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 1.3 by root, Sat Oct 26 11:08:34 2013 UTC vs.
Revision 1.40 by root, Tue Dec 10 13:54:40 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// compatibility with perl <5.18
15#ifndef HvNAMELEN_get
16# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
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
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
14#define F_SHRINK 0x00000200UL 99#define F_SHRINK 0x00000001UL
15#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
16 105
17#define INIT_SIZE 32 // initial scalar size to be allocated 106#define INIT_SIZE 32 // initial scalar size to be allocated
18 107
19#define SB do { 108#define SB do {
20#define SE } while (0) 109#define SE } while (0)
21
22#if __GNUC__ >= 3
23# define expect(expr,value) __builtin_expect ((expr), (value))
24# define INLINE static inline
25#else
26# define expect(expr,value) (expr)
27# define INLINE static
28#endif
29
30#define expect_false(expr) expect ((expr) != 0, 0)
31#define expect_true(expr) expect ((expr) != 0, 1)
32 110
33#define IN_RANGE_INC(type,val,beg,end) \ 111#define IN_RANGE_INC(type,val,beg,end) \
34 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \ 112 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \
35 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg))) 113 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
36 114
42#else 120#else
43# define CBOR_SLOW 0 121# define CBOR_SLOW 0
44# define CBOR_STASH cbor_stash 122# define CBOR_STASH cbor_stash
45#endif 123#endif
46 124
47static HV *cbor_stash, *cbor_boolean_stash; // CBOR::XS:: 125static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
48static SV *cbor_true, *cbor_false; 126static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
49 127
50typedef struct { 128typedef struct {
51 U32 flags; 129 U32 flags;
52 U32 max_depth; 130 U32 max_depth;
53 STRLEN max_size; 131 STRLEN max_size;
132 SV *filter;
54 133
55 SV *cb_object; 134 // for the incremental parser
56 HV *cb_sk_object; 135 STRLEN incr_pos; // the current offset into the text
136 STRLEN incr_need; // minimum bytes needed to decode
137 AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
57} CBOR; 138} CBOR;
58 139
59INLINE void 140ecb_inline void
60cbor_init (CBOR *cbor) 141cbor_init (CBOR *cbor)
61{ 142{
62 Zero (cbor, 1, CBOR); 143 Zero (cbor, 1, CBOR);
63 cbor->max_depth = 512; 144 cbor->max_depth = 512;
64} 145}
65 146
147ecb_inline void
148cbor_free (CBOR *cbor)
149{
150 SvREFCNT_dec (cbor->filter);
151 SvREFCNT_dec (cbor->incr_count);
152}
153
66///////////////////////////////////////////////////////////////////////////// 154/////////////////////////////////////////////////////////////////////////////
67// utility functions 155// utility functions
68 156
69INLINE SV * 157ecb_inline SV *
70get_bool (const char *name) 158get_bool (const char *name)
71{ 159{
72 SV *sv = get_sv (name, 1); 160 SV *sv = get_sv (name, 1);
73 161
74 SvREADONLY_on (sv); 162 SvREADONLY_on (sv);
75 SvREADONLY_on (SvRV (sv)); 163 SvREADONLY_on (SvRV (sv));
76 164
77 return sv; 165 return sv;
78} 166}
79 167
80INLINE void 168ecb_inline void
81shrink (SV *sv) 169shrink (SV *sv)
82{ 170{
83 sv_utf8_downgrade (sv, 1); 171 sv_utf8_downgrade (sv, 1);
84 172
85 if (SvLEN (sv) > SvCUR (sv) + 1) 173 if (SvLEN (sv) > SvCUR (sv) + 1)
90 SvPV_renew (sv, SvCUR (sv) + 1); 178 SvPV_renew (sv, SvCUR (sv) + 1);
91#endif 179#endif
92 } 180 }
93} 181}
94 182
95///////////////////////////////////////////////////////////////////////////// 183// minimum length of a string to be registered for stringref
96// fp hell 184ecb_inline int
97 185minimum_string_length (UV idx)
98//TODO 186{
187 return idx > 23
188 ? idx > 0xffU
189 ? idx > 0xffffU
190 ? idx > 0xffffffffU
191 ? 11
192 : 7
193 : 5
194 : 4
195 : 3;
196}
99 197
100///////////////////////////////////////////////////////////////////////////// 198/////////////////////////////////////////////////////////////////////////////
101// encoder 199// encoder
102 200
103// structure used for encoding CBOR 201// structure used for encoding CBOR
106 char *cur; // SvPVX (sv) + current output position 204 char *cur; // SvPVX (sv) + current output position
107 char *end; // SvEND (sv) 205 char *end; // SvEND (sv)
108 SV *sv; // result scalar 206 SV *sv; // result scalar
109 CBOR cbor; 207 CBOR cbor;
110 U32 depth; // recursion level 208 U32 depth; // recursion level
209 HV *stringref[2]; // string => index, or 0 ([0] = bytes, [1] = utf-8)
210 UV stringref_idx;
211 HV *shareable; // ptr => index, or 0
212 UV shareable_idx;
111} enc_t; 213} enc_t;
112 214
113INLINE void 215ecb_inline void
114need (enc_t *enc, STRLEN len) 216need (enc_t *enc, STRLEN len)
115{ 217{
116 if (expect_false (enc->cur + len >= enc->end)) 218 if (ecb_expect_false (enc->cur + len >= enc->end))
117 { 219 {
118 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 220 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
119 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 221 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
120 enc->cur = SvPVX (enc->sv) + cur; 222 enc->cur = SvPVX (enc->sv) + cur;
121 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 223 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
122 } 224 }
123} 225}
124 226
125INLINE void 227ecb_inline void
126encode_ch (enc_t *enc, char ch) 228encode_ch (enc_t *enc, char ch)
127{ 229{
128 need (enc, 1); 230 need (enc, 1);
129 *enc->cur++ = ch; 231 *enc->cur++ = ch;
130} 232}
132static void 234static void
133encode_uint (enc_t *enc, int major, UV len) 235encode_uint (enc_t *enc, int major, UV len)
134{ 236{
135 need (enc, 9); 237 need (enc, 9);
136 238
137 if (len < 24) 239 if (ecb_expect_true (len < LENGTH_EXT1))
138 *enc->cur++ = major | len; 240 *enc->cur++ = major | len;
139 else if (len < 0x100) 241 else if (ecb_expect_true (len <= 0xffU))
140 { 242 {
141 *enc->cur++ = major | 24; 243 *enc->cur++ = major | LENGTH_EXT1;
142 *enc->cur++ = len; 244 *enc->cur++ = len;
143 } 245 }
144 else if (len < 0x10000) 246 else if (len <= 0xffffU)
145 { 247 {
146 *enc->cur++ = major | 25; 248 *enc->cur++ = major | LENGTH_EXT2;
147 *enc->cur++ = len >> 8; 249 *enc->cur++ = len >> 8;
148 *enc->cur++ = len; 250 *enc->cur++ = len;
149 } 251 }
150 else if (len < 0x100000000) 252 else if (len <= 0xffffffffU)
151 { 253 {
152 *enc->cur++ = major | 26; 254 *enc->cur++ = major | LENGTH_EXT4;
153 *enc->cur++ = len >> 24; 255 *enc->cur++ = len >> 24;
154 *enc->cur++ = len >> 16; 256 *enc->cur++ = len >> 16;
155 *enc->cur++ = len >> 8; 257 *enc->cur++ = len >> 8;
156 *enc->cur++ = len; 258 *enc->cur++ = len;
157 } 259 }
158 else if (len) 260 else
159 { 261 {
160 *enc->cur++ = major | 27; 262 *enc->cur++ = major | LENGTH_EXT8;
161 *enc->cur++ = len >> 56; 263 *enc->cur++ = len >> 56;
162 *enc->cur++ = len >> 48; 264 *enc->cur++ = len >> 48;
163 *enc->cur++ = len >> 40; 265 *enc->cur++ = len >> 40;
164 *enc->cur++ = len >> 32; 266 *enc->cur++ = len >> 32;
165 *enc->cur++ = len >> 24; 267 *enc->cur++ = len >> 24;
167 *enc->cur++ = len >> 8; 269 *enc->cur++ = len >> 8;
168 *enc->cur++ = len; 270 *enc->cur++ = len;
169 } 271 }
170} 272}
171 273
172static void 274ecb_inline void
275encode_tag (enc_t *enc, UV tag)
276{
277 encode_uint (enc, MAJOR_TAG, tag);
278}
279
280ecb_inline void
173encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 281encode_str (enc_t *enc, int utf8, char *str, STRLEN len)
174{ 282{
175 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 283 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
176 need (enc, len); 284 need (enc, len);
177 memcpy (enc->cur, str, len); 285 memcpy (enc->cur, str, len);
178 enc->cur += len; 286 enc->cur += len;
179} 287}
180 288
289static void
290encode_strref (enc_t *enc, int utf8, char *str, STRLEN len)
291{
292 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
293 {
294 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
295
296 if (SvOK (*svp))
297 {
298 // already registered, use stringref
299 encode_tag (enc, CBOR_TAG_STRINGREF);
300 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
301 return;
302 }
303 else if (len >= minimum_string_length (enc->stringref_idx))
304 {
305 // register only
306 sv_setuv (*svp, enc->stringref_idx);
307 ++enc->stringref_idx;
308 }
309 }
310
311 encode_str (enc, utf8, str, len);
312}
313
181static void encode_sv (enc_t *enc, SV *sv); 314static void encode_sv (enc_t *enc, SV *sv);
182 315
183static void 316static void
184encode_av (enc_t *enc, AV *av) 317encode_av (enc_t *enc, AV *av)
185{ 318{
188 if (enc->depth >= enc->cbor.max_depth) 321 if (enc->depth >= enc->cbor.max_depth)
189 croak (ERR_NESTING_EXCEEDED); 322 croak (ERR_NESTING_EXCEEDED);
190 323
191 ++enc->depth; 324 ++enc->depth;
192 325
193 encode_uint (enc, 0x80, len + 1); 326 encode_uint (enc, MAJOR_ARRAY, len + 1);
194 327
195 for (i = 0; i <= len; ++i) 328 for (i = 0; i <= len; ++i)
196 { 329 {
197 SV **svp = av_fetch (av, i, 0); 330 SV **svp = av_fetch (av, i, 0);
198 encode_sv (enc, svp ? *svp : &PL_sv_undef); 331 encode_sv (enc, svp ? *svp : &PL_sv_undef);
213 346
214 int pairs = hv_iterinit (hv); 347 int pairs = hv_iterinit (hv);
215 int mg = SvMAGICAL (hv); 348 int mg = SvMAGICAL (hv);
216 349
217 if (mg) 350 if (mg)
218 encode_ch (enc, 0xa0 | 31); 351 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
219 else 352 else
220 encode_uint (enc, 0xa0, pairs); 353 encode_uint (enc, MAJOR_MAP, pairs);
221 354
222 while ((he = hv_iternext (hv))) 355 while ((he = hv_iternext (hv)))
223 { 356 {
224 if (HeKLEN (he) == HEf_SVKEY) 357 if (HeKLEN (he) == HEf_SVKEY)
225 encode_sv (enc, HeSVKEY (he)); 358 encode_sv (enc, HeSVKEY (he));
226 else 359 else
227 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 360 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
228 361
229 encode_sv (enc, expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 362 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
230 } 363 }
231 364
232 if (mg) 365 if (mg)
233 encode_ch (enc, 0xe0 | 31); 366 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
234 367
235 --enc->depth; 368 --enc->depth;
236} 369}
237 370
238// encode objects, arrays and special \0=false and \1=true values. 371// encode objects, arrays and special \0=false and \1=true values.
239static void 372static void
240encode_rv (enc_t *enc, SV *sv) 373encode_rv (enc_t *enc, SV *sv)
241{ 374{
242 svtype svt;
243
244 SvGETMAGIC (sv); 375 SvGETMAGIC (sv);
376
245 svt = SvTYPE (sv); 377 svtype svt = SvTYPE (sv);
246 378
247 if (expect_false (SvOBJECT (sv))) 379 if (ecb_expect_false (SvOBJECT (sv)))
248 { 380 {
249 HV *stash = !CBOR_SLOW || cbor_boolean_stash 381 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
250 ? cbor_boolean_stash 382 ? types_boolean_stash
383 : gv_stashpv ("Types::Serialiser::Boolean", 1);
384 HV *error_stash = !CBOR_SLOW || types_error_stash
385 ? types_error_stash
386 : gv_stashpv ("Types::Serialiser::Error", 1);
387 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
388 ? cbor_tagged_stash
251 : gv_stashpv ("CBOR::XS::Boolean", 1); 389 : gv_stashpv ("CBOR::XS::Tagged" , 1);
252 390
253 if (SvSTASH (sv) == stash) 391 HV *stash = SvSTASH (sv);
254 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 392
393 if (stash == boolean_stash)
394 {
395 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
396 return;
397 }
398 else if (stash == error_stash)
399 {
400 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
401 return;
402 }
403 else if (stash == tagged_stash)
404 {
405 if (svt != SVt_PVAV)
406 croak ("encountered CBOR::XS::Tagged object that isn't an array");
407
408 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
409 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
410
411 return;
412 }
413 }
414
415 if (ecb_expect_false (SvREFCNT (sv) > 1)
416 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
417 {
418 if (!enc->shareable)
419 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
420
421 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
422
423 if (SvOK (*svp))
424 {
425 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
426 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
427 return;
428 }
255 else 429 else
256 { 430 {
257#if 0 //TODO 431 sv_setuv (*svp, enc->shareable_idx);
258 if (enc->cbor.flags & F_CONV_BLESSED) 432 ++enc->shareable_idx;
433 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
434 }
435 }
436
437 if (ecb_expect_false (SvOBJECT (sv)))
438 {
439 HV *stash = SvSTASH (sv);
440 GV *method;
441
442 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
259 { 443 {
444 dSP;
445
446 ENTER; SAVETMPS; PUSHMARK (SP);
260 // we re-bless the reference to get overload and other niceties right 447 // we re-bless the reference to get overload and other niceties right
261 GV *to_cbor = gv_fetchmethod_autoload (SvSTASH (sv), "TO_CBOR", 0);
262
263 if (to_cbor)
264 {
265 dSP;
266
267 ENTER; SAVETMPS; PUSHMARK (SP);
268 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv))); 448 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
269 449
270 // calling with G_SCALAR ensures that we always get a 1 return value
271 PUTBACK; 450 PUTBACK;
451 // G_SCALAR ensures that return value is 1
272 call_sv ((SV *)GvCV (to_cbor), G_SCALAR); 452 call_sv ((SV *)GvCV (method), G_SCALAR);
273 SPAGAIN; 453 SPAGAIN;
274 454
275 // catch this surprisingly common error 455 // catch this surprisingly common error
276 if (SvROK (TOPs) && SvRV (TOPs) == sv) 456 if (SvROK (TOPs) && SvRV (TOPs) == sv)
277 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv))); 457 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (stash));
278 458
279 sv = POPs;
280 PUTBACK;
281
282 encode_sv (enc, sv); 459 encode_sv (enc, POPs);
283 460
461 PUTBACK;
462
284 FREETMPS; LEAVE; 463 FREETMPS; LEAVE;
285 }
286 else if (enc->cbor.flags & F_ALLOW_BLESSED)
287 encode_str (enc, "null", 4, 0);
288 else
289 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_CBOR method available on it",
290 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
291 } 464 }
292 else if (enc->cbor.flags & F_ALLOW_BLESSED) 465 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
293 encode_str (enc, "null", 4, 0); 466 {
467 dSP;
468
469 ENTER; SAVETMPS; PUSHMARK (SP);
470 EXTEND (SP, 2);
471 // we re-bless the reference to get overload and other niceties right
472 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
473 PUSHs (sv_cbor);
474
475 PUTBACK;
476 int count = call_sv ((SV *)GvCV (method), G_ARRAY);
477 SPAGAIN;
478
479 // catch this surprisingly common error
480 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
481 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
482
483 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
484 encode_uint (enc, MAJOR_ARRAY, count + 1);
485 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
486
487 while (count)
488 encode_sv (enc, SP[1 - count--]);
489
490 PUTBACK;
491
492 FREETMPS; LEAVE;
493 }
294 else 494 else
295 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled", 495 croak ("encountered object '%s', but no TO_CBOR or FREEZE methods available on it",
296 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 496 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
297#endif
298 }
299 } 497 }
300 else if (svt == SVt_PVHV) 498 else if (svt == SVt_PVHV)
301 encode_hv (enc, (HV *)sv); 499 encode_hv (enc, (HV *)sv);
302 else if (svt == SVt_PVAV) 500 else if (svt == SVt_PVAV)
303 encode_av (enc, (AV *)sv); 501 encode_av (enc, (AV *)sv);
304 else if (svt < SVt_PVAV)
305 {
306 STRLEN len = 0;
307 char *pv = svt ? SvPV (sv, len) : 0;
308
309 if (len == 1 && *pv == '1')
310 encode_ch (enc, 0xe0 | 21);
311 else if (len == 1 && *pv == '0')
312 encode_ch (enc, 0xe0 | 20);
313 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
314 encode_ch (enc, 0xe0 | 23);
315 else
316 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
317 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
318 }
319 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
320 encode_ch (enc, 0xe0 | 23);
321 else 502 else
322 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 503 {
323 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 504 encode_tag (enc, CBOR_TAG_INDIRECTION);
505 encode_sv (enc, sv);
506 }
324} 507}
325 508
326static void 509static void
327encode_nv (enc_t *enc, SV *sv) 510encode_nv (enc_t *enc, SV *sv)
328{ 511{
329 double nv = SvNVX (sv); 512 double nv = SvNVX (sv);
330 513
331 need (enc, 9); 514 need (enc, 9);
332 515
333 if (expect_false (nv == (U32)nv)) 516 if (ecb_expect_false (nv == (NV)(U32)nv))
334 encode_uint (enc, 0x00, (U32)nv); 517 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
335 //TODO: maybe I32? 518 //TODO: maybe I32?
336 else if (expect_false (nv == (float)nv)) 519 else if (ecb_expect_false (nv == (float)nv))
337 { 520 {
338 uint32_t fp = ecb_float_to_binary32 (nv); 521 uint32_t fp = ecb_float_to_binary32 (nv);
339 522
340 *enc->cur++ = 0xe0 | 26; 523 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
341 524
342 if (!ecb_big_endian ()) 525 if (!ecb_big_endian ())
343 fp = ecb_bswap32 (fp); 526 fp = ecb_bswap32 (fp);
344 527
345 memcpy (enc->cur, &fp, 4); 528 memcpy (enc->cur, &fp, 4);
347 } 530 }
348 else 531 else
349 { 532 {
350 uint64_t fp = ecb_double_to_binary64 (nv); 533 uint64_t fp = ecb_double_to_binary64 (nv);
351 534
352 *enc->cur++ = 0xe0 | 27; 535 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
353 536
354 if (!ecb_big_endian ()) 537 if (!ecb_big_endian ())
355 fp = ecb_bswap64 (fp); 538 fp = ecb_bswap64 (fp);
356 539
357 memcpy (enc->cur, &fp, 8); 540 memcpy (enc->cur, &fp, 8);
366 549
367 if (SvPOKp (sv)) 550 if (SvPOKp (sv))
368 { 551 {
369 STRLEN len; 552 STRLEN len;
370 char *str = SvPV (sv, len); 553 char *str = SvPV (sv, len);
371 encode_str (enc, SvUTF8 (sv), str, len); 554 encode_strref (enc, SvUTF8 (sv), str, len);
372 } 555 }
373 else if (SvNOKp (sv)) 556 else if (SvNOKp (sv))
374 encode_nv (enc, sv); 557 encode_nv (enc, sv);
375 else if (SvIOKp (sv)) 558 else if (SvIOKp (sv))
376 { 559 {
377 if (SvIsUV (sv)) 560 if (SvIsUV (sv))
378 encode_uint (enc, 0x00, SvUVX (sv)); 561 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
379 else if (SvIVX (sv) >= 0) 562 else if (SvIVX (sv) >= 0)
380 encode_uint (enc, 0x00, SvIVX (sv)); 563 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
381 else 564 else
382 encode_uint (enc, 0x20, -(SvIVX (sv) + 1)); 565 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
383 } 566 }
384 else if (SvROK (sv)) 567 else if (SvROK (sv))
385 encode_rv (enc, SvRV (sv)); 568 encode_rv (enc, SvRV (sv));
386 else if (!SvOK (sv)) 569 else if (!SvOK (sv))
387 encode_ch (enc, 0xe0 | 22); 570 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
388 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 571 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
389 encode_ch (enc, 0xe0 | 23); 572 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
390 else 573 else
391 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 574 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
392 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 575 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
393} 576}
394 577
395static SV * 578static SV *
396encode_cbor (SV *scalar, CBOR *cbor) 579encode_cbor (SV *scalar, CBOR *cbor)
397{ 580{
398 enc_t enc; 581 enc_t enc = { };
399 582
400 enc.cbor = *cbor; 583 enc.cbor = *cbor;
401 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 584 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
402 enc.cur = SvPVX (enc.sv); 585 enc.cur = SvPVX (enc.sv);
403 enc.end = SvEND (enc.sv); 586 enc.end = SvEND (enc.sv);
404 enc.depth = 0;
405 587
406 SvPOK_only (enc.sv); 588 SvPOK_only (enc.sv);
589
590 if (cbor->flags & F_PACK_STRINGS)
591 {
592 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
593 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
594 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
595 }
596
407 encode_sv (&enc, scalar); 597 encode_sv (&enc, scalar);
408 598
409 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 599 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
410 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 600 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
411 601
425 U8 *end; // end of input string 615 U8 *end; // end of input string
426 const char *err; // parse error, if != 0 616 const char *err; // parse error, if != 0
427 CBOR cbor; 617 CBOR cbor;
428 U32 depth; // recursion depth 618 U32 depth; // recursion depth
429 U32 maxdepth; // recursion depth limit 619 U32 maxdepth; // recursion depth limit
620 AV *shareable;
621 AV *stringref;
622 SV *decode_tagged;
430} dec_t; 623} dec_t;
431 624
432#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 625#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE
433 626
434#define WANT(len) if (expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data"); 627#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data")
435 628
436#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED) 629#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED)
437#define DEC_DEC_DEPTH --dec->depth 630#define DEC_DEC_DEPTH --dec->depth
438 631
439static UV 632static UV
440decode_uint (dec_t *dec) 633decode_uint (dec_t *dec)
441{ 634{
442 switch (*dec->cur & 31) 635 U8 m = *dec->cur & MINOR_MASK;
443 { 636 ++dec->cur;
444 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
445 case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
446 case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
447 return *dec->cur++ & 31;
448 637
449 case 24: 638 if (ecb_expect_true (m < LENGTH_EXT1))
639 return m;
640 else if (ecb_expect_true (m == LENGTH_EXT1))
641 {
450 WANT (2); 642 WANT (1);
451 dec->cur += 2; 643 dec->cur += 1;
452 return dec->cur[-1]; 644 return dec->cur[-1];
453 645 }
454 case 25: 646 else if (ecb_expect_true (m == LENGTH_EXT2))
647 {
455 WANT (3); 648 WANT (2);
456 dec->cur += 3; 649 dec->cur += 2;
457 return (((UV)dec->cur[-2]) << 8) 650 return (((UV)dec->cur[-2]) << 8)
458 | ((UV)dec->cur[-1]); 651 | ((UV)dec->cur[-1]);
459 652 }
460 case 26: 653 else if (ecb_expect_true (m == LENGTH_EXT4))
654 {
461 WANT (5); 655 WANT (4);
462 dec->cur += 5; 656 dec->cur += 4;
463 return (((UV)dec->cur[-4]) << 24) 657 return (((UV)dec->cur[-4]) << 24)
464 | (((UV)dec->cur[-3]) << 16) 658 | (((UV)dec->cur[-3]) << 16)
465 | (((UV)dec->cur[-2]) << 8) 659 | (((UV)dec->cur[-2]) << 8)
466 | ((UV)dec->cur[-1]); 660 | ((UV)dec->cur[-1]);
467 661 }
468 case 27: 662 else if (ecb_expect_true (m == LENGTH_EXT8))
663 {
469 WANT (9); 664 WANT (8);
470 dec->cur += 9; 665 dec->cur += 8;
666
667 return
668#if UVSIZE < 8
669 0
670#else
471 return (((UV)dec->cur[-8]) << 56) 671 (((UV)dec->cur[-8]) << 56)
472 | (((UV)dec->cur[-7]) << 48) 672 | (((UV)dec->cur[-7]) << 48)
473 | (((UV)dec->cur[-6]) << 40) 673 | (((UV)dec->cur[-6]) << 40)
474 | (((UV)dec->cur[-5]) << 32) 674 | (((UV)dec->cur[-5]) << 32)
675#endif
475 | (((UV)dec->cur[-4]) << 24) 676 | (((UV)dec->cur[-4]) << 24)
476 | (((UV)dec->cur[-3]) << 16) 677 | (((UV)dec->cur[-3]) << 16)
477 | (((UV)dec->cur[-2]) << 8) 678 | (((UV)dec->cur[-2]) << 8)
478 | ((UV)dec->cur[-1]); 679 | ((UV)dec->cur[-1]);
479 680 }
480 default: 681 else
481 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 682 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
482 }
483 683
484fail: 684fail:
485 return 0; 685 return 0;
486} 686}
487 687
492{ 692{
493 AV *av = newAV (); 693 AV *av = newAV ();
494 694
495 DEC_INC_DEPTH; 695 DEC_INC_DEPTH;
496 696
497 if ((*dec->cur & 31) == 31) 697 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
498 { 698 {
499 ++dec->cur; 699 ++dec->cur;
500 700
501 for (;;) 701 for (;;)
502 { 702 {
503 WANT (1); 703 WANT (1);
504 704
505 if (*dec->cur == (0xe0 | 31)) 705 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
506 { 706 {
507 ++dec->cur; 707 ++dec->cur;
508 break; 708 break;
509 } 709 }
510 710
513 } 713 }
514 else 714 else
515 { 715 {
516 int i, len = decode_uint (dec); 716 int i, len = decode_uint (dec);
517 717
718 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
518 av_fill (av, len - 1); 719 av_fill (av, len - 1);
519 720
520 for (i = 0; i < len; ++i) 721 for (i = 0; i < len; ++i)
521 AvARRAY (av)[i] = decode_sv (dec); 722 AvARRAY (av)[i] = decode_sv (dec);
522 } 723 }
528 SvREFCNT_dec (av); 729 SvREFCNT_dec (av);
529 DEC_DEC_DEPTH; 730 DEC_DEC_DEPTH;
530 return &PL_sv_undef; 731 return &PL_sv_undef;
531} 732}
532 733
734static void
735decode_he (dec_t *dec, HV *hv)
736{
737 // for speed reasons, we specialcase single-string
738 // byte or utf-8 strings as keys, but only when !stringref
739
740 if (ecb_expect_true (!dec->stringref))
741 if (ecb_expect_true ((*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
742 {
743 I32 len = decode_uint (dec);
744 char *key = (char *)dec->cur;
745
746 dec->cur += len;
747
748 hv_store (hv, key, len, decode_sv (dec), 0);
749
750 return;
751 }
752 else if (ecb_expect_true ((*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
753 {
754 I32 len = decode_uint (dec);
755 char *key = (char *)dec->cur;
756
757 dec->cur += len;
758
759 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
760 if (!is_utf8_string (key, len))
761 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
762
763 hv_store (hv, key, -len, decode_sv (dec), 0);
764
765 return;
766 }
767
768 SV *k = decode_sv (dec);
769 SV *v = decode_sv (dec);
770
771 hv_store_ent (hv, k, v, 0);
772 SvREFCNT_dec (k);
773
774fail:
775 ;
776}
777
533static SV * 778static SV *
534decode_hv (dec_t *dec) 779decode_hv (dec_t *dec)
535{ 780{
536 HV *hv = newHV (); 781 HV *hv = newHV ();
537 782
538 DEC_INC_DEPTH; 783 DEC_INC_DEPTH;
539 784
540 if ((*dec->cur & 31) == 31) 785 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
541 { 786 {
542 ++dec->cur; 787 ++dec->cur;
543 788
544 for (;;) 789 for (;;)
545 { 790 {
546 WANT (1); 791 WANT (1);
547 792
548 if (*dec->cur == (0xe0 | 31)) 793 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
549 { 794 {
550 ++dec->cur; 795 ++dec->cur;
551 break; 796 break;
552 } 797 }
553 798
554 SV *k = decode_sv (dec); 799 decode_he (dec, hv);
555 SV *v = decode_sv (dec);
556
557 hv_store_ent (hv, k, v, 0);
558 } 800 }
559 } 801 }
560 else 802 else
561 { 803 {
562 int len = decode_uint (dec); 804 int pairs = decode_uint (dec);
563 805
564 while (len--) 806 while (pairs--)
565 { 807 decode_he (dec, hv);
566 SV *k = decode_sv (dec);
567 SV *v = decode_sv (dec);
568
569 hv_store_ent (hv, k, v, 0);
570 }
571 } 808 }
572 809
573 DEC_DEC_DEPTH; 810 DEC_DEC_DEPTH;
574 return newRV_noinc ((SV *)hv); 811 return newRV_noinc ((SV *)hv);
575
576#if 0
577 SV *sv;
578 HV *hv = newHV ();
579
580 DEC_INC_DEPTH;
581 decode_ws (dec);
582
583 for (;;)
584 {
585 // heuristic: assume that
586 // a) decode_str + hv_store_ent are abysmally slow.
587 // b) most hash keys are short, simple ascii text.
588 // => try to "fast-match" such strings to avoid
589 // the overhead of decode_str + hv_store_ent.
590 {
591 SV *value;
592 char *p = dec->cur;
593 char *e = p + 24; // only try up to 24 bytes
594
595 for (;;)
596 {
597 // the >= 0x80 is false on most architectures
598 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
599 {
600 // slow path, back up and use decode_str
601 SV *key = decode_str (dec);
602 if (!key)
603 goto fail;
604
605 decode_ws (dec); EXPECT_CH (':');
606
607 decode_ws (dec);
608 value = decode_sv (dec);
609 if (!value)
610 {
611 SvREFCNT_dec (key);
612 goto fail;
613 }
614
615 hv_store_ent (hv, key, value, 0);
616 SvREFCNT_dec (key);
617
618 break;
619 }
620 else if (*p == '"')
621 {
622 // fast path, got a simple key
623 char *key = dec->cur;
624 int len = p - key;
625 dec->cur = p + 1;
626
627 decode_ws (dec); EXPECT_CH (':');
628
629 decode_ws (dec);
630 value = decode_sv (dec);
631 if (!value)
632 goto fail;
633
634 hv_store (hv, key, len, value, 0);
635
636 break;
637 }
638
639 ++p;
640 }
641 }
642
643 decode_ws (dec);
644
645 if (*dec->cur == '}')
646 {
647 ++dec->cur;
648 break;
649 }
650
651 if (*dec->cur != ',')
652 ERR (", or } expected while parsing object/hash");
653
654 ++dec->cur;
655
656 decode_ws (dec);
657
658 if (*dec->cur == '}' && dec->cbor.flags & F_RELAXED)
659 {
660 ++dec->cur;
661 break;
662 }
663 }
664
665 DEC_DEC_DEPTH;
666 sv = newRV_noinc ((SV *)hv);
667
668 // check filter callbacks
669 if (dec->cbor.flags & F_HOOK)
670 {
671 if (dec->cbor.cb_sk_object && HvKEYS (hv) == 1)
672 {
673 HE *cb, *he;
674
675 hv_iterinit (hv);
676 he = hv_iternext (hv);
677 hv_iterinit (hv);
678
679 // the next line creates a mortal sv each time its called.
680 // might want to optimise this for common cases.
681 cb = hv_fetch_ent (dec->cbor.cb_sk_object, hv_iterkeysv (he), 0, 0);
682
683 if (cb)
684 {
685 dSP;
686 int count;
687
688 ENTER; SAVETMPS; PUSHMARK (SP);
689 XPUSHs (HeVAL (he));
690 sv_2mortal (sv);
691
692 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
693
694 if (count == 1)
695 {
696 sv = newSVsv (POPs);
697 FREETMPS; LEAVE;
698 return sv;
699 }
700
701 SvREFCNT_inc (sv);
702 FREETMPS; LEAVE;
703 }
704 }
705
706 if (dec->cbor.cb_object)
707 {
708 dSP;
709 int count;
710
711 ENTER; SAVETMPS; PUSHMARK (SP);
712 XPUSHs (sv_2mortal (sv));
713
714 PUTBACK; count = call_sv (dec->cbor.cb_object, G_ARRAY); SPAGAIN;
715
716 if (count == 1)
717 {
718 sv = newSVsv (POPs);
719 FREETMPS; LEAVE;
720 return sv;
721 }
722
723 SvREFCNT_inc (sv);
724 FREETMPS; LEAVE;
725 }
726 }
727
728 return sv;
729#endif
730 812
731fail: 813fail:
732 SvREFCNT_dec (hv); 814 SvREFCNT_dec (hv);
733 DEC_DEC_DEPTH; 815 DEC_DEC_DEPTH;
734 return &PL_sv_undef; 816 return &PL_sv_undef;
735} 817}
736 818
737static SV * 819static SV *
738decode_str (dec_t *dec, int utf8) 820decode_str (dec_t *dec, int utf8)
739{ 821{
740 SV *sv; 822 SV *sv = 0;
741 823
742 if ((*dec->cur & 31) == 31) 824 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF)
743 { 825 {
826 // indefinite length strings
744 ++dec->cur; 827 ++dec->cur;
745 828
829 U8 major = *dec->cur & MAJOR_MISC;
830
746 sv = newSVpvn ("", 0); 831 sv = newSVpvn ("", 0);
747 832
748 // not very fast, and certainly not robust against illegal input
749 for (;;) 833 for (;;)
750 { 834 {
751 WANT (1); 835 WANT (1);
752 836
753 if (*dec->cur == (0xe0 | 31)) 837 if ((*dec->cur - major) > LENGTH_EXT8)
838 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
754 { 839 {
755 ++dec->cur; 840 ++dec->cur;
756 break; 841 break;
757 } 842 }
843 else
844 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
758 845
759 SV *sv2 = decode_sv (dec); 846 STRLEN len = decode_uint (dec);
760 sv_catsv (sv, sv2); 847
848 WANT (len);
849 sv_catpvn (sv, dec->cur, len);
850 dec->cur += len;
761 } 851 }
762 } 852 }
763 else 853 else
764 { 854 {
765 STRLEN len = decode_uint (dec); 855 STRLEN len = decode_uint (dec);
766 856
767 WANT (len); 857 WANT (len);
768 sv = newSVpvn (dec->cur, len); 858 sv = newSVpvn (dec->cur, len);
769 dec->cur += len; 859 dec->cur += len;
860
861 if (ecb_expect_false (dec->stringref)
862 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
863 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
770 } 864 }
771 865
772 if (utf8) 866 if (utf8)
867 {
868 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
869 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
870 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
871
773 SvUTF8_on (sv); 872 SvUTF8_on (sv);
873 }
774 874
775 return sv; 875 return sv;
776 876
777fail: 877fail:
878 SvREFCNT_dec (sv);
778 return &PL_sv_undef; 879 return &PL_sv_undef;
779} 880}
780 881
781static SV * 882static SV *
782decode_tagged (dec_t *dec) 883decode_tagged (dec_t *dec)
783{ 884{
885 SV *sv = 0;
784 UV tag = decode_uint (dec); 886 UV tag = decode_uint (dec);
887
888 WANT (1);
889
890 switch (tag)
891 {
892 case CBOR_TAG_MAGIC:
785 SV *sv = decode_sv (dec); 893 sv = decode_sv (dec);
894 break;
786 895
787 if (tag == 55799) // 2.4.5 Self-Describe CBOR 896 case CBOR_TAG_INDIRECTION:
897 sv = newRV_noinc (decode_sv (dec));
898 break;
899
900 case CBOR_TAG_STRINGREF_NAMESPACE:
901 {
902 ENTER; SAVETMPS;
903
904 SAVESPTR (dec->stringref);
905 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
906
907 sv = decode_sv (dec);
908
909 FREETMPS; LEAVE;
910 }
911 break;
912
913 case CBOR_TAG_STRINGREF:
914 {
915 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
916 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
917
918 UV idx = decode_uint (dec);
919
920 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
921 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
922
923 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
924 }
925 break;
926
927 case CBOR_TAG_VALUE_SHAREABLE:
928 {
929 if (ecb_expect_false (!dec->shareable))
930 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
931
932 if (dec->cbor.flags & F_ALLOW_CYCLES)
933 {
934 sv = newSV (0);
935 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
936
937 SV *osv = decode_sv (dec);
938 sv_setsv (sv, osv);
939 SvREFCNT_dec_NN (osv);
940 }
941 else
942 {
943 av_push (dec->shareable, &PL_sv_undef);
944 int idx = AvFILLp (dec->shareable);
945 sv = decode_sv (dec);
946 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
947 }
948 }
949 break;
950
951 case CBOR_TAG_VALUE_SHAREDREF:
952 {
953 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
954 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
955
956 UV idx = decode_uint (dec);
957
958 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
959 ERR ("corrupted CBOR data (sharedref index out of bounds)");
960
961 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
962
963 if (sv == &PL_sv_undef)
964 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
965 }
966 break;
967
968 case CBOR_TAG_PERL_OBJECT:
969 {
970 sv = decode_sv (dec);
971
972 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
973 ERR ("corrupted CBOR data (non-array perl object)");
974
975 AV *av = (AV *)SvRV (sv);
976 int len = av_len (av) + 1;
977 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
978
979 if (!stash)
980 ERR ("cannot decode perl-object (package does not exist)");
981
982 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
983
984 if (!method)
985 ERR ("cannot decode perl-object (package does not have a THAW method)");
986
987 dSP;
988
989 ENTER; SAVETMPS; PUSHMARK (SP);
990 EXTEND (SP, len + 1);
991 // we re-bless the reference to get overload and other niceties right
992 PUSHs (*av_fetch (av, 0, 1));
993 PUSHs (sv_cbor);
994
995 int i;
996
997 for (i = 1; i < len; ++i)
998 PUSHs (*av_fetch (av, i, 1));
999
1000 PUTBACK;
1001 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
1002 SPAGAIN;
1003
1004 if (SvTRUE (ERRSV))
1005 {
1006 FREETMPS; LEAVE;
1007 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1008 }
1009
1010 SvREFCNT_dec (sv);
1011 sv = SvREFCNT_inc (POPs);
1012
1013 PUTBACK;
1014
1015 FREETMPS; LEAVE;
1016 }
1017 break;
1018
1019 default:
1020 {
1021 sv = decode_sv (dec);
1022
1023 dSP;
1024 ENTER; SAVETMPS; PUSHMARK (SP);
1025 EXTEND (SP, 2);
1026 PUSHs (newSVuv (tag));
1027 PUSHs (sv);
1028
1029 PUTBACK;
1030 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1031 SPAGAIN;
1032
1033 if (SvTRUE (ERRSV))
1034 {
1035 FREETMPS; LEAVE;
1036 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
1037 }
1038
1039 if (count)
1040 {
1041 SvREFCNT_dec (sv);
1042 sv = SvREFCNT_inc (POPs);
1043 }
1044 else
1045 {
1046 AV *av = newAV ();
1047 av_push (av, newSVuv (tag));
1048 av_push (av, sv);
1049
1050 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
1051 ? cbor_tagged_stash
1052 : gv_stashpv ("CBOR::XS::Tagged" , 1);
1053 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
1054 }
1055
1056 PUTBACK;
1057
1058 FREETMPS; LEAVE;
1059 }
1060 break;
1061 }
1062
788 return sv; 1063 return sv;
789 1064
790 AV *av = newAV (); 1065fail:
791 av_push (av, newSVuv (tag)); 1066 SvREFCNT_dec (sv);
792 av_push (av, sv); 1067 return &PL_sv_undef;
793 return newRV_noinc ((SV *)av);
794} 1068}
795 1069
796static SV * 1070static SV *
797decode_sv (dec_t *dec) 1071decode_sv (dec_t *dec)
798{ 1072{
799 WANT (1); 1073 WANT (1);
800 1074
801 switch (*dec->cur >> 5) 1075 switch (*dec->cur >> MAJOR_SHIFT)
802 { 1076 {
803 case 0: // unsigned int 1077 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
804 //TODO: 64 bit values on 3 2bit perls 1078 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
805 return newSVuv (decode_uint (dec)); 1079 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
806 case 1: // negative int 1080 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
807 return newSViv (-1 - (IV)decode_uint (dec)); 1081 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
808 case 2: // octet string 1082 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
809 return decode_str (dec, 0); 1083 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
810 case 3: // utf-8 string 1084
811 return decode_str (dec, 1); 1085 case MAJOR_MISC >> MAJOR_SHIFT:
812 case 4: // array
813 return decode_av (dec);
814 case 5: // map
815 return decode_hv (dec);
816 case 6: // tag
817 return decode_tagged (dec);
818 case 7: // misc
819 switch (*dec->cur++ & 31) 1086 switch (*dec->cur++ & MINOR_MASK)
820 { 1087 {
821 case 20: 1088 case SIMPLE_FALSE:
822#if CBOR_SLOW 1089#if CBOR_SLOW
823 cbor_false = get_bool ("CBOR::XS::false"); 1090 types_false = get_bool ("Types::Serialiser::false");
824#endif 1091#endif
825 return newSVsv (cbor_false); 1092 return newSVsv (types_false);
826 case 21: 1093 case SIMPLE_TRUE:
827#if CBOR_SLOW 1094#if CBOR_SLOW
828 cbor_true = get_bool ("CBOR::XS::true"); 1095 types_true = get_bool ("Types::Serialiser::true");
829#endif 1096#endif
830 return newSVsv (cbor_true); 1097 return newSVsv (types_true);
831 case 22: 1098 case SIMPLE_NULL:
832 return newSVsv (&PL_sv_undef); 1099 return newSVsv (&PL_sv_undef);
1100 case SIMPLE_UNDEF:
1101#if CBOR_SLOW
1102 types_error = get_bool ("Types::Serialiser::error");
1103#endif
1104 return newSVsv (types_error);
833 1105
834 case 25: 1106 case MISC_FLOAT16:
835 { 1107 {
836 WANT (2); 1108 WANT (2);
837 1109
838 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1110 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
839 dec->cur += 2; 1111 dec->cur += 2;
840 1112
841 return newSVnv (ecb_binary16_to_float (fp)); 1113 return newSVnv (ecb_binary16_to_float (fp));
842 } 1114 }
843 1115
844 case 26: 1116 case MISC_FLOAT32:
845 { 1117 {
846 uint32_t fp; 1118 uint32_t fp;
847 WANT (4); 1119 WANT (4);
848 memcpy (&fp, dec->cur, 4); 1120 memcpy (&fp, dec->cur, 4);
849 dec->cur += 4; 1121 dec->cur += 4;
852 fp = ecb_bswap32 (fp); 1124 fp = ecb_bswap32 (fp);
853 1125
854 return newSVnv (ecb_binary32_to_float (fp)); 1126 return newSVnv (ecb_binary32_to_float (fp));
855 } 1127 }
856 1128
857 case 27: 1129 case MISC_FLOAT64:
858 { 1130 {
859 uint64_t fp; 1131 uint64_t fp;
860 WANT (8); 1132 WANT (8);
861 memcpy (&fp, dec->cur, 8); 1133 memcpy (&fp, dec->cur, 8);
862 dec->cur += 8; 1134 dec->cur += 8;
865 fp = ecb_bswap64 (fp); 1137 fp = ecb_bswap64 (fp);
866 1138
867 return newSVnv (ecb_binary64_to_double (fp)); 1139 return newSVnv (ecb_binary64_to_double (fp));
868 } 1140 }
869 1141
870 // 0..19 unassigned 1142 // 0..19 unassigned simple
871 // 24 reserved + unassigned (reserved values are not encodable) 1143 // 24 reserved + unassigned simple (reserved values are not encodable)
1144 // 28-30 unassigned misc
1145 // 31 break code
872 default: 1146 default:
873 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1147 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
874 } 1148 }
875 1149
876 break; 1150 break;
877 } 1151 }
878#if 0
879 switch (*dec->cur)
880 {
881 //case '"': ++dec->cur; return decode_str (dec);
882 case '[': ++dec->cur; return decode_av (dec);
883 case '{': ++dec->cur; return decode_hv (dec);
884
885 case '-':
886 case '0': case '1': case '2': case '3': case '4':
887 case '5': case '6': case '7': case '8': case '9':
888 //TODO return decode_num (dec);
889
890 case 't':
891 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
892 {
893 dec->cur += 4;
894#if CBOR_SLOW
895 cbor_true = get_bool ("CBOR::XS::true");
896#endif
897 return newSVsv (cbor_true);
898 }
899 else
900 ERR ("'true' expected");
901
902 break;
903
904 case 'f':
905 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
906 {
907 dec->cur += 5;
908#if CBOR_SLOW
909 cbor_false = get_bool ("CBOR::XS::false");
910#endif
911 return newSVsv (cbor_false);
912 }
913 else
914 ERR ("'false' expected");
915
916 break;
917
918 case 'n':
919 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4))
920 {
921 dec->cur += 4;
922 return newSVsv (&PL_sv_undef);
923 }
924 else
925 ERR ("'null' expected");
926
927 break;
928
929 default:
930 ERR ("malformed CBOR string, neither array, object, number, string or atom");
931 break;
932 }
933#endif
934 1152
935fail: 1153fail:
936 return &PL_sv_undef; 1154 return &PL_sv_undef;
937} 1155}
938 1156
939static SV * 1157static SV *
940decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1158decode_cbor (SV *string, CBOR *cbor, char **offset_return)
941{ 1159{
942 dec_t dec; 1160 dec_t dec = { };
943 SV *sv; 1161 SV *sv;
1162 STRLEN len;
1163 char *data = SvPVbyte (string, len);
944 1164
945 /* work around bugs in 5.10 where manipulating magic values
946 * makes perl ignore the magic in subsequent accesses.
947 * also make a copy of non-PV values, to get them into a clean
948 * state (SvPV should do that, but it's buggy, see below).
949 */
950 /*SvGETMAGIC (string);*/
951 if (SvMAGICAL (string) || !SvPOK (string))
952 string = sv_2mortal (newSVsv (string));
953
954 SvUPGRADE (string, SVt_PV);
955
956 /* work around a bug in perl 5.10, which causes SvCUR to fail an
957 * assertion with -DDEBUGGING, although SvCUR is documented to
958 * return the xpv_cur field which certainly exists after upgrading.
959 * according to nicholas clark, calling SvPOK fixes this.
960 * But it doesn't fix it, so try another workaround, call SvPV_nolen
961 * and hope for the best.
962 * Damnit, SvPV_nolen still trips over yet another assertion. This
963 * assertion business is seriously broken, try yet another workaround
964 * for the broken -DDEBUGGING.
965 */
966 {
967#ifdef DEBUGGING
968 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
969#else
970 STRLEN offset = SvCUR (string);
971#endif
972
973 if (offset > cbor->max_size && cbor->max_size) 1165 if (len > cbor->max_size && cbor->max_size)
974 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu", 1166 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
975 (unsigned long)SvCUR (string), (unsigned long)cbor->max_size); 1167 (unsigned long)len, (unsigned long)cbor->max_size);
976 }
977
978 sv_utf8_downgrade (string, 0);
979 1168
980 dec.cbor = *cbor; 1169 dec.cbor = *cbor;
981 dec.cur = (U8 *)SvPVX (string); 1170 dec.cur = (U8 *)data;
982 dec.end = (U8 *)SvEND (string); 1171 dec.end = (U8 *)data + len;
983 dec.err = 0;
984 dec.depth = 0;
985
986 if (dec.cbor.cb_object || dec.cbor.cb_sk_object)
987 ;//TODO dec.cbor.flags |= F_HOOK;
988 1172
989 sv = decode_sv (&dec); 1173 sv = decode_sv (&dec);
990 1174
991 if (offset_return) 1175 if (offset_return)
992 *offset_return = dec.cur; 1176 *offset_return = dec.cur;
995 if (dec.cur != dec.end && !dec.err) 1179 if (dec.cur != dec.end && !dec.err)
996 dec.err = "garbage after CBOR object"; 1180 dec.err = "garbage after CBOR object";
997 1181
998 if (dec.err) 1182 if (dec.err)
999 { 1183 {
1184 if (dec.shareable)
1185 {
1186 // need to break cyclic links, which whould all be in shareable
1187 int i;
1188 SV **svp;
1189
1190 for (i = av_len (dec.shareable) + 1; i--; )
1191 if ((svp = av_fetch (dec.shareable, i, 0)))
1192 sv_setsv (*svp, &PL_sv_undef);
1193 }
1194
1000 SvREFCNT_dec (sv); 1195 SvREFCNT_dec (sv);
1001 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)SvPVX (string), (int)(uint8_t)*dec.cur); 1196 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
1002 } 1197 }
1003 1198
1004 sv = sv_2mortal (sv); 1199 sv = sv_2mortal (sv);
1005 1200
1006 return sv; 1201 return sv;
1007} 1202}
1008 1203
1204/////////////////////////////////////////////////////////////////////////////
1205// incremental parser
1206
1207#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1208
1209// returns 0 for notyet, 1 for success or error
1210static int
1211incr_parse (CBOR *self, SV *cborstr)
1212{
1213 STRLEN cur;
1214 SvPV (cborstr, cur);
1215
1216 while (ecb_expect_true (self->incr_need <= cur))
1217 {
1218 // table of integer count bytes
1219 static I8 incr_len[MINOR_MASK + 1] = {
1220 0, 0, 0, 0, 0, 0, 0, 0,
1221 0, 0, 0, 0, 0, 0, 0, 0,
1222 0, 0, 0, 0, 0, 0, 0, 0,
1223 1, 2, 4, 8,-1,-1,-1,-2
1224 };
1225
1226 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1227 U8 m = *p & MINOR_MASK;
1228 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1229 I8 ilen = incr_len[m];
1230
1231 self->incr_need = self->incr_pos + 1;
1232
1233 if (ecb_expect_false (ilen < 0))
1234 {
1235 if (m != MINOR_INDEF)
1236 return 1; // error
1237
1238 if (*p == (MAJOR_MISC | MINOR_INDEF))
1239 {
1240 if (count >= 0)
1241 return 1; // error
1242
1243 count = 1;
1244 }
1245 else
1246 {
1247 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1248 count = -1;
1249 }
1250 }
1251 else
1252 {
1253 self->incr_need += ilen;
1254 if (ecb_expect_false (self->incr_need > cur))
1255 return 0;
1256
1257 int major = *p >> MAJOR_SHIFT;
1258
1259 switch (major)
1260 {
1261 case MAJOR_BYTES >> MAJOR_SHIFT:
1262 case MAJOR_TEXT >> MAJOR_SHIFT:
1263 case MAJOR_ARRAY >> MAJOR_SHIFT:
1264 case MAJOR_MAP >> MAJOR_SHIFT:
1265 {
1266 UV len;
1267
1268 if (ecb_expect_false (ilen))
1269 {
1270 len = 0;
1271
1272 do {
1273 len = (len << 8) | *++p;
1274 } while (--ilen);
1275 }
1276 else
1277 len = m;
1278
1279 switch (major)
1280 {
1281 case MAJOR_BYTES >> MAJOR_SHIFT:
1282 case MAJOR_TEXT >> MAJOR_SHIFT:
1283 self->incr_need += len;
1284 if (ecb_expect_false (self->incr_need > cur))
1285 return 0;
1286
1287 break;
1288
1289 case MAJOR_MAP >> MAJOR_SHIFT:
1290 len <<= 1;
1291 case MAJOR_ARRAY >> MAJOR_SHIFT:
1292 if (len)
1293 {
1294 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1295 count = len + 1;
1296 }
1297 break;
1298 }
1299 }
1300 }
1301 }
1302
1303 self->incr_pos = self->incr_need;
1304
1305 if (count > 0)
1306 {
1307 while (!--count)
1308 {
1309 if (!AvFILLp (self->incr_count))
1310 return 1; // done
1311
1312 SvREFCNT_dec_NN (av_pop (self->incr_count));
1313 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1314 }
1315
1316 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1317 }
1318 }
1319
1320 return 0;
1321}
1322
1323
1009///////////////////////////////////////////////////////////////////////////// 1324/////////////////////////////////////////////////////////////////////////////
1010// XS interface functions 1325// XS interface functions
1011 1326
1012MODULE = CBOR::XS PACKAGE = CBOR::XS 1327MODULE = CBOR::XS PACKAGE = CBOR::XS
1013 1328
1014BOOT: 1329BOOT:
1015{ 1330{
1016 cbor_stash = gv_stashpv ("CBOR::XS" , 1); 1331 cbor_stash = gv_stashpv ("CBOR::XS" , 1);
1017 cbor_boolean_stash = gv_stashpv ("CBOR::XS::Boolean", 1); 1332 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1);
1018 1333
1019 cbor_true = get_bool ("CBOR::XS::true"); 1334 types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
1020 cbor_false = get_bool ("CBOR::XS::false"); 1335 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
1336
1337 types_true = get_bool ("Types::Serialiser::true" );
1338 types_false = get_bool ("Types::Serialiser::false");
1339 types_error = get_bool ("Types::Serialiser::error");
1340
1341 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1342
1343 sv_cbor = newSVpv ("CBOR", 0);
1344 SvREADONLY_on (sv_cbor);
1021} 1345}
1022 1346
1023PROTOTYPES: DISABLE 1347PROTOTYPES: DISABLE
1024 1348
1025void CLONE (...) 1349void CLONE (...)
1026 CODE: 1350 CODE:
1027 cbor_stash = 0; 1351 cbor_stash = 0;
1352 cbor_tagged_stash = 0;
1353 types_error_stash = 0;
1028 cbor_boolean_stash = 0; 1354 types_boolean_stash = 0;
1029 1355
1030void new (char *klass) 1356void new (char *klass)
1031 PPCODE: 1357 PPCODE:
1032{ 1358{
1033 SV *pv = NEWSV (0, sizeof (CBOR)); 1359 SV *pv = NEWSV (0, sizeof (CBOR));
1041 1367
1042void shrink (CBOR *self, int enable = 1) 1368void shrink (CBOR *self, int enable = 1)
1043 ALIAS: 1369 ALIAS:
1044 shrink = F_SHRINK 1370 shrink = F_SHRINK
1045 allow_unknown = F_ALLOW_UNKNOWN 1371 allow_unknown = F_ALLOW_UNKNOWN
1372 allow_sharing = F_ALLOW_SHARING
1373 allow_cycles = F_ALLOW_CYCLES
1374 pack_strings = F_PACK_STRINGS
1375 validate_utf8 = F_VALIDATE_UTF8
1046 PPCODE: 1376 PPCODE:
1047{ 1377{
1048 if (enable) 1378 if (enable)
1049 self->flags |= ix; 1379 self->flags |= ix;
1050 else 1380 else
1055 1385
1056void get_shrink (CBOR *self) 1386void get_shrink (CBOR *self)
1057 ALIAS: 1387 ALIAS:
1058 get_shrink = F_SHRINK 1388 get_shrink = F_SHRINK
1059 get_allow_unknown = F_ALLOW_UNKNOWN 1389 get_allow_unknown = F_ALLOW_UNKNOWN
1390 get_allow_sharing = F_ALLOW_SHARING
1391 get_allow_cycles = F_ALLOW_CYCLES
1392 get_pack_strings = F_PACK_STRINGS
1393 get_validate_utf8 = F_VALIDATE_UTF8
1060 PPCODE: 1394 PPCODE:
1061 XPUSHs (boolSV (self->flags & ix)); 1395 XPUSHs (boolSV (self->flags & ix));
1062 1396
1063void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1397void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1064 PPCODE: 1398 PPCODE:
1080 CODE: 1414 CODE:
1081 RETVAL = self->max_size; 1415 RETVAL = self->max_size;
1082 OUTPUT: 1416 OUTPUT:
1083 RETVAL 1417 RETVAL
1084 1418
1085#if 0 //TODO 1419void filter (CBOR *self, SV *filter = 0)
1086
1087void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef)
1088 PPCODE: 1420 PPCODE:
1089{
1090 SvREFCNT_dec (self->cb_object); 1421 SvREFCNT_dec (self->filter);
1091 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; 1422 self->filter = filter ? newSVsv (filter) : filter;
1092
1093 XPUSHs (ST (0)); 1423 XPUSHs (ST (0));
1094}
1095 1424
1096void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef) 1425SV *get_filter (CBOR *self)
1097 PPCODE: 1426 CODE:
1098{ 1427 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
1099 if (!self->cb_sk_object) 1428 OUTPUT:
1100 self->cb_sk_object = newHV (); 1429 RETVAL
1101
1102 if (SvOK (cb))
1103 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1104 else
1105 {
1106 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1107
1108 if (!HvKEYS (self->cb_sk_object))
1109 {
1110 SvREFCNT_dec (self->cb_sk_object);
1111 self->cb_sk_object = 0;
1112 }
1113 }
1114
1115 XPUSHs (ST (0));
1116}
1117
1118#endif
1119 1430
1120void encode (CBOR *self, SV *scalar) 1431void encode (CBOR *self, SV *scalar)
1121 PPCODE: 1432 PPCODE:
1122 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; 1433 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
1123 XPUSHs (scalar); 1434 XPUSHs (scalar);
1136 EXTEND (SP, 2); 1447 EXTEND (SP, 2);
1137 PUSHs (sv); 1448 PUSHs (sv);
1138 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1449 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1139} 1450}
1140 1451
1452void incr_parse (CBOR *self, SV *cborstr, int chop = 0)
1453 PPCODE:
1454{
1455 if (SvUTF8 (cborstr))
1456 sv_utf8_downgrade (cborstr, 0);
1457
1458 if (!self->incr_count)
1459 {
1460 self->incr_count = newAV ();
1461 self->incr_pos = 0;
1462 self->incr_need = 1;
1463
1464 av_push (self->incr_count, newSViv (1));
1465 }
1466
1467 for (;;)
1468 {
1469 if (!incr_parse (self, cborstr))
1470 {
1471 if (self->incr_need > self->max_size && self->max_size)
1472 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1473 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1474
1475 break;
1476 }
1477
1478 SV *sv;
1479 char *offset;
1480
1481 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1482 XPUSHs (sv);
1483
1484 av_clear (self->incr_count);
1485 av_push (self->incr_count, newSViv (1));
1486
1487 if (chop)
1488 {
1489 self->incr_pos = 0;
1490 sv_chop (cborstr, offset);
1491 }
1492 else
1493 self->incr_pos = offset - SvPVX (cborstr);
1494
1495 self->incr_need = self->incr_pos + 1;
1496 }
1497}
1498
1499void incr_reset (CBOR *self)
1500 CODE:
1501{
1502 SvREFCNT_dec (self->incr_count);
1503 self->incr_count = 0;
1504}
1505
1141void DESTROY (CBOR *self) 1506void DESTROY (CBOR *self)
1142 CODE: 1507 PPCODE:
1143 SvREFCNT_dec (self->cb_sk_object); 1508 cbor_free (self);
1144 SvREFCNT_dec (self->cb_object);
1145 1509
1146PROTOTYPES: ENABLE 1510PROTOTYPES: ENABLE
1147 1511
1148void encode_cbor (SV *scalar) 1512void encode_cbor (SV *scalar)
1513 ALIAS:
1514 encode_cbor = 0
1515 encode_cbor_sharing = F_ALLOW_SHARING
1149 PPCODE: 1516 PPCODE:
1150{ 1517{
1151 CBOR cbor; 1518 CBOR cbor;
1152 cbor_init (&cbor); 1519 cbor_init (&cbor);
1520 cbor.flags |= ix;
1153 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1521 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1154 XPUSHs (scalar); 1522 XPUSHs (scalar);
1155} 1523}
1156 1524
1157void decode_cbor (SV *cborstr) 1525void decode_cbor (SV *cborstr)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines