… | |
… | |
23 | #endif |
23 | #endif |
24 | #ifndef SvREFCNT_dec_NN |
24 | #ifndef SvREFCNT_dec_NN |
25 | # define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv) |
25 | # define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv) |
26 | #endif |
26 | #endif |
27 | |
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 | }; |
|
|
64 | |
28 | // known tags |
65 | // known tags |
29 | enum cbor_tag |
66 | enum cbor_tag |
30 | { |
67 | { |
31 | // extensions |
68 | // extensions |
32 | CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref |
69 | CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref |
33 | CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object |
70 | CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object |
34 | CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object |
71 | CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object |
35 | CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing |
72 | CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing |
36 | CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing |
73 | CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing |
37 | CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref |
74 | CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref |
38 | CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection |
75 | CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection |
39 | |
76 | |
40 | // rfc7049 |
77 | // rfc7049 |
41 | CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 |
78 | CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 |
42 | CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any |
79 | CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any |
43 | CBOR_TAG_POS_BIGNUM = 2, // byte string |
80 | CBOR_TAG_POS_BIGNUM = 2, // byte string |
44 | CBOR_TAG_NEG_BIGNUM = 3, // byte string |
81 | CBOR_TAG_NEG_BIGNUM = 3, // byte string |
45 | CBOR_TAG_DECIMAL = 4, // decimal fraction, array |
82 | CBOR_TAG_DECIMAL = 4, // decimal fraction, array |
46 | CBOR_TAG_BIGFLOAT = 5, // array |
83 | CBOR_TAG_BIGFLOAT = 5, // array |
47 | |
84 | |
48 | CBOR_TAG_CONV_B64U = 21, // base64url, any |
85 | CBOR_TAG_CONV_B64U = 21, // base64url, any |
49 | CBOR_TAG_CONV_B64 = 22, // base64, any |
86 | CBOR_TAG_CONV_B64 = 22, // base64, any |
50 | CBOR_TAG_CONV_HEX = 23, // base16, any |
87 | CBOR_TAG_CONV_HEX = 23, // base16, any |
51 | CBOR_TAG_CBOR = 24, // embedded cbor, byte string |
88 | CBOR_TAG_CBOR = 24, // embedded cbor, byte string |
52 | |
89 | |
53 | CBOR_TAG_URI = 32, // URI rfc3986, utf-8 |
90 | CBOR_TAG_URI = 32, // URI rfc3986, utf-8 |
54 | CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8 |
91 | CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8 |
55 | CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8 |
92 | CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8 |
56 | CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8 |
93 | CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8 |
57 | CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 |
94 | CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 |
58 | |
95 | |
59 | CBOR_TAG_MAGIC = 55799 // self-describe cbor |
96 | CBOR_TAG_MAGIC = 55799, // self-describe cbor |
60 | }; |
97 | }; |
61 | |
98 | |
62 | #define F_SHRINK 0x00000001UL |
99 | #define F_SHRINK 0x00000001UL |
63 | #define F_ALLOW_UNKNOWN 0x00000002UL |
100 | #define F_ALLOW_UNKNOWN 0x00000002UL |
64 | #define F_ALLOW_SHARING 0x00000004UL |
101 | #define F_ALLOW_SHARING 0x00000004UL |
… | |
… | |
189 | static void |
226 | static void |
190 | encode_uint (enc_t *enc, int major, UV len) |
227 | encode_uint (enc_t *enc, int major, UV len) |
191 | { |
228 | { |
192 | need (enc, 9); |
229 | need (enc, 9); |
193 | |
230 | |
194 | if (ecb_expect_true (len < 24)) |
231 | if (ecb_expect_true (len < LENGTH_EXT1)) |
195 | *enc->cur++ = major | len; |
232 | *enc->cur++ = major | len; |
196 | else if (ecb_expect_true (len <= 0xff)) |
233 | else if (ecb_expect_true (len <= 0xff)) |
197 | { |
234 | { |
198 | *enc->cur++ = major | 24; |
235 | *enc->cur++ = major | LENGTH_EXT1; |
199 | *enc->cur++ = len; |
236 | *enc->cur++ = len; |
200 | } |
237 | } |
201 | else if (len <= 0xffff) |
238 | else if (len <= 0xffff) |
202 | { |
239 | { |
203 | *enc->cur++ = major | 25; |
240 | *enc->cur++ = major | LENGTH_EXT2; |
204 | *enc->cur++ = len >> 8; |
241 | *enc->cur++ = len >> 8; |
205 | *enc->cur++ = len; |
242 | *enc->cur++ = len; |
206 | } |
243 | } |
207 | else if (len <= 0xffffffff) |
244 | else if (len <= 0xffffffff) |
208 | { |
245 | { |
209 | *enc->cur++ = major | 26; |
246 | *enc->cur++ = major | LENGTH_EXT4; |
210 | *enc->cur++ = len >> 24; |
247 | *enc->cur++ = len >> 24; |
211 | *enc->cur++ = len >> 16; |
248 | *enc->cur++ = len >> 16; |
212 | *enc->cur++ = len >> 8; |
249 | *enc->cur++ = len >> 8; |
213 | *enc->cur++ = len; |
250 | *enc->cur++ = len; |
214 | } |
251 | } |
215 | else |
252 | else |
216 | { |
253 | { |
217 | *enc->cur++ = major | 27; |
254 | *enc->cur++ = major | LENGTH_EXT8; |
218 | *enc->cur++ = len >> 56; |
255 | *enc->cur++ = len >> 56; |
219 | *enc->cur++ = len >> 48; |
256 | *enc->cur++ = len >> 48; |
220 | *enc->cur++ = len >> 40; |
257 | *enc->cur++ = len >> 40; |
221 | *enc->cur++ = len >> 32; |
258 | *enc->cur++ = len >> 32; |
222 | *enc->cur++ = len >> 24; |
259 | *enc->cur++ = len >> 24; |
… | |
… | |
227 | } |
264 | } |
228 | |
265 | |
229 | ecb_inline void |
266 | ecb_inline void |
230 | encode_tag (enc_t *enc, UV tag) |
267 | encode_tag (enc_t *enc, UV tag) |
231 | { |
268 | { |
232 | encode_uint (enc, 0xc0, tag); |
269 | encode_uint (enc, MAJOR_TAG, tag); |
233 | } |
270 | } |
234 | |
271 | |
235 | ecb_inline void |
272 | ecb_inline void |
236 | encode_str (enc_t *enc, int utf8, char *str, STRLEN len) |
273 | encode_str (enc_t *enc, int utf8, char *str, STRLEN len) |
237 | { |
274 | { |
238 | encode_uint (enc, utf8 ? 0x60 : 0x40, len); |
275 | encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len); |
239 | need (enc, len); |
276 | need (enc, len); |
240 | memcpy (enc->cur, str, len); |
277 | memcpy (enc->cur, str, len); |
241 | enc->cur += len; |
278 | enc->cur += len; |
242 | } |
279 | } |
243 | |
280 | |
… | |
… | |
250 | |
287 | |
251 | if (SvOK (*svp)) |
288 | if (SvOK (*svp)) |
252 | { |
289 | { |
253 | // already registered, use stringref |
290 | // already registered, use stringref |
254 | encode_tag (enc, CBOR_TAG_STRINGREF); |
291 | encode_tag (enc, CBOR_TAG_STRINGREF); |
255 | encode_uint (enc, 0x00, SvUV (*svp)); |
292 | encode_uint (enc, MAJOR_POS_INT, SvUV (*svp)); |
256 | return; |
293 | return; |
257 | } |
294 | } |
258 | else if (len >= minimum_string_length (enc->stringref_idx)) |
295 | else if (len >= minimum_string_length (enc->stringref_idx)) |
259 | { |
296 | { |
260 | // register only |
297 | // register only |
… | |
… | |
276 | if (enc->depth >= enc->cbor.max_depth) |
313 | if (enc->depth >= enc->cbor.max_depth) |
277 | croak (ERR_NESTING_EXCEEDED); |
314 | croak (ERR_NESTING_EXCEEDED); |
278 | |
315 | |
279 | ++enc->depth; |
316 | ++enc->depth; |
280 | |
317 | |
281 | encode_uint (enc, 0x80, len + 1); |
318 | encode_uint (enc, MAJOR_ARRAY, len + 1); |
282 | |
319 | |
283 | for (i = 0; i <= len; ++i) |
320 | for (i = 0; i <= len; ++i) |
284 | { |
321 | { |
285 | SV **svp = av_fetch (av, i, 0); |
322 | SV **svp = av_fetch (av, i, 0); |
286 | encode_sv (enc, svp ? *svp : &PL_sv_undef); |
323 | encode_sv (enc, svp ? *svp : &PL_sv_undef); |
… | |
… | |
301 | |
338 | |
302 | int pairs = hv_iterinit (hv); |
339 | int pairs = hv_iterinit (hv); |
303 | int mg = SvMAGICAL (hv); |
340 | int mg = SvMAGICAL (hv); |
304 | |
341 | |
305 | if (mg) |
342 | if (mg) |
306 | encode_ch (enc, 0xa0 | 31); |
343 | encode_ch (enc, MAJOR_MAP | MINOR_INDEF); |
307 | else |
344 | else |
308 | encode_uint (enc, 0xa0, pairs); |
345 | encode_uint (enc, MAJOR_MAP, pairs); |
309 | |
346 | |
310 | while ((he = hv_iternext (hv))) |
347 | while ((he = hv_iternext (hv))) |
311 | { |
348 | { |
312 | if (HeKLEN (he) == HEf_SVKEY) |
349 | if (HeKLEN (he) == HEf_SVKEY) |
313 | encode_sv (enc, HeSVKEY (he)); |
350 | encode_sv (enc, HeSVKEY (he)); |
… | |
… | |
316 | |
353 | |
317 | encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); |
354 | encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); |
318 | } |
355 | } |
319 | |
356 | |
320 | if (mg) |
357 | if (mg) |
321 | encode_ch (enc, 0xe0 | 31); |
358 | encode_ch (enc, MAJOR_MISC | MINOR_INDEF); |
322 | |
359 | |
323 | --enc->depth; |
360 | --enc->depth; |
324 | } |
361 | } |
325 | |
362 | |
326 | // encode objects, arrays and special \0=false and \1=true values. |
363 | // encode objects, arrays and special \0=false and \1=true values. |
… | |
… | |
345 | |
382 | |
346 | HV *stash = SvSTASH (sv); |
383 | HV *stash = SvSTASH (sv); |
347 | |
384 | |
348 | if (stash == boolean_stash) |
385 | if (stash == boolean_stash) |
349 | { |
386 | { |
350 | encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); |
387 | encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE); |
351 | return; |
388 | return; |
352 | } |
389 | } |
353 | else if (stash == error_stash) |
390 | else if (stash == error_stash) |
354 | { |
391 | { |
355 | encode_ch (enc, 0xe0 | 23); |
392 | encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF); |
356 | return; |
393 | return; |
357 | } |
394 | } |
358 | else if (stash == tagged_stash) |
395 | else if (stash == tagged_stash) |
359 | { |
396 | { |
360 | if (svt != SVt_PVAV) |
397 | if (svt != SVt_PVAV) |
361 | croak ("encountered CBOR::XS::Tagged object that isn't an array"); |
398 | croak ("encountered CBOR::XS::Tagged object that isn't an array"); |
362 | |
399 | |
363 | encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); |
400 | encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1))); |
364 | encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); |
401 | encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); |
365 | |
402 | |
366 | return; |
403 | return; |
367 | } |
404 | } |
368 | } |
405 | } |
… | |
… | |
376 | SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1); |
413 | SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1); |
377 | |
414 | |
378 | if (SvOK (*svp)) |
415 | if (SvOK (*svp)) |
379 | { |
416 | { |
380 | encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF); |
417 | encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF); |
381 | encode_uint (enc, 0x00, SvUV (*svp)); |
418 | encode_uint (enc, MAJOR_POS_INT, SvUV (*svp)); |
382 | return; |
419 | return; |
383 | } |
420 | } |
384 | else |
421 | else |
385 | { |
422 | { |
386 | sv_setuv (*svp, enc->shareable_idx); |
423 | sv_setuv (*svp, enc->shareable_idx); |
… | |
… | |
434 | // catch this surprisingly common error |
471 | // catch this surprisingly common error |
435 | if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) |
472 | if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) |
436 | croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); |
473 | croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); |
437 | |
474 | |
438 | encode_tag (enc, CBOR_TAG_PERL_OBJECT); |
475 | encode_tag (enc, CBOR_TAG_PERL_OBJECT); |
439 | encode_uint (enc, 0x80, count + 1); |
476 | encode_uint (enc, MAJOR_ARRAY, count + 1); |
440 | encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); |
477 | encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); |
441 | |
478 | |
442 | while (count) |
479 | while (count) |
443 | encode_sv (enc, SP[1 - count--]); |
480 | encode_sv (enc, SP[1 - count--]); |
444 | |
481 | |
… | |
… | |
466 | { |
503 | { |
467 | double nv = SvNVX (sv); |
504 | double nv = SvNVX (sv); |
468 | |
505 | |
469 | need (enc, 9); |
506 | need (enc, 9); |
470 | |
507 | |
471 | if (ecb_expect_false (nv == (U32)nv)) |
508 | if (ecb_expect_false (nv == (NV)(U32)nv)) |
472 | encode_uint (enc, 0x00, (U32)nv); |
509 | encode_uint (enc, MAJOR_POS_INT, (U32)nv); |
473 | //TODO: maybe I32? |
510 | //TODO: maybe I32? |
474 | else if (ecb_expect_false (nv == (float)nv)) |
511 | else if (ecb_expect_false (nv == (float)nv)) |
475 | { |
512 | { |
476 | uint32_t fp = ecb_float_to_binary32 (nv); |
513 | uint32_t fp = ecb_float_to_binary32 (nv); |
477 | |
514 | |
478 | *enc->cur++ = 0xe0 | 26; |
515 | *enc->cur++ = MAJOR_MISC | MISC_FLOAT32; |
479 | |
516 | |
480 | if (!ecb_big_endian ()) |
517 | if (!ecb_big_endian ()) |
481 | fp = ecb_bswap32 (fp); |
518 | fp = ecb_bswap32 (fp); |
482 | |
519 | |
483 | memcpy (enc->cur, &fp, 4); |
520 | memcpy (enc->cur, &fp, 4); |
… | |
… | |
485 | } |
522 | } |
486 | else |
523 | else |
487 | { |
524 | { |
488 | uint64_t fp = ecb_double_to_binary64 (nv); |
525 | uint64_t fp = ecb_double_to_binary64 (nv); |
489 | |
526 | |
490 | *enc->cur++ = 0xe0 | 27; |
527 | *enc->cur++ = MAJOR_MISC | MISC_FLOAT64; |
491 | |
528 | |
492 | if (!ecb_big_endian ()) |
529 | if (!ecb_big_endian ()) |
493 | fp = ecb_bswap64 (fp); |
530 | fp = ecb_bswap64 (fp); |
494 | |
531 | |
495 | memcpy (enc->cur, &fp, 8); |
532 | memcpy (enc->cur, &fp, 8); |
… | |
… | |
511 | else if (SvNOKp (sv)) |
548 | else if (SvNOKp (sv)) |
512 | encode_nv (enc, sv); |
549 | encode_nv (enc, sv); |
513 | else if (SvIOKp (sv)) |
550 | else if (SvIOKp (sv)) |
514 | { |
551 | { |
515 | if (SvIsUV (sv)) |
552 | if (SvIsUV (sv)) |
516 | encode_uint (enc, 0x00, SvUVX (sv)); |
553 | encode_uint (enc, MAJOR_POS_INT, SvUVX (sv)); |
517 | else if (SvIVX (sv) >= 0) |
554 | else if (SvIVX (sv) >= 0) |
518 | encode_uint (enc, 0x00, SvIVX (sv)); |
555 | encode_uint (enc, MAJOR_POS_INT, SvIVX (sv)); |
519 | else |
556 | else |
520 | encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); |
557 | encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1)); |
521 | } |
558 | } |
522 | else if (SvROK (sv)) |
559 | else if (SvROK (sv)) |
523 | encode_rv (enc, SvRV (sv)); |
560 | encode_rv (enc, SvRV (sv)); |
524 | else if (!SvOK (sv)) |
561 | else if (!SvOK (sv)) |
525 | encode_ch (enc, 0xe0 | 22); |
562 | encode_ch (enc, MAJOR_MISC | SIMPLE_NULL); |
526 | else if (enc->cbor.flags & F_ALLOW_UNKNOWN) |
563 | else if (enc->cbor.flags & F_ALLOW_UNKNOWN) |
527 | encode_ch (enc, 0xe0 | 23); |
564 | encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF); |
528 | else |
565 | else |
529 | croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", |
566 | croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", |
530 | SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); |
567 | SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); |
531 | } |
568 | } |
532 | |
569 | |
… | |
… | |
585 | #define DEC_DEC_DEPTH --dec->depth |
622 | #define DEC_DEC_DEPTH --dec->depth |
586 | |
623 | |
587 | static UV |
624 | static UV |
588 | decode_uint (dec_t *dec) |
625 | decode_uint (dec_t *dec) |
589 | { |
626 | { |
590 | switch (*dec->cur & 31) |
627 | U8 m = *dec->cur & MINOR_MASK; |
|
|
628 | ++dec->cur; |
|
|
629 | |
|
|
630 | if (ecb_expect_true (m < LENGTH_EXT1)) |
|
|
631 | return m; |
|
|
632 | |
|
|
633 | switch (m) |
591 | { |
634 | { |
592 | case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7: |
635 | case LENGTH_EXT1: |
593 | case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15: |
636 | WANT (1); |
594 | case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23: |
637 | dec->cur += 1; |
595 | return *dec->cur++ & 31; |
638 | return dec->cur[-1]; |
596 | |
639 | |
597 | case 24: |
640 | case LENGTH_EXT2: |
598 | WANT (2); |
641 | WANT (2); |
599 | dec->cur += 2; |
642 | dec->cur += 2; |
600 | return dec->cur[-1]; |
|
|
601 | |
|
|
602 | case 25: |
|
|
603 | WANT (3); |
|
|
604 | dec->cur += 3; |
|
|
605 | return (((UV)dec->cur[-2]) << 8) |
643 | return (((UV)dec->cur[-2]) << 8) |
606 | | ((UV)dec->cur[-1]); |
644 | | ((UV)dec->cur[-1]); |
607 | |
645 | |
608 | case 26: |
646 | case LENGTH_EXT4: |
609 | WANT (5); |
647 | WANT (4); |
610 | dec->cur += 5; |
648 | dec->cur += 4; |
611 | return (((UV)dec->cur[-4]) << 24) |
649 | return (((UV)dec->cur[-4]) << 24) |
612 | | (((UV)dec->cur[-3]) << 16) |
650 | | (((UV)dec->cur[-3]) << 16) |
613 | | (((UV)dec->cur[-2]) << 8) |
651 | | (((UV)dec->cur[-2]) << 8) |
614 | | ((UV)dec->cur[-1]); |
652 | | ((UV)dec->cur[-1]); |
615 | |
653 | |
616 | case 27: |
654 | case LENGTH_EXT8: |
617 | WANT (9); |
655 | WANT (8); |
618 | dec->cur += 9; |
656 | dec->cur += 8; |
619 | |
657 | |
620 | return |
658 | return |
621 | #if UVSIZE < 8 |
659 | #if UVSIZE < 8 |
622 | 0 |
660 | 0 |
623 | #else |
661 | #else |
… | |
… | |
646 | { |
684 | { |
647 | AV *av = newAV (); |
685 | AV *av = newAV (); |
648 | |
686 | |
649 | DEC_INC_DEPTH; |
687 | DEC_INC_DEPTH; |
650 | |
688 | |
651 | if ((*dec->cur & 31) == 31) |
689 | if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF)) |
652 | { |
690 | { |
653 | ++dec->cur; |
691 | ++dec->cur; |
654 | |
692 | |
655 | for (;;) |
693 | for (;;) |
656 | { |
694 | { |
657 | WANT (1); |
695 | WANT (1); |
658 | |
696 | |
659 | if (*dec->cur == (0xe0 | 31)) |
697 | if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) |
660 | { |
698 | { |
661 | ++dec->cur; |
699 | ++dec->cur; |
662 | break; |
700 | break; |
663 | } |
701 | } |
664 | |
702 | |
… | |
… | |
689 | { |
727 | { |
690 | // for speed reasons, we specialcase single-string |
728 | // for speed reasons, we specialcase single-string |
691 | // byte or utf-8 strings as keys, but only when !stringref |
729 | // byte or utf-8 strings as keys, but only when !stringref |
692 | |
730 | |
693 | if (ecb_expect_true (!dec->stringref)) |
731 | if (ecb_expect_true (!dec->stringref)) |
694 | if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27) |
732 | if ((*dec->cur - MAJOR_BYTES) <= 27) |
695 | { |
733 | { |
696 | I32 len = decode_uint (dec); |
734 | I32 len = decode_uint (dec); |
697 | char *key = (char *)dec->cur; |
735 | char *key = (char *)dec->cur; |
698 | |
736 | |
699 | dec->cur += len; |
737 | dec->cur += len; |
… | |
… | |
703 | |
741 | |
704 | hv_store (hv, key, len, decode_sv (dec), 0); |
742 | hv_store (hv, key, len, decode_sv (dec), 0); |
705 | |
743 | |
706 | return; |
744 | return; |
707 | } |
745 | } |
708 | else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27) |
746 | else if ((*dec->cur - MAJOR_TEXT) <= 27) |
709 | { |
747 | { |
710 | I32 len = decode_uint (dec); |
748 | I32 len = decode_uint (dec); |
711 | char *key = (char *)dec->cur; |
749 | char *key = (char *)dec->cur; |
712 | |
750 | |
713 | dec->cur += len; |
751 | dec->cur += len; |
… | |
… | |
732 | { |
770 | { |
733 | HV *hv = newHV (); |
771 | HV *hv = newHV (); |
734 | |
772 | |
735 | DEC_INC_DEPTH; |
773 | DEC_INC_DEPTH; |
736 | |
774 | |
737 | if ((*dec->cur & 31) == 31) |
775 | if (*dec->cur == (MAJOR_MAP | MINOR_INDEF)) |
738 | { |
776 | { |
739 | ++dec->cur; |
777 | ++dec->cur; |
740 | |
778 | |
741 | for (;;) |
779 | for (;;) |
742 | { |
780 | { |
743 | WANT (1); |
781 | WANT (1); |
744 | |
782 | |
745 | if (*dec->cur == (0xe0 | 31)) |
783 | if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) |
746 | { |
784 | { |
747 | ++dec->cur; |
785 | ++dec->cur; |
748 | break; |
786 | break; |
749 | } |
787 | } |
750 | |
788 | |
… | |
… | |
771 | static SV * |
809 | static SV * |
772 | decode_str (dec_t *dec, int utf8) |
810 | decode_str (dec_t *dec, int utf8) |
773 | { |
811 | { |
774 | SV *sv = 0; |
812 | SV *sv = 0; |
775 | |
813 | |
776 | if ((*dec->cur & 31) == 31) |
814 | if ((*dec->cur & MINOR_MASK) == MINOR_INDEF) |
777 | { |
815 | { |
778 | // indefinite length strings |
816 | // indefinite length strings |
779 | ++dec->cur; |
817 | ++dec->cur; |
780 | |
818 | |
781 | unsigned char major = *dec->cur & 0xe0; |
819 | U8 major = *dec->cur & MAJOR_MISC; |
782 | |
820 | |
783 | sv = newSVpvn ("", 0); |
821 | sv = newSVpvn ("", 0); |
784 | |
822 | |
785 | for (;;) |
823 | for (;;) |
786 | { |
824 | { |
787 | WANT (1); |
825 | WANT (1); |
788 | |
826 | |
789 | if ((*dec->cur ^ major) >= 31) |
827 | if ((*dec->cur - major) > LENGTH_EXT8) |
790 | if (*dec->cur == (0xe0 | 31)) |
828 | if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) |
791 | { |
829 | { |
792 | ++dec->cur; |
830 | ++dec->cur; |
793 | break; |
831 | break; |
794 | } |
832 | } |
795 | else |
833 | else |
… | |
… | |
856 | } |
894 | } |
857 | break; |
895 | break; |
858 | |
896 | |
859 | case CBOR_TAG_STRINGREF: |
897 | case CBOR_TAG_STRINGREF: |
860 | { |
898 | { |
861 | if ((*dec->cur >> 5) != 0) |
899 | if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) |
862 | ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); |
900 | ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); |
863 | |
901 | |
864 | UV idx = decode_uint (dec); |
902 | UV idx = decode_uint (dec); |
865 | |
903 | |
866 | if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) |
904 | if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) |
… | |
… | |
884 | } |
922 | } |
885 | break; |
923 | break; |
886 | |
924 | |
887 | case CBOR_TAG_VALUE_SHAREDREF: |
925 | case CBOR_TAG_VALUE_SHAREDREF: |
888 | { |
926 | { |
889 | if ((*dec->cur >> 5) != 0) |
927 | if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) |
890 | ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); |
928 | ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); |
891 | |
929 | |
892 | UV idx = decode_uint (dec); |
930 | UV idx = decode_uint (dec); |
893 | |
931 | |
894 | if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) |
932 | if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) |
… | |
… | |
1003 | static SV * |
1041 | static SV * |
1004 | decode_sv (dec_t *dec) |
1042 | decode_sv (dec_t *dec) |
1005 | { |
1043 | { |
1006 | WANT (1); |
1044 | WANT (1); |
1007 | |
1045 | |
1008 | switch (*dec->cur >> 5) |
1046 | switch (*dec->cur >> MAJOR_SHIFT) |
1009 | { |
1047 | { |
1010 | case 0: // unsigned int |
1048 | case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec)); |
1011 | return newSVuv (decode_uint (dec)); |
1049 | case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec)); |
1012 | case 1: // negative int |
1050 | case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0); |
1013 | return newSViv (-1 - (IV)decode_uint (dec)); |
1051 | case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1); |
1014 | case 2: // octet string |
1052 | case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec); |
1015 | return decode_str (dec, 0); |
1053 | case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec); |
1016 | case 3: // utf-8 string |
1054 | case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec); |
1017 | return decode_str (dec, 1); |
1055 | |
1018 | case 4: // array |
1056 | case MAJOR_MISC >> MAJOR_SHIFT: |
1019 | return decode_av (dec); |
|
|
1020 | case 5: // map |
|
|
1021 | return decode_hv (dec); |
|
|
1022 | case 6: // tag |
|
|
1023 | return decode_tagged (dec); |
|
|
1024 | case 7: // misc |
|
|
1025 | switch (*dec->cur++ & 31) |
1057 | switch (*dec->cur++ & MINOR_MASK) |
1026 | { |
1058 | { |
1027 | case 20: |
1059 | case SIMPLE_FALSE: |
1028 | #if CBOR_SLOW |
1060 | #if CBOR_SLOW |
1029 | types_false = get_bool ("Types::Serialiser::false"); |
1061 | types_false = get_bool ("Types::Serialiser::false"); |
1030 | #endif |
1062 | #endif |
1031 | return newSVsv (types_false); |
1063 | return newSVsv (types_false); |
1032 | case 21: |
1064 | case SIMPLE_TRUE: |
1033 | #if CBOR_SLOW |
1065 | #if CBOR_SLOW |
1034 | types_true = get_bool ("Types::Serialiser::true"); |
1066 | types_true = get_bool ("Types::Serialiser::true"); |
1035 | #endif |
1067 | #endif |
1036 | return newSVsv (types_true); |
1068 | return newSVsv (types_true); |
1037 | case 22: |
1069 | case SIMPLE_NULL: |
1038 | return newSVsv (&PL_sv_undef); |
1070 | return newSVsv (&PL_sv_undef); |
1039 | case 23: |
1071 | case SIMPLE_UNDEF: |
1040 | #if CBOR_SLOW |
1072 | #if CBOR_SLOW |
1041 | types_error = get_bool ("Types::Serialiser::error"); |
1073 | types_error = get_bool ("Types::Serialiser::error"); |
1042 | #endif |
1074 | #endif |
1043 | return newSVsv (types_error); |
1075 | return newSVsv (types_error); |
1044 | |
1076 | |
1045 | case 25: |
1077 | case MISC_FLOAT16: |
1046 | { |
1078 | { |
1047 | WANT (2); |
1079 | WANT (2); |
1048 | |
1080 | |
1049 | uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; |
1081 | uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; |
1050 | dec->cur += 2; |
1082 | dec->cur += 2; |
1051 | |
1083 | |
1052 | return newSVnv (ecb_binary16_to_float (fp)); |
1084 | return newSVnv (ecb_binary16_to_float (fp)); |
1053 | } |
1085 | } |
1054 | |
1086 | |
1055 | case 26: |
1087 | case MISC_FLOAT32: |
1056 | { |
1088 | { |
1057 | uint32_t fp; |
1089 | uint32_t fp; |
1058 | WANT (4); |
1090 | WANT (4); |
1059 | memcpy (&fp, dec->cur, 4); |
1091 | memcpy (&fp, dec->cur, 4); |
1060 | dec->cur += 4; |
1092 | dec->cur += 4; |
… | |
… | |
1063 | fp = ecb_bswap32 (fp); |
1095 | fp = ecb_bswap32 (fp); |
1064 | |
1096 | |
1065 | return newSVnv (ecb_binary32_to_float (fp)); |
1097 | return newSVnv (ecb_binary32_to_float (fp)); |
1066 | } |
1098 | } |
1067 | |
1099 | |
1068 | case 27: |
1100 | case MISC_FLOAT64: |
1069 | { |
1101 | { |
1070 | uint64_t fp; |
1102 | uint64_t fp; |
1071 | WANT (8); |
1103 | WANT (8); |
1072 | memcpy (&fp, dec->cur, 8); |
1104 | memcpy (&fp, dec->cur, 8); |
1073 | dec->cur += 8; |
1105 | dec->cur += 8; |
… | |
… | |
1076 | fp = ecb_bswap64 (fp); |
1108 | fp = ecb_bswap64 (fp); |
1077 | |
1109 | |
1078 | return newSVnv (ecb_binary64_to_double (fp)); |
1110 | return newSVnv (ecb_binary64_to_double (fp)); |
1079 | } |
1111 | } |
1080 | |
1112 | |
1081 | // 0..19 unassigned |
1113 | // 0..19 unassigned simple |
1082 | // 24 reserved + unassigned (reserved values are not encodable) |
1114 | // 24 reserved + unassigned (reserved values are not encodable) |
1083 | default: |
1115 | default: |
1084 | ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); |
1116 | ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); |
1085 | } |
1117 | } |
1086 | |
1118 | |