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 1.49 by root, Mon Feb 8 04:11:11 2016 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 {
338 SV *sv = AvARRAY (av)[i];
339 encode_sv (enc, sv ? sv : &PL_sv_undef);
340 }
186 341
187 --enc->depth; 342 --enc->depth;
188} 343}
189 344
190static void 345static void
199 354
200 int pairs = hv_iterinit (hv); 355 int pairs = hv_iterinit (hv);
201 int mg = SvMAGICAL (hv); 356 int mg = SvMAGICAL (hv);
202 357
203 if (mg) 358 if (mg)
204 encode_ch (enc, 0xa0 | 31); 359 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
205 else 360 else
206 encode_uint (enc, 0xa0, pairs); 361 encode_uint (enc, MAJOR_MAP, pairs);
207 362
208 while ((he = hv_iternext (hv))) 363 while ((he = hv_iternext (hv)))
209 { 364 {
210 if (HeKLEN (he) == HEf_SVKEY) 365 if (HeKLEN (he) == HEf_SVKEY)
211 encode_sv (enc, HeSVKEY (he)); 366 encode_sv (enc, HeSVKEY (he));
212 else 367 else
213 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 368 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
214 369
215 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 370 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
216 } 371 }
217 372
218 if (mg) 373 if (mg)
219 encode_ch (enc, 0xe0 | 31); 374 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
220 375
221 --enc->depth; 376 --enc->depth;
222} 377}
223 378
224// encode objects, arrays and special \0=false and \1=true values. 379// encode objects, arrays and special \0=false and \1=true values.
225static void 380static void
226encode_rv (enc_t *enc, SV *sv) 381encode_rv (enc_t *enc, SV *sv)
227{ 382{
228 svtype svt;
229
230 SvGETMAGIC (sv); 383 SvGETMAGIC (sv);
384
231 svt = SvTYPE (sv); 385 svtype svt = SvTYPE (sv);
232 386
233 if (ecb_expect_false (SvOBJECT (sv))) 387 if (ecb_expect_false (SvOBJECT (sv)))
234 { 388 {
235 HV *boolean_stash = !CBOR_SLOW || cbor_boolean_stash 389 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
236 ? cbor_boolean_stash 390 ? types_boolean_stash
237 : gv_stashpv ("CBOR::XS::Boolean", 1); 391 : gv_stashpv ("Types::Serialiser::Boolean", 1);
392 HV *error_stash = !CBOR_SLOW || types_error_stash
393 ? types_error_stash
394 : gv_stashpv ("Types::Serialiser::Error", 1);
238 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 395 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
239 ? cbor_tagged_stash 396 ? cbor_tagged_stash
240 : gv_stashpv ("CBOR::XS::Tagged" , 1); 397 : gv_stashpv ("CBOR::XS::Tagged" , 1);
241 398
399 HV *stash = SvSTASH (sv);
400
242 if (SvSTASH (sv) == boolean_stash) 401 if (stash == boolean_stash)
243 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 402 {
403 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
404 return;
405 }
406 else if (stash == error_stash)
407 {
408 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
409 return;
410 }
244 else if (SvSTASH (sv) == tagged_stash) 411 else if (stash == tagged_stash)
245 { 412 {
246 if (svt != SVt_PVAV) 413 if (svt != SVt_PVAV)
247 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 414 croak ("encountered CBOR::XS::Tagged object that isn't an array");
248 415
249 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 416 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
250 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 417 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
418
419 return;
420 }
421 }
422
423 if (ecb_expect_false (SvREFCNT (sv) > 1)
424 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
425 {
426 if (!enc->shareable)
427 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
428
429 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
430
431 if (SvOK (*svp))
432 {
433 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
434 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
435 return;
251 } 436 }
252 else 437 else
253 { 438 {
439 sv_setuv (*svp, enc->shareable_idx);
440 ++enc->shareable_idx;
441 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
442 }
443 }
444
445 if (ecb_expect_false (SvOBJECT (sv)))
446 {
447 HV *stash = SvSTASH (sv);
448 GV *method;
449
450 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
451 {
452 dSP;
453
454 ENTER; SAVETMPS; PUSHMARK (SP);
254 // we re-bless the reference to get overload and other niceties right 455 // 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))); 456 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
263 457
264 // calling with G_SCALAR ensures that we always get a 1 return value
265 PUTBACK; 458 PUTBACK;
459 // G_SCALAR ensures that return value is 1
266 call_sv ((SV *)GvCV (to_cbor), G_SCALAR); 460 call_sv ((SV *)GvCV (method), G_SCALAR);
267 SPAGAIN; 461 SPAGAIN;
268 462
269 // catch this surprisingly common error 463 // catch this surprisingly common error
270 if (SvROK (TOPs) && SvRV (TOPs) == sv) 464 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))); 465 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (stash));
272 466
273 sv = POPs; 467 encode_sv (enc, POPs);
468
274 PUTBACK; 469 PUTBACK;
275 470
276 encode_sv (enc, sv);
277
278 FREETMPS; LEAVE; 471 FREETMPS; LEAVE;
279 } 472 }
473 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
474 {
475 dSP;
476
477 ENTER; SAVETMPS; PUSHMARK (SP);
478 EXTEND (SP, 2);
479 // we re-bless the reference to get overload and other niceties right
480 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
481 PUSHs (sv_cbor);
482
483 PUTBACK;
484 int count = call_sv ((SV *)GvCV (method), G_ARRAY);
485 SPAGAIN;
486
487 // catch this surprisingly common error
488 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
489 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
490
491 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
492 encode_uint (enc, MAJOR_ARRAY, count + 1);
493 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
494
495 while (count)
496 encode_sv (enc, SP[1 - count--]);
497
498 PUTBACK;
499
500 FREETMPS; LEAVE;
501 }
280 else 502 else
281 croak ("encountered object '%s', but no TO_CBOR method available on it", 503 croak ("encountered object '%s', but no TO_CBOR or FREEZE methods available on it",
282 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 504 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
283 }
284 } 505 }
285 else if (svt == SVt_PVHV) 506 else if (svt == SVt_PVHV)
286 encode_hv (enc, (HV *)sv); 507 encode_hv (enc, (HV *)sv);
287 else if (svt == SVt_PVAV) 508 else if (svt == SVt_PVAV)
288 encode_av (enc, (AV *)sv); 509 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 510 else
307 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 511 {
308 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 512 encode_tag (enc, CBOR_TAG_INDIRECTION);
513 encode_sv (enc, sv);
514 }
309} 515}
310 516
311static void 517static void
312encode_nv (enc_t *enc, SV *sv) 518encode_nv (enc_t *enc, SV *sv)
313{ 519{
314 double nv = SvNVX (sv); 520 double nv = SvNVX (sv);
315 521
316 need (enc, 9); 522 need (enc, 9);
317 523
318 if (ecb_expect_false (nv == (U32)nv)) 524 if (ecb_expect_false (nv == (NV)(U32)nv))
319 encode_uint (enc, 0x00, (U32)nv); 525 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
320 //TODO: maybe I32? 526 //TODO: maybe I32?
321 else if (ecb_expect_false (nv == (float)nv)) 527 else if (ecb_expect_false (nv == (float)nv))
322 { 528 {
323 uint32_t fp = ecb_float_to_binary32 (nv); 529 uint32_t fp = ecb_float_to_binary32 (nv);
324 530
325 *enc->cur++ = 0xe0 | 26; 531 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
326 532
327 if (!ecb_big_endian ()) 533 if (!ecb_big_endian ())
328 fp = ecb_bswap32 (fp); 534 fp = ecb_bswap32 (fp);
329 535
330 memcpy (enc->cur, &fp, 4); 536 memcpy (enc->cur, &fp, 4);
332 } 538 }
333 else 539 else
334 { 540 {
335 uint64_t fp = ecb_double_to_binary64 (nv); 541 uint64_t fp = ecb_double_to_binary64 (nv);
336 542
337 *enc->cur++ = 0xe0 | 27; 543 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
338 544
339 if (!ecb_big_endian ()) 545 if (!ecb_big_endian ())
340 fp = ecb_bswap64 (fp); 546 fp = ecb_bswap64 (fp);
341 547
342 memcpy (enc->cur, &fp, 8); 548 memcpy (enc->cur, &fp, 8);
351 557
352 if (SvPOKp (sv)) 558 if (SvPOKp (sv))
353 { 559 {
354 STRLEN len; 560 STRLEN len;
355 char *str = SvPV (sv, len); 561 char *str = SvPV (sv, len);
356 encode_str (enc, SvUTF8 (sv), str, len); 562 encode_strref (enc, SvUTF8 (sv), str, len);
357 } 563 }
358 else if (SvNOKp (sv)) 564 else if (SvNOKp (sv))
359 encode_nv (enc, sv); 565 encode_nv (enc, sv);
360 else if (SvIOKp (sv)) 566 else if (SvIOKp (sv))
361 { 567 {
362 if (SvIsUV (sv)) 568 if (SvIsUV (sv))
363 encode_uint (enc, 0x00, SvUVX (sv)); 569 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
364 else if (SvIVX (sv) >= 0) 570 else if (SvIVX (sv) >= 0)
365 encode_uint (enc, 0x00, SvIVX (sv)); 571 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
366 else 572 else
367 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 573 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
368 } 574 }
369 else if (SvROK (sv)) 575 else if (SvROK (sv))
370 encode_rv (enc, SvRV (sv)); 576 encode_rv (enc, SvRV (sv));
371 else if (!SvOK (sv)) 577 else if (!SvOK (sv))
372 encode_ch (enc, 0xe0 | 22); 578 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
373 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 579 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
374 encode_ch (enc, 0xe0 | 23); 580 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
375 else 581 else
376 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 582 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
377 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 583 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
378} 584}
379 585
380static SV * 586static SV *
381encode_cbor (SV *scalar, CBOR *cbor) 587encode_cbor (SV *scalar, CBOR *cbor)
382{ 588{
383 enc_t enc; 589 enc_t enc = { 0 };
384 590
385 enc.cbor = *cbor; 591 enc.cbor = *cbor;
386 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 592 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
387 enc.cur = SvPVX (enc.sv); 593 enc.cur = SvPVX (enc.sv);
388 enc.end = SvEND (enc.sv); 594 enc.end = SvEND (enc.sv);
389 enc.depth = 0;
390 595
391 SvPOK_only (enc.sv); 596 SvPOK_only (enc.sv);
597
598 if (cbor->flags & F_PACK_STRINGS)
599 {
600 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
601 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
602 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
603 }
604
392 encode_sv (&enc, scalar); 605 encode_sv (&enc, scalar);
393 606
394 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 607 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
395 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 608 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
396 609
410 U8 *end; // end of input string 623 U8 *end; // end of input string
411 const char *err; // parse error, if != 0 624 const char *err; // parse error, if != 0
412 CBOR cbor; 625 CBOR cbor;
413 U32 depth; // recursion depth 626 U32 depth; // recursion depth
414 U32 maxdepth; // recursion depth limit 627 U32 maxdepth; // recursion depth limit
628 AV *shareable;
629 AV *stringref;
630 SV *decode_tagged;
415} dec_t; 631} dec_t;
416 632
417#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 633#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE
418 634
419#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 635#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 638#define DEC_DEC_DEPTH --dec->depth
423 639
424static UV 640static UV
425decode_uint (dec_t *dec) 641decode_uint (dec_t *dec)
426{ 642{
427 switch (*dec->cur & 31) 643 U8 m = *dec->cur & MINOR_MASK;
428 { 644 ++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;
433 645
434 case 24: 646 if (ecb_expect_true (m < LENGTH_EXT1))
647 return m;
648 else if (ecb_expect_true (m == LENGTH_EXT1))
649 {
435 WANT (2); 650 WANT (1);
436 dec->cur += 2; 651 dec->cur += 1;
437 return dec->cur[-1]; 652 return dec->cur[-1];
438 653 }
439 case 25: 654 else if (ecb_expect_true (m == LENGTH_EXT2))
655 {
440 WANT (3); 656 WANT (2);
441 dec->cur += 3; 657 dec->cur += 2;
442 return (((UV)dec->cur[-2]) << 8) 658 return (((UV)dec->cur[-2]) << 8)
443 | ((UV)dec->cur[-1]); 659 | ((UV)dec->cur[-1]);
444 660 }
445 case 26: 661 else if (ecb_expect_true (m == LENGTH_EXT4))
662 {
446 WANT (5); 663 WANT (4);
447 dec->cur += 5; 664 dec->cur += 4;
448 return (((UV)dec->cur[-4]) << 24) 665 return (((UV)dec->cur[-4]) << 24)
449 | (((UV)dec->cur[-3]) << 16) 666 | (((UV)dec->cur[-3]) << 16)
450 | (((UV)dec->cur[-2]) << 8) 667 | (((UV)dec->cur[-2]) << 8)
451 | ((UV)dec->cur[-1]); 668 | ((UV)dec->cur[-1]);
452 669 }
453 case 27: 670 else if (ecb_expect_true (m == LENGTH_EXT8))
671 {
454 WANT (9); 672 WANT (8);
455 dec->cur += 9; 673 dec->cur += 8;
674
675 return
676#if UVSIZE < 8
677 0
678#else
456 return (((UV)dec->cur[-8]) << 56) 679 (((UV)dec->cur[-8]) << 56)
457 | (((UV)dec->cur[-7]) << 48) 680 | (((UV)dec->cur[-7]) << 48)
458 | (((UV)dec->cur[-6]) << 40) 681 | (((UV)dec->cur[-6]) << 40)
459 | (((UV)dec->cur[-5]) << 32) 682 | (((UV)dec->cur[-5]) << 32)
683#endif
460 | (((UV)dec->cur[-4]) << 24) 684 | (((UV)dec->cur[-4]) << 24)
461 | (((UV)dec->cur[-3]) << 16) 685 | (((UV)dec->cur[-3]) << 16)
462 | (((UV)dec->cur[-2]) << 8) 686 | (((UV)dec->cur[-2]) << 8)
463 | ((UV)dec->cur[-1]); 687 | ((UV)dec->cur[-1]);
464 688 }
465 default: 689 else
466 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 690 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
467 }
468 691
469fail: 692fail:
470 return 0; 693 return 0;
471} 694}
472 695
477{ 700{
478 AV *av = newAV (); 701 AV *av = newAV ();
479 702
480 DEC_INC_DEPTH; 703 DEC_INC_DEPTH;
481 704
482 if ((*dec->cur & 31) == 31) 705 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
483 { 706 {
484 ++dec->cur; 707 ++dec->cur;
485 708
486 for (;;) 709 for (;;)
487 { 710 {
488 WANT (1); 711 WANT (1);
489 712
490 if (*dec->cur == (0xe0 | 31)) 713 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
491 { 714 {
492 ++dec->cur; 715 ++dec->cur;
493 break; 716 break;
494 } 717 }
495 718
498 } 721 }
499 else 722 else
500 { 723 {
501 int i, len = decode_uint (dec); 724 int i, len = decode_uint (dec);
502 725
726 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); 727 av_fill (av, len - 1);
504 728
505 for (i = 0; i < len; ++i) 729 for (i = 0; i < len; ++i)
506 AvARRAY (av)[i] = decode_sv (dec); 730 AvARRAY (av)[i] = decode_sv (dec);
507 } 731 }
513 SvREFCNT_dec (av); 737 SvREFCNT_dec (av);
514 DEC_DEC_DEPTH; 738 DEC_DEC_DEPTH;
515 return &PL_sv_undef; 739 return &PL_sv_undef;
516} 740}
517 741
742static void
743decode_he (dec_t *dec, HV *hv)
744{
745 // for speed reasons, we specialcase single-string
746 // byte or utf-8 strings as keys, but only when !stringref
747
748 if (ecb_expect_true (!dec->stringref))
749 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
750 {
751 I32 len = decode_uint (dec);
752 char *key = (char *)dec->cur;
753
754 WANT (len);
755 dec->cur += len;
756
757 hv_store (hv, key, len, decode_sv (dec), 0);
758
759 return;
760 }
761 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
762 {
763 I32 len = decode_uint (dec);
764 char *key = (char *)dec->cur;
765
766 WANT (len);
767 dec->cur += len;
768
769 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
770 if (!is_utf8_string (key, len))
771 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
772
773 hv_store (hv, key, -len, decode_sv (dec), 0);
774
775 return;
776 }
777
778 SV *k = decode_sv (dec);
779 SV *v = decode_sv (dec);
780
781 hv_store_ent (hv, k, v, 0);
782 SvREFCNT_dec (k);
783
784fail:
785 ;
786}
787
518static SV * 788static SV *
519decode_hv (dec_t *dec) 789decode_hv (dec_t *dec)
520{ 790{
521 HV *hv = newHV (); 791 HV *hv = newHV ();
522 792
523 DEC_INC_DEPTH; 793 DEC_INC_DEPTH;
524 794
525 if ((*dec->cur & 31) == 31) 795 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
526 { 796 {
527 ++dec->cur; 797 ++dec->cur;
528 798
529 for (;;) 799 for (;;)
530 { 800 {
531 WANT (1); 801 WANT (1);
532 802
533 if (*dec->cur == (0xe0 | 31)) 803 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
534 { 804 {
535 ++dec->cur; 805 ++dec->cur;
536 break; 806 break;
537 } 807 }
538 808
539 SV *k = decode_sv (dec); 809 decode_he (dec, hv);
540 SV *v = decode_sv (dec);
541
542 hv_store_ent (hv, k, v, 0);
543 } 810 }
544 } 811 }
545 else 812 else
546 { 813 {
547 int len = decode_uint (dec); 814 int pairs = decode_uint (dec);
548 815
549 while (len--) 816 while (pairs--)
550 { 817 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 } 818 }
557 819
558 DEC_DEC_DEPTH; 820 DEC_DEC_DEPTH;
559 return newRV_noinc ((SV *)hv); 821 return newRV_noinc ((SV *)hv);
560 822
567static SV * 829static SV *
568decode_str (dec_t *dec, int utf8) 830decode_str (dec_t *dec, int utf8)
569{ 831{
570 SV *sv = 0; 832 SV *sv = 0;
571 833
572 if ((*dec->cur & 31) == 31) 834 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
573 { 835 {
836 // indefinite length strings
574 ++dec->cur; 837 ++dec->cur;
575 838
839 U8 major = *dec->cur & MAJOR_MISC;
840
576 sv = newSVpvn ("", 0); 841 sv = newSVpvn ("", 0);
577 842
578 // not very fast, and certainly not robust against illegal input
579 for (;;) 843 for (;;)
580 { 844 {
581 WANT (1); 845 WANT (1);
582 846
583 if (*dec->cur == (0xe0 | 31)) 847 if ((*dec->cur - major) > LENGTH_EXT8)
848 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
584 { 849 {
585 ++dec->cur; 850 ++dec->cur;
586 break; 851 break;
587 } 852 }
853 else
854 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
588 855
589 sv_catsv (sv, decode_sv (dec)); 856 STRLEN len = decode_uint (dec);
857
858 WANT (len);
859 sv_catpvn (sv, dec->cur, len);
860 dec->cur += len;
590 } 861 }
591 } 862 }
592 else 863 else
593 { 864 {
594 STRLEN len = decode_uint (dec); 865 STRLEN len = decode_uint (dec);
595 866
596 WANT (len); 867 WANT (len);
597 sv = newSVpvn (dec->cur, len); 868 sv = newSVpvn (dec->cur, len);
598 dec->cur += len; 869 dec->cur += len;
870
871 if (ecb_expect_false (dec->stringref)
872 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
873 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
599 } 874 }
600 875
601 if (utf8) 876 if (utf8)
877 {
878 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
879 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
880 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
881
602 SvUTF8_on (sv); 882 SvUTF8_on (sv);
883 }
603 884
604 return sv; 885 return sv;
605 886
606fail: 887fail:
607 SvREFCNT_dec (sv); 888 SvREFCNT_dec (sv);
609} 890}
610 891
611static SV * 892static SV *
612decode_tagged (dec_t *dec) 893decode_tagged (dec_t *dec)
613{ 894{
895 SV *sv = 0;
614 UV tag = decode_uint (dec); 896 UV tag = decode_uint (dec);
897
898 WANT (1);
899
900 switch (tag)
901 {
902 case CBOR_TAG_MAGIC:
615 SV *sv = decode_sv (dec); 903 sv = decode_sv (dec);
904 break;
616 905
617 if (tag == 55799) // 2.4.5 Self-Describe CBOR 906 case CBOR_TAG_INDIRECTION:
907 sv = newRV_noinc (decode_sv (dec));
908 break;
909
910 case CBOR_TAG_STRINGREF_NAMESPACE:
911 {
912 ENTER; SAVETMPS;
913
914 SAVESPTR (dec->stringref);
915 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
916
917 sv = decode_sv (dec);
918
919 FREETMPS; LEAVE;
920 }
921 break;
922
923 case CBOR_TAG_STRINGREF:
924 {
925 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
926 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
927
928 UV idx = decode_uint (dec);
929
930 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
931 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
932
933 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
934 }
935 break;
936
937 case CBOR_TAG_VALUE_SHAREABLE:
938 {
939 if (ecb_expect_false (!dec->shareable))
940 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
941
942 if (dec->cbor.flags & F_ALLOW_CYCLES)
943 {
944 sv = newSV (0);
945 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
946
947 SV *osv = decode_sv (dec);
948 sv_setsv (sv, osv);
949 SvREFCNT_dec_NN (osv);
950 }
951 else
952 {
953 av_push (dec->shareable, &PL_sv_undef);
954 int idx = AvFILLp (dec->shareable);
955 sv = decode_sv (dec);
956 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
957 }
958 }
959 break;
960
961 case CBOR_TAG_VALUE_SHAREDREF:
962 {
963 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
964 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
965
966 UV idx = decode_uint (dec);
967
968 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
969 ERR ("corrupted CBOR data (sharedref index out of bounds)");
970
971 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
972
973 if (sv == &PL_sv_undef)
974 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
975 }
976 break;
977
978 case CBOR_TAG_PERL_OBJECT:
979 {
980 sv = decode_sv (dec);
981
982 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
983 ERR ("corrupted CBOR data (non-array perl object)");
984
985 AV *av = (AV *)SvRV (sv);
986 int len = av_len (av) + 1;
987 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
988
989 if (!stash)
990 ERR ("cannot decode perl-object (package does not exist)");
991
992 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
993
994 if (!method)
995 ERR ("cannot decode perl-object (package does not have a THAW method)");
996
997 dSP;
998
999 ENTER; SAVETMPS; PUSHMARK (SP);
1000 EXTEND (SP, len + 1);
1001 // we re-bless the reference to get overload and other niceties right
1002 PUSHs (*av_fetch (av, 0, 1));
1003 PUSHs (sv_cbor);
1004
1005 int i;
1006
1007 for (i = 1; i < len; ++i)
1008 PUSHs (*av_fetch (av, i, 1));
1009
1010 PUTBACK;
1011 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
1012 SPAGAIN;
1013
1014 if (SvTRUE (ERRSV))
1015 {
1016 FREETMPS; LEAVE;
1017 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1018 }
1019
1020 SvREFCNT_dec (sv);
1021 sv = SvREFCNT_inc (POPs);
1022
1023 PUTBACK;
1024
1025 FREETMPS; LEAVE;
1026 }
1027 break;
1028
1029 default:
1030 {
1031 sv = decode_sv (dec);
1032
1033 dSP;
1034 ENTER; SAVETMPS; PUSHMARK (SP);
1035 EXTEND (SP, 2);
1036 PUSHs (newSVuv (tag));
1037 PUSHs (sv);
1038
1039 PUTBACK;
1040 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1041 SPAGAIN;
1042
1043 if (SvTRUE (ERRSV))
1044 {
1045 FREETMPS; LEAVE;
1046 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1047 }
1048
1049 if (count)
1050 {
1051 SvREFCNT_dec (sv);
1052 sv = SvREFCNT_inc (POPs);
1053 }
1054 else
1055 {
1056 AV *av = newAV ();
1057 av_push (av, newSVuv (tag));
1058 av_push (av, sv);
1059
1060 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
1061 ? cbor_tagged_stash
1062 : gv_stashpv ("CBOR::XS::Tagged" , 1);
1063 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
1064 }
1065
1066 PUTBACK;
1067
1068 FREETMPS; LEAVE;
1069 }
1070 break;
1071 }
1072
618 return sv; 1073 return sv;
619 1074
620 AV *av = newAV (); 1075fail:
621 av_push (av, newSVuv (tag)); 1076 SvREFCNT_dec (sv);
622 av_push (av, sv); 1077 return &PL_sv_undef;
623 return newRV_noinc ((SV *)av);
624} 1078}
625 1079
626static SV * 1080static SV *
627decode_sv (dec_t *dec) 1081decode_sv (dec_t *dec)
628{ 1082{
629 WANT (1); 1083 WANT (1);
630 1084
631 switch (*dec->cur >> 5) 1085 switch (*dec->cur >> MAJOR_SHIFT)
632 { 1086 {
633 case 0: // unsigned int 1087 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
634 return newSVuv (decode_uint (dec)); 1088 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
635 case 1: // negative int 1089 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
636 return newSViv (-1 - (IV)decode_uint (dec)); 1090 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
637 case 2: // octet string 1091 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
638 return decode_str (dec, 0); 1092 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
639 case 3: // utf-8 string 1093 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
640 return decode_str (dec, 1); 1094
641 case 4: // array 1095 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) 1096 switch (*dec->cur++ & MINOR_MASK)
649 { 1097 {
650 case 20: 1098 case SIMPLE_FALSE:
651#if CBOR_SLOW 1099#if CBOR_SLOW
652 cbor_false = get_bool ("CBOR::XS::false"); 1100 types_false = get_bool ("Types::Serialiser::false");
653#endif 1101#endif
654 return newSVsv (cbor_false); 1102 return newSVsv (types_false);
655 case 21: 1103 case SIMPLE_TRUE:
656#if CBOR_SLOW 1104#if CBOR_SLOW
657 cbor_true = get_bool ("CBOR::XS::true"); 1105 types_true = get_bool ("Types::Serialiser::true");
658#endif 1106#endif
659 return newSVsv (cbor_true); 1107 return newSVsv (types_true);
660 case 22: 1108 case SIMPLE_NULL:
661 return newSVsv (&PL_sv_undef); 1109 return newSVsv (&PL_sv_undef);
1110 case SIMPLE_UNDEF:
1111#if CBOR_SLOW
1112 types_error = get_bool ("Types::Serialiser::error");
1113#endif
1114 return newSVsv (types_error);
662 1115
663 case 25: 1116 case MISC_FLOAT16:
664 { 1117 {
665 WANT (2); 1118 WANT (2);
666 1119
667 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1120 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
668 dec->cur += 2; 1121 dec->cur += 2;
669 1122
670 return newSVnv (ecb_binary16_to_float (fp)); 1123 return newSVnv (ecb_binary16_to_float (fp));
671 } 1124 }
672 1125
673 case 26: 1126 case MISC_FLOAT32:
674 { 1127 {
675 uint32_t fp; 1128 uint32_t fp;
676 WANT (4); 1129 WANT (4);
677 memcpy (&fp, dec->cur, 4); 1130 memcpy (&fp, dec->cur, 4);
678 dec->cur += 4; 1131 dec->cur += 4;
681 fp = ecb_bswap32 (fp); 1134 fp = ecb_bswap32 (fp);
682 1135
683 return newSVnv (ecb_binary32_to_float (fp)); 1136 return newSVnv (ecb_binary32_to_float (fp));
684 } 1137 }
685 1138
686 case 27: 1139 case MISC_FLOAT64:
687 { 1140 {
688 uint64_t fp; 1141 uint64_t fp;
689 WANT (8); 1142 WANT (8);
690 memcpy (&fp, dec->cur, 8); 1143 memcpy (&fp, dec->cur, 8);
691 dec->cur += 8; 1144 dec->cur += 8;
694 fp = ecb_bswap64 (fp); 1147 fp = ecb_bswap64 (fp);
695 1148
696 return newSVnv (ecb_binary64_to_double (fp)); 1149 return newSVnv (ecb_binary64_to_double (fp));
697 } 1150 }
698 1151
699 // 0..19 unassigned 1152 // 0..19 unassigned simple
700 // 24 reserved + unassigned (reserved values are not encodable) 1153 // 24 reserved + unassigned simple (reserved values are not encodable)
1154 // 28-30 unassigned misc
1155 // 31 break code
701 default: 1156 default:
702 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1157 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
703 } 1158 }
704 1159
705 break; 1160 break;
706 } 1161 }
707 1162
710} 1165}
711 1166
712static SV * 1167static SV *
713decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1168decode_cbor (SV *string, CBOR *cbor, char **offset_return)
714{ 1169{
715 dec_t dec; 1170 dec_t dec = { 0 };
716 SV *sv; 1171 SV *sv;
1172 STRLEN len;
1173 char *data = SvPVbyte (string, len);
717 1174
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) 1175 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", 1176 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); 1177 (unsigned long)len, (unsigned long)cbor->max_size);
749 }
750
751 sv_utf8_downgrade (string, 0);
752 1178
753 dec.cbor = *cbor; 1179 dec.cbor = *cbor;
754 dec.cur = (U8 *)SvPVX (string); 1180 dec.cur = (U8 *)data;
755 dec.end = (U8 *)SvEND (string); 1181 dec.end = (U8 *)data + len;
756 dec.err = 0;
757 dec.depth = 0;
758 1182
759 sv = decode_sv (&dec); 1183 sv = decode_sv (&dec);
760 1184
761 if (offset_return) 1185 if (offset_return)
762 *offset_return = dec.cur; 1186 *offset_return = dec.cur;
765 if (dec.cur != dec.end && !dec.err) 1189 if (dec.cur != dec.end && !dec.err)
766 dec.err = "garbage after CBOR object"; 1190 dec.err = "garbage after CBOR object";
767 1191
768 if (dec.err) 1192 if (dec.err)
769 { 1193 {
1194 if (dec.shareable)
1195 {
1196 // need to break cyclic links, which whould all be in shareable
1197 int i;
1198 SV **svp;
1199
1200 for (i = av_len (dec.shareable) + 1; i--; )
1201 if ((svp = av_fetch (dec.shareable, i, 0)))
1202 sv_setsv (*svp, &PL_sv_undef);
1203 }
1204
770 SvREFCNT_dec (sv); 1205 SvREFCNT_dec (sv);
771 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)SvPVX (string), (int)(uint8_t)*dec.cur); 1206 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
772 } 1207 }
773 1208
774 sv = sv_2mortal (sv); 1209 sv = sv_2mortal (sv);
775 1210
776 return sv; 1211 return sv;
777} 1212}
778 1213
1214/////////////////////////////////////////////////////////////////////////////
1215// incremental parser
1216
1217#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1218
1219// returns 0 for notyet, 1 for success or error
1220static int
1221incr_parse (CBOR *self, SV *cborstr)
1222{
1223 STRLEN cur;
1224 SvPV (cborstr, cur);
1225
1226 while (ecb_expect_true (self->incr_need <= cur))
1227 {
1228 // table of integer count bytes
1229 static I8 incr_len[MINOR_MASK + 1] = {
1230 0, 0, 0, 0, 0, 0, 0, 0,
1231 0, 0, 0, 0, 0, 0, 0, 0,
1232 0, 0, 0, 0, 0, 0, 0, 0,
1233 1, 2, 4, 8,-1,-1,-1,-2
1234 };
1235
1236 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1237 U8 m = *p & MINOR_MASK;
1238 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1239 I8 ilen = incr_len[m];
1240
1241 self->incr_need = self->incr_pos + 1;
1242
1243 if (ecb_expect_false (ilen < 0))
1244 {
1245 if (m != MINOR_INDEF)
1246 return 1; // error
1247
1248 if (*p == (MAJOR_MISC | MINOR_INDEF))
1249 {
1250 if (count >= 0)
1251 return 1; // error
1252
1253 count = 1;
1254 }
1255 else
1256 {
1257 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1258 count = -1;
1259 }
1260 }
1261 else
1262 {
1263 self->incr_need += ilen;
1264 if (ecb_expect_false (self->incr_need > cur))
1265 return 0;
1266
1267 int major = *p >> MAJOR_SHIFT;
1268
1269 switch (major)
1270 {
1271 case MAJOR_TAG >> MAJOR_SHIFT:
1272 ++count; // tags merely prefix another value
1273 break;
1274
1275 case MAJOR_BYTES >> MAJOR_SHIFT:
1276 case MAJOR_TEXT >> MAJOR_SHIFT:
1277 case MAJOR_ARRAY >> MAJOR_SHIFT:
1278 case MAJOR_MAP >> MAJOR_SHIFT:
1279 {
1280 UV len;
1281
1282 if (ecb_expect_false (ilen))
1283 {
1284 len = 0;
1285
1286 do {
1287 len = (len << 8) | *++p;
1288 } while (--ilen);
1289 }
1290 else
1291 len = m;
1292
1293 switch (major)
1294 {
1295 case MAJOR_BYTES >> MAJOR_SHIFT:
1296 case MAJOR_TEXT >> MAJOR_SHIFT:
1297 self->incr_need += len;
1298 if (ecb_expect_false (self->incr_need > cur))
1299 return 0;
1300
1301 break;
1302
1303 case MAJOR_MAP >> MAJOR_SHIFT:
1304 len <<= 1;
1305 case MAJOR_ARRAY >> MAJOR_SHIFT:
1306 if (len)
1307 {
1308 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1309 count = len + 1;
1310 }
1311 break;
1312 }
1313 }
1314 }
1315 }
1316
1317 self->incr_pos = self->incr_need;
1318
1319 if (count > 0)
1320 {
1321 while (!--count)
1322 {
1323 if (!AvFILLp (self->incr_count))
1324 return 1; // done
1325
1326 SvREFCNT_dec_NN (av_pop (self->incr_count));
1327 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1328 }
1329
1330 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1331 }
1332 }
1333
1334 return 0;
1335}
1336
1337
779///////////////////////////////////////////////////////////////////////////// 1338/////////////////////////////////////////////////////////////////////////////
780// XS interface functions 1339// XS interface functions
781 1340
782MODULE = CBOR::XS PACKAGE = CBOR::XS 1341MODULE = CBOR::XS PACKAGE = CBOR::XS
783 1342
784BOOT: 1343BOOT:
785{ 1344{
786 cbor_stash = gv_stashpv ("CBOR::XS" , 1); 1345 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); 1346 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1);
789 1347
790 cbor_true = get_bool ("CBOR::XS::true"); 1348 types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
791 cbor_false = get_bool ("CBOR::XS::false"); 1349 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
1350
1351 types_true = get_bool ("Types::Serialiser::true" );
1352 types_false = get_bool ("Types::Serialiser::false");
1353 types_error = get_bool ("Types::Serialiser::error");
1354
1355 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1356
1357 sv_cbor = newSVpv ("CBOR", 0);
1358 SvREADONLY_on (sv_cbor);
792} 1359}
793 1360
794PROTOTYPES: DISABLE 1361PROTOTYPES: DISABLE
795 1362
796void CLONE (...) 1363void CLONE (...)
797 CODE: 1364 CODE:
798 cbor_stash = 0; 1365 cbor_stash = 0;
799 cbor_boolean_stash = 0;
800 cbor_tagged_stash = 0; 1366 cbor_tagged_stash = 0;
1367 types_error_stash = 0;
1368 types_boolean_stash = 0;
801 1369
802void new (char *klass) 1370void new (char *klass)
803 PPCODE: 1371 PPCODE:
804{ 1372{
805 SV *pv = NEWSV (0, sizeof (CBOR)); 1373 SV *pv = NEWSV (0, sizeof (CBOR));
813 1381
814void shrink (CBOR *self, int enable = 1) 1382void shrink (CBOR *self, int enable = 1)
815 ALIAS: 1383 ALIAS:
816 shrink = F_SHRINK 1384 shrink = F_SHRINK
817 allow_unknown = F_ALLOW_UNKNOWN 1385 allow_unknown = F_ALLOW_UNKNOWN
1386 allow_sharing = F_ALLOW_SHARING
1387 allow_cycles = F_ALLOW_CYCLES
1388 pack_strings = F_PACK_STRINGS
1389 validate_utf8 = F_VALIDATE_UTF8
818 PPCODE: 1390 PPCODE:
819{ 1391{
820 if (enable) 1392 if (enable)
821 self->flags |= ix; 1393 self->flags |= ix;
822 else 1394 else
827 1399
828void get_shrink (CBOR *self) 1400void get_shrink (CBOR *self)
829 ALIAS: 1401 ALIAS:
830 get_shrink = F_SHRINK 1402 get_shrink = F_SHRINK
831 get_allow_unknown = F_ALLOW_UNKNOWN 1403 get_allow_unknown = F_ALLOW_UNKNOWN
1404 get_allow_sharing = F_ALLOW_SHARING
1405 get_allow_cycles = F_ALLOW_CYCLES
1406 get_pack_strings = F_PACK_STRINGS
1407 get_validate_utf8 = F_VALIDATE_UTF8
832 PPCODE: 1408 PPCODE:
833 XPUSHs (boolSV (self->flags & ix)); 1409 XPUSHs (boolSV (self->flags & ix));
834 1410
835void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1411void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
836 PPCODE: 1412 PPCODE:
852 CODE: 1428 CODE:
853 RETVAL = self->max_size; 1429 RETVAL = self->max_size;
854 OUTPUT: 1430 OUTPUT:
855 RETVAL 1431 RETVAL
856 1432
857#if 0 //TODO 1433void filter (CBOR *self, SV *filter = 0)
858
859void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef)
860 PPCODE: 1434 PPCODE:
861{
862 SvREFCNT_dec (self->cb_object); 1435 SvREFCNT_dec (self->filter);
863 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; 1436 self->filter = filter ? newSVsv (filter) : filter;
864
865 XPUSHs (ST (0)); 1437 XPUSHs (ST (0));
866}
867 1438
868void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef) 1439SV *get_filter (CBOR *self)
869 PPCODE: 1440 CODE:
870{ 1441 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
871 if (!self->cb_sk_object) 1442 OUTPUT:
872 self->cb_sk_object = newHV (); 1443 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 1444
892void encode (CBOR *self, SV *scalar) 1445void encode (CBOR *self, SV *scalar)
893 PPCODE: 1446 PPCODE:
894 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; 1447 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
895 XPUSHs (scalar); 1448 XPUSHs (scalar);
908 EXTEND (SP, 2); 1461 EXTEND (SP, 2);
909 PUSHs (sv); 1462 PUSHs (sv);
910 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1463 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
911} 1464}
912 1465
913#if 0 1466void incr_parse (CBOR *self, SV *cborstr)
1467 ALIAS:
1468 incr_parse_multiple = 1
1469 PPCODE:
1470{
1471 if (SvUTF8 (cborstr))
1472 sv_utf8_downgrade (cborstr, 0);
1473
1474 if (!self->incr_count)
1475 {
1476 self->incr_count = newAV ();
1477 self->incr_pos = 0;
1478 self->incr_need = 1;
1479
1480 av_push (self->incr_count, newSViv (1));
1481 }
1482
1483 do
1484 {
1485 if (!incr_parse (self, cborstr))
1486 {
1487 if (self->incr_need > self->max_size && self->max_size)
1488 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1489 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1490
1491 break;
1492 }
1493
1494 SV *sv;
1495 char *offset;
1496
1497 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1498 XPUSHs (sv);
1499
1500 sv_chop (cborstr, offset);
1501
1502 av_clear (self->incr_count);
1503 av_push (self->incr_count, newSViv (1));
1504
1505 self->incr_pos = 0;
1506 self->incr_need = self->incr_pos + 1;
1507 }
1508 while (ix);
1509}
1510
1511void incr_reset (CBOR *self)
1512 CODE:
1513{
1514 SvREFCNT_dec (self->incr_count);
1515 self->incr_count = 0;
1516}
914 1517
915void DESTROY (CBOR *self) 1518void DESTROY (CBOR *self)
916 CODE: 1519 PPCODE:
917 SvREFCNT_dec (self->cb_sk_object); 1520 cbor_free (self);
918 SvREFCNT_dec (self->cb_object);
919
920#endif
921 1521
922PROTOTYPES: ENABLE 1522PROTOTYPES: ENABLE
923 1523
924void encode_cbor (SV *scalar) 1524void encode_cbor (SV *scalar)
1525 ALIAS:
1526 encode_cbor = 0
1527 encode_cbor_sharing = F_ALLOW_SHARING
925 PPCODE: 1528 PPCODE:
926{ 1529{
927 CBOR cbor; 1530 CBOR cbor;
928 cbor_init (&cbor); 1531 cbor_init (&cbor);
1532 cbor.flags |= ix;
929 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1533 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
930 XPUSHs (scalar); 1534 XPUSHs (scalar);
931} 1535}
932 1536
933void decode_cbor (SV *cborstr) 1537void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines