… | |
… | |
7 | #include <stdlib.h> |
7 | #include <stdlib.h> |
8 | #include <stdio.h> |
8 | #include <stdio.h> |
9 | #include <limits.h> |
9 | #include <limits.h> |
10 | #include <float.h> |
10 | #include <float.h> |
11 | |
11 | |
|
|
12 | #define ECB_NO_THREADS 1 |
12 | #include "ecb.h" |
13 | #include "ecb.h" |
13 | |
14 | |
14 | // compatibility with perl <5.18 |
15 | // compatibility with perl <5.18 |
15 | #ifndef HvNAMELEN_get |
16 | #ifndef HvNAMELEN_get |
16 | # define HvNAMELEN_get(hv) strlen (HvNAME (hv)) |
17 | # define HvNAMELEN_get(hv) strlen (HvNAME (hv)) |
… | |
… | |
19 | # define HvNAMELEN(hv) HvNAMELEN_get (hv) |
20 | # define HvNAMELEN(hv) HvNAMELEN_get (hv) |
20 | #endif |
21 | #endif |
21 | #ifndef HvNAMEUTF8 |
22 | #ifndef HvNAMEUTF8 |
22 | # define HvNAMEUTF8(hv) 0 |
23 | # define HvNAMEUTF8(hv) 0 |
23 | #endif |
24 | #endif |
|
|
25 | #ifndef SvREFCNT_dec_NN |
|
|
26 | # define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv) |
|
|
27 | #endif |
|
|
28 | |
|
|
29 | // known major and minor types |
|
|
30 | enum cbor_type |
|
|
31 | { |
|
|
32 | MAJOR_SHIFT = 5, |
|
|
33 | MINOR_MASK = 0x1f, |
|
|
34 | |
|
|
35 | MAJOR_POS_INT = 0 << MAJOR_SHIFT, |
|
|
36 | MAJOR_NEG_INT = 1 << MAJOR_SHIFT, |
|
|
37 | MAJOR_BYTES = 2 << MAJOR_SHIFT, |
|
|
38 | MAJOR_TEXT = 3 << MAJOR_SHIFT, |
|
|
39 | MAJOR_ARRAY = 4 << MAJOR_SHIFT, |
|
|
40 | MAJOR_MAP = 5 << MAJOR_SHIFT, |
|
|
41 | MAJOR_TAG = 6 << MAJOR_SHIFT, |
|
|
42 | MAJOR_MISC = 7 << MAJOR_SHIFT, |
|
|
43 | |
|
|
44 | // INT/STRING/ARRAY/MAP subtypes |
|
|
45 | LENGTH_EXT1 = 24, |
|
|
46 | LENGTH_EXT2 = 25, |
|
|
47 | LENGTH_EXT4 = 26, |
|
|
48 | LENGTH_EXT8 = 27, |
|
|
49 | |
|
|
50 | // SIMPLE types (effectively MISC subtypes) |
|
|
51 | SIMPLE_FALSE = 20, |
|
|
52 | SIMPLE_TRUE = 21, |
|
|
53 | SIMPLE_NULL = 22, |
|
|
54 | SIMPLE_UNDEF = 23, |
|
|
55 | |
|
|
56 | // MISC subtype (unused) |
|
|
57 | MISC_EXT1 = 24, |
|
|
58 | MISC_FLOAT16 = 25, |
|
|
59 | MISC_FLOAT32 = 26, |
|
|
60 | MISC_FLOAT64 = 27, |
|
|
61 | |
|
|
62 | // BYTES/TEXT/ARRAY/MAP |
|
|
63 | MINOR_INDEF = 31, |
|
|
64 | }; |
24 | |
65 | |
25 | // known tags |
66 | // known tags |
26 | enum cbor_tag |
67 | enum cbor_tag |
27 | { |
68 | { |
28 | // inofficial extensions (pending iana registration) |
69 | // extensions |
|
|
70 | CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref |
29 | CBOR_TAG_PERL_OBJECT = 24, // http://cbor.schmorp.de/perl-object |
71 | CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object |
30 | CBOR_TAG_GENERIC_OBJECT = 25, // http://cbor.schmorp.de/generic-object |
72 | CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object |
31 | CBOR_TAG_VALUE_SHAREABLE = 26, // http://cbor.schmorp.de/value-sharing |
73 | CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing |
32 | CBOR_TAG_VALUE_SHAREDREF = 27, // http://cbor.schmorp.de/value-sharing |
74 | CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing |
33 | CBOR_TAG_STRINGREF_NAMESPACE = 65537, // http://cbor.schmorp.de/stringref |
75 | CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref |
34 | CBOR_TAG_STRINGREF = 28, // http://cbor.schmorp.de/stringref |
|
|
35 | CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection |
76 | CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection |
36 | |
77 | |
37 | // rfc7049 |
78 | // rfc7049 |
38 | CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 |
79 | CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 |
39 | CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any |
80 | CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any |
40 | CBOR_TAG_POS_BIGNUM = 2, // byte string |
81 | CBOR_TAG_POS_BIGNUM = 2, // byte string |
41 | CBOR_TAG_NEG_BIGNUM = 3, // byte string |
82 | CBOR_TAG_NEG_BIGNUM = 3, // byte string |
42 | CBOR_TAG_DECIMAL = 4, // decimal fraction, array |
83 | CBOR_TAG_DECIMAL = 4, // decimal fraction, array |
43 | CBOR_TAG_BIGFLOAT = 5, // array |
84 | CBOR_TAG_BIGFLOAT = 5, // array |
44 | |
85 | |
45 | CBOR_TAG_CONV_B64U = 21, // base64url, any |
86 | CBOR_TAG_CONV_B64U = 21, // base64url, any |
46 | CBOR_TAG_CONV_B64 = 22, // base64, any |
87 | CBOR_TAG_CONV_B64 = 22, // base64, any |
47 | CBOR_TAG_CONV_HEX = 23, // base16, any |
88 | CBOR_TAG_CONV_HEX = 23, // base16, any |
48 | CBOR_TAG_CBOR = 24, // embedded cbor, byte string |
89 | CBOR_TAG_CBOR = 24, // embedded cbor, byte string |
49 | |
90 | |
50 | CBOR_TAG_URI = 32, // URI rfc3986, utf-8 |
91 | CBOR_TAG_URI = 32, // URI rfc3986, utf-8 |
51 | CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8 |
92 | CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8 |
52 | CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8 |
93 | CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8 |
53 | CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8 |
94 | CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8 |
54 | CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 |
95 | CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 |
55 | |
96 | |
56 | CBOR_TAG_MAGIC = 55799 // self-describe cbor |
97 | CBOR_TAG_MAGIC = 55799, // self-describe cbor |
57 | }; |
98 | }; |
58 | |
99 | |
59 | #define F_SHRINK 0x00000001UL |
100 | #define F_SHRINK 0x00000001UL |
60 | #define F_ALLOW_UNKNOWN 0x00000002UL |
101 | #define F_ALLOW_UNKNOWN 0x00000002UL |
61 | #define F_ALLOW_SHARING 0x00000004UL //TODO |
102 | #define F_ALLOW_SHARING 0x00000004UL |
|
|
103 | #define F_ALLOW_CYCLES 0x00000008UL |
62 | #define F_DEDUP_STRINGS 0x00000008UL //TODO |
104 | #define F_PACK_STRINGS 0x00000010UL |
|
|
105 | #define F_UTF8_STRINGS 0x00000020UL |
|
|
106 | #define F_UTF8_KEYS 0x00000040UL |
|
|
107 | #define F_VALIDATE_UTF8 0x00000080UL |
63 | |
108 | |
64 | #define INIT_SIZE 32 // initial scalar size to be allocated |
109 | #define INIT_SIZE 32 // initial scalar size to be allocated |
65 | |
110 | |
66 | #define SB do { |
111 | #define SB do { |
67 | #define SE } while (0) |
112 | #define SE } while (0) |
… | |
… | |
79 | # define CBOR_SLOW 0 |
124 | # define CBOR_SLOW 0 |
80 | # define CBOR_STASH cbor_stash |
125 | # define CBOR_STASH cbor_stash |
81 | #endif |
126 | #endif |
82 | |
127 | |
83 | static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS:: |
128 | static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS:: |
84 | static SV *types_true, *types_false, *types_error, *sv_cbor; |
129 | static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter; |
85 | |
130 | |
86 | typedef struct { |
131 | typedef struct { |
87 | U32 flags; |
132 | U32 flags; |
88 | U32 max_depth; |
133 | U32 max_depth; |
89 | STRLEN max_size; |
134 | STRLEN max_size; |
|
|
135 | SV *filter; |
|
|
136 | |
|
|
137 | // for the incremental parser |
|
|
138 | STRLEN incr_pos; // the current offset into the text |
|
|
139 | STRLEN incr_need; // minimum bytes needed to decode |
|
|
140 | AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef. |
90 | } CBOR; |
141 | } CBOR; |
91 | |
142 | |
92 | ecb_inline void |
143 | ecb_inline void |
93 | cbor_init (CBOR *cbor) |
144 | cbor_init (CBOR *cbor) |
94 | { |
145 | { |
95 | Zero (cbor, 1, CBOR); |
146 | Zero (cbor, 1, CBOR); |
96 | cbor->max_depth = 512; |
147 | cbor->max_depth = 512; |
|
|
148 | } |
|
|
149 | |
|
|
150 | ecb_inline void |
|
|
151 | cbor_free (CBOR *cbor) |
|
|
152 | { |
|
|
153 | SvREFCNT_dec (cbor->filter); |
|
|
154 | SvREFCNT_dec (cbor->incr_count); |
97 | } |
155 | } |
98 | |
156 | |
99 | ///////////////////////////////////////////////////////////////////////////// |
157 | ///////////////////////////////////////////////////////////////////////////// |
100 | // utility functions |
158 | // utility functions |
101 | |
159 | |
… | |
… | |
131 | { |
189 | { |
132 | return idx > 23 |
190 | return idx > 23 |
133 | ? idx > 0xffU |
191 | ? idx > 0xffU |
134 | ? idx > 0xffffU |
192 | ? idx > 0xffffU |
135 | ? idx > 0xffffffffU |
193 | ? idx > 0xffffffffU |
|
|
194 | ? 11 |
136 | ? 7 |
195 | : 7 |
137 | : 6 |
|
|
138 | : 5 |
196 | : 5 |
139 | : 4 |
197 | : 4 |
140 | : 3; |
198 | : 3; |
141 | } |
199 | } |
142 | |
200 | |
… | |
… | |
179 | static void |
237 | static void |
180 | encode_uint (enc_t *enc, int major, UV len) |
238 | encode_uint (enc_t *enc, int major, UV len) |
181 | { |
239 | { |
182 | need (enc, 9); |
240 | need (enc, 9); |
183 | |
241 | |
184 | if (len < 24) |
242 | if (ecb_expect_true (len < LENGTH_EXT1)) |
185 | *enc->cur++ = major | len; |
243 | *enc->cur++ = major | len; |
186 | else if (len <= 0xff) |
244 | else if (ecb_expect_true (len <= 0xffU)) |
187 | { |
245 | { |
188 | *enc->cur++ = major | 24; |
246 | *enc->cur++ = major | LENGTH_EXT1; |
189 | *enc->cur++ = len; |
247 | *enc->cur++ = len; |
190 | } |
248 | } |
191 | else if (len <= 0xffff) |
249 | else if (len <= 0xffffU) |
192 | { |
250 | { |
193 | *enc->cur++ = major | 25; |
251 | *enc->cur++ = major | LENGTH_EXT2; |
194 | *enc->cur++ = len >> 8; |
252 | *enc->cur++ = len >> 8; |
195 | *enc->cur++ = len; |
253 | *enc->cur++ = len; |
196 | } |
254 | } |
197 | else if (len <= 0xffffffff) |
255 | else if (len <= 0xffffffffU) |
198 | { |
256 | { |
199 | *enc->cur++ = major | 26; |
257 | *enc->cur++ = major | LENGTH_EXT4; |
200 | *enc->cur++ = len >> 24; |
258 | *enc->cur++ = len >> 24; |
201 | *enc->cur++ = len >> 16; |
259 | *enc->cur++ = len >> 16; |
202 | *enc->cur++ = len >> 8; |
260 | *enc->cur++ = len >> 8; |
203 | *enc->cur++ = len; |
261 | *enc->cur++ = len; |
204 | } |
262 | } |
205 | else |
263 | else |
206 | { |
264 | { |
207 | *enc->cur++ = major | 27; |
265 | *enc->cur++ = major | LENGTH_EXT8; |
208 | *enc->cur++ = len >> 56; |
266 | *enc->cur++ = len >> 56; |
209 | *enc->cur++ = len >> 48; |
267 | *enc->cur++ = len >> 48; |
210 | *enc->cur++ = len >> 40; |
268 | *enc->cur++ = len >> 40; |
211 | *enc->cur++ = len >> 32; |
269 | *enc->cur++ = len >> 32; |
212 | *enc->cur++ = len >> 24; |
270 | *enc->cur++ = len >> 24; |
… | |
… | |
217 | } |
275 | } |
218 | |
276 | |
219 | ecb_inline void |
277 | ecb_inline void |
220 | encode_tag (enc_t *enc, UV tag) |
278 | encode_tag (enc_t *enc, UV tag) |
221 | { |
279 | { |
222 | encode_uint (enc, 0xc0, tag); |
280 | encode_uint (enc, MAJOR_TAG, tag); |
223 | } |
281 | } |
224 | |
282 | |
225 | static void |
283 | // exceptional (hopefully) slow path for byte strings that need to be utf8-encoded |
|
|
284 | ecb_noinline static void |
226 | encode_str (enc_t *enc, int utf8, char *str, STRLEN len) |
285 | encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len) |
227 | { |
286 | { |
|
|
287 | STRLEN ulen = len; |
|
|
288 | U8 *p, *pend = (U8 *)str + len; |
|
|
289 | |
|
|
290 | for (p = (U8 *)str; p < pend; ++p) |
|
|
291 | ulen += *p >> 7; // count set high bits |
|
|
292 | |
|
|
293 | encode_uint (enc, MAJOR_TEXT, ulen); |
|
|
294 | |
|
|
295 | need (enc, ulen); |
|
|
296 | for (p = (U8 *)str; p < pend; ++p) |
|
|
297 | if (*p < 0x80) |
|
|
298 | *enc->cur++ = *p; |
|
|
299 | else |
|
|
300 | { |
|
|
301 | *enc->cur++ = 0xc0 + (*p >> 6); |
|
|
302 | *enc->cur++ = 0x80 + (*p & 63); |
|
|
303 | } |
|
|
304 | } |
|
|
305 | |
|
|
306 | ecb_inline void |
|
|
307 | encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len) |
|
|
308 | { |
|
|
309 | if (ecb_expect_false (upgrade_utf8)) |
|
|
310 | if (!utf8) |
|
|
311 | { |
|
|
312 | encode_str_utf8 (enc, utf8, str, len); |
|
|
313 | return; |
|
|
314 | } |
|
|
315 | |
|
|
316 | encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len); |
|
|
317 | need (enc, len); |
|
|
318 | memcpy (enc->cur, str, len); |
|
|
319 | enc->cur += len; |
|
|
320 | } |
|
|
321 | |
|
|
322 | ecb_inline |
|
|
323 | encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len) |
|
|
324 | { |
228 | if (ecb_expect_false (enc->cbor.flags & F_DEDUP_STRINGS)) |
325 | if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS)) |
229 | { |
326 | { |
230 | SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1); |
327 | SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1); |
231 | |
328 | |
232 | if (SvOK (*svp)) |
329 | if (SvOK (*svp)) |
233 | { |
330 | { |
234 | // already registered, use stringref |
331 | // already registered, use stringref |
235 | encode_tag (enc, CBOR_TAG_STRINGREF); |
332 | encode_tag (enc, CBOR_TAG_STRINGREF); |
236 | encode_uint (enc, 0x00, SvUV (*svp)); |
333 | encode_uint (enc, MAJOR_POS_INT, SvUV (*svp)); |
237 | return; |
334 | return; |
238 | } |
335 | } |
239 | else if (len >= minimum_string_length (enc->stringref_idx)) |
336 | else if (len >= minimum_string_length (enc->stringref_idx)) |
240 | { |
337 | { |
241 | // register only |
338 | // register only |
242 | sv_setuv (*svp, enc->stringref_idx); |
339 | sv_setuv (*svp, enc->stringref_idx); |
243 | ++enc->stringref_idx; |
340 | ++enc->stringref_idx; |
244 | } |
341 | } |
245 | } |
342 | } |
246 | |
343 | |
247 | encode_uint (enc, utf8 ? 0x60 : 0x40, len); |
344 | encode_str (enc, upgrade_utf8, utf8, str, len); |
248 | need (enc, len); |
|
|
249 | memcpy (enc->cur, str, len); |
|
|
250 | enc->cur += len; |
|
|
251 | } |
345 | } |
252 | |
346 | |
253 | static void encode_sv (enc_t *enc, SV *sv); |
347 | static void encode_sv (enc_t *enc, SV *sv); |
254 | |
348 | |
255 | static void |
349 | static void |
… | |
… | |
260 | if (enc->depth >= enc->cbor.max_depth) |
354 | if (enc->depth >= enc->cbor.max_depth) |
261 | croak (ERR_NESTING_EXCEEDED); |
355 | croak (ERR_NESTING_EXCEEDED); |
262 | |
356 | |
263 | ++enc->depth; |
357 | ++enc->depth; |
264 | |
358 | |
265 | encode_uint (enc, 0x80, len + 1); |
359 | encode_uint (enc, MAJOR_ARRAY, len + 1); |
266 | |
360 | |
|
|
361 | if (SvMAGICAL (av)) |
267 | for (i = 0; i <= len; ++i) |
362 | for (i = 0; i <= len; ++i) |
268 | { |
363 | { |
269 | SV **svp = av_fetch (av, i, 0); |
364 | SV **svp = av_fetch (av, i, 0); |
270 | encode_sv (enc, svp ? *svp : &PL_sv_undef); |
365 | encode_sv (enc, svp ? *svp : &PL_sv_undef); |
271 | } |
366 | } |
|
|
367 | else |
|
|
368 | for (i = 0; i <= len; ++i) |
|
|
369 | { |
|
|
370 | SV *sv = AvARRAY (av)[i]; |
|
|
371 | encode_sv (enc, sv ? sv : &PL_sv_undef); |
|
|
372 | } |
272 | |
373 | |
273 | --enc->depth; |
374 | --enc->depth; |
274 | } |
|
|
275 | |
|
|
276 | ecb_inline void |
|
|
277 | encode_he (enc_t *enc, HE *he) |
|
|
278 | { |
|
|
279 | } |
375 | } |
280 | |
376 | |
281 | static void |
377 | static void |
282 | encode_hv (enc_t *enc, HV *hv) |
378 | encode_hv (enc_t *enc, HV *hv) |
283 | { |
379 | { |
… | |
… | |
290 | |
386 | |
291 | int pairs = hv_iterinit (hv); |
387 | int pairs = hv_iterinit (hv); |
292 | int mg = SvMAGICAL (hv); |
388 | int mg = SvMAGICAL (hv); |
293 | |
389 | |
294 | if (mg) |
390 | if (mg) |
295 | encode_ch (enc, 0xa0 | 31); |
391 | encode_ch (enc, MAJOR_MAP | MINOR_INDEF); |
296 | else |
392 | else |
297 | encode_uint (enc, 0xa0, pairs); |
393 | encode_uint (enc, MAJOR_MAP, pairs); |
298 | |
394 | |
299 | while ((he = hv_iternext (hv))) |
395 | while ((he = hv_iternext (hv))) |
300 | { |
396 | { |
301 | if (HeKLEN (he) == HEf_SVKEY) |
397 | if (HeKLEN (he) == HEf_SVKEY) |
302 | encode_sv (enc, HeSVKEY (he)); |
398 | encode_sv (enc, HeSVKEY (he)); |
303 | else |
399 | else |
304 | encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); |
400 | encode_strref (enc, enc->cbor.flags & (F_UTF8_KEYS | F_UTF8_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he)); |
305 | |
401 | |
306 | encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); |
402 | encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); |
307 | } |
403 | } |
308 | |
404 | |
309 | if (mg) |
405 | if (mg) |
310 | encode_ch (enc, 0xe0 | 31); |
406 | encode_ch (enc, MAJOR_MISC | MINOR_INDEF); |
311 | |
407 | |
312 | --enc->depth; |
408 | --enc->depth; |
313 | } |
409 | } |
314 | |
410 | |
315 | // encode objects, arrays and special \0=false and \1=true values. |
411 | // encode objects, arrays and special \0=false and \1=true values. |
316 | static void |
412 | static void |
317 | encode_rv (enc_t *enc, SV *sv) |
413 | encode_rv (enc_t *enc, SV *sv) |
318 | { |
414 | { |
319 | SvGETMAGIC (sv); |
415 | SvGETMAGIC (sv); |
320 | |
|
|
321 | if (ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING) |
|
|
322 | && ecb_expect_false (SvREFCNT (sv) > 1)) |
|
|
323 | { |
|
|
324 | if (!enc->shareable) |
|
|
325 | enc->shareable = (HV *)sv_2mortal ((SV *)newHV ()); |
|
|
326 | |
|
|
327 | SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1); |
|
|
328 | |
|
|
329 | if (SvOK (*svp)) |
|
|
330 | { |
|
|
331 | encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF); |
|
|
332 | encode_uint (enc, 0x00, SvUV (*svp)); |
|
|
333 | return; |
|
|
334 | } |
|
|
335 | else |
|
|
336 | { |
|
|
337 | sv_setuv (*svp, enc->shareable_idx); |
|
|
338 | ++enc->shareable_idx; |
|
|
339 | encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE); |
|
|
340 | } |
|
|
341 | } |
|
|
342 | |
416 | |
343 | svtype svt = SvTYPE (sv); |
417 | svtype svt = SvTYPE (sv); |
344 | |
418 | |
345 | if (ecb_expect_false (SvOBJECT (sv))) |
419 | if (ecb_expect_false (SvOBJECT (sv))) |
346 | { |
420 | { |
… | |
… | |
353 | HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash |
427 | HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash |
354 | ? cbor_tagged_stash |
428 | ? cbor_tagged_stash |
355 | : gv_stashpv ("CBOR::XS::Tagged" , 1); |
429 | : gv_stashpv ("CBOR::XS::Tagged" , 1); |
356 | |
430 | |
357 | HV *stash = SvSTASH (sv); |
431 | HV *stash = SvSTASH (sv); |
358 | GV *method; |
|
|
359 | |
432 | |
360 | if (stash == boolean_stash) |
433 | if (stash == boolean_stash) |
361 | encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); |
434 | { |
|
|
435 | encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE); |
|
|
436 | return; |
|
|
437 | } |
362 | else if (stash == error_stash) |
438 | else if (stash == error_stash) |
363 | encode_ch (enc, 0xe0 | 23); |
439 | { |
|
|
440 | encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF); |
|
|
441 | return; |
|
|
442 | } |
364 | else if (stash == tagged_stash) |
443 | else if (stash == tagged_stash) |
365 | { |
444 | { |
366 | if (svt != SVt_PVAV) |
445 | if (svt != SVt_PVAV) |
367 | croak ("encountered CBOR::XS::Tagged object that isn't an array"); |
446 | croak ("encountered CBOR::XS::Tagged object that isn't an array"); |
368 | |
447 | |
369 | encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); |
448 | encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1))); |
370 | encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); |
449 | encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); |
|
|
450 | |
|
|
451 | return; |
|
|
452 | } |
|
|
453 | } |
|
|
454 | |
|
|
455 | if (ecb_expect_false (SvREFCNT (sv) > 1) |
|
|
456 | && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)) |
|
|
457 | { |
|
|
458 | if (!enc->shareable) |
|
|
459 | enc->shareable = (HV *)sv_2mortal ((SV *)newHV ()); |
|
|
460 | |
|
|
461 | SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1); |
|
|
462 | |
|
|
463 | if (SvOK (*svp)) |
371 | } |
464 | { |
|
|
465 | encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF); |
|
|
466 | encode_uint (enc, MAJOR_POS_INT, SvUV (*svp)); |
|
|
467 | return; |
|
|
468 | } |
|
|
469 | else |
|
|
470 | { |
|
|
471 | sv_setuv (*svp, enc->shareable_idx); |
|
|
472 | ++enc->shareable_idx; |
|
|
473 | encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE); |
|
|
474 | } |
|
|
475 | } |
|
|
476 | |
|
|
477 | if (ecb_expect_false (SvOBJECT (sv))) |
|
|
478 | { |
|
|
479 | HV *stash = SvSTASH (sv); |
|
|
480 | GV *method; |
|
|
481 | |
372 | else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) |
482 | if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) |
373 | { |
483 | { |
374 | dSP; |
484 | dSP; |
375 | |
485 | |
376 | ENTER; SAVETMPS; PUSHMARK (SP); |
486 | ENTER; SAVETMPS; |
|
|
487 | PUSHMARK (SP); |
377 | // we re-bless the reference to get overload and other niceties right |
488 | // we re-bless the reference to get overload and other niceties right |
378 | XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); |
489 | XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); |
379 | |
490 | |
380 | PUTBACK; |
491 | PUTBACK; |
381 | // G_SCALAR ensures that return value is 1 |
492 | // G_SCALAR ensures that return value is 1 |
… | |
… | |
394 | } |
505 | } |
395 | else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) |
506 | else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) |
396 | { |
507 | { |
397 | dSP; |
508 | dSP; |
398 | |
509 | |
399 | ENTER; SAVETMPS; PUSHMARK (SP); |
510 | ENTER; SAVETMPS; |
|
|
511 | SAVESTACK_POS (); |
|
|
512 | PUSHMARK (SP); |
400 | EXTEND (SP, 2); |
513 | EXTEND (SP, 2); |
401 | // we re-bless the reference to get overload and other niceties right |
514 | // we re-bless the reference to get overload and other niceties right |
402 | PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); |
515 | PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); |
403 | PUSHs (sv_cbor); |
516 | PUSHs (sv_cbor); |
404 | |
517 | |
… | |
… | |
409 | // catch this surprisingly common error |
522 | // catch this surprisingly common error |
410 | if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) |
523 | if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) |
411 | croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); |
524 | croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); |
412 | |
525 | |
413 | encode_tag (enc, CBOR_TAG_PERL_OBJECT); |
526 | encode_tag (enc, CBOR_TAG_PERL_OBJECT); |
414 | encode_uint (enc, 0x80, count + 1); |
527 | encode_uint (enc, MAJOR_ARRAY, count + 1); |
415 | encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); |
528 | encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); |
416 | |
529 | |
417 | while (count) |
530 | while (count) |
418 | encode_sv (enc, SP[1 - count--]); |
531 | encode_sv (enc, SP[1 - count--]); |
419 | |
532 | |
420 | PUTBACK; |
533 | PUTBACK; |
… | |
… | |
441 | { |
554 | { |
442 | double nv = SvNVX (sv); |
555 | double nv = SvNVX (sv); |
443 | |
556 | |
444 | need (enc, 9); |
557 | need (enc, 9); |
445 | |
558 | |
446 | if (ecb_expect_false (nv == (U32)nv)) |
559 | if (ecb_expect_false (nv == (NV)(U32)nv)) |
447 | encode_uint (enc, 0x00, (U32)nv); |
560 | encode_uint (enc, MAJOR_POS_INT, (U32)nv); |
448 | //TODO: maybe I32? |
561 | //TODO: maybe I32? |
449 | else if (ecb_expect_false (nv == (float)nv)) |
562 | else if (ecb_expect_false (nv == (float)nv)) |
450 | { |
563 | { |
451 | uint32_t fp = ecb_float_to_binary32 (nv); |
564 | uint32_t fp = ecb_float_to_binary32 (nv); |
452 | |
565 | |
453 | *enc->cur++ = 0xe0 | 26; |
566 | *enc->cur++ = MAJOR_MISC | MISC_FLOAT32; |
454 | |
567 | |
455 | if (!ecb_big_endian ()) |
568 | if (!ecb_big_endian ()) |
456 | fp = ecb_bswap32 (fp); |
569 | fp = ecb_bswap32 (fp); |
457 | |
570 | |
458 | memcpy (enc->cur, &fp, 4); |
571 | memcpy (enc->cur, &fp, 4); |
… | |
… | |
460 | } |
573 | } |
461 | else |
574 | else |
462 | { |
575 | { |
463 | uint64_t fp = ecb_double_to_binary64 (nv); |
576 | uint64_t fp = ecb_double_to_binary64 (nv); |
464 | |
577 | |
465 | *enc->cur++ = 0xe0 | 27; |
578 | *enc->cur++ = MAJOR_MISC | MISC_FLOAT64; |
466 | |
579 | |
467 | if (!ecb_big_endian ()) |
580 | if (!ecb_big_endian ()) |
468 | fp = ecb_bswap64 (fp); |
581 | fp = ecb_bswap64 (fp); |
469 | |
582 | |
470 | memcpy (enc->cur, &fp, 8); |
583 | memcpy (enc->cur, &fp, 8); |
… | |
… | |
479 | |
592 | |
480 | if (SvPOKp (sv)) |
593 | if (SvPOKp (sv)) |
481 | { |
594 | { |
482 | STRLEN len; |
595 | STRLEN len; |
483 | char *str = SvPV (sv, len); |
596 | char *str = SvPV (sv, len); |
484 | encode_str (enc, SvUTF8 (sv), str, len); |
597 | encode_strref (enc, enc->cbor.flags & F_UTF8_STRINGS, SvUTF8 (sv), str, len); |
485 | } |
598 | } |
486 | else if (SvNOKp (sv)) |
599 | else if (SvNOKp (sv)) |
487 | encode_nv (enc, sv); |
600 | encode_nv (enc, sv); |
488 | else if (SvIOKp (sv)) |
601 | else if (SvIOKp (sv)) |
489 | { |
602 | { |
490 | if (SvIsUV (sv)) |
603 | if (SvIsUV (sv)) |
491 | encode_uint (enc, 0x00, SvUVX (sv)); |
604 | encode_uint (enc, MAJOR_POS_INT, SvUVX (sv)); |
492 | else if (SvIVX (sv) >= 0) |
605 | else if (SvIVX (sv) >= 0) |
493 | encode_uint (enc, 0x00, SvIVX (sv)); |
606 | encode_uint (enc, MAJOR_POS_INT, SvIVX (sv)); |
494 | else |
607 | else |
495 | encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); |
608 | encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1)); |
496 | } |
609 | } |
497 | else if (SvROK (sv)) |
610 | else if (SvROK (sv)) |
498 | encode_rv (enc, SvRV (sv)); |
611 | encode_rv (enc, SvRV (sv)); |
499 | else if (!SvOK (sv)) |
612 | else if (!SvOK (sv)) |
500 | encode_ch (enc, 0xe0 | 22); |
613 | encode_ch (enc, MAJOR_MISC | SIMPLE_NULL); |
501 | else if (enc->cbor.flags & F_ALLOW_UNKNOWN) |
614 | else if (enc->cbor.flags & F_ALLOW_UNKNOWN) |
502 | encode_ch (enc, 0xe0 | 23); |
615 | encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF); |
503 | else |
616 | else |
504 | croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", |
617 | croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", |
505 | SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); |
618 | SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); |
506 | } |
619 | } |
507 | |
620 | |
508 | static SV * |
621 | static SV * |
509 | encode_cbor (SV *scalar, CBOR *cbor) |
622 | encode_cbor (SV *scalar, CBOR *cbor) |
510 | { |
623 | { |
511 | enc_t enc = { }; |
624 | enc_t enc = { 0 }; |
512 | |
625 | |
513 | enc.cbor = *cbor; |
626 | enc.cbor = *cbor; |
514 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
627 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
515 | enc.cur = SvPVX (enc.sv); |
628 | enc.cur = SvPVX (enc.sv); |
516 | enc.end = SvEND (enc.sv); |
629 | enc.end = SvEND (enc.sv); |
517 | |
630 | |
518 | SvPOK_only (enc.sv); |
631 | SvPOK_only (enc.sv); |
519 | |
632 | |
520 | if (cbor->flags & F_DEDUP_STRINGS) |
633 | if (cbor->flags & F_PACK_STRINGS) |
521 | { |
634 | { |
522 | encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE); |
635 | encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE); |
523 | enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ()); |
636 | enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ()); |
524 | enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ()); |
637 | enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ()); |
525 | } |
638 | } |
… | |
… | |
547 | CBOR cbor; |
660 | CBOR cbor; |
548 | U32 depth; // recursion depth |
661 | U32 depth; // recursion depth |
549 | U32 maxdepth; // recursion depth limit |
662 | U32 maxdepth; // recursion depth limit |
550 | AV *shareable; |
663 | AV *shareable; |
551 | AV *stringref; |
664 | AV *stringref; |
|
|
665 | SV *decode_tagged; |
552 | } dec_t; |
666 | } dec_t; |
553 | |
667 | |
554 | #define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE |
668 | #define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE |
555 | |
669 | |
556 | #define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") |
670 | #define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") |
… | |
… | |
559 | #define DEC_DEC_DEPTH --dec->depth |
673 | #define DEC_DEC_DEPTH --dec->depth |
560 | |
674 | |
561 | static UV |
675 | static UV |
562 | decode_uint (dec_t *dec) |
676 | decode_uint (dec_t *dec) |
563 | { |
677 | { |
564 | switch (*dec->cur & 31) |
678 | U8 m = *dec->cur & MINOR_MASK; |
565 | { |
679 | ++dec->cur; |
566 | case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7: |
|
|
567 | case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15: |
|
|
568 | case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23: |
|
|
569 | return *dec->cur++ & 31; |
|
|
570 | |
680 | |
571 | case 24: |
681 | if (ecb_expect_true (m < LENGTH_EXT1)) |
|
|
682 | return m; |
|
|
683 | else if (ecb_expect_true (m == LENGTH_EXT1)) |
|
|
684 | { |
572 | WANT (2); |
685 | WANT (1); |
573 | dec->cur += 2; |
686 | dec->cur += 1; |
574 | return dec->cur[-1]; |
687 | return dec->cur[-1]; |
575 | |
688 | } |
576 | case 25: |
689 | else if (ecb_expect_true (m == LENGTH_EXT2)) |
|
|
690 | { |
577 | WANT (3); |
691 | WANT (2); |
578 | dec->cur += 3; |
692 | dec->cur += 2; |
579 | return (((UV)dec->cur[-2]) << 8) |
693 | return (((UV)dec->cur[-2]) << 8) |
580 | | ((UV)dec->cur[-1]); |
694 | | ((UV)dec->cur[-1]); |
581 | |
695 | } |
582 | case 26: |
696 | else if (ecb_expect_true (m == LENGTH_EXT4)) |
|
|
697 | { |
583 | WANT (5); |
698 | WANT (4); |
584 | dec->cur += 5; |
699 | dec->cur += 4; |
585 | return (((UV)dec->cur[-4]) << 24) |
700 | return (((UV)dec->cur[-4]) << 24) |
586 | | (((UV)dec->cur[-3]) << 16) |
701 | | (((UV)dec->cur[-3]) << 16) |
587 | | (((UV)dec->cur[-2]) << 8) |
702 | | (((UV)dec->cur[-2]) << 8) |
588 | | ((UV)dec->cur[-1]); |
703 | | ((UV)dec->cur[-1]); |
589 | |
704 | } |
590 | case 27: |
705 | else if (ecb_expect_true (m == LENGTH_EXT8)) |
|
|
706 | { |
591 | WANT (9); |
707 | WANT (8); |
592 | dec->cur += 9; |
708 | dec->cur += 8; |
|
|
709 | |
|
|
710 | return |
|
|
711 | #if UVSIZE < 8 |
|
|
712 | 0 |
|
|
713 | #else |
593 | return (((UV)dec->cur[-8]) << 56) |
714 | (((UV)dec->cur[-8]) << 56) |
594 | | (((UV)dec->cur[-7]) << 48) |
715 | | (((UV)dec->cur[-7]) << 48) |
595 | | (((UV)dec->cur[-6]) << 40) |
716 | | (((UV)dec->cur[-6]) << 40) |
596 | | (((UV)dec->cur[-5]) << 32) |
717 | | (((UV)dec->cur[-5]) << 32) |
|
|
718 | #endif |
597 | | (((UV)dec->cur[-4]) << 24) |
719 | | (((UV)dec->cur[-4]) << 24) |
598 | | (((UV)dec->cur[-3]) << 16) |
720 | | (((UV)dec->cur[-3]) << 16) |
599 | | (((UV)dec->cur[-2]) << 8) |
721 | | (((UV)dec->cur[-2]) << 8) |
600 | | ((UV)dec->cur[-1]); |
722 | | ((UV)dec->cur[-1]); |
601 | |
723 | } |
602 | default: |
724 | else |
603 | ERR ("corrupted CBOR data (unsupported integer minor encoding)"); |
725 | ERR ("corrupted CBOR data (unsupported integer minor encoding)"); |
604 | } |
|
|
605 | |
726 | |
606 | fail: |
727 | fail: |
607 | return 0; |
728 | return 0; |
608 | } |
729 | } |
609 | |
730 | |
… | |
… | |
614 | { |
735 | { |
615 | AV *av = newAV (); |
736 | AV *av = newAV (); |
616 | |
737 | |
617 | DEC_INC_DEPTH; |
738 | DEC_INC_DEPTH; |
618 | |
739 | |
619 | if ((*dec->cur & 31) == 31) |
740 | if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF)) |
620 | { |
741 | { |
621 | ++dec->cur; |
742 | ++dec->cur; |
622 | |
743 | |
623 | for (;;) |
744 | for (;;) |
624 | { |
745 | { |
625 | WANT (1); |
746 | WANT (1); |
626 | |
747 | |
627 | if (*dec->cur == (0xe0 | 31)) |
748 | if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) |
628 | { |
749 | { |
629 | ++dec->cur; |
750 | ++dec->cur; |
630 | break; |
751 | break; |
631 | } |
752 | } |
632 | |
753 | |
… | |
… | |
635 | } |
756 | } |
636 | else |
757 | else |
637 | { |
758 | { |
638 | int i, len = decode_uint (dec); |
759 | int i, len = decode_uint (dec); |
639 | |
760 | |
|
|
761 | WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays |
640 | av_fill (av, len - 1); |
762 | av_fill (av, len - 1); |
641 | |
763 | |
642 | for (i = 0; i < len; ++i) |
764 | for (i = 0; i < len; ++i) |
643 | AvARRAY (av)[i] = decode_sv (dec); |
765 | AvARRAY (av)[i] = decode_sv (dec); |
644 | } |
766 | } |
… | |
… | |
656 | decode_he (dec_t *dec, HV *hv) |
778 | decode_he (dec_t *dec, HV *hv) |
657 | { |
779 | { |
658 | // for speed reasons, we specialcase single-string |
780 | // for speed reasons, we specialcase single-string |
659 | // byte or utf-8 strings as keys, but only when !stringref |
781 | // byte or utf-8 strings as keys, but only when !stringref |
660 | |
782 | |
661 | if (expect_true (!dec->stringref)) |
783 | if (ecb_expect_true (!dec->stringref)) |
662 | if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27) |
784 | if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8)) |
663 | { |
785 | { |
664 | I32 len = decode_uint (dec); |
786 | I32 len = decode_uint (dec); |
665 | char *key = (char *)dec->cur; |
787 | char *key = (char *)dec->cur; |
666 | |
788 | |
|
|
789 | WANT (len); |
667 | dec->cur += len; |
790 | dec->cur += len; |
668 | |
791 | |
669 | if (ecb_expect_false (dec->stringref)) |
|
|
670 | av_push (dec->stringref, newSVpvn (key, len)); |
|
|
671 | |
|
|
672 | hv_store (hv, key, len, decode_sv (dec), 0); |
792 | hv_store (hv, key, len, decode_sv (dec), 0); |
673 | |
793 | |
674 | return; |
794 | return; |
675 | } |
795 | } |
676 | else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27) |
796 | else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8)) |
677 | { |
797 | { |
678 | I32 len = decode_uint (dec); |
798 | I32 len = decode_uint (dec); |
679 | char *key = (char *)dec->cur; |
799 | char *key = (char *)dec->cur; |
680 | |
800 | |
|
|
801 | WANT (len); |
681 | dec->cur += len; |
802 | dec->cur += len; |
682 | |
803 | |
683 | if (ecb_expect_false (dec->stringref)) |
804 | if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8)) |
684 | av_push (dec->stringref, newSVpvn_utf8 (key, len, 1)); |
805 | if (!is_utf8_string (key, len)) |
|
|
806 | ERR ("corrupted CBOR data (invalid UTF-8 in map key)"); |
685 | |
807 | |
686 | hv_store (hv, key, -len, decode_sv (dec), 0); |
808 | hv_store (hv, key, -len, decode_sv (dec), 0); |
687 | |
809 | |
688 | return; |
810 | return; |
689 | } |
811 | } |
… | |
… | |
691 | SV *k = decode_sv (dec); |
813 | SV *k = decode_sv (dec); |
692 | SV *v = decode_sv (dec); |
814 | SV *v = decode_sv (dec); |
693 | |
815 | |
694 | hv_store_ent (hv, k, v, 0); |
816 | hv_store_ent (hv, k, v, 0); |
695 | SvREFCNT_dec (k); |
817 | SvREFCNT_dec (k); |
|
|
818 | |
|
|
819 | fail: |
|
|
820 | ; |
696 | } |
821 | } |
697 | |
822 | |
698 | static SV * |
823 | static SV * |
699 | decode_hv (dec_t *dec) |
824 | decode_hv (dec_t *dec) |
700 | { |
825 | { |
701 | HV *hv = newHV (); |
826 | HV *hv = newHV (); |
702 | |
827 | |
703 | DEC_INC_DEPTH; |
828 | DEC_INC_DEPTH; |
704 | |
829 | |
705 | if ((*dec->cur & 31) == 31) |
830 | if (*dec->cur == (MAJOR_MAP | MINOR_INDEF)) |
706 | { |
831 | { |
707 | ++dec->cur; |
832 | ++dec->cur; |
708 | |
833 | |
709 | for (;;) |
834 | for (;;) |
710 | { |
835 | { |
711 | WANT (1); |
836 | WANT (1); |
712 | |
837 | |
713 | if (*dec->cur == (0xe0 | 31)) |
838 | if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) |
714 | { |
839 | { |
715 | ++dec->cur; |
840 | ++dec->cur; |
716 | break; |
841 | break; |
717 | } |
842 | } |
718 | |
843 | |
… | |
… | |
739 | static SV * |
864 | static SV * |
740 | decode_str (dec_t *dec, int utf8) |
865 | decode_str (dec_t *dec, int utf8) |
741 | { |
866 | { |
742 | SV *sv = 0; |
867 | SV *sv = 0; |
743 | |
868 | |
744 | if ((*dec->cur & 31) == 31) |
869 | if ((*dec->cur & MINOR_MASK) == MINOR_INDEF) |
745 | { |
870 | { |
|
|
871 | // indefinite length strings |
746 | ++dec->cur; |
872 | ++dec->cur; |
747 | |
873 | |
|
|
874 | U8 major = *dec->cur & MAJOR_MISC; |
|
|
875 | |
748 | sv = newSVpvn ("", 0); |
876 | sv = newSVpvn ("", 0); |
749 | |
877 | |
750 | // not very fast, and certainly not robust against illegal input |
|
|
751 | for (;;) |
878 | for (;;) |
752 | { |
879 | { |
753 | WANT (1); |
880 | WANT (1); |
754 | |
881 | |
755 | if (*dec->cur == (0xe0 | 31)) |
882 | if ((*dec->cur - major) > LENGTH_EXT8) |
|
|
883 | if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) |
756 | { |
884 | { |
757 | ++dec->cur; |
885 | ++dec->cur; |
758 | break; |
886 | break; |
759 | } |
887 | } |
|
|
888 | else |
|
|
889 | ERR ("corrupted CBOR data (invalid chunks in indefinite length string)"); |
760 | |
890 | |
761 | sv_catsv (sv, decode_sv (dec)); |
891 | STRLEN len = decode_uint (dec); |
|
|
892 | |
|
|
893 | WANT (len); |
|
|
894 | sv_catpvn (sv, dec->cur, len); |
|
|
895 | dec->cur += len; |
762 | } |
896 | } |
763 | } |
897 | } |
764 | else |
898 | else |
765 | { |
899 | { |
766 | STRLEN len = decode_uint (dec); |
900 | STRLEN len = decode_uint (dec); |
767 | |
901 | |
768 | WANT (len); |
902 | WANT (len); |
769 | sv = newSVpvn (dec->cur, len); |
903 | sv = newSVpvn (dec->cur, len); |
770 | dec->cur += len; |
904 | dec->cur += len; |
|
|
905 | |
|
|
906 | if (ecb_expect_false (dec->stringref) |
|
|
907 | && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1)) |
|
|
908 | av_push (dec->stringref, SvREFCNT_inc_NN (sv)); |
771 | } |
909 | } |
772 | |
910 | |
773 | if (utf8) |
911 | if (utf8) |
|
|
912 | { |
|
|
913 | if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8)) |
|
|
914 | if (!is_utf8_string (SvPVX (sv), SvCUR (sv))) |
|
|
915 | ERR ("corrupted CBOR data (invalid UTF-8 in text string)"); |
|
|
916 | |
774 | SvUTF8_on (sv); |
917 | SvUTF8_on (sv); |
775 | |
918 | } |
776 | if (ecb_expect_false (dec->stringref) |
|
|
777 | && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1)) |
|
|
778 | av_push (dec->stringref, SvREFCNT_inc_NN (sv)); |
|
|
779 | |
919 | |
780 | return sv; |
920 | return sv; |
781 | |
921 | |
782 | fail: |
922 | fail: |
783 | SvREFCNT_dec (sv); |
923 | SvREFCNT_dec (sv); |
… | |
… | |
815 | } |
955 | } |
816 | break; |
956 | break; |
817 | |
957 | |
818 | case CBOR_TAG_STRINGREF: |
958 | case CBOR_TAG_STRINGREF: |
819 | { |
959 | { |
820 | if ((*dec->cur >> 5) != 0) |
960 | if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) |
821 | ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); |
961 | ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); |
822 | |
962 | |
823 | UV idx = decode_uint (dec); |
963 | UV idx = decode_uint (dec); |
824 | |
964 | |
825 | if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) |
965 | if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) |
… | |
… | |
832 | case CBOR_TAG_VALUE_SHAREABLE: |
972 | case CBOR_TAG_VALUE_SHAREABLE: |
833 | { |
973 | { |
834 | if (ecb_expect_false (!dec->shareable)) |
974 | if (ecb_expect_false (!dec->shareable)) |
835 | dec->shareable = (AV *)sv_2mortal ((SV *)newAV ()); |
975 | dec->shareable = (AV *)sv_2mortal ((SV *)newAV ()); |
836 | |
976 | |
|
|
977 | if (dec->cbor.flags & F_ALLOW_CYCLES) |
|
|
978 | { |
837 | sv = newSV (0); |
979 | sv = newSV (0); |
838 | av_push (dec->shareable, SvREFCNT_inc_NN (sv)); |
980 | av_push (dec->shareable, SvREFCNT_inc_NN (sv)); |
839 | |
981 | |
840 | SV *osv = decode_sv (dec); |
982 | SV *osv = decode_sv (dec); |
841 | sv_setsv (sv, osv); |
983 | sv_setsv (sv, osv); |
842 | SvREFCNT_dec_NN (osv); |
984 | SvREFCNT_dec_NN (osv); |
|
|
985 | } |
|
|
986 | else |
|
|
987 | { |
|
|
988 | av_push (dec->shareable, &PL_sv_undef); |
|
|
989 | int idx = AvFILLp (dec->shareable); |
|
|
990 | sv = decode_sv (dec); |
|
|
991 | av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv)); |
|
|
992 | } |
843 | } |
993 | } |
844 | break; |
994 | break; |
845 | |
995 | |
846 | case CBOR_TAG_VALUE_SHAREDREF: |
996 | case CBOR_TAG_VALUE_SHAREDREF: |
847 | { |
997 | { |
848 | if ((*dec->cur >> 5) != 0) |
998 | if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) |
849 | ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); |
999 | ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); |
850 | |
1000 | |
851 | UV idx = decode_uint (dec); |
1001 | UV idx = decode_uint (dec); |
852 | |
1002 | |
853 | if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) |
1003 | if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) |
854 | ERR ("corrupted CBOR data (sharedref index out of bounds)"); |
1004 | ERR ("corrupted CBOR data (sharedref index out of bounds)"); |
855 | |
1005 | |
856 | sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); |
1006 | sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); |
|
|
1007 | |
|
|
1008 | if (sv == &PL_sv_undef) |
|
|
1009 | ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled"); |
857 | } |
1010 | } |
858 | break; |
1011 | break; |
859 | |
1012 | |
860 | case CBOR_TAG_PERL_OBJECT: |
1013 | case CBOR_TAG_PERL_OBJECT: |
861 | { |
1014 | { |
… | |
… | |
876 | if (!method) |
1029 | if (!method) |
877 | ERR ("cannot decode perl-object (package does not have a THAW method)"); |
1030 | ERR ("cannot decode perl-object (package does not have a THAW method)"); |
878 | |
1031 | |
879 | dSP; |
1032 | dSP; |
880 | |
1033 | |
881 | ENTER; SAVETMPS; PUSHMARK (SP); |
1034 | ENTER; SAVETMPS; |
|
|
1035 | PUSHMARK (SP); |
882 | EXTEND (SP, len + 1); |
1036 | EXTEND (SP, len + 1); |
883 | // we re-bless the reference to get overload and other niceties right |
1037 | // we re-bless the reference to get overload and other niceties right |
884 | PUSHs (*av_fetch (av, 0, 1)); |
1038 | PUSHs (*av_fetch (av, 0, 1)); |
885 | PUSHs (sv_cbor); |
1039 | PUSHs (sv_cbor); |
886 | |
1040 | |
… | |
… | |
910 | |
1064 | |
911 | default: |
1065 | default: |
912 | { |
1066 | { |
913 | sv = decode_sv (dec); |
1067 | sv = decode_sv (dec); |
914 | |
1068 | |
|
|
1069 | dSP; |
|
|
1070 | ENTER; SAVETMPS; |
|
|
1071 | SAVESTACK_POS (); |
|
|
1072 | PUSHMARK (SP); |
|
|
1073 | EXTEND (SP, 2); |
|
|
1074 | PUSHs (newSVuv (tag)); |
|
|
1075 | PUSHs (sv); |
|
|
1076 | |
|
|
1077 | PUTBACK; |
|
|
1078 | int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL); |
|
|
1079 | SPAGAIN; |
|
|
1080 | |
|
|
1081 | if (SvTRUE (ERRSV)) |
|
|
1082 | { |
|
|
1083 | FREETMPS; LEAVE; |
|
|
1084 | ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); |
|
|
1085 | } |
|
|
1086 | |
|
|
1087 | if (count) |
|
|
1088 | { |
|
|
1089 | SvREFCNT_dec (sv); |
|
|
1090 | sv = SvREFCNT_inc (POPs); |
|
|
1091 | } |
|
|
1092 | else |
|
|
1093 | { |
915 | AV *av = newAV (); |
1094 | AV *av = newAV (); |
916 | av_push (av, newSVuv (tag)); |
1095 | av_push (av, newSVuv (tag)); |
917 | av_push (av, sv); |
1096 | av_push (av, sv); |
918 | |
1097 | |
919 | HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash |
1098 | HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash |
920 | ? cbor_tagged_stash |
1099 | ? cbor_tagged_stash |
921 | : gv_stashpv ("CBOR::XS::Tagged" , 1); |
1100 | : gv_stashpv ("CBOR::XS::Tagged" , 1); |
922 | |
|
|
923 | sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash); |
1101 | sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash); |
|
|
1102 | } |
|
|
1103 | |
|
|
1104 | PUTBACK; |
|
|
1105 | |
|
|
1106 | FREETMPS; LEAVE; |
924 | } |
1107 | } |
925 | break; |
1108 | break; |
926 | } |
1109 | } |
927 | |
1110 | |
928 | return sv; |
1111 | return sv; |
… | |
… | |
935 | static SV * |
1118 | static SV * |
936 | decode_sv (dec_t *dec) |
1119 | decode_sv (dec_t *dec) |
937 | { |
1120 | { |
938 | WANT (1); |
1121 | WANT (1); |
939 | |
1122 | |
940 | switch (*dec->cur >> 5) |
1123 | switch (*dec->cur >> MAJOR_SHIFT) |
941 | { |
1124 | { |
942 | case 0: // unsigned int |
1125 | case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec)); |
943 | return newSVuv (decode_uint (dec)); |
1126 | case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec)); |
944 | case 1: // negative int |
1127 | case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0); |
945 | return newSViv (-1 - (IV)decode_uint (dec)); |
1128 | case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1); |
946 | case 2: // octet string |
1129 | case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec); |
947 | return decode_str (dec, 0); |
1130 | case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec); |
948 | case 3: // utf-8 string |
1131 | case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec); |
949 | return decode_str (dec, 1); |
1132 | |
950 | case 4: // array |
1133 | case MAJOR_MISC >> MAJOR_SHIFT: |
951 | return decode_av (dec); |
|
|
952 | case 5: // map |
|
|
953 | return decode_hv (dec); |
|
|
954 | case 6: // tag |
|
|
955 | return decode_tagged (dec); |
|
|
956 | case 7: // misc |
|
|
957 | switch (*dec->cur++ & 31) |
1134 | switch (*dec->cur++ & MINOR_MASK) |
958 | { |
1135 | { |
959 | case 20: |
1136 | case SIMPLE_FALSE: |
960 | #if CBOR_SLOW |
1137 | #if CBOR_SLOW |
961 | types_false = get_bool ("Types::Serialiser::false"); |
1138 | types_false = get_bool ("Types::Serialiser::false"); |
962 | #endif |
1139 | #endif |
963 | return newSVsv (types_false); |
1140 | return newSVsv (types_false); |
964 | case 21: |
1141 | case SIMPLE_TRUE: |
965 | #if CBOR_SLOW |
1142 | #if CBOR_SLOW |
966 | types_true = get_bool ("Types::Serialiser::true"); |
1143 | types_true = get_bool ("Types::Serialiser::true"); |
967 | #endif |
1144 | #endif |
968 | return newSVsv (types_true); |
1145 | return newSVsv (types_true); |
969 | case 22: |
1146 | case SIMPLE_NULL: |
970 | return newSVsv (&PL_sv_undef); |
1147 | return newSVsv (&PL_sv_undef); |
971 | case 23: |
1148 | case SIMPLE_UNDEF: |
972 | #if CBOR_SLOW |
1149 | #if CBOR_SLOW |
973 | types_error = get_bool ("Types::Serialiser::error"); |
1150 | types_error = get_bool ("Types::Serialiser::error"); |
974 | #endif |
1151 | #endif |
975 | return newSVsv (types_error); |
1152 | return newSVsv (types_error); |
976 | |
1153 | |
977 | case 25: |
1154 | case MISC_FLOAT16: |
978 | { |
1155 | { |
979 | WANT (2); |
1156 | WANT (2); |
980 | |
1157 | |
981 | uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; |
1158 | uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; |
982 | dec->cur += 2; |
1159 | dec->cur += 2; |
983 | |
1160 | |
984 | return newSVnv (ecb_binary16_to_float (fp)); |
1161 | return newSVnv (ecb_binary16_to_float (fp)); |
985 | } |
1162 | } |
986 | |
1163 | |
987 | case 26: |
1164 | case MISC_FLOAT32: |
988 | { |
1165 | { |
989 | uint32_t fp; |
1166 | uint32_t fp; |
990 | WANT (4); |
1167 | WANT (4); |
991 | memcpy (&fp, dec->cur, 4); |
1168 | memcpy (&fp, dec->cur, 4); |
992 | dec->cur += 4; |
1169 | dec->cur += 4; |
… | |
… | |
995 | fp = ecb_bswap32 (fp); |
1172 | fp = ecb_bswap32 (fp); |
996 | |
1173 | |
997 | return newSVnv (ecb_binary32_to_float (fp)); |
1174 | return newSVnv (ecb_binary32_to_float (fp)); |
998 | } |
1175 | } |
999 | |
1176 | |
1000 | case 27: |
1177 | case MISC_FLOAT64: |
1001 | { |
1178 | { |
1002 | uint64_t fp; |
1179 | uint64_t fp; |
1003 | WANT (8); |
1180 | WANT (8); |
1004 | memcpy (&fp, dec->cur, 8); |
1181 | memcpy (&fp, dec->cur, 8); |
1005 | dec->cur += 8; |
1182 | dec->cur += 8; |
… | |
… | |
1008 | fp = ecb_bswap64 (fp); |
1185 | fp = ecb_bswap64 (fp); |
1009 | |
1186 | |
1010 | return newSVnv (ecb_binary64_to_double (fp)); |
1187 | return newSVnv (ecb_binary64_to_double (fp)); |
1011 | } |
1188 | } |
1012 | |
1189 | |
1013 | // 0..19 unassigned |
1190 | // 0..19 unassigned simple |
1014 | // 24 reserved + unassigned (reserved values are not encodable) |
1191 | // 24 reserved + unassigned simple (reserved values are not encodable) |
|
|
1192 | // 28-30 unassigned misc |
|
|
1193 | // 31 break code |
1015 | default: |
1194 | default: |
1016 | ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); |
1195 | ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)"); |
1017 | } |
1196 | } |
1018 | |
1197 | |
1019 | break; |
1198 | break; |
1020 | } |
1199 | } |
1021 | |
1200 | |
… | |
… | |
1024 | } |
1203 | } |
1025 | |
1204 | |
1026 | static SV * |
1205 | static SV * |
1027 | decode_cbor (SV *string, CBOR *cbor, char **offset_return) |
1206 | decode_cbor (SV *string, CBOR *cbor, char **offset_return) |
1028 | { |
1207 | { |
1029 | dec_t dec = { }; |
1208 | dec_t dec = { 0 }; |
1030 | SV *sv; |
1209 | SV *sv; |
1031 | STRLEN len; |
1210 | STRLEN len; |
1032 | char *data = SvPVbyte (string, len); |
1211 | char *data = SvPVbyte (string, len); |
1033 | |
1212 | |
1034 | if (len > cbor->max_size && cbor->max_size) |
1213 | if (len > cbor->max_size && cbor->max_size) |
… | |
… | |
1048 | if (dec.cur != dec.end && !dec.err) |
1227 | if (dec.cur != dec.end && !dec.err) |
1049 | dec.err = "garbage after CBOR object"; |
1228 | dec.err = "garbage after CBOR object"; |
1050 | |
1229 | |
1051 | if (dec.err) |
1230 | if (dec.err) |
1052 | { |
1231 | { |
|
|
1232 | if (dec.shareable) |
|
|
1233 | { |
|
|
1234 | // need to break cyclic links, which whould all be in shareable |
|
|
1235 | int i; |
|
|
1236 | SV **svp; |
|
|
1237 | |
|
|
1238 | for (i = av_len (dec.shareable) + 1; i--; ) |
|
|
1239 | if ((svp = av_fetch (dec.shareable, i, 0))) |
|
|
1240 | sv_setsv (*svp, &PL_sv_undef); |
|
|
1241 | } |
|
|
1242 | |
1053 | SvREFCNT_dec (sv); |
1243 | SvREFCNT_dec (sv); |
1054 | croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); |
1244 | croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); |
1055 | } |
1245 | } |
1056 | |
1246 | |
1057 | sv = sv_2mortal (sv); |
1247 | sv = sv_2mortal (sv); |
1058 | |
1248 | |
1059 | return sv; |
1249 | return sv; |
1060 | } |
1250 | } |
1061 | |
1251 | |
|
|
1252 | ///////////////////////////////////////////////////////////////////////////// |
|
|
1253 | // incremental parser |
|
|
1254 | |
|
|
1255 | #define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0) |
|
|
1256 | |
|
|
1257 | // returns 0 for notyet, 1 for success or error |
|
|
1258 | static int |
|
|
1259 | incr_parse (CBOR *self, SV *cborstr) |
|
|
1260 | { |
|
|
1261 | STRLEN cur; |
|
|
1262 | SvPV (cborstr, cur); |
|
|
1263 | |
|
|
1264 | while (ecb_expect_true (self->incr_need <= cur)) |
|
|
1265 | { |
|
|
1266 | // table of integer count bytes |
|
|
1267 | static I8 incr_len[MINOR_MASK + 1] = { |
|
|
1268 | 0, 0, 0, 0, 0, 0, 0, 0, |
|
|
1269 | 0, 0, 0, 0, 0, 0, 0, 0, |
|
|
1270 | 0, 0, 0, 0, 0, 0, 0, 0, |
|
|
1271 | 1, 2, 4, 8,-1,-1,-1,-2 |
|
|
1272 | }; |
|
|
1273 | |
|
|
1274 | const U8 *p = SvPVX (cborstr) + self->incr_pos; |
|
|
1275 | U8 m = *p & MINOR_MASK; |
|
|
1276 | IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]); |
|
|
1277 | I8 ilen = incr_len[m]; |
|
|
1278 | |
|
|
1279 | self->incr_need = self->incr_pos + 1; |
|
|
1280 | |
|
|
1281 | if (ecb_expect_false (ilen < 0)) |
|
|
1282 | { |
|
|
1283 | if (m != MINOR_INDEF) |
|
|
1284 | return 1; // error |
|
|
1285 | |
|
|
1286 | if (*p == (MAJOR_MISC | MINOR_INDEF)) |
|
|
1287 | { |
|
|
1288 | if (count >= 0) |
|
|
1289 | return 1; // error |
|
|
1290 | |
|
|
1291 | count = 1; |
|
|
1292 | } |
|
|
1293 | else |
|
|
1294 | { |
|
|
1295 | av_push (self->incr_count, newSViv (-1)); //TODO: nest |
|
|
1296 | count = -1; |
|
|
1297 | } |
|
|
1298 | } |
|
|
1299 | else |
|
|
1300 | { |
|
|
1301 | self->incr_need += ilen; |
|
|
1302 | if (ecb_expect_false (self->incr_need > cur)) |
|
|
1303 | return 0; |
|
|
1304 | |
|
|
1305 | int major = *p >> MAJOR_SHIFT; |
|
|
1306 | |
|
|
1307 | switch (major) |
|
|
1308 | { |
|
|
1309 | case MAJOR_TAG >> MAJOR_SHIFT: |
|
|
1310 | ++count; // tags merely prefix another value |
|
|
1311 | break; |
|
|
1312 | |
|
|
1313 | case MAJOR_BYTES >> MAJOR_SHIFT: |
|
|
1314 | case MAJOR_TEXT >> MAJOR_SHIFT: |
|
|
1315 | case MAJOR_ARRAY >> MAJOR_SHIFT: |
|
|
1316 | case MAJOR_MAP >> MAJOR_SHIFT: |
|
|
1317 | { |
|
|
1318 | UV len; |
|
|
1319 | |
|
|
1320 | if (ecb_expect_false (ilen)) |
|
|
1321 | { |
|
|
1322 | len = 0; |
|
|
1323 | |
|
|
1324 | do { |
|
|
1325 | len = (len << 8) | *++p; |
|
|
1326 | } while (--ilen); |
|
|
1327 | } |
|
|
1328 | else |
|
|
1329 | len = m; |
|
|
1330 | |
|
|
1331 | switch (major) |
|
|
1332 | { |
|
|
1333 | case MAJOR_BYTES >> MAJOR_SHIFT: |
|
|
1334 | case MAJOR_TEXT >> MAJOR_SHIFT: |
|
|
1335 | self->incr_need += len; |
|
|
1336 | if (ecb_expect_false (self->incr_need > cur)) |
|
|
1337 | return 0; |
|
|
1338 | |
|
|
1339 | break; |
|
|
1340 | |
|
|
1341 | case MAJOR_MAP >> MAJOR_SHIFT: |
|
|
1342 | len <<= 1; |
|
|
1343 | case MAJOR_ARRAY >> MAJOR_SHIFT: |
|
|
1344 | if (len) |
|
|
1345 | { |
|
|
1346 | av_push (self->incr_count, newSViv (len + 1)); //TODO: nest |
|
|
1347 | count = len + 1; |
|
|
1348 | } |
|
|
1349 | break; |
|
|
1350 | } |
|
|
1351 | } |
|
|
1352 | } |
|
|
1353 | } |
|
|
1354 | |
|
|
1355 | self->incr_pos = self->incr_need; |
|
|
1356 | |
|
|
1357 | if (count > 0) |
|
|
1358 | { |
|
|
1359 | while (!--count) |
|
|
1360 | { |
|
|
1361 | if (!AvFILLp (self->incr_count)) |
|
|
1362 | return 1; // done |
|
|
1363 | |
|
|
1364 | SvREFCNT_dec_NN (av_pop (self->incr_count)); |
|
|
1365 | count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]); |
|
|
1366 | } |
|
|
1367 | |
|
|
1368 | SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count; |
|
|
1369 | } |
|
|
1370 | } |
|
|
1371 | |
|
|
1372 | return 0; |
|
|
1373 | } |
|
|
1374 | |
|
|
1375 | |
1062 | ///////////////////////////////////////////////////////////////////////////// |
1376 | ///////////////////////////////////////////////////////////////////////////// |
1063 | // XS interface functions |
1377 | // XS interface functions |
1064 | |
1378 | |
1065 | MODULE = CBOR::XS PACKAGE = CBOR::XS |
1379 | MODULE = CBOR::XS PACKAGE = CBOR::XS |
1066 | |
1380 | |
… | |
… | |
1073 | types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1); |
1387 | types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1); |
1074 | |
1388 | |
1075 | types_true = get_bool ("Types::Serialiser::true" ); |
1389 | types_true = get_bool ("Types::Serialiser::true" ); |
1076 | types_false = get_bool ("Types::Serialiser::false"); |
1390 | types_false = get_bool ("Types::Serialiser::false"); |
1077 | types_error = get_bool ("Types::Serialiser::error"); |
1391 | types_error = get_bool ("Types::Serialiser::error"); |
|
|
1392 | |
|
|
1393 | default_filter = newSVpv ("CBOR::XS::default_filter", 0); |
1078 | |
1394 | |
1079 | sv_cbor = newSVpv ("CBOR", 0); |
1395 | sv_cbor = newSVpv ("CBOR", 0); |
1080 | SvREADONLY_on (sv_cbor); |
1396 | SvREADONLY_on (sv_cbor); |
1081 | } |
1397 | } |
1082 | |
1398 | |
… | |
… | |
1104 | void shrink (CBOR *self, int enable = 1) |
1420 | void shrink (CBOR *self, int enable = 1) |
1105 | ALIAS: |
1421 | ALIAS: |
1106 | shrink = F_SHRINK |
1422 | shrink = F_SHRINK |
1107 | allow_unknown = F_ALLOW_UNKNOWN |
1423 | allow_unknown = F_ALLOW_UNKNOWN |
1108 | allow_sharing = F_ALLOW_SHARING |
1424 | allow_sharing = F_ALLOW_SHARING |
|
|
1425 | allow_cycles = F_ALLOW_CYCLES |
1109 | dedup_strings = F_DEDUP_STRINGS |
1426 | pack_strings = F_PACK_STRINGS |
|
|
1427 | utf8_strings = F_UTF8_STRINGS |
|
|
1428 | validate_utf8 = F_VALIDATE_UTF8 |
1110 | PPCODE: |
1429 | PPCODE: |
1111 | { |
1430 | { |
1112 | if (enable) |
1431 | if (enable) |
1113 | self->flags |= ix; |
1432 | self->flags |= ix; |
1114 | else |
1433 | else |
… | |
… | |
1120 | void get_shrink (CBOR *self) |
1439 | void get_shrink (CBOR *self) |
1121 | ALIAS: |
1440 | ALIAS: |
1122 | get_shrink = F_SHRINK |
1441 | get_shrink = F_SHRINK |
1123 | get_allow_unknown = F_ALLOW_UNKNOWN |
1442 | get_allow_unknown = F_ALLOW_UNKNOWN |
1124 | get_allow_sharing = F_ALLOW_SHARING |
1443 | get_allow_sharing = F_ALLOW_SHARING |
|
|
1444 | get_allow_cycles = F_ALLOW_CYCLES |
1125 | get_dedup_strings = F_DEDUP_STRINGS |
1445 | get_pack_strings = F_PACK_STRINGS |
|
|
1446 | get_validate_utf8 = F_VALIDATE_UTF8 |
1126 | PPCODE: |
1447 | PPCODE: |
1127 | XPUSHs (boolSV (self->flags & ix)); |
1448 | XPUSHs (boolSV (self->flags & ix)); |
1128 | |
1449 | |
1129 | void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) |
1450 | void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) |
1130 | PPCODE: |
1451 | PPCODE: |
… | |
… | |
1143 | XPUSHs (ST (0)); |
1464 | XPUSHs (ST (0)); |
1144 | |
1465 | |
1145 | int get_max_size (CBOR *self) |
1466 | int get_max_size (CBOR *self) |
1146 | CODE: |
1467 | CODE: |
1147 | RETVAL = self->max_size; |
1468 | RETVAL = self->max_size; |
|
|
1469 | OUTPUT: |
|
|
1470 | RETVAL |
|
|
1471 | |
|
|
1472 | void filter (CBOR *self, SV *filter = 0) |
|
|
1473 | PPCODE: |
|
|
1474 | SvREFCNT_dec (self->filter); |
|
|
1475 | self->filter = filter ? newSVsv (filter) : filter; |
|
|
1476 | XPUSHs (ST (0)); |
|
|
1477 | |
|
|
1478 | SV *get_filter (CBOR *self) |
|
|
1479 | CODE: |
|
|
1480 | RETVAL = self->filter ? self->filter : NEWSV (0, 0); |
1148 | OUTPUT: |
1481 | OUTPUT: |
1149 | RETVAL |
1482 | RETVAL |
1150 | |
1483 | |
1151 | void encode (CBOR *self, SV *scalar) |
1484 | void encode (CBOR *self, SV *scalar) |
1152 | PPCODE: |
1485 | PPCODE: |
… | |
… | |
1167 | EXTEND (SP, 2); |
1500 | EXTEND (SP, 2); |
1168 | PUSHs (sv); |
1501 | PUSHs (sv); |
1169 | PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); |
1502 | PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); |
1170 | } |
1503 | } |
1171 | |
1504 | |
|
|
1505 | void incr_parse (CBOR *self, SV *cborstr) |
|
|
1506 | ALIAS: |
|
|
1507 | incr_parse_multiple = 1 |
|
|
1508 | PPCODE: |
|
|
1509 | { |
|
|
1510 | if (SvUTF8 (cborstr)) |
|
|
1511 | sv_utf8_downgrade (cborstr, 0); |
|
|
1512 | |
|
|
1513 | if (!self->incr_count) |
|
|
1514 | { |
|
|
1515 | self->incr_count = newAV (); |
|
|
1516 | self->incr_pos = 0; |
|
|
1517 | self->incr_need = 1; |
|
|
1518 | |
|
|
1519 | av_push (self->incr_count, newSViv (1)); |
|
|
1520 | } |
|
|
1521 | |
|
|
1522 | do |
|
|
1523 | { |
|
|
1524 | if (!incr_parse (self, cborstr)) |
|
|
1525 | { |
|
|
1526 | if (self->incr_need > self->max_size && self->max_size) |
|
|
1527 | croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu", |
|
|
1528 | (unsigned long)self->incr_need, (unsigned long)self->max_size); |
|
|
1529 | |
|
|
1530 | break; |
|
|
1531 | } |
|
|
1532 | |
|
|
1533 | SV *sv; |
|
|
1534 | char *offset; |
|
|
1535 | |
|
|
1536 | PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN; |
|
|
1537 | XPUSHs (sv); |
|
|
1538 | |
|
|
1539 | sv_chop (cborstr, offset); |
|
|
1540 | |
|
|
1541 | av_clear (self->incr_count); |
|
|
1542 | av_push (self->incr_count, newSViv (1)); |
|
|
1543 | |
|
|
1544 | self->incr_pos = 0; |
|
|
1545 | self->incr_need = self->incr_pos + 1; |
|
|
1546 | } |
|
|
1547 | while (ix); |
|
|
1548 | } |
|
|
1549 | |
|
|
1550 | void incr_reset (CBOR *self) |
|
|
1551 | CODE: |
|
|
1552 | { |
|
|
1553 | SvREFCNT_dec (self->incr_count); |
|
|
1554 | self->incr_count = 0; |
|
|
1555 | } |
|
|
1556 | |
|
|
1557 | void DESTROY (CBOR *self) |
|
|
1558 | PPCODE: |
|
|
1559 | cbor_free (self); |
|
|
1560 | |
1172 | PROTOTYPES: ENABLE |
1561 | PROTOTYPES: ENABLE |
1173 | |
1562 | |
1174 | void encode_cbor (SV *scalar) |
1563 | void encode_cbor (SV *scalar) |
|
|
1564 | ALIAS: |
|
|
1565 | encode_cbor = 0 |
|
|
1566 | encode_cbor_sharing = F_ALLOW_SHARING |
1175 | PPCODE: |
1567 | PPCODE: |
1176 | { |
1568 | { |
1177 | CBOR cbor; |
1569 | CBOR cbor; |
1178 | cbor_init (&cbor); |
1570 | cbor_init (&cbor); |
|
|
1571 | cbor.flags |= ix; |
1179 | PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; |
1572 | PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; |
1180 | XPUSHs (scalar); |
1573 | XPUSHs (scalar); |
1181 | } |
1574 | } |
1182 | |
1575 | |
1183 | void decode_cbor (SV *cborstr) |
1576 | void decode_cbor (SV *cborstr) |