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 1.12 by root, Mon Oct 28 19:06:45 2013 UTC vs.
Revision 1.50 by root, Thu Feb 25 02:29:22 2016 UTC

7#include <stdlib.h> 7#include <stdlib.h>
8#include <stdio.h> 8#include <stdio.h>
9#include <limits.h> 9#include <limits.h>
10#include <float.h> 10#include <float.h>
11 11
12#define ECB_NO_THREADS 1
12#include "ecb.h" 13#include "ecb.h"
14
15// compatibility with perl <5.18
16#ifndef HvNAMELEN_get
17# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
18#endif
19#ifndef HvNAMELEN
20# define HvNAMELEN(hv) HvNAMELEN_get (hv)
21#endif
22#ifndef HvNAMEUTF8
23# define HvNAMEUTF8(hv) 0
24#endif
25#ifndef SvREFCNT_dec_NN
26# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)
27#endif
28
29// known major and minor types
30enum cbor_type
31{
32 MAJOR_SHIFT = 5,
33 MINOR_MASK = 0x1f,
34
35 MAJOR_POS_INT = 0 << MAJOR_SHIFT,
36 MAJOR_NEG_INT = 1 << MAJOR_SHIFT,
37 MAJOR_BYTES = 2 << MAJOR_SHIFT,
38 MAJOR_TEXT = 3 << MAJOR_SHIFT,
39 MAJOR_ARRAY = 4 << MAJOR_SHIFT,
40 MAJOR_MAP = 5 << MAJOR_SHIFT,
41 MAJOR_TAG = 6 << MAJOR_SHIFT,
42 MAJOR_MISC = 7 << MAJOR_SHIFT,
43
44 // INT/STRING/ARRAY/MAP subtypes
45 LENGTH_EXT1 = 24,
46 LENGTH_EXT2 = 25,
47 LENGTH_EXT4 = 26,
48 LENGTH_EXT8 = 27,
49
50 // SIMPLE types (effectively MISC subtypes)
51 SIMPLE_FALSE = 20,
52 SIMPLE_TRUE = 21,
53 SIMPLE_NULL = 22,
54 SIMPLE_UNDEF = 23,
55
56 // MISC subtype (unused)
57 MISC_EXT1 = 24,
58 MISC_FLOAT16 = 25,
59 MISC_FLOAT32 = 26,
60 MISC_FLOAT64 = 27,
61
62 // BYTES/TEXT/ARRAY/MAP
63 MINOR_INDEF = 31,
64};
13 65
14// known tags 66// known tags
15enum cbor_tag 67enum cbor_tag
16{ 68{
17 // inofficial extensions (pending iana registration) 69 // extensions
18 CBOR_TAG_PERL_OBJECT = 256, 70 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
19 CBOR_TAG_GENERIC_OBJECT = 257, 71 CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object
72 CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object
73 CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing
74 CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing
75 CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref
76 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection
20 77
21 // rfc7049 78 // rfc7049
22 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 79 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
23 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any 80 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
24 CBOR_TAG_POS_BIGNUM = 2, // byte string 81 CBOR_TAG_POS_BIGNUM = 2, // byte string
25 CBOR_TAG_NEG_BIGNUM = 3, // byte string 82 CBOR_TAG_NEG_BIGNUM = 3, // byte string
26 CBOR_TAG_DECIMAL = 4, // decimal fraction, array 83 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
27 CBOR_TAG_BIGFLOAT = 5, // array 84 CBOR_TAG_BIGFLOAT = 5, // array
28 85
29 CBOR_TAG_CONV_B64U = 21, // base64url, any 86 CBOR_TAG_CONV_B64U = 21, // base64url, any
30 CBOR_TAG_CONV_B64 = 22, // base64, any 87 CBOR_TAG_CONV_B64 = 22, // base64, any
31 CBOR_TAG_CONV_HEX = 23, // base16, any 88 CBOR_TAG_CONV_HEX = 23, // base16, any
32 CBOR_TAG_CBOR = 24, // embedded cbor, byte string 89 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
33 90
34 CBOR_TAG_URI = 32, // URI rfc3986, utf-8 91 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
35 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8 92 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
36 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8 93 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
37 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8 94 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
38 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 95 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
39 96
40 CBOR_TAG_MAGIC = 55799 // self-describe cbor 97 CBOR_TAG_MAGIC = 55799, // self-describe cbor
41}; 98};
42 99
43#define F_SHRINK 0x00000200UL 100#define F_SHRINK 0x00000001UL
44#define F_ALLOW_UNKNOWN 0x00002000UL 101#define F_ALLOW_UNKNOWN 0x00000002UL
102#define F_ALLOW_SHARING 0x00000004UL
103#define F_ALLOW_CYCLES 0x00000008UL
104#define F_PACK_STRINGS 0x00000010UL
105#define F_VALIDATE_UTF8 0x00000020UL
45 106
46#define INIT_SIZE 32 // initial scalar size to be allocated 107#define INIT_SIZE 32 // initial scalar size to be allocated
47 108
48#define SB do { 109#define SB do {
49#define SE } while (0) 110#define SE } while (0)
61# define CBOR_SLOW 0 122# define CBOR_SLOW 0
62# define CBOR_STASH cbor_stash 123# define CBOR_STASH cbor_stash
63#endif 124#endif
64 125
65static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS:: 126static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
66static SV *types_true, *types_false, *types_error, *sv_cbor; 127static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
67 128
68typedef struct { 129typedef struct {
69 U32 flags; 130 U32 flags;
70 U32 max_depth; 131 U32 max_depth;
71 STRLEN max_size; 132 STRLEN max_size;
133 SV *filter;
134
135 // for the incremental parser
136 STRLEN incr_pos; // the current offset into the text
137 STRLEN incr_need; // minimum bytes needed to decode
138 AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
72} CBOR; 139} CBOR;
73 140
74ecb_inline void 141ecb_inline void
75cbor_init (CBOR *cbor) 142cbor_init (CBOR *cbor)
76{ 143{
77 Zero (cbor, 1, CBOR); 144 Zero (cbor, 1, CBOR);
78 cbor->max_depth = 512; 145 cbor->max_depth = 512;
146}
147
148ecb_inline void
149cbor_free (CBOR *cbor)
150{
151 SvREFCNT_dec (cbor->filter);
152 SvREFCNT_dec (cbor->incr_count);
79} 153}
80 154
81///////////////////////////////////////////////////////////////////////////// 155/////////////////////////////////////////////////////////////////////////////
82// utility functions 156// utility functions
83 157
105 SvPV_renew (sv, SvCUR (sv) + 1); 179 SvPV_renew (sv, SvCUR (sv) + 1);
106#endif 180#endif
107 } 181 }
108} 182}
109 183
110///////////////////////////////////////////////////////////////////////////// 184// minimum length of a string to be registered for stringref
111// fp hell 185ecb_inline int
112 186minimum_string_length (UV idx)
113//TODO 187{
188 return idx > 23
189 ? idx > 0xffU
190 ? idx > 0xffffU
191 ? idx > 0xffffffffU
192 ? 11
193 : 7
194 : 5
195 : 4
196 : 3;
197}
114 198
115///////////////////////////////////////////////////////////////////////////// 199/////////////////////////////////////////////////////////////////////////////
116// encoder 200// encoder
117 201
118// structure used for encoding CBOR 202// structure used for encoding CBOR
121 char *cur; // SvPVX (sv) + current output position 205 char *cur; // SvPVX (sv) + current output position
122 char *end; // SvEND (sv) 206 char *end; // SvEND (sv)
123 SV *sv; // result scalar 207 SV *sv; // result scalar
124 CBOR cbor; 208 CBOR cbor;
125 U32 depth; // recursion level 209 U32 depth; // recursion level
210 HV *stringref[2]; // string => index, or 0 ([0] = bytes, [1] = utf-8)
211 UV stringref_idx;
212 HV *shareable; // ptr => index, or 0
213 UV shareable_idx;
126} enc_t; 214} enc_t;
127 215
128ecb_inline void 216ecb_inline void
129need (enc_t *enc, STRLEN len) 217need (enc_t *enc, STRLEN len)
130{ 218{
147static void 235static void
148encode_uint (enc_t *enc, int major, UV len) 236encode_uint (enc_t *enc, int major, UV len)
149{ 237{
150 need (enc, 9); 238 need (enc, 9);
151 239
152 if (len < 24) 240 if (ecb_expect_true (len < LENGTH_EXT1))
153 *enc->cur++ = major | len; 241 *enc->cur++ = major | len;
154 else if (len <= 0xff) 242 else if (ecb_expect_true (len <= 0xffU))
155 { 243 {
156 *enc->cur++ = major | 24; 244 *enc->cur++ = major | LENGTH_EXT1;
157 *enc->cur++ = len; 245 *enc->cur++ = len;
158 } 246 }
159 else if (len <= 0xffff) 247 else if (len <= 0xffffU)
160 { 248 {
161 *enc->cur++ = major | 25; 249 *enc->cur++ = major | LENGTH_EXT2;
162 *enc->cur++ = len >> 8; 250 *enc->cur++ = len >> 8;
163 *enc->cur++ = len; 251 *enc->cur++ = len;
164 } 252 }
165 else if (len <= 0xffffffff) 253 else if (len <= 0xffffffffU)
166 { 254 {
167 *enc->cur++ = major | 26; 255 *enc->cur++ = major | LENGTH_EXT4;
168 *enc->cur++ = len >> 24; 256 *enc->cur++ = len >> 24;
169 *enc->cur++ = len >> 16; 257 *enc->cur++ = len >> 16;
170 *enc->cur++ = len >> 8; 258 *enc->cur++ = len >> 8;
171 *enc->cur++ = len; 259 *enc->cur++ = len;
172 } 260 }
173 else 261 else
174 { 262 {
175 *enc->cur++ = major | 27; 263 *enc->cur++ = major | LENGTH_EXT8;
176 *enc->cur++ = len >> 56; 264 *enc->cur++ = len >> 56;
177 *enc->cur++ = len >> 48; 265 *enc->cur++ = len >> 48;
178 *enc->cur++ = len >> 40; 266 *enc->cur++ = len >> 40;
179 *enc->cur++ = len >> 32; 267 *enc->cur++ = len >> 32;
180 *enc->cur++ = len >> 24; 268 *enc->cur++ = len >> 24;
182 *enc->cur++ = len >> 8; 270 *enc->cur++ = len >> 8;
183 *enc->cur++ = len; 271 *enc->cur++ = len;
184 } 272 }
185} 273}
186 274
187static void 275ecb_inline void
276encode_tag (enc_t *enc, UV tag)
277{
278 encode_uint (enc, MAJOR_TAG, tag);
279}
280
281ecb_inline void
188encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 282encode_str (enc_t *enc, int utf8, char *str, STRLEN len)
189{ 283{
190 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 284 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
191 need (enc, len); 285 need (enc, len);
192 memcpy (enc->cur, str, len); 286 memcpy (enc->cur, str, len);
193 enc->cur += len; 287 enc->cur += len;
194} 288}
195 289
290static void
291encode_strref (enc_t *enc, int utf8, char *str, STRLEN len)
292{
293 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
294 {
295 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
296
297 if (SvOK (*svp))
298 {
299 // already registered, use stringref
300 encode_tag (enc, CBOR_TAG_STRINGREF);
301 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
302 return;
303 }
304 else if (len >= minimum_string_length (enc->stringref_idx))
305 {
306 // register only
307 sv_setuv (*svp, enc->stringref_idx);
308 ++enc->stringref_idx;
309 }
310 }
311
312 encode_str (enc, utf8, str, len);
313}
314
196static void encode_sv (enc_t *enc, SV *sv); 315static void encode_sv (enc_t *enc, SV *sv);
197 316
198static void 317static void
199encode_av (enc_t *enc, AV *av) 318encode_av (enc_t *enc, AV *av)
200{ 319{
203 if (enc->depth >= enc->cbor.max_depth) 322 if (enc->depth >= enc->cbor.max_depth)
204 croak (ERR_NESTING_EXCEEDED); 323 croak (ERR_NESTING_EXCEEDED);
205 324
206 ++enc->depth; 325 ++enc->depth;
207 326
208 encode_uint (enc, 0x80, len + 1); 327 encode_uint (enc, MAJOR_ARRAY, len + 1);
209 328
329 if (SvMAGICAL (av))
210 for (i = 0; i <= len; ++i) 330 for (i = 0; i <= len; ++i)
211 { 331 {
212 SV **svp = av_fetch (av, i, 0); 332 SV **svp = av_fetch (av, i, 0);
213 encode_sv (enc, svp ? *svp : &PL_sv_undef); 333 encode_sv (enc, svp ? *svp : &PL_sv_undef);
214 } 334 }
335 else
336 for (i = 0; i <= len; ++i)
337 {
338 SV *sv = AvARRAY (av)[i];
339 encode_sv (enc, sv ? sv : &PL_sv_undef);
340 }
215 341
216 --enc->depth; 342 --enc->depth;
217} 343}
218 344
219static void 345static void
228 354
229 int pairs = hv_iterinit (hv); 355 int pairs = hv_iterinit (hv);
230 int mg = SvMAGICAL (hv); 356 int mg = SvMAGICAL (hv);
231 357
232 if (mg) 358 if (mg)
233 encode_ch (enc, 0xa0 | 31); 359 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
234 else 360 else
235 encode_uint (enc, 0xa0, pairs); 361 encode_uint (enc, MAJOR_MAP, pairs);
236 362
237 while ((he = hv_iternext (hv))) 363 while ((he = hv_iternext (hv)))
238 { 364 {
239 if (HeKLEN (he) == HEf_SVKEY) 365 if (HeKLEN (he) == HEf_SVKEY)
240 encode_sv (enc, HeSVKEY (he)); 366 encode_sv (enc, HeSVKEY (he));
241 else 367 else
242 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 368 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
243 369
244 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 370 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
245 } 371 }
246 372
247 if (mg) 373 if (mg)
248 encode_ch (enc, 0xe0 | 31); 374 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
249 375
250 --enc->depth; 376 --enc->depth;
251} 377}
252 378
253// encode objects, arrays and special \0=false and \1=true values. 379// encode objects, arrays and special \0=false and \1=true values.
254static void 380static void
255encode_rv (enc_t *enc, SV *sv) 381encode_rv (enc_t *enc, SV *sv)
256{ 382{
257 svtype svt;
258
259 SvGETMAGIC (sv); 383 SvGETMAGIC (sv);
384
260 svt = SvTYPE (sv); 385 svtype svt = SvTYPE (sv);
261 386
262 if (ecb_expect_false (SvOBJECT (sv))) 387 if (ecb_expect_false (SvOBJECT (sv)))
263 { 388 {
264 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash 389 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
265 ? types_boolean_stash 390 ? types_boolean_stash
270 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 395 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
271 ? cbor_tagged_stash 396 ? cbor_tagged_stash
272 : gv_stashpv ("CBOR::XS::Tagged" , 1); 397 : gv_stashpv ("CBOR::XS::Tagged" , 1);
273 398
274 HV *stash = SvSTASH (sv); 399 HV *stash = SvSTASH (sv);
275 GV *method;
276 400
277 if (stash == boolean_stash) 401 if (stash == boolean_stash)
278 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 }
279 else if (stash == error_stash) 406 else if (stash == error_stash)
280 encode_ch (enc, 0xe0 | 23); 407 {
408 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
409 return;
410 }
281 else if (stash == tagged_stash) 411 else if (stash == tagged_stash)
282 { 412 {
283 if (svt != SVt_PVAV) 413 if (svt != SVt_PVAV)
284 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 414 croak ("encountered CBOR::XS::Tagged object that isn't an array");
285 415
286 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 416 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
287 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))
288 } 432 {
433 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
434 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
435 return;
436 }
437 else
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
289 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 450 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
290 { 451 {
291 dSP; 452 dSP;
292 453
293 ENTER; SAVETMPS; PUSHMARK (SP); 454 ENTER; SAVETMPS;
455 PUSHMARK (SP);
294 // we re-bless the reference to get overload and other niceties right 456 // we re-bless the reference to get overload and other niceties right
295 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 457 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
296 458
297 PUTBACK; 459 PUTBACK;
298 // G_SCALAR ensures that return value is 1 460 // G_SCALAR ensures that return value is 1
311 } 473 }
312 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) 474 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
313 { 475 {
314 dSP; 476 dSP;
315 477
316 ENTER; SAVETMPS; PUSHMARK (SP); 478 ENTER; SAVETMPS;
479 SAVESTACK_POS ();
480 PUSHMARK (SP);
317 EXTEND (SP, 2); 481 EXTEND (SP, 2);
318 // we re-bless the reference to get overload and other niceties right 482 // we re-bless the reference to get overload and other niceties right
319 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 483 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
320 PUSHs (sv_cbor); 484 PUSHs (sv_cbor);
321 485
325 489
326 // catch this surprisingly common error 490 // catch this surprisingly common error
327 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 491 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
328 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); 492 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
329 493
330 encode_uint (enc, 0xc0, CBOR_TAG_PERL_OBJECT); 494 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
331 encode_uint (enc, 0x80, count + 1); 495 encode_uint (enc, MAJOR_ARRAY, count + 1);
332 encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 496 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
333 497
334 while (count) 498 while (count)
335 encode_sv (enc, SP[1 - count--]); 499 encode_sv (enc, SP[1 - count--]);
336 500
337 PUTBACK; 501 PUTBACK;
344 } 508 }
345 else if (svt == SVt_PVHV) 509 else if (svt == SVt_PVHV)
346 encode_hv (enc, (HV *)sv); 510 encode_hv (enc, (HV *)sv);
347 else if (svt == SVt_PVAV) 511 else if (svt == SVt_PVAV)
348 encode_av (enc, (AV *)sv); 512 encode_av (enc, (AV *)sv);
349 else if (svt < SVt_PVAV)
350 {
351 STRLEN len = 0;
352 char *pv = svt ? SvPV (sv, len) : 0;
353
354 if (len == 1 && *pv == '1')
355 encode_ch (enc, 0xe0 | 21);
356 else if (len == 1 && *pv == '0')
357 encode_ch (enc, 0xe0 | 20);
358 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
359 encode_ch (enc, 0xe0 | 23);
360 else
361 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
362 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
363 }
364 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
365 encode_ch (enc, 0xe0 | 23);
366 else 513 else
367 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 514 {
368 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 515 encode_tag (enc, CBOR_TAG_INDIRECTION);
516 encode_sv (enc, sv);
517 }
369} 518}
370 519
371static void 520static void
372encode_nv (enc_t *enc, SV *sv) 521encode_nv (enc_t *enc, SV *sv)
373{ 522{
374 double nv = SvNVX (sv); 523 double nv = SvNVX (sv);
375 524
376 need (enc, 9); 525 need (enc, 9);
377 526
378 if (ecb_expect_false (nv == (U32)nv)) 527 if (ecb_expect_false (nv == (NV)(U32)nv))
379 encode_uint (enc, 0x00, (U32)nv); 528 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
380 //TODO: maybe I32? 529 //TODO: maybe I32?
381 else if (ecb_expect_false (nv == (float)nv)) 530 else if (ecb_expect_false (nv == (float)nv))
382 { 531 {
383 uint32_t fp = ecb_float_to_binary32 (nv); 532 uint32_t fp = ecb_float_to_binary32 (nv);
384 533
385 *enc->cur++ = 0xe0 | 26; 534 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
386 535
387 if (!ecb_big_endian ()) 536 if (!ecb_big_endian ())
388 fp = ecb_bswap32 (fp); 537 fp = ecb_bswap32 (fp);
389 538
390 memcpy (enc->cur, &fp, 4); 539 memcpy (enc->cur, &fp, 4);
392 } 541 }
393 else 542 else
394 { 543 {
395 uint64_t fp = ecb_double_to_binary64 (nv); 544 uint64_t fp = ecb_double_to_binary64 (nv);
396 545
397 *enc->cur++ = 0xe0 | 27; 546 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
398 547
399 if (!ecb_big_endian ()) 548 if (!ecb_big_endian ())
400 fp = ecb_bswap64 (fp); 549 fp = ecb_bswap64 (fp);
401 550
402 memcpy (enc->cur, &fp, 8); 551 memcpy (enc->cur, &fp, 8);
411 560
412 if (SvPOKp (sv)) 561 if (SvPOKp (sv))
413 { 562 {
414 STRLEN len; 563 STRLEN len;
415 char *str = SvPV (sv, len); 564 char *str = SvPV (sv, len);
416 encode_str (enc, SvUTF8 (sv), str, len); 565 encode_strref (enc, SvUTF8 (sv), str, len);
417 } 566 }
418 else if (SvNOKp (sv)) 567 else if (SvNOKp (sv))
419 encode_nv (enc, sv); 568 encode_nv (enc, sv);
420 else if (SvIOKp (sv)) 569 else if (SvIOKp (sv))
421 { 570 {
422 if (SvIsUV (sv)) 571 if (SvIsUV (sv))
423 encode_uint (enc, 0x00, SvUVX (sv)); 572 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
424 else if (SvIVX (sv) >= 0) 573 else if (SvIVX (sv) >= 0)
425 encode_uint (enc, 0x00, SvIVX (sv)); 574 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
426 else 575 else
427 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 576 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
428 } 577 }
429 else if (SvROK (sv)) 578 else if (SvROK (sv))
430 encode_rv (enc, SvRV (sv)); 579 encode_rv (enc, SvRV (sv));
431 else if (!SvOK (sv)) 580 else if (!SvOK (sv))
432 encode_ch (enc, 0xe0 | 22); 581 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
433 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 582 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
434 encode_ch (enc, 0xe0 | 23); 583 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
435 else 584 else
436 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 585 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
437 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 586 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
438} 587}
439 588
440static SV * 589static SV *
441encode_cbor (SV *scalar, CBOR *cbor) 590encode_cbor (SV *scalar, CBOR *cbor)
442{ 591{
443 enc_t enc; 592 enc_t enc = { 0 };
444 593
445 enc.cbor = *cbor; 594 enc.cbor = *cbor;
446 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 595 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
447 enc.cur = SvPVX (enc.sv); 596 enc.cur = SvPVX (enc.sv);
448 enc.end = SvEND (enc.sv); 597 enc.end = SvEND (enc.sv);
449 enc.depth = 0;
450 598
451 SvPOK_only (enc.sv); 599 SvPOK_only (enc.sv);
600
601 if (cbor->flags & F_PACK_STRINGS)
602 {
603 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
604 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
605 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
606 }
607
452 encode_sv (&enc, scalar); 608 encode_sv (&enc, scalar);
453 609
454 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 610 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
455 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 611 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
456 612
470 U8 *end; // end of input string 626 U8 *end; // end of input string
471 const char *err; // parse error, if != 0 627 const char *err; // parse error, if != 0
472 CBOR cbor; 628 CBOR cbor;
473 U32 depth; // recursion depth 629 U32 depth; // recursion depth
474 U32 maxdepth; // recursion depth limit 630 U32 maxdepth; // recursion depth limit
631 AV *shareable;
632 AV *stringref;
633 SV *decode_tagged;
475} dec_t; 634} dec_t;
476 635
477#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 636#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE
478 637
479#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 638#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data")
482#define DEC_DEC_DEPTH --dec->depth 641#define DEC_DEC_DEPTH --dec->depth
483 642
484static UV 643static UV
485decode_uint (dec_t *dec) 644decode_uint (dec_t *dec)
486{ 645{
487 switch (*dec->cur & 31) 646 U8 m = *dec->cur & MINOR_MASK;
488 { 647 ++dec->cur;
489 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
490 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
491 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
492 return *dec->cur++ & 31;
493 648
494 case 24: 649 if (ecb_expect_true (m < LENGTH_EXT1))
650 return m;
651 else if (ecb_expect_true (m == LENGTH_EXT1))
652 {
495 WANT (2); 653 WANT (1);
496 dec->cur += 2; 654 dec->cur += 1;
497 return dec->cur[-1]; 655 return dec->cur[-1];
498 656 }
499 case 25: 657 else if (ecb_expect_true (m == LENGTH_EXT2))
658 {
500 WANT (3); 659 WANT (2);
501 dec->cur += 3; 660 dec->cur += 2;
502 return (((UV)dec->cur[-2]) << 8) 661 return (((UV)dec->cur[-2]) << 8)
503 | ((UV)dec->cur[-1]); 662 | ((UV)dec->cur[-1]);
504 663 }
505 case 26: 664 else if (ecb_expect_true (m == LENGTH_EXT4))
665 {
506 WANT (5); 666 WANT (4);
507 dec->cur += 5; 667 dec->cur += 4;
508 return (((UV)dec->cur[-4]) << 24) 668 return (((UV)dec->cur[-4]) << 24)
509 | (((UV)dec->cur[-3]) << 16) 669 | (((UV)dec->cur[-3]) << 16)
510 | (((UV)dec->cur[-2]) << 8) 670 | (((UV)dec->cur[-2]) << 8)
511 | ((UV)dec->cur[-1]); 671 | ((UV)dec->cur[-1]);
512 672 }
513 case 27: 673 else if (ecb_expect_true (m == LENGTH_EXT8))
674 {
514 WANT (9); 675 WANT (8);
515 dec->cur += 9; 676 dec->cur += 8;
677
678 return
679#if UVSIZE < 8
680 0
681#else
516 return (((UV)dec->cur[-8]) << 56) 682 (((UV)dec->cur[-8]) << 56)
517 | (((UV)dec->cur[-7]) << 48) 683 | (((UV)dec->cur[-7]) << 48)
518 | (((UV)dec->cur[-6]) << 40) 684 | (((UV)dec->cur[-6]) << 40)
519 | (((UV)dec->cur[-5]) << 32) 685 | (((UV)dec->cur[-5]) << 32)
686#endif
520 | (((UV)dec->cur[-4]) << 24) 687 | (((UV)dec->cur[-4]) << 24)
521 | (((UV)dec->cur[-3]) << 16) 688 | (((UV)dec->cur[-3]) << 16)
522 | (((UV)dec->cur[-2]) << 8) 689 | (((UV)dec->cur[-2]) << 8)
523 | ((UV)dec->cur[-1]); 690 | ((UV)dec->cur[-1]);
524 691 }
525 default: 692 else
526 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 693 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
527 }
528 694
529fail: 695fail:
530 return 0; 696 return 0;
531} 697}
532 698
537{ 703{
538 AV *av = newAV (); 704 AV *av = newAV ();
539 705
540 DEC_INC_DEPTH; 706 DEC_INC_DEPTH;
541 707
542 if ((*dec->cur & 31) == 31) 708 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
543 { 709 {
544 ++dec->cur; 710 ++dec->cur;
545 711
546 for (;;) 712 for (;;)
547 { 713 {
548 WANT (1); 714 WANT (1);
549 715
550 if (*dec->cur == (0xe0 | 31)) 716 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
551 { 717 {
552 ++dec->cur; 718 ++dec->cur;
553 break; 719 break;
554 } 720 }
555 721
558 } 724 }
559 else 725 else
560 { 726 {
561 int i, len = decode_uint (dec); 727 int i, len = decode_uint (dec);
562 728
729 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
563 av_fill (av, len - 1); 730 av_fill (av, len - 1);
564 731
565 for (i = 0; i < len; ++i) 732 for (i = 0; i < len; ++i)
566 AvARRAY (av)[i] = decode_sv (dec); 733 AvARRAY (av)[i] = decode_sv (dec);
567 } 734 }
573 SvREFCNT_dec (av); 740 SvREFCNT_dec (av);
574 DEC_DEC_DEPTH; 741 DEC_DEC_DEPTH;
575 return &PL_sv_undef; 742 return &PL_sv_undef;
576} 743}
577 744
745static void
746decode_he (dec_t *dec, HV *hv)
747{
748 // for speed reasons, we specialcase single-string
749 // byte or utf-8 strings as keys, but only when !stringref
750
751 if (ecb_expect_true (!dec->stringref))
752 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
753 {
754 I32 len = decode_uint (dec);
755 char *key = (char *)dec->cur;
756
757 WANT (len);
758 dec->cur += len;
759
760 hv_store (hv, key, len, decode_sv (dec), 0);
761
762 return;
763 }
764 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
765 {
766 I32 len = decode_uint (dec);
767 char *key = (char *)dec->cur;
768
769 WANT (len);
770 dec->cur += len;
771
772 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
773 if (!is_utf8_string (key, len))
774 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
775
776 hv_store (hv, key, -len, decode_sv (dec), 0);
777
778 return;
779 }
780
781 SV *k = decode_sv (dec);
782 SV *v = decode_sv (dec);
783
784 hv_store_ent (hv, k, v, 0);
785 SvREFCNT_dec (k);
786
787fail:
788 ;
789}
790
578static SV * 791static SV *
579decode_hv (dec_t *dec) 792decode_hv (dec_t *dec)
580{ 793{
581 HV *hv = newHV (); 794 HV *hv = newHV ();
582 795
583 DEC_INC_DEPTH; 796 DEC_INC_DEPTH;
584 797
585 if ((*dec->cur & 31) == 31) 798 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
586 { 799 {
587 ++dec->cur; 800 ++dec->cur;
588 801
589 for (;;) 802 for (;;)
590 { 803 {
591 WANT (1); 804 WANT (1);
592 805
593 if (*dec->cur == (0xe0 | 31)) 806 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
594 { 807 {
595 ++dec->cur; 808 ++dec->cur;
596 break; 809 break;
597 } 810 }
598 811
599 SV *k = decode_sv (dec); 812 decode_he (dec, hv);
600 SV *v = decode_sv (dec);
601
602 hv_store_ent (hv, k, v, 0);
603 SvREFCNT_dec (k);
604 } 813 }
605 } 814 }
606 else 815 else
607 { 816 {
608 int len = decode_uint (dec); 817 int pairs = decode_uint (dec);
609 818
610 while (len--) 819 while (pairs--)
611 { 820 decode_he (dec, hv);
612 SV *k = decode_sv (dec);
613 SV *v = decode_sv (dec);
614
615 hv_store_ent (hv, k, v, 0);
616 SvREFCNT_dec (k);
617 }
618 } 821 }
619 822
620 DEC_DEC_DEPTH; 823 DEC_DEC_DEPTH;
621 return newRV_noinc ((SV *)hv); 824 return newRV_noinc ((SV *)hv);
622 825
629static SV * 832static SV *
630decode_str (dec_t *dec, int utf8) 833decode_str (dec_t *dec, int utf8)
631{ 834{
632 SV *sv = 0; 835 SV *sv = 0;
633 836
634 if ((*dec->cur & 31) == 31) 837 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
635 { 838 {
839 // indefinite length strings
636 ++dec->cur; 840 ++dec->cur;
637 841
842 U8 major = *dec->cur & MAJOR_MISC;
843
638 sv = newSVpvn ("", 0); 844 sv = newSVpvn ("", 0);
639 845
640 // not very fast, and certainly not robust against illegal input
641 for (;;) 846 for (;;)
642 { 847 {
643 WANT (1); 848 WANT (1);
644 849
645 if (*dec->cur == (0xe0 | 31)) 850 if ((*dec->cur - major) > LENGTH_EXT8)
851 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
646 { 852 {
647 ++dec->cur; 853 ++dec->cur;
648 break; 854 break;
649 } 855 }
856 else
857 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
650 858
651 sv_catsv (sv, decode_sv (dec)); 859 STRLEN len = decode_uint (dec);
860
861 WANT (len);
862 sv_catpvn (sv, dec->cur, len);
863 dec->cur += len;
652 } 864 }
653 } 865 }
654 else 866 else
655 { 867 {
656 STRLEN len = decode_uint (dec); 868 STRLEN len = decode_uint (dec);
657 869
658 WANT (len); 870 WANT (len);
659 sv = newSVpvn (dec->cur, len); 871 sv = newSVpvn (dec->cur, len);
660 dec->cur += len; 872 dec->cur += len;
873
874 if (ecb_expect_false (dec->stringref)
875 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
876 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
661 } 877 }
662 878
663 if (utf8) 879 if (utf8)
880 {
881 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
882 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
883 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
884
664 SvUTF8_on (sv); 885 SvUTF8_on (sv);
886 }
665 887
666 return sv; 888 return sv;
667 889
668fail: 890fail:
669 SvREFCNT_dec (sv); 891 SvREFCNT_dec (sv);
671} 893}
672 894
673static SV * 895static SV *
674decode_tagged (dec_t *dec) 896decode_tagged (dec_t *dec)
675{ 897{
898 SV *sv = 0;
676 UV tag = decode_uint (dec); 899 UV tag = decode_uint (dec);
900
901 WANT (1);
902
903 switch (tag)
904 {
905 case CBOR_TAG_MAGIC:
677 SV *sv = decode_sv (dec); 906 sv = decode_sv (dec);
907 break;
678 908
679 if (tag == CBOR_TAG_MAGIC) 909 case CBOR_TAG_INDIRECTION:
680 return sv; 910 sv = newRV_noinc (decode_sv (dec));
681 else if (tag == CBOR_TAG_PERL_OBJECT) 911 break;
682 { 912
913 case CBOR_TAG_STRINGREF_NAMESPACE:
914 {
915 ENTER; SAVETMPS;
916
917 SAVESPTR (dec->stringref);
918 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
919
920 sv = decode_sv (dec);
921
922 FREETMPS; LEAVE;
923 }
924 break;
925
926 case CBOR_TAG_STRINGREF:
927 {
928 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
929 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
930
931 UV idx = decode_uint (dec);
932
933 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
934 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
935
936 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
937 }
938 break;
939
940 case CBOR_TAG_VALUE_SHAREABLE:
941 {
942 if (ecb_expect_false (!dec->shareable))
943 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
944
945 if (dec->cbor.flags & F_ALLOW_CYCLES)
946 {
947 sv = newSV (0);
948 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
949
950 SV *osv = decode_sv (dec);
951 sv_setsv (sv, osv);
952 SvREFCNT_dec_NN (osv);
953 }
954 else
955 {
956 av_push (dec->shareable, &PL_sv_undef);
957 int idx = AvFILLp (dec->shareable);
958 sv = decode_sv (dec);
959 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
960 }
961 }
962 break;
963
964 case CBOR_TAG_VALUE_SHAREDREF:
965 {
966 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
967 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
968
969 UV idx = decode_uint (dec);
970
971 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
972 ERR ("corrupted CBOR data (sharedref index out of bounds)");
973
974 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
975
976 if (sv == &PL_sv_undef)
977 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
978 }
979 break;
980
981 case CBOR_TAG_PERL_OBJECT:
982 {
983 sv = decode_sv (dec);
984
683 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 985 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
684 ERR ("corrupted CBOR data (non-array perl object)"); 986 ERR ("corrupted CBOR data (non-array perl object)");
685 987
686 AV *av = (AV *)SvRV (sv); 988 AV *av = (AV *)SvRV (sv);
687 int len = av_len (av) + 1; 989 int len = av_len (av) + 1;
688 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0); 990 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
689 991
690 if (!stash) 992 if (!stash)
691 ERR ("cannot decode perl-object (package does not exist)"); 993 ERR ("cannot decode perl-object (package does not exist)");
692 994
693 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0); 995 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
694 996
695 if (!method) 997 if (!method)
696 ERR ("cannot decode perl-object (package does not have a THAW method)"); 998 ERR ("cannot decode perl-object (package does not have a THAW method)");
697 999
698 dSP; 1000 dSP;
699 1001
700 ENTER; SAVETMPS; PUSHMARK (SP); 1002 ENTER; SAVETMPS;
1003 PUSHMARK (SP);
701 EXTEND (SP, len + 1); 1004 EXTEND (SP, len + 1);
702 // we re-bless the reference to get overload and other niceties right 1005 // we re-bless the reference to get overload and other niceties right
703 PUSHs (*av_fetch (av, 0, 1)); 1006 PUSHs (*av_fetch (av, 0, 1));
704 PUSHs (sv_cbor); 1007 PUSHs (sv_cbor);
705 1008
706 int i; 1009 int i;
707 1010
708 for (i = 1; i < len; ++i) 1011 for (i = 1; i < len; ++i)
709 PUSHs (*av_fetch (av, i, 1)); 1012 PUSHs (*av_fetch (av, i, 1));
710 1013
711 PUTBACK; 1014 PUTBACK;
712 call_sv ((SV *)GvCV (method), G_SCALAR); 1015 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
713 SPAGAIN; 1016 SPAGAIN;
714 1017
1018 if (SvTRUE (ERRSV))
1019 {
1020 FREETMPS; LEAVE;
1021 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1022 }
1023
1024 SvREFCNT_dec (sv);
715 sv = SvREFCNT_inc (POPs); 1025 sv = SvREFCNT_inc (POPs);
716 1026
717 PUTBACK; 1027 PUTBACK;
718 1028
719 FREETMPS; LEAVE; 1029 FREETMPS; LEAVE;
1030 }
1031 break;
720 1032
721 return sv; 1033 default:
722 } 1034 {
723 else 1035 sv = decode_sv (dec);
724 { 1036
1037 dSP;
1038 ENTER; SAVETMPS;
1039 SAVESTACK_POS ();
1040 PUSHMARK (SP);
1041 EXTEND (SP, 2);
1042 PUSHs (newSVuv (tag));
1043 PUSHs (sv);
1044
1045 PUTBACK;
1046 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1047 SPAGAIN;
1048
1049 if (SvTRUE (ERRSV))
1050 {
1051 FREETMPS; LEAVE;
1052 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1053 }
1054
1055 if (count)
1056 {
1057 SvREFCNT_dec (sv);
1058 sv = SvREFCNT_inc (POPs);
1059 }
1060 else
1061 {
725 AV *av = newAV (); 1062 AV *av = newAV ();
726 av_push (av, newSVuv (tag)); 1063 av_push (av, newSVuv (tag));
727 av_push (av, sv); 1064 av_push (av, sv);
728 1065
729 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1066 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
730 ? cbor_tagged_stash 1067 ? cbor_tagged_stash
731 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1068 : gv_stashpv ("CBOR::XS::Tagged" , 1);
732
733 return sv_bless (newRV_noinc ((SV *)av), tagged_stash); 1069 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
1070 }
1071
1072 PUTBACK;
1073
1074 FREETMPS; LEAVE;
1075 }
1076 break;
734 } 1077 }
1078
1079 return sv;
735 1080
736fail: 1081fail:
737 SvREFCNT_dec (sv); 1082 SvREFCNT_dec (sv);
738 return &PL_sv_undef; 1083 return &PL_sv_undef;
739} 1084}
741static SV * 1086static SV *
742decode_sv (dec_t *dec) 1087decode_sv (dec_t *dec)
743{ 1088{
744 WANT (1); 1089 WANT (1);
745 1090
746 switch (*dec->cur >> 5) 1091 switch (*dec->cur >> MAJOR_SHIFT)
747 { 1092 {
748 case 0: // unsigned int 1093 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
749 return newSVuv (decode_uint (dec)); 1094 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
750 case 1: // negative int 1095 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
751 return newSViv (-1 - (IV)decode_uint (dec)); 1096 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
752 case 2: // octet string 1097 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
753 return decode_str (dec, 0); 1098 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
754 case 3: // utf-8 string 1099 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
755 return decode_str (dec, 1); 1100
756 case 4: // array 1101 case MAJOR_MISC >> MAJOR_SHIFT:
757 return decode_av (dec);
758 case 5: // map
759 return decode_hv (dec);
760 case 6: // tag
761 return decode_tagged (dec);
762 case 7: // misc
763 switch (*dec->cur++ & 31) 1102 switch (*dec->cur++ & MINOR_MASK)
764 { 1103 {
765 case 20: 1104 case SIMPLE_FALSE:
766#if CBOR_SLOW 1105#if CBOR_SLOW
767 types_false = get_bool ("Types::Serialiser::false"); 1106 types_false = get_bool ("Types::Serialiser::false");
768#endif 1107#endif
769 return newSVsv (types_false); 1108 return newSVsv (types_false);
770 case 21: 1109 case SIMPLE_TRUE:
771#if CBOR_SLOW 1110#if CBOR_SLOW
772 types_true = get_bool ("Types::Serialiser::true"); 1111 types_true = get_bool ("Types::Serialiser::true");
773#endif 1112#endif
774 return newSVsv (types_true); 1113 return newSVsv (types_true);
775 case 22: 1114 case SIMPLE_NULL:
776 return newSVsv (&PL_sv_undef); 1115 return newSVsv (&PL_sv_undef);
777 case 23: 1116 case SIMPLE_UNDEF:
778#if CBOR_SLOW 1117#if CBOR_SLOW
779 types_error = get_bool ("Types::Serialiser::error"); 1118 types_error = get_bool ("Types::Serialiser::error");
780#endif 1119#endif
781 return newSVsv (types_error); 1120 return newSVsv (types_error);
782 1121
783 case 25: 1122 case MISC_FLOAT16:
784 { 1123 {
785 WANT (2); 1124 WANT (2);
786 1125
787 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1126 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
788 dec->cur += 2; 1127 dec->cur += 2;
789 1128
790 return newSVnv (ecb_binary16_to_float (fp)); 1129 return newSVnv (ecb_binary16_to_float (fp));
791 } 1130 }
792 1131
793 case 26: 1132 case MISC_FLOAT32:
794 { 1133 {
795 uint32_t fp; 1134 uint32_t fp;
796 WANT (4); 1135 WANT (4);
797 memcpy (&fp, dec->cur, 4); 1136 memcpy (&fp, dec->cur, 4);
798 dec->cur += 4; 1137 dec->cur += 4;
801 fp = ecb_bswap32 (fp); 1140 fp = ecb_bswap32 (fp);
802 1141
803 return newSVnv (ecb_binary32_to_float (fp)); 1142 return newSVnv (ecb_binary32_to_float (fp));
804 } 1143 }
805 1144
806 case 27: 1145 case MISC_FLOAT64:
807 { 1146 {
808 uint64_t fp; 1147 uint64_t fp;
809 WANT (8); 1148 WANT (8);
810 memcpy (&fp, dec->cur, 8); 1149 memcpy (&fp, dec->cur, 8);
811 dec->cur += 8; 1150 dec->cur += 8;
814 fp = ecb_bswap64 (fp); 1153 fp = ecb_bswap64 (fp);
815 1154
816 return newSVnv (ecb_binary64_to_double (fp)); 1155 return newSVnv (ecb_binary64_to_double (fp));
817 } 1156 }
818 1157
819 // 0..19 unassigned 1158 // 0..19 unassigned simple
820 // 24 reserved + unassigned (reserved values are not encodable) 1159 // 24 reserved + unassigned simple (reserved values are not encodable)
1160 // 28-30 unassigned misc
1161 // 31 break code
821 default: 1162 default:
822 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1163 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
823 } 1164 }
824 1165
825 break; 1166 break;
826 } 1167 }
827 1168
830} 1171}
831 1172
832static SV * 1173static SV *
833decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1174decode_cbor (SV *string, CBOR *cbor, char **offset_return)
834{ 1175{
835 dec_t dec; 1176 dec_t dec = { 0 };
836 SV *sv; 1177 SV *sv;
1178 STRLEN len;
1179 char *data = SvPVbyte (string, len);
837 1180
838 /* work around bugs in 5.10 where manipulating magic values
839 * makes perl ignore the magic in subsequent accesses.
840 * also make a copy of non-PV values, to get them into a clean
841 * state (SvPV should do that, but it's buggy, see below).
842 */
843 /*SvGETMAGIC (string);*/
844 if (SvMAGICAL (string) || !SvPOK (string))
845 string = sv_2mortal (newSVsv (string));
846
847 SvUPGRADE (string, SVt_PV);
848
849 /* work around a bug in perl 5.10, which causes SvCUR to fail an
850 * assertion with -DDEBUGGING, although SvCUR is documented to
851 * return the xpv_cur field which certainly exists after upgrading.
852 * according to nicholas clark, calling SvPOK fixes this.
853 * But it doesn't fix it, so try another workaround, call SvPV_nolen
854 * and hope for the best.
855 * Damnit, SvPV_nolen still trips over yet another assertion. This
856 * assertion business is seriously broken, try yet another workaround
857 * for the broken -DDEBUGGING.
858 */
859 {
860#ifdef DEBUGGING
861 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
862#else
863 STRLEN offset = SvCUR (string);
864#endif
865
866 if (offset > cbor->max_size && cbor->max_size) 1181 if (len > cbor->max_size && cbor->max_size)
867 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu", 1182 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
868 (unsigned long)SvCUR (string), (unsigned long)cbor->max_size); 1183 (unsigned long)len, (unsigned long)cbor->max_size);
869 }
870
871 sv_utf8_downgrade (string, 0);
872 1184
873 dec.cbor = *cbor; 1185 dec.cbor = *cbor;
874 dec.cur = (U8 *)SvPVX (string); 1186 dec.cur = (U8 *)data;
875 dec.end = (U8 *)SvEND (string); 1187 dec.end = (U8 *)data + len;
876 dec.err = 0;
877 dec.depth = 0;
878 1188
879 sv = decode_sv (&dec); 1189 sv = decode_sv (&dec);
880 1190
881 if (offset_return) 1191 if (offset_return)
882 *offset_return = dec.cur; 1192 *offset_return = dec.cur;
885 if (dec.cur != dec.end && !dec.err) 1195 if (dec.cur != dec.end && !dec.err)
886 dec.err = "garbage after CBOR object"; 1196 dec.err = "garbage after CBOR object";
887 1197
888 if (dec.err) 1198 if (dec.err)
889 { 1199 {
1200 if (dec.shareable)
1201 {
1202 // need to break cyclic links, which whould all be in shareable
1203 int i;
1204 SV **svp;
1205
1206 for (i = av_len (dec.shareable) + 1; i--; )
1207 if ((svp = av_fetch (dec.shareable, i, 0)))
1208 sv_setsv (*svp, &PL_sv_undef);
1209 }
1210
890 SvREFCNT_dec (sv); 1211 SvREFCNT_dec (sv);
891 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)SvPVX (string), (int)(uint8_t)*dec.cur); 1212 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
892 } 1213 }
893 1214
894 sv = sv_2mortal (sv); 1215 sv = sv_2mortal (sv);
895 1216
896 return sv; 1217 return sv;
897} 1218}
898 1219
1220/////////////////////////////////////////////////////////////////////////////
1221// incremental parser
1222
1223#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1224
1225// returns 0 for notyet, 1 for success or error
1226static int
1227incr_parse (CBOR *self, SV *cborstr)
1228{
1229 STRLEN cur;
1230 SvPV (cborstr, cur);
1231
1232 while (ecb_expect_true (self->incr_need <= cur))
1233 {
1234 // table of integer count bytes
1235 static I8 incr_len[MINOR_MASK + 1] = {
1236 0, 0, 0, 0, 0, 0, 0, 0,
1237 0, 0, 0, 0, 0, 0, 0, 0,
1238 0, 0, 0, 0, 0, 0, 0, 0,
1239 1, 2, 4, 8,-1,-1,-1,-2
1240 };
1241
1242 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1243 U8 m = *p & MINOR_MASK;
1244 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1245 I8 ilen = incr_len[m];
1246
1247 self->incr_need = self->incr_pos + 1;
1248
1249 if (ecb_expect_false (ilen < 0))
1250 {
1251 if (m != MINOR_INDEF)
1252 return 1; // error
1253
1254 if (*p == (MAJOR_MISC | MINOR_INDEF))
1255 {
1256 if (count >= 0)
1257 return 1; // error
1258
1259 count = 1;
1260 }
1261 else
1262 {
1263 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1264 count = -1;
1265 }
1266 }
1267 else
1268 {
1269 self->incr_need += ilen;
1270 if (ecb_expect_false (self->incr_need > cur))
1271 return 0;
1272
1273 int major = *p >> MAJOR_SHIFT;
1274
1275 switch (major)
1276 {
1277 case MAJOR_TAG >> MAJOR_SHIFT:
1278 ++count; // tags merely prefix another value
1279 break;
1280
1281 case MAJOR_BYTES >> MAJOR_SHIFT:
1282 case MAJOR_TEXT >> MAJOR_SHIFT:
1283 case MAJOR_ARRAY >> MAJOR_SHIFT:
1284 case MAJOR_MAP >> MAJOR_SHIFT:
1285 {
1286 UV len;
1287
1288 if (ecb_expect_false (ilen))
1289 {
1290 len = 0;
1291
1292 do {
1293 len = (len << 8) | *++p;
1294 } while (--ilen);
1295 }
1296 else
1297 len = m;
1298
1299 switch (major)
1300 {
1301 case MAJOR_BYTES >> MAJOR_SHIFT:
1302 case MAJOR_TEXT >> MAJOR_SHIFT:
1303 self->incr_need += len;
1304 if (ecb_expect_false (self->incr_need > cur))
1305 return 0;
1306
1307 break;
1308
1309 case MAJOR_MAP >> MAJOR_SHIFT:
1310 len <<= 1;
1311 case MAJOR_ARRAY >> MAJOR_SHIFT:
1312 if (len)
1313 {
1314 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1315 count = len + 1;
1316 }
1317 break;
1318 }
1319 }
1320 }
1321 }
1322
1323 self->incr_pos = self->incr_need;
1324
1325 if (count > 0)
1326 {
1327 while (!--count)
1328 {
1329 if (!AvFILLp (self->incr_count))
1330 return 1; // done
1331
1332 SvREFCNT_dec_NN (av_pop (self->incr_count));
1333 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1334 }
1335
1336 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1337 }
1338 }
1339
1340 return 0;
1341}
1342
1343
899///////////////////////////////////////////////////////////////////////////// 1344/////////////////////////////////////////////////////////////////////////////
900// XS interface functions 1345// XS interface functions
901 1346
902MODULE = CBOR::XS PACKAGE = CBOR::XS 1347MODULE = CBOR::XS PACKAGE = CBOR::XS
903 1348
910 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1); 1355 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
911 1356
912 types_true = get_bool ("Types::Serialiser::true" ); 1357 types_true = get_bool ("Types::Serialiser::true" );
913 types_false = get_bool ("Types::Serialiser::false"); 1358 types_false = get_bool ("Types::Serialiser::false");
914 types_error = get_bool ("Types::Serialiser::error"); 1359 types_error = get_bool ("Types::Serialiser::error");
1360
1361 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
915 1362
916 sv_cbor = newSVpv ("CBOR", 0); 1363 sv_cbor = newSVpv ("CBOR", 0);
917 SvREADONLY_on (sv_cbor); 1364 SvREADONLY_on (sv_cbor);
918} 1365}
919 1366
940 1387
941void shrink (CBOR *self, int enable = 1) 1388void shrink (CBOR *self, int enable = 1)
942 ALIAS: 1389 ALIAS:
943 shrink = F_SHRINK 1390 shrink = F_SHRINK
944 allow_unknown = F_ALLOW_UNKNOWN 1391 allow_unknown = F_ALLOW_UNKNOWN
1392 allow_sharing = F_ALLOW_SHARING
1393 allow_cycles = F_ALLOW_CYCLES
1394 pack_strings = F_PACK_STRINGS
1395 validate_utf8 = F_VALIDATE_UTF8
945 PPCODE: 1396 PPCODE:
946{ 1397{
947 if (enable) 1398 if (enable)
948 self->flags |= ix; 1399 self->flags |= ix;
949 else 1400 else
954 1405
955void get_shrink (CBOR *self) 1406void get_shrink (CBOR *self)
956 ALIAS: 1407 ALIAS:
957 get_shrink = F_SHRINK 1408 get_shrink = F_SHRINK
958 get_allow_unknown = F_ALLOW_UNKNOWN 1409 get_allow_unknown = F_ALLOW_UNKNOWN
1410 get_allow_sharing = F_ALLOW_SHARING
1411 get_allow_cycles = F_ALLOW_CYCLES
1412 get_pack_strings = F_PACK_STRINGS
1413 get_validate_utf8 = F_VALIDATE_UTF8
959 PPCODE: 1414 PPCODE:
960 XPUSHs (boolSV (self->flags & ix)); 1415 XPUSHs (boolSV (self->flags & ix));
961 1416
962void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1417void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
963 PPCODE: 1418 PPCODE:
979 CODE: 1434 CODE:
980 RETVAL = self->max_size; 1435 RETVAL = self->max_size;
981 OUTPUT: 1436 OUTPUT:
982 RETVAL 1437 RETVAL
983 1438
984#if 0 //TODO 1439void filter (CBOR *self, SV *filter = 0)
985
986void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef)
987 PPCODE: 1440 PPCODE:
988{
989 SvREFCNT_dec (self->cb_object); 1441 SvREFCNT_dec (self->filter);
990 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; 1442 self->filter = filter ? newSVsv (filter) : filter;
991
992 XPUSHs (ST (0)); 1443 XPUSHs (ST (0));
993}
994 1444
995void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef) 1445SV *get_filter (CBOR *self)
996 PPCODE: 1446 CODE:
997{ 1447 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
998 if (!self->cb_sk_object) 1448 OUTPUT:
999 self->cb_sk_object = newHV (); 1449 RETVAL
1000
1001 if (SvOK (cb))
1002 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1003 else
1004 {
1005 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1006
1007 if (!HvKEYS (self->cb_sk_object))
1008 {
1009 SvREFCNT_dec (self->cb_sk_object);
1010 self->cb_sk_object = 0;
1011 }
1012 }
1013
1014 XPUSHs (ST (0));
1015}
1016
1017#endif
1018 1450
1019void encode (CBOR *self, SV *scalar) 1451void encode (CBOR *self, SV *scalar)
1020 PPCODE: 1452 PPCODE:
1021 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; 1453 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
1022 XPUSHs (scalar); 1454 XPUSHs (scalar);
1035 EXTEND (SP, 2); 1467 EXTEND (SP, 2);
1036 PUSHs (sv); 1468 PUSHs (sv);
1037 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1469 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1038} 1470}
1039 1471
1040#if 0 1472void incr_parse (CBOR *self, SV *cborstr)
1473 ALIAS:
1474 incr_parse_multiple = 1
1475 PPCODE:
1476{
1477 if (SvUTF8 (cborstr))
1478 sv_utf8_downgrade (cborstr, 0);
1479
1480 if (!self->incr_count)
1481 {
1482 self->incr_count = newAV ();
1483 self->incr_pos = 0;
1484 self->incr_need = 1;
1485
1486 av_push (self->incr_count, newSViv (1));
1487 }
1488
1489 do
1490 {
1491 if (!incr_parse (self, cborstr))
1492 {
1493 if (self->incr_need > self->max_size && self->max_size)
1494 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1495 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1496
1497 break;
1498 }
1499
1500 SV *sv;
1501 char *offset;
1502
1503 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1504 XPUSHs (sv);
1505
1506 sv_chop (cborstr, offset);
1507
1508 av_clear (self->incr_count);
1509 av_push (self->incr_count, newSViv (1));
1510
1511 self->incr_pos = 0;
1512 self->incr_need = self->incr_pos + 1;
1513 }
1514 while (ix);
1515}
1516
1517void incr_reset (CBOR *self)
1518 CODE:
1519{
1520 SvREFCNT_dec (self->incr_count);
1521 self->incr_count = 0;
1522}
1041 1523
1042void DESTROY (CBOR *self) 1524void DESTROY (CBOR *self)
1043 CODE: 1525 PPCODE:
1044 SvREFCNT_dec (self->cb_sk_object); 1526 cbor_free (self);
1045 SvREFCNT_dec (self->cb_object);
1046
1047#endif
1048 1527
1049PROTOTYPES: ENABLE 1528PROTOTYPES: ENABLE
1050 1529
1051void encode_cbor (SV *scalar) 1530void encode_cbor (SV *scalar)
1531 ALIAS:
1532 encode_cbor = 0
1533 encode_cbor_sharing = F_ALLOW_SHARING
1052 PPCODE: 1534 PPCODE:
1053{ 1535{
1054 CBOR cbor; 1536 CBOR cbor;
1055 cbor_init (&cbor); 1537 cbor_init (&cbor);
1538 cbor.flags |= ix;
1056 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1539 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1057 XPUSHs (scalar); 1540 XPUSHs (scalar);
1058} 1541}
1059 1542
1060void decode_cbor (SV *cborstr) 1543void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines