… | |
… | |
19 | # define HvNAMELEN(hv) HvNAMELEN_get (hv) |
19 | # define HvNAMELEN(hv) HvNAMELEN_get (hv) |
20 | #endif |
20 | #endif |
21 | #ifndef HvNAMEUTF8 |
21 | #ifndef HvNAMEUTF8 |
22 | # define HvNAMEUTF8(hv) 0 |
22 | # define HvNAMEUTF8(hv) 0 |
23 | #endif |
23 | #endif |
|
|
24 | #ifndef SvREFCNT_dec_NN |
|
|
25 | # define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv) |
|
|
26 | #endif |
|
|
27 | |
|
|
28 | // known major and minor types |
|
|
29 | enum cbor_type |
|
|
30 | { |
|
|
31 | MAJOR_SHIFT = 5, |
|
|
32 | MINOR_MASK = 0x1f, |
|
|
33 | |
|
|
34 | MAJOR_POS_INT = 0 << MAJOR_SHIFT, |
|
|
35 | MAJOR_NEG_INT = 1 << MAJOR_SHIFT, |
|
|
36 | MAJOR_BYTES = 2 << MAJOR_SHIFT, |
|
|
37 | MAJOR_TEXT = 3 << MAJOR_SHIFT, |
|
|
38 | MAJOR_ARRAY = 4 << MAJOR_SHIFT, |
|
|
39 | MAJOR_MAP = 5 << MAJOR_SHIFT, |
|
|
40 | MAJOR_TAG = 6 << MAJOR_SHIFT, |
|
|
41 | MAJOR_MISC = 7 << MAJOR_SHIFT, |
|
|
42 | |
|
|
43 | // INT/STRING/ARRAY/MAP subtypes |
|
|
44 | LENGTH_EXT1 = 24, |
|
|
45 | LENGTH_EXT2 = 25, |
|
|
46 | LENGTH_EXT4 = 26, |
|
|
47 | LENGTH_EXT8 = 27, |
|
|
48 | |
|
|
49 | // SIMPLE types (effectively MISC subtypes) |
|
|
50 | SIMPLE_FALSE = 20, |
|
|
51 | SIMPLE_TRUE = 21, |
|
|
52 | SIMPLE_NULL = 22, |
|
|
53 | SIMPLE_UNDEF = 23, |
|
|
54 | |
|
|
55 | // MISC subtype (unused) |
|
|
56 | MISC_EXT1 = 24, |
|
|
57 | MISC_FLOAT16 = 25, |
|
|
58 | MISC_FLOAT32 = 26, |
|
|
59 | MISC_FLOAT64 = 27, |
|
|
60 | |
|
|
61 | // BYTES/TEXT/ARRAY/MAP |
|
|
62 | MINOR_INDEF = 31, |
|
|
63 | }; |
24 | |
64 | |
25 | // known tags |
65 | // known tags |
26 | enum cbor_tag |
66 | enum cbor_tag |
27 | { |
67 | { |
28 | // inofficial extensions (pending iana registration) |
68 | // extensions |
|
|
69 | CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref |
29 | CBOR_TAG_PERL_OBJECT = 24, // http://cbor.schmorp.de/perl-object |
70 | CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object |
30 | CBOR_TAG_GENERIC_OBJECT = 25, // http://cbor.schmorp.de/generic-object |
71 | CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object |
31 | CBOR_TAG_VALUE_SHAREABLE = 26, // http://cbor.schmorp.de/value-sharing |
72 | CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing |
32 | CBOR_TAG_VALUE_SHAREDREF = 27, // http://cbor.schmorp.de/value-sharing |
73 | CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing |
33 | CBOR_TAG_STRINGREF_NAMESPACE = 65537, // http://cbor.schmorp.de/stringref |
74 | 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 |
75 | CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection |
36 | |
76 | |
37 | // rfc7049 |
77 | // rfc7049 |
38 | CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 |
78 | CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 |
39 | CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any |
79 | CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any |
40 | CBOR_TAG_POS_BIGNUM = 2, // byte string |
80 | CBOR_TAG_POS_BIGNUM = 2, // byte string |
41 | CBOR_TAG_NEG_BIGNUM = 3, // byte string |
81 | CBOR_TAG_NEG_BIGNUM = 3, // byte string |
42 | CBOR_TAG_DECIMAL = 4, // decimal fraction, array |
82 | CBOR_TAG_DECIMAL = 4, // decimal fraction, array |
43 | CBOR_TAG_BIGFLOAT = 5, // array |
83 | CBOR_TAG_BIGFLOAT = 5, // array |
44 | |
84 | |
45 | CBOR_TAG_CONV_B64U = 21, // base64url, any |
85 | CBOR_TAG_CONV_B64U = 21, // base64url, any |
46 | CBOR_TAG_CONV_B64 = 22, // base64, any |
86 | CBOR_TAG_CONV_B64 = 22, // base64, any |
47 | CBOR_TAG_CONV_HEX = 23, // base16, any |
87 | CBOR_TAG_CONV_HEX = 23, // base16, any |
48 | CBOR_TAG_CBOR = 24, // embedded cbor, byte string |
88 | CBOR_TAG_CBOR = 24, // embedded cbor, byte string |
49 | |
89 | |
50 | CBOR_TAG_URI = 32, // URI rfc3986, utf-8 |
90 | CBOR_TAG_URI = 32, // URI rfc3986, utf-8 |
51 | CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8 |
91 | CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8 |
52 | CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8 |
92 | CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8 |
53 | CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8 |
93 | CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8 |
54 | CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 |
94 | CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 |
55 | |
95 | |
56 | CBOR_TAG_MAGIC = 55799 // self-describe cbor |
96 | CBOR_TAG_MAGIC = 55799, // self-describe cbor |
57 | }; |
97 | }; |
58 | |
98 | |
59 | #define F_SHRINK 0x00000001UL |
99 | #define F_SHRINK 0x00000001UL |
60 | #define F_ALLOW_UNKNOWN 0x00000002UL |
100 | #define F_ALLOW_UNKNOWN 0x00000002UL |
61 | #define F_ALLOW_SHARING 0x00000004UL //TODO |
101 | #define F_ALLOW_SHARING 0x00000004UL |
62 | #define F_ALLOW_STRINGREF 0x00000008UL //TODO |
102 | #define F_ALLOW_CYCLES 0x00000008UL |
|
|
103 | #define F_PACK_STRINGS 0x00000010UL |
|
|
104 | #define F_VALIDATE_UTF8 0x00000020UL |
63 | |
105 | |
64 | #define INIT_SIZE 32 // initial scalar size to be allocated |
106 | #define INIT_SIZE 32 // initial scalar size to be allocated |
65 | |
107 | |
66 | #define SB do { |
108 | #define SB do { |
67 | #define SE } while (0) |
109 | #define SE } while (0) |
… | |
… | |
86 | typedef struct { |
128 | typedef struct { |
87 | U32 flags; |
129 | U32 flags; |
88 | U32 max_depth; |
130 | U32 max_depth; |
89 | STRLEN max_size; |
131 | STRLEN max_size; |
90 | SV *filter; |
132 | SV *filter; |
|
|
133 | |
|
|
134 | // for the incremental parser |
|
|
135 | STRLEN incr_pos; // the current offset into the text |
|
|
136 | STRLEN incr_need; // minimum bytes needed to decode |
|
|
137 | AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef. |
91 | } CBOR; |
138 | } CBOR; |
92 | |
139 | |
93 | ecb_inline void |
140 | ecb_inline void |
94 | cbor_init (CBOR *cbor) |
141 | cbor_init (CBOR *cbor) |
95 | { |
142 | { |
… | |
… | |
99 | |
146 | |
100 | ecb_inline void |
147 | ecb_inline void |
101 | cbor_free (CBOR *cbor) |
148 | cbor_free (CBOR *cbor) |
102 | { |
149 | { |
103 | SvREFCNT_dec (cbor->filter); |
150 | SvREFCNT_dec (cbor->filter); |
|
|
151 | SvREFCNT_dec (cbor->incr_count); |
104 | } |
152 | } |
105 | |
153 | |
106 | ///////////////////////////////////////////////////////////////////////////// |
154 | ///////////////////////////////////////////////////////////////////////////// |
107 | // utility functions |
155 | // utility functions |
108 | |
156 | |
… | |
… | |
138 | { |
186 | { |
139 | return idx > 23 |
187 | return idx > 23 |
140 | ? idx > 0xffU |
188 | ? idx > 0xffU |
141 | ? idx > 0xffffU |
189 | ? idx > 0xffffU |
142 | ? idx > 0xffffffffU |
190 | ? idx > 0xffffffffU |
|
|
191 | ? 11 |
143 | ? 7 |
192 | : 7 |
144 | : 6 |
|
|
145 | : 5 |
193 | : 5 |
146 | : 4 |
194 | : 4 |
147 | : 3; |
195 | : 3; |
148 | } |
196 | } |
149 | |
197 | |
… | |
… | |
186 | static void |
234 | static void |
187 | encode_uint (enc_t *enc, int major, UV len) |
235 | encode_uint (enc_t *enc, int major, UV len) |
188 | { |
236 | { |
189 | need (enc, 9); |
237 | need (enc, 9); |
190 | |
238 | |
191 | if (len < 24) |
239 | if (ecb_expect_true (len < LENGTH_EXT1)) |
192 | *enc->cur++ = major | len; |
240 | *enc->cur++ = major | len; |
193 | else if (len <= 0xff) |
241 | else if (ecb_expect_true (len <= 0xffU)) |
194 | { |
242 | { |
195 | *enc->cur++ = major | 24; |
243 | *enc->cur++ = major | LENGTH_EXT1; |
196 | *enc->cur++ = len; |
244 | *enc->cur++ = len; |
197 | } |
245 | } |
198 | else if (len <= 0xffff) |
246 | else if (len <= 0xffffU) |
199 | { |
247 | { |
200 | *enc->cur++ = major | 25; |
248 | *enc->cur++ = major | LENGTH_EXT2; |
201 | *enc->cur++ = len >> 8; |
249 | *enc->cur++ = len >> 8; |
202 | *enc->cur++ = len; |
250 | *enc->cur++ = len; |
203 | } |
251 | } |
204 | else if (len <= 0xffffffff) |
252 | else if (len <= 0xffffffffU) |
205 | { |
253 | { |
206 | *enc->cur++ = major | 26; |
254 | *enc->cur++ = major | LENGTH_EXT4; |
207 | *enc->cur++ = len >> 24; |
255 | *enc->cur++ = len >> 24; |
208 | *enc->cur++ = len >> 16; |
256 | *enc->cur++ = len >> 16; |
209 | *enc->cur++ = len >> 8; |
257 | *enc->cur++ = len >> 8; |
210 | *enc->cur++ = len; |
258 | *enc->cur++ = len; |
211 | } |
259 | } |
212 | else |
260 | else |
213 | { |
261 | { |
214 | *enc->cur++ = major | 27; |
262 | *enc->cur++ = major | LENGTH_EXT8; |
215 | *enc->cur++ = len >> 56; |
263 | *enc->cur++ = len >> 56; |
216 | *enc->cur++ = len >> 48; |
264 | *enc->cur++ = len >> 48; |
217 | *enc->cur++ = len >> 40; |
265 | *enc->cur++ = len >> 40; |
218 | *enc->cur++ = len >> 32; |
266 | *enc->cur++ = len >> 32; |
219 | *enc->cur++ = len >> 24; |
267 | *enc->cur++ = len >> 24; |
… | |
… | |
224 | } |
272 | } |
225 | |
273 | |
226 | ecb_inline void |
274 | ecb_inline void |
227 | encode_tag (enc_t *enc, UV tag) |
275 | encode_tag (enc_t *enc, UV tag) |
228 | { |
276 | { |
229 | encode_uint (enc, 0xc0, tag); |
277 | encode_uint (enc, MAJOR_TAG, tag); |
|
|
278 | } |
|
|
279 | |
|
|
280 | ecb_inline void |
|
|
281 | encode_str (enc_t *enc, int utf8, char *str, STRLEN len) |
|
|
282 | { |
|
|
283 | encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len); |
|
|
284 | need (enc, len); |
|
|
285 | memcpy (enc->cur, str, len); |
|
|
286 | enc->cur += len; |
230 | } |
287 | } |
231 | |
288 | |
232 | static void |
289 | static void |
233 | encode_str (enc_t *enc, int utf8, char *str, STRLEN len) |
290 | encode_strref (enc_t *enc, int utf8, char *str, STRLEN len) |
234 | { |
291 | { |
235 | if (ecb_expect_false (enc->cbor.flags & F_ALLOW_STRINGREF)) |
292 | if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS)) |
236 | { |
293 | { |
237 | SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1); |
294 | SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1); |
238 | |
295 | |
239 | if (SvOK (*svp)) |
296 | if (SvOK (*svp)) |
240 | { |
297 | { |
241 | // already registered, use stringref |
298 | // already registered, use stringref |
242 | encode_tag (enc, CBOR_TAG_STRINGREF); |
299 | encode_tag (enc, CBOR_TAG_STRINGREF); |
243 | encode_uint (enc, 0x00, SvUV (*svp)); |
300 | encode_uint (enc, MAJOR_POS_INT, SvUV (*svp)); |
244 | return; |
301 | return; |
245 | } |
302 | } |
246 | else if (len >= minimum_string_length (enc->stringref_idx)) |
303 | else if (len >= minimum_string_length (enc->stringref_idx)) |
247 | { |
304 | { |
248 | // register only |
305 | // register only |
249 | sv_setuv (*svp, enc->stringref_idx); |
306 | sv_setuv (*svp, enc->stringref_idx); |
250 | ++enc->stringref_idx; |
307 | ++enc->stringref_idx; |
251 | } |
308 | } |
252 | } |
309 | } |
253 | |
310 | |
254 | encode_uint (enc, utf8 ? 0x60 : 0x40, len); |
311 | encode_str (enc, utf8, str, len); |
255 | need (enc, len); |
|
|
256 | memcpy (enc->cur, str, len); |
|
|
257 | enc->cur += len; |
|
|
258 | } |
312 | } |
259 | |
313 | |
260 | static void encode_sv (enc_t *enc, SV *sv); |
314 | static void encode_sv (enc_t *enc, SV *sv); |
261 | |
315 | |
262 | static void |
316 | static void |
… | |
… | |
267 | if (enc->depth >= enc->cbor.max_depth) |
321 | if (enc->depth >= enc->cbor.max_depth) |
268 | croak (ERR_NESTING_EXCEEDED); |
322 | croak (ERR_NESTING_EXCEEDED); |
269 | |
323 | |
270 | ++enc->depth; |
324 | ++enc->depth; |
271 | |
325 | |
272 | encode_uint (enc, 0x80, len + 1); |
326 | encode_uint (enc, MAJOR_ARRAY, len + 1); |
273 | |
327 | |
274 | for (i = 0; i <= len; ++i) |
328 | for (i = 0; i <= len; ++i) |
275 | { |
329 | { |
276 | SV **svp = av_fetch (av, i, 0); |
330 | SV **svp = av_fetch (av, i, 0); |
277 | encode_sv (enc, svp ? *svp : &PL_sv_undef); |
331 | encode_sv (enc, svp ? *svp : &PL_sv_undef); |
… | |
… | |
292 | |
346 | |
293 | int pairs = hv_iterinit (hv); |
347 | int pairs = hv_iterinit (hv); |
294 | int mg = SvMAGICAL (hv); |
348 | int mg = SvMAGICAL (hv); |
295 | |
349 | |
296 | if (mg) |
350 | if (mg) |
297 | encode_ch (enc, 0xa0 | 31); |
351 | encode_ch (enc, MAJOR_MAP | MINOR_INDEF); |
298 | else |
352 | else |
299 | encode_uint (enc, 0xa0, pairs); |
353 | encode_uint (enc, MAJOR_MAP, pairs); |
300 | |
354 | |
301 | while ((he = hv_iternext (hv))) |
355 | while ((he = hv_iternext (hv))) |
302 | { |
356 | { |
303 | if (HeKLEN (he) == HEf_SVKEY) |
357 | if (HeKLEN (he) == HEf_SVKEY) |
304 | encode_sv (enc, HeSVKEY (he)); |
358 | encode_sv (enc, HeSVKEY (he)); |
305 | else |
359 | else |
306 | encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); |
360 | encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); |
307 | |
361 | |
308 | encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); |
362 | encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); |
309 | } |
363 | } |
310 | |
364 | |
311 | if (mg) |
365 | if (mg) |
312 | encode_ch (enc, 0xe0 | 31); |
366 | encode_ch (enc, MAJOR_MISC | MINOR_INDEF); |
313 | |
367 | |
314 | --enc->depth; |
368 | --enc->depth; |
315 | } |
369 | } |
316 | |
370 | |
317 | // encode objects, arrays and special \0=false and \1=true values. |
371 | // encode objects, arrays and special \0=false and \1=true values. |
318 | static void |
372 | static void |
319 | encode_rv (enc_t *enc, SV *sv) |
373 | encode_rv (enc_t *enc, SV *sv) |
320 | { |
374 | { |
321 | SvGETMAGIC (sv); |
375 | SvGETMAGIC (sv); |
322 | |
|
|
323 | if (ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING) |
|
|
324 | && ecb_expect_false (SvREFCNT (sv) > 1)) |
|
|
325 | { |
|
|
326 | if (!enc->shareable) |
|
|
327 | enc->shareable = (HV *)sv_2mortal ((SV *)newHV ()); |
|
|
328 | |
|
|
329 | SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1); |
|
|
330 | |
|
|
331 | if (SvOK (*svp)) |
|
|
332 | { |
|
|
333 | encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF); |
|
|
334 | encode_uint (enc, 0x00, SvUV (*svp)); |
|
|
335 | return; |
|
|
336 | } |
|
|
337 | else |
|
|
338 | { |
|
|
339 | sv_setuv (*svp, enc->shareable_idx); |
|
|
340 | ++enc->shareable_idx; |
|
|
341 | encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE); |
|
|
342 | } |
|
|
343 | } |
|
|
344 | |
376 | |
345 | svtype svt = SvTYPE (sv); |
377 | svtype svt = SvTYPE (sv); |
346 | |
378 | |
347 | if (ecb_expect_false (SvOBJECT (sv))) |
379 | if (ecb_expect_false (SvOBJECT (sv))) |
348 | { |
380 | { |
… | |
… | |
355 | HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash |
387 | HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash |
356 | ? cbor_tagged_stash |
388 | ? cbor_tagged_stash |
357 | : gv_stashpv ("CBOR::XS::Tagged" , 1); |
389 | : gv_stashpv ("CBOR::XS::Tagged" , 1); |
358 | |
390 | |
359 | HV *stash = SvSTASH (sv); |
391 | HV *stash = SvSTASH (sv); |
360 | GV *method; |
|
|
361 | |
392 | |
362 | if (stash == boolean_stash) |
393 | if (stash == boolean_stash) |
363 | encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); |
394 | { |
|
|
395 | encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE); |
|
|
396 | return; |
|
|
397 | } |
364 | else if (stash == error_stash) |
398 | else if (stash == error_stash) |
365 | encode_ch (enc, 0xe0 | 23); |
399 | { |
|
|
400 | encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF); |
|
|
401 | return; |
|
|
402 | } |
366 | else if (stash == tagged_stash) |
403 | else if (stash == tagged_stash) |
367 | { |
404 | { |
368 | if (svt != SVt_PVAV) |
405 | if (svt != SVt_PVAV) |
369 | croak ("encountered CBOR::XS::Tagged object that isn't an array"); |
406 | croak ("encountered CBOR::XS::Tagged object that isn't an array"); |
370 | |
407 | |
371 | encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); |
408 | encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1))); |
372 | encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); |
409 | encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); |
|
|
410 | |
|
|
411 | return; |
|
|
412 | } |
|
|
413 | } |
|
|
414 | |
|
|
415 | if (ecb_expect_false (SvREFCNT (sv) > 1) |
|
|
416 | && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)) |
|
|
417 | { |
|
|
418 | if (!enc->shareable) |
|
|
419 | enc->shareable = (HV *)sv_2mortal ((SV *)newHV ()); |
|
|
420 | |
|
|
421 | SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1); |
|
|
422 | |
|
|
423 | if (SvOK (*svp)) |
373 | } |
424 | { |
|
|
425 | encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF); |
|
|
426 | encode_uint (enc, MAJOR_POS_INT, SvUV (*svp)); |
|
|
427 | return; |
|
|
428 | } |
|
|
429 | else |
|
|
430 | { |
|
|
431 | sv_setuv (*svp, enc->shareable_idx); |
|
|
432 | ++enc->shareable_idx; |
|
|
433 | encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE); |
|
|
434 | } |
|
|
435 | } |
|
|
436 | |
|
|
437 | if (ecb_expect_false (SvOBJECT (sv))) |
|
|
438 | { |
|
|
439 | HV *stash = SvSTASH (sv); |
|
|
440 | GV *method; |
|
|
441 | |
374 | else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) |
442 | if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) |
375 | { |
443 | { |
376 | dSP; |
444 | dSP; |
377 | |
445 | |
378 | ENTER; SAVETMPS; PUSHMARK (SP); |
446 | ENTER; SAVETMPS; PUSHMARK (SP); |
379 | // we re-bless the reference to get overload and other niceties right |
447 | // we re-bless the reference to get overload and other niceties right |
… | |
… | |
411 | // catch this surprisingly common error |
479 | // catch this surprisingly common error |
412 | if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) |
480 | if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) |
413 | croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); |
481 | croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); |
414 | |
482 | |
415 | encode_tag (enc, CBOR_TAG_PERL_OBJECT); |
483 | encode_tag (enc, CBOR_TAG_PERL_OBJECT); |
416 | encode_uint (enc, 0x80, count + 1); |
484 | encode_uint (enc, MAJOR_ARRAY, count + 1); |
417 | encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); |
485 | encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); |
418 | |
486 | |
419 | while (count) |
487 | while (count) |
420 | encode_sv (enc, SP[1 - count--]); |
488 | encode_sv (enc, SP[1 - count--]); |
421 | |
489 | |
422 | PUTBACK; |
490 | PUTBACK; |
… | |
… | |
443 | { |
511 | { |
444 | double nv = SvNVX (sv); |
512 | double nv = SvNVX (sv); |
445 | |
513 | |
446 | need (enc, 9); |
514 | need (enc, 9); |
447 | |
515 | |
448 | if (ecb_expect_false (nv == (U32)nv)) |
516 | if (ecb_expect_false (nv == (NV)(U32)nv)) |
449 | encode_uint (enc, 0x00, (U32)nv); |
517 | encode_uint (enc, MAJOR_POS_INT, (U32)nv); |
450 | //TODO: maybe I32? |
518 | //TODO: maybe I32? |
451 | else if (ecb_expect_false (nv == (float)nv)) |
519 | else if (ecb_expect_false (nv == (float)nv)) |
452 | { |
520 | { |
453 | uint32_t fp = ecb_float_to_binary32 (nv); |
521 | uint32_t fp = ecb_float_to_binary32 (nv); |
454 | |
522 | |
455 | *enc->cur++ = 0xe0 | 26; |
523 | *enc->cur++ = MAJOR_MISC | MISC_FLOAT32; |
456 | |
524 | |
457 | if (!ecb_big_endian ()) |
525 | if (!ecb_big_endian ()) |
458 | fp = ecb_bswap32 (fp); |
526 | fp = ecb_bswap32 (fp); |
459 | |
527 | |
460 | memcpy (enc->cur, &fp, 4); |
528 | memcpy (enc->cur, &fp, 4); |
… | |
… | |
462 | } |
530 | } |
463 | else |
531 | else |
464 | { |
532 | { |
465 | uint64_t fp = ecb_double_to_binary64 (nv); |
533 | uint64_t fp = ecb_double_to_binary64 (nv); |
466 | |
534 | |
467 | *enc->cur++ = 0xe0 | 27; |
535 | *enc->cur++ = MAJOR_MISC | MISC_FLOAT64; |
468 | |
536 | |
469 | if (!ecb_big_endian ()) |
537 | if (!ecb_big_endian ()) |
470 | fp = ecb_bswap64 (fp); |
538 | fp = ecb_bswap64 (fp); |
471 | |
539 | |
472 | memcpy (enc->cur, &fp, 8); |
540 | memcpy (enc->cur, &fp, 8); |
… | |
… | |
481 | |
549 | |
482 | if (SvPOKp (sv)) |
550 | if (SvPOKp (sv)) |
483 | { |
551 | { |
484 | STRLEN len; |
552 | STRLEN len; |
485 | char *str = SvPV (sv, len); |
553 | char *str = SvPV (sv, len); |
486 | encode_str (enc, SvUTF8 (sv), str, len); |
554 | encode_strref (enc, SvUTF8 (sv), str, len); |
487 | } |
555 | } |
488 | else if (SvNOKp (sv)) |
556 | else if (SvNOKp (sv)) |
489 | encode_nv (enc, sv); |
557 | encode_nv (enc, sv); |
490 | else if (SvIOKp (sv)) |
558 | else if (SvIOKp (sv)) |
491 | { |
559 | { |
492 | if (SvIsUV (sv)) |
560 | if (SvIsUV (sv)) |
493 | encode_uint (enc, 0x00, SvUVX (sv)); |
561 | encode_uint (enc, MAJOR_POS_INT, SvUVX (sv)); |
494 | else if (SvIVX (sv) >= 0) |
562 | else if (SvIVX (sv) >= 0) |
495 | encode_uint (enc, 0x00, SvIVX (sv)); |
563 | encode_uint (enc, MAJOR_POS_INT, SvIVX (sv)); |
496 | else |
564 | else |
497 | encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); |
565 | encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1)); |
498 | } |
566 | } |
499 | else if (SvROK (sv)) |
567 | else if (SvROK (sv)) |
500 | encode_rv (enc, SvRV (sv)); |
568 | encode_rv (enc, SvRV (sv)); |
501 | else if (!SvOK (sv)) |
569 | else if (!SvOK (sv)) |
502 | encode_ch (enc, 0xe0 | 22); |
570 | encode_ch (enc, MAJOR_MISC | SIMPLE_NULL); |
503 | else if (enc->cbor.flags & F_ALLOW_UNKNOWN) |
571 | else if (enc->cbor.flags & F_ALLOW_UNKNOWN) |
504 | encode_ch (enc, 0xe0 | 23); |
572 | encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF); |
505 | else |
573 | else |
506 | croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", |
574 | croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", |
507 | SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); |
575 | SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); |
508 | } |
576 | } |
509 | |
577 | |
… | |
… | |
517 | enc.cur = SvPVX (enc.sv); |
585 | enc.cur = SvPVX (enc.sv); |
518 | enc.end = SvEND (enc.sv); |
586 | enc.end = SvEND (enc.sv); |
519 | |
587 | |
520 | SvPOK_only (enc.sv); |
588 | SvPOK_only (enc.sv); |
521 | |
589 | |
522 | if (cbor->flags & F_ALLOW_STRINGREF) |
590 | if (cbor->flags & F_PACK_STRINGS) |
523 | { |
591 | { |
524 | encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE); |
592 | encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE); |
525 | enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ()); |
593 | enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ()); |
526 | enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ()); |
594 | enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ()); |
527 | } |
595 | } |
… | |
… | |
562 | #define DEC_DEC_DEPTH --dec->depth |
630 | #define DEC_DEC_DEPTH --dec->depth |
563 | |
631 | |
564 | static UV |
632 | static UV |
565 | decode_uint (dec_t *dec) |
633 | decode_uint (dec_t *dec) |
566 | { |
634 | { |
567 | switch (*dec->cur & 31) |
635 | U8 m = *dec->cur & MINOR_MASK; |
568 | { |
636 | ++dec->cur; |
569 | case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7: |
|
|
570 | case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15: |
|
|
571 | case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23: |
|
|
572 | return *dec->cur++ & 31; |
|
|
573 | |
637 | |
574 | case 24: |
638 | if (ecb_expect_true (m < LENGTH_EXT1)) |
|
|
639 | return m; |
|
|
640 | else if (ecb_expect_true (m == LENGTH_EXT1)) |
|
|
641 | { |
575 | WANT (2); |
642 | WANT (1); |
576 | dec->cur += 2; |
643 | dec->cur += 1; |
577 | return dec->cur[-1]; |
644 | return dec->cur[-1]; |
578 | |
645 | } |
579 | case 25: |
646 | else if (ecb_expect_true (m == LENGTH_EXT2)) |
|
|
647 | { |
580 | WANT (3); |
648 | WANT (2); |
581 | dec->cur += 3; |
649 | dec->cur += 2; |
582 | return (((UV)dec->cur[-2]) << 8) |
650 | return (((UV)dec->cur[-2]) << 8) |
583 | | ((UV)dec->cur[-1]); |
651 | | ((UV)dec->cur[-1]); |
584 | |
652 | } |
585 | case 26: |
653 | else if (ecb_expect_true (m == LENGTH_EXT4)) |
|
|
654 | { |
586 | WANT (5); |
655 | WANT (4); |
587 | dec->cur += 5; |
656 | dec->cur += 4; |
588 | return (((UV)dec->cur[-4]) << 24) |
657 | return (((UV)dec->cur[-4]) << 24) |
589 | | (((UV)dec->cur[-3]) << 16) |
658 | | (((UV)dec->cur[-3]) << 16) |
590 | | (((UV)dec->cur[-2]) << 8) |
659 | | (((UV)dec->cur[-2]) << 8) |
591 | | ((UV)dec->cur[-1]); |
660 | | ((UV)dec->cur[-1]); |
592 | |
661 | } |
593 | case 27: |
662 | else if (ecb_expect_true (m == LENGTH_EXT8)) |
|
|
663 | { |
594 | WANT (9); |
664 | WANT (8); |
595 | dec->cur += 9; |
665 | dec->cur += 8; |
|
|
666 | |
|
|
667 | return |
|
|
668 | #if UVSIZE < 8 |
|
|
669 | 0 |
|
|
670 | #else |
596 | return (((UV)dec->cur[-8]) << 56) |
671 | (((UV)dec->cur[-8]) << 56) |
597 | | (((UV)dec->cur[-7]) << 48) |
672 | | (((UV)dec->cur[-7]) << 48) |
598 | | (((UV)dec->cur[-6]) << 40) |
673 | | (((UV)dec->cur[-6]) << 40) |
599 | | (((UV)dec->cur[-5]) << 32) |
674 | | (((UV)dec->cur[-5]) << 32) |
|
|
675 | #endif |
600 | | (((UV)dec->cur[-4]) << 24) |
676 | | (((UV)dec->cur[-4]) << 24) |
601 | | (((UV)dec->cur[-3]) << 16) |
677 | | (((UV)dec->cur[-3]) << 16) |
602 | | (((UV)dec->cur[-2]) << 8) |
678 | | (((UV)dec->cur[-2]) << 8) |
603 | | ((UV)dec->cur[-1]); |
679 | | ((UV)dec->cur[-1]); |
604 | |
680 | } |
605 | default: |
681 | else |
606 | ERR ("corrupted CBOR data (unsupported integer minor encoding)"); |
682 | ERR ("corrupted CBOR data (unsupported integer minor encoding)"); |
607 | } |
|
|
608 | |
683 | |
609 | fail: |
684 | fail: |
610 | return 0; |
685 | return 0; |
611 | } |
686 | } |
612 | |
687 | |
… | |
… | |
617 | { |
692 | { |
618 | AV *av = newAV (); |
693 | AV *av = newAV (); |
619 | |
694 | |
620 | DEC_INC_DEPTH; |
695 | DEC_INC_DEPTH; |
621 | |
696 | |
622 | if ((*dec->cur & 31) == 31) |
697 | if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF)) |
623 | { |
698 | { |
624 | ++dec->cur; |
699 | ++dec->cur; |
625 | |
700 | |
626 | for (;;) |
701 | for (;;) |
627 | { |
702 | { |
628 | WANT (1); |
703 | WANT (1); |
629 | |
704 | |
630 | if (*dec->cur == (0xe0 | 31)) |
705 | if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) |
631 | { |
706 | { |
632 | ++dec->cur; |
707 | ++dec->cur; |
633 | break; |
708 | break; |
634 | } |
709 | } |
635 | |
710 | |
… | |
… | |
638 | } |
713 | } |
639 | else |
714 | else |
640 | { |
715 | { |
641 | int i, len = decode_uint (dec); |
716 | int i, len = decode_uint (dec); |
642 | |
717 | |
|
|
718 | WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays |
643 | av_fill (av, len - 1); |
719 | av_fill (av, len - 1); |
644 | |
720 | |
645 | for (i = 0; i < len; ++i) |
721 | for (i = 0; i < len; ++i) |
646 | AvARRAY (av)[i] = decode_sv (dec); |
722 | AvARRAY (av)[i] = decode_sv (dec); |
647 | } |
723 | } |
… | |
… | |
660 | { |
736 | { |
661 | // for speed reasons, we specialcase single-string |
737 | // for speed reasons, we specialcase single-string |
662 | // byte or utf-8 strings as keys, but only when !stringref |
738 | // byte or utf-8 strings as keys, but only when !stringref |
663 | |
739 | |
664 | if (ecb_expect_true (!dec->stringref)) |
740 | if (ecb_expect_true (!dec->stringref)) |
665 | if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27) |
741 | if (ecb_expect_true ((*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8)) |
666 | { |
742 | { |
667 | I32 len = decode_uint (dec); |
743 | I32 len = decode_uint (dec); |
668 | char *key = (char *)dec->cur; |
744 | char *key = (char *)dec->cur; |
669 | |
745 | |
670 | dec->cur += len; |
746 | dec->cur += len; |
671 | |
747 | |
672 | if (ecb_expect_false (dec->stringref)) |
|
|
673 | av_push (dec->stringref, newSVpvn (key, len)); |
|
|
674 | |
|
|
675 | hv_store (hv, key, len, decode_sv (dec), 0); |
748 | hv_store (hv, key, len, decode_sv (dec), 0); |
676 | |
749 | |
677 | return; |
750 | return; |
678 | } |
751 | } |
679 | else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27) |
752 | else if (ecb_expect_true ((*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8)) |
680 | { |
753 | { |
681 | I32 len = decode_uint (dec); |
754 | I32 len = decode_uint (dec); |
682 | char *key = (char *)dec->cur; |
755 | char *key = (char *)dec->cur; |
683 | |
756 | |
684 | dec->cur += len; |
757 | dec->cur += len; |
685 | |
758 | |
686 | if (ecb_expect_false (dec->stringref)) |
759 | if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8)) |
687 | av_push (dec->stringref, newSVpvn_utf8 (key, len, 1)); |
760 | if (!is_utf8_string (key, len)) |
|
|
761 | ERR ("corrupted CBOR data (invalid UTF-8 in map key)"); |
688 | |
762 | |
689 | hv_store (hv, key, -len, decode_sv (dec), 0); |
763 | hv_store (hv, key, -len, decode_sv (dec), 0); |
690 | |
764 | |
691 | return; |
765 | return; |
692 | } |
766 | } |
… | |
… | |
694 | SV *k = decode_sv (dec); |
768 | SV *k = decode_sv (dec); |
695 | SV *v = decode_sv (dec); |
769 | SV *v = decode_sv (dec); |
696 | |
770 | |
697 | hv_store_ent (hv, k, v, 0); |
771 | hv_store_ent (hv, k, v, 0); |
698 | SvREFCNT_dec (k); |
772 | SvREFCNT_dec (k); |
|
|
773 | |
|
|
774 | fail: |
|
|
775 | ; |
699 | } |
776 | } |
700 | |
777 | |
701 | static SV * |
778 | static SV * |
702 | decode_hv (dec_t *dec) |
779 | decode_hv (dec_t *dec) |
703 | { |
780 | { |
704 | HV *hv = newHV (); |
781 | HV *hv = newHV (); |
705 | |
782 | |
706 | DEC_INC_DEPTH; |
783 | DEC_INC_DEPTH; |
707 | |
784 | |
708 | if ((*dec->cur & 31) == 31) |
785 | if (*dec->cur == (MAJOR_MAP | MINOR_INDEF)) |
709 | { |
786 | { |
710 | ++dec->cur; |
787 | ++dec->cur; |
711 | |
788 | |
712 | for (;;) |
789 | for (;;) |
713 | { |
790 | { |
714 | WANT (1); |
791 | WANT (1); |
715 | |
792 | |
716 | if (*dec->cur == (0xe0 | 31)) |
793 | if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) |
717 | { |
794 | { |
718 | ++dec->cur; |
795 | ++dec->cur; |
719 | break; |
796 | break; |
720 | } |
797 | } |
721 | |
798 | |
… | |
… | |
742 | static SV * |
819 | static SV * |
743 | decode_str (dec_t *dec, int utf8) |
820 | decode_str (dec_t *dec, int utf8) |
744 | { |
821 | { |
745 | SV *sv = 0; |
822 | SV *sv = 0; |
746 | |
823 | |
747 | if ((*dec->cur & 31) == 31) |
824 | if ((*dec->cur & MINOR_MASK) == MINOR_INDEF) |
748 | { |
825 | { |
|
|
826 | // indefinite length strings |
749 | ++dec->cur; |
827 | ++dec->cur; |
750 | |
828 | |
|
|
829 | U8 major = *dec->cur & MAJOR_MISC; |
|
|
830 | |
751 | sv = newSVpvn ("", 0); |
831 | sv = newSVpvn ("", 0); |
752 | |
832 | |
753 | // not very fast, and certainly not robust against illegal input |
|
|
754 | for (;;) |
833 | for (;;) |
755 | { |
834 | { |
756 | WANT (1); |
835 | WANT (1); |
757 | |
836 | |
758 | if (*dec->cur == (0xe0 | 31)) |
837 | if ((*dec->cur - major) > LENGTH_EXT8) |
|
|
838 | if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) |
759 | { |
839 | { |
760 | ++dec->cur; |
840 | ++dec->cur; |
761 | break; |
841 | break; |
762 | } |
842 | } |
|
|
843 | else |
|
|
844 | ERR ("corrupted CBOR data (invalid chunks in indefinite length string)"); |
763 | |
845 | |
764 | sv_catsv (sv, decode_sv (dec)); |
846 | STRLEN len = decode_uint (dec); |
|
|
847 | |
|
|
848 | WANT (len); |
|
|
849 | sv_catpvn (sv, dec->cur, len); |
|
|
850 | dec->cur += len; |
765 | } |
851 | } |
766 | } |
852 | } |
767 | else |
853 | else |
768 | { |
854 | { |
769 | STRLEN len = decode_uint (dec); |
855 | STRLEN len = decode_uint (dec); |
… | |
… | |
776 | && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1)) |
862 | && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1)) |
777 | av_push (dec->stringref, SvREFCNT_inc_NN (sv)); |
863 | av_push (dec->stringref, SvREFCNT_inc_NN (sv)); |
778 | } |
864 | } |
779 | |
865 | |
780 | if (utf8) |
866 | if (utf8) |
|
|
867 | { |
|
|
868 | if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8)) |
|
|
869 | if (!is_utf8_string (SvPVX (sv), SvCUR (sv))) |
|
|
870 | ERR ("corrupted CBOR data (invalid UTF-8 in text string)"); |
|
|
871 | |
781 | SvUTF8_on (sv); |
872 | SvUTF8_on (sv); |
|
|
873 | } |
782 | |
874 | |
783 | return sv; |
875 | return sv; |
784 | |
876 | |
785 | fail: |
877 | fail: |
786 | SvREFCNT_dec (sv); |
878 | SvREFCNT_dec (sv); |
… | |
… | |
818 | } |
910 | } |
819 | break; |
911 | break; |
820 | |
912 | |
821 | case CBOR_TAG_STRINGREF: |
913 | case CBOR_TAG_STRINGREF: |
822 | { |
914 | { |
823 | if ((*dec->cur >> 5) != 0) |
915 | if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) |
824 | ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); |
916 | ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); |
825 | |
917 | |
826 | UV idx = decode_uint (dec); |
918 | UV idx = decode_uint (dec); |
827 | |
919 | |
828 | if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) |
920 | if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) |
… | |
… | |
835 | case CBOR_TAG_VALUE_SHAREABLE: |
927 | case CBOR_TAG_VALUE_SHAREABLE: |
836 | { |
928 | { |
837 | if (ecb_expect_false (!dec->shareable)) |
929 | if (ecb_expect_false (!dec->shareable)) |
838 | dec->shareable = (AV *)sv_2mortal ((SV *)newAV ()); |
930 | dec->shareable = (AV *)sv_2mortal ((SV *)newAV ()); |
839 | |
931 | |
|
|
932 | if (dec->cbor.flags & F_ALLOW_CYCLES) |
|
|
933 | { |
840 | sv = newSV (0); |
934 | sv = newSV (0); |
841 | av_push (dec->shareable, SvREFCNT_inc_NN (sv)); |
935 | av_push (dec->shareable, SvREFCNT_inc_NN (sv)); |
842 | |
936 | |
843 | SV *osv = decode_sv (dec); |
937 | SV *osv = decode_sv (dec); |
844 | sv_setsv (sv, osv); |
938 | sv_setsv (sv, osv); |
845 | SvREFCNT_dec_NN (osv); |
939 | SvREFCNT_dec_NN (osv); |
|
|
940 | } |
|
|
941 | else |
|
|
942 | { |
|
|
943 | av_push (dec->shareable, &PL_sv_undef); |
|
|
944 | int idx = AvFILLp (dec->shareable); |
|
|
945 | sv = decode_sv (dec); |
|
|
946 | av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv)); |
|
|
947 | } |
846 | } |
948 | } |
847 | break; |
949 | break; |
848 | |
950 | |
849 | case CBOR_TAG_VALUE_SHAREDREF: |
951 | case CBOR_TAG_VALUE_SHAREDREF: |
850 | { |
952 | { |
851 | if ((*dec->cur >> 5) != 0) |
953 | if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) |
852 | ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); |
954 | ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); |
853 | |
955 | |
854 | UV idx = decode_uint (dec); |
956 | UV idx = decode_uint (dec); |
855 | |
957 | |
856 | if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) |
958 | if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) |
857 | ERR ("corrupted CBOR data (sharedref index out of bounds)"); |
959 | ERR ("corrupted CBOR data (sharedref index out of bounds)"); |
858 | |
960 | |
859 | sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); |
961 | sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); |
|
|
962 | |
|
|
963 | if (sv == &PL_sv_undef) |
|
|
964 | ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled"); |
860 | } |
965 | } |
861 | break; |
966 | break; |
862 | |
967 | |
863 | case CBOR_TAG_PERL_OBJECT: |
968 | case CBOR_TAG_PERL_OBJECT: |
864 | { |
969 | { |
… | |
… | |
965 | static SV * |
1070 | static SV * |
966 | decode_sv (dec_t *dec) |
1071 | decode_sv (dec_t *dec) |
967 | { |
1072 | { |
968 | WANT (1); |
1073 | WANT (1); |
969 | |
1074 | |
970 | switch (*dec->cur >> 5) |
1075 | switch (*dec->cur >> MAJOR_SHIFT) |
971 | { |
1076 | { |
972 | case 0: // unsigned int |
1077 | case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec)); |
973 | return newSVuv (decode_uint (dec)); |
1078 | case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec)); |
974 | case 1: // negative int |
1079 | case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0); |
975 | return newSViv (-1 - (IV)decode_uint (dec)); |
1080 | case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1); |
976 | case 2: // octet string |
1081 | case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec); |
977 | return decode_str (dec, 0); |
1082 | case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec); |
978 | case 3: // utf-8 string |
1083 | case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec); |
979 | return decode_str (dec, 1); |
1084 | |
980 | case 4: // array |
1085 | case MAJOR_MISC >> MAJOR_SHIFT: |
981 | return decode_av (dec); |
|
|
982 | case 5: // map |
|
|
983 | return decode_hv (dec); |
|
|
984 | case 6: // tag |
|
|
985 | return decode_tagged (dec); |
|
|
986 | case 7: // misc |
|
|
987 | switch (*dec->cur++ & 31) |
1086 | switch (*dec->cur++ & MINOR_MASK) |
988 | { |
1087 | { |
989 | case 20: |
1088 | case SIMPLE_FALSE: |
990 | #if CBOR_SLOW |
1089 | #if CBOR_SLOW |
991 | types_false = get_bool ("Types::Serialiser::false"); |
1090 | types_false = get_bool ("Types::Serialiser::false"); |
992 | #endif |
1091 | #endif |
993 | return newSVsv (types_false); |
1092 | return newSVsv (types_false); |
994 | case 21: |
1093 | case SIMPLE_TRUE: |
995 | #if CBOR_SLOW |
1094 | #if CBOR_SLOW |
996 | types_true = get_bool ("Types::Serialiser::true"); |
1095 | types_true = get_bool ("Types::Serialiser::true"); |
997 | #endif |
1096 | #endif |
998 | return newSVsv (types_true); |
1097 | return newSVsv (types_true); |
999 | case 22: |
1098 | case SIMPLE_NULL: |
1000 | return newSVsv (&PL_sv_undef); |
1099 | return newSVsv (&PL_sv_undef); |
1001 | case 23: |
1100 | case SIMPLE_UNDEF: |
1002 | #if CBOR_SLOW |
1101 | #if CBOR_SLOW |
1003 | types_error = get_bool ("Types::Serialiser::error"); |
1102 | types_error = get_bool ("Types::Serialiser::error"); |
1004 | #endif |
1103 | #endif |
1005 | return newSVsv (types_error); |
1104 | return newSVsv (types_error); |
1006 | |
1105 | |
1007 | case 25: |
1106 | case MISC_FLOAT16: |
1008 | { |
1107 | { |
1009 | WANT (2); |
1108 | WANT (2); |
1010 | |
1109 | |
1011 | uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; |
1110 | uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; |
1012 | dec->cur += 2; |
1111 | dec->cur += 2; |
1013 | |
1112 | |
1014 | return newSVnv (ecb_binary16_to_float (fp)); |
1113 | return newSVnv (ecb_binary16_to_float (fp)); |
1015 | } |
1114 | } |
1016 | |
1115 | |
1017 | case 26: |
1116 | case MISC_FLOAT32: |
1018 | { |
1117 | { |
1019 | uint32_t fp; |
1118 | uint32_t fp; |
1020 | WANT (4); |
1119 | WANT (4); |
1021 | memcpy (&fp, dec->cur, 4); |
1120 | memcpy (&fp, dec->cur, 4); |
1022 | dec->cur += 4; |
1121 | dec->cur += 4; |
… | |
… | |
1025 | fp = ecb_bswap32 (fp); |
1124 | fp = ecb_bswap32 (fp); |
1026 | |
1125 | |
1027 | return newSVnv (ecb_binary32_to_float (fp)); |
1126 | return newSVnv (ecb_binary32_to_float (fp)); |
1028 | } |
1127 | } |
1029 | |
1128 | |
1030 | case 27: |
1129 | case MISC_FLOAT64: |
1031 | { |
1130 | { |
1032 | uint64_t fp; |
1131 | uint64_t fp; |
1033 | WANT (8); |
1132 | WANT (8); |
1034 | memcpy (&fp, dec->cur, 8); |
1133 | memcpy (&fp, dec->cur, 8); |
1035 | dec->cur += 8; |
1134 | dec->cur += 8; |
… | |
… | |
1038 | fp = ecb_bswap64 (fp); |
1137 | fp = ecb_bswap64 (fp); |
1039 | |
1138 | |
1040 | return newSVnv (ecb_binary64_to_double (fp)); |
1139 | return newSVnv (ecb_binary64_to_double (fp)); |
1041 | } |
1140 | } |
1042 | |
1141 | |
1043 | // 0..19 unassigned |
1142 | // 0..19 unassigned simple |
1044 | // 24 reserved + unassigned (reserved values are not encodable) |
1143 | // 24 reserved + unassigned simple (reserved values are not encodable) |
|
|
1144 | // 28-30 unassigned misc |
|
|
1145 | // 31 break code |
1045 | default: |
1146 | default: |
1046 | ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); |
1147 | ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)"); |
1047 | } |
1148 | } |
1048 | |
1149 | |
1049 | break; |
1150 | break; |
1050 | } |
1151 | } |
1051 | |
1152 | |
… | |
… | |
1078 | if (dec.cur != dec.end && !dec.err) |
1179 | if (dec.cur != dec.end && !dec.err) |
1079 | dec.err = "garbage after CBOR object"; |
1180 | dec.err = "garbage after CBOR object"; |
1080 | |
1181 | |
1081 | if (dec.err) |
1182 | if (dec.err) |
1082 | { |
1183 | { |
|
|
1184 | if (dec.shareable) |
|
|
1185 | { |
|
|
1186 | // need to break cyclic links, which whould all be in shareable |
|
|
1187 | int i; |
|
|
1188 | SV **svp; |
|
|
1189 | |
|
|
1190 | for (i = av_len (dec.shareable) + 1; i--; ) |
|
|
1191 | if ((svp = av_fetch (dec.shareable, i, 0))) |
|
|
1192 | sv_setsv (*svp, &PL_sv_undef); |
|
|
1193 | } |
|
|
1194 | |
1083 | SvREFCNT_dec (sv); |
1195 | SvREFCNT_dec (sv); |
1084 | croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); |
1196 | croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); |
1085 | } |
1197 | } |
1086 | |
1198 | |
1087 | sv = sv_2mortal (sv); |
1199 | sv = sv_2mortal (sv); |
1088 | |
1200 | |
1089 | return sv; |
1201 | return sv; |
1090 | } |
1202 | } |
1091 | |
1203 | |
|
|
1204 | ///////////////////////////////////////////////////////////////////////////// |
|
|
1205 | // incremental parser |
|
|
1206 | |
|
|
1207 | #define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0) |
|
|
1208 | |
|
|
1209 | // returns 0 for notyet, 1 for success or error |
|
|
1210 | static int |
|
|
1211 | incr_parse (CBOR *self, SV *cborstr) |
|
|
1212 | { |
|
|
1213 | STRLEN cur; |
|
|
1214 | SvPV (cborstr, cur); |
|
|
1215 | |
|
|
1216 | while (ecb_expect_true (self->incr_need <= cur)) |
|
|
1217 | { |
|
|
1218 | // table of integer count bytes |
|
|
1219 | static I8 incr_len[MINOR_MASK + 1] = { |
|
|
1220 | 0, 0, 0, 0, 0, 0, 0, 0, |
|
|
1221 | 0, 0, 0, 0, 0, 0, 0, 0, |
|
|
1222 | 0, 0, 0, 0, 0, 0, 0, 0, |
|
|
1223 | 1, 2, 4, 8,-1,-1,-1,-2 |
|
|
1224 | }; |
|
|
1225 | |
|
|
1226 | const U8 *p = SvPVX (cborstr) + self->incr_pos; |
|
|
1227 | U8 m = *p & MINOR_MASK; |
|
|
1228 | IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]); |
|
|
1229 | I8 ilen = incr_len[m]; |
|
|
1230 | |
|
|
1231 | self->incr_need = self->incr_pos + 1; |
|
|
1232 | |
|
|
1233 | if (ecb_expect_false (ilen < 0)) |
|
|
1234 | { |
|
|
1235 | if (m != MINOR_INDEF) |
|
|
1236 | return 1; // error |
|
|
1237 | |
|
|
1238 | if (*p == (MAJOR_MISC | MINOR_INDEF)) |
|
|
1239 | { |
|
|
1240 | if (count >= 0) |
|
|
1241 | return 1; // error |
|
|
1242 | |
|
|
1243 | count = 1; |
|
|
1244 | } |
|
|
1245 | else |
|
|
1246 | { |
|
|
1247 | av_push (self->incr_count, newSViv (-1)); //TODO: nest |
|
|
1248 | count = -1; |
|
|
1249 | } |
|
|
1250 | } |
|
|
1251 | else |
|
|
1252 | { |
|
|
1253 | self->incr_need += ilen; |
|
|
1254 | if (ecb_expect_false (self->incr_need > cur)) |
|
|
1255 | return 0; |
|
|
1256 | |
|
|
1257 | int major = *p >> MAJOR_SHIFT; |
|
|
1258 | |
|
|
1259 | switch (major) |
|
|
1260 | { |
|
|
1261 | case MAJOR_BYTES >> MAJOR_SHIFT: |
|
|
1262 | case MAJOR_TEXT >> MAJOR_SHIFT: |
|
|
1263 | case MAJOR_ARRAY >> MAJOR_SHIFT: |
|
|
1264 | case MAJOR_MAP >> MAJOR_SHIFT: |
|
|
1265 | { |
|
|
1266 | UV len; |
|
|
1267 | |
|
|
1268 | if (ecb_expect_false (ilen)) |
|
|
1269 | { |
|
|
1270 | len = 0; |
|
|
1271 | |
|
|
1272 | do { |
|
|
1273 | len = (len << 8) | *++p; |
|
|
1274 | } while (--ilen); |
|
|
1275 | } |
|
|
1276 | else |
|
|
1277 | len = m; |
|
|
1278 | |
|
|
1279 | switch (major) |
|
|
1280 | { |
|
|
1281 | case MAJOR_BYTES >> MAJOR_SHIFT: |
|
|
1282 | case MAJOR_TEXT >> MAJOR_SHIFT: |
|
|
1283 | self->incr_need += len; |
|
|
1284 | if (ecb_expect_false (self->incr_need > cur)) |
|
|
1285 | return 0; |
|
|
1286 | |
|
|
1287 | break; |
|
|
1288 | |
|
|
1289 | case MAJOR_MAP >> MAJOR_SHIFT: |
|
|
1290 | len <<= 1; |
|
|
1291 | case MAJOR_ARRAY >> MAJOR_SHIFT: |
|
|
1292 | if (len) |
|
|
1293 | { |
|
|
1294 | av_push (self->incr_count, newSViv (len + 1)); //TODO: nest |
|
|
1295 | count = len + 1; |
|
|
1296 | } |
|
|
1297 | break; |
|
|
1298 | } |
|
|
1299 | } |
|
|
1300 | } |
|
|
1301 | } |
|
|
1302 | |
|
|
1303 | self->incr_pos = self->incr_need; |
|
|
1304 | |
|
|
1305 | if (count > 0) |
|
|
1306 | { |
|
|
1307 | while (!--count) |
|
|
1308 | { |
|
|
1309 | if (!AvFILLp (self->incr_count)) |
|
|
1310 | return 1; // done |
|
|
1311 | |
|
|
1312 | SvREFCNT_dec_NN (av_pop (self->incr_count)); |
|
|
1313 | count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]); |
|
|
1314 | } |
|
|
1315 | |
|
|
1316 | SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count; |
|
|
1317 | } |
|
|
1318 | } |
|
|
1319 | |
|
|
1320 | return 0; |
|
|
1321 | } |
|
|
1322 | |
|
|
1323 | |
1092 | ///////////////////////////////////////////////////////////////////////////// |
1324 | ///////////////////////////////////////////////////////////////////////////// |
1093 | // XS interface functions |
1325 | // XS interface functions |
1094 | |
1326 | |
1095 | MODULE = CBOR::XS PACKAGE = CBOR::XS |
1327 | MODULE = CBOR::XS PACKAGE = CBOR::XS |
1096 | |
1328 | |
… | |
… | |
1136 | void shrink (CBOR *self, int enable = 1) |
1368 | void shrink (CBOR *self, int enable = 1) |
1137 | ALIAS: |
1369 | ALIAS: |
1138 | shrink = F_SHRINK |
1370 | shrink = F_SHRINK |
1139 | allow_unknown = F_ALLOW_UNKNOWN |
1371 | allow_unknown = F_ALLOW_UNKNOWN |
1140 | allow_sharing = F_ALLOW_SHARING |
1372 | allow_sharing = F_ALLOW_SHARING |
1141 | allow_stringref = F_ALLOW_STRINGREF |
1373 | allow_cycles = F_ALLOW_CYCLES |
|
|
1374 | pack_strings = F_PACK_STRINGS |
|
|
1375 | validate_utf8 = F_VALIDATE_UTF8 |
1142 | PPCODE: |
1376 | PPCODE: |
1143 | { |
1377 | { |
1144 | if (enable) |
1378 | if (enable) |
1145 | self->flags |= ix; |
1379 | self->flags |= ix; |
1146 | else |
1380 | else |
… | |
… | |
1152 | void get_shrink (CBOR *self) |
1386 | void get_shrink (CBOR *self) |
1153 | ALIAS: |
1387 | ALIAS: |
1154 | get_shrink = F_SHRINK |
1388 | get_shrink = F_SHRINK |
1155 | get_allow_unknown = F_ALLOW_UNKNOWN |
1389 | get_allow_unknown = F_ALLOW_UNKNOWN |
1156 | get_allow_sharing = F_ALLOW_SHARING |
1390 | get_allow_sharing = F_ALLOW_SHARING |
1157 | get_allow_stringref = F_ALLOW_STRINGREF |
1391 | get_allow_cycles = F_ALLOW_CYCLES |
|
|
1392 | get_pack_strings = F_PACK_STRINGS |
|
|
1393 | get_validate_utf8 = F_VALIDATE_UTF8 |
1158 | PPCODE: |
1394 | PPCODE: |
1159 | XPUSHs (boolSV (self->flags & ix)); |
1395 | XPUSHs (boolSV (self->flags & ix)); |
1160 | |
1396 | |
1161 | void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) |
1397 | void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) |
1162 | PPCODE: |
1398 | PPCODE: |
… | |
… | |
1211 | EXTEND (SP, 2); |
1447 | EXTEND (SP, 2); |
1212 | PUSHs (sv); |
1448 | PUSHs (sv); |
1213 | PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); |
1449 | PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); |
1214 | } |
1450 | } |
1215 | |
1451 | |
|
|
1452 | void incr_parse (CBOR *self, SV *cborstr) |
|
|
1453 | PPCODE: |
|
|
1454 | { |
|
|
1455 | if (SvUTF8 (cborstr)) |
|
|
1456 | sv_utf8_downgrade (cborstr, 0); |
|
|
1457 | |
|
|
1458 | if (!self->incr_count) |
|
|
1459 | { |
|
|
1460 | self->incr_count = newAV (); |
|
|
1461 | self->incr_pos = 0; |
|
|
1462 | self->incr_need = 1; |
|
|
1463 | |
|
|
1464 | av_push (self->incr_count, newSViv (1)); |
|
|
1465 | } |
|
|
1466 | |
|
|
1467 | do |
|
|
1468 | { |
|
|
1469 | if (!incr_parse (self, cborstr)) |
|
|
1470 | { |
|
|
1471 | if (self->incr_need > self->max_size && self->max_size) |
|
|
1472 | croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu", |
|
|
1473 | (unsigned long)self->incr_need, (unsigned long)self->max_size); |
|
|
1474 | |
|
|
1475 | break; |
|
|
1476 | } |
|
|
1477 | |
|
|
1478 | SV *sv; |
|
|
1479 | char *offset; |
|
|
1480 | |
|
|
1481 | PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN; |
|
|
1482 | XPUSHs (sv); |
|
|
1483 | |
|
|
1484 | sv_chop (cborstr, offset); |
|
|
1485 | |
|
|
1486 | av_clear (self->incr_count); |
|
|
1487 | av_push (self->incr_count, newSViv (1)); |
|
|
1488 | |
|
|
1489 | self->incr_pos = 0; |
|
|
1490 | self->incr_need = self->incr_pos + 1; |
|
|
1491 | } |
|
|
1492 | while (GIMME_V == G_ARRAY); |
|
|
1493 | } |
|
|
1494 | |
|
|
1495 | void incr_reset (CBOR *self) |
|
|
1496 | CODE: |
|
|
1497 | { |
|
|
1498 | SvREFCNT_dec (self->incr_count); |
|
|
1499 | self->incr_count = 0; |
|
|
1500 | } |
|
|
1501 | |
1216 | void DESTROY (CBOR *self) |
1502 | void DESTROY (CBOR *self) |
1217 | PPCODE: |
1503 | PPCODE: |
1218 | cbor_free (self); |
1504 | cbor_free (self); |
1219 | |
1505 | |
1220 | PROTOTYPES: ENABLE |
1506 | PROTOTYPES: ENABLE |
1221 | |
1507 | |
1222 | void encode_cbor (SV *scalar) |
1508 | void encode_cbor (SV *scalar) |
|
|
1509 | ALIAS: |
|
|
1510 | encode_cbor = 0 |
|
|
1511 | encode_cbor_sharing = F_ALLOW_SHARING |
1223 | PPCODE: |
1512 | PPCODE: |
1224 | { |
1513 | { |
1225 | CBOR cbor; |
1514 | CBOR cbor; |
1226 | cbor_init (&cbor); |
1515 | cbor_init (&cbor); |
|
|
1516 | cbor.flags |= ix; |
1227 | PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; |
1517 | PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; |
1228 | XPUSHs (scalar); |
1518 | XPUSHs (scalar); |
1229 | } |
1519 | } |
1230 | |
1520 | |
1231 | void decode_cbor (SV *cborstr) |
1521 | void decode_cbor (SV *cborstr) |