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 1.12 by root, Mon Oct 28 19:06:45 2013 UTC vs.
Revision 1.49 by root, Mon Feb 8 04:11:11 2016 UTC

7#include <stdlib.h> 7#include <stdlib.h>
8#include <stdio.h> 8#include <stdio.h>
9#include <limits.h> 9#include <limits.h>
10#include <float.h> 10#include <float.h>
11 11
12#define ECB_NO_THREADS 1
12#include "ecb.h" 13#include "ecb.h"
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 = { 0 };
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 WANT (len);
755 dec->cur += len;
756
757 hv_store (hv, key, len, decode_sv (dec), 0);
758
759 return;
760 }
761 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
762 {
763 I32 len = decode_uint (dec);
764 char *key = (char *)dec->cur;
765
766 WANT (len);
767 dec->cur += len;
768
769 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
770 if (!is_utf8_string (key, len))
771 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
772
773 hv_store (hv, key, -len, decode_sv (dec), 0);
774
775 return;
776 }
777
778 SV *k = decode_sv (dec);
779 SV *v = decode_sv (dec);
780
781 hv_store_ent (hv, k, v, 0);
782 SvREFCNT_dec (k);
783
784fail:
785 ;
786}
787
578static SV * 788static SV *
579decode_hv (dec_t *dec) 789decode_hv (dec_t *dec)
580{ 790{
581 HV *hv = newHV (); 791 HV *hv = newHV ();
582 792
583 DEC_INC_DEPTH; 793 DEC_INC_DEPTH;
584 794
585 if ((*dec->cur & 31) == 31) 795 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
586 { 796 {
587 ++dec->cur; 797 ++dec->cur;
588 798
589 for (;;) 799 for (;;)
590 { 800 {
591 WANT (1); 801 WANT (1);
592 802
593 if (*dec->cur == (0xe0 | 31)) 803 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
594 { 804 {
595 ++dec->cur; 805 ++dec->cur;
596 break; 806 break;
597 } 807 }
598 808
599 SV *k = decode_sv (dec); 809 decode_he (dec, hv);
600 SV *v = decode_sv (dec);
601
602 hv_store_ent (hv, k, v, 0);
603 SvREFCNT_dec (k);
604 } 810 }
605 } 811 }
606 else 812 else
607 { 813 {
608 int len = decode_uint (dec); 814 int pairs = decode_uint (dec);
609 815
610 while (len--) 816 while (pairs--)
611 { 817 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 } 818 }
619 819
620 DEC_DEC_DEPTH; 820 DEC_DEC_DEPTH;
621 return newRV_noinc ((SV *)hv); 821 return newRV_noinc ((SV *)hv);
622 822
629static SV * 829static SV *
630decode_str (dec_t *dec, int utf8) 830decode_str (dec_t *dec, int utf8)
631{ 831{
632 SV *sv = 0; 832 SV *sv = 0;
633 833
634 if ((*dec->cur & 31) == 31) 834 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
635 { 835 {
836 // indefinite length strings
636 ++dec->cur; 837 ++dec->cur;
637 838
839 U8 major = *dec->cur & MAJOR_MISC;
840
638 sv = newSVpvn ("", 0); 841 sv = newSVpvn ("", 0);
639 842
640 // not very fast, and certainly not robust against illegal input
641 for (;;) 843 for (;;)
642 { 844 {
643 WANT (1); 845 WANT (1);
644 846
645 if (*dec->cur == (0xe0 | 31)) 847 if ((*dec->cur - major) > LENGTH_EXT8)
848 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
646 { 849 {
647 ++dec->cur; 850 ++dec->cur;
648 break; 851 break;
649 } 852 }
853 else
854 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
650 855
651 sv_catsv (sv, decode_sv (dec)); 856 STRLEN len = decode_uint (dec);
857
858 WANT (len);
859 sv_catpvn (sv, dec->cur, len);
860 dec->cur += len;
652 } 861 }
653 } 862 }
654 else 863 else
655 { 864 {
656 STRLEN len = decode_uint (dec); 865 STRLEN len = decode_uint (dec);
657 866
658 WANT (len); 867 WANT (len);
659 sv = newSVpvn (dec->cur, len); 868 sv = newSVpvn (dec->cur, len);
660 dec->cur += len; 869 dec->cur += len;
870
871 if (ecb_expect_false (dec->stringref)
872 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
873 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
661 } 874 }
662 875
663 if (utf8) 876 if (utf8)
877 {
878 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
879 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
880 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
881
664 SvUTF8_on (sv); 882 SvUTF8_on (sv);
883 }
665 884
666 return sv; 885 return sv;
667 886
668fail: 887fail:
669 SvREFCNT_dec (sv); 888 SvREFCNT_dec (sv);
671} 890}
672 891
673static SV * 892static SV *
674decode_tagged (dec_t *dec) 893decode_tagged (dec_t *dec)
675{ 894{
895 SV *sv = 0;
676 UV tag = decode_uint (dec); 896 UV tag = decode_uint (dec);
897
898 WANT (1);
899
900 switch (tag)
901 {
902 case CBOR_TAG_MAGIC:
677 SV *sv = decode_sv (dec); 903 sv = decode_sv (dec);
904 break;
678 905
679 if (tag == CBOR_TAG_MAGIC) 906 case CBOR_TAG_INDIRECTION:
680 return sv; 907 sv = newRV_noinc (decode_sv (dec));
681 else if (tag == CBOR_TAG_PERL_OBJECT) 908 break;
682 { 909
910 case CBOR_TAG_STRINGREF_NAMESPACE:
911 {
912 ENTER; SAVETMPS;
913
914 SAVESPTR (dec->stringref);
915 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
916
917 sv = decode_sv (dec);
918
919 FREETMPS; LEAVE;
920 }
921 break;
922
923 case CBOR_TAG_STRINGREF:
924 {
925 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
926 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
927
928 UV idx = decode_uint (dec);
929
930 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
931 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
932
933 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
934 }
935 break;
936
937 case CBOR_TAG_VALUE_SHAREABLE:
938 {
939 if (ecb_expect_false (!dec->shareable))
940 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
941
942 if (dec->cbor.flags & F_ALLOW_CYCLES)
943 {
944 sv = newSV (0);
945 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
946
947 SV *osv = decode_sv (dec);
948 sv_setsv (sv, osv);
949 SvREFCNT_dec_NN (osv);
950 }
951 else
952 {
953 av_push (dec->shareable, &PL_sv_undef);
954 int idx = AvFILLp (dec->shareable);
955 sv = decode_sv (dec);
956 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
957 }
958 }
959 break;
960
961 case CBOR_TAG_VALUE_SHAREDREF:
962 {
963 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
964 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
965
966 UV idx = decode_uint (dec);
967
968 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
969 ERR ("corrupted CBOR data (sharedref index out of bounds)");
970
971 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
972
973 if (sv == &PL_sv_undef)
974 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
975 }
976 break;
977
978 case CBOR_TAG_PERL_OBJECT:
979 {
980 sv = decode_sv (dec);
981
683 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 982 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
684 ERR ("corrupted CBOR data (non-array perl object)"); 983 ERR ("corrupted CBOR data (non-array perl object)");
685 984
686 AV *av = (AV *)SvRV (sv); 985 AV *av = (AV *)SvRV (sv);
687 int len = av_len (av) + 1; 986 int len = av_len (av) + 1;
688 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0); 987 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
689 988
690 if (!stash) 989 if (!stash)
691 ERR ("cannot decode perl-object (package does not exist)"); 990 ERR ("cannot decode perl-object (package does not exist)");
692 991
693 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0); 992 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
694 993
695 if (!method) 994 if (!method)
696 ERR ("cannot decode perl-object (package does not have a THAW method)"); 995 ERR ("cannot decode perl-object (package does not have a THAW method)");
697 996
698 dSP; 997 dSP;
699 998
700 ENTER; SAVETMPS; PUSHMARK (SP); 999 ENTER; SAVETMPS; PUSHMARK (SP);
701 EXTEND (SP, len + 1); 1000 EXTEND (SP, len + 1);
702 // we re-bless the reference to get overload and other niceties right 1001 // we re-bless the reference to get overload and other niceties right
703 PUSHs (*av_fetch (av, 0, 1)); 1002 PUSHs (*av_fetch (av, 0, 1));
704 PUSHs (sv_cbor); 1003 PUSHs (sv_cbor);
705 1004
706 int i; 1005 int i;
707 1006
708 for (i = 1; i < len; ++i) 1007 for (i = 1; i < len; ++i)
709 PUSHs (*av_fetch (av, i, 1)); 1008 PUSHs (*av_fetch (av, i, 1));
710 1009
711 PUTBACK; 1010 PUTBACK;
712 call_sv ((SV *)GvCV (method), G_SCALAR); 1011 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
713 SPAGAIN; 1012 SPAGAIN;
714 1013
1014 if (SvTRUE (ERRSV))
1015 {
1016 FREETMPS; LEAVE;
1017 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1018 }
1019
1020 SvREFCNT_dec (sv);
715 sv = SvREFCNT_inc (POPs); 1021 sv = SvREFCNT_inc (POPs);
716 1022
717 PUTBACK; 1023 PUTBACK;
718 1024
719 FREETMPS; LEAVE; 1025 FREETMPS; LEAVE;
1026 }
1027 break;
720 1028
721 return sv; 1029 default:
722 } 1030 {
723 else 1031 sv = decode_sv (dec);
724 { 1032
1033 dSP;
1034 ENTER; SAVETMPS; PUSHMARK (SP);
1035 EXTEND (SP, 2);
1036 PUSHs (newSVuv (tag));
1037 PUSHs (sv);
1038
1039 PUTBACK;
1040 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1041 SPAGAIN;
1042
1043 if (SvTRUE (ERRSV))
1044 {
1045 FREETMPS; LEAVE;
1046 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1047 }
1048
1049 if (count)
1050 {
1051 SvREFCNT_dec (sv);
1052 sv = SvREFCNT_inc (POPs);
1053 }
1054 else
1055 {
725 AV *av = newAV (); 1056 AV *av = newAV ();
726 av_push (av, newSVuv (tag)); 1057 av_push (av, newSVuv (tag));
727 av_push (av, sv); 1058 av_push (av, sv);
728 1059
729 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1060 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
730 ? cbor_tagged_stash 1061 ? cbor_tagged_stash
731 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1062 : gv_stashpv ("CBOR::XS::Tagged" , 1);
732
733 return sv_bless (newRV_noinc ((SV *)av), tagged_stash); 1063 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
1064 }
1065
1066 PUTBACK;
1067
1068 FREETMPS; LEAVE;
1069 }
1070 break;
734 } 1071 }
1072
1073 return sv;
735 1074
736fail: 1075fail:
737 SvREFCNT_dec (sv); 1076 SvREFCNT_dec (sv);
738 return &PL_sv_undef; 1077 return &PL_sv_undef;
739} 1078}
741static SV * 1080static SV *
742decode_sv (dec_t *dec) 1081decode_sv (dec_t *dec)
743{ 1082{
744 WANT (1); 1083 WANT (1);
745 1084
746 switch (*dec->cur >> 5) 1085 switch (*dec->cur >> MAJOR_SHIFT)
747 { 1086 {
748 case 0: // unsigned int 1087 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
749 return newSVuv (decode_uint (dec)); 1088 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
750 case 1: // negative int 1089 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
751 return newSViv (-1 - (IV)decode_uint (dec)); 1090 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
752 case 2: // octet string 1091 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
753 return decode_str (dec, 0); 1092 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
754 case 3: // utf-8 string 1093 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
755 return decode_str (dec, 1); 1094
756 case 4: // array 1095 case MAJOR_MISC >> MAJOR_SHIFT:
757 return decode_av (dec);
758 case 5: // map
759 return decode_hv (dec);
760 case 6: // tag
761 return decode_tagged (dec);
762 case 7: // misc
763 switch (*dec->cur++ & 31) 1096 switch (*dec->cur++ & MINOR_MASK)
764 { 1097 {
765 case 20: 1098 case SIMPLE_FALSE:
766#if CBOR_SLOW 1099#if CBOR_SLOW
767 types_false = get_bool ("Types::Serialiser::false"); 1100 types_false = get_bool ("Types::Serialiser::false");
768#endif 1101#endif
769 return newSVsv (types_false); 1102 return newSVsv (types_false);
770 case 21: 1103 case SIMPLE_TRUE:
771#if CBOR_SLOW 1104#if CBOR_SLOW
772 types_true = get_bool ("Types::Serialiser::true"); 1105 types_true = get_bool ("Types::Serialiser::true");
773#endif 1106#endif
774 return newSVsv (types_true); 1107 return newSVsv (types_true);
775 case 22: 1108 case SIMPLE_NULL:
776 return newSVsv (&PL_sv_undef); 1109 return newSVsv (&PL_sv_undef);
777 case 23: 1110 case SIMPLE_UNDEF:
778#if CBOR_SLOW 1111#if CBOR_SLOW
779 types_error = get_bool ("Types::Serialiser::error"); 1112 types_error = get_bool ("Types::Serialiser::error");
780#endif 1113#endif
781 return newSVsv (types_error); 1114 return newSVsv (types_error);
782 1115
783 case 25: 1116 case MISC_FLOAT16:
784 { 1117 {
785 WANT (2); 1118 WANT (2);
786 1119
787 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1120 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
788 dec->cur += 2; 1121 dec->cur += 2;
789 1122
790 return newSVnv (ecb_binary16_to_float (fp)); 1123 return newSVnv (ecb_binary16_to_float (fp));
791 } 1124 }
792 1125
793 case 26: 1126 case MISC_FLOAT32:
794 { 1127 {
795 uint32_t fp; 1128 uint32_t fp;
796 WANT (4); 1129 WANT (4);
797 memcpy (&fp, dec->cur, 4); 1130 memcpy (&fp, dec->cur, 4);
798 dec->cur += 4; 1131 dec->cur += 4;
801 fp = ecb_bswap32 (fp); 1134 fp = ecb_bswap32 (fp);
802 1135
803 return newSVnv (ecb_binary32_to_float (fp)); 1136 return newSVnv (ecb_binary32_to_float (fp));
804 } 1137 }
805 1138
806 case 27: 1139 case MISC_FLOAT64:
807 { 1140 {
808 uint64_t fp; 1141 uint64_t fp;
809 WANT (8); 1142 WANT (8);
810 memcpy (&fp, dec->cur, 8); 1143 memcpy (&fp, dec->cur, 8);
811 dec->cur += 8; 1144 dec->cur += 8;
814 fp = ecb_bswap64 (fp); 1147 fp = ecb_bswap64 (fp);
815 1148
816 return newSVnv (ecb_binary64_to_double (fp)); 1149 return newSVnv (ecb_binary64_to_double (fp));
817 } 1150 }
818 1151
819 // 0..19 unassigned 1152 // 0..19 unassigned simple
820 // 24 reserved + unassigned (reserved values are not encodable) 1153 // 24 reserved + unassigned simple (reserved values are not encodable)
1154 // 28-30 unassigned misc
1155 // 31 break code
821 default: 1156 default:
822 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1157 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
823 } 1158 }
824 1159
825 break; 1160 break;
826 } 1161 }
827 1162
830} 1165}
831 1166
832static SV * 1167static SV *
833decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1168decode_cbor (SV *string, CBOR *cbor, char **offset_return)
834{ 1169{
835 dec_t dec; 1170 dec_t dec = { 0 };
836 SV *sv; 1171 SV *sv;
1172 STRLEN len;
1173 char *data = SvPVbyte (string, len);
837 1174
838 /* work around bugs in 5.10 where manipulating magic values
839 * makes perl ignore the magic in subsequent accesses.
840 * also make a copy of non-PV values, to get them into a clean
841 * state (SvPV should do that, but it's buggy, see below).
842 */
843 /*SvGETMAGIC (string);*/
844 if (SvMAGICAL (string) || !SvPOK (string))
845 string = sv_2mortal (newSVsv (string));
846
847 SvUPGRADE (string, SVt_PV);
848
849 /* work around a bug in perl 5.10, which causes SvCUR to fail an
850 * assertion with -DDEBUGGING, although SvCUR is documented to
851 * return the xpv_cur field which certainly exists after upgrading.
852 * according to nicholas clark, calling SvPOK fixes this.
853 * But it doesn't fix it, so try another workaround, call SvPV_nolen
854 * and hope for the best.
855 * Damnit, SvPV_nolen still trips over yet another assertion. This
856 * assertion business is seriously broken, try yet another workaround
857 * for the broken -DDEBUGGING.
858 */
859 {
860#ifdef DEBUGGING
861 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
862#else
863 STRLEN offset = SvCUR (string);
864#endif
865
866 if (offset > cbor->max_size && cbor->max_size) 1175 if (len > cbor->max_size && cbor->max_size)
867 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu", 1176 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
868 (unsigned long)SvCUR (string), (unsigned long)cbor->max_size); 1177 (unsigned long)len, (unsigned long)cbor->max_size);
869 }
870
871 sv_utf8_downgrade (string, 0);
872 1178
873 dec.cbor = *cbor; 1179 dec.cbor = *cbor;
874 dec.cur = (U8 *)SvPVX (string); 1180 dec.cur = (U8 *)data;
875 dec.end = (U8 *)SvEND (string); 1181 dec.end = (U8 *)data + len;
876 dec.err = 0;
877 dec.depth = 0;
878 1182
879 sv = decode_sv (&dec); 1183 sv = decode_sv (&dec);
880 1184
881 if (offset_return) 1185 if (offset_return)
882 *offset_return = dec.cur; 1186 *offset_return = dec.cur;
885 if (dec.cur != dec.end && !dec.err) 1189 if (dec.cur != dec.end && !dec.err)
886 dec.err = "garbage after CBOR object"; 1190 dec.err = "garbage after CBOR object";
887 1191
888 if (dec.err) 1192 if (dec.err)
889 { 1193 {
1194 if (dec.shareable)
1195 {
1196 // need to break cyclic links, which whould all be in shareable
1197 int i;
1198 SV **svp;
1199
1200 for (i = av_len (dec.shareable) + 1; i--; )
1201 if ((svp = av_fetch (dec.shareable, i, 0)))
1202 sv_setsv (*svp, &PL_sv_undef);
1203 }
1204
890 SvREFCNT_dec (sv); 1205 SvREFCNT_dec (sv);
891 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)SvPVX (string), (int)(uint8_t)*dec.cur); 1206 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
892 } 1207 }
893 1208
894 sv = sv_2mortal (sv); 1209 sv = sv_2mortal (sv);
895 1210
896 return sv; 1211 return sv;
897} 1212}
898 1213
1214/////////////////////////////////////////////////////////////////////////////
1215// incremental parser
1216
1217#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1218
1219// returns 0 for notyet, 1 for success or error
1220static int
1221incr_parse (CBOR *self, SV *cborstr)
1222{
1223 STRLEN cur;
1224 SvPV (cborstr, cur);
1225
1226 while (ecb_expect_true (self->incr_need <= cur))
1227 {
1228 // table of integer count bytes
1229 static I8 incr_len[MINOR_MASK + 1] = {
1230 0, 0, 0, 0, 0, 0, 0, 0,
1231 0, 0, 0, 0, 0, 0, 0, 0,
1232 0, 0, 0, 0, 0, 0, 0, 0,
1233 1, 2, 4, 8,-1,-1,-1,-2
1234 };
1235
1236 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1237 U8 m = *p & MINOR_MASK;
1238 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1239 I8 ilen = incr_len[m];
1240
1241 self->incr_need = self->incr_pos + 1;
1242
1243 if (ecb_expect_false (ilen < 0))
1244 {
1245 if (m != MINOR_INDEF)
1246 return 1; // error
1247
1248 if (*p == (MAJOR_MISC | MINOR_INDEF))
1249 {
1250 if (count >= 0)
1251 return 1; // error
1252
1253 count = 1;
1254 }
1255 else
1256 {
1257 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1258 count = -1;
1259 }
1260 }
1261 else
1262 {
1263 self->incr_need += ilen;
1264 if (ecb_expect_false (self->incr_need > cur))
1265 return 0;
1266
1267 int major = *p >> MAJOR_SHIFT;
1268
1269 switch (major)
1270 {
1271 case MAJOR_TAG >> MAJOR_SHIFT:
1272 ++count; // tags merely prefix another value
1273 break;
1274
1275 case MAJOR_BYTES >> MAJOR_SHIFT:
1276 case MAJOR_TEXT >> MAJOR_SHIFT:
1277 case MAJOR_ARRAY >> MAJOR_SHIFT:
1278 case MAJOR_MAP >> MAJOR_SHIFT:
1279 {
1280 UV len;
1281
1282 if (ecb_expect_false (ilen))
1283 {
1284 len = 0;
1285
1286 do {
1287 len = (len << 8) | *++p;
1288 } while (--ilen);
1289 }
1290 else
1291 len = m;
1292
1293 switch (major)
1294 {
1295 case MAJOR_BYTES >> MAJOR_SHIFT:
1296 case MAJOR_TEXT >> MAJOR_SHIFT:
1297 self->incr_need += len;
1298 if (ecb_expect_false (self->incr_need > cur))
1299 return 0;
1300
1301 break;
1302
1303 case MAJOR_MAP >> MAJOR_SHIFT:
1304 len <<= 1;
1305 case MAJOR_ARRAY >> MAJOR_SHIFT:
1306 if (len)
1307 {
1308 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1309 count = len + 1;
1310 }
1311 break;
1312 }
1313 }
1314 }
1315 }
1316
1317 self->incr_pos = self->incr_need;
1318
1319 if (count > 0)
1320 {
1321 while (!--count)
1322 {
1323 if (!AvFILLp (self->incr_count))
1324 return 1; // done
1325
1326 SvREFCNT_dec_NN (av_pop (self->incr_count));
1327 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1328 }
1329
1330 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1331 }
1332 }
1333
1334 return 0;
1335}
1336
1337
899///////////////////////////////////////////////////////////////////////////// 1338/////////////////////////////////////////////////////////////////////////////
900// XS interface functions 1339// XS interface functions
901 1340
902MODULE = CBOR::XS PACKAGE = CBOR::XS 1341MODULE = CBOR::XS PACKAGE = CBOR::XS
903 1342
910 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1); 1349 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
911 1350
912 types_true = get_bool ("Types::Serialiser::true" ); 1351 types_true = get_bool ("Types::Serialiser::true" );
913 types_false = get_bool ("Types::Serialiser::false"); 1352 types_false = get_bool ("Types::Serialiser::false");
914 types_error = get_bool ("Types::Serialiser::error"); 1353 types_error = get_bool ("Types::Serialiser::error");
1354
1355 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
915 1356
916 sv_cbor = newSVpv ("CBOR", 0); 1357 sv_cbor = newSVpv ("CBOR", 0);
917 SvREADONLY_on (sv_cbor); 1358 SvREADONLY_on (sv_cbor);
918} 1359}
919 1360
940 1381
941void shrink (CBOR *self, int enable = 1) 1382void shrink (CBOR *self, int enable = 1)
942 ALIAS: 1383 ALIAS:
943 shrink = F_SHRINK 1384 shrink = F_SHRINK
944 allow_unknown = F_ALLOW_UNKNOWN 1385 allow_unknown = F_ALLOW_UNKNOWN
1386 allow_sharing = F_ALLOW_SHARING
1387 allow_cycles = F_ALLOW_CYCLES
1388 pack_strings = F_PACK_STRINGS
1389 validate_utf8 = F_VALIDATE_UTF8
945 PPCODE: 1390 PPCODE:
946{ 1391{
947 if (enable) 1392 if (enable)
948 self->flags |= ix; 1393 self->flags |= ix;
949 else 1394 else
954 1399
955void get_shrink (CBOR *self) 1400void get_shrink (CBOR *self)
956 ALIAS: 1401 ALIAS:
957 get_shrink = F_SHRINK 1402 get_shrink = F_SHRINK
958 get_allow_unknown = F_ALLOW_UNKNOWN 1403 get_allow_unknown = F_ALLOW_UNKNOWN
1404 get_allow_sharing = F_ALLOW_SHARING
1405 get_allow_cycles = F_ALLOW_CYCLES
1406 get_pack_strings = F_PACK_STRINGS
1407 get_validate_utf8 = F_VALIDATE_UTF8
959 PPCODE: 1408 PPCODE:
960 XPUSHs (boolSV (self->flags & ix)); 1409 XPUSHs (boolSV (self->flags & ix));
961 1410
962void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1411void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
963 PPCODE: 1412 PPCODE:
979 CODE: 1428 CODE:
980 RETVAL = self->max_size; 1429 RETVAL = self->max_size;
981 OUTPUT: 1430 OUTPUT:
982 RETVAL 1431 RETVAL
983 1432
984#if 0 //TODO 1433void filter (CBOR *self, SV *filter = 0)
985
986void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef)
987 PPCODE: 1434 PPCODE:
988{
989 SvREFCNT_dec (self->cb_object); 1435 SvREFCNT_dec (self->filter);
990 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; 1436 self->filter = filter ? newSVsv (filter) : filter;
991
992 XPUSHs (ST (0)); 1437 XPUSHs (ST (0));
993}
994 1438
995void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef) 1439SV *get_filter (CBOR *self)
996 PPCODE: 1440 CODE:
997{ 1441 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
998 if (!self->cb_sk_object) 1442 OUTPUT:
999 self->cb_sk_object = newHV (); 1443 RETVAL
1000
1001 if (SvOK (cb))
1002 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1003 else
1004 {
1005 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1006
1007 if (!HvKEYS (self->cb_sk_object))
1008 {
1009 SvREFCNT_dec (self->cb_sk_object);
1010 self->cb_sk_object = 0;
1011 }
1012 }
1013
1014 XPUSHs (ST (0));
1015}
1016
1017#endif
1018 1444
1019void encode (CBOR *self, SV *scalar) 1445void encode (CBOR *self, SV *scalar)
1020 PPCODE: 1446 PPCODE:
1021 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; 1447 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
1022 XPUSHs (scalar); 1448 XPUSHs (scalar);
1035 EXTEND (SP, 2); 1461 EXTEND (SP, 2);
1036 PUSHs (sv); 1462 PUSHs (sv);
1037 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1463 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1038} 1464}
1039 1465
1040#if 0 1466void incr_parse (CBOR *self, SV *cborstr)
1467 ALIAS:
1468 incr_parse_multiple = 1
1469 PPCODE:
1470{
1471 if (SvUTF8 (cborstr))
1472 sv_utf8_downgrade (cborstr, 0);
1473
1474 if (!self->incr_count)
1475 {
1476 self->incr_count = newAV ();
1477 self->incr_pos = 0;
1478 self->incr_need = 1;
1479
1480 av_push (self->incr_count, newSViv (1));
1481 }
1482
1483 do
1484 {
1485 if (!incr_parse (self, cborstr))
1486 {
1487 if (self->incr_need > self->max_size && self->max_size)
1488 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1489 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1490
1491 break;
1492 }
1493
1494 SV *sv;
1495 char *offset;
1496
1497 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1498 XPUSHs (sv);
1499
1500 sv_chop (cborstr, offset);
1501
1502 av_clear (self->incr_count);
1503 av_push (self->incr_count, newSViv (1));
1504
1505 self->incr_pos = 0;
1506 self->incr_need = self->incr_pos + 1;
1507 }
1508 while (ix);
1509}
1510
1511void incr_reset (CBOR *self)
1512 CODE:
1513{
1514 SvREFCNT_dec (self->incr_count);
1515 self->incr_count = 0;
1516}
1041 1517
1042void DESTROY (CBOR *self) 1518void DESTROY (CBOR *self)
1043 CODE: 1519 PPCODE:
1044 SvREFCNT_dec (self->cb_sk_object); 1520 cbor_free (self);
1045 SvREFCNT_dec (self->cb_object);
1046
1047#endif
1048 1521
1049PROTOTYPES: ENABLE 1522PROTOTYPES: ENABLE
1050 1523
1051void encode_cbor (SV *scalar) 1524void encode_cbor (SV *scalar)
1525 ALIAS:
1526 encode_cbor = 0
1527 encode_cbor_sharing = F_ALLOW_SHARING
1052 PPCODE: 1528 PPCODE:
1053{ 1529{
1054 CBOR cbor; 1530 CBOR cbor;
1055 cbor_init (&cbor); 1531 cbor_init (&cbor);
1532 cbor.flags |= ix;
1056 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1533 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1057 XPUSHs (scalar); 1534 XPUSHs (scalar);
1058} 1535}
1059 1536
1060void decode_cbor (SV *cborstr) 1537void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines