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 1.13 by root, Mon Oct 28 22:50:50 2013 UTC vs.
Revision 1.47 by root, Mon Apr 27 20:21:53 2015 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"
14
15// compatibility with perl <5.18
16#ifndef HvNAMELEN_get
17# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
18#endif
19#ifndef HvNAMELEN
20# define HvNAMELEN(hv) HvNAMELEN_get (hv)
21#endif
22#ifndef HvNAMEUTF8
23# define HvNAMEUTF8(hv) 0
24#endif
25#ifndef SvREFCNT_dec_NN
26# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)
27#endif
28
29// known major and minor types
30enum cbor_type
31{
32 MAJOR_SHIFT = 5,
33 MINOR_MASK = 0x1f,
34
35 MAJOR_POS_INT = 0 << MAJOR_SHIFT,
36 MAJOR_NEG_INT = 1 << MAJOR_SHIFT,
37 MAJOR_BYTES = 2 << MAJOR_SHIFT,
38 MAJOR_TEXT = 3 << MAJOR_SHIFT,
39 MAJOR_ARRAY = 4 << MAJOR_SHIFT,
40 MAJOR_MAP = 5 << MAJOR_SHIFT,
41 MAJOR_TAG = 6 << MAJOR_SHIFT,
42 MAJOR_MISC = 7 << MAJOR_SHIFT,
43
44 // INT/STRING/ARRAY/MAP subtypes
45 LENGTH_EXT1 = 24,
46 LENGTH_EXT2 = 25,
47 LENGTH_EXT4 = 26,
48 LENGTH_EXT8 = 27,
49
50 // SIMPLE types (effectively MISC subtypes)
51 SIMPLE_FALSE = 20,
52 SIMPLE_TRUE = 21,
53 SIMPLE_NULL = 22,
54 SIMPLE_UNDEF = 23,
55
56 // MISC subtype (unused)
57 MISC_EXT1 = 24,
58 MISC_FLOAT16 = 25,
59 MISC_FLOAT32 = 26,
60 MISC_FLOAT64 = 27,
61
62 // BYTES/TEXT/ARRAY/MAP
63 MINOR_INDEF = 31,
64};
13 65
14// known tags 66// known tags
15enum cbor_tag 67enum cbor_tag
16{ 68{
17 // inofficial extensions (pending iana registration) 69 // extensions
18 CBOR_TAG_PERL_OBJECT = 256, 70 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
19 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
20 77
21 // rfc7049 78 // rfc7049
22 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 79 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
23 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any 80 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
24 CBOR_TAG_POS_BIGNUM = 2, // byte string 81 CBOR_TAG_POS_BIGNUM = 2, // byte string
25 CBOR_TAG_NEG_BIGNUM = 3, // byte string 82 CBOR_TAG_NEG_BIGNUM = 3, // byte string
26 CBOR_TAG_DECIMAL = 4, // decimal fraction, array 83 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
27 CBOR_TAG_BIGFLOAT = 5, // array 84 CBOR_TAG_BIGFLOAT = 5, // array
28 85
29 CBOR_TAG_CONV_B64U = 21, // base64url, any 86 CBOR_TAG_CONV_B64U = 21, // base64url, any
30 CBOR_TAG_CONV_B64 = 22, // base64, any 87 CBOR_TAG_CONV_B64 = 22, // base64, any
31 CBOR_TAG_CONV_HEX = 23, // base16, any 88 CBOR_TAG_CONV_HEX = 23, // base16, any
32 CBOR_TAG_CBOR = 24, // embedded cbor, byte string 89 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
33 90
34 CBOR_TAG_URI = 32, // URI rfc3986, utf-8 91 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
35 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8 92 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
36 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8 93 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
37 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8 94 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
38 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 95 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
39 96
40 CBOR_TAG_MAGIC = 55799 // self-describe cbor 97 CBOR_TAG_MAGIC = 55799, // self-describe cbor
41}; 98};
42 99
43#define F_SHRINK 0x00000200UL 100#define F_SHRINK 0x00000001UL
44#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
45 106
46#define INIT_SIZE 32 // initial scalar size to be allocated 107#define INIT_SIZE 32 // initial scalar size to be allocated
47 108
48#define SB do { 109#define SB do {
49#define SE } while (0) 110#define SE } while (0)
61# define CBOR_SLOW 0 122# define CBOR_SLOW 0
62# define CBOR_STASH cbor_stash 123# define CBOR_STASH cbor_stash
63#endif 124#endif
64 125
65static 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::
66static SV *types_true, *types_false, *types_error, *sv_cbor; 127static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
67 128
68typedef struct { 129typedef struct {
69 U32 flags; 130 U32 flags;
70 U32 max_depth; 131 U32 max_depth;
71 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.
72} CBOR; 139} CBOR;
73 140
74ecb_inline void 141ecb_inline void
75cbor_init (CBOR *cbor) 142cbor_init (CBOR *cbor)
76{ 143{
77 Zero (cbor, 1, CBOR); 144 Zero (cbor, 1, CBOR);
78 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);
79} 153}
80 154
81///////////////////////////////////////////////////////////////////////////// 155/////////////////////////////////////////////////////////////////////////////
82// utility functions 156// utility functions
83 157
105 SvPV_renew (sv, SvCUR (sv) + 1); 179 SvPV_renew (sv, SvCUR (sv) + 1);
106#endif 180#endif
107 } 181 }
108} 182}
109 183
110///////////////////////////////////////////////////////////////////////////// 184// minimum length of a string to be registered for stringref
111// fp hell 185ecb_inline int
112 186minimum_string_length (UV idx)
113//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}
114 198
115///////////////////////////////////////////////////////////////////////////// 199/////////////////////////////////////////////////////////////////////////////
116// encoder 200// encoder
117 201
118// structure used for encoding CBOR 202// structure used for encoding CBOR
121 char *cur; // SvPVX (sv) + current output position 205 char *cur; // SvPVX (sv) + current output position
122 char *end; // SvEND (sv) 206 char *end; // SvEND (sv)
123 SV *sv; // result scalar 207 SV *sv; // result scalar
124 CBOR cbor; 208 CBOR cbor;
125 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;
126} enc_t; 214} enc_t;
127 215
128ecb_inline void 216ecb_inline void
129need (enc_t *enc, STRLEN len) 217need (enc_t *enc, STRLEN len)
130{ 218{
147static void 235static void
148encode_uint (enc_t *enc, int major, UV len) 236encode_uint (enc_t *enc, int major, UV len)
149{ 237{
150 need (enc, 9); 238 need (enc, 9);
151 239
152 if (len < 24) 240 if (ecb_expect_true (len < LENGTH_EXT1))
153 *enc->cur++ = major | len; 241 *enc->cur++ = major | len;
154 else if (len <= 0xff) 242 else if (ecb_expect_true (len <= 0xffU))
155 { 243 {
156 *enc->cur++ = major | 24; 244 *enc->cur++ = major | LENGTH_EXT1;
157 *enc->cur++ = len; 245 *enc->cur++ = len;
158 } 246 }
159 else if (len <= 0xffff) 247 else if (len <= 0xffffU)
160 { 248 {
161 *enc->cur++ = major | 25; 249 *enc->cur++ = major | LENGTH_EXT2;
162 *enc->cur++ = len >> 8; 250 *enc->cur++ = len >> 8;
163 *enc->cur++ = len; 251 *enc->cur++ = len;
164 } 252 }
165 else if (len <= 0xffffffff) 253 else if (len <= 0xffffffffU)
166 { 254 {
167 *enc->cur++ = major | 26; 255 *enc->cur++ = major | LENGTH_EXT4;
168 *enc->cur++ = len >> 24; 256 *enc->cur++ = len >> 24;
169 *enc->cur++ = len >> 16; 257 *enc->cur++ = len >> 16;
170 *enc->cur++ = len >> 8; 258 *enc->cur++ = len >> 8;
171 *enc->cur++ = len; 259 *enc->cur++ = len;
172 } 260 }
173 else 261 else
174 { 262 {
175 *enc->cur++ = major | 27; 263 *enc->cur++ = major | LENGTH_EXT8;
176 *enc->cur++ = len >> 56; 264 *enc->cur++ = len >> 56;
177 *enc->cur++ = len >> 48; 265 *enc->cur++ = len >> 48;
178 *enc->cur++ = len >> 40; 266 *enc->cur++ = len >> 40;
179 *enc->cur++ = len >> 32; 267 *enc->cur++ = len >> 32;
180 *enc->cur++ = len >> 24; 268 *enc->cur++ = len >> 24;
182 *enc->cur++ = len >> 8; 270 *enc->cur++ = len >> 8;
183 *enc->cur++ = len; 271 *enc->cur++ = len;
184 } 272 }
185} 273}
186 274
187static void 275ecb_inline void
276encode_tag (enc_t *enc, UV tag)
277{
278 encode_uint (enc, MAJOR_TAG, tag);
279}
280
281ecb_inline void
188encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 282encode_str (enc_t *enc, int utf8, char *str, STRLEN len)
189{ 283{
190 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 284 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
191 need (enc, len); 285 need (enc, len);
192 memcpy (enc->cur, str, len); 286 memcpy (enc->cur, str, len);
193 enc->cur += len; 287 enc->cur += len;
194} 288}
195 289
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
196static void encode_sv (enc_t *enc, SV *sv); 315static void encode_sv (enc_t *enc, SV *sv);
197 316
198static void 317static void
199encode_av (enc_t *enc, AV *av) 318encode_av (enc_t *enc, AV *av)
200{ 319{
203 if (enc->depth >= enc->cbor.max_depth) 322 if (enc->depth >= enc->cbor.max_depth)
204 croak (ERR_NESTING_EXCEEDED); 323 croak (ERR_NESTING_EXCEEDED);
205 324
206 ++enc->depth; 325 ++enc->depth;
207 326
208 encode_uint (enc, 0x80, len + 1); 327 encode_uint (enc, MAJOR_ARRAY, len + 1);
209 328
329 if (SvMAGICAL (av))
210 for (i = 0; i <= len; ++i) 330 for (i = 0; i <= len; ++i)
211 { 331 {
212 SV **svp = av_fetch (av, i, 0); 332 SV **svp = av_fetch (av, i, 0);
213 encode_sv (enc, svp ? *svp : &PL_sv_undef); 333 encode_sv (enc, svp ? *svp : &PL_sv_undef);
214 } 334 }
335 else
336 for (i = 0; i <= len; ++i)
337 {
338 SV *sv = AvARRAY (av)[i];
339 encode_sv (enc, sv ? sv : &PL_sv_undef);
340 }
215 341
216 --enc->depth; 342 --enc->depth;
217} 343}
218 344
219static void 345static void
228 354
229 int pairs = hv_iterinit (hv); 355 int pairs = hv_iterinit (hv);
230 int mg = SvMAGICAL (hv); 356 int mg = SvMAGICAL (hv);
231 357
232 if (mg) 358 if (mg)
233 encode_ch (enc, 0xa0 | 31); 359 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
234 else 360 else
235 encode_uint (enc, 0xa0, pairs); 361 encode_uint (enc, MAJOR_MAP, pairs);
236 362
237 while ((he = hv_iternext (hv))) 363 while ((he = hv_iternext (hv)))
238 { 364 {
239 if (HeKLEN (he) == HEf_SVKEY) 365 if (HeKLEN (he) == HEf_SVKEY)
240 encode_sv (enc, HeSVKEY (he)); 366 encode_sv (enc, HeSVKEY (he));
241 else 367 else
242 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 368 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
243 369
244 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 370 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
245 } 371 }
246 372
247 if (mg) 373 if (mg)
248 encode_ch (enc, 0xe0 | 31); 374 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
249 375
250 --enc->depth; 376 --enc->depth;
251} 377}
252 378
253// encode objects, arrays and special \0=false and \1=true values. 379// encode objects, arrays and special \0=false and \1=true values.
254static void 380static void
255encode_rv (enc_t *enc, SV *sv) 381encode_rv (enc_t *enc, SV *sv)
256{ 382{
257 svtype svt;
258
259 SvGETMAGIC (sv); 383 SvGETMAGIC (sv);
384
260 svt = SvTYPE (sv); 385 svtype svt = SvTYPE (sv);
261 386
262 if (ecb_expect_false (SvOBJECT (sv))) 387 if (ecb_expect_false (SvOBJECT (sv)))
263 { 388 {
264 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash 389 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
265 ? types_boolean_stash 390 ? types_boolean_stash
270 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 395 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
271 ? cbor_tagged_stash 396 ? cbor_tagged_stash
272 : gv_stashpv ("CBOR::XS::Tagged" , 1); 397 : gv_stashpv ("CBOR::XS::Tagged" , 1);
273 398
274 HV *stash = SvSTASH (sv); 399 HV *stash = SvSTASH (sv);
275 GV *method;
276 400
277 if (stash == boolean_stash) 401 if (stash == boolean_stash)
278 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 402 {
403 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
404 return;
405 }
279 else if (stash == error_stash) 406 else if (stash == error_stash)
280 encode_ch (enc, 0xe0 | 23); 407 {
408 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
409 return;
410 }
281 else if (stash == tagged_stash) 411 else if (stash == tagged_stash)
282 { 412 {
283 if (svt != SVt_PVAV) 413 if (svt != SVt_PVAV)
284 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 414 croak ("encountered CBOR::XS::Tagged object that isn't an array");
285 415
286 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 416 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
287 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 417 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
418
419 return;
420 }
421 }
422
423 if (ecb_expect_false (SvREFCNT (sv) > 1)
424 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
425 {
426 if (!enc->shareable)
427 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
428
429 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
430
431 if (SvOK (*svp))
288 } 432 {
433 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
434 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
435 return;
436 }
437 else
438 {
439 sv_setuv (*svp, enc->shareable_idx);
440 ++enc->shareable_idx;
441 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
442 }
443 }
444
445 if (ecb_expect_false (SvOBJECT (sv)))
446 {
447 HV *stash = SvSTASH (sv);
448 GV *method;
449
289 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 450 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
290 { 451 {
291 dSP; 452 dSP;
292 453
293 ENTER; SAVETMPS; PUSHMARK (SP); 454 ENTER; SAVETMPS; PUSHMARK (SP);
294 // we re-bless the reference to get overload and other niceties right 455 // we re-bless the reference to get overload and other niceties right
325 486
326 // catch this surprisingly common error 487 // catch this surprisingly common error
327 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 488 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
328 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); 489 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
329 490
330 encode_uint (enc, 0xc0, CBOR_TAG_PERL_OBJECT); 491 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
331 encode_uint (enc, 0x80, count + 1); 492 encode_uint (enc, MAJOR_ARRAY, count + 1);
332 encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 493 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
333 494
334 while (count) 495 while (count)
335 encode_sv (enc, SP[1 - count--]); 496 encode_sv (enc, SP[1 - count--]);
336 497
337 PUTBACK; 498 PUTBACK;
344 } 505 }
345 else if (svt == SVt_PVHV) 506 else if (svt == SVt_PVHV)
346 encode_hv (enc, (HV *)sv); 507 encode_hv (enc, (HV *)sv);
347 else if (svt == SVt_PVAV) 508 else if (svt == SVt_PVAV)
348 encode_av (enc, (AV *)sv); 509 encode_av (enc, (AV *)sv);
349 else if (svt < SVt_PVAV)
350 {
351 STRLEN len = 0;
352 char *pv = svt ? SvPV (sv, len) : 0;
353
354 if (len == 1 && *pv == '1')
355 encode_ch (enc, 0xe0 | 21);
356 else if (len == 1 && *pv == '0')
357 encode_ch (enc, 0xe0 | 20);
358 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
359 encode_ch (enc, 0xe0 | 23);
360 else
361 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
362 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
363 }
364 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
365 encode_ch (enc, 0xe0 | 23);
366 else 510 else
367 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 511 {
368 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 512 encode_tag (enc, CBOR_TAG_INDIRECTION);
513 encode_sv (enc, sv);
514 }
369} 515}
370 516
371static void 517static void
372encode_nv (enc_t *enc, SV *sv) 518encode_nv (enc_t *enc, SV *sv)
373{ 519{
374 double nv = SvNVX (sv); 520 double nv = SvNVX (sv);
375 521
376 need (enc, 9); 522 need (enc, 9);
377 523
378 if (ecb_expect_false (nv == (U32)nv)) 524 if (ecb_expect_false (nv == (NV)(U32)nv))
379 encode_uint (enc, 0x00, (U32)nv); 525 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
380 //TODO: maybe I32? 526 //TODO: maybe I32?
381 else if (ecb_expect_false (nv == (float)nv)) 527 else if (ecb_expect_false (nv == (float)nv))
382 { 528 {
383 uint32_t fp = ecb_float_to_binary32 (nv); 529 uint32_t fp = ecb_float_to_binary32 (nv);
384 530
385 *enc->cur++ = 0xe0 | 26; 531 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
386 532
387 if (!ecb_big_endian ()) 533 if (!ecb_big_endian ())
388 fp = ecb_bswap32 (fp); 534 fp = ecb_bswap32 (fp);
389 535
390 memcpy (enc->cur, &fp, 4); 536 memcpy (enc->cur, &fp, 4);
392 } 538 }
393 else 539 else
394 { 540 {
395 uint64_t fp = ecb_double_to_binary64 (nv); 541 uint64_t fp = ecb_double_to_binary64 (nv);
396 542
397 *enc->cur++ = 0xe0 | 27; 543 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
398 544
399 if (!ecb_big_endian ()) 545 if (!ecb_big_endian ())
400 fp = ecb_bswap64 (fp); 546 fp = ecb_bswap64 (fp);
401 547
402 memcpy (enc->cur, &fp, 8); 548 memcpy (enc->cur, &fp, 8);
411 557
412 if (SvPOKp (sv)) 558 if (SvPOKp (sv))
413 { 559 {
414 STRLEN len; 560 STRLEN len;
415 char *str = SvPV (sv, len); 561 char *str = SvPV (sv, len);
416 encode_str (enc, SvUTF8 (sv), str, len); 562 encode_strref (enc, SvUTF8 (sv), str, len);
417 } 563 }
418 else if (SvNOKp (sv)) 564 else if (SvNOKp (sv))
419 encode_nv (enc, sv); 565 encode_nv (enc, sv);
420 else if (SvIOKp (sv)) 566 else if (SvIOKp (sv))
421 { 567 {
422 if (SvIsUV (sv)) 568 if (SvIsUV (sv))
423 encode_uint (enc, 0x00, SvUVX (sv)); 569 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
424 else if (SvIVX (sv) >= 0) 570 else if (SvIVX (sv) >= 0)
425 encode_uint (enc, 0x00, SvIVX (sv)); 571 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
426 else 572 else
427 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 573 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
428 } 574 }
429 else if (SvROK (sv)) 575 else if (SvROK (sv))
430 encode_rv (enc, SvRV (sv)); 576 encode_rv (enc, SvRV (sv));
431 else if (!SvOK (sv)) 577 else if (!SvOK (sv))
432 encode_ch (enc, 0xe0 | 22); 578 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
433 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 579 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
434 encode_ch (enc, 0xe0 | 23); 580 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
435 else 581 else
436 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 582 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
437 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 583 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
438} 584}
439 585
440static SV * 586static SV *
441encode_cbor (SV *scalar, CBOR *cbor) 587encode_cbor (SV *scalar, CBOR *cbor)
442{ 588{
443 enc_t enc; 589 enc_t enc = { };
444 590
445 enc.cbor = *cbor; 591 enc.cbor = *cbor;
446 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 592 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
447 enc.cur = SvPVX (enc.sv); 593 enc.cur = SvPVX (enc.sv);
448 enc.end = SvEND (enc.sv); 594 enc.end = SvEND (enc.sv);
449 enc.depth = 0;
450 595
451 SvPOK_only (enc.sv); 596 SvPOK_only (enc.sv);
597
598 if (cbor->flags & F_PACK_STRINGS)
599 {
600 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
601 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
602 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
603 }
604
452 encode_sv (&enc, scalar); 605 encode_sv (&enc, scalar);
453 606
454 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 607 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
455 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 608 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
456 609
470 U8 *end; // end of input string 623 U8 *end; // end of input string
471 const char *err; // parse error, if != 0 624 const char *err; // parse error, if != 0
472 CBOR cbor; 625 CBOR cbor;
473 U32 depth; // recursion depth 626 U32 depth; // recursion depth
474 U32 maxdepth; // recursion depth limit 627 U32 maxdepth; // recursion depth limit
628 AV *shareable;
629 AV *stringref;
630 SV *decode_tagged;
475} dec_t; 631} dec_t;
476 632
477#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 633#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE
478 634
479#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 635#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data")
482#define DEC_DEC_DEPTH --dec->depth 638#define DEC_DEC_DEPTH --dec->depth
483 639
484static UV 640static UV
485decode_uint (dec_t *dec) 641decode_uint (dec_t *dec)
486{ 642{
487 switch (*dec->cur & 31) 643 U8 m = *dec->cur & MINOR_MASK;
488 { 644 ++dec->cur;
489 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
490 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
491 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
492 return *dec->cur++ & 31;
493 645
494 case 24: 646 if (ecb_expect_true (m < LENGTH_EXT1))
647 return m;
648 else if (ecb_expect_true (m == LENGTH_EXT1))
649 {
495 WANT (2); 650 WANT (1);
496 dec->cur += 2; 651 dec->cur += 1;
497 return dec->cur[-1]; 652 return dec->cur[-1];
498 653 }
499 case 25: 654 else if (ecb_expect_true (m == LENGTH_EXT2))
655 {
500 WANT (3); 656 WANT (2);
501 dec->cur += 3; 657 dec->cur += 2;
502 return (((UV)dec->cur[-2]) << 8) 658 return (((UV)dec->cur[-2]) << 8)
503 | ((UV)dec->cur[-1]); 659 | ((UV)dec->cur[-1]);
504 660 }
505 case 26: 661 else if (ecb_expect_true (m == LENGTH_EXT4))
662 {
506 WANT (5); 663 WANT (4);
507 dec->cur += 5; 664 dec->cur += 4;
508 return (((UV)dec->cur[-4]) << 24) 665 return (((UV)dec->cur[-4]) << 24)
509 | (((UV)dec->cur[-3]) << 16) 666 | (((UV)dec->cur[-3]) << 16)
510 | (((UV)dec->cur[-2]) << 8) 667 | (((UV)dec->cur[-2]) << 8)
511 | ((UV)dec->cur[-1]); 668 | ((UV)dec->cur[-1]);
512 669 }
513 case 27: 670 else if (ecb_expect_true (m == LENGTH_EXT8))
671 {
514 WANT (9); 672 WANT (8);
515 dec->cur += 9; 673 dec->cur += 8;
674
675 return
676#if UVSIZE < 8
677 0
678#else
516 return (((UV)dec->cur[-8]) << 56) 679 (((UV)dec->cur[-8]) << 56)
517 | (((UV)dec->cur[-7]) << 48) 680 | (((UV)dec->cur[-7]) << 48)
518 | (((UV)dec->cur[-6]) << 40) 681 | (((UV)dec->cur[-6]) << 40)
519 | (((UV)dec->cur[-5]) << 32) 682 | (((UV)dec->cur[-5]) << 32)
683#endif
520 | (((UV)dec->cur[-4]) << 24) 684 | (((UV)dec->cur[-4]) << 24)
521 | (((UV)dec->cur[-3]) << 16) 685 | (((UV)dec->cur[-3]) << 16)
522 | (((UV)dec->cur[-2]) << 8) 686 | (((UV)dec->cur[-2]) << 8)
523 | ((UV)dec->cur[-1]); 687 | ((UV)dec->cur[-1]);
524 688 }
525 default: 689 else
526 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 690 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
527 }
528 691
529fail: 692fail:
530 return 0; 693 return 0;
531} 694}
532 695
537{ 700{
538 AV *av = newAV (); 701 AV *av = newAV ();
539 702
540 DEC_INC_DEPTH; 703 DEC_INC_DEPTH;
541 704
542 if ((*dec->cur & 31) == 31) 705 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
543 { 706 {
544 ++dec->cur; 707 ++dec->cur;
545 708
546 for (;;) 709 for (;;)
547 { 710 {
548 WANT (1); 711 WANT (1);
549 712
550 if (*dec->cur == (0xe0 | 31)) 713 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
551 { 714 {
552 ++dec->cur; 715 ++dec->cur;
553 break; 716 break;
554 } 717 }
555 718
558 } 721 }
559 else 722 else
560 { 723 {
561 int i, len = decode_uint (dec); 724 int i, len = decode_uint (dec);
562 725
726 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
563 av_fill (av, len - 1); 727 av_fill (av, len - 1);
564 728
565 for (i = 0; i < len; ++i) 729 for (i = 0; i < len; ++i)
566 AvARRAY (av)[i] = decode_sv (dec); 730 AvARRAY (av)[i] = decode_sv (dec);
567 } 731 }
573 SvREFCNT_dec (av); 737 SvREFCNT_dec (av);
574 DEC_DEC_DEPTH; 738 DEC_DEC_DEPTH;
575 return &PL_sv_undef; 739 return &PL_sv_undef;
576} 740}
577 741
742static void
743decode_he (dec_t *dec, HV *hv)
744{
745 // for speed reasons, we specialcase single-string
746 // byte or utf-8 strings as keys, but only when !stringref
747
748 if (ecb_expect_true (!dec->stringref))
749 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
750 {
751 I32 len = decode_uint (dec);
752 char *key = (char *)dec->cur;
753
754 dec->cur += len;
755
756 hv_store (hv, key, len, decode_sv (dec), 0);
757
758 return;
759 }
760 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
761 {
762 I32 len = decode_uint (dec);
763 char *key = (char *)dec->cur;
764
765 dec->cur += len;
766
767 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
768 if (!is_utf8_string (key, len))
769 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
770
771 hv_store (hv, key, -len, decode_sv (dec), 0);
772
773 return;
774 }
775
776 SV *k = decode_sv (dec);
777 SV *v = decode_sv (dec);
778
779 hv_store_ent (hv, k, v, 0);
780 SvREFCNT_dec (k);
781
782fail:
783 ;
784}
785
578static SV * 786static SV *
579decode_hv (dec_t *dec) 787decode_hv (dec_t *dec)
580{ 788{
581 HV *hv = newHV (); 789 HV *hv = newHV ();
582 790
583 DEC_INC_DEPTH; 791 DEC_INC_DEPTH;
584 792
585 if ((*dec->cur & 31) == 31) 793 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
586 { 794 {
587 ++dec->cur; 795 ++dec->cur;
588 796
589 for (;;) 797 for (;;)
590 { 798 {
591 WANT (1); 799 WANT (1);
592 800
593 if (*dec->cur == (0xe0 | 31)) 801 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
594 { 802 {
595 ++dec->cur; 803 ++dec->cur;
596 break; 804 break;
597 } 805 }
598 806
599 SV *k = decode_sv (dec); 807 decode_he (dec, hv);
600 SV *v = decode_sv (dec);
601
602 hv_store_ent (hv, k, v, 0);
603 SvREFCNT_dec (k);
604 } 808 }
605 } 809 }
606 else 810 else
607 { 811 {
608 int len = decode_uint (dec); 812 int pairs = decode_uint (dec);
609 813
610 while (len--) 814 while (pairs--)
611 { 815 decode_he (dec, hv);
612 SV *k = decode_sv (dec);
613 SV *v = decode_sv (dec);
614
615 hv_store_ent (hv, k, v, 0);
616 SvREFCNT_dec (k);
617 }
618 } 816 }
619 817
620 DEC_DEC_DEPTH; 818 DEC_DEC_DEPTH;
621 return newRV_noinc ((SV *)hv); 819 return newRV_noinc ((SV *)hv);
622 820
629static SV * 827static SV *
630decode_str (dec_t *dec, int utf8) 828decode_str (dec_t *dec, int utf8)
631{ 829{
632 SV *sv = 0; 830 SV *sv = 0;
633 831
634 if ((*dec->cur & 31) == 31) 832 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
635 { 833 {
834 // indefinite length strings
636 ++dec->cur; 835 ++dec->cur;
637 836
837 U8 major = *dec->cur & MAJOR_MISC;
838
638 sv = newSVpvn ("", 0); 839 sv = newSVpvn ("", 0);
639 840
640 // not very fast, and certainly not robust against illegal input
641 for (;;) 841 for (;;)
642 { 842 {
643 WANT (1); 843 WANT (1);
644 844
645 if (*dec->cur == (0xe0 | 31)) 845 if ((*dec->cur - major) > LENGTH_EXT8)
846 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
646 { 847 {
647 ++dec->cur; 848 ++dec->cur;
648 break; 849 break;
649 } 850 }
851 else
852 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
650 853
651 sv_catsv (sv, decode_sv (dec)); 854 STRLEN len = decode_uint (dec);
855
856 WANT (len);
857 sv_catpvn (sv, dec->cur, len);
858 dec->cur += len;
652 } 859 }
653 } 860 }
654 else 861 else
655 { 862 {
656 STRLEN len = decode_uint (dec); 863 STRLEN len = decode_uint (dec);
657 864
658 WANT (len); 865 WANT (len);
659 sv = newSVpvn (dec->cur, len); 866 sv = newSVpvn (dec->cur, len);
660 dec->cur += len; 867 dec->cur += len;
868
869 if (ecb_expect_false (dec->stringref)
870 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
871 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
661 } 872 }
662 873
663 if (utf8) 874 if (utf8)
875 {
876 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
877 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
878 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
879
664 SvUTF8_on (sv); 880 SvUTF8_on (sv);
881 }
665 882
666 return sv; 883 return sv;
667 884
668fail: 885fail:
669 SvREFCNT_dec (sv); 886 SvREFCNT_dec (sv);
671} 888}
672 889
673static SV * 890static SV *
674decode_tagged (dec_t *dec) 891decode_tagged (dec_t *dec)
675{ 892{
893 SV *sv = 0;
676 UV tag = decode_uint (dec); 894 UV tag = decode_uint (dec);
895
896 WANT (1);
897
898 switch (tag)
899 {
900 case CBOR_TAG_MAGIC:
677 SV *sv = decode_sv (dec); 901 sv = decode_sv (dec);
902 break;
678 903
679 if (tag == CBOR_TAG_MAGIC) 904 case CBOR_TAG_INDIRECTION:
680 return sv; 905 sv = newRV_noinc (decode_sv (dec));
681 else if (tag == CBOR_TAG_PERL_OBJECT) 906 break;
682 { 907
908 case CBOR_TAG_STRINGREF_NAMESPACE:
909 {
910 ENTER; SAVETMPS;
911
912 SAVESPTR (dec->stringref);
913 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
914
915 sv = decode_sv (dec);
916
917 FREETMPS; LEAVE;
918 }
919 break;
920
921 case CBOR_TAG_STRINGREF:
922 {
923 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
924 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
925
926 UV idx = decode_uint (dec);
927
928 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
929 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
930
931 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
932 }
933 break;
934
935 case CBOR_TAG_VALUE_SHAREABLE:
936 {
937 if (ecb_expect_false (!dec->shareable))
938 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
939
940 if (dec->cbor.flags & F_ALLOW_CYCLES)
941 {
942 sv = newSV (0);
943 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
944
945 SV *osv = decode_sv (dec);
946 sv_setsv (sv, osv);
947 SvREFCNT_dec_NN (osv);
948 }
949 else
950 {
951 av_push (dec->shareable, &PL_sv_undef);
952 int idx = AvFILLp (dec->shareable);
953 sv = decode_sv (dec);
954 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
955 }
956 }
957 break;
958
959 case CBOR_TAG_VALUE_SHAREDREF:
960 {
961 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
962 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
963
964 UV idx = decode_uint (dec);
965
966 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
967 ERR ("corrupted CBOR data (sharedref index out of bounds)");
968
969 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
970
971 if (sv == &PL_sv_undef)
972 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
973 }
974 break;
975
976 case CBOR_TAG_PERL_OBJECT:
977 {
978 sv = decode_sv (dec);
979
683 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 980 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
684 ERR ("corrupted CBOR data (non-array perl object)"); 981 ERR ("corrupted CBOR data (non-array perl object)");
685 982
686 AV *av = (AV *)SvRV (sv); 983 AV *av = (AV *)SvRV (sv);
687 int len = av_len (av) + 1; 984 int len = av_len (av) + 1;
688 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0); 985 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
689 986
690 if (!stash) 987 if (!stash)
691 ERR ("cannot decode perl-object (package does not exist)"); 988 ERR ("cannot decode perl-object (package does not exist)");
692 989
693 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0); 990 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
694 991
695 if (!method) 992 if (!method)
696 ERR ("cannot decode perl-object (package does not have a THAW method)"); 993 ERR ("cannot decode perl-object (package does not have a THAW method)");
697 994
698 dSP; 995 dSP;
699 996
700 ENTER; SAVETMPS; PUSHMARK (SP); 997 ENTER; SAVETMPS; PUSHMARK (SP);
701 EXTEND (SP, len + 1); 998 EXTEND (SP, len + 1);
702 // we re-bless the reference to get overload and other niceties right 999 // we re-bless the reference to get overload and other niceties right
703 PUSHs (*av_fetch (av, 0, 1)); 1000 PUSHs (*av_fetch (av, 0, 1));
704 PUSHs (sv_cbor); 1001 PUSHs (sv_cbor);
705 1002
706 int i; 1003 int i;
707 1004
708 for (i = 1; i < len; ++i) 1005 for (i = 1; i < len; ++i)
709 PUSHs (*av_fetch (av, i, 1)); 1006 PUSHs (*av_fetch (av, i, 1));
710 1007
711 PUTBACK; 1008 PUTBACK;
712 call_sv ((SV *)GvCV (method), G_SCALAR); 1009 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
713 SPAGAIN; 1010 SPAGAIN;
714 1011
1012 if (SvTRUE (ERRSV))
1013 {
1014 FREETMPS; LEAVE;
1015 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1016 }
1017
715 SvREFCNT_dec (sv); 1018 SvREFCNT_dec (sv);
716 sv = SvREFCNT_inc (POPs); 1019 sv = SvREFCNT_inc (POPs);
717 1020
718 PUTBACK; 1021 PUTBACK;
719 1022
720 FREETMPS; LEAVE; 1023 FREETMPS; LEAVE;
1024 }
1025 break;
721 1026
722 return sv; 1027 default:
723 } 1028 {
724 else 1029 sv = decode_sv (dec);
725 { 1030
1031 dSP;
1032 ENTER; SAVETMPS; PUSHMARK (SP);
1033 EXTEND (SP, 2);
1034 PUSHs (newSVuv (tag));
1035 PUSHs (sv);
1036
1037 PUTBACK;
1038 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1039 SPAGAIN;
1040
1041 if (SvTRUE (ERRSV))
1042 {
1043 FREETMPS; LEAVE;
1044 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1045 }
1046
1047 if (count)
1048 {
1049 SvREFCNT_dec (sv);
1050 sv = SvREFCNT_inc (POPs);
1051 }
1052 else
1053 {
726 AV *av = newAV (); 1054 AV *av = newAV ();
727 av_push (av, newSVuv (tag)); 1055 av_push (av, newSVuv (tag));
728 av_push (av, sv); 1056 av_push (av, sv);
729 1057
730 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1058 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
731 ? cbor_tagged_stash 1059 ? cbor_tagged_stash
732 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1060 : gv_stashpv ("CBOR::XS::Tagged" , 1);
733
734 return sv_bless (newRV_noinc ((SV *)av), tagged_stash); 1061 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
1062 }
1063
1064 PUTBACK;
1065
1066 FREETMPS; LEAVE;
1067 }
1068 break;
735 } 1069 }
1070
1071 return sv;
736 1072
737fail: 1073fail:
738 SvREFCNT_dec (sv); 1074 SvREFCNT_dec (sv);
739 return &PL_sv_undef; 1075 return &PL_sv_undef;
740} 1076}
742static SV * 1078static SV *
743decode_sv (dec_t *dec) 1079decode_sv (dec_t *dec)
744{ 1080{
745 WANT (1); 1081 WANT (1);
746 1082
747 switch (*dec->cur >> 5) 1083 switch (*dec->cur >> MAJOR_SHIFT)
748 { 1084 {
749 case 0: // unsigned int 1085 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
750 return newSVuv (decode_uint (dec)); 1086 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
751 case 1: // negative int 1087 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
752 return newSViv (-1 - (IV)decode_uint (dec)); 1088 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
753 case 2: // octet string 1089 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
754 return decode_str (dec, 0); 1090 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
755 case 3: // utf-8 string 1091 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
756 return decode_str (dec, 1); 1092
757 case 4: // array 1093 case MAJOR_MISC >> MAJOR_SHIFT:
758 return decode_av (dec);
759 case 5: // map
760 return decode_hv (dec);
761 case 6: // tag
762 return decode_tagged (dec);
763 case 7: // misc
764 switch (*dec->cur++ & 31) 1094 switch (*dec->cur++ & MINOR_MASK)
765 { 1095 {
766 case 20: 1096 case SIMPLE_FALSE:
767#if CBOR_SLOW 1097#if CBOR_SLOW
768 types_false = get_bool ("Types::Serialiser::false"); 1098 types_false = get_bool ("Types::Serialiser::false");
769#endif 1099#endif
770 return newSVsv (types_false); 1100 return newSVsv (types_false);
771 case 21: 1101 case SIMPLE_TRUE:
772#if CBOR_SLOW 1102#if CBOR_SLOW
773 types_true = get_bool ("Types::Serialiser::true"); 1103 types_true = get_bool ("Types::Serialiser::true");
774#endif 1104#endif
775 return newSVsv (types_true); 1105 return newSVsv (types_true);
776 case 22: 1106 case SIMPLE_NULL:
777 return newSVsv (&PL_sv_undef); 1107 return newSVsv (&PL_sv_undef);
778 case 23: 1108 case SIMPLE_UNDEF:
779#if CBOR_SLOW 1109#if CBOR_SLOW
780 types_error = get_bool ("Types::Serialiser::error"); 1110 types_error = get_bool ("Types::Serialiser::error");
781#endif 1111#endif
782 return newSVsv (types_error); 1112 return newSVsv (types_error);
783 1113
784 case 25: 1114 case MISC_FLOAT16:
785 { 1115 {
786 WANT (2); 1116 WANT (2);
787 1117
788 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1118 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
789 dec->cur += 2; 1119 dec->cur += 2;
790 1120
791 return newSVnv (ecb_binary16_to_float (fp)); 1121 return newSVnv (ecb_binary16_to_float (fp));
792 } 1122 }
793 1123
794 case 26: 1124 case MISC_FLOAT32:
795 { 1125 {
796 uint32_t fp; 1126 uint32_t fp;
797 WANT (4); 1127 WANT (4);
798 memcpy (&fp, dec->cur, 4); 1128 memcpy (&fp, dec->cur, 4);
799 dec->cur += 4; 1129 dec->cur += 4;
802 fp = ecb_bswap32 (fp); 1132 fp = ecb_bswap32 (fp);
803 1133
804 return newSVnv (ecb_binary32_to_float (fp)); 1134 return newSVnv (ecb_binary32_to_float (fp));
805 } 1135 }
806 1136
807 case 27: 1137 case MISC_FLOAT64:
808 { 1138 {
809 uint64_t fp; 1139 uint64_t fp;
810 WANT (8); 1140 WANT (8);
811 memcpy (&fp, dec->cur, 8); 1141 memcpy (&fp, dec->cur, 8);
812 dec->cur += 8; 1142 dec->cur += 8;
815 fp = ecb_bswap64 (fp); 1145 fp = ecb_bswap64 (fp);
816 1146
817 return newSVnv (ecb_binary64_to_double (fp)); 1147 return newSVnv (ecb_binary64_to_double (fp));
818 } 1148 }
819 1149
820 // 0..19 unassigned 1150 // 0..19 unassigned simple
821 // 24 reserved + unassigned (reserved values are not encodable) 1151 // 24 reserved + unassigned simple (reserved values are not encodable)
1152 // 28-30 unassigned misc
1153 // 31 break code
822 default: 1154 default:
823 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1155 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
824 } 1156 }
825 1157
826 break; 1158 break;
827 } 1159 }
828 1160
831} 1163}
832 1164
833static SV * 1165static SV *
834decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1166decode_cbor (SV *string, CBOR *cbor, char **offset_return)
835{ 1167{
836 dec_t dec; 1168 dec_t dec = { };
837 SV *sv; 1169 SV *sv;
1170 STRLEN len;
1171 char *data = SvPVbyte (string, len);
838 1172
839 /* work around bugs in 5.10 where manipulating magic values
840 * makes perl ignore the magic in subsequent accesses.
841 * also make a copy of non-PV values, to get them into a clean
842 * state (SvPV should do that, but it's buggy, see below).
843 */
844 /*SvGETMAGIC (string);*/
845 if (SvMAGICAL (string) || !SvPOK (string))
846 string = sv_2mortal (newSVsv (string));
847
848 SvUPGRADE (string, SVt_PV);
849
850 /* work around a bug in perl 5.10, which causes SvCUR to fail an
851 * assertion with -DDEBUGGING, although SvCUR is documented to
852 * return the xpv_cur field which certainly exists after upgrading.
853 * according to nicholas clark, calling SvPOK fixes this.
854 * But it doesn't fix it, so try another workaround, call SvPV_nolen
855 * and hope for the best.
856 * Damnit, SvPV_nolen still trips over yet another assertion. This
857 * assertion business is seriously broken, try yet another workaround
858 * for the broken -DDEBUGGING.
859 */
860 {
861#ifdef DEBUGGING
862 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
863#else
864 STRLEN offset = SvCUR (string);
865#endif
866
867 if (offset > cbor->max_size && cbor->max_size) 1173 if (len > cbor->max_size && cbor->max_size)
868 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu", 1174 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
869 (unsigned long)SvCUR (string), (unsigned long)cbor->max_size); 1175 (unsigned long)len, (unsigned long)cbor->max_size);
870 }
871
872 sv_utf8_downgrade (string, 0);
873 1176
874 dec.cbor = *cbor; 1177 dec.cbor = *cbor;
875 dec.cur = (U8 *)SvPVX (string); 1178 dec.cur = (U8 *)data;
876 dec.end = (U8 *)SvEND (string); 1179 dec.end = (U8 *)data + len;
877 dec.err = 0;
878 dec.depth = 0;
879 1180
880 sv = decode_sv (&dec); 1181 sv = decode_sv (&dec);
881 1182
882 if (offset_return) 1183 if (offset_return)
883 *offset_return = dec.cur; 1184 *offset_return = dec.cur;
886 if (dec.cur != dec.end && !dec.err) 1187 if (dec.cur != dec.end && !dec.err)
887 dec.err = "garbage after CBOR object"; 1188 dec.err = "garbage after CBOR object";
888 1189
889 if (dec.err) 1190 if (dec.err)
890 { 1191 {
1192 if (dec.shareable)
1193 {
1194 // need to break cyclic links, which whould all be in shareable
1195 int i;
1196 SV **svp;
1197
1198 for (i = av_len (dec.shareable) + 1; i--; )
1199 if ((svp = av_fetch (dec.shareable, i, 0)))
1200 sv_setsv (*svp, &PL_sv_undef);
1201 }
1202
891 SvREFCNT_dec (sv); 1203 SvREFCNT_dec (sv);
892 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)SvPVX (string), (int)(uint8_t)*dec.cur); 1204 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
893 } 1205 }
894 1206
895 sv = sv_2mortal (sv); 1207 sv = sv_2mortal (sv);
896 1208
897 return sv; 1209 return sv;
898} 1210}
899 1211
1212/////////////////////////////////////////////////////////////////////////////
1213// incremental parser
1214
1215#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1216
1217// returns 0 for notyet, 1 for success or error
1218static int
1219incr_parse (CBOR *self, SV *cborstr)
1220{
1221 STRLEN cur;
1222 SvPV (cborstr, cur);
1223
1224 while (ecb_expect_true (self->incr_need <= cur))
1225 {
1226 // table of integer count bytes
1227 static I8 incr_len[MINOR_MASK + 1] = {
1228 0, 0, 0, 0, 0, 0, 0, 0,
1229 0, 0, 0, 0, 0, 0, 0, 0,
1230 0, 0, 0, 0, 0, 0, 0, 0,
1231 1, 2, 4, 8,-1,-1,-1,-2
1232 };
1233
1234 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1235 U8 m = *p & MINOR_MASK;
1236 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1237 I8 ilen = incr_len[m];
1238
1239 self->incr_need = self->incr_pos + 1;
1240
1241 if (ecb_expect_false (ilen < 0))
1242 {
1243 if (m != MINOR_INDEF)
1244 return 1; // error
1245
1246 if (*p == (MAJOR_MISC | MINOR_INDEF))
1247 {
1248 if (count >= 0)
1249 return 1; // error
1250
1251 count = 1;
1252 }
1253 else
1254 {
1255 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1256 count = -1;
1257 }
1258 }
1259 else
1260 {
1261 self->incr_need += ilen;
1262 if (ecb_expect_false (self->incr_need > cur))
1263 return 0;
1264
1265 int major = *p >> MAJOR_SHIFT;
1266
1267 switch (major)
1268 {
1269 case MAJOR_TAG >> MAJOR_SHIFT:
1270 ++count; // tags merely prefix another value
1271 break;
1272
1273 case MAJOR_BYTES >> MAJOR_SHIFT:
1274 case MAJOR_TEXT >> MAJOR_SHIFT:
1275 case MAJOR_ARRAY >> MAJOR_SHIFT:
1276 case MAJOR_MAP >> MAJOR_SHIFT:
1277 {
1278 UV len;
1279
1280 if (ecb_expect_false (ilen))
1281 {
1282 len = 0;
1283
1284 do {
1285 len = (len << 8) | *++p;
1286 } while (--ilen);
1287 }
1288 else
1289 len = m;
1290
1291 switch (major)
1292 {
1293 case MAJOR_BYTES >> MAJOR_SHIFT:
1294 case MAJOR_TEXT >> MAJOR_SHIFT:
1295 self->incr_need += len;
1296 if (ecb_expect_false (self->incr_need > cur))
1297 return 0;
1298
1299 break;
1300
1301 case MAJOR_MAP >> MAJOR_SHIFT:
1302 len <<= 1;
1303 case MAJOR_ARRAY >> MAJOR_SHIFT:
1304 if (len)
1305 {
1306 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1307 count = len + 1;
1308 }
1309 break;
1310 }
1311 }
1312 }
1313 }
1314
1315 self->incr_pos = self->incr_need;
1316
1317 if (count > 0)
1318 {
1319 while (!--count)
1320 {
1321 if (!AvFILLp (self->incr_count))
1322 return 1; // done
1323
1324 SvREFCNT_dec_NN (av_pop (self->incr_count));
1325 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1326 }
1327
1328 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1329 }
1330 }
1331
1332 return 0;
1333}
1334
1335
900///////////////////////////////////////////////////////////////////////////// 1336/////////////////////////////////////////////////////////////////////////////
901// XS interface functions 1337// XS interface functions
902 1338
903MODULE = CBOR::XS PACKAGE = CBOR::XS 1339MODULE = CBOR::XS PACKAGE = CBOR::XS
904 1340
911 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1); 1347 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
912 1348
913 types_true = get_bool ("Types::Serialiser::true" ); 1349 types_true = get_bool ("Types::Serialiser::true" );
914 types_false = get_bool ("Types::Serialiser::false"); 1350 types_false = get_bool ("Types::Serialiser::false");
915 types_error = get_bool ("Types::Serialiser::error"); 1351 types_error = get_bool ("Types::Serialiser::error");
1352
1353 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
916 1354
917 sv_cbor = newSVpv ("CBOR", 0); 1355 sv_cbor = newSVpv ("CBOR", 0);
918 SvREADONLY_on (sv_cbor); 1356 SvREADONLY_on (sv_cbor);
919} 1357}
920 1358
941 1379
942void shrink (CBOR *self, int enable = 1) 1380void shrink (CBOR *self, int enable = 1)
943 ALIAS: 1381 ALIAS:
944 shrink = F_SHRINK 1382 shrink = F_SHRINK
945 allow_unknown = F_ALLOW_UNKNOWN 1383 allow_unknown = F_ALLOW_UNKNOWN
1384 allow_sharing = F_ALLOW_SHARING
1385 allow_cycles = F_ALLOW_CYCLES
1386 pack_strings = F_PACK_STRINGS
1387 validate_utf8 = F_VALIDATE_UTF8
946 PPCODE: 1388 PPCODE:
947{ 1389{
948 if (enable) 1390 if (enable)
949 self->flags |= ix; 1391 self->flags |= ix;
950 else 1392 else
955 1397
956void get_shrink (CBOR *self) 1398void get_shrink (CBOR *self)
957 ALIAS: 1399 ALIAS:
958 get_shrink = F_SHRINK 1400 get_shrink = F_SHRINK
959 get_allow_unknown = F_ALLOW_UNKNOWN 1401 get_allow_unknown = F_ALLOW_UNKNOWN
1402 get_allow_sharing = F_ALLOW_SHARING
1403 get_allow_cycles = F_ALLOW_CYCLES
1404 get_pack_strings = F_PACK_STRINGS
1405 get_validate_utf8 = F_VALIDATE_UTF8
960 PPCODE: 1406 PPCODE:
961 XPUSHs (boolSV (self->flags & ix)); 1407 XPUSHs (boolSV (self->flags & ix));
962 1408
963void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1409void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
964 PPCODE: 1410 PPCODE:
980 CODE: 1426 CODE:
981 RETVAL = self->max_size; 1427 RETVAL = self->max_size;
982 OUTPUT: 1428 OUTPUT:
983 RETVAL 1429 RETVAL
984 1430
985#if 0 //TODO 1431void filter (CBOR *self, SV *filter = 0)
986
987void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef)
988 PPCODE: 1432 PPCODE:
989{
990 SvREFCNT_dec (self->cb_object); 1433 SvREFCNT_dec (self->filter);
991 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; 1434 self->filter = filter ? newSVsv (filter) : filter;
992
993 XPUSHs (ST (0)); 1435 XPUSHs (ST (0));
994}
995 1436
996void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef) 1437SV *get_filter (CBOR *self)
997 PPCODE: 1438 CODE:
998{ 1439 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
999 if (!self->cb_sk_object) 1440 OUTPUT:
1000 self->cb_sk_object = newHV (); 1441 RETVAL
1001
1002 if (SvOK (cb))
1003 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1004 else
1005 {
1006 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1007
1008 if (!HvKEYS (self->cb_sk_object))
1009 {
1010 SvREFCNT_dec (self->cb_sk_object);
1011 self->cb_sk_object = 0;
1012 }
1013 }
1014
1015 XPUSHs (ST (0));
1016}
1017
1018#endif
1019 1442
1020void encode (CBOR *self, SV *scalar) 1443void encode (CBOR *self, SV *scalar)
1021 PPCODE: 1444 PPCODE:
1022 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; 1445 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
1023 XPUSHs (scalar); 1446 XPUSHs (scalar);
1036 EXTEND (SP, 2); 1459 EXTEND (SP, 2);
1037 PUSHs (sv); 1460 PUSHs (sv);
1038 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1461 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1039} 1462}
1040 1463
1041#if 0 1464void incr_parse (CBOR *self, SV *cborstr)
1465 ALIAS:
1466 incr_parse_multiple = 1
1467 PPCODE:
1468{
1469 if (SvUTF8 (cborstr))
1470 sv_utf8_downgrade (cborstr, 0);
1471
1472 if (!self->incr_count)
1473 {
1474 self->incr_count = newAV ();
1475 self->incr_pos = 0;
1476 self->incr_need = 1;
1477
1478 av_push (self->incr_count, newSViv (1));
1479 }
1480
1481 do
1482 {
1483 if (!incr_parse (self, cborstr))
1484 {
1485 if (self->incr_need > self->max_size && self->max_size)
1486 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1487 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1488
1489 break;
1490 }
1491
1492 SV *sv;
1493 char *offset;
1494
1495 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1496 XPUSHs (sv);
1497
1498 sv_chop (cborstr, offset);
1499
1500 av_clear (self->incr_count);
1501 av_push (self->incr_count, newSViv (1));
1502
1503 self->incr_pos = 0;
1504 self->incr_need = self->incr_pos + 1;
1505 }
1506 while (ix);
1507}
1508
1509void incr_reset (CBOR *self)
1510 CODE:
1511{
1512 SvREFCNT_dec (self->incr_count);
1513 self->incr_count = 0;
1514}
1042 1515
1043void DESTROY (CBOR *self) 1516void DESTROY (CBOR *self)
1044 CODE: 1517 PPCODE:
1045 SvREFCNT_dec (self->cb_sk_object); 1518 cbor_free (self);
1046 SvREFCNT_dec (self->cb_object);
1047
1048#endif
1049 1519
1050PROTOTYPES: ENABLE 1520PROTOTYPES: ENABLE
1051 1521
1052void encode_cbor (SV *scalar) 1522void encode_cbor (SV *scalar)
1523 ALIAS:
1524 encode_cbor = 0
1525 encode_cbor_sharing = F_ALLOW_SHARING
1053 PPCODE: 1526 PPCODE:
1054{ 1527{
1055 CBOR cbor; 1528 CBOR cbor;
1056 cbor_init (&cbor); 1529 cbor_init (&cbor);
1530 cbor.flags |= ix;
1057 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1531 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1058 XPUSHs (scalar); 1532 XPUSHs (scalar);
1059} 1533}
1060 1534
1061void decode_cbor (SV *cborstr) 1535void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines