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 1.44 by root, Tue Feb 18 22:12:12 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
210 for (i = 0; i <= len; ++i) 329 for (i = 0; i <= len; ++i)
211 { 330 {
212 SV **svp = av_fetch (av, i, 0); 331 SV **svp = av_fetch (av, i, 0);
213 encode_sv (enc, svp ? *svp : &PL_sv_undef); 332 encode_sv (enc, svp ? *svp : &PL_sv_undef);
228 347
229 int pairs = hv_iterinit (hv); 348 int pairs = hv_iterinit (hv);
230 int mg = SvMAGICAL (hv); 349 int mg = SvMAGICAL (hv);
231 350
232 if (mg) 351 if (mg)
233 encode_ch (enc, 0xa0 | 31); 352 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
234 else 353 else
235 encode_uint (enc, 0xa0, pairs); 354 encode_uint (enc, MAJOR_MAP, pairs);
236 355
237 while ((he = hv_iternext (hv))) 356 while ((he = hv_iternext (hv)))
238 { 357 {
239 if (HeKLEN (he) == HEf_SVKEY) 358 if (HeKLEN (he) == HEf_SVKEY)
240 encode_sv (enc, HeSVKEY (he)); 359 encode_sv (enc, HeSVKEY (he));
241 else 360 else
242 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 361 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
243 362
244 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 363 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
245 } 364 }
246 365
247 if (mg) 366 if (mg)
248 encode_ch (enc, 0xe0 | 31); 367 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
249 368
250 --enc->depth; 369 --enc->depth;
251} 370}
252 371
253// encode objects, arrays and special \0=false and \1=true values. 372// encode objects, arrays and special \0=false and \1=true values.
254static void 373static void
255encode_rv (enc_t *enc, SV *sv) 374encode_rv (enc_t *enc, SV *sv)
256{ 375{
257 svtype svt;
258
259 SvGETMAGIC (sv); 376 SvGETMAGIC (sv);
377
260 svt = SvTYPE (sv); 378 svtype svt = SvTYPE (sv);
261 379
262 if (ecb_expect_false (SvOBJECT (sv))) 380 if (ecb_expect_false (SvOBJECT (sv)))
263 { 381 {
264 HV *boolean_stash = !CBOR_SLOW || cbor_boolean_stash 382 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
265 ? cbor_boolean_stash 383 ? types_boolean_stash
266 : gv_stashpv ("CBOR::XS::Boolean", 1); 384 : gv_stashpv ("Types::Serialiser::Boolean", 1);
385 HV *error_stash = !CBOR_SLOW || types_error_stash
386 ? types_error_stash
387 : gv_stashpv ("Types::Serialiser::Error", 1);
267 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 388 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
268 ? cbor_tagged_stash 389 ? cbor_tagged_stash
269 : gv_stashpv ("CBOR::XS::Tagged" , 1); 390 : gv_stashpv ("CBOR::XS::Tagged" , 1);
270 391
392 HV *stash = SvSTASH (sv);
393
271 if (SvSTASH (sv) == boolean_stash) 394 if (stash == boolean_stash)
272 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 395 {
396 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
397 return;
398 }
399 else if (stash == error_stash)
400 {
401 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
402 return;
403 }
273 else if (SvSTASH (sv) == tagged_stash) 404 else if (stash == tagged_stash)
274 { 405 {
275 if (svt != SVt_PVAV) 406 if (svt != SVt_PVAV)
276 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 407 croak ("encountered CBOR::XS::Tagged object that isn't an array");
277 408
278 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 409 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
279 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 410 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
411
412 return;
413 }
414 }
415
416 if (ecb_expect_false (SvREFCNT (sv) > 1)
417 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
418 {
419 if (!enc->shareable)
420 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
421
422 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
423
424 if (SvOK (*svp))
425 {
426 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
427 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
428 return;
280 } 429 }
281 else 430 else
282 { 431 {
432 sv_setuv (*svp, enc->shareable_idx);
433 ++enc->shareable_idx;
434 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
435 }
436 }
437
438 if (ecb_expect_false (SvOBJECT (sv)))
439 {
440 HV *stash = SvSTASH (sv);
441 GV *method;
442
443 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
444 {
445 dSP;
446
447 ENTER; SAVETMPS; PUSHMARK (SP);
283 // we re-bless the reference to get overload and other niceties right 448 // 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))); 449 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
292 450
293 // calling with G_SCALAR ensures that we always get a 1 return value
294 PUTBACK; 451 PUTBACK;
452 // G_SCALAR ensures that return value is 1
295 call_sv ((SV *)GvCV (to_cbor), G_SCALAR); 453 call_sv ((SV *)GvCV (method), G_SCALAR);
296 SPAGAIN; 454 SPAGAIN;
297 455
298 // catch this surprisingly common error 456 // catch this surprisingly common error
299 if (SvROK (TOPs) && SvRV (TOPs) == sv) 457 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))); 458 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (stash));
301 459
302 sv = POPs; 460 encode_sv (enc, POPs);
461
303 PUTBACK; 462 PUTBACK;
304 463
305 encode_sv (enc, sv);
306
307 FREETMPS; LEAVE; 464 FREETMPS; LEAVE;
308 } 465 }
466 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
467 {
468 dSP;
469
470 ENTER; SAVETMPS; PUSHMARK (SP);
471 EXTEND (SP, 2);
472 // we re-bless the reference to get overload and other niceties right
473 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
474 PUSHs (sv_cbor);
475
476 PUTBACK;
477 int count = call_sv ((SV *)GvCV (method), G_ARRAY);
478 SPAGAIN;
479
480 // catch this surprisingly common error
481 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
482 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
483
484 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
485 encode_uint (enc, MAJOR_ARRAY, count + 1);
486 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
487
488 while (count)
489 encode_sv (enc, SP[1 - count--]);
490
491 PUTBACK;
492
493 FREETMPS; LEAVE;
494 }
309 else 495 else
310 croak ("encountered object '%s', but no TO_CBOR method available on it", 496 croak ("encountered object '%s', but no TO_CBOR or FREEZE methods available on it",
311 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 497 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
312 }
313 } 498 }
314 else if (svt == SVt_PVHV) 499 else if (svt == SVt_PVHV)
315 encode_hv (enc, (HV *)sv); 500 encode_hv (enc, (HV *)sv);
316 else if (svt == SVt_PVAV) 501 else if (svt == SVt_PVAV)
317 encode_av (enc, (AV *)sv); 502 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 503 else
336 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 504 {
337 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 505 encode_tag (enc, CBOR_TAG_INDIRECTION);
506 encode_sv (enc, sv);
507 }
338} 508}
339 509
340static void 510static void
341encode_nv (enc_t *enc, SV *sv) 511encode_nv (enc_t *enc, SV *sv)
342{ 512{
343 double nv = SvNVX (sv); 513 double nv = SvNVX (sv);
344 514
345 need (enc, 9); 515 need (enc, 9);
346 516
347 if (ecb_expect_false (nv == (U32)nv)) 517 if (ecb_expect_false (nv == (NV)(U32)nv))
348 encode_uint (enc, 0x00, (U32)nv); 518 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
349 //TODO: maybe I32? 519 //TODO: maybe I32?
350 else if (ecb_expect_false (nv == (float)nv)) 520 else if (ecb_expect_false (nv == (float)nv))
351 { 521 {
352 uint32_t fp = ecb_float_to_binary32 (nv); 522 uint32_t fp = ecb_float_to_binary32 (nv);
353 523
354 *enc->cur++ = 0xe0 | 26; 524 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
355 525
356 if (!ecb_big_endian ()) 526 if (!ecb_big_endian ())
357 fp = ecb_bswap32 (fp); 527 fp = ecb_bswap32 (fp);
358 528
359 memcpy (enc->cur, &fp, 4); 529 memcpy (enc->cur, &fp, 4);
361 } 531 }
362 else 532 else
363 { 533 {
364 uint64_t fp = ecb_double_to_binary64 (nv); 534 uint64_t fp = ecb_double_to_binary64 (nv);
365 535
366 *enc->cur++ = 0xe0 | 27; 536 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
367 537
368 if (!ecb_big_endian ()) 538 if (!ecb_big_endian ())
369 fp = ecb_bswap64 (fp); 539 fp = ecb_bswap64 (fp);
370 540
371 memcpy (enc->cur, &fp, 8); 541 memcpy (enc->cur, &fp, 8);
380 550
381 if (SvPOKp (sv)) 551 if (SvPOKp (sv))
382 { 552 {
383 STRLEN len; 553 STRLEN len;
384 char *str = SvPV (sv, len); 554 char *str = SvPV (sv, len);
385 encode_str (enc, SvUTF8 (sv), str, len); 555 encode_strref (enc, SvUTF8 (sv), str, len);
386 } 556 }
387 else if (SvNOKp (sv)) 557 else if (SvNOKp (sv))
388 encode_nv (enc, sv); 558 encode_nv (enc, sv);
389 else if (SvIOKp (sv)) 559 else if (SvIOKp (sv))
390 { 560 {
391 if (SvIsUV (sv)) 561 if (SvIsUV (sv))
392 encode_uint (enc, 0x00, SvUVX (sv)); 562 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
393 else if (SvIVX (sv) >= 0) 563 else if (SvIVX (sv) >= 0)
394 encode_uint (enc, 0x00, SvIVX (sv)); 564 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
395 else 565 else
396 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 566 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
397 } 567 }
398 else if (SvROK (sv)) 568 else if (SvROK (sv))
399 encode_rv (enc, SvRV (sv)); 569 encode_rv (enc, SvRV (sv));
400 else if (!SvOK (sv)) 570 else if (!SvOK (sv))
401 encode_ch (enc, 0xe0 | 22); 571 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
402 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 572 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
403 encode_ch (enc, 0xe0 | 23); 573 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
404 else 574 else
405 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 575 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
406 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 576 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
407} 577}
408 578
409static SV * 579static SV *
410encode_cbor (SV *scalar, CBOR *cbor) 580encode_cbor (SV *scalar, CBOR *cbor)
411{ 581{
412 enc_t enc; 582 enc_t enc = { };
413 583
414 enc.cbor = *cbor; 584 enc.cbor = *cbor;
415 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 585 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
416 enc.cur = SvPVX (enc.sv); 586 enc.cur = SvPVX (enc.sv);
417 enc.end = SvEND (enc.sv); 587 enc.end = SvEND (enc.sv);
418 enc.depth = 0;
419 588
420 SvPOK_only (enc.sv); 589 SvPOK_only (enc.sv);
590
591 if (cbor->flags & F_PACK_STRINGS)
592 {
593 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
594 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
595 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
596 }
597
421 encode_sv (&enc, scalar); 598 encode_sv (&enc, scalar);
422 599
423 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 600 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
424 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 601 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
425 602
439 U8 *end; // end of input string 616 U8 *end; // end of input string
440 const char *err; // parse error, if != 0 617 const char *err; // parse error, if != 0
441 CBOR cbor; 618 CBOR cbor;
442 U32 depth; // recursion depth 619 U32 depth; // recursion depth
443 U32 maxdepth; // recursion depth limit 620 U32 maxdepth; // recursion depth limit
621 AV *shareable;
622 AV *stringref;
623 SV *decode_tagged;
444} dec_t; 624} dec_t;
445 625
446#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 626#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE
447 627
448#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 628#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 631#define DEC_DEC_DEPTH --dec->depth
452 632
453static UV 633static UV
454decode_uint (dec_t *dec) 634decode_uint (dec_t *dec)
455{ 635{
456 switch (*dec->cur & 31) 636 U8 m = *dec->cur & MINOR_MASK;
457 { 637 ++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;
462 638
463 case 24: 639 if (ecb_expect_true (m < LENGTH_EXT1))
640 return m;
641 else if (ecb_expect_true (m == LENGTH_EXT1))
642 {
464 WANT (2); 643 WANT (1);
465 dec->cur += 2; 644 dec->cur += 1;
466 return dec->cur[-1]; 645 return dec->cur[-1];
467 646 }
468 case 25: 647 else if (ecb_expect_true (m == LENGTH_EXT2))
648 {
469 WANT (3); 649 WANT (2);
470 dec->cur += 3; 650 dec->cur += 2;
471 return (((UV)dec->cur[-2]) << 8) 651 return (((UV)dec->cur[-2]) << 8)
472 | ((UV)dec->cur[-1]); 652 | ((UV)dec->cur[-1]);
473 653 }
474 case 26: 654 else if (ecb_expect_true (m == LENGTH_EXT4))
655 {
475 WANT (5); 656 WANT (4);
476 dec->cur += 5; 657 dec->cur += 4;
477 return (((UV)dec->cur[-4]) << 24) 658 return (((UV)dec->cur[-4]) << 24)
478 | (((UV)dec->cur[-3]) << 16) 659 | (((UV)dec->cur[-3]) << 16)
479 | (((UV)dec->cur[-2]) << 8) 660 | (((UV)dec->cur[-2]) << 8)
480 | ((UV)dec->cur[-1]); 661 | ((UV)dec->cur[-1]);
481 662 }
482 case 27: 663 else if (ecb_expect_true (m == LENGTH_EXT8))
664 {
483 WANT (9); 665 WANT (8);
484 dec->cur += 9; 666 dec->cur += 8;
667
668 return
669#if UVSIZE < 8
670 0
671#else
485 return (((UV)dec->cur[-8]) << 56) 672 (((UV)dec->cur[-8]) << 56)
486 | (((UV)dec->cur[-7]) << 48) 673 | (((UV)dec->cur[-7]) << 48)
487 | (((UV)dec->cur[-6]) << 40) 674 | (((UV)dec->cur[-6]) << 40)
488 | (((UV)dec->cur[-5]) << 32) 675 | (((UV)dec->cur[-5]) << 32)
676#endif
489 | (((UV)dec->cur[-4]) << 24) 677 | (((UV)dec->cur[-4]) << 24)
490 | (((UV)dec->cur[-3]) << 16) 678 | (((UV)dec->cur[-3]) << 16)
491 | (((UV)dec->cur[-2]) << 8) 679 | (((UV)dec->cur[-2]) << 8)
492 | ((UV)dec->cur[-1]); 680 | ((UV)dec->cur[-1]);
493 681 }
494 default: 682 else
495 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 683 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
496 }
497 684
498fail: 685fail:
499 return 0; 686 return 0;
500} 687}
501 688
506{ 693{
507 AV *av = newAV (); 694 AV *av = newAV ();
508 695
509 DEC_INC_DEPTH; 696 DEC_INC_DEPTH;
510 697
511 if ((*dec->cur & 31) == 31) 698 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
512 { 699 {
513 ++dec->cur; 700 ++dec->cur;
514 701
515 for (;;) 702 for (;;)
516 { 703 {
517 WANT (1); 704 WANT (1);
518 705
519 if (*dec->cur == (0xe0 | 31)) 706 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
520 { 707 {
521 ++dec->cur; 708 ++dec->cur;
522 break; 709 break;
523 } 710 }
524 711
527 } 714 }
528 else 715 else
529 { 716 {
530 int i, len = decode_uint (dec); 717 int i, len = decode_uint (dec);
531 718
719 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); 720 av_fill (av, len - 1);
533 721
534 for (i = 0; i < len; ++i) 722 for (i = 0; i < len; ++i)
535 AvARRAY (av)[i] = decode_sv (dec); 723 AvARRAY (av)[i] = decode_sv (dec);
536 } 724 }
542 SvREFCNT_dec (av); 730 SvREFCNT_dec (av);
543 DEC_DEC_DEPTH; 731 DEC_DEC_DEPTH;
544 return &PL_sv_undef; 732 return &PL_sv_undef;
545} 733}
546 734
735static void
736decode_he (dec_t *dec, HV *hv)
737{
738 // for speed reasons, we specialcase single-string
739 // byte or utf-8 strings as keys, but only when !stringref
740
741 if (ecb_expect_true (!dec->stringref))
742 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
743 {
744 I32 len = decode_uint (dec);
745 char *key = (char *)dec->cur;
746
747 dec->cur += len;
748
749 hv_store (hv, key, len, decode_sv (dec), 0);
750
751 return;
752 }
753 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
754 {
755 I32 len = decode_uint (dec);
756 char *key = (char *)dec->cur;
757
758 dec->cur += len;
759
760 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
761 if (!is_utf8_string (key, len))
762 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
763
764 hv_store (hv, key, -len, decode_sv (dec), 0);
765
766 return;
767 }
768
769 SV *k = decode_sv (dec);
770 SV *v = decode_sv (dec);
771
772 hv_store_ent (hv, k, v, 0);
773 SvREFCNT_dec (k);
774
775fail:
776 ;
777}
778
547static SV * 779static SV *
548decode_hv (dec_t *dec) 780decode_hv (dec_t *dec)
549{ 781{
550 HV *hv = newHV (); 782 HV *hv = newHV ();
551 783
552 DEC_INC_DEPTH; 784 DEC_INC_DEPTH;
553 785
554 if ((*dec->cur & 31) == 31) 786 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
555 { 787 {
556 ++dec->cur; 788 ++dec->cur;
557 789
558 for (;;) 790 for (;;)
559 { 791 {
560 WANT (1); 792 WANT (1);
561 793
562 if (*dec->cur == (0xe0 | 31)) 794 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
563 { 795 {
564 ++dec->cur; 796 ++dec->cur;
565 break; 797 break;
566 } 798 }
567 799
568 SV *k = decode_sv (dec); 800 decode_he (dec, hv);
569 SV *v = decode_sv (dec);
570
571 hv_store_ent (hv, k, v, 0);
572 } 801 }
573 } 802 }
574 else 803 else
575 { 804 {
576 int len = decode_uint (dec); 805 int pairs = decode_uint (dec);
577 806
578 while (len--) 807 while (pairs--)
579 { 808 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 } 809 }
586 810
587 DEC_DEC_DEPTH; 811 DEC_DEC_DEPTH;
588 return newRV_noinc ((SV *)hv); 812 return newRV_noinc ((SV *)hv);
589 813
596static SV * 820static SV *
597decode_str (dec_t *dec, int utf8) 821decode_str (dec_t *dec, int utf8)
598{ 822{
599 SV *sv = 0; 823 SV *sv = 0;
600 824
601 if ((*dec->cur & 31) == 31) 825 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
602 { 826 {
827 // indefinite length strings
603 ++dec->cur; 828 ++dec->cur;
604 829
830 U8 major = *dec->cur & MAJOR_MISC;
831
605 sv = newSVpvn ("", 0); 832 sv = newSVpvn ("", 0);
606 833
607 // not very fast, and certainly not robust against illegal input
608 for (;;) 834 for (;;)
609 { 835 {
610 WANT (1); 836 WANT (1);
611 837
612 if (*dec->cur == (0xe0 | 31)) 838 if ((*dec->cur - major) > LENGTH_EXT8)
839 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
613 { 840 {
614 ++dec->cur; 841 ++dec->cur;
615 break; 842 break;
616 } 843 }
844 else
845 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
617 846
618 sv_catsv (sv, decode_sv (dec)); 847 STRLEN len = decode_uint (dec);
848
849 WANT (len);
850 sv_catpvn (sv, dec->cur, len);
851 dec->cur += len;
619 } 852 }
620 } 853 }
621 else 854 else
622 { 855 {
623 STRLEN len = decode_uint (dec); 856 STRLEN len = decode_uint (dec);
624 857
625 WANT (len); 858 WANT (len);
626 sv = newSVpvn (dec->cur, len); 859 sv = newSVpvn (dec->cur, len);
627 dec->cur += len; 860 dec->cur += len;
861
862 if (ecb_expect_false (dec->stringref)
863 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
864 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
628 } 865 }
629 866
630 if (utf8) 867 if (utf8)
868 {
869 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
870 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
871 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
872
631 SvUTF8_on (sv); 873 SvUTF8_on (sv);
874 }
632 875
633 return sv; 876 return sv;
634 877
635fail: 878fail:
636 SvREFCNT_dec (sv); 879 SvREFCNT_dec (sv);
638} 881}
639 882
640static SV * 883static SV *
641decode_tagged (dec_t *dec) 884decode_tagged (dec_t *dec)
642{ 885{
886 SV *sv = 0;
643 UV tag = decode_uint (dec); 887 UV tag = decode_uint (dec);
888
889 WANT (1);
890
891 switch (tag)
892 {
893 case CBOR_TAG_MAGIC:
644 SV *sv = decode_sv (dec); 894 sv = decode_sv (dec);
895 break;
645 896
646 if (tag == CBOR_TAG_MAGIC) 897 case CBOR_TAG_INDIRECTION:
647 return sv; 898 sv = newRV_noinc (decode_sv (dec));
899 break;
648 900
649 if (tag == CBOR_TAG_PERL_OBJECT) 901 case CBOR_TAG_STRINGREF_NAMESPACE:
650 { 902 {
903 ENTER; SAVETMPS;
904
905 SAVESPTR (dec->stringref);
906 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
907
908 sv = decode_sv (dec);
909
910 FREETMPS; LEAVE;
911 }
912 break;
913
914 case CBOR_TAG_STRINGREF:
915 {
916 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
917 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
918
919 UV idx = decode_uint (dec);
920
921 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
922 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
923
924 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
925 }
926 break;
927
928 case CBOR_TAG_VALUE_SHAREABLE:
929 {
930 if (ecb_expect_false (!dec->shareable))
931 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
932
933 if (dec->cbor.flags & F_ALLOW_CYCLES)
934 {
935 sv = newSV (0);
936 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
937
938 SV *osv = decode_sv (dec);
939 sv_setsv (sv, osv);
940 SvREFCNT_dec_NN (osv);
941 }
942 else
943 {
944 av_push (dec->shareable, &PL_sv_undef);
945 int idx = AvFILLp (dec->shareable);
946 sv = decode_sv (dec);
947 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
948 }
949 }
950 break;
951
952 case CBOR_TAG_VALUE_SHAREDREF:
953 {
954 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
955 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
956
957 UV idx = decode_uint (dec);
958
959 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
960 ERR ("corrupted CBOR data (sharedref index out of bounds)");
961
962 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
963
964 if (sv == &PL_sv_undef)
965 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
966 }
967 break;
968
969 case CBOR_TAG_PERL_OBJECT:
970 {
971 sv = decode_sv (dec);
972
651 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 973 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
652 ERR ("corrupted CBOR data (non-array perl object)"); 974 ERR ("corrupted CBOR data (non-array perl object)");
975
976 AV *av = (AV *)SvRV (sv);
977 int len = av_len (av) + 1;
978 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
979
980 if (!stash)
981 ERR ("cannot decode perl-object (package does not exist)");
982
983 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
984
985 if (!method)
986 ERR ("cannot decode perl-object (package does not have a THAW method)");
987
988 dSP;
989
990 ENTER; SAVETMPS; PUSHMARK (SP);
991 EXTEND (SP, len + 1);
992 // we re-bless the reference to get overload and other niceties right
993 PUSHs (*av_fetch (av, 0, 1));
994 PUSHs (sv_cbor);
995
996 int i;
997
998 for (i = 1; i < len; ++i)
999 PUSHs (*av_fetch (av, i, 1));
1000
1001 PUTBACK;
1002 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
1003 SPAGAIN;
1004
1005 if (SvTRUE (ERRSV))
1006 {
1007 FREETMPS; LEAVE;
1008 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1009 }
1010
1011 SvREFCNT_dec (sv);
1012 sv = SvREFCNT_inc (POPs);
1013
1014 PUTBACK;
1015
1016 FREETMPS; LEAVE;
653 1017 }
654 // TODO 1018 break;
655 }
656 1019
1020 default:
1021 {
1022 sv = decode_sv (dec);
1023
1024 dSP;
1025 ENTER; SAVETMPS; PUSHMARK (SP);
1026 EXTEND (SP, 2);
1027 PUSHs (newSVuv (tag));
1028 PUSHs (sv);
1029
1030 PUTBACK;
1031 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1032 SPAGAIN;
1033
1034 if (SvTRUE (ERRSV))
1035 {
1036 FREETMPS; LEAVE;
1037 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1038 }
1039
1040 if (count)
1041 {
1042 SvREFCNT_dec (sv);
1043 sv = SvREFCNT_inc (POPs);
1044 }
1045 else
1046 {
657 AV *av = newAV (); 1047 AV *av = newAV ();
658 av_push (av, newSVuv (tag)); 1048 av_push (av, newSVuv (tag));
659 av_push (av, sv); 1049 av_push (av, sv);
660 1050
661 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1051 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
662 ? cbor_tagged_stash 1052 ? cbor_tagged_stash
663 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1053 : gv_stashpv ("CBOR::XS::Tagged" , 1);
664
665 return sv_bless (newRV_noinc ((SV *)av), tagged_stash); 1054 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
1055 }
1056
1057 PUTBACK;
1058
1059 FREETMPS; LEAVE;
1060 }
1061 break;
1062 }
1063
1064 return sv;
666 1065
667fail: 1066fail:
668 SvREFCNT_dec (sv); 1067 SvREFCNT_dec (sv);
669 return &PL_sv_undef; 1068 return &PL_sv_undef;
670} 1069}
672static SV * 1071static SV *
673decode_sv (dec_t *dec) 1072decode_sv (dec_t *dec)
674{ 1073{
675 WANT (1); 1074 WANT (1);
676 1075
677 switch (*dec->cur >> 5) 1076 switch (*dec->cur >> MAJOR_SHIFT)
678 { 1077 {
679 case 0: // unsigned int 1078 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
680 return newSVuv (decode_uint (dec)); 1079 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
681 case 1: // negative int 1080 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
682 return newSViv (-1 - (IV)decode_uint (dec)); 1081 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
683 case 2: // octet string 1082 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
684 return decode_str (dec, 0); 1083 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
685 case 3: // utf-8 string 1084 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
686 return decode_str (dec, 1); 1085
687 case 4: // array 1086 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) 1087 switch (*dec->cur++ & MINOR_MASK)
695 { 1088 {
696 case 20: 1089 case SIMPLE_FALSE:
697#if CBOR_SLOW 1090#if CBOR_SLOW
698 cbor_false = get_bool ("CBOR::XS::false"); 1091 types_false = get_bool ("Types::Serialiser::false");
699#endif 1092#endif
700 return newSVsv (cbor_false); 1093 return newSVsv (types_false);
701 case 21: 1094 case SIMPLE_TRUE:
702#if CBOR_SLOW 1095#if CBOR_SLOW
703 cbor_true = get_bool ("CBOR::XS::true"); 1096 types_true = get_bool ("Types::Serialiser::true");
704#endif 1097#endif
705 return newSVsv (cbor_true); 1098 return newSVsv (types_true);
706 case 22: 1099 case SIMPLE_NULL:
707 return newSVsv (&PL_sv_undef); 1100 return newSVsv (&PL_sv_undef);
1101 case SIMPLE_UNDEF:
1102#if CBOR_SLOW
1103 types_error = get_bool ("Types::Serialiser::error");
1104#endif
1105 return newSVsv (types_error);
708 1106
709 case 25: 1107 case MISC_FLOAT16:
710 { 1108 {
711 WANT (2); 1109 WANT (2);
712 1110
713 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1111 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
714 dec->cur += 2; 1112 dec->cur += 2;
715 1113
716 return newSVnv (ecb_binary16_to_float (fp)); 1114 return newSVnv (ecb_binary16_to_float (fp));
717 } 1115 }
718 1116
719 case 26: 1117 case MISC_FLOAT32:
720 { 1118 {
721 uint32_t fp; 1119 uint32_t fp;
722 WANT (4); 1120 WANT (4);
723 memcpy (&fp, dec->cur, 4); 1121 memcpy (&fp, dec->cur, 4);
724 dec->cur += 4; 1122 dec->cur += 4;
727 fp = ecb_bswap32 (fp); 1125 fp = ecb_bswap32 (fp);
728 1126
729 return newSVnv (ecb_binary32_to_float (fp)); 1127 return newSVnv (ecb_binary32_to_float (fp));
730 } 1128 }
731 1129
732 case 27: 1130 case MISC_FLOAT64:
733 { 1131 {
734 uint64_t fp; 1132 uint64_t fp;
735 WANT (8); 1133 WANT (8);
736 memcpy (&fp, dec->cur, 8); 1134 memcpy (&fp, dec->cur, 8);
737 dec->cur += 8; 1135 dec->cur += 8;
740 fp = ecb_bswap64 (fp); 1138 fp = ecb_bswap64 (fp);
741 1139
742 return newSVnv (ecb_binary64_to_double (fp)); 1140 return newSVnv (ecb_binary64_to_double (fp));
743 } 1141 }
744 1142
745 // 0..19 unassigned 1143 // 0..19 unassigned simple
746 // 24 reserved + unassigned (reserved values are not encodable) 1144 // 24 reserved + unassigned simple (reserved values are not encodable)
1145 // 28-30 unassigned misc
1146 // 31 break code
747 default: 1147 default:
748 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1148 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
749 } 1149 }
750 1150
751 break; 1151 break;
752 } 1152 }
753 1153
756} 1156}
757 1157
758static SV * 1158static SV *
759decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1159decode_cbor (SV *string, CBOR *cbor, char **offset_return)
760{ 1160{
761 dec_t dec; 1161 dec_t dec = { };
762 SV *sv; 1162 SV *sv;
1163 STRLEN len;
1164 char *data = SvPVbyte (string, len);
763 1165
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) 1166 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", 1167 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); 1168 (unsigned long)len, (unsigned long)cbor->max_size);
795 }
796
797 sv_utf8_downgrade (string, 0);
798 1169
799 dec.cbor = *cbor; 1170 dec.cbor = *cbor;
800 dec.cur = (U8 *)SvPVX (string); 1171 dec.cur = (U8 *)data;
801 dec.end = (U8 *)SvEND (string); 1172 dec.end = (U8 *)data + len;
802 dec.err = 0;
803 dec.depth = 0;
804 1173
805 sv = decode_sv (&dec); 1174 sv = decode_sv (&dec);
806 1175
807 if (offset_return) 1176 if (offset_return)
808 *offset_return = dec.cur; 1177 *offset_return = dec.cur;
811 if (dec.cur != dec.end && !dec.err) 1180 if (dec.cur != dec.end && !dec.err)
812 dec.err = "garbage after CBOR object"; 1181 dec.err = "garbage after CBOR object";
813 1182
814 if (dec.err) 1183 if (dec.err)
815 { 1184 {
1185 if (dec.shareable)
1186 {
1187 // need to break cyclic links, which whould all be in shareable
1188 int i;
1189 SV **svp;
1190
1191 for (i = av_len (dec.shareable) + 1; i--; )
1192 if ((svp = av_fetch (dec.shareable, i, 0)))
1193 sv_setsv (*svp, &PL_sv_undef);
1194 }
1195
816 SvREFCNT_dec (sv); 1196 SvREFCNT_dec (sv);
817 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)SvPVX (string), (int)(uint8_t)*dec.cur); 1197 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
818 } 1198 }
819 1199
820 sv = sv_2mortal (sv); 1200 sv = sv_2mortal (sv);
821 1201
822 return sv; 1202 return sv;
823} 1203}
824 1204
1205/////////////////////////////////////////////////////////////////////////////
1206// incremental parser
1207
1208#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1209
1210// returns 0 for notyet, 1 for success or error
1211static int
1212incr_parse (CBOR *self, SV *cborstr)
1213{
1214 STRLEN cur;
1215 SvPV (cborstr, cur);
1216
1217 while (ecb_expect_true (self->incr_need <= cur))
1218 {
1219 // table of integer count bytes
1220 static I8 incr_len[MINOR_MASK + 1] = {
1221 0, 0, 0, 0, 0, 0, 0, 0,
1222 0, 0, 0, 0, 0, 0, 0, 0,
1223 0, 0, 0, 0, 0, 0, 0, 0,
1224 1, 2, 4, 8,-1,-1,-1,-2
1225 };
1226
1227 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1228 U8 m = *p & MINOR_MASK;
1229 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1230 I8 ilen = incr_len[m];
1231
1232 self->incr_need = self->incr_pos + 1;
1233
1234 if (ecb_expect_false (ilen < 0))
1235 {
1236 if (m != MINOR_INDEF)
1237 return 1; // error
1238
1239 if (*p == (MAJOR_MISC | MINOR_INDEF))
1240 {
1241 if (count >= 0)
1242 return 1; // error
1243
1244 count = 1;
1245 }
1246 else
1247 {
1248 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1249 count = -1;
1250 }
1251 }
1252 else
1253 {
1254 self->incr_need += ilen;
1255 if (ecb_expect_false (self->incr_need > cur))
1256 return 0;
1257
1258 int major = *p >> MAJOR_SHIFT;
1259
1260 switch (major)
1261 {
1262 case MAJOR_BYTES >> MAJOR_SHIFT:
1263 case MAJOR_TEXT >> MAJOR_SHIFT:
1264 case MAJOR_ARRAY >> MAJOR_SHIFT:
1265 case MAJOR_MAP >> MAJOR_SHIFT:
1266 {
1267 UV len;
1268
1269 if (ecb_expect_false (ilen))
1270 {
1271 len = 0;
1272
1273 do {
1274 len = (len << 8) | *++p;
1275 } while (--ilen);
1276 }
1277 else
1278 len = m;
1279
1280 switch (major)
1281 {
1282 case MAJOR_BYTES >> MAJOR_SHIFT:
1283 case MAJOR_TEXT >> MAJOR_SHIFT:
1284 self->incr_need += len;
1285 if (ecb_expect_false (self->incr_need > cur))
1286 return 0;
1287
1288 break;
1289
1290 case MAJOR_MAP >> MAJOR_SHIFT:
1291 len <<= 1;
1292 case MAJOR_ARRAY >> MAJOR_SHIFT:
1293 if (len)
1294 {
1295 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1296 count = len + 1;
1297 }
1298 break;
1299 }
1300 }
1301 }
1302 }
1303
1304 self->incr_pos = self->incr_need;
1305
1306 if (count > 0)
1307 {
1308 while (!--count)
1309 {
1310 if (!AvFILLp (self->incr_count))
1311 return 1; // done
1312
1313 SvREFCNT_dec_NN (av_pop (self->incr_count));
1314 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1315 }
1316
1317 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1318 }
1319 }
1320
1321 return 0;
1322}
1323
1324
825///////////////////////////////////////////////////////////////////////////// 1325/////////////////////////////////////////////////////////////////////////////
826// XS interface functions 1326// XS interface functions
827 1327
828MODULE = CBOR::XS PACKAGE = CBOR::XS 1328MODULE = CBOR::XS PACKAGE = CBOR::XS
829 1329
830BOOT: 1330BOOT:
831{ 1331{
832 cbor_stash = gv_stashpv ("CBOR::XS" , 1); 1332 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); 1333 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1);
835 1334
836 cbor_true = get_bool ("CBOR::XS::true"); 1335 types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
837 cbor_false = get_bool ("CBOR::XS::false"); 1336 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
1337
1338 types_true = get_bool ("Types::Serialiser::true" );
1339 types_false = get_bool ("Types::Serialiser::false");
1340 types_error = get_bool ("Types::Serialiser::error");
1341
1342 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1343
1344 sv_cbor = newSVpv ("CBOR", 0);
1345 SvREADONLY_on (sv_cbor);
838} 1346}
839 1347
840PROTOTYPES: DISABLE 1348PROTOTYPES: DISABLE
841 1349
842void CLONE (...) 1350void CLONE (...)
843 CODE: 1351 CODE:
844 cbor_stash = 0; 1352 cbor_stash = 0;
845 cbor_boolean_stash = 0;
846 cbor_tagged_stash = 0; 1353 cbor_tagged_stash = 0;
1354 types_error_stash = 0;
1355 types_boolean_stash = 0;
847 1356
848void new (char *klass) 1357void new (char *klass)
849 PPCODE: 1358 PPCODE:
850{ 1359{
851 SV *pv = NEWSV (0, sizeof (CBOR)); 1360 SV *pv = NEWSV (0, sizeof (CBOR));
859 1368
860void shrink (CBOR *self, int enable = 1) 1369void shrink (CBOR *self, int enable = 1)
861 ALIAS: 1370 ALIAS:
862 shrink = F_SHRINK 1371 shrink = F_SHRINK
863 allow_unknown = F_ALLOW_UNKNOWN 1372 allow_unknown = F_ALLOW_UNKNOWN
1373 allow_sharing = F_ALLOW_SHARING
1374 allow_cycles = F_ALLOW_CYCLES
1375 pack_strings = F_PACK_STRINGS
1376 validate_utf8 = F_VALIDATE_UTF8
864 PPCODE: 1377 PPCODE:
865{ 1378{
866 if (enable) 1379 if (enable)
867 self->flags |= ix; 1380 self->flags |= ix;
868 else 1381 else
873 1386
874void get_shrink (CBOR *self) 1387void get_shrink (CBOR *self)
875 ALIAS: 1388 ALIAS:
876 get_shrink = F_SHRINK 1389 get_shrink = F_SHRINK
877 get_allow_unknown = F_ALLOW_UNKNOWN 1390 get_allow_unknown = F_ALLOW_UNKNOWN
1391 get_allow_sharing = F_ALLOW_SHARING
1392 get_allow_cycles = F_ALLOW_CYCLES
1393 get_pack_strings = F_PACK_STRINGS
1394 get_validate_utf8 = F_VALIDATE_UTF8
878 PPCODE: 1395 PPCODE:
879 XPUSHs (boolSV (self->flags & ix)); 1396 XPUSHs (boolSV (self->flags & ix));
880 1397
881void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1398void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
882 PPCODE: 1399 PPCODE:
898 CODE: 1415 CODE:
899 RETVAL = self->max_size; 1416 RETVAL = self->max_size;
900 OUTPUT: 1417 OUTPUT:
901 RETVAL 1418 RETVAL
902 1419
903#if 0 //TODO 1420void filter (CBOR *self, SV *filter = 0)
904
905void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef)
906 PPCODE: 1421 PPCODE:
907{
908 SvREFCNT_dec (self->cb_object); 1422 SvREFCNT_dec (self->filter);
909 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; 1423 self->filter = filter ? newSVsv (filter) : filter;
910
911 XPUSHs (ST (0)); 1424 XPUSHs (ST (0));
912}
913 1425
914void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef) 1426SV *get_filter (CBOR *self)
915 PPCODE: 1427 CODE:
916{ 1428 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
917 if (!self->cb_sk_object) 1429 OUTPUT:
918 self->cb_sk_object = newHV (); 1430 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 1431
938void encode (CBOR *self, SV *scalar) 1432void encode (CBOR *self, SV *scalar)
939 PPCODE: 1433 PPCODE:
940 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; 1434 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
941 XPUSHs (scalar); 1435 XPUSHs (scalar);
954 EXTEND (SP, 2); 1448 EXTEND (SP, 2);
955 PUSHs (sv); 1449 PUSHs (sv);
956 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1450 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
957} 1451}
958 1452
959#if 0 1453void incr_parse (CBOR *self, SV *cborstr)
1454 ALIAS:
1455 incr_parse_multiple = 1
1456 PPCODE:
1457{
1458 if (SvUTF8 (cborstr))
1459 sv_utf8_downgrade (cborstr, 0);
1460
1461 if (!self->incr_count)
1462 {
1463 self->incr_count = newAV ();
1464 self->incr_pos = 0;
1465 self->incr_need = 1;
1466
1467 av_push (self->incr_count, newSViv (1));
1468 }
1469
1470 do
1471 {
1472 if (!incr_parse (self, cborstr))
1473 {
1474 if (self->incr_need > self->max_size && self->max_size)
1475 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1476 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1477
1478 break;
1479 }
1480
1481 SV *sv;
1482 char *offset;
1483
1484 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1485 XPUSHs (sv);
1486
1487 sv_chop (cborstr, offset);
1488
1489 av_clear (self->incr_count);
1490 av_push (self->incr_count, newSViv (1));
1491
1492 self->incr_pos = 0;
1493 self->incr_need = self->incr_pos + 1;
1494 }
1495 while (ix);
1496}
1497
1498void incr_reset (CBOR *self)
1499 CODE:
1500{
1501 SvREFCNT_dec (self->incr_count);
1502 self->incr_count = 0;
1503}
960 1504
961void DESTROY (CBOR *self) 1505void DESTROY (CBOR *self)
962 CODE: 1506 PPCODE:
963 SvREFCNT_dec (self->cb_sk_object); 1507 cbor_free (self);
964 SvREFCNT_dec (self->cb_object);
965
966#endif
967 1508
968PROTOTYPES: ENABLE 1509PROTOTYPES: ENABLE
969 1510
970void encode_cbor (SV *scalar) 1511void encode_cbor (SV *scalar)
1512 ALIAS:
1513 encode_cbor = 0
1514 encode_cbor_sharing = F_ALLOW_SHARING
971 PPCODE: 1515 PPCODE:
972{ 1516{
973 CBOR cbor; 1517 CBOR cbor;
974 cbor_init (&cbor); 1518 cbor_init (&cbor);
1519 cbor.flags |= ix;
975 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1520 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
976 XPUSHs (scalar); 1521 XPUSHs (scalar);
977} 1522}
978 1523
979void decode_cbor (SV *cborstr) 1524void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines