… | |
… | |
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 | // compatibility with perl <5.18 |
|
|
23 | #ifndef HvNAMELEN_get |
|
|
24 | # define HvNAMELEN_get(hv) strlen (HvNAME (hv)) |
|
|
25 | #endif |
|
|
26 | #ifndef HvNAMELEN |
|
|
27 | # define HvNAMELEN(hv) HvNAMELEN_get (hv) |
|
|
28 | #endif |
|
|
29 | #ifndef HvNAMEUTF8 |
|
|
30 | # define HvNAMEUTF8(hv) 0 |
|
|
31 | #endif |
|
|
32 | |
|
|
33 | // three extra for rounding, sign, and end of string |
22 | #define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 2) |
34 | #define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 3) |
23 | |
35 | |
24 | #define F_ASCII 0x00000001UL |
36 | #define F_ASCII 0x00000001UL |
25 | #define F_LATIN1 0x00000002UL |
37 | #define F_LATIN1 0x00000002UL |
26 | #define F_UTF8 0x00000004UL |
38 | #define F_UTF8 0x00000004UL |
27 | #define F_INDENT 0x00000008UL |
39 | #define F_INDENT 0x00000008UL |
… | |
… | |
31 | #define F_ALLOW_NONREF 0x00000100UL |
43 | #define F_ALLOW_NONREF 0x00000100UL |
32 | #define F_SHRINK 0x00000200UL |
44 | #define F_SHRINK 0x00000200UL |
33 | #define F_ALLOW_BLESSED 0x00000400UL |
45 | #define F_ALLOW_BLESSED 0x00000400UL |
34 | #define F_CONV_BLESSED 0x00000800UL |
46 | #define F_CONV_BLESSED 0x00000800UL |
35 | #define F_RELAXED 0x00001000UL |
47 | #define F_RELAXED 0x00001000UL |
36 | |
48 | #define F_ALLOW_UNKNOWN 0x00002000UL |
37 | #define F_MAXDEPTH 0xf8000000UL |
49 | #define F_ALLOW_TAGS 0x00004000UL |
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 |
50 | #define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing |
42 | |
51 | |
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 |
52 | #define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER |
47 | #define F_DEFAULT (9UL << S_MAXDEPTH) |
|
|
48 | |
53 | |
49 | #define INIT_SIZE 32 // initial scalar size to be allocated |
54 | #define INIT_SIZE 32 // initial scalar size to be allocated |
50 | #define INDENT_STEP 3 // spaces per indentation level |
55 | #define INDENT_STEP 3 // spaces per indentation level |
51 | |
56 | |
52 | #define SHORT_STRING_LEN 16384 // special-case strings of up to this size |
57 | #define SHORT_STRING_LEN 16384 // special-case strings of up to this size |
|
|
58 | |
|
|
59 | #define DECODE_WANTS_OCTETS(json) ((json)->flags & F_UTF8) |
53 | |
60 | |
54 | #define SB do { |
61 | #define SB do { |
55 | #define SE } while (0) |
62 | #define SE } while (0) |
56 | |
63 | |
57 | #if __GNUC__ >= 3 |
64 | #if __GNUC__ >= 3 |
… | |
… | |
67 | |
74 | |
68 | #define IN_RANGE_INC(type,val,beg,end) \ |
75 | #define IN_RANGE_INC(type,val,beg,end) \ |
69 | ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \ |
76 | ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \ |
70 | <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg))) |
77 | <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg))) |
71 | |
78 | |
|
|
79 | #define ERR_NESTING_EXCEEDED "json text or perl structure exceeds maximum nesting level (max_depth set too low?)" |
|
|
80 | |
72 | #ifdef USE_ITHREADS |
81 | #ifdef USE_ITHREADS |
73 | # define JSON_SLOW 1 |
82 | # define JSON_SLOW 1 |
74 | # define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1)) |
83 | # define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1)) |
|
|
84 | # define BOOL_STASH (bool_stash ? bool_stash : gv_stashpv ("Types::Serialiser::Boolean", 1)) |
75 | #else |
85 | #else |
76 | # define JSON_SLOW 0 |
86 | # define JSON_SLOW 0 |
77 | # define JSON_STASH json_stash |
87 | # define JSON_STASH json_stash |
|
|
88 | # define BOOL_STASH bool_stash |
78 | #endif |
89 | #endif |
79 | |
90 | |
80 | static HV *json_stash, *json_boolean_stash; // JSON::XS:: |
91 | // the amount of HEs to allocate on the stack, when sorting keys |
81 | static SV *json_true, *json_false; |
92 | #define STACK_HES 64 |
|
|
93 | |
|
|
94 | static HV *json_stash, *bool_stash; // JSON::XS::, Types::Serialiser::Boolean:: |
|
|
95 | static SV *bool_true, *bool_false, *sv_json; |
|
|
96 | |
|
|
97 | enum { |
|
|
98 | INCR_M_WS = 0, // initial whitespace skipping, must be 0 |
|
|
99 | INCR_M_STR, // inside string |
|
|
100 | INCR_M_BS, // inside backslash |
|
|
101 | INCR_M_C0, // inside comment in initial whitespace sequence |
|
|
102 | INCR_M_C1, // inside comment in other places |
|
|
103 | INCR_M_JSON // outside anything, count nesting |
|
|
104 | }; |
|
|
105 | |
|
|
106 | #define INCR_DONE(json) ((json)->incr_nest <= 0 && (json)->incr_mode == INCR_M_JSON) |
82 | |
107 | |
83 | typedef struct { |
108 | typedef struct { |
84 | U32 flags; |
109 | U32 flags; |
|
|
110 | U32 max_depth; |
|
|
111 | STRLEN max_size; |
|
|
112 | |
85 | SV *cb_object; |
113 | SV *cb_object; |
86 | HV *cb_sk_object; |
114 | HV *cb_sk_object; |
|
|
115 | |
|
|
116 | // for the incremental parser |
|
|
117 | SV *incr_text; // the source text so far |
|
|
118 | STRLEN incr_pos; // the current offset into the text |
|
|
119 | int incr_nest; // {[]}-nesting level |
|
|
120 | unsigned char incr_mode; |
87 | } JSON; |
121 | } JSON; |
|
|
122 | |
|
|
123 | INLINE void |
|
|
124 | json_init (JSON *json) |
|
|
125 | { |
|
|
126 | Zero (json, 1, JSON); |
|
|
127 | json->max_depth = 512; |
|
|
128 | } |
88 | |
129 | |
89 | ///////////////////////////////////////////////////////////////////////////// |
130 | ///////////////////////////////////////////////////////////////////////////// |
90 | // utility functions |
131 | // utility functions |
91 | |
132 | |
|
|
133 | INLINE SV * |
|
|
134 | get_bool (const char *name) |
|
|
135 | { |
|
|
136 | SV *sv = get_sv (name, 1); |
|
|
137 | |
|
|
138 | SvREADONLY_on (sv); |
|
|
139 | SvREADONLY_on (SvRV (sv)); |
|
|
140 | |
|
|
141 | return sv; |
|
|
142 | } |
|
|
143 | |
92 | INLINE void |
144 | INLINE void |
93 | shrink (SV *sv) |
145 | shrink (SV *sv) |
94 | { |
146 | { |
95 | sv_utf8_downgrade (sv, 1); |
147 | sv_utf8_downgrade (sv, 1); |
|
|
148 | |
96 | if (SvLEN (sv) > SvCUR (sv) + 1) |
149 | if (SvLEN (sv) > SvCUR (sv) + 1) |
97 | { |
150 | { |
98 | #ifdef SvPV_shrink_to_cur |
151 | #ifdef SvPV_shrink_to_cur |
99 | SvPV_shrink_to_cur (sv); |
152 | SvPV_shrink_to_cur (sv); |
100 | #elif defined (SvPV_renew) |
153 | #elif defined (SvPV_renew) |
… | |
… | |
144 | *s++ = 0x80 | ( ch & 0x3f); |
197 | *s++ = 0x80 | ( ch & 0x3f); |
145 | |
198 | |
146 | return s; |
199 | return s; |
147 | } |
200 | } |
148 | |
201 | |
|
|
202 | // convert offset pointer to character index, sv must be string |
|
|
203 | static STRLEN |
|
|
204 | ptr_to_index (SV *sv, char *offset) |
|
|
205 | { |
|
|
206 | return SvUTF8 (sv) |
|
|
207 | ? utf8_distance (offset, SvPVX (sv)) |
|
|
208 | : offset - SvPVX (sv); |
|
|
209 | } |
|
|
210 | |
|
|
211 | ///////////////////////////////////////////////////////////////////////////// |
|
|
212 | // fp hell |
|
|
213 | |
|
|
214 | // scan a group of digits, and a trailing exponent |
|
|
215 | static void |
|
|
216 | json_atof_scan1 (const char *s, NV *accum, int *expo, int postdp, int maxdepth) |
|
|
217 | { |
|
|
218 | UV uaccum = 0; |
|
|
219 | int eaccum = 0; |
|
|
220 | |
|
|
221 | // if we recurse too deep, skip all remaining digits |
|
|
222 | // to avoid a stack overflow attack |
|
|
223 | if (expect_false (--maxdepth <= 0)) |
|
|
224 | while (((U8)*s - '0') < 10) |
|
|
225 | ++s; |
|
|
226 | |
|
|
227 | for (;;) |
|
|
228 | { |
|
|
229 | U8 dig = (U8)*s - '0'; |
|
|
230 | |
|
|
231 | if (expect_false (dig >= 10)) |
|
|
232 | { |
|
|
233 | if (dig == (U8)((U8)'.' - (U8)'0')) |
|
|
234 | { |
|
|
235 | ++s; |
|
|
236 | json_atof_scan1 (s, accum, expo, 1, maxdepth); |
|
|
237 | } |
|
|
238 | else if ((dig | ' ') == 'e' - '0') |
|
|
239 | { |
|
|
240 | int exp2 = 0; |
|
|
241 | int neg = 0; |
|
|
242 | |
|
|
243 | ++s; |
|
|
244 | |
|
|
245 | if (*s == '-') |
|
|
246 | { |
|
|
247 | ++s; |
|
|
248 | neg = 1; |
|
|
249 | } |
|
|
250 | else if (*s == '+') |
|
|
251 | ++s; |
|
|
252 | |
|
|
253 | while ((dig = (U8)*s - '0') < 10) |
|
|
254 | exp2 = exp2 * 10 + *s++ - '0'; |
|
|
255 | |
|
|
256 | *expo += neg ? -exp2 : exp2; |
|
|
257 | } |
|
|
258 | |
|
|
259 | break; |
|
|
260 | } |
|
|
261 | |
|
|
262 | ++s; |
|
|
263 | |
|
|
264 | uaccum = uaccum * 10 + dig; |
|
|
265 | ++eaccum; |
|
|
266 | |
|
|
267 | // if we have too many digits, then recurse for more |
|
|
268 | // we actually do this for rather few digits |
|
|
269 | if (uaccum >= (UV_MAX - 9) / 10) |
|
|
270 | { |
|
|
271 | if (postdp) *expo -= eaccum; |
|
|
272 | json_atof_scan1 (s, accum, expo, postdp, maxdepth); |
|
|
273 | if (postdp) *expo += eaccum; |
|
|
274 | |
|
|
275 | break; |
|
|
276 | } |
|
|
277 | } |
|
|
278 | |
|
|
279 | // this relies greatly on the quality of the pow () |
|
|
280 | // implementation of the platform, but a good |
|
|
281 | // implementation is hard to beat. |
|
|
282 | // (IEEE 754 conformant ones are required to be exact) |
|
|
283 | if (postdp) *expo -= eaccum; |
|
|
284 | *accum += uaccum * Perl_pow (10., *expo); |
|
|
285 | *expo += eaccum; |
|
|
286 | } |
|
|
287 | |
|
|
288 | static NV |
|
|
289 | json_atof (const char *s) |
|
|
290 | { |
|
|
291 | NV accum = 0.; |
|
|
292 | int expo = 0; |
|
|
293 | int neg = 0; |
|
|
294 | |
|
|
295 | if (*s == '-') |
|
|
296 | { |
|
|
297 | ++s; |
|
|
298 | neg = 1; |
|
|
299 | } |
|
|
300 | |
|
|
301 | // a recursion depth of ten gives us >>500 bits |
|
|
302 | json_atof_scan1 (s, &accum, &expo, 0, 10); |
|
|
303 | |
|
|
304 | return neg ? -accum : accum; |
|
|
305 | } |
|
|
306 | |
|
|
307 | // target of scalar reference is bool? -1 == nope, 0 == false, 1 == true |
|
|
308 | static int |
|
|
309 | ref_bool_type (SV *sv) |
|
|
310 | { |
|
|
311 | svtype svt = SvTYPE (sv); |
|
|
312 | |
|
|
313 | if (svt < SVt_PVAV) |
|
|
314 | { |
|
|
315 | STRLEN len = 0; |
|
|
316 | char *pv = svt ? SvPV (sv, len) : 0; |
|
|
317 | |
|
|
318 | if (len == 1) |
|
|
319 | if (*pv == '1') |
|
|
320 | return 1; |
|
|
321 | else if (*pv == '0') |
|
|
322 | return 0; |
|
|
323 | } |
|
|
324 | |
|
|
325 | return -1; |
|
|
326 | } |
|
|
327 | |
|
|
328 | // returns whether scalar is not a reference in the sense of allow_nonref |
|
|
329 | static int |
|
|
330 | json_nonref (SV *scalar) |
|
|
331 | { |
|
|
332 | if (!SvROK (scalar)) |
|
|
333 | return 1; |
|
|
334 | |
|
|
335 | scalar = SvRV (scalar); |
|
|
336 | |
|
|
337 | if (SvTYPE (scalar) >= SVt_PVMG) |
|
|
338 | { |
|
|
339 | if (SvSTASH (scalar) == bool_stash) |
|
|
340 | return 1; |
|
|
341 | |
|
|
342 | if (!SvOBJECT (scalar) && ref_bool_type (scalar) >= 0) |
|
|
343 | return 1; |
|
|
344 | } |
|
|
345 | |
|
|
346 | return 0; |
|
|
347 | } |
|
|
348 | |
149 | ///////////////////////////////////////////////////////////////////////////// |
349 | ///////////////////////////////////////////////////////////////////////////// |
150 | // encoder |
350 | // encoder |
151 | |
351 | |
152 | // structure used for encoding JSON |
352 | // structure used for encoding JSON |
153 | typedef struct |
353 | typedef struct |
… | |
… | |
155 | char *cur; // SvPVX (sv) + current output position |
355 | char *cur; // SvPVX (sv) + current output position |
156 | char *end; // SvEND (sv) |
356 | char *end; // SvEND (sv) |
157 | SV *sv; // result scalar |
357 | SV *sv; // result scalar |
158 | JSON json; |
358 | JSON json; |
159 | U32 indent; // indentation level |
359 | U32 indent; // indentation level |
160 | U32 maxdepth; // max. indentation/recursion level |
|
|
161 | UV limit; // escape character values >= this value when encoding |
360 | UV limit; // escape character values >= this value when encoding |
162 | } enc_t; |
361 | } enc_t; |
163 | |
362 | |
164 | INLINE void |
363 | INLINE void |
165 | need (enc_t *enc, STRLEN len) |
364 | need (enc_t *enc, STRLEN len) |
166 | { |
365 | { |
167 | if (expect_false (enc->cur + len >= enc->end)) |
366 | if (expect_false (enc->cur + len >= enc->end)) |
168 | { |
367 | { |
169 | STRLEN cur = enc->cur - SvPVX (enc->sv); |
368 | STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); |
170 | SvGROW (enc->sv, cur + len + 1); |
369 | SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); |
171 | enc->cur = SvPVX (enc->sv) + cur; |
370 | enc->cur = SvPVX (enc->sv) + cur; |
172 | enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; |
371 | enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; |
173 | } |
372 | } |
174 | } |
373 | } |
175 | |
374 | |
… | |
… | |
250 | (int)((uch - 0x10000) % 0x400 + 0xDC00)); |
449 | (int)((uch - 0x10000) % 0x400 + 0xDC00)); |
251 | enc->cur += 12; |
450 | enc->cur += 12; |
252 | } |
451 | } |
253 | else |
452 | else |
254 | { |
453 | { |
255 | static char hexdigit [16] = "0123456789abcdef"; |
|
|
256 | need (enc, len += 5); |
454 | need (enc, len += 5); |
257 | *enc->cur++ = '\\'; |
455 | *enc->cur++ = '\\'; |
258 | *enc->cur++ = 'u'; |
456 | *enc->cur++ = 'u'; |
259 | *enc->cur++ = hexdigit [ uch >> 12 ]; |
457 | *enc->cur++ = PL_hexdigit [ uch >> 12 ]; |
260 | *enc->cur++ = hexdigit [(uch >> 8) & 15]; |
458 | *enc->cur++ = PL_hexdigit [(uch >> 8) & 15]; |
261 | *enc->cur++ = hexdigit [(uch >> 4) & 15]; |
459 | *enc->cur++ = PL_hexdigit [(uch >> 4) & 15]; |
262 | *enc->cur++ = hexdigit [(uch >> 0) & 15]; |
460 | *enc->cur++ = PL_hexdigit [(uch >> 0) & 15]; |
263 | } |
461 | } |
264 | |
462 | |
265 | str += clen; |
463 | str += clen; |
266 | } |
464 | } |
267 | else if (enc->json.flags & F_LATIN1) |
465 | else if (enc->json.flags & F_LATIN1) |
… | |
… | |
338 | static void |
536 | static void |
339 | encode_av (enc_t *enc, AV *av) |
537 | encode_av (enc_t *enc, AV *av) |
340 | { |
538 | { |
341 | int i, len = av_len (av); |
539 | int i, len = av_len (av); |
342 | |
540 | |
343 | if (enc->indent >= enc->maxdepth) |
541 | if (enc->indent >= enc->json.max_depth) |
344 | croak ("data structure too deep (hit recursion limit)"); |
542 | croak (ERR_NESTING_EXCEEDED); |
345 | |
543 | |
346 | encode_ch (enc, '['); |
544 | encode_ch (enc, '['); |
347 | |
545 | |
348 | if (len >= 0) |
546 | if (len >= 0) |
349 | { |
547 | { |
350 | encode_nl (enc); ++enc->indent; |
548 | encode_nl (enc); ++enc->indent; |
351 | |
549 | |
352 | for (i = 0; i <= len; ++i) |
550 | for (i = 0; i <= len; ++i) |
… | |
… | |
364 | encode_comma (enc); |
562 | encode_comma (enc); |
365 | } |
563 | } |
366 | |
564 | |
367 | encode_nl (enc); --enc->indent; encode_indent (enc); |
565 | encode_nl (enc); --enc->indent; encode_indent (enc); |
368 | } |
566 | } |
369 | |
567 | |
370 | encode_ch (enc, ']'); |
568 | encode_ch (enc, ']'); |
371 | } |
569 | } |
372 | |
570 | |
373 | static void |
571 | static void |
374 | encode_hk (enc_t *enc, HE *he) |
572 | encode_hk (enc_t *enc, HE *he) |
… | |
… | |
378 | if (HeKLEN (he) == HEf_SVKEY) |
576 | if (HeKLEN (he) == HEf_SVKEY) |
379 | { |
577 | { |
380 | SV *sv = HeSVKEY (he); |
578 | SV *sv = HeSVKEY (he); |
381 | STRLEN len; |
579 | STRLEN len; |
382 | char *str; |
580 | char *str; |
383 | |
581 | |
384 | SvGETMAGIC (sv); |
582 | SvGETMAGIC (sv); |
385 | str = SvPV (sv, len); |
583 | str = SvPV (sv, len); |
386 | |
584 | |
387 | encode_str (enc, str, len, SvUTF8 (sv)); |
585 | encode_str (enc, str, len, SvUTF8 (sv)); |
388 | } |
586 | } |
… | |
… | |
423 | |
621 | |
424 | static void |
622 | static void |
425 | encode_hv (enc_t *enc, HV *hv) |
623 | encode_hv (enc_t *enc, HV *hv) |
426 | { |
624 | { |
427 | HE *he; |
625 | HE *he; |
428 | int count; |
|
|
429 | |
626 | |
430 | if (enc->indent >= enc->maxdepth) |
627 | if (enc->indent >= enc->json.max_depth) |
431 | croak ("data structure too deep (hit recursion limit)"); |
628 | croak (ERR_NESTING_EXCEEDED); |
432 | |
629 | |
433 | encode_ch (enc, '{'); |
630 | encode_ch (enc, '{'); |
434 | |
631 | |
435 | // for canonical output we have to sort by keys first |
632 | // for canonical output we have to sort by keys first |
436 | // actually, this is mostly due to the stupid so-called |
633 | // actually, this is mostly due to the stupid so-called |
437 | // security workaround added somewhere in 5.8.x. |
634 | // security workaround added somewhere in 5.8.x |
438 | // that randomises hash orderings |
635 | // that randomises hash orderings |
439 | if (enc->json.flags & F_CANONICAL) |
636 | if (enc->json.flags & F_CANONICAL && !SvRMAGICAL (hv)) |
440 | { |
637 | { |
441 | int count = hv_iterinit (hv); |
638 | int count = hv_iterinit (hv); |
442 | |
639 | |
443 | if (SvMAGICAL (hv)) |
640 | if (SvMAGICAL (hv)) |
444 | { |
641 | { |
… | |
… | |
455 | } |
652 | } |
456 | |
653 | |
457 | if (count) |
654 | if (count) |
458 | { |
655 | { |
459 | int i, fast = 1; |
656 | int i, fast = 1; |
460 | #if defined(__BORLANDC__) || defined(_MSC_VER) |
657 | HE *hes_stack [STACK_HES]; |
461 | HE **hes = _alloca (count * sizeof (HE)); |
658 | HE **hes = hes_stack; |
462 | #else |
659 | |
463 | HE *hes [count]; // if your compiler dies here, you need to enable C99 mode |
660 | // allocate larger arrays on the heap |
464 | #endif |
661 | if (count > STACK_HES) |
|
|
662 | { |
|
|
663 | SV *sv = sv_2mortal (NEWSV (0, count * sizeof (*hes))); |
|
|
664 | hes = (HE **)SvPVX (sv); |
|
|
665 | } |
465 | |
666 | |
466 | i = 0; |
667 | i = 0; |
467 | while ((he = hv_iternext (hv))) |
668 | while ((he = hv_iternext (hv))) |
468 | { |
669 | { |
469 | hes [i++] = he; |
670 | hes [i++] = he; |
… | |
… | |
538 | // encode objects, arrays and special \0=false and \1=true values. |
739 | // encode objects, arrays and special \0=false and \1=true values. |
539 | static void |
740 | static void |
540 | encode_rv (enc_t *enc, SV *sv) |
741 | encode_rv (enc_t *enc, SV *sv) |
541 | { |
742 | { |
542 | svtype svt; |
743 | svtype svt; |
|
|
744 | GV *method; |
543 | |
745 | |
544 | SvGETMAGIC (sv); |
746 | SvGETMAGIC (sv); |
545 | svt = SvTYPE (sv); |
747 | svt = SvTYPE (sv); |
546 | |
748 | |
547 | if (expect_false (SvOBJECT (sv))) |
749 | if (expect_false (SvOBJECT (sv))) |
548 | { |
750 | { |
549 | HV *stash = !JSON_SLOW || json_boolean_stash |
751 | HV *stash = SvSTASH (sv); |
550 | ? json_boolean_stash |
|
|
551 | : gv_stashpv ("JSON::XS::Boolean", 1); |
|
|
552 | |
752 | |
553 | if (SvSTASH (sv) == stash) |
753 | if (stash == bool_stash) |
554 | { |
754 | { |
555 | if (SvIV (sv)) |
755 | if (SvIV (sv)) |
556 | encode_str (enc, "true", 4, 0); |
756 | encode_str (enc, "true", 4, 0); |
557 | else |
757 | else |
558 | encode_str (enc, "false", 5, 0); |
758 | encode_str (enc, "false", 5, 0); |
559 | } |
759 | } |
|
|
760 | else if ((enc->json.flags & F_ALLOW_TAGS) && (method = gv_fetchmethod_autoload (stash, "FREEZE", 0))) |
|
|
761 | { |
|
|
762 | int count; |
|
|
763 | dSP; |
|
|
764 | |
|
|
765 | ENTER; SAVETMPS; |
|
|
766 | SAVESTACK_POS (); |
|
|
767 | PUSHMARK (SP); |
|
|
768 | EXTEND (SP, 2); |
|
|
769 | // we re-bless the reference to get overload and other niceties right |
|
|
770 | PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); |
|
|
771 | PUSHs (sv_json); |
|
|
772 | |
|
|
773 | PUTBACK; |
|
|
774 | count = call_sv ((SV *)GvCV (method), G_ARRAY); |
|
|
775 | SPAGAIN; |
|
|
776 | |
|
|
777 | // catch this surprisingly common error |
|
|
778 | if (SvROK (TOPs) && SvRV (TOPs) == sv) |
|
|
779 | croak ("%s::FREEZE method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv))); |
|
|
780 | |
|
|
781 | encode_ch (enc, '('); |
|
|
782 | encode_ch (enc, '"'); |
|
|
783 | encode_str (enc, HvNAME (stash), HvNAMELEN (stash), HvNAMEUTF8 (stash)); |
|
|
784 | encode_ch (enc, '"'); |
|
|
785 | encode_ch (enc, ')'); |
|
|
786 | encode_ch (enc, '['); |
|
|
787 | |
|
|
788 | while (count) |
|
|
789 | { |
|
|
790 | encode_sv (enc, SP[1 - count--]); |
|
|
791 | |
|
|
792 | if (count) |
|
|
793 | encode_ch (enc, ','); |
|
|
794 | } |
|
|
795 | |
|
|
796 | encode_ch (enc, ']'); |
|
|
797 | |
|
|
798 | FREETMPS; LEAVE; |
|
|
799 | } |
|
|
800 | else if ((enc->json.flags & F_CONV_BLESSED) && (method = gv_fetchmethod_autoload (stash, "TO_JSON", 0))) |
|
|
801 | { |
|
|
802 | dSP; |
|
|
803 | |
|
|
804 | ENTER; SAVETMPS; |
|
|
805 | PUSHMARK (SP); |
|
|
806 | // we re-bless the reference to get overload and other niceties right |
|
|
807 | XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); |
|
|
808 | |
|
|
809 | // calling with G_SCALAR ensures that we always get a 1 return value |
|
|
810 | PUTBACK; |
|
|
811 | call_sv ((SV *)GvCV (method), G_SCALAR); |
|
|
812 | SPAGAIN; |
|
|
813 | |
|
|
814 | // catch this surprisingly common error |
|
|
815 | if (SvROK (TOPs) && SvRV (TOPs) == sv) |
|
|
816 | croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv))); |
|
|
817 | |
|
|
818 | sv = POPs; |
|
|
819 | PUTBACK; |
|
|
820 | |
|
|
821 | encode_sv (enc, sv); |
|
|
822 | |
|
|
823 | FREETMPS; LEAVE; |
|
|
824 | } |
|
|
825 | else if (enc->json.flags & F_ALLOW_BLESSED) |
|
|
826 | encode_str (enc, "null", 4, 0); |
560 | else |
827 | else |
561 | { |
|
|
562 | #if 0 |
|
|
563 | if (0 && sv_derived_from (rv, "JSON::Literal")) |
|
|
564 | { |
|
|
565 | // not yet |
|
|
566 | } |
|
|
567 | #endif |
|
|
568 | if (enc->json.flags & F_CONV_BLESSED) |
|
|
569 | { |
|
|
570 | // we re-bless the reference to get overload and other niceties right |
|
|
571 | GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0); |
|
|
572 | |
|
|
573 | if (to_json) |
|
|
574 | { |
|
|
575 | dSP; |
|
|
576 | |
|
|
577 | ENTER; SAVETMPS; PUSHMARK (SP); |
|
|
578 | XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv))); |
|
|
579 | |
|
|
580 | // calling with G_SCALAR ensures that we always get a 1 return value |
|
|
581 | PUTBACK; |
|
|
582 | call_sv ((SV *)GvCV (to_json), G_SCALAR); |
|
|
583 | SPAGAIN; |
|
|
584 | |
|
|
585 | // catch this surprisingly common error |
|
|
586 | if (SvROK (TOPs) && SvRV (TOPs) == sv) |
|
|
587 | croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv))); |
|
|
588 | |
|
|
589 | sv = POPs; |
|
|
590 | PUTBACK; |
|
|
591 | |
|
|
592 | encode_sv (enc, sv); |
|
|
593 | |
|
|
594 | FREETMPS; LEAVE; |
|
|
595 | } |
|
|
596 | else if (enc->json.flags & F_ALLOW_BLESSED) |
|
|
597 | encode_str (enc, "null", 4, 0); |
|
|
598 | else |
|
|
599 | croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it", |
|
|
600 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
|
|
601 | } |
|
|
602 | else if (enc->json.flags & F_ALLOW_BLESSED) |
|
|
603 | encode_str (enc, "null", 4, 0); |
|
|
604 | else |
|
|
605 | croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled", |
828 | croak ("encountered object '%s', but neither allow_blessed, convert_blessed nor allow_tags settings are enabled (or TO_JSON/FREEZE method missing)", |
606 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
829 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
607 | } |
|
|
608 | } |
830 | } |
609 | else if (svt == SVt_PVHV) |
831 | else if (svt == SVt_PVHV) |
610 | encode_hv (enc, (HV *)sv); |
832 | encode_hv (enc, (HV *)sv); |
611 | else if (svt == SVt_PVAV) |
833 | else if (svt == SVt_PVAV) |
612 | encode_av (enc, (AV *)sv); |
834 | encode_av (enc, (AV *)sv); |
613 | else if (svt < SVt_PVAV) |
835 | else if (svt < SVt_PVAV) |
614 | { |
836 | { |
615 | STRLEN len = 0; |
837 | int bool_type = ref_bool_type (sv); |
616 | char *pv = svt ? SvPV (sv, len) : 0; |
|
|
617 | |
838 | |
618 | if (len == 1 && *pv == '1') |
839 | if (bool_type == 1) |
619 | encode_str (enc, "true", 4, 0); |
840 | encode_str (enc, "true", 4, 0); |
620 | else if (len == 1 && *pv == '0') |
841 | else if (bool_type == 0) |
621 | encode_str (enc, "false", 5, 0); |
842 | encode_str (enc, "false", 5, 0); |
|
|
843 | else if (enc->json.flags & F_ALLOW_UNKNOWN) |
|
|
844 | encode_str (enc, "null", 4, 0); |
622 | else |
845 | else |
623 | croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", |
846 | croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", |
624 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
847 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
625 | } |
848 | } |
|
|
849 | else if (enc->json.flags & F_ALLOW_UNKNOWN) |
|
|
850 | encode_str (enc, "null", 4, 0); |
626 | else |
851 | else |
627 | croak ("encountered %s, but JSON can only represent references to arrays or hashes", |
852 | croak ("encountered %s, but JSON can only represent references to arrays or hashes", |
628 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
853 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
629 | } |
854 | } |
630 | |
855 | |
… | |
… | |
684 | } |
909 | } |
685 | else |
910 | else |
686 | { |
911 | { |
687 | // large integer, use the (rather slow) snprintf way. |
912 | // large integer, use the (rather slow) snprintf way. |
688 | need (enc, IVUV_MAXCHARS); |
913 | need (enc, IVUV_MAXCHARS); |
689 | enc->cur += |
914 | enc->cur += |
690 | SvIsUV(sv) |
915 | SvIsUV(sv) |
691 | ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv)) |
916 | ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv)) |
692 | : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv)); |
917 | : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv)); |
693 | } |
918 | } |
694 | } |
919 | } |
695 | else if (SvROK (sv)) |
920 | else if (SvROK (sv)) |
696 | encode_rv (enc, SvRV (sv)); |
921 | encode_rv (enc, SvRV (sv)); |
697 | else if (!SvOK (sv)) |
922 | else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN) |
698 | encode_str (enc, "null", 4, 0); |
923 | encode_str (enc, "null", 4, 0); |
699 | else |
924 | else |
700 | croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", |
925 | croak ("encountered perl type (%s,0x%x) that JSON cannot handle, check your input data", |
701 | SvPV_nolen (sv), SvFLAGS (sv)); |
926 | SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); |
702 | } |
927 | } |
703 | |
928 | |
704 | static SV * |
929 | static SV * |
705 | encode_json (SV *scalar, JSON *json) |
930 | encode_json (SV *scalar, JSON *json) |
706 | { |
931 | { |
707 | enc_t enc; |
932 | enc_t enc; |
708 | |
933 | |
709 | if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar)) |
934 | if (!(json->flags & F_ALLOW_NONREF) && json_nonref (scalar)) |
710 | croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); |
935 | croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); |
711 | |
936 | |
712 | enc.json = *json; |
937 | enc.json = *json; |
713 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
938 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
714 | enc.cur = SvPVX (enc.sv); |
939 | enc.cur = SvPVX (enc.sv); |
715 | enc.end = SvEND (enc.sv); |
940 | enc.end = SvEND (enc.sv); |
716 | enc.indent = 0; |
941 | enc.indent = 0; |
717 | enc.maxdepth = DEC_DEPTH (enc.json.flags); |
|
|
718 | enc.limit = enc.json.flags & F_ASCII ? 0x000080UL |
942 | enc.limit = enc.json.flags & F_ASCII ? 0x000080UL |
719 | : enc.json.flags & F_LATIN1 ? 0x000100UL |
943 | : enc.json.flags & F_LATIN1 ? 0x000100UL |
720 | : 0x110000UL; |
944 | : 0x110000UL; |
721 | |
945 | |
722 | SvPOK_only (enc.sv); |
946 | SvPOK_only (enc.sv); |
723 | encode_sv (&enc, scalar); |
947 | encode_sv (&enc, scalar); |
|
|
948 | encode_nl (&enc); |
724 | |
949 | |
725 | SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); |
950 | SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); |
726 | *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings |
951 | *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings |
727 | |
952 | |
728 | if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8))) |
953 | if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8))) |
… | |
… | |
789 | if (*dec->cur != ch) \ |
1014 | if (*dec->cur != ch) \ |
790 | ERR (# ch " expected"); \ |
1015 | ERR (# ch " expected"); \ |
791 | ++dec->cur; \ |
1016 | ++dec->cur; \ |
792 | SE |
1017 | SE |
793 | |
1018 | |
794 | #define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)") |
1019 | #define DEC_INC_DEPTH if (++dec->depth > dec->json.max_depth) ERR (ERR_NESTING_EXCEEDED) |
795 | #define DEC_DEC_DEPTH --dec->depth |
1020 | #define DEC_DEC_DEPTH --dec->depth |
796 | |
1021 | |
797 | static SV *decode_sv (dec_t *dec); |
1022 | static SV *decode_sv (dec_t *dec); |
798 | |
1023 | |
799 | static signed char decode_hexdigit[256]; |
1024 | static signed char decode_hexdigit[256]; |
… | |
… | |
908 | else if (expect_true (ch >= 0x20 && ch < 0x80)) |
1133 | else if (expect_true (ch >= 0x20 && ch < 0x80)) |
909 | *cur++ = ch; |
1134 | *cur++ = ch; |
910 | else if (ch >= 0x80) |
1135 | else if (ch >= 0x80) |
911 | { |
1136 | { |
912 | STRLEN clen; |
1137 | STRLEN clen; |
913 | UV uch; |
|
|
914 | |
1138 | |
915 | --dec_cur; |
1139 | --dec_cur; |
916 | |
1140 | |
917 | uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen); |
1141 | decode_utf8 (dec_cur, dec->end - dec_cur, &clen); |
918 | if (clen == (STRLEN)-1) |
1142 | if (clen == (STRLEN)-1) |
919 | ERR ("malformed UTF-8 character in JSON string"); |
1143 | ERR ("malformed UTF-8 character in JSON string"); |
920 | |
1144 | |
921 | do |
1145 | do |
922 | *cur++ = *dec_cur++; |
1146 | *cur++ = *dec_cur++; |
923 | while (--clen); |
1147 | while (--clen); |
924 | |
1148 | |
925 | utf8 = 1; |
1149 | utf8 = 1; |
926 | } |
1150 | } |
|
|
1151 | else if (ch == '\t' && dec->json.flags & F_RELAXED) |
|
|
1152 | *cur++ = ch; |
927 | else |
1153 | else |
928 | { |
1154 | { |
929 | --dec_cur; |
1155 | --dec_cur; |
930 | |
1156 | |
931 | if (!ch) |
1157 | if (!ch) |
… | |
… | |
939 | { |
1165 | { |
940 | STRLEN len = cur - buf; |
1166 | STRLEN len = cur - buf; |
941 | |
1167 | |
942 | if (sv) |
1168 | if (sv) |
943 | { |
1169 | { |
944 | SvGROW (sv, SvCUR (sv) + len + 1); |
1170 | STRLEN cur = SvCUR (sv); |
|
|
1171 | |
|
|
1172 | if (SvLEN (sv) <= cur + len) |
|
|
1173 | SvGROW (sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); |
|
|
1174 | |
945 | memcpy (SvPVX (sv) + SvCUR (sv), buf, len); |
1175 | memcpy (SvPVX (sv) + SvCUR (sv), buf, len); |
946 | SvCUR_set (sv, SvCUR (sv) + len); |
1176 | SvCUR_set (sv, SvCUR (sv) + len); |
947 | } |
1177 | } |
948 | else |
1178 | else |
949 | sv = newSVpvn (buf, len); |
1179 | sv = newSVpvn (buf, len); |
… | |
… | |
1040 | |
1270 | |
1041 | // special case the rather common 1..5-digit-int case |
1271 | // special case the rather common 1..5-digit-int case |
1042 | if (*start == '-') |
1272 | if (*start == '-') |
1043 | switch (len) |
1273 | switch (len) |
1044 | { |
1274 | { |
1045 | case 2: return newSViv (-( start [1] - '0' * 1)); |
1275 | case 2: return newSViv (-(IV)( start [1] - '0' * 1)); |
1046 | case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); |
1276 | case 3: return newSViv (-(IV)( start [1] * 10 + start [2] - '0' * 11)); |
1047 | case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); |
1277 | case 4: return newSViv (-(IV)( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); |
1048 | case 5: return newSViv (-( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); |
1278 | case 5: return newSViv (-(IV)( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); |
1049 | case 6: return newSViv (-(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111)); |
1279 | case 6: return newSViv (-(IV)(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111)); |
1050 | } |
1280 | } |
1051 | else |
1281 | else |
1052 | switch (len) |
1282 | switch (len) |
1053 | { |
1283 | { |
1054 | case 1: return newSViv ( start [0] - '0' * 1); |
1284 | case 1: return newSViv ( start [0] - '0' * 1); |
1055 | case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); |
1285 | case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); |
1056 | case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); |
1286 | case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); |
1057 | case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); |
1287 | case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); |
1058 | case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111); |
1288 | case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111); |
1059 | } |
1289 | } |
1060 | |
1290 | |
1061 | { |
1291 | { |
1062 | UV uv; |
1292 | UV uv; |
1063 | int numtype = grok_number (start, len, &uv); |
1293 | int numtype = grok_number (start, len, &uv); |
… | |
… | |
1072 | } |
1302 | } |
1073 | |
1303 | |
1074 | len -= *start == '-' ? 1 : 0; |
1304 | len -= *start == '-' ? 1 : 0; |
1075 | |
1305 | |
1076 | // does not fit into IV or UV, try NV |
1306 | // does not fit into IV or UV, try NV |
1077 | if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len) |
1307 | if (len <= NV_DIG) |
1078 | #if defined (LDBL_DIG) |
|
|
1079 | || (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len) |
|
|
1080 | #endif |
|
|
1081 | ) |
|
|
1082 | // fits into NV without loss of precision |
1308 | // fits into NV without loss of precision |
1083 | return newSVnv (Atof (start)); |
1309 | return newSVnv (json_atof (start)); |
1084 | |
1310 | |
1085 | // everything else fails, convert it to a string |
1311 | // everything else fails, convert it to a string |
1086 | return newSVpvn (start, dec->cur - start); |
1312 | return newSVpvn (start, dec->cur - start); |
1087 | } |
1313 | } |
1088 | |
1314 | |
1089 | // loss of precision here |
1315 | // loss of precision here |
1090 | return newSVnv (Atof (start)); |
1316 | return newSVnv (json_atof (start)); |
1091 | |
1317 | |
1092 | fail: |
1318 | fail: |
1093 | return 0; |
1319 | return 0; |
1094 | } |
1320 | } |
1095 | |
1321 | |
… | |
… | |
1119 | if (*dec->cur == ']') |
1345 | if (*dec->cur == ']') |
1120 | { |
1346 | { |
1121 | ++dec->cur; |
1347 | ++dec->cur; |
1122 | break; |
1348 | break; |
1123 | } |
1349 | } |
1124 | |
1350 | |
1125 | if (*dec->cur != ',') |
1351 | if (*dec->cur != ',') |
1126 | ERR (", or ] expected while parsing array"); |
1352 | ERR (", or ] expected while parsing array"); |
1127 | |
1353 | |
1128 | ++dec->cur; |
1354 | ++dec->cur; |
1129 | |
1355 | |
… | |
… | |
1171 | char *p = dec->cur; |
1397 | char *p = dec->cur; |
1172 | char *e = p + 24; // only try up to 24 bytes |
1398 | char *e = p + 24; // only try up to 24 bytes |
1173 | |
1399 | |
1174 | for (;;) |
1400 | for (;;) |
1175 | { |
1401 | { |
1176 | // the >= 0x80 is true on most architectures |
1402 | // the >= 0x80 is false on most architectures |
1177 | if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\') |
1403 | if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\') |
1178 | { |
1404 | { |
1179 | // slow path, back up and use decode_str |
1405 | // slow path, back up and use decode_str |
1180 | SV *key = decode_str (dec); |
1406 | SV *key = decode_str (dec); |
1181 | if (!key) |
1407 | if (!key) |
… | |
… | |
1253 | |
1479 | |
1254 | hv_iterinit (hv); |
1480 | hv_iterinit (hv); |
1255 | he = hv_iternext (hv); |
1481 | he = hv_iternext (hv); |
1256 | hv_iterinit (hv); |
1482 | hv_iterinit (hv); |
1257 | |
1483 | |
1258 | // the next line creates a mortal sv each time its called. |
1484 | // the next line creates a mortal sv each time it's called. |
1259 | // might want to optimise this for common cases. |
1485 | // might want to optimise this for common cases. |
1260 | cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0); |
1486 | cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0); |
1261 | |
1487 | |
1262 | if (cb) |
1488 | if (cb) |
1263 | { |
1489 | { |
1264 | dSP; |
1490 | dSP; |
1265 | int count; |
1491 | int count; |
1266 | |
1492 | |
1267 | ENTER; SAVETMPS; PUSHMARK (SP); |
1493 | ENTER; SAVETMPS; |
|
|
1494 | SAVESTACK_POS (); |
|
|
1495 | PUSHMARK (SP); |
1268 | XPUSHs (HeVAL (he)); |
1496 | XPUSHs (HeVAL (he)); |
|
|
1497 | sv_2mortal (sv); |
1269 | |
1498 | |
1270 | PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN; |
1499 | PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN; |
1271 | |
1500 | |
1272 | if (count == 1) |
1501 | if (count == 1) |
1273 | { |
1502 | { |
1274 | sv = newSVsv (POPs); |
1503 | sv = newSVsv (POPs); |
1275 | FREETMPS; LEAVE; |
1504 | FREETMPS; LEAVE; |
1276 | return sv; |
1505 | return sv; |
1277 | } |
1506 | } |
1278 | |
1507 | |
|
|
1508 | SvREFCNT_inc (sv); |
1279 | FREETMPS; LEAVE; |
1509 | FREETMPS; LEAVE; |
1280 | } |
1510 | } |
1281 | } |
1511 | } |
1282 | |
1512 | |
1283 | if (dec->json.cb_object) |
1513 | if (dec->json.cb_object) |
1284 | { |
1514 | { |
1285 | dSP; |
1515 | dSP; |
1286 | int count; |
1516 | int count; |
1287 | |
1517 | |
1288 | ENTER; SAVETMPS; PUSHMARK (SP); |
1518 | ENTER; SAVETMPS; |
|
|
1519 | SAVESTACK_POS (); |
|
|
1520 | PUSHMARK (SP); |
1289 | XPUSHs (sv_2mortal (sv)); |
1521 | XPUSHs (sv_2mortal (sv)); |
1290 | |
1522 | |
1291 | PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN; |
1523 | PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN; |
1292 | |
1524 | |
1293 | if (count == 1) |
1525 | if (count == 1) |
… | |
… | |
1309 | DEC_DEC_DEPTH; |
1541 | DEC_DEC_DEPTH; |
1310 | return 0; |
1542 | return 0; |
1311 | } |
1543 | } |
1312 | |
1544 | |
1313 | static SV * |
1545 | static SV * |
|
|
1546 | decode_tag (dec_t *dec) |
|
|
1547 | { |
|
|
1548 | SV *tag = 0; |
|
|
1549 | SV *val = 0; |
|
|
1550 | |
|
|
1551 | if (!(dec->json.flags & F_ALLOW_TAGS)) |
|
|
1552 | ERR ("malformed JSON string, neither array, object, number, string or atom"); |
|
|
1553 | |
|
|
1554 | ++dec->cur; |
|
|
1555 | |
|
|
1556 | decode_ws (dec); |
|
|
1557 | |
|
|
1558 | tag = decode_sv (dec); |
|
|
1559 | if (!tag) |
|
|
1560 | goto fail; |
|
|
1561 | |
|
|
1562 | if (!SvPOK (tag)) |
|
|
1563 | ERR ("malformed JSON string, (tag) must be a string"); |
|
|
1564 | |
|
|
1565 | decode_ws (dec); |
|
|
1566 | |
|
|
1567 | if (*dec->cur != ')') |
|
|
1568 | ERR (") expected after tag"); |
|
|
1569 | |
|
|
1570 | ++dec->cur; |
|
|
1571 | |
|
|
1572 | decode_ws (dec); |
|
|
1573 | |
|
|
1574 | val = decode_sv (dec); |
|
|
1575 | if (!val) |
|
|
1576 | goto fail; |
|
|
1577 | |
|
|
1578 | if (!SvROK (val) || SvTYPE (SvRV (val)) != SVt_PVAV) |
|
|
1579 | ERR ("malformed JSON string, tag value must be an array"); |
|
|
1580 | |
|
|
1581 | { |
|
|
1582 | AV *av = (AV *)SvRV (val); |
|
|
1583 | int i, len = av_len (av) + 1; |
|
|
1584 | HV *stash = gv_stashsv (tag, 0); |
|
|
1585 | SV *sv; |
|
|
1586 | |
|
|
1587 | if (!stash) |
|
|
1588 | ERR ("cannot decode perl-object (package does not exist)"); |
|
|
1589 | |
|
|
1590 | GV *method = gv_fetchmethod_autoload (stash, "THAW", 0); |
|
|
1591 | |
|
|
1592 | if (!method) |
|
|
1593 | ERR ("cannot decode perl-object (package does not have a THAW method)"); |
|
|
1594 | |
|
|
1595 | dSP; |
|
|
1596 | |
|
|
1597 | ENTER; SAVETMPS; |
|
|
1598 | PUSHMARK (SP); |
|
|
1599 | EXTEND (SP, len + 2); |
|
|
1600 | // we re-bless the reference to get overload and other niceties right |
|
|
1601 | PUSHs (tag); |
|
|
1602 | PUSHs (sv_json); |
|
|
1603 | |
|
|
1604 | for (i = 0; i < len; ++i) |
|
|
1605 | PUSHs (*av_fetch (av, i, 1)); |
|
|
1606 | |
|
|
1607 | PUTBACK; |
|
|
1608 | call_sv ((SV *)GvCV (method), G_SCALAR); |
|
|
1609 | SPAGAIN; |
|
|
1610 | |
|
|
1611 | SvREFCNT_dec (tag); |
|
|
1612 | SvREFCNT_dec (val); |
|
|
1613 | sv = SvREFCNT_inc (POPs); |
|
|
1614 | |
|
|
1615 | PUTBACK; |
|
|
1616 | |
|
|
1617 | FREETMPS; LEAVE; |
|
|
1618 | |
|
|
1619 | return sv; |
|
|
1620 | } |
|
|
1621 | |
|
|
1622 | fail: |
|
|
1623 | SvREFCNT_dec (tag); |
|
|
1624 | SvREFCNT_dec (val); |
|
|
1625 | return 0; |
|
|
1626 | } |
|
|
1627 | |
|
|
1628 | static SV * |
1314 | decode_sv (dec_t *dec) |
1629 | decode_sv (dec_t *dec) |
1315 | { |
1630 | { |
1316 | // the beauty of JSON: you need exactly one character lookahead |
1631 | // the beauty of JSON: you need exactly one character lookahead |
1317 | // to parse everything. |
1632 | // to parse everything. |
1318 | switch (*dec->cur) |
1633 | switch (*dec->cur) |
1319 | { |
1634 | { |
1320 | case '"': ++dec->cur; return decode_str (dec); |
1635 | case '"': ++dec->cur; return decode_str (dec); |
1321 | case '[': ++dec->cur; return decode_av (dec); |
1636 | case '[': ++dec->cur; return decode_av (dec); |
1322 | case '{': ++dec->cur; return decode_hv (dec); |
1637 | case '{': ++dec->cur; return decode_hv (dec); |
|
|
1638 | case '(': return decode_tag (dec); |
1323 | |
1639 | |
1324 | case '-': |
1640 | case '-': |
1325 | case '0': case '1': case '2': case '3': case '4': |
1641 | case '0': case '1': case '2': case '3': case '4': |
1326 | case '5': case '6': case '7': case '8': case '9': |
1642 | case '5': case '6': case '7': case '8': case '9': |
1327 | return decode_num (dec); |
1643 | return decode_num (dec); |
… | |
… | |
1329 | case 't': |
1645 | case 't': |
1330 | if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) |
1646 | if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) |
1331 | { |
1647 | { |
1332 | dec->cur += 4; |
1648 | dec->cur += 4; |
1333 | #if JSON_SLOW |
1649 | #if JSON_SLOW |
1334 | json_true = get_sv ("JSON::XS::true", 1); SvREADONLY_on (json_true); |
1650 | bool_true = get_bool ("Types::Serialiser::true"); |
1335 | #endif |
1651 | #endif |
1336 | return SvREFCNT_inc (json_true); |
1652 | return newSVsv (bool_true); |
1337 | } |
1653 | } |
1338 | else |
1654 | else |
1339 | ERR ("'true' expected"); |
1655 | ERR ("'true' expected"); |
1340 | |
1656 | |
1341 | break; |
1657 | break; |
… | |
… | |
1343 | case 'f': |
1659 | case 'f': |
1344 | if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) |
1660 | if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) |
1345 | { |
1661 | { |
1346 | dec->cur += 5; |
1662 | dec->cur += 5; |
1347 | #if JSON_SLOW |
1663 | #if JSON_SLOW |
1348 | json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); |
1664 | bool_false = get_bool ("Types::Serialiser::false"); |
1349 | #endif |
1665 | #endif |
1350 | return SvREFCNT_inc (json_false); |
1666 | return newSVsv (bool_false); |
1351 | } |
1667 | } |
1352 | else |
1668 | else |
1353 | ERR ("'false' expected"); |
1669 | ERR ("'false' expected"); |
1354 | |
1670 | |
1355 | break; |
1671 | break; |
… | |
… | |
1364 | ERR ("'null' expected"); |
1680 | ERR ("'null' expected"); |
1365 | |
1681 | |
1366 | break; |
1682 | break; |
1367 | |
1683 | |
1368 | default: |
1684 | default: |
1369 | ERR ("malformed JSON string, neither array, object, number, string or atom"); |
1685 | ERR ("malformed JSON string, neither tag, array, object, number, string or atom"); |
1370 | break; |
1686 | break; |
1371 | } |
1687 | } |
1372 | |
1688 | |
1373 | fail: |
1689 | fail: |
1374 | return 0; |
1690 | return 0; |
1375 | } |
1691 | } |
1376 | |
1692 | |
1377 | static SV * |
1693 | static SV * |
1378 | decode_json (SV *string, JSON *json, UV *offset_return) |
1694 | decode_json (SV *string, JSON *json, STRLEN *offset_return) |
1379 | { |
1695 | { |
1380 | dec_t dec; |
1696 | dec_t dec; |
1381 | UV offset; |
|
|
1382 | SV *sv; |
1697 | SV *sv; |
1383 | |
1698 | |
|
|
1699 | /* work around bugs in 5.10 where manipulating magic values |
|
|
1700 | * makes perl ignore the magic in subsequent accesses. |
|
|
1701 | * also make a copy of non-PV values, to get them into a clean |
|
|
1702 | * state (SvPV should do that, but it's buggy, see below). |
|
|
1703 | * |
|
|
1704 | * SvIsCOW_shared_hash works around a bug in perl (possibly 5.16), |
|
|
1705 | * as reported by Reini Urban. |
|
|
1706 | */ |
1384 | SvGETMAGIC (string); |
1707 | /*SvGETMAGIC (string);*/ |
|
|
1708 | if (SvMAGICAL (string) || !SvPOK (string) || SvIsCOW_shared_hash (string)) |
|
|
1709 | string = sv_2mortal (newSVsv (string)); |
|
|
1710 | |
1385 | SvUPGRADE (string, SVt_PV); |
1711 | SvUPGRADE (string, SVt_PV); |
1386 | |
1712 | |
1387 | if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags)) |
1713 | /* work around a bug in perl 5.10, which causes SvCUR to fail an |
|
|
1714 | * assertion with -DDEBUGGING, although SvCUR is documented to |
|
|
1715 | * return the xpv_cur field which certainly exists after upgrading. |
|
|
1716 | * according to nicholas clark, calling SvPOK fixes this. |
|
|
1717 | * But it doesn't fix it, so try another workaround, call SvPV_nolen |
|
|
1718 | * and hope for the best. |
|
|
1719 | * Damnit, SvPV_nolen still trips over yet another assertion. This |
|
|
1720 | * assertion business is seriously broken, try yet another workaround |
|
|
1721 | * for the broken -DDEBUGGING. |
|
|
1722 | */ |
|
|
1723 | { |
|
|
1724 | #ifdef DEBUGGING |
|
|
1725 | STRLEN offset = SvOK (string) ? sv_len (string) : 0; |
|
|
1726 | #else |
|
|
1727 | STRLEN offset = SvCUR (string); |
|
|
1728 | #endif |
|
|
1729 | |
|
|
1730 | if (offset > json->max_size && json->max_size) |
1388 | croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", |
1731 | croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", |
1389 | (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags)); |
1732 | (unsigned long)SvCUR (string), (unsigned long)json->max_size); |
|
|
1733 | } |
1390 | |
1734 | |
1391 | if (json->flags & F_UTF8) |
1735 | if (DECODE_WANTS_OCTETS (json)) |
1392 | sv_utf8_downgrade (string, 0); |
1736 | sv_utf8_downgrade (string, 0); |
1393 | else |
1737 | else |
1394 | sv_utf8_upgrade (string); |
1738 | sv_utf8_upgrade (string); |
1395 | |
1739 | |
1396 | SvGROW (string, SvCUR (string) + 1); // should basically be a NOP |
1740 | SvGROW (string, SvCUR (string) + 1); // should basically be a NOP |
1397 | |
1741 | |
1398 | dec.json = *json; |
1742 | dec.json = *json; |
1399 | dec.cur = SvPVX (string); |
1743 | dec.cur = SvPVX (string); |
1400 | dec.end = SvEND (string); |
1744 | dec.end = SvEND (string); |
1401 | dec.err = 0; |
1745 | dec.err = 0; |
1402 | dec.depth = 0; |
1746 | dec.depth = 0; |
1403 | dec.maxdepth = DEC_DEPTH (dec.json.flags); |
|
|
1404 | |
1747 | |
1405 | if (dec.json.cb_object || dec.json.cb_sk_object) |
1748 | if (dec.json.cb_object || dec.json.cb_sk_object) |
1406 | dec.json.flags |= F_HOOK; |
1749 | dec.json.flags |= F_HOOK; |
1407 | |
1750 | |
1408 | *dec.end = 0; // this should basically be a nop, too, but make sure it's there |
1751 | *dec.end = 0; // this should basically be a nop, too, but make sure it's there |
1409 | |
1752 | |
1410 | decode_ws (&dec); |
1753 | decode_ws (&dec); |
1411 | sv = decode_sv (&dec); |
1754 | sv = decode_sv (&dec); |
1412 | |
1755 | |
1413 | if (!(offset_return || !sv)) |
1756 | if (offset_return) |
|
|
1757 | *offset_return = dec.cur - SvPVX (string); |
|
|
1758 | else if (sv) |
1414 | { |
1759 | { |
1415 | // check for trailing garbage |
1760 | // check for trailing garbage |
1416 | decode_ws (&dec); |
1761 | decode_ws (&dec); |
1417 | |
1762 | |
1418 | if (*dec.cur) |
1763 | if (*dec.cur) |
1419 | { |
1764 | { |
1420 | dec.err = "garbage after JSON object"; |
1765 | dec.err = "garbage after JSON object"; |
1421 | SvREFCNT_dec (sv); |
1766 | SvREFCNT_dec (sv); |
1422 | sv = 0; |
1767 | sv = 0; |
1423 | } |
1768 | } |
1424 | } |
|
|
1425 | |
|
|
1426 | if (offset_return || !sv) |
|
|
1427 | { |
|
|
1428 | offset = dec.json.flags & F_UTF8 |
|
|
1429 | ? dec.cur - SvPVX (string) |
|
|
1430 | : utf8_distance (dec.cur, SvPVX (string)); |
|
|
1431 | |
|
|
1432 | if (offset_return) |
|
|
1433 | *offset_return = offset; |
|
|
1434 | } |
1769 | } |
1435 | |
1770 | |
1436 | if (!sv) |
1771 | if (!sv) |
1437 | { |
1772 | { |
1438 | SV *uni = sv_newmortal (); |
1773 | SV *uni = sv_newmortal (); |
… | |
… | |
1444 | SAVEVPTR (PL_curcop); |
1779 | SAVEVPTR (PL_curcop); |
1445 | PL_curcop = &cop; |
1780 | PL_curcop = &cop; |
1446 | pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); |
1781 | pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); |
1447 | LEAVE; |
1782 | LEAVE; |
1448 | |
1783 | |
1449 | croak ("%s, at character offset %d [\"%s\"]", |
1784 | croak ("%s, at character offset %d (before \"%s\")", |
1450 | dec.err, |
1785 | dec.err, |
1451 | (int)offset, |
1786 | (int)ptr_to_index (string, dec.cur), |
1452 | dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); |
1787 | dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); |
1453 | } |
1788 | } |
1454 | |
1789 | |
1455 | sv = sv_2mortal (sv); |
1790 | sv = sv_2mortal (sv); |
1456 | |
1791 | |
1457 | if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv)) |
1792 | if (!(dec.json.flags & F_ALLOW_NONREF) && json_nonref (sv)) |
1458 | croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); |
1793 | croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); |
1459 | |
1794 | |
1460 | return sv; |
1795 | return sv; |
|
|
1796 | } |
|
|
1797 | |
|
|
1798 | ///////////////////////////////////////////////////////////////////////////// |
|
|
1799 | // incremental parser |
|
|
1800 | |
|
|
1801 | static void |
|
|
1802 | incr_parse (JSON *self) |
|
|
1803 | { |
|
|
1804 | const char *p = SvPVX (self->incr_text) + self->incr_pos; |
|
|
1805 | |
|
|
1806 | // the state machine here is a bit convoluted and could be simplified a lot |
|
|
1807 | // but this would make it slower, so... |
|
|
1808 | |
|
|
1809 | for (;;) |
|
|
1810 | { |
|
|
1811 | //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 |
|
|
1812 | switch (self->incr_mode) |
|
|
1813 | { |
|
|
1814 | // only used for initial whitespace skipping |
|
|
1815 | case INCR_M_WS: |
|
|
1816 | for (;;) |
|
|
1817 | { |
|
|
1818 | if (*p > 0x20) |
|
|
1819 | { |
|
|
1820 | if (*p == '#') |
|
|
1821 | { |
|
|
1822 | self->incr_mode = INCR_M_C0; |
|
|
1823 | goto incr_m_c; |
|
|
1824 | } |
|
|
1825 | else |
|
|
1826 | { |
|
|
1827 | self->incr_mode = INCR_M_JSON; |
|
|
1828 | goto incr_m_json; |
|
|
1829 | } |
|
|
1830 | } |
|
|
1831 | else if (!*p) |
|
|
1832 | goto interrupt; |
|
|
1833 | |
|
|
1834 | ++p; |
|
|
1835 | } |
|
|
1836 | |
|
|
1837 | // skip a single char inside a string (for \\-processing) |
|
|
1838 | case INCR_M_BS: |
|
|
1839 | if (!*p) |
|
|
1840 | goto interrupt; |
|
|
1841 | |
|
|
1842 | ++p; |
|
|
1843 | self->incr_mode = INCR_M_STR; |
|
|
1844 | goto incr_m_str; |
|
|
1845 | |
|
|
1846 | // inside #-style comments |
|
|
1847 | case INCR_M_C0: |
|
|
1848 | case INCR_M_C1: |
|
|
1849 | incr_m_c: |
|
|
1850 | for (;;) |
|
|
1851 | { |
|
|
1852 | if (*p == '\n') |
|
|
1853 | { |
|
|
1854 | self->incr_mode = self->incr_mode == INCR_M_C0 ? INCR_M_WS : INCR_M_JSON; |
|
|
1855 | break; |
|
|
1856 | } |
|
|
1857 | else if (!*p) |
|
|
1858 | goto interrupt; |
|
|
1859 | |
|
|
1860 | ++p; |
|
|
1861 | } |
|
|
1862 | |
|
|
1863 | break; |
|
|
1864 | |
|
|
1865 | // inside a string |
|
|
1866 | case INCR_M_STR: |
|
|
1867 | incr_m_str: |
|
|
1868 | for (;;) |
|
|
1869 | { |
|
|
1870 | if (*p == '"') |
|
|
1871 | { |
|
|
1872 | ++p; |
|
|
1873 | self->incr_mode = INCR_M_JSON; |
|
|
1874 | |
|
|
1875 | if (!self->incr_nest) |
|
|
1876 | goto interrupt; |
|
|
1877 | |
|
|
1878 | goto incr_m_json; |
|
|
1879 | } |
|
|
1880 | else if (*p == '\\') |
|
|
1881 | { |
|
|
1882 | ++p; // "virtually" consumes character after \ |
|
|
1883 | |
|
|
1884 | if (!*p) // if at end of string we have to switch modes |
|
|
1885 | { |
|
|
1886 | self->incr_mode = INCR_M_BS; |
|
|
1887 | goto interrupt; |
|
|
1888 | } |
|
|
1889 | } |
|
|
1890 | else if (!*p) |
|
|
1891 | goto interrupt; |
|
|
1892 | |
|
|
1893 | ++p; |
|
|
1894 | } |
|
|
1895 | |
|
|
1896 | // after initial ws, outside string |
|
|
1897 | case INCR_M_JSON: |
|
|
1898 | incr_m_json: |
|
|
1899 | for (;;) |
|
|
1900 | { |
|
|
1901 | switch (*p++) |
|
|
1902 | { |
|
|
1903 | case 0: |
|
|
1904 | --p; |
|
|
1905 | goto interrupt; |
|
|
1906 | |
|
|
1907 | case 0x09: |
|
|
1908 | case 0x0a: |
|
|
1909 | case 0x0d: |
|
|
1910 | case 0x20: |
|
|
1911 | if (!self->incr_nest) |
|
|
1912 | { |
|
|
1913 | --p; // do not eat the whitespace, let the next round do it |
|
|
1914 | goto interrupt; |
|
|
1915 | } |
|
|
1916 | break; |
|
|
1917 | |
|
|
1918 | case '"': |
|
|
1919 | self->incr_mode = INCR_M_STR; |
|
|
1920 | goto incr_m_str; |
|
|
1921 | |
|
|
1922 | case '[': |
|
|
1923 | case '{': |
|
|
1924 | case '(': |
|
|
1925 | if (++self->incr_nest > self->max_depth) |
|
|
1926 | croak (ERR_NESTING_EXCEEDED); |
|
|
1927 | break; |
|
|
1928 | |
|
|
1929 | case ']': |
|
|
1930 | case '}': |
|
|
1931 | if (--self->incr_nest <= 0) |
|
|
1932 | goto interrupt; |
|
|
1933 | break; |
|
|
1934 | |
|
|
1935 | case ')': |
|
|
1936 | --self->incr_nest; |
|
|
1937 | break; |
|
|
1938 | |
|
|
1939 | case '#': |
|
|
1940 | self->incr_mode = INCR_M_C1; |
|
|
1941 | goto incr_m_c; |
|
|
1942 | } |
|
|
1943 | } |
|
|
1944 | } |
|
|
1945 | |
|
|
1946 | modechange: |
|
|
1947 | ; |
|
|
1948 | } |
|
|
1949 | |
|
|
1950 | interrupt: |
|
|
1951 | self->incr_pos = p - SvPVX (self->incr_text); |
|
|
1952 | //printf ("interrupt<%.*s>\n", self->incr_pos, SvPVX(self->incr_text));//D |
|
|
1953 | //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D |
1461 | } |
1954 | } |
1462 | |
1955 | |
1463 | ///////////////////////////////////////////////////////////////////////////// |
1956 | ///////////////////////////////////////////////////////////////////////////// |
1464 | // XS interface functions |
1957 | // XS interface functions |
1465 | |
1958 | |
… | |
… | |
1474 | i >= '0' && i <= '9' ? i - '0' |
1967 | i >= '0' && i <= '9' ? i - '0' |
1475 | : i >= 'a' && i <= 'f' ? i - 'a' + 10 |
1968 | : i >= 'a' && i <= 'f' ? i - 'a' + 10 |
1476 | : i >= 'A' && i <= 'F' ? i - 'A' + 10 |
1969 | : i >= 'A' && i <= 'F' ? i - 'A' + 10 |
1477 | : -1; |
1970 | : -1; |
1478 | |
1971 | |
1479 | json_stash = gv_stashpv ("JSON::XS" , 1); |
1972 | json_stash = gv_stashpv ("JSON::XS" , 1); |
1480 | json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); |
1973 | bool_stash = gv_stashpv ("Types::Serialiser::Boolean", 1); |
|
|
1974 | bool_true = get_bool ("Types::Serialiser::true"); |
|
|
1975 | bool_false = get_bool ("Types::Serialiser::false"); |
1481 | |
1976 | |
1482 | json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true ); |
1977 | sv_json = newSVpv ("JSON", 0); |
1483 | json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); |
1978 | SvREADONLY_on (sv_json); |
|
|
1979 | |
|
|
1980 | CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */ |
1484 | } |
1981 | } |
1485 | |
1982 | |
1486 | PROTOTYPES: DISABLE |
1983 | PROTOTYPES: DISABLE |
1487 | |
1984 | |
1488 | void CLONE (...) |
1985 | void CLONE (...) |
1489 | CODE: |
1986 | CODE: |
1490 | json_stash = 0; |
1987 | json_stash = 0; |
1491 | json_boolean_stash = 0; |
1988 | bool_stash = 0; |
1492 | |
1989 | |
1493 | void new (char *klass) |
1990 | void new (char *klass) |
1494 | PPCODE: |
1991 | PPCODE: |
1495 | { |
1992 | { |
1496 | SV *pv = NEWSV (0, sizeof (JSON)); |
1993 | SV *pv = NEWSV (0, sizeof (JSON)); |
1497 | SvPOK_only (pv); |
1994 | SvPOK_only (pv); |
1498 | Zero (SvPVX (pv), 1, JSON); |
1995 | json_init ((JSON *)SvPVX (pv)); |
1499 | ((JSON *)SvPVX (pv))->flags = F_DEFAULT; |
|
|
1500 | XPUSHs (sv_2mortal (sv_bless ( |
1996 | XPUSHs (sv_2mortal (sv_bless ( |
1501 | newRV_noinc (pv), |
1997 | newRV_noinc (pv), |
1502 | strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1) |
1998 | strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1) |
1503 | ))); |
1999 | ))); |
1504 | } |
2000 | } |
… | |
… | |
1516 | allow_nonref = F_ALLOW_NONREF |
2012 | allow_nonref = F_ALLOW_NONREF |
1517 | shrink = F_SHRINK |
2013 | shrink = F_SHRINK |
1518 | allow_blessed = F_ALLOW_BLESSED |
2014 | allow_blessed = F_ALLOW_BLESSED |
1519 | convert_blessed = F_CONV_BLESSED |
2015 | convert_blessed = F_CONV_BLESSED |
1520 | relaxed = F_RELAXED |
2016 | relaxed = F_RELAXED |
|
|
2017 | allow_unknown = F_ALLOW_UNKNOWN |
|
|
2018 | allow_tags = F_ALLOW_TAGS |
1521 | PPCODE: |
2019 | PPCODE: |
1522 | { |
2020 | { |
1523 | if (enable) |
2021 | if (enable) |
1524 | self->flags |= ix; |
2022 | self->flags |= ix; |
1525 | else |
2023 | else |
… | |
… | |
1540 | get_allow_nonref = F_ALLOW_NONREF |
2038 | get_allow_nonref = F_ALLOW_NONREF |
1541 | get_shrink = F_SHRINK |
2039 | get_shrink = F_SHRINK |
1542 | get_allow_blessed = F_ALLOW_BLESSED |
2040 | get_allow_blessed = F_ALLOW_BLESSED |
1543 | get_convert_blessed = F_CONV_BLESSED |
2041 | get_convert_blessed = F_CONV_BLESSED |
1544 | get_relaxed = F_RELAXED |
2042 | get_relaxed = F_RELAXED |
|
|
2043 | get_allow_unknown = F_ALLOW_UNKNOWN |
|
|
2044 | get_allow_tags = F_ALLOW_TAGS |
1545 | PPCODE: |
2045 | PPCODE: |
1546 | XPUSHs (boolSV (self->flags & ix)); |
2046 | XPUSHs (boolSV (self->flags & ix)); |
1547 | |
2047 | |
1548 | void max_depth (JSON *self, UV max_depth = 0x80000000UL) |
2048 | void max_depth (JSON *self, U32 max_depth = 0x80000000UL) |
1549 | PPCODE: |
2049 | PPCODE: |
1550 | { |
2050 | self->max_depth = max_depth; |
1551 | UV log2 = 0; |
|
|
1552 | |
|
|
1553 | if (max_depth > 0x80000000UL) max_depth = 0x80000000UL; |
|
|
1554 | |
|
|
1555 | while ((1UL << log2) < max_depth) |
|
|
1556 | ++log2; |
|
|
1557 | |
|
|
1558 | self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH); |
|
|
1559 | |
|
|
1560 | XPUSHs (ST (0)); |
2051 | XPUSHs (ST (0)); |
1561 | } |
|
|
1562 | |
2052 | |
1563 | U32 get_max_depth (JSON *self) |
2053 | U32 get_max_depth (JSON *self) |
1564 | CODE: |
2054 | CODE: |
1565 | RETVAL = DEC_DEPTH (self->flags); |
2055 | RETVAL = self->max_depth; |
1566 | OUTPUT: |
2056 | OUTPUT: |
1567 | RETVAL |
2057 | RETVAL |
1568 | |
2058 | |
1569 | void max_size (JSON *self, UV max_size = 0) |
2059 | void max_size (JSON *self, U32 max_size = 0) |
1570 | PPCODE: |
2060 | PPCODE: |
1571 | { |
2061 | self->max_size = max_size; |
1572 | UV log2 = 0; |
|
|
1573 | |
|
|
1574 | if (max_size > 0x80000000UL) max_size = 0x80000000UL; |
|
|
1575 | if (max_size == 1) max_size = 2; |
|
|
1576 | |
|
|
1577 | while ((1UL << log2) < max_size) |
|
|
1578 | ++log2; |
|
|
1579 | |
|
|
1580 | self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE); |
|
|
1581 | |
|
|
1582 | XPUSHs (ST (0)); |
2062 | XPUSHs (ST (0)); |
1583 | } |
|
|
1584 | |
2063 | |
1585 | int get_max_size (JSON *self) |
2064 | int get_max_size (JSON *self) |
1586 | CODE: |
2065 | CODE: |
1587 | RETVAL = DEC_SIZE (self->flags); |
2066 | RETVAL = self->max_size; |
1588 | OUTPUT: |
2067 | OUTPUT: |
1589 | RETVAL |
2068 | RETVAL |
1590 | |
2069 | |
1591 | void filter_json_object (JSON *self, SV *cb = &PL_sv_undef) |
2070 | void filter_json_object (JSON *self, SV *cb = &PL_sv_undef) |
1592 | PPCODE: |
2071 | PPCODE: |
… | |
… | |
1598 | } |
2077 | } |
1599 | |
2078 | |
1600 | void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef) |
2079 | void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef) |
1601 | PPCODE: |
2080 | PPCODE: |
1602 | { |
2081 | { |
1603 | if (!self->cb_sk_object) |
2082 | if (!self->cb_sk_object) |
1604 | self->cb_sk_object = newHV (); |
2083 | self->cb_sk_object = newHV (); |
1605 | |
2084 | |
1606 | if (SvOK (cb)) |
2085 | if (SvOK (cb)) |
1607 | hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0); |
2086 | hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0); |
1608 | else |
2087 | else |
… | |
… | |
1619 | XPUSHs (ST (0)); |
2098 | XPUSHs (ST (0)); |
1620 | } |
2099 | } |
1621 | |
2100 | |
1622 | void encode (JSON *self, SV *scalar) |
2101 | void encode (JSON *self, SV *scalar) |
1623 | PPCODE: |
2102 | PPCODE: |
1624 | XPUSHs (encode_json (scalar, self)); |
2103 | PUTBACK; scalar = encode_json (scalar, self); SPAGAIN; |
|
|
2104 | XPUSHs (scalar); |
1625 | |
2105 | |
1626 | void decode (JSON *self, SV *jsonstr) |
2106 | void decode (JSON *self, SV *jsonstr) |
1627 | PPCODE: |
2107 | PPCODE: |
1628 | XPUSHs (decode_json (jsonstr, self, 0)); |
2108 | PUTBACK; jsonstr = decode_json (jsonstr, self, 0); SPAGAIN; |
|
|
2109 | XPUSHs (jsonstr); |
1629 | |
2110 | |
1630 | void decode_prefix (JSON *self, SV *jsonstr) |
2111 | void decode_prefix (JSON *self, SV *jsonstr) |
1631 | PPCODE: |
2112 | PPCODE: |
1632 | { |
2113 | { |
1633 | UV offset; |
2114 | SV *sv; |
|
|
2115 | STRLEN offset; |
|
|
2116 | PUTBACK; sv = decode_json (jsonstr, self, &offset); SPAGAIN; |
1634 | EXTEND (SP, 2); |
2117 | EXTEND (SP, 2); |
1635 | PUSHs (decode_json (jsonstr, self, &offset)); |
2118 | PUSHs (sv); |
1636 | PUSHs (sv_2mortal (newSVuv (offset))); |
2119 | PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, SvPV_nolen (jsonstr) + offset)))); |
|
|
2120 | } |
|
|
2121 | |
|
|
2122 | void incr_parse (JSON *self, SV *jsonstr = 0) |
|
|
2123 | PPCODE: |
|
|
2124 | { |
|
|
2125 | if (!self->incr_text) |
|
|
2126 | self->incr_text = newSVpvn ("", 0); |
|
|
2127 | |
|
|
2128 | /* if utf8-ness doesn't match the decoder, need to upgrade/downgrade */ |
|
|
2129 | if (!DECODE_WANTS_OCTETS (self) == !SvUTF8 (self->incr_text)) |
|
|
2130 | if (DECODE_WANTS_OCTETS (self)) |
|
|
2131 | { |
|
|
2132 | if (self->incr_pos) |
|
|
2133 | self->incr_pos = utf8_length ((U8 *)SvPVX (self->incr_text), |
|
|
2134 | (U8 *)SvPVX (self->incr_text) + self->incr_pos); |
|
|
2135 | |
|
|
2136 | sv_utf8_downgrade (self->incr_text, 0); |
|
|
2137 | } |
|
|
2138 | else |
|
|
2139 | { |
|
|
2140 | sv_utf8_upgrade (self->incr_text); |
|
|
2141 | |
|
|
2142 | if (self->incr_pos) |
|
|
2143 | self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos) |
|
|
2144 | - (U8 *)SvPVX (self->incr_text); |
|
|
2145 | } |
|
|
2146 | |
|
|
2147 | // append data, if any |
|
|
2148 | if (jsonstr) |
|
|
2149 | { |
|
|
2150 | /* make sure both strings have same encoding */ |
|
|
2151 | if (SvUTF8 (jsonstr) != SvUTF8 (self->incr_text)) |
|
|
2152 | if (SvUTF8 (jsonstr)) |
|
|
2153 | sv_utf8_downgrade (jsonstr, 0); |
|
|
2154 | else |
|
|
2155 | sv_utf8_upgrade (jsonstr); |
|
|
2156 | |
|
|
2157 | /* and then just blindly append */ |
|
|
2158 | { |
|
|
2159 | STRLEN len; |
|
|
2160 | const char *str = SvPV (jsonstr, len); |
|
|
2161 | STRLEN cur = SvCUR (self->incr_text); |
|
|
2162 | |
|
|
2163 | if (SvLEN (self->incr_text) <= cur + len) |
|
|
2164 | SvGROW (self->incr_text, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); |
|
|
2165 | |
|
|
2166 | Move (str, SvEND (self->incr_text), len, char); |
|
|
2167 | SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len); |
|
|
2168 | *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there |
|
|
2169 | } |
|
|
2170 | } |
|
|
2171 | |
|
|
2172 | if (GIMME_V != G_VOID) |
|
|
2173 | do |
|
|
2174 | { |
|
|
2175 | SV *sv; |
|
|
2176 | STRLEN offset; |
|
|
2177 | |
|
|
2178 | if (!INCR_DONE (self)) |
|
|
2179 | { |
|
|
2180 | incr_parse (self); |
|
|
2181 | |
|
|
2182 | if (self->incr_pos > self->max_size && self->max_size) |
|
|
2183 | croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", |
|
|
2184 | (unsigned long)self->incr_pos, (unsigned long)self->max_size); |
|
|
2185 | |
|
|
2186 | if (!INCR_DONE (self)) |
|
|
2187 | { |
|
|
2188 | // as an optimisation, do not accumulate white space in the incr buffer |
|
|
2189 | if (self->incr_mode == INCR_M_WS && self->incr_pos) |
|
|
2190 | { |
|
|
2191 | self->incr_pos = 0; |
|
|
2192 | SvCUR_set (self->incr_text, 0); |
|
|
2193 | } |
|
|
2194 | |
|
|
2195 | break; |
|
|
2196 | } |
|
|
2197 | } |
|
|
2198 | |
|
|
2199 | PUTBACK; sv = decode_json (self->incr_text, self, &offset); SPAGAIN; |
|
|
2200 | XPUSHs (sv); |
|
|
2201 | |
|
|
2202 | self->incr_pos -= offset; |
|
|
2203 | self->incr_nest = 0; |
|
|
2204 | self->incr_mode = 0; |
|
|
2205 | |
|
|
2206 | sv_chop (self->incr_text, SvPVX (self->incr_text) + offset); |
|
|
2207 | } |
|
|
2208 | while (GIMME_V == G_ARRAY); |
|
|
2209 | } |
|
|
2210 | |
|
|
2211 | SV *incr_text (JSON *self) |
|
|
2212 | ATTRS: lvalue |
|
|
2213 | CODE: |
|
|
2214 | { |
|
|
2215 | if (self->incr_pos) |
|
|
2216 | croak ("incr_text can not be called when the incremental parser already started parsing"); |
|
|
2217 | |
|
|
2218 | RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef; |
|
|
2219 | } |
|
|
2220 | OUTPUT: |
|
|
2221 | RETVAL |
|
|
2222 | |
|
|
2223 | void incr_skip (JSON *self) |
|
|
2224 | CODE: |
|
|
2225 | { |
|
|
2226 | if (self->incr_pos) |
|
|
2227 | { |
|
|
2228 | sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + self->incr_pos); |
|
|
2229 | self->incr_pos = 0; |
|
|
2230 | self->incr_nest = 0; |
|
|
2231 | self->incr_mode = 0; |
|
|
2232 | } |
|
|
2233 | } |
|
|
2234 | |
|
|
2235 | void incr_reset (JSON *self) |
|
|
2236 | CODE: |
|
|
2237 | { |
|
|
2238 | SvREFCNT_dec (self->incr_text); |
|
|
2239 | self->incr_text = 0; |
|
|
2240 | self->incr_pos = 0; |
|
|
2241 | self->incr_nest = 0; |
|
|
2242 | self->incr_mode = 0; |
1637 | } |
2243 | } |
1638 | |
2244 | |
1639 | void DESTROY (JSON *self) |
2245 | void DESTROY (JSON *self) |
1640 | CODE: |
2246 | CODE: |
1641 | SvREFCNT_dec (self->cb_sk_object); |
2247 | SvREFCNT_dec (self->cb_sk_object); |
1642 | SvREFCNT_dec (self->cb_object); |
2248 | SvREFCNT_dec (self->cb_object); |
|
|
2249 | SvREFCNT_dec (self->incr_text); |
1643 | |
2250 | |
1644 | PROTOTYPES: ENABLE |
2251 | PROTOTYPES: ENABLE |
1645 | |
2252 | |
1646 | void encode_json (SV *scalar) |
2253 | void encode_json (SV *scalar) |
1647 | PPCODE: |
2254 | PPCODE: |
1648 | { |
2255 | { |
1649 | JSON json = { F_DEFAULT | F_UTF8 }; |
2256 | JSON json; |
1650 | XPUSHs (encode_json (scalar, &json)); |
2257 | json_init (&json); |
|
|
2258 | json.flags |= F_UTF8; |
|
|
2259 | PUTBACK; scalar = encode_json (scalar, &json); SPAGAIN; |
|
|
2260 | XPUSHs (scalar); |
1651 | } |
2261 | } |
1652 | |
2262 | |
1653 | void decode_json (SV *jsonstr) |
2263 | void decode_json (SV *jsonstr) |
1654 | PPCODE: |
2264 | PPCODE: |
1655 | { |
2265 | { |
1656 | JSON json = { F_DEFAULT | F_UTF8 }; |
2266 | JSON json; |
1657 | XPUSHs (decode_json (jsonstr, &json, 0)); |
2267 | json_init (&json); |
|
|
2268 | json.flags |= F_UTF8; |
|
|
2269 | PUTBACK; jsonstr = decode_json (jsonstr, &json, 0); SPAGAIN; |
|
|
2270 | XPUSHs (jsonstr); |
1658 | } |
2271 | } |
1659 | |
2272 | |