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 1.10 by root, Sun Oct 27 20:40:25 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; 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
269 : gv_stashpv ("Types::Serialiser::Error", 1); 394 : gv_stashpv ("Types::Serialiser::Error", 1);
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
399 HV *stash = SvSTASH (sv);
400
274 if (SvSTASH (sv) == boolean_stash) 401 if (stash == boolean_stash)
275 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 }
276 else if (SvSTASH (sv) == error_stash) 406 else if (stash == error_stash)
277 encode_ch (enc, 0xe0 | 23); 407 {
408 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
409 return;
410 }
278 else if (SvSTASH (sv) == tagged_stash) 411 else if (stash == tagged_stash)
279 { 412 {
280 if (svt != SVt_PVAV) 413 if (svt != SVt_PVAV)
281 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 414 croak ("encountered CBOR::XS::Tagged object that isn't an array");
282 415
283 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 416 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
284 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))
432 {
433 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
434 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
435 return;
285 } 436 }
286 else 437 else
287 { 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
450 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
451 {
452 dSP;
453
454 ENTER; SAVETMPS; PUSHMARK (SP);
288 // 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
289 GV *to_cbor = gv_fetchmethod_autoload (SvSTASH (sv), "TO_CBOR", 0);
290
291 if (to_cbor)
292 {
293 dSP;
294
295 ENTER; SAVETMPS; PUSHMARK (SP);
296 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv))); 456 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
297 457
298 // calling with G_SCALAR ensures that we always get a 1 return value
299 PUTBACK; 458 PUTBACK;
459 // G_SCALAR ensures that return value is 1
300 call_sv ((SV *)GvCV (to_cbor), G_SCALAR); 460 call_sv ((SV *)GvCV (method), G_SCALAR);
301 SPAGAIN; 461 SPAGAIN;
302 462
303 // catch this surprisingly common error 463 // catch this surprisingly common error
304 if (SvROK (TOPs) && SvRV (TOPs) == sv) 464 if (SvROK (TOPs) && SvRV (TOPs) == sv)
305 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv))); 465 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (stash));
306 466
307 sv = POPs; 467 encode_sv (enc, POPs);
468
308 PUTBACK; 469 PUTBACK;
309 470
310 encode_sv (enc, sv);
311
312 FREETMPS; LEAVE; 471 FREETMPS; LEAVE;
313 } 472 }
473 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
474 {
475 dSP;
476
477 ENTER; SAVETMPS; PUSHMARK (SP);
478 EXTEND (SP, 2);
479 // we re-bless the reference to get overload and other niceties right
480 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
481 PUSHs (sv_cbor);
482
483 PUTBACK;
484 int count = call_sv ((SV *)GvCV (method), G_ARRAY);
485 SPAGAIN;
486
487 // catch this surprisingly common error
488 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
489 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
490
491 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
492 encode_uint (enc, MAJOR_ARRAY, count + 1);
493 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
494
495 while (count)
496 encode_sv (enc, SP[1 - count--]);
497
498 PUTBACK;
499
500 FREETMPS; LEAVE;
501 }
314 else 502 else
315 croak ("encountered object '%s', but no TO_CBOR method available on it", 503 croak ("encountered object '%s', but no TO_CBOR or FREEZE methods available on it",
316 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 504 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
317 }
318 } 505 }
319 else if (svt == SVt_PVHV) 506 else if (svt == SVt_PVHV)
320 encode_hv (enc, (HV *)sv); 507 encode_hv (enc, (HV *)sv);
321 else if (svt == SVt_PVAV) 508 else if (svt == SVt_PVAV)
322 encode_av (enc, (AV *)sv); 509 encode_av (enc, (AV *)sv);
323 else if (svt < SVt_PVAV)
324 {
325 STRLEN len = 0;
326 char *pv = svt ? SvPV (sv, len) : 0;
327
328 if (len == 1 && *pv == '1')
329 encode_ch (enc, 0xe0 | 21);
330 else if (len == 1 && *pv == '0')
331 encode_ch (enc, 0xe0 | 20);
332 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
333 encode_ch (enc, 0xe0 | 23);
334 else
335 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
336 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
337 }
338 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
339 encode_ch (enc, 0xe0 | 23);
340 else 510 else
341 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 511 {
342 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 512 encode_tag (enc, CBOR_TAG_INDIRECTION);
513 encode_sv (enc, sv);
514 }
343} 515}
344 516
345static void 517static void
346encode_nv (enc_t *enc, SV *sv) 518encode_nv (enc_t *enc, SV *sv)
347{ 519{
348 double nv = SvNVX (sv); 520 double nv = SvNVX (sv);
349 521
350 need (enc, 9); 522 need (enc, 9);
351 523
352 if (ecb_expect_false (nv == (U32)nv)) 524 if (ecb_expect_false (nv == (NV)(U32)nv))
353 encode_uint (enc, 0x00, (U32)nv); 525 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
354 //TODO: maybe I32? 526 //TODO: maybe I32?
355 else if (ecb_expect_false (nv == (float)nv)) 527 else if (ecb_expect_false (nv == (float)nv))
356 { 528 {
357 uint32_t fp = ecb_float_to_binary32 (nv); 529 uint32_t fp = ecb_float_to_binary32 (nv);
358 530
359 *enc->cur++ = 0xe0 | 26; 531 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
360 532
361 if (!ecb_big_endian ()) 533 if (!ecb_big_endian ())
362 fp = ecb_bswap32 (fp); 534 fp = ecb_bswap32 (fp);
363 535
364 memcpy (enc->cur, &fp, 4); 536 memcpy (enc->cur, &fp, 4);
366 } 538 }
367 else 539 else
368 { 540 {
369 uint64_t fp = ecb_double_to_binary64 (nv); 541 uint64_t fp = ecb_double_to_binary64 (nv);
370 542
371 *enc->cur++ = 0xe0 | 27; 543 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
372 544
373 if (!ecb_big_endian ()) 545 if (!ecb_big_endian ())
374 fp = ecb_bswap64 (fp); 546 fp = ecb_bswap64 (fp);
375 547
376 memcpy (enc->cur, &fp, 8); 548 memcpy (enc->cur, &fp, 8);
385 557
386 if (SvPOKp (sv)) 558 if (SvPOKp (sv))
387 { 559 {
388 STRLEN len; 560 STRLEN len;
389 char *str = SvPV (sv, len); 561 char *str = SvPV (sv, len);
390 encode_str (enc, SvUTF8 (sv), str, len); 562 encode_strref (enc, SvUTF8 (sv), str, len);
391 } 563 }
392 else if (SvNOKp (sv)) 564 else if (SvNOKp (sv))
393 encode_nv (enc, sv); 565 encode_nv (enc, sv);
394 else if (SvIOKp (sv)) 566 else if (SvIOKp (sv))
395 { 567 {
396 if (SvIsUV (sv)) 568 if (SvIsUV (sv))
397 encode_uint (enc, 0x00, SvUVX (sv)); 569 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
398 else if (SvIVX (sv) >= 0) 570 else if (SvIVX (sv) >= 0)
399 encode_uint (enc, 0x00, SvIVX (sv)); 571 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
400 else 572 else
401 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 573 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
402 } 574 }
403 else if (SvROK (sv)) 575 else if (SvROK (sv))
404 encode_rv (enc, SvRV (sv)); 576 encode_rv (enc, SvRV (sv));
405 else if (!SvOK (sv)) 577 else if (!SvOK (sv))
406 encode_ch (enc, 0xe0 | 22); 578 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
407 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 579 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
408 encode_ch (enc, 0xe0 | 23); 580 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
409 else 581 else
410 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",
411 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 583 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
412} 584}
413 585
414static SV * 586static SV *
415encode_cbor (SV *scalar, CBOR *cbor) 587encode_cbor (SV *scalar, CBOR *cbor)
416{ 588{
417 enc_t enc; 589 enc_t enc = { 0 };
418 590
419 enc.cbor = *cbor; 591 enc.cbor = *cbor;
420 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 592 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
421 enc.cur = SvPVX (enc.sv); 593 enc.cur = SvPVX (enc.sv);
422 enc.end = SvEND (enc.sv); 594 enc.end = SvEND (enc.sv);
423 enc.depth = 0;
424 595
425 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
426 encode_sv (&enc, scalar); 605 encode_sv (&enc, scalar);
427 606
428 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 607 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
429 *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
430 609
444 U8 *end; // end of input string 623 U8 *end; // end of input string
445 const char *err; // parse error, if != 0 624 const char *err; // parse error, if != 0
446 CBOR cbor; 625 CBOR cbor;
447 U32 depth; // recursion depth 626 U32 depth; // recursion depth
448 U32 maxdepth; // recursion depth limit 627 U32 maxdepth; // recursion depth limit
628 AV *shareable;
629 AV *stringref;
630 SV *decode_tagged;
449} dec_t; 631} dec_t;
450 632
451#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
452 634
453#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")
456#define DEC_DEC_DEPTH --dec->depth 638#define DEC_DEC_DEPTH --dec->depth
457 639
458static UV 640static UV
459decode_uint (dec_t *dec) 641decode_uint (dec_t *dec)
460{ 642{
461 switch (*dec->cur & 31) 643 U8 m = *dec->cur & MINOR_MASK;
462 { 644 ++dec->cur;
463 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
464 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
465 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
466 return *dec->cur++ & 31;
467 645
468 case 24: 646 if (ecb_expect_true (m < LENGTH_EXT1))
647 return m;
648 else if (ecb_expect_true (m == LENGTH_EXT1))
649 {
469 WANT (2); 650 WANT (1);
470 dec->cur += 2; 651 dec->cur += 1;
471 return dec->cur[-1]; 652 return dec->cur[-1];
472 653 }
473 case 25: 654 else if (ecb_expect_true (m == LENGTH_EXT2))
655 {
474 WANT (3); 656 WANT (2);
475 dec->cur += 3; 657 dec->cur += 2;
476 return (((UV)dec->cur[-2]) << 8) 658 return (((UV)dec->cur[-2]) << 8)
477 | ((UV)dec->cur[-1]); 659 | ((UV)dec->cur[-1]);
478 660 }
479 case 26: 661 else if (ecb_expect_true (m == LENGTH_EXT4))
662 {
480 WANT (5); 663 WANT (4);
481 dec->cur += 5; 664 dec->cur += 4;
482 return (((UV)dec->cur[-4]) << 24) 665 return (((UV)dec->cur[-4]) << 24)
483 | (((UV)dec->cur[-3]) << 16) 666 | (((UV)dec->cur[-3]) << 16)
484 | (((UV)dec->cur[-2]) << 8) 667 | (((UV)dec->cur[-2]) << 8)
485 | ((UV)dec->cur[-1]); 668 | ((UV)dec->cur[-1]);
486 669 }
487 case 27: 670 else if (ecb_expect_true (m == LENGTH_EXT8))
671 {
488 WANT (9); 672 WANT (8);
489 dec->cur += 9; 673 dec->cur += 8;
674
675 return
676#if UVSIZE < 8
677 0
678#else
490 return (((UV)dec->cur[-8]) << 56) 679 (((UV)dec->cur[-8]) << 56)
491 | (((UV)dec->cur[-7]) << 48) 680 | (((UV)dec->cur[-7]) << 48)
492 | (((UV)dec->cur[-6]) << 40) 681 | (((UV)dec->cur[-6]) << 40)
493 | (((UV)dec->cur[-5]) << 32) 682 | (((UV)dec->cur[-5]) << 32)
683#endif
494 | (((UV)dec->cur[-4]) << 24) 684 | (((UV)dec->cur[-4]) << 24)
495 | (((UV)dec->cur[-3]) << 16) 685 | (((UV)dec->cur[-3]) << 16)
496 | (((UV)dec->cur[-2]) << 8) 686 | (((UV)dec->cur[-2]) << 8)
497 | ((UV)dec->cur[-1]); 687 | ((UV)dec->cur[-1]);
498 688 }
499 default: 689 else
500 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 690 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
501 }
502 691
503fail: 692fail:
504 return 0; 693 return 0;
505} 694}
506 695
511{ 700{
512 AV *av = newAV (); 701 AV *av = newAV ();
513 702
514 DEC_INC_DEPTH; 703 DEC_INC_DEPTH;
515 704
516 if ((*dec->cur & 31) == 31) 705 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
517 { 706 {
518 ++dec->cur; 707 ++dec->cur;
519 708
520 for (;;) 709 for (;;)
521 { 710 {
522 WANT (1); 711 WANT (1);
523 712
524 if (*dec->cur == (0xe0 | 31)) 713 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
525 { 714 {
526 ++dec->cur; 715 ++dec->cur;
527 break; 716 break;
528 } 717 }
529 718
532 } 721 }
533 else 722 else
534 { 723 {
535 int i, len = decode_uint (dec); 724 int i, len = decode_uint (dec);
536 725
726 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
537 av_fill (av, len - 1); 727 av_fill (av, len - 1);
538 728
539 for (i = 0; i < len; ++i) 729 for (i = 0; i < len; ++i)
540 AvARRAY (av)[i] = decode_sv (dec); 730 AvARRAY (av)[i] = decode_sv (dec);
541 } 731 }
547 SvREFCNT_dec (av); 737 SvREFCNT_dec (av);
548 DEC_DEC_DEPTH; 738 DEC_DEC_DEPTH;
549 return &PL_sv_undef; 739 return &PL_sv_undef;
550} 740}
551 741
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
552static SV * 788static SV *
553decode_hv (dec_t *dec) 789decode_hv (dec_t *dec)
554{ 790{
555 HV *hv = newHV (); 791 HV *hv = newHV ();
556 792
557 DEC_INC_DEPTH; 793 DEC_INC_DEPTH;
558 794
559 if ((*dec->cur & 31) == 31) 795 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
560 { 796 {
561 ++dec->cur; 797 ++dec->cur;
562 798
563 for (;;) 799 for (;;)
564 { 800 {
565 WANT (1); 801 WANT (1);
566 802
567 if (*dec->cur == (0xe0 | 31)) 803 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
568 { 804 {
569 ++dec->cur; 805 ++dec->cur;
570 break; 806 break;
571 } 807 }
572 808
573 SV *k = decode_sv (dec); 809 decode_he (dec, hv);
574 SV *v = decode_sv (dec);
575
576 hv_store_ent (hv, k, v, 0);
577 } 810 }
578 } 811 }
579 else 812 else
580 { 813 {
581 int len = decode_uint (dec); 814 int pairs = decode_uint (dec);
582 815
583 while (len--) 816 while (pairs--)
584 { 817 decode_he (dec, hv);
585 SV *k = decode_sv (dec);
586 SV *v = decode_sv (dec);
587
588 hv_store_ent (hv, k, v, 0);
589 }
590 } 818 }
591 819
592 DEC_DEC_DEPTH; 820 DEC_DEC_DEPTH;
593 return newRV_noinc ((SV *)hv); 821 return newRV_noinc ((SV *)hv);
594 822
601static SV * 829static SV *
602decode_str (dec_t *dec, int utf8) 830decode_str (dec_t *dec, int utf8)
603{ 831{
604 SV *sv = 0; 832 SV *sv = 0;
605 833
606 if ((*dec->cur & 31) == 31) 834 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
607 { 835 {
836 // indefinite length strings
608 ++dec->cur; 837 ++dec->cur;
609 838
839 U8 major = *dec->cur & MAJOR_MISC;
840
610 sv = newSVpvn ("", 0); 841 sv = newSVpvn ("", 0);
611 842
612 // not very fast, and certainly not robust against illegal input
613 for (;;) 843 for (;;)
614 { 844 {
615 WANT (1); 845 WANT (1);
616 846
617 if (*dec->cur == (0xe0 | 31)) 847 if ((*dec->cur - major) > LENGTH_EXT8)
848 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
618 { 849 {
619 ++dec->cur; 850 ++dec->cur;
620 break; 851 break;
621 } 852 }
853 else
854 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
622 855
623 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;
624 } 861 }
625 } 862 }
626 else 863 else
627 { 864 {
628 STRLEN len = decode_uint (dec); 865 STRLEN len = decode_uint (dec);
629 866
630 WANT (len); 867 WANT (len);
631 sv = newSVpvn (dec->cur, len); 868 sv = newSVpvn (dec->cur, len);
632 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));
633 } 874 }
634 875
635 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
636 SvUTF8_on (sv); 882 SvUTF8_on (sv);
883 }
637 884
638 return sv; 885 return sv;
639 886
640fail: 887fail:
641 SvREFCNT_dec (sv); 888 SvREFCNT_dec (sv);
643} 890}
644 891
645static SV * 892static SV *
646decode_tagged (dec_t *dec) 893decode_tagged (dec_t *dec)
647{ 894{
895 SV *sv = 0;
648 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:
649 SV *sv = decode_sv (dec); 903 sv = decode_sv (dec);
904 break;
650 905
651 if (tag == CBOR_TAG_MAGIC) 906 case CBOR_TAG_INDIRECTION:
652 return sv; 907 sv = newRV_noinc (decode_sv (dec));
908 break;
653 909
654 if (tag == CBOR_TAG_PERL_OBJECT) 910 case CBOR_TAG_STRINGREF_NAMESPACE:
655 { 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
656 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 982 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
657 ERR ("corrupted CBOR data (non-array perl object)"); 983 ERR ("corrupted CBOR data (non-array perl object)");
984
985 AV *av = (AV *)SvRV (sv);
986 int len = av_len (av) + 1;
987 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
988
989 if (!stash)
990 ERR ("cannot decode perl-object (package does not exist)");
991
992 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
993
994 if (!method)
995 ERR ("cannot decode perl-object (package does not have a THAW method)");
996
997 dSP;
998
999 ENTER; SAVETMPS; PUSHMARK (SP);
1000 EXTEND (SP, len + 1);
1001 // we re-bless the reference to get overload and other niceties right
1002 PUSHs (*av_fetch (av, 0, 1));
1003 PUSHs (sv_cbor);
1004
1005 int i;
1006
1007 for (i = 1; i < len; ++i)
1008 PUSHs (*av_fetch (av, i, 1));
1009
1010 PUTBACK;
1011 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
1012 SPAGAIN;
1013
1014 if (SvTRUE (ERRSV))
1015 {
1016 FREETMPS; LEAVE;
1017 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1018 }
1019
1020 SvREFCNT_dec (sv);
1021 sv = SvREFCNT_inc (POPs);
1022
1023 PUTBACK;
1024
1025 FREETMPS; LEAVE;
658 1026 }
659 // TODO 1027 break;
660 }
661 1028
1029 default:
1030 {
1031 sv = decode_sv (dec);
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 {
662 AV *av = newAV (); 1056 AV *av = newAV ();
663 av_push (av, newSVuv (tag)); 1057 av_push (av, newSVuv (tag));
664 av_push (av, sv); 1058 av_push (av, sv);
665 1059
666 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1060 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
667 ? cbor_tagged_stash 1061 ? cbor_tagged_stash
668 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1062 : gv_stashpv ("CBOR::XS::Tagged" , 1);
669
670 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;
1071 }
1072
1073 return sv;
671 1074
672fail: 1075fail:
673 SvREFCNT_dec (sv); 1076 SvREFCNT_dec (sv);
674 return &PL_sv_undef; 1077 return &PL_sv_undef;
675} 1078}
677static SV * 1080static SV *
678decode_sv (dec_t *dec) 1081decode_sv (dec_t *dec)
679{ 1082{
680 WANT (1); 1083 WANT (1);
681 1084
682 switch (*dec->cur >> 5) 1085 switch (*dec->cur >> MAJOR_SHIFT)
683 { 1086 {
684 case 0: // unsigned int 1087 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
685 return newSVuv (decode_uint (dec)); 1088 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
686 case 1: // negative int 1089 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
687 return newSViv (-1 - (IV)decode_uint (dec)); 1090 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
688 case 2: // octet string 1091 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
689 return decode_str (dec, 0); 1092 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
690 case 3: // utf-8 string 1093 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
691 return decode_str (dec, 1); 1094
692 case 4: // array 1095 case MAJOR_MISC >> MAJOR_SHIFT:
693 return decode_av (dec);
694 case 5: // map
695 return decode_hv (dec);
696 case 6: // tag
697 return decode_tagged (dec);
698 case 7: // misc
699 switch (*dec->cur++ & 31) 1096 switch (*dec->cur++ & MINOR_MASK)
700 { 1097 {
701 case 20: 1098 case SIMPLE_FALSE:
702#if CBOR_SLOW 1099#if CBOR_SLOW
703 types_false = get_bool ("Types::Serialiser::false"); 1100 types_false = get_bool ("Types::Serialiser::false");
704#endif 1101#endif
705 return newSVsv (types_false); 1102 return newSVsv (types_false);
706 case 21: 1103 case SIMPLE_TRUE:
707#if CBOR_SLOW 1104#if CBOR_SLOW
708 types_true = get_bool ("Types::Serialiser::true"); 1105 types_true = get_bool ("Types::Serialiser::true");
709#endif 1106#endif
710 return newSVsv (types_true); 1107 return newSVsv (types_true);
711 case 22: 1108 case SIMPLE_NULL:
712 return newSVsv (&PL_sv_undef); 1109 return newSVsv (&PL_sv_undef);
713 case 23: 1110 case SIMPLE_UNDEF:
714#if CBOR_SLOW 1111#if CBOR_SLOW
715 types_error = get_bool ("Types::Serialiser::error"); 1112 types_error = get_bool ("Types::Serialiser::error");
716#endif 1113#endif
717 return newSVsv (types_error); 1114 return newSVsv (types_error);
718 1115
719 case 25: 1116 case MISC_FLOAT16:
720 { 1117 {
721 WANT (2); 1118 WANT (2);
722 1119
723 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1120 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
724 dec->cur += 2; 1121 dec->cur += 2;
725 1122
726 return newSVnv (ecb_binary16_to_float (fp)); 1123 return newSVnv (ecb_binary16_to_float (fp));
727 } 1124 }
728 1125
729 case 26: 1126 case MISC_FLOAT32:
730 { 1127 {
731 uint32_t fp; 1128 uint32_t fp;
732 WANT (4); 1129 WANT (4);
733 memcpy (&fp, dec->cur, 4); 1130 memcpy (&fp, dec->cur, 4);
734 dec->cur += 4; 1131 dec->cur += 4;
737 fp = ecb_bswap32 (fp); 1134 fp = ecb_bswap32 (fp);
738 1135
739 return newSVnv (ecb_binary32_to_float (fp)); 1136 return newSVnv (ecb_binary32_to_float (fp));
740 } 1137 }
741 1138
742 case 27: 1139 case MISC_FLOAT64:
743 { 1140 {
744 uint64_t fp; 1141 uint64_t fp;
745 WANT (8); 1142 WANT (8);
746 memcpy (&fp, dec->cur, 8); 1143 memcpy (&fp, dec->cur, 8);
747 dec->cur += 8; 1144 dec->cur += 8;
750 fp = ecb_bswap64 (fp); 1147 fp = ecb_bswap64 (fp);
751 1148
752 return newSVnv (ecb_binary64_to_double (fp)); 1149 return newSVnv (ecb_binary64_to_double (fp));
753 } 1150 }
754 1151
755 // 0..19 unassigned 1152 // 0..19 unassigned simple
756 // 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
757 default: 1156 default:
758 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1157 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
759 } 1158 }
760 1159
761 break; 1160 break;
762 } 1161 }
763 1162
766} 1165}
767 1166
768static SV * 1167static SV *
769decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1168decode_cbor (SV *string, CBOR *cbor, char **offset_return)
770{ 1169{
771 dec_t dec; 1170 dec_t dec = { 0 };
772 SV *sv; 1171 SV *sv;
1172 STRLEN len;
1173 char *data = SvPVbyte (string, len);
773 1174
774 /* work around bugs in 5.10 where manipulating magic values
775 * makes perl ignore the magic in subsequent accesses.
776 * also make a copy of non-PV values, to get them into a clean
777 * state (SvPV should do that, but it's buggy, see below).
778 */
779 /*SvGETMAGIC (string);*/
780 if (SvMAGICAL (string) || !SvPOK (string))
781 string = sv_2mortal (newSVsv (string));
782
783 SvUPGRADE (string, SVt_PV);
784
785 /* work around a bug in perl 5.10, which causes SvCUR to fail an
786 * assertion with -DDEBUGGING, although SvCUR is documented to
787 * return the xpv_cur field which certainly exists after upgrading.
788 * according to nicholas clark, calling SvPOK fixes this.
789 * But it doesn't fix it, so try another workaround, call SvPV_nolen
790 * and hope for the best.
791 * Damnit, SvPV_nolen still trips over yet another assertion. This
792 * assertion business is seriously broken, try yet another workaround
793 * for the broken -DDEBUGGING.
794 */
795 {
796#ifdef DEBUGGING
797 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
798#else
799 STRLEN offset = SvCUR (string);
800#endif
801
802 if (offset > cbor->max_size && cbor->max_size) 1175 if (len > cbor->max_size && cbor->max_size)
803 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",
804 (unsigned long)SvCUR (string), (unsigned long)cbor->max_size); 1177 (unsigned long)len, (unsigned long)cbor->max_size);
805 }
806
807 sv_utf8_downgrade (string, 0);
808 1178
809 dec.cbor = *cbor; 1179 dec.cbor = *cbor;
810 dec.cur = (U8 *)SvPVX (string); 1180 dec.cur = (U8 *)data;
811 dec.end = (U8 *)SvEND (string); 1181 dec.end = (U8 *)data + len;
812 dec.err = 0;
813 dec.depth = 0;
814 1182
815 sv = decode_sv (&dec); 1183 sv = decode_sv (&dec);
816 1184
817 if (offset_return) 1185 if (offset_return)
818 *offset_return = dec.cur; 1186 *offset_return = dec.cur;
821 if (dec.cur != dec.end && !dec.err) 1189 if (dec.cur != dec.end && !dec.err)
822 dec.err = "garbage after CBOR object"; 1190 dec.err = "garbage after CBOR object";
823 1191
824 if (dec.err) 1192 if (dec.err)
825 { 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
826 SvREFCNT_dec (sv); 1205 SvREFCNT_dec (sv);
827 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);
828 } 1207 }
829 1208
830 sv = sv_2mortal (sv); 1209 sv = sv_2mortal (sv);
831 1210
832 return sv; 1211 return sv;
833} 1212}
834 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
835///////////////////////////////////////////////////////////////////////////// 1338/////////////////////////////////////////////////////////////////////////////
836// XS interface functions 1339// XS interface functions
837 1340
838MODULE = CBOR::XS PACKAGE = CBOR::XS 1341MODULE = CBOR::XS PACKAGE = CBOR::XS
839 1342
846 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1); 1349 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
847 1350
848 types_true = get_bool ("Types::Serialiser::true" ); 1351 types_true = get_bool ("Types::Serialiser::true" );
849 types_false = get_bool ("Types::Serialiser::false"); 1352 types_false = get_bool ("Types::Serialiser::false");
850 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);
1356
1357 sv_cbor = newSVpv ("CBOR", 0);
1358 SvREADONLY_on (sv_cbor);
851} 1359}
852 1360
853PROTOTYPES: DISABLE 1361PROTOTYPES: DISABLE
854 1362
855void CLONE (...) 1363void CLONE (...)
873 1381
874void shrink (CBOR *self, int enable = 1) 1382void shrink (CBOR *self, int enable = 1)
875 ALIAS: 1383 ALIAS:
876 shrink = F_SHRINK 1384 shrink = F_SHRINK
877 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
878 PPCODE: 1390 PPCODE:
879{ 1391{
880 if (enable) 1392 if (enable)
881 self->flags |= ix; 1393 self->flags |= ix;
882 else 1394 else
887 1399
888void get_shrink (CBOR *self) 1400void get_shrink (CBOR *self)
889 ALIAS: 1401 ALIAS:
890 get_shrink = F_SHRINK 1402 get_shrink = F_SHRINK
891 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
892 PPCODE: 1408 PPCODE:
893 XPUSHs (boolSV (self->flags & ix)); 1409 XPUSHs (boolSV (self->flags & ix));
894 1410
895void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1411void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
896 PPCODE: 1412 PPCODE:
912 CODE: 1428 CODE:
913 RETVAL = self->max_size; 1429 RETVAL = self->max_size;
914 OUTPUT: 1430 OUTPUT:
915 RETVAL 1431 RETVAL
916 1432
917#if 0 //TODO 1433void filter (CBOR *self, SV *filter = 0)
918
919void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef)
920 PPCODE: 1434 PPCODE:
921{
922 SvREFCNT_dec (self->cb_object); 1435 SvREFCNT_dec (self->filter);
923 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; 1436 self->filter = filter ? newSVsv (filter) : filter;
924
925 XPUSHs (ST (0)); 1437 XPUSHs (ST (0));
926}
927 1438
928void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef) 1439SV *get_filter (CBOR *self)
929 PPCODE: 1440 CODE:
930{ 1441 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
931 if (!self->cb_sk_object) 1442 OUTPUT:
932 self->cb_sk_object = newHV (); 1443 RETVAL
933
934 if (SvOK (cb))
935 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
936 else
937 {
938 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
939
940 if (!HvKEYS (self->cb_sk_object))
941 {
942 SvREFCNT_dec (self->cb_sk_object);
943 self->cb_sk_object = 0;
944 }
945 }
946
947 XPUSHs (ST (0));
948}
949
950#endif
951 1444
952void encode (CBOR *self, SV *scalar) 1445void encode (CBOR *self, SV *scalar)
953 PPCODE: 1446 PPCODE:
954 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; 1447 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
955 XPUSHs (scalar); 1448 XPUSHs (scalar);
968 EXTEND (SP, 2); 1461 EXTEND (SP, 2);
969 PUSHs (sv); 1462 PUSHs (sv);
970 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1463 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
971} 1464}
972 1465
973#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}
974 1517
975void DESTROY (CBOR *self) 1518void DESTROY (CBOR *self)
976 CODE: 1519 PPCODE:
977 SvREFCNT_dec (self->cb_sk_object); 1520 cbor_free (self);
978 SvREFCNT_dec (self->cb_object);
979
980#endif
981 1521
982PROTOTYPES: ENABLE 1522PROTOTYPES: ENABLE
983 1523
984void encode_cbor (SV *scalar) 1524void encode_cbor (SV *scalar)
1525 ALIAS:
1526 encode_cbor = 0
1527 encode_cbor_sharing = F_ALLOW_SHARING
985 PPCODE: 1528 PPCODE:
986{ 1529{
987 CBOR cbor; 1530 CBOR cbor;
988 cbor_init (&cbor); 1531 cbor_init (&cbor);
1532 cbor.flags |= ix;
989 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1533 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
990 XPUSHs (scalar); 1534 XPUSHs (scalar);
991} 1535}
992 1536
993void decode_cbor (SV *cborstr) 1537void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines