… | |
… | |
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 |
24 | |
27 | |
25 | // known tags |
28 | // known tags |
26 | enum cbor_tag |
29 | enum cbor_tag |
27 | { |
30 | { |
28 | // inofficial extensions (pending iana registration) |
31 | // extensions |
29 | CBOR_TAG_PERL_OBJECT = 256, |
32 | CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref |
30 | CBOR_TAG_GENERIC_OBJECT = 257, |
33 | CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object |
|
|
34 | CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object |
|
|
35 | CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing |
|
|
36 | CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing |
|
|
37 | CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref |
|
|
38 | CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection |
31 | |
39 | |
32 | // rfc7049 |
40 | // rfc7049 |
33 | CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 |
41 | CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 |
34 | CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any |
42 | CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any |
35 | CBOR_TAG_POS_BIGNUM = 2, // byte string |
43 | CBOR_TAG_POS_BIGNUM = 2, // byte string |
… | |
… | |
49 | CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 |
57 | CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 |
50 | |
58 | |
51 | CBOR_TAG_MAGIC = 55799 // self-describe cbor |
59 | CBOR_TAG_MAGIC = 55799 // self-describe cbor |
52 | }; |
60 | }; |
53 | |
61 | |
54 | #define F_SHRINK 0x00000200UL |
62 | #define F_SHRINK 0x00000001UL |
55 | #define F_ALLOW_UNKNOWN 0x00002000UL |
63 | #define F_ALLOW_UNKNOWN 0x00000002UL |
|
|
64 | #define F_ALLOW_SHARING 0x00000004UL |
|
|
65 | #define F_PACK_STRINGS 0x00000008UL |
56 | |
66 | |
57 | #define INIT_SIZE 32 // initial scalar size to be allocated |
67 | #define INIT_SIZE 32 // initial scalar size to be allocated |
58 | |
68 | |
59 | #define SB do { |
69 | #define SB do { |
60 | #define SE } while (0) |
70 | #define SE } while (0) |
… | |
… | |
72 | # define CBOR_SLOW 0 |
82 | # define CBOR_SLOW 0 |
73 | # define CBOR_STASH cbor_stash |
83 | # define CBOR_STASH cbor_stash |
74 | #endif |
84 | #endif |
75 | |
85 | |
76 | static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS:: |
86 | static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS:: |
77 | static SV *types_true, *types_false, *types_error, *sv_cbor; |
87 | static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter; |
78 | |
88 | |
79 | typedef struct { |
89 | typedef struct { |
80 | U32 flags; |
90 | U32 flags; |
81 | U32 max_depth; |
91 | U32 max_depth; |
82 | STRLEN max_size; |
92 | STRLEN max_size; |
|
|
93 | SV *filter; |
83 | } CBOR; |
94 | } CBOR; |
84 | |
95 | |
85 | ecb_inline void |
96 | ecb_inline void |
86 | cbor_init (CBOR *cbor) |
97 | cbor_init (CBOR *cbor) |
87 | { |
98 | { |
88 | Zero (cbor, 1, CBOR); |
99 | Zero (cbor, 1, CBOR); |
89 | cbor->max_depth = 512; |
100 | cbor->max_depth = 512; |
|
|
101 | } |
|
|
102 | |
|
|
103 | ecb_inline void |
|
|
104 | cbor_free (CBOR *cbor) |
|
|
105 | { |
|
|
106 | SvREFCNT_dec (cbor->filter); |
90 | } |
107 | } |
91 | |
108 | |
92 | ///////////////////////////////////////////////////////////////////////////// |
109 | ///////////////////////////////////////////////////////////////////////////// |
93 | // utility functions |
110 | // utility functions |
94 | |
111 | |
… | |
… | |
116 | SvPV_renew (sv, SvCUR (sv) + 1); |
133 | SvPV_renew (sv, SvCUR (sv) + 1); |
117 | #endif |
134 | #endif |
118 | } |
135 | } |
119 | } |
136 | } |
120 | |
137 | |
121 | ///////////////////////////////////////////////////////////////////////////// |
138 | // minimum length of a string to be registered for stringref |
122 | // fp hell |
139 | ecb_inline int |
123 | |
140 | minimum_string_length (UV idx) |
124 | //TODO |
141 | { |
|
|
142 | return idx > 23 |
|
|
143 | ? idx > 0xffU |
|
|
144 | ? idx > 0xffffU |
|
|
145 | ? idx > 0xffffffffU |
|
|
146 | ? 11 |
|
|
147 | : 7 |
|
|
148 | : 5 |
|
|
149 | : 4 |
|
|
150 | : 3; |
|
|
151 | } |
125 | |
152 | |
126 | ///////////////////////////////////////////////////////////////////////////// |
153 | ///////////////////////////////////////////////////////////////////////////// |
127 | // encoder |
154 | // encoder |
128 | |
155 | |
129 | // structure used for encoding CBOR |
156 | // structure used for encoding CBOR |
… | |
… | |
132 | char *cur; // SvPVX (sv) + current output position |
159 | char *cur; // SvPVX (sv) + current output position |
133 | char *end; // SvEND (sv) |
160 | char *end; // SvEND (sv) |
134 | SV *sv; // result scalar |
161 | SV *sv; // result scalar |
135 | CBOR cbor; |
162 | CBOR cbor; |
136 | U32 depth; // recursion level |
163 | U32 depth; // recursion level |
|
|
164 | HV *stringref[2]; // string => index, or 0 ([0] = bytes, [1] = utf-8) |
|
|
165 | UV stringref_idx; |
|
|
166 | HV *shareable; // ptr => index, or 0 |
|
|
167 | UV shareable_idx; |
137 | } enc_t; |
168 | } enc_t; |
138 | |
169 | |
139 | ecb_inline void |
170 | ecb_inline void |
140 | need (enc_t *enc, STRLEN len) |
171 | need (enc_t *enc, STRLEN len) |
141 | { |
172 | { |
… | |
… | |
158 | static void |
189 | static void |
159 | encode_uint (enc_t *enc, int major, UV len) |
190 | encode_uint (enc_t *enc, int major, UV len) |
160 | { |
191 | { |
161 | need (enc, 9); |
192 | need (enc, 9); |
162 | |
193 | |
163 | if (len < 24) |
194 | if (ecb_expect_true (len < 24)) |
164 | *enc->cur++ = major | len; |
195 | *enc->cur++ = major | len; |
165 | else if (len <= 0xff) |
196 | else if (ecb_expect_true (len <= 0xff)) |
166 | { |
197 | { |
167 | *enc->cur++ = major | 24; |
198 | *enc->cur++ = major | 24; |
168 | *enc->cur++ = len; |
199 | *enc->cur++ = len; |
169 | } |
200 | } |
170 | else if (len <= 0xffff) |
201 | else if (len <= 0xffff) |
… | |
… | |
193 | *enc->cur++ = len >> 8; |
224 | *enc->cur++ = len >> 8; |
194 | *enc->cur++ = len; |
225 | *enc->cur++ = len; |
195 | } |
226 | } |
196 | } |
227 | } |
197 | |
228 | |
198 | static void |
229 | ecb_inline void |
|
|
230 | encode_tag (enc_t *enc, UV tag) |
|
|
231 | { |
|
|
232 | encode_uint (enc, 0xc0, tag); |
|
|
233 | } |
|
|
234 | |
|
|
235 | ecb_inline void |
199 | encode_str (enc_t *enc, int utf8, char *str, STRLEN len) |
236 | encode_str (enc_t *enc, int utf8, char *str, STRLEN len) |
200 | { |
237 | { |
201 | encode_uint (enc, utf8 ? 0x60 : 0x40, len); |
238 | encode_uint (enc, utf8 ? 0x60 : 0x40, len); |
202 | need (enc, len); |
239 | need (enc, len); |
203 | memcpy (enc->cur, str, len); |
240 | memcpy (enc->cur, str, len); |
204 | enc->cur += len; |
241 | enc->cur += len; |
205 | } |
242 | } |
206 | |
243 | |
|
|
244 | static void |
|
|
245 | encode_strref (enc_t *enc, int utf8, char *str, STRLEN len) |
|
|
246 | { |
|
|
247 | if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS)) |
|
|
248 | { |
|
|
249 | SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1); |
|
|
250 | |
|
|
251 | if (SvOK (*svp)) |
|
|
252 | { |
|
|
253 | // already registered, use stringref |
|
|
254 | encode_tag (enc, CBOR_TAG_STRINGREF); |
|
|
255 | encode_uint (enc, 0x00, SvUV (*svp)); |
|
|
256 | return; |
|
|
257 | } |
|
|
258 | else if (len >= minimum_string_length (enc->stringref_idx)) |
|
|
259 | { |
|
|
260 | // register only |
|
|
261 | sv_setuv (*svp, enc->stringref_idx); |
|
|
262 | ++enc->stringref_idx; |
|
|
263 | } |
|
|
264 | } |
|
|
265 | |
|
|
266 | encode_str (enc, utf8, str, len); |
|
|
267 | } |
|
|
268 | |
207 | static void encode_sv (enc_t *enc, SV *sv); |
269 | static void encode_sv (enc_t *enc, SV *sv); |
208 | |
270 | |
209 | static void |
271 | static void |
210 | encode_av (enc_t *enc, AV *av) |
272 | encode_av (enc_t *enc, AV *av) |
211 | { |
273 | { |
… | |
… | |
248 | while ((he = hv_iternext (hv))) |
310 | while ((he = hv_iternext (hv))) |
249 | { |
311 | { |
250 | if (HeKLEN (he) == HEf_SVKEY) |
312 | if (HeKLEN (he) == HEf_SVKEY) |
251 | encode_sv (enc, HeSVKEY (he)); |
313 | encode_sv (enc, HeSVKEY (he)); |
252 | else |
314 | else |
253 | encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); |
315 | encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); |
254 | |
316 | |
255 | encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); |
317 | encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); |
256 | } |
318 | } |
257 | |
319 | |
258 | if (mg) |
320 | if (mg) |
… | |
… | |
263 | |
325 | |
264 | // encode objects, arrays and special \0=false and \1=true values. |
326 | // encode objects, arrays and special \0=false and \1=true values. |
265 | static void |
327 | static void |
266 | encode_rv (enc_t *enc, SV *sv) |
328 | encode_rv (enc_t *enc, SV *sv) |
267 | { |
329 | { |
268 | svtype svt; |
|
|
269 | |
|
|
270 | SvGETMAGIC (sv); |
330 | 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 | |
271 | svt = SvTYPE (sv); |
354 | svtype svt = SvTYPE (sv); |
272 | |
355 | |
273 | if (ecb_expect_false (SvOBJECT (sv))) |
356 | if (ecb_expect_false (SvOBJECT (sv))) |
274 | { |
357 | { |
275 | HV *boolean_stash = !CBOR_SLOW || types_boolean_stash |
358 | HV *boolean_stash = !CBOR_SLOW || types_boolean_stash |
276 | ? types_boolean_stash |
359 | ? types_boolean_stash |
… | |
… | |
336 | |
419 | |
337 | // catch this surprisingly common error |
420 | // catch this surprisingly common error |
338 | if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) |
421 | if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) |
339 | croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); |
422 | croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); |
340 | |
423 | |
341 | encode_uint (enc, 0xc0, CBOR_TAG_PERL_OBJECT); |
424 | encode_tag (enc, CBOR_TAG_PERL_OBJECT); |
342 | encode_uint (enc, 0x80, count + 1); |
425 | encode_uint (enc, 0x80, count + 1); |
343 | encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); |
426 | encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); |
344 | |
427 | |
345 | while (count) |
428 | while (count) |
346 | encode_sv (enc, SP[1 - count--]); |
429 | encode_sv (enc, SP[1 - count--]); |
347 | |
430 | |
348 | PUTBACK; |
431 | PUTBACK; |
… | |
… | |
355 | } |
438 | } |
356 | else if (svt == SVt_PVHV) |
439 | else if (svt == SVt_PVHV) |
357 | encode_hv (enc, (HV *)sv); |
440 | encode_hv (enc, (HV *)sv); |
358 | else if (svt == SVt_PVAV) |
441 | else if (svt == SVt_PVAV) |
359 | encode_av (enc, (AV *)sv); |
442 | encode_av (enc, (AV *)sv); |
360 | else if (svt < SVt_PVAV) |
|
|
361 | { |
|
|
362 | STRLEN len = 0; |
|
|
363 | char *pv = svt ? SvPV (sv, len) : 0; |
|
|
364 | |
|
|
365 | if (len == 1 && *pv == '1') |
|
|
366 | encode_ch (enc, 0xe0 | 21); |
|
|
367 | else if (len == 1 && *pv == '0') |
|
|
368 | encode_ch (enc, 0xe0 | 20); |
|
|
369 | else if (enc->cbor.flags & F_ALLOW_UNKNOWN) |
|
|
370 | encode_ch (enc, 0xe0 | 23); |
|
|
371 | else |
|
|
372 | croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", |
|
|
373 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
|
|
374 | } |
|
|
375 | else if (enc->cbor.flags & F_ALLOW_UNKNOWN) |
|
|
376 | encode_ch (enc, 0xe0 | 23); |
|
|
377 | else |
443 | else |
378 | croak ("encountered %s, but CBOR can only represent references to arrays or hashes", |
444 | { |
379 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
445 | encode_tag (enc, CBOR_TAG_INDIRECTION); |
|
|
446 | encode_sv (enc, sv); |
|
|
447 | } |
380 | } |
448 | } |
381 | |
449 | |
382 | static void |
450 | static void |
383 | encode_nv (enc_t *enc, SV *sv) |
451 | encode_nv (enc_t *enc, SV *sv) |
384 | { |
452 | { |
… | |
… | |
422 | |
490 | |
423 | if (SvPOKp (sv)) |
491 | if (SvPOKp (sv)) |
424 | { |
492 | { |
425 | STRLEN len; |
493 | STRLEN len; |
426 | char *str = SvPV (sv, len); |
494 | char *str = SvPV (sv, len); |
427 | encode_str (enc, SvUTF8 (sv), str, len); |
495 | encode_strref (enc, SvUTF8 (sv), str, len); |
428 | } |
496 | } |
429 | else if (SvNOKp (sv)) |
497 | else if (SvNOKp (sv)) |
430 | encode_nv (enc, sv); |
498 | encode_nv (enc, sv); |
431 | else if (SvIOKp (sv)) |
499 | else if (SvIOKp (sv)) |
432 | { |
500 | { |
… | |
… | |
449 | } |
517 | } |
450 | |
518 | |
451 | static SV * |
519 | static SV * |
452 | encode_cbor (SV *scalar, CBOR *cbor) |
520 | encode_cbor (SV *scalar, CBOR *cbor) |
453 | { |
521 | { |
454 | enc_t enc; |
522 | enc_t enc = { }; |
455 | |
523 | |
456 | enc.cbor = *cbor; |
524 | enc.cbor = *cbor; |
457 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
525 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
458 | enc.cur = SvPVX (enc.sv); |
526 | enc.cur = SvPVX (enc.sv); |
459 | enc.end = SvEND (enc.sv); |
527 | enc.end = SvEND (enc.sv); |
460 | enc.depth = 0; |
|
|
461 | |
528 | |
462 | SvPOK_only (enc.sv); |
529 | SvPOK_only (enc.sv); |
|
|
530 | |
|
|
531 | if (cbor->flags & F_PACK_STRINGS) |
|
|
532 | { |
|
|
533 | encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE); |
|
|
534 | enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ()); |
|
|
535 | enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ()); |
|
|
536 | } |
|
|
537 | |
463 | encode_sv (&enc, scalar); |
538 | encode_sv (&enc, scalar); |
464 | |
539 | |
465 | SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); |
540 | SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); |
466 | *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings |
541 | *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings |
467 | |
542 | |
… | |
… | |
481 | U8 *end; // end of input string |
556 | U8 *end; // end of input string |
482 | const char *err; // parse error, if != 0 |
557 | const char *err; // parse error, if != 0 |
483 | CBOR cbor; |
558 | CBOR cbor; |
484 | U32 depth; // recursion depth |
559 | U32 depth; // recursion depth |
485 | U32 maxdepth; // recursion depth limit |
560 | U32 maxdepth; // recursion depth limit |
|
|
561 | AV *shareable; |
|
|
562 | AV *stringref; |
|
|
563 | SV *decode_tagged; |
486 | } dec_t; |
564 | } dec_t; |
487 | |
565 | |
488 | #define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE |
566 | #define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE |
489 | |
567 | |
490 | #define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") |
568 | #define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") |
… | |
… | |
584 | SvREFCNT_dec (av); |
662 | SvREFCNT_dec (av); |
585 | DEC_DEC_DEPTH; |
663 | DEC_DEC_DEPTH; |
586 | return &PL_sv_undef; |
664 | return &PL_sv_undef; |
587 | } |
665 | } |
588 | |
666 | |
|
|
667 | static void |
|
|
668 | decode_he (dec_t *dec, HV *hv) |
|
|
669 | { |
|
|
670 | // for speed reasons, we specialcase single-string |
|
|
671 | // byte or utf-8 strings as keys, but only when !stringref |
|
|
672 | |
|
|
673 | if (ecb_expect_true (!dec->stringref)) |
|
|
674 | if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27) |
|
|
675 | { |
|
|
676 | I32 len = decode_uint (dec); |
|
|
677 | char *key = (char *)dec->cur; |
|
|
678 | |
|
|
679 | dec->cur += len; |
|
|
680 | |
|
|
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); |
|
|
685 | |
|
|
686 | return; |
|
|
687 | } |
|
|
688 | else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27) |
|
|
689 | { |
|
|
690 | I32 len = decode_uint (dec); |
|
|
691 | char *key = (char *)dec->cur; |
|
|
692 | |
|
|
693 | dec->cur += len; |
|
|
694 | |
|
|
695 | if (ecb_expect_false (dec->stringref)) |
|
|
696 | av_push (dec->stringref, newSVpvn_utf8 (key, len, 1)); |
|
|
697 | |
|
|
698 | hv_store (hv, key, -len, decode_sv (dec), 0); |
|
|
699 | |
|
|
700 | return; |
|
|
701 | } |
|
|
702 | |
|
|
703 | SV *k = decode_sv (dec); |
|
|
704 | SV *v = decode_sv (dec); |
|
|
705 | |
|
|
706 | hv_store_ent (hv, k, v, 0); |
|
|
707 | SvREFCNT_dec (k); |
|
|
708 | } |
|
|
709 | |
589 | static SV * |
710 | static SV * |
590 | decode_hv (dec_t *dec) |
711 | decode_hv (dec_t *dec) |
591 | { |
712 | { |
592 | HV *hv = newHV (); |
713 | HV *hv = newHV (); |
593 | |
714 | |
… | |
… | |
605 | { |
726 | { |
606 | ++dec->cur; |
727 | ++dec->cur; |
607 | break; |
728 | break; |
608 | } |
729 | } |
609 | |
730 | |
610 | SV *k = decode_sv (dec); |
731 | decode_he (dec, hv); |
611 | SV *v = decode_sv (dec); |
|
|
612 | |
|
|
613 | hv_store_ent (hv, k, v, 0); |
|
|
614 | SvREFCNT_dec (k); |
|
|
615 | } |
732 | } |
616 | } |
733 | } |
617 | else |
734 | else |
618 | { |
735 | { |
619 | int len = decode_uint (dec); |
736 | int pairs = decode_uint (dec); |
620 | |
737 | |
621 | while (len--) |
738 | while (pairs--) |
622 | { |
739 | decode_he (dec, hv); |
623 | SV *k = decode_sv (dec); |
|
|
624 | SV *v = decode_sv (dec); |
|
|
625 | |
|
|
626 | hv_store_ent (hv, k, v, 0); |
|
|
627 | SvREFCNT_dec (k); |
|
|
628 | } |
|
|
629 | } |
740 | } |
630 | |
741 | |
631 | DEC_DEC_DEPTH; |
742 | DEC_DEC_DEPTH; |
632 | return newRV_noinc ((SV *)hv); |
743 | return newRV_noinc ((SV *)hv); |
633 | |
744 | |
… | |
… | |
667 | STRLEN len = decode_uint (dec); |
778 | STRLEN len = decode_uint (dec); |
668 | |
779 | |
669 | WANT (len); |
780 | WANT (len); |
670 | sv = newSVpvn (dec->cur, len); |
781 | sv = newSVpvn (dec->cur, len); |
671 | dec->cur += len; |
782 | dec->cur += len; |
|
|
783 | |
|
|
784 | if (ecb_expect_false (dec->stringref) |
|
|
785 | && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1)) |
|
|
786 | av_push (dec->stringref, SvREFCNT_inc_NN (sv)); |
672 | } |
787 | } |
673 | |
788 | |
674 | if (utf8) |
789 | if (utf8) |
675 | SvUTF8_on (sv); |
790 | SvUTF8_on (sv); |
676 | |
791 | |
… | |
… | |
682 | } |
797 | } |
683 | |
798 | |
684 | static SV * |
799 | static SV * |
685 | decode_tagged (dec_t *dec) |
800 | decode_tagged (dec_t *dec) |
686 | { |
801 | { |
|
|
802 | SV *sv = 0; |
687 | UV tag = decode_uint (dec); |
803 | UV tag = decode_uint (dec); |
|
|
804 | |
|
|
805 | WANT (1); |
|
|
806 | |
|
|
807 | switch (tag) |
|
|
808 | { |
|
|
809 | case CBOR_TAG_MAGIC: |
|
|
810 | sv = decode_sv (dec); |
|
|
811 | break; |
|
|
812 | |
|
|
813 | case CBOR_TAG_INDIRECTION: |
|
|
814 | sv = newRV_noinc (decode_sv (dec)); |
|
|
815 | break; |
|
|
816 | |
|
|
817 | case CBOR_TAG_STRINGREF_NAMESPACE: |
|
|
818 | { |
|
|
819 | ENTER; SAVETMPS; |
|
|
820 | |
|
|
821 | SAVESPTR (dec->stringref); |
|
|
822 | dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); |
|
|
823 | |
|
|
824 | sv = decode_sv (dec); |
|
|
825 | |
|
|
826 | FREETMPS; LEAVE; |
|
|
827 | } |
|
|
828 | break; |
|
|
829 | |
|
|
830 | case CBOR_TAG_STRINGREF: |
|
|
831 | { |
|
|
832 | if ((*dec->cur >> 5) != 0) |
|
|
833 | ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); |
|
|
834 | |
|
|
835 | UV idx = decode_uint (dec); |
|
|
836 | |
|
|
837 | if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) |
|
|
838 | ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)"); |
|
|
839 | |
|
|
840 | sv = newSVsv (AvARRAY (dec->stringref)[idx]); |
|
|
841 | } |
|
|
842 | break; |
|
|
843 | |
|
|
844 | case CBOR_TAG_VALUE_SHAREABLE: |
|
|
845 | { |
|
|
846 | if (ecb_expect_false (!dec->shareable)) |
|
|
847 | dec->shareable = (AV *)sv_2mortal ((SV *)newAV ()); |
|
|
848 | |
|
|
849 | sv = newSV (0); |
|
|
850 | av_push (dec->shareable, SvREFCNT_inc_NN (sv)); |
|
|
851 | |
688 | SV *sv = decode_sv (dec); |
852 | SV *osv = decode_sv (dec); |
|
|
853 | sv_setsv (sv, osv); |
|
|
854 | SvREFCNT_dec_NN (osv); |
|
|
855 | } |
|
|
856 | break; |
689 | |
857 | |
690 | if (tag == CBOR_TAG_MAGIC) |
858 | case CBOR_TAG_VALUE_SHAREDREF: |
691 | return sv; |
859 | { |
692 | else if (tag == CBOR_TAG_PERL_OBJECT) |
860 | if ((*dec->cur >> 5) != 0) |
693 | { |
861 | ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); |
|
|
862 | |
|
|
863 | UV idx = decode_uint (dec); |
|
|
864 | |
|
|
865 | if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) |
|
|
866 | ERR ("corrupted CBOR data (sharedref index out of bounds)"); |
|
|
867 | |
|
|
868 | sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); |
|
|
869 | } |
|
|
870 | break; |
|
|
871 | |
|
|
872 | case CBOR_TAG_PERL_OBJECT: |
|
|
873 | { |
|
|
874 | sv = decode_sv (dec); |
|
|
875 | |
694 | if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) |
876 | if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) |
695 | ERR ("corrupted CBOR data (non-array perl object)"); |
877 | ERR ("corrupted CBOR data (non-array perl object)"); |
696 | |
878 | |
697 | AV *av = (AV *)SvRV (sv); |
879 | AV *av = (AV *)SvRV (sv); |
698 | int len = av_len (av) + 1; |
880 | int len = av_len (av) + 1; |
699 | HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0); |
881 | HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0); |
700 | |
882 | |
701 | if (!stash) |
883 | if (!stash) |
702 | ERR ("cannot decode perl-object (package does not exist)"); |
884 | ERR ("cannot decode perl-object (package does not exist)"); |
703 | |
885 | |
704 | GV *method = gv_fetchmethod_autoload (stash, "THAW", 0); |
886 | GV *method = gv_fetchmethod_autoload (stash, "THAW", 0); |
705 | |
887 | |
706 | if (!method) |
888 | if (!method) |
707 | ERR ("cannot decode perl-object (package does not have a THAW method)"); |
889 | ERR ("cannot decode perl-object (package does not have a THAW method)"); |
708 | |
890 | |
709 | dSP; |
891 | dSP; |
710 | |
892 | |
711 | ENTER; SAVETMPS; PUSHMARK (SP); |
893 | ENTER; SAVETMPS; PUSHMARK (SP); |
712 | EXTEND (SP, len + 1); |
894 | EXTEND (SP, len + 1); |
713 | // we re-bless the reference to get overload and other niceties right |
895 | // we re-bless the reference to get overload and other niceties right |
714 | PUSHs (*av_fetch (av, 0, 1)); |
896 | PUSHs (*av_fetch (av, 0, 1)); |
715 | PUSHs (sv_cbor); |
897 | PUSHs (sv_cbor); |
716 | |
898 | |
717 | int i; |
899 | int i; |
718 | |
900 | |
719 | for (i = 1; i < len; ++i) |
901 | for (i = 1; i < len; ++i) |
720 | PUSHs (*av_fetch (av, i, 1)); |
902 | PUSHs (*av_fetch (av, i, 1)); |
721 | |
903 | |
722 | PUTBACK; |
904 | PUTBACK; |
723 | call_sv ((SV *)GvCV (method), G_SCALAR); |
905 | call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL); |
724 | SPAGAIN; |
906 | SPAGAIN; |
725 | |
907 | |
|
|
908 | if (SvTRUE (ERRSV)) |
|
|
909 | { |
|
|
910 | FREETMPS; LEAVE; |
|
|
911 | ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); |
|
|
912 | } |
|
|
913 | |
726 | SvREFCNT_dec (sv); |
914 | SvREFCNT_dec (sv); |
727 | sv = SvREFCNT_inc (POPs); |
915 | sv = SvREFCNT_inc (POPs); |
728 | |
916 | |
729 | PUTBACK; |
917 | PUTBACK; |
730 | |
918 | |
731 | FREETMPS; LEAVE; |
919 | FREETMPS; LEAVE; |
|
|
920 | } |
|
|
921 | break; |
732 | |
922 | |
733 | return sv; |
923 | default: |
734 | } |
924 | { |
735 | else |
925 | sv = decode_sv (dec); |
736 | { |
926 | |
|
|
927 | dSP; |
|
|
928 | ENTER; SAVETMPS; PUSHMARK (SP); |
|
|
929 | EXTEND (SP, 2); |
|
|
930 | PUSHs (newSVuv (tag)); |
|
|
931 | PUSHs (sv); |
|
|
932 | |
|
|
933 | PUTBACK; |
|
|
934 | int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL); |
|
|
935 | SPAGAIN; |
|
|
936 | |
|
|
937 | if (SvTRUE (ERRSV)) |
|
|
938 | { |
|
|
939 | FREETMPS; LEAVE; |
|
|
940 | ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); |
|
|
941 | } |
|
|
942 | |
|
|
943 | if (count) |
|
|
944 | { |
|
|
945 | SvREFCNT_dec (sv); |
|
|
946 | sv = SvREFCNT_inc (POPs); |
|
|
947 | } |
|
|
948 | else |
|
|
949 | { |
737 | AV *av = newAV (); |
950 | AV *av = newAV (); |
738 | av_push (av, newSVuv (tag)); |
951 | av_push (av, newSVuv (tag)); |
739 | av_push (av, sv); |
952 | av_push (av, sv); |
740 | |
953 | |
741 | HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash |
954 | HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash |
742 | ? cbor_tagged_stash |
955 | ? cbor_tagged_stash |
743 | : gv_stashpv ("CBOR::XS::Tagged" , 1); |
956 | : gv_stashpv ("CBOR::XS::Tagged" , 1); |
744 | |
|
|
745 | return sv_bless (newRV_noinc ((SV *)av), tagged_stash); |
957 | sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash); |
|
|
958 | } |
|
|
959 | |
|
|
960 | PUTBACK; |
|
|
961 | |
|
|
962 | FREETMPS; LEAVE; |
|
|
963 | } |
|
|
964 | break; |
746 | } |
965 | } |
|
|
966 | |
|
|
967 | return sv; |
747 | |
968 | |
748 | fail: |
969 | fail: |
749 | SvREFCNT_dec (sv); |
970 | SvREFCNT_dec (sv); |
750 | return &PL_sv_undef; |
971 | return &PL_sv_undef; |
751 | } |
972 | } |
… | |
… | |
842 | } |
1063 | } |
843 | |
1064 | |
844 | static SV * |
1065 | static SV * |
845 | decode_cbor (SV *string, CBOR *cbor, char **offset_return) |
1066 | decode_cbor (SV *string, CBOR *cbor, char **offset_return) |
846 | { |
1067 | { |
847 | dec_t dec; |
1068 | dec_t dec = { }; |
848 | SV *sv; |
1069 | SV *sv; |
|
|
1070 | STRLEN len; |
|
|
1071 | char *data = SvPVbyte (string, len); |
849 | |
1072 | |
850 | /* work around bugs in 5.10 where manipulating magic values |
|
|
851 | * makes perl ignore the magic in subsequent accesses. |
|
|
852 | * also make a copy of non-PV values, to get them into a clean |
|
|
853 | * state (SvPV should do that, but it's buggy, see below). |
|
|
854 | */ |
|
|
855 | /*SvGETMAGIC (string);*/ |
|
|
856 | if (SvMAGICAL (string) || !SvPOK (string)) |
|
|
857 | string = sv_2mortal (newSVsv (string)); |
|
|
858 | |
|
|
859 | SvUPGRADE (string, SVt_PV); |
|
|
860 | |
|
|
861 | /* work around a bug in perl 5.10, which causes SvCUR to fail an |
|
|
862 | * assertion with -DDEBUGGING, although SvCUR is documented to |
|
|
863 | * return the xpv_cur field which certainly exists after upgrading. |
|
|
864 | * according to nicholas clark, calling SvPOK fixes this. |
|
|
865 | * But it doesn't fix it, so try another workaround, call SvPV_nolen |
|
|
866 | * and hope for the best. |
|
|
867 | * Damnit, SvPV_nolen still trips over yet another assertion. This |
|
|
868 | * assertion business is seriously broken, try yet another workaround |
|
|
869 | * for the broken -DDEBUGGING. |
|
|
870 | */ |
|
|
871 | { |
|
|
872 | #ifdef DEBUGGING |
|
|
873 | STRLEN offset = SvOK (string) ? sv_len (string) : 0; |
|
|
874 | #else |
|
|
875 | STRLEN offset = SvCUR (string); |
|
|
876 | #endif |
|
|
877 | |
|
|
878 | if (offset > cbor->max_size && cbor->max_size) |
1073 | if (len > cbor->max_size && cbor->max_size) |
879 | croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu", |
1074 | croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu", |
880 | (unsigned long)SvCUR (string), (unsigned long)cbor->max_size); |
1075 | (unsigned long)len, (unsigned long)cbor->max_size); |
881 | } |
|
|
882 | |
|
|
883 | sv_utf8_downgrade (string, 0); |
|
|
884 | |
1076 | |
885 | dec.cbor = *cbor; |
1077 | dec.cbor = *cbor; |
886 | dec.cur = (U8 *)SvPVX (string); |
1078 | dec.cur = (U8 *)data; |
887 | dec.end = (U8 *)SvEND (string); |
1079 | dec.end = (U8 *)data + len; |
888 | dec.err = 0; |
|
|
889 | dec.depth = 0; |
|
|
890 | |
1080 | |
891 | sv = decode_sv (&dec); |
1081 | sv = decode_sv (&dec); |
892 | |
1082 | |
893 | if (offset_return) |
1083 | if (offset_return) |
894 | *offset_return = dec.cur; |
1084 | *offset_return = dec.cur; |
… | |
… | |
898 | dec.err = "garbage after CBOR object"; |
1088 | dec.err = "garbage after CBOR object"; |
899 | |
1089 | |
900 | if (dec.err) |
1090 | if (dec.err) |
901 | { |
1091 | { |
902 | SvREFCNT_dec (sv); |
1092 | SvREFCNT_dec (sv); |
903 | croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)SvPVX (string), (int)(uint8_t)*dec.cur); |
1093 | croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); |
904 | } |
1094 | } |
905 | |
1095 | |
906 | sv = sv_2mortal (sv); |
1096 | sv = sv_2mortal (sv); |
907 | |
1097 | |
908 | return sv; |
1098 | return sv; |
… | |
… | |
922 | types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1); |
1112 | types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1); |
923 | |
1113 | |
924 | types_true = get_bool ("Types::Serialiser::true" ); |
1114 | types_true = get_bool ("Types::Serialiser::true" ); |
925 | types_false = get_bool ("Types::Serialiser::false"); |
1115 | types_false = get_bool ("Types::Serialiser::false"); |
926 | types_error = get_bool ("Types::Serialiser::error"); |
1116 | types_error = get_bool ("Types::Serialiser::error"); |
|
|
1117 | |
|
|
1118 | default_filter = newSVpv ("CBOR::XS::default_filter", 0); |
927 | |
1119 | |
928 | sv_cbor = newSVpv ("CBOR", 0); |
1120 | sv_cbor = newSVpv ("CBOR", 0); |
929 | SvREADONLY_on (sv_cbor); |
1121 | SvREADONLY_on (sv_cbor); |
930 | } |
1122 | } |
931 | |
1123 | |
… | |
… | |
952 | |
1144 | |
953 | void shrink (CBOR *self, int enable = 1) |
1145 | void shrink (CBOR *self, int enable = 1) |
954 | ALIAS: |
1146 | ALIAS: |
955 | shrink = F_SHRINK |
1147 | shrink = F_SHRINK |
956 | allow_unknown = F_ALLOW_UNKNOWN |
1148 | allow_unknown = F_ALLOW_UNKNOWN |
|
|
1149 | allow_sharing = F_ALLOW_SHARING |
|
|
1150 | pack_strings = F_PACK_STRINGS |
957 | PPCODE: |
1151 | PPCODE: |
958 | { |
1152 | { |
959 | if (enable) |
1153 | if (enable) |
960 | self->flags |= ix; |
1154 | self->flags |= ix; |
961 | else |
1155 | else |
… | |
… | |
966 | |
1160 | |
967 | void get_shrink (CBOR *self) |
1161 | void get_shrink (CBOR *self) |
968 | ALIAS: |
1162 | ALIAS: |
969 | get_shrink = F_SHRINK |
1163 | get_shrink = F_SHRINK |
970 | get_allow_unknown = F_ALLOW_UNKNOWN |
1164 | get_allow_unknown = F_ALLOW_UNKNOWN |
|
|
1165 | get_allow_sharing = F_ALLOW_SHARING |
|
|
1166 | get_pack_strings = F_PACK_STRINGS |
971 | PPCODE: |
1167 | PPCODE: |
972 | XPUSHs (boolSV (self->flags & ix)); |
1168 | XPUSHs (boolSV (self->flags & ix)); |
973 | |
1169 | |
974 | void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) |
1170 | void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) |
975 | PPCODE: |
1171 | PPCODE: |
… | |
… | |
988 | XPUSHs (ST (0)); |
1184 | XPUSHs (ST (0)); |
989 | |
1185 | |
990 | int get_max_size (CBOR *self) |
1186 | int get_max_size (CBOR *self) |
991 | CODE: |
1187 | CODE: |
992 | RETVAL = self->max_size; |
1188 | RETVAL = self->max_size; |
|
|
1189 | OUTPUT: |
|
|
1190 | RETVAL |
|
|
1191 | |
|
|
1192 | void filter (CBOR *self, SV *filter = 0) |
|
|
1193 | PPCODE: |
|
|
1194 | SvREFCNT_dec (self->filter); |
|
|
1195 | self->filter = filter ? newSVsv (filter) : filter; |
|
|
1196 | XPUSHs (ST (0)); |
|
|
1197 | |
|
|
1198 | SV *get_filter (CBOR *self) |
|
|
1199 | CODE: |
|
|
1200 | RETVAL = self->filter ? self->filter : NEWSV (0, 0); |
993 | OUTPUT: |
1201 | OUTPUT: |
994 | RETVAL |
1202 | RETVAL |
995 | |
1203 | |
996 | void encode (CBOR *self, SV *scalar) |
1204 | void encode (CBOR *self, SV *scalar) |
997 | PPCODE: |
1205 | PPCODE: |
… | |
… | |
1012 | EXTEND (SP, 2); |
1220 | EXTEND (SP, 2); |
1013 | PUSHs (sv); |
1221 | PUSHs (sv); |
1014 | PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); |
1222 | PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); |
1015 | } |
1223 | } |
1016 | |
1224 | |
|
|
1225 | void DESTROY (CBOR *self) |
|
|
1226 | PPCODE: |
|
|
1227 | cbor_free (self); |
|
|
1228 | |
1017 | PROTOTYPES: ENABLE |
1229 | PROTOTYPES: ENABLE |
1018 | |
1230 | |
1019 | void encode_cbor (SV *scalar) |
1231 | void encode_cbor (SV *scalar) |
1020 | PPCODE: |
1232 | PPCODE: |
1021 | { |
1233 | { |