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 1.8 by root, Sun Oct 27 10:12:01 2013 UTC vs.
Revision 1.46 by root, Sun Dec 14 05:57:22 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"
13 14
14// known tags, rfc7049 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};
65
66// known tags
15enum cbor_tag 67enum cbor_tag
16{ 68{
69 // extensions
70 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
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
77
78 // rfc7049
17 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 79 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
18 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any 80 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
19 CBOR_TAG_POS_BIGNUM = 2, // byte string 81 CBOR_TAG_POS_BIGNUM = 2, // byte string
20 CBOR_TAG_NEG_BIGNUM = 3, // byte string 82 CBOR_TAG_NEG_BIGNUM = 3, // byte string
21 CBOR_TAG_DECIMAL = 4, // decimal fraction, array 83 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
22 CBOR_TAG_BIGFLOAT = 5, // array 84 CBOR_TAG_BIGFLOAT = 5, // array
23 85
24 CBOR_TAG_CONV_B64U = 21, // base64url, any 86 CBOR_TAG_CONV_B64U = 21, // base64url, any
25 CBOR_TAG_CONV_B64 = 22, // base64, any 87 CBOR_TAG_CONV_B64 = 22, // base64, any
26 CBOR_TAG_CONV_HEX = 23, // base16, any 88 CBOR_TAG_CONV_HEX = 23, // base16, any
27 CBOR_TAG_CBOR = 24, // embedded cbor, byte string 89 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
28 90
29 CBOR_TAG_URI = 32, // URI rfc3986, utf-8 91 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
30 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8 92 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
31 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8 93 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
32 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8 94 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
33 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 95 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
34 96
35 CBOR_TAG_MAGIC = 55799 97 CBOR_TAG_MAGIC = 55799, // self-describe cbor
36}; 98};
37 99
38#define F_SHRINK 0x00000200UL 100#define F_SHRINK 0x00000001UL
39#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
40 106
41#define INIT_SIZE 32 // initial scalar size to be allocated 107#define INIT_SIZE 32 // initial scalar size to be allocated
42 108
43#define SB do { 109#define SB do {
44#define SE } while (0) 110#define SE } while (0)
55#else 121#else
56# define CBOR_SLOW 0 122# define CBOR_SLOW 0
57# define CBOR_STASH cbor_stash 123# define CBOR_STASH cbor_stash
58#endif 124#endif
59 125
60static 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::
61static SV *cbor_true, *cbor_false; 127static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
62 128
63typedef struct { 129typedef struct {
64 U32 flags; 130 U32 flags;
65 U32 max_depth; 131 U32 max_depth;
66 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.
67} CBOR; 139} CBOR;
68 140
69ecb_inline void 141ecb_inline void
70cbor_init (CBOR *cbor) 142cbor_init (CBOR *cbor)
71{ 143{
72 Zero (cbor, 1, CBOR); 144 Zero (cbor, 1, CBOR);
73 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);
74} 153}
75 154
76///////////////////////////////////////////////////////////////////////////// 155/////////////////////////////////////////////////////////////////////////////
77// utility functions 156// utility functions
78 157
100 SvPV_renew (sv, SvCUR (sv) + 1); 179 SvPV_renew (sv, SvCUR (sv) + 1);
101#endif 180#endif
102 } 181 }
103} 182}
104 183
105///////////////////////////////////////////////////////////////////////////// 184// minimum length of a string to be registered for stringref
106// fp hell 185ecb_inline int
107 186minimum_string_length (UV idx)
108//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}
109 198
110///////////////////////////////////////////////////////////////////////////// 199/////////////////////////////////////////////////////////////////////////////
111// encoder 200// encoder
112 201
113// structure used for encoding CBOR 202// structure used for encoding CBOR
116 char *cur; // SvPVX (sv) + current output position 205 char *cur; // SvPVX (sv) + current output position
117 char *end; // SvEND (sv) 206 char *end; // SvEND (sv)
118 SV *sv; // result scalar 207 SV *sv; // result scalar
119 CBOR cbor; 208 CBOR cbor;
120 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;
121} enc_t; 214} enc_t;
122 215
123ecb_inline void 216ecb_inline void
124need (enc_t *enc, STRLEN len) 217need (enc_t *enc, STRLEN len)
125{ 218{
142static void 235static void
143encode_uint (enc_t *enc, int major, UV len) 236encode_uint (enc_t *enc, int major, UV len)
144{ 237{
145 need (enc, 9); 238 need (enc, 9);
146 239
147 if (len < 24) 240 if (ecb_expect_true (len < LENGTH_EXT1))
148 *enc->cur++ = major | len; 241 *enc->cur++ = major | len;
149 else if (len <= 0xff) 242 else if (ecb_expect_true (len <= 0xffU))
150 { 243 {
151 *enc->cur++ = major | 24; 244 *enc->cur++ = major | LENGTH_EXT1;
152 *enc->cur++ = len; 245 *enc->cur++ = len;
153 } 246 }
154 else if (len <= 0xffff) 247 else if (len <= 0xffffU)
155 { 248 {
156 *enc->cur++ = major | 25; 249 *enc->cur++ = major | LENGTH_EXT2;
157 *enc->cur++ = len >> 8; 250 *enc->cur++ = len >> 8;
158 *enc->cur++ = len; 251 *enc->cur++ = len;
159 } 252 }
160 else if (len <= 0xffffffff) 253 else if (len <= 0xffffffffU)
161 { 254 {
162 *enc->cur++ = major | 26; 255 *enc->cur++ = major | LENGTH_EXT4;
163 *enc->cur++ = len >> 24; 256 *enc->cur++ = len >> 24;
164 *enc->cur++ = len >> 16; 257 *enc->cur++ = len >> 16;
165 *enc->cur++ = len >> 8; 258 *enc->cur++ = len >> 8;
166 *enc->cur++ = len; 259 *enc->cur++ = len;
167 } 260 }
168 else 261 else
169 { 262 {
170 *enc->cur++ = major | 27; 263 *enc->cur++ = major | LENGTH_EXT8;
171 *enc->cur++ = len >> 56; 264 *enc->cur++ = len >> 56;
172 *enc->cur++ = len >> 48; 265 *enc->cur++ = len >> 48;
173 *enc->cur++ = len >> 40; 266 *enc->cur++ = len >> 40;
174 *enc->cur++ = len >> 32; 267 *enc->cur++ = len >> 32;
175 *enc->cur++ = len >> 24; 268 *enc->cur++ = len >> 24;
177 *enc->cur++ = len >> 8; 270 *enc->cur++ = len >> 8;
178 *enc->cur++ = len; 271 *enc->cur++ = len;
179 } 272 }
180} 273}
181 274
182static void 275ecb_inline void
276encode_tag (enc_t *enc, UV tag)
277{
278 encode_uint (enc, MAJOR_TAG, tag);
279}
280
281ecb_inline void
183encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 282encode_str (enc_t *enc, int utf8, char *str, STRLEN len)
184{ 283{
185 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 284 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
186 need (enc, len); 285 need (enc, len);
187 memcpy (enc->cur, str, len); 286 memcpy (enc->cur, str, len);
188 enc->cur += len; 287 enc->cur += len;
189} 288}
190 289
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
191static void encode_sv (enc_t *enc, SV *sv); 315static void encode_sv (enc_t *enc, SV *sv);
192 316
193static void 317static void
194encode_av (enc_t *enc, AV *av) 318encode_av (enc_t *enc, AV *av)
195{ 319{
198 if (enc->depth >= enc->cbor.max_depth) 322 if (enc->depth >= enc->cbor.max_depth)
199 croak (ERR_NESTING_EXCEEDED); 323 croak (ERR_NESTING_EXCEEDED);
200 324
201 ++enc->depth; 325 ++enc->depth;
202 326
203 encode_uint (enc, 0x80, len + 1); 327 encode_uint (enc, MAJOR_ARRAY, len + 1);
204 328
329 if (SvMAGICAL (av))
205 for (i = 0; i <= len; ++i) 330 for (i = 0; i <= len; ++i)
206 { 331 {
207 SV **svp = av_fetch (av, i, 0); 332 SV **svp = av_fetch (av, i, 0);
208 encode_sv (enc, svp ? *svp : &PL_sv_undef); 333 encode_sv (enc, svp ? *svp : &PL_sv_undef);
209 } 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 }
210 341
211 --enc->depth; 342 --enc->depth;
212} 343}
213 344
214static void 345static void
223 354
224 int pairs = hv_iterinit (hv); 355 int pairs = hv_iterinit (hv);
225 int mg = SvMAGICAL (hv); 356 int mg = SvMAGICAL (hv);
226 357
227 if (mg) 358 if (mg)
228 encode_ch (enc, 0xa0 | 31); 359 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
229 else 360 else
230 encode_uint (enc, 0xa0, pairs); 361 encode_uint (enc, MAJOR_MAP, pairs);
231 362
232 while ((he = hv_iternext (hv))) 363 while ((he = hv_iternext (hv)))
233 { 364 {
234 if (HeKLEN (he) == HEf_SVKEY) 365 if (HeKLEN (he) == HEf_SVKEY)
235 encode_sv (enc, HeSVKEY (he)); 366 encode_sv (enc, HeSVKEY (he));
236 else 367 else
237 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 368 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
238 369
239 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));
240 } 371 }
241 372
242 if (mg) 373 if (mg)
243 encode_ch (enc, 0xe0 | 31); 374 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
244 375
245 --enc->depth; 376 --enc->depth;
246} 377}
247 378
248// encode objects, arrays and special \0=false and \1=true values. 379// encode objects, arrays and special \0=false and \1=true values.
249static void 380static void
250encode_rv (enc_t *enc, SV *sv) 381encode_rv (enc_t *enc, SV *sv)
251{ 382{
252 svtype svt;
253
254 SvGETMAGIC (sv); 383 SvGETMAGIC (sv);
384
255 svt = SvTYPE (sv); 385 svtype svt = SvTYPE (sv);
256 386
257 if (ecb_expect_false (SvOBJECT (sv))) 387 if (ecb_expect_false (SvOBJECT (sv)))
258 { 388 {
259 HV *boolean_stash = !CBOR_SLOW || cbor_boolean_stash 389 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
260 ? cbor_boolean_stash 390 ? types_boolean_stash
261 : gv_stashpv ("CBOR::XS::Boolean", 1); 391 : gv_stashpv ("Types::Serialiser::Boolean", 1);
392 HV *error_stash = !CBOR_SLOW || types_error_stash
393 ? types_error_stash
394 : gv_stashpv ("Types::Serialiser::Error", 1);
262 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 395 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
263 ? cbor_tagged_stash 396 ? cbor_tagged_stash
264 : gv_stashpv ("CBOR::XS::Tagged" , 1); 397 : gv_stashpv ("CBOR::XS::Tagged" , 1);
265 398
399 HV *stash = SvSTASH (sv);
400
266 if (SvSTASH (sv) == boolean_stash) 401 if (stash == boolean_stash)
267 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 }
406 else if (stash == error_stash)
407 {
408 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
409 return;
410 }
268 else if (SvSTASH (sv) == tagged_stash) 411 else if (stash == tagged_stash)
269 { 412 {
270 if (svt != SVt_PVAV) 413 if (svt != SVt_PVAV)
271 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 414 croak ("encountered CBOR::XS::Tagged object that isn't an array");
272 415
273 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 416 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
274 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;
275 } 436 }
276 else 437 else
277 { 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);
278 // 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
279 GV *to_cbor = gv_fetchmethod_autoload (SvSTASH (sv), "TO_CBOR", 0);
280
281 if (to_cbor)
282 {
283 dSP;
284
285 ENTER; SAVETMPS; PUSHMARK (SP);
286 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv))); 456 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
287 457
288 // calling with G_SCALAR ensures that we always get a 1 return value
289 PUTBACK; 458 PUTBACK;
459 // G_SCALAR ensures that return value is 1
290 call_sv ((SV *)GvCV (to_cbor), G_SCALAR); 460 call_sv ((SV *)GvCV (method), G_SCALAR);
291 SPAGAIN; 461 SPAGAIN;
292 462
293 // catch this surprisingly common error 463 // catch this surprisingly common error
294 if (SvROK (TOPs) && SvRV (TOPs) == sv) 464 if (SvROK (TOPs) && SvRV (TOPs) == sv)
295 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));
296 466
297 sv = POPs; 467 encode_sv (enc, POPs);
468
298 PUTBACK; 469 PUTBACK;
299 470
300 encode_sv (enc, sv);
301
302 FREETMPS; LEAVE; 471 FREETMPS; LEAVE;
303 } 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 }
304 else 502 else
305 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",
306 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 504 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
307 }
308 } 505 }
309 else if (svt == SVt_PVHV) 506 else if (svt == SVt_PVHV)
310 encode_hv (enc, (HV *)sv); 507 encode_hv (enc, (HV *)sv);
311 else if (svt == SVt_PVAV) 508 else if (svt == SVt_PVAV)
312 encode_av (enc, (AV *)sv); 509 encode_av (enc, (AV *)sv);
313 else if (svt < SVt_PVAV)
314 {
315 STRLEN len = 0;
316 char *pv = svt ? SvPV (sv, len) : 0;
317
318 if (len == 1 && *pv == '1')
319 encode_ch (enc, 0xe0 | 21);
320 else if (len == 1 && *pv == '0')
321 encode_ch (enc, 0xe0 | 20);
322 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
323 encode_ch (enc, 0xe0 | 23);
324 else
325 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
326 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
327 }
328 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
329 encode_ch (enc, 0xe0 | 23);
330 else 510 else
331 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 511 {
332 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 512 encode_tag (enc, CBOR_TAG_INDIRECTION);
513 encode_sv (enc, sv);
514 }
333} 515}
334 516
335static void 517static void
336encode_nv (enc_t *enc, SV *sv) 518encode_nv (enc_t *enc, SV *sv)
337{ 519{
338 double nv = SvNVX (sv); 520 double nv = SvNVX (sv);
339 521
340 need (enc, 9); 522 need (enc, 9);
341 523
342 if (ecb_expect_false (nv == (U32)nv)) 524 if (ecb_expect_false (nv == (NV)(U32)nv))
343 encode_uint (enc, 0x00, (U32)nv); 525 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
344 //TODO: maybe I32? 526 //TODO: maybe I32?
345 else if (ecb_expect_false (nv == (float)nv)) 527 else if (ecb_expect_false (nv == (float)nv))
346 { 528 {
347 uint32_t fp = ecb_float_to_binary32 (nv); 529 uint32_t fp = ecb_float_to_binary32 (nv);
348 530
349 *enc->cur++ = 0xe0 | 26; 531 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
350 532
351 if (!ecb_big_endian ()) 533 if (!ecb_big_endian ())
352 fp = ecb_bswap32 (fp); 534 fp = ecb_bswap32 (fp);
353 535
354 memcpy (enc->cur, &fp, 4); 536 memcpy (enc->cur, &fp, 4);
356 } 538 }
357 else 539 else
358 { 540 {
359 uint64_t fp = ecb_double_to_binary64 (nv); 541 uint64_t fp = ecb_double_to_binary64 (nv);
360 542
361 *enc->cur++ = 0xe0 | 27; 543 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
362 544
363 if (!ecb_big_endian ()) 545 if (!ecb_big_endian ())
364 fp = ecb_bswap64 (fp); 546 fp = ecb_bswap64 (fp);
365 547
366 memcpy (enc->cur, &fp, 8); 548 memcpy (enc->cur, &fp, 8);
375 557
376 if (SvPOKp (sv)) 558 if (SvPOKp (sv))
377 { 559 {
378 STRLEN len; 560 STRLEN len;
379 char *str = SvPV (sv, len); 561 char *str = SvPV (sv, len);
380 encode_str (enc, SvUTF8 (sv), str, len); 562 encode_strref (enc, SvUTF8 (sv), str, len);
381 } 563 }
382 else if (SvNOKp (sv)) 564 else if (SvNOKp (sv))
383 encode_nv (enc, sv); 565 encode_nv (enc, sv);
384 else if (SvIOKp (sv)) 566 else if (SvIOKp (sv))
385 { 567 {
386 if (SvIsUV (sv)) 568 if (SvIsUV (sv))
387 encode_uint (enc, 0x00, SvUVX (sv)); 569 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
388 else if (SvIVX (sv) >= 0) 570 else if (SvIVX (sv) >= 0)
389 encode_uint (enc, 0x00, SvIVX (sv)); 571 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
390 else 572 else
391 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 573 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
392 } 574 }
393 else if (SvROK (sv)) 575 else if (SvROK (sv))
394 encode_rv (enc, SvRV (sv)); 576 encode_rv (enc, SvRV (sv));
395 else if (!SvOK (sv)) 577 else if (!SvOK (sv))
396 encode_ch (enc, 0xe0 | 22); 578 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
397 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 579 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
398 encode_ch (enc, 0xe0 | 23); 580 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
399 else 581 else
400 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",
401 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 583 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
402} 584}
403 585
404static SV * 586static SV *
405encode_cbor (SV *scalar, CBOR *cbor) 587encode_cbor (SV *scalar, CBOR *cbor)
406{ 588{
407 enc_t enc; 589 enc_t enc = { };
408 590
409 enc.cbor = *cbor; 591 enc.cbor = *cbor;
410 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 592 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
411 enc.cur = SvPVX (enc.sv); 593 enc.cur = SvPVX (enc.sv);
412 enc.end = SvEND (enc.sv); 594 enc.end = SvEND (enc.sv);
413 enc.depth = 0;
414 595
415 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
416 encode_sv (&enc, scalar); 605 encode_sv (&enc, scalar);
417 606
418 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 607 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
419 *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
420 609
434 U8 *end; // end of input string 623 U8 *end; // end of input string
435 const char *err; // parse error, if != 0 624 const char *err; // parse error, if != 0
436 CBOR cbor; 625 CBOR cbor;
437 U32 depth; // recursion depth 626 U32 depth; // recursion depth
438 U32 maxdepth; // recursion depth limit 627 U32 maxdepth; // recursion depth limit
628 AV *shareable;
629 AV *stringref;
630 SV *decode_tagged;
439} dec_t; 631} dec_t;
440 632
441#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
442 634
443#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")
446#define DEC_DEC_DEPTH --dec->depth 638#define DEC_DEC_DEPTH --dec->depth
447 639
448static UV 640static UV
449decode_uint (dec_t *dec) 641decode_uint (dec_t *dec)
450{ 642{
451 switch (*dec->cur & 31) 643 U8 m = *dec->cur & MINOR_MASK;
452 { 644 ++dec->cur;
453 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
454 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
455 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
456 return *dec->cur++ & 31;
457 645
458 case 24: 646 if (ecb_expect_true (m < LENGTH_EXT1))
647 return m;
648 else if (ecb_expect_true (m == LENGTH_EXT1))
649 {
459 WANT (2); 650 WANT (1);
460 dec->cur += 2; 651 dec->cur += 1;
461 return dec->cur[-1]; 652 return dec->cur[-1];
462 653 }
463 case 25: 654 else if (ecb_expect_true (m == LENGTH_EXT2))
655 {
464 WANT (3); 656 WANT (2);
465 dec->cur += 3; 657 dec->cur += 2;
466 return (((UV)dec->cur[-2]) << 8) 658 return (((UV)dec->cur[-2]) << 8)
467 | ((UV)dec->cur[-1]); 659 | ((UV)dec->cur[-1]);
468 660 }
469 case 26: 661 else if (ecb_expect_true (m == LENGTH_EXT4))
662 {
470 WANT (5); 663 WANT (4);
471 dec->cur += 5; 664 dec->cur += 4;
472 return (((UV)dec->cur[-4]) << 24) 665 return (((UV)dec->cur[-4]) << 24)
473 | (((UV)dec->cur[-3]) << 16) 666 | (((UV)dec->cur[-3]) << 16)
474 | (((UV)dec->cur[-2]) << 8) 667 | (((UV)dec->cur[-2]) << 8)
475 | ((UV)dec->cur[-1]); 668 | ((UV)dec->cur[-1]);
476 669 }
477 case 27: 670 else if (ecb_expect_true (m == LENGTH_EXT8))
671 {
478 WANT (9); 672 WANT (8);
479 dec->cur += 9; 673 dec->cur += 8;
674
675 return
676#if UVSIZE < 8
677 0
678#else
480 return (((UV)dec->cur[-8]) << 56) 679 (((UV)dec->cur[-8]) << 56)
481 | (((UV)dec->cur[-7]) << 48) 680 | (((UV)dec->cur[-7]) << 48)
482 | (((UV)dec->cur[-6]) << 40) 681 | (((UV)dec->cur[-6]) << 40)
483 | (((UV)dec->cur[-5]) << 32) 682 | (((UV)dec->cur[-5]) << 32)
683#endif
484 | (((UV)dec->cur[-4]) << 24) 684 | (((UV)dec->cur[-4]) << 24)
485 | (((UV)dec->cur[-3]) << 16) 685 | (((UV)dec->cur[-3]) << 16)
486 | (((UV)dec->cur[-2]) << 8) 686 | (((UV)dec->cur[-2]) << 8)
487 | ((UV)dec->cur[-1]); 687 | ((UV)dec->cur[-1]);
488 688 }
489 default: 689 else
490 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 690 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
491 }
492 691
493fail: 692fail:
494 return 0; 693 return 0;
495} 694}
496 695
501{ 700{
502 AV *av = newAV (); 701 AV *av = newAV ();
503 702
504 DEC_INC_DEPTH; 703 DEC_INC_DEPTH;
505 704
506 if ((*dec->cur & 31) == 31) 705 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
507 { 706 {
508 ++dec->cur; 707 ++dec->cur;
509 708
510 for (;;) 709 for (;;)
511 { 710 {
512 WANT (1); 711 WANT (1);
513 712
514 if (*dec->cur == (0xe0 | 31)) 713 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
515 { 714 {
516 ++dec->cur; 715 ++dec->cur;
517 break; 716 break;
518 } 717 }
519 718
522 } 721 }
523 else 722 else
524 { 723 {
525 int i, len = decode_uint (dec); 724 int i, len = decode_uint (dec);
526 725
726 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
527 av_fill (av, len - 1); 727 av_fill (av, len - 1);
528 728
529 for (i = 0; i < len; ++i) 729 for (i = 0; i < len; ++i)
530 AvARRAY (av)[i] = decode_sv (dec); 730 AvARRAY (av)[i] = decode_sv (dec);
531 } 731 }
537 SvREFCNT_dec (av); 737 SvREFCNT_dec (av);
538 DEC_DEC_DEPTH; 738 DEC_DEC_DEPTH;
539 return &PL_sv_undef; 739 return &PL_sv_undef;
540} 740}
541 741
742static void
743decode_he (dec_t *dec, HV *hv)
744{
745 // for speed reasons, we specialcase single-string
746 // byte or utf-8 strings as keys, but only when !stringref
747
748 if (ecb_expect_true (!dec->stringref))
749 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
750 {
751 I32 len = decode_uint (dec);
752 char *key = (char *)dec->cur;
753
754 dec->cur += len;
755
756 hv_store (hv, key, len, decode_sv (dec), 0);
757
758 return;
759 }
760 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
761 {
762 I32 len = decode_uint (dec);
763 char *key = (char *)dec->cur;
764
765 dec->cur += len;
766
767 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
768 if (!is_utf8_string (key, len))
769 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
770
771 hv_store (hv, key, -len, decode_sv (dec), 0);
772
773 return;
774 }
775
776 SV *k = decode_sv (dec);
777 SV *v = decode_sv (dec);
778
779 hv_store_ent (hv, k, v, 0);
780 SvREFCNT_dec (k);
781
782fail:
783 ;
784}
785
542static SV * 786static SV *
543decode_hv (dec_t *dec) 787decode_hv (dec_t *dec)
544{ 788{
545 HV *hv = newHV (); 789 HV *hv = newHV ();
546 790
547 DEC_INC_DEPTH; 791 DEC_INC_DEPTH;
548 792
549 if ((*dec->cur & 31) == 31) 793 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
550 { 794 {
551 ++dec->cur; 795 ++dec->cur;
552 796
553 for (;;) 797 for (;;)
554 { 798 {
555 WANT (1); 799 WANT (1);
556 800
557 if (*dec->cur == (0xe0 | 31)) 801 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
558 { 802 {
559 ++dec->cur; 803 ++dec->cur;
560 break; 804 break;
561 } 805 }
562 806
563 SV *k = decode_sv (dec); 807 decode_he (dec, hv);
564 SV *v = decode_sv (dec);
565
566 hv_store_ent (hv, k, v, 0);
567 } 808 }
568 } 809 }
569 else 810 else
570 { 811 {
571 int len = decode_uint (dec); 812 int pairs = decode_uint (dec);
572 813
573 while (len--) 814 while (pairs--)
574 { 815 decode_he (dec, hv);
575 SV *k = decode_sv (dec);
576 SV *v = decode_sv (dec);
577
578 hv_store_ent (hv, k, v, 0);
579 }
580 } 816 }
581 817
582 DEC_DEC_DEPTH; 818 DEC_DEC_DEPTH;
583 return newRV_noinc ((SV *)hv); 819 return newRV_noinc ((SV *)hv);
584 820
591static SV * 827static SV *
592decode_str (dec_t *dec, int utf8) 828decode_str (dec_t *dec, int utf8)
593{ 829{
594 SV *sv = 0; 830 SV *sv = 0;
595 831
596 if ((*dec->cur & 31) == 31) 832 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
597 { 833 {
834 // indefinite length strings
598 ++dec->cur; 835 ++dec->cur;
599 836
837 U8 major = *dec->cur & MAJOR_MISC;
838
600 sv = newSVpvn ("", 0); 839 sv = newSVpvn ("", 0);
601 840
602 // not very fast, and certainly not robust against illegal input
603 for (;;) 841 for (;;)
604 { 842 {
605 WANT (1); 843 WANT (1);
606 844
607 if (*dec->cur == (0xe0 | 31)) 845 if ((*dec->cur - major) > LENGTH_EXT8)
846 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
608 { 847 {
609 ++dec->cur; 848 ++dec->cur;
610 break; 849 break;
611 } 850 }
851 else
852 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
612 853
613 sv_catsv (sv, decode_sv (dec)); 854 STRLEN len = decode_uint (dec);
855
856 WANT (len);
857 sv_catpvn (sv, dec->cur, len);
858 dec->cur += len;
614 } 859 }
615 } 860 }
616 else 861 else
617 { 862 {
618 STRLEN len = decode_uint (dec); 863 STRLEN len = decode_uint (dec);
619 864
620 WANT (len); 865 WANT (len);
621 sv = newSVpvn (dec->cur, len); 866 sv = newSVpvn (dec->cur, len);
622 dec->cur += len; 867 dec->cur += len;
868
869 if (ecb_expect_false (dec->stringref)
870 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
871 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
623 } 872 }
624 873
625 if (utf8) 874 if (utf8)
875 {
876 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
877 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
878 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
879
626 SvUTF8_on (sv); 880 SvUTF8_on (sv);
881 }
627 882
628 return sv; 883 return sv;
629 884
630fail: 885fail:
631 SvREFCNT_dec (sv); 886 SvREFCNT_dec (sv);
633} 888}
634 889
635static SV * 890static SV *
636decode_tagged (dec_t *dec) 891decode_tagged (dec_t *dec)
637{ 892{
893 SV *sv = 0;
638 UV tag = decode_uint (dec); 894 UV tag = decode_uint (dec);
895
896 WANT (1);
897
898 switch (tag)
899 {
900 case CBOR_TAG_MAGIC:
639 SV *sv = decode_sv (dec); 901 sv = decode_sv (dec);
902 break;
640 903
641 if (tag == CBOR_TAG_MAGIC) // 2.4.5 Self-Describe CBOR 904 case CBOR_TAG_INDIRECTION:
905 sv = newRV_noinc (decode_sv (dec));
906 break;
907
908 case CBOR_TAG_STRINGREF_NAMESPACE:
909 {
910 ENTER; SAVETMPS;
911
912 SAVESPTR (dec->stringref);
913 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
914
915 sv = decode_sv (dec);
916
917 FREETMPS; LEAVE;
918 }
919 break;
920
921 case CBOR_TAG_STRINGREF:
922 {
923 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
924 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
925
926 UV idx = decode_uint (dec);
927
928 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
929 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
930
931 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
932 }
933 break;
934
935 case CBOR_TAG_VALUE_SHAREABLE:
936 {
937 if (ecb_expect_false (!dec->shareable))
938 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
939
940 if (dec->cbor.flags & F_ALLOW_CYCLES)
941 {
942 sv = newSV (0);
943 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
944
945 SV *osv = decode_sv (dec);
946 sv_setsv (sv, osv);
947 SvREFCNT_dec_NN (osv);
948 }
949 else
950 {
951 av_push (dec->shareable, &PL_sv_undef);
952 int idx = AvFILLp (dec->shareable);
953 sv = decode_sv (dec);
954 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
955 }
956 }
957 break;
958
959 case CBOR_TAG_VALUE_SHAREDREF:
960 {
961 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
962 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
963
964 UV idx = decode_uint (dec);
965
966 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
967 ERR ("corrupted CBOR data (sharedref index out of bounds)");
968
969 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
970
971 if (sv == &PL_sv_undef)
972 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
973 }
974 break;
975
976 case CBOR_TAG_PERL_OBJECT:
977 {
978 sv = decode_sv (dec);
979
980 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
981 ERR ("corrupted CBOR data (non-array perl object)");
982
983 AV *av = (AV *)SvRV (sv);
984 int len = av_len (av) + 1;
985 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
986
987 if (!stash)
988 ERR ("cannot decode perl-object (package does not exist)");
989
990 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
991
992 if (!method)
993 ERR ("cannot decode perl-object (package does not have a THAW method)");
994
995 dSP;
996
997 ENTER; SAVETMPS; PUSHMARK (SP);
998 EXTEND (SP, len + 1);
999 // we re-bless the reference to get overload and other niceties right
1000 PUSHs (*av_fetch (av, 0, 1));
1001 PUSHs (sv_cbor);
1002
1003 int i;
1004
1005 for (i = 1; i < len; ++i)
1006 PUSHs (*av_fetch (av, i, 1));
1007
1008 PUTBACK;
1009 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
1010 SPAGAIN;
1011
1012 if (SvTRUE (ERRSV))
1013 {
1014 FREETMPS; LEAVE;
1015 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1016 }
1017
1018 SvREFCNT_dec (sv);
1019 sv = SvREFCNT_inc (POPs);
1020
1021 PUTBACK;
1022
1023 FREETMPS; LEAVE;
1024 }
1025 break;
1026
1027 default:
1028 {
1029 sv = decode_sv (dec);
1030
1031 dSP;
1032 ENTER; SAVETMPS; PUSHMARK (SP);
1033 EXTEND (SP, 2);
1034 PUSHs (newSVuv (tag));
1035 PUSHs (sv);
1036
1037 PUTBACK;
1038 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1039 SPAGAIN;
1040
1041 if (SvTRUE (ERRSV))
1042 {
1043 FREETMPS; LEAVE;
1044 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1045 }
1046
1047 if (count)
1048 {
1049 SvREFCNT_dec (sv);
1050 sv = SvREFCNT_inc (POPs);
1051 }
1052 else
1053 {
1054 AV *av = newAV ();
1055 av_push (av, newSVuv (tag));
1056 av_push (av, sv);
1057
1058 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
1059 ? cbor_tagged_stash
1060 : gv_stashpv ("CBOR::XS::Tagged" , 1);
1061 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
1062 }
1063
1064 PUTBACK;
1065
1066 FREETMPS; LEAVE;
1067 }
1068 break;
1069 }
1070
642 return sv; 1071 return sv;
643 1072
644 AV *av = newAV (); 1073fail:
645 av_push (av, newSVuv (tag)); 1074 SvREFCNT_dec (sv);
646 av_push (av, sv); 1075 return &PL_sv_undef;
647
648 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
649 ? cbor_tagged_stash
650 : gv_stashpv ("CBOR::XS::Tagged" , 1);
651
652 return sv_bless (newRV_noinc ((SV *)av), tagged_stash);
653} 1076}
654 1077
655static SV * 1078static SV *
656decode_sv (dec_t *dec) 1079decode_sv (dec_t *dec)
657{ 1080{
658 WANT (1); 1081 WANT (1);
659 1082
660 switch (*dec->cur >> 5) 1083 switch (*dec->cur >> MAJOR_SHIFT)
661 { 1084 {
662 case 0: // unsigned int 1085 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
663 return newSVuv (decode_uint (dec)); 1086 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
664 case 1: // negative int 1087 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
665 return newSViv (-1 - (IV)decode_uint (dec)); 1088 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
666 case 2: // octet string 1089 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
667 return decode_str (dec, 0); 1090 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
668 case 3: // utf-8 string 1091 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
669 return decode_str (dec, 1); 1092
670 case 4: // array 1093 case MAJOR_MISC >> MAJOR_SHIFT:
671 return decode_av (dec);
672 case 5: // map
673 return decode_hv (dec);
674 case 6: // tag
675 return decode_tagged (dec);
676 case 7: // misc
677 switch (*dec->cur++ & 31) 1094 switch (*dec->cur++ & MINOR_MASK)
678 { 1095 {
679 case 20: 1096 case SIMPLE_FALSE:
680#if CBOR_SLOW 1097#if CBOR_SLOW
681 cbor_false = get_bool ("CBOR::XS::false"); 1098 types_false = get_bool ("Types::Serialiser::false");
682#endif 1099#endif
683 return newSVsv (cbor_false); 1100 return newSVsv (types_false);
684 case 21: 1101 case SIMPLE_TRUE:
685#if CBOR_SLOW 1102#if CBOR_SLOW
686 cbor_true = get_bool ("CBOR::XS::true"); 1103 types_true = get_bool ("Types::Serialiser::true");
687#endif 1104#endif
688 return newSVsv (cbor_true); 1105 return newSVsv (types_true);
689 case 22: 1106 case SIMPLE_NULL:
690 return newSVsv (&PL_sv_undef); 1107 return newSVsv (&PL_sv_undef);
1108 case SIMPLE_UNDEF:
1109#if CBOR_SLOW
1110 types_error = get_bool ("Types::Serialiser::error");
1111#endif
1112 return newSVsv (types_error);
691 1113
692 case 25: 1114 case MISC_FLOAT16:
693 { 1115 {
694 WANT (2); 1116 WANT (2);
695 1117
696 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1118 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
697 dec->cur += 2; 1119 dec->cur += 2;
698 1120
699 return newSVnv (ecb_binary16_to_float (fp)); 1121 return newSVnv (ecb_binary16_to_float (fp));
700 } 1122 }
701 1123
702 case 26: 1124 case MISC_FLOAT32:
703 { 1125 {
704 uint32_t fp; 1126 uint32_t fp;
705 WANT (4); 1127 WANT (4);
706 memcpy (&fp, dec->cur, 4); 1128 memcpy (&fp, dec->cur, 4);
707 dec->cur += 4; 1129 dec->cur += 4;
710 fp = ecb_bswap32 (fp); 1132 fp = ecb_bswap32 (fp);
711 1133
712 return newSVnv (ecb_binary32_to_float (fp)); 1134 return newSVnv (ecb_binary32_to_float (fp));
713 } 1135 }
714 1136
715 case 27: 1137 case MISC_FLOAT64:
716 { 1138 {
717 uint64_t fp; 1139 uint64_t fp;
718 WANT (8); 1140 WANT (8);
719 memcpy (&fp, dec->cur, 8); 1141 memcpy (&fp, dec->cur, 8);
720 dec->cur += 8; 1142 dec->cur += 8;
723 fp = ecb_bswap64 (fp); 1145 fp = ecb_bswap64 (fp);
724 1146
725 return newSVnv (ecb_binary64_to_double (fp)); 1147 return newSVnv (ecb_binary64_to_double (fp));
726 } 1148 }
727 1149
728 // 0..19 unassigned 1150 // 0..19 unassigned simple
729 // 24 reserved + unassigned (reserved values are not encodable) 1151 // 24 reserved + unassigned simple (reserved values are not encodable)
1152 // 28-30 unassigned misc
1153 // 31 break code
730 default: 1154 default:
731 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1155 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
732 } 1156 }
733 1157
734 break; 1158 break;
735 } 1159 }
736 1160
739} 1163}
740 1164
741static SV * 1165static SV *
742decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1166decode_cbor (SV *string, CBOR *cbor, char **offset_return)
743{ 1167{
744 dec_t dec; 1168 dec_t dec = { };
745 SV *sv; 1169 SV *sv;
1170 STRLEN len;
1171 char *data = SvPVbyte (string, len);
746 1172
747 /* work around bugs in 5.10 where manipulating magic values
748 * makes perl ignore the magic in subsequent accesses.
749 * also make a copy of non-PV values, to get them into a clean
750 * state (SvPV should do that, but it's buggy, see below).
751 */
752 /*SvGETMAGIC (string);*/
753 if (SvMAGICAL (string) || !SvPOK (string))
754 string = sv_2mortal (newSVsv (string));
755
756 SvUPGRADE (string, SVt_PV);
757
758 /* work around a bug in perl 5.10, which causes SvCUR to fail an
759 * assertion with -DDEBUGGING, although SvCUR is documented to
760 * return the xpv_cur field which certainly exists after upgrading.
761 * according to nicholas clark, calling SvPOK fixes this.
762 * But it doesn't fix it, so try another workaround, call SvPV_nolen
763 * and hope for the best.
764 * Damnit, SvPV_nolen still trips over yet another assertion. This
765 * assertion business is seriously broken, try yet another workaround
766 * for the broken -DDEBUGGING.
767 */
768 {
769#ifdef DEBUGGING
770 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
771#else
772 STRLEN offset = SvCUR (string);
773#endif
774
775 if (offset > cbor->max_size && cbor->max_size) 1173 if (len > cbor->max_size && cbor->max_size)
776 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu", 1174 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
777 (unsigned long)SvCUR (string), (unsigned long)cbor->max_size); 1175 (unsigned long)len, (unsigned long)cbor->max_size);
778 }
779
780 sv_utf8_downgrade (string, 0);
781 1176
782 dec.cbor = *cbor; 1177 dec.cbor = *cbor;
783 dec.cur = (U8 *)SvPVX (string); 1178 dec.cur = (U8 *)data;
784 dec.end = (U8 *)SvEND (string); 1179 dec.end = (U8 *)data + len;
785 dec.err = 0;
786 dec.depth = 0;
787 1180
788 sv = decode_sv (&dec); 1181 sv = decode_sv (&dec);
789 1182
790 if (offset_return) 1183 if (offset_return)
791 *offset_return = dec.cur; 1184 *offset_return = dec.cur;
794 if (dec.cur != dec.end && !dec.err) 1187 if (dec.cur != dec.end && !dec.err)
795 dec.err = "garbage after CBOR object"; 1188 dec.err = "garbage after CBOR object";
796 1189
797 if (dec.err) 1190 if (dec.err)
798 { 1191 {
1192 if (dec.shareable)
1193 {
1194 // need to break cyclic links, which whould all be in shareable
1195 int i;
1196 SV **svp;
1197
1198 for (i = av_len (dec.shareable) + 1; i--; )
1199 if ((svp = av_fetch (dec.shareable, i, 0)))
1200 sv_setsv (*svp, &PL_sv_undef);
1201 }
1202
799 SvREFCNT_dec (sv); 1203 SvREFCNT_dec (sv);
800 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)SvPVX (string), (int)(uint8_t)*dec.cur); 1204 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
801 } 1205 }
802 1206
803 sv = sv_2mortal (sv); 1207 sv = sv_2mortal (sv);
804 1208
805 return sv; 1209 return sv;
806} 1210}
807 1211
1212/////////////////////////////////////////////////////////////////////////////
1213// incremental parser
1214
1215#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1216
1217// returns 0 for notyet, 1 for success or error
1218static int
1219incr_parse (CBOR *self, SV *cborstr)
1220{
1221 STRLEN cur;
1222 SvPV (cborstr, cur);
1223
1224 while (ecb_expect_true (self->incr_need <= cur))
1225 {
1226 // table of integer count bytes
1227 static I8 incr_len[MINOR_MASK + 1] = {
1228 0, 0, 0, 0, 0, 0, 0, 0,
1229 0, 0, 0, 0, 0, 0, 0, 0,
1230 0, 0, 0, 0, 0, 0, 0, 0,
1231 1, 2, 4, 8,-1,-1,-1,-2
1232 };
1233
1234 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1235 U8 m = *p & MINOR_MASK;
1236 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1237 I8 ilen = incr_len[m];
1238
1239 self->incr_need = self->incr_pos + 1;
1240
1241 if (ecb_expect_false (ilen < 0))
1242 {
1243 if (m != MINOR_INDEF)
1244 return 1; // error
1245
1246 if (*p == (MAJOR_MISC | MINOR_INDEF))
1247 {
1248 if (count >= 0)
1249 return 1; // error
1250
1251 count = 1;
1252 }
1253 else
1254 {
1255 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1256 count = -1;
1257 }
1258 }
1259 else
1260 {
1261 self->incr_need += ilen;
1262 if (ecb_expect_false (self->incr_need > cur))
1263 return 0;
1264
1265 int major = *p >> MAJOR_SHIFT;
1266
1267 switch (major)
1268 {
1269 case MAJOR_BYTES >> MAJOR_SHIFT:
1270 case MAJOR_TEXT >> MAJOR_SHIFT:
1271 case MAJOR_ARRAY >> MAJOR_SHIFT:
1272 case MAJOR_MAP >> MAJOR_SHIFT:
1273 {
1274 UV len;
1275
1276 if (ecb_expect_false (ilen))
1277 {
1278 len = 0;
1279
1280 do {
1281 len = (len << 8) | *++p;
1282 } while (--ilen);
1283 }
1284 else
1285 len = m;
1286
1287 switch (major)
1288 {
1289 case MAJOR_BYTES >> MAJOR_SHIFT:
1290 case MAJOR_TEXT >> MAJOR_SHIFT:
1291 self->incr_need += len;
1292 if (ecb_expect_false (self->incr_need > cur))
1293 return 0;
1294
1295 break;
1296
1297 case MAJOR_MAP >> MAJOR_SHIFT:
1298 len <<= 1;
1299 case MAJOR_ARRAY >> MAJOR_SHIFT:
1300 if (len)
1301 {
1302 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1303 count = len + 1;
1304 }
1305 break;
1306 }
1307 }
1308 }
1309 }
1310
1311 self->incr_pos = self->incr_need;
1312
1313 if (count > 0)
1314 {
1315 while (!--count)
1316 {
1317 if (!AvFILLp (self->incr_count))
1318 return 1; // done
1319
1320 SvREFCNT_dec_NN (av_pop (self->incr_count));
1321 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1322 }
1323
1324 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1325 }
1326 }
1327
1328 return 0;
1329}
1330
1331
808///////////////////////////////////////////////////////////////////////////// 1332/////////////////////////////////////////////////////////////////////////////
809// XS interface functions 1333// XS interface functions
810 1334
811MODULE = CBOR::XS PACKAGE = CBOR::XS 1335MODULE = CBOR::XS PACKAGE = CBOR::XS
812 1336
813BOOT: 1337BOOT:
814{ 1338{
815 cbor_stash = gv_stashpv ("CBOR::XS" , 1); 1339 cbor_stash = gv_stashpv ("CBOR::XS" , 1);
816 cbor_boolean_stash = gv_stashpv ("CBOR::XS::Boolean", 1);
817 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1); 1340 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1);
818 1341
819 cbor_true = get_bool ("CBOR::XS::true"); 1342 types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
820 cbor_false = get_bool ("CBOR::XS::false"); 1343 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
1344
1345 types_true = get_bool ("Types::Serialiser::true" );
1346 types_false = get_bool ("Types::Serialiser::false");
1347 types_error = get_bool ("Types::Serialiser::error");
1348
1349 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1350
1351 sv_cbor = newSVpv ("CBOR", 0);
1352 SvREADONLY_on (sv_cbor);
821} 1353}
822 1354
823PROTOTYPES: DISABLE 1355PROTOTYPES: DISABLE
824 1356
825void CLONE (...) 1357void CLONE (...)
826 CODE: 1358 CODE:
827 cbor_stash = 0; 1359 cbor_stash = 0;
828 cbor_boolean_stash = 0;
829 cbor_tagged_stash = 0; 1360 cbor_tagged_stash = 0;
1361 types_error_stash = 0;
1362 types_boolean_stash = 0;
830 1363
831void new (char *klass) 1364void new (char *klass)
832 PPCODE: 1365 PPCODE:
833{ 1366{
834 SV *pv = NEWSV (0, sizeof (CBOR)); 1367 SV *pv = NEWSV (0, sizeof (CBOR));
842 1375
843void shrink (CBOR *self, int enable = 1) 1376void shrink (CBOR *self, int enable = 1)
844 ALIAS: 1377 ALIAS:
845 shrink = F_SHRINK 1378 shrink = F_SHRINK
846 allow_unknown = F_ALLOW_UNKNOWN 1379 allow_unknown = F_ALLOW_UNKNOWN
1380 allow_sharing = F_ALLOW_SHARING
1381 allow_cycles = F_ALLOW_CYCLES
1382 pack_strings = F_PACK_STRINGS
1383 validate_utf8 = F_VALIDATE_UTF8
847 PPCODE: 1384 PPCODE:
848{ 1385{
849 if (enable) 1386 if (enable)
850 self->flags |= ix; 1387 self->flags |= ix;
851 else 1388 else
856 1393
857void get_shrink (CBOR *self) 1394void get_shrink (CBOR *self)
858 ALIAS: 1395 ALIAS:
859 get_shrink = F_SHRINK 1396 get_shrink = F_SHRINK
860 get_allow_unknown = F_ALLOW_UNKNOWN 1397 get_allow_unknown = F_ALLOW_UNKNOWN
1398 get_allow_sharing = F_ALLOW_SHARING
1399 get_allow_cycles = F_ALLOW_CYCLES
1400 get_pack_strings = F_PACK_STRINGS
1401 get_validate_utf8 = F_VALIDATE_UTF8
861 PPCODE: 1402 PPCODE:
862 XPUSHs (boolSV (self->flags & ix)); 1403 XPUSHs (boolSV (self->flags & ix));
863 1404
864void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1405void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
865 PPCODE: 1406 PPCODE:
881 CODE: 1422 CODE:
882 RETVAL = self->max_size; 1423 RETVAL = self->max_size;
883 OUTPUT: 1424 OUTPUT:
884 RETVAL 1425 RETVAL
885 1426
886#if 0 //TODO 1427void filter (CBOR *self, SV *filter = 0)
887
888void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef)
889 PPCODE: 1428 PPCODE:
890{
891 SvREFCNT_dec (self->cb_object); 1429 SvREFCNT_dec (self->filter);
892 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; 1430 self->filter = filter ? newSVsv (filter) : filter;
893
894 XPUSHs (ST (0)); 1431 XPUSHs (ST (0));
895}
896 1432
897void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef) 1433SV *get_filter (CBOR *self)
898 PPCODE: 1434 CODE:
899{ 1435 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
900 if (!self->cb_sk_object) 1436 OUTPUT:
901 self->cb_sk_object = newHV (); 1437 RETVAL
902
903 if (SvOK (cb))
904 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
905 else
906 {
907 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
908
909 if (!HvKEYS (self->cb_sk_object))
910 {
911 SvREFCNT_dec (self->cb_sk_object);
912 self->cb_sk_object = 0;
913 }
914 }
915
916 XPUSHs (ST (0));
917}
918
919#endif
920 1438
921void encode (CBOR *self, SV *scalar) 1439void encode (CBOR *self, SV *scalar)
922 PPCODE: 1440 PPCODE:
923 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; 1441 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
924 XPUSHs (scalar); 1442 XPUSHs (scalar);
937 EXTEND (SP, 2); 1455 EXTEND (SP, 2);
938 PUSHs (sv); 1456 PUSHs (sv);
939 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1457 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
940} 1458}
941 1459
942#if 0 1460void incr_parse (CBOR *self, SV *cborstr)
1461 ALIAS:
1462 incr_parse_multiple = 1
1463 PPCODE:
1464{
1465 if (SvUTF8 (cborstr))
1466 sv_utf8_downgrade (cborstr, 0);
1467
1468 if (!self->incr_count)
1469 {
1470 self->incr_count = newAV ();
1471 self->incr_pos = 0;
1472 self->incr_need = 1;
1473
1474 av_push (self->incr_count, newSViv (1));
1475 }
1476
1477 do
1478 {
1479 if (!incr_parse (self, cborstr))
1480 {
1481 if (self->incr_need > self->max_size && self->max_size)
1482 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1483 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1484
1485 break;
1486 }
1487
1488 SV *sv;
1489 char *offset;
1490
1491 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1492 XPUSHs (sv);
1493
1494 sv_chop (cborstr, offset);
1495
1496 av_clear (self->incr_count);
1497 av_push (self->incr_count, newSViv (1));
1498
1499 self->incr_pos = 0;
1500 self->incr_need = self->incr_pos + 1;
1501 }
1502 while (ix);
1503}
1504
1505void incr_reset (CBOR *self)
1506 CODE:
1507{
1508 SvREFCNT_dec (self->incr_count);
1509 self->incr_count = 0;
1510}
943 1511
944void DESTROY (CBOR *self) 1512void DESTROY (CBOR *self)
945 CODE: 1513 PPCODE:
946 SvREFCNT_dec (self->cb_sk_object); 1514 cbor_free (self);
947 SvREFCNT_dec (self->cb_object);
948
949#endif
950 1515
951PROTOTYPES: ENABLE 1516PROTOTYPES: ENABLE
952 1517
953void encode_cbor (SV *scalar) 1518void encode_cbor (SV *scalar)
1519 ALIAS:
1520 encode_cbor = 0
1521 encode_cbor_sharing = F_ALLOW_SHARING
954 PPCODE: 1522 PPCODE:
955{ 1523{
956 CBOR cbor; 1524 CBOR cbor;
957 cbor_init (&cbor); 1525 cbor_init (&cbor);
1526 cbor.flags |= ix;
958 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1527 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
959 XPUSHs (scalar); 1528 XPUSHs (scalar);
960} 1529}
961 1530
962void decode_cbor (SV *cborstr) 1531void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines