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 1.11 by root, Sun Oct 27 22:35:15 2013 UTC vs.
Revision 1.45 by root, Sun Dec 14 05:48:39 2014 UTC

7#include <stdlib.h> 7#include <stdlib.h>
8#include <stdio.h> 8#include <stdio.h>
9#include <limits.h> 9#include <limits.h>
10#include <float.h> 10#include <float.h>
11 11
12#define ECB_NO_THREADS 1
12#include "ecb.h" 13#include "ecb.h"
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 encode_sv (enc, AvARRAY (av)[i]);
215 338
216 --enc->depth; 339 --enc->depth;
217} 340}
218 341
219static void 342static void
228 351
229 int pairs = hv_iterinit (hv); 352 int pairs = hv_iterinit (hv);
230 int mg = SvMAGICAL (hv); 353 int mg = SvMAGICAL (hv);
231 354
232 if (mg) 355 if (mg)
233 encode_ch (enc, 0xa0 | 31); 356 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
234 else 357 else
235 encode_uint (enc, 0xa0, pairs); 358 encode_uint (enc, MAJOR_MAP, pairs);
236 359
237 while ((he = hv_iternext (hv))) 360 while ((he = hv_iternext (hv)))
238 { 361 {
239 if (HeKLEN (he) == HEf_SVKEY) 362 if (HeKLEN (he) == HEf_SVKEY)
240 encode_sv (enc, HeSVKEY (he)); 363 encode_sv (enc, HeSVKEY (he));
241 else 364 else
242 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 365 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
243 366
244 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 367 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
245 } 368 }
246 369
247 if (mg) 370 if (mg)
248 encode_ch (enc, 0xe0 | 31); 371 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
249 372
250 --enc->depth; 373 --enc->depth;
251} 374}
252 375
253// encode objects, arrays and special \0=false and \1=true values. 376// encode objects, arrays and special \0=false and \1=true values.
254static void 377static void
255encode_rv (enc_t *enc, SV *sv) 378encode_rv (enc_t *enc, SV *sv)
256{ 379{
257 svtype svt;
258
259 SvGETMAGIC (sv); 380 SvGETMAGIC (sv);
381
260 svt = SvTYPE (sv); 382 svtype svt = SvTYPE (sv);
261 383
262 if (ecb_expect_false (SvOBJECT (sv))) 384 if (ecb_expect_false (SvOBJECT (sv)))
263 { 385 {
264 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash 386 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
265 ? types_boolean_stash 387 ? types_boolean_stash
270 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 392 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
271 ? cbor_tagged_stash 393 ? cbor_tagged_stash
272 : gv_stashpv ("CBOR::XS::Tagged" , 1); 394 : gv_stashpv ("CBOR::XS::Tagged" , 1);
273 395
274 HV *stash = SvSTASH (sv); 396 HV *stash = SvSTASH (sv);
275 GV *method;
276 397
277 if (stash == boolean_stash) 398 if (stash == boolean_stash)
278 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 399 {
400 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
401 return;
402 }
279 else if (stash == error_stash) 403 else if (stash == error_stash)
280 encode_ch (enc, 0xe0 | 23); 404 {
405 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
406 return;
407 }
281 else if (stash == tagged_stash) 408 else if (stash == tagged_stash)
282 { 409 {
283 if (svt != SVt_PVAV) 410 if (svt != SVt_PVAV)
284 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 411 croak ("encountered CBOR::XS::Tagged object that isn't an array");
285 412
286 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 413 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
287 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 414 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
415
416 return;
417 }
418 }
419
420 if (ecb_expect_false (SvREFCNT (sv) > 1)
421 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
422 {
423 if (!enc->shareable)
424 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
425
426 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
427
428 if (SvOK (*svp))
288 } 429 {
430 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
431 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
432 return;
433 }
434 else
435 {
436 sv_setuv (*svp, enc->shareable_idx);
437 ++enc->shareable_idx;
438 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
439 }
440 }
441
442 if (ecb_expect_false (SvOBJECT (sv)))
443 {
444 HV *stash = SvSTASH (sv);
445 GV *method;
446
289 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 447 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
290 { 448 {
291 dSP; 449 dSP;
292 450
293 ENTER; SAVETMPS; PUSHMARK (SP); 451 ENTER; SAVETMPS; PUSHMARK (SP);
294 // we re-bless the reference to get overload and other niceties right 452 // we re-bless the reference to get overload and other niceties right
325 483
326 // catch this surprisingly common error 484 // catch this surprisingly common error
327 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 485 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)); 486 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
329 487
330 encode_uint (enc, 0xc0, CBOR_TAG_PERL_OBJECT); 488 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
331 encode_uint (enc, 0x80, count + 1); 489 encode_uint (enc, MAJOR_ARRAY, count + 1);
332 encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 490 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
333 491
334 while (count) 492 while (count)
335 encode_sv (enc, SP[1 - count--]); 493 encode_sv (enc, SP[1 - count--]);
336 494
337 PUTBACK; 495 PUTBACK;
344 } 502 }
345 else if (svt == SVt_PVHV) 503 else if (svt == SVt_PVHV)
346 encode_hv (enc, (HV *)sv); 504 encode_hv (enc, (HV *)sv);
347 else if (svt == SVt_PVAV) 505 else if (svt == SVt_PVAV)
348 encode_av (enc, (AV *)sv); 506 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 507 else
367 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 508 {
368 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 509 encode_tag (enc, CBOR_TAG_INDIRECTION);
510 encode_sv (enc, sv);
511 }
369} 512}
370 513
371static void 514static void
372encode_nv (enc_t *enc, SV *sv) 515encode_nv (enc_t *enc, SV *sv)
373{ 516{
374 double nv = SvNVX (sv); 517 double nv = SvNVX (sv);
375 518
376 need (enc, 9); 519 need (enc, 9);
377 520
378 if (ecb_expect_false (nv == (U32)nv)) 521 if (ecb_expect_false (nv == (NV)(U32)nv))
379 encode_uint (enc, 0x00, (U32)nv); 522 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
380 //TODO: maybe I32? 523 //TODO: maybe I32?
381 else if (ecb_expect_false (nv == (float)nv)) 524 else if (ecb_expect_false (nv == (float)nv))
382 { 525 {
383 uint32_t fp = ecb_float_to_binary32 (nv); 526 uint32_t fp = ecb_float_to_binary32 (nv);
384 527
385 *enc->cur++ = 0xe0 | 26; 528 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
386 529
387 if (!ecb_big_endian ()) 530 if (!ecb_big_endian ())
388 fp = ecb_bswap32 (fp); 531 fp = ecb_bswap32 (fp);
389 532
390 memcpy (enc->cur, &fp, 4); 533 memcpy (enc->cur, &fp, 4);
392 } 535 }
393 else 536 else
394 { 537 {
395 uint64_t fp = ecb_double_to_binary64 (nv); 538 uint64_t fp = ecb_double_to_binary64 (nv);
396 539
397 *enc->cur++ = 0xe0 | 27; 540 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
398 541
399 if (!ecb_big_endian ()) 542 if (!ecb_big_endian ())
400 fp = ecb_bswap64 (fp); 543 fp = ecb_bswap64 (fp);
401 544
402 memcpy (enc->cur, &fp, 8); 545 memcpy (enc->cur, &fp, 8);
411 554
412 if (SvPOKp (sv)) 555 if (SvPOKp (sv))
413 { 556 {
414 STRLEN len; 557 STRLEN len;
415 char *str = SvPV (sv, len); 558 char *str = SvPV (sv, len);
416 encode_str (enc, SvUTF8 (sv), str, len); 559 encode_strref (enc, SvUTF8 (sv), str, len);
417 } 560 }
418 else if (SvNOKp (sv)) 561 else if (SvNOKp (sv))
419 encode_nv (enc, sv); 562 encode_nv (enc, sv);
420 else if (SvIOKp (sv)) 563 else if (SvIOKp (sv))
421 { 564 {
422 if (SvIsUV (sv)) 565 if (SvIsUV (sv))
423 encode_uint (enc, 0x00, SvUVX (sv)); 566 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
424 else if (SvIVX (sv) >= 0) 567 else if (SvIVX (sv) >= 0)
425 encode_uint (enc, 0x00, SvIVX (sv)); 568 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
426 else 569 else
427 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 570 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
428 } 571 }
429 else if (SvROK (sv)) 572 else if (SvROK (sv))
430 encode_rv (enc, SvRV (sv)); 573 encode_rv (enc, SvRV (sv));
431 else if (!SvOK (sv)) 574 else if (!SvOK (sv))
432 encode_ch (enc, 0xe0 | 22); 575 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
433 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 576 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
434 encode_ch (enc, 0xe0 | 23); 577 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
435 else 578 else
436 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 579 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
437 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 580 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
438} 581}
439 582
440static SV * 583static SV *
441encode_cbor (SV *scalar, CBOR *cbor) 584encode_cbor (SV *scalar, CBOR *cbor)
442{ 585{
443 enc_t enc; 586 enc_t enc = { };
444 587
445 enc.cbor = *cbor; 588 enc.cbor = *cbor;
446 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 589 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
447 enc.cur = SvPVX (enc.sv); 590 enc.cur = SvPVX (enc.sv);
448 enc.end = SvEND (enc.sv); 591 enc.end = SvEND (enc.sv);
449 enc.depth = 0;
450 592
451 SvPOK_only (enc.sv); 593 SvPOK_only (enc.sv);
594
595 if (cbor->flags & F_PACK_STRINGS)
596 {
597 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
598 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
599 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
600 }
601
452 encode_sv (&enc, scalar); 602 encode_sv (&enc, scalar);
453 603
454 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 604 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
455 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 605 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
456 606
470 U8 *end; // end of input string 620 U8 *end; // end of input string
471 const char *err; // parse error, if != 0 621 const char *err; // parse error, if != 0
472 CBOR cbor; 622 CBOR cbor;
473 U32 depth; // recursion depth 623 U32 depth; // recursion depth
474 U32 maxdepth; // recursion depth limit 624 U32 maxdepth; // recursion depth limit
625 AV *shareable;
626 AV *stringref;
627 SV *decode_tagged;
475} dec_t; 628} dec_t;
476 629
477#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 630#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE
478 631
479#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 632#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data")
482#define DEC_DEC_DEPTH --dec->depth 635#define DEC_DEC_DEPTH --dec->depth
483 636
484static UV 637static UV
485decode_uint (dec_t *dec) 638decode_uint (dec_t *dec)
486{ 639{
487 switch (*dec->cur & 31) 640 U8 m = *dec->cur & MINOR_MASK;
488 { 641 ++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;
493 642
494 case 24: 643 if (ecb_expect_true (m < LENGTH_EXT1))
644 return m;
645 else if (ecb_expect_true (m == LENGTH_EXT1))
646 {
495 WANT (2); 647 WANT (1);
496 dec->cur += 2; 648 dec->cur += 1;
497 return dec->cur[-1]; 649 return dec->cur[-1];
498 650 }
499 case 25: 651 else if (ecb_expect_true (m == LENGTH_EXT2))
652 {
500 WANT (3); 653 WANT (2);
501 dec->cur += 3; 654 dec->cur += 2;
502 return (((UV)dec->cur[-2]) << 8) 655 return (((UV)dec->cur[-2]) << 8)
503 | ((UV)dec->cur[-1]); 656 | ((UV)dec->cur[-1]);
504 657 }
505 case 26: 658 else if (ecb_expect_true (m == LENGTH_EXT4))
659 {
506 WANT (5); 660 WANT (4);
507 dec->cur += 5; 661 dec->cur += 4;
508 return (((UV)dec->cur[-4]) << 24) 662 return (((UV)dec->cur[-4]) << 24)
509 | (((UV)dec->cur[-3]) << 16) 663 | (((UV)dec->cur[-3]) << 16)
510 | (((UV)dec->cur[-2]) << 8) 664 | (((UV)dec->cur[-2]) << 8)
511 | ((UV)dec->cur[-1]); 665 | ((UV)dec->cur[-1]);
512 666 }
513 case 27: 667 else if (ecb_expect_true (m == LENGTH_EXT8))
668 {
514 WANT (9); 669 WANT (8);
515 dec->cur += 9; 670 dec->cur += 8;
671
672 return
673#if UVSIZE < 8
674 0
675#else
516 return (((UV)dec->cur[-8]) << 56) 676 (((UV)dec->cur[-8]) << 56)
517 | (((UV)dec->cur[-7]) << 48) 677 | (((UV)dec->cur[-7]) << 48)
518 | (((UV)dec->cur[-6]) << 40) 678 | (((UV)dec->cur[-6]) << 40)
519 | (((UV)dec->cur[-5]) << 32) 679 | (((UV)dec->cur[-5]) << 32)
680#endif
520 | (((UV)dec->cur[-4]) << 24) 681 | (((UV)dec->cur[-4]) << 24)
521 | (((UV)dec->cur[-3]) << 16) 682 | (((UV)dec->cur[-3]) << 16)
522 | (((UV)dec->cur[-2]) << 8) 683 | (((UV)dec->cur[-2]) << 8)
523 | ((UV)dec->cur[-1]); 684 | ((UV)dec->cur[-1]);
524 685 }
525 default: 686 else
526 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 687 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
527 }
528 688
529fail: 689fail:
530 return 0; 690 return 0;
531} 691}
532 692
537{ 697{
538 AV *av = newAV (); 698 AV *av = newAV ();
539 699
540 DEC_INC_DEPTH; 700 DEC_INC_DEPTH;
541 701
542 if ((*dec->cur & 31) == 31) 702 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
543 { 703 {
544 ++dec->cur; 704 ++dec->cur;
545 705
546 for (;;) 706 for (;;)
547 { 707 {
548 WANT (1); 708 WANT (1);
549 709
550 if (*dec->cur == (0xe0 | 31)) 710 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
551 { 711 {
552 ++dec->cur; 712 ++dec->cur;
553 break; 713 break;
554 } 714 }
555 715
558 } 718 }
559 else 719 else
560 { 720 {
561 int i, len = decode_uint (dec); 721 int i, len = decode_uint (dec);
562 722
723 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); 724 av_fill (av, len - 1);
564 725
565 for (i = 0; i < len; ++i) 726 for (i = 0; i < len; ++i)
566 AvARRAY (av)[i] = decode_sv (dec); 727 AvARRAY (av)[i] = decode_sv (dec);
567 } 728 }
573 SvREFCNT_dec (av); 734 SvREFCNT_dec (av);
574 DEC_DEC_DEPTH; 735 DEC_DEC_DEPTH;
575 return &PL_sv_undef; 736 return &PL_sv_undef;
576} 737}
577 738
739static void
740decode_he (dec_t *dec, HV *hv)
741{
742 // for speed reasons, we specialcase single-string
743 // byte or utf-8 strings as keys, but only when !stringref
744
745 if (ecb_expect_true (!dec->stringref))
746 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
747 {
748 I32 len = decode_uint (dec);
749 char *key = (char *)dec->cur;
750
751 dec->cur += len;
752
753 hv_store (hv, key, len, decode_sv (dec), 0);
754
755 return;
756 }
757 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
758 {
759 I32 len = decode_uint (dec);
760 char *key = (char *)dec->cur;
761
762 dec->cur += len;
763
764 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
765 if (!is_utf8_string (key, len))
766 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
767
768 hv_store (hv, key, -len, decode_sv (dec), 0);
769
770 return;
771 }
772
773 SV *k = decode_sv (dec);
774 SV *v = decode_sv (dec);
775
776 hv_store_ent (hv, k, v, 0);
777 SvREFCNT_dec (k);
778
779fail:
780 ;
781}
782
578static SV * 783static SV *
579decode_hv (dec_t *dec) 784decode_hv (dec_t *dec)
580{ 785{
581 HV *hv = newHV (); 786 HV *hv = newHV ();
582 787
583 DEC_INC_DEPTH; 788 DEC_INC_DEPTH;
584 789
585 if ((*dec->cur & 31) == 31) 790 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
586 { 791 {
587 ++dec->cur; 792 ++dec->cur;
588 793
589 for (;;) 794 for (;;)
590 { 795 {
591 WANT (1); 796 WANT (1);
592 797
593 if (*dec->cur == (0xe0 | 31)) 798 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
594 { 799 {
595 ++dec->cur; 800 ++dec->cur;
596 break; 801 break;
597 } 802 }
598 803
599 SV *k = decode_sv (dec); 804 decode_he (dec, hv);
600 SV *v = decode_sv (dec);
601
602 hv_store_ent (hv, k, v, 0);
603 } 805 }
604 } 806 }
605 else 807 else
606 { 808 {
607 int len = decode_uint (dec); 809 int pairs = decode_uint (dec);
608 810
609 while (len--) 811 while (pairs--)
610 { 812 decode_he (dec, hv);
611 SV *k = decode_sv (dec);
612 SV *v = decode_sv (dec);
613
614 hv_store_ent (hv, k, v, 0);
615 }
616 } 813 }
617 814
618 DEC_DEC_DEPTH; 815 DEC_DEC_DEPTH;
619 return newRV_noinc ((SV *)hv); 816 return newRV_noinc ((SV *)hv);
620 817
627static SV * 824static SV *
628decode_str (dec_t *dec, int utf8) 825decode_str (dec_t *dec, int utf8)
629{ 826{
630 SV *sv = 0; 827 SV *sv = 0;
631 828
632 if ((*dec->cur & 31) == 31) 829 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
633 { 830 {
831 // indefinite length strings
634 ++dec->cur; 832 ++dec->cur;
635 833
834 U8 major = *dec->cur & MAJOR_MISC;
835
636 sv = newSVpvn ("", 0); 836 sv = newSVpvn ("", 0);
637 837
638 // not very fast, and certainly not robust against illegal input
639 for (;;) 838 for (;;)
640 { 839 {
641 WANT (1); 840 WANT (1);
642 841
643 if (*dec->cur == (0xe0 | 31)) 842 if ((*dec->cur - major) > LENGTH_EXT8)
843 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
644 { 844 {
645 ++dec->cur; 845 ++dec->cur;
646 break; 846 break;
647 } 847 }
848 else
849 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
648 850
649 sv_catsv (sv, decode_sv (dec)); 851 STRLEN len = decode_uint (dec);
852
853 WANT (len);
854 sv_catpvn (sv, dec->cur, len);
855 dec->cur += len;
650 } 856 }
651 } 857 }
652 else 858 else
653 { 859 {
654 STRLEN len = decode_uint (dec); 860 STRLEN len = decode_uint (dec);
655 861
656 WANT (len); 862 WANT (len);
657 sv = newSVpvn (dec->cur, len); 863 sv = newSVpvn (dec->cur, len);
658 dec->cur += len; 864 dec->cur += len;
865
866 if (ecb_expect_false (dec->stringref)
867 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
868 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
659 } 869 }
660 870
661 if (utf8) 871 if (utf8)
872 {
873 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
874 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
875 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
876
662 SvUTF8_on (sv); 877 SvUTF8_on (sv);
878 }
663 879
664 return sv; 880 return sv;
665 881
666fail: 882fail:
667 SvREFCNT_dec (sv); 883 SvREFCNT_dec (sv);
669} 885}
670 886
671static SV * 887static SV *
672decode_tagged (dec_t *dec) 888decode_tagged (dec_t *dec)
673{ 889{
890 SV *sv = 0;
674 UV tag = decode_uint (dec); 891 UV tag = decode_uint (dec);
892
893 WANT (1);
894
895 switch (tag)
896 {
897 case CBOR_TAG_MAGIC:
675 SV *sv = decode_sv (dec); 898 sv = decode_sv (dec);
899 break;
676 900
677 if (tag == CBOR_TAG_MAGIC) 901 case CBOR_TAG_INDIRECTION:
678 return sv; 902 sv = newRV_noinc (decode_sv (dec));
679 else if (tag == CBOR_TAG_PERL_OBJECT) 903 break;
680 { 904
905 case CBOR_TAG_STRINGREF_NAMESPACE:
906 {
907 ENTER; SAVETMPS;
908
909 SAVESPTR (dec->stringref);
910 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
911
912 sv = decode_sv (dec);
913
914 FREETMPS; LEAVE;
915 }
916 break;
917
918 case CBOR_TAG_STRINGREF:
919 {
920 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
921 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
922
923 UV idx = decode_uint (dec);
924
925 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
926 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
927
928 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
929 }
930 break;
931
932 case CBOR_TAG_VALUE_SHAREABLE:
933 {
934 if (ecb_expect_false (!dec->shareable))
935 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
936
937 if (dec->cbor.flags & F_ALLOW_CYCLES)
938 {
939 sv = newSV (0);
940 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
941
942 SV *osv = decode_sv (dec);
943 sv_setsv (sv, osv);
944 SvREFCNT_dec_NN (osv);
945 }
946 else
947 {
948 av_push (dec->shareable, &PL_sv_undef);
949 int idx = AvFILLp (dec->shareable);
950 sv = decode_sv (dec);
951 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
952 }
953 }
954 break;
955
956 case CBOR_TAG_VALUE_SHAREDREF:
957 {
958 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
959 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
960
961 UV idx = decode_uint (dec);
962
963 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
964 ERR ("corrupted CBOR data (sharedref index out of bounds)");
965
966 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
967
968 if (sv == &PL_sv_undef)
969 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
970 }
971 break;
972
973 case CBOR_TAG_PERL_OBJECT:
974 {
975 sv = decode_sv (dec);
976
681 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 977 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
682 ERR ("corrupted CBOR data (non-array perl object)"); 978 ERR ("corrupted CBOR data (non-array perl object)");
683 979
684 AV *av = (AV *)SvRV (sv); 980 AV *av = (AV *)SvRV (sv);
685 int len = av_len (av) + 1; 981 int len = av_len (av) + 1;
686 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0); 982 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
687 983
688 if (!stash) 984 if (!stash)
689 ERR ("cannot decode perl-object (package does not exist)"); 985 ERR ("cannot decode perl-object (package does not exist)");
690 986
691 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0); 987 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
692 988
693 if (!method) 989 if (!method)
694 ERR ("cannot decode perl-object (package does not have a THAW method)"); 990 ERR ("cannot decode perl-object (package does not have a THAW method)");
695 991
696 dSP; 992 dSP;
697 993
698 ENTER; SAVETMPS; PUSHMARK (SP); 994 ENTER; SAVETMPS; PUSHMARK (SP);
699 EXTEND (SP, len + 1); 995 EXTEND (SP, len + 1);
700 // we re-bless the reference to get overload and other niceties right 996 // we re-bless the reference to get overload and other niceties right
701 PUSHs (*av_fetch (av, 0, 1)); 997 PUSHs (*av_fetch (av, 0, 1));
702 PUSHs (sv_cbor); 998 PUSHs (sv_cbor);
703 999
704 int i; 1000 int i;
705 1001
706 for (i = 1; i < len; ++i) 1002 for (i = 1; i < len; ++i)
707 PUSHs (*av_fetch (av, i, 1)); 1003 PUSHs (*av_fetch (av, i, 1));
708 1004
709 PUTBACK; 1005 PUTBACK;
710 call_sv ((SV *)GvCV (method), G_SCALAR); 1006 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
711 SPAGAIN; 1007 SPAGAIN;
712 1008
1009 if (SvTRUE (ERRSV))
1010 {
1011 FREETMPS; LEAVE;
1012 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1013 }
1014
1015 SvREFCNT_dec (sv);
713 sv = SvREFCNT_inc (POPs); 1016 sv = SvREFCNT_inc (POPs);
714 1017
715 PUTBACK; 1018 PUTBACK;
716 1019
717 FREETMPS; LEAVE; 1020 FREETMPS; LEAVE;
1021 }
1022 break;
718 1023
719 return sv; 1024 default:
720 } 1025 {
721 else 1026 sv = decode_sv (dec);
722 { 1027
1028 dSP;
1029 ENTER; SAVETMPS; PUSHMARK (SP);
1030 EXTEND (SP, 2);
1031 PUSHs (newSVuv (tag));
1032 PUSHs (sv);
1033
1034 PUTBACK;
1035 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1036 SPAGAIN;
1037
1038 if (SvTRUE (ERRSV))
1039 {
1040 FREETMPS; LEAVE;
1041 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1042 }
1043
1044 if (count)
1045 {
1046 SvREFCNT_dec (sv);
1047 sv = SvREFCNT_inc (POPs);
1048 }
1049 else
1050 {
723 AV *av = newAV (); 1051 AV *av = newAV ();
724 av_push (av, newSVuv (tag)); 1052 av_push (av, newSVuv (tag));
725 av_push (av, sv); 1053 av_push (av, sv);
726 1054
727 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1055 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
728 ? cbor_tagged_stash 1056 ? cbor_tagged_stash
729 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1057 : gv_stashpv ("CBOR::XS::Tagged" , 1);
730
731 return sv_bless (newRV_noinc ((SV *)av), tagged_stash); 1058 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
1059 }
1060
1061 PUTBACK;
1062
1063 FREETMPS; LEAVE;
1064 }
1065 break;
732 } 1066 }
1067
1068 return sv;
733 1069
734fail: 1070fail:
735 SvREFCNT_dec (sv); 1071 SvREFCNT_dec (sv);
736 return &PL_sv_undef; 1072 return &PL_sv_undef;
737} 1073}
739static SV * 1075static SV *
740decode_sv (dec_t *dec) 1076decode_sv (dec_t *dec)
741{ 1077{
742 WANT (1); 1078 WANT (1);
743 1079
744 switch (*dec->cur >> 5) 1080 switch (*dec->cur >> MAJOR_SHIFT)
745 { 1081 {
746 case 0: // unsigned int 1082 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
747 return newSVuv (decode_uint (dec)); 1083 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
748 case 1: // negative int 1084 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
749 return newSViv (-1 - (IV)decode_uint (dec)); 1085 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
750 case 2: // octet string 1086 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
751 return decode_str (dec, 0); 1087 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
752 case 3: // utf-8 string 1088 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
753 return decode_str (dec, 1); 1089
754 case 4: // array 1090 case MAJOR_MISC >> MAJOR_SHIFT:
755 return decode_av (dec);
756 case 5: // map
757 return decode_hv (dec);
758 case 6: // tag
759 return decode_tagged (dec);
760 case 7: // misc
761 switch (*dec->cur++ & 31) 1091 switch (*dec->cur++ & MINOR_MASK)
762 { 1092 {
763 case 20: 1093 case SIMPLE_FALSE:
764#if CBOR_SLOW 1094#if CBOR_SLOW
765 types_false = get_bool ("Types::Serialiser::false"); 1095 types_false = get_bool ("Types::Serialiser::false");
766#endif 1096#endif
767 return newSVsv (types_false); 1097 return newSVsv (types_false);
768 case 21: 1098 case SIMPLE_TRUE:
769#if CBOR_SLOW 1099#if CBOR_SLOW
770 types_true = get_bool ("Types::Serialiser::true"); 1100 types_true = get_bool ("Types::Serialiser::true");
771#endif 1101#endif
772 return newSVsv (types_true); 1102 return newSVsv (types_true);
773 case 22: 1103 case SIMPLE_NULL:
774 return newSVsv (&PL_sv_undef); 1104 return newSVsv (&PL_sv_undef);
775 case 23: 1105 case SIMPLE_UNDEF:
776#if CBOR_SLOW 1106#if CBOR_SLOW
777 types_error = get_bool ("Types::Serialiser::error"); 1107 types_error = get_bool ("Types::Serialiser::error");
778#endif 1108#endif
779 return newSVsv (types_error); 1109 return newSVsv (types_error);
780 1110
781 case 25: 1111 case MISC_FLOAT16:
782 { 1112 {
783 WANT (2); 1113 WANT (2);
784 1114
785 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1115 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
786 dec->cur += 2; 1116 dec->cur += 2;
787 1117
788 return newSVnv (ecb_binary16_to_float (fp)); 1118 return newSVnv (ecb_binary16_to_float (fp));
789 } 1119 }
790 1120
791 case 26: 1121 case MISC_FLOAT32:
792 { 1122 {
793 uint32_t fp; 1123 uint32_t fp;
794 WANT (4); 1124 WANT (4);
795 memcpy (&fp, dec->cur, 4); 1125 memcpy (&fp, dec->cur, 4);
796 dec->cur += 4; 1126 dec->cur += 4;
799 fp = ecb_bswap32 (fp); 1129 fp = ecb_bswap32 (fp);
800 1130
801 return newSVnv (ecb_binary32_to_float (fp)); 1131 return newSVnv (ecb_binary32_to_float (fp));
802 } 1132 }
803 1133
804 case 27: 1134 case MISC_FLOAT64:
805 { 1135 {
806 uint64_t fp; 1136 uint64_t fp;
807 WANT (8); 1137 WANT (8);
808 memcpy (&fp, dec->cur, 8); 1138 memcpy (&fp, dec->cur, 8);
809 dec->cur += 8; 1139 dec->cur += 8;
812 fp = ecb_bswap64 (fp); 1142 fp = ecb_bswap64 (fp);
813 1143
814 return newSVnv (ecb_binary64_to_double (fp)); 1144 return newSVnv (ecb_binary64_to_double (fp));
815 } 1145 }
816 1146
817 // 0..19 unassigned 1147 // 0..19 unassigned simple
818 // 24 reserved + unassigned (reserved values are not encodable) 1148 // 24 reserved + unassigned simple (reserved values are not encodable)
1149 // 28-30 unassigned misc
1150 // 31 break code
819 default: 1151 default:
820 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1152 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
821 } 1153 }
822 1154
823 break; 1155 break;
824 } 1156 }
825 1157
828} 1160}
829 1161
830static SV * 1162static SV *
831decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1163decode_cbor (SV *string, CBOR *cbor, char **offset_return)
832{ 1164{
833 dec_t dec; 1165 dec_t dec = { };
834 SV *sv; 1166 SV *sv;
1167 STRLEN len;
1168 char *data = SvPVbyte (string, len);
835 1169
836 /* work around bugs in 5.10 where manipulating magic values
837 * makes perl ignore the magic in subsequent accesses.
838 * also make a copy of non-PV values, to get them into a clean
839 * state (SvPV should do that, but it's buggy, see below).
840 */
841 /*SvGETMAGIC (string);*/
842 if (SvMAGICAL (string) || !SvPOK (string))
843 string = sv_2mortal (newSVsv (string));
844
845 SvUPGRADE (string, SVt_PV);
846
847 /* work around a bug in perl 5.10, which causes SvCUR to fail an
848 * assertion with -DDEBUGGING, although SvCUR is documented to
849 * return the xpv_cur field which certainly exists after upgrading.
850 * according to nicholas clark, calling SvPOK fixes this.
851 * But it doesn't fix it, so try another workaround, call SvPV_nolen
852 * and hope for the best.
853 * Damnit, SvPV_nolen still trips over yet another assertion. This
854 * assertion business is seriously broken, try yet another workaround
855 * for the broken -DDEBUGGING.
856 */
857 {
858#ifdef DEBUGGING
859 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
860#else
861 STRLEN offset = SvCUR (string);
862#endif
863
864 if (offset > cbor->max_size && cbor->max_size) 1170 if (len > cbor->max_size && cbor->max_size)
865 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu", 1171 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
866 (unsigned long)SvCUR (string), (unsigned long)cbor->max_size); 1172 (unsigned long)len, (unsigned long)cbor->max_size);
867 }
868
869 sv_utf8_downgrade (string, 0);
870 1173
871 dec.cbor = *cbor; 1174 dec.cbor = *cbor;
872 dec.cur = (U8 *)SvPVX (string); 1175 dec.cur = (U8 *)data;
873 dec.end = (U8 *)SvEND (string); 1176 dec.end = (U8 *)data + len;
874 dec.err = 0;
875 dec.depth = 0;
876 1177
877 sv = decode_sv (&dec); 1178 sv = decode_sv (&dec);
878 1179
879 if (offset_return) 1180 if (offset_return)
880 *offset_return = dec.cur; 1181 *offset_return = dec.cur;
883 if (dec.cur != dec.end && !dec.err) 1184 if (dec.cur != dec.end && !dec.err)
884 dec.err = "garbage after CBOR object"; 1185 dec.err = "garbage after CBOR object";
885 1186
886 if (dec.err) 1187 if (dec.err)
887 { 1188 {
1189 if (dec.shareable)
1190 {
1191 // need to break cyclic links, which whould all be in shareable
1192 int i;
1193 SV **svp;
1194
1195 for (i = av_len (dec.shareable) + 1; i--; )
1196 if ((svp = av_fetch (dec.shareable, i, 0)))
1197 sv_setsv (*svp, &PL_sv_undef);
1198 }
1199
888 SvREFCNT_dec (sv); 1200 SvREFCNT_dec (sv);
889 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)SvPVX (string), (int)(uint8_t)*dec.cur); 1201 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
890 } 1202 }
891 1203
892 sv = sv_2mortal (sv); 1204 sv = sv_2mortal (sv);
893 1205
894 return sv; 1206 return sv;
895} 1207}
896 1208
1209/////////////////////////////////////////////////////////////////////////////
1210// incremental parser
1211
1212#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1213
1214// returns 0 for notyet, 1 for success or error
1215static int
1216incr_parse (CBOR *self, SV *cborstr)
1217{
1218 STRLEN cur;
1219 SvPV (cborstr, cur);
1220
1221 while (ecb_expect_true (self->incr_need <= cur))
1222 {
1223 // table of integer count bytes
1224 static I8 incr_len[MINOR_MASK + 1] = {
1225 0, 0, 0, 0, 0, 0, 0, 0,
1226 0, 0, 0, 0, 0, 0, 0, 0,
1227 0, 0, 0, 0, 0, 0, 0, 0,
1228 1, 2, 4, 8,-1,-1,-1,-2
1229 };
1230
1231 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1232 U8 m = *p & MINOR_MASK;
1233 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1234 I8 ilen = incr_len[m];
1235
1236 self->incr_need = self->incr_pos + 1;
1237
1238 if (ecb_expect_false (ilen < 0))
1239 {
1240 if (m != MINOR_INDEF)
1241 return 1; // error
1242
1243 if (*p == (MAJOR_MISC | MINOR_INDEF))
1244 {
1245 if (count >= 0)
1246 return 1; // error
1247
1248 count = 1;
1249 }
1250 else
1251 {
1252 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1253 count = -1;
1254 }
1255 }
1256 else
1257 {
1258 self->incr_need += ilen;
1259 if (ecb_expect_false (self->incr_need > cur))
1260 return 0;
1261
1262 int major = *p >> MAJOR_SHIFT;
1263
1264 switch (major)
1265 {
1266 case MAJOR_BYTES >> MAJOR_SHIFT:
1267 case MAJOR_TEXT >> MAJOR_SHIFT:
1268 case MAJOR_ARRAY >> MAJOR_SHIFT:
1269 case MAJOR_MAP >> MAJOR_SHIFT:
1270 {
1271 UV len;
1272
1273 if (ecb_expect_false (ilen))
1274 {
1275 len = 0;
1276
1277 do {
1278 len = (len << 8) | *++p;
1279 } while (--ilen);
1280 }
1281 else
1282 len = m;
1283
1284 switch (major)
1285 {
1286 case MAJOR_BYTES >> MAJOR_SHIFT:
1287 case MAJOR_TEXT >> MAJOR_SHIFT:
1288 self->incr_need += len;
1289 if (ecb_expect_false (self->incr_need > cur))
1290 return 0;
1291
1292 break;
1293
1294 case MAJOR_MAP >> MAJOR_SHIFT:
1295 len <<= 1;
1296 case MAJOR_ARRAY >> MAJOR_SHIFT:
1297 if (len)
1298 {
1299 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1300 count = len + 1;
1301 }
1302 break;
1303 }
1304 }
1305 }
1306 }
1307
1308 self->incr_pos = self->incr_need;
1309
1310 if (count > 0)
1311 {
1312 while (!--count)
1313 {
1314 if (!AvFILLp (self->incr_count))
1315 return 1; // done
1316
1317 SvREFCNT_dec_NN (av_pop (self->incr_count));
1318 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1319 }
1320
1321 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1322 }
1323 }
1324
1325 return 0;
1326}
1327
1328
897///////////////////////////////////////////////////////////////////////////// 1329/////////////////////////////////////////////////////////////////////////////
898// XS interface functions 1330// XS interface functions
899 1331
900MODULE = CBOR::XS PACKAGE = CBOR::XS 1332MODULE = CBOR::XS PACKAGE = CBOR::XS
901 1333
908 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1); 1340 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
909 1341
910 types_true = get_bool ("Types::Serialiser::true" ); 1342 types_true = get_bool ("Types::Serialiser::true" );
911 types_false = get_bool ("Types::Serialiser::false"); 1343 types_false = get_bool ("Types::Serialiser::false");
912 types_error = get_bool ("Types::Serialiser::error"); 1344 types_error = get_bool ("Types::Serialiser::error");
1345
1346 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
913 1347
914 sv_cbor = newSVpv ("CBOR", 0); 1348 sv_cbor = newSVpv ("CBOR", 0);
915 SvREADONLY_on (sv_cbor); 1349 SvREADONLY_on (sv_cbor);
916} 1350}
917 1351
938 1372
939void shrink (CBOR *self, int enable = 1) 1373void shrink (CBOR *self, int enable = 1)
940 ALIAS: 1374 ALIAS:
941 shrink = F_SHRINK 1375 shrink = F_SHRINK
942 allow_unknown = F_ALLOW_UNKNOWN 1376 allow_unknown = F_ALLOW_UNKNOWN
1377 allow_sharing = F_ALLOW_SHARING
1378 allow_cycles = F_ALLOW_CYCLES
1379 pack_strings = F_PACK_STRINGS
1380 validate_utf8 = F_VALIDATE_UTF8
943 PPCODE: 1381 PPCODE:
944{ 1382{
945 if (enable) 1383 if (enable)
946 self->flags |= ix; 1384 self->flags |= ix;
947 else 1385 else
952 1390
953void get_shrink (CBOR *self) 1391void get_shrink (CBOR *self)
954 ALIAS: 1392 ALIAS:
955 get_shrink = F_SHRINK 1393 get_shrink = F_SHRINK
956 get_allow_unknown = F_ALLOW_UNKNOWN 1394 get_allow_unknown = F_ALLOW_UNKNOWN
1395 get_allow_sharing = F_ALLOW_SHARING
1396 get_allow_cycles = F_ALLOW_CYCLES
1397 get_pack_strings = F_PACK_STRINGS
1398 get_validate_utf8 = F_VALIDATE_UTF8
957 PPCODE: 1399 PPCODE:
958 XPUSHs (boolSV (self->flags & ix)); 1400 XPUSHs (boolSV (self->flags & ix));
959 1401
960void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1402void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
961 PPCODE: 1403 PPCODE:
977 CODE: 1419 CODE:
978 RETVAL = self->max_size; 1420 RETVAL = self->max_size;
979 OUTPUT: 1421 OUTPUT:
980 RETVAL 1422 RETVAL
981 1423
982#if 0 //TODO 1424void filter (CBOR *self, SV *filter = 0)
983
984void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef)
985 PPCODE: 1425 PPCODE:
986{
987 SvREFCNT_dec (self->cb_object); 1426 SvREFCNT_dec (self->filter);
988 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; 1427 self->filter = filter ? newSVsv (filter) : filter;
989
990 XPUSHs (ST (0)); 1428 XPUSHs (ST (0));
991}
992 1429
993void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef) 1430SV *get_filter (CBOR *self)
994 PPCODE: 1431 CODE:
995{ 1432 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
996 if (!self->cb_sk_object) 1433 OUTPUT:
997 self->cb_sk_object = newHV (); 1434 RETVAL
998
999 if (SvOK (cb))
1000 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1001 else
1002 {
1003 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1004
1005 if (!HvKEYS (self->cb_sk_object))
1006 {
1007 SvREFCNT_dec (self->cb_sk_object);
1008 self->cb_sk_object = 0;
1009 }
1010 }
1011
1012 XPUSHs (ST (0));
1013}
1014
1015#endif
1016 1435
1017void encode (CBOR *self, SV *scalar) 1436void encode (CBOR *self, SV *scalar)
1018 PPCODE: 1437 PPCODE:
1019 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; 1438 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
1020 XPUSHs (scalar); 1439 XPUSHs (scalar);
1033 EXTEND (SP, 2); 1452 EXTEND (SP, 2);
1034 PUSHs (sv); 1453 PUSHs (sv);
1035 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1454 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1036} 1455}
1037 1456
1038#if 0 1457void incr_parse (CBOR *self, SV *cborstr)
1458 ALIAS:
1459 incr_parse_multiple = 1
1460 PPCODE:
1461{
1462 if (SvUTF8 (cborstr))
1463 sv_utf8_downgrade (cborstr, 0);
1464
1465 if (!self->incr_count)
1466 {
1467 self->incr_count = newAV ();
1468 self->incr_pos = 0;
1469 self->incr_need = 1;
1470
1471 av_push (self->incr_count, newSViv (1));
1472 }
1473
1474 do
1475 {
1476 if (!incr_parse (self, cborstr))
1477 {
1478 if (self->incr_need > self->max_size && self->max_size)
1479 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1480 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1481
1482 break;
1483 }
1484
1485 SV *sv;
1486 char *offset;
1487
1488 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1489 XPUSHs (sv);
1490
1491 sv_chop (cborstr, offset);
1492
1493 av_clear (self->incr_count);
1494 av_push (self->incr_count, newSViv (1));
1495
1496 self->incr_pos = 0;
1497 self->incr_need = self->incr_pos + 1;
1498 }
1499 while (ix);
1500}
1501
1502void incr_reset (CBOR *self)
1503 CODE:
1504{
1505 SvREFCNT_dec (self->incr_count);
1506 self->incr_count = 0;
1507}
1039 1508
1040void DESTROY (CBOR *self) 1509void DESTROY (CBOR *self)
1041 CODE: 1510 PPCODE:
1042 SvREFCNT_dec (self->cb_sk_object); 1511 cbor_free (self);
1043 SvREFCNT_dec (self->cb_object);
1044
1045#endif
1046 1512
1047PROTOTYPES: ENABLE 1513PROTOTYPES: ENABLE
1048 1514
1049void encode_cbor (SV *scalar) 1515void encode_cbor (SV *scalar)
1516 ALIAS:
1517 encode_cbor = 0
1518 encode_cbor_sharing = F_ALLOW_SHARING
1050 PPCODE: 1519 PPCODE:
1051{ 1520{
1052 CBOR cbor; 1521 CBOR cbor;
1053 cbor_init (&cbor); 1522 cbor_init (&cbor);
1523 cbor.flags |= ix;
1054 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1524 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1055 XPUSHs (scalar); 1525 XPUSHs (scalar);
1056} 1526}
1057 1527
1058void decode_cbor (SV *cborstr) 1528void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines