… | |
… | |
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)) |
… | |
… | |
23 | #endif |
24 | #endif |
24 | #ifndef SvREFCNT_dec_NN |
25 | #ifndef SvREFCNT_dec_NN |
25 | # define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv) |
26 | # define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv) |
26 | #endif |
27 | #endif |
27 | |
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 | }; |
|
|
65 | |
28 | // known tags |
66 | // known tags |
29 | enum cbor_tag |
67 | enum cbor_tag |
30 | { |
68 | { |
31 | // extensions |
69 | // extensions |
32 | CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref |
70 | CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref |
33 | CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object |
71 | CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object |
34 | CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object |
72 | CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object |
35 | CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing |
73 | CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing |
36 | CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing |
74 | CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing |
37 | CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref |
75 | CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref |
38 | CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection |
76 | CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection |
39 | |
77 | |
40 | // rfc7049 |
78 | // rfc7049 |
41 | CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 |
79 | CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 |
42 | CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any |
80 | CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any |
43 | CBOR_TAG_POS_BIGNUM = 2, // byte string |
81 | CBOR_TAG_POS_BIGNUM = 2, // byte string |
44 | CBOR_TAG_NEG_BIGNUM = 3, // byte string |
82 | CBOR_TAG_NEG_BIGNUM = 3, // byte string |
45 | CBOR_TAG_DECIMAL = 4, // decimal fraction, array |
83 | CBOR_TAG_DECIMAL = 4, // decimal fraction, array |
46 | CBOR_TAG_BIGFLOAT = 5, // array |
84 | CBOR_TAG_BIGFLOAT = 5, // array |
47 | |
85 | |
48 | CBOR_TAG_CONV_B64U = 21, // base64url, any |
86 | CBOR_TAG_CONV_B64U = 21, // base64url, any |
49 | CBOR_TAG_CONV_B64 = 22, // base64, any |
87 | CBOR_TAG_CONV_B64 = 22, // base64, any |
50 | CBOR_TAG_CONV_HEX = 23, // base16, any |
88 | CBOR_TAG_CONV_HEX = 23, // base16, any |
51 | CBOR_TAG_CBOR = 24, // embedded cbor, byte string |
89 | CBOR_TAG_CBOR = 24, // embedded cbor, byte string |
52 | |
90 | |
53 | CBOR_TAG_URI = 32, // URI rfc3986, utf-8 |
91 | CBOR_TAG_URI = 32, // URI rfc3986, utf-8 |
54 | CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8 |
92 | CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8 |
55 | CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8 |
93 | CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8 |
56 | CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8 |
94 | CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8 |
57 | CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 |
95 | CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 |
58 | |
96 | |
59 | CBOR_TAG_MAGIC = 55799 // self-describe cbor |
97 | CBOR_TAG_MAGIC = 55799, // self-describe cbor |
60 | }; |
98 | }; |
61 | |
99 | |
62 | #define F_SHRINK 0x00000001UL |
100 | #define F_SHRINK 0x00000001UL |
63 | #define F_ALLOW_UNKNOWN 0x00000002UL |
101 | #define F_ALLOW_UNKNOWN 0x00000002UL |
64 | #define F_ALLOW_SHARING 0x00000004UL //TODO |
102 | #define F_ALLOW_SHARING 0x00000004UL |
65 | #define F_ALLOW_STRINGREF 0x00000008UL //TODO |
103 | #define F_ALLOW_CYCLES 0x00000008UL |
|
|
104 | #define F_PACK_STRINGS 0x00000010UL |
|
|
105 | #define F_VALIDATE_UTF8 0x00000020UL |
66 | |
106 | |
67 | #define INIT_SIZE 32 // initial scalar size to be allocated |
107 | #define INIT_SIZE 32 // initial scalar size to be allocated |
68 | |
108 | |
69 | #define SB do { |
109 | #define SB do { |
70 | #define SE } while (0) |
110 | #define SE } while (0) |
… | |
… | |
89 | typedef struct { |
129 | typedef struct { |
90 | U32 flags; |
130 | U32 flags; |
91 | U32 max_depth; |
131 | U32 max_depth; |
92 | STRLEN max_size; |
132 | STRLEN max_size; |
93 | SV *filter; |
133 | SV *filter; |
|
|
134 | |
|
|
135 | // for the incremental parser |
|
|
136 | STRLEN incr_pos; // the current offset into the text |
|
|
137 | STRLEN incr_need; // minimum bytes needed to decode |
|
|
138 | AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef. |
94 | } CBOR; |
139 | } CBOR; |
95 | |
140 | |
96 | ecb_inline void |
141 | ecb_inline void |
97 | cbor_init (CBOR *cbor) |
142 | cbor_init (CBOR *cbor) |
98 | { |
143 | { |
… | |
… | |
102 | |
147 | |
103 | ecb_inline void |
148 | ecb_inline void |
104 | cbor_free (CBOR *cbor) |
149 | cbor_free (CBOR *cbor) |
105 | { |
150 | { |
106 | SvREFCNT_dec (cbor->filter); |
151 | SvREFCNT_dec (cbor->filter); |
|
|
152 | SvREFCNT_dec (cbor->incr_count); |
107 | } |
153 | } |
108 | |
154 | |
109 | ///////////////////////////////////////////////////////////////////////////// |
155 | ///////////////////////////////////////////////////////////////////////////// |
110 | // utility functions |
156 | // utility functions |
111 | |
157 | |
… | |
… | |
189 | static void |
235 | static void |
190 | encode_uint (enc_t *enc, int major, UV len) |
236 | encode_uint (enc_t *enc, int major, UV len) |
191 | { |
237 | { |
192 | need (enc, 9); |
238 | need (enc, 9); |
193 | |
239 | |
194 | if (len < 24) |
240 | if (ecb_expect_true (len < LENGTH_EXT1)) |
195 | *enc->cur++ = major | len; |
241 | *enc->cur++ = major | len; |
196 | else if (len <= 0xff) |
242 | else if (ecb_expect_true (len <= 0xffU)) |
197 | { |
243 | { |
198 | *enc->cur++ = major | 24; |
244 | *enc->cur++ = major | LENGTH_EXT1; |
199 | *enc->cur++ = len; |
245 | *enc->cur++ = len; |
200 | } |
246 | } |
201 | else if (len <= 0xffff) |
247 | else if (len <= 0xffffU) |
202 | { |
248 | { |
203 | *enc->cur++ = major | 25; |
249 | *enc->cur++ = major | LENGTH_EXT2; |
204 | *enc->cur++ = len >> 8; |
250 | *enc->cur++ = len >> 8; |
205 | *enc->cur++ = len; |
251 | *enc->cur++ = len; |
206 | } |
252 | } |
207 | else if (len <= 0xffffffff) |
253 | else if (len <= 0xffffffffU) |
208 | { |
254 | { |
209 | *enc->cur++ = major | 26; |
255 | *enc->cur++ = major | LENGTH_EXT4; |
210 | *enc->cur++ = len >> 24; |
256 | *enc->cur++ = len >> 24; |
211 | *enc->cur++ = len >> 16; |
257 | *enc->cur++ = len >> 16; |
212 | *enc->cur++ = len >> 8; |
258 | *enc->cur++ = len >> 8; |
213 | *enc->cur++ = len; |
259 | *enc->cur++ = len; |
214 | } |
260 | } |
215 | else |
261 | else |
216 | { |
262 | { |
217 | *enc->cur++ = major | 27; |
263 | *enc->cur++ = major | LENGTH_EXT8; |
218 | *enc->cur++ = len >> 56; |
264 | *enc->cur++ = len >> 56; |
219 | *enc->cur++ = len >> 48; |
265 | *enc->cur++ = len >> 48; |
220 | *enc->cur++ = len >> 40; |
266 | *enc->cur++ = len >> 40; |
221 | *enc->cur++ = len >> 32; |
267 | *enc->cur++ = len >> 32; |
222 | *enc->cur++ = len >> 24; |
268 | *enc->cur++ = len >> 24; |
… | |
… | |
227 | } |
273 | } |
228 | |
274 | |
229 | ecb_inline void |
275 | ecb_inline void |
230 | encode_tag (enc_t *enc, UV tag) |
276 | encode_tag (enc_t *enc, UV tag) |
231 | { |
277 | { |
232 | encode_uint (enc, 0xc0, tag); |
278 | encode_uint (enc, MAJOR_TAG, tag); |
233 | } |
279 | } |
234 | |
280 | |
235 | ecb_inline void |
281 | ecb_inline void |
236 | encode_str (enc_t *enc, int utf8, char *str, STRLEN len) |
282 | encode_str (enc_t *enc, int utf8, char *str, STRLEN len) |
237 | { |
283 | { |
238 | encode_uint (enc, utf8 ? 0x60 : 0x40, len); |
284 | encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len); |
239 | need (enc, len); |
285 | need (enc, len); |
240 | memcpy (enc->cur, str, len); |
286 | memcpy (enc->cur, str, len); |
241 | enc->cur += len; |
287 | enc->cur += len; |
242 | } |
288 | } |
243 | |
289 | |
244 | static void |
290 | static void |
245 | encode_strref (enc_t *enc, int utf8, char *str, STRLEN len) |
291 | encode_strref (enc_t *enc, int utf8, char *str, STRLEN len) |
246 | { |
292 | { |
247 | if (ecb_expect_false (enc->cbor.flags & F_ALLOW_STRINGREF)) |
293 | if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS)) |
248 | { |
294 | { |
249 | SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1); |
295 | SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1); |
250 | |
296 | |
251 | if (SvOK (*svp)) |
297 | if (SvOK (*svp)) |
252 | { |
298 | { |
253 | // already registered, use stringref |
299 | // already registered, use stringref |
254 | encode_tag (enc, CBOR_TAG_STRINGREF); |
300 | encode_tag (enc, CBOR_TAG_STRINGREF); |
255 | encode_uint (enc, 0x00, SvUV (*svp)); |
301 | encode_uint (enc, MAJOR_POS_INT, SvUV (*svp)); |
256 | return; |
302 | return; |
257 | } |
303 | } |
258 | else if (len >= minimum_string_length (enc->stringref_idx)) |
304 | else if (len >= minimum_string_length (enc->stringref_idx)) |
259 | { |
305 | { |
260 | // register only |
306 | // register only |
… | |
… | |
276 | if (enc->depth >= enc->cbor.max_depth) |
322 | if (enc->depth >= enc->cbor.max_depth) |
277 | croak (ERR_NESTING_EXCEEDED); |
323 | croak (ERR_NESTING_EXCEEDED); |
278 | |
324 | |
279 | ++enc->depth; |
325 | ++enc->depth; |
280 | |
326 | |
281 | encode_uint (enc, 0x80, len + 1); |
327 | encode_uint (enc, MAJOR_ARRAY, len + 1); |
282 | |
328 | |
|
|
329 | if (SvMAGICAL (av)) |
283 | for (i = 0; i <= len; ++i) |
330 | for (i = 0; i <= len; ++i) |
284 | { |
331 | { |
285 | SV **svp = av_fetch (av, i, 0); |
332 | SV **svp = av_fetch (av, i, 0); |
286 | encode_sv (enc, svp ? *svp : &PL_sv_undef); |
333 | encode_sv (enc, svp ? *svp : &PL_sv_undef); |
287 | } |
334 | } |
|
|
335 | else |
|
|
336 | for (i = 0; i <= len; ++i) |
|
|
337 | { |
|
|
338 | SV *sv = AvARRAY (av)[i]; |
|
|
339 | encode_sv (enc, sv ? sv : &PL_sv_undef); |
|
|
340 | } |
288 | |
341 | |
289 | --enc->depth; |
342 | --enc->depth; |
290 | } |
343 | } |
291 | |
344 | |
292 | static void |
345 | static void |
… | |
… | |
301 | |
354 | |
302 | int pairs = hv_iterinit (hv); |
355 | int pairs = hv_iterinit (hv); |
303 | int mg = SvMAGICAL (hv); |
356 | int mg = SvMAGICAL (hv); |
304 | |
357 | |
305 | if (mg) |
358 | if (mg) |
306 | encode_ch (enc, 0xa0 | 31); |
359 | encode_ch (enc, MAJOR_MAP | MINOR_INDEF); |
307 | else |
360 | else |
308 | encode_uint (enc, 0xa0, pairs); |
361 | encode_uint (enc, MAJOR_MAP, pairs); |
309 | |
362 | |
310 | while ((he = hv_iternext (hv))) |
363 | while ((he = hv_iternext (hv))) |
311 | { |
364 | { |
312 | if (HeKLEN (he) == HEf_SVKEY) |
365 | if (HeKLEN (he) == HEf_SVKEY) |
313 | encode_sv (enc, HeSVKEY (he)); |
366 | encode_sv (enc, HeSVKEY (he)); |
… | |
… | |
316 | |
369 | |
317 | encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); |
370 | encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); |
318 | } |
371 | } |
319 | |
372 | |
320 | if (mg) |
373 | if (mg) |
321 | encode_ch (enc, 0xe0 | 31); |
374 | encode_ch (enc, MAJOR_MISC | MINOR_INDEF); |
322 | |
375 | |
323 | --enc->depth; |
376 | --enc->depth; |
324 | } |
377 | } |
325 | |
378 | |
326 | // encode objects, arrays and special \0=false and \1=true values. |
379 | // encode objects, arrays and special \0=false and \1=true values. |
327 | static void |
380 | static void |
328 | encode_rv (enc_t *enc, SV *sv) |
381 | encode_rv (enc_t *enc, SV *sv) |
329 | { |
382 | { |
330 | SvGETMAGIC (sv); |
383 | SvGETMAGIC (sv); |
331 | |
|
|
332 | if (ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING) |
|
|
333 | && ecb_expect_false (SvREFCNT (sv) > 1)) |
|
|
334 | { |
|
|
335 | if (!enc->shareable) |
|
|
336 | enc->shareable = (HV *)sv_2mortal ((SV *)newHV ()); |
|
|
337 | |
|
|
338 | SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1); |
|
|
339 | |
|
|
340 | if (SvOK (*svp)) |
|
|
341 | { |
|
|
342 | encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF); |
|
|
343 | encode_uint (enc, 0x00, SvUV (*svp)); |
|
|
344 | return; |
|
|
345 | } |
|
|
346 | else |
|
|
347 | { |
|
|
348 | sv_setuv (*svp, enc->shareable_idx); |
|
|
349 | ++enc->shareable_idx; |
|
|
350 | encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE); |
|
|
351 | } |
|
|
352 | } |
|
|
353 | |
384 | |
354 | svtype svt = SvTYPE (sv); |
385 | svtype svt = SvTYPE (sv); |
355 | |
386 | |
356 | if (ecb_expect_false (SvOBJECT (sv))) |
387 | if (ecb_expect_false (SvOBJECT (sv))) |
357 | { |
388 | { |
… | |
… | |
364 | HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash |
395 | HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash |
365 | ? cbor_tagged_stash |
396 | ? cbor_tagged_stash |
366 | : gv_stashpv ("CBOR::XS::Tagged" , 1); |
397 | : gv_stashpv ("CBOR::XS::Tagged" , 1); |
367 | |
398 | |
368 | HV *stash = SvSTASH (sv); |
399 | HV *stash = SvSTASH (sv); |
369 | GV *method; |
|
|
370 | |
400 | |
371 | if (stash == boolean_stash) |
401 | if (stash == boolean_stash) |
372 | encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); |
402 | { |
|
|
403 | encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE); |
|
|
404 | return; |
|
|
405 | } |
373 | else if (stash == error_stash) |
406 | else if (stash == error_stash) |
374 | encode_ch (enc, 0xe0 | 23); |
407 | { |
|
|
408 | encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF); |
|
|
409 | return; |
|
|
410 | } |
375 | else if (stash == tagged_stash) |
411 | else if (stash == tagged_stash) |
376 | { |
412 | { |
377 | if (svt != SVt_PVAV) |
413 | if (svt != SVt_PVAV) |
378 | croak ("encountered CBOR::XS::Tagged object that isn't an array"); |
414 | croak ("encountered CBOR::XS::Tagged object that isn't an array"); |
379 | |
415 | |
380 | encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); |
416 | encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1))); |
381 | encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); |
417 | encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); |
|
|
418 | |
|
|
419 | return; |
|
|
420 | } |
|
|
421 | } |
|
|
422 | |
|
|
423 | if (ecb_expect_false (SvREFCNT (sv) > 1) |
|
|
424 | && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)) |
|
|
425 | { |
|
|
426 | if (!enc->shareable) |
|
|
427 | enc->shareable = (HV *)sv_2mortal ((SV *)newHV ()); |
|
|
428 | |
|
|
429 | SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1); |
|
|
430 | |
|
|
431 | if (SvOK (*svp)) |
382 | } |
432 | { |
|
|
433 | encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF); |
|
|
434 | encode_uint (enc, MAJOR_POS_INT, SvUV (*svp)); |
|
|
435 | return; |
|
|
436 | } |
|
|
437 | else |
|
|
438 | { |
|
|
439 | sv_setuv (*svp, enc->shareable_idx); |
|
|
440 | ++enc->shareable_idx; |
|
|
441 | encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE); |
|
|
442 | } |
|
|
443 | } |
|
|
444 | |
|
|
445 | if (ecb_expect_false (SvOBJECT (sv))) |
|
|
446 | { |
|
|
447 | HV *stash = SvSTASH (sv); |
|
|
448 | GV *method; |
|
|
449 | |
383 | else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) |
450 | if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) |
384 | { |
451 | { |
385 | dSP; |
452 | dSP; |
386 | |
453 | |
387 | ENTER; SAVETMPS; PUSHMARK (SP); |
454 | ENTER; SAVETMPS; PUSHMARK (SP); |
388 | // we re-bless the reference to get overload and other niceties right |
455 | // we re-bless the reference to get overload and other niceties right |
… | |
… | |
420 | // catch this surprisingly common error |
487 | // catch this surprisingly common error |
421 | if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) |
488 | if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) |
422 | croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); |
489 | croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); |
423 | |
490 | |
424 | encode_tag (enc, CBOR_TAG_PERL_OBJECT); |
491 | encode_tag (enc, CBOR_TAG_PERL_OBJECT); |
425 | encode_uint (enc, 0x80, count + 1); |
492 | encode_uint (enc, MAJOR_ARRAY, count + 1); |
426 | encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); |
493 | encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); |
427 | |
494 | |
428 | while (count) |
495 | while (count) |
429 | encode_sv (enc, SP[1 - count--]); |
496 | encode_sv (enc, SP[1 - count--]); |
430 | |
497 | |
… | |
… | |
452 | { |
519 | { |
453 | double nv = SvNVX (sv); |
520 | double nv = SvNVX (sv); |
454 | |
521 | |
455 | need (enc, 9); |
522 | need (enc, 9); |
456 | |
523 | |
457 | if (ecb_expect_false (nv == (U32)nv)) |
524 | if (ecb_expect_false (nv == (NV)(U32)nv)) |
458 | encode_uint (enc, 0x00, (U32)nv); |
525 | encode_uint (enc, MAJOR_POS_INT, (U32)nv); |
459 | //TODO: maybe I32? |
526 | //TODO: maybe I32? |
460 | else if (ecb_expect_false (nv == (float)nv)) |
527 | else if (ecb_expect_false (nv == (float)nv)) |
461 | { |
528 | { |
462 | uint32_t fp = ecb_float_to_binary32 (nv); |
529 | uint32_t fp = ecb_float_to_binary32 (nv); |
463 | |
530 | |
464 | *enc->cur++ = 0xe0 | 26; |
531 | *enc->cur++ = MAJOR_MISC | MISC_FLOAT32; |
465 | |
532 | |
466 | if (!ecb_big_endian ()) |
533 | if (!ecb_big_endian ()) |
467 | fp = ecb_bswap32 (fp); |
534 | fp = ecb_bswap32 (fp); |
468 | |
535 | |
469 | memcpy (enc->cur, &fp, 4); |
536 | memcpy (enc->cur, &fp, 4); |
… | |
… | |
471 | } |
538 | } |
472 | else |
539 | else |
473 | { |
540 | { |
474 | uint64_t fp = ecb_double_to_binary64 (nv); |
541 | uint64_t fp = ecb_double_to_binary64 (nv); |
475 | |
542 | |
476 | *enc->cur++ = 0xe0 | 27; |
543 | *enc->cur++ = MAJOR_MISC | MISC_FLOAT64; |
477 | |
544 | |
478 | if (!ecb_big_endian ()) |
545 | if (!ecb_big_endian ()) |
479 | fp = ecb_bswap64 (fp); |
546 | fp = ecb_bswap64 (fp); |
480 | |
547 | |
481 | memcpy (enc->cur, &fp, 8); |
548 | memcpy (enc->cur, &fp, 8); |
… | |
… | |
497 | else if (SvNOKp (sv)) |
564 | else if (SvNOKp (sv)) |
498 | encode_nv (enc, sv); |
565 | encode_nv (enc, sv); |
499 | else if (SvIOKp (sv)) |
566 | else if (SvIOKp (sv)) |
500 | { |
567 | { |
501 | if (SvIsUV (sv)) |
568 | if (SvIsUV (sv)) |
502 | encode_uint (enc, 0x00, SvUVX (sv)); |
569 | encode_uint (enc, MAJOR_POS_INT, SvUVX (sv)); |
503 | else if (SvIVX (sv) >= 0) |
570 | else if (SvIVX (sv) >= 0) |
504 | encode_uint (enc, 0x00, SvIVX (sv)); |
571 | encode_uint (enc, MAJOR_POS_INT, SvIVX (sv)); |
505 | else |
572 | else |
506 | encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); |
573 | encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1)); |
507 | } |
574 | } |
508 | else if (SvROK (sv)) |
575 | else if (SvROK (sv)) |
509 | encode_rv (enc, SvRV (sv)); |
576 | encode_rv (enc, SvRV (sv)); |
510 | else if (!SvOK (sv)) |
577 | else if (!SvOK (sv)) |
511 | encode_ch (enc, 0xe0 | 22); |
578 | encode_ch (enc, MAJOR_MISC | SIMPLE_NULL); |
512 | else if (enc->cbor.flags & F_ALLOW_UNKNOWN) |
579 | else if (enc->cbor.flags & F_ALLOW_UNKNOWN) |
513 | encode_ch (enc, 0xe0 | 23); |
580 | encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF); |
514 | else |
581 | else |
515 | croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", |
582 | croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", |
516 | SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); |
583 | SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); |
517 | } |
584 | } |
518 | |
585 | |
519 | static SV * |
586 | static SV * |
520 | encode_cbor (SV *scalar, CBOR *cbor) |
587 | encode_cbor (SV *scalar, CBOR *cbor) |
521 | { |
588 | { |
522 | enc_t enc = { }; |
589 | enc_t enc = { 0 }; |
523 | |
590 | |
524 | enc.cbor = *cbor; |
591 | enc.cbor = *cbor; |
525 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
592 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
526 | enc.cur = SvPVX (enc.sv); |
593 | enc.cur = SvPVX (enc.sv); |
527 | enc.end = SvEND (enc.sv); |
594 | enc.end = SvEND (enc.sv); |
528 | |
595 | |
529 | SvPOK_only (enc.sv); |
596 | SvPOK_only (enc.sv); |
530 | |
597 | |
531 | if (cbor->flags & F_ALLOW_STRINGREF) |
598 | if (cbor->flags & F_PACK_STRINGS) |
532 | { |
599 | { |
533 | encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE); |
600 | encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE); |
534 | enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ()); |
601 | enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ()); |
535 | enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ()); |
602 | enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ()); |
536 | } |
603 | } |
… | |
… | |
571 | #define DEC_DEC_DEPTH --dec->depth |
638 | #define DEC_DEC_DEPTH --dec->depth |
572 | |
639 | |
573 | static UV |
640 | static UV |
574 | decode_uint (dec_t *dec) |
641 | decode_uint (dec_t *dec) |
575 | { |
642 | { |
576 | switch (*dec->cur & 31) |
643 | U8 m = *dec->cur & MINOR_MASK; |
577 | { |
644 | ++dec->cur; |
578 | case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7: |
|
|
579 | case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15: |
|
|
580 | case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23: |
|
|
581 | return *dec->cur++ & 31; |
|
|
582 | |
645 | |
583 | case 24: |
646 | if (ecb_expect_true (m < LENGTH_EXT1)) |
|
|
647 | return m; |
|
|
648 | else if (ecb_expect_true (m == LENGTH_EXT1)) |
|
|
649 | { |
584 | WANT (2); |
650 | WANT (1); |
585 | dec->cur += 2; |
651 | dec->cur += 1; |
586 | return dec->cur[-1]; |
652 | return dec->cur[-1]; |
587 | |
653 | } |
588 | case 25: |
654 | else if (ecb_expect_true (m == LENGTH_EXT2)) |
|
|
655 | { |
589 | WANT (3); |
656 | WANT (2); |
590 | dec->cur += 3; |
657 | dec->cur += 2; |
591 | return (((UV)dec->cur[-2]) << 8) |
658 | return (((UV)dec->cur[-2]) << 8) |
592 | | ((UV)dec->cur[-1]); |
659 | | ((UV)dec->cur[-1]); |
593 | |
660 | } |
594 | case 26: |
661 | else if (ecb_expect_true (m == LENGTH_EXT4)) |
|
|
662 | { |
595 | WANT (5); |
663 | WANT (4); |
596 | dec->cur += 5; |
664 | dec->cur += 4; |
597 | return (((UV)dec->cur[-4]) << 24) |
665 | return (((UV)dec->cur[-4]) << 24) |
598 | | (((UV)dec->cur[-3]) << 16) |
666 | | (((UV)dec->cur[-3]) << 16) |
599 | | (((UV)dec->cur[-2]) << 8) |
667 | | (((UV)dec->cur[-2]) << 8) |
600 | | ((UV)dec->cur[-1]); |
668 | | ((UV)dec->cur[-1]); |
601 | |
669 | } |
602 | case 27: |
670 | else if (ecb_expect_true (m == LENGTH_EXT8)) |
|
|
671 | { |
603 | WANT (9); |
672 | WANT (8); |
604 | dec->cur += 9; |
673 | dec->cur += 8; |
|
|
674 | |
|
|
675 | return |
|
|
676 | #if UVSIZE < 8 |
|
|
677 | 0 |
|
|
678 | #else |
605 | return (((UV)dec->cur[-8]) << 56) |
679 | (((UV)dec->cur[-8]) << 56) |
606 | | (((UV)dec->cur[-7]) << 48) |
680 | | (((UV)dec->cur[-7]) << 48) |
607 | | (((UV)dec->cur[-6]) << 40) |
681 | | (((UV)dec->cur[-6]) << 40) |
608 | | (((UV)dec->cur[-5]) << 32) |
682 | | (((UV)dec->cur[-5]) << 32) |
|
|
683 | #endif |
609 | | (((UV)dec->cur[-4]) << 24) |
684 | | (((UV)dec->cur[-4]) << 24) |
610 | | (((UV)dec->cur[-3]) << 16) |
685 | | (((UV)dec->cur[-3]) << 16) |
611 | | (((UV)dec->cur[-2]) << 8) |
686 | | (((UV)dec->cur[-2]) << 8) |
612 | | ((UV)dec->cur[-1]); |
687 | | ((UV)dec->cur[-1]); |
613 | |
688 | } |
614 | default: |
689 | else |
615 | ERR ("corrupted CBOR data (unsupported integer minor encoding)"); |
690 | ERR ("corrupted CBOR data (unsupported integer minor encoding)"); |
616 | } |
|
|
617 | |
691 | |
618 | fail: |
692 | fail: |
619 | return 0; |
693 | return 0; |
620 | } |
694 | } |
621 | |
695 | |
… | |
… | |
626 | { |
700 | { |
627 | AV *av = newAV (); |
701 | AV *av = newAV (); |
628 | |
702 | |
629 | DEC_INC_DEPTH; |
703 | DEC_INC_DEPTH; |
630 | |
704 | |
631 | if ((*dec->cur & 31) == 31) |
705 | if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF)) |
632 | { |
706 | { |
633 | ++dec->cur; |
707 | ++dec->cur; |
634 | |
708 | |
635 | for (;;) |
709 | for (;;) |
636 | { |
710 | { |
637 | WANT (1); |
711 | WANT (1); |
638 | |
712 | |
639 | if (*dec->cur == (0xe0 | 31)) |
713 | if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) |
640 | { |
714 | { |
641 | ++dec->cur; |
715 | ++dec->cur; |
642 | break; |
716 | break; |
643 | } |
717 | } |
644 | |
718 | |
… | |
… | |
647 | } |
721 | } |
648 | else |
722 | else |
649 | { |
723 | { |
650 | int i, len = decode_uint (dec); |
724 | int i, len = decode_uint (dec); |
651 | |
725 | |
|
|
726 | WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays |
652 | av_fill (av, len - 1); |
727 | av_fill (av, len - 1); |
653 | |
728 | |
654 | for (i = 0; i < len; ++i) |
729 | for (i = 0; i < len; ++i) |
655 | AvARRAY (av)[i] = decode_sv (dec); |
730 | AvARRAY (av)[i] = decode_sv (dec); |
656 | } |
731 | } |
… | |
… | |
669 | { |
744 | { |
670 | // for speed reasons, we specialcase single-string |
745 | // for speed reasons, we specialcase single-string |
671 | // byte or utf-8 strings as keys, but only when !stringref |
746 | // byte or utf-8 strings as keys, but only when !stringref |
672 | |
747 | |
673 | if (ecb_expect_true (!dec->stringref)) |
748 | if (ecb_expect_true (!dec->stringref)) |
674 | if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27) |
749 | if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8)) |
675 | { |
750 | { |
676 | I32 len = decode_uint (dec); |
751 | I32 len = decode_uint (dec); |
677 | char *key = (char *)dec->cur; |
752 | char *key = (char *)dec->cur; |
678 | |
753 | |
679 | dec->cur += len; |
754 | dec->cur += len; |
680 | |
755 | |
681 | if (ecb_expect_false (dec->stringref)) |
|
|
682 | av_push (dec->stringref, newSVpvn (key, len)); |
|
|
683 | |
|
|
684 | hv_store (hv, key, len, decode_sv (dec), 0); |
756 | hv_store (hv, key, len, decode_sv (dec), 0); |
685 | |
757 | |
686 | return; |
758 | return; |
687 | } |
759 | } |
688 | else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27) |
760 | else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8)) |
689 | { |
761 | { |
690 | I32 len = decode_uint (dec); |
762 | I32 len = decode_uint (dec); |
691 | char *key = (char *)dec->cur; |
763 | char *key = (char *)dec->cur; |
692 | |
764 | |
693 | dec->cur += len; |
765 | dec->cur += len; |
694 | |
766 | |
695 | if (ecb_expect_false (dec->stringref)) |
767 | if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8)) |
696 | av_push (dec->stringref, newSVpvn_utf8 (key, len, 1)); |
768 | if (!is_utf8_string (key, len)) |
|
|
769 | ERR ("corrupted CBOR data (invalid UTF-8 in map key)"); |
697 | |
770 | |
698 | hv_store (hv, key, -len, decode_sv (dec), 0); |
771 | hv_store (hv, key, -len, decode_sv (dec), 0); |
699 | |
772 | |
700 | return; |
773 | return; |
701 | } |
774 | } |
… | |
… | |
703 | SV *k = decode_sv (dec); |
776 | SV *k = decode_sv (dec); |
704 | SV *v = decode_sv (dec); |
777 | SV *v = decode_sv (dec); |
705 | |
778 | |
706 | hv_store_ent (hv, k, v, 0); |
779 | hv_store_ent (hv, k, v, 0); |
707 | SvREFCNT_dec (k); |
780 | SvREFCNT_dec (k); |
|
|
781 | |
|
|
782 | fail: |
|
|
783 | ; |
708 | } |
784 | } |
709 | |
785 | |
710 | static SV * |
786 | static SV * |
711 | decode_hv (dec_t *dec) |
787 | decode_hv (dec_t *dec) |
712 | { |
788 | { |
713 | HV *hv = newHV (); |
789 | HV *hv = newHV (); |
714 | |
790 | |
715 | DEC_INC_DEPTH; |
791 | DEC_INC_DEPTH; |
716 | |
792 | |
717 | if ((*dec->cur & 31) == 31) |
793 | if (*dec->cur == (MAJOR_MAP | MINOR_INDEF)) |
718 | { |
794 | { |
719 | ++dec->cur; |
795 | ++dec->cur; |
720 | |
796 | |
721 | for (;;) |
797 | for (;;) |
722 | { |
798 | { |
723 | WANT (1); |
799 | WANT (1); |
724 | |
800 | |
725 | if (*dec->cur == (0xe0 | 31)) |
801 | if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) |
726 | { |
802 | { |
727 | ++dec->cur; |
803 | ++dec->cur; |
728 | break; |
804 | break; |
729 | } |
805 | } |
730 | |
806 | |
… | |
… | |
751 | static SV * |
827 | static SV * |
752 | decode_str (dec_t *dec, int utf8) |
828 | decode_str (dec_t *dec, int utf8) |
753 | { |
829 | { |
754 | SV *sv = 0; |
830 | SV *sv = 0; |
755 | |
831 | |
756 | if ((*dec->cur & 31) == 31) |
832 | if ((*dec->cur & MINOR_MASK) == MINOR_INDEF) |
757 | { |
833 | { |
|
|
834 | // indefinite length strings |
758 | ++dec->cur; |
835 | ++dec->cur; |
759 | |
836 | |
|
|
837 | U8 major = *dec->cur & MAJOR_MISC; |
|
|
838 | |
760 | sv = newSVpvn ("", 0); |
839 | sv = newSVpvn ("", 0); |
761 | |
840 | |
762 | // not very fast, and certainly not robust against illegal input |
|
|
763 | for (;;) |
841 | for (;;) |
764 | { |
842 | { |
765 | WANT (1); |
843 | WANT (1); |
766 | |
844 | |
767 | if (*dec->cur == (0xe0 | 31)) |
845 | if ((*dec->cur - major) > LENGTH_EXT8) |
|
|
846 | if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) |
768 | { |
847 | { |
769 | ++dec->cur; |
848 | ++dec->cur; |
770 | break; |
849 | break; |
771 | } |
850 | } |
|
|
851 | else |
|
|
852 | ERR ("corrupted CBOR data (invalid chunks in indefinite length string)"); |
772 | |
853 | |
773 | sv_catsv (sv, decode_sv (dec)); |
854 | STRLEN len = decode_uint (dec); |
|
|
855 | |
|
|
856 | WANT (len); |
|
|
857 | sv_catpvn (sv, dec->cur, len); |
|
|
858 | dec->cur += len; |
774 | } |
859 | } |
775 | } |
860 | } |
776 | else |
861 | else |
777 | { |
862 | { |
778 | STRLEN len = decode_uint (dec); |
863 | STRLEN len = decode_uint (dec); |
… | |
… | |
785 | && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1)) |
870 | && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1)) |
786 | av_push (dec->stringref, SvREFCNT_inc_NN (sv)); |
871 | av_push (dec->stringref, SvREFCNT_inc_NN (sv)); |
787 | } |
872 | } |
788 | |
873 | |
789 | if (utf8) |
874 | if (utf8) |
|
|
875 | { |
|
|
876 | if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8)) |
|
|
877 | if (!is_utf8_string (SvPVX (sv), SvCUR (sv))) |
|
|
878 | ERR ("corrupted CBOR data (invalid UTF-8 in text string)"); |
|
|
879 | |
790 | SvUTF8_on (sv); |
880 | SvUTF8_on (sv); |
|
|
881 | } |
791 | |
882 | |
792 | return sv; |
883 | return sv; |
793 | |
884 | |
794 | fail: |
885 | fail: |
795 | SvREFCNT_dec (sv); |
886 | SvREFCNT_dec (sv); |
… | |
… | |
827 | } |
918 | } |
828 | break; |
919 | break; |
829 | |
920 | |
830 | case CBOR_TAG_STRINGREF: |
921 | case CBOR_TAG_STRINGREF: |
831 | { |
922 | { |
832 | if ((*dec->cur >> 5) != 0) |
923 | if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) |
833 | ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); |
924 | ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); |
834 | |
925 | |
835 | UV idx = decode_uint (dec); |
926 | UV idx = decode_uint (dec); |
836 | |
927 | |
837 | if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) |
928 | if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) |
… | |
… | |
844 | case CBOR_TAG_VALUE_SHAREABLE: |
935 | case CBOR_TAG_VALUE_SHAREABLE: |
845 | { |
936 | { |
846 | if (ecb_expect_false (!dec->shareable)) |
937 | if (ecb_expect_false (!dec->shareable)) |
847 | dec->shareable = (AV *)sv_2mortal ((SV *)newAV ()); |
938 | dec->shareable = (AV *)sv_2mortal ((SV *)newAV ()); |
848 | |
939 | |
|
|
940 | if (dec->cbor.flags & F_ALLOW_CYCLES) |
|
|
941 | { |
849 | sv = newSV (0); |
942 | sv = newSV (0); |
850 | av_push (dec->shareable, SvREFCNT_inc_NN (sv)); |
943 | av_push (dec->shareable, SvREFCNT_inc_NN (sv)); |
851 | |
944 | |
852 | SV *osv = decode_sv (dec); |
945 | SV *osv = decode_sv (dec); |
853 | sv_setsv (sv, osv); |
946 | sv_setsv (sv, osv); |
854 | SvREFCNT_dec_NN (osv); |
947 | SvREFCNT_dec_NN (osv); |
|
|
948 | } |
|
|
949 | else |
|
|
950 | { |
|
|
951 | av_push (dec->shareable, &PL_sv_undef); |
|
|
952 | int idx = AvFILLp (dec->shareable); |
|
|
953 | sv = decode_sv (dec); |
|
|
954 | av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv)); |
|
|
955 | } |
855 | } |
956 | } |
856 | break; |
957 | break; |
857 | |
958 | |
858 | case CBOR_TAG_VALUE_SHAREDREF: |
959 | case CBOR_TAG_VALUE_SHAREDREF: |
859 | { |
960 | { |
860 | if ((*dec->cur >> 5) != 0) |
961 | if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) |
861 | ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); |
962 | ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); |
862 | |
963 | |
863 | UV idx = decode_uint (dec); |
964 | UV idx = decode_uint (dec); |
864 | |
965 | |
865 | if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) |
966 | if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) |
866 | ERR ("corrupted CBOR data (sharedref index out of bounds)"); |
967 | ERR ("corrupted CBOR data (sharedref index out of bounds)"); |
867 | |
968 | |
868 | sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); |
969 | sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); |
|
|
970 | |
|
|
971 | if (sv == &PL_sv_undef) |
|
|
972 | ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled"); |
869 | } |
973 | } |
870 | break; |
974 | break; |
871 | |
975 | |
872 | case CBOR_TAG_PERL_OBJECT: |
976 | case CBOR_TAG_PERL_OBJECT: |
873 | { |
977 | { |
… | |
… | |
974 | static SV * |
1078 | static SV * |
975 | decode_sv (dec_t *dec) |
1079 | decode_sv (dec_t *dec) |
976 | { |
1080 | { |
977 | WANT (1); |
1081 | WANT (1); |
978 | |
1082 | |
979 | switch (*dec->cur >> 5) |
1083 | switch (*dec->cur >> MAJOR_SHIFT) |
980 | { |
1084 | { |
981 | case 0: // unsigned int |
1085 | case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec)); |
982 | return newSVuv (decode_uint (dec)); |
1086 | case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec)); |
983 | case 1: // negative int |
1087 | case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0); |
984 | return newSViv (-1 - (IV)decode_uint (dec)); |
1088 | case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1); |
985 | case 2: // octet string |
1089 | case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec); |
986 | return decode_str (dec, 0); |
1090 | case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec); |
987 | case 3: // utf-8 string |
1091 | case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec); |
988 | return decode_str (dec, 1); |
1092 | |
989 | case 4: // array |
1093 | case MAJOR_MISC >> MAJOR_SHIFT: |
990 | return decode_av (dec); |
|
|
991 | case 5: // map |
|
|
992 | return decode_hv (dec); |
|
|
993 | case 6: // tag |
|
|
994 | return decode_tagged (dec); |
|
|
995 | case 7: // misc |
|
|
996 | switch (*dec->cur++ & 31) |
1094 | switch (*dec->cur++ & MINOR_MASK) |
997 | { |
1095 | { |
998 | case 20: |
1096 | case SIMPLE_FALSE: |
999 | #if CBOR_SLOW |
1097 | #if CBOR_SLOW |
1000 | types_false = get_bool ("Types::Serialiser::false"); |
1098 | types_false = get_bool ("Types::Serialiser::false"); |
1001 | #endif |
1099 | #endif |
1002 | return newSVsv (types_false); |
1100 | return newSVsv (types_false); |
1003 | case 21: |
1101 | case SIMPLE_TRUE: |
1004 | #if CBOR_SLOW |
1102 | #if CBOR_SLOW |
1005 | types_true = get_bool ("Types::Serialiser::true"); |
1103 | types_true = get_bool ("Types::Serialiser::true"); |
1006 | #endif |
1104 | #endif |
1007 | return newSVsv (types_true); |
1105 | return newSVsv (types_true); |
1008 | case 22: |
1106 | case SIMPLE_NULL: |
1009 | return newSVsv (&PL_sv_undef); |
1107 | return newSVsv (&PL_sv_undef); |
1010 | case 23: |
1108 | case SIMPLE_UNDEF: |
1011 | #if CBOR_SLOW |
1109 | #if CBOR_SLOW |
1012 | types_error = get_bool ("Types::Serialiser::error"); |
1110 | types_error = get_bool ("Types::Serialiser::error"); |
1013 | #endif |
1111 | #endif |
1014 | return newSVsv (types_error); |
1112 | return newSVsv (types_error); |
1015 | |
1113 | |
1016 | case 25: |
1114 | case MISC_FLOAT16: |
1017 | { |
1115 | { |
1018 | WANT (2); |
1116 | WANT (2); |
1019 | |
1117 | |
1020 | uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; |
1118 | uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; |
1021 | dec->cur += 2; |
1119 | dec->cur += 2; |
1022 | |
1120 | |
1023 | return newSVnv (ecb_binary16_to_float (fp)); |
1121 | return newSVnv (ecb_binary16_to_float (fp)); |
1024 | } |
1122 | } |
1025 | |
1123 | |
1026 | case 26: |
1124 | case MISC_FLOAT32: |
1027 | { |
1125 | { |
1028 | uint32_t fp; |
1126 | uint32_t fp; |
1029 | WANT (4); |
1127 | WANT (4); |
1030 | memcpy (&fp, dec->cur, 4); |
1128 | memcpy (&fp, dec->cur, 4); |
1031 | dec->cur += 4; |
1129 | dec->cur += 4; |
… | |
… | |
1034 | fp = ecb_bswap32 (fp); |
1132 | fp = ecb_bswap32 (fp); |
1035 | |
1133 | |
1036 | return newSVnv (ecb_binary32_to_float (fp)); |
1134 | return newSVnv (ecb_binary32_to_float (fp)); |
1037 | } |
1135 | } |
1038 | |
1136 | |
1039 | case 27: |
1137 | case MISC_FLOAT64: |
1040 | { |
1138 | { |
1041 | uint64_t fp; |
1139 | uint64_t fp; |
1042 | WANT (8); |
1140 | WANT (8); |
1043 | memcpy (&fp, dec->cur, 8); |
1141 | memcpy (&fp, dec->cur, 8); |
1044 | dec->cur += 8; |
1142 | dec->cur += 8; |
… | |
… | |
1047 | fp = ecb_bswap64 (fp); |
1145 | fp = ecb_bswap64 (fp); |
1048 | |
1146 | |
1049 | return newSVnv (ecb_binary64_to_double (fp)); |
1147 | return newSVnv (ecb_binary64_to_double (fp)); |
1050 | } |
1148 | } |
1051 | |
1149 | |
1052 | // 0..19 unassigned |
1150 | // 0..19 unassigned simple |
1053 | // 24 reserved + unassigned (reserved values are not encodable) |
1151 | // 24 reserved + unassigned simple (reserved values are not encodable) |
|
|
1152 | // 28-30 unassigned misc |
|
|
1153 | // 31 break code |
1054 | default: |
1154 | default: |
1055 | ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); |
1155 | ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)"); |
1056 | } |
1156 | } |
1057 | |
1157 | |
1058 | break; |
1158 | break; |
1059 | } |
1159 | } |
1060 | |
1160 | |
… | |
… | |
1063 | } |
1163 | } |
1064 | |
1164 | |
1065 | static SV * |
1165 | static SV * |
1066 | decode_cbor (SV *string, CBOR *cbor, char **offset_return) |
1166 | decode_cbor (SV *string, CBOR *cbor, char **offset_return) |
1067 | { |
1167 | { |
1068 | dec_t dec = { }; |
1168 | dec_t dec = { 0 }; |
1069 | SV *sv; |
1169 | SV *sv; |
1070 | STRLEN len; |
1170 | STRLEN len; |
1071 | char *data = SvPVbyte (string, len); |
1171 | char *data = SvPVbyte (string, len); |
1072 | |
1172 | |
1073 | if (len > cbor->max_size && cbor->max_size) |
1173 | if (len > cbor->max_size && cbor->max_size) |
… | |
… | |
1087 | if (dec.cur != dec.end && !dec.err) |
1187 | if (dec.cur != dec.end && !dec.err) |
1088 | dec.err = "garbage after CBOR object"; |
1188 | dec.err = "garbage after CBOR object"; |
1089 | |
1189 | |
1090 | if (dec.err) |
1190 | if (dec.err) |
1091 | { |
1191 | { |
|
|
1192 | if (dec.shareable) |
|
|
1193 | { |
|
|
1194 | // need to break cyclic links, which whould all be in shareable |
|
|
1195 | int i; |
|
|
1196 | SV **svp; |
|
|
1197 | |
|
|
1198 | for (i = av_len (dec.shareable) + 1; i--; ) |
|
|
1199 | if ((svp = av_fetch (dec.shareable, i, 0))) |
|
|
1200 | sv_setsv (*svp, &PL_sv_undef); |
|
|
1201 | } |
|
|
1202 | |
1092 | SvREFCNT_dec (sv); |
1203 | SvREFCNT_dec (sv); |
1093 | croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); |
1204 | croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); |
1094 | } |
1205 | } |
1095 | |
1206 | |
1096 | sv = sv_2mortal (sv); |
1207 | sv = sv_2mortal (sv); |
1097 | |
1208 | |
1098 | return sv; |
1209 | return sv; |
1099 | } |
1210 | } |
1100 | |
1211 | |
|
|
1212 | ///////////////////////////////////////////////////////////////////////////// |
|
|
1213 | // incremental parser |
|
|
1214 | |
|
|
1215 | #define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0) |
|
|
1216 | |
|
|
1217 | // returns 0 for notyet, 1 for success or error |
|
|
1218 | static int |
|
|
1219 | incr_parse (CBOR *self, SV *cborstr) |
|
|
1220 | { |
|
|
1221 | STRLEN cur; |
|
|
1222 | SvPV (cborstr, cur); |
|
|
1223 | |
|
|
1224 | while (ecb_expect_true (self->incr_need <= cur)) |
|
|
1225 | { |
|
|
1226 | // table of integer count bytes |
|
|
1227 | static I8 incr_len[MINOR_MASK + 1] = { |
|
|
1228 | 0, 0, 0, 0, 0, 0, 0, 0, |
|
|
1229 | 0, 0, 0, 0, 0, 0, 0, 0, |
|
|
1230 | 0, 0, 0, 0, 0, 0, 0, 0, |
|
|
1231 | 1, 2, 4, 8,-1,-1,-1,-2 |
|
|
1232 | }; |
|
|
1233 | |
|
|
1234 | const U8 *p = SvPVX (cborstr) + self->incr_pos; |
|
|
1235 | U8 m = *p & MINOR_MASK; |
|
|
1236 | IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]); |
|
|
1237 | I8 ilen = incr_len[m]; |
|
|
1238 | |
|
|
1239 | self->incr_need = self->incr_pos + 1; |
|
|
1240 | |
|
|
1241 | if (ecb_expect_false (ilen < 0)) |
|
|
1242 | { |
|
|
1243 | if (m != MINOR_INDEF) |
|
|
1244 | return 1; // error |
|
|
1245 | |
|
|
1246 | if (*p == (MAJOR_MISC | MINOR_INDEF)) |
|
|
1247 | { |
|
|
1248 | if (count >= 0) |
|
|
1249 | return 1; // error |
|
|
1250 | |
|
|
1251 | count = 1; |
|
|
1252 | } |
|
|
1253 | else |
|
|
1254 | { |
|
|
1255 | av_push (self->incr_count, newSViv (-1)); //TODO: nest |
|
|
1256 | count = -1; |
|
|
1257 | } |
|
|
1258 | } |
|
|
1259 | else |
|
|
1260 | { |
|
|
1261 | self->incr_need += ilen; |
|
|
1262 | if (ecb_expect_false (self->incr_need > cur)) |
|
|
1263 | return 0; |
|
|
1264 | |
|
|
1265 | int major = *p >> MAJOR_SHIFT; |
|
|
1266 | |
|
|
1267 | switch (major) |
|
|
1268 | { |
|
|
1269 | case MAJOR_TAG >> MAJOR_SHIFT: |
|
|
1270 | ++count; // tags merely prefix another value |
|
|
1271 | break; |
|
|
1272 | |
|
|
1273 | case MAJOR_BYTES >> MAJOR_SHIFT: |
|
|
1274 | case MAJOR_TEXT >> MAJOR_SHIFT: |
|
|
1275 | case MAJOR_ARRAY >> MAJOR_SHIFT: |
|
|
1276 | case MAJOR_MAP >> MAJOR_SHIFT: |
|
|
1277 | { |
|
|
1278 | UV len; |
|
|
1279 | |
|
|
1280 | if (ecb_expect_false (ilen)) |
|
|
1281 | { |
|
|
1282 | len = 0; |
|
|
1283 | |
|
|
1284 | do { |
|
|
1285 | len = (len << 8) | *++p; |
|
|
1286 | } while (--ilen); |
|
|
1287 | } |
|
|
1288 | else |
|
|
1289 | len = m; |
|
|
1290 | |
|
|
1291 | switch (major) |
|
|
1292 | { |
|
|
1293 | case MAJOR_BYTES >> MAJOR_SHIFT: |
|
|
1294 | case MAJOR_TEXT >> MAJOR_SHIFT: |
|
|
1295 | self->incr_need += len; |
|
|
1296 | if (ecb_expect_false (self->incr_need > cur)) |
|
|
1297 | return 0; |
|
|
1298 | |
|
|
1299 | break; |
|
|
1300 | |
|
|
1301 | case MAJOR_MAP >> MAJOR_SHIFT: |
|
|
1302 | len <<= 1; |
|
|
1303 | case MAJOR_ARRAY >> MAJOR_SHIFT: |
|
|
1304 | if (len) |
|
|
1305 | { |
|
|
1306 | av_push (self->incr_count, newSViv (len + 1)); //TODO: nest |
|
|
1307 | count = len + 1; |
|
|
1308 | } |
|
|
1309 | break; |
|
|
1310 | } |
|
|
1311 | } |
|
|
1312 | } |
|
|
1313 | } |
|
|
1314 | |
|
|
1315 | self->incr_pos = self->incr_need; |
|
|
1316 | |
|
|
1317 | if (count > 0) |
|
|
1318 | { |
|
|
1319 | while (!--count) |
|
|
1320 | { |
|
|
1321 | if (!AvFILLp (self->incr_count)) |
|
|
1322 | return 1; // done |
|
|
1323 | |
|
|
1324 | SvREFCNT_dec_NN (av_pop (self->incr_count)); |
|
|
1325 | count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]); |
|
|
1326 | } |
|
|
1327 | |
|
|
1328 | SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count; |
|
|
1329 | } |
|
|
1330 | } |
|
|
1331 | |
|
|
1332 | return 0; |
|
|
1333 | } |
|
|
1334 | |
|
|
1335 | |
1101 | ///////////////////////////////////////////////////////////////////////////// |
1336 | ///////////////////////////////////////////////////////////////////////////// |
1102 | // XS interface functions |
1337 | // XS interface functions |
1103 | |
1338 | |
1104 | MODULE = CBOR::XS PACKAGE = CBOR::XS |
1339 | MODULE = CBOR::XS PACKAGE = CBOR::XS |
1105 | |
1340 | |
… | |
… | |
1145 | void shrink (CBOR *self, int enable = 1) |
1380 | void shrink (CBOR *self, int enable = 1) |
1146 | ALIAS: |
1381 | ALIAS: |
1147 | shrink = F_SHRINK |
1382 | shrink = F_SHRINK |
1148 | allow_unknown = F_ALLOW_UNKNOWN |
1383 | allow_unknown = F_ALLOW_UNKNOWN |
1149 | allow_sharing = F_ALLOW_SHARING |
1384 | allow_sharing = F_ALLOW_SHARING |
1150 | allow_stringref = F_ALLOW_STRINGREF |
1385 | allow_cycles = F_ALLOW_CYCLES |
|
|
1386 | pack_strings = F_PACK_STRINGS |
|
|
1387 | validate_utf8 = F_VALIDATE_UTF8 |
1151 | PPCODE: |
1388 | PPCODE: |
1152 | { |
1389 | { |
1153 | if (enable) |
1390 | if (enable) |
1154 | self->flags |= ix; |
1391 | self->flags |= ix; |
1155 | else |
1392 | else |
… | |
… | |
1161 | void get_shrink (CBOR *self) |
1398 | void get_shrink (CBOR *self) |
1162 | ALIAS: |
1399 | ALIAS: |
1163 | get_shrink = F_SHRINK |
1400 | get_shrink = F_SHRINK |
1164 | get_allow_unknown = F_ALLOW_UNKNOWN |
1401 | get_allow_unknown = F_ALLOW_UNKNOWN |
1165 | get_allow_sharing = F_ALLOW_SHARING |
1402 | get_allow_sharing = F_ALLOW_SHARING |
1166 | get_allow_stringref = F_ALLOW_STRINGREF |
1403 | get_allow_cycles = F_ALLOW_CYCLES |
|
|
1404 | get_pack_strings = F_PACK_STRINGS |
|
|
1405 | get_validate_utf8 = F_VALIDATE_UTF8 |
1167 | PPCODE: |
1406 | PPCODE: |
1168 | XPUSHs (boolSV (self->flags & ix)); |
1407 | XPUSHs (boolSV (self->flags & ix)); |
1169 | |
1408 | |
1170 | void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) |
1409 | void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) |
1171 | PPCODE: |
1410 | PPCODE: |
… | |
… | |
1220 | EXTEND (SP, 2); |
1459 | EXTEND (SP, 2); |
1221 | PUSHs (sv); |
1460 | PUSHs (sv); |
1222 | PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); |
1461 | PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); |
1223 | } |
1462 | } |
1224 | |
1463 | |
|
|
1464 | void incr_parse (CBOR *self, SV *cborstr) |
|
|
1465 | ALIAS: |
|
|
1466 | incr_parse_multiple = 1 |
|
|
1467 | PPCODE: |
|
|
1468 | { |
|
|
1469 | if (SvUTF8 (cborstr)) |
|
|
1470 | sv_utf8_downgrade (cborstr, 0); |
|
|
1471 | |
|
|
1472 | if (!self->incr_count) |
|
|
1473 | { |
|
|
1474 | self->incr_count = newAV (); |
|
|
1475 | self->incr_pos = 0; |
|
|
1476 | self->incr_need = 1; |
|
|
1477 | |
|
|
1478 | av_push (self->incr_count, newSViv (1)); |
|
|
1479 | } |
|
|
1480 | |
|
|
1481 | do |
|
|
1482 | { |
|
|
1483 | if (!incr_parse (self, cborstr)) |
|
|
1484 | { |
|
|
1485 | if (self->incr_need > self->max_size && self->max_size) |
|
|
1486 | croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu", |
|
|
1487 | (unsigned long)self->incr_need, (unsigned long)self->max_size); |
|
|
1488 | |
|
|
1489 | break; |
|
|
1490 | } |
|
|
1491 | |
|
|
1492 | SV *sv; |
|
|
1493 | char *offset; |
|
|
1494 | |
|
|
1495 | PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN; |
|
|
1496 | XPUSHs (sv); |
|
|
1497 | |
|
|
1498 | sv_chop (cborstr, offset); |
|
|
1499 | |
|
|
1500 | av_clear (self->incr_count); |
|
|
1501 | av_push (self->incr_count, newSViv (1)); |
|
|
1502 | |
|
|
1503 | self->incr_pos = 0; |
|
|
1504 | self->incr_need = self->incr_pos + 1; |
|
|
1505 | } |
|
|
1506 | while (ix); |
|
|
1507 | } |
|
|
1508 | |
|
|
1509 | void incr_reset (CBOR *self) |
|
|
1510 | CODE: |
|
|
1511 | { |
|
|
1512 | SvREFCNT_dec (self->incr_count); |
|
|
1513 | self->incr_count = 0; |
|
|
1514 | } |
|
|
1515 | |
1225 | void DESTROY (CBOR *self) |
1516 | void DESTROY (CBOR *self) |
1226 | PPCODE: |
1517 | PPCODE: |
1227 | cbor_free (self); |
1518 | cbor_free (self); |
1228 | |
1519 | |
1229 | PROTOTYPES: ENABLE |
1520 | PROTOTYPES: ENABLE |
1230 | |
1521 | |
1231 | void encode_cbor (SV *scalar) |
1522 | void encode_cbor (SV *scalar) |
|
|
1523 | ALIAS: |
|
|
1524 | encode_cbor = 0 |
|
|
1525 | encode_cbor_sharing = F_ALLOW_SHARING |
1232 | PPCODE: |
1526 | PPCODE: |
1233 | { |
1527 | { |
1234 | CBOR cbor; |
1528 | CBOR cbor; |
1235 | cbor_init (&cbor); |
1529 | cbor_init (&cbor); |
|
|
1530 | cbor.flags |= ix; |
1236 | PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; |
1531 | PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; |
1237 | XPUSHs (scalar); |
1532 | XPUSHs (scalar); |
1238 | } |
1533 | } |
1239 | |
1534 | |
1240 | void decode_cbor (SV *cborstr) |
1535 | void decode_cbor (SV *cborstr) |