ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/CBOR-XS/XS.xs
(Generate patch)

Comparing CBOR-XS/XS.xs (file contents):
Revision 1.6 by root, Sat Oct 26 22:25:47 2013 UTC vs.
Revision 1.45 by root, Sun Dec 14 05:48:39 2014 UTC

7#include <stdlib.h> 7#include <stdlib.h>
8#include <stdio.h> 8#include <stdio.h>
9#include <limits.h> 9#include <limits.h>
10#include <float.h> 10#include <float.h>
11 11
12#define ECB_NO_THREADS 1
12#include "ecb.h" 13#include "ecb.h"
13 14
15// compatibility with perl <5.18
16#ifndef HvNAMELEN_get
17# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
18#endif
19#ifndef HvNAMELEN
20# define HvNAMELEN(hv) HvNAMELEN_get (hv)
21#endif
22#ifndef HvNAMEUTF8
23# define HvNAMEUTF8(hv) 0
24#endif
25#ifndef SvREFCNT_dec_NN
26# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)
27#endif
28
29// known major and minor types
30enum cbor_type
31{
32 MAJOR_SHIFT = 5,
33 MINOR_MASK = 0x1f,
34
35 MAJOR_POS_INT = 0 << MAJOR_SHIFT,
36 MAJOR_NEG_INT = 1 << MAJOR_SHIFT,
37 MAJOR_BYTES = 2 << MAJOR_SHIFT,
38 MAJOR_TEXT = 3 << MAJOR_SHIFT,
39 MAJOR_ARRAY = 4 << MAJOR_SHIFT,
40 MAJOR_MAP = 5 << MAJOR_SHIFT,
41 MAJOR_TAG = 6 << MAJOR_SHIFT,
42 MAJOR_MISC = 7 << MAJOR_SHIFT,
43
44 // INT/STRING/ARRAY/MAP subtypes
45 LENGTH_EXT1 = 24,
46 LENGTH_EXT2 = 25,
47 LENGTH_EXT4 = 26,
48 LENGTH_EXT8 = 27,
49
50 // SIMPLE types (effectively MISC subtypes)
51 SIMPLE_FALSE = 20,
52 SIMPLE_TRUE = 21,
53 SIMPLE_NULL = 22,
54 SIMPLE_UNDEF = 23,
55
56 // MISC subtype (unused)
57 MISC_EXT1 = 24,
58 MISC_FLOAT16 = 25,
59 MISC_FLOAT32 = 26,
60 MISC_FLOAT64 = 27,
61
62 // BYTES/TEXT/ARRAY/MAP
63 MINOR_INDEF = 31,
64};
65
66// known tags
67enum cbor_tag
68{
69 // extensions
70 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
71 CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object
72 CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object
73 CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing
74 CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing
75 CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref
76 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection
77
78 // rfc7049
79 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
80 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
81 CBOR_TAG_POS_BIGNUM = 2, // byte string
82 CBOR_TAG_NEG_BIGNUM = 3, // byte string
83 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
84 CBOR_TAG_BIGFLOAT = 5, // array
85
86 CBOR_TAG_CONV_B64U = 21, // base64url, any
87 CBOR_TAG_CONV_B64 = 22, // base64, any
88 CBOR_TAG_CONV_HEX = 23, // base16, any
89 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
90
91 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
92 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
93 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
94 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
95 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
96
97 CBOR_TAG_MAGIC = 55799, // self-describe cbor
98};
99
14#define F_SHRINK 0x00000200UL 100#define F_SHRINK 0x00000001UL
15#define F_ALLOW_UNKNOWN 0x00002000UL 101#define F_ALLOW_UNKNOWN 0x00000002UL
102#define F_ALLOW_SHARING 0x00000004UL
103#define F_ALLOW_CYCLES 0x00000008UL
104#define F_PACK_STRINGS 0x00000010UL
105#define F_VALIDATE_UTF8 0x00000020UL
16 106
17#define INIT_SIZE 32 // initial scalar size to be allocated 107#define INIT_SIZE 32 // initial scalar size to be allocated
18 108
19#define SB do { 109#define SB do {
20#define SE } while (0) 110#define SE } while (0)
31#else 121#else
32# define CBOR_SLOW 0 122# define CBOR_SLOW 0
33# define CBOR_STASH cbor_stash 123# define CBOR_STASH cbor_stash
34#endif 124#endif
35 125
36static HV *cbor_stash, *cbor_boolean_stash, *cbor_tagged_stash; // CBOR::XS:: 126static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
37static SV *cbor_true, *cbor_false; 127static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
38 128
39typedef struct { 129typedef struct {
40 U32 flags; 130 U32 flags;
41 U32 max_depth; 131 U32 max_depth;
42 STRLEN max_size; 132 STRLEN max_size;
133 SV *filter;
134
135 // for the incremental parser
136 STRLEN incr_pos; // the current offset into the text
137 STRLEN incr_need; // minimum bytes needed to decode
138 AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
43} CBOR; 139} CBOR;
44 140
45ecb_inline void 141ecb_inline void
46cbor_init (CBOR *cbor) 142cbor_init (CBOR *cbor)
47{ 143{
48 Zero (cbor, 1, CBOR); 144 Zero (cbor, 1, CBOR);
49 cbor->max_depth = 512; 145 cbor->max_depth = 512;
146}
147
148ecb_inline void
149cbor_free (CBOR *cbor)
150{
151 SvREFCNT_dec (cbor->filter);
152 SvREFCNT_dec (cbor->incr_count);
50} 153}
51 154
52///////////////////////////////////////////////////////////////////////////// 155/////////////////////////////////////////////////////////////////////////////
53// utility functions 156// utility functions
54 157
76 SvPV_renew (sv, SvCUR (sv) + 1); 179 SvPV_renew (sv, SvCUR (sv) + 1);
77#endif 180#endif
78 } 181 }
79} 182}
80 183
81///////////////////////////////////////////////////////////////////////////// 184// minimum length of a string to be registered for stringref
82// fp hell 185ecb_inline int
83 186minimum_string_length (UV idx)
84//TODO 187{
188 return idx > 23
189 ? idx > 0xffU
190 ? idx > 0xffffU
191 ? idx > 0xffffffffU
192 ? 11
193 : 7
194 : 5
195 : 4
196 : 3;
197}
85 198
86///////////////////////////////////////////////////////////////////////////// 199/////////////////////////////////////////////////////////////////////////////
87// encoder 200// encoder
88 201
89// structure used for encoding CBOR 202// structure used for encoding CBOR
92 char *cur; // SvPVX (sv) + current output position 205 char *cur; // SvPVX (sv) + current output position
93 char *end; // SvEND (sv) 206 char *end; // SvEND (sv)
94 SV *sv; // result scalar 207 SV *sv; // result scalar
95 CBOR cbor; 208 CBOR cbor;
96 U32 depth; // recursion level 209 U32 depth; // recursion level
210 HV *stringref[2]; // string => index, or 0 ([0] = bytes, [1] = utf-8)
211 UV stringref_idx;
212 HV *shareable; // ptr => index, or 0
213 UV shareable_idx;
97} enc_t; 214} enc_t;
98 215
99ecb_inline void 216ecb_inline void
100need (enc_t *enc, STRLEN len) 217need (enc_t *enc, STRLEN len)
101{ 218{
118static void 235static void
119encode_uint (enc_t *enc, int major, UV len) 236encode_uint (enc_t *enc, int major, UV len)
120{ 237{
121 need (enc, 9); 238 need (enc, 9);
122 239
123 if (len < 24) 240 if (ecb_expect_true (len < LENGTH_EXT1))
124 *enc->cur++ = major | len; 241 *enc->cur++ = major | len;
125 else if (len <= 0xff) 242 else if (ecb_expect_true (len <= 0xffU))
126 { 243 {
127 *enc->cur++ = major | 24; 244 *enc->cur++ = major | LENGTH_EXT1;
128 *enc->cur++ = len; 245 *enc->cur++ = len;
129 } 246 }
130 else if (len <= 0xffff) 247 else if (len <= 0xffffU)
131 { 248 {
132 *enc->cur++ = major | 25; 249 *enc->cur++ = major | LENGTH_EXT2;
133 *enc->cur++ = len >> 8; 250 *enc->cur++ = len >> 8;
134 *enc->cur++ = len; 251 *enc->cur++ = len;
135 } 252 }
136 else if (len <= 0xffffffff) 253 else if (len <= 0xffffffffU)
137 { 254 {
138 *enc->cur++ = major | 26; 255 *enc->cur++ = major | LENGTH_EXT4;
139 *enc->cur++ = len >> 24; 256 *enc->cur++ = len >> 24;
140 *enc->cur++ = len >> 16; 257 *enc->cur++ = len >> 16;
141 *enc->cur++ = len >> 8; 258 *enc->cur++ = len >> 8;
142 *enc->cur++ = len; 259 *enc->cur++ = len;
143 } 260 }
144 else 261 else
145 { 262 {
146 *enc->cur++ = major | 27; 263 *enc->cur++ = major | LENGTH_EXT8;
147 *enc->cur++ = len >> 56; 264 *enc->cur++ = len >> 56;
148 *enc->cur++ = len >> 48; 265 *enc->cur++ = len >> 48;
149 *enc->cur++ = len >> 40; 266 *enc->cur++ = len >> 40;
150 *enc->cur++ = len >> 32; 267 *enc->cur++ = len >> 32;
151 *enc->cur++ = len >> 24; 268 *enc->cur++ = len >> 24;
153 *enc->cur++ = len >> 8; 270 *enc->cur++ = len >> 8;
154 *enc->cur++ = len; 271 *enc->cur++ = len;
155 } 272 }
156} 273}
157 274
158static void 275ecb_inline void
276encode_tag (enc_t *enc, UV tag)
277{
278 encode_uint (enc, MAJOR_TAG, tag);
279}
280
281ecb_inline void
159encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 282encode_str (enc_t *enc, int utf8, char *str, STRLEN len)
160{ 283{
161 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 284 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
162 need (enc, len); 285 need (enc, len);
163 memcpy (enc->cur, str, len); 286 memcpy (enc->cur, str, len);
164 enc->cur += len; 287 enc->cur += len;
165} 288}
166 289
290static void
291encode_strref (enc_t *enc, int utf8, char *str, STRLEN len)
292{
293 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
294 {
295 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
296
297 if (SvOK (*svp))
298 {
299 // already registered, use stringref
300 encode_tag (enc, CBOR_TAG_STRINGREF);
301 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
302 return;
303 }
304 else if (len >= minimum_string_length (enc->stringref_idx))
305 {
306 // register only
307 sv_setuv (*svp, enc->stringref_idx);
308 ++enc->stringref_idx;
309 }
310 }
311
312 encode_str (enc, utf8, str, len);
313}
314
167static void encode_sv (enc_t *enc, SV *sv); 315static void encode_sv (enc_t *enc, SV *sv);
168 316
169static void 317static void
170encode_av (enc_t *enc, AV *av) 318encode_av (enc_t *enc, AV *av)
171{ 319{
174 if (enc->depth >= enc->cbor.max_depth) 322 if (enc->depth >= enc->cbor.max_depth)
175 croak (ERR_NESTING_EXCEEDED); 323 croak (ERR_NESTING_EXCEEDED);
176 324
177 ++enc->depth; 325 ++enc->depth;
178 326
179 encode_uint (enc, 0x80, len + 1); 327 encode_uint (enc, MAJOR_ARRAY, len + 1);
180 328
329 if (SvMAGICAL (av))
181 for (i = 0; i <= len; ++i) 330 for (i = 0; i <= len; ++i)
182 { 331 {
183 SV **svp = av_fetch (av, i, 0); 332 SV **svp = av_fetch (av, i, 0);
184 encode_sv (enc, svp ? *svp : &PL_sv_undef); 333 encode_sv (enc, svp ? *svp : &PL_sv_undef);
185 } 334 }
335 else
336 for (i = 0; i <= len; ++i)
337 encode_sv (enc, AvARRAY (av)[i]);
186 338
187 --enc->depth; 339 --enc->depth;
188} 340}
189 341
190static void 342static void
199 351
200 int pairs = hv_iterinit (hv); 352 int pairs = hv_iterinit (hv);
201 int mg = SvMAGICAL (hv); 353 int mg = SvMAGICAL (hv);
202 354
203 if (mg) 355 if (mg)
204 encode_ch (enc, 0xa0 | 31); 356 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
205 else 357 else
206 encode_uint (enc, 0xa0, pairs); 358 encode_uint (enc, MAJOR_MAP, pairs);
207 359
208 while ((he = hv_iternext (hv))) 360 while ((he = hv_iternext (hv)))
209 { 361 {
210 if (HeKLEN (he) == HEf_SVKEY) 362 if (HeKLEN (he) == HEf_SVKEY)
211 encode_sv (enc, HeSVKEY (he)); 363 encode_sv (enc, HeSVKEY (he));
212 else 364 else
213 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 365 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
214 366
215 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 367 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
216 } 368 }
217 369
218 if (mg) 370 if (mg)
219 encode_ch (enc, 0xe0 | 31); 371 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
220 372
221 --enc->depth; 373 --enc->depth;
222} 374}
223 375
224// encode objects, arrays and special \0=false and \1=true values. 376// encode objects, arrays and special \0=false and \1=true values.
225static void 377static void
226encode_rv (enc_t *enc, SV *sv) 378encode_rv (enc_t *enc, SV *sv)
227{ 379{
228 svtype svt;
229
230 SvGETMAGIC (sv); 380 SvGETMAGIC (sv);
381
231 svt = SvTYPE (sv); 382 svtype svt = SvTYPE (sv);
232 383
233 if (ecb_expect_false (SvOBJECT (sv))) 384 if (ecb_expect_false (SvOBJECT (sv)))
234 { 385 {
235 HV *boolean_stash = !CBOR_SLOW || cbor_boolean_stash 386 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
236 ? cbor_boolean_stash 387 ? types_boolean_stash
237 : gv_stashpv ("CBOR::XS::Boolean", 1); 388 : gv_stashpv ("Types::Serialiser::Boolean", 1);
389 HV *error_stash = !CBOR_SLOW || types_error_stash
390 ? types_error_stash
391 : gv_stashpv ("Types::Serialiser::Error", 1);
238 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 392 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
239 ? cbor_tagged_stash 393 ? cbor_tagged_stash
240 : gv_stashpv ("CBOR::XS::Tagged" , 1); 394 : gv_stashpv ("CBOR::XS::Tagged" , 1);
241 395
396 HV *stash = SvSTASH (sv);
397
242 if (SvSTASH (sv) == boolean_stash) 398 if (stash == boolean_stash)
243 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 399 {
400 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
401 return;
402 }
403 else if (stash == error_stash)
404 {
405 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
406 return;
407 }
244 else if (SvSTASH (sv) == tagged_stash) 408 else if (stash == tagged_stash)
245 { 409 {
246 if (svt != SVt_PVAV) 410 if (svt != SVt_PVAV)
247 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 411 croak ("encountered CBOR::XS::Tagged object that isn't an array");
248 412
249 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 413 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
250 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 414 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
415
416 return;
417 }
418 }
419
420 if (ecb_expect_false (SvREFCNT (sv) > 1)
421 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
422 {
423 if (!enc->shareable)
424 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
425
426 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
427
428 if (SvOK (*svp))
429 {
430 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
431 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
432 return;
251 } 433 }
252 else 434 else
253 { 435 {
436 sv_setuv (*svp, enc->shareable_idx);
437 ++enc->shareable_idx;
438 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
439 }
440 }
441
442 if (ecb_expect_false (SvOBJECT (sv)))
443 {
444 HV *stash = SvSTASH (sv);
445 GV *method;
446
447 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
448 {
449 dSP;
450
451 ENTER; SAVETMPS; PUSHMARK (SP);
254 // we re-bless the reference to get overload and other niceties right 452 // we re-bless the reference to get overload and other niceties right
255 GV *to_cbor = gv_fetchmethod_autoload (SvSTASH (sv), "TO_CBOR", 0);
256
257 if (to_cbor)
258 {
259 dSP;
260
261 ENTER; SAVETMPS; PUSHMARK (SP);
262 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv))); 453 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
263 454
264 // calling with G_SCALAR ensures that we always get a 1 return value
265 PUTBACK; 455 PUTBACK;
456 // G_SCALAR ensures that return value is 1
266 call_sv ((SV *)GvCV (to_cbor), G_SCALAR); 457 call_sv ((SV *)GvCV (method), G_SCALAR);
267 SPAGAIN; 458 SPAGAIN;
268 459
269 // catch this surprisingly common error 460 // catch this surprisingly common error
270 if (SvROK (TOPs) && SvRV (TOPs) == sv) 461 if (SvROK (TOPs) && SvRV (TOPs) == sv)
271 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv))); 462 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (stash));
272 463
273 sv = POPs; 464 encode_sv (enc, POPs);
465
274 PUTBACK; 466 PUTBACK;
275 467
276 encode_sv (enc, sv);
277
278 FREETMPS; LEAVE; 468 FREETMPS; LEAVE;
279 } 469 }
470 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
471 {
472 dSP;
473
474 ENTER; SAVETMPS; PUSHMARK (SP);
475 EXTEND (SP, 2);
476 // we re-bless the reference to get overload and other niceties right
477 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
478 PUSHs (sv_cbor);
479
480 PUTBACK;
481 int count = call_sv ((SV *)GvCV (method), G_ARRAY);
482 SPAGAIN;
483
484 // catch this surprisingly common error
485 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
486 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
487
488 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
489 encode_uint (enc, MAJOR_ARRAY, count + 1);
490 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
491
492 while (count)
493 encode_sv (enc, SP[1 - count--]);
494
495 PUTBACK;
496
497 FREETMPS; LEAVE;
498 }
280 else 499 else
281 croak ("encountered object '%s', but no TO_CBOR method available on it", 500 croak ("encountered object '%s', but no TO_CBOR or FREEZE methods available on it",
282 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 501 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
283 }
284 } 502 }
285 else if (svt == SVt_PVHV) 503 else if (svt == SVt_PVHV)
286 encode_hv (enc, (HV *)sv); 504 encode_hv (enc, (HV *)sv);
287 else if (svt == SVt_PVAV) 505 else if (svt == SVt_PVAV)
288 encode_av (enc, (AV *)sv); 506 encode_av (enc, (AV *)sv);
289 else if (svt < SVt_PVAV)
290 {
291 STRLEN len = 0;
292 char *pv = svt ? SvPV (sv, len) : 0;
293
294 if (len == 1 && *pv == '1')
295 encode_ch (enc, 0xe0 | 21);
296 else if (len == 1 && *pv == '0')
297 encode_ch (enc, 0xe0 | 20);
298 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
299 encode_ch (enc, 0xe0 | 23);
300 else
301 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
302 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
303 }
304 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
305 encode_ch (enc, 0xe0 | 23);
306 else 507 else
307 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 508 {
308 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 509 encode_tag (enc, CBOR_TAG_INDIRECTION);
510 encode_sv (enc, sv);
511 }
309} 512}
310 513
311static void 514static void
312encode_nv (enc_t *enc, SV *sv) 515encode_nv (enc_t *enc, SV *sv)
313{ 516{
314 double nv = SvNVX (sv); 517 double nv = SvNVX (sv);
315 518
316 need (enc, 9); 519 need (enc, 9);
317 520
318 if (ecb_expect_false (nv == (U32)nv)) 521 if (ecb_expect_false (nv == (NV)(U32)nv))
319 encode_uint (enc, 0x00, (U32)nv); 522 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
320 //TODO: maybe I32? 523 //TODO: maybe I32?
321 else if (ecb_expect_false (nv == (float)nv)) 524 else if (ecb_expect_false (nv == (float)nv))
322 { 525 {
323 uint32_t fp = ecb_float_to_binary32 (nv); 526 uint32_t fp = ecb_float_to_binary32 (nv);
324 527
325 *enc->cur++ = 0xe0 | 26; 528 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
326 529
327 if (!ecb_big_endian ()) 530 if (!ecb_big_endian ())
328 fp = ecb_bswap32 (fp); 531 fp = ecb_bswap32 (fp);
329 532
330 memcpy (enc->cur, &fp, 4); 533 memcpy (enc->cur, &fp, 4);
332 } 535 }
333 else 536 else
334 { 537 {
335 uint64_t fp = ecb_double_to_binary64 (nv); 538 uint64_t fp = ecb_double_to_binary64 (nv);
336 539
337 *enc->cur++ = 0xe0 | 27; 540 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
338 541
339 if (!ecb_big_endian ()) 542 if (!ecb_big_endian ())
340 fp = ecb_bswap64 (fp); 543 fp = ecb_bswap64 (fp);
341 544
342 memcpy (enc->cur, &fp, 8); 545 memcpy (enc->cur, &fp, 8);
351 554
352 if (SvPOKp (sv)) 555 if (SvPOKp (sv))
353 { 556 {
354 STRLEN len; 557 STRLEN len;
355 char *str = SvPV (sv, len); 558 char *str = SvPV (sv, len);
356 encode_str (enc, SvUTF8 (sv), str, len); 559 encode_strref (enc, SvUTF8 (sv), str, len);
357 } 560 }
358 else if (SvNOKp (sv)) 561 else if (SvNOKp (sv))
359 encode_nv (enc, sv); 562 encode_nv (enc, sv);
360 else if (SvIOKp (sv)) 563 else if (SvIOKp (sv))
361 { 564 {
362 if (SvIsUV (sv)) 565 if (SvIsUV (sv))
363 encode_uint (enc, 0x00, SvUVX (sv)); 566 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
364 else if (SvIVX (sv) >= 0) 567 else if (SvIVX (sv) >= 0)
365 encode_uint (enc, 0x00, SvIVX (sv)); 568 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
366 else 569 else
367 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 570 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
368 } 571 }
369 else if (SvROK (sv)) 572 else if (SvROK (sv))
370 encode_rv (enc, SvRV (sv)); 573 encode_rv (enc, SvRV (sv));
371 else if (!SvOK (sv)) 574 else if (!SvOK (sv))
372 encode_ch (enc, 0xe0 | 22); 575 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
373 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 576 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
374 encode_ch (enc, 0xe0 | 23); 577 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
375 else 578 else
376 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 579 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
377 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 580 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
378} 581}
379 582
380static SV * 583static SV *
381encode_cbor (SV *scalar, CBOR *cbor) 584encode_cbor (SV *scalar, CBOR *cbor)
382{ 585{
383 enc_t enc; 586 enc_t enc = { };
384 587
385 enc.cbor = *cbor; 588 enc.cbor = *cbor;
386 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 589 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
387 enc.cur = SvPVX (enc.sv); 590 enc.cur = SvPVX (enc.sv);
388 enc.end = SvEND (enc.sv); 591 enc.end = SvEND (enc.sv);
389 enc.depth = 0;
390 592
391 SvPOK_only (enc.sv); 593 SvPOK_only (enc.sv);
594
595 if (cbor->flags & F_PACK_STRINGS)
596 {
597 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
598 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
599 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
600 }
601
392 encode_sv (&enc, scalar); 602 encode_sv (&enc, scalar);
393 603
394 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 604 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
395 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 605 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
396 606
410 U8 *end; // end of input string 620 U8 *end; // end of input string
411 const char *err; // parse error, if != 0 621 const char *err; // parse error, if != 0
412 CBOR cbor; 622 CBOR cbor;
413 U32 depth; // recursion depth 623 U32 depth; // recursion depth
414 U32 maxdepth; // recursion depth limit 624 U32 maxdepth; // recursion depth limit
625 AV *shareable;
626 AV *stringref;
627 SV *decode_tagged;
415} dec_t; 628} dec_t;
416 629
417#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 630#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE
418 631
419#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 632#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data")
422#define DEC_DEC_DEPTH --dec->depth 635#define DEC_DEC_DEPTH --dec->depth
423 636
424static UV 637static UV
425decode_uint (dec_t *dec) 638decode_uint (dec_t *dec)
426{ 639{
427 switch (*dec->cur & 31) 640 U8 m = *dec->cur & MINOR_MASK;
428 { 641 ++dec->cur;
429 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
430 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
431 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
432 return *dec->cur++ & 31;
433 642
434 case 24: 643 if (ecb_expect_true (m < LENGTH_EXT1))
644 return m;
645 else if (ecb_expect_true (m == LENGTH_EXT1))
646 {
435 WANT (2); 647 WANT (1);
436 dec->cur += 2; 648 dec->cur += 1;
437 return dec->cur[-1]; 649 return dec->cur[-1];
438 650 }
439 case 25: 651 else if (ecb_expect_true (m == LENGTH_EXT2))
652 {
440 WANT (3); 653 WANT (2);
441 dec->cur += 3; 654 dec->cur += 2;
442 return (((UV)dec->cur[-2]) << 8) 655 return (((UV)dec->cur[-2]) << 8)
443 | ((UV)dec->cur[-1]); 656 | ((UV)dec->cur[-1]);
444 657 }
445 case 26: 658 else if (ecb_expect_true (m == LENGTH_EXT4))
659 {
446 WANT (5); 660 WANT (4);
447 dec->cur += 5; 661 dec->cur += 4;
448 return (((UV)dec->cur[-4]) << 24) 662 return (((UV)dec->cur[-4]) << 24)
449 | (((UV)dec->cur[-3]) << 16) 663 | (((UV)dec->cur[-3]) << 16)
450 | (((UV)dec->cur[-2]) << 8) 664 | (((UV)dec->cur[-2]) << 8)
451 | ((UV)dec->cur[-1]); 665 | ((UV)dec->cur[-1]);
452 666 }
453 case 27: 667 else if (ecb_expect_true (m == LENGTH_EXT8))
668 {
454 WANT (9); 669 WANT (8);
455 dec->cur += 9; 670 dec->cur += 8;
671
672 return
673#if UVSIZE < 8
674 0
675#else
456 return (((UV)dec->cur[-8]) << 56) 676 (((UV)dec->cur[-8]) << 56)
457 | (((UV)dec->cur[-7]) << 48) 677 | (((UV)dec->cur[-7]) << 48)
458 | (((UV)dec->cur[-6]) << 40) 678 | (((UV)dec->cur[-6]) << 40)
459 | (((UV)dec->cur[-5]) << 32) 679 | (((UV)dec->cur[-5]) << 32)
680#endif
460 | (((UV)dec->cur[-4]) << 24) 681 | (((UV)dec->cur[-4]) << 24)
461 | (((UV)dec->cur[-3]) << 16) 682 | (((UV)dec->cur[-3]) << 16)
462 | (((UV)dec->cur[-2]) << 8) 683 | (((UV)dec->cur[-2]) << 8)
463 | ((UV)dec->cur[-1]); 684 | ((UV)dec->cur[-1]);
464 685 }
465 default: 686 else
466 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 687 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
467 }
468 688
469fail: 689fail:
470 return 0; 690 return 0;
471} 691}
472 692
477{ 697{
478 AV *av = newAV (); 698 AV *av = newAV ();
479 699
480 DEC_INC_DEPTH; 700 DEC_INC_DEPTH;
481 701
482 if ((*dec->cur & 31) == 31) 702 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
483 { 703 {
484 ++dec->cur; 704 ++dec->cur;
485 705
486 for (;;) 706 for (;;)
487 { 707 {
488 WANT (1); 708 WANT (1);
489 709
490 if (*dec->cur == (0xe0 | 31)) 710 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
491 { 711 {
492 ++dec->cur; 712 ++dec->cur;
493 break; 713 break;
494 } 714 }
495 715
498 } 718 }
499 else 719 else
500 { 720 {
501 int i, len = decode_uint (dec); 721 int i, len = decode_uint (dec);
502 722
723 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
503 av_fill (av, len - 1); 724 av_fill (av, len - 1);
504 725
505 for (i = 0; i < len; ++i) 726 for (i = 0; i < len; ++i)
506 AvARRAY (av)[i] = decode_sv (dec); 727 AvARRAY (av)[i] = decode_sv (dec);
507 } 728 }
513 SvREFCNT_dec (av); 734 SvREFCNT_dec (av);
514 DEC_DEC_DEPTH; 735 DEC_DEC_DEPTH;
515 return &PL_sv_undef; 736 return &PL_sv_undef;
516} 737}
517 738
739static void
740decode_he (dec_t *dec, HV *hv)
741{
742 // for speed reasons, we specialcase single-string
743 // byte or utf-8 strings as keys, but only when !stringref
744
745 if (ecb_expect_true (!dec->stringref))
746 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
747 {
748 I32 len = decode_uint (dec);
749 char *key = (char *)dec->cur;
750
751 dec->cur += len;
752
753 hv_store (hv, key, len, decode_sv (dec), 0);
754
755 return;
756 }
757 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
758 {
759 I32 len = decode_uint (dec);
760 char *key = (char *)dec->cur;
761
762 dec->cur += len;
763
764 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
765 if (!is_utf8_string (key, len))
766 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
767
768 hv_store (hv, key, -len, decode_sv (dec), 0);
769
770 return;
771 }
772
773 SV *k = decode_sv (dec);
774 SV *v = decode_sv (dec);
775
776 hv_store_ent (hv, k, v, 0);
777 SvREFCNT_dec (k);
778
779fail:
780 ;
781}
782
518static SV * 783static SV *
519decode_hv (dec_t *dec) 784decode_hv (dec_t *dec)
520{ 785{
521 HV *hv = newHV (); 786 HV *hv = newHV ();
522 787
523 DEC_INC_DEPTH; 788 DEC_INC_DEPTH;
524 789
525 if ((*dec->cur & 31) == 31) 790 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
526 { 791 {
527 ++dec->cur; 792 ++dec->cur;
528 793
529 for (;;) 794 for (;;)
530 { 795 {
531 WANT (1); 796 WANT (1);
532 797
533 if (*dec->cur == (0xe0 | 31)) 798 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
534 { 799 {
535 ++dec->cur; 800 ++dec->cur;
536 break; 801 break;
537 } 802 }
538 803
539 SV *k = decode_sv (dec); 804 decode_he (dec, hv);
540 SV *v = decode_sv (dec);
541
542 hv_store_ent (hv, k, v, 0);
543 } 805 }
544 } 806 }
545 else 807 else
546 { 808 {
547 int len = decode_uint (dec); 809 int pairs = decode_uint (dec);
548 810
549 while (len--) 811 while (pairs--)
550 { 812 decode_he (dec, hv);
551 SV *k = decode_sv (dec);
552 SV *v = decode_sv (dec);
553
554 hv_store_ent (hv, k, v, 0);
555 }
556 } 813 }
557 814
558 DEC_DEC_DEPTH; 815 DEC_DEC_DEPTH;
559 return newRV_noinc ((SV *)hv); 816 return newRV_noinc ((SV *)hv);
560 817
567static SV * 824static SV *
568decode_str (dec_t *dec, int utf8) 825decode_str (dec_t *dec, int utf8)
569{ 826{
570 SV *sv = 0; 827 SV *sv = 0;
571 828
572 if ((*dec->cur & 31) == 31) 829 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
573 { 830 {
831 // indefinite length strings
574 ++dec->cur; 832 ++dec->cur;
575 833
834 U8 major = *dec->cur & MAJOR_MISC;
835
576 sv = newSVpvn ("", 0); 836 sv = newSVpvn ("", 0);
577 837
578 // not very fast, and certainly not robust against illegal input
579 for (;;) 838 for (;;)
580 { 839 {
581 WANT (1); 840 WANT (1);
582 841
583 if (*dec->cur == (0xe0 | 31)) 842 if ((*dec->cur - major) > LENGTH_EXT8)
843 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
584 { 844 {
585 ++dec->cur; 845 ++dec->cur;
586 break; 846 break;
587 } 847 }
848 else
849 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
588 850
589 sv_catsv (sv, decode_sv (dec)); 851 STRLEN len = decode_uint (dec);
852
853 WANT (len);
854 sv_catpvn (sv, dec->cur, len);
855 dec->cur += len;
590 } 856 }
591 } 857 }
592 else 858 else
593 { 859 {
594 STRLEN len = decode_uint (dec); 860 STRLEN len = decode_uint (dec);
595 861
596 WANT (len); 862 WANT (len);
597 sv = newSVpvn (dec->cur, len); 863 sv = newSVpvn (dec->cur, len);
598 dec->cur += len; 864 dec->cur += len;
865
866 if (ecb_expect_false (dec->stringref)
867 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
868 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
599 } 869 }
600 870
601 if (utf8) 871 if (utf8)
872 {
873 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
874 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
875 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
876
602 SvUTF8_on (sv); 877 SvUTF8_on (sv);
878 }
603 879
604 return sv; 880 return sv;
605 881
606fail: 882fail:
607 SvREFCNT_dec (sv); 883 SvREFCNT_dec (sv);
609} 885}
610 886
611static SV * 887static SV *
612decode_tagged (dec_t *dec) 888decode_tagged (dec_t *dec)
613{ 889{
890 SV *sv = 0;
614 UV tag = decode_uint (dec); 891 UV tag = decode_uint (dec);
892
893 WANT (1);
894
895 switch (tag)
896 {
897 case CBOR_TAG_MAGIC:
615 SV *sv = decode_sv (dec); 898 sv = decode_sv (dec);
899 break;
616 900
617 if (tag == 55799) // 2.4.5 Self-Describe CBOR 901 case CBOR_TAG_INDIRECTION:
902 sv = newRV_noinc (decode_sv (dec));
903 break;
904
905 case CBOR_TAG_STRINGREF_NAMESPACE:
906 {
907 ENTER; SAVETMPS;
908
909 SAVESPTR (dec->stringref);
910 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
911
912 sv = decode_sv (dec);
913
914 FREETMPS; LEAVE;
915 }
916 break;
917
918 case CBOR_TAG_STRINGREF:
919 {
920 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
921 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
922
923 UV idx = decode_uint (dec);
924
925 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
926 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
927
928 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
929 }
930 break;
931
932 case CBOR_TAG_VALUE_SHAREABLE:
933 {
934 if (ecb_expect_false (!dec->shareable))
935 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
936
937 if (dec->cbor.flags & F_ALLOW_CYCLES)
938 {
939 sv = newSV (0);
940 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
941
942 SV *osv = decode_sv (dec);
943 sv_setsv (sv, osv);
944 SvREFCNT_dec_NN (osv);
945 }
946 else
947 {
948 av_push (dec->shareable, &PL_sv_undef);
949 int idx = AvFILLp (dec->shareable);
950 sv = decode_sv (dec);
951 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
952 }
953 }
954 break;
955
956 case CBOR_TAG_VALUE_SHAREDREF:
957 {
958 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
959 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
960
961 UV idx = decode_uint (dec);
962
963 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
964 ERR ("corrupted CBOR data (sharedref index out of bounds)");
965
966 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
967
968 if (sv == &PL_sv_undef)
969 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
970 }
971 break;
972
973 case CBOR_TAG_PERL_OBJECT:
974 {
975 sv = decode_sv (dec);
976
977 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
978 ERR ("corrupted CBOR data (non-array perl object)");
979
980 AV *av = (AV *)SvRV (sv);
981 int len = av_len (av) + 1;
982 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
983
984 if (!stash)
985 ERR ("cannot decode perl-object (package does not exist)");
986
987 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
988
989 if (!method)
990 ERR ("cannot decode perl-object (package does not have a THAW method)");
991
992 dSP;
993
994 ENTER; SAVETMPS; PUSHMARK (SP);
995 EXTEND (SP, len + 1);
996 // we re-bless the reference to get overload and other niceties right
997 PUSHs (*av_fetch (av, 0, 1));
998 PUSHs (sv_cbor);
999
1000 int i;
1001
1002 for (i = 1; i < len; ++i)
1003 PUSHs (*av_fetch (av, i, 1));
1004
1005 PUTBACK;
1006 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
1007 SPAGAIN;
1008
1009 if (SvTRUE (ERRSV))
1010 {
1011 FREETMPS; LEAVE;
1012 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1013 }
1014
1015 SvREFCNT_dec (sv);
1016 sv = SvREFCNT_inc (POPs);
1017
1018 PUTBACK;
1019
1020 FREETMPS; LEAVE;
1021 }
1022 break;
1023
1024 default:
1025 {
1026 sv = decode_sv (dec);
1027
1028 dSP;
1029 ENTER; SAVETMPS; PUSHMARK (SP);
1030 EXTEND (SP, 2);
1031 PUSHs (newSVuv (tag));
1032 PUSHs (sv);
1033
1034 PUTBACK;
1035 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1036 SPAGAIN;
1037
1038 if (SvTRUE (ERRSV))
1039 {
1040 FREETMPS; LEAVE;
1041 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1042 }
1043
1044 if (count)
1045 {
1046 SvREFCNT_dec (sv);
1047 sv = SvREFCNT_inc (POPs);
1048 }
1049 else
1050 {
1051 AV *av = newAV ();
1052 av_push (av, newSVuv (tag));
1053 av_push (av, sv);
1054
1055 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
1056 ? cbor_tagged_stash
1057 : gv_stashpv ("CBOR::XS::Tagged" , 1);
1058 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
1059 }
1060
1061 PUTBACK;
1062
1063 FREETMPS; LEAVE;
1064 }
1065 break;
1066 }
1067
618 return sv; 1068 return sv;
619 1069
620 AV *av = newAV (); 1070fail:
621 av_push (av, newSVuv (tag)); 1071 SvREFCNT_dec (sv);
622 av_push (av, sv); 1072 return &PL_sv_undef;
623 return newRV_noinc ((SV *)av);
624} 1073}
625 1074
626static SV * 1075static SV *
627decode_sv (dec_t *dec) 1076decode_sv (dec_t *dec)
628{ 1077{
629 WANT (1); 1078 WANT (1);
630 1079
631 switch (*dec->cur >> 5) 1080 switch (*dec->cur >> MAJOR_SHIFT)
632 { 1081 {
633 case 0: // unsigned int 1082 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
634 return newSVuv (decode_uint (dec)); 1083 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
635 case 1: // negative int 1084 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
636 return newSViv (-1 - (IV)decode_uint (dec)); 1085 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
637 case 2: // octet string 1086 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
638 return decode_str (dec, 0); 1087 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
639 case 3: // utf-8 string 1088 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
640 return decode_str (dec, 1); 1089
641 case 4: // array 1090 case MAJOR_MISC >> MAJOR_SHIFT:
642 return decode_av (dec);
643 case 5: // map
644 return decode_hv (dec);
645 case 6: // tag
646 return decode_tagged (dec);
647 case 7: // misc
648 switch (*dec->cur++ & 31) 1091 switch (*dec->cur++ & MINOR_MASK)
649 { 1092 {
650 case 20: 1093 case SIMPLE_FALSE:
651#if CBOR_SLOW 1094#if CBOR_SLOW
652 cbor_false = get_bool ("CBOR::XS::false"); 1095 types_false = get_bool ("Types::Serialiser::false");
653#endif 1096#endif
654 return newSVsv (cbor_false); 1097 return newSVsv (types_false);
655 case 21: 1098 case SIMPLE_TRUE:
656#if CBOR_SLOW 1099#if CBOR_SLOW
657 cbor_true = get_bool ("CBOR::XS::true"); 1100 types_true = get_bool ("Types::Serialiser::true");
658#endif 1101#endif
659 return newSVsv (cbor_true); 1102 return newSVsv (types_true);
660 case 22: 1103 case SIMPLE_NULL:
661 return newSVsv (&PL_sv_undef); 1104 return newSVsv (&PL_sv_undef);
1105 case SIMPLE_UNDEF:
1106#if CBOR_SLOW
1107 types_error = get_bool ("Types::Serialiser::error");
1108#endif
1109 return newSVsv (types_error);
662 1110
663 case 25: 1111 case MISC_FLOAT16:
664 { 1112 {
665 WANT (2); 1113 WANT (2);
666 1114
667 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1115 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
668 dec->cur += 2; 1116 dec->cur += 2;
669 1117
670 return newSVnv (ecb_binary16_to_float (fp)); 1118 return newSVnv (ecb_binary16_to_float (fp));
671 } 1119 }
672 1120
673 case 26: 1121 case MISC_FLOAT32:
674 { 1122 {
675 uint32_t fp; 1123 uint32_t fp;
676 WANT (4); 1124 WANT (4);
677 memcpy (&fp, dec->cur, 4); 1125 memcpy (&fp, dec->cur, 4);
678 dec->cur += 4; 1126 dec->cur += 4;
681 fp = ecb_bswap32 (fp); 1129 fp = ecb_bswap32 (fp);
682 1130
683 return newSVnv (ecb_binary32_to_float (fp)); 1131 return newSVnv (ecb_binary32_to_float (fp));
684 } 1132 }
685 1133
686 case 27: 1134 case MISC_FLOAT64:
687 { 1135 {
688 uint64_t fp; 1136 uint64_t fp;
689 WANT (8); 1137 WANT (8);
690 memcpy (&fp, dec->cur, 8); 1138 memcpy (&fp, dec->cur, 8);
691 dec->cur += 8; 1139 dec->cur += 8;
694 fp = ecb_bswap64 (fp); 1142 fp = ecb_bswap64 (fp);
695 1143
696 return newSVnv (ecb_binary64_to_double (fp)); 1144 return newSVnv (ecb_binary64_to_double (fp));
697 } 1145 }
698 1146
699 // 0..19 unassigned 1147 // 0..19 unassigned simple
700 // 24 reserved + unassigned (reserved values are not encodable) 1148 // 24 reserved + unassigned simple (reserved values are not encodable)
1149 // 28-30 unassigned misc
1150 // 31 break code
701 default: 1151 default:
702 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1152 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
703 } 1153 }
704 1154
705 break; 1155 break;
706 } 1156 }
707 1157
710} 1160}
711 1161
712static SV * 1162static SV *
713decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1163decode_cbor (SV *string, CBOR *cbor, char **offset_return)
714{ 1164{
715 dec_t dec; 1165 dec_t dec = { };
716 SV *sv; 1166 SV *sv;
1167 STRLEN len;
1168 char *data = SvPVbyte (string, len);
717 1169
718 /* work around bugs in 5.10 where manipulating magic values
719 * makes perl ignore the magic in subsequent accesses.
720 * also make a copy of non-PV values, to get them into a clean
721 * state (SvPV should do that, but it's buggy, see below).
722 */
723 /*SvGETMAGIC (string);*/
724 if (SvMAGICAL (string) || !SvPOK (string))
725 string = sv_2mortal (newSVsv (string));
726
727 SvUPGRADE (string, SVt_PV);
728
729 /* work around a bug in perl 5.10, which causes SvCUR to fail an
730 * assertion with -DDEBUGGING, although SvCUR is documented to
731 * return the xpv_cur field which certainly exists after upgrading.
732 * according to nicholas clark, calling SvPOK fixes this.
733 * But it doesn't fix it, so try another workaround, call SvPV_nolen
734 * and hope for the best.
735 * Damnit, SvPV_nolen still trips over yet another assertion. This
736 * assertion business is seriously broken, try yet another workaround
737 * for the broken -DDEBUGGING.
738 */
739 {
740#ifdef DEBUGGING
741 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
742#else
743 STRLEN offset = SvCUR (string);
744#endif
745
746 if (offset > cbor->max_size && cbor->max_size) 1170 if (len > cbor->max_size && cbor->max_size)
747 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu", 1171 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
748 (unsigned long)SvCUR (string), (unsigned long)cbor->max_size); 1172 (unsigned long)len, (unsigned long)cbor->max_size);
749 }
750
751 sv_utf8_downgrade (string, 0);
752 1173
753 dec.cbor = *cbor; 1174 dec.cbor = *cbor;
754 dec.cur = (U8 *)SvPVX (string); 1175 dec.cur = (U8 *)data;
755 dec.end = (U8 *)SvEND (string); 1176 dec.end = (U8 *)data + len;
756 dec.err = 0;
757 dec.depth = 0;
758 1177
759 sv = decode_sv (&dec); 1178 sv = decode_sv (&dec);
760 1179
761 if (offset_return) 1180 if (offset_return)
762 *offset_return = dec.cur; 1181 *offset_return = dec.cur;
765 if (dec.cur != dec.end && !dec.err) 1184 if (dec.cur != dec.end && !dec.err)
766 dec.err = "garbage after CBOR object"; 1185 dec.err = "garbage after CBOR object";
767 1186
768 if (dec.err) 1187 if (dec.err)
769 { 1188 {
1189 if (dec.shareable)
1190 {
1191 // need to break cyclic links, which whould all be in shareable
1192 int i;
1193 SV **svp;
1194
1195 for (i = av_len (dec.shareable) + 1; i--; )
1196 if ((svp = av_fetch (dec.shareable, i, 0)))
1197 sv_setsv (*svp, &PL_sv_undef);
1198 }
1199
770 SvREFCNT_dec (sv); 1200 SvREFCNT_dec (sv);
771 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)SvPVX (string), (int)(uint8_t)*dec.cur); 1201 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
772 } 1202 }
773 1203
774 sv = sv_2mortal (sv); 1204 sv = sv_2mortal (sv);
775 1205
776 return sv; 1206 return sv;
777} 1207}
778 1208
1209/////////////////////////////////////////////////////////////////////////////
1210// incremental parser
1211
1212#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1213
1214// returns 0 for notyet, 1 for success or error
1215static int
1216incr_parse (CBOR *self, SV *cborstr)
1217{
1218 STRLEN cur;
1219 SvPV (cborstr, cur);
1220
1221 while (ecb_expect_true (self->incr_need <= cur))
1222 {
1223 // table of integer count bytes
1224 static I8 incr_len[MINOR_MASK + 1] = {
1225 0, 0, 0, 0, 0, 0, 0, 0,
1226 0, 0, 0, 0, 0, 0, 0, 0,
1227 0, 0, 0, 0, 0, 0, 0, 0,
1228 1, 2, 4, 8,-1,-1,-1,-2
1229 };
1230
1231 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1232 U8 m = *p & MINOR_MASK;
1233 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1234 I8 ilen = incr_len[m];
1235
1236 self->incr_need = self->incr_pos + 1;
1237
1238 if (ecb_expect_false (ilen < 0))
1239 {
1240 if (m != MINOR_INDEF)
1241 return 1; // error
1242
1243 if (*p == (MAJOR_MISC | MINOR_INDEF))
1244 {
1245 if (count >= 0)
1246 return 1; // error
1247
1248 count = 1;
1249 }
1250 else
1251 {
1252 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1253 count = -1;
1254 }
1255 }
1256 else
1257 {
1258 self->incr_need += ilen;
1259 if (ecb_expect_false (self->incr_need > cur))
1260 return 0;
1261
1262 int major = *p >> MAJOR_SHIFT;
1263
1264 switch (major)
1265 {
1266 case MAJOR_BYTES >> MAJOR_SHIFT:
1267 case MAJOR_TEXT >> MAJOR_SHIFT:
1268 case MAJOR_ARRAY >> MAJOR_SHIFT:
1269 case MAJOR_MAP >> MAJOR_SHIFT:
1270 {
1271 UV len;
1272
1273 if (ecb_expect_false (ilen))
1274 {
1275 len = 0;
1276
1277 do {
1278 len = (len << 8) | *++p;
1279 } while (--ilen);
1280 }
1281 else
1282 len = m;
1283
1284 switch (major)
1285 {
1286 case MAJOR_BYTES >> MAJOR_SHIFT:
1287 case MAJOR_TEXT >> MAJOR_SHIFT:
1288 self->incr_need += len;
1289 if (ecb_expect_false (self->incr_need > cur))
1290 return 0;
1291
1292 break;
1293
1294 case MAJOR_MAP >> MAJOR_SHIFT:
1295 len <<= 1;
1296 case MAJOR_ARRAY >> MAJOR_SHIFT:
1297 if (len)
1298 {
1299 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1300 count = len + 1;
1301 }
1302 break;
1303 }
1304 }
1305 }
1306 }
1307
1308 self->incr_pos = self->incr_need;
1309
1310 if (count > 0)
1311 {
1312 while (!--count)
1313 {
1314 if (!AvFILLp (self->incr_count))
1315 return 1; // done
1316
1317 SvREFCNT_dec_NN (av_pop (self->incr_count));
1318 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1319 }
1320
1321 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1322 }
1323 }
1324
1325 return 0;
1326}
1327
1328
779///////////////////////////////////////////////////////////////////////////// 1329/////////////////////////////////////////////////////////////////////////////
780// XS interface functions 1330// XS interface functions
781 1331
782MODULE = CBOR::XS PACKAGE = CBOR::XS 1332MODULE = CBOR::XS PACKAGE = CBOR::XS
783 1333
784BOOT: 1334BOOT:
785{ 1335{
786 cbor_stash = gv_stashpv ("CBOR::XS" , 1); 1336 cbor_stash = gv_stashpv ("CBOR::XS" , 1);
787 cbor_boolean_stash = gv_stashpv ("CBOR::XS::Boolean", 1);
788 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1); 1337 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1);
789 1338
790 cbor_true = get_bool ("CBOR::XS::true"); 1339 types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
791 cbor_false = get_bool ("CBOR::XS::false"); 1340 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
1341
1342 types_true = get_bool ("Types::Serialiser::true" );
1343 types_false = get_bool ("Types::Serialiser::false");
1344 types_error = get_bool ("Types::Serialiser::error");
1345
1346 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1347
1348 sv_cbor = newSVpv ("CBOR", 0);
1349 SvREADONLY_on (sv_cbor);
792} 1350}
793 1351
794PROTOTYPES: DISABLE 1352PROTOTYPES: DISABLE
795 1353
796void CLONE (...) 1354void CLONE (...)
797 CODE: 1355 CODE:
798 cbor_stash = 0; 1356 cbor_stash = 0;
799 cbor_boolean_stash = 0;
800 cbor_tagged_stash = 0; 1357 cbor_tagged_stash = 0;
1358 types_error_stash = 0;
1359 types_boolean_stash = 0;
801 1360
802void new (char *klass) 1361void new (char *klass)
803 PPCODE: 1362 PPCODE:
804{ 1363{
805 SV *pv = NEWSV (0, sizeof (CBOR)); 1364 SV *pv = NEWSV (0, sizeof (CBOR));
813 1372
814void shrink (CBOR *self, int enable = 1) 1373void shrink (CBOR *self, int enable = 1)
815 ALIAS: 1374 ALIAS:
816 shrink = F_SHRINK 1375 shrink = F_SHRINK
817 allow_unknown = F_ALLOW_UNKNOWN 1376 allow_unknown = F_ALLOW_UNKNOWN
1377 allow_sharing = F_ALLOW_SHARING
1378 allow_cycles = F_ALLOW_CYCLES
1379 pack_strings = F_PACK_STRINGS
1380 validate_utf8 = F_VALIDATE_UTF8
818 PPCODE: 1381 PPCODE:
819{ 1382{
820 if (enable) 1383 if (enable)
821 self->flags |= ix; 1384 self->flags |= ix;
822 else 1385 else
827 1390
828void get_shrink (CBOR *self) 1391void get_shrink (CBOR *self)
829 ALIAS: 1392 ALIAS:
830 get_shrink = F_SHRINK 1393 get_shrink = F_SHRINK
831 get_allow_unknown = F_ALLOW_UNKNOWN 1394 get_allow_unknown = F_ALLOW_UNKNOWN
1395 get_allow_sharing = F_ALLOW_SHARING
1396 get_allow_cycles = F_ALLOW_CYCLES
1397 get_pack_strings = F_PACK_STRINGS
1398 get_validate_utf8 = F_VALIDATE_UTF8
832 PPCODE: 1399 PPCODE:
833 XPUSHs (boolSV (self->flags & ix)); 1400 XPUSHs (boolSV (self->flags & ix));
834 1401
835void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1402void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
836 PPCODE: 1403 PPCODE:
852 CODE: 1419 CODE:
853 RETVAL = self->max_size; 1420 RETVAL = self->max_size;
854 OUTPUT: 1421 OUTPUT:
855 RETVAL 1422 RETVAL
856 1423
857#if 0 //TODO 1424void filter (CBOR *self, SV *filter = 0)
858
859void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef)
860 PPCODE: 1425 PPCODE:
861{
862 SvREFCNT_dec (self->cb_object); 1426 SvREFCNT_dec (self->filter);
863 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; 1427 self->filter = filter ? newSVsv (filter) : filter;
864
865 XPUSHs (ST (0)); 1428 XPUSHs (ST (0));
866}
867 1429
868void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef) 1430SV *get_filter (CBOR *self)
869 PPCODE: 1431 CODE:
870{ 1432 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
871 if (!self->cb_sk_object) 1433 OUTPUT:
872 self->cb_sk_object = newHV (); 1434 RETVAL
873
874 if (SvOK (cb))
875 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
876 else
877 {
878 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
879
880 if (!HvKEYS (self->cb_sk_object))
881 {
882 SvREFCNT_dec (self->cb_sk_object);
883 self->cb_sk_object = 0;
884 }
885 }
886
887 XPUSHs (ST (0));
888}
889
890#endif
891 1435
892void encode (CBOR *self, SV *scalar) 1436void encode (CBOR *self, SV *scalar)
893 PPCODE: 1437 PPCODE:
894 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; 1438 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
895 XPUSHs (scalar); 1439 XPUSHs (scalar);
908 EXTEND (SP, 2); 1452 EXTEND (SP, 2);
909 PUSHs (sv); 1453 PUSHs (sv);
910 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1454 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
911} 1455}
912 1456
913#if 0 1457void incr_parse (CBOR *self, SV *cborstr)
1458 ALIAS:
1459 incr_parse_multiple = 1
1460 PPCODE:
1461{
1462 if (SvUTF8 (cborstr))
1463 sv_utf8_downgrade (cborstr, 0);
1464
1465 if (!self->incr_count)
1466 {
1467 self->incr_count = newAV ();
1468 self->incr_pos = 0;
1469 self->incr_need = 1;
1470
1471 av_push (self->incr_count, newSViv (1));
1472 }
1473
1474 do
1475 {
1476 if (!incr_parse (self, cborstr))
1477 {
1478 if (self->incr_need > self->max_size && self->max_size)
1479 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1480 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1481
1482 break;
1483 }
1484
1485 SV *sv;
1486 char *offset;
1487
1488 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1489 XPUSHs (sv);
1490
1491 sv_chop (cborstr, offset);
1492
1493 av_clear (self->incr_count);
1494 av_push (self->incr_count, newSViv (1));
1495
1496 self->incr_pos = 0;
1497 self->incr_need = self->incr_pos + 1;
1498 }
1499 while (ix);
1500}
1501
1502void incr_reset (CBOR *self)
1503 CODE:
1504{
1505 SvREFCNT_dec (self->incr_count);
1506 self->incr_count = 0;
1507}
914 1508
915void DESTROY (CBOR *self) 1509void DESTROY (CBOR *self)
916 CODE: 1510 PPCODE:
917 SvREFCNT_dec (self->cb_sk_object); 1511 cbor_free (self);
918 SvREFCNT_dec (self->cb_object);
919
920#endif
921 1512
922PROTOTYPES: ENABLE 1513PROTOTYPES: ENABLE
923 1514
924void encode_cbor (SV *scalar) 1515void encode_cbor (SV *scalar)
1516 ALIAS:
1517 encode_cbor = 0
1518 encode_cbor_sharing = F_ALLOW_SHARING
925 PPCODE: 1519 PPCODE:
926{ 1520{
927 CBOR cbor; 1521 CBOR cbor;
928 cbor_init (&cbor); 1522 cbor_init (&cbor);
1523 cbor.flags |= ix;
929 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1524 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
930 XPUSHs (scalar); 1525 XPUSHs (scalar);
931} 1526}
932 1527
933void decode_cbor (SV *cborstr) 1528void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines