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 1.9 by root, Sun Oct 27 10:17:12 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)
60#else 121#else
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, *cbor_boolean_stash, *cbor_tagged_stash; // CBOR::XS:: 126static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
66static SV *cbor_true, *cbor_false; 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 || cbor_boolean_stash 386 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
265 ? cbor_boolean_stash 387 ? types_boolean_stash
266 : gv_stashpv ("CBOR::XS::Boolean", 1); 388 : gv_stashpv ("Types::Serialiser::Boolean", 1);
389 HV *error_stash = !CBOR_SLOW || types_error_stash
390 ? types_error_stash
391 : gv_stashpv ("Types::Serialiser::Error", 1);
267 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 392 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
268 ? cbor_tagged_stash 393 ? cbor_tagged_stash
269 : gv_stashpv ("CBOR::XS::Tagged" , 1); 394 : gv_stashpv ("CBOR::XS::Tagged" , 1);
270 395
396 HV *stash = SvSTASH (sv);
397
271 if (SvSTASH (sv) == boolean_stash) 398 if (stash == boolean_stash)
272 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 }
403 else if (stash == error_stash)
404 {
405 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
406 return;
407 }
273 else if (SvSTASH (sv) == tagged_stash) 408 else if (stash == tagged_stash)
274 { 409 {
275 if (svt != SVt_PVAV) 410 if (svt != SVt_PVAV)
276 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 411 croak ("encountered CBOR::XS::Tagged object that isn't an array");
277 412
278 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 413 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
279 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))
429 {
430 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
431 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
432 return;
280 } 433 }
281 else 434 else
282 { 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
447 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
448 {
449 dSP;
450
451 ENTER; SAVETMPS; PUSHMARK (SP);
283 // 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
284 GV *to_cbor = gv_fetchmethod_autoload (SvSTASH (sv), "TO_CBOR", 0);
285
286 if (to_cbor)
287 {
288 dSP;
289
290 ENTER; SAVETMPS; PUSHMARK (SP);
291 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv))); 453 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
292 454
293 // calling with G_SCALAR ensures that we always get a 1 return value
294 PUTBACK; 455 PUTBACK;
456 // G_SCALAR ensures that return value is 1
295 call_sv ((SV *)GvCV (to_cbor), G_SCALAR); 457 call_sv ((SV *)GvCV (method), G_SCALAR);
296 SPAGAIN; 458 SPAGAIN;
297 459
298 // catch this surprisingly common error 460 // catch this surprisingly common error
299 if (SvROK (TOPs) && SvRV (TOPs) == sv) 461 if (SvROK (TOPs) && SvRV (TOPs) == sv)
300 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv))); 462 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (stash));
301 463
302 sv = POPs; 464 encode_sv (enc, POPs);
465
303 PUTBACK; 466 PUTBACK;
304 467
305 encode_sv (enc, sv);
306
307 FREETMPS; LEAVE; 468 FREETMPS; LEAVE;
308 } 469 }
470 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
471 {
472 dSP;
473
474 ENTER; SAVETMPS; PUSHMARK (SP);
475 EXTEND (SP, 2);
476 // we re-bless the reference to get overload and other niceties right
477 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
478 PUSHs (sv_cbor);
479
480 PUTBACK;
481 int count = call_sv ((SV *)GvCV (method), G_ARRAY);
482 SPAGAIN;
483
484 // catch this surprisingly common error
485 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
486 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
487
488 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
489 encode_uint (enc, MAJOR_ARRAY, count + 1);
490 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
491
492 while (count)
493 encode_sv (enc, SP[1 - count--]);
494
495 PUTBACK;
496
497 FREETMPS; LEAVE;
498 }
309 else 499 else
310 croak ("encountered object '%s', but no TO_CBOR method available on it", 500 croak ("encountered object '%s', but no TO_CBOR or FREEZE methods available on it",
311 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 501 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
312 }
313 } 502 }
314 else if (svt == SVt_PVHV) 503 else if (svt == SVt_PVHV)
315 encode_hv (enc, (HV *)sv); 504 encode_hv (enc, (HV *)sv);
316 else if (svt == SVt_PVAV) 505 else if (svt == SVt_PVAV)
317 encode_av (enc, (AV *)sv); 506 encode_av (enc, (AV *)sv);
318 else if (svt < SVt_PVAV)
319 {
320 STRLEN len = 0;
321 char *pv = svt ? SvPV (sv, len) : 0;
322
323 if (len == 1 && *pv == '1')
324 encode_ch (enc, 0xe0 | 21);
325 else if (len == 1 && *pv == '0')
326 encode_ch (enc, 0xe0 | 20);
327 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
328 encode_ch (enc, 0xe0 | 23);
329 else
330 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
331 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
332 }
333 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
334 encode_ch (enc, 0xe0 | 23);
335 else 507 else
336 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 508 {
337 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 509 encode_tag (enc, CBOR_TAG_INDIRECTION);
510 encode_sv (enc, sv);
511 }
338} 512}
339 513
340static void 514static void
341encode_nv (enc_t *enc, SV *sv) 515encode_nv (enc_t *enc, SV *sv)
342{ 516{
343 double nv = SvNVX (sv); 517 double nv = SvNVX (sv);
344 518
345 need (enc, 9); 519 need (enc, 9);
346 520
347 if (ecb_expect_false (nv == (U32)nv)) 521 if (ecb_expect_false (nv == (NV)(U32)nv))
348 encode_uint (enc, 0x00, (U32)nv); 522 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
349 //TODO: maybe I32? 523 //TODO: maybe I32?
350 else if (ecb_expect_false (nv == (float)nv)) 524 else if (ecb_expect_false (nv == (float)nv))
351 { 525 {
352 uint32_t fp = ecb_float_to_binary32 (nv); 526 uint32_t fp = ecb_float_to_binary32 (nv);
353 527
354 *enc->cur++ = 0xe0 | 26; 528 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
355 529
356 if (!ecb_big_endian ()) 530 if (!ecb_big_endian ())
357 fp = ecb_bswap32 (fp); 531 fp = ecb_bswap32 (fp);
358 532
359 memcpy (enc->cur, &fp, 4); 533 memcpy (enc->cur, &fp, 4);
361 } 535 }
362 else 536 else
363 { 537 {
364 uint64_t fp = ecb_double_to_binary64 (nv); 538 uint64_t fp = ecb_double_to_binary64 (nv);
365 539
366 *enc->cur++ = 0xe0 | 27; 540 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
367 541
368 if (!ecb_big_endian ()) 542 if (!ecb_big_endian ())
369 fp = ecb_bswap64 (fp); 543 fp = ecb_bswap64 (fp);
370 544
371 memcpy (enc->cur, &fp, 8); 545 memcpy (enc->cur, &fp, 8);
380 554
381 if (SvPOKp (sv)) 555 if (SvPOKp (sv))
382 { 556 {
383 STRLEN len; 557 STRLEN len;
384 char *str = SvPV (sv, len); 558 char *str = SvPV (sv, len);
385 encode_str (enc, SvUTF8 (sv), str, len); 559 encode_strref (enc, SvUTF8 (sv), str, len);
386 } 560 }
387 else if (SvNOKp (sv)) 561 else if (SvNOKp (sv))
388 encode_nv (enc, sv); 562 encode_nv (enc, sv);
389 else if (SvIOKp (sv)) 563 else if (SvIOKp (sv))
390 { 564 {
391 if (SvIsUV (sv)) 565 if (SvIsUV (sv))
392 encode_uint (enc, 0x00, SvUVX (sv)); 566 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
393 else if (SvIVX (sv) >= 0) 567 else if (SvIVX (sv) >= 0)
394 encode_uint (enc, 0x00, SvIVX (sv)); 568 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
395 else 569 else
396 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 570 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
397 } 571 }
398 else if (SvROK (sv)) 572 else if (SvROK (sv))
399 encode_rv (enc, SvRV (sv)); 573 encode_rv (enc, SvRV (sv));
400 else if (!SvOK (sv)) 574 else if (!SvOK (sv))
401 encode_ch (enc, 0xe0 | 22); 575 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
402 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 576 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
403 encode_ch (enc, 0xe0 | 23); 577 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
404 else 578 else
405 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",
406 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 580 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
407} 581}
408 582
409static SV * 583static SV *
410encode_cbor (SV *scalar, CBOR *cbor) 584encode_cbor (SV *scalar, CBOR *cbor)
411{ 585{
412 enc_t enc; 586 enc_t enc = { };
413 587
414 enc.cbor = *cbor; 588 enc.cbor = *cbor;
415 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 589 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
416 enc.cur = SvPVX (enc.sv); 590 enc.cur = SvPVX (enc.sv);
417 enc.end = SvEND (enc.sv); 591 enc.end = SvEND (enc.sv);
418 enc.depth = 0;
419 592
420 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
421 encode_sv (&enc, scalar); 602 encode_sv (&enc, scalar);
422 603
423 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 604 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
424 *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
425 606
439 U8 *end; // end of input string 620 U8 *end; // end of input string
440 const char *err; // parse error, if != 0 621 const char *err; // parse error, if != 0
441 CBOR cbor; 622 CBOR cbor;
442 U32 depth; // recursion depth 623 U32 depth; // recursion depth
443 U32 maxdepth; // recursion depth limit 624 U32 maxdepth; // recursion depth limit
625 AV *shareable;
626 AV *stringref;
627 SV *decode_tagged;
444} dec_t; 628} dec_t;
445 629
446#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
447 631
448#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")
451#define DEC_DEC_DEPTH --dec->depth 635#define DEC_DEC_DEPTH --dec->depth
452 636
453static UV 637static UV
454decode_uint (dec_t *dec) 638decode_uint (dec_t *dec)
455{ 639{
456 switch (*dec->cur & 31) 640 U8 m = *dec->cur & MINOR_MASK;
457 { 641 ++dec->cur;
458 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
459 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
460 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
461 return *dec->cur++ & 31;
462 642
463 case 24: 643 if (ecb_expect_true (m < LENGTH_EXT1))
644 return m;
645 else if (ecb_expect_true (m == LENGTH_EXT1))
646 {
464 WANT (2); 647 WANT (1);
465 dec->cur += 2; 648 dec->cur += 1;
466 return dec->cur[-1]; 649 return dec->cur[-1];
467 650 }
468 case 25: 651 else if (ecb_expect_true (m == LENGTH_EXT2))
652 {
469 WANT (3); 653 WANT (2);
470 dec->cur += 3; 654 dec->cur += 2;
471 return (((UV)dec->cur[-2]) << 8) 655 return (((UV)dec->cur[-2]) << 8)
472 | ((UV)dec->cur[-1]); 656 | ((UV)dec->cur[-1]);
473 657 }
474 case 26: 658 else if (ecb_expect_true (m == LENGTH_EXT4))
659 {
475 WANT (5); 660 WANT (4);
476 dec->cur += 5; 661 dec->cur += 4;
477 return (((UV)dec->cur[-4]) << 24) 662 return (((UV)dec->cur[-4]) << 24)
478 | (((UV)dec->cur[-3]) << 16) 663 | (((UV)dec->cur[-3]) << 16)
479 | (((UV)dec->cur[-2]) << 8) 664 | (((UV)dec->cur[-2]) << 8)
480 | ((UV)dec->cur[-1]); 665 | ((UV)dec->cur[-1]);
481 666 }
482 case 27: 667 else if (ecb_expect_true (m == LENGTH_EXT8))
668 {
483 WANT (9); 669 WANT (8);
484 dec->cur += 9; 670 dec->cur += 8;
671
672 return
673#if UVSIZE < 8
674 0
675#else
485 return (((UV)dec->cur[-8]) << 56) 676 (((UV)dec->cur[-8]) << 56)
486 | (((UV)dec->cur[-7]) << 48) 677 | (((UV)dec->cur[-7]) << 48)
487 | (((UV)dec->cur[-6]) << 40) 678 | (((UV)dec->cur[-6]) << 40)
488 | (((UV)dec->cur[-5]) << 32) 679 | (((UV)dec->cur[-5]) << 32)
680#endif
489 | (((UV)dec->cur[-4]) << 24) 681 | (((UV)dec->cur[-4]) << 24)
490 | (((UV)dec->cur[-3]) << 16) 682 | (((UV)dec->cur[-3]) << 16)
491 | (((UV)dec->cur[-2]) << 8) 683 | (((UV)dec->cur[-2]) << 8)
492 | ((UV)dec->cur[-1]); 684 | ((UV)dec->cur[-1]);
493 685 }
494 default: 686 else
495 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 687 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
496 }
497 688
498fail: 689fail:
499 return 0; 690 return 0;
500} 691}
501 692
506{ 697{
507 AV *av = newAV (); 698 AV *av = newAV ();
508 699
509 DEC_INC_DEPTH; 700 DEC_INC_DEPTH;
510 701
511 if ((*dec->cur & 31) == 31) 702 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
512 { 703 {
513 ++dec->cur; 704 ++dec->cur;
514 705
515 for (;;) 706 for (;;)
516 { 707 {
517 WANT (1); 708 WANT (1);
518 709
519 if (*dec->cur == (0xe0 | 31)) 710 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
520 { 711 {
521 ++dec->cur; 712 ++dec->cur;
522 break; 713 break;
523 } 714 }
524 715
527 } 718 }
528 else 719 else
529 { 720 {
530 int i, len = decode_uint (dec); 721 int i, len = decode_uint (dec);
531 722
723 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
532 av_fill (av, len - 1); 724 av_fill (av, len - 1);
533 725
534 for (i = 0; i < len; ++i) 726 for (i = 0; i < len; ++i)
535 AvARRAY (av)[i] = decode_sv (dec); 727 AvARRAY (av)[i] = decode_sv (dec);
536 } 728 }
542 SvREFCNT_dec (av); 734 SvREFCNT_dec (av);
543 DEC_DEC_DEPTH; 735 DEC_DEC_DEPTH;
544 return &PL_sv_undef; 736 return &PL_sv_undef;
545} 737}
546 738
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
547static SV * 783static SV *
548decode_hv (dec_t *dec) 784decode_hv (dec_t *dec)
549{ 785{
550 HV *hv = newHV (); 786 HV *hv = newHV ();
551 787
552 DEC_INC_DEPTH; 788 DEC_INC_DEPTH;
553 789
554 if ((*dec->cur & 31) == 31) 790 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
555 { 791 {
556 ++dec->cur; 792 ++dec->cur;
557 793
558 for (;;) 794 for (;;)
559 { 795 {
560 WANT (1); 796 WANT (1);
561 797
562 if (*dec->cur == (0xe0 | 31)) 798 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
563 { 799 {
564 ++dec->cur; 800 ++dec->cur;
565 break; 801 break;
566 } 802 }
567 803
568 SV *k = decode_sv (dec); 804 decode_he (dec, hv);
569 SV *v = decode_sv (dec);
570
571 hv_store_ent (hv, k, v, 0);
572 } 805 }
573 } 806 }
574 else 807 else
575 { 808 {
576 int len = decode_uint (dec); 809 int pairs = decode_uint (dec);
577 810
578 while (len--) 811 while (pairs--)
579 { 812 decode_he (dec, hv);
580 SV *k = decode_sv (dec);
581 SV *v = decode_sv (dec);
582
583 hv_store_ent (hv, k, v, 0);
584 }
585 } 813 }
586 814
587 DEC_DEC_DEPTH; 815 DEC_DEC_DEPTH;
588 return newRV_noinc ((SV *)hv); 816 return newRV_noinc ((SV *)hv);
589 817
596static SV * 824static SV *
597decode_str (dec_t *dec, int utf8) 825decode_str (dec_t *dec, int utf8)
598{ 826{
599 SV *sv = 0; 827 SV *sv = 0;
600 828
601 if ((*dec->cur & 31) == 31) 829 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
602 { 830 {
831 // indefinite length strings
603 ++dec->cur; 832 ++dec->cur;
604 833
834 U8 major = *dec->cur & MAJOR_MISC;
835
605 sv = newSVpvn ("", 0); 836 sv = newSVpvn ("", 0);
606 837
607 // not very fast, and certainly not robust against illegal input
608 for (;;) 838 for (;;)
609 { 839 {
610 WANT (1); 840 WANT (1);
611 841
612 if (*dec->cur == (0xe0 | 31)) 842 if ((*dec->cur - major) > LENGTH_EXT8)
843 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
613 { 844 {
614 ++dec->cur; 845 ++dec->cur;
615 break; 846 break;
616 } 847 }
848 else
849 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
617 850
618 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;
619 } 856 }
620 } 857 }
621 else 858 else
622 { 859 {
623 STRLEN len = decode_uint (dec); 860 STRLEN len = decode_uint (dec);
624 861
625 WANT (len); 862 WANT (len);
626 sv = newSVpvn (dec->cur, len); 863 sv = newSVpvn (dec->cur, len);
627 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));
628 } 869 }
629 870
630 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
631 SvUTF8_on (sv); 877 SvUTF8_on (sv);
878 }
632 879
633 return sv; 880 return sv;
634 881
635fail: 882fail:
636 SvREFCNT_dec (sv); 883 SvREFCNT_dec (sv);
638} 885}
639 886
640static SV * 887static SV *
641decode_tagged (dec_t *dec) 888decode_tagged (dec_t *dec)
642{ 889{
890 SV *sv = 0;
643 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:
644 SV *sv = decode_sv (dec); 898 sv = decode_sv (dec);
899 break;
645 900
646 if (tag == CBOR_TAG_MAGIC) 901 case CBOR_TAG_INDIRECTION:
647 return sv; 902 sv = newRV_noinc (decode_sv (dec));
903 break;
648 904
649 if (tag == CBOR_TAG_PERL_OBJECT) 905 case CBOR_TAG_STRINGREF_NAMESPACE:
650 { 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
651 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 977 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
652 ERR ("corrupted CBOR data (non-array perl object)"); 978 ERR ("corrupted CBOR data (non-array perl object)");
979
980 AV *av = (AV *)SvRV (sv);
981 int len = av_len (av) + 1;
982 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
983
984 if (!stash)
985 ERR ("cannot decode perl-object (package does not exist)");
986
987 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
988
989 if (!method)
990 ERR ("cannot decode perl-object (package does not have a THAW method)");
991
992 dSP;
993
994 ENTER; SAVETMPS; PUSHMARK (SP);
995 EXTEND (SP, len + 1);
996 // we re-bless the reference to get overload and other niceties right
997 PUSHs (*av_fetch (av, 0, 1));
998 PUSHs (sv_cbor);
999
1000 int i;
1001
1002 for (i = 1; i < len; ++i)
1003 PUSHs (*av_fetch (av, i, 1));
1004
1005 PUTBACK;
1006 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
1007 SPAGAIN;
1008
1009 if (SvTRUE (ERRSV))
1010 {
1011 FREETMPS; LEAVE;
1012 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1013 }
1014
1015 SvREFCNT_dec (sv);
1016 sv = SvREFCNT_inc (POPs);
1017
1018 PUTBACK;
1019
1020 FREETMPS; LEAVE;
653 1021 }
654 // TODO 1022 break;
655 }
656 1023
1024 default:
1025 {
1026 sv = decode_sv (dec);
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 {
657 AV *av = newAV (); 1051 AV *av = newAV ();
658 av_push (av, newSVuv (tag)); 1052 av_push (av, newSVuv (tag));
659 av_push (av, sv); 1053 av_push (av, sv);
660 1054
661 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1055 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
662 ? cbor_tagged_stash 1056 ? cbor_tagged_stash
663 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1057 : gv_stashpv ("CBOR::XS::Tagged" , 1);
664
665 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;
1066 }
1067
1068 return sv;
666 1069
667fail: 1070fail:
668 SvREFCNT_dec (sv); 1071 SvREFCNT_dec (sv);
669 return &PL_sv_undef; 1072 return &PL_sv_undef;
670} 1073}
672static SV * 1075static SV *
673decode_sv (dec_t *dec) 1076decode_sv (dec_t *dec)
674{ 1077{
675 WANT (1); 1078 WANT (1);
676 1079
677 switch (*dec->cur >> 5) 1080 switch (*dec->cur >> MAJOR_SHIFT)
678 { 1081 {
679 case 0: // unsigned int 1082 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
680 return newSVuv (decode_uint (dec)); 1083 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
681 case 1: // negative int 1084 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
682 return newSViv (-1 - (IV)decode_uint (dec)); 1085 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
683 case 2: // octet string 1086 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
684 return decode_str (dec, 0); 1087 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
685 case 3: // utf-8 string 1088 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
686 return decode_str (dec, 1); 1089
687 case 4: // array 1090 case MAJOR_MISC >> MAJOR_SHIFT:
688 return decode_av (dec);
689 case 5: // map
690 return decode_hv (dec);
691 case 6: // tag
692 return decode_tagged (dec);
693 case 7: // misc
694 switch (*dec->cur++ & 31) 1091 switch (*dec->cur++ & MINOR_MASK)
695 { 1092 {
696 case 20: 1093 case SIMPLE_FALSE:
697#if CBOR_SLOW 1094#if CBOR_SLOW
698 cbor_false = get_bool ("CBOR::XS::false"); 1095 types_false = get_bool ("Types::Serialiser::false");
699#endif 1096#endif
700 return newSVsv (cbor_false); 1097 return newSVsv (types_false);
701 case 21: 1098 case SIMPLE_TRUE:
702#if CBOR_SLOW 1099#if CBOR_SLOW
703 cbor_true = get_bool ("CBOR::XS::true"); 1100 types_true = get_bool ("Types::Serialiser::true");
704#endif 1101#endif
705 return newSVsv (cbor_true); 1102 return newSVsv (types_true);
706 case 22: 1103 case SIMPLE_NULL:
707 return newSVsv (&PL_sv_undef); 1104 return newSVsv (&PL_sv_undef);
1105 case SIMPLE_UNDEF:
1106#if CBOR_SLOW
1107 types_error = get_bool ("Types::Serialiser::error");
1108#endif
1109 return newSVsv (types_error);
708 1110
709 case 25: 1111 case MISC_FLOAT16:
710 { 1112 {
711 WANT (2); 1113 WANT (2);
712 1114
713 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1115 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
714 dec->cur += 2; 1116 dec->cur += 2;
715 1117
716 return newSVnv (ecb_binary16_to_float (fp)); 1118 return newSVnv (ecb_binary16_to_float (fp));
717 } 1119 }
718 1120
719 case 26: 1121 case MISC_FLOAT32:
720 { 1122 {
721 uint32_t fp; 1123 uint32_t fp;
722 WANT (4); 1124 WANT (4);
723 memcpy (&fp, dec->cur, 4); 1125 memcpy (&fp, dec->cur, 4);
724 dec->cur += 4; 1126 dec->cur += 4;
727 fp = ecb_bswap32 (fp); 1129 fp = ecb_bswap32 (fp);
728 1130
729 return newSVnv (ecb_binary32_to_float (fp)); 1131 return newSVnv (ecb_binary32_to_float (fp));
730 } 1132 }
731 1133
732 case 27: 1134 case MISC_FLOAT64:
733 { 1135 {
734 uint64_t fp; 1136 uint64_t fp;
735 WANT (8); 1137 WANT (8);
736 memcpy (&fp, dec->cur, 8); 1138 memcpy (&fp, dec->cur, 8);
737 dec->cur += 8; 1139 dec->cur += 8;
740 fp = ecb_bswap64 (fp); 1142 fp = ecb_bswap64 (fp);
741 1143
742 return newSVnv (ecb_binary64_to_double (fp)); 1144 return newSVnv (ecb_binary64_to_double (fp));
743 } 1145 }
744 1146
745 // 0..19 unassigned 1147 // 0..19 unassigned simple
746 // 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
747 default: 1151 default:
748 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1152 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
749 } 1153 }
750 1154
751 break; 1155 break;
752 } 1156 }
753 1157
756} 1160}
757 1161
758static SV * 1162static SV *
759decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1163decode_cbor (SV *string, CBOR *cbor, char **offset_return)
760{ 1164{
761 dec_t dec; 1165 dec_t dec = { };
762 SV *sv; 1166 SV *sv;
1167 STRLEN len;
1168 char *data = SvPVbyte (string, len);
763 1169
764 /* work around bugs in 5.10 where manipulating magic values
765 * makes perl ignore the magic in subsequent accesses.
766 * also make a copy of non-PV values, to get them into a clean
767 * state (SvPV should do that, but it's buggy, see below).
768 */
769 /*SvGETMAGIC (string);*/
770 if (SvMAGICAL (string) || !SvPOK (string))
771 string = sv_2mortal (newSVsv (string));
772
773 SvUPGRADE (string, SVt_PV);
774
775 /* work around a bug in perl 5.10, which causes SvCUR to fail an
776 * assertion with -DDEBUGGING, although SvCUR is documented to
777 * return the xpv_cur field which certainly exists after upgrading.
778 * according to nicholas clark, calling SvPOK fixes this.
779 * But it doesn't fix it, so try another workaround, call SvPV_nolen
780 * and hope for the best.
781 * Damnit, SvPV_nolen still trips over yet another assertion. This
782 * assertion business is seriously broken, try yet another workaround
783 * for the broken -DDEBUGGING.
784 */
785 {
786#ifdef DEBUGGING
787 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
788#else
789 STRLEN offset = SvCUR (string);
790#endif
791
792 if (offset > cbor->max_size && cbor->max_size) 1170 if (len > cbor->max_size && cbor->max_size)
793 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",
794 (unsigned long)SvCUR (string), (unsigned long)cbor->max_size); 1172 (unsigned long)len, (unsigned long)cbor->max_size);
795 }
796
797 sv_utf8_downgrade (string, 0);
798 1173
799 dec.cbor = *cbor; 1174 dec.cbor = *cbor;
800 dec.cur = (U8 *)SvPVX (string); 1175 dec.cur = (U8 *)data;
801 dec.end = (U8 *)SvEND (string); 1176 dec.end = (U8 *)data + len;
802 dec.err = 0;
803 dec.depth = 0;
804 1177
805 sv = decode_sv (&dec); 1178 sv = decode_sv (&dec);
806 1179
807 if (offset_return) 1180 if (offset_return)
808 *offset_return = dec.cur; 1181 *offset_return = dec.cur;
811 if (dec.cur != dec.end && !dec.err) 1184 if (dec.cur != dec.end && !dec.err)
812 dec.err = "garbage after CBOR object"; 1185 dec.err = "garbage after CBOR object";
813 1186
814 if (dec.err) 1187 if (dec.err)
815 { 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
816 SvREFCNT_dec (sv); 1200 SvREFCNT_dec (sv);
817 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);
818 } 1202 }
819 1203
820 sv = sv_2mortal (sv); 1204 sv = sv_2mortal (sv);
821 1205
822 return sv; 1206 return sv;
823} 1207}
824 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
825///////////////////////////////////////////////////////////////////////////// 1329/////////////////////////////////////////////////////////////////////////////
826// XS interface functions 1330// XS interface functions
827 1331
828MODULE = CBOR::XS PACKAGE = CBOR::XS 1332MODULE = CBOR::XS PACKAGE = CBOR::XS
829 1333
830BOOT: 1334BOOT:
831{ 1335{
832 cbor_stash = gv_stashpv ("CBOR::XS" , 1); 1336 cbor_stash = gv_stashpv ("CBOR::XS" , 1);
833 cbor_boolean_stash = gv_stashpv ("CBOR::XS::Boolean", 1);
834 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1); 1337 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1);
835 1338
836 cbor_true = get_bool ("CBOR::XS::true"); 1339 types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
837 cbor_false = get_bool ("CBOR::XS::false"); 1340 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
1341
1342 types_true = get_bool ("Types::Serialiser::true" );
1343 types_false = get_bool ("Types::Serialiser::false");
1344 types_error = get_bool ("Types::Serialiser::error");
1345
1346 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1347
1348 sv_cbor = newSVpv ("CBOR", 0);
1349 SvREADONLY_on (sv_cbor);
838} 1350}
839 1351
840PROTOTYPES: DISABLE 1352PROTOTYPES: DISABLE
841 1353
842void CLONE (...) 1354void CLONE (...)
843 CODE: 1355 CODE:
844 cbor_stash = 0; 1356 cbor_stash = 0;
845 cbor_boolean_stash = 0;
846 cbor_tagged_stash = 0; 1357 cbor_tagged_stash = 0;
1358 types_error_stash = 0;
1359 types_boolean_stash = 0;
847 1360
848void new (char *klass) 1361void new (char *klass)
849 PPCODE: 1362 PPCODE:
850{ 1363{
851 SV *pv = NEWSV (0, sizeof (CBOR)); 1364 SV *pv = NEWSV (0, sizeof (CBOR));
859 1372
860void shrink (CBOR *self, int enable = 1) 1373void shrink (CBOR *self, int enable = 1)
861 ALIAS: 1374 ALIAS:
862 shrink = F_SHRINK 1375 shrink = F_SHRINK
863 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
864 PPCODE: 1381 PPCODE:
865{ 1382{
866 if (enable) 1383 if (enable)
867 self->flags |= ix; 1384 self->flags |= ix;
868 else 1385 else
873 1390
874void get_shrink (CBOR *self) 1391void get_shrink (CBOR *self)
875 ALIAS: 1392 ALIAS:
876 get_shrink = F_SHRINK 1393 get_shrink = F_SHRINK
877 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
878 PPCODE: 1399 PPCODE:
879 XPUSHs (boolSV (self->flags & ix)); 1400 XPUSHs (boolSV (self->flags & ix));
880 1401
881void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1402void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
882 PPCODE: 1403 PPCODE:
898 CODE: 1419 CODE:
899 RETVAL = self->max_size; 1420 RETVAL = self->max_size;
900 OUTPUT: 1421 OUTPUT:
901 RETVAL 1422 RETVAL
902 1423
903#if 0 //TODO 1424void filter (CBOR *self, SV *filter = 0)
904
905void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef)
906 PPCODE: 1425 PPCODE:
907{
908 SvREFCNT_dec (self->cb_object); 1426 SvREFCNT_dec (self->filter);
909 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; 1427 self->filter = filter ? newSVsv (filter) : filter;
910
911 XPUSHs (ST (0)); 1428 XPUSHs (ST (0));
912}
913 1429
914void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef) 1430SV *get_filter (CBOR *self)
915 PPCODE: 1431 CODE:
916{ 1432 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
917 if (!self->cb_sk_object) 1433 OUTPUT:
918 self->cb_sk_object = newHV (); 1434 RETVAL
919
920 if (SvOK (cb))
921 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
922 else
923 {
924 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
925
926 if (!HvKEYS (self->cb_sk_object))
927 {
928 SvREFCNT_dec (self->cb_sk_object);
929 self->cb_sk_object = 0;
930 }
931 }
932
933 XPUSHs (ST (0));
934}
935
936#endif
937 1435
938void encode (CBOR *self, SV *scalar) 1436void encode (CBOR *self, SV *scalar)
939 PPCODE: 1437 PPCODE:
940 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; 1438 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
941 XPUSHs (scalar); 1439 XPUSHs (scalar);
954 EXTEND (SP, 2); 1452 EXTEND (SP, 2);
955 PUSHs (sv); 1453 PUSHs (sv);
956 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1454 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
957} 1455}
958 1456
959#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}
960 1508
961void DESTROY (CBOR *self) 1509void DESTROY (CBOR *self)
962 CODE: 1510 PPCODE:
963 SvREFCNT_dec (self->cb_sk_object); 1511 cbor_free (self);
964 SvREFCNT_dec (self->cb_object);
965
966#endif
967 1512
968PROTOTYPES: ENABLE 1513PROTOTYPES: ENABLE
969 1514
970void encode_cbor (SV *scalar) 1515void encode_cbor (SV *scalar)
1516 ALIAS:
1517 encode_cbor = 0
1518 encode_cbor_sharing = F_ALLOW_SHARING
971 PPCODE: 1519 PPCODE:
972{ 1520{
973 CBOR cbor; 1521 CBOR cbor;
974 cbor_init (&cbor); 1522 cbor_init (&cbor);
1523 cbor.flags |= ix;
975 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1524 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
976 XPUSHs (scalar); 1525 XPUSHs (scalar);
977} 1526}
978 1527
979void decode_cbor (SV *cborstr) 1528void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines