… | |
… | |
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 | // known tags, rfc7049 |
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 | |
|
|
25 | // known tags |
15 | enum cbor_tag |
26 | enum cbor_tag |
16 | { |
27 | { |
|
|
28 | // inofficial extensions (pending iana registration) |
|
|
29 | CBOR_TAG_PERL_OBJECT = 24, // http://cbor.schmorp.de/perl-object |
|
|
30 | CBOR_TAG_GENERIC_OBJECT = 25, // http://cbor.schmorp.de/generic-object |
|
|
31 | CBOR_TAG_VALUE_SHARABLE = 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 |
|
|
36 | |
|
|
37 | // rfc7049 |
17 | CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 |
38 | CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 |
18 | CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any |
39 | CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any |
19 | CBOR_TAG_POS_BIGNUM = 2, // byte string |
40 | CBOR_TAG_POS_BIGNUM = 2, // byte string |
20 | CBOR_TAG_NEG_BIGNUM = 3, // byte string |
41 | CBOR_TAG_NEG_BIGNUM = 3, // byte string |
21 | CBOR_TAG_DECIMAL = 4, // decimal fraction, array |
42 | CBOR_TAG_DECIMAL = 4, // decimal fraction, array |
22 | CBOR_TAG_BIGFLOAT = 5, // array |
43 | CBOR_TAG_BIGFLOAT = 5, // array |
23 | |
44 | |
24 | CBOR_TAG_CONV_B64U = 21, // base64url, any |
45 | CBOR_TAG_CONV_B64U = 21, // base64url, any |
25 | CBOR_TAG_CONV_B64 = 22, // base64, any |
46 | CBOR_TAG_CONV_B64 = 22, // base64, any |
26 | CBOR_TAG_CONV_HEX = 23, // base16, any |
47 | CBOR_TAG_CONV_HEX = 23, // base16, any |
27 | CBOR_TAG_CBOR = 24, // embedded cbor, byte string |
48 | CBOR_TAG_CBOR = 24, // embedded cbor, byte string |
28 | |
49 | |
29 | CBOR_TAG_URI = 32, // URI rfc3986, utf-8 |
50 | CBOR_TAG_URI = 32, // URI rfc3986, utf-8 |
30 | CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8 |
51 | CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8 |
31 | CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8 |
52 | CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8 |
32 | CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8 |
53 | CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8 |
33 | CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 |
54 | CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 |
34 | |
55 | |
35 | CBOR_TAG_MAGIC = 55799 |
56 | CBOR_TAG_MAGIC = 55799 // self-describe cbor |
36 | }; |
57 | }; |
37 | |
58 | |
38 | #define F_SHRINK 0x00000200UL |
59 | #define F_SHRINK 0x00000001UL |
39 | #define F_ALLOW_UNKNOWN 0x00002000UL |
60 | #define F_ALLOW_UNKNOWN 0x00000002UL |
|
|
61 | #define F_ALLOW_SHARING 0x00000004UL //TODO |
|
|
62 | #define F_DEDUP_STRINGS 0x00000008UL //TODO |
|
|
63 | #define F_DEDUP_KEYS 0x00000010UL //TODO |
40 | |
64 | |
41 | #define INIT_SIZE 32 // initial scalar size to be allocated |
65 | #define INIT_SIZE 32 // initial scalar size to be allocated |
42 | |
66 | |
43 | #define SB do { |
67 | #define SB do { |
44 | #define SE } while (0) |
68 | #define SE } while (0) |
… | |
… | |
55 | #else |
79 | #else |
56 | # define CBOR_SLOW 0 |
80 | # define CBOR_SLOW 0 |
57 | # define CBOR_STASH cbor_stash |
81 | # define CBOR_STASH cbor_stash |
58 | #endif |
82 | #endif |
59 | |
83 | |
60 | static HV *cbor_stash, *cbor_boolean_stash, *cbor_tagged_stash; // CBOR::XS:: |
84 | static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS:: |
61 | static SV *cbor_true, *cbor_false; |
85 | static SV *types_true, *types_false, *types_error, *sv_cbor; |
62 | |
86 | |
63 | typedef struct { |
87 | typedef struct { |
64 | U32 flags; |
88 | U32 flags; |
65 | U32 max_depth; |
89 | U32 max_depth; |
66 | STRLEN max_size; |
90 | STRLEN max_size; |
… | |
… | |
101 | #endif |
125 | #endif |
102 | } |
126 | } |
103 | } |
127 | } |
104 | |
128 | |
105 | ///////////////////////////////////////////////////////////////////////////// |
129 | ///////////////////////////////////////////////////////////////////////////// |
106 | // fp hell |
|
|
107 | |
|
|
108 | //TODO |
|
|
109 | |
|
|
110 | ///////////////////////////////////////////////////////////////////////////// |
|
|
111 | // encoder |
130 | // encoder |
112 | |
131 | |
113 | // structure used for encoding CBOR |
132 | // structure used for encoding CBOR |
114 | typedef struct |
133 | typedef struct |
115 | { |
134 | { |
116 | char *cur; // SvPVX (sv) + current output position |
135 | char *cur; // SvPVX (sv) + current output position |
117 | char *end; // SvEND (sv) |
136 | char *end; // SvEND (sv) |
118 | SV *sv; // result scalar |
137 | SV *sv; // result scalar |
119 | CBOR cbor; |
138 | CBOR cbor; |
120 | U32 depth; // recursion level |
139 | U32 depth; // recursion level |
|
|
140 | HV *stringref; // string => index, or 0 |
|
|
141 | HV *sharable; // ptr => index, or 0 |
|
|
142 | HV *sharable_idx; |
121 | } enc_t; |
143 | } enc_t; |
122 | |
144 | |
123 | ecb_inline void |
145 | ecb_inline void |
124 | need (enc_t *enc, STRLEN len) |
146 | need (enc_t *enc, STRLEN len) |
125 | { |
147 | { |
… | |
… | |
186 | need (enc, len); |
208 | need (enc, len); |
187 | memcpy (enc->cur, str, len); |
209 | memcpy (enc->cur, str, len); |
188 | enc->cur += len; |
210 | enc->cur += len; |
189 | } |
211 | } |
190 | |
212 | |
|
|
213 | ecb_inline void |
|
|
214 | encode_tag (enc_t *enc, UV tag) |
|
|
215 | { |
|
|
216 | encode_uint (enc, 0xc0, tag); |
|
|
217 | } |
|
|
218 | |
|
|
219 | static int |
|
|
220 | encode_sharable2 (enc_t *enc, SV *sv) |
|
|
221 | { |
|
|
222 | if (!enc->sharable) |
|
|
223 | enc->sharable = (HV *)sv_2mortal ((SV *)newHV ()); |
|
|
224 | |
|
|
225 | SV **svp = hv_fetch (enc->sharable, &sv, sizeof (sv), 1); |
|
|
226 | |
|
|
227 | if (SvOK (*svp)) |
|
|
228 | { |
|
|
229 | encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF); |
|
|
230 | encode_uint (enc, 0x00, SvUV (*svp)); |
|
|
231 | |
|
|
232 | return 1; |
|
|
233 | } |
|
|
234 | else |
|
|
235 | { |
|
|
236 | sv_setuv (*svp, enc->sharable_idx++); |
|
|
237 | encode_tag (enc, CBOR_TAG_VALUE_SHARABLE); |
|
|
238 | |
|
|
239 | return 0; |
|
|
240 | } |
|
|
241 | } |
|
|
242 | |
|
|
243 | ecb_inline int |
|
|
244 | encode_sharable (enc_t *enc, SV *sv) |
|
|
245 | { |
|
|
246 | if (ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING) |
|
|
247 | && ecb_expect_false (SvREFCNT (sv) > 1)) |
|
|
248 | return encode_sharable2 (enc, sv); |
|
|
249 | |
|
|
250 | return 0; |
|
|
251 | } |
|
|
252 | |
191 | static void encode_sv (enc_t *enc, SV *sv); |
253 | static void encode_sv (enc_t *enc, SV *sv); |
192 | |
254 | |
193 | static void |
255 | static void |
194 | encode_av (enc_t *enc, AV *av) |
256 | encode_av (enc_t *enc, AV *av) |
195 | { |
257 | { |
… | |
… | |
252 | svtype svt; |
314 | svtype svt; |
253 | |
315 | |
254 | SvGETMAGIC (sv); |
316 | SvGETMAGIC (sv); |
255 | svt = SvTYPE (sv); |
317 | svt = SvTYPE (sv); |
256 | |
318 | |
|
|
319 | if (encode_sharable (enc, sv)) |
|
|
320 | return; |
|
|
321 | |
257 | if (ecb_expect_false (SvOBJECT (sv))) |
322 | if (ecb_expect_false (SvOBJECT (sv))) |
258 | { |
323 | { |
259 | HV *boolean_stash = !CBOR_SLOW || cbor_boolean_stash |
324 | HV *boolean_stash = !CBOR_SLOW || types_boolean_stash |
260 | ? cbor_boolean_stash |
325 | ? types_boolean_stash |
261 | : gv_stashpv ("CBOR::XS::Boolean", 1); |
326 | : gv_stashpv ("Types::Serialiser::Boolean", 1); |
|
|
327 | HV *error_stash = !CBOR_SLOW || types_error_stash |
|
|
328 | ? types_error_stash |
|
|
329 | : gv_stashpv ("Types::Serialiser::Error", 1); |
262 | HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash |
330 | HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash |
263 | ? cbor_tagged_stash |
331 | ? cbor_tagged_stash |
264 | : gv_stashpv ("CBOR::XS::Tagged" , 1); |
332 | : gv_stashpv ("CBOR::XS::Tagged" , 1); |
265 | |
333 | |
|
|
334 | HV *stash = SvSTASH (sv); |
|
|
335 | GV *method; |
|
|
336 | |
266 | if (SvSTASH (sv) == boolean_stash) |
337 | if (stash == boolean_stash) |
267 | encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); |
338 | encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); |
|
|
339 | else if (stash == error_stash) |
|
|
340 | encode_ch (enc, 0xe0 | 23); |
268 | else if (SvSTASH (sv) == tagged_stash) |
341 | else if (stash == tagged_stash) |
269 | { |
342 | { |
270 | if (svt != SVt_PVAV) |
343 | if (svt != SVt_PVAV) |
271 | croak ("encountered CBOR::XS::Tagged object that isn't an array"); |
344 | croak ("encountered CBOR::XS::Tagged object that isn't an array"); |
272 | |
345 | |
273 | encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); |
346 | encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); |
274 | encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); |
347 | encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); |
275 | } |
348 | } |
|
|
349 | else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) |
|
|
350 | { |
|
|
351 | dSP; |
|
|
352 | |
|
|
353 | ENTER; SAVETMPS; PUSHMARK (SP); |
|
|
354 | // we re-bless the reference to get overload and other niceties right |
|
|
355 | XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); |
|
|
356 | |
|
|
357 | PUTBACK; |
|
|
358 | // G_SCALAR ensures that return value is 1 |
|
|
359 | call_sv ((SV *)GvCV (method), G_SCALAR); |
|
|
360 | SPAGAIN; |
|
|
361 | |
|
|
362 | // catch this surprisingly common error |
|
|
363 | if (SvROK (TOPs) && SvRV (TOPs) == sv) |
|
|
364 | croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (stash)); |
|
|
365 | |
|
|
366 | encode_sv (enc, POPs); |
|
|
367 | |
|
|
368 | PUTBACK; |
|
|
369 | |
|
|
370 | FREETMPS; LEAVE; |
|
|
371 | } |
|
|
372 | else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) |
|
|
373 | { |
|
|
374 | dSP; |
|
|
375 | |
|
|
376 | ENTER; SAVETMPS; PUSHMARK (SP); |
|
|
377 | EXTEND (SP, 2); |
|
|
378 | // we re-bless the reference to get overload and other niceties right |
|
|
379 | PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); |
|
|
380 | PUSHs (sv_cbor); |
|
|
381 | |
|
|
382 | PUTBACK; |
|
|
383 | int count = call_sv ((SV *)GvCV (method), G_ARRAY); |
|
|
384 | SPAGAIN; |
|
|
385 | |
|
|
386 | // catch this surprisingly common error |
|
|
387 | if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) |
|
|
388 | croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); |
|
|
389 | |
|
|
390 | encode_tag (enc, CBOR_TAG_PERL_OBJECT); |
|
|
391 | encode_uint (enc, 0x80, count + 1); |
|
|
392 | encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); |
|
|
393 | |
|
|
394 | while (count) |
|
|
395 | encode_sv (enc, SP[1 - count--]); |
|
|
396 | |
|
|
397 | PUTBACK; |
|
|
398 | |
|
|
399 | FREETMPS; LEAVE; |
|
|
400 | } |
276 | else |
401 | else |
277 | { |
|
|
278 | // we re-bless the reference to get overload and other niceties right |
|
|
279 | GV *to_cbor = gv_fetchmethod_autoload (SvSTASH (sv), "TO_CBOR", 0); |
|
|
280 | |
|
|
281 | if (to_cbor) |
|
|
282 | { |
|
|
283 | dSP; |
|
|
284 | |
|
|
285 | ENTER; SAVETMPS; PUSHMARK (SP); |
|
|
286 | XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv))); |
|
|
287 | |
|
|
288 | // calling with G_SCALAR ensures that we always get a 1 return value |
|
|
289 | PUTBACK; |
|
|
290 | call_sv ((SV *)GvCV (to_cbor), G_SCALAR); |
|
|
291 | SPAGAIN; |
|
|
292 | |
|
|
293 | // catch this surprisingly common error |
|
|
294 | if (SvROK (TOPs) && SvRV (TOPs) == sv) |
|
|
295 | croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv))); |
|
|
296 | |
|
|
297 | sv = POPs; |
|
|
298 | PUTBACK; |
|
|
299 | |
|
|
300 | encode_sv (enc, sv); |
|
|
301 | |
|
|
302 | FREETMPS; LEAVE; |
|
|
303 | } |
|
|
304 | else |
|
|
305 | croak ("encountered object '%s', but no TO_CBOR method available on it", |
402 | croak ("encountered object '%s', but no TO_CBOR or FREEZE methods available on it", |
306 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
403 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
307 | } |
|
|
308 | } |
404 | } |
309 | else if (svt == SVt_PVHV) |
405 | else if (svt == SVt_PVHV) |
310 | encode_hv (enc, (HV *)sv); |
406 | encode_hv (enc, (HV *)sv); |
311 | else if (svt == SVt_PVAV) |
407 | else if (svt == SVt_PVAV) |
312 | encode_av (enc, (AV *)sv); |
408 | encode_av (enc, (AV *)sv); |
313 | else if (svt < SVt_PVAV) |
|
|
314 | { |
|
|
315 | STRLEN len = 0; |
|
|
316 | char *pv = svt ? SvPV (sv, len) : 0; |
|
|
317 | |
|
|
318 | if (len == 1 && *pv == '1') |
|
|
319 | encode_ch (enc, 0xe0 | 21); |
|
|
320 | else if (len == 1 && *pv == '0') |
|
|
321 | encode_ch (enc, 0xe0 | 20); |
|
|
322 | else if (enc->cbor.flags & F_ALLOW_UNKNOWN) |
|
|
323 | encode_ch (enc, 0xe0 | 23); |
|
|
324 | else |
|
|
325 | croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", |
|
|
326 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
|
|
327 | } |
|
|
328 | else if (enc->cbor.flags & F_ALLOW_UNKNOWN) |
|
|
329 | encode_ch (enc, 0xe0 | 23); |
|
|
330 | else |
409 | else |
331 | croak ("encountered %s, but CBOR can only represent references to arrays or hashes", |
410 | { |
332 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
411 | encode_tag (enc, CBOR_TAG_INDIRECTION); |
|
|
412 | encode_sv (enc, sv); |
|
|
413 | } |
333 | } |
414 | } |
334 | |
415 | |
335 | static void |
416 | static void |
336 | encode_nv (enc_t *enc, SV *sv) |
417 | encode_nv (enc_t *enc, SV *sv) |
337 | { |
418 | { |
… | |
… | |
370 | |
451 | |
371 | static void |
452 | static void |
372 | encode_sv (enc_t *enc, SV *sv) |
453 | encode_sv (enc_t *enc, SV *sv) |
373 | { |
454 | { |
374 | SvGETMAGIC (sv); |
455 | SvGETMAGIC (sv); |
|
|
456 | |
|
|
457 | if (encode_sharable (enc, sv)) |
|
|
458 | return; |
375 | |
459 | |
376 | if (SvPOKp (sv)) |
460 | if (SvPOKp (sv)) |
377 | { |
461 | { |
378 | STRLEN len; |
462 | STRLEN len; |
379 | char *str = SvPV (sv, len); |
463 | char *str = SvPV (sv, len); |
… | |
… | |
402 | } |
486 | } |
403 | |
487 | |
404 | static SV * |
488 | static SV * |
405 | encode_cbor (SV *scalar, CBOR *cbor) |
489 | encode_cbor (SV *scalar, CBOR *cbor) |
406 | { |
490 | { |
407 | enc_t enc; |
491 | enc_t enc = { }; |
408 | |
492 | |
409 | enc.cbor = *cbor; |
493 | enc.cbor = *cbor; |
410 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
494 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
411 | enc.cur = SvPVX (enc.sv); |
495 | enc.cur = SvPVX (enc.sv); |
412 | enc.end = SvEND (enc.sv); |
496 | enc.end = SvEND (enc.sv); |
413 | enc.depth = 0; |
|
|
414 | |
497 | |
415 | SvPOK_only (enc.sv); |
498 | SvPOK_only (enc.sv); |
416 | encode_sv (&enc, scalar); |
499 | encode_sv (&enc, scalar); |
417 | |
500 | |
418 | SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); |
501 | SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); |
… | |
… | |
434 | U8 *end; // end of input string |
517 | U8 *end; // end of input string |
435 | const char *err; // parse error, if != 0 |
518 | const char *err; // parse error, if != 0 |
436 | CBOR cbor; |
519 | CBOR cbor; |
437 | U32 depth; // recursion depth |
520 | U32 depth; // recursion depth |
438 | U32 maxdepth; // recursion depth limit |
521 | U32 maxdepth; // recursion depth limit |
|
|
522 | AV *sharable; |
439 | } dec_t; |
523 | } dec_t; |
440 | |
524 | |
441 | #define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE |
525 | #define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE |
442 | |
526 | |
443 | #define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") |
527 | #define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") |
… | |
… | |
537 | SvREFCNT_dec (av); |
621 | SvREFCNT_dec (av); |
538 | DEC_DEC_DEPTH; |
622 | DEC_DEC_DEPTH; |
539 | return &PL_sv_undef; |
623 | return &PL_sv_undef; |
540 | } |
624 | } |
541 | |
625 | |
|
|
626 | static void |
|
|
627 | decode_he (dec_t *dec, HV *hv) |
|
|
628 | { |
|
|
629 | // for speed reasons, we specialcase single-string |
|
|
630 | // byte or utf-8 strings as keys. |
|
|
631 | |
|
|
632 | if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27) |
|
|
633 | { |
|
|
634 | I32 len = decode_uint (dec); |
|
|
635 | char *key = (char *)dec->cur; |
|
|
636 | |
|
|
637 | dec->cur += len; |
|
|
638 | |
|
|
639 | hv_store (hv, key, len, decode_sv (dec), 0); |
|
|
640 | } |
|
|
641 | else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27) |
|
|
642 | { |
|
|
643 | I32 len = decode_uint (dec); |
|
|
644 | char *key = (char *)dec->cur; |
|
|
645 | |
|
|
646 | dec->cur += len; |
|
|
647 | |
|
|
648 | hv_store (hv, key, -len, decode_sv (dec), 0); |
|
|
649 | } |
|
|
650 | else |
|
|
651 | { |
|
|
652 | SV *k = decode_sv (dec); |
|
|
653 | SV *v = decode_sv (dec); |
|
|
654 | |
|
|
655 | hv_store_ent (hv, k, v, 0); |
|
|
656 | SvREFCNT_dec (k); |
|
|
657 | } |
|
|
658 | } |
|
|
659 | |
542 | static SV * |
660 | static SV * |
543 | decode_hv (dec_t *dec) |
661 | decode_hv (dec_t *dec) |
544 | { |
662 | { |
545 | HV *hv = newHV (); |
663 | HV *hv = newHV (); |
546 | |
664 | |
… | |
… | |
558 | { |
676 | { |
559 | ++dec->cur; |
677 | ++dec->cur; |
560 | break; |
678 | break; |
561 | } |
679 | } |
562 | |
680 | |
563 | SV *k = decode_sv (dec); |
681 | decode_he (dec, hv); |
564 | SV *v = decode_sv (dec); |
|
|
565 | |
|
|
566 | hv_store_ent (hv, k, v, 0); |
|
|
567 | } |
682 | } |
568 | } |
683 | } |
569 | else |
684 | else |
570 | { |
685 | { |
571 | int len = decode_uint (dec); |
686 | int pairs = decode_uint (dec); |
572 | |
687 | |
573 | while (len--) |
688 | while (pairs--) |
574 | { |
689 | decode_he (dec, hv); |
575 | SV *k = decode_sv (dec); |
|
|
576 | SV *v = decode_sv (dec); |
|
|
577 | |
|
|
578 | hv_store_ent (hv, k, v, 0); |
|
|
579 | } |
|
|
580 | } |
690 | } |
581 | |
691 | |
582 | DEC_DEC_DEPTH; |
692 | DEC_DEC_DEPTH; |
583 | return newRV_noinc ((SV *)hv); |
693 | return newRV_noinc ((SV *)hv); |
584 | |
694 | |
… | |
… | |
636 | decode_tagged (dec_t *dec) |
746 | decode_tagged (dec_t *dec) |
637 | { |
747 | { |
638 | UV tag = decode_uint (dec); |
748 | UV tag = decode_uint (dec); |
639 | SV *sv = decode_sv (dec); |
749 | SV *sv = decode_sv (dec); |
640 | |
750 | |
641 | if (tag == CBOR_TAG_MAGIC) // 2.4.5 Self-Describe CBOR |
751 | switch (tag) |
|
|
752 | { |
|
|
753 | case CBOR_TAG_MAGIC: |
642 | return sv; |
754 | return sv; |
643 | |
755 | |
|
|
756 | case CBOR_TAG_INDIRECTION: |
|
|
757 | return newRV_noinc (sv); |
|
|
758 | |
|
|
759 | case CBOR_TAG_VALUE_SHARABLE: |
|
|
760 | if (ecb_expect_false (!dec->sharable)) |
|
|
761 | dec->sharable = (AV *)sv_2mortal ((SV *)newAV ()); |
|
|
762 | |
|
|
763 | av_push (dec->sharable, SvREFCNT_inc_NN (sv)); |
|
|
764 | |
|
|
765 | return sv; |
|
|
766 | |
|
|
767 | case CBOR_TAG_VALUE_SHAREDREF: |
|
|
768 | { |
|
|
769 | // TODO: should verify that the sv atcually was a CBOR unsigned integer |
|
|
770 | UV idx = SvUV (sv); |
|
|
771 | |
|
|
772 | if (!dec->sharable || idx > AvFILLp (dec->sharable)) |
|
|
773 | ERR ("corrupted CBOR data (sharedref index out of bounds)"); |
|
|
774 | |
|
|
775 | SvREFCNT_dec (sv); |
|
|
776 | |
|
|
777 | return SvREFCNT_inc_NN (AvARRAY (dec->sharable)[idx]); |
|
|
778 | } |
|
|
779 | |
|
|
780 | case CBOR_TAG_PERL_OBJECT: |
|
|
781 | { |
|
|
782 | if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) |
|
|
783 | ERR ("corrupted CBOR data (non-array perl object)"); |
|
|
784 | |
|
|
785 | AV *av = (AV *)SvRV (sv); |
|
|
786 | int len = av_len (av) + 1; |
|
|
787 | HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0); |
|
|
788 | |
|
|
789 | if (!stash) |
|
|
790 | ERR ("cannot decode perl-object (package does not exist)"); |
|
|
791 | |
|
|
792 | GV *method = gv_fetchmethod_autoload (stash, "THAW", 0); |
|
|
793 | |
|
|
794 | if (!method) |
|
|
795 | ERR ("cannot decode perl-object (package does not have a THAW method)"); |
|
|
796 | |
|
|
797 | dSP; |
|
|
798 | |
|
|
799 | ENTER; SAVETMPS; PUSHMARK (SP); |
|
|
800 | EXTEND (SP, len + 1); |
|
|
801 | // we re-bless the reference to get overload and other niceties right |
|
|
802 | PUSHs (*av_fetch (av, 0, 1)); |
|
|
803 | PUSHs (sv_cbor); |
|
|
804 | |
|
|
805 | int i; |
|
|
806 | |
|
|
807 | for (i = 1; i < len; ++i) |
|
|
808 | PUSHs (*av_fetch (av, i, 1)); |
|
|
809 | |
|
|
810 | PUTBACK; |
|
|
811 | call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL); |
|
|
812 | SPAGAIN; |
|
|
813 | |
|
|
814 | if (SvTRUE (ERRSV)) |
|
|
815 | { |
|
|
816 | FREETMPS; LEAVE; |
|
|
817 | ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); |
|
|
818 | } |
|
|
819 | |
|
|
820 | SvREFCNT_dec (sv); |
|
|
821 | sv = SvREFCNT_inc (POPs); |
|
|
822 | |
|
|
823 | PUTBACK; |
|
|
824 | |
|
|
825 | FREETMPS; LEAVE; |
|
|
826 | |
|
|
827 | return sv; |
|
|
828 | } |
|
|
829 | |
|
|
830 | default: |
|
|
831 | { |
644 | AV *av = newAV (); |
832 | AV *av = newAV (); |
645 | av_push (av, newSVuv (tag)); |
833 | av_push (av, newSVuv (tag)); |
646 | av_push (av, sv); |
834 | av_push (av, sv); |
647 | |
835 | |
648 | HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash |
836 | HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash |
649 | ? cbor_tagged_stash |
837 | ? cbor_tagged_stash |
650 | : gv_stashpv ("CBOR::XS::Tagged" , 1); |
838 | : gv_stashpv ("CBOR::XS::Tagged" , 1); |
651 | |
839 | |
652 | return sv_bless (newRV_noinc ((SV *)av), tagged_stash); |
840 | return sv_bless (newRV_noinc ((SV *)av), tagged_stash); |
|
|
841 | } |
|
|
842 | } |
|
|
843 | |
|
|
844 | fail: |
|
|
845 | SvREFCNT_dec (sv); |
|
|
846 | return &PL_sv_undef; |
653 | } |
847 | } |
654 | |
848 | |
655 | static SV * |
849 | static SV * |
656 | decode_sv (dec_t *dec) |
850 | decode_sv (dec_t *dec) |
657 | { |
851 | { |
… | |
… | |
676 | case 7: // misc |
870 | case 7: // misc |
677 | switch (*dec->cur++ & 31) |
871 | switch (*dec->cur++ & 31) |
678 | { |
872 | { |
679 | case 20: |
873 | case 20: |
680 | #if CBOR_SLOW |
874 | #if CBOR_SLOW |
681 | cbor_false = get_bool ("CBOR::XS::false"); |
875 | types_false = get_bool ("Types::Serialiser::false"); |
682 | #endif |
876 | #endif |
683 | return newSVsv (cbor_false); |
877 | return newSVsv (types_false); |
684 | case 21: |
878 | case 21: |
685 | #if CBOR_SLOW |
879 | #if CBOR_SLOW |
686 | cbor_true = get_bool ("CBOR::XS::true"); |
880 | types_true = get_bool ("Types::Serialiser::true"); |
687 | #endif |
881 | #endif |
688 | return newSVsv (cbor_true); |
882 | return newSVsv (types_true); |
689 | case 22: |
883 | case 22: |
690 | return newSVsv (&PL_sv_undef); |
884 | return newSVsv (&PL_sv_undef); |
|
|
885 | case 23: |
|
|
886 | #if CBOR_SLOW |
|
|
887 | types_error = get_bool ("Types::Serialiser::error"); |
|
|
888 | #endif |
|
|
889 | return newSVsv (types_error); |
691 | |
890 | |
692 | case 25: |
891 | case 25: |
693 | { |
892 | { |
694 | WANT (2); |
893 | WANT (2); |
695 | |
894 | |
… | |
… | |
739 | } |
938 | } |
740 | |
939 | |
741 | static SV * |
940 | static SV * |
742 | decode_cbor (SV *string, CBOR *cbor, char **offset_return) |
941 | decode_cbor (SV *string, CBOR *cbor, char **offset_return) |
743 | { |
942 | { |
744 | dec_t dec; |
943 | dec_t dec = { }; |
745 | SV *sv; |
944 | SV *sv; |
|
|
945 | STRLEN len; |
|
|
946 | char *data = SvPVbyte (string, len); |
746 | |
947 | |
747 | /* work around bugs in 5.10 where manipulating magic values |
|
|
748 | * makes perl ignore the magic in subsequent accesses. |
|
|
749 | * also make a copy of non-PV values, to get them into a clean |
|
|
750 | * state (SvPV should do that, but it's buggy, see below). |
|
|
751 | */ |
|
|
752 | /*SvGETMAGIC (string);*/ |
|
|
753 | if (SvMAGICAL (string) || !SvPOK (string)) |
|
|
754 | string = sv_2mortal (newSVsv (string)); |
|
|
755 | |
|
|
756 | SvUPGRADE (string, SVt_PV); |
|
|
757 | |
|
|
758 | /* work around a bug in perl 5.10, which causes SvCUR to fail an |
|
|
759 | * assertion with -DDEBUGGING, although SvCUR is documented to |
|
|
760 | * return the xpv_cur field which certainly exists after upgrading. |
|
|
761 | * according to nicholas clark, calling SvPOK fixes this. |
|
|
762 | * But it doesn't fix it, so try another workaround, call SvPV_nolen |
|
|
763 | * and hope for the best. |
|
|
764 | * Damnit, SvPV_nolen still trips over yet another assertion. This |
|
|
765 | * assertion business is seriously broken, try yet another workaround |
|
|
766 | * for the broken -DDEBUGGING. |
|
|
767 | */ |
|
|
768 | { |
|
|
769 | #ifdef DEBUGGING |
|
|
770 | STRLEN offset = SvOK (string) ? sv_len (string) : 0; |
|
|
771 | #else |
|
|
772 | STRLEN offset = SvCUR (string); |
|
|
773 | #endif |
|
|
774 | |
|
|
775 | if (offset > cbor->max_size && cbor->max_size) |
948 | if (len > cbor->max_size && cbor->max_size) |
776 | croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu", |
949 | croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu", |
777 | (unsigned long)SvCUR (string), (unsigned long)cbor->max_size); |
950 | (unsigned long)len, (unsigned long)cbor->max_size); |
778 | } |
|
|
779 | |
|
|
780 | sv_utf8_downgrade (string, 0); |
|
|
781 | |
951 | |
782 | dec.cbor = *cbor; |
952 | dec.cbor = *cbor; |
783 | dec.cur = (U8 *)SvPVX (string); |
953 | dec.cur = (U8 *)data; |
784 | dec.end = (U8 *)SvEND (string); |
954 | dec.end = (U8 *)data + len; |
785 | dec.err = 0; |
|
|
786 | dec.depth = 0; |
|
|
787 | |
955 | |
788 | sv = decode_sv (&dec); |
956 | sv = decode_sv (&dec); |
789 | |
957 | |
790 | if (offset_return) |
958 | if (offset_return) |
791 | *offset_return = dec.cur; |
959 | *offset_return = dec.cur; |
… | |
… | |
795 | dec.err = "garbage after CBOR object"; |
963 | dec.err = "garbage after CBOR object"; |
796 | |
964 | |
797 | if (dec.err) |
965 | if (dec.err) |
798 | { |
966 | { |
799 | SvREFCNT_dec (sv); |
967 | SvREFCNT_dec (sv); |
800 | croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)SvPVX (string), (int)(uint8_t)*dec.cur); |
968 | croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); |
801 | } |
969 | } |
802 | |
970 | |
803 | sv = sv_2mortal (sv); |
971 | sv = sv_2mortal (sv); |
804 | |
972 | |
805 | return sv; |
973 | return sv; |
… | |
… | |
811 | MODULE = CBOR::XS PACKAGE = CBOR::XS |
979 | MODULE = CBOR::XS PACKAGE = CBOR::XS |
812 | |
980 | |
813 | BOOT: |
981 | BOOT: |
814 | { |
982 | { |
815 | cbor_stash = gv_stashpv ("CBOR::XS" , 1); |
983 | cbor_stash = gv_stashpv ("CBOR::XS" , 1); |
816 | cbor_boolean_stash = gv_stashpv ("CBOR::XS::Boolean", 1); |
|
|
817 | cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1); |
984 | cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1); |
818 | |
985 | |
819 | cbor_true = get_bool ("CBOR::XS::true"); |
986 | types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1); |
820 | cbor_false = get_bool ("CBOR::XS::false"); |
987 | types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1); |
|
|
988 | |
|
|
989 | types_true = get_bool ("Types::Serialiser::true" ); |
|
|
990 | types_false = get_bool ("Types::Serialiser::false"); |
|
|
991 | types_error = get_bool ("Types::Serialiser::error"); |
|
|
992 | |
|
|
993 | sv_cbor = newSVpv ("CBOR", 0); |
|
|
994 | SvREADONLY_on (sv_cbor); |
821 | } |
995 | } |
822 | |
996 | |
823 | PROTOTYPES: DISABLE |
997 | PROTOTYPES: DISABLE |
824 | |
998 | |
825 | void CLONE (...) |
999 | void CLONE (...) |
826 | CODE: |
1000 | CODE: |
827 | cbor_stash = 0; |
1001 | cbor_stash = 0; |
828 | cbor_boolean_stash = 0; |
|
|
829 | cbor_tagged_stash = 0; |
1002 | cbor_tagged_stash = 0; |
|
|
1003 | types_error_stash = 0; |
|
|
1004 | types_boolean_stash = 0; |
830 | |
1005 | |
831 | void new (char *klass) |
1006 | void new (char *klass) |
832 | PPCODE: |
1007 | PPCODE: |
833 | { |
1008 | { |
834 | SV *pv = NEWSV (0, sizeof (CBOR)); |
1009 | SV *pv = NEWSV (0, sizeof (CBOR)); |
… | |
… | |
842 | |
1017 | |
843 | void shrink (CBOR *self, int enable = 1) |
1018 | void shrink (CBOR *self, int enable = 1) |
844 | ALIAS: |
1019 | ALIAS: |
845 | shrink = F_SHRINK |
1020 | shrink = F_SHRINK |
846 | allow_unknown = F_ALLOW_UNKNOWN |
1021 | allow_unknown = F_ALLOW_UNKNOWN |
|
|
1022 | allow_sharing = F_ALLOW_SHARING |
|
|
1023 | dedup_keys = F_DEDUP_KEYS |
|
|
1024 | dedup_strings = F_DEDUP_STRINGS |
847 | PPCODE: |
1025 | PPCODE: |
848 | { |
1026 | { |
849 | if (enable) |
1027 | if (enable) |
850 | self->flags |= ix; |
1028 | self->flags |= ix; |
851 | else |
1029 | else |
… | |
… | |
856 | |
1034 | |
857 | void get_shrink (CBOR *self) |
1035 | void get_shrink (CBOR *self) |
858 | ALIAS: |
1036 | ALIAS: |
859 | get_shrink = F_SHRINK |
1037 | get_shrink = F_SHRINK |
860 | get_allow_unknown = F_ALLOW_UNKNOWN |
1038 | get_allow_unknown = F_ALLOW_UNKNOWN |
|
|
1039 | get_allow_sharing = F_ALLOW_SHARING |
|
|
1040 | get_dedup_keys = F_DEDUP_KEYS |
|
|
1041 | get_dedup_strings = F_DEDUP_STRINGS |
861 | PPCODE: |
1042 | PPCODE: |
862 | XPUSHs (boolSV (self->flags & ix)); |
1043 | XPUSHs (boolSV (self->flags & ix)); |
863 | |
1044 | |
864 | void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) |
1045 | void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) |
865 | PPCODE: |
1046 | PPCODE: |
… | |
… | |
880 | int get_max_size (CBOR *self) |
1061 | int get_max_size (CBOR *self) |
881 | CODE: |
1062 | CODE: |
882 | RETVAL = self->max_size; |
1063 | RETVAL = self->max_size; |
883 | OUTPUT: |
1064 | OUTPUT: |
884 | RETVAL |
1065 | RETVAL |
885 | |
|
|
886 | #if 0 //TODO |
|
|
887 | |
|
|
888 | void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef) |
|
|
889 | PPCODE: |
|
|
890 | { |
|
|
891 | SvREFCNT_dec (self->cb_object); |
|
|
892 | self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; |
|
|
893 | |
|
|
894 | XPUSHs (ST (0)); |
|
|
895 | } |
|
|
896 | |
|
|
897 | void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef) |
|
|
898 | PPCODE: |
|
|
899 | { |
|
|
900 | if (!self->cb_sk_object) |
|
|
901 | self->cb_sk_object = newHV (); |
|
|
902 | |
|
|
903 | if (SvOK (cb)) |
|
|
904 | hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0); |
|
|
905 | else |
|
|
906 | { |
|
|
907 | hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0); |
|
|
908 | |
|
|
909 | if (!HvKEYS (self->cb_sk_object)) |
|
|
910 | { |
|
|
911 | SvREFCNT_dec (self->cb_sk_object); |
|
|
912 | self->cb_sk_object = 0; |
|
|
913 | } |
|
|
914 | } |
|
|
915 | |
|
|
916 | XPUSHs (ST (0)); |
|
|
917 | } |
|
|
918 | |
|
|
919 | #endif |
|
|
920 | |
1066 | |
921 | void encode (CBOR *self, SV *scalar) |
1067 | void encode (CBOR *self, SV *scalar) |
922 | PPCODE: |
1068 | PPCODE: |
923 | PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; |
1069 | PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; |
924 | XPUSHs (scalar); |
1070 | XPUSHs (scalar); |
… | |
… | |
937 | EXTEND (SP, 2); |
1083 | EXTEND (SP, 2); |
938 | PUSHs (sv); |
1084 | PUSHs (sv); |
939 | PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); |
1085 | PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); |
940 | } |
1086 | } |
941 | |
1087 | |
942 | #if 0 |
|
|
943 | |
|
|
944 | void DESTROY (CBOR *self) |
|
|
945 | CODE: |
|
|
946 | SvREFCNT_dec (self->cb_sk_object); |
|
|
947 | SvREFCNT_dec (self->cb_object); |
|
|
948 | |
|
|
949 | #endif |
|
|
950 | |
|
|
951 | PROTOTYPES: ENABLE |
1088 | PROTOTYPES: ENABLE |
952 | |
1089 | |
953 | void encode_cbor (SV *scalar) |
1090 | void encode_cbor (SV *scalar) |
954 | PPCODE: |
1091 | PPCODE: |
955 | { |
1092 | { |