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 1.18 by root, Wed Nov 20 01:09:46 2013 UTC vs.
Revision 1.44 by root, Tue Feb 18 22:12:12 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
70 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
29 CBOR_TAG_PERL_OBJECT = 24, // http://cbor.schmorp.de/perl-object 71 CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object
30 CBOR_TAG_GENERIC_OBJECT = 25, // http://cbor.schmorp.de/generic-object 72 CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object
31 CBOR_TAG_VALUE_SHARABLE = 26, // http://cbor.schmorp.de/value-sharing 73 CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing
32 CBOR_TAG_VALUE_SHAREDREF = 27, // http://cbor.schmorp.de/value-sharing 74 CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing
33 CBOR_TAG_STRINGREF_NAMESPACE = 65537, // http://cbor.schmorp.de/stringref 75 CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref
34 CBOR_TAG_STRINGREF = 28, // http://cbor.schmorp.de/stringref
35 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection 76 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection
36 77
37 // rfc7049 78 // rfc7049
38 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 79 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
39 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any 80 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
40 CBOR_TAG_POS_BIGNUM = 2, // byte string 81 CBOR_TAG_POS_BIGNUM = 2, // byte string
41 CBOR_TAG_NEG_BIGNUM = 3, // byte string 82 CBOR_TAG_NEG_BIGNUM = 3, // byte string
42 CBOR_TAG_DECIMAL = 4, // decimal fraction, array 83 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
43 CBOR_TAG_BIGFLOAT = 5, // array 84 CBOR_TAG_BIGFLOAT = 5, // array
44 85
45 CBOR_TAG_CONV_B64U = 21, // base64url, any 86 CBOR_TAG_CONV_B64U = 21, // base64url, any
46 CBOR_TAG_CONV_B64 = 22, // base64, any 87 CBOR_TAG_CONV_B64 = 22, // base64, any
47 CBOR_TAG_CONV_HEX = 23, // base16, any 88 CBOR_TAG_CONV_HEX = 23, // base16, any
48 CBOR_TAG_CBOR = 24, // embedded cbor, byte string 89 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
49 90
50 CBOR_TAG_URI = 32, // URI rfc3986, utf-8 91 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
51 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8 92 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
52 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8 93 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
53 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8 94 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
54 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 95 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
55 96
56 CBOR_TAG_MAGIC = 55799 // self-describe cbor 97 CBOR_TAG_MAGIC = 55799, // self-describe cbor
57}; 98};
58 99
59#define F_SHRINK 0x00000001UL 100#define F_SHRINK 0x00000001UL
60#define F_ALLOW_UNKNOWN 0x00000002UL 101#define F_ALLOW_UNKNOWN 0x00000002UL
61#define F_ALLOW_SHARING 0x00000004UL //TODO 102#define F_ALLOW_SHARING 0x00000004UL
103#define F_ALLOW_CYCLES 0x00000008UL
62#define F_DEDUP_STRINGS 0x00000008UL //TODO 104#define F_PACK_STRINGS 0x00000010UL
63#define F_DEDUP_KEYS 0x00000010UL //TODO 105#define F_VALIDATE_UTF8 0x00000020UL
64 106
65#define INIT_SIZE 32 // initial scalar size to be allocated 107#define INIT_SIZE 32 // initial scalar size to be allocated
66 108
67#define SB do { 109#define SB do {
68#define SE } while (0) 110#define SE } while (0)
80# define CBOR_SLOW 0 122# define CBOR_SLOW 0
81# define CBOR_STASH cbor_stash 123# define CBOR_STASH cbor_stash
82#endif 124#endif
83 125
84static 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::
85static SV *types_true, *types_false, *types_error, *sv_cbor; 127static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
86 128
87typedef struct { 129typedef struct {
88 U32 flags; 130 U32 flags;
89 U32 max_depth; 131 U32 max_depth;
90 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.
91} CBOR; 139} CBOR;
92 140
93ecb_inline void 141ecb_inline void
94cbor_init (CBOR *cbor) 142cbor_init (CBOR *cbor)
95{ 143{
96 Zero (cbor, 1, CBOR); 144 Zero (cbor, 1, CBOR);
97 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);
98} 153}
99 154
100///////////////////////////////////////////////////////////////////////////// 155/////////////////////////////////////////////////////////////////////////////
101// utility functions 156// utility functions
102 157
124 SvPV_renew (sv, SvCUR (sv) + 1); 179 SvPV_renew (sv, SvCUR (sv) + 1);
125#endif 180#endif
126 } 181 }
127} 182}
128 183
184// minimum length of a string to be registered for stringref
185ecb_inline int
186minimum_string_length (UV idx)
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}
198
129///////////////////////////////////////////////////////////////////////////// 199/////////////////////////////////////////////////////////////////////////////
130// encoder 200// encoder
131 201
132// structure used for encoding CBOR 202// structure used for encoding CBOR
133typedef struct 203typedef struct
135 char *cur; // SvPVX (sv) + current output position 205 char *cur; // SvPVX (sv) + current output position
136 char *end; // SvEND (sv) 206 char *end; // SvEND (sv)
137 SV *sv; // result scalar 207 SV *sv; // result scalar
138 CBOR cbor; 208 CBOR cbor;
139 U32 depth; // recursion level 209 U32 depth; // recursion level
140 HV *stringref; // string => index, or 0 210 HV *stringref[2]; // string => index, or 0 ([0] = bytes, [1] = utf-8)
211 UV stringref_idx;
141 HV *sharable; // ptr => index, or 0 212 HV *shareable; // ptr => index, or 0
142 HV *sharable_idx; 213 UV shareable_idx;
143} enc_t; 214} enc_t;
144 215
145ecb_inline void 216ecb_inline void
146need (enc_t *enc, STRLEN len) 217need (enc_t *enc, STRLEN len)
147{ 218{
164static void 235static void
165encode_uint (enc_t *enc, int major, UV len) 236encode_uint (enc_t *enc, int major, UV len)
166{ 237{
167 need (enc, 9); 238 need (enc, 9);
168 239
169 if (len < 24) 240 if (ecb_expect_true (len < LENGTH_EXT1))
170 *enc->cur++ = major | len; 241 *enc->cur++ = major | len;
171 else if (len <= 0xff) 242 else if (ecb_expect_true (len <= 0xffU))
172 { 243 {
173 *enc->cur++ = major | 24; 244 *enc->cur++ = major | LENGTH_EXT1;
174 *enc->cur++ = len; 245 *enc->cur++ = len;
175 } 246 }
176 else if (len <= 0xffff) 247 else if (len <= 0xffffU)
177 { 248 {
178 *enc->cur++ = major | 25; 249 *enc->cur++ = major | LENGTH_EXT2;
179 *enc->cur++ = len >> 8; 250 *enc->cur++ = len >> 8;
180 *enc->cur++ = len; 251 *enc->cur++ = len;
181 } 252 }
182 else if (len <= 0xffffffff) 253 else if (len <= 0xffffffffU)
183 { 254 {
184 *enc->cur++ = major | 26; 255 *enc->cur++ = major | LENGTH_EXT4;
185 *enc->cur++ = len >> 24; 256 *enc->cur++ = len >> 24;
186 *enc->cur++ = len >> 16; 257 *enc->cur++ = len >> 16;
187 *enc->cur++ = len >> 8; 258 *enc->cur++ = len >> 8;
188 *enc->cur++ = len; 259 *enc->cur++ = len;
189 } 260 }
190 else 261 else
191 { 262 {
192 *enc->cur++ = major | 27; 263 *enc->cur++ = major | LENGTH_EXT8;
193 *enc->cur++ = len >> 56; 264 *enc->cur++ = len >> 56;
194 *enc->cur++ = len >> 48; 265 *enc->cur++ = len >> 48;
195 *enc->cur++ = len >> 40; 266 *enc->cur++ = len >> 40;
196 *enc->cur++ = len >> 32; 267 *enc->cur++ = len >> 32;
197 *enc->cur++ = len >> 24; 268 *enc->cur++ = len >> 24;
199 *enc->cur++ = len >> 8; 270 *enc->cur++ = len >> 8;
200 *enc->cur++ = len; 271 *enc->cur++ = len;
201 } 272 }
202} 273}
203 274
204static void 275ecb_inline void
276encode_tag (enc_t *enc, UV tag)
277{
278 encode_uint (enc, MAJOR_TAG, tag);
279}
280
281ecb_inline void
205encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 282encode_str (enc_t *enc, int utf8, char *str, STRLEN len)
206{ 283{
207 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 284 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
208 need (enc, len); 285 need (enc, len);
209 memcpy (enc->cur, str, len); 286 memcpy (enc->cur, str, len);
210 enc->cur += len; 287 enc->cur += len;
211} 288}
212 289
213ecb_inline void
214encode_tag (enc_t *enc, UV tag)
215{
216 encode_uint (enc, 0xc0, tag);
217}
218
219static int 290static void
220encode_sharable2 (enc_t *enc, SV *sv) 291encode_strref (enc_t *enc, int utf8, char *str, STRLEN len)
221{ 292{
222 if (!enc->sharable) 293 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
223 enc->sharable = (HV *)sv_2mortal ((SV *)newHV ()); 294 {
295 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
224 296
225 SV **svp = hv_fetch (enc->sharable, &sv, sizeof (sv), 1);
226
227 if (SvOK (*svp)) 297 if (SvOK (*svp))
228 { 298 {
299 // already registered, use stringref
229 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF); 300 encode_tag (enc, CBOR_TAG_STRINGREF);
230 encode_uint (enc, 0x00, SvUV (*svp)); 301 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
231
232 return 1; 302 return;
233 } 303 }
234 else 304 else if (len >= minimum_string_length (enc->stringref_idx))
235 { 305 {
306 // register only
236 sv_setuv (*svp, enc->sharable_idx++); 307 sv_setuv (*svp, enc->stringref_idx);
237 encode_tag (enc, CBOR_TAG_VALUE_SHARABLE); 308 ++enc->stringref_idx;
238 309 }
239 return 0;
240 } 310 }
241}
242 311
243ecb_inline int 312 encode_str (enc, utf8, str, len);
244encode_sharable (enc_t *enc, SV *sv)
245{
246 if (ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)
247 && ecb_expect_false (SvREFCNT (sv) > 1))
248 return encode_sharable2 (enc, sv);
249
250 return 0;
251} 313}
252 314
253static void encode_sv (enc_t *enc, SV *sv); 315static void encode_sv (enc_t *enc, SV *sv);
254 316
255static void 317static void
260 if (enc->depth >= enc->cbor.max_depth) 322 if (enc->depth >= enc->cbor.max_depth)
261 croak (ERR_NESTING_EXCEEDED); 323 croak (ERR_NESTING_EXCEEDED);
262 324
263 ++enc->depth; 325 ++enc->depth;
264 326
265 encode_uint (enc, 0x80, len + 1); 327 encode_uint (enc, MAJOR_ARRAY, len + 1);
266 328
267 for (i = 0; i <= len; ++i) 329 for (i = 0; i <= len; ++i)
268 { 330 {
269 SV **svp = av_fetch (av, i, 0); 331 SV **svp = av_fetch (av, i, 0);
270 encode_sv (enc, svp ? *svp : &PL_sv_undef); 332 encode_sv (enc, svp ? *svp : &PL_sv_undef);
285 347
286 int pairs = hv_iterinit (hv); 348 int pairs = hv_iterinit (hv);
287 int mg = SvMAGICAL (hv); 349 int mg = SvMAGICAL (hv);
288 350
289 if (mg) 351 if (mg)
290 encode_ch (enc, 0xa0 | 31); 352 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
291 else 353 else
292 encode_uint (enc, 0xa0, pairs); 354 encode_uint (enc, MAJOR_MAP, pairs);
293 355
294 while ((he = hv_iternext (hv))) 356 while ((he = hv_iternext (hv)))
295 { 357 {
296 if (HeKLEN (he) == HEf_SVKEY) 358 if (HeKLEN (he) == HEf_SVKEY)
297 encode_sv (enc, HeSVKEY (he)); 359 encode_sv (enc, HeSVKEY (he));
298 else 360 else
299 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 361 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
300 362
301 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 363 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
302 } 364 }
303 365
304 if (mg) 366 if (mg)
305 encode_ch (enc, 0xe0 | 31); 367 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
306 368
307 --enc->depth; 369 --enc->depth;
308} 370}
309 371
310// encode objects, arrays and special \0=false and \1=true values. 372// encode objects, arrays and special \0=false and \1=true values.
311static void 373static void
312encode_rv (enc_t *enc, SV *sv) 374encode_rv (enc_t *enc, SV *sv)
313{ 375{
314 svtype svt;
315
316 SvGETMAGIC (sv); 376 SvGETMAGIC (sv);
377
317 svt = SvTYPE (sv); 378 svtype svt = SvTYPE (sv);
318
319 if (encode_sharable (enc, sv))
320 return;
321 379
322 if (ecb_expect_false (SvOBJECT (sv))) 380 if (ecb_expect_false (SvOBJECT (sv)))
323 { 381 {
324 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash 382 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
325 ? types_boolean_stash 383 ? types_boolean_stash
330 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 388 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
331 ? cbor_tagged_stash 389 ? cbor_tagged_stash
332 : gv_stashpv ("CBOR::XS::Tagged" , 1); 390 : gv_stashpv ("CBOR::XS::Tagged" , 1);
333 391
334 HV *stash = SvSTASH (sv); 392 HV *stash = SvSTASH (sv);
335 GV *method;
336 393
337 if (stash == boolean_stash) 394 if (stash == boolean_stash)
338 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 395 {
396 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
397 return;
398 }
339 else if (stash == error_stash) 399 else if (stash == error_stash)
340 encode_ch (enc, 0xe0 | 23); 400 {
401 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
402 return;
403 }
341 else if (stash == tagged_stash) 404 else if (stash == tagged_stash)
342 { 405 {
343 if (svt != SVt_PVAV) 406 if (svt != SVt_PVAV)
344 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 407 croak ("encountered CBOR::XS::Tagged object that isn't an array");
345 408
346 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 409 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
347 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 410 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
411
412 return;
413 }
414 }
415
416 if (ecb_expect_false (SvREFCNT (sv) > 1)
417 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
418 {
419 if (!enc->shareable)
420 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
421
422 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
423
424 if (SvOK (*svp))
348 } 425 {
426 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
427 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
428 return;
429 }
430 else
431 {
432 sv_setuv (*svp, enc->shareable_idx);
433 ++enc->shareable_idx;
434 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
435 }
436 }
437
438 if (ecb_expect_false (SvOBJECT (sv)))
439 {
440 HV *stash = SvSTASH (sv);
441 GV *method;
442
349 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 443 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
350 { 444 {
351 dSP; 445 dSP;
352 446
353 ENTER; SAVETMPS; PUSHMARK (SP); 447 ENTER; SAVETMPS; PUSHMARK (SP);
354 // we re-bless the reference to get overload and other niceties right 448 // we re-bless the reference to get overload and other niceties right
386 // catch this surprisingly common error 480 // catch this surprisingly common error
387 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 481 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
388 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); 482 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
389 483
390 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 484 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
391 encode_uint (enc, 0x80, count + 1); 485 encode_uint (enc, MAJOR_ARRAY, count + 1);
392 encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 486 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
393 487
394 while (count) 488 while (count)
395 encode_sv (enc, SP[1 - count--]); 489 encode_sv (enc, SP[1 - count--]);
396 490
397 PUTBACK; 491 PUTBACK;
418{ 512{
419 double nv = SvNVX (sv); 513 double nv = SvNVX (sv);
420 514
421 need (enc, 9); 515 need (enc, 9);
422 516
423 if (ecb_expect_false (nv == (U32)nv)) 517 if (ecb_expect_false (nv == (NV)(U32)nv))
424 encode_uint (enc, 0x00, (U32)nv); 518 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
425 //TODO: maybe I32? 519 //TODO: maybe I32?
426 else if (ecb_expect_false (nv == (float)nv)) 520 else if (ecb_expect_false (nv == (float)nv))
427 { 521 {
428 uint32_t fp = ecb_float_to_binary32 (nv); 522 uint32_t fp = ecb_float_to_binary32 (nv);
429 523
430 *enc->cur++ = 0xe0 | 26; 524 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
431 525
432 if (!ecb_big_endian ()) 526 if (!ecb_big_endian ())
433 fp = ecb_bswap32 (fp); 527 fp = ecb_bswap32 (fp);
434 528
435 memcpy (enc->cur, &fp, 4); 529 memcpy (enc->cur, &fp, 4);
437 } 531 }
438 else 532 else
439 { 533 {
440 uint64_t fp = ecb_double_to_binary64 (nv); 534 uint64_t fp = ecb_double_to_binary64 (nv);
441 535
442 *enc->cur++ = 0xe0 | 27; 536 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
443 537
444 if (!ecb_big_endian ()) 538 if (!ecb_big_endian ())
445 fp = ecb_bswap64 (fp); 539 fp = ecb_bswap64 (fp);
446 540
447 memcpy (enc->cur, &fp, 8); 541 memcpy (enc->cur, &fp, 8);
452static void 546static void
453encode_sv (enc_t *enc, SV *sv) 547encode_sv (enc_t *enc, SV *sv)
454{ 548{
455 SvGETMAGIC (sv); 549 SvGETMAGIC (sv);
456 550
457 if (encode_sharable (enc, sv))
458 return;
459
460 if (SvPOKp (sv)) 551 if (SvPOKp (sv))
461 { 552 {
462 STRLEN len; 553 STRLEN len;
463 char *str = SvPV (sv, len); 554 char *str = SvPV (sv, len);
464 encode_str (enc, SvUTF8 (sv), str, len); 555 encode_strref (enc, SvUTF8 (sv), str, len);
465 } 556 }
466 else if (SvNOKp (sv)) 557 else if (SvNOKp (sv))
467 encode_nv (enc, sv); 558 encode_nv (enc, sv);
468 else if (SvIOKp (sv)) 559 else if (SvIOKp (sv))
469 { 560 {
470 if (SvIsUV (sv)) 561 if (SvIsUV (sv))
471 encode_uint (enc, 0x00, SvUVX (sv)); 562 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
472 else if (SvIVX (sv) >= 0) 563 else if (SvIVX (sv) >= 0)
473 encode_uint (enc, 0x00, SvIVX (sv)); 564 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
474 else 565 else
475 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 566 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
476 } 567 }
477 else if (SvROK (sv)) 568 else if (SvROK (sv))
478 encode_rv (enc, SvRV (sv)); 569 encode_rv (enc, SvRV (sv));
479 else if (!SvOK (sv)) 570 else if (!SvOK (sv))
480 encode_ch (enc, 0xe0 | 22); 571 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
481 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 572 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
482 encode_ch (enc, 0xe0 | 23); 573 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
483 else 574 else
484 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 575 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
485 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 576 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
486} 577}
487 578
494 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 585 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
495 enc.cur = SvPVX (enc.sv); 586 enc.cur = SvPVX (enc.sv);
496 enc.end = SvEND (enc.sv); 587 enc.end = SvEND (enc.sv);
497 588
498 SvPOK_only (enc.sv); 589 SvPOK_only (enc.sv);
590
591 if (cbor->flags & F_PACK_STRINGS)
592 {
593 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
594 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
595 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
596 }
597
499 encode_sv (&enc, scalar); 598 encode_sv (&enc, scalar);
500 599
501 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 600 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
502 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 601 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
503 602
517 U8 *end; // end of input string 616 U8 *end; // end of input string
518 const char *err; // parse error, if != 0 617 const char *err; // parse error, if != 0
519 CBOR cbor; 618 CBOR cbor;
520 U32 depth; // recursion depth 619 U32 depth; // recursion depth
521 U32 maxdepth; // recursion depth limit 620 U32 maxdepth; // recursion depth limit
522 AV *sharable; 621 AV *shareable;
622 AV *stringref;
623 SV *decode_tagged;
523} dec_t; 624} dec_t;
524 625
525#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 626#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE
526 627
527#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 628#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data")
530#define DEC_DEC_DEPTH --dec->depth 631#define DEC_DEC_DEPTH --dec->depth
531 632
532static UV 633static UV
533decode_uint (dec_t *dec) 634decode_uint (dec_t *dec)
534{ 635{
535 switch (*dec->cur & 31) 636 U8 m = *dec->cur & MINOR_MASK;
536 { 637 ++dec->cur;
537 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
538 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
539 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
540 return *dec->cur++ & 31;
541 638
542 case 24: 639 if (ecb_expect_true (m < LENGTH_EXT1))
640 return m;
641 else if (ecb_expect_true (m == LENGTH_EXT1))
642 {
543 WANT (2); 643 WANT (1);
544 dec->cur += 2; 644 dec->cur += 1;
545 return dec->cur[-1]; 645 return dec->cur[-1];
546 646 }
547 case 25: 647 else if (ecb_expect_true (m == LENGTH_EXT2))
648 {
548 WANT (3); 649 WANT (2);
549 dec->cur += 3; 650 dec->cur += 2;
550 return (((UV)dec->cur[-2]) << 8) 651 return (((UV)dec->cur[-2]) << 8)
551 | ((UV)dec->cur[-1]); 652 | ((UV)dec->cur[-1]);
552 653 }
553 case 26: 654 else if (ecb_expect_true (m == LENGTH_EXT4))
655 {
554 WANT (5); 656 WANT (4);
555 dec->cur += 5; 657 dec->cur += 4;
556 return (((UV)dec->cur[-4]) << 24) 658 return (((UV)dec->cur[-4]) << 24)
557 | (((UV)dec->cur[-3]) << 16) 659 | (((UV)dec->cur[-3]) << 16)
558 | (((UV)dec->cur[-2]) << 8) 660 | (((UV)dec->cur[-2]) << 8)
559 | ((UV)dec->cur[-1]); 661 | ((UV)dec->cur[-1]);
560 662 }
561 case 27: 663 else if (ecb_expect_true (m == LENGTH_EXT8))
664 {
562 WANT (9); 665 WANT (8);
563 dec->cur += 9; 666 dec->cur += 8;
667
668 return
669#if UVSIZE < 8
670 0
671#else
564 return (((UV)dec->cur[-8]) << 56) 672 (((UV)dec->cur[-8]) << 56)
565 | (((UV)dec->cur[-7]) << 48) 673 | (((UV)dec->cur[-7]) << 48)
566 | (((UV)dec->cur[-6]) << 40) 674 | (((UV)dec->cur[-6]) << 40)
567 | (((UV)dec->cur[-5]) << 32) 675 | (((UV)dec->cur[-5]) << 32)
676#endif
568 | (((UV)dec->cur[-4]) << 24) 677 | (((UV)dec->cur[-4]) << 24)
569 | (((UV)dec->cur[-3]) << 16) 678 | (((UV)dec->cur[-3]) << 16)
570 | (((UV)dec->cur[-2]) << 8) 679 | (((UV)dec->cur[-2]) << 8)
571 | ((UV)dec->cur[-1]); 680 | ((UV)dec->cur[-1]);
572 681 }
573 default: 682 else
574 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 683 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
575 }
576 684
577fail: 685fail:
578 return 0; 686 return 0;
579} 687}
580 688
585{ 693{
586 AV *av = newAV (); 694 AV *av = newAV ();
587 695
588 DEC_INC_DEPTH; 696 DEC_INC_DEPTH;
589 697
590 if ((*dec->cur & 31) == 31) 698 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
591 { 699 {
592 ++dec->cur; 700 ++dec->cur;
593 701
594 for (;;) 702 for (;;)
595 { 703 {
596 WANT (1); 704 WANT (1);
597 705
598 if (*dec->cur == (0xe0 | 31)) 706 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
599 { 707 {
600 ++dec->cur; 708 ++dec->cur;
601 break; 709 break;
602 } 710 }
603 711
606 } 714 }
607 else 715 else
608 { 716 {
609 int i, len = decode_uint (dec); 717 int i, len = decode_uint (dec);
610 718
719 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
611 av_fill (av, len - 1); 720 av_fill (av, len - 1);
612 721
613 for (i = 0; i < len; ++i) 722 for (i = 0; i < len; ++i)
614 AvARRAY (av)[i] = decode_sv (dec); 723 AvARRAY (av)[i] = decode_sv (dec);
615 } 724 }
625 734
626static void 735static void
627decode_he (dec_t *dec, HV *hv) 736decode_he (dec_t *dec, HV *hv)
628{ 737{
629 // for speed reasons, we specialcase single-string 738 // for speed reasons, we specialcase single-string
630 // byte or utf-8 strings as keys. 739 // byte or utf-8 strings as keys, but only when !stringref
631 740
632 if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27) 741 if (ecb_expect_true (!dec->stringref))
742 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
633 { 743 {
634 I32 len = decode_uint (dec); 744 I32 len = decode_uint (dec);
635 char *key = (char *)dec->cur; 745 char *key = (char *)dec->cur;
636 746
637 dec->cur += len; 747 dec->cur += len;
638 748
639 hv_store (hv, key, len, decode_sv (dec), 0); 749 hv_store (hv, key, len, decode_sv (dec), 0);
750
751 return;
640 } 752 }
641 else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27) 753 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
642 { 754 {
643 I32 len = decode_uint (dec); 755 I32 len = decode_uint (dec);
644 char *key = (char *)dec->cur; 756 char *key = (char *)dec->cur;
645 757
646 dec->cur += len; 758 dec->cur += len;
647 759
760 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
761 if (!is_utf8_string (key, len))
762 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
763
648 hv_store (hv, key, -len, decode_sv (dec), 0); 764 hv_store (hv, key, -len, decode_sv (dec), 0);
765
766 return;
649 } 767 }
650 else 768
651 {
652 SV *k = decode_sv (dec); 769 SV *k = decode_sv (dec);
653 SV *v = decode_sv (dec); 770 SV *v = decode_sv (dec);
654 771
655 hv_store_ent (hv, k, v, 0); 772 hv_store_ent (hv, k, v, 0);
656 SvREFCNT_dec (k); 773 SvREFCNT_dec (k);
657 } 774
775fail:
776 ;
658} 777}
659 778
660static SV * 779static SV *
661decode_hv (dec_t *dec) 780decode_hv (dec_t *dec)
662{ 781{
663 HV *hv = newHV (); 782 HV *hv = newHV ();
664 783
665 DEC_INC_DEPTH; 784 DEC_INC_DEPTH;
666 785
667 if ((*dec->cur & 31) == 31) 786 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
668 { 787 {
669 ++dec->cur; 788 ++dec->cur;
670 789
671 for (;;) 790 for (;;)
672 { 791 {
673 WANT (1); 792 WANT (1);
674 793
675 if (*dec->cur == (0xe0 | 31)) 794 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
676 { 795 {
677 ++dec->cur; 796 ++dec->cur;
678 break; 797 break;
679 } 798 }
680 799
701static SV * 820static SV *
702decode_str (dec_t *dec, int utf8) 821decode_str (dec_t *dec, int utf8)
703{ 822{
704 SV *sv = 0; 823 SV *sv = 0;
705 824
706 if ((*dec->cur & 31) == 31) 825 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
707 { 826 {
827 // indefinite length strings
708 ++dec->cur; 828 ++dec->cur;
709 829
830 U8 major = *dec->cur & MAJOR_MISC;
831
710 sv = newSVpvn ("", 0); 832 sv = newSVpvn ("", 0);
711 833
712 // not very fast, and certainly not robust against illegal input
713 for (;;) 834 for (;;)
714 { 835 {
715 WANT (1); 836 WANT (1);
716 837
717 if (*dec->cur == (0xe0 | 31)) 838 if ((*dec->cur - major) > LENGTH_EXT8)
839 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
718 { 840 {
719 ++dec->cur; 841 ++dec->cur;
720 break; 842 break;
721 } 843 }
844 else
845 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
722 846
723 sv_catsv (sv, decode_sv (dec)); 847 STRLEN len = decode_uint (dec);
848
849 WANT (len);
850 sv_catpvn (sv, dec->cur, len);
851 dec->cur += len;
724 } 852 }
725 } 853 }
726 else 854 else
727 { 855 {
728 STRLEN len = decode_uint (dec); 856 STRLEN len = decode_uint (dec);
729 857
730 WANT (len); 858 WANT (len);
731 sv = newSVpvn (dec->cur, len); 859 sv = newSVpvn (dec->cur, len);
732 dec->cur += len; 860 dec->cur += len;
861
862 if (ecb_expect_false (dec->stringref)
863 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
864 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
733 } 865 }
734 866
735 if (utf8) 867 if (utf8)
868 {
869 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
870 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
871 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
872
736 SvUTF8_on (sv); 873 SvUTF8_on (sv);
874 }
737 875
738 return sv; 876 return sv;
739 877
740fail: 878fail:
741 SvREFCNT_dec (sv); 879 SvREFCNT_dec (sv);
743} 881}
744 882
745static SV * 883static SV *
746decode_tagged (dec_t *dec) 884decode_tagged (dec_t *dec)
747{ 885{
886 SV *sv = 0;
748 UV tag = decode_uint (dec); 887 UV tag = decode_uint (dec);
749 SV *sv = decode_sv (dec); 888
889 WANT (1);
750 890
751 switch (tag) 891 switch (tag)
752 { 892 {
753 case CBOR_TAG_MAGIC: 893 case CBOR_TAG_MAGIC:
754 return sv; 894 sv = decode_sv (dec);
895 break;
755 896
756 case CBOR_TAG_INDIRECTION: 897 case CBOR_TAG_INDIRECTION:
757 return newRV_noinc (sv); 898 sv = newRV_noinc (decode_sv (dec));
899 break;
758 900
901 case CBOR_TAG_STRINGREF_NAMESPACE:
902 {
903 ENTER; SAVETMPS;
904
905 SAVESPTR (dec->stringref);
906 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
907
908 sv = decode_sv (dec);
909
910 FREETMPS; LEAVE;
911 }
912 break;
913
914 case CBOR_TAG_STRINGREF:
915 {
916 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
917 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
918
919 UV idx = decode_uint (dec);
920
921 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
922 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
923
924 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
925 }
926 break;
927
759 case CBOR_TAG_VALUE_SHARABLE: 928 case CBOR_TAG_VALUE_SHAREABLE:
929 {
760 if (ecb_expect_false (!dec->sharable)) 930 if (ecb_expect_false (!dec->shareable))
761 dec->sharable = (AV *)sv_2mortal ((SV *)newAV ()); 931 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
762 932
933 if (dec->cbor.flags & F_ALLOW_CYCLES)
934 {
935 sv = newSV (0);
763 av_push (dec->sharable, SvREFCNT_inc_NN (sv)); 936 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
764 937
765 return sv; 938 SV *osv = decode_sv (dec);
939 sv_setsv (sv, osv);
940 SvREFCNT_dec_NN (osv);
941 }
942 else
943 {
944 av_push (dec->shareable, &PL_sv_undef);
945 int idx = AvFILLp (dec->shareable);
946 sv = decode_sv (dec);
947 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
948 }
949 }
950 break;
766 951
767 case CBOR_TAG_VALUE_SHAREDREF: 952 case CBOR_TAG_VALUE_SHAREDREF:
768 { 953 {
769 // TODO: should verify that the sv atcually was a CBOR unsigned integer 954 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
770 UV idx = SvUV (sv); 955 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
771 956
957 UV idx = decode_uint (dec);
958
772 if (!dec->sharable || idx > AvFILLp (dec->sharable)) 959 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
773 ERR ("corrupted CBOR data (sharedref index out of bounds)"); 960 ERR ("corrupted CBOR data (sharedref index out of bounds)");
774 961
775 SvREFCNT_dec (sv);
776
777 return SvREFCNT_inc_NN (AvARRAY (dec->sharable)[idx]); 962 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
963
964 if (sv == &PL_sv_undef)
965 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
778 } 966 }
967 break;
779 968
780 case CBOR_TAG_PERL_OBJECT: 969 case CBOR_TAG_PERL_OBJECT:
781 { 970 {
971 sv = decode_sv (dec);
972
782 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 973 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
783 ERR ("corrupted CBOR data (non-array perl object)"); 974 ERR ("corrupted CBOR data (non-array perl object)");
784 975
785 AV *av = (AV *)SvRV (sv); 976 AV *av = (AV *)SvRV (sv);
786 int len = av_len (av) + 1; 977 int len = av_len (av) + 1;
821 sv = SvREFCNT_inc (POPs); 1012 sv = SvREFCNT_inc (POPs);
822 1013
823 PUTBACK; 1014 PUTBACK;
824 1015
825 FREETMPS; LEAVE; 1016 FREETMPS; LEAVE;
826
827 return sv;
828 } 1017 }
1018 break;
829 1019
830 default: 1020 default:
831 { 1021 {
1022 sv = decode_sv (dec);
1023
1024 dSP;
1025 ENTER; SAVETMPS; PUSHMARK (SP);
1026 EXTEND (SP, 2);
1027 PUSHs (newSVuv (tag));
1028 PUSHs (sv);
1029
1030 PUTBACK;
1031 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1032 SPAGAIN;
1033
1034 if (SvTRUE (ERRSV))
1035 {
1036 FREETMPS; LEAVE;
1037 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1038 }
1039
1040 if (count)
1041 {
1042 SvREFCNT_dec (sv);
1043 sv = SvREFCNT_inc (POPs);
1044 }
1045 else
1046 {
832 AV *av = newAV (); 1047 AV *av = newAV ();
833 av_push (av, newSVuv (tag)); 1048 av_push (av, newSVuv (tag));
834 av_push (av, sv); 1049 av_push (av, sv);
835 1050
836 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1051 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
837 ? cbor_tagged_stash 1052 ? cbor_tagged_stash
838 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1053 : gv_stashpv ("CBOR::XS::Tagged" , 1);
839
840 return sv_bless (newRV_noinc ((SV *)av), tagged_stash); 1054 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
841 } 1055 }
1056
1057 PUTBACK;
1058
1059 FREETMPS; LEAVE;
1060 }
1061 break;
842 } 1062 }
1063
1064 return sv;
843 1065
844fail: 1066fail:
845 SvREFCNT_dec (sv); 1067 SvREFCNT_dec (sv);
846 return &PL_sv_undef; 1068 return &PL_sv_undef;
847} 1069}
849static SV * 1071static SV *
850decode_sv (dec_t *dec) 1072decode_sv (dec_t *dec)
851{ 1073{
852 WANT (1); 1074 WANT (1);
853 1075
854 switch (*dec->cur >> 5) 1076 switch (*dec->cur >> MAJOR_SHIFT)
855 { 1077 {
856 case 0: // unsigned int 1078 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
857 return newSVuv (decode_uint (dec)); 1079 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
858 case 1: // negative int 1080 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
859 return newSViv (-1 - (IV)decode_uint (dec)); 1081 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
860 case 2: // octet string 1082 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
861 return decode_str (dec, 0); 1083 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
862 case 3: // utf-8 string 1084 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
863 return decode_str (dec, 1); 1085
864 case 4: // array 1086 case MAJOR_MISC >> MAJOR_SHIFT:
865 return decode_av (dec);
866 case 5: // map
867 return decode_hv (dec);
868 case 6: // tag
869 return decode_tagged (dec);
870 case 7: // misc
871 switch (*dec->cur++ & 31) 1087 switch (*dec->cur++ & MINOR_MASK)
872 { 1088 {
873 case 20: 1089 case SIMPLE_FALSE:
874#if CBOR_SLOW 1090#if CBOR_SLOW
875 types_false = get_bool ("Types::Serialiser::false"); 1091 types_false = get_bool ("Types::Serialiser::false");
876#endif 1092#endif
877 return newSVsv (types_false); 1093 return newSVsv (types_false);
878 case 21: 1094 case SIMPLE_TRUE:
879#if CBOR_SLOW 1095#if CBOR_SLOW
880 types_true = get_bool ("Types::Serialiser::true"); 1096 types_true = get_bool ("Types::Serialiser::true");
881#endif 1097#endif
882 return newSVsv (types_true); 1098 return newSVsv (types_true);
883 case 22: 1099 case SIMPLE_NULL:
884 return newSVsv (&PL_sv_undef); 1100 return newSVsv (&PL_sv_undef);
885 case 23: 1101 case SIMPLE_UNDEF:
886#if CBOR_SLOW 1102#if CBOR_SLOW
887 types_error = get_bool ("Types::Serialiser::error"); 1103 types_error = get_bool ("Types::Serialiser::error");
888#endif 1104#endif
889 return newSVsv (types_error); 1105 return newSVsv (types_error);
890 1106
891 case 25: 1107 case MISC_FLOAT16:
892 { 1108 {
893 WANT (2); 1109 WANT (2);
894 1110
895 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1111 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
896 dec->cur += 2; 1112 dec->cur += 2;
897 1113
898 return newSVnv (ecb_binary16_to_float (fp)); 1114 return newSVnv (ecb_binary16_to_float (fp));
899 } 1115 }
900 1116
901 case 26: 1117 case MISC_FLOAT32:
902 { 1118 {
903 uint32_t fp; 1119 uint32_t fp;
904 WANT (4); 1120 WANT (4);
905 memcpy (&fp, dec->cur, 4); 1121 memcpy (&fp, dec->cur, 4);
906 dec->cur += 4; 1122 dec->cur += 4;
909 fp = ecb_bswap32 (fp); 1125 fp = ecb_bswap32 (fp);
910 1126
911 return newSVnv (ecb_binary32_to_float (fp)); 1127 return newSVnv (ecb_binary32_to_float (fp));
912 } 1128 }
913 1129
914 case 27: 1130 case MISC_FLOAT64:
915 { 1131 {
916 uint64_t fp; 1132 uint64_t fp;
917 WANT (8); 1133 WANT (8);
918 memcpy (&fp, dec->cur, 8); 1134 memcpy (&fp, dec->cur, 8);
919 dec->cur += 8; 1135 dec->cur += 8;
922 fp = ecb_bswap64 (fp); 1138 fp = ecb_bswap64 (fp);
923 1139
924 return newSVnv (ecb_binary64_to_double (fp)); 1140 return newSVnv (ecb_binary64_to_double (fp));
925 } 1141 }
926 1142
927 // 0..19 unassigned 1143 // 0..19 unassigned simple
928 // 24 reserved + unassigned (reserved values are not encodable) 1144 // 24 reserved + unassigned simple (reserved values are not encodable)
1145 // 28-30 unassigned misc
1146 // 31 break code
929 default: 1147 default:
930 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1148 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
931 } 1149 }
932 1150
933 break; 1151 break;
934 } 1152 }
935 1153
962 if (dec.cur != dec.end && !dec.err) 1180 if (dec.cur != dec.end && !dec.err)
963 dec.err = "garbage after CBOR object"; 1181 dec.err = "garbage after CBOR object";
964 1182
965 if (dec.err) 1183 if (dec.err)
966 { 1184 {
1185 if (dec.shareable)
1186 {
1187 // need to break cyclic links, which whould all be in shareable
1188 int i;
1189 SV **svp;
1190
1191 for (i = av_len (dec.shareable) + 1; i--; )
1192 if ((svp = av_fetch (dec.shareable, i, 0)))
1193 sv_setsv (*svp, &PL_sv_undef);
1194 }
1195
967 SvREFCNT_dec (sv); 1196 SvREFCNT_dec (sv);
968 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); 1197 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
969 } 1198 }
970 1199
971 sv = sv_2mortal (sv); 1200 sv = sv_2mortal (sv);
972 1201
973 return sv; 1202 return sv;
974} 1203}
975 1204
1205/////////////////////////////////////////////////////////////////////////////
1206// incremental parser
1207
1208#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1209
1210// returns 0 for notyet, 1 for success or error
1211static int
1212incr_parse (CBOR *self, SV *cborstr)
1213{
1214 STRLEN cur;
1215 SvPV (cborstr, cur);
1216
1217 while (ecb_expect_true (self->incr_need <= cur))
1218 {
1219 // table of integer count bytes
1220 static I8 incr_len[MINOR_MASK + 1] = {
1221 0, 0, 0, 0, 0, 0, 0, 0,
1222 0, 0, 0, 0, 0, 0, 0, 0,
1223 0, 0, 0, 0, 0, 0, 0, 0,
1224 1, 2, 4, 8,-1,-1,-1,-2
1225 };
1226
1227 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1228 U8 m = *p & MINOR_MASK;
1229 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1230 I8 ilen = incr_len[m];
1231
1232 self->incr_need = self->incr_pos + 1;
1233
1234 if (ecb_expect_false (ilen < 0))
1235 {
1236 if (m != MINOR_INDEF)
1237 return 1; // error
1238
1239 if (*p == (MAJOR_MISC | MINOR_INDEF))
1240 {
1241 if (count >= 0)
1242 return 1; // error
1243
1244 count = 1;
1245 }
1246 else
1247 {
1248 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1249 count = -1;
1250 }
1251 }
1252 else
1253 {
1254 self->incr_need += ilen;
1255 if (ecb_expect_false (self->incr_need > cur))
1256 return 0;
1257
1258 int major = *p >> MAJOR_SHIFT;
1259
1260 switch (major)
1261 {
1262 case MAJOR_BYTES >> MAJOR_SHIFT:
1263 case MAJOR_TEXT >> MAJOR_SHIFT:
1264 case MAJOR_ARRAY >> MAJOR_SHIFT:
1265 case MAJOR_MAP >> MAJOR_SHIFT:
1266 {
1267 UV len;
1268
1269 if (ecb_expect_false (ilen))
1270 {
1271 len = 0;
1272
1273 do {
1274 len = (len << 8) | *++p;
1275 } while (--ilen);
1276 }
1277 else
1278 len = m;
1279
1280 switch (major)
1281 {
1282 case MAJOR_BYTES >> MAJOR_SHIFT:
1283 case MAJOR_TEXT >> MAJOR_SHIFT:
1284 self->incr_need += len;
1285 if (ecb_expect_false (self->incr_need > cur))
1286 return 0;
1287
1288 break;
1289
1290 case MAJOR_MAP >> MAJOR_SHIFT:
1291 len <<= 1;
1292 case MAJOR_ARRAY >> MAJOR_SHIFT:
1293 if (len)
1294 {
1295 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1296 count = len + 1;
1297 }
1298 break;
1299 }
1300 }
1301 }
1302 }
1303
1304 self->incr_pos = self->incr_need;
1305
1306 if (count > 0)
1307 {
1308 while (!--count)
1309 {
1310 if (!AvFILLp (self->incr_count))
1311 return 1; // done
1312
1313 SvREFCNT_dec_NN (av_pop (self->incr_count));
1314 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1315 }
1316
1317 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1318 }
1319 }
1320
1321 return 0;
1322}
1323
1324
976///////////////////////////////////////////////////////////////////////////// 1325/////////////////////////////////////////////////////////////////////////////
977// XS interface functions 1326// XS interface functions
978 1327
979MODULE = CBOR::XS PACKAGE = CBOR::XS 1328MODULE = CBOR::XS PACKAGE = CBOR::XS
980 1329
987 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1); 1336 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
988 1337
989 types_true = get_bool ("Types::Serialiser::true" ); 1338 types_true = get_bool ("Types::Serialiser::true" );
990 types_false = get_bool ("Types::Serialiser::false"); 1339 types_false = get_bool ("Types::Serialiser::false");
991 types_error = get_bool ("Types::Serialiser::error"); 1340 types_error = get_bool ("Types::Serialiser::error");
1341
1342 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
992 1343
993 sv_cbor = newSVpv ("CBOR", 0); 1344 sv_cbor = newSVpv ("CBOR", 0);
994 SvREADONLY_on (sv_cbor); 1345 SvREADONLY_on (sv_cbor);
995} 1346}
996 1347
1018void shrink (CBOR *self, int enable = 1) 1369void shrink (CBOR *self, int enable = 1)
1019 ALIAS: 1370 ALIAS:
1020 shrink = F_SHRINK 1371 shrink = F_SHRINK
1021 allow_unknown = F_ALLOW_UNKNOWN 1372 allow_unknown = F_ALLOW_UNKNOWN
1022 allow_sharing = F_ALLOW_SHARING 1373 allow_sharing = F_ALLOW_SHARING
1023 dedup_keys = F_DEDUP_KEYS 1374 allow_cycles = F_ALLOW_CYCLES
1024 dedup_strings = F_DEDUP_STRINGS 1375 pack_strings = F_PACK_STRINGS
1376 validate_utf8 = F_VALIDATE_UTF8
1025 PPCODE: 1377 PPCODE:
1026{ 1378{
1027 if (enable) 1379 if (enable)
1028 self->flags |= ix; 1380 self->flags |= ix;
1029 else 1381 else
1035void get_shrink (CBOR *self) 1387void get_shrink (CBOR *self)
1036 ALIAS: 1388 ALIAS:
1037 get_shrink = F_SHRINK 1389 get_shrink = F_SHRINK
1038 get_allow_unknown = F_ALLOW_UNKNOWN 1390 get_allow_unknown = F_ALLOW_UNKNOWN
1039 get_allow_sharing = F_ALLOW_SHARING 1391 get_allow_sharing = F_ALLOW_SHARING
1040 get_dedup_keys = F_DEDUP_KEYS 1392 get_allow_cycles = F_ALLOW_CYCLES
1041 get_dedup_strings = F_DEDUP_STRINGS 1393 get_pack_strings = F_PACK_STRINGS
1394 get_validate_utf8 = F_VALIDATE_UTF8
1042 PPCODE: 1395 PPCODE:
1043 XPUSHs (boolSV (self->flags & ix)); 1396 XPUSHs (boolSV (self->flags & ix));
1044 1397
1045void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1398void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1046 PPCODE: 1399 PPCODE:
1059 XPUSHs (ST (0)); 1412 XPUSHs (ST (0));
1060 1413
1061int get_max_size (CBOR *self) 1414int get_max_size (CBOR *self)
1062 CODE: 1415 CODE:
1063 RETVAL = self->max_size; 1416 RETVAL = self->max_size;
1417 OUTPUT:
1418 RETVAL
1419
1420void filter (CBOR *self, SV *filter = 0)
1421 PPCODE:
1422 SvREFCNT_dec (self->filter);
1423 self->filter = filter ? newSVsv (filter) : filter;
1424 XPUSHs (ST (0));
1425
1426SV *get_filter (CBOR *self)
1427 CODE:
1428 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
1064 OUTPUT: 1429 OUTPUT:
1065 RETVAL 1430 RETVAL
1066 1431
1067void encode (CBOR *self, SV *scalar) 1432void encode (CBOR *self, SV *scalar)
1068 PPCODE: 1433 PPCODE:
1083 EXTEND (SP, 2); 1448 EXTEND (SP, 2);
1084 PUSHs (sv); 1449 PUSHs (sv);
1085 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1450 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1086} 1451}
1087 1452
1453void incr_parse (CBOR *self, SV *cborstr)
1454 ALIAS:
1455 incr_parse_multiple = 1
1456 PPCODE:
1457{
1458 if (SvUTF8 (cborstr))
1459 sv_utf8_downgrade (cborstr, 0);
1460
1461 if (!self->incr_count)
1462 {
1463 self->incr_count = newAV ();
1464 self->incr_pos = 0;
1465 self->incr_need = 1;
1466
1467 av_push (self->incr_count, newSViv (1));
1468 }
1469
1470 do
1471 {
1472 if (!incr_parse (self, cborstr))
1473 {
1474 if (self->incr_need > self->max_size && self->max_size)
1475 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1476 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1477
1478 break;
1479 }
1480
1481 SV *sv;
1482 char *offset;
1483
1484 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1485 XPUSHs (sv);
1486
1487 sv_chop (cborstr, offset);
1488
1489 av_clear (self->incr_count);
1490 av_push (self->incr_count, newSViv (1));
1491
1492 self->incr_pos = 0;
1493 self->incr_need = self->incr_pos + 1;
1494 }
1495 while (ix);
1496}
1497
1498void incr_reset (CBOR *self)
1499 CODE:
1500{
1501 SvREFCNT_dec (self->incr_count);
1502 self->incr_count = 0;
1503}
1504
1505void DESTROY (CBOR *self)
1506 PPCODE:
1507 cbor_free (self);
1508
1088PROTOTYPES: ENABLE 1509PROTOTYPES: ENABLE
1089 1510
1090void encode_cbor (SV *scalar) 1511void encode_cbor (SV *scalar)
1512 ALIAS:
1513 encode_cbor = 0
1514 encode_cbor_sharing = F_ALLOW_SHARING
1091 PPCODE: 1515 PPCODE:
1092{ 1516{
1093 CBOR cbor; 1517 CBOR cbor;
1094 cbor_init (&cbor); 1518 cbor_init (&cbor);
1519 cbor.flags |= ix;
1095 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1520 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1096 XPUSHs (scalar); 1521 XPUSHs (scalar);
1097} 1522}
1098 1523
1099void decode_cbor (SV *cborstr) 1524void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines