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 1.16 by root, Tue Oct 29 20:59:16 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
14// compatibility with perl <5.18 15// compatibility with perl <5.18
15#ifndef HvNAMELEN_get 16#ifndef HvNAMELEN_get
16# define HvNAMELEN_get(hv) strlen (HvNAME (hv)) 17# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
19# define HvNAMELEN(hv) HvNAMELEN_get (hv) 20# define HvNAMELEN(hv) HvNAMELEN_get (hv)
20#endif 21#endif
21#ifndef HvNAMEUTF8 22#ifndef HvNAMEUTF8
22# define HvNAMEUTF8(hv) 0 23# define HvNAMEUTF8(hv) 0
23#endif 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};
24 65
25// known tags 66// known tags
26enum cbor_tag 67enum cbor_tag
27{ 68{
28 // inofficial extensions (pending iana registration) 69 // extensions
29 CBOR_TAG_PERL_OBJECT = 256, 70 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
30 CBOR_TAG_GENERIC_OBJECT = 257, 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
31 77
32 // rfc7049 78 // rfc7049
33 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 79 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
34 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any 80 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
35 CBOR_TAG_POS_BIGNUM = 2, // byte string 81 CBOR_TAG_POS_BIGNUM = 2, // byte string
36 CBOR_TAG_NEG_BIGNUM = 3, // byte string 82 CBOR_TAG_NEG_BIGNUM = 3, // byte string
37 CBOR_TAG_DECIMAL = 4, // decimal fraction, array 83 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
38 CBOR_TAG_BIGFLOAT = 5, // array 84 CBOR_TAG_BIGFLOAT = 5, // array
39 85
40 CBOR_TAG_CONV_B64U = 21, // base64url, any 86 CBOR_TAG_CONV_B64U = 21, // base64url, any
41 CBOR_TAG_CONV_B64 = 22, // base64, any 87 CBOR_TAG_CONV_B64 = 22, // base64, any
42 CBOR_TAG_CONV_HEX = 23, // base16, any 88 CBOR_TAG_CONV_HEX = 23, // base16, any
43 CBOR_TAG_CBOR = 24, // embedded cbor, byte string 89 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
44 90
45 CBOR_TAG_URI = 32, // URI rfc3986, utf-8 91 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
46 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8 92 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
47 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8 93 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
48 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8 94 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
49 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 95 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
50 96
51 CBOR_TAG_MAGIC = 55799 // self-describe cbor 97 CBOR_TAG_MAGIC = 55799, // self-describe cbor
52}; 98};
53 99
54#define F_SHRINK 0x00000200UL 100#define F_SHRINK 0x00000001UL
55#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
56 106
57#define INIT_SIZE 32 // initial scalar size to be allocated 107#define INIT_SIZE 32 // initial scalar size to be allocated
58 108
59#define SB do { 109#define SB do {
60#define SE } while (0) 110#define SE } while (0)
72# define CBOR_SLOW 0 122# define CBOR_SLOW 0
73# define CBOR_STASH cbor_stash 123# define CBOR_STASH cbor_stash
74#endif 124#endif
75 125
76static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS:: 126static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
77static SV *types_true, *types_false, *types_error, *sv_cbor; 127static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
78 128
79typedef struct { 129typedef struct {
80 U32 flags; 130 U32 flags;
81 U32 max_depth; 131 U32 max_depth;
82 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.
83} CBOR; 139} CBOR;
84 140
85ecb_inline void 141ecb_inline void
86cbor_init (CBOR *cbor) 142cbor_init (CBOR *cbor)
87{ 143{
88 Zero (cbor, 1, CBOR); 144 Zero (cbor, 1, CBOR);
89 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);
90} 153}
91 154
92///////////////////////////////////////////////////////////////////////////// 155/////////////////////////////////////////////////////////////////////////////
93// utility functions 156// utility functions
94 157
116 SvPV_renew (sv, SvCUR (sv) + 1); 179 SvPV_renew (sv, SvCUR (sv) + 1);
117#endif 180#endif
118 } 181 }
119} 182}
120 183
121///////////////////////////////////////////////////////////////////////////// 184// minimum length of a string to be registered for stringref
122// fp hell 185ecb_inline int
123 186minimum_string_length (UV idx)
124//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}
125 198
126///////////////////////////////////////////////////////////////////////////// 199/////////////////////////////////////////////////////////////////////////////
127// encoder 200// encoder
128 201
129// structure used for encoding CBOR 202// structure used for encoding CBOR
132 char *cur; // SvPVX (sv) + current output position 205 char *cur; // SvPVX (sv) + current output position
133 char *end; // SvEND (sv) 206 char *end; // SvEND (sv)
134 SV *sv; // result scalar 207 SV *sv; // result scalar
135 CBOR cbor; 208 CBOR cbor;
136 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;
137} enc_t; 214} enc_t;
138 215
139ecb_inline void 216ecb_inline void
140need (enc_t *enc, STRLEN len) 217need (enc_t *enc, STRLEN len)
141{ 218{
158static void 235static void
159encode_uint (enc_t *enc, int major, UV len) 236encode_uint (enc_t *enc, int major, UV len)
160{ 237{
161 need (enc, 9); 238 need (enc, 9);
162 239
163 if (len < 24) 240 if (ecb_expect_true (len < LENGTH_EXT1))
164 *enc->cur++ = major | len; 241 *enc->cur++ = major | len;
165 else if (len <= 0xff) 242 else if (ecb_expect_true (len <= 0xffU))
166 { 243 {
167 *enc->cur++ = major | 24; 244 *enc->cur++ = major | LENGTH_EXT1;
168 *enc->cur++ = len; 245 *enc->cur++ = len;
169 } 246 }
170 else if (len <= 0xffff) 247 else if (len <= 0xffffU)
171 { 248 {
172 *enc->cur++ = major | 25; 249 *enc->cur++ = major | LENGTH_EXT2;
173 *enc->cur++ = len >> 8; 250 *enc->cur++ = len >> 8;
174 *enc->cur++ = len; 251 *enc->cur++ = len;
175 } 252 }
176 else if (len <= 0xffffffff) 253 else if (len <= 0xffffffffU)
177 { 254 {
178 *enc->cur++ = major | 26; 255 *enc->cur++ = major | LENGTH_EXT4;
179 *enc->cur++ = len >> 24; 256 *enc->cur++ = len >> 24;
180 *enc->cur++ = len >> 16; 257 *enc->cur++ = len >> 16;
181 *enc->cur++ = len >> 8; 258 *enc->cur++ = len >> 8;
182 *enc->cur++ = len; 259 *enc->cur++ = len;
183 } 260 }
184 else 261 else
185 { 262 {
186 *enc->cur++ = major | 27; 263 *enc->cur++ = major | LENGTH_EXT8;
187 *enc->cur++ = len >> 56; 264 *enc->cur++ = len >> 56;
188 *enc->cur++ = len >> 48; 265 *enc->cur++ = len >> 48;
189 *enc->cur++ = len >> 40; 266 *enc->cur++ = len >> 40;
190 *enc->cur++ = len >> 32; 267 *enc->cur++ = len >> 32;
191 *enc->cur++ = len >> 24; 268 *enc->cur++ = len >> 24;
193 *enc->cur++ = len >> 8; 270 *enc->cur++ = len >> 8;
194 *enc->cur++ = len; 271 *enc->cur++ = len;
195 } 272 }
196} 273}
197 274
198static void 275ecb_inline void
276encode_tag (enc_t *enc, UV tag)
277{
278 encode_uint (enc, MAJOR_TAG, tag);
279}
280
281ecb_inline void
199encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 282encode_str (enc_t *enc, int utf8, char *str, STRLEN len)
200{ 283{
201 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 284 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
202 need (enc, len); 285 need (enc, len);
203 memcpy (enc->cur, str, len); 286 memcpy (enc->cur, str, len);
204 enc->cur += len; 287 enc->cur += len;
205} 288}
206 289
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
207static void encode_sv (enc_t *enc, SV *sv); 315static void encode_sv (enc_t *enc, SV *sv);
208 316
209static void 317static void
210encode_av (enc_t *enc, AV *av) 318encode_av (enc_t *enc, AV *av)
211{ 319{
214 if (enc->depth >= enc->cbor.max_depth) 322 if (enc->depth >= enc->cbor.max_depth)
215 croak (ERR_NESTING_EXCEEDED); 323 croak (ERR_NESTING_EXCEEDED);
216 324
217 ++enc->depth; 325 ++enc->depth;
218 326
219 encode_uint (enc, 0x80, len + 1); 327 encode_uint (enc, MAJOR_ARRAY, len + 1);
220 328
329 if (SvMAGICAL (av))
221 for (i = 0; i <= len; ++i) 330 for (i = 0; i <= len; ++i)
222 { 331 {
223 SV **svp = av_fetch (av, i, 0); 332 SV **svp = av_fetch (av, i, 0);
224 encode_sv (enc, svp ? *svp : &PL_sv_undef); 333 encode_sv (enc, svp ? *svp : &PL_sv_undef);
225 } 334 }
335 else
336 for (i = 0; i <= len; ++i)
337 encode_sv (enc, AvARRAY (av)[i]);
226 338
227 --enc->depth; 339 --enc->depth;
228} 340}
229 341
230static void 342static void
239 351
240 int pairs = hv_iterinit (hv); 352 int pairs = hv_iterinit (hv);
241 int mg = SvMAGICAL (hv); 353 int mg = SvMAGICAL (hv);
242 354
243 if (mg) 355 if (mg)
244 encode_ch (enc, 0xa0 | 31); 356 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
245 else 357 else
246 encode_uint (enc, 0xa0, pairs); 358 encode_uint (enc, MAJOR_MAP, pairs);
247 359
248 while ((he = hv_iternext (hv))) 360 while ((he = hv_iternext (hv)))
249 { 361 {
250 if (HeKLEN (he) == HEf_SVKEY) 362 if (HeKLEN (he) == HEf_SVKEY)
251 encode_sv (enc, HeSVKEY (he)); 363 encode_sv (enc, HeSVKEY (he));
252 else 364 else
253 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 365 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
254 366
255 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));
256 } 368 }
257 369
258 if (mg) 370 if (mg)
259 encode_ch (enc, 0xe0 | 31); 371 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
260 372
261 --enc->depth; 373 --enc->depth;
262} 374}
263 375
264// encode objects, arrays and special \0=false and \1=true values. 376// encode objects, arrays and special \0=false and \1=true values.
265static void 377static void
266encode_rv (enc_t *enc, SV *sv) 378encode_rv (enc_t *enc, SV *sv)
267{ 379{
268 svtype svt;
269
270 SvGETMAGIC (sv); 380 SvGETMAGIC (sv);
381
271 svt = SvTYPE (sv); 382 svtype svt = SvTYPE (sv);
272 383
273 if (ecb_expect_false (SvOBJECT (sv))) 384 if (ecb_expect_false (SvOBJECT (sv)))
274 { 385 {
275 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash 386 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
276 ? types_boolean_stash 387 ? types_boolean_stash
281 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 392 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
282 ? cbor_tagged_stash 393 ? cbor_tagged_stash
283 : gv_stashpv ("CBOR::XS::Tagged" , 1); 394 : gv_stashpv ("CBOR::XS::Tagged" , 1);
284 395
285 HV *stash = SvSTASH (sv); 396 HV *stash = SvSTASH (sv);
286 GV *method;
287 397
288 if (stash == boolean_stash) 398 if (stash == boolean_stash)
289 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 }
290 else if (stash == error_stash) 403 else if (stash == error_stash)
291 encode_ch (enc, 0xe0 | 23); 404 {
405 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
406 return;
407 }
292 else if (stash == tagged_stash) 408 else if (stash == tagged_stash)
293 { 409 {
294 if (svt != SVt_PVAV) 410 if (svt != SVt_PVAV)
295 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 411 croak ("encountered CBOR::XS::Tagged object that isn't an array");
296 412
297 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 413 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
298 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))
299 } 429 {
430 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
431 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
432 return;
433 }
434 else
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
300 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 447 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
301 { 448 {
302 dSP; 449 dSP;
303 450
304 ENTER; SAVETMPS; PUSHMARK (SP); 451 ENTER; SAVETMPS; PUSHMARK (SP);
305 // 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
336 483
337 // catch this surprisingly common error 484 // catch this surprisingly common error
338 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 485 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
339 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); 486 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
340 487
341 encode_uint (enc, 0xc0, CBOR_TAG_PERL_OBJECT); 488 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
342 encode_uint (enc, 0x80, count + 1); 489 encode_uint (enc, MAJOR_ARRAY, count + 1);
343 encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 490 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
344 491
345 while (count) 492 while (count)
346 encode_sv (enc, SP[1 - count--]); 493 encode_sv (enc, SP[1 - count--]);
347 494
348 PUTBACK; 495 PUTBACK;
355 } 502 }
356 else if (svt == SVt_PVHV) 503 else if (svt == SVt_PVHV)
357 encode_hv (enc, (HV *)sv); 504 encode_hv (enc, (HV *)sv);
358 else if (svt == SVt_PVAV) 505 else if (svt == SVt_PVAV)
359 encode_av (enc, (AV *)sv); 506 encode_av (enc, (AV *)sv);
360 else if (svt < SVt_PVAV)
361 {
362 STRLEN len = 0;
363 char *pv = svt ? SvPV (sv, len) : 0;
364
365 if (len == 1 && *pv == '1')
366 encode_ch (enc, 0xe0 | 21);
367 else if (len == 1 && *pv == '0')
368 encode_ch (enc, 0xe0 | 20);
369 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
370 encode_ch (enc, 0xe0 | 23);
371 else
372 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
373 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
374 }
375 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
376 encode_ch (enc, 0xe0 | 23);
377 else 507 else
378 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 508 {
379 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 509 encode_tag (enc, CBOR_TAG_INDIRECTION);
510 encode_sv (enc, sv);
511 }
380} 512}
381 513
382static void 514static void
383encode_nv (enc_t *enc, SV *sv) 515encode_nv (enc_t *enc, SV *sv)
384{ 516{
385 double nv = SvNVX (sv); 517 double nv = SvNVX (sv);
386 518
387 need (enc, 9); 519 need (enc, 9);
388 520
389 if (ecb_expect_false (nv == (U32)nv)) 521 if (ecb_expect_false (nv == (NV)(U32)nv))
390 encode_uint (enc, 0x00, (U32)nv); 522 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
391 //TODO: maybe I32? 523 //TODO: maybe I32?
392 else if (ecb_expect_false (nv == (float)nv)) 524 else if (ecb_expect_false (nv == (float)nv))
393 { 525 {
394 uint32_t fp = ecb_float_to_binary32 (nv); 526 uint32_t fp = ecb_float_to_binary32 (nv);
395 527
396 *enc->cur++ = 0xe0 | 26; 528 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
397 529
398 if (!ecb_big_endian ()) 530 if (!ecb_big_endian ())
399 fp = ecb_bswap32 (fp); 531 fp = ecb_bswap32 (fp);
400 532
401 memcpy (enc->cur, &fp, 4); 533 memcpy (enc->cur, &fp, 4);
403 } 535 }
404 else 536 else
405 { 537 {
406 uint64_t fp = ecb_double_to_binary64 (nv); 538 uint64_t fp = ecb_double_to_binary64 (nv);
407 539
408 *enc->cur++ = 0xe0 | 27; 540 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
409 541
410 if (!ecb_big_endian ()) 542 if (!ecb_big_endian ())
411 fp = ecb_bswap64 (fp); 543 fp = ecb_bswap64 (fp);
412 544
413 memcpy (enc->cur, &fp, 8); 545 memcpy (enc->cur, &fp, 8);
422 554
423 if (SvPOKp (sv)) 555 if (SvPOKp (sv))
424 { 556 {
425 STRLEN len; 557 STRLEN len;
426 char *str = SvPV (sv, len); 558 char *str = SvPV (sv, len);
427 encode_str (enc, SvUTF8 (sv), str, len); 559 encode_strref (enc, SvUTF8 (sv), str, len);
428 } 560 }
429 else if (SvNOKp (sv)) 561 else if (SvNOKp (sv))
430 encode_nv (enc, sv); 562 encode_nv (enc, sv);
431 else if (SvIOKp (sv)) 563 else if (SvIOKp (sv))
432 { 564 {
433 if (SvIsUV (sv)) 565 if (SvIsUV (sv))
434 encode_uint (enc, 0x00, SvUVX (sv)); 566 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
435 else if (SvIVX (sv) >= 0) 567 else if (SvIVX (sv) >= 0)
436 encode_uint (enc, 0x00, SvIVX (sv)); 568 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
437 else 569 else
438 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 570 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
439 } 571 }
440 else if (SvROK (sv)) 572 else if (SvROK (sv))
441 encode_rv (enc, SvRV (sv)); 573 encode_rv (enc, SvRV (sv));
442 else if (!SvOK (sv)) 574 else if (!SvOK (sv))
443 encode_ch (enc, 0xe0 | 22); 575 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
444 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 576 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
445 encode_ch (enc, 0xe0 | 23); 577 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
446 else 578 else
447 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",
448 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 580 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
449} 581}
450 582
451static SV * 583static SV *
452encode_cbor (SV *scalar, CBOR *cbor) 584encode_cbor (SV *scalar, CBOR *cbor)
453{ 585{
454 enc_t enc; 586 enc_t enc = { };
455 587
456 enc.cbor = *cbor; 588 enc.cbor = *cbor;
457 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 589 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
458 enc.cur = SvPVX (enc.sv); 590 enc.cur = SvPVX (enc.sv);
459 enc.end = SvEND (enc.sv); 591 enc.end = SvEND (enc.sv);
460 enc.depth = 0;
461 592
462 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
463 encode_sv (&enc, scalar); 602 encode_sv (&enc, scalar);
464 603
465 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 604 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
466 *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
467 606
481 U8 *end; // end of input string 620 U8 *end; // end of input string
482 const char *err; // parse error, if != 0 621 const char *err; // parse error, if != 0
483 CBOR cbor; 622 CBOR cbor;
484 U32 depth; // recursion depth 623 U32 depth; // recursion depth
485 U32 maxdepth; // recursion depth limit 624 U32 maxdepth; // recursion depth limit
625 AV *shareable;
626 AV *stringref;
627 SV *decode_tagged;
486} dec_t; 628} dec_t;
487 629
488#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
489 631
490#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")
493#define DEC_DEC_DEPTH --dec->depth 635#define DEC_DEC_DEPTH --dec->depth
494 636
495static UV 637static UV
496decode_uint (dec_t *dec) 638decode_uint (dec_t *dec)
497{ 639{
498 switch (*dec->cur & 31) 640 U8 m = *dec->cur & MINOR_MASK;
499 { 641 ++dec->cur;
500 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
501 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
502 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
503 return *dec->cur++ & 31;
504 642
505 case 24: 643 if (ecb_expect_true (m < LENGTH_EXT1))
644 return m;
645 else if (ecb_expect_true (m == LENGTH_EXT1))
646 {
506 WANT (2); 647 WANT (1);
507 dec->cur += 2; 648 dec->cur += 1;
508 return dec->cur[-1]; 649 return dec->cur[-1];
509 650 }
510 case 25: 651 else if (ecb_expect_true (m == LENGTH_EXT2))
652 {
511 WANT (3); 653 WANT (2);
512 dec->cur += 3; 654 dec->cur += 2;
513 return (((UV)dec->cur[-2]) << 8) 655 return (((UV)dec->cur[-2]) << 8)
514 | ((UV)dec->cur[-1]); 656 | ((UV)dec->cur[-1]);
515 657 }
516 case 26: 658 else if (ecb_expect_true (m == LENGTH_EXT4))
659 {
517 WANT (5); 660 WANT (4);
518 dec->cur += 5; 661 dec->cur += 4;
519 return (((UV)dec->cur[-4]) << 24) 662 return (((UV)dec->cur[-4]) << 24)
520 | (((UV)dec->cur[-3]) << 16) 663 | (((UV)dec->cur[-3]) << 16)
521 | (((UV)dec->cur[-2]) << 8) 664 | (((UV)dec->cur[-2]) << 8)
522 | ((UV)dec->cur[-1]); 665 | ((UV)dec->cur[-1]);
523 666 }
524 case 27: 667 else if (ecb_expect_true (m == LENGTH_EXT8))
668 {
525 WANT (9); 669 WANT (8);
526 dec->cur += 9; 670 dec->cur += 8;
671
672 return
673#if UVSIZE < 8
674 0
675#else
527 return (((UV)dec->cur[-8]) << 56) 676 (((UV)dec->cur[-8]) << 56)
528 | (((UV)dec->cur[-7]) << 48) 677 | (((UV)dec->cur[-7]) << 48)
529 | (((UV)dec->cur[-6]) << 40) 678 | (((UV)dec->cur[-6]) << 40)
530 | (((UV)dec->cur[-5]) << 32) 679 | (((UV)dec->cur[-5]) << 32)
680#endif
531 | (((UV)dec->cur[-4]) << 24) 681 | (((UV)dec->cur[-4]) << 24)
532 | (((UV)dec->cur[-3]) << 16) 682 | (((UV)dec->cur[-3]) << 16)
533 | (((UV)dec->cur[-2]) << 8) 683 | (((UV)dec->cur[-2]) << 8)
534 | ((UV)dec->cur[-1]); 684 | ((UV)dec->cur[-1]);
535 685 }
536 default: 686 else
537 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 687 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
538 }
539 688
540fail: 689fail:
541 return 0; 690 return 0;
542} 691}
543 692
548{ 697{
549 AV *av = newAV (); 698 AV *av = newAV ();
550 699
551 DEC_INC_DEPTH; 700 DEC_INC_DEPTH;
552 701
553 if ((*dec->cur & 31) == 31) 702 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
554 { 703 {
555 ++dec->cur; 704 ++dec->cur;
556 705
557 for (;;) 706 for (;;)
558 { 707 {
559 WANT (1); 708 WANT (1);
560 709
561 if (*dec->cur == (0xe0 | 31)) 710 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
562 { 711 {
563 ++dec->cur; 712 ++dec->cur;
564 break; 713 break;
565 } 714 }
566 715
569 } 718 }
570 else 719 else
571 { 720 {
572 int i, len = decode_uint (dec); 721 int i, len = decode_uint (dec);
573 722
723 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
574 av_fill (av, len - 1); 724 av_fill (av, len - 1);
575 725
576 for (i = 0; i < len; ++i) 726 for (i = 0; i < len; ++i)
577 AvARRAY (av)[i] = decode_sv (dec); 727 AvARRAY (av)[i] = decode_sv (dec);
578 } 728 }
588 738
589static void 739static void
590decode_he (dec_t *dec, HV *hv) 740decode_he (dec_t *dec, HV *hv)
591{ 741{
592 // for speed reasons, we specialcase single-string 742 // for speed reasons, we specialcase single-string
593 // byte or utf-8 strings as keys. 743 // byte or utf-8 strings as keys, but only when !stringref
594 744
595 if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27) 745 if (ecb_expect_true (!dec->stringref))
746 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
596 { 747 {
597 I32 len = decode_uint (dec); 748 I32 len = decode_uint (dec);
598 char *key = (char *)dec->cur; 749 char *key = (char *)dec->cur;
599 750
600 dec->cur += len; 751 dec->cur += len;
601 752
602 hv_store (hv, key, len, decode_sv (dec), 0); 753 hv_store (hv, key, len, decode_sv (dec), 0);
754
755 return;
603 } 756 }
604 else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27) 757 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
605 { 758 {
606 I32 len = decode_uint (dec); 759 I32 len = decode_uint (dec);
607 char *key = (char *)dec->cur; 760 char *key = (char *)dec->cur;
608 761
609 dec->cur += len; 762 dec->cur += len;
610 763
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
611 hv_store (hv, key, -len, decode_sv (dec), 0); 768 hv_store (hv, key, -len, decode_sv (dec), 0);
769
770 return;
612 } 771 }
613 else 772
614 {
615 SV *k = decode_sv (dec); 773 SV *k = decode_sv (dec);
616 SV *v = decode_sv (dec); 774 SV *v = decode_sv (dec);
617 775
618 hv_store_ent (hv, k, v, 0); 776 hv_store_ent (hv, k, v, 0);
619 SvREFCNT_dec (k); 777 SvREFCNT_dec (k);
620 } 778
779fail:
780 ;
621} 781}
622 782
623static SV * 783static SV *
624decode_hv (dec_t *dec) 784decode_hv (dec_t *dec)
625{ 785{
626 HV *hv = newHV (); 786 HV *hv = newHV ();
627 787
628 DEC_INC_DEPTH; 788 DEC_INC_DEPTH;
629 789
630 if ((*dec->cur & 31) == 31) 790 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
631 { 791 {
632 ++dec->cur; 792 ++dec->cur;
633 793
634 for (;;) 794 for (;;)
635 { 795 {
636 WANT (1); 796 WANT (1);
637 797
638 if (*dec->cur == (0xe0 | 31)) 798 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
639 { 799 {
640 ++dec->cur; 800 ++dec->cur;
641 break; 801 break;
642 } 802 }
643 803
664static SV * 824static SV *
665decode_str (dec_t *dec, int utf8) 825decode_str (dec_t *dec, int utf8)
666{ 826{
667 SV *sv = 0; 827 SV *sv = 0;
668 828
669 if ((*dec->cur & 31) == 31) 829 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
670 { 830 {
831 // indefinite length strings
671 ++dec->cur; 832 ++dec->cur;
672 833
834 U8 major = *dec->cur & MAJOR_MISC;
835
673 sv = newSVpvn ("", 0); 836 sv = newSVpvn ("", 0);
674 837
675 // not very fast, and certainly not robust against illegal input
676 for (;;) 838 for (;;)
677 { 839 {
678 WANT (1); 840 WANT (1);
679 841
680 if (*dec->cur == (0xe0 | 31)) 842 if ((*dec->cur - major) > LENGTH_EXT8)
843 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
681 { 844 {
682 ++dec->cur; 845 ++dec->cur;
683 break; 846 break;
684 } 847 }
848 else
849 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
685 850
686 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;
687 } 856 }
688 } 857 }
689 else 858 else
690 { 859 {
691 STRLEN len = decode_uint (dec); 860 STRLEN len = decode_uint (dec);
692 861
693 WANT (len); 862 WANT (len);
694 sv = newSVpvn (dec->cur, len); 863 sv = newSVpvn (dec->cur, len);
695 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));
696 } 869 }
697 870
698 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
699 SvUTF8_on (sv); 877 SvUTF8_on (sv);
878 }
700 879
701 return sv; 880 return sv;
702 881
703fail: 882fail:
704 SvREFCNT_dec (sv); 883 SvREFCNT_dec (sv);
706} 885}
707 886
708static SV * 887static SV *
709decode_tagged (dec_t *dec) 888decode_tagged (dec_t *dec)
710{ 889{
890 SV *sv = 0;
711 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:
712 SV *sv = decode_sv (dec); 898 sv = decode_sv (dec);
899 break;
713 900
714 if (tag == CBOR_TAG_MAGIC) 901 case CBOR_TAG_INDIRECTION:
715 return sv; 902 sv = newRV_noinc (decode_sv (dec));
716 else if (tag == CBOR_TAG_PERL_OBJECT) 903 break;
717 { 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
718 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 977 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
719 ERR ("corrupted CBOR data (non-array perl object)"); 978 ERR ("corrupted CBOR data (non-array perl object)");
720 979
721 AV *av = (AV *)SvRV (sv); 980 AV *av = (AV *)SvRV (sv);
722 int len = av_len (av) + 1; 981 int len = av_len (av) + 1;
723 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0); 982 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
724 983
725 if (!stash) 984 if (!stash)
726 ERR ("cannot decode perl-object (package does not exist)"); 985 ERR ("cannot decode perl-object (package does not exist)");
727 986
728 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0); 987 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
729 988
730 if (!method) 989 if (!method)
731 ERR ("cannot decode perl-object (package does not have a THAW method)"); 990 ERR ("cannot decode perl-object (package does not have a THAW method)");
732 991
733 dSP; 992 dSP;
734 993
735 ENTER; SAVETMPS; PUSHMARK (SP); 994 ENTER; SAVETMPS; PUSHMARK (SP);
736 EXTEND (SP, len + 1); 995 EXTEND (SP, len + 1);
737 // we re-bless the reference to get overload and other niceties right 996 // we re-bless the reference to get overload and other niceties right
738 PUSHs (*av_fetch (av, 0, 1)); 997 PUSHs (*av_fetch (av, 0, 1));
739 PUSHs (sv_cbor); 998 PUSHs (sv_cbor);
740 999
741 int i; 1000 int i;
742 1001
743 for (i = 1; i < len; ++i) 1002 for (i = 1; i < len; ++i)
744 PUSHs (*av_fetch (av, i, 1)); 1003 PUSHs (*av_fetch (av, i, 1));
745 1004
746 PUTBACK; 1005 PUTBACK;
747 call_sv ((SV *)GvCV (method), G_SCALAR); 1006 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
748 SPAGAIN; 1007 SPAGAIN;
749 1008
1009 if (SvTRUE (ERRSV))
1010 {
1011 FREETMPS; LEAVE;
1012 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1013 }
1014
750 SvREFCNT_dec (sv); 1015 SvREFCNT_dec (sv);
751 sv = SvREFCNT_inc (POPs); 1016 sv = SvREFCNT_inc (POPs);
752 1017
753 PUTBACK; 1018 PUTBACK;
754 1019
755 FREETMPS; LEAVE; 1020 FREETMPS; LEAVE;
1021 }
1022 break;
756 1023
757 return sv; 1024 default:
758 } 1025 {
759 else 1026 sv = decode_sv (dec);
760 { 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 {
761 AV *av = newAV (); 1051 AV *av = newAV ();
762 av_push (av, newSVuv (tag)); 1052 av_push (av, newSVuv (tag));
763 av_push (av, sv); 1053 av_push (av, sv);
764 1054
765 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1055 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
766 ? cbor_tagged_stash 1056 ? cbor_tagged_stash
767 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1057 : gv_stashpv ("CBOR::XS::Tagged" , 1);
768
769 return sv_bless (newRV_noinc ((SV *)av), tagged_stash); 1058 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
1059 }
1060
1061 PUTBACK;
1062
1063 FREETMPS; LEAVE;
1064 }
1065 break;
770 } 1066 }
1067
1068 return sv;
771 1069
772fail: 1070fail:
773 SvREFCNT_dec (sv); 1071 SvREFCNT_dec (sv);
774 return &PL_sv_undef; 1072 return &PL_sv_undef;
775} 1073}
777static SV * 1075static SV *
778decode_sv (dec_t *dec) 1076decode_sv (dec_t *dec)
779{ 1077{
780 WANT (1); 1078 WANT (1);
781 1079
782 switch (*dec->cur >> 5) 1080 switch (*dec->cur >> MAJOR_SHIFT)
783 { 1081 {
784 case 0: // unsigned int 1082 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
785 return newSVuv (decode_uint (dec)); 1083 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
786 case 1: // negative int 1084 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
787 return newSViv (-1 - (IV)decode_uint (dec)); 1085 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
788 case 2: // octet string 1086 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
789 return decode_str (dec, 0); 1087 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
790 case 3: // utf-8 string 1088 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
791 return decode_str (dec, 1); 1089
792 case 4: // array 1090 case MAJOR_MISC >> MAJOR_SHIFT:
793 return decode_av (dec);
794 case 5: // map
795 return decode_hv (dec);
796 case 6: // tag
797 return decode_tagged (dec);
798 case 7: // misc
799 switch (*dec->cur++ & 31) 1091 switch (*dec->cur++ & MINOR_MASK)
800 { 1092 {
801 case 20: 1093 case SIMPLE_FALSE:
802#if CBOR_SLOW 1094#if CBOR_SLOW
803 types_false = get_bool ("Types::Serialiser::false"); 1095 types_false = get_bool ("Types::Serialiser::false");
804#endif 1096#endif
805 return newSVsv (types_false); 1097 return newSVsv (types_false);
806 case 21: 1098 case SIMPLE_TRUE:
807#if CBOR_SLOW 1099#if CBOR_SLOW
808 types_true = get_bool ("Types::Serialiser::true"); 1100 types_true = get_bool ("Types::Serialiser::true");
809#endif 1101#endif
810 return newSVsv (types_true); 1102 return newSVsv (types_true);
811 case 22: 1103 case SIMPLE_NULL:
812 return newSVsv (&PL_sv_undef); 1104 return newSVsv (&PL_sv_undef);
813 case 23: 1105 case SIMPLE_UNDEF:
814#if CBOR_SLOW 1106#if CBOR_SLOW
815 types_error = get_bool ("Types::Serialiser::error"); 1107 types_error = get_bool ("Types::Serialiser::error");
816#endif 1108#endif
817 return newSVsv (types_error); 1109 return newSVsv (types_error);
818 1110
819 case 25: 1111 case MISC_FLOAT16:
820 { 1112 {
821 WANT (2); 1113 WANT (2);
822 1114
823 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1115 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
824 dec->cur += 2; 1116 dec->cur += 2;
825 1117
826 return newSVnv (ecb_binary16_to_float (fp)); 1118 return newSVnv (ecb_binary16_to_float (fp));
827 } 1119 }
828 1120
829 case 26: 1121 case MISC_FLOAT32:
830 { 1122 {
831 uint32_t fp; 1123 uint32_t fp;
832 WANT (4); 1124 WANT (4);
833 memcpy (&fp, dec->cur, 4); 1125 memcpy (&fp, dec->cur, 4);
834 dec->cur += 4; 1126 dec->cur += 4;
837 fp = ecb_bswap32 (fp); 1129 fp = ecb_bswap32 (fp);
838 1130
839 return newSVnv (ecb_binary32_to_float (fp)); 1131 return newSVnv (ecb_binary32_to_float (fp));
840 } 1132 }
841 1133
842 case 27: 1134 case MISC_FLOAT64:
843 { 1135 {
844 uint64_t fp; 1136 uint64_t fp;
845 WANT (8); 1137 WANT (8);
846 memcpy (&fp, dec->cur, 8); 1138 memcpy (&fp, dec->cur, 8);
847 dec->cur += 8; 1139 dec->cur += 8;
850 fp = ecb_bswap64 (fp); 1142 fp = ecb_bswap64 (fp);
851 1143
852 return newSVnv (ecb_binary64_to_double (fp)); 1144 return newSVnv (ecb_binary64_to_double (fp));
853 } 1145 }
854 1146
855 // 0..19 unassigned 1147 // 0..19 unassigned simple
856 // 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
857 default: 1151 default:
858 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1152 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
859 } 1153 }
860 1154
861 break; 1155 break;
862 } 1156 }
863 1157
866} 1160}
867 1161
868static SV * 1162static SV *
869decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1163decode_cbor (SV *string, CBOR *cbor, char **offset_return)
870{ 1164{
871 dec_t dec; 1165 dec_t dec = { };
872 SV *sv; 1166 SV *sv;
873 STRLEN len; 1167 STRLEN len;
874 char *data = SvPVbyte (string, len); 1168 char *data = SvPVbyte (string, len);
875 1169
876 if (len > cbor->max_size && cbor->max_size) 1170 if (len > cbor->max_size && cbor->max_size)
878 (unsigned long)len, (unsigned long)cbor->max_size); 1172 (unsigned long)len, (unsigned long)cbor->max_size);
879 1173
880 dec.cbor = *cbor; 1174 dec.cbor = *cbor;
881 dec.cur = (U8 *)data; 1175 dec.cur = (U8 *)data;
882 dec.end = (U8 *)data + len; 1176 dec.end = (U8 *)data + len;
883 dec.err = 0;
884 dec.depth = 0;
885 1177
886 sv = decode_sv (&dec); 1178 sv = decode_sv (&dec);
887 1179
888 if (offset_return) 1180 if (offset_return)
889 *offset_return = dec.cur; 1181 *offset_return = dec.cur;
892 if (dec.cur != dec.end && !dec.err) 1184 if (dec.cur != dec.end && !dec.err)
893 dec.err = "garbage after CBOR object"; 1185 dec.err = "garbage after CBOR object";
894 1186
895 if (dec.err) 1187 if (dec.err)
896 { 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
897 SvREFCNT_dec (sv); 1200 SvREFCNT_dec (sv);
898 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (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);
899 } 1202 }
900 1203
901 sv = sv_2mortal (sv); 1204 sv = sv_2mortal (sv);
902 1205
903 return sv; 1206 return sv;
904} 1207}
905 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
906///////////////////////////////////////////////////////////////////////////// 1329/////////////////////////////////////////////////////////////////////////////
907// XS interface functions 1330// XS interface functions
908 1331
909MODULE = CBOR::XS PACKAGE = CBOR::XS 1332MODULE = CBOR::XS PACKAGE = CBOR::XS
910 1333
917 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1); 1340 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
918 1341
919 types_true = get_bool ("Types::Serialiser::true" ); 1342 types_true = get_bool ("Types::Serialiser::true" );
920 types_false = get_bool ("Types::Serialiser::false"); 1343 types_false = get_bool ("Types::Serialiser::false");
921 types_error = get_bool ("Types::Serialiser::error"); 1344 types_error = get_bool ("Types::Serialiser::error");
1345
1346 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
922 1347
923 sv_cbor = newSVpv ("CBOR", 0); 1348 sv_cbor = newSVpv ("CBOR", 0);
924 SvREADONLY_on (sv_cbor); 1349 SvREADONLY_on (sv_cbor);
925} 1350}
926 1351
947 1372
948void shrink (CBOR *self, int enable = 1) 1373void shrink (CBOR *self, int enable = 1)
949 ALIAS: 1374 ALIAS:
950 shrink = F_SHRINK 1375 shrink = F_SHRINK
951 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
952 PPCODE: 1381 PPCODE:
953{ 1382{
954 if (enable) 1383 if (enable)
955 self->flags |= ix; 1384 self->flags |= ix;
956 else 1385 else
961 1390
962void get_shrink (CBOR *self) 1391void get_shrink (CBOR *self)
963 ALIAS: 1392 ALIAS:
964 get_shrink = F_SHRINK 1393 get_shrink = F_SHRINK
965 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
966 PPCODE: 1399 PPCODE:
967 XPUSHs (boolSV (self->flags & ix)); 1400 XPUSHs (boolSV (self->flags & ix));
968 1401
969void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1402void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
970 PPCODE: 1403 PPCODE:
983 XPUSHs (ST (0)); 1416 XPUSHs (ST (0));
984 1417
985int get_max_size (CBOR *self) 1418int get_max_size (CBOR *self)
986 CODE: 1419 CODE:
987 RETVAL = self->max_size; 1420 RETVAL = self->max_size;
1421 OUTPUT:
1422 RETVAL
1423
1424void filter (CBOR *self, SV *filter = 0)
1425 PPCODE:
1426 SvREFCNT_dec (self->filter);
1427 self->filter = filter ? newSVsv (filter) : filter;
1428 XPUSHs (ST (0));
1429
1430SV *get_filter (CBOR *self)
1431 CODE:
1432 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
988 OUTPUT: 1433 OUTPUT:
989 RETVAL 1434 RETVAL
990 1435
991void encode (CBOR *self, SV *scalar) 1436void encode (CBOR *self, SV *scalar)
992 PPCODE: 1437 PPCODE:
1007 EXTEND (SP, 2); 1452 EXTEND (SP, 2);
1008 PUSHs (sv); 1453 PUSHs (sv);
1009 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1454 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1010} 1455}
1011 1456
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}
1508
1509void DESTROY (CBOR *self)
1510 PPCODE:
1511 cbor_free (self);
1512
1012PROTOTYPES: ENABLE 1513PROTOTYPES: ENABLE
1013 1514
1014void encode_cbor (SV *scalar) 1515void encode_cbor (SV *scalar)
1516 ALIAS:
1517 encode_cbor = 0
1518 encode_cbor_sharing = F_ALLOW_SHARING
1015 PPCODE: 1519 PPCODE:
1016{ 1520{
1017 CBOR cbor; 1521 CBOR cbor;
1018 cbor_init (&cbor); 1522 cbor_init (&cbor);
1523 cbor.flags |= ix;
1019 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1524 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1020 XPUSHs (scalar); 1525 XPUSHs (scalar);
1021} 1526}
1022 1527
1023void decode_cbor (SV *cborstr) 1528void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines