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 1.2 by root, Sat Oct 26 10:41:12 2013 UTC vs.
Revision 1.38 by root, Sun Dec 1 14:30:52 2013 UTC

9#include <limits.h> 9#include <limits.h>
10#include <float.h> 10#include <float.h>
11 11
12#include "ecb.h" 12#include "ecb.h"
13 13
14#if defined(__BORLANDC__) || defined(_MSC_VER) 14// compatibility with perl <5.18
15# define snprintf _snprintf // C compilers have this in stdio.h 15#ifndef HvNAMELEN_get
16# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
16#endif 17#endif
18#ifndef HvNAMELEN
19# define HvNAMELEN(hv) HvNAMELEN_get (hv)
20#endif
21#ifndef HvNAMEUTF8
22# define HvNAMEUTF8(hv) 0
23#endif
24#ifndef SvREFCNT_dec_NN
25# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)
26#endif
17 27
28// known major and minor types
29enum cbor_type
30{
31 MAJOR_SHIFT = 5,
32 MINOR_MASK = 0x1f,
33
34 MAJOR_POS_INT = 0 << MAJOR_SHIFT,
35 MAJOR_NEG_INT = 1 << MAJOR_SHIFT,
36 MAJOR_BYTES = 2 << MAJOR_SHIFT,
37 MAJOR_TEXT = 3 << MAJOR_SHIFT,
38 MAJOR_ARRAY = 4 << MAJOR_SHIFT,
39 MAJOR_MAP = 5 << MAJOR_SHIFT,
40 MAJOR_TAG = 6 << MAJOR_SHIFT,
41 MAJOR_MISC = 7 << MAJOR_SHIFT,
42
43 // INT/STRING/ARRAY/MAP subtypes
44 LENGTH_EXT1 = 24,
45 LENGTH_EXT2 = 25,
46 LENGTH_EXT4 = 26,
47 LENGTH_EXT8 = 27,
48
49 // SIMPLE types (effectively MISC subtypes)
50 SIMPLE_FALSE = 20,
51 SIMPLE_TRUE = 21,
52 SIMPLE_NULL = 22,
53 SIMPLE_UNDEF = 23,
54
55 // MISC subtype (unused)
56 MISC_EXT1 = 24,
57 MISC_FLOAT16 = 25,
58 MISC_FLOAT32 = 26,
59 MISC_FLOAT64 = 27,
60
61 // BYTES/TEXT/ARRAY/MAP
62 MINOR_INDEF = 31,
63};
64
65// known tags
66enum cbor_tag
67{
68 // extensions
69 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
70 CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object
71 CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object
72 CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing
73 CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing
74 CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref
75 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection
76
77 // rfc7049
78 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
79 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
80 CBOR_TAG_POS_BIGNUM = 2, // byte string
81 CBOR_TAG_NEG_BIGNUM = 3, // byte string
82 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
83 CBOR_TAG_BIGFLOAT = 5, // array
84
85 CBOR_TAG_CONV_B64U = 21, // base64url, any
86 CBOR_TAG_CONV_B64 = 22, // base64, any
87 CBOR_TAG_CONV_HEX = 23, // base16, any
88 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
89
90 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
91 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
92 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
93 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
94 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
95
96 CBOR_TAG_MAGIC = 55799, // self-describe cbor
97};
98
18#define F_SHRINK 0x00000200UL 99#define F_SHRINK 0x00000001UL
19#define F_ALLOW_UNKNOWN 0x00002000UL 100#define F_ALLOW_UNKNOWN 0x00000002UL
101#define F_ALLOW_SHARING 0x00000004UL
102#define F_ALLOW_CYCLES 0x00000008UL
103#define F_PACK_STRINGS 0x00000010UL
104#define F_VALIDATE_UTF8 0x00000020UL
20 105
21#define INIT_SIZE 32 // initial scalar size to be allocated 106#define INIT_SIZE 32 // initial scalar size to be allocated
22 107
23#define SB do { 108#define SB do {
24#define SE } while (0) 109#define SE } while (0)
25
26#if __GNUC__ >= 3
27# define expect(expr,value) __builtin_expect ((expr), (value))
28# define INLINE static inline
29#else
30# define expect(expr,value) (expr)
31# define INLINE static
32#endif
33
34#define expect_false(expr) expect ((expr) != 0, 0)
35#define expect_true(expr) expect ((expr) != 0, 1)
36 110
37#define IN_RANGE_INC(type,val,beg,end) \ 111#define IN_RANGE_INC(type,val,beg,end) \
38 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \ 112 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \
39 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg))) 113 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
40 114
46#else 120#else
47# define CBOR_SLOW 0 121# define CBOR_SLOW 0
48# define CBOR_STASH cbor_stash 122# define CBOR_STASH cbor_stash
49#endif 123#endif
50 124
51static HV *cbor_stash, *cbor_boolean_stash; // CBOR::XS:: 125static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
52static SV *cbor_true, *cbor_false; 126static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
53 127
54typedef struct { 128typedef struct {
55 U32 flags; 129 U32 flags;
56 U32 max_depth; 130 U32 max_depth;
57 STRLEN max_size; 131 STRLEN max_size;
58 132 SV *filter;
59 SV *cb_object;
60 HV *cb_sk_object;
61} CBOR; 133} CBOR;
62 134
63INLINE void 135ecb_inline void
64cbor_init (CBOR *cbor) 136cbor_init (CBOR *cbor)
65{ 137{
66 Zero (cbor, 1, CBOR); 138 Zero (cbor, 1, CBOR);
67 cbor->max_depth = 512; 139 cbor->max_depth = 512;
68} 140}
69 141
142ecb_inline void
143cbor_free (CBOR *cbor)
144{
145 SvREFCNT_dec (cbor->filter);
146}
147
70///////////////////////////////////////////////////////////////////////////// 148/////////////////////////////////////////////////////////////////////////////
71// utility functions 149// utility functions
72 150
73INLINE SV * 151ecb_inline SV *
74get_bool (const char *name) 152get_bool (const char *name)
75{ 153{
76 SV *sv = get_sv (name, 1); 154 SV *sv = get_sv (name, 1);
77 155
78 SvREADONLY_on (sv); 156 SvREADONLY_on (sv);
79 SvREADONLY_on (SvRV (sv)); 157 SvREADONLY_on (SvRV (sv));
80 158
81 return sv; 159 return sv;
82} 160}
83 161
84INLINE void 162ecb_inline void
85shrink (SV *sv) 163shrink (SV *sv)
86{ 164{
87 sv_utf8_downgrade (sv, 1); 165 sv_utf8_downgrade (sv, 1);
88 166
89 if (SvLEN (sv) > SvCUR (sv) + 1) 167 if (SvLEN (sv) > SvCUR (sv) + 1)
94 SvPV_renew (sv, SvCUR (sv) + 1); 172 SvPV_renew (sv, SvCUR (sv) + 1);
95#endif 173#endif
96 } 174 }
97} 175}
98 176
99///////////////////////////////////////////////////////////////////////////// 177// minimum length of a string to be registered for stringref
100// fp hell 178ecb_inline int
101 179minimum_string_length (UV idx)
102//TODO 180{
181 return idx > 23
182 ? idx > 0xffU
183 ? idx > 0xffffU
184 ? idx > 0xffffffffU
185 ? 11
186 : 7
187 : 5
188 : 4
189 : 3;
190}
103 191
104///////////////////////////////////////////////////////////////////////////// 192/////////////////////////////////////////////////////////////////////////////
105// encoder 193// encoder
106 194
107// structure used for encoding CBOR 195// structure used for encoding CBOR
110 char *cur; // SvPVX (sv) + current output position 198 char *cur; // SvPVX (sv) + current output position
111 char *end; // SvEND (sv) 199 char *end; // SvEND (sv)
112 SV *sv; // result scalar 200 SV *sv; // result scalar
113 CBOR cbor; 201 CBOR cbor;
114 U32 depth; // recursion level 202 U32 depth; // recursion level
203 HV *stringref[2]; // string => index, or 0 ([0] = bytes, [1] = utf-8)
204 UV stringref_idx;
205 HV *shareable; // ptr => index, or 0
206 UV shareable_idx;
115} enc_t; 207} enc_t;
116 208
117INLINE void 209ecb_inline void
118need (enc_t *enc, STRLEN len) 210need (enc_t *enc, STRLEN len)
119{ 211{
120 if (expect_false (enc->cur + len >= enc->end)) 212 if (ecb_expect_false (enc->cur + len >= enc->end))
121 { 213 {
122 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 214 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
123 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 215 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
124 enc->cur = SvPVX (enc->sv) + cur; 216 enc->cur = SvPVX (enc->sv) + cur;
125 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 217 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
126 } 218 }
127} 219}
128 220
129INLINE void 221ecb_inline void
130encode_ch (enc_t *enc, char ch) 222encode_ch (enc_t *enc, char ch)
131{ 223{
132 need (enc, 1); 224 need (enc, 1);
133 *enc->cur++ = ch; 225 *enc->cur++ = ch;
134} 226}
136static void 228static void
137encode_uint (enc_t *enc, int major, UV len) 229encode_uint (enc_t *enc, int major, UV len)
138{ 230{
139 need (enc, 9); 231 need (enc, 9);
140 232
141 if (len < 24) 233 if (ecb_expect_true (len < LENGTH_EXT1))
142 *enc->cur++ = major | len; 234 *enc->cur++ = major | len;
143 else if (len < 0x100) 235 else if (ecb_expect_true (len <= 0xffU))
144 { 236 {
145 *enc->cur++ = major | 24; 237 *enc->cur++ = major | LENGTH_EXT1;
146 *enc->cur++ = len; 238 *enc->cur++ = len;
147 } 239 }
148 else if (len < 0x10000) 240 else if (len <= 0xffffU)
149 { 241 {
150 *enc->cur++ = major | 25; 242 *enc->cur++ = major | LENGTH_EXT2;
151 *enc->cur++ = len >> 8; 243 *enc->cur++ = len >> 8;
152 *enc->cur++ = len; 244 *enc->cur++ = len;
153 } 245 }
154 else if (len < 0x100000000) 246 else if (len <= 0xffffffffU)
155 { 247 {
156 *enc->cur++ = major | 26; 248 *enc->cur++ = major | LENGTH_EXT4;
157 *enc->cur++ = len >> 24; 249 *enc->cur++ = len >> 24;
158 *enc->cur++ = len >> 16; 250 *enc->cur++ = len >> 16;
159 *enc->cur++ = len >> 8; 251 *enc->cur++ = len >> 8;
160 *enc->cur++ = len; 252 *enc->cur++ = len;
161 } 253 }
162 else if (len) 254 else
163 { 255 {
164 *enc->cur++ = major | 27; 256 *enc->cur++ = major | LENGTH_EXT8;
165 *enc->cur++ = len >> 56; 257 *enc->cur++ = len >> 56;
166 *enc->cur++ = len >> 48; 258 *enc->cur++ = len >> 48;
167 *enc->cur++ = len >> 40; 259 *enc->cur++ = len >> 40;
168 *enc->cur++ = len >> 32; 260 *enc->cur++ = len >> 32;
169 *enc->cur++ = len >> 24; 261 *enc->cur++ = len >> 24;
171 *enc->cur++ = len >> 8; 263 *enc->cur++ = len >> 8;
172 *enc->cur++ = len; 264 *enc->cur++ = len;
173 } 265 }
174} 266}
175 267
176static void 268ecb_inline void
269encode_tag (enc_t *enc, UV tag)
270{
271 encode_uint (enc, MAJOR_TAG, tag);
272}
273
274ecb_inline void
177encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 275encode_str (enc_t *enc, int utf8, char *str, STRLEN len)
178{ 276{
179 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 277 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
180 need (enc, len); 278 need (enc, len);
181 memcpy (enc->cur, str, len); 279 memcpy (enc->cur, str, len);
182 enc->cur += len; 280 enc->cur += len;
183} 281}
184 282
283static void
284encode_strref (enc_t *enc, int utf8, char *str, STRLEN len)
285{
286 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
287 {
288 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
289
290 if (SvOK (*svp))
291 {
292 // already registered, use stringref
293 encode_tag (enc, CBOR_TAG_STRINGREF);
294 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
295 return;
296 }
297 else if (len >= minimum_string_length (enc->stringref_idx))
298 {
299 // register only
300 sv_setuv (*svp, enc->stringref_idx);
301 ++enc->stringref_idx;
302 }
303 }
304
305 encode_str (enc, utf8, str, len);
306}
307
185static void encode_sv (enc_t *enc, SV *sv); 308static void encode_sv (enc_t *enc, SV *sv);
186 309
187static void 310static void
188encode_av (enc_t *enc, AV *av) 311encode_av (enc_t *enc, AV *av)
189{ 312{
192 if (enc->depth >= enc->cbor.max_depth) 315 if (enc->depth >= enc->cbor.max_depth)
193 croak (ERR_NESTING_EXCEEDED); 316 croak (ERR_NESTING_EXCEEDED);
194 317
195 ++enc->depth; 318 ++enc->depth;
196 319
197 encode_uint (enc, 0x80, len + 1); 320 encode_uint (enc, MAJOR_ARRAY, len + 1);
198 321
199 for (i = 0; i <= len; ++i) 322 for (i = 0; i <= len; ++i)
200 { 323 {
201 SV **svp = av_fetch (av, i, 0); 324 SV **svp = av_fetch (av, i, 0);
202 encode_sv (enc, svp ? *svp : &PL_sv_undef); 325 encode_sv (enc, svp ? *svp : &PL_sv_undef);
217 340
218 int pairs = hv_iterinit (hv); 341 int pairs = hv_iterinit (hv);
219 int mg = SvMAGICAL (hv); 342 int mg = SvMAGICAL (hv);
220 343
221 if (mg) 344 if (mg)
222 encode_ch (enc, 0xa0 | 31); 345 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
223 else 346 else
224 encode_uint (enc, 0xa0, pairs); 347 encode_uint (enc, MAJOR_MAP, pairs);
225 348
226 while ((he = hv_iternext (hv))) 349 while ((he = hv_iternext (hv)))
227 { 350 {
228 if (HeKLEN (he) == HEf_SVKEY) 351 if (HeKLEN (he) == HEf_SVKEY)
229 encode_sv (enc, HeSVKEY (he)); 352 encode_sv (enc, HeSVKEY (he));
230 else 353 else
231 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 354 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
232 355
233 encode_sv (enc, expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 356 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
234 } 357 }
235 358
236 if (mg) 359 if (mg)
237 encode_ch (enc, 0xe0 | 31); 360 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
238 361
239 --enc->depth; 362 --enc->depth;
240} 363}
241 364
242// encode objects, arrays and special \0=false and \1=true values. 365// encode objects, arrays and special \0=false and \1=true values.
243static void 366static void
244encode_rv (enc_t *enc, SV *sv) 367encode_rv (enc_t *enc, SV *sv)
245{ 368{
246 svtype svt;
247
248 SvGETMAGIC (sv); 369 SvGETMAGIC (sv);
370
249 svt = SvTYPE (sv); 371 svtype svt = SvTYPE (sv);
250 372
251 if (expect_false (SvOBJECT (sv))) 373 if (ecb_expect_false (SvOBJECT (sv)))
252 { 374 {
253 HV *stash = !CBOR_SLOW || cbor_boolean_stash 375 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
254 ? cbor_boolean_stash 376 ? types_boolean_stash
377 : gv_stashpv ("Types::Serialiser::Boolean", 1);
378 HV *error_stash = !CBOR_SLOW || types_error_stash
379 ? types_error_stash
380 : gv_stashpv ("Types::Serialiser::Error", 1);
381 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
382 ? cbor_tagged_stash
255 : gv_stashpv ("CBOR::XS::Boolean", 1); 383 : gv_stashpv ("CBOR::XS::Tagged" , 1);
256 384
257 if (SvSTASH (sv) == stash) 385 HV *stash = SvSTASH (sv);
258 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 386
387 if (stash == boolean_stash)
388 {
389 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
390 return;
391 }
392 else if (stash == error_stash)
393 {
394 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
395 return;
396 }
397 else if (stash == tagged_stash)
398 {
399 if (svt != SVt_PVAV)
400 croak ("encountered CBOR::XS::Tagged object that isn't an array");
401
402 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
403 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
404
405 return;
406 }
407 }
408
409 if (ecb_expect_false (SvREFCNT (sv) > 1)
410 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
411 {
412 if (!enc->shareable)
413 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
414
415 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
416
417 if (SvOK (*svp))
418 {
419 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
420 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
421 return;
422 }
259 else 423 else
260 { 424 {
261#if 0 //TODO 425 sv_setuv (*svp, enc->shareable_idx);
262 if (enc->cbor.flags & F_CONV_BLESSED) 426 ++enc->shareable_idx;
427 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
428 }
429 }
430
431 if (ecb_expect_false (SvOBJECT (sv)))
432 {
433 HV *stash = SvSTASH (sv);
434 GV *method;
435
436 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
263 { 437 {
438 dSP;
439
440 ENTER; SAVETMPS; PUSHMARK (SP);
264 // we re-bless the reference to get overload and other niceties right 441 // we re-bless the reference to get overload and other niceties right
265 GV *to_cbor = gv_fetchmethod_autoload (SvSTASH (sv), "TO_CBOR", 0);
266
267 if (to_cbor)
268 {
269 dSP;
270
271 ENTER; SAVETMPS; PUSHMARK (SP);
272 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv))); 442 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
273 443
274 // calling with G_SCALAR ensures that we always get a 1 return value
275 PUTBACK; 444 PUTBACK;
445 // G_SCALAR ensures that return value is 1
276 call_sv ((SV *)GvCV (to_cbor), G_SCALAR); 446 call_sv ((SV *)GvCV (method), G_SCALAR);
277 SPAGAIN; 447 SPAGAIN;
278 448
279 // catch this surprisingly common error 449 // catch this surprisingly common error
280 if (SvROK (TOPs) && SvRV (TOPs) == sv) 450 if (SvROK (TOPs) && SvRV (TOPs) == sv)
281 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv))); 451 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (stash));
282 452
283 sv = POPs;
284 PUTBACK;
285
286 encode_sv (enc, sv); 453 encode_sv (enc, POPs);
287 454
455 PUTBACK;
456
288 FREETMPS; LEAVE; 457 FREETMPS; LEAVE;
289 }
290 else if (enc->cbor.flags & F_ALLOW_BLESSED)
291 encode_str (enc, "null", 4, 0);
292 else
293 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_CBOR method available on it",
294 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
295 } 458 }
296 else if (enc->cbor.flags & F_ALLOW_BLESSED) 459 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
297 encode_str (enc, "null", 4, 0); 460 {
461 dSP;
462
463 ENTER; SAVETMPS; PUSHMARK (SP);
464 EXTEND (SP, 2);
465 // we re-bless the reference to get overload and other niceties right
466 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
467 PUSHs (sv_cbor);
468
469 PUTBACK;
470 int count = call_sv ((SV *)GvCV (method), G_ARRAY);
471 SPAGAIN;
472
473 // catch this surprisingly common error
474 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
475 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
476
477 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
478 encode_uint (enc, MAJOR_ARRAY, count + 1);
479 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
480
481 while (count)
482 encode_sv (enc, SP[1 - count--]);
483
484 PUTBACK;
485
486 FREETMPS; LEAVE;
487 }
298 else 488 else
299 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled", 489 croak ("encountered object '%s', but no TO_CBOR or FREEZE methods available on it",
300 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 490 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
301#endif
302 }
303 } 491 }
304 else if (svt == SVt_PVHV) 492 else if (svt == SVt_PVHV)
305 encode_hv (enc, (HV *)sv); 493 encode_hv (enc, (HV *)sv);
306 else if (svt == SVt_PVAV) 494 else if (svt == SVt_PVAV)
307 encode_av (enc, (AV *)sv); 495 encode_av (enc, (AV *)sv);
308 else if (svt < SVt_PVAV)
309 {
310 STRLEN len = 0;
311 char *pv = svt ? SvPV (sv, len) : 0;
312
313 if (len == 1 && *pv == '1')
314 encode_ch (enc, 0xe0 | 21);
315 else if (len == 1 && *pv == '0')
316 encode_ch (enc, 0xe0 | 20);
317 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
318 encode_ch (enc, 0xe0 | 23);
319 else
320 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
321 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
322 }
323 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
324 encode_ch (enc, 0xe0 | 23);
325 else 496 else
326 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 497 {
327 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 498 encode_tag (enc, CBOR_TAG_INDIRECTION);
499 encode_sv (enc, sv);
500 }
328} 501}
329 502
330static void 503static void
331encode_nv (enc_t *enc, SV *sv) 504encode_nv (enc_t *enc, SV *sv)
332{ 505{
333 double nv = SvNVX (sv); 506 double nv = SvNVX (sv);
334 507
335 need (enc, 9); 508 need (enc, 9);
336 509
337 if (expect_false (nv == (U32)nv)) 510 if (ecb_expect_false (nv == (NV)(U32)nv))
338 encode_uint (enc, 0x00, (U32)nv); 511 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
339 //TODO: maybe I32? 512 //TODO: maybe I32?
340 else if (expect_false (nv == (float)nv)) 513 else if (ecb_expect_false (nv == (float)nv))
341 { 514 {
342 uint32_t fp = ecb_float_to_binary32 (nv); 515 uint32_t fp = ecb_float_to_binary32 (nv);
343 516
344 *enc->cur++ = 0xe0 | 26; 517 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
345 518
346 if (!ecb_big_endian ()) 519 if (!ecb_big_endian ())
347 fp = ecb_bswap32 (fp); 520 fp = ecb_bswap32 (fp);
348 521
349 memcpy (enc->cur, &fp, 4); 522 memcpy (enc->cur, &fp, 4);
351 } 524 }
352 else 525 else
353 { 526 {
354 uint64_t fp = ecb_double_to_binary64 (nv); 527 uint64_t fp = ecb_double_to_binary64 (nv);
355 528
356 *enc->cur++ = 0xe0 | 27; 529 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
357 530
358 if (!ecb_big_endian ()) 531 if (!ecb_big_endian ())
359 fp = ecb_bswap64 (fp); 532 fp = ecb_bswap64 (fp);
360 533
361 memcpy (enc->cur, &fp, 8); 534 memcpy (enc->cur, &fp, 8);
370 543
371 if (SvPOKp (sv)) 544 if (SvPOKp (sv))
372 { 545 {
373 STRLEN len; 546 STRLEN len;
374 char *str = SvPV (sv, len); 547 char *str = SvPV (sv, len);
375 encode_str (enc, SvUTF8 (sv), str, len); 548 encode_strref (enc, SvUTF8 (sv), str, len);
376 } 549 }
377 else if (SvNOKp (sv)) 550 else if (SvNOKp (sv))
378 encode_nv (enc, sv); 551 encode_nv (enc, sv);
379 else if (SvIOKp (sv)) 552 else if (SvIOKp (sv))
380 { 553 {
381 if (SvIsUV (sv)) 554 if (SvIsUV (sv))
382 encode_uint (enc, 0x00, SvUVX (sv)); 555 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
383 else if (SvIVX (sv) >= 0) 556 else if (SvIVX (sv) >= 0)
384 encode_uint (enc, 0x00, SvIVX (sv)); 557 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
385 else 558 else
386 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 559 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
387 } 560 }
388 else if (SvROK (sv)) 561 else if (SvROK (sv))
389 encode_rv (enc, SvRV (sv)); 562 encode_rv (enc, SvRV (sv));
390 else if (!SvOK (sv)) 563 else if (!SvOK (sv))
391 encode_ch (enc, 0xe0 | 22); 564 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
392 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 565 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
393 encode_ch (enc, 0xe0 | 23); 566 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
394 else 567 else
395 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 568 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
396 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 569 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
397} 570}
398 571
399static SV * 572static SV *
400encode_cbor (SV *scalar, CBOR *cbor) 573encode_cbor (SV *scalar, CBOR *cbor)
401{ 574{
402 enc_t enc; 575 enc_t enc = { };
403 576
404 enc.cbor = *cbor; 577 enc.cbor = *cbor;
405 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 578 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
406 enc.cur = SvPVX (enc.sv); 579 enc.cur = SvPVX (enc.sv);
407 enc.end = SvEND (enc.sv); 580 enc.end = SvEND (enc.sv);
408 enc.depth = 0;
409 581
410 SvPOK_only (enc.sv); 582 SvPOK_only (enc.sv);
583
584 if (cbor->flags & F_PACK_STRINGS)
585 {
586 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
587 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
588 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
589 }
590
411 encode_sv (&enc, scalar); 591 encode_sv (&enc, scalar);
412 592
413 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 593 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
414 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 594 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
415 595
429 U8 *end; // end of input string 609 U8 *end; // end of input string
430 const char *err; // parse error, if != 0 610 const char *err; // parse error, if != 0
431 CBOR cbor; 611 CBOR cbor;
432 U32 depth; // recursion depth 612 U32 depth; // recursion depth
433 U32 maxdepth; // recursion depth limit 613 U32 maxdepth; // recursion depth limit
614 AV *shareable;
615 AV *stringref;
616 SV *decode_tagged;
434} dec_t; 617} dec_t;
435 618
436#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 619#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE
437 620
438#define WANT(len) if (expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data"); 621#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data")
439 622
440#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED) 623#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED)
441#define DEC_DEC_DEPTH --dec->depth 624#define DEC_DEC_DEPTH --dec->depth
442 625
443static UV 626static UV
444decode_uint (dec_t *dec) 627decode_uint (dec_t *dec)
445{ 628{
446 switch (*dec->cur & 31) 629 U8 m = *dec->cur & MINOR_MASK;
447 { 630 ++dec->cur;
448 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
449 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
450 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
451 return *dec->cur++ & 31;
452 631
453 case 24: 632 if (ecb_expect_true (m < LENGTH_EXT1))
633 return m;
634 else if (ecb_expect_true (m == LENGTH_EXT1))
635 {
454 WANT (2); 636 WANT (1);
455 dec->cur += 2; 637 dec->cur += 1;
456 return dec->cur[-1]; 638 return dec->cur[-1];
457 639 }
458 case 25: 640 else if (ecb_expect_true (m == LENGTH_EXT2))
641 {
459 WANT (3); 642 WANT (2);
460 dec->cur += 3; 643 dec->cur += 2;
461 return (((UV)dec->cur[-2]) << 8) 644 return (((UV)dec->cur[-2]) << 8)
462 | ((UV)dec->cur[-1]); 645 | ((UV)dec->cur[-1]);
463 646 }
464 case 26: 647 else if (ecb_expect_true (m == LENGTH_EXT4))
648 {
465 WANT (5); 649 WANT (4);
466 dec->cur += 5; 650 dec->cur += 4;
467 return (((UV)dec->cur[-4]) << 24) 651 return (((UV)dec->cur[-4]) << 24)
468 | (((UV)dec->cur[-3]) << 16) 652 | (((UV)dec->cur[-3]) << 16)
469 | (((UV)dec->cur[-2]) << 8) 653 | (((UV)dec->cur[-2]) << 8)
470 | ((UV)dec->cur[-1]); 654 | ((UV)dec->cur[-1]);
471 655 }
472 case 27: 656 else if (ecb_expect_true (m == LENGTH_EXT8))
657 {
473 WANT (9); 658 WANT (8);
474 dec->cur += 9; 659 dec->cur += 8;
660
661 return
662#if UVSIZE < 8
663 0
664#else
475 return (((UV)dec->cur[-8]) << 56) 665 (((UV)dec->cur[-8]) << 56)
476 | (((UV)dec->cur[-7]) << 48) 666 | (((UV)dec->cur[-7]) << 48)
477 | (((UV)dec->cur[-6]) << 40) 667 | (((UV)dec->cur[-6]) << 40)
478 | (((UV)dec->cur[-5]) << 32) 668 | (((UV)dec->cur[-5]) << 32)
669#endif
479 | (((UV)dec->cur[-4]) << 24) 670 | (((UV)dec->cur[-4]) << 24)
480 | (((UV)dec->cur[-3]) << 16) 671 | (((UV)dec->cur[-3]) << 16)
481 | (((UV)dec->cur[-2]) << 8) 672 | (((UV)dec->cur[-2]) << 8)
482 | ((UV)dec->cur[-1]); 673 | ((UV)dec->cur[-1]);
483 674 }
484 default: 675 else
485 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 676 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
486 }
487 677
488fail: 678fail:
489 return 0; 679 return 0;
490} 680}
491 681
496{ 686{
497 AV *av = newAV (); 687 AV *av = newAV ();
498 688
499 DEC_INC_DEPTH; 689 DEC_INC_DEPTH;
500 690
501 if ((*dec->cur & 31) == 31) 691 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
502 { 692 {
503 ++dec->cur; 693 ++dec->cur;
504 694
505 for (;;) 695 for (;;)
506 { 696 {
507 WANT (1); 697 WANT (1);
508 698
509 if (*dec->cur == (0xe0 | 31)) 699 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
510 { 700 {
511 ++dec->cur; 701 ++dec->cur;
512 break; 702 break;
513 } 703 }
514 704
517 } 707 }
518 else 708 else
519 { 709 {
520 int i, len = decode_uint (dec); 710 int i, len = decode_uint (dec);
521 711
712 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
522 av_fill (av, len - 1); 713 av_fill (av, len - 1);
523 714
524 for (i = 0; i < len; ++i) 715 for (i = 0; i < len; ++i)
525 AvARRAY (av)[i] = decode_sv (dec); 716 AvARRAY (av)[i] = decode_sv (dec);
526 } 717 }
532 SvREFCNT_dec (av); 723 SvREFCNT_dec (av);
533 DEC_DEC_DEPTH; 724 DEC_DEC_DEPTH;
534 return &PL_sv_undef; 725 return &PL_sv_undef;
535} 726}
536 727
728static void
729decode_he (dec_t *dec, HV *hv)
730{
731 // for speed reasons, we specialcase single-string
732 // byte or utf-8 strings as keys, but only when !stringref
733
734 if (ecb_expect_true (!dec->stringref))
735 if (ecb_expect_true ((*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
736 {
737 I32 len = decode_uint (dec);
738 char *key = (char *)dec->cur;
739
740 dec->cur += len;
741
742 hv_store (hv, key, len, decode_sv (dec), 0);
743
744 return;
745 }
746 else if (ecb_expect_true ((*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
747 {
748 I32 len = decode_uint (dec);
749 char *key = (char *)dec->cur;
750
751 dec->cur += len;
752
753 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
754 if (!is_utf8_string (key, len))
755 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
756
757 hv_store (hv, key, -len, decode_sv (dec), 0);
758
759 return;
760 }
761
762 SV *k = decode_sv (dec);
763 SV *v = decode_sv (dec);
764
765 hv_store_ent (hv, k, v, 0);
766 SvREFCNT_dec (k);
767
768fail:
769 ;
770}
771
537static SV * 772static SV *
538decode_hv (dec_t *dec) 773decode_hv (dec_t *dec)
539{ 774{
540 HV *hv = newHV (); 775 HV *hv = newHV ();
541 776
542 DEC_INC_DEPTH; 777 DEC_INC_DEPTH;
543 778
544 if ((*dec->cur & 31) == 31) 779 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
545 { 780 {
546 ++dec->cur; 781 ++dec->cur;
547 782
548 for (;;) 783 for (;;)
549 { 784 {
550 WANT (1); 785 WANT (1);
551 786
552 if (*dec->cur == (0xe0 | 31)) 787 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
553 { 788 {
554 ++dec->cur; 789 ++dec->cur;
555 break; 790 break;
556 } 791 }
557 792
558 SV *k = decode_sv (dec); 793 decode_he (dec, hv);
559 SV *v = decode_sv (dec);
560
561 hv_store_ent (hv, k, v, 0);
562 } 794 }
563 } 795 }
564 else 796 else
565 { 797 {
566 int len = decode_uint (dec); 798 int pairs = decode_uint (dec);
567 799
568 while (len--) 800 while (pairs--)
569 { 801 decode_he (dec, hv);
570 SV *k = decode_sv (dec);
571 SV *v = decode_sv (dec);
572
573 hv_store_ent (hv, k, v, 0);
574 }
575 } 802 }
576 803
577 DEC_DEC_DEPTH; 804 DEC_DEC_DEPTH;
578 return newRV_noinc ((SV *)hv); 805 return newRV_noinc ((SV *)hv);
579
580#if 0
581 SV *sv;
582 HV *hv = newHV ();
583
584 DEC_INC_DEPTH;
585 decode_ws (dec);
586
587 for (;;)
588 {
589 // heuristic: assume that
590 // a) decode_str + hv_store_ent are abysmally slow.
591 // b) most hash keys are short, simple ascii text.
592 // => try to "fast-match" such strings to avoid
593 // the overhead of decode_str + hv_store_ent.
594 {
595 SV *value;
596 char *p = dec->cur;
597 char *e = p + 24; // only try up to 24 bytes
598
599 for (;;)
600 {
601 // the >= 0x80 is false on most architectures
602 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
603 {
604 // slow path, back up and use decode_str
605 SV *key = decode_str (dec);
606 if (!key)
607 goto fail;
608
609 decode_ws (dec); EXPECT_CH (':');
610
611 decode_ws (dec);
612 value = decode_sv (dec);
613 if (!value)
614 {
615 SvREFCNT_dec (key);
616 goto fail;
617 }
618
619 hv_store_ent (hv, key, value, 0);
620 SvREFCNT_dec (key);
621
622 break;
623 }
624 else if (*p == '"')
625 {
626 // fast path, got a simple key
627 char *key = dec->cur;
628 int len = p - key;
629 dec->cur = p + 1;
630
631 decode_ws (dec); EXPECT_CH (':');
632
633 decode_ws (dec);
634 value = decode_sv (dec);
635 if (!value)
636 goto fail;
637
638 hv_store (hv, key, len, value, 0);
639
640 break;
641 }
642
643 ++p;
644 }
645 }
646
647 decode_ws (dec);
648
649 if (*dec->cur == '}')
650 {
651 ++dec->cur;
652 break;
653 }
654
655 if (*dec->cur != ',')
656 ERR (", or } expected while parsing object/hash");
657
658 ++dec->cur;
659
660 decode_ws (dec);
661
662 if (*dec->cur == '}' && dec->cbor.flags & F_RELAXED)
663 {
664 ++dec->cur;
665 break;
666 }
667 }
668
669 DEC_DEC_DEPTH;
670 sv = newRV_noinc ((SV *)hv);
671
672 // check filter callbacks
673 if (dec->cbor.flags & F_HOOK)
674 {
675 if (dec->cbor.cb_sk_object && HvKEYS (hv) == 1)
676 {
677 HE *cb, *he;
678
679 hv_iterinit (hv);
680 he = hv_iternext (hv);
681 hv_iterinit (hv);
682
683 // the next line creates a mortal sv each time its called.
684 // might want to optimise this for common cases.
685 cb = hv_fetch_ent (dec->cbor.cb_sk_object, hv_iterkeysv (he), 0, 0);
686
687 if (cb)
688 {
689 dSP;
690 int count;
691
692 ENTER; SAVETMPS; PUSHMARK (SP);
693 XPUSHs (HeVAL (he));
694 sv_2mortal (sv);
695
696 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
697
698 if (count == 1)
699 {
700 sv = newSVsv (POPs);
701 FREETMPS; LEAVE;
702 return sv;
703 }
704
705 SvREFCNT_inc (sv);
706 FREETMPS; LEAVE;
707 }
708 }
709
710 if (dec->cbor.cb_object)
711 {
712 dSP;
713 int count;
714
715 ENTER; SAVETMPS; PUSHMARK (SP);
716 XPUSHs (sv_2mortal (sv));
717
718 PUTBACK; count = call_sv (dec->cbor.cb_object, G_ARRAY); SPAGAIN;
719
720 if (count == 1)
721 {
722 sv = newSVsv (POPs);
723 FREETMPS; LEAVE;
724 return sv;
725 }
726
727 SvREFCNT_inc (sv);
728 FREETMPS; LEAVE;
729 }
730 }
731
732 return sv;
733#endif
734 806
735fail: 807fail:
736 SvREFCNT_dec (hv); 808 SvREFCNT_dec (hv);
737 DEC_DEC_DEPTH; 809 DEC_DEC_DEPTH;
738 return &PL_sv_undef; 810 return &PL_sv_undef;
739} 811}
740 812
741static SV * 813static SV *
742decode_str (dec_t *dec, int utf8) 814decode_str (dec_t *dec, int utf8)
743{ 815{
744 SV *sv; 816 SV *sv = 0;
745 817
746 if ((*dec->cur & 31) == 31) 818 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
747 { 819 {
820 // indefinite length strings
748 ++dec->cur; 821 ++dec->cur;
749 822
823 U8 major = *dec->cur & MAJOR_MISC;
824
750 sv = newSVpvn ("", 0); 825 sv = newSVpvn ("", 0);
751 826
752 // not very fast, and certainly not robust against illegal input
753 for (;;) 827 for (;;)
754 { 828 {
755 WANT (1); 829 WANT (1);
756 830
757 if (*dec->cur == (0xe0 | 31)) 831 if ((*dec->cur - major) > LENGTH_EXT8)
832 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
758 { 833 {
759 ++dec->cur; 834 ++dec->cur;
760 break; 835 break;
761 } 836 }
837 else
838 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
762 839
763 SV *sv2 = decode_sv (dec); 840 STRLEN len = decode_uint (dec);
764 sv_catsv (sv, sv2); 841
842 WANT (len);
843 sv_catpvn (sv, dec->cur, len);
844 dec->cur += len;
765 } 845 }
766 } 846 }
767 else 847 else
768 { 848 {
769 STRLEN len = decode_uint (dec); 849 STRLEN len = decode_uint (dec);
770 850
771 WANT (len); 851 WANT (len);
772 sv = newSVpvn (dec->cur, len); 852 sv = newSVpvn (dec->cur, len);
773 dec->cur += len; 853 dec->cur += len;
854
855 if (ecb_expect_false (dec->stringref)
856 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
857 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
774 } 858 }
775 859
776 if (utf8) 860 if (utf8)
861 {
862 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
863 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
864 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
865
777 SvUTF8_on (sv); 866 SvUTF8_on (sv);
867 }
778 868
779 return sv; 869 return sv;
780 870
781fail: 871fail:
872 SvREFCNT_dec (sv);
873 return &PL_sv_undef;
874}
875
876static SV *
877decode_tagged (dec_t *dec)
878{
879 SV *sv = 0;
880 UV tag = decode_uint (dec);
881
882 WANT (1);
883
884 switch (tag)
885 {
886 case CBOR_TAG_MAGIC:
887 sv = decode_sv (dec);
888 break;
889
890 case CBOR_TAG_INDIRECTION:
891 sv = newRV_noinc (decode_sv (dec));
892 break;
893
894 case CBOR_TAG_STRINGREF_NAMESPACE:
895 {
896 ENTER; SAVETMPS;
897
898 SAVESPTR (dec->stringref);
899 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
900
901 sv = decode_sv (dec);
902
903 FREETMPS; LEAVE;
904 }
905 break;
906
907 case CBOR_TAG_STRINGREF:
908 {
909 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
910 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
911
912 UV idx = decode_uint (dec);
913
914 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
915 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
916
917 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
918 }
919 break;
920
921 case CBOR_TAG_VALUE_SHAREABLE:
922 {
923 if (ecb_expect_false (!dec->shareable))
924 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
925
926 if (dec->cbor.flags & F_ALLOW_CYCLES)
927 {
928 sv = newSV (0);
929 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
930
931 SV *osv = decode_sv (dec);
932 sv_setsv (sv, osv);
933 SvREFCNT_dec_NN (osv);
934 }
935 else
936 {
937 av_push (dec->shareable, &PL_sv_undef);
938 int idx = AvFILLp (dec->shareable);
939 sv = decode_sv (dec);
940 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
941 }
942 }
943 break;
944
945 case CBOR_TAG_VALUE_SHAREDREF:
946 {
947 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
948 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
949
950 UV idx = decode_uint (dec);
951
952 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
953 ERR ("corrupted CBOR data (sharedref index out of bounds)");
954
955 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
956
957 if (sv == &PL_sv_undef)
958 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
959 }
960 break;
961
962 case CBOR_TAG_PERL_OBJECT:
963 {
964 sv = decode_sv (dec);
965
966 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
967 ERR ("corrupted CBOR data (non-array perl object)");
968
969 AV *av = (AV *)SvRV (sv);
970 int len = av_len (av) + 1;
971 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
972
973 if (!stash)
974 ERR ("cannot decode perl-object (package does not exist)");
975
976 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
977
978 if (!method)
979 ERR ("cannot decode perl-object (package does not have a THAW method)");
980
981 dSP;
982
983 ENTER; SAVETMPS; PUSHMARK (SP);
984 EXTEND (SP, len + 1);
985 // we re-bless the reference to get overload and other niceties right
986 PUSHs (*av_fetch (av, 0, 1));
987 PUSHs (sv_cbor);
988
989 int i;
990
991 for (i = 1; i < len; ++i)
992 PUSHs (*av_fetch (av, i, 1));
993
994 PUTBACK;
995 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
996 SPAGAIN;
997
998 if (SvTRUE (ERRSV))
999 {
1000 FREETMPS; LEAVE;
1001 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1002 }
1003
1004 SvREFCNT_dec (sv);
1005 sv = SvREFCNT_inc (POPs);
1006
1007 PUTBACK;
1008
1009 FREETMPS; LEAVE;
1010 }
1011 break;
1012
1013 default:
1014 {
1015 sv = decode_sv (dec);
1016
1017 dSP;
1018 ENTER; SAVETMPS; PUSHMARK (SP);
1019 EXTEND (SP, 2);
1020 PUSHs (newSVuv (tag));
1021 PUSHs (sv);
1022
1023 PUTBACK;
1024 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1025 SPAGAIN;
1026
1027 if (SvTRUE (ERRSV))
1028 {
1029 FREETMPS; LEAVE;
1030 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1031 }
1032
1033 if (count)
1034 {
1035 SvREFCNT_dec (sv);
1036 sv = SvREFCNT_inc (POPs);
1037 }
1038 else
1039 {
1040 AV *av = newAV ();
1041 av_push (av, newSVuv (tag));
1042 av_push (av, sv);
1043
1044 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
1045 ? cbor_tagged_stash
1046 : gv_stashpv ("CBOR::XS::Tagged" , 1);
1047 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
1048 }
1049
1050 PUTBACK;
1051
1052 FREETMPS; LEAVE;
1053 }
1054 break;
1055 }
1056
1057 return sv;
1058
1059fail:
1060 SvREFCNT_dec (sv);
782 return &PL_sv_undef; 1061 return &PL_sv_undef;
783} 1062}
784 1063
785static SV * 1064static SV *
786decode_sv (dec_t *dec) 1065decode_sv (dec_t *dec)
787{ 1066{
788 WANT (1); 1067 WANT (1);
789 1068
790 switch (*dec->cur >> 5) 1069 switch (*dec->cur >> MAJOR_SHIFT)
791 { 1070 {
792 case 0: // unsigned int 1071 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
793 //TODO: 64 bit values on 3 2bit perls 1072 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
794 return newSVuv (decode_uint (dec)); 1073 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
795 case 1: // negative int 1074 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
796 return newSViv (-1 - (IV)decode_uint (dec)); 1075 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
797 case 2: // octet string 1076 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
798 return decode_str (dec, 0); 1077 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
799 case 3: // utf-8 string 1078
800 return decode_str (dec, 1); 1079 case MAJOR_MISC >> MAJOR_SHIFT:
801 case 4: // array
802 return decode_av (dec);
803 case 5: // map
804 return decode_hv (dec);
805 case 6: // tag
806 abort ();
807 break;
808 case 7: // misc
809 switch (*dec->cur++ & 31) 1080 switch (*dec->cur++ & MINOR_MASK)
810 { 1081 {
811 case 20: 1082 case SIMPLE_FALSE:
812#if CBOR_SLOW 1083#if CBOR_SLOW
813 cbor_false = get_bool ("CBOR::XS::false"); 1084 types_false = get_bool ("Types::Serialiser::false");
814#endif 1085#endif
815 return newSVsv (cbor_false); 1086 return newSVsv (types_false);
816 case 21: 1087 case SIMPLE_TRUE:
817#if CBOR_SLOW 1088#if CBOR_SLOW
818 cbor_true = get_bool ("CBOR::XS::true"); 1089 types_true = get_bool ("Types::Serialiser::true");
819#endif 1090#endif
820 return newSVsv (cbor_true); 1091 return newSVsv (types_true);
821 case 22: 1092 case SIMPLE_NULL:
822 return newSVsv (&PL_sv_undef); 1093 return newSVsv (&PL_sv_undef);
1094 case SIMPLE_UNDEF:
1095#if CBOR_SLOW
1096 types_error = get_bool ("Types::Serialiser::error");
1097#endif
1098 return newSVsv (types_error);
823 1099
824 case 25: 1100 case MISC_FLOAT16:
825 { 1101 {
826 WANT (2); 1102 WANT (2);
827 1103
828 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1104 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
829 dec->cur += 2; 1105 dec->cur += 2;
830 1106
831 return newSVnv (ecb_binary16_to_float (fp)); 1107 return newSVnv (ecb_binary16_to_float (fp));
832 } 1108 }
833 1109
834 case 26: 1110 case MISC_FLOAT32:
835 { 1111 {
836 uint32_t fp; 1112 uint32_t fp;
837 WANT (4); 1113 WANT (4);
838 memcpy (&fp, dec->cur, 4); 1114 memcpy (&fp, dec->cur, 4);
839 dec->cur += 4; 1115 dec->cur += 4;
842 fp = ecb_bswap32 (fp); 1118 fp = ecb_bswap32 (fp);
843 1119
844 return newSVnv (ecb_binary32_to_float (fp)); 1120 return newSVnv (ecb_binary32_to_float (fp));
845 } 1121 }
846 1122
847 case 27: 1123 case MISC_FLOAT64:
848 { 1124 {
849 uint64_t fp; 1125 uint64_t fp;
850 WANT (8); 1126 WANT (8);
851 memcpy (&fp, dec->cur, 8); 1127 memcpy (&fp, dec->cur, 8);
852 dec->cur += 8; 1128 dec->cur += 8;
855 fp = ecb_bswap64 (fp); 1131 fp = ecb_bswap64 (fp);
856 1132
857 return newSVnv (ecb_binary64_to_double (fp)); 1133 return newSVnv (ecb_binary64_to_double (fp));
858 } 1134 }
859 1135
860 // 0..19 unassigned 1136 // 0..19 unassigned simple
861 // 24 reserved + unassigned (reserved values are not encodable) 1137 // 24 reserved + unassigned (reserved values are not encodable)
862 default: 1138 default:
863 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1139 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)");
864 } 1140 }
865 1141
866 break; 1142 break;
867 } 1143 }
868#if 0
869 switch (*dec->cur)
870 {
871 //case '"': ++dec->cur; return decode_str (dec);
872 case '[': ++dec->cur; return decode_av (dec);
873 case '{': ++dec->cur; return decode_hv (dec);
874
875 case '-':
876 case '0': case '1': case '2': case '3': case '4':
877 case '5': case '6': case '7': case '8': case '9':
878 //TODO return decode_num (dec);
879
880 case 't':
881 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
882 {
883 dec->cur += 4;
884#if CBOR_SLOW
885 cbor_true = get_bool ("CBOR::XS::true");
886#endif
887 return newSVsv (cbor_true);
888 }
889 else
890 ERR ("'true' expected");
891
892 break;
893
894 case 'f':
895 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
896 {
897 dec->cur += 5;
898#if CBOR_SLOW
899 cbor_false = get_bool ("CBOR::XS::false");
900#endif
901 return newSVsv (cbor_false);
902 }
903 else
904 ERR ("'false' expected");
905
906 break;
907
908 case 'n':
909 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4))
910 {
911 dec->cur += 4;
912 return newSVsv (&PL_sv_undef);
913 }
914 else
915 ERR ("'null' expected");
916
917 break;
918
919 default:
920 ERR ("malformed CBOR string, neither array, object, number, string or atom");
921 break;
922 }
923#endif
924 1144
925fail: 1145fail:
926 return &PL_sv_undef; 1146 return &PL_sv_undef;
927} 1147}
928 1148
929static SV * 1149static SV *
930decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1150decode_cbor (SV *string, CBOR *cbor, char **offset_return)
931{ 1151{
932 dec_t dec; 1152 dec_t dec = { };
933 SV *sv; 1153 SV *sv;
1154 STRLEN len;
1155 char *data = SvPVbyte (string, len);
934 1156
935 /* work around bugs in 5.10 where manipulating magic values
936 * makes perl ignore the magic in subsequent accesses.
937 * also make a copy of non-PV values, to get them into a clean
938 * state (SvPV should do that, but it's buggy, see below).
939 */
940 /*SvGETMAGIC (string);*/
941 if (SvMAGICAL (string) || !SvPOK (string))
942 string = sv_2mortal (newSVsv (string));
943
944 SvUPGRADE (string, SVt_PV);
945
946 /* work around a bug in perl 5.10, which causes SvCUR to fail an
947 * assertion with -DDEBUGGING, although SvCUR is documented to
948 * return the xpv_cur field which certainly exists after upgrading.
949 * according to nicholas clark, calling SvPOK fixes this.
950 * But it doesn't fix it, so try another workaround, call SvPV_nolen
951 * and hope for the best.
952 * Damnit, SvPV_nolen still trips over yet another assertion. This
953 * assertion business is seriously broken, try yet another workaround
954 * for the broken -DDEBUGGING.
955 */
956 {
957#ifdef DEBUGGING
958 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
959#else
960 STRLEN offset = SvCUR (string);
961#endif
962
963 if (offset > cbor->max_size && cbor->max_size) 1157 if (len > cbor->max_size && cbor->max_size)
964 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu", 1158 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
965 (unsigned long)SvCUR (string), (unsigned long)cbor->max_size); 1159 (unsigned long)len, (unsigned long)cbor->max_size);
966 }
967
968 sv_utf8_downgrade (string, 0);
969 1160
970 dec.cbor = *cbor; 1161 dec.cbor = *cbor;
971 dec.cur = (U8 *)SvPVX (string); 1162 dec.cur = (U8 *)data;
972 dec.end = (U8 *)SvEND (string); 1163 dec.end = (U8 *)data + len;
973 dec.err = 0;
974 dec.depth = 0;
975
976 if (dec.cbor.cb_object || dec.cbor.cb_sk_object)
977 ;//TODO dec.cbor.flags |= F_HOOK;
978 1164
979 sv = decode_sv (&dec); 1165 sv = decode_sv (&dec);
980 1166
981 if (offset_return) 1167 if (offset_return)
982 *offset_return = dec.cur; 1168 *offset_return = dec.cur;
986 dec.err = "garbage after CBOR object"; 1172 dec.err = "garbage after CBOR object";
987 1173
988 if (dec.err) 1174 if (dec.err)
989 { 1175 {
990 SvREFCNT_dec (sv); 1176 SvREFCNT_dec (sv);
991 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)SvPVX (string), (int)(uint8_t)*dec.cur); 1177 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
992 } 1178 }
993 1179
994 sv = sv_2mortal (sv); 1180 sv = sv_2mortal (sv);
995 1181
996 return sv; 1182 return sv;
1002MODULE = CBOR::XS PACKAGE = CBOR::XS 1188MODULE = CBOR::XS PACKAGE = CBOR::XS
1003 1189
1004BOOT: 1190BOOT:
1005{ 1191{
1006 cbor_stash = gv_stashpv ("CBOR::XS" , 1); 1192 cbor_stash = gv_stashpv ("CBOR::XS" , 1);
1007 cbor_boolean_stash = gv_stashpv ("CBOR::XS::Boolean", 1); 1193 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1);
1008 1194
1009 cbor_true = get_bool ("CBOR::XS::true"); 1195 types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
1010 cbor_false = get_bool ("CBOR::XS::false"); 1196 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
1197
1198 types_true = get_bool ("Types::Serialiser::true" );
1199 types_false = get_bool ("Types::Serialiser::false");
1200 types_error = get_bool ("Types::Serialiser::error");
1201
1202 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1203
1204 sv_cbor = newSVpv ("CBOR", 0);
1205 SvREADONLY_on (sv_cbor);
1011} 1206}
1012 1207
1013PROTOTYPES: DISABLE 1208PROTOTYPES: DISABLE
1014 1209
1015void CLONE (...) 1210void CLONE (...)
1016 CODE: 1211 CODE:
1017 cbor_stash = 0; 1212 cbor_stash = 0;
1213 cbor_tagged_stash = 0;
1214 types_error_stash = 0;
1018 cbor_boolean_stash = 0; 1215 types_boolean_stash = 0;
1019 1216
1020void new (char *klass) 1217void new (char *klass)
1021 PPCODE: 1218 PPCODE:
1022{ 1219{
1023 SV *pv = NEWSV (0, sizeof (CBOR)); 1220 SV *pv = NEWSV (0, sizeof (CBOR));
1031 1228
1032void shrink (CBOR *self, int enable = 1) 1229void shrink (CBOR *self, int enable = 1)
1033 ALIAS: 1230 ALIAS:
1034 shrink = F_SHRINK 1231 shrink = F_SHRINK
1035 allow_unknown = F_ALLOW_UNKNOWN 1232 allow_unknown = F_ALLOW_UNKNOWN
1233 allow_sharing = F_ALLOW_SHARING
1234 allow_cycles = F_ALLOW_CYCLES
1235 pack_strings = F_PACK_STRINGS
1236 validate_utf8 = F_VALIDATE_UTF8
1036 PPCODE: 1237 PPCODE:
1037{ 1238{
1038 if (enable) 1239 if (enable)
1039 self->flags |= ix; 1240 self->flags |= ix;
1040 else 1241 else
1045 1246
1046void get_shrink (CBOR *self) 1247void get_shrink (CBOR *self)
1047 ALIAS: 1248 ALIAS:
1048 get_shrink = F_SHRINK 1249 get_shrink = F_SHRINK
1049 get_allow_unknown = F_ALLOW_UNKNOWN 1250 get_allow_unknown = F_ALLOW_UNKNOWN
1251 get_allow_sharing = F_ALLOW_SHARING
1252 get_allow_cycles = F_ALLOW_CYCLES
1253 get_pack_strings = F_PACK_STRINGS
1254 get_validate_utf8 = F_VALIDATE_UTF8
1050 PPCODE: 1255 PPCODE:
1051 XPUSHs (boolSV (self->flags & ix)); 1256 XPUSHs (boolSV (self->flags & ix));
1052 1257
1053void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1258void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1054 PPCODE: 1259 PPCODE:
1070 CODE: 1275 CODE:
1071 RETVAL = self->max_size; 1276 RETVAL = self->max_size;
1072 OUTPUT: 1277 OUTPUT:
1073 RETVAL 1278 RETVAL
1074 1279
1075#if 0 //TODO 1280void filter (CBOR *self, SV *filter = 0)
1076
1077void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef)
1078 PPCODE: 1281 PPCODE:
1079{
1080 SvREFCNT_dec (self->cb_object); 1282 SvREFCNT_dec (self->filter);
1081 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; 1283 self->filter = filter ? newSVsv (filter) : filter;
1082
1083 XPUSHs (ST (0)); 1284 XPUSHs (ST (0));
1084}
1085 1285
1086void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef) 1286SV *get_filter (CBOR *self)
1087 PPCODE: 1287 CODE:
1088{ 1288 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
1089 if (!self->cb_sk_object) 1289 OUTPUT:
1090 self->cb_sk_object = newHV (); 1290 RETVAL
1091
1092 if (SvOK (cb))
1093 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1094 else
1095 {
1096 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1097
1098 if (!HvKEYS (self->cb_sk_object))
1099 {
1100 SvREFCNT_dec (self->cb_sk_object);
1101 self->cb_sk_object = 0;
1102 }
1103 }
1104
1105 XPUSHs (ST (0));
1106}
1107
1108#endif
1109 1291
1110void encode (CBOR *self, SV *scalar) 1292void encode (CBOR *self, SV *scalar)
1111 PPCODE: 1293 PPCODE:
1112 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; 1294 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
1113 XPUSHs (scalar); 1295 XPUSHs (scalar);
1127 PUSHs (sv); 1309 PUSHs (sv);
1128 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1310 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1129} 1311}
1130 1312
1131void DESTROY (CBOR *self) 1313void DESTROY (CBOR *self)
1132 CODE: 1314 PPCODE:
1133 SvREFCNT_dec (self->cb_sk_object); 1315 cbor_free (self);
1134 SvREFCNT_dec (self->cb_object);
1135 1316
1136PROTOTYPES: ENABLE 1317PROTOTYPES: ENABLE
1137 1318
1138void encode_cbor (SV *scalar) 1319void encode_cbor (SV *scalar)
1320 ALIAS:
1321 encode_cbor = 0
1322 encode_cbor_sharing = F_ALLOW_SHARING
1139 PPCODE: 1323 PPCODE:
1140{ 1324{
1141 CBOR cbor; 1325 CBOR cbor;
1142 cbor_init (&cbor); 1326 cbor_init (&cbor);
1327 cbor.flags |= ix;
1143 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1328 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1144 XPUSHs (scalar); 1329 XPUSHs (scalar);
1145} 1330}
1146 1331
1147void decode_cbor (SV *cborstr) 1332void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines