… | |
… | |
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 | |
14 | // known tags |
25 | // known tags |
15 | enum cbor_tag |
26 | enum cbor_tag |
16 | { |
27 | { |
17 | // inofficial extensions (pending iana registration) |
28 | // inofficial extensions (pending iana registration) |
18 | CBOR_TAG_PERL_OBJECT = 256, |
29 | CBOR_TAG_PERL_OBJECT = 24, // http://cbor.schmorp.de/perl-object |
19 | CBOR_TAG_GENERIC_OBJECT = 257, |
30 | CBOR_TAG_GENERIC_OBJECT = 25, // http://cbor.schmorp.de/generic-object |
|
|
31 | CBOR_TAG_VALUE_SHAREABLE = 26, // http://cbor.schmorp.de/value-sharing |
|
|
32 | CBOR_TAG_VALUE_SHAREDREF = 27, // http://cbor.schmorp.de/value-sharing |
|
|
33 | CBOR_TAG_STRINGREF_NAMESPACE = 65537, // http://cbor.schmorp.de/stringref |
|
|
34 | CBOR_TAG_STRINGREF = 28, // http://cbor.schmorp.de/stringref |
|
|
35 | CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection |
20 | |
36 | |
21 | // rfc7049 |
37 | // rfc7049 |
22 | CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 |
38 | CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 |
23 | CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any |
39 | CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any |
24 | CBOR_TAG_POS_BIGNUM = 2, // byte string |
40 | CBOR_TAG_POS_BIGNUM = 2, // byte string |
… | |
… | |
38 | CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 |
54 | CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 |
39 | |
55 | |
40 | CBOR_TAG_MAGIC = 55799 // self-describe cbor |
56 | CBOR_TAG_MAGIC = 55799 // self-describe cbor |
41 | }; |
57 | }; |
42 | |
58 | |
43 | #define F_SHRINK 0x00000200UL |
59 | #define F_SHRINK 0x00000001UL |
44 | #define F_ALLOW_UNKNOWN 0x00002000UL |
60 | #define F_ALLOW_UNKNOWN 0x00000002UL |
|
|
61 | #define F_ALLOW_SHARING 0x00000004UL //TODO |
|
|
62 | #define F_ALLOW_STRINGREF 0x00000008UL //TODO |
45 | |
63 | |
46 | #define INIT_SIZE 32 // initial scalar size to be allocated |
64 | #define INIT_SIZE 32 // initial scalar size to be allocated |
47 | |
65 | |
48 | #define SB do { |
66 | #define SB do { |
49 | #define SE } while (0) |
67 | #define SE } while (0) |
… | |
… | |
60 | #else |
78 | #else |
61 | # define CBOR_SLOW 0 |
79 | # define CBOR_SLOW 0 |
62 | # define CBOR_STASH cbor_stash |
80 | # define CBOR_STASH cbor_stash |
63 | #endif |
81 | #endif |
64 | |
82 | |
65 | static HV *cbor_stash, *cbor_boolean_stash, *cbor_tagged_stash; // CBOR::XS:: |
83 | static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS:: |
66 | static SV *cbor_true, *cbor_false; |
84 | static SV *types_true, *types_false, *types_error, *sv_cbor; |
67 | |
85 | |
68 | typedef struct { |
86 | typedef struct { |
69 | U32 flags; |
87 | U32 flags; |
70 | U32 max_depth; |
88 | U32 max_depth; |
71 | STRLEN max_size; |
89 | STRLEN max_size; |
… | |
… | |
105 | SvPV_renew (sv, SvCUR (sv) + 1); |
123 | SvPV_renew (sv, SvCUR (sv) + 1); |
106 | #endif |
124 | #endif |
107 | } |
125 | } |
108 | } |
126 | } |
109 | |
127 | |
110 | ///////////////////////////////////////////////////////////////////////////// |
128 | // minimum length of a string to be registered for stringref |
111 | // fp hell |
129 | ecb_inline int |
112 | |
130 | minimum_string_length (UV idx) |
113 | //TODO |
131 | { |
|
|
132 | return idx > 23 |
|
|
133 | ? idx > 0xffU |
|
|
134 | ? idx > 0xffffU |
|
|
135 | ? idx > 0xffffffffU |
|
|
136 | ? 7 |
|
|
137 | : 6 |
|
|
138 | : 5 |
|
|
139 | : 4 |
|
|
140 | : 3; |
|
|
141 | } |
114 | |
142 | |
115 | ///////////////////////////////////////////////////////////////////////////// |
143 | ///////////////////////////////////////////////////////////////////////////// |
116 | // encoder |
144 | // encoder |
117 | |
145 | |
118 | // structure used for encoding CBOR |
146 | // structure used for encoding CBOR |
… | |
… | |
121 | char *cur; // SvPVX (sv) + current output position |
149 | char *cur; // SvPVX (sv) + current output position |
122 | char *end; // SvEND (sv) |
150 | char *end; // SvEND (sv) |
123 | SV *sv; // result scalar |
151 | SV *sv; // result scalar |
124 | CBOR cbor; |
152 | CBOR cbor; |
125 | U32 depth; // recursion level |
153 | U32 depth; // recursion level |
|
|
154 | HV *stringref[2]; // string => index, or 0 ([0] = bytes, [1] = utf-8) |
|
|
155 | UV stringref_idx; |
|
|
156 | HV *shareable; // ptr => index, or 0 |
|
|
157 | UV shareable_idx; |
126 | } enc_t; |
158 | } enc_t; |
127 | |
159 | |
128 | ecb_inline void |
160 | ecb_inline void |
129 | need (enc_t *enc, STRLEN len) |
161 | need (enc_t *enc, STRLEN len) |
130 | { |
162 | { |
… | |
… | |
182 | *enc->cur++ = len >> 8; |
214 | *enc->cur++ = len >> 8; |
183 | *enc->cur++ = len; |
215 | *enc->cur++ = len; |
184 | } |
216 | } |
185 | } |
217 | } |
186 | |
218 | |
|
|
219 | ecb_inline void |
|
|
220 | encode_tag (enc_t *enc, UV tag) |
|
|
221 | { |
|
|
222 | encode_uint (enc, 0xc0, tag); |
|
|
223 | } |
|
|
224 | |
187 | static void |
225 | static void |
188 | encode_str (enc_t *enc, int utf8, char *str, STRLEN len) |
226 | encode_str (enc_t *enc, int utf8, char *str, STRLEN len) |
189 | { |
227 | { |
|
|
228 | if (ecb_expect_false (enc->cbor.flags & F_ALLOW_STRINGREF)) |
|
|
229 | { |
|
|
230 | SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1); |
|
|
231 | |
|
|
232 | if (SvOK (*svp)) |
|
|
233 | { |
|
|
234 | // already registered, use stringref |
|
|
235 | encode_tag (enc, CBOR_TAG_STRINGREF); |
|
|
236 | encode_uint (enc, 0x00, SvUV (*svp)); |
|
|
237 | return; |
|
|
238 | } |
|
|
239 | else if (len >= minimum_string_length (enc->stringref_idx)) |
|
|
240 | { |
|
|
241 | // register only |
|
|
242 | sv_setuv (*svp, enc->stringref_idx); |
|
|
243 | ++enc->stringref_idx; |
|
|
244 | } |
|
|
245 | } |
|
|
246 | |
190 | encode_uint (enc, utf8 ? 0x60 : 0x40, len); |
247 | encode_uint (enc, utf8 ? 0x60 : 0x40, len); |
191 | need (enc, len); |
248 | need (enc, len); |
192 | memcpy (enc->cur, str, len); |
249 | memcpy (enc->cur, str, len); |
193 | enc->cur += len; |
250 | enc->cur += len; |
194 | } |
251 | } |
… | |
… | |
252 | |
309 | |
253 | // encode objects, arrays and special \0=false and \1=true values. |
310 | // encode objects, arrays and special \0=false and \1=true values. |
254 | static void |
311 | static void |
255 | encode_rv (enc_t *enc, SV *sv) |
312 | encode_rv (enc_t *enc, SV *sv) |
256 | { |
313 | { |
257 | svtype svt; |
|
|
258 | |
|
|
259 | SvGETMAGIC (sv); |
314 | SvGETMAGIC (sv); |
|
|
315 | |
|
|
316 | if (ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING) |
|
|
317 | && ecb_expect_false (SvREFCNT (sv) > 1)) |
|
|
318 | { |
|
|
319 | if (!enc->shareable) |
|
|
320 | enc->shareable = (HV *)sv_2mortal ((SV *)newHV ()); |
|
|
321 | |
|
|
322 | SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1); |
|
|
323 | |
|
|
324 | if (SvOK (*svp)) |
|
|
325 | { |
|
|
326 | encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF); |
|
|
327 | encode_uint (enc, 0x00, SvUV (*svp)); |
|
|
328 | return; |
|
|
329 | } |
|
|
330 | else |
|
|
331 | { |
|
|
332 | sv_setuv (*svp, enc->shareable_idx); |
|
|
333 | ++enc->shareable_idx; |
|
|
334 | encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE); |
|
|
335 | } |
|
|
336 | } |
|
|
337 | |
260 | svt = SvTYPE (sv); |
338 | svtype svt = SvTYPE (sv); |
261 | |
339 | |
262 | if (ecb_expect_false (SvOBJECT (sv))) |
340 | if (ecb_expect_false (SvOBJECT (sv))) |
263 | { |
341 | { |
264 | HV *boolean_stash = !CBOR_SLOW || cbor_boolean_stash |
342 | HV *boolean_stash = !CBOR_SLOW || types_boolean_stash |
265 | ? cbor_boolean_stash |
343 | ? types_boolean_stash |
266 | : gv_stashpv ("CBOR::XS::Boolean", 1); |
344 | : gv_stashpv ("Types::Serialiser::Boolean", 1); |
|
|
345 | HV *error_stash = !CBOR_SLOW || types_error_stash |
|
|
346 | ? types_error_stash |
|
|
347 | : gv_stashpv ("Types::Serialiser::Error", 1); |
267 | HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash |
348 | HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash |
268 | ? cbor_tagged_stash |
349 | ? cbor_tagged_stash |
269 | : gv_stashpv ("CBOR::XS::Tagged" , 1); |
350 | : gv_stashpv ("CBOR::XS::Tagged" , 1); |
270 | |
351 | |
|
|
352 | HV *stash = SvSTASH (sv); |
|
|
353 | GV *method; |
|
|
354 | |
271 | if (SvSTASH (sv) == boolean_stash) |
355 | if (stash == boolean_stash) |
272 | encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); |
356 | encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); |
|
|
357 | else if (stash == error_stash) |
|
|
358 | encode_ch (enc, 0xe0 | 23); |
273 | else if (SvSTASH (sv) == tagged_stash) |
359 | else if (stash == tagged_stash) |
274 | { |
360 | { |
275 | if (svt != SVt_PVAV) |
361 | if (svt != SVt_PVAV) |
276 | croak ("encountered CBOR::XS::Tagged object that isn't an array"); |
362 | croak ("encountered CBOR::XS::Tagged object that isn't an array"); |
277 | |
363 | |
278 | encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); |
364 | encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); |
279 | encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); |
365 | encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); |
280 | } |
366 | } |
|
|
367 | else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) |
|
|
368 | { |
|
|
369 | dSP; |
|
|
370 | |
|
|
371 | ENTER; SAVETMPS; PUSHMARK (SP); |
|
|
372 | // we re-bless the reference to get overload and other niceties right |
|
|
373 | XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); |
|
|
374 | |
|
|
375 | PUTBACK; |
|
|
376 | // G_SCALAR ensures that return value is 1 |
|
|
377 | call_sv ((SV *)GvCV (method), G_SCALAR); |
|
|
378 | SPAGAIN; |
|
|
379 | |
|
|
380 | // catch this surprisingly common error |
|
|
381 | if (SvROK (TOPs) && SvRV (TOPs) == sv) |
|
|
382 | croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (stash)); |
|
|
383 | |
|
|
384 | encode_sv (enc, POPs); |
|
|
385 | |
|
|
386 | PUTBACK; |
|
|
387 | |
|
|
388 | FREETMPS; LEAVE; |
|
|
389 | } |
|
|
390 | else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) |
|
|
391 | { |
|
|
392 | dSP; |
|
|
393 | |
|
|
394 | ENTER; SAVETMPS; PUSHMARK (SP); |
|
|
395 | EXTEND (SP, 2); |
|
|
396 | // we re-bless the reference to get overload and other niceties right |
|
|
397 | PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); |
|
|
398 | PUSHs (sv_cbor); |
|
|
399 | |
|
|
400 | PUTBACK; |
|
|
401 | int count = call_sv ((SV *)GvCV (method), G_ARRAY); |
|
|
402 | SPAGAIN; |
|
|
403 | |
|
|
404 | // catch this surprisingly common error |
|
|
405 | if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) |
|
|
406 | croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); |
|
|
407 | |
|
|
408 | encode_tag (enc, CBOR_TAG_PERL_OBJECT); |
|
|
409 | encode_uint (enc, 0x80, count + 1); |
|
|
410 | encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); |
|
|
411 | |
|
|
412 | while (count) |
|
|
413 | encode_sv (enc, SP[1 - count--]); |
|
|
414 | |
|
|
415 | PUTBACK; |
|
|
416 | |
|
|
417 | FREETMPS; LEAVE; |
|
|
418 | } |
281 | else |
419 | else |
282 | { |
|
|
283 | // we re-bless the reference to get overload and other niceties right |
|
|
284 | GV *to_cbor = gv_fetchmethod_autoload (SvSTASH (sv), "TO_CBOR", 0); |
|
|
285 | |
|
|
286 | if (to_cbor) |
|
|
287 | { |
|
|
288 | dSP; |
|
|
289 | |
|
|
290 | ENTER; SAVETMPS; PUSHMARK (SP); |
|
|
291 | XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv))); |
|
|
292 | |
|
|
293 | // calling with G_SCALAR ensures that we always get a 1 return value |
|
|
294 | PUTBACK; |
|
|
295 | call_sv ((SV *)GvCV (to_cbor), G_SCALAR); |
|
|
296 | SPAGAIN; |
|
|
297 | |
|
|
298 | // catch this surprisingly common error |
|
|
299 | if (SvROK (TOPs) && SvRV (TOPs) == sv) |
|
|
300 | croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv))); |
|
|
301 | |
|
|
302 | sv = POPs; |
|
|
303 | PUTBACK; |
|
|
304 | |
|
|
305 | encode_sv (enc, sv); |
|
|
306 | |
|
|
307 | FREETMPS; LEAVE; |
|
|
308 | } |
|
|
309 | else |
|
|
310 | croak ("encountered object '%s', but no TO_CBOR method available on it", |
420 | croak ("encountered object '%s', but no TO_CBOR or FREEZE methods available on it", |
311 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
421 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
312 | } |
|
|
313 | } |
422 | } |
314 | else if (svt == SVt_PVHV) |
423 | else if (svt == SVt_PVHV) |
315 | encode_hv (enc, (HV *)sv); |
424 | encode_hv (enc, (HV *)sv); |
316 | else if (svt == SVt_PVAV) |
425 | else if (svt == SVt_PVAV) |
317 | encode_av (enc, (AV *)sv); |
426 | encode_av (enc, (AV *)sv); |
318 | else if (svt < SVt_PVAV) |
|
|
319 | { |
|
|
320 | STRLEN len = 0; |
|
|
321 | char *pv = svt ? SvPV (sv, len) : 0; |
|
|
322 | |
|
|
323 | if (len == 1 && *pv == '1') |
|
|
324 | encode_ch (enc, 0xe0 | 21); |
|
|
325 | else if (len == 1 && *pv == '0') |
|
|
326 | encode_ch (enc, 0xe0 | 20); |
|
|
327 | else if (enc->cbor.flags & F_ALLOW_UNKNOWN) |
|
|
328 | encode_ch (enc, 0xe0 | 23); |
|
|
329 | else |
|
|
330 | croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", |
|
|
331 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
|
|
332 | } |
|
|
333 | else if (enc->cbor.flags & F_ALLOW_UNKNOWN) |
|
|
334 | encode_ch (enc, 0xe0 | 23); |
|
|
335 | else |
427 | else |
336 | croak ("encountered %s, but CBOR can only represent references to arrays or hashes", |
428 | { |
337 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
429 | encode_tag (enc, CBOR_TAG_INDIRECTION); |
|
|
430 | encode_sv (enc, sv); |
|
|
431 | } |
338 | } |
432 | } |
339 | |
433 | |
340 | static void |
434 | static void |
341 | encode_nv (enc_t *enc, SV *sv) |
435 | encode_nv (enc_t *enc, SV *sv) |
342 | { |
436 | { |
… | |
… | |
407 | } |
501 | } |
408 | |
502 | |
409 | static SV * |
503 | static SV * |
410 | encode_cbor (SV *scalar, CBOR *cbor) |
504 | encode_cbor (SV *scalar, CBOR *cbor) |
411 | { |
505 | { |
412 | enc_t enc; |
506 | enc_t enc = { }; |
413 | |
507 | |
414 | enc.cbor = *cbor; |
508 | enc.cbor = *cbor; |
415 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
509 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
416 | enc.cur = SvPVX (enc.sv); |
510 | enc.cur = SvPVX (enc.sv); |
417 | enc.end = SvEND (enc.sv); |
511 | enc.end = SvEND (enc.sv); |
418 | enc.depth = 0; |
|
|
419 | |
512 | |
420 | SvPOK_only (enc.sv); |
513 | SvPOK_only (enc.sv); |
|
|
514 | |
|
|
515 | if (cbor->flags & F_ALLOW_STRINGREF) |
|
|
516 | { |
|
|
517 | encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE); |
|
|
518 | enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ()); |
|
|
519 | enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ()); |
|
|
520 | } |
|
|
521 | |
421 | encode_sv (&enc, scalar); |
522 | encode_sv (&enc, scalar); |
422 | |
523 | |
423 | SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); |
524 | SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); |
424 | *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings |
525 | *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings |
425 | |
526 | |
… | |
… | |
439 | U8 *end; // end of input string |
540 | U8 *end; // end of input string |
440 | const char *err; // parse error, if != 0 |
541 | const char *err; // parse error, if != 0 |
441 | CBOR cbor; |
542 | CBOR cbor; |
442 | U32 depth; // recursion depth |
543 | U32 depth; // recursion depth |
443 | U32 maxdepth; // recursion depth limit |
544 | U32 maxdepth; // recursion depth limit |
|
|
545 | AV *shareable; |
|
|
546 | AV *stringref; |
444 | } dec_t; |
547 | } dec_t; |
445 | |
548 | |
446 | #define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE |
549 | #define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE |
447 | |
550 | |
448 | #define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") |
551 | #define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") |
… | |
… | |
542 | SvREFCNT_dec (av); |
645 | SvREFCNT_dec (av); |
543 | DEC_DEC_DEPTH; |
646 | DEC_DEC_DEPTH; |
544 | return &PL_sv_undef; |
647 | return &PL_sv_undef; |
545 | } |
648 | } |
546 | |
649 | |
|
|
650 | static void |
|
|
651 | decode_he (dec_t *dec, HV *hv) |
|
|
652 | { |
|
|
653 | // for speed reasons, we specialcase single-string |
|
|
654 | // byte or utf-8 strings as keys, but only when !stringref |
|
|
655 | |
|
|
656 | if (ecb_expect_true (!dec->stringref)) |
|
|
657 | if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27) |
|
|
658 | { |
|
|
659 | I32 len = decode_uint (dec); |
|
|
660 | char *key = (char *)dec->cur; |
|
|
661 | |
|
|
662 | dec->cur += len; |
|
|
663 | |
|
|
664 | if (ecb_expect_false (dec->stringref)) |
|
|
665 | av_push (dec->stringref, newSVpvn (key, len)); |
|
|
666 | |
|
|
667 | hv_store (hv, key, len, decode_sv (dec), 0); |
|
|
668 | |
|
|
669 | return; |
|
|
670 | } |
|
|
671 | else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27) |
|
|
672 | { |
|
|
673 | I32 len = decode_uint (dec); |
|
|
674 | char *key = (char *)dec->cur; |
|
|
675 | |
|
|
676 | dec->cur += len; |
|
|
677 | |
|
|
678 | if (ecb_expect_false (dec->stringref)) |
|
|
679 | av_push (dec->stringref, newSVpvn_utf8 (key, len, 1)); |
|
|
680 | |
|
|
681 | hv_store (hv, key, -len, decode_sv (dec), 0); |
|
|
682 | |
|
|
683 | return; |
|
|
684 | } |
|
|
685 | |
|
|
686 | SV *k = decode_sv (dec); |
|
|
687 | SV *v = decode_sv (dec); |
|
|
688 | |
|
|
689 | hv_store_ent (hv, k, v, 0); |
|
|
690 | SvREFCNT_dec (k); |
|
|
691 | } |
|
|
692 | |
547 | static SV * |
693 | static SV * |
548 | decode_hv (dec_t *dec) |
694 | decode_hv (dec_t *dec) |
549 | { |
695 | { |
550 | HV *hv = newHV (); |
696 | HV *hv = newHV (); |
551 | |
697 | |
… | |
… | |
563 | { |
709 | { |
564 | ++dec->cur; |
710 | ++dec->cur; |
565 | break; |
711 | break; |
566 | } |
712 | } |
567 | |
713 | |
568 | SV *k = decode_sv (dec); |
714 | decode_he (dec, hv); |
569 | SV *v = decode_sv (dec); |
|
|
570 | |
|
|
571 | hv_store_ent (hv, k, v, 0); |
|
|
572 | } |
715 | } |
573 | } |
716 | } |
574 | else |
717 | else |
575 | { |
718 | { |
576 | int len = decode_uint (dec); |
719 | int pairs = decode_uint (dec); |
577 | |
720 | |
578 | while (len--) |
721 | while (pairs--) |
579 | { |
722 | decode_he (dec, hv); |
580 | SV *k = decode_sv (dec); |
|
|
581 | SV *v = decode_sv (dec); |
|
|
582 | |
|
|
583 | hv_store_ent (hv, k, v, 0); |
|
|
584 | } |
|
|
585 | } |
723 | } |
586 | |
724 | |
587 | DEC_DEC_DEPTH; |
725 | DEC_DEC_DEPTH; |
588 | return newRV_noinc ((SV *)hv); |
726 | return newRV_noinc ((SV *)hv); |
589 | |
727 | |
… | |
… | |
623 | STRLEN len = decode_uint (dec); |
761 | STRLEN len = decode_uint (dec); |
624 | |
762 | |
625 | WANT (len); |
763 | WANT (len); |
626 | sv = newSVpvn (dec->cur, len); |
764 | sv = newSVpvn (dec->cur, len); |
627 | dec->cur += len; |
765 | dec->cur += len; |
|
|
766 | |
|
|
767 | if (ecb_expect_false (dec->stringref) |
|
|
768 | && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1)) |
|
|
769 | av_push (dec->stringref, SvREFCNT_inc_NN (sv)); |
628 | } |
770 | } |
629 | |
771 | |
630 | if (utf8) |
772 | if (utf8) |
631 | SvUTF8_on (sv); |
773 | SvUTF8_on (sv); |
632 | |
774 | |
… | |
… | |
638 | } |
780 | } |
639 | |
781 | |
640 | static SV * |
782 | static SV * |
641 | decode_tagged (dec_t *dec) |
783 | decode_tagged (dec_t *dec) |
642 | { |
784 | { |
|
|
785 | SV *sv = 0; |
643 | UV tag = decode_uint (dec); |
786 | UV tag = decode_uint (dec); |
|
|
787 | |
|
|
788 | WANT (1); |
|
|
789 | |
|
|
790 | switch (tag) |
|
|
791 | { |
|
|
792 | case CBOR_TAG_MAGIC: |
|
|
793 | sv = decode_sv (dec); |
|
|
794 | break; |
|
|
795 | |
|
|
796 | case CBOR_TAG_INDIRECTION: |
|
|
797 | sv = newRV_noinc (decode_sv (dec)); |
|
|
798 | break; |
|
|
799 | |
|
|
800 | case CBOR_TAG_STRINGREF_NAMESPACE: |
|
|
801 | { |
|
|
802 | ENTER; SAVETMPS; |
|
|
803 | |
|
|
804 | SAVESPTR (dec->stringref); |
|
|
805 | dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); |
|
|
806 | |
|
|
807 | sv = decode_sv (dec); |
|
|
808 | |
|
|
809 | FREETMPS; LEAVE; |
|
|
810 | } |
|
|
811 | break; |
|
|
812 | |
|
|
813 | case CBOR_TAG_STRINGREF: |
|
|
814 | { |
|
|
815 | if ((*dec->cur >> 5) != 0) |
|
|
816 | ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); |
|
|
817 | |
|
|
818 | UV idx = decode_uint (dec); |
|
|
819 | |
|
|
820 | if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) |
|
|
821 | ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)"); |
|
|
822 | |
|
|
823 | sv = newSVsv (AvARRAY (dec->stringref)[idx]); |
|
|
824 | } |
|
|
825 | break; |
|
|
826 | |
|
|
827 | case CBOR_TAG_VALUE_SHAREABLE: |
|
|
828 | { |
|
|
829 | if (ecb_expect_false (!dec->shareable)) |
|
|
830 | dec->shareable = (AV *)sv_2mortal ((SV *)newAV ()); |
|
|
831 | |
|
|
832 | sv = newSV (0); |
|
|
833 | av_push (dec->shareable, SvREFCNT_inc_NN (sv)); |
|
|
834 | |
644 | SV *sv = decode_sv (dec); |
835 | SV *osv = decode_sv (dec); |
|
|
836 | sv_setsv (sv, osv); |
|
|
837 | SvREFCNT_dec_NN (osv); |
|
|
838 | } |
|
|
839 | break; |
645 | |
840 | |
646 | if (tag == CBOR_TAG_MAGIC) |
841 | case CBOR_TAG_VALUE_SHAREDREF: |
647 | return sv; |
842 | { |
|
|
843 | if ((*dec->cur >> 5) != 0) |
|
|
844 | ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); |
648 | |
845 | |
649 | if (tag == CBOR_TAG_PERL_OBJECT) |
846 | UV idx = decode_uint (dec); |
650 | { |
847 | |
|
|
848 | if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) |
|
|
849 | ERR ("corrupted CBOR data (sharedref index out of bounds)"); |
|
|
850 | |
|
|
851 | sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); |
|
|
852 | } |
|
|
853 | break; |
|
|
854 | |
|
|
855 | case CBOR_TAG_PERL_OBJECT: |
|
|
856 | { |
|
|
857 | sv = decode_sv (dec); |
|
|
858 | |
651 | if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) |
859 | if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) |
652 | ERR ("corrupted CBOR data (non-array perl object)"); |
860 | ERR ("corrupted CBOR data (non-array perl object)"); |
|
|
861 | |
|
|
862 | AV *av = (AV *)SvRV (sv); |
|
|
863 | int len = av_len (av) + 1; |
|
|
864 | HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0); |
|
|
865 | |
|
|
866 | if (!stash) |
|
|
867 | ERR ("cannot decode perl-object (package does not exist)"); |
|
|
868 | |
|
|
869 | GV *method = gv_fetchmethod_autoload (stash, "THAW", 0); |
|
|
870 | |
|
|
871 | if (!method) |
|
|
872 | ERR ("cannot decode perl-object (package does not have a THAW method)"); |
|
|
873 | |
|
|
874 | dSP; |
|
|
875 | |
|
|
876 | ENTER; SAVETMPS; PUSHMARK (SP); |
|
|
877 | EXTEND (SP, len + 1); |
|
|
878 | // we re-bless the reference to get overload and other niceties right |
|
|
879 | PUSHs (*av_fetch (av, 0, 1)); |
|
|
880 | PUSHs (sv_cbor); |
|
|
881 | |
|
|
882 | int i; |
|
|
883 | |
|
|
884 | for (i = 1; i < len; ++i) |
|
|
885 | PUSHs (*av_fetch (av, i, 1)); |
|
|
886 | |
|
|
887 | PUTBACK; |
|
|
888 | call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL); |
|
|
889 | SPAGAIN; |
|
|
890 | |
|
|
891 | if (SvTRUE (ERRSV)) |
|
|
892 | { |
|
|
893 | FREETMPS; LEAVE; |
|
|
894 | ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); |
|
|
895 | } |
|
|
896 | |
|
|
897 | SvREFCNT_dec (sv); |
|
|
898 | sv = SvREFCNT_inc (POPs); |
|
|
899 | |
|
|
900 | PUTBACK; |
|
|
901 | |
|
|
902 | FREETMPS; LEAVE; |
653 | |
903 | } |
654 | // TODO |
904 | break; |
655 | } |
|
|
656 | |
905 | |
|
|
906 | default: |
|
|
907 | { |
|
|
908 | sv = decode_sv (dec); |
|
|
909 | |
657 | AV *av = newAV (); |
910 | AV *av = newAV (); |
658 | av_push (av, newSVuv (tag)); |
911 | av_push (av, newSVuv (tag)); |
659 | av_push (av, sv); |
912 | av_push (av, sv); |
660 | |
913 | |
661 | HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash |
914 | HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash |
662 | ? cbor_tagged_stash |
915 | ? cbor_tagged_stash |
663 | : gv_stashpv ("CBOR::XS::Tagged" , 1); |
916 | : gv_stashpv ("CBOR::XS::Tagged" , 1); |
664 | |
917 | |
665 | return sv_bless (newRV_noinc ((SV *)av), tagged_stash); |
918 | sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash); |
|
|
919 | } |
|
|
920 | break; |
|
|
921 | } |
|
|
922 | |
|
|
923 | return sv; |
666 | |
924 | |
667 | fail: |
925 | fail: |
668 | SvREFCNT_dec (sv); |
926 | SvREFCNT_dec (sv); |
669 | return &PL_sv_undef; |
927 | return &PL_sv_undef; |
670 | } |
928 | } |
… | |
… | |
693 | case 7: // misc |
951 | case 7: // misc |
694 | switch (*dec->cur++ & 31) |
952 | switch (*dec->cur++ & 31) |
695 | { |
953 | { |
696 | case 20: |
954 | case 20: |
697 | #if CBOR_SLOW |
955 | #if CBOR_SLOW |
698 | cbor_false = get_bool ("CBOR::XS::false"); |
956 | types_false = get_bool ("Types::Serialiser::false"); |
699 | #endif |
957 | #endif |
700 | return newSVsv (cbor_false); |
958 | return newSVsv (types_false); |
701 | case 21: |
959 | case 21: |
702 | #if CBOR_SLOW |
960 | #if CBOR_SLOW |
703 | cbor_true = get_bool ("CBOR::XS::true"); |
961 | types_true = get_bool ("Types::Serialiser::true"); |
704 | #endif |
962 | #endif |
705 | return newSVsv (cbor_true); |
963 | return newSVsv (types_true); |
706 | case 22: |
964 | case 22: |
707 | return newSVsv (&PL_sv_undef); |
965 | return newSVsv (&PL_sv_undef); |
|
|
966 | case 23: |
|
|
967 | #if CBOR_SLOW |
|
|
968 | types_error = get_bool ("Types::Serialiser::error"); |
|
|
969 | #endif |
|
|
970 | return newSVsv (types_error); |
708 | |
971 | |
709 | case 25: |
972 | case 25: |
710 | { |
973 | { |
711 | WANT (2); |
974 | WANT (2); |
712 | |
975 | |
… | |
… | |
756 | } |
1019 | } |
757 | |
1020 | |
758 | static SV * |
1021 | static SV * |
759 | decode_cbor (SV *string, CBOR *cbor, char **offset_return) |
1022 | decode_cbor (SV *string, CBOR *cbor, char **offset_return) |
760 | { |
1023 | { |
761 | dec_t dec; |
1024 | dec_t dec = { }; |
762 | SV *sv; |
1025 | SV *sv; |
|
|
1026 | STRLEN len; |
|
|
1027 | char *data = SvPVbyte (string, len); |
763 | |
1028 | |
764 | /* work around bugs in 5.10 where manipulating magic values |
|
|
765 | * makes perl ignore the magic in subsequent accesses. |
|
|
766 | * also make a copy of non-PV values, to get them into a clean |
|
|
767 | * state (SvPV should do that, but it's buggy, see below). |
|
|
768 | */ |
|
|
769 | /*SvGETMAGIC (string);*/ |
|
|
770 | if (SvMAGICAL (string) || !SvPOK (string)) |
|
|
771 | string = sv_2mortal (newSVsv (string)); |
|
|
772 | |
|
|
773 | SvUPGRADE (string, SVt_PV); |
|
|
774 | |
|
|
775 | /* work around a bug in perl 5.10, which causes SvCUR to fail an |
|
|
776 | * assertion with -DDEBUGGING, although SvCUR is documented to |
|
|
777 | * return the xpv_cur field which certainly exists after upgrading. |
|
|
778 | * according to nicholas clark, calling SvPOK fixes this. |
|
|
779 | * But it doesn't fix it, so try another workaround, call SvPV_nolen |
|
|
780 | * and hope for the best. |
|
|
781 | * Damnit, SvPV_nolen still trips over yet another assertion. This |
|
|
782 | * assertion business is seriously broken, try yet another workaround |
|
|
783 | * for the broken -DDEBUGGING. |
|
|
784 | */ |
|
|
785 | { |
|
|
786 | #ifdef DEBUGGING |
|
|
787 | STRLEN offset = SvOK (string) ? sv_len (string) : 0; |
|
|
788 | #else |
|
|
789 | STRLEN offset = SvCUR (string); |
|
|
790 | #endif |
|
|
791 | |
|
|
792 | if (offset > cbor->max_size && cbor->max_size) |
1029 | if (len > cbor->max_size && cbor->max_size) |
793 | croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu", |
1030 | croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu", |
794 | (unsigned long)SvCUR (string), (unsigned long)cbor->max_size); |
1031 | (unsigned long)len, (unsigned long)cbor->max_size); |
795 | } |
|
|
796 | |
|
|
797 | sv_utf8_downgrade (string, 0); |
|
|
798 | |
1032 | |
799 | dec.cbor = *cbor; |
1033 | dec.cbor = *cbor; |
800 | dec.cur = (U8 *)SvPVX (string); |
1034 | dec.cur = (U8 *)data; |
801 | dec.end = (U8 *)SvEND (string); |
1035 | dec.end = (U8 *)data + len; |
802 | dec.err = 0; |
|
|
803 | dec.depth = 0; |
|
|
804 | |
1036 | |
805 | sv = decode_sv (&dec); |
1037 | sv = decode_sv (&dec); |
806 | |
1038 | |
807 | if (offset_return) |
1039 | if (offset_return) |
808 | *offset_return = dec.cur; |
1040 | *offset_return = dec.cur; |
… | |
… | |
812 | dec.err = "garbage after CBOR object"; |
1044 | dec.err = "garbage after CBOR object"; |
813 | |
1045 | |
814 | if (dec.err) |
1046 | if (dec.err) |
815 | { |
1047 | { |
816 | SvREFCNT_dec (sv); |
1048 | SvREFCNT_dec (sv); |
817 | croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)SvPVX (string), (int)(uint8_t)*dec.cur); |
1049 | croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); |
818 | } |
1050 | } |
819 | |
1051 | |
820 | sv = sv_2mortal (sv); |
1052 | sv = sv_2mortal (sv); |
821 | |
1053 | |
822 | return sv; |
1054 | return sv; |
… | |
… | |
828 | MODULE = CBOR::XS PACKAGE = CBOR::XS |
1060 | MODULE = CBOR::XS PACKAGE = CBOR::XS |
829 | |
1061 | |
830 | BOOT: |
1062 | BOOT: |
831 | { |
1063 | { |
832 | cbor_stash = gv_stashpv ("CBOR::XS" , 1); |
1064 | cbor_stash = gv_stashpv ("CBOR::XS" , 1); |
833 | cbor_boolean_stash = gv_stashpv ("CBOR::XS::Boolean", 1); |
|
|
834 | cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1); |
1065 | cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1); |
835 | |
1066 | |
836 | cbor_true = get_bool ("CBOR::XS::true"); |
1067 | types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1); |
837 | cbor_false = get_bool ("CBOR::XS::false"); |
1068 | types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1); |
|
|
1069 | |
|
|
1070 | types_true = get_bool ("Types::Serialiser::true" ); |
|
|
1071 | types_false = get_bool ("Types::Serialiser::false"); |
|
|
1072 | types_error = get_bool ("Types::Serialiser::error"); |
|
|
1073 | |
|
|
1074 | sv_cbor = newSVpv ("CBOR", 0); |
|
|
1075 | SvREADONLY_on (sv_cbor); |
838 | } |
1076 | } |
839 | |
1077 | |
840 | PROTOTYPES: DISABLE |
1078 | PROTOTYPES: DISABLE |
841 | |
1079 | |
842 | void CLONE (...) |
1080 | void CLONE (...) |
843 | CODE: |
1081 | CODE: |
844 | cbor_stash = 0; |
1082 | cbor_stash = 0; |
845 | cbor_boolean_stash = 0; |
|
|
846 | cbor_tagged_stash = 0; |
1083 | cbor_tagged_stash = 0; |
|
|
1084 | types_error_stash = 0; |
|
|
1085 | types_boolean_stash = 0; |
847 | |
1086 | |
848 | void new (char *klass) |
1087 | void new (char *klass) |
849 | PPCODE: |
1088 | PPCODE: |
850 | { |
1089 | { |
851 | SV *pv = NEWSV (0, sizeof (CBOR)); |
1090 | SV *pv = NEWSV (0, sizeof (CBOR)); |
… | |
… | |
859 | |
1098 | |
860 | void shrink (CBOR *self, int enable = 1) |
1099 | void shrink (CBOR *self, int enable = 1) |
861 | ALIAS: |
1100 | ALIAS: |
862 | shrink = F_SHRINK |
1101 | shrink = F_SHRINK |
863 | allow_unknown = F_ALLOW_UNKNOWN |
1102 | allow_unknown = F_ALLOW_UNKNOWN |
|
|
1103 | allow_sharing = F_ALLOW_SHARING |
|
|
1104 | allow_stringref = F_ALLOW_STRINGREF |
864 | PPCODE: |
1105 | PPCODE: |
865 | { |
1106 | { |
866 | if (enable) |
1107 | if (enable) |
867 | self->flags |= ix; |
1108 | self->flags |= ix; |
868 | else |
1109 | else |
… | |
… | |
873 | |
1114 | |
874 | void get_shrink (CBOR *self) |
1115 | void get_shrink (CBOR *self) |
875 | ALIAS: |
1116 | ALIAS: |
876 | get_shrink = F_SHRINK |
1117 | get_shrink = F_SHRINK |
877 | get_allow_unknown = F_ALLOW_UNKNOWN |
1118 | get_allow_unknown = F_ALLOW_UNKNOWN |
|
|
1119 | get_allow_sharing = F_ALLOW_SHARING |
|
|
1120 | get_allow_stringref = F_ALLOW_STRINGREF |
878 | PPCODE: |
1121 | PPCODE: |
879 | XPUSHs (boolSV (self->flags & ix)); |
1122 | XPUSHs (boolSV (self->flags & ix)); |
880 | |
1123 | |
881 | void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) |
1124 | void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) |
882 | PPCODE: |
1125 | PPCODE: |
… | |
… | |
897 | int get_max_size (CBOR *self) |
1140 | int get_max_size (CBOR *self) |
898 | CODE: |
1141 | CODE: |
899 | RETVAL = self->max_size; |
1142 | RETVAL = self->max_size; |
900 | OUTPUT: |
1143 | OUTPUT: |
901 | RETVAL |
1144 | RETVAL |
902 | |
|
|
903 | #if 0 //TODO |
|
|
904 | |
|
|
905 | void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef) |
|
|
906 | PPCODE: |
|
|
907 | { |
|
|
908 | SvREFCNT_dec (self->cb_object); |
|
|
909 | self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; |
|
|
910 | |
|
|
911 | XPUSHs (ST (0)); |
|
|
912 | } |
|
|
913 | |
|
|
914 | void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef) |
|
|
915 | PPCODE: |
|
|
916 | { |
|
|
917 | if (!self->cb_sk_object) |
|
|
918 | self->cb_sk_object = newHV (); |
|
|
919 | |
|
|
920 | if (SvOK (cb)) |
|
|
921 | hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0); |
|
|
922 | else |
|
|
923 | { |
|
|
924 | hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0); |
|
|
925 | |
|
|
926 | if (!HvKEYS (self->cb_sk_object)) |
|
|
927 | { |
|
|
928 | SvREFCNT_dec (self->cb_sk_object); |
|
|
929 | self->cb_sk_object = 0; |
|
|
930 | } |
|
|
931 | } |
|
|
932 | |
|
|
933 | XPUSHs (ST (0)); |
|
|
934 | } |
|
|
935 | |
|
|
936 | #endif |
|
|
937 | |
1145 | |
938 | void encode (CBOR *self, SV *scalar) |
1146 | void encode (CBOR *self, SV *scalar) |
939 | PPCODE: |
1147 | PPCODE: |
940 | PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; |
1148 | PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; |
941 | XPUSHs (scalar); |
1149 | XPUSHs (scalar); |
… | |
… | |
954 | EXTEND (SP, 2); |
1162 | EXTEND (SP, 2); |
955 | PUSHs (sv); |
1163 | PUSHs (sv); |
956 | PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); |
1164 | PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); |
957 | } |
1165 | } |
958 | |
1166 | |
959 | #if 0 |
|
|
960 | |
|
|
961 | void DESTROY (CBOR *self) |
|
|
962 | CODE: |
|
|
963 | SvREFCNT_dec (self->cb_sk_object); |
|
|
964 | SvREFCNT_dec (self->cb_object); |
|
|
965 | |
|
|
966 | #endif |
|
|
967 | |
|
|
968 | PROTOTYPES: ENABLE |
1167 | PROTOTYPES: ENABLE |
969 | |
1168 | |
970 | void encode_cbor (SV *scalar) |
1169 | void encode_cbor (SV *scalar) |
971 | PPCODE: |
1170 | PPCODE: |
972 | { |
1171 | { |