->cur = SvPVX (enc->sv) + cur; |
310 | enc->cur = SvPVX (enc->sv) + cur; |
187 | enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; |
311 | enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; |
188 | } |
312 | } |
189 | } |
313 | } |
190 | |
314 | |
… | |
… | |
265 | (int)((uch - 0x10000) % 0x400 + 0xDC00)); |
389 | (int)((uch - 0x10000) % 0x400 + 0xDC00)); |
266 | enc->cur += 12; |
390 | enc->cur += 12; |
267 | } |
391 | } |
268 | else |
392 | else |
269 | { |
393 | { |
270 | static char hexdigit [16] = "0123456789abcdef"; |
|
|
271 | need (enc, len += 5); |
394 | need (enc, len += 5); |
272 | *enc->cur++ = '\\'; |
395 | *enc->cur++ = '\\'; |
273 | *enc->cur++ = 'u'; |
396 | *enc->cur++ = 'u'; |
274 | *enc->cur++ = hexdigit [ uch >> 12 ]; |
397 | *enc->cur++ = PL_hexdigit [ uch >> 12 ]; |
275 | *enc->cur++ = hexdigit [(uch >> 8) & 15]; |
398 | *enc->cur++ = PL_hexdigit [(uch >> 8) & 15]; |
276 | *enc->cur++ = hexdigit [(uch >> 4) & 15]; |
399 | *enc->cur++ = PL_hexdigit [(uch >> 4) & 15]; |
277 | *enc->cur++ = hexdigit [(uch >> 0) & 15]; |
400 | *enc->cur++ = PL_hexdigit [(uch >> 0) & 15]; |
278 | } |
401 | } |
279 | |
402 | |
280 | str += clen; |
403 | str += clen; |
281 | } |
404 | } |
282 | else if (enc->json.flags & F_LATIN1) |
405 | else if (enc->json.flags & F_LATIN1) |
… | |
… | |
353 | static void |
476 | static void |
354 | encode_av (enc_t *enc, AV *av) |
477 | encode_av (enc_t *enc, AV *av) |
355 | { |
478 | { |
356 | int i, len = av_len (av); |
479 | int i, len = av_len (av); |
357 | |
480 | |
358 | if (enc->indent >= enc->maxdepth) |
481 | if (enc->indent >= enc->json.max_depth) |
359 | croak ("data structure too deep (hit recursion limit)"); |
482 | croak (ERR_NESTING_EXCEEDED); |
360 | |
483 | |
361 | encode_ch (enc, '['); |
484 | encode_ch (enc, '['); |
362 | |
485 | |
363 | if (len >= 0) |
486 | if (len >= 0) |
364 | { |
487 | { |
… | |
… | |
439 | static void |
562 | static void |
440 | encode_hv (enc_t *enc, HV *hv) |
563 | encode_hv (enc_t *enc, HV *hv) |
441 | { |
564 | { |
442 | HE *he; |
565 | HE *he; |
443 | |
566 | |
444 | if (enc->indent >= enc->maxdepth) |
567 | if (enc->indent >= enc->json.max_depth) |
445 | croak ("data structure too deep (hit recursion limit)"); |
568 | croak (ERR_NESTING_EXCEEDED); |
446 | |
569 | |
447 | encode_ch (enc, '{'); |
570 | encode_ch (enc, '{'); |
448 | |
571 | |
449 | // for canonical output we have to sort by keys first |
572 | // for canonical output we have to sort by keys first |
450 | // actually, this is mostly due to the stupid so-called |
573 | // actually, this is mostly due to the stupid so-called |
451 | // security workaround added somewhere in 5.8.x. |
574 | // security workaround added somewhere in 5.8.x |
452 | // that randomises hash orderings |
575 | // that randomises hash orderings |
453 | if (enc->json.flags & F_CANONICAL) |
576 | if (enc->json.flags & F_CANONICAL && !SvRMAGICAL (hv)) |
454 | { |
577 | { |
455 | int count = hv_iterinit (hv); |
578 | int count = hv_iterinit (hv); |
456 | |
579 | |
457 | if (SvMAGICAL (hv)) |
580 | if (SvMAGICAL (hv)) |
458 | { |
581 | { |
… | |
… | |
631 | |
754 | |
632 | if (len == 1 && *pv == '1') |
755 | if (len == 1 && *pv == '1') |
633 | encode_str (enc, "true", 4, 0); |
756 | encode_str (enc, "true", 4, 0); |
634 | else if (len == 1 && *pv == '0') |
757 | else if (len == 1 && *pv == '0') |
635 | encode_str (enc, "false", 5, 0); |
758 | encode_str (enc, "false", 5, 0); |
|
|
759 | else if (enc->json.flags & F_ALLOW_UNKNOWN) |
|
|
760 | encode_str (enc, "null", 4, 0); |
636 | else |
761 | else |
637 | croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", |
762 | croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", |
638 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
763 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
639 | } |
764 | } |
|
|
765 | else if (enc->json.flags & F_ALLOW_UNKNOWN) |
|
|
766 | encode_str (enc, "null", 4, 0); |
640 | else |
767 | else |
641 | croak ("encountered %s, but JSON can only represent references to arrays or hashes", |
768 | croak ("encountered %s, but JSON can only represent references to arrays or hashes", |
642 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
769 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
643 | } |
770 | } |
644 | |
771 | |
… | |
… | |
706 | : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv)); |
833 | : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv)); |
707 | } |
834 | } |
708 | } |
835 | } |
709 | else if (SvROK (sv)) |
836 | else if (SvROK (sv)) |
710 | encode_rv (enc, SvRV (sv)); |
837 | encode_rv (enc, SvRV (sv)); |
711 | else if (!SvOK (sv)) |
838 | else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN) |
712 | encode_str (enc, "null", 4, 0); |
839 | encode_str (enc, "null", 4, 0); |
713 | else |
840 | else |
714 | croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", |
841 | croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", |
715 | SvPV_nolen (sv), SvFLAGS (sv)); |
842 | SvPV_nolen (sv), SvFLAGS (sv)); |
716 | } |
843 | } |
… | |
… | |
726 | enc.json = *json; |
853 | enc.json = *json; |
727 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
854 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
728 | enc.cur = SvPVX (enc.sv); |
855 | enc.cur = SvPVX (enc.sv); |
729 | enc.end = SvEND (enc.sv); |
856 | enc.end = SvEND (enc.sv); |
730 | enc.indent = 0; |
857 | enc.indent = 0; |
731 | enc.maxdepth = DEC_DEPTH (enc.json.flags); |
|
|
732 | enc.limit = enc.json.flags & F_ASCII ? 0x000080UL |
858 | enc.limit = enc.json.flags & F_ASCII ? 0x000080UL |
733 | : enc.json.flags & F_LATIN1 ? 0x000100UL |
859 | : enc.json.flags & F_LATIN1 ? 0x000100UL |
734 | : 0x110000UL; |
860 | : 0x110000UL; |
735 | |
861 | |
736 | SvPOK_only (enc.sv); |
862 | SvPOK_only (enc.sv); |
737 | encode_sv (&enc, scalar); |
863 | encode_sv (&enc, scalar); |
|
|
864 | encode_nl (&enc); |
738 | |
865 | |
739 | SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); |
866 | SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); |
740 | *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings |
867 | *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings |
741 | |
868 | |
742 | if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8))) |
869 | if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8))) |
… | |
… | |
803 | if (*dec->cur != ch) \ |
930 | if (*dec->cur != ch) \ |
804 | ERR (# ch " expected"); \ |
931 | ERR (# ch " expected"); \ |
805 | ++dec->cur; \ |
932 | ++dec->cur; \ |
806 | SE |
933 | SE |
807 | |
934 | |
808 | #define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)") |
935 | #define DEC_INC_DEPTH if (++dec->depth > dec->json.max_depth) ERR (ERR_NESTING_EXCEEDED) |
809 | #define DEC_DEC_DEPTH --dec->depth |
936 | #define DEC_DEC_DEPTH --dec->depth |
810 | |
937 | |
811 | static SV *decode_sv (dec_t *dec); |
938 | static SV *decode_sv (dec_t *dec); |
812 | |
939 | |
813 | static signed char decode_hexdigit[256]; |
940 | static signed char decode_hexdigit[256]; |
… | |
… | |
922 | else if (expect_true (ch >= 0x20 && ch < 0x80)) |
1049 | else if (expect_true (ch >= 0x20 && ch < 0x80)) |
923 | *cur++ = ch; |
1050 | *cur++ = ch; |
924 | else if (ch >= 0x80) |
1051 | else if (ch >= 0x80) |
925 | { |
1052 | { |
926 | STRLEN clen; |
1053 | STRLEN clen; |
927 | UV uch; |
|
|
928 | |
1054 | |
929 | --dec_cur; |
1055 | --dec_cur; |
930 | |
1056 | |
931 | uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen); |
1057 | decode_utf8 (dec_cur, dec->end - dec_cur, &clen); |
932 | if (clen == (STRLEN)-1) |
1058 | if (clen == (STRLEN)-1) |
933 | ERR ("malformed UTF-8 character in JSON string"); |
1059 | ERR ("malformed UTF-8 character in JSON string"); |
934 | |
1060 | |
935 | do |
1061 | do |
936 | *cur++ = *dec_cur++; |
1062 | *cur++ = *dec_cur++; |
… | |
… | |
953 | { |
1079 | { |
954 | STRLEN len = cur - buf; |
1080 | STRLEN len = cur - buf; |
955 | |
1081 | |
956 | if (sv) |
1082 | if (sv) |
957 | { |
1083 | { |
958 | SvGROW (sv, SvCUR (sv) + len + 1); |
1084 | STRLEN cur = SvCUR (sv); |
|
|
1085 | |
|
|
1086 | if (SvLEN (sv) <= cur + len) |
|
|
1087 | SvGROW (sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); |
|
|
1088 | |
959 | memcpy (SvPVX (sv) + SvCUR (sv), buf, len); |
1089 | memcpy (SvPVX (sv) + SvCUR (sv), buf, len); |
960 | SvCUR_set (sv, SvCUR (sv) + len); |
1090 | SvCUR_set (sv, SvCUR (sv) + len); |
961 | } |
1091 | } |
962 | else |
1092 | else |
963 | sv = newSVpvn (buf, len); |
1093 | sv = newSVpvn (buf, len); |
… | |
… | |
1054 | |
1184 | |
1055 | // special case the rather common 1..5-digit-int case |
1185 | // special case the rather common 1..5-digit-int case |
1056 | if (*start == '-') |
1186 | if (*start == '-') |
1057 | switch (len) |
1187 | switch (len) |
1058 | { |
1188 | { |
1059 | case 2: return newSViv (-( start [1] - '0' * 1)); |
1189 | case 2: return newSViv (-(IV)( start [1] - '0' * 1)); |
1060 | case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); |
1190 | case 3: return newSViv (-(IV)( start [1] * 10 + start [2] - '0' * 11)); |
1061 | case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); |
1191 | case 4: return newSViv (-(IV)( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); |
1062 | case 5: return newSViv (-( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); |
1192 | case 5: return newSViv (-(IV)( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); |
1063 | case 6: return newSViv (-(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111)); |
1193 | case 6: return newSViv (-(IV)(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111)); |
1064 | } |
1194 | } |
1065 | else |
1195 | else |
1066 | switch (len) |
1196 | switch (len) |
1067 | { |
1197 | { |
1068 | case 1: return newSViv ( start [0] - '0' * 1); |
1198 | case 1: return newSViv ( start [0] - '0' * 1); |
1069 | case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); |
1199 | case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); |
1070 | case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); |
1200 | case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); |
1071 | case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); |
1201 | case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); |
1072 | case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111); |
1202 | case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111); |
1073 | } |
1203 | } |
1074 | |
1204 | |
1075 | { |
1205 | { |
1076 | UV uv; |
1206 | UV uv; |
1077 | int numtype = grok_number (start, len, &uv); |
1207 | int numtype = grok_number (start, len, &uv); |
… | |
… | |
1086 | } |
1216 | } |
1087 | |
1217 | |
1088 | len -= *start == '-' ? 1 : 0; |
1218 | len -= *start == '-' ? 1 : 0; |
1089 | |
1219 | |
1090 | // does not fit into IV or UV, try NV |
1220 | // does not fit into IV or UV, try NV |
1091 | if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len) |
1221 | if (len <= NV_DIG) |
1092 | #if defined (LDBL_DIG) |
|
|
1093 | || (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len) |
|
|
1094 | #endif |
|
|
1095 | ) |
|
|
1096 | // fits into NV without loss of precision |
1222 | // fits into NV without loss of precision |
1097 | return newSVnv (Atof (start)); |
1223 | return newSVnv (json_atof (start)); |
1098 | |
1224 | |
1099 | // everything else fails, convert it to a string |
1225 | // everything else fails, convert it to a string |
1100 | return newSVpvn (start, dec->cur - start); |
1226 | return newSVpvn (start, dec->cur - start); |
1101 | } |
1227 | } |
1102 | |
1228 | |
1103 | // loss of precision here |
1229 | // loss of precision here |
1104 | return newSVnv (Atof (start)); |
1230 | return newSVnv (json_atof (start)); |
1105 | |
1231 | |
1106 | fail: |
1232 | fail: |
1107 | return 0; |
1233 | return 0; |
1108 | } |
1234 | } |
1109 | |
1235 | |
… | |
… | |
1278 | dSP; |
1404 | dSP; |
1279 | int count; |
1405 | int count; |
1280 | |
1406 | |
1281 | ENTER; SAVETMPS; PUSHMARK (SP); |
1407 | ENTER; SAVETMPS; PUSHMARK (SP); |
1282 | XPUSHs (HeVAL (he)); |
1408 | XPUSHs (HeVAL (he)); |
|
|
1409 | sv_2mortal (sv); |
1283 | |
1410 | |
1284 | PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN; |
1411 | PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN; |
1285 | |
1412 | |
1286 | if (count == 1) |
1413 | if (count == 1) |
1287 | { |
1414 | { |
1288 | sv = newSVsv (POPs); |
1415 | sv = newSVsv (POPs); |
1289 | FREETMPS; LEAVE; |
1416 | FREETMPS; LEAVE; |
1290 | return sv; |
1417 | return sv; |
1291 | } |
1418 | } |
1292 | |
1419 | |
|
|
1420 | SvREFCNT_inc (sv); |
1293 | FREETMPS; LEAVE; |
1421 | FREETMPS; LEAVE; |
1294 | } |
1422 | } |
1295 | } |
1423 | } |
1296 | |
1424 | |
1297 | if (dec->json.cb_object) |
1425 | if (dec->json.cb_object) |
… | |
… | |
1343 | case 't': |
1471 | case 't': |
1344 | if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) |
1472 | if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) |
1345 | { |
1473 | { |
1346 | dec->cur += 4; |
1474 | dec->cur += 4; |
1347 | #if JSON_SLOW |
1475 | #if JSON_SLOW |
1348 | json_true = get_sv ("JSON::XS::true", 1); SvREADONLY_on (json_true); |
1476 | json_true = get_bool ("JSON::XS::true"); |
1349 | #endif |
1477 | #endif |
1350 | return SvREFCNT_inc (json_true); |
1478 | return newSVsv (json_true); |
1351 | } |
1479 | } |
1352 | else |
1480 | else |
1353 | ERR ("'true' expected"); |
1481 | ERR ("'true' expected"); |
1354 | |
1482 | |
1355 | break; |
1483 | break; |
… | |
… | |
1357 | case 'f': |
1485 | case 'f': |
1358 | if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) |
1486 | if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) |
1359 | { |
1487 | { |
1360 | dec->cur += 5; |
1488 | dec->cur += 5; |
1361 | #if JSON_SLOW |
1489 | #if JSON_SLOW |
1362 | json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); |
1490 | json_false = get_bool ("JSON::XS::false"); |
1363 | #endif |
1491 | #endif |
1364 | return SvREFCNT_inc (json_false); |
1492 | return newSVsv (json_false); |
1365 | } |
1493 | } |
1366 | else |
1494 | else |
1367 | ERR ("'false' expected"); |
1495 | ERR ("'false' expected"); |
1368 | |
1496 | |
1369 | break; |
1497 | break; |
… | |
… | |
1387 | fail: |
1515 | fail: |
1388 | return 0; |
1516 | return 0; |
1389 | } |
1517 | } |
1390 | |
1518 | |
1391 | static SV * |
1519 | static SV * |
1392 | decode_json (SV *string, JSON *json, STRLEN *offset_return) |
1520 | decode_json (SV *string, JSON *json, char **offset_return) |
1393 | { |
1521 | { |
1394 | dec_t dec; |
1522 | dec_t dec; |
1395 | STRLEN offset; |
|
|
1396 | SV *sv; |
1523 | SV *sv; |
1397 | |
1524 | |
|
|
1525 | /* work around bugs in 5.10 where manipulating magic values |
|
|
1526 | * will perl ignore the magic in subsequent accesses. |
|
|
1527 | * also make a copy of non-PV values, to get them into a clean |
|
|
1528 | * state (SvPV should do that, but it's buggy, see below). |
|
|
1529 | */ |
1398 | SvGETMAGIC (string); |
1530 | /*SvGETMAGIC (string);*/ |
|
|
1531 | if (SvMAGICAL (string) || !SvPOK (string)) |
|
|
1532 | string = sv_2mortal (newSVsv (string)); |
|
|
1533 | |
1399 | SvUPGRADE (string, SVt_PV); |
1534 | SvUPGRADE (string, SVt_PV); |
1400 | |
1535 | |
1401 | if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags)) |
1536 | /* work around a bug in perl 5.10, which causes SvCUR to fail an |
|
|
1537 | * assertion with -DDEBUGGING, although SvCUR is documented to |
|
|
1538 | * return the xpv_cur field which certainly exists after upgrading. |
|
|
1539 | * according to nicholas clark, calling SvPOK fixes this. |
|
|
1540 | * But it doesn't fix it, so try another workaround, call SvPV_nolen |
|
|
1541 | * and hope for the best. |
|
|
1542 | * Damnit, SvPV_nolen still trips over yet another assertion. This |
|
|
1543 | * assertion business is seriously broken, try yet another workaround |
|
|
1544 | * for the broken -DDEBUGGING. |
|
|
1545 | */ |
|
|
1546 | { |
|
|
1547 | #ifdef DEBUGGING |
|
|
1548 | STRLEN offset = SvOK (string) ? sv_len (string) : 0; |
|
|
1549 | #else |
|
|
1550 | STRLEN offset = SvCUR (string); |
|
|
1551 | #endif |
|
|
1552 | |
|
|
1553 | if (offset > json->max_size && json->max_size) |
1402 | croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", |
1554 | croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", |
1403 | (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags)); |
1555 | (unsigned long)SvCUR (string), (unsigned long)json->max_size); |
|
|
1556 | } |
1404 | |
1557 | |
1405 | if (json->flags & F_UTF8) |
1558 | if (DECODE_WANTS_OCTETS (json)) |
1406 | sv_utf8_downgrade (string, 0); |
1559 | sv_utf8_downgrade (string, 0); |
1407 | else |
1560 | else |
1408 | sv_utf8_upgrade (string); |
1561 | sv_utf8_upgrade (string); |
1409 | |
1562 | |
1410 | SvGROW (string, SvCUR (string) + 1); // should basically be a NOP |
1563 | SvGROW (string, SvCUR (string) + 1); // should basically be a NOP |
1411 | |
1564 | |
1412 | dec.json = *json; |
1565 | dec.json = *json; |
1413 | dec.cur = SvPVX (string); |
1566 | dec.cur = SvPVX (string); |
1414 | dec.end = SvEND (string); |
1567 | dec.end = SvEND (string); |
1415 | dec.err = 0; |
1568 | dec.err = 0; |
1416 | dec.depth = 0; |
1569 | dec.depth = 0; |
1417 | dec.maxdepth = DEC_DEPTH (dec.json.flags); |
|
|
1418 | |
1570 | |
1419 | if (dec.json.cb_object || dec.json.cb_sk_object) |
1571 | if (dec.json.cb_object || dec.json.cb_sk_object) |
1420 | dec.json.flags |= F_HOOK; |
1572 | dec.json.flags |= F_HOOK; |
1421 | |
1573 | |
1422 | *dec.end = 0; // this should basically be a nop, too, but make sure it's there |
1574 | *dec.end = 0; // this should basically be a nop, too, but make sure it's there |
1423 | |
1575 | |
1424 | decode_ws (&dec); |
1576 | decode_ws (&dec); |
1425 | sv = decode_sv (&dec); |
1577 | sv = decode_sv (&dec); |
|
|
1578 | |
|
|
1579 | if (offset_return) |
|
|
1580 | *offset_return = dec.cur; |
1426 | |
1581 | |
1427 | if (!(offset_return || !sv)) |
1582 | if (!(offset_return || !sv)) |
1428 | { |
1583 | { |
1429 | // check for trailing garbage |
1584 | // check for trailing garbage |
1430 | decode_ws (&dec); |
1585 | decode_ws (&dec); |
… | |
… | |
1433 | { |
1588 | { |
1434 | dec.err = "garbage after JSON object"; |
1589 | dec.err = "garbage after JSON object"; |
1435 | SvREFCNT_dec (sv); |
1590 | SvREFCNT_dec (sv); |
1436 | sv = 0; |
1591 | sv = 0; |
1437 | } |
1592 | } |
1438 | } |
|
|
1439 | |
|
|
1440 | if (offset_return || !sv) |
|
|
1441 | { |
|
|
1442 | offset = dec.json.flags & F_UTF8 |
|
|
1443 | ? dec.cur - SvPVX (string) |
|
|
1444 | : utf8_distance (dec.cur, SvPVX (string)); |
|
|
1445 | |
|
|
1446 | if (offset_return) |
|
|
1447 | *offset_return = offset; |
|
|
1448 | } |
1593 | } |
1449 | |
1594 | |
1450 | if (!sv) |
1595 | if (!sv) |
1451 | { |
1596 | { |
1452 | SV *uni = sv_newmortal (); |
1597 | SV *uni = sv_newmortal (); |
… | |
… | |
1458 | SAVEVPTR (PL_curcop); |
1603 | SAVEVPTR (PL_curcop); |
1459 | PL_curcop = &cop; |
1604 | PL_curcop = &cop; |
1460 | pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); |
1605 | pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); |
1461 | LEAVE; |
1606 | LEAVE; |
1462 | |
1607 | |
1463 | croak ("%s, at character offset %d [\"%s\"]", |
1608 | croak ("%s, at character offset %d (before \"%s\")", |
1464 | dec.err, |
1609 | dec.err, |
1465 | (int)offset, |
1610 | ptr_to_index (string, dec.cur), |
1466 | dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); |
1611 | dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); |
1467 | } |
1612 | } |
1468 | |
1613 | |
1469 | sv = sv_2mortal (sv); |
1614 | sv = sv_2mortal (sv); |
1470 | |
1615 | |
… | |
… | |
1479 | |
1624 | |
1480 | static void |
1625 | static void |
1481 | incr_parse (JSON *self) |
1626 | incr_parse (JSON *self) |
1482 | { |
1627 | { |
1483 | const char *p = SvPVX (self->incr_text) + self->incr_pos; |
1628 | const char *p = SvPVX (self->incr_text) + self->incr_pos; |
|
|
1629 | |
|
|
1630 | // the state machine here is a bit convoluted and could be simplified a lot |
|
|
1631 | // but this would make it slower, so... |
1484 | |
1632 | |
1485 | for (;;) |
1633 | for (;;) |
1486 | { |
1634 | { |
1487 | //printf ("loop pod %d *p<%c><%s>, mode %d nest %d\n", p - SvPVX (self->incr_text), *p, p, self->incr_mode, self->incr_nest);//D |
1635 | //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 |
1488 | switch (self->incr_mode) |
1636 | switch (self->incr_mode) |
1489 | { |
1637 | { |
1490 | // only used for intiial whitespace skipping |
1638 | // only used for initial whitespace skipping |
1491 | case INCR_M_WS: |
1639 | case INCR_M_WS: |
1492 | for (;;) |
1640 | for (;;) |
1493 | { |
1641 | { |
1494 | if (*p > 0x20) |
1642 | if (*p > 0x20) |
1495 | { |
1643 | { |
|
|
1644 | if (*p == '#') |
|
|
1645 | { |
|
|
1646 | self->incr_mode = INCR_M_C0; |
|
|
1647 | goto incr_m_c; |
|
|
1648 | } |
|
|
1649 | else |
|
|
1650 | { |
1496 | self->incr_mode = INCR_M_JSON; |
1651 | self->incr_mode = INCR_M_JSON; |
1497 | goto incr_m_json; |
1652 | goto incr_m_json; |
|
|
1653 | } |
1498 | } |
1654 | } |
1499 | else if (!*p) |
1655 | else if (!*p) |
1500 | goto interrupt; |
1656 | goto interrupt; |
1501 | |
1657 | |
1502 | ++p; |
1658 | ++p; |
… | |
… | |
1508 | goto interrupt; |
1664 | goto interrupt; |
1509 | |
1665 | |
1510 | ++p; |
1666 | ++p; |
1511 | self->incr_mode = INCR_M_STR; |
1667 | self->incr_mode = INCR_M_STR; |
1512 | goto incr_m_str; |
1668 | goto incr_m_str; |
|
|
1669 | |
|
|
1670 | // inside #-style comments |
|
|
1671 | case INCR_M_C0: |
|
|
1672 | case INCR_M_C1: |
|
|
1673 | incr_m_c: |
|
|
1674 | for (;;) |
|
|
1675 | { |
|
|
1676 | if (*p == '\n') |
|
|
1677 | { |
|
|
1678 | self->incr_mode = self->incr_mode == INCR_M_C0 ? INCR_M_WS : INCR_M_JSON; |
|
|
1679 | break; |
|
|
1680 | } |
|
|
1681 | else if (!*p) |
|
|
1682 | goto interrupt; |
|
|
1683 | |
|
|
1684 | ++p; |
|
|
1685 | } |
|
|
1686 | |
|
|
1687 | break; |
1513 | |
1688 | |
1514 | // inside a string |
1689 | // inside a string |
1515 | case INCR_M_STR: |
1690 | case INCR_M_STR: |
1516 | incr_m_str: |
1691 | incr_m_str: |
1517 | for (;;) |
1692 | for (;;) |
… | |
… | |
1568 | self->incr_mode = INCR_M_STR; |
1743 | self->incr_mode = INCR_M_STR; |
1569 | goto incr_m_str; |
1744 | goto incr_m_str; |
1570 | |
1745 | |
1571 | case '[': |
1746 | case '[': |
1572 | case '{': |
1747 | case '{': |
1573 | ++self->incr_nest; |
1748 | if (++self->incr_nest > self->max_depth) |
|
|
1749 | croak (ERR_NESTING_EXCEEDED); |
1574 | break; |
1750 | break; |
1575 | |
1751 | |
1576 | case ']': |
1752 | case ']': |
1577 | case '}': |
1753 | case '}': |
1578 | if (!--self->incr_nest) |
1754 | if (--self->incr_nest <= 0) |
1579 | goto interrupt; |
1755 | goto interrupt; |
|
|
1756 | break; |
|
|
1757 | |
|
|
1758 | case '#': |
|
|
1759 | self->incr_mode = INCR_M_C1; |
|
|
1760 | goto incr_m_c; |
1580 | } |
1761 | } |
1581 | } |
1762 | } |
1582 | } |
1763 | } |
1583 | |
1764 | |
1584 | modechange: |
1765 | modechange: |
1585 | ; |
1766 | ; |
1586 | } |
1767 | } |
1587 | |
1768 | |
1588 | interrupt: |
1769 | interrupt: |
1589 | self->incr_pos = p - SvPVX (self->incr_text); |
1770 | self->incr_pos = p - SvPVX (self->incr_text); |
|
|
1771 | //printf ("interrupt<%.*s>\n", self->incr_pos, SvPVX(self->incr_text));//D |
1590 | //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D |
1772 | //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D |
1591 | } |
1773 | } |
1592 | |
1774 | |
1593 | ///////////////////////////////////////////////////////////////////////////// |
1775 | ///////////////////////////////////////////////////////////////////////////// |
1594 | // XS interface functions |
1776 | // XS interface functions |
… | |
… | |
1607 | : -1; |
1789 | : -1; |
1608 | |
1790 | |
1609 | json_stash = gv_stashpv ("JSON::XS" , 1); |
1791 | json_stash = gv_stashpv ("JSON::XS" , 1); |
1610 | json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); |
1792 | json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); |
1611 | |
1793 | |
1612 | json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true ); |
1794 | json_true = get_bool ("JSON::XS::true"); |
1613 | json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); |
1795 | json_false = get_bool ("JSON::XS::false"); |
|
|
1796 | |
|
|
1797 | CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */ |
1614 | } |
1798 | } |
1615 | |
1799 | |
1616 | PROTOTYPES: DISABLE |
1800 | PROTOTYPES: DISABLE |
1617 | |
1801 | |
1618 | void CLONE (...) |
1802 | void CLONE (...) |
… | |
… | |
1623 | void new (char *klass) |
1807 | void new (char *klass) |
1624 | PPCODE: |
1808 | PPCODE: |
1625 | { |
1809 | { |
1626 | SV *pv = NEWSV (0, sizeof (JSON)); |
1810 | SV *pv = NEWSV (0, sizeof (JSON)); |
1627 | SvPOK_only (pv); |
1811 | SvPOK_only (pv); |
1628 | Zero (SvPVX (pv), 1, JSON); |
1812 | json_init ((JSON *)SvPVX (pv)); |
1629 | ((JSON *)SvPVX (pv))->flags = F_DEFAULT; |
|
|
1630 | XPUSHs (sv_2mortal (sv_bless ( |
1813 | XPUSHs (sv_2mortal (sv_bless ( |
1631 | newRV_noinc (pv), |
1814 | newRV_noinc (pv), |
1632 | strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1) |
1815 | strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1) |
1633 | ))); |
1816 | ))); |
1634 | } |
1817 | } |
… | |
… | |
1646 | allow_nonref = F_ALLOW_NONREF |
1829 | allow_nonref = F_ALLOW_NONREF |
1647 | shrink = F_SHRINK |
1830 | shrink = F_SHRINK |
1648 | allow_blessed = F_ALLOW_BLESSED |
1831 | allow_blessed = F_ALLOW_BLESSED |
1649 | convert_blessed = F_CONV_BLESSED |
1832 | convert_blessed = F_CONV_BLESSED |
1650 | relaxed = F_RELAXED |
1833 | relaxed = F_RELAXED |
|
|
1834 | allow_unknown = F_ALLOW_UNKNOWN |
1651 | PPCODE: |
1835 | PPCODE: |
1652 | { |
1836 | { |
1653 | if (enable) |
1837 | if (enable) |
1654 | self->flags |= ix; |
1838 | self->flags |= ix; |
1655 | else |
1839 | else |
… | |
… | |
1670 | get_allow_nonref = F_ALLOW_NONREF |
1854 | get_allow_nonref = F_ALLOW_NONREF |
1671 | get_shrink = F_SHRINK |
1855 | get_shrink = F_SHRINK |
1672 | get_allow_blessed = F_ALLOW_BLESSED |
1856 | get_allow_blessed = F_ALLOW_BLESSED |
1673 | get_convert_blessed = F_CONV_BLESSED |
1857 | get_convert_blessed = F_CONV_BLESSED |
1674 | get_relaxed = F_RELAXED |
1858 | get_relaxed = F_RELAXED |
|
|
1859 | get_allow_unknown = F_ALLOW_UNKNOWN |
1675 | PPCODE: |
1860 | PPCODE: |
1676 | XPUSHs (boolSV (self->flags & ix)); |
1861 | XPUSHs (boolSV (self->flags & ix)); |
1677 | |
1862 | |
1678 | void max_depth (JSON *self, UV max_depth = 0x80000000UL) |
1863 | void max_depth (JSON *self, U32 max_depth = 0x80000000UL) |
1679 | PPCODE: |
1864 | PPCODE: |
1680 | { |
1865 | self->max_depth = max_depth; |
1681 | UV log2 = 0; |
|
|
1682 | |
|
|
1683 | if (max_depth > 0x80000000UL) max_depth = 0x80000000UL; |
|
|
1684 | |
|
|
1685 | while ((1UL << log2) < max_depth) |
|
|
1686 | ++log2; |
|
|
1687 | |
|
|
1688 | self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH); |
|
|
1689 | |
|
|
1690 | XPUSHs (ST (0)); |
1866 | XPUSHs (ST (0)); |
1691 | } |
|
|
1692 | |
1867 | |
1693 | U32 get_max_depth (JSON *self) |
1868 | U32 get_max_depth (JSON *self) |
1694 | CODE: |
1869 | CODE: |
1695 | RETVAL = DEC_DEPTH (self->flags); |
1870 | RETVAL = self->max_depth; |
1696 | OUTPUT: |
1871 | OUTPUT: |
1697 | RETVAL |
1872 | RETVAL |
1698 | |
1873 | |
1699 | void max_size (JSON *self, UV max_size = 0) |
1874 | void max_size (JSON *self, U32 max_size = 0) |
1700 | PPCODE: |
1875 | PPCODE: |
1701 | { |
1876 | self->max_size = max_size; |
1702 | UV log2 = 0; |
|
|
1703 | |
|
|
1704 | if (max_size > 0x80000000UL) max_size = 0x80000000UL; |
|
|
1705 | if (max_size == 1) max_size = 2; |
|
|
1706 | |
|
|
1707 | while ((1UL << log2) < max_size) |
|
|
1708 | ++log2; |
|
|
1709 | |
|
|
1710 | self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE); |
|
|
1711 | |
|
|
1712 | XPUSHs (ST (0)); |
1877 | XPUSHs (ST (0)); |
1713 | } |
|
|
1714 | |
1878 | |
1715 | int get_max_size (JSON *self) |
1879 | int get_max_size (JSON *self) |
1716 | CODE: |
1880 | CODE: |
1717 | RETVAL = DEC_SIZE (self->flags); |
1881 | RETVAL = self->max_size; |
1718 | OUTPUT: |
1882 | OUTPUT: |
1719 | RETVAL |
1883 | RETVAL |
1720 | |
1884 | |
1721 | void filter_json_object (JSON *self, SV *cb = &PL_sv_undef) |
1885 | void filter_json_object (JSON *self, SV *cb = &PL_sv_undef) |
1722 | PPCODE: |
1886 | PPCODE: |
… | |
… | |
1758 | XPUSHs (decode_json (jsonstr, self, 0)); |
1922 | XPUSHs (decode_json (jsonstr, self, 0)); |
1759 | |
1923 | |
1760 | void decode_prefix (JSON *self, SV *jsonstr) |
1924 | void decode_prefix (JSON *self, SV *jsonstr) |
1761 | PPCODE: |
1925 | PPCODE: |
1762 | { |
1926 | { |
1763 | STRLEN offset; |
1927 | char *offset; |
1764 | EXTEND (SP, 2); |
1928 | EXTEND (SP, 2); |
1765 | PUSHs (decode_json (jsonstr, self, &offset)); |
1929 | PUSHs (decode_json (jsonstr, self, &offset)); |
1766 | PUSHs (sv_2mortal (newSVuv (offset))); |
1930 | PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset)))); |
1767 | } |
1931 | } |
1768 | |
1932 | |
1769 | void incr_parse (JSON *self, SV *jsonstr = 0) |
1933 | void incr_parse (JSON *self, SV *jsonstr = 0) |
1770 | PPCODE: |
1934 | PPCODE: |
1771 | { |
1935 | { |
1772 | if (!self->incr_text) |
1936 | if (!self->incr_text) |
1773 | self->incr_text = newSVpvn ("", 0); |
1937 | self->incr_text = newSVpvn ("", 0); |
|
|
1938 | |
|
|
1939 | /* if utf8-ness doesn't match the decoder, need to upgrade/downgrade */ |
|
|
1940 | if (!DECODE_WANTS_OCTETS (self) == !SvUTF8 (self->incr_text)) |
|
|
1941 | if (DECODE_WANTS_OCTETS (self)) |
|
|
1942 | { |
|
|
1943 | if (self->incr_pos) |
|
|
1944 | self->incr_pos = utf8_length ((U8 *)SvPVX (self->incr_text), |
|
|
1945 | (U8 *)SvPVX (self->incr_text) + self->incr_pos); |
|
|
1946 | |
|
|
1947 | sv_utf8_downgrade (self->incr_text, 0); |
|
|
1948 | } |
|
|
1949 | else |
|
|
1950 | { |
|
|
1951 | sv_utf8_upgrade (self->incr_text); |
|
|
1952 | |
|
|
1953 | if (self->incr_pos) |
|
|
1954 | self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos) |
|
|
1955 | - (U8 *)SvPVX (self->incr_text); |
|
|
1956 | } |
1774 | |
1957 | |
1775 | // append data, if any |
1958 | // append data, if any |
1776 | if (jsonstr) |
1959 | if (jsonstr) |
1777 | { |
1960 | { |
|
|
1961 | /* make sure both strings have same encoding */ |
1778 | if (SvUTF8 (jsonstr) && !SvUTF8 (self->incr_text)) |
1962 | if (SvUTF8 (jsonstr) != SvUTF8 (self->incr_text)) |
|
|
1963 | if (SvUTF8 (jsonstr)) |
|
|
1964 | sv_utf8_downgrade (jsonstr, 0); |
1779 | { |
1965 | else |
1780 | /* utf-8-ness differs, need to upgrade */ |
|
|
1781 | sv_utf8_upgrade (self->incr_text); |
1966 | sv_utf8_upgrade (jsonstr); |
1782 | |
1967 | |
1783 | if (self->incr_pos) |
1968 | /* and then just blindly append */ |
1784 | self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos) |
|
|
1785 | - (U8 *)SvPVX (self->incr_text); |
|
|
1786 | } |
|
|
1787 | |
|
|
1788 | { |
1969 | { |
1789 | STRLEN len; |
1970 | STRLEN len; |
1790 | const char *str = SvPV (jsonstr, len); |
1971 | const char *str = SvPV (jsonstr, len); |
1791 | SvGROW (self->incr_text, SvCUR (self->incr_text) + len + 1); |
1972 | STRLEN cur = SvCUR (self->incr_text); |
|
|
1973 | |
|
|
1974 | if (SvLEN (self->incr_text) <= cur + len) |
|
|
1975 | SvGROW (self->incr_text, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); |
|
|
1976 | |
1792 | Move (str, SvEND (self->incr_text), len, char); |
1977 | Move (str, SvEND (self->incr_text), len, char); |
1793 | SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len); |
1978 | SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len); |
1794 | *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there |
1979 | *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there |
1795 | } |
1980 | } |
1796 | } |
1981 | } |
1797 | |
1982 | |
1798 | if (GIMME_V != G_VOID) |
1983 | if (GIMME_V != G_VOID) |
1799 | do |
1984 | do |
1800 | { |
1985 | { |
1801 | STRLEN offset; |
1986 | char *offset; |
1802 | |
|
|
1803 | incr_parse (self); |
|
|
1804 | |
1987 | |
1805 | if (!INCR_DONE (self)) |
1988 | if (!INCR_DONE (self)) |
|
|
1989 | { |
|
|
1990 | incr_parse (self); |
|
|
1991 | |
|
|
1992 | if (self->incr_pos > self->max_size && self->max_size) |
|
|
1993 | croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", |
|
|
1994 | (unsigned long)self->incr_pos, (unsigned long)self->max_size); |
|
|
1995 | |
|
|
1996 | if (!INCR_DONE (self)) |
1806 | break; |
1997 | break; |
|
|
1998 | } |
1807 | |
1999 | |
1808 | XPUSHs (decode_json (self->incr_text, self, &offset)); |
2000 | XPUSHs (decode_json (self->incr_text, self, &offset)); |
1809 | |
2001 | |
1810 | sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + offset); |
|
|
1811 | self->incr_pos -= offset; |
2002 | self->incr_pos -= offset - SvPVX (self->incr_text); |
1812 | self->incr_nest = 0; |
2003 | self->incr_nest = 0; |
1813 | self->incr_mode = 0; |
2004 | self->incr_mode = 0; |
|
|
2005 | |
|
|
2006 | sv_chop (self->incr_text, offset); |
1814 | } |
2007 | } |
1815 | while (GIMME_V == G_ARRAY); |
2008 | while (GIMME_V == G_ARRAY); |
1816 | } |
2009 | } |
1817 | |
2010 | |
1818 | SV *incr_text (JSON *self) |
2011 | SV *incr_text (JSON *self) |
1819 | ATTRS: lvalue |
2012 | ATTRS: lvalue |
1820 | CODE: |
2013 | CODE: |
1821 | { |
2014 | { |
1822 | if (self->incr_pos) |
2015 | if (self->incr_pos) |
1823 | croak ("incr_text can only be called after a successful incr_parse call in scalar context"); |
2016 | croak ("incr_text can not be called when the incremental parser already started parsing"); |
1824 | |
2017 | |
1825 | RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef; |
2018 | RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef; |
1826 | } |
2019 | } |
1827 | OUTPUT: |
2020 | OUTPUT: |
1828 | RETVAL |
2021 | RETVAL |
… | |
… | |
1837 | self->incr_nest = 0; |
2030 | self->incr_nest = 0; |
1838 | self->incr_mode = 0; |
2031 | self->incr_mode = 0; |
1839 | } |
2032 | } |
1840 | } |
2033 | } |
1841 | |
2034 | |
|
|
2035 | void incr_reset (JSON *self) |
|
|
2036 | CODE: |
|
|
2037 | { |
|
|
2038 | SvREFCNT_dec (self->incr_text); |
|
|
2039 | self->incr_text = 0; |
|
|
2040 | self->incr_pos = 0; |
|
|
2041 | self->incr_nest = 0; |
|
|
2042 | self->incr_mode = 0; |
|
|
2043 | } |
|
|
2044 | |
1842 | void DESTROY (JSON *self) |
2045 | void DESTROY (JSON *self) |
1843 | CODE: |
2046 | CODE: |
1844 | SvREFCNT_dec (self->cb_sk_object); |
2047 | SvREFCNT_dec (self->cb_sk_object); |
1845 | SvREFCNT_dec (self->cb_object); |
2048 | SvREFCNT_dec (self->cb_object); |
1846 | SvREFCNT_dec (self->incr_text); |
2049 | SvREFCNT_dec (self->incr_text); |
… | |
… | |
1851 | ALIAS: |
2054 | ALIAS: |
1852 | to_json_ = 0 |
2055 | to_json_ = 0 |
1853 | encode_json = F_UTF8 |
2056 | encode_json = F_UTF8 |
1854 | PPCODE: |
2057 | PPCODE: |
1855 | { |
2058 | { |
1856 | JSON json = { F_DEFAULT | ix }; |
2059 | JSON json; |
|
|
2060 | json_init (&json); |
|
|
2061 | json.flags |= ix; |
1857 | XPUSHs (encode_json (scalar, &json)); |
2062 | XPUSHs (encode_json (scalar, &json)); |
1858 | } |
2063 | } |
1859 | |
2064 | |
1860 | void decode_json (SV *jsonstr) |
2065 | void decode_json (SV *jsonstr) |
1861 | ALIAS: |
2066 | ALIAS: |
1862 | from_json_ = 0 |
2067 | from_json_ = 0 |
1863 | decode_json = F_UTF8 |
2068 | decode_json = F_UTF8 |
1864 | PPCODE: |
2069 | PPCODE: |
1865 | { |
2070 | { |
1866 | JSON json = { F_DEFAULT | ix }; |
2071 | JSON json; |
|
|
2072 | json_init (&json); |
|
|
2073 | json.flags |= ix; |
1867 | XPUSHs (decode_json (jsonstr, &json, 0)); |
2074 | XPUSHs (decode_json (jsonstr, &json, 0)); |
1868 | } |
2075 | } |
1869 | |
2076 | |
1870 | |
|
|