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 1.19 by root, Wed Nov 20 02:03:09 2013 UTC vs.
Revision 1.57 by root, Fri Nov 25 12:16:12 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"
13 14
14// compatibility with perl <5.18 15// compatibility with perl <5.18
15#ifndef HvNAMELEN_get 16#ifndef HvNAMELEN_get
16# define HvNAMELEN_get(hv) strlen (HvNAME (hv)) 17# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
19# define HvNAMELEN(hv) HvNAMELEN_get (hv) 20# define HvNAMELEN(hv) HvNAMELEN_get (hv)
20#endif 21#endif
21#ifndef HvNAMEUTF8 22#ifndef HvNAMEUTF8
22# define HvNAMEUTF8(hv) 0 23# define HvNAMEUTF8(hv) 0
23#endif 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};
24 65
25// known tags 66// known tags
26enum cbor_tag 67enum cbor_tag
27{ 68{
28 // inofficial extensions (pending iana registration) 69 // extensions
70 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
29 CBOR_TAG_PERL_OBJECT = 24, // http://cbor.schmorp.de/perl-object 71 CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object
30 CBOR_TAG_GENERIC_OBJECT = 25, // http://cbor.schmorp.de/generic-object 72 CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object
31 CBOR_TAG_VALUE_SHAREABLE = 26, // http://cbor.schmorp.de/value-sharing 73 CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing
32 CBOR_TAG_VALUE_SHAREDREF = 27, // http://cbor.schmorp.de/value-sharing 74 CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing
33 CBOR_TAG_STRINGREF_NAMESPACE = 65537, // http://cbor.schmorp.de/stringref 75 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 76 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection
36 77
37 // rfc7049 78 // rfc7049
38 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 79 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
39 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any 80 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
40 CBOR_TAG_POS_BIGNUM = 2, // byte string 81 CBOR_TAG_POS_BIGNUM = 2, // byte string
41 CBOR_TAG_NEG_BIGNUM = 3, // byte string 82 CBOR_TAG_NEG_BIGNUM = 3, // byte string
42 CBOR_TAG_DECIMAL = 4, // decimal fraction, array 83 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
43 CBOR_TAG_BIGFLOAT = 5, // array 84 CBOR_TAG_BIGFLOAT = 5, // array
44 85
45 CBOR_TAG_CONV_B64U = 21, // base64url, any 86 CBOR_TAG_CONV_B64U = 21, // base64url, any
46 CBOR_TAG_CONV_B64 = 22, // base64, any 87 CBOR_TAG_CONV_B64 = 22, // base64, any
47 CBOR_TAG_CONV_HEX = 23, // base16, any 88 CBOR_TAG_CONV_HEX = 23, // base16, any
48 CBOR_TAG_CBOR = 24, // embedded cbor, byte string 89 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
49 90
50 CBOR_TAG_URI = 32, // URI rfc3986, utf-8 91 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
51 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8 92 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
52 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8 93 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
53 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8 94 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
54 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 95 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
55 96
56 CBOR_TAG_MAGIC = 55799 // self-describe cbor 97 CBOR_TAG_MAGIC = 55799, // self-describe cbor
57}; 98};
58 99
59#define F_SHRINK 0x00000001UL 100#define F_SHRINK 0x00000001UL
60#define F_ALLOW_UNKNOWN 0x00000002UL 101#define F_ALLOW_UNKNOWN 0x00000002UL
61#define F_ALLOW_SHARING 0x00000004UL //TODO 102#define F_ALLOW_SHARING 0x00000004UL
103#define F_ALLOW_CYCLES 0x00000008UL
62#define F_DEDUP_STRINGS 0x00000008UL //TODO 104#define F_PACK_STRINGS 0x00000010UL
63#define F_DEDUP_KEYS 0x00000010UL //TODO 105#define F_TEXT_KEYS 0x00000020UL
106#define F_TEXT_STRINGS 0x00000040UL
107#define F_VALIDATE_UTF8 0x00000080UL
64 108
65#define INIT_SIZE 32 // initial scalar size to be allocated 109#define INIT_SIZE 32 // initial scalar size to be allocated
66 110
67#define SB do { 111#define SB do {
68#define SE } while (0) 112#define SE } while (0)
80# define CBOR_SLOW 0 124# define CBOR_SLOW 0
81# define CBOR_STASH cbor_stash 125# define CBOR_STASH cbor_stash
82#endif 126#endif
83 127
84static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS:: 128static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
85static SV *types_true, *types_false, *types_error, *sv_cbor; 129static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
86 130
87typedef struct { 131typedef struct {
88 U32 flags; 132 U32 flags;
89 U32 max_depth; 133 U32 max_depth;
90 STRLEN max_size; 134 STRLEN max_size;
135 SV *filter;
136
137 // for the incremental parser
138 STRLEN incr_pos; // the current offset into the text
139 STRLEN incr_need; // minimum bytes needed to decode
140 AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
91} CBOR; 141} CBOR;
92 142
93ecb_inline void 143ecb_inline void
94cbor_init (CBOR *cbor) 144cbor_init (CBOR *cbor)
95{ 145{
96 Zero (cbor, 1, CBOR); 146 Zero (cbor, 1, CBOR);
97 cbor->max_depth = 512; 147 cbor->max_depth = 512;
148}
149
150ecb_inline void
151cbor_free (CBOR *cbor)
152{
153 SvREFCNT_dec (cbor->filter);
154 SvREFCNT_dec (cbor->incr_count);
98} 155}
99 156
100///////////////////////////////////////////////////////////////////////////// 157/////////////////////////////////////////////////////////////////////////////
101// utility functions 158// utility functions
102 159
124 SvPV_renew (sv, SvCUR (sv) + 1); 181 SvPV_renew (sv, SvCUR (sv) + 1);
125#endif 182#endif
126 } 183 }
127} 184}
128 185
186// minimum length of a string to be registered for stringref
187ecb_inline int
188minimum_string_length (UV idx)
189{
190 return idx > 23
191 ? idx > 0xffU
192 ? idx > 0xffffU
193 ? idx > 0xffffffffU
194 ? 11
195 : 7
196 : 5
197 : 4
198 : 3;
199}
200
129///////////////////////////////////////////////////////////////////////////// 201/////////////////////////////////////////////////////////////////////////////
130// encoder 202// encoder
131 203
132// structure used for encoding CBOR 204// structure used for encoding CBOR
133typedef struct 205typedef struct
135 char *cur; // SvPVX (sv) + current output position 207 char *cur; // SvPVX (sv) + current output position
136 char *end; // SvEND (sv) 208 char *end; // SvEND (sv)
137 SV *sv; // result scalar 209 SV *sv; // result scalar
138 CBOR cbor; 210 CBOR cbor;
139 U32 depth; // recursion level 211 U32 depth; // recursion level
140 HV *stringref; // string => index, or 0 212 HV *stringref[2]; // string => index, or 0 ([0] = bytes, [1] = utf-8)
213 UV stringref_idx;
141 HV *shareable; // ptr => index, or 0 214 HV *shareable; // ptr => index, or 0
142 UV shareable_idx; 215 UV shareable_idx;
143} enc_t; 216} enc_t;
144 217
145ecb_inline void 218ecb_inline void
164static void 237static void
165encode_uint (enc_t *enc, int major, UV len) 238encode_uint (enc_t *enc, int major, UV len)
166{ 239{
167 need (enc, 9); 240 need (enc, 9);
168 241
169 if (len < 24) 242 if (ecb_expect_true (len < LENGTH_EXT1))
170 *enc->cur++ = major | len; 243 *enc->cur++ = major | len;
171 else if (len <= 0xff) 244 else if (ecb_expect_true (len <= 0xffU))
172 { 245 {
173 *enc->cur++ = major | 24; 246 *enc->cur++ = major | LENGTH_EXT1;
174 *enc->cur++ = len; 247 *enc->cur++ = len;
175 } 248 }
176 else if (len <= 0xffff) 249 else if (len <= 0xffffU)
177 { 250 {
178 *enc->cur++ = major | 25; 251 *enc->cur++ = major | LENGTH_EXT2;
179 *enc->cur++ = len >> 8; 252 *enc->cur++ = len >> 8;
180 *enc->cur++ = len; 253 *enc->cur++ = len;
181 } 254 }
182 else if (len <= 0xffffffff) 255 else if (len <= 0xffffffffU)
183 { 256 {
184 *enc->cur++ = major | 26; 257 *enc->cur++ = major | LENGTH_EXT4;
185 *enc->cur++ = len >> 24; 258 *enc->cur++ = len >> 24;
186 *enc->cur++ = len >> 16; 259 *enc->cur++ = len >> 16;
187 *enc->cur++ = len >> 8; 260 *enc->cur++ = len >> 8;
188 *enc->cur++ = len; 261 *enc->cur++ = len;
189 } 262 }
190 else 263 else
191 { 264 {
192 *enc->cur++ = major | 27; 265 *enc->cur++ = major | LENGTH_EXT8;
193 *enc->cur++ = len >> 56; 266 *enc->cur++ = len >> 56;
194 *enc->cur++ = len >> 48; 267 *enc->cur++ = len >> 48;
195 *enc->cur++ = len >> 40; 268 *enc->cur++ = len >> 40;
196 *enc->cur++ = len >> 32; 269 *enc->cur++ = len >> 32;
197 *enc->cur++ = len >> 24; 270 *enc->cur++ = len >> 24;
199 *enc->cur++ = len >> 8; 272 *enc->cur++ = len >> 8;
200 *enc->cur++ = len; 273 *enc->cur++ = len;
201 } 274 }
202} 275}
203 276
204static void 277ecb_inline void
278encode_tag (enc_t *enc, UV tag)
279{
280 encode_uint (enc, MAJOR_TAG, tag);
281}
282
283// exceptional (hopefully) slow path for byte strings that need to be utf8-encoded
284ecb_noinline static void
205encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 285encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len)
206{ 286{
207 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 287 STRLEN ulen = len;
288 U8 *p, *pend = (U8 *)str + len;
289
290 for (p = (U8 *)str; p < pend; ++p)
291 ulen += *p >> 7; // count set high bits
292
293 encode_uint (enc, MAJOR_TEXT, ulen);
294
295 need (enc, ulen);
296 for (p = (U8 *)str; p < pend; ++p)
297 if (*p < 0x80)
298 *enc->cur++ = *p;
299 else
300 {
301 *enc->cur++ = 0xc0 + (*p >> 6);
302 *enc->cur++ = 0x80 + (*p & 63);
303 }
304}
305
306ecb_inline void
307encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
308{
309 if (ecb_expect_false (upgrade_utf8))
310 if (!utf8)
311 {
312 encode_str_utf8 (enc, utf8, str, len);
313 return;
314 }
315
316 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
208 need (enc, len); 317 need (enc, len);
209 memcpy (enc->cur, str, len); 318 memcpy (enc->cur, str, len);
210 enc->cur += len; 319 enc->cur += len;
211} 320}
212 321
213ecb_inline void 322ecb_inline void
214encode_tag (enc_t *enc, UV tag) 323encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
215{ 324{
216 encode_uint (enc, 0xc0, tag); 325 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
326 {
327 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
328
329 if (SvOK (*svp))
330 {
331 // already registered, use stringref
332 encode_tag (enc, CBOR_TAG_STRINGREF);
333 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
334 return;
335 }
336 else if (len >= minimum_string_length (enc->stringref_idx))
337 {
338 // register only
339 sv_setuv (*svp, enc->stringref_idx);
340 ++enc->stringref_idx;
341 }
342 }
343
344 encode_str (enc, upgrade_utf8, utf8, str, len);
217} 345}
218 346
219static void encode_sv (enc_t *enc, SV *sv); 347static void encode_sv (enc_t *enc, SV *sv);
220 348
221static void 349static void
226 if (enc->depth >= enc->cbor.max_depth) 354 if (enc->depth >= enc->cbor.max_depth)
227 croak (ERR_NESTING_EXCEEDED); 355 croak (ERR_NESTING_EXCEEDED);
228 356
229 ++enc->depth; 357 ++enc->depth;
230 358
231 encode_uint (enc, 0x80, len + 1); 359 encode_uint (enc, MAJOR_ARRAY, len + 1);
232 360
361 if (SvMAGICAL (av))
233 for (i = 0; i <= len; ++i) 362 for (i = 0; i <= len; ++i)
234 { 363 {
235 SV **svp = av_fetch (av, i, 0); 364 SV **svp = av_fetch (av, i, 0);
236 encode_sv (enc, svp ? *svp : &PL_sv_undef); 365 encode_sv (enc, svp ? *svp : &PL_sv_undef);
237 } 366 }
367 else
368 for (i = 0; i <= len; ++i)
369 {
370 SV *sv = AvARRAY (av)[i];
371 encode_sv (enc, sv ? sv : &PL_sv_undef);
372 }
238 373
239 --enc->depth; 374 --enc->depth;
240} 375}
241 376
242static void 377static void
251 386
252 int pairs = hv_iterinit (hv); 387 int pairs = hv_iterinit (hv);
253 int mg = SvMAGICAL (hv); 388 int mg = SvMAGICAL (hv);
254 389
255 if (mg) 390 if (mg)
256 encode_ch (enc, 0xa0 | 31); 391 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
257 else 392 else
258 encode_uint (enc, 0xa0, pairs); 393 encode_uint (enc, MAJOR_MAP, pairs);
259 394
260 while ((he = hv_iternext (hv))) 395 while ((he = hv_iternext (hv)))
261 { 396 {
262 if (HeKLEN (he) == HEf_SVKEY) 397 if (HeKLEN (he) == HEf_SVKEY)
263 encode_sv (enc, HeSVKEY (he)); 398 encode_sv (enc, HeSVKEY (he));
264 else 399 else
265 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 400 encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS | F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he));
266 401
267 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 402 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
268 } 403 }
269 404
270 if (mg) 405 if (mg)
271 encode_ch (enc, 0xe0 | 31); 406 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
272 407
273 --enc->depth; 408 --enc->depth;
274} 409}
275 410
276// encode objects, arrays and special \0=false and \1=true values. 411// encode objects, arrays and special \0=false and \1=true values.
277static void 412static void
278encode_rv (enc_t *enc, SV *sv) 413encode_rv (enc_t *enc, SV *sv)
279{ 414{
280 SvGETMAGIC (sv); 415 SvGETMAGIC (sv);
281
282 if (ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)
283 && ecb_expect_false (SvREFCNT (sv) > 1))
284 {
285 if (!enc->shareable)
286 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
287
288 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
289
290 if (SvOK (*svp))
291 {
292 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
293 encode_uint (enc, 0x00, SvUV (*svp));
294 return;
295 }
296 else
297 {
298 sv_setuv (*svp, enc->shareable_idx);
299 ++enc->shareable_idx;
300 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
301 }
302 }
303 416
304 svtype svt = SvTYPE (sv); 417 svtype svt = SvTYPE (sv);
305 418
306 if (ecb_expect_false (SvOBJECT (sv))) 419 if (ecb_expect_false (SvOBJECT (sv)))
307 { 420 {
314 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 427 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
315 ? cbor_tagged_stash 428 ? cbor_tagged_stash
316 : gv_stashpv ("CBOR::XS::Tagged" , 1); 429 : gv_stashpv ("CBOR::XS::Tagged" , 1);
317 430
318 HV *stash = SvSTASH (sv); 431 HV *stash = SvSTASH (sv);
319 GV *method;
320 432
321 if (stash == boolean_stash) 433 if (stash == boolean_stash)
322 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 434 {
435 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
436 return;
437 }
323 else if (stash == error_stash) 438 else if (stash == error_stash)
324 encode_ch (enc, 0xe0 | 23); 439 {
440 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
441 return;
442 }
325 else if (stash == tagged_stash) 443 else if (stash == tagged_stash)
326 { 444 {
327 if (svt != SVt_PVAV) 445 if (svt != SVt_PVAV)
328 croak ("encountered CBOR::XS::Tagged object that isn't an array"); 446 croak ("encountered CBOR::XS::Tagged object that isn't an array");
329 447
330 encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1))); 448 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
331 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); 449 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
450
451 return;
452 }
453 }
454
455 if (ecb_expect_false (SvREFCNT (sv) > 1)
456 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
457 {
458 if (!enc->shareable)
459 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
460
461 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
462
463 if (SvOK (*svp))
332 } 464 {
465 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
466 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
467 return;
468 }
469 else
470 {
471 sv_setuv (*svp, enc->shareable_idx);
472 ++enc->shareable_idx;
473 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
474 }
475 }
476
477 if (ecb_expect_false (SvOBJECT (sv)))
478 {
479 HV *stash = SvSTASH (sv);
480 GV *method;
481
333 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 482 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
334 { 483 {
335 dSP; 484 dSP;
336 485
337 ENTER; SAVETMPS; PUSHMARK (SP); 486 ENTER; SAVETMPS;
487 PUSHMARK (SP);
338 // we re-bless the reference to get overload and other niceties right 488 // we re-bless the reference to get overload and other niceties right
339 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 489 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
340 490
341 PUTBACK; 491 PUTBACK;
342 // G_SCALAR ensures that return value is 1 492 // G_SCALAR ensures that return value is 1
355 } 505 }
356 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) 506 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
357 { 507 {
358 dSP; 508 dSP;
359 509
360 ENTER; SAVETMPS; PUSHMARK (SP); 510 ENTER; SAVETMPS;
511 SAVESTACK_POS ();
512 PUSHMARK (SP);
361 EXTEND (SP, 2); 513 EXTEND (SP, 2);
362 // we re-bless the reference to get overload and other niceties right 514 // we re-bless the reference to get overload and other niceties right
363 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 515 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
364 PUSHs (sv_cbor); 516 PUSHs (sv_cbor);
365 517
370 // catch this surprisingly common error 522 // catch this surprisingly common error
371 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 523 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
372 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); 524 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
373 525
374 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 526 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
375 encode_uint (enc, 0x80, count + 1); 527 encode_uint (enc, MAJOR_ARRAY, count + 1);
376 encode_str (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 528 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
377 529
378 while (count) 530 while (count)
379 encode_sv (enc, SP[1 - count--]); 531 encode_sv (enc, SP[1 - count--]);
380 532
381 PUTBACK; 533 PUTBACK;
402{ 554{
403 double nv = SvNVX (sv); 555 double nv = SvNVX (sv);
404 556
405 need (enc, 9); 557 need (enc, 9);
406 558
407 if (ecb_expect_false (nv == (U32)nv)) 559 if (ecb_expect_false (nv == (NV)(U32)nv))
408 encode_uint (enc, 0x00, (U32)nv); 560 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
409 //TODO: maybe I32? 561 //TODO: maybe I32?
410 else if (ecb_expect_false (nv == (float)nv)) 562 else if (ecb_expect_false (nv == (float)nv))
411 { 563 {
412 uint32_t fp = ecb_float_to_binary32 (nv); 564 uint32_t fp = ecb_float_to_binary32 (nv);
413 565
414 *enc->cur++ = 0xe0 | 26; 566 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
415 567
416 if (!ecb_big_endian ()) 568 if (!ecb_big_endian ())
417 fp = ecb_bswap32 (fp); 569 fp = ecb_bswap32 (fp);
418 570
419 memcpy (enc->cur, &fp, 4); 571 memcpy (enc->cur, &fp, 4);
421 } 573 }
422 else 574 else
423 { 575 {
424 uint64_t fp = ecb_double_to_binary64 (nv); 576 uint64_t fp = ecb_double_to_binary64 (nv);
425 577
426 *enc->cur++ = 0xe0 | 27; 578 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
427 579
428 if (!ecb_big_endian ()) 580 if (!ecb_big_endian ())
429 fp = ecb_bswap64 (fp); 581 fp = ecb_bswap64 (fp);
430 582
431 memcpy (enc->cur, &fp, 8); 583 memcpy (enc->cur, &fp, 8);
440 592
441 if (SvPOKp (sv)) 593 if (SvPOKp (sv))
442 { 594 {
443 STRLEN len; 595 STRLEN len;
444 char *str = SvPV (sv, len); 596 char *str = SvPV (sv, len);
445 encode_str (enc, SvUTF8 (sv), str, len); 597 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
446 } 598 }
447 else if (SvNOKp (sv)) 599 else if (SvNOKp (sv))
448 encode_nv (enc, sv); 600 encode_nv (enc, sv);
449 else if (SvIOKp (sv)) 601 else if (SvIOKp (sv))
450 { 602 {
451 if (SvIsUV (sv)) 603 if (SvIsUV (sv))
452 encode_uint (enc, 0x00, SvUVX (sv)); 604 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
453 else if (SvIVX (sv) >= 0) 605 else if (SvIVX (sv) >= 0)
454 encode_uint (enc, 0x00, SvIVX (sv)); 606 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
455 else 607 else
456 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 608 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
457 } 609 }
458 else if (SvROK (sv)) 610 else if (SvROK (sv))
459 encode_rv (enc, SvRV (sv)); 611 encode_rv (enc, SvRV (sv));
460 else if (!SvOK (sv)) 612 else if (!SvOK (sv))
461 encode_ch (enc, 0xe0 | 22); 613 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
462 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 614 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
463 encode_ch (enc, 0xe0 | 23); 615 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
464 else 616 else
465 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 617 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
466 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 618 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
467} 619}
468 620
469static SV * 621static SV *
470encode_cbor (SV *scalar, CBOR *cbor) 622encode_cbor (SV *scalar, CBOR *cbor)
471{ 623{
472 enc_t enc = { }; 624 enc_t enc = { 0 };
473 625
474 enc.cbor = *cbor; 626 enc.cbor = *cbor;
475 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 627 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
476 enc.cur = SvPVX (enc.sv); 628 enc.cur = SvPVX (enc.sv);
477 enc.end = SvEND (enc.sv); 629 enc.end = SvEND (enc.sv);
478 630
479 SvPOK_only (enc.sv); 631 SvPOK_only (enc.sv);
632
633 if (cbor->flags & F_PACK_STRINGS)
634 {
635 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
636 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
637 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
638 }
639
480 encode_sv (&enc, scalar); 640 encode_sv (&enc, scalar);
481 641
482 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 642 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
483 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 643 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
484 644
499 const char *err; // parse error, if != 0 659 const char *err; // parse error, if != 0
500 CBOR cbor; 660 CBOR cbor;
501 U32 depth; // recursion depth 661 U32 depth; // recursion depth
502 U32 maxdepth; // recursion depth limit 662 U32 maxdepth; // recursion depth limit
503 AV *shareable; 663 AV *shareable;
664 AV *stringref;
665 SV *decode_tagged;
504} dec_t; 666} dec_t;
505 667
506#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 668#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE
507 669
508#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 670#define WANT(len) if (ecb_expect_false ((UV)(dec->end - dec->cur) < (UV)len)) ERR ("unexpected end of CBOR data")
509 671
510#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED) 672#define DEC_INC_DEPTH if (ecb_expect_false (++dec->depth > dec->cbor.max_depth)) ERR (ERR_NESTING_EXCEEDED)
511#define DEC_DEC_DEPTH --dec->depth 673#define DEC_DEC_DEPTH --dec->depth
512 674
513static UV 675static UV
514decode_uint (dec_t *dec) 676decode_uint (dec_t *dec)
515{ 677{
516 switch (*dec->cur & 31) 678 U8 m = *dec->cur & MINOR_MASK;
517 { 679 ++dec->cur;
518 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
519 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
520 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
521 return *dec->cur++ & 31;
522 680
523 case 24: 681 if (ecb_expect_true (m < LENGTH_EXT1))
682 return m;
683 else if (ecb_expect_true (m == LENGTH_EXT1))
684 {
524 WANT (2); 685 WANT (1);
525 dec->cur += 2; 686 dec->cur += 1;
526 return dec->cur[-1]; 687 return dec->cur[-1];
527 688 }
528 case 25: 689 else if (ecb_expect_true (m == LENGTH_EXT2))
690 {
529 WANT (3); 691 WANT (2);
530 dec->cur += 3; 692 dec->cur += 2;
531 return (((UV)dec->cur[-2]) << 8) 693 return (((UV)dec->cur[-2]) << 8)
532 | ((UV)dec->cur[-1]); 694 | ((UV)dec->cur[-1]);
533 695 }
534 case 26: 696 else if (ecb_expect_true (m == LENGTH_EXT4))
697 {
535 WANT (5); 698 WANT (4);
536 dec->cur += 5; 699 dec->cur += 4;
537 return (((UV)dec->cur[-4]) << 24) 700 return (((UV)dec->cur[-4]) << 24)
538 | (((UV)dec->cur[-3]) << 16) 701 | (((UV)dec->cur[-3]) << 16)
539 | (((UV)dec->cur[-2]) << 8) 702 | (((UV)dec->cur[-2]) << 8)
540 | ((UV)dec->cur[-1]); 703 | ((UV)dec->cur[-1]);
541 704 }
542 case 27: 705 else if (ecb_expect_true (m == LENGTH_EXT8))
706 {
543 WANT (9); 707 WANT (8);
544 dec->cur += 9; 708 dec->cur += 8;
709
710 return
711#if UVSIZE < 8
712 0
713#else
545 return (((UV)dec->cur[-8]) << 56) 714 (((UV)dec->cur[-8]) << 56)
546 | (((UV)dec->cur[-7]) << 48) 715 | (((UV)dec->cur[-7]) << 48)
547 | (((UV)dec->cur[-6]) << 40) 716 | (((UV)dec->cur[-6]) << 40)
548 | (((UV)dec->cur[-5]) << 32) 717 | (((UV)dec->cur[-5]) << 32)
718#endif
549 | (((UV)dec->cur[-4]) << 24) 719 | (((UV)dec->cur[-4]) << 24)
550 | (((UV)dec->cur[-3]) << 16) 720 | (((UV)dec->cur[-3]) << 16)
551 | (((UV)dec->cur[-2]) << 8) 721 | (((UV)dec->cur[-2]) << 8)
552 | ((UV)dec->cur[-1]); 722 | ((UV)dec->cur[-1]);
553 723 }
554 default: 724 else
555 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 725 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
556 }
557 726
558fail: 727fail:
559 return 0; 728 return 0;
560} 729}
561 730
566{ 735{
567 AV *av = newAV (); 736 AV *av = newAV ();
568 737
569 DEC_INC_DEPTH; 738 DEC_INC_DEPTH;
570 739
571 if ((*dec->cur & 31) == 31) 740 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
572 { 741 {
573 ++dec->cur; 742 ++dec->cur;
574 743
575 for (;;) 744 for (;;)
576 { 745 {
577 WANT (1); 746 WANT (1);
578 747
579 if (*dec->cur == (0xe0 | 31)) 748 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
580 { 749 {
581 ++dec->cur; 750 ++dec->cur;
582 break; 751 break;
583 } 752 }
584 753
585 av_push (av, decode_sv (dec)); 754 av_push (av, decode_sv (dec));
586 } 755 }
587 } 756 }
588 else 757 else
589 { 758 {
590 int i, len = decode_uint (dec); 759 UV i, len = decode_uint (dec);
591 760
761 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
592 av_fill (av, len - 1); 762 av_fill (av, len - 1);
593 763
594 for (i = 0; i < len; ++i) 764 for (i = 0; i < len; ++i)
595 AvARRAY (av)[i] = decode_sv (dec); 765 AvARRAY (av)[i] = decode_sv (dec);
596 } 766 }
606 776
607static void 777static void
608decode_he (dec_t *dec, HV *hv) 778decode_he (dec_t *dec, HV *hv)
609{ 779{
610 // for speed reasons, we specialcase single-string 780 // for speed reasons, we specialcase single-string
611 // byte or utf-8 strings as keys. 781 // byte or utf-8 strings as keys, but only when !stringref
612 782
613 if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27) 783 if (ecb_expect_true (!dec->stringref))
784 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
614 { 785 {
615 I32 len = decode_uint (dec); 786 STRLEN len = decode_uint (dec);
616 char *key = (char *)dec->cur; 787 char *key = (char *)dec->cur;
617 788
789 WANT (len);
618 dec->cur += len; 790 dec->cur += len;
619 791
620 hv_store (hv, key, len, decode_sv (dec), 0); 792 hv_store (hv, key, len, decode_sv (dec), 0);
793
794 return;
621 } 795 }
622 else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27) 796 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
623 { 797 {
624 I32 len = decode_uint (dec); 798 STRLEN len = decode_uint (dec);
625 char *key = (char *)dec->cur; 799 char *key = (char *)dec->cur;
626 800
801 WANT (len);
627 dec->cur += len; 802 dec->cur += len;
628 803
804 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
805 if (!is_utf8_string (key, len))
806 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
807
629 hv_store (hv, key, -len, decode_sv (dec), 0); 808 hv_store (hv, key, -len, decode_sv (dec), 0);
809
810 return;
630 } 811 }
631 else 812
632 {
633 SV *k = decode_sv (dec); 813 SV *k = decode_sv (dec);
634 SV *v = decode_sv (dec); 814 SV *v = decode_sv (dec);
635 815
636 hv_store_ent (hv, k, v, 0); 816 hv_store_ent (hv, k, v, 0);
637 SvREFCNT_dec (k); 817 SvREFCNT_dec (k);
638 } 818
819fail:
820 ;
639} 821}
640 822
641static SV * 823static SV *
642decode_hv (dec_t *dec) 824decode_hv (dec_t *dec)
643{ 825{
644 HV *hv = newHV (); 826 HV *hv = newHV ();
645 827
646 DEC_INC_DEPTH; 828 DEC_INC_DEPTH;
647 829
648 if ((*dec->cur & 31) == 31) 830 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
649 { 831 {
650 ++dec->cur; 832 ++dec->cur;
651 833
652 for (;;) 834 for (;;)
653 { 835 {
654 WANT (1); 836 WANT (1);
655 837
656 if (*dec->cur == (0xe0 | 31)) 838 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
657 { 839 {
658 ++dec->cur; 840 ++dec->cur;
659 break; 841 break;
660 } 842 }
661 843
662 decode_he (dec, hv); 844 decode_he (dec, hv);
663 } 845 }
664 } 846 }
665 else 847 else
666 { 848 {
667 int pairs = decode_uint (dec); 849 UV pairs = decode_uint (dec);
850
851 WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes
668 852
669 while (pairs--) 853 while (pairs--)
670 decode_he (dec, hv); 854 decode_he (dec, hv);
671 } 855 }
672 856
682static SV * 866static SV *
683decode_str (dec_t *dec, int utf8) 867decode_str (dec_t *dec, int utf8)
684{ 868{
685 SV *sv = 0; 869 SV *sv = 0;
686 870
687 if ((*dec->cur & 31) == 31) 871 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
688 { 872 {
873 // indefinite length strings
689 ++dec->cur; 874 ++dec->cur;
690 875
876 U8 major = *dec->cur & MAJOR_MISC;
877
691 sv = newSVpvn ("", 0); 878 sv = newSVpvn ("", 0);
692 879
693 // not very fast, and certainly not robust against illegal input
694 for (;;) 880 for (;;)
695 { 881 {
696 WANT (1); 882 WANT (1);
697 883
698 if (*dec->cur == (0xe0 | 31)) 884 if ((*dec->cur - major) > LENGTH_EXT8)
885 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
699 { 886 {
700 ++dec->cur; 887 ++dec->cur;
701 break; 888 break;
702 } 889 }
890 else
891 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
703 892
704 sv_catsv (sv, decode_sv (dec)); 893 STRLEN len = decode_uint (dec);
894
895 WANT (len);
896 sv_catpvn (sv, dec->cur, len);
897 dec->cur += len;
705 } 898 }
706 } 899 }
707 else 900 else
708 { 901 {
709 STRLEN len = decode_uint (dec); 902 STRLEN len = decode_uint (dec);
710 903
711 WANT (len); 904 WANT (len);
712 sv = newSVpvn (dec->cur, len); 905 sv = newSVpvn (dec->cur, len);
713 dec->cur += len; 906 dec->cur += len;
907
908 if (ecb_expect_false (dec->stringref)
909 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
910 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
714 } 911 }
715 912
716 if (utf8) 913 if (utf8)
914 {
915 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
916 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
917 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
918
717 SvUTF8_on (sv); 919 SvUTF8_on (sv);
920 }
718 921
719 return sv; 922 return sv;
720 923
721fail: 924fail:
722 SvREFCNT_dec (sv); 925 SvREFCNT_dec (sv);
732 WANT (1); 935 WANT (1);
733 936
734 switch (tag) 937 switch (tag)
735 { 938 {
736 case CBOR_TAG_MAGIC: 939 case CBOR_TAG_MAGIC:
737 return decode_sv (dec); 940 sv = decode_sv (dec);
941 break;
738 942
739 case CBOR_TAG_INDIRECTION: 943 case CBOR_TAG_INDIRECTION:
740 return newRV_noinc (decode_sv (dec)); 944 sv = newRV_noinc (decode_sv (dec));
945 break;
946
947 case CBOR_TAG_STRINGREF_NAMESPACE:
948 {
949 // do nmot use SAVETMPS/FREETMPS, as these will
950 // erase mortalised caches, e.g. "shareable"
951 ENTER;
952
953 SAVESPTR (dec->stringref);
954 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
955
956 sv = decode_sv (dec);
957
958 LEAVE;
959 }
960 break;
961
962 case CBOR_TAG_STRINGREF:
963 {
964 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
965 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
966
967 UV idx = decode_uint (dec);
968
969 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
970 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
971
972 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
973 }
974 break;
741 975
742 case CBOR_TAG_VALUE_SHAREABLE: 976 case CBOR_TAG_VALUE_SHAREABLE:
743 { 977 {
744 if (ecb_expect_false (!dec->shareable)) 978 if (ecb_expect_false (!dec->shareable))
745 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ()); 979 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
746 980
981 if (dec->cbor.flags & F_ALLOW_CYCLES)
982 {
747 sv = newSV (0); 983 sv = newSV (0);
748 av_push (dec->shareable, SvREFCNT_inc_NN (sv)); 984 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
749 985
750 SV *osv = decode_sv (dec); 986 SV *osv = decode_sv (dec);
751 sv_setsv (sv, osv); 987 sv_setsv (sv, osv);
752 SvREFCNT_dec_NN (osv); 988 SvREFCNT_dec_NN (osv);
753 } 989 }
754 990 else
755 return sv; 991 {
992 av_push (dec->shareable, &PL_sv_undef);
993 int idx = AvFILLp (dec->shareable);
994 sv = decode_sv (dec);
995 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
996 }
997 }
998 break;
756 999
757 case CBOR_TAG_VALUE_SHAREDREF: 1000 case CBOR_TAG_VALUE_SHAREDREF:
758 { 1001 {
759 if ((*dec->cur >> 5) != 0) 1002 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
760 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); 1003 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
761 1004
762 UV idx = decode_uint (dec); 1005 UV idx = decode_uint (dec);
763 1006
764 if (!dec->shareable || idx > AvFILLp (dec->shareable)) 1007 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
765 ERR ("corrupted CBOR data (sharedref index out of bounds)"); 1008 ERR ("corrupted CBOR data (sharedref index out of bounds)");
766 1009
767 return SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); 1010 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
1011
1012 if (sv == &PL_sv_undef)
1013 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
768 } 1014 }
1015 break;
769 1016
770 case CBOR_TAG_PERL_OBJECT: 1017 case CBOR_TAG_PERL_OBJECT:
771 { 1018 {
772 sv = decode_sv (dec); 1019 sv = decode_sv (dec);
773 1020
786 if (!method) 1033 if (!method)
787 ERR ("cannot decode perl-object (package does not have a THAW method)"); 1034 ERR ("cannot decode perl-object (package does not have a THAW method)");
788 1035
789 dSP; 1036 dSP;
790 1037
791 ENTER; SAVETMPS; PUSHMARK (SP); 1038 ENTER; SAVETMPS;
1039 PUSHMARK (SP);
792 EXTEND (SP, len + 1); 1040 EXTEND (SP, len + 1);
793 // we re-bless the reference to get overload and other niceties right 1041 // we re-bless the reference to get overload and other niceties right
794 PUSHs (*av_fetch (av, 0, 1)); 1042 PUSHs (*av_fetch (av, 0, 1));
795 PUSHs (sv_cbor); 1043 PUSHs (sv_cbor);
796 1044
813 sv = SvREFCNT_inc (POPs); 1061 sv = SvREFCNT_inc (POPs);
814 1062
815 PUTBACK; 1063 PUTBACK;
816 1064
817 FREETMPS; LEAVE; 1065 FREETMPS; LEAVE;
818
819 return sv;
820 } 1066 }
1067 break;
821 1068
822 default: 1069 default:
823 { 1070 {
824 sv = decode_sv (dec); 1071 sv = decode_sv (dec);
825 1072
1073 dSP;
1074 ENTER; SAVETMPS;
1075 SAVESTACK_POS ();
1076 PUSHMARK (SP);
1077 EXTEND (SP, 2);
1078 PUSHs (newSVuv (tag));
1079 PUSHs (sv);
1080
1081 PUTBACK;
1082 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1083 SPAGAIN;
1084
1085 if (SvTRUE (ERRSV))
1086 {
1087 FREETMPS; LEAVE;
1088 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1089 }
1090
1091 if (count)
1092 {
1093 SvREFCNT_dec (sv);
1094 sv = SvREFCNT_inc (POPs);
1095 }
1096 else
1097 {
826 AV *av = newAV (); 1098 AV *av = newAV ();
827 av_push (av, newSVuv (tag)); 1099 av_push (av, newSVuv (tag));
828 av_push (av, sv); 1100 av_push (av, sv);
829 1101
830 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1102 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
831 ? cbor_tagged_stash 1103 ? cbor_tagged_stash
832 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1104 : gv_stashpv ("CBOR::XS::Tagged" , 1);
833
834 return sv_bless (newRV_noinc ((SV *)av), tagged_stash); 1105 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
835 } 1106 }
1107
1108 PUTBACK;
1109
1110 FREETMPS; LEAVE;
1111 }
1112 break;
836 } 1113 }
1114
1115 return sv;
837 1116
838fail: 1117fail:
839 SvREFCNT_dec (sv); 1118 SvREFCNT_dec (sv);
840 return &PL_sv_undef; 1119 return &PL_sv_undef;
841} 1120}
843static SV * 1122static SV *
844decode_sv (dec_t *dec) 1123decode_sv (dec_t *dec)
845{ 1124{
846 WANT (1); 1125 WANT (1);
847 1126
848 switch (*dec->cur >> 5) 1127 switch (*dec->cur >> MAJOR_SHIFT)
849 { 1128 {
850 case 0: // unsigned int 1129 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
851 return newSVuv (decode_uint (dec)); 1130 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
852 case 1: // negative int 1131 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
853 return newSViv (-1 - (IV)decode_uint (dec)); 1132 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
854 case 2: // octet string 1133 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
855 return decode_str (dec, 0); 1134 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
856 case 3: // utf-8 string 1135 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
857 return decode_str (dec, 1); 1136
858 case 4: // array 1137 case MAJOR_MISC >> MAJOR_SHIFT:
859 return decode_av (dec);
860 case 5: // map
861 return decode_hv (dec);
862 case 6: // tag
863 return decode_tagged (dec);
864 case 7: // misc
865 switch (*dec->cur++ & 31) 1138 switch (*dec->cur++ & MINOR_MASK)
866 { 1139 {
867 case 20: 1140 case SIMPLE_FALSE:
868#if CBOR_SLOW 1141#if CBOR_SLOW
869 types_false = get_bool ("Types::Serialiser::false"); 1142 types_false = get_bool ("Types::Serialiser::false");
870#endif 1143#endif
871 return newSVsv (types_false); 1144 return newSVsv (types_false);
872 case 21: 1145 case SIMPLE_TRUE:
873#if CBOR_SLOW 1146#if CBOR_SLOW
874 types_true = get_bool ("Types::Serialiser::true"); 1147 types_true = get_bool ("Types::Serialiser::true");
875#endif 1148#endif
876 return newSVsv (types_true); 1149 return newSVsv (types_true);
877 case 22: 1150 case SIMPLE_NULL:
878 return newSVsv (&PL_sv_undef); 1151 return newSVsv (&PL_sv_undef);
879 case 23: 1152 case SIMPLE_UNDEF:
880#if CBOR_SLOW 1153#if CBOR_SLOW
881 types_error = get_bool ("Types::Serialiser::error"); 1154 types_error = get_bool ("Types::Serialiser::error");
882#endif 1155#endif
883 return newSVsv (types_error); 1156 return newSVsv (types_error);
884 1157
885 case 25: 1158 case MISC_FLOAT16:
886 { 1159 {
887 WANT (2); 1160 WANT (2);
888 1161
889 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1162 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
890 dec->cur += 2; 1163 dec->cur += 2;
891 1164
892 return newSVnv (ecb_binary16_to_float (fp)); 1165 return newSVnv (ecb_binary16_to_float (fp));
893 } 1166 }
894 1167
895 case 26: 1168 case MISC_FLOAT32:
896 { 1169 {
897 uint32_t fp; 1170 uint32_t fp;
898 WANT (4); 1171 WANT (4);
899 memcpy (&fp, dec->cur, 4); 1172 memcpy (&fp, dec->cur, 4);
900 dec->cur += 4; 1173 dec->cur += 4;
903 fp = ecb_bswap32 (fp); 1176 fp = ecb_bswap32 (fp);
904 1177
905 return newSVnv (ecb_binary32_to_float (fp)); 1178 return newSVnv (ecb_binary32_to_float (fp));
906 } 1179 }
907 1180
908 case 27: 1181 case MISC_FLOAT64:
909 { 1182 {
910 uint64_t fp; 1183 uint64_t fp;
911 WANT (8); 1184 WANT (8);
912 memcpy (&fp, dec->cur, 8); 1185 memcpy (&fp, dec->cur, 8);
913 dec->cur += 8; 1186 dec->cur += 8;
916 fp = ecb_bswap64 (fp); 1189 fp = ecb_bswap64 (fp);
917 1190
918 return newSVnv (ecb_binary64_to_double (fp)); 1191 return newSVnv (ecb_binary64_to_double (fp));
919 } 1192 }
920 1193
921 // 0..19 unassigned 1194 // 0..19 unassigned simple
922 // 24 reserved + unassigned (reserved values are not encodable) 1195 // 24 reserved + unassigned simple (reserved values are not encodable)
1196 // 28-30 unassigned misc
1197 // 31 break code
923 default: 1198 default:
924 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1199 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
925 } 1200 }
926 1201
927 break; 1202 break;
928 } 1203 }
929 1204
932} 1207}
933 1208
934static SV * 1209static SV *
935decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1210decode_cbor (SV *string, CBOR *cbor, char **offset_return)
936{ 1211{
937 dec_t dec = { }; 1212 dec_t dec = { 0 };
938 SV *sv; 1213 SV *sv;
939 STRLEN len; 1214 STRLEN len;
940 char *data = SvPVbyte (string, len); 1215 char *data = SvPVbyte (string, len);
941 1216
942 if (len > cbor->max_size && cbor->max_size) 1217 if (len > cbor->max_size && cbor->max_size)
956 if (dec.cur != dec.end && !dec.err) 1231 if (dec.cur != dec.end && !dec.err)
957 dec.err = "garbage after CBOR object"; 1232 dec.err = "garbage after CBOR object";
958 1233
959 if (dec.err) 1234 if (dec.err)
960 { 1235 {
1236 if (dec.shareable)
1237 {
1238 // need to break cyclic links, which whould all be in shareable
1239 int i;
1240 SV **svp;
1241
1242 for (i = av_len (dec.shareable) + 1; i--; )
1243 if ((svp = av_fetch (dec.shareable, i, 0)))
1244 sv_setsv (*svp, &PL_sv_undef);
1245 }
1246
961 SvREFCNT_dec (sv); 1247 SvREFCNT_dec (sv);
962 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); 1248 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
963 } 1249 }
964 1250
965 sv = sv_2mortal (sv); 1251 sv = sv_2mortal (sv);
966 1252
967 return sv; 1253 return sv;
968} 1254}
969 1255
1256/////////////////////////////////////////////////////////////////////////////
1257// incremental parser
1258
1259#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1260
1261// returns 0 for notyet, 1 for success or error
1262static int
1263incr_parse (CBOR *self, SV *cborstr)
1264{
1265 STRLEN cur;
1266 SvPV (cborstr, cur);
1267
1268 while (ecb_expect_true (self->incr_need <= cur))
1269 {
1270 // table of integer count bytes
1271 static I8 incr_len[MINOR_MASK + 1] = {
1272 0, 0, 0, 0, 0, 0, 0, 0,
1273 0, 0, 0, 0, 0, 0, 0, 0,
1274 0, 0, 0, 0, 0, 0, 0, 0,
1275 1, 2, 4, 8,-1,-1,-1,-2
1276 };
1277
1278 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1279 U8 m = *p & MINOR_MASK;
1280 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1281 I8 ilen = incr_len[m];
1282
1283 self->incr_need = self->incr_pos + 1;
1284
1285 if (ecb_expect_false (ilen < 0))
1286 {
1287 if (m != MINOR_INDEF)
1288 return 1; // error
1289
1290 if (*p == (MAJOR_MISC | MINOR_INDEF))
1291 {
1292 if (count >= 0)
1293 return 1; // error
1294
1295 count = 1;
1296 }
1297 else
1298 {
1299 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1300 count = -1;
1301 }
1302 }
1303 else
1304 {
1305 self->incr_need += ilen;
1306 if (ecb_expect_false (self->incr_need > cur))
1307 return 0;
1308
1309 int major = *p >> MAJOR_SHIFT;
1310
1311 switch (major)
1312 {
1313 case MAJOR_TAG >> MAJOR_SHIFT:
1314 ++count; // tags merely prefix another value
1315 break;
1316
1317 case MAJOR_BYTES >> MAJOR_SHIFT:
1318 case MAJOR_TEXT >> MAJOR_SHIFT:
1319 case MAJOR_ARRAY >> MAJOR_SHIFT:
1320 case MAJOR_MAP >> MAJOR_SHIFT:
1321 {
1322 UV len;
1323
1324 if (ecb_expect_false (ilen))
1325 {
1326 len = 0;
1327
1328 do {
1329 len = (len << 8) | *++p;
1330 } while (--ilen);
1331 }
1332 else
1333 len = m;
1334
1335 switch (major)
1336 {
1337 case MAJOR_BYTES >> MAJOR_SHIFT:
1338 case MAJOR_TEXT >> MAJOR_SHIFT:
1339 self->incr_need += len;
1340 if (ecb_expect_false (self->incr_need > cur))
1341 return 0;
1342
1343 break;
1344
1345 case MAJOR_MAP >> MAJOR_SHIFT:
1346 len <<= 1;
1347 case MAJOR_ARRAY >> MAJOR_SHIFT:
1348 if (len)
1349 {
1350 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1351 count = len + 1;
1352 }
1353 break;
1354 }
1355 }
1356 }
1357 }
1358
1359 self->incr_pos = self->incr_need;
1360
1361 if (count > 0)
1362 {
1363 while (!--count)
1364 {
1365 if (!AvFILLp (self->incr_count))
1366 return 1; // done
1367
1368 SvREFCNT_dec_NN (av_pop (self->incr_count));
1369 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1370 }
1371
1372 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1373 }
1374 }
1375
1376 return 0;
1377}
1378
1379
970///////////////////////////////////////////////////////////////////////////// 1380/////////////////////////////////////////////////////////////////////////////
971// XS interface functions 1381// XS interface functions
972 1382
973MODULE = CBOR::XS PACKAGE = CBOR::XS 1383MODULE = CBOR::XS PACKAGE = CBOR::XS
974 1384
982 1392
983 types_true = get_bool ("Types::Serialiser::true" ); 1393 types_true = get_bool ("Types::Serialiser::true" );
984 types_false = get_bool ("Types::Serialiser::false"); 1394 types_false = get_bool ("Types::Serialiser::false");
985 types_error = get_bool ("Types::Serialiser::error"); 1395 types_error = get_bool ("Types::Serialiser::error");
986 1396
1397 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1398
987 sv_cbor = newSVpv ("CBOR", 0); 1399 sv_cbor = newSVpv ("CBOR", 0);
988 SvREADONLY_on (sv_cbor); 1400 SvREADONLY_on (sv_cbor);
1401
1402 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
989} 1403}
990 1404
991PROTOTYPES: DISABLE 1405PROTOTYPES: DISABLE
992 1406
993void CLONE (...) 1407void CLONE (...)
1012void shrink (CBOR *self, int enable = 1) 1426void shrink (CBOR *self, int enable = 1)
1013 ALIAS: 1427 ALIAS:
1014 shrink = F_SHRINK 1428 shrink = F_SHRINK
1015 allow_unknown = F_ALLOW_UNKNOWN 1429 allow_unknown = F_ALLOW_UNKNOWN
1016 allow_sharing = F_ALLOW_SHARING 1430 allow_sharing = F_ALLOW_SHARING
1017 dedup_keys = F_DEDUP_KEYS 1431 allow_cycles = F_ALLOW_CYCLES
1018 dedup_strings = F_DEDUP_STRINGS 1432 pack_strings = F_PACK_STRINGS
1433 text_keys = F_TEXT_KEYS
1434 text_strings = F_TEXT_STRINGS
1435 validate_utf8 = F_VALIDATE_UTF8
1019 PPCODE: 1436 PPCODE:
1020{ 1437{
1021 if (enable) 1438 if (enable)
1022 self->flags |= ix; 1439 self->flags |= ix;
1023 else 1440 else
1029void get_shrink (CBOR *self) 1446void get_shrink (CBOR *self)
1030 ALIAS: 1447 ALIAS:
1031 get_shrink = F_SHRINK 1448 get_shrink = F_SHRINK
1032 get_allow_unknown = F_ALLOW_UNKNOWN 1449 get_allow_unknown = F_ALLOW_UNKNOWN
1033 get_allow_sharing = F_ALLOW_SHARING 1450 get_allow_sharing = F_ALLOW_SHARING
1034 get_dedup_keys = F_DEDUP_KEYS 1451 get_allow_cycles = F_ALLOW_CYCLES
1035 get_dedup_strings = F_DEDUP_STRINGS 1452 get_pack_strings = F_PACK_STRINGS
1453 get_text_keys = F_TEXT_KEYS
1454 get_text_strings = F_TEXT_STRINGS
1455 get_validate_utf8 = F_VALIDATE_UTF8
1036 PPCODE: 1456 PPCODE:
1037 XPUSHs (boolSV (self->flags & ix)); 1457 XPUSHs (boolSV (self->flags & ix));
1038 1458
1039void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1459void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1040 PPCODE: 1460 PPCODE:
1053 XPUSHs (ST (0)); 1473 XPUSHs (ST (0));
1054 1474
1055int get_max_size (CBOR *self) 1475int get_max_size (CBOR *self)
1056 CODE: 1476 CODE:
1057 RETVAL = self->max_size; 1477 RETVAL = self->max_size;
1478 OUTPUT:
1479 RETVAL
1480
1481void filter (CBOR *self, SV *filter = 0)
1482 PPCODE:
1483 SvREFCNT_dec (self->filter);
1484 self->filter = filter ? newSVsv (filter) : filter;
1485 XPUSHs (ST (0));
1486
1487SV *get_filter (CBOR *self)
1488 CODE:
1489 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
1058 OUTPUT: 1490 OUTPUT:
1059 RETVAL 1491 RETVAL
1060 1492
1061void encode (CBOR *self, SV *scalar) 1493void encode (CBOR *self, SV *scalar)
1062 PPCODE: 1494 PPCODE:
1077 EXTEND (SP, 2); 1509 EXTEND (SP, 2);
1078 PUSHs (sv); 1510 PUSHs (sv);
1079 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1511 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1080} 1512}
1081 1513
1514void incr_parse (CBOR *self, SV *cborstr)
1515 ALIAS:
1516 incr_parse_multiple = 1
1517 PPCODE:
1518{
1519 if (SvUTF8 (cborstr))
1520 sv_utf8_downgrade (cborstr, 0);
1521
1522 if (!self->incr_count)
1523 {
1524 self->incr_count = newAV ();
1525 self->incr_pos = 0;
1526 self->incr_need = 1;
1527
1528 av_push (self->incr_count, newSViv (1));
1529 }
1530
1531 do
1532 {
1533 if (!incr_parse (self, cborstr))
1534 {
1535 if (self->incr_need > self->max_size && self->max_size)
1536 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1537 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1538
1539 break;
1540 }
1541
1542 SV *sv;
1543 char *offset;
1544
1545 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1546 XPUSHs (sv);
1547
1548 sv_chop (cborstr, offset);
1549
1550 av_clear (self->incr_count);
1551 av_push (self->incr_count, newSViv (1));
1552
1553 self->incr_pos = 0;
1554 self->incr_need = self->incr_pos + 1;
1555 }
1556 while (ix);
1557}
1558
1559void incr_reset (CBOR *self)
1560 CODE:
1561{
1562 SvREFCNT_dec (self->incr_count);
1563 self->incr_count = 0;
1564}
1565
1566void DESTROY (CBOR *self)
1567 PPCODE:
1568 cbor_free (self);
1569
1082PROTOTYPES: ENABLE 1570PROTOTYPES: ENABLE
1083 1571
1084void encode_cbor (SV *scalar) 1572void encode_cbor (SV *scalar)
1573 ALIAS:
1574 encode_cbor = 0
1575 encode_cbor_sharing = F_ALLOW_SHARING
1085 PPCODE: 1576 PPCODE:
1086{ 1577{
1087 CBOR cbor; 1578 CBOR cbor;
1088 cbor_init (&cbor); 1579 cbor_init (&cbor);
1580 cbor.flags |= ix;
1089 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1581 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1090 XPUSHs (scalar); 1582 XPUSHs (scalar);
1091} 1583}
1092 1584
1093void decode_cbor (SV *cborstr) 1585void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines