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 1.20 by root, Wed Nov 20 11:06:42 2013 UTC vs.
Revision 1.62 by root, Sat Nov 26 02:11:52 2016 UTC

6#include <string.h> 6#include <string.h>
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#include <inttypes.h>
11 12
13#define ECB_NO_THREADS 1
12#include "ecb.h" 14#include "ecb.h"
13 15
14// compatibility with perl <5.18 16// compatibility with perl <5.18
15#ifndef HvNAMELEN_get 17#ifndef HvNAMELEN_get
16# define HvNAMELEN_get(hv) strlen (HvNAME (hv)) 18# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
19# define HvNAMELEN(hv) HvNAMELEN_get (hv) 21# define HvNAMELEN(hv) HvNAMELEN_get (hv)
20#endif 22#endif
21#ifndef HvNAMEUTF8 23#ifndef HvNAMEUTF8
22# define HvNAMEUTF8(hv) 0 24# define HvNAMEUTF8(hv) 0
23#endif 25#endif
26#ifndef SvREFCNT_dec_NN
27# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)
28#endif
29
30// known major and minor types
31enum cbor_type
32{
33 MAJOR_SHIFT = 5,
34 MINOR_MASK = 0x1f,
35
36 MAJOR_POS_INT = 0 << MAJOR_SHIFT,
37 MAJOR_NEG_INT = 1 << MAJOR_SHIFT,
38 MAJOR_BYTES = 2 << MAJOR_SHIFT,
39 MAJOR_TEXT = 3 << MAJOR_SHIFT,
40 MAJOR_ARRAY = 4 << MAJOR_SHIFT,
41 MAJOR_MAP = 5 << MAJOR_SHIFT,
42 MAJOR_TAG = 6 << MAJOR_SHIFT,
43 MAJOR_MISC = 7 << MAJOR_SHIFT,
44
45 // INT/STRING/ARRAY/MAP subtypes
46 LENGTH_EXT1 = 24,
47 LENGTH_EXT2 = 25,
48 LENGTH_EXT4 = 26,
49 LENGTH_EXT8 = 27,
50
51 // SIMPLE types (effectively MISC subtypes)
52 SIMPLE_FALSE = 20,
53 SIMPLE_TRUE = 21,
54 SIMPLE_NULL = 22,
55 SIMPLE_UNDEF = 23,
56
57 // MISC subtype (unused)
58 MISC_EXT1 = 24,
59 MISC_FLOAT16 = 25,
60 MISC_FLOAT32 = 26,
61 MISC_FLOAT64 = 27,
62
63 // BYTES/TEXT/ARRAY/MAP
64 MINOR_INDEF = 31,
65};
24 66
25// known tags 67// known tags
26enum cbor_tag 68enum cbor_tag
27{ 69{
28 // inofficial extensions (pending iana registration) 70 // extensions
71 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
29 CBOR_TAG_PERL_OBJECT = 24, // http://cbor.schmorp.de/perl-object 72 CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object
30 CBOR_TAG_GENERIC_OBJECT = 25, // http://cbor.schmorp.de/generic-object 73 CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object
31 CBOR_TAG_VALUE_SHAREABLE = 26, // http://cbor.schmorp.de/value-sharing 74 CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing
32 CBOR_TAG_VALUE_SHAREDREF = 27, // http://cbor.schmorp.de/value-sharing 75 CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing
33 CBOR_TAG_STRINGREF_NAMESPACE = 65537, // http://cbor.schmorp.de/stringref 76 CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref
34 CBOR_TAG_STRINGREF = 28, // http://cbor.schmorp.de/stringref
35 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection 77 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection
36 78
37 // rfc7049 79 // rfc7049
38 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 80 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
39 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any 81 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
40 CBOR_TAG_POS_BIGNUM = 2, // byte string 82 CBOR_TAG_POS_BIGNUM = 2, // byte string
41 CBOR_TAG_NEG_BIGNUM = 3, // byte string 83 CBOR_TAG_NEG_BIGNUM = 3, // byte string
42 CBOR_TAG_DECIMAL = 4, // decimal fraction, array 84 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
43 CBOR_TAG_BIGFLOAT = 5, // array 85 CBOR_TAG_BIGFLOAT = 5, // array
44 86
45 CBOR_TAG_CONV_B64U = 21, // base64url, any 87 CBOR_TAG_CONV_B64U = 21, // base64url, any
46 CBOR_TAG_CONV_B64 = 22, // base64, any 88 CBOR_TAG_CONV_B64 = 22, // base64, any
47 CBOR_TAG_CONV_HEX = 23, // base16, any 89 CBOR_TAG_CONV_HEX = 23, // base16, any
48 CBOR_TAG_CBOR = 24, // embedded cbor, byte string 90 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
49 91
50 CBOR_TAG_URI = 32, // URI rfc3986, utf-8 92 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
51 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8 93 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
52 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8 94 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
53 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8 95 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
54 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 96 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
55 97
56 CBOR_TAG_MAGIC = 55799 // self-describe cbor 98 CBOR_TAG_MAGIC = 55799, // self-describe cbor
57}; 99};
58 100
59#define F_SHRINK 0x00000001UL 101#define F_SHRINK 0x00000001UL
60#define F_ALLOW_UNKNOWN 0x00000002UL 102#define F_ALLOW_UNKNOWN 0x00000002UL
61#define F_ALLOW_SHARING 0x00000004UL //TODO 103#define F_ALLOW_SHARING 0x00000004UL
104#define F_ALLOW_CYCLES 0x00000008UL
105#define F_FORBID_OBJECTS 0x00000010UL
62#define F_DEDUP_STRINGS 0x00000008UL //TODO 106#define F_PACK_STRINGS 0x00000020UL
63#define F_DEDUP_KEYS 0x00000010UL //TODO 107#define F_TEXT_KEYS 0x00000040UL
108#define F_TEXT_STRINGS 0x00000080UL
109#define F_VALIDATE_UTF8 0x00000100UL
64 110
65#define INIT_SIZE 32 // initial scalar size to be allocated 111#define INIT_SIZE 32 // initial scalar size to be allocated
66 112
67#define SB do { 113#define SB do {
68#define SE } while (0) 114#define SE } while (0)
80# define CBOR_SLOW 0 126# define CBOR_SLOW 0
81# define CBOR_STASH cbor_stash 127# define CBOR_STASH cbor_stash
82#endif 128#endif
83 129
84static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS:: 130static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
85static SV *types_true, *types_false, *types_error, *sv_cbor; 131static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
86 132
87typedef struct { 133typedef struct {
88 U32 flags; 134 U32 flags;
89 U32 max_depth; 135 U32 max_depth;
90 STRLEN max_size; 136 STRLEN max_size;
137 SV *filter;
138
139 // for the incremental parser
140 STRLEN incr_pos; // the current offset into the text
141 STRLEN incr_need; // minimum bytes needed to decode
142 AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
91} CBOR; 143} CBOR;
92 144
93ecb_inline void 145ecb_inline void
94cbor_init (CBOR *cbor) 146cbor_init (CBOR *cbor)
95{ 147{
96 Zero (cbor, 1, CBOR); 148 Zero (cbor, 1, CBOR);
97 cbor->max_depth = 512; 149 cbor->max_depth = 512;
150}
151
152ecb_inline void
153cbor_free (CBOR *cbor)
154{
155 SvREFCNT_dec (cbor->filter);
156 SvREFCNT_dec (cbor->incr_count);
98} 157}
99 158
100///////////////////////////////////////////////////////////////////////////// 159/////////////////////////////////////////////////////////////////////////////
101// utility functions 160// utility functions
102 161
122 SvPV_shrink_to_cur (sv); 181 SvPV_shrink_to_cur (sv);
123#elif defined (SvPV_renew) 182#elif defined (SvPV_renew)
124 SvPV_renew (sv, SvCUR (sv) + 1); 183 SvPV_renew (sv, SvCUR (sv) + 1);
125#endif 184#endif
126 } 185 }
186}
187
188// minimum length of a string to be registered for stringref
189ecb_inline int
190minimum_string_length (UV idx)
191{
192 return idx <= 23 ? 3
193 : idx <= 0xffU ? 4
194 : idx <= 0xffffU ? 5
195 : idx <= 0xffffffffU ? 7
196 : 11;
127} 197}
128 198
129///////////////////////////////////////////////////////////////////////////// 199/////////////////////////////////////////////////////////////////////////////
130// encoder 200// encoder
131 201
144} enc_t; 214} enc_t;
145 215
146ecb_inline void 216ecb_inline void
147need (enc_t *enc, STRLEN len) 217need (enc_t *enc, STRLEN len)
148{ 218{
149 if (ecb_expect_false (enc->cur + len >= enc->end)) 219 if (ecb_expect_false ((uintptr_t)(enc->end - enc->cur) < len))
150 { 220 {
151 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 221 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
152 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 222 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
153 enc->cur = SvPVX (enc->sv) + cur; 223 enc->cur = SvPVX (enc->sv) + cur;
154 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 224 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
165static void 235static void
166encode_uint (enc_t *enc, int major, UV len) 236encode_uint (enc_t *enc, int major, UV len)
167{ 237{
168 need (enc, 9); 238 need (enc, 9);
169 239
170 if (len < 24) 240 if (ecb_expect_true (len < LENGTH_EXT1))
171 *enc->cur++ = major | len; 241 *enc->cur++ = major | len;
172 else if (len <= 0xff) 242 else if (ecb_expect_true (len <= 0xffU))
173 { 243 {
174 *enc->cur++ = major | 24; 244 *enc->cur++ = major | LENGTH_EXT1;
175 *enc->cur++ = len; 245 *enc->cur++ = len;
176 } 246 }
177 else if (len <= 0xffff) 247 else if (len <= 0xffffU)
178 { 248 {
179 *enc->cur++ = major | 25; 249 *enc->cur++ = major | LENGTH_EXT2;
180 *enc->cur++ = len >> 8; 250 *enc->cur++ = len >> 8;
181 *enc->cur++ = len; 251 *enc->cur++ = len;
182 } 252 }
183 else if (len <= 0xffffffff) 253 else if (len <= 0xffffffffU)
184 { 254 {
185 *enc->cur++ = major | 26; 255 *enc->cur++ = major | LENGTH_EXT4;
186 *enc->cur++ = len >> 24; 256 *enc->cur++ = len >> 24;
187 *enc->cur++ = len >> 16; 257 *enc->cur++ = len >> 16;
188 *enc->cur++ = len >> 8; 258 *enc->cur++ = len >> 8;
189 *enc->cur++ = len; 259 *enc->cur++ = len;
190 } 260 }
191 else 261 else
192 { 262 {
193 *enc->cur++ = major | 27; 263 *enc->cur++ = major | LENGTH_EXT8;
194 *enc->cur++ = len >> 56; 264 *enc->cur++ = len >> 56;
195 *enc->cur++ = len >> 48; 265 *enc->cur++ = len >> 48;
196 *enc->cur++ = len >> 40; 266 *enc->cur++ = len >> 40;
197 *enc->cur++ = len >> 32; 267 *enc->cur++ = len >> 32;
198 *enc->cur++ = len >> 24; 268 *enc->cur++ = len >> 24;
200 *enc->cur++ = len >> 8; 270 *enc->cur++ = len >> 8;
201 *enc->cur++ = len; 271 *enc->cur++ = len;
202 } 272 }
203} 273}
204 274
205static void 275ecb_inline void
276encode_tag (enc_t *enc, UV tag)
277{
278 encode_uint (enc, MAJOR_TAG, tag);
279}
280
281// exceptional (hopefully) slow path for byte strings that need to be utf8-encoded
282ecb_noinline static void
206encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 283encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len)
207{ 284{
208 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 285 STRLEN ulen = len;
286 U8 *p, *pend = (U8 *)str + len;
287
288 for (p = (U8 *)str; p < pend; ++p)
289 ulen += *p >> 7; // count set high bits
290
291 encode_uint (enc, MAJOR_TEXT, ulen);
292
293 need (enc, ulen);
294 for (p = (U8 *)str; p < pend; ++p)
295 if (*p < 0x80)
296 *enc->cur++ = *p;
297 else
298 {
299 *enc->cur++ = 0xc0 + (*p >> 6);
300 *enc->cur++ = 0x80 + (*p & 63);
301 }
302}
303
304ecb_inline void
305encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
306{
307 if (ecb_expect_false (upgrade_utf8))
308 if (!utf8)
309 {
310 encode_str_utf8 (enc, utf8, str, len);
311 return;
312 }
313
314 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
209 need (enc, len); 315 need (enc, len);
210 memcpy (enc->cur, str, len); 316 memcpy (enc->cur, str, len);
211 enc->cur += len; 317 enc->cur += len;
212} 318}
213 319
214ecb_inline void 320ecb_inline void
215encode_tag (enc_t *enc, UV tag) 321encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
216{ 322{
217 encode_uint (enc, 0xc0, tag); 323 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
324 {
325 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
326
327 if (SvOK (*svp))
328 {
329 // already registered, use stringref
330 encode_tag (enc, CBOR_TAG_STRINGREF);
331 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
332 return;
333 }
334 else if (len >= minimum_string_length (enc->stringref_idx))
335 {
336 // register only
337 sv_setuv (*svp, enc->stringref_idx);
338 ++enc->stringref_idx;
339 }
340 }
341
342 encode_str (enc, upgrade_utf8, utf8, str, len);
218} 343}
219 344
220static void encode_sv (enc_t *enc, SV *sv); 345static void encode_sv (enc_t *enc, SV *sv);
221 346
222static void 347static void
227 if (enc->depth >= enc->cbor.max_depth) 352 if (enc->depth >= enc->cbor.max_depth)
228 croak (ERR_NESTING_EXCEEDED); 353 croak (ERR_NESTING_EXCEEDED);
229 354
230 ++enc->depth; 355 ++enc->depth;
231 356
232 encode_uint (enc, 0x80, len + 1); 357 encode_uint (enc, MAJOR_ARRAY, len + 1);
233 358
359 if (SvMAGICAL (av))
234 for (i = 0; i <= len; ++i) 360 for (i = 0; i <= len; ++i)
235 { 361 {
236 SV **svp = av_fetch (av, i, 0); 362 SV **svp = av_fetch (av, i, 0);
237 encode_sv (enc, svp ? *svp : &PL_sv_undef); 363 encode_sv (enc, svp ? *svp : &PL_sv_undef);
238 } 364 }
365 else
366 for (i = 0; i <= len; ++i)
367 {
368 SV *sv = AvARRAY (av)[i];
369 encode_sv (enc, sv ? sv : &PL_sv_undef);
370 }
239 371
240 --enc->depth; 372 --enc->depth;
241}
242
243ecb_inline void
244encode_he (enc_t *enc, HE *he)
245{
246 if (HeKLEN (he) == HEf_SVKEY)
247 encode_sv (enc, HeSVKEY (he));
248 else
249 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
250} 373}
251 374
252static void 375static void
253encode_hv (enc_t *enc, HV *hv) 376encode_hv (enc_t *enc, HV *hv)
254{ 377{
261 384
262 int pairs = hv_iterinit (hv); 385 int pairs = hv_iterinit (hv);
263 int mg = SvMAGICAL (hv); 386 int mg = SvMAGICAL (hv);
264 387
265 if (mg) 388 if (mg)
266 encode_ch (enc, 0xa0 | 31); 389 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
267 else 390 else
268 encode_uint (enc, 0xa0, pairs); 391 encode_uint (enc, MAJOR_MAP, pairs);
269 392
270 while ((he = hv_iternext (hv))) 393 while ((he = hv_iternext (hv)))
271 { 394 {
272 if (ecb_expect_false (enc->cbor.flags & (F_DEDUP_STRINGS | F_DEDUP_KEYS)))
273 {
274 SV **svp;
275
276 if (HeKLEN (he) == HEf_SVKEY) 395 if (HeKLEN (he) == HEf_SVKEY)
277 svp = hv_fetch_ent (enc->stringref[!! SvUTF8 (HeSVKEY (he))], HeSVKEY (he) , 1, 0);//TODO return HE :/ 396 encode_sv (enc, HeSVKEY (he));
278 else
279 svp = hv_fetch (enc->stringref[!! HeKUTF8 (he) ], HeKEY (he), HeKLEN (he), 1);
280
281 if (SvOK (*svp))
282 {
283 encode_tag (enc, CBOR_TAG_STRINGREF);
284 encode_uint (enc, 0x00, SvUV (*svp));
285 }
286 else
287 {
288 sv_setuv (*svp, enc->stringref_idx);
289 ++enc->stringref_idx;
290 encode_he (enc, he);
291 }
292 }
293 else 397 else
294 encode_he (enc, he); 398 encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS | F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he));
295 399
296 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 400 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
297 } 401 }
298 402
299 if (mg) 403 if (mg)
300 encode_ch (enc, 0xe0 | 31); 404 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
301 405
302 --enc->depth; 406 --enc->depth;
303} 407}
304 408
305// encode objects, arrays and special \0=false and \1=true values. 409// encode objects, arrays and special \0=false and \1=true values.
306static void 410static void
307encode_rv (enc_t *enc, SV *sv) 411encode_rv (enc_t *enc, SV *sv)
308{ 412{
309 SvGETMAGIC (sv); 413 SvGETMAGIC (sv);
310
311 if (ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)
312 && ecb_expect_false (SvREFCNT (sv) > 1))
313 {
314 if (!enc->shareable)
315 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
316
317 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
318
319 if (SvOK (*svp))
320 {
321 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
322 encode_uint (enc, 0x00, SvUV (*svp));
323 return;
324 }
325 else
326 {
327 sv_setuv (*svp, enc->shareable_idx);
328 ++enc->shareable_idx;
329 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
330 }
331 }
332 414
333 svtype svt = SvTYPE (sv); 415 svtype svt = SvTYPE (sv);
334 416
335 if (ecb_expect_false (SvOBJECT (sv))) 417 if (ecb_expect_false (SvOBJECT (sv)))
336 { 418 {
343 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 425 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
344 ? cbor_tagged_stash 426 ? cbor_tagged_stash
345 : gv_stashpv ("CBOR::XS::Tagged" , 1); 427 : gv_stashpv ("CBOR::XS::Tagged" , 1);
346 428
347 HV *stash = SvSTASH (sv); 429 HV *stash = SvSTASH (sv);
348 GV *method;
349 430
350 if (stash == boolean_stash) 431 if (stash == boolean_stash)
351 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 432 {
433 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
434 return;
435 }
352 else if (stash == error_stash) 436 else if (stash == error_stash)
353 encode_ch (enc, 0xe0 | 23); 437 {
438 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
439 return;
440 }
354 else if (stash == tagged_stash) 441 else if (stash == tagged_stash)
355 { 442 {
356 if (svt != SVt_PVAV) 443 if (svt != SVt_PVAV)
357 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 444 croak ("encountered CBOR::XS::Tagged object that isn't an array");
358 445
359 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 446 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
360 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 447 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
448
449 return;
450 }
451 }
452
453 if (ecb_expect_false (SvREFCNT (sv) > 1)
454 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
455 {
456 if (!enc->shareable)
457 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
458
459 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
460
461 if (SvOK (*svp))
361 } 462 {
463 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
464 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
465 return;
466 }
467 else
468 {
469 sv_setuv (*svp, enc->shareable_idx);
470 ++enc->shareable_idx;
471 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
472 }
473 }
474
475 if (ecb_expect_false (SvOBJECT (sv)))
476 {
477 HV *stash = SvSTASH (sv);
478 GV *method;
479
480 if (enc->cbor.flags & F_FORBID_OBJECTS)
481 croak ("encountered object '%s', but forbid_objects is enabled",
482 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
362 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 483 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
363 { 484 {
364 dSP; 485 dSP;
365 486
366 ENTER; SAVETMPS; PUSHMARK (SP); 487 ENTER; SAVETMPS;
488 PUSHMARK (SP);
367 // we re-bless the reference to get overload and other niceties right 489 // we re-bless the reference to get overload and other niceties right
368 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 490 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
369 491
370 PUTBACK; 492 PUTBACK;
371 // G_SCALAR ensures that return value is 1 493 // G_SCALAR ensures that return value is 1
384 } 506 }
385 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) 507 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
386 { 508 {
387 dSP; 509 dSP;
388 510
389 ENTER; SAVETMPS; PUSHMARK (SP); 511 ENTER; SAVETMPS;
512 SAVESTACK_POS ();
513 PUSHMARK (SP);
390 EXTEND (SP, 2); 514 EXTEND (SP, 2);
391 // we re-bless the reference to get overload and other niceties right 515 // we re-bless the reference to get overload and other niceties right
392 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 516 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
393 PUSHs (sv_cbor); 517 PUSHs (sv_cbor);
394 518
399 // catch this surprisingly common error 523 // catch this surprisingly common error
400 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 524 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
401 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); 525 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
402 526
403 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 527 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
404 encode_uint (enc, 0x80, count + 1); 528 encode_uint (enc, MAJOR_ARRAY, count + 1);
405 encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 529 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
406 530
407 while (count) 531 while (count)
408 encode_sv (enc, SP[1 - count--]); 532 encode_sv (enc, SP[1 - count--]);
409 533
410 PUTBACK; 534 PUTBACK;
431{ 555{
432 double nv = SvNVX (sv); 556 double nv = SvNVX (sv);
433 557
434 need (enc, 9); 558 need (enc, 9);
435 559
436 if (ecb_expect_false (nv == (U32)nv)) 560 if (ecb_expect_false (nv == (NV)(U32)nv))
437 encode_uint (enc, 0x00, (U32)nv); 561 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
438 //TODO: maybe I32? 562 //TODO: maybe I32?
439 else if (ecb_expect_false (nv == (float)nv)) 563 else if (ecb_expect_false (nv == (float)nv))
440 { 564 {
441 uint32_t fp = ecb_float_to_binary32 (nv); 565 uint32_t fp = ecb_float_to_binary32 (nv);
442 566
443 *enc->cur++ = 0xe0 | 26; 567 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
444 568
445 if (!ecb_big_endian ()) 569 if (!ecb_big_endian ())
446 fp = ecb_bswap32 (fp); 570 fp = ecb_bswap32 (fp);
447 571
448 memcpy (enc->cur, &fp, 4); 572 memcpy (enc->cur, &fp, 4);
450 } 574 }
451 else 575 else
452 { 576 {
453 uint64_t fp = ecb_double_to_binary64 (nv); 577 uint64_t fp = ecb_double_to_binary64 (nv);
454 578
455 *enc->cur++ = 0xe0 | 27; 579 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
456 580
457 if (!ecb_big_endian ()) 581 if (!ecb_big_endian ())
458 fp = ecb_bswap64 (fp); 582 fp = ecb_bswap64 (fp);
459 583
460 memcpy (enc->cur, &fp, 8); 584 memcpy (enc->cur, &fp, 8);
469 593
470 if (SvPOKp (sv)) 594 if (SvPOKp (sv))
471 { 595 {
472 STRLEN len; 596 STRLEN len;
473 char *str = SvPV (sv, len); 597 char *str = SvPV (sv, len);
474 encode_str (enc, SvUTF8 (sv), str, len); 598 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
475 } 599 }
476 else if (SvNOKp (sv)) 600 else if (SvNOKp (sv))
477 encode_nv (enc, sv); 601 encode_nv (enc, sv);
478 else if (SvIOKp (sv)) 602 else if (SvIOKp (sv))
479 { 603 {
480 if (SvIsUV (sv)) 604 if (SvIsUV (sv))
481 encode_uint (enc, 0x00, SvUVX (sv)); 605 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
482 else if (SvIVX (sv) >= 0) 606 else if (SvIVX (sv) >= 0)
483 encode_uint (enc, 0x00, SvIVX (sv)); 607 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
484 else 608 else
485 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 609 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
486 } 610 }
487 else if (SvROK (sv)) 611 else if (SvROK (sv))
488 encode_rv (enc, SvRV (sv)); 612 encode_rv (enc, SvRV (sv));
489 else if (!SvOK (sv)) 613 else if (!SvOK (sv))
490 encode_ch (enc, 0xe0 | 22); 614 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
491 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 615 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
492 encode_ch (enc, 0xe0 | 23); 616 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
493 else 617 else
494 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 618 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
495 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 619 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
496} 620}
497 621
498static SV * 622static SV *
499encode_cbor (SV *scalar, CBOR *cbor) 623encode_cbor (SV *scalar, CBOR *cbor)
500{ 624{
501 enc_t enc = { }; 625 enc_t enc = { 0 };
502 626
503 enc.cbor = *cbor; 627 enc.cbor = *cbor;
504 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 628 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
505 enc.cur = SvPVX (enc.sv); 629 enc.cur = SvPVX (enc.sv);
506 enc.end = SvEND (enc.sv); 630 enc.end = SvEND (enc.sv);
507 631
508 SvPOK_only (enc.sv); 632 SvPOK_only (enc.sv);
509 633
510 if (cbor->flags & (F_DEDUP_STRINGS | F_DEDUP_KEYS)) 634 if (cbor->flags & F_PACK_STRINGS)
511 { 635 {
512 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE); 636 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
513 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ()); 637 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
514 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ()); 638 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
515 } 639 }
537 CBOR cbor; 661 CBOR cbor;
538 U32 depth; // recursion depth 662 U32 depth; // recursion depth
539 U32 maxdepth; // recursion depth limit 663 U32 maxdepth; // recursion depth limit
540 AV *shareable; 664 AV *shareable;
541 AV *stringref; 665 AV *stringref;
666 SV *decode_tagged;
667 SV *err_sv; // optional sv for error, needs to be freed
542} dec_t; 668} dec_t;
543 669
544#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 670// set dec->err to ERRSV
671ecb_cold static void
672err_errsv (dec_t *dec)
673{
674 if (!dec->err)
675 {
676 dec->err_sv = newSVsv (ERRSV);
545 677
546#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 678 // chop off the trailing \n
679 SvCUR_set (dec->err_sv, SvCUR (dec->err_sv) - 1);
680 *SvEND (dec->err_sv) = 0;
547 681
682 dec->err = SvPVutf8_nolen (dec->err_sv);
683 }
684}
685
686// the following functions are used to reduce code size and help the compiler to optimise
687ecb_cold static void
688err_set (dec_t *dec, const char *reason)
689{
690 if (!dec->err)
691 dec->err = reason;
692}
693
694ecb_cold static void
695err_unexpected_end (dec_t *dec)
696{
697 err_set (dec, "unexpected end of CBOR data");
698}
699
700ecb_cold static void
701err_nesting_exceeded (dec_t *dec)
702{
703 err_set (dec, ERR_NESTING_EXCEEDED);
704}
705
706#define ERR_DO(do) SB do; goto fail; SE
707#define ERR(reason) ERR_DO (err_set (dec, reason))
708#define ERR_ERRSV ERR_DO (err_errsv (dec))
709
710#define WANT(len) if (ecb_expect_false ((uintptr_t)(dec->end - dec->cur) < (STRLEN)len)) ERR_DO (err_unexpected_end (dec))
711
548#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED) 712#define DEC_INC_DEPTH if (ecb_expect_false (++dec->depth > dec->cbor.max_depth)) ERR (ERR_NESTING_EXCEEDED)
549#define DEC_DEC_DEPTH --dec->depth 713#define DEC_DEC_DEPTH --dec->depth
550 714
551static UV 715static UV
552decode_uint (dec_t *dec) 716decode_uint (dec_t *dec)
553{ 717{
554 switch (*dec->cur & 31) 718 U8 m = *dec->cur & MINOR_MASK;
555 { 719 ++dec->cur;
556 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
557 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
558 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
559 return *dec->cur++ & 31;
560 720
561 case 24: 721 if (ecb_expect_true (m < LENGTH_EXT1))
722 return m;
723 else if (ecb_expect_true (m == LENGTH_EXT1))
724 {
562 WANT (2); 725 WANT (1);
563 dec->cur += 2; 726 dec->cur += 1;
564 return dec->cur[-1]; 727 return dec->cur[-1];
565 728 }
566 case 25: 729 else if (ecb_expect_true (m == LENGTH_EXT2))
730 {
567 WANT (3); 731 WANT (2);
568 dec->cur += 3; 732 dec->cur += 2;
569 return (((UV)dec->cur[-2]) << 8) 733 return (((UV)dec->cur[-2]) << 8)
570 | ((UV)dec->cur[-1]); 734 | ((UV)dec->cur[-1]);
571 735 }
572 case 26: 736 else if (ecb_expect_true (m == LENGTH_EXT4))
737 {
573 WANT (5); 738 WANT (4);
574 dec->cur += 5; 739 dec->cur += 4;
575 return (((UV)dec->cur[-4]) << 24) 740 return (((UV)dec->cur[-4]) << 24)
576 | (((UV)dec->cur[-3]) << 16) 741 | (((UV)dec->cur[-3]) << 16)
577 | (((UV)dec->cur[-2]) << 8) 742 | (((UV)dec->cur[-2]) << 8)
578 | ((UV)dec->cur[-1]); 743 | ((UV)dec->cur[-1]);
579 744 }
580 case 27: 745 else if (ecb_expect_true (m == LENGTH_EXT8))
746 {
581 WANT (9); 747 WANT (8);
582 dec->cur += 9; 748 dec->cur += 8;
749
750 return
751#if UVSIZE < 8
752 0
753#else
583 return (((UV)dec->cur[-8]) << 56) 754 (((UV)dec->cur[-8]) << 56)
584 | (((UV)dec->cur[-7]) << 48) 755 | (((UV)dec->cur[-7]) << 48)
585 | (((UV)dec->cur[-6]) << 40) 756 | (((UV)dec->cur[-6]) << 40)
586 | (((UV)dec->cur[-5]) << 32) 757 | (((UV)dec->cur[-5]) << 32)
758#endif
587 | (((UV)dec->cur[-4]) << 24) 759 | (((UV)dec->cur[-4]) << 24)
588 | (((UV)dec->cur[-3]) << 16) 760 | (((UV)dec->cur[-3]) << 16)
589 | (((UV)dec->cur[-2]) << 8) 761 | (((UV)dec->cur[-2]) << 8)
590 | ((UV)dec->cur[-1]); 762 | ((UV)dec->cur[-1]);
591 763 }
592 default: 764 else
593 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 765 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
594 }
595 766
596fail: 767fail:
597 return 0; 768 return 0;
598} 769}
599 770
604{ 775{
605 AV *av = newAV (); 776 AV *av = newAV ();
606 777
607 DEC_INC_DEPTH; 778 DEC_INC_DEPTH;
608 779
609 if ((*dec->cur & 31) == 31) 780 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
610 { 781 {
611 ++dec->cur; 782 ++dec->cur;
612 783
613 for (;;) 784 for (;;)
614 { 785 {
615 WANT (1); 786 WANT (1);
616 787
617 if (*dec->cur == (0xe0 | 31)) 788 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
618 { 789 {
619 ++dec->cur; 790 ++dec->cur;
620 break; 791 break;
621 } 792 }
622 793
623 av_push (av, decode_sv (dec)); 794 av_push (av, decode_sv (dec));
624 } 795 }
625 } 796 }
626 else 797 else
627 { 798 {
628 int i, len = decode_uint (dec); 799 UV i, len = decode_uint (dec);
629 800
801 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
630 av_fill (av, len - 1); 802 av_fill (av, len - 1);
631 803
632 for (i = 0; i < len; ++i) 804 for (i = 0; i < len; ++i)
633 AvARRAY (av)[i] = decode_sv (dec); 805 AvARRAY (av)[i] = decode_sv (dec);
634 } 806 }
644 816
645static void 817static void
646decode_he (dec_t *dec, HV *hv) 818decode_he (dec_t *dec, HV *hv)
647{ 819{
648 // for speed reasons, we specialcase single-string 820 // for speed reasons, we specialcase single-string
649 // byte or utf-8 strings as keys. 821 // byte or utf-8 strings as keys, but only when !stringref
650 822
651 if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27) 823 if (ecb_expect_true (!dec->stringref))
824 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
652 { 825 {
653 I32 len = decode_uint (dec); 826 STRLEN len = decode_uint (dec);
654 char *key = (char *)dec->cur; 827 char *key = (char *)dec->cur;
655 828
829 WANT (len);
656 dec->cur += len; 830 dec->cur += len;
657 831
658 if (ecb_expect_false (dec->stringref))
659 av_push (dec->stringref, newSVpvn (key, len));
660
661 hv_store (hv, key, len, decode_sv (dec), 0); 832 hv_store (hv, key, len, decode_sv (dec), 0);
833
834 return;
662 } 835 }
663 else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27) 836 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
664 { 837 {
665 I32 len = decode_uint (dec); 838 STRLEN len = decode_uint (dec);
666 char *key = (char *)dec->cur; 839 char *key = (char *)dec->cur;
667 840
841 WANT (len);
668 dec->cur += len; 842 dec->cur += len;
669 843
670 if (ecb_expect_false (dec->stringref)) 844 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
671 av_push (dec->stringref, newSVpvn_utf8 (key, len, 1)); 845 if (!is_utf8_string (key, len))
846 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
672 847
673 hv_store (hv, key, -len, decode_sv (dec), 0); 848 hv_store (hv, key, -len, decode_sv (dec), 0);
849
850 return;
674 } 851 }
675 else 852
676 {
677 SV *k = decode_sv (dec); 853 SV *k = decode_sv (dec);
678 SV *v = decode_sv (dec); 854 SV *v = decode_sv (dec);
679 855
856 // we leak memory if uncaught exceptions are thrown by random magical
857 // methods, and this is hopefully the only place where it can happen,
858 // so if there is a chance of an exception, take the very slow path.
859 // since catching exceptions is "undocumented/internal/forbidden" by
860 // the new p5p powers, we need to call out to a perl function :/
861 if (ecb_expect_false (SvAMAGIC (k)))
862 {
863 dSP;
864
865 ENTER; SAVETMPS;
866 PUSHMARK (SP);
867 EXTEND (SP, 3);
868 PUSHs (sv_2mortal (newRV_inc ((SV *)hv)));
869 PUSHs (sv_2mortal (k));
870 PUSHs (sv_2mortal (v));
871
872 PUTBACK;
873 call_pv ("CBOR::XS::_hv_store", G_VOID | G_DISCARD | G_EVAL);
874 SPAGAIN;
875
876 FREETMPS; LEAVE;
877
878 if (SvTRUE (ERRSV))
879 ERR_ERRSV;
880
881 return;
882 }
883
680 hv_store_ent (hv, k, v, 0); 884 hv_store_ent (hv, k, v, 0);
681 SvREFCNT_dec (k); 885 SvREFCNT_dec (k);
682 } 886
887fail:
888 ;
683} 889}
684 890
685static SV * 891static SV *
686decode_hv (dec_t *dec) 892decode_hv (dec_t *dec)
687{ 893{
688 HV *hv = newHV (); 894 HV *hv = newHV ();
689 895
690 DEC_INC_DEPTH; 896 DEC_INC_DEPTH;
691 897
692 if ((*dec->cur & 31) == 31) 898 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
693 { 899 {
694 ++dec->cur; 900 ++dec->cur;
695 901
696 for (;;) 902 for (;;)
697 { 903 {
698 WANT (1); 904 WANT (1);
699 905
700 if (*dec->cur == (0xe0 | 31)) 906 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
701 { 907 {
702 ++dec->cur; 908 ++dec->cur;
703 break; 909 break;
704 } 910 }
705 911
706 decode_he (dec, hv); 912 decode_he (dec, hv);
707 } 913 }
708 } 914 }
709 else 915 else
710 { 916 {
711 int pairs = decode_uint (dec); 917 UV pairs = decode_uint (dec);
918
919 WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes
712 920
713 while (pairs--) 921 while (pairs--)
714 decode_he (dec, hv); 922 decode_he (dec, hv);
715 } 923 }
716 924
726static SV * 934static SV *
727decode_str (dec_t *dec, int utf8) 935decode_str (dec_t *dec, int utf8)
728{ 936{
729 SV *sv = 0; 937 SV *sv = 0;
730 938
731 if ((*dec->cur & 31) == 31) 939 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
732 { 940 {
941 // indefinite length strings
733 ++dec->cur; 942 ++dec->cur;
734 943
944 U8 major = *dec->cur & MAJOR_MISC;
945
735 sv = newSVpvn ("", 0); 946 sv = newSVpvn ("", 0);
736 947
737 // not very fast, and certainly not robust against illegal input
738 for (;;) 948 for (;;)
739 { 949 {
740 WANT (1); 950 WANT (1);
741 951
742 if (*dec->cur == (0xe0 | 31)) 952 if ((*dec->cur - major) > LENGTH_EXT8)
953 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
743 { 954 {
744 ++dec->cur; 955 ++dec->cur;
745 break; 956 break;
746 } 957 }
958 else
959 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
747 960
748 sv_catsv (sv, decode_sv (dec)); 961 STRLEN len = decode_uint (dec);
962
963 WANT (len);
964 sv_catpvn (sv, dec->cur, len);
965 dec->cur += len;
749 } 966 }
750 } 967 }
751 else 968 else
752 { 969 {
753 STRLEN len = decode_uint (dec); 970 STRLEN len = decode_uint (dec);
754 971
755 WANT (len); 972 WANT (len);
756 sv = newSVpvn (dec->cur, len); 973 sv = newSVpvn (dec->cur, len);
757 dec->cur += len; 974 dec->cur += len;
975
976 if (ecb_expect_false (dec->stringref)
977 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
978 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
758 } 979 }
759 980
760 if (utf8) 981 if (utf8)
982 {
983 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
984 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
985 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
986
761 SvUTF8_on (sv); 987 SvUTF8_on (sv);
762 988 }
763 if (ecb_expect_false (dec->stringref))
764 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
765 989
766 return sv; 990 return sv;
767 991
768fail: 992fail:
769 SvREFCNT_dec (sv); 993 SvREFCNT_dec (sv);
788 sv = newRV_noinc (decode_sv (dec)); 1012 sv = newRV_noinc (decode_sv (dec));
789 break; 1013 break;
790 1014
791 case CBOR_TAG_STRINGREF_NAMESPACE: 1015 case CBOR_TAG_STRINGREF_NAMESPACE:
792 { 1016 {
1017 // do nmot use SAVETMPS/FREETMPS, as these will
1018 // erase mortalised caches, e.g. "shareable"
793 ENTER; SAVETMPS; 1019 ENTER;
794 1020
795 SAVESPTR (dec->stringref); 1021 SAVESPTR (dec->stringref);
796 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); 1022 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
797 1023
798 sv = decode_sv (dec); 1024 sv = decode_sv (dec);
799 1025
800 FREETMPS; LEAVE; 1026 LEAVE;
801 } 1027 }
802 break; 1028 break;
803 1029
804 case CBOR_TAG_STRINGREF: 1030 case CBOR_TAG_STRINGREF:
805 { 1031 {
806 if ((*dec->cur >> 5) != 0) 1032 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
807 ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); 1033 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
808 1034
809 UV idx = decode_uint (dec); 1035 UV idx = decode_uint (dec);
810 1036
811 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) 1037 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
818 case CBOR_TAG_VALUE_SHAREABLE: 1044 case CBOR_TAG_VALUE_SHAREABLE:
819 { 1045 {
820 if (ecb_expect_false (!dec->shareable)) 1046 if (ecb_expect_false (!dec->shareable))
821 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ()); 1047 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
822 1048
1049 if (dec->cbor.flags & F_ALLOW_CYCLES)
1050 {
823 sv = newSV (0); 1051 sv = newSV (0);
824 av_push (dec->shareable, SvREFCNT_inc_NN (sv)); 1052 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
825 1053
826 SV *osv = decode_sv (dec); 1054 SV *osv = decode_sv (dec);
827 sv_setsv (sv, osv); 1055 sv_setsv (sv, osv);
828 SvREFCNT_dec_NN (osv); 1056 SvREFCNT_dec_NN (osv);
1057 }
1058 else
1059 {
1060 av_push (dec->shareable, &PL_sv_undef);
1061 int idx = AvFILLp (dec->shareable);
1062 sv = decode_sv (dec);
1063 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
1064 }
829 } 1065 }
830 break; 1066 break;
831 1067
832 case CBOR_TAG_VALUE_SHAREDREF: 1068 case CBOR_TAG_VALUE_SHAREDREF:
833 { 1069 {
834 if ((*dec->cur >> 5) != 0) 1070 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
835 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); 1071 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
836 1072
837 UV idx = decode_uint (dec); 1073 UV idx = decode_uint (dec);
838 1074
839 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) 1075 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
840 ERR ("corrupted CBOR data (sharedref index out of bounds)"); 1076 ERR ("corrupted CBOR data (sharedref index out of bounds)");
841 1077
842 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); 1078 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
1079
1080 if (sv == &PL_sv_undef)
1081 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
843 } 1082 }
844 break; 1083 break;
845 1084
846 case CBOR_TAG_PERL_OBJECT: 1085 case CBOR_TAG_PERL_OBJECT:
847 { 1086 {
1087 if (dec->cbor.flags & F_FORBID_OBJECTS)
1088 goto filter;
1089
848 sv = decode_sv (dec); 1090 sv = decode_sv (dec);
849 1091
850 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 1092 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
851 ERR ("corrupted CBOR data (non-array perl object)"); 1093 ERR ("corrupted CBOR data (non-array perl object)");
852 1094
862 if (!method) 1104 if (!method)
863 ERR ("cannot decode perl-object (package does not have a THAW method)"); 1105 ERR ("cannot decode perl-object (package does not have a THAW method)");
864 1106
865 dSP; 1107 dSP;
866 1108
867 ENTER; SAVETMPS; PUSHMARK (SP); 1109 ENTER; SAVETMPS;
1110 PUSHMARK (SP);
868 EXTEND (SP, len + 1); 1111 EXTEND (SP, len + 1);
869 // we re-bless the reference to get overload and other niceties right 1112 // we re-bless the reference to get overload and other niceties right
870 PUSHs (*av_fetch (av, 0, 1)); 1113 PUSHs (*av_fetch (av, 0, 1));
871 PUSHs (sv_cbor); 1114 PUSHs (sv_cbor);
872 1115
880 SPAGAIN; 1123 SPAGAIN;
881 1124
882 if (SvTRUE (ERRSV)) 1125 if (SvTRUE (ERRSV))
883 { 1126 {
884 FREETMPS; LEAVE; 1127 FREETMPS; LEAVE;
885 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1128 ERR_ERRSV;
886 } 1129 }
887 1130
888 SvREFCNT_dec (sv); 1131 SvREFCNT_dec (sv);
889 sv = SvREFCNT_inc (POPs); 1132 sv = SvREFCNT_inc (POPs);
890 1133
893 FREETMPS; LEAVE; 1136 FREETMPS; LEAVE;
894 } 1137 }
895 break; 1138 break;
896 1139
897 default: 1140 default:
1141 filter:
898 { 1142 {
1143 SV *tag_sv = newSVuv (tag);
1144
899 sv = decode_sv (dec); 1145 sv = decode_sv (dec);
900 1146
1147 dSP;
1148 ENTER; SAVETMPS;
1149 SAVESTACK_POS ();
1150 PUSHMARK (SP);
1151 EXTEND (SP, 2);
1152 PUSHs (tag_sv);
1153 PUSHs (sv);
1154
1155 PUTBACK;
1156 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1157 SPAGAIN;
1158
1159 if (SvTRUE (ERRSV))
1160 {
1161 SvREFCNT_dec (tag_sv);
1162 FREETMPS; LEAVE;
1163 ERR_ERRSV;
1164 }
1165
1166 if (count)
1167 {
1168 SvREFCNT_dec (tag_sv);
1169 SvREFCNT_dec (sv);
1170 sv = SvREFCNT_inc (POPs);
1171 }
1172 else
1173 {
901 AV *av = newAV (); 1174 AV *av = newAV ();
902 av_push (av, newSVuv (tag)); 1175 av_push (av, tag_sv);
903 av_push (av, sv); 1176 av_push (av, sv);
904 1177
905 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1178 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
906 ? cbor_tagged_stash 1179 ? cbor_tagged_stash
907 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1180 : gv_stashpv ("CBOR::XS::Tagged" , 1);
908
909 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash); 1181 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
1182 }
1183
1184 PUTBACK;
1185
1186 FREETMPS; LEAVE;
910 } 1187 }
911 break; 1188 break;
912 } 1189 }
913 1190
914 return sv; 1191 return sv;
921static SV * 1198static SV *
922decode_sv (dec_t *dec) 1199decode_sv (dec_t *dec)
923{ 1200{
924 WANT (1); 1201 WANT (1);
925 1202
926 switch (*dec->cur >> 5) 1203 switch (*dec->cur >> MAJOR_SHIFT)
927 { 1204 {
928 case 0: // unsigned int 1205 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
929 return newSVuv (decode_uint (dec)); 1206 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
930 case 1: // negative int 1207 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
931 return newSViv (-1 - (IV)decode_uint (dec)); 1208 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
932 case 2: // octet string 1209 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
933 return decode_str (dec, 0); 1210 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
934 case 3: // utf-8 string 1211 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
935 return decode_str (dec, 1); 1212
936 case 4: // array 1213 case MAJOR_MISC >> MAJOR_SHIFT:
937 return decode_av (dec);
938 case 5: // map
939 return decode_hv (dec);
940 case 6: // tag
941 return decode_tagged (dec);
942 case 7: // misc
943 switch (*dec->cur++ & 31) 1214 switch (*dec->cur++ & MINOR_MASK)
944 { 1215 {
945 case 20: 1216 case SIMPLE_FALSE:
946#if CBOR_SLOW 1217#if CBOR_SLOW
947 types_false = get_bool ("Types::Serialiser::false"); 1218 types_false = get_bool ("Types::Serialiser::false");
948#endif 1219#endif
949 return newSVsv (types_false); 1220 return newSVsv (types_false);
950 case 21: 1221 case SIMPLE_TRUE:
951#if CBOR_SLOW 1222#if CBOR_SLOW
952 types_true = get_bool ("Types::Serialiser::true"); 1223 types_true = get_bool ("Types::Serialiser::true");
953#endif 1224#endif
954 return newSVsv (types_true); 1225 return newSVsv (types_true);
955 case 22: 1226 case SIMPLE_NULL:
956 return newSVsv (&PL_sv_undef); 1227 return newSVsv (&PL_sv_undef);
957 case 23: 1228 case SIMPLE_UNDEF:
958#if CBOR_SLOW 1229#if CBOR_SLOW
959 types_error = get_bool ("Types::Serialiser::error"); 1230 types_error = get_bool ("Types::Serialiser::error");
960#endif 1231#endif
961 return newSVsv (types_error); 1232 return newSVsv (types_error);
962 1233
963 case 25: 1234 case MISC_FLOAT16:
964 { 1235 {
965 WANT (2); 1236 WANT (2);
966 1237
967 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1238 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
968 dec->cur += 2; 1239 dec->cur += 2;
969 1240
970 return newSVnv (ecb_binary16_to_float (fp)); 1241 return newSVnv (ecb_binary16_to_float (fp));
971 } 1242 }
972 1243
973 case 26: 1244 case MISC_FLOAT32:
974 { 1245 {
975 uint32_t fp; 1246 uint32_t fp;
976 WANT (4); 1247 WANT (4);
977 memcpy (&fp, dec->cur, 4); 1248 memcpy (&fp, dec->cur, 4);
978 dec->cur += 4; 1249 dec->cur += 4;
981 fp = ecb_bswap32 (fp); 1252 fp = ecb_bswap32 (fp);
982 1253
983 return newSVnv (ecb_binary32_to_float (fp)); 1254 return newSVnv (ecb_binary32_to_float (fp));
984 } 1255 }
985 1256
986 case 27: 1257 case MISC_FLOAT64:
987 { 1258 {
988 uint64_t fp; 1259 uint64_t fp;
989 WANT (8); 1260 WANT (8);
990 memcpy (&fp, dec->cur, 8); 1261 memcpy (&fp, dec->cur, 8);
991 dec->cur += 8; 1262 dec->cur += 8;
994 fp = ecb_bswap64 (fp); 1265 fp = ecb_bswap64 (fp);
995 1266
996 return newSVnv (ecb_binary64_to_double (fp)); 1267 return newSVnv (ecb_binary64_to_double (fp));
997 } 1268 }
998 1269
999 // 0..19 unassigned 1270 // 0..19 unassigned simple
1000 // 24 reserved + unassigned (reserved values are not encodable) 1271 // 24 reserved + unassigned simple (reserved values are not encodable)
1272 // 28-30 unassigned misc
1273 // 31 break code
1001 default: 1274 default:
1002 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1275 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
1003 } 1276 }
1004 1277
1005 break; 1278 break;
1006 } 1279 }
1007 1280
1010} 1283}
1011 1284
1012static SV * 1285static SV *
1013decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1286decode_cbor (SV *string, CBOR *cbor, char **offset_return)
1014{ 1287{
1015 dec_t dec = { }; 1288 dec_t dec = { 0 };
1016 SV *sv; 1289 SV *sv;
1017 STRLEN len; 1290 STRLEN len;
1018 char *data = SvPVbyte (string, len); 1291 char *data = SvPVbyte (string, len);
1019 1292
1020 if (len > cbor->max_size && cbor->max_size) 1293 if (len > cbor->max_size && cbor->max_size)
1034 if (dec.cur != dec.end && !dec.err) 1307 if (dec.cur != dec.end && !dec.err)
1035 dec.err = "garbage after CBOR object"; 1308 dec.err = "garbage after CBOR object";
1036 1309
1037 if (dec.err) 1310 if (dec.err)
1038 { 1311 {
1312 if (dec.shareable)
1313 {
1314 // need to break cyclic links, which would all be in shareable
1315 int i;
1316 SV **svp;
1317
1318 for (i = av_len (dec.shareable) + 1; i--; )
1319 if ((svp = av_fetch (dec.shareable, i, 0)))
1320 sv_setsv (*svp, &PL_sv_undef);
1321 }
1322
1039 SvREFCNT_dec (sv); 1323 SvREFCNT_dec (sv);
1324
1325 if (dec.err_sv)
1326 sv_2mortal (dec.err_sv);
1327
1040 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); 1328 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
1041 } 1329 }
1042 1330
1043 sv = sv_2mortal (sv); 1331 sv = sv_2mortal (sv);
1044 1332
1045 return sv; 1333 return sv;
1046} 1334}
1047 1335
1336/////////////////////////////////////////////////////////////////////////////
1337// incremental parser
1338
1339#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1340
1341// returns 0 for notyet, 1 for success or error
1342static int
1343incr_parse (CBOR *self, SV *cborstr)
1344{
1345 STRLEN cur;
1346 SvPV (cborstr, cur);
1347
1348 while (ecb_expect_true (self->incr_need <= cur))
1349 {
1350 // table of integer count bytes
1351 static I8 incr_len[MINOR_MASK + 1] = {
1352 0, 0, 0, 0, 0, 0, 0, 0,
1353 0, 0, 0, 0, 0, 0, 0, 0,
1354 0, 0, 0, 0, 0, 0, 0, 0,
1355 1, 2, 4, 8,-1,-1,-1,-2
1356 };
1357
1358 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1359 U8 m = *p & MINOR_MASK;
1360 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1361 I8 ilen = incr_len[m];
1362
1363 self->incr_need = self->incr_pos + 1;
1364
1365 if (ecb_expect_false (ilen < 0))
1366 {
1367 if (m != MINOR_INDEF)
1368 return 1; // error
1369
1370 if (*p == (MAJOR_MISC | MINOR_INDEF))
1371 {
1372 if (count >= 0)
1373 return 1; // error
1374
1375 count = 1;
1376 }
1377 else
1378 {
1379 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1380 count = -1;
1381 }
1382 }
1383 else
1384 {
1385 self->incr_need += ilen;
1386 if (ecb_expect_false (self->incr_need > cur))
1387 return 0;
1388
1389 int major = *p >> MAJOR_SHIFT;
1390
1391 switch (major)
1392 {
1393 case MAJOR_TAG >> MAJOR_SHIFT:
1394 ++count; // tags merely prefix another value
1395 break;
1396
1397 case MAJOR_BYTES >> MAJOR_SHIFT:
1398 case MAJOR_TEXT >> MAJOR_SHIFT:
1399 case MAJOR_ARRAY >> MAJOR_SHIFT:
1400 case MAJOR_MAP >> MAJOR_SHIFT:
1401 {
1402 UV len;
1403
1404 if (ecb_expect_false (ilen))
1405 {
1406 len = 0;
1407
1408 do {
1409 len = (len << 8) | *++p;
1410 } while (--ilen);
1411 }
1412 else
1413 len = m;
1414
1415 switch (major)
1416 {
1417 case MAJOR_BYTES >> MAJOR_SHIFT:
1418 case MAJOR_TEXT >> MAJOR_SHIFT:
1419 self->incr_need += len;
1420 if (ecb_expect_false (self->incr_need > cur))
1421 return 0;
1422
1423 break;
1424
1425 case MAJOR_MAP >> MAJOR_SHIFT:
1426 len <<= 1;
1427 case MAJOR_ARRAY >> MAJOR_SHIFT:
1428 if (len)
1429 {
1430 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1431 count = len + 1;
1432 }
1433 break;
1434 }
1435 }
1436 }
1437 }
1438
1439 self->incr_pos = self->incr_need;
1440
1441 if (count > 0)
1442 {
1443 while (!--count)
1444 {
1445 if (!AvFILLp (self->incr_count))
1446 return 1; // done
1447
1448 SvREFCNT_dec_NN (av_pop (self->incr_count));
1449 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1450 }
1451
1452 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1453 }
1454 }
1455
1456 return 0;
1457}
1458
1459
1048///////////////////////////////////////////////////////////////////////////// 1460/////////////////////////////////////////////////////////////////////////////
1049// XS interface functions 1461// XS interface functions
1050 1462
1051MODULE = CBOR::XS PACKAGE = CBOR::XS 1463MODULE = CBOR::XS PACKAGE = CBOR::XS
1052 1464
1060 1472
1061 types_true = get_bool ("Types::Serialiser::true" ); 1473 types_true = get_bool ("Types::Serialiser::true" );
1062 types_false = get_bool ("Types::Serialiser::false"); 1474 types_false = get_bool ("Types::Serialiser::false");
1063 types_error = get_bool ("Types::Serialiser::error"); 1475 types_error = get_bool ("Types::Serialiser::error");
1064 1476
1477 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1478
1065 sv_cbor = newSVpv ("CBOR", 0); 1479 sv_cbor = newSVpv ("CBOR", 0);
1066 SvREADONLY_on (sv_cbor); 1480 SvREADONLY_on (sv_cbor);
1481
1482 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1067} 1483}
1068 1484
1069PROTOTYPES: DISABLE 1485PROTOTYPES: DISABLE
1070 1486
1071void CLONE (...) 1487void CLONE (...)
1090void shrink (CBOR *self, int enable = 1) 1506void shrink (CBOR *self, int enable = 1)
1091 ALIAS: 1507 ALIAS:
1092 shrink = F_SHRINK 1508 shrink = F_SHRINK
1093 allow_unknown = F_ALLOW_UNKNOWN 1509 allow_unknown = F_ALLOW_UNKNOWN
1094 allow_sharing = F_ALLOW_SHARING 1510 allow_sharing = F_ALLOW_SHARING
1095 dedup_keys = F_DEDUP_KEYS 1511 allow_cycles = F_ALLOW_CYCLES
1512 forbid_objects = F_FORBID_OBJECTS
1096 dedup_strings = F_DEDUP_STRINGS 1513 pack_strings = F_PACK_STRINGS
1514 text_keys = F_TEXT_KEYS
1515 text_strings = F_TEXT_STRINGS
1516 validate_utf8 = F_VALIDATE_UTF8
1097 PPCODE: 1517 PPCODE:
1098{ 1518{
1099 if (enable) 1519 if (enable)
1100 self->flags |= ix; 1520 self->flags |= ix;
1101 else 1521 else
1107void get_shrink (CBOR *self) 1527void get_shrink (CBOR *self)
1108 ALIAS: 1528 ALIAS:
1109 get_shrink = F_SHRINK 1529 get_shrink = F_SHRINK
1110 get_allow_unknown = F_ALLOW_UNKNOWN 1530 get_allow_unknown = F_ALLOW_UNKNOWN
1111 get_allow_sharing = F_ALLOW_SHARING 1531 get_allow_sharing = F_ALLOW_SHARING
1112 get_dedup_keys = F_DEDUP_KEYS 1532 get_allow_cycles = F_ALLOW_CYCLES
1533 get_forbid_objects = F_FORBID_OBJECTS
1113 get_dedup_strings = F_DEDUP_STRINGS 1534 get_pack_strings = F_PACK_STRINGS
1535 get_text_keys = F_TEXT_KEYS
1536 get_text_strings = F_TEXT_STRINGS
1537 get_validate_utf8 = F_VALIDATE_UTF8
1114 PPCODE: 1538 PPCODE:
1115 XPUSHs (boolSV (self->flags & ix)); 1539 XPUSHs (boolSV (self->flags & ix));
1116 1540
1117void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1541void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1118 PPCODE: 1542 PPCODE:
1131 XPUSHs (ST (0)); 1555 XPUSHs (ST (0));
1132 1556
1133int get_max_size (CBOR *self) 1557int get_max_size (CBOR *self)
1134 CODE: 1558 CODE:
1135 RETVAL = self->max_size; 1559 RETVAL = self->max_size;
1560 OUTPUT:
1561 RETVAL
1562
1563void filter (CBOR *self, SV *filter = 0)
1564 PPCODE:
1565 SvREFCNT_dec (self->filter);
1566 self->filter = filter ? newSVsv (filter) : filter;
1567 XPUSHs (ST (0));
1568
1569SV *get_filter (CBOR *self)
1570 CODE:
1571 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
1136 OUTPUT: 1572 OUTPUT:
1137 RETVAL 1573 RETVAL
1138 1574
1139void encode (CBOR *self, SV *scalar) 1575void encode (CBOR *self, SV *scalar)
1140 PPCODE: 1576 PPCODE:
1155 EXTEND (SP, 2); 1591 EXTEND (SP, 2);
1156 PUSHs (sv); 1592 PUSHs (sv);
1157 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1593 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1158} 1594}
1159 1595
1596void incr_parse (CBOR *self, SV *cborstr)
1597 ALIAS:
1598 incr_parse_multiple = 1
1599 PPCODE:
1600{
1601 if (SvUTF8 (cborstr))
1602 sv_utf8_downgrade (cborstr, 0);
1603
1604 if (!self->incr_count)
1605 {
1606 self->incr_count = newAV ();
1607 self->incr_pos = 0;
1608 self->incr_need = 1;
1609
1610 av_push (self->incr_count, newSViv (1));
1611 }
1612
1613 do
1614 {
1615 if (!incr_parse (self, cborstr))
1616 {
1617 if (self->incr_need > self->max_size && self->max_size)
1618 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1619 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1620
1621 break;
1622 }
1623
1624 SV *sv;
1625 char *offset;
1626
1627 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1628 XPUSHs (sv);
1629
1630 sv_chop (cborstr, offset);
1631
1632 av_clear (self->incr_count);
1633 av_push (self->incr_count, newSViv (1));
1634
1635 self->incr_pos = 0;
1636 self->incr_need = self->incr_pos + 1;
1637 }
1638 while (ix);
1639}
1640
1641void incr_reset (CBOR *self)
1642 CODE:
1643{
1644 SvREFCNT_dec (self->incr_count);
1645 self->incr_count = 0;
1646}
1647
1648void DESTROY (CBOR *self)
1649 PPCODE:
1650 cbor_free (self);
1651
1160PROTOTYPES: ENABLE 1652PROTOTYPES: ENABLE
1161 1653
1162void encode_cbor (SV *scalar) 1654void encode_cbor (SV *scalar)
1655 ALIAS:
1656 encode_cbor = 0
1657 encode_cbor_sharing = F_ALLOW_SHARING
1163 PPCODE: 1658 PPCODE:
1164{ 1659{
1165 CBOR cbor; 1660 CBOR cbor;
1166 cbor_init (&cbor); 1661 cbor_init (&cbor);
1662 cbor.flags |= ix;
1167 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1663 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1168 XPUSHs (scalar); 1664 XPUSHs (scalar);
1169} 1665}
1170 1666
1171void decode_cbor (SV *cborstr) 1667void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines