… | |
… | |
9 | #include <limits.h> |
9 | #include <limits.h> |
10 | #include <float.h> |
10 | #include <float.h> |
11 | |
11 | |
12 | #include "ecb.h" |
12 | #include "ecb.h" |
13 | |
13 | |
|
|
14 | // compatibility with perl <5.18 |
|
|
15 | #ifndef HvNAMELEN_get |
|
|
16 | # define HvNAMELEN_get(hv) strlen (HvNAME (hv)) |
|
|
17 | #endif |
|
|
18 | #ifndef HvNAMELEN |
|
|
19 | # define HvNAMELEN(hv) HvNAMELEN_get (hv) |
|
|
20 | #endif |
|
|
21 | #ifndef HvNAMEUTF8 |
|
|
22 | # define HvNAMEUTF8(hv) 0 |
|
|
23 | #endif |
|
|
24 | #ifndef SvREFCNT_dec_NN |
|
|
25 | # define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv) |
|
|
26 | #endif |
|
|
27 | |
14 | // known tags |
28 | // known tags |
15 | enum cbor_tag |
29 | enum cbor_tag |
16 | { |
30 | { |
17 | // inofficial extensions (pending iana registration) |
31 | // extensions |
18 | CBOR_TAG_PERL_OBJECT = 256, |
32 | CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref |
19 | 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 |
20 | |
39 | |
21 | // rfc7049 |
40 | // rfc7049 |
22 | CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 |
41 | CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 |
23 | CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any |
42 | CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any |
24 | CBOR_TAG_POS_BIGNUM = 2, // byte string |
43 | CBOR_TAG_POS_BIGNUM = 2, // byte string |
… | |
… | |
38 | CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 |
57 | CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 |
39 | |
58 | |
40 | CBOR_TAG_MAGIC = 55799 // self-describe cbor |
59 | CBOR_TAG_MAGIC = 55799 // self-describe cbor |
41 | }; |
60 | }; |
42 | |
61 | |
43 | #define F_SHRINK 0x00000200UL |
62 | #define F_SHRINK 0x00000001UL |
44 | #define F_ALLOW_UNKNOWN 0x00002000UL |
63 | #define F_ALLOW_UNKNOWN 0x00000002UL |
|
|
64 | #define F_ALLOW_SHARING 0x00000004UL |
|
|
65 | #define F_PACK_STRINGS 0x00000008UL |
45 | |
66 | |
46 | #define INIT_SIZE 32 // initial scalar size to be allocated |
67 | #define INIT_SIZE 32 // initial scalar size to be allocated |
47 | |
68 | |
48 | #define SB do { |
69 | #define SB do { |
49 | #define SE } while (0) |
70 | #define SE } while (0) |
… | |
… | |
61 | # define CBOR_SLOW 0 |
82 | # define CBOR_SLOW 0 |
62 | # define CBOR_STASH cbor_stash |
83 | # define CBOR_STASH cbor_stash |
63 | #endif |
84 | #endif |
64 | |
85 | |
65 | 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:: |
66 | static SV *types_true, *types_false, *types_error, *sv_cbor; |
87 | static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter; |
67 | |
88 | |
68 | typedef struct { |
89 | typedef struct { |
69 | U32 flags; |
90 | U32 flags; |
70 | U32 max_depth; |
91 | U32 max_depth; |
71 | STRLEN max_size; |
92 | STRLEN max_size; |
|
|
93 | SV *filter; |
72 | } CBOR; |
94 | } CBOR; |
73 | |
95 | |
74 | ecb_inline void |
96 | ecb_inline void |
75 | cbor_init (CBOR *cbor) |
97 | cbor_init (CBOR *cbor) |
76 | { |
98 | { |
77 | Zero (cbor, 1, CBOR); |
99 | Zero (cbor, 1, CBOR); |
78 | 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); |
79 | } |
107 | } |
80 | |
108 | |
81 | ///////////////////////////////////////////////////////////////////////////// |
109 | ///////////////////////////////////////////////////////////////////////////// |
82 | // utility functions |
110 | // utility functions |
83 | |
111 | |
… | |
… | |
105 | SvPV_renew (sv, SvCUR (sv) + 1); |
133 | SvPV_renew (sv, SvCUR (sv) + 1); |
106 | #endif |
134 | #endif |
107 | } |
135 | } |
108 | } |
136 | } |
109 | |
137 | |
110 | ///////////////////////////////////////////////////////////////////////////// |
138 | // minimum length of a string to be registered for stringref |
111 | // fp hell |
139 | ecb_inline int |
112 | |
140 | minimum_string_length (UV idx) |
113 | //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 | } |
114 | |
152 | |
115 | ///////////////////////////////////////////////////////////////////////////// |
153 | ///////////////////////////////////////////////////////////////////////////// |
116 | // encoder |
154 | // encoder |
117 | |
155 | |
118 | // structure used for encoding CBOR |
156 | // structure used for encoding CBOR |
… | |
… | |
121 | char *cur; // SvPVX (sv) + current output position |
159 | char *cur; // SvPVX (sv) + current output position |
122 | char *end; // SvEND (sv) |
160 | char *end; // SvEND (sv) |
123 | SV *sv; // result scalar |
161 | SV *sv; // result scalar |
124 | CBOR cbor; |
162 | CBOR cbor; |
125 | 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; |
126 | } enc_t; |
168 | } enc_t; |
127 | |
169 | |
128 | ecb_inline void |
170 | ecb_inline void |
129 | need (enc_t *enc, STRLEN len) |
171 | need (enc_t *enc, STRLEN len) |
130 | { |
172 | { |
… | |
… | |
147 | static void |
189 | static void |
148 | encode_uint (enc_t *enc, int major, UV len) |
190 | encode_uint (enc_t *enc, int major, UV len) |
149 | { |
191 | { |
150 | need (enc, 9); |
192 | need (enc, 9); |
151 | |
193 | |
152 | if (len < 24) |
194 | if (ecb_expect_true (len < 24)) |
153 | *enc->cur++ = major | len; |
195 | *enc->cur++ = major | len; |
154 | else if (len <= 0xff) |
196 | else if (ecb_expect_true (len <= 0xff)) |
155 | { |
197 | { |
156 | *enc->cur++ = major | 24; |
198 | *enc->cur++ = major | 24; |
157 | *enc->cur++ = len; |
199 | *enc->cur++ = len; |
158 | } |
200 | } |
159 | else if (len <= 0xffff) |
201 | else if (len <= 0xffff) |
… | |
… | |
182 | *enc->cur++ = len >> 8; |
224 | *enc->cur++ = len >> 8; |
183 | *enc->cur++ = len; |
225 | *enc->cur++ = len; |
184 | } |
226 | } |
185 | } |
227 | } |
186 | |
228 | |
187 | 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 |
188 | encode_str (enc_t *enc, int utf8, char *str, STRLEN len) |
236 | encode_str (enc_t *enc, int utf8, char *str, STRLEN len) |
189 | { |
237 | { |
190 | encode_uint (enc, utf8 ? 0x60 : 0x40, len); |
238 | encode_uint (enc, utf8 ? 0x60 : 0x40, len); |
191 | need (enc, len); |
239 | need (enc, len); |
192 | memcpy (enc->cur, str, len); |
240 | memcpy (enc->cur, str, len); |
193 | enc->cur += len; |
241 | enc->cur += len; |
194 | } |
242 | } |
195 | |
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 | |
196 | static void encode_sv (enc_t *enc, SV *sv); |
269 | static void encode_sv (enc_t *enc, SV *sv); |
197 | |
270 | |
198 | static void |
271 | static void |
199 | encode_av (enc_t *enc, AV *av) |
272 | encode_av (enc_t *enc, AV *av) |
200 | { |
273 | { |
… | |
… | |
237 | while ((he = hv_iternext (hv))) |
310 | while ((he = hv_iternext (hv))) |
238 | { |
311 | { |
239 | if (HeKLEN (he) == HEf_SVKEY) |
312 | if (HeKLEN (he) == HEf_SVKEY) |
240 | encode_sv (enc, HeSVKEY (he)); |
313 | encode_sv (enc, HeSVKEY (he)); |
241 | else |
314 | else |
242 | encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); |
315 | encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); |
243 | |
316 | |
244 | 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)); |
245 | } |
318 | } |
246 | |
319 | |
247 | if (mg) |
320 | if (mg) |
… | |
… | |
252 | |
325 | |
253 | // encode objects, arrays and special \0=false and \1=true values. |
326 | // encode objects, arrays and special \0=false and \1=true values. |
254 | static void |
327 | static void |
255 | encode_rv (enc_t *enc, SV *sv) |
328 | encode_rv (enc_t *enc, SV *sv) |
256 | { |
329 | { |
257 | svtype svt; |
|
|
258 | |
|
|
259 | 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 | |
260 | svt = SvTYPE (sv); |
354 | svtype svt = SvTYPE (sv); |
261 | |
355 | |
262 | if (ecb_expect_false (SvOBJECT (sv))) |
356 | if (ecb_expect_false (SvOBJECT (sv))) |
263 | { |
357 | { |
264 | HV *boolean_stash = !CBOR_SLOW || types_boolean_stash |
358 | HV *boolean_stash = !CBOR_SLOW || types_boolean_stash |
265 | ? types_boolean_stash |
359 | ? types_boolean_stash |
… | |
… | |
325 | |
419 | |
326 | // catch this surprisingly common error |
420 | // catch this surprisingly common error |
327 | if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) |
421 | if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) |
328 | 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)); |
329 | |
423 | |
330 | encode_uint (enc, 0xc0, CBOR_TAG_PERL_OBJECT); |
424 | encode_tag (enc, CBOR_TAG_PERL_OBJECT); |
331 | encode_uint (enc, 0x80, count + 1); |
425 | encode_uint (enc, 0x80, count + 1); |
332 | encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); |
426 | encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); |
333 | |
427 | |
334 | while (count) |
428 | while (count) |
335 | encode_sv (enc, SP[1 - count--]); |
429 | encode_sv (enc, SP[1 - count--]); |
336 | |
430 | |
337 | PUTBACK; |
431 | PUTBACK; |
… | |
… | |
344 | } |
438 | } |
345 | else if (svt == SVt_PVHV) |
439 | else if (svt == SVt_PVHV) |
346 | encode_hv (enc, (HV *)sv); |
440 | encode_hv (enc, (HV *)sv); |
347 | else if (svt == SVt_PVAV) |
441 | else if (svt == SVt_PVAV) |
348 | encode_av (enc, (AV *)sv); |
442 | encode_av (enc, (AV *)sv); |
349 | else if (svt < SVt_PVAV) |
|
|
350 | { |
|
|
351 | STRLEN len = 0; |
|
|
352 | char *pv = svt ? SvPV (sv, len) : 0; |
|
|
353 | |
|
|
354 | if (len == 1 && *pv == '1') |
|
|
355 | encode_ch (enc, 0xe0 | 21); |
|
|
356 | else if (len == 1 && *pv == '0') |
|
|
357 | encode_ch (enc, 0xe0 | 20); |
|
|
358 | else if (enc->cbor.flags & F_ALLOW_UNKNOWN) |
|
|
359 | encode_ch (enc, 0xe0 | 23); |
|
|
360 | else |
|
|
361 | croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", |
|
|
362 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
|
|
363 | } |
|
|
364 | else if (enc->cbor.flags & F_ALLOW_UNKNOWN) |
|
|
365 | encode_ch (enc, 0xe0 | 23); |
|
|
366 | else |
443 | else |
367 | croak ("encountered %s, but CBOR can only represent references to arrays or hashes", |
444 | { |
368 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
445 | encode_tag (enc, CBOR_TAG_INDIRECTION); |
|
|
446 | encode_sv (enc, sv); |
|
|
447 | } |
369 | } |
448 | } |
370 | |
449 | |
371 | static void |
450 | static void |
372 | encode_nv (enc_t *enc, SV *sv) |
451 | encode_nv (enc_t *enc, SV *sv) |
373 | { |
452 | { |
… | |
… | |
411 | |
490 | |
412 | if (SvPOKp (sv)) |
491 | if (SvPOKp (sv)) |
413 | { |
492 | { |
414 | STRLEN len; |
493 | STRLEN len; |
415 | char *str = SvPV (sv, len); |
494 | char *str = SvPV (sv, len); |
416 | encode_str (enc, SvUTF8 (sv), str, len); |
495 | encode_strref (enc, SvUTF8 (sv), str, len); |
417 | } |
496 | } |
418 | else if (SvNOKp (sv)) |
497 | else if (SvNOKp (sv)) |
419 | encode_nv (enc, sv); |
498 | encode_nv (enc, sv); |
420 | else if (SvIOKp (sv)) |
499 | else if (SvIOKp (sv)) |
421 | { |
500 | { |
… | |
… | |
438 | } |
517 | } |
439 | |
518 | |
440 | static SV * |
519 | static SV * |
441 | encode_cbor (SV *scalar, CBOR *cbor) |
520 | encode_cbor (SV *scalar, CBOR *cbor) |
442 | { |
521 | { |
443 | enc_t enc; |
522 | enc_t enc = { }; |
444 | |
523 | |
445 | enc.cbor = *cbor; |
524 | enc.cbor = *cbor; |
446 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
525 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
447 | enc.cur = SvPVX (enc.sv); |
526 | enc.cur = SvPVX (enc.sv); |
448 | enc.end = SvEND (enc.sv); |
527 | enc.end = SvEND (enc.sv); |
449 | enc.depth = 0; |
|
|
450 | |
528 | |
451 | 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 | |
452 | encode_sv (&enc, scalar); |
538 | encode_sv (&enc, scalar); |
453 | |
539 | |
454 | SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); |
540 | SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); |
455 | *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 |
456 | |
542 | |
… | |
… | |
470 | U8 *end; // end of input string |
556 | U8 *end; // end of input string |
471 | const char *err; // parse error, if != 0 |
557 | const char *err; // parse error, if != 0 |
472 | CBOR cbor; |
558 | CBOR cbor; |
473 | U32 depth; // recursion depth |
559 | U32 depth; // recursion depth |
474 | U32 maxdepth; // recursion depth limit |
560 | U32 maxdepth; // recursion depth limit |
|
|
561 | AV *shareable; |
|
|
562 | AV *stringref; |
|
|
563 | SV *decode_tagged; |
475 | } dec_t; |
564 | } dec_t; |
476 | |
565 | |
477 | #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 |
478 | |
567 | |
479 | #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") |
… | |
… | |
573 | SvREFCNT_dec (av); |
662 | SvREFCNT_dec (av); |
574 | DEC_DEC_DEPTH; |
663 | DEC_DEC_DEPTH; |
575 | return &PL_sv_undef; |
664 | return &PL_sv_undef; |
576 | } |
665 | } |
577 | |
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 | |
578 | static SV * |
710 | static SV * |
579 | decode_hv (dec_t *dec) |
711 | decode_hv (dec_t *dec) |
580 | { |
712 | { |
581 | HV *hv = newHV (); |
713 | HV *hv = newHV (); |
582 | |
714 | |
… | |
… | |
594 | { |
726 | { |
595 | ++dec->cur; |
727 | ++dec->cur; |
596 | break; |
728 | break; |
597 | } |
729 | } |
598 | |
730 | |
599 | SV *k = decode_sv (dec); |
731 | decode_he (dec, hv); |
600 | SV *v = decode_sv (dec); |
|
|
601 | |
|
|
602 | hv_store_ent (hv, k, v, 0); |
|
|
603 | } |
732 | } |
604 | } |
733 | } |
605 | else |
734 | else |
606 | { |
735 | { |
607 | int len = decode_uint (dec); |
736 | int pairs = decode_uint (dec); |
608 | |
737 | |
609 | while (len--) |
738 | while (pairs--) |
610 | { |
739 | decode_he (dec, hv); |
611 | SV *k = decode_sv (dec); |
|
|
612 | SV *v = decode_sv (dec); |
|
|
613 | |
|
|
614 | hv_store_ent (hv, k, v, 0); |
|
|
615 | } |
|
|
616 | } |
740 | } |
617 | |
741 | |
618 | DEC_DEC_DEPTH; |
742 | DEC_DEC_DEPTH; |
619 | return newRV_noinc ((SV *)hv); |
743 | return newRV_noinc ((SV *)hv); |
620 | |
744 | |
… | |
… | |
654 | STRLEN len = decode_uint (dec); |
778 | STRLEN len = decode_uint (dec); |
655 | |
779 | |
656 | WANT (len); |
780 | WANT (len); |
657 | sv = newSVpvn (dec->cur, len); |
781 | sv = newSVpvn (dec->cur, len); |
658 | 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)); |
659 | } |
787 | } |
660 | |
788 | |
661 | if (utf8) |
789 | if (utf8) |
662 | SvUTF8_on (sv); |
790 | SvUTF8_on (sv); |
663 | |
791 | |
… | |
… | |
669 | } |
797 | } |
670 | |
798 | |
671 | static SV * |
799 | static SV * |
672 | decode_tagged (dec_t *dec) |
800 | decode_tagged (dec_t *dec) |
673 | { |
801 | { |
|
|
802 | SV *sv = 0; |
674 | 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 | |
675 | SV *sv = decode_sv (dec); |
852 | SV *osv = decode_sv (dec); |
|
|
853 | sv_setsv (sv, osv); |
|
|
854 | SvREFCNT_dec_NN (osv); |
|
|
855 | } |
|
|
856 | break; |
676 | |
857 | |
677 | if (tag == CBOR_TAG_MAGIC) |
858 | case CBOR_TAG_VALUE_SHAREDREF: |
678 | return sv; |
859 | { |
679 | else if (tag == CBOR_TAG_PERL_OBJECT) |
860 | if ((*dec->cur >> 5) != 0) |
680 | { |
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 | |
681 | if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) |
876 | if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) |
682 | ERR ("corrupted CBOR data (non-array perl object)"); |
877 | ERR ("corrupted CBOR data (non-array perl object)"); |
683 | |
878 | |
684 | AV *av = (AV *)SvRV (sv); |
879 | AV *av = (AV *)SvRV (sv); |
685 | int len = av_len (av) + 1; |
880 | int len = av_len (av) + 1; |
686 | HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0); |
881 | HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0); |
687 | |
882 | |
688 | if (!stash) |
883 | if (!stash) |
689 | ERR ("cannot decode perl-object (package does not exist)"); |
884 | ERR ("cannot decode perl-object (package does not exist)"); |
690 | |
885 | |
691 | GV *method = gv_fetchmethod_autoload (stash, "THAW", 0); |
886 | GV *method = gv_fetchmethod_autoload (stash, "THAW", 0); |
692 | |
887 | |
693 | if (!method) |
888 | if (!method) |
694 | 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)"); |
695 | |
890 | |
696 | dSP; |
891 | dSP; |
697 | |
892 | |
698 | ENTER; SAVETMPS; PUSHMARK (SP); |
893 | ENTER; SAVETMPS; PUSHMARK (SP); |
699 | EXTEND (SP, len + 1); |
894 | EXTEND (SP, len + 1); |
700 | // 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 |
701 | PUSHs (*av_fetch (av, 0, 1)); |
896 | PUSHs (*av_fetch (av, 0, 1)); |
702 | PUSHs (sv_cbor); |
897 | PUSHs (sv_cbor); |
703 | |
898 | |
704 | int i; |
899 | int i; |
705 | |
900 | |
706 | for (i = 1; i < len; ++i) |
901 | for (i = 1; i < len; ++i) |
707 | PUSHs (*av_fetch (av, i, 1)); |
902 | PUSHs (*av_fetch (av, i, 1)); |
708 | |
903 | |
709 | PUTBACK; |
904 | PUTBACK; |
710 | call_sv ((SV *)GvCV (method), G_SCALAR); |
905 | call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL); |
711 | SPAGAIN; |
906 | SPAGAIN; |
712 | |
907 | |
|
|
908 | if (SvTRUE (ERRSV)) |
|
|
909 | { |
|
|
910 | FREETMPS; LEAVE; |
|
|
911 | ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); |
|
|
912 | } |
|
|
913 | |
|
|
914 | SvREFCNT_dec (sv); |
713 | sv = SvREFCNT_inc (POPs); |
915 | sv = SvREFCNT_inc (POPs); |
714 | |
916 | |
715 | PUTBACK; |
917 | PUTBACK; |
716 | |
918 | |
717 | FREETMPS; LEAVE; |
919 | FREETMPS; LEAVE; |
|
|
920 | } |
|
|
921 | break; |
718 | |
922 | |
719 | return sv; |
923 | default: |
720 | } |
924 | { |
721 | else |
925 | sv = decode_sv (dec); |
722 | { |
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 | { |
723 | AV *av = newAV (); |
950 | AV *av = newAV (); |
724 | av_push (av, newSVuv (tag)); |
951 | av_push (av, newSVuv (tag)); |
725 | av_push (av, sv); |
952 | av_push (av, sv); |
726 | |
953 | |
727 | HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash |
954 | HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash |
728 | ? cbor_tagged_stash |
955 | ? cbor_tagged_stash |
729 | : gv_stashpv ("CBOR::XS::Tagged" , 1); |
956 | : gv_stashpv ("CBOR::XS::Tagged" , 1); |
730 | |
|
|
731 | 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; |
732 | } |
965 | } |
|
|
966 | |
|
|
967 | return sv; |
733 | |
968 | |
734 | fail: |
969 | fail: |
735 | SvREFCNT_dec (sv); |
970 | SvREFCNT_dec (sv); |
736 | return &PL_sv_undef; |
971 | return &PL_sv_undef; |
737 | } |
972 | } |
… | |
… | |
828 | } |
1063 | } |
829 | |
1064 | |
830 | static SV * |
1065 | static SV * |
831 | decode_cbor (SV *string, CBOR *cbor, char **offset_return) |
1066 | decode_cbor (SV *string, CBOR *cbor, char **offset_return) |
832 | { |
1067 | { |
833 | dec_t dec; |
1068 | dec_t dec = { }; |
834 | SV *sv; |
1069 | SV *sv; |
|
|
1070 | STRLEN len; |
|
|
1071 | char *data = SvPVbyte (string, len); |
835 | |
1072 | |
836 | /* work around bugs in 5.10 where manipulating magic values |
|
|
837 | * makes perl ignore the magic in subsequent accesses. |
|
|
838 | * also make a copy of non-PV values, to get them into a clean |
|
|
839 | * state (SvPV should do that, but it's buggy, see below). |
|
|
840 | */ |
|
|
841 | /*SvGETMAGIC (string);*/ |
|
|
842 | if (SvMAGICAL (string) || !SvPOK (string)) |
|
|
843 | string = sv_2mortal (newSVsv (string)); |
|
|
844 | |
|
|
845 | SvUPGRADE (string, SVt_PV); |
|
|
846 | |
|
|
847 | /* work around a bug in perl 5.10, which causes SvCUR to fail an |
|
|
848 | * assertion with -DDEBUGGING, although SvCUR is documented to |
|
|
849 | * return the xpv_cur field which certainly exists after upgrading. |
|
|
850 | * according to nicholas clark, calling SvPOK fixes this. |
|
|
851 | * But it doesn't fix it, so try another workaround, call SvPV_nolen |
|
|
852 | * and hope for the best. |
|
|
853 | * Damnit, SvPV_nolen still trips over yet another assertion. This |
|
|
854 | * assertion business is seriously broken, try yet another workaround |
|
|
855 | * for the broken -DDEBUGGING. |
|
|
856 | */ |
|
|
857 | { |
|
|
858 | #ifdef DEBUGGING |
|
|
859 | STRLEN offset = SvOK (string) ? sv_len (string) : 0; |
|
|
860 | #else |
|
|
861 | STRLEN offset = SvCUR (string); |
|
|
862 | #endif |
|
|
863 | |
|
|
864 | if (offset > cbor->max_size && cbor->max_size) |
1073 | if (len > cbor->max_size && cbor->max_size) |
865 | 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", |
866 | (unsigned long)SvCUR (string), (unsigned long)cbor->max_size); |
1075 | (unsigned long)len, (unsigned long)cbor->max_size); |
867 | } |
|
|
868 | |
|
|
869 | sv_utf8_downgrade (string, 0); |
|
|
870 | |
1076 | |
871 | dec.cbor = *cbor; |
1077 | dec.cbor = *cbor; |
872 | dec.cur = (U8 *)SvPVX (string); |
1078 | dec.cur = (U8 *)data; |
873 | dec.end = (U8 *)SvEND (string); |
1079 | dec.end = (U8 *)data + len; |
874 | dec.err = 0; |
|
|
875 | dec.depth = 0; |
|
|
876 | |
1080 | |
877 | sv = decode_sv (&dec); |
1081 | sv = decode_sv (&dec); |
878 | |
1082 | |
879 | if (offset_return) |
1083 | if (offset_return) |
880 | *offset_return = dec.cur; |
1084 | *offset_return = dec.cur; |
… | |
… | |
884 | dec.err = "garbage after CBOR object"; |
1088 | dec.err = "garbage after CBOR object"; |
885 | |
1089 | |
886 | if (dec.err) |
1090 | if (dec.err) |
887 | { |
1091 | { |
888 | SvREFCNT_dec (sv); |
1092 | SvREFCNT_dec (sv); |
889 | 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); |
890 | } |
1094 | } |
891 | |
1095 | |
892 | sv = sv_2mortal (sv); |
1096 | sv = sv_2mortal (sv); |
893 | |
1097 | |
894 | return sv; |
1098 | return sv; |
… | |
… | |
908 | types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1); |
1112 | types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1); |
909 | |
1113 | |
910 | types_true = get_bool ("Types::Serialiser::true" ); |
1114 | types_true = get_bool ("Types::Serialiser::true" ); |
911 | types_false = get_bool ("Types::Serialiser::false"); |
1115 | types_false = get_bool ("Types::Serialiser::false"); |
912 | 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); |
913 | |
1119 | |
914 | sv_cbor = newSVpv ("CBOR", 0); |
1120 | sv_cbor = newSVpv ("CBOR", 0); |
915 | SvREADONLY_on (sv_cbor); |
1121 | SvREADONLY_on (sv_cbor); |
916 | } |
1122 | } |
917 | |
1123 | |
… | |
… | |
938 | |
1144 | |
939 | void shrink (CBOR *self, int enable = 1) |
1145 | void shrink (CBOR *self, int enable = 1) |
940 | ALIAS: |
1146 | ALIAS: |
941 | shrink = F_SHRINK |
1147 | shrink = F_SHRINK |
942 | allow_unknown = F_ALLOW_UNKNOWN |
1148 | allow_unknown = F_ALLOW_UNKNOWN |
|
|
1149 | allow_sharing = F_ALLOW_SHARING |
|
|
1150 | pack_strings = F_PACK_STRINGS |
943 | PPCODE: |
1151 | PPCODE: |
944 | { |
1152 | { |
945 | if (enable) |
1153 | if (enable) |
946 | self->flags |= ix; |
1154 | self->flags |= ix; |
947 | else |
1155 | else |
… | |
… | |
952 | |
1160 | |
953 | void get_shrink (CBOR *self) |
1161 | void get_shrink (CBOR *self) |
954 | ALIAS: |
1162 | ALIAS: |
955 | get_shrink = F_SHRINK |
1163 | get_shrink = F_SHRINK |
956 | 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 |
957 | PPCODE: |
1167 | PPCODE: |
958 | XPUSHs (boolSV (self->flags & ix)); |
1168 | XPUSHs (boolSV (self->flags & ix)); |
959 | |
1169 | |
960 | void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) |
1170 | void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) |
961 | PPCODE: |
1171 | PPCODE: |
… | |
… | |
977 | CODE: |
1187 | CODE: |
978 | RETVAL = self->max_size; |
1188 | RETVAL = self->max_size; |
979 | OUTPUT: |
1189 | OUTPUT: |
980 | RETVAL |
1190 | RETVAL |
981 | |
1191 | |
982 | #if 0 //TODO |
1192 | void filter (CBOR *self, SV *filter = 0) |
983 | |
|
|
984 | void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef) |
|
|
985 | PPCODE: |
1193 | PPCODE: |
986 | { |
|
|
987 | SvREFCNT_dec (self->cb_object); |
1194 | SvREFCNT_dec (self->filter); |
988 | self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; |
1195 | self->filter = filter ? newSVsv (filter) : filter; |
989 | |
|
|
990 | XPUSHs (ST (0)); |
1196 | XPUSHs (ST (0)); |
991 | } |
|
|
992 | |
1197 | |
993 | void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef) |
1198 | SV *get_filter (CBOR *self) |
994 | PPCODE: |
1199 | CODE: |
995 | { |
1200 | RETVAL = self->filter ? self->filter : NEWSV (0, 0); |
996 | if (!self->cb_sk_object) |
1201 | OUTPUT: |
997 | self->cb_sk_object = newHV (); |
1202 | RETVAL |
998 | |
|
|
999 | if (SvOK (cb)) |
|
|
1000 | hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0); |
|
|
1001 | else |
|
|
1002 | { |
|
|
1003 | hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0); |
|
|
1004 | |
|
|
1005 | if (!HvKEYS (self->cb_sk_object)) |
|
|
1006 | { |
|
|
1007 | SvREFCNT_dec (self->cb_sk_object); |
|
|
1008 | self->cb_sk_object = 0; |
|
|
1009 | } |
|
|
1010 | } |
|
|
1011 | |
|
|
1012 | XPUSHs (ST (0)); |
|
|
1013 | } |
|
|
1014 | |
|
|
1015 | #endif |
|
|
1016 | |
1203 | |
1017 | void encode (CBOR *self, SV *scalar) |
1204 | void encode (CBOR *self, SV *scalar) |
1018 | PPCODE: |
1205 | PPCODE: |
1019 | PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; |
1206 | PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; |
1020 | XPUSHs (scalar); |
1207 | XPUSHs (scalar); |
… | |
… | |
1033 | EXTEND (SP, 2); |
1220 | EXTEND (SP, 2); |
1034 | PUSHs (sv); |
1221 | PUSHs (sv); |
1035 | PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); |
1222 | PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); |
1036 | } |
1223 | } |
1037 | |
1224 | |
1038 | #if 0 |
|
|
1039 | |
|
|
1040 | void DESTROY (CBOR *self) |
1225 | void DESTROY (CBOR *self) |
1041 | CODE: |
1226 | PPCODE: |
1042 | SvREFCNT_dec (self->cb_sk_object); |
1227 | cbor_free (self); |
1043 | SvREFCNT_dec (self->cb_object); |
|
|
1044 | |
|
|
1045 | #endif |
|
|
1046 | |
1228 | |
1047 | PROTOTYPES: ENABLE |
1229 | PROTOTYPES: ENABLE |
1048 | |
1230 | |
1049 | void encode_cbor (SV *scalar) |
1231 | void encode_cbor (SV *scalar) |
1050 | PPCODE: |
1232 | PPCODE: |