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 1.66 by root, Sun Nov 29 21:32:01 2020 UTC

6#include <string.h> 6#include <string.h>
7#include <stdlib.h> 7#include <stdlib.h>
8#include <stdio.h> 8#include <stdio.h>
9#include <limits.h> 9#include <limits.h>
10#include <float.h> 10#include <float.h>
11#include <inttypes.h>
11 12
13#define ECB_NO_THREADS 1
12#include "ecb.h" 14#include "ecb.h"
13 15
16// compatibility with perl <5.18
17#ifndef HvNAMELEN_get
18# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
19#endif
20#ifndef HvNAMELEN
21# define HvNAMELEN(hv) HvNAMELEN_get (hv)
22#endif
23#ifndef HvNAMEUTF8
24# define HvNAMEUTF8(hv) 0
25#endif
26#ifndef SvREFCNT_inc_NN
27# define SvREFCNT_inc_NN(sv) SvREFCNT_inc (sv)
28#endif
29#ifndef SvREFCNT_dec_NN
30# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)
31#endif
32
33// known major and minor types
34enum cbor_type
35{
36 MAJOR_SHIFT = 5,
37 MINOR_MASK = 0x1f,
38
39 MAJOR_POS_INT = 0 << MAJOR_SHIFT,
40 MAJOR_NEG_INT = 1 << MAJOR_SHIFT,
41 MAJOR_BYTES = 2 << MAJOR_SHIFT,
42 MAJOR_TEXT = 3 << MAJOR_SHIFT,
43 MAJOR_ARRAY = 4 << MAJOR_SHIFT,
44 MAJOR_MAP = 5 << MAJOR_SHIFT,
45 MAJOR_TAG = 6 << MAJOR_SHIFT,
46 MAJOR_MISC = 7 << MAJOR_SHIFT,
47
48 // INT/STRING/ARRAY/MAP subtypes
49 LENGTH_EXT1 = 24,
50 LENGTH_EXT2 = 25,
51 LENGTH_EXT4 = 26,
52 LENGTH_EXT8 = 27,
53
54 // SIMPLE types (effectively MISC subtypes)
55 SIMPLE_FALSE = 20,
56 SIMPLE_TRUE = 21,
57 SIMPLE_NULL = 22,
58 SIMPLE_UNDEF = 23,
59
60 // MISC subtype (unused)
61 MISC_EXT1 = 24,
62 MISC_FLOAT16 = 25,
63 MISC_FLOAT32 = 26,
64 MISC_FLOAT64 = 27,
65
66 // BYTES/TEXT/ARRAY/MAP
67 MINOR_INDEF = 31,
68};
69
70// known tags
71enum cbor_tag
72{
73 // extensions
74 CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
75 CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object
76 CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object
77 CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing
78 CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing
79 CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref
80 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection
81
82 // rfc7049
83 CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
84 CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
85 CBOR_TAG_POS_BIGNUM = 2, // byte string
86 CBOR_TAG_NEG_BIGNUM = 3, // byte string
87 CBOR_TAG_DECIMAL = 4, // decimal fraction, array
88 CBOR_TAG_BIGFLOAT = 5, // array
89
90 CBOR_TAG_CONV_B64U = 21, // base64url, any
91 CBOR_TAG_CONV_B64 = 22, // base64, any
92 CBOR_TAG_CONV_HEX = 23, // base16, any
93 CBOR_TAG_CBOR = 24, // embedded cbor, byte string
94
95 CBOR_TAG_URI = 32, // URI rfc3986, utf-8
96 CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
97 CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
98 CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
99 CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
100
101 CBOR_TAG_MAGIC = 55799, // self-describe cbor
102};
103
104// known forced types, also hardcoded in CBOR.pm
105enum
106{
107 AS_CBOR = 0,
108 AS_BYTES = 1,
109 AS_TEXT = 2,
110 AS_FLOAT16 = 3,
111 AS_FLOAT32 = 4,
112 AS_FLOAT64 = 5,
113 // possibly future enhancements: float, integer
114};
115
14#define F_SHRINK 0x00000200UL 116#define F_SHRINK 0x00000001UL
15#define F_ALLOW_UNKNOWN 0x00002000UL 117#define F_ALLOW_UNKNOWN 0x00000002UL
118#define F_ALLOW_SHARING 0x00000004UL
119#define F_ALLOW_CYCLES 0x00000008UL
120#define F_FORBID_OBJECTS 0x00000010UL
121#define F_PACK_STRINGS 0x00000020UL
122#define F_TEXT_KEYS 0x00000040UL
123#define F_TEXT_STRINGS 0x00000080UL
124#define F_VALIDATE_UTF8 0x00000100UL
16 125
17#define INIT_SIZE 32 // initial scalar size to be allocated 126#define INIT_SIZE 32 // initial scalar size to be allocated
18 127
19#define SB do { 128#define SB do {
20#define SE } while (0) 129#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)
32 130
33#define IN_RANGE_INC(type,val,beg,end) \ 131#define IN_RANGE_INC(type,val,beg,end) \
34 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \ 132 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \
35 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg))) 133 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
36 134
42#else 140#else
43# define CBOR_SLOW 0 141# define CBOR_SLOW 0
44# define CBOR_STASH cbor_stash 142# define CBOR_STASH cbor_stash
45#endif 143#endif
46 144
47static HV *cbor_stash, *cbor_boolean_stash; // CBOR::XS:: 145static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
48static SV *cbor_true, *cbor_false; 146static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
49 147
50typedef struct { 148typedef struct {
51 U32 flags; 149 U32 flags;
52 U32 max_depth; 150 U32 max_depth;
53 STRLEN max_size; 151 STRLEN max_size;
152 SV *filter;
54 153
55 SV *cb_object; 154 // for the incremental parser
56 HV *cb_sk_object; 155 STRLEN incr_pos; // the current offset into the text
156 STRLEN incr_need; // minimum bytes needed to decode
157 AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef.
57} CBOR; 158} CBOR;
58 159
59INLINE void 160ecb_inline void
60cbor_init (CBOR *cbor) 161cbor_init (CBOR *cbor)
61{ 162{
62 Zero (cbor, 1, CBOR); 163 Zero (cbor, 1, CBOR);
63 cbor->max_depth = 512; 164 cbor->max_depth = 512;
64} 165}
65 166
167ecb_inline void
168cbor_free (CBOR *cbor)
169{
170 SvREFCNT_dec (cbor->filter);
171 SvREFCNT_dec (cbor->incr_count);
172}
173
66///////////////////////////////////////////////////////////////////////////// 174/////////////////////////////////////////////////////////////////////////////
67// utility functions 175// utility functions
68 176
69INLINE SV * 177ecb_inline SV *
70get_bool (const char *name) 178get_bool (const char *name)
71{ 179{
72 SV *sv = get_sv (name, 1); 180 SV *sv = get_sv (name, 1);
73 181
74 SvREADONLY_on (sv); 182 SvREADONLY_on (sv);
75 SvREADONLY_on (SvRV (sv)); 183 SvREADONLY_on (SvRV (sv));
76 184
77 return sv; 185 return sv;
78} 186}
79 187
80INLINE void 188ecb_inline void
81shrink (SV *sv) 189shrink (SV *sv)
82{ 190{
83 sv_utf8_downgrade (sv, 1); 191 sv_utf8_downgrade (sv, 1);
84 192
85 if (SvLEN (sv) > SvCUR (sv) + 1) 193 if (SvLEN (sv) > SvCUR (sv) + 1)
90 SvPV_renew (sv, SvCUR (sv) + 1); 198 SvPV_renew (sv, SvCUR (sv) + 1);
91#endif 199#endif
92 } 200 }
93} 201}
94 202
95///////////////////////////////////////////////////////////////////////////// 203// minimum length of a string to be registered for stringref
96// fp hell 204ecb_inline int
97 205minimum_string_length (UV idx)
98//TODO 206{
207 return idx <= 23 ? 3
208 : idx <= 0xffU ? 4
209 : idx <= 0xffffU ? 5
210 : idx <= 0xffffffffU ? 7
211 : 11;
212}
99 213
100///////////////////////////////////////////////////////////////////////////// 214/////////////////////////////////////////////////////////////////////////////
101// encoder 215// encoder
102 216
103// structure used for encoding CBOR 217// structure used for encoding CBOR
106 char *cur; // SvPVX (sv) + current output position 220 char *cur; // SvPVX (sv) + current output position
107 char *end; // SvEND (sv) 221 char *end; // SvEND (sv)
108 SV *sv; // result scalar 222 SV *sv; // result scalar
109 CBOR cbor; 223 CBOR cbor;
110 U32 depth; // recursion level 224 U32 depth; // recursion level
225 HV *stringref[2]; // string => index, or 0 ([0] = bytes, [1] = utf-8)
226 UV stringref_idx;
227 HV *shareable; // ptr => index, or 0
228 UV shareable_idx;
111} enc_t; 229} enc_t;
112 230
113INLINE void 231ecb_inline void
114need (enc_t *enc, STRLEN len) 232need (enc_t *enc, STRLEN len)
115{ 233{
116 if (expect_false (enc->cur + len >= enc->end)) 234 if (ecb_expect_false ((uintptr_t)(enc->end - enc->cur) < len))
117 { 235 {
118 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 236 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
119 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 237 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
120 enc->cur = SvPVX (enc->sv) + cur; 238 enc->cur = SvPVX (enc->sv) + cur;
121 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 239 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
122 } 240 }
123} 241}
124 242
125INLINE void 243ecb_inline void
126encode_ch (enc_t *enc, char ch) 244encode_ch (enc_t *enc, char ch)
127{ 245{
128 need (enc, 1); 246 need (enc, 1);
129 *enc->cur++ = ch; 247 *enc->cur++ = ch;
130} 248}
131 249
250// used for tags, intregers, element counts and so on
132static void 251static void
133encode_uint (enc_t *enc, int major, UV len) 252encode_uint (enc_t *enc, int major, UV len)
134{ 253{
135 need (enc, 9); 254 need (enc, 9);
136 255
137 if (len < 24) 256 if (ecb_expect_true (len < LENGTH_EXT1))
138 *enc->cur++ = major | len; 257 *enc->cur++ = major | len;
139 else if (len < 0x100) 258 else if (ecb_expect_true (len <= 0xffU))
140 { 259 {
141 *enc->cur++ = major | 24; 260 *enc->cur++ = major | LENGTH_EXT1;
142 *enc->cur++ = len; 261 *enc->cur++ = len;
143 } 262 }
144 else if (len < 0x10000) 263 else if (len <= 0xffffU)
145 { 264 {
146 *enc->cur++ = major | 25; 265 *enc->cur++ = major | LENGTH_EXT2;
147 *enc->cur++ = len >> 8; 266 *enc->cur++ = len >> 8;
148 *enc->cur++ = len; 267 *enc->cur++ = len;
149 } 268 }
150 else if (len < 0x100000000) 269 else if (len <= 0xffffffffU)
151 { 270 {
152 *enc->cur++ = major | 26; 271 *enc->cur++ = major | LENGTH_EXT4;
153 *enc->cur++ = len >> 24; 272 *enc->cur++ = len >> 24;
154 *enc->cur++ = len >> 16; 273 *enc->cur++ = len >> 16;
155 *enc->cur++ = len >> 8; 274 *enc->cur++ = len >> 8;
156 *enc->cur++ = len; 275 *enc->cur++ = len;
157 } 276 }
158 else if (len) 277 else
159 { 278 {
160 *enc->cur++ = major | 27; 279 *enc->cur++ = major | LENGTH_EXT8;
161 *enc->cur++ = len >> 56; 280 *enc->cur++ = len >> 56;
162 *enc->cur++ = len >> 48; 281 *enc->cur++ = len >> 48;
163 *enc->cur++ = len >> 40; 282 *enc->cur++ = len >> 40;
164 *enc->cur++ = len >> 32; 283 *enc->cur++ = len >> 32;
165 *enc->cur++ = len >> 24; 284 *enc->cur++ = len >> 24;
167 *enc->cur++ = len >> 8; 286 *enc->cur++ = len >> 8;
168 *enc->cur++ = len; 287 *enc->cur++ = len;
169 } 288 }
170} 289}
171 290
172static void 291// encodes a perl value into a CBOR integer
292ecb_inline
293encode_int (enc_t *enc, SV *sv)
294{
295 if (SvIsUV (sv))
296 encode_uint (enc, MAJOR_POS_INT, SvUVX (sv));
297 else if (SvIVX (sv) >= 0)
298 encode_uint (enc, MAJOR_POS_INT, SvIVX (sv));
299 else
300 encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1));
301}
302
303ecb_inline void
304encode_tag (enc_t *enc, UV tag)
305{
306 encode_uint (enc, MAJOR_TAG, tag);
307}
308
309// exceptional (hopefully) slow path for byte strings that need to be utf8-encoded
310ecb_noinline static void
173encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 311encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len)
174{ 312{
175 encode_uint (enc, utf8 ? 0x60 : 0x40, len); 313 STRLEN ulen = len;
314 U8 *p, *pend = (U8 *)str + len;
315
316 for (p = (U8 *)str; p < pend; ++p)
317 ulen += *p >> 7; // count set high bits
318
319 encode_uint (enc, MAJOR_TEXT, ulen);
320
321 need (enc, ulen);
322 for (p = (U8 *)str; p < pend; ++p)
323 if (*p < 0x80)
324 *enc->cur++ = *p;
325 else
326 {
327 *enc->cur++ = 0xc0 + (*p >> 6);
328 *enc->cur++ = 0x80 + (*p & 63);
329 }
330}
331
332ecb_inline void
333encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
334{
335 if (ecb_expect_false (upgrade_utf8))
336 if (!utf8)
337 {
338 encode_str_utf8 (enc, utf8, str, len);
339 return;
340 }
341
342 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
176 need (enc, len); 343 need (enc, len);
177 memcpy (enc->cur, str, len); 344 memcpy (enc->cur, str, len);
178 enc->cur += len; 345 enc->cur += len;
179} 346}
180 347
348ecb_inline void
349encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
350{
351 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
352 {
353 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
354
355 if (SvOK (*svp))
356 {
357 // already registered, use stringref
358 encode_tag (enc, CBOR_TAG_STRINGREF);
359 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
360 return;
361 }
362 else if (len >= minimum_string_length (enc->stringref_idx))
363 {
364 // register only
365 sv_setuv (*svp, enc->stringref_idx);
366 ++enc->stringref_idx;
367 }
368 }
369
370 encode_str (enc, upgrade_utf8, utf8, str, len);
371}
372
373ecb_inline void
374encode_float16 (enc_t *enc, NV nv)
375{
376 need (enc, 1+2);
377
378 *enc->cur++ = MAJOR_MISC | MISC_FLOAT16;
379
380 uint16_t fp = ecb_float_to_binary16 (nv);
381
382 if (!ecb_big_endian ())
383 fp = ecb_bswap16 (fp);
384
385 memcpy (enc->cur, &fp, 2);
386 enc->cur += 2;
387}
388
389ecb_inline void
390encode_float32 (enc_t *enc, NV nv)
391{
392 need (enc, 1+4);
393
394 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
395
396 uint32_t fp = ecb_float_to_binary32 (nv);
397
398 if (!ecb_big_endian ())
399 fp = ecb_bswap32 (fp);
400
401 memcpy (enc->cur, &fp, 4);
402 enc->cur += 4;
403}
404
405ecb_inline void
406encode_float64 (enc_t *enc, NV nv)
407{
408 need (enc, 1+8);
409
410 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
411
412 uint64_t fp = ecb_double_to_binary64 (nv);
413
414 if (!ecb_big_endian ())
415 fp = ecb_bswap64 (fp);
416
417 memcpy (enc->cur, &fp, 8);
418 enc->cur += 8;
419}
420
421ecb_inline void
422encode_forced (enc_t *enc, UV type, SV *sv)
423{
424 switch (type)
425 {
426 case AS_CBOR:
427 {
428 STRLEN len;
429 char *str = SvPVbyte (sv, len);
430
431 need (enc, len);
432 memcpy (enc->cur, str, len);
433 enc->cur += len;
434 }
435 break;
436
437 case AS_BYTES:
438 {
439 STRLEN len;
440 char *str = SvPVbyte (sv, len);
441 encode_strref (enc, 0, 0, str, len);
442 }
443 break;
444
445 case AS_TEXT:
446 {
447 STRLEN len;
448 char *str = SvPVutf8 (sv, len);
449 encode_strref (enc, 1, 1, str, len);
450 }
451 break;
452
453 case AS_FLOAT16: encode_float16 (enc, SvNV (sv)); break;
454 case AS_FLOAT32: encode_float32 (enc, SvNV (sv)); break;
455 case AS_FLOAT64: encode_float64 (enc, SvNV (sv)); break;
456
457 default:
458 croak ("encountered malformed CBOR::XS::Tagged object");
459 }
460}
461
181static void encode_sv (enc_t *enc, SV *sv); 462static void encode_sv (enc_t *enc, SV *sv);
182 463
183static void 464static void
184encode_av (enc_t *enc, AV *av) 465encode_av (enc_t *enc, AV *av)
185{ 466{
188 if (enc->depth >= enc->cbor.max_depth) 469 if (enc->depth >= enc->cbor.max_depth)
189 croak (ERR_NESTING_EXCEEDED); 470 croak (ERR_NESTING_EXCEEDED);
190 471
191 ++enc->depth; 472 ++enc->depth;
192 473
193 encode_uint (enc, 0x80, len + 1); 474 encode_uint (enc, MAJOR_ARRAY, len + 1);
194 475
476 if (ecb_expect_false (SvMAGICAL (av)))
195 for (i = 0; i <= len; ++i) 477 for (i = 0; i <= len; ++i)
196 { 478 {
197 SV **svp = av_fetch (av, i, 0); 479 SV **svp = av_fetch (av, i, 0);
198 encode_sv (enc, svp ? *svp : &PL_sv_undef); 480 encode_sv (enc, svp ? *svp : &PL_sv_undef);
199 } 481 }
482 else
483 for (i = 0; i <= len; ++i)
484 {
485 SV *sv = AvARRAY (av)[i];
486 encode_sv (enc, sv ? sv : &PL_sv_undef);
487 }
200 488
201 --enc->depth; 489 --enc->depth;
202} 490}
203 491
204static void 492static void
212 ++enc->depth; 500 ++enc->depth;
213 501
214 int pairs = hv_iterinit (hv); 502 int pairs = hv_iterinit (hv);
215 int mg = SvMAGICAL (hv); 503 int mg = SvMAGICAL (hv);
216 504
217 if (mg) 505 if (ecb_expect_false (mg))
218 encode_ch (enc, 0xa0 | 31); 506 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
219 else 507 else
220 encode_uint (enc, 0xa0, pairs); 508 encode_uint (enc, MAJOR_MAP, pairs);
221 509
222 while ((he = hv_iternext (hv))) 510 while ((he = hv_iternext (hv)))
223 { 511 {
224 if (HeKLEN (he) == HEf_SVKEY) 512 if (HeKLEN (he) == HEf_SVKEY)
225 encode_sv (enc, HeSVKEY (he)); 513 encode_sv (enc, HeSVKEY (he));
226 else 514 else
227 encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 515 encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS | F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he));
228 516
229 encode_sv (enc, expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 517 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
230 } 518 }
231 519
232 if (mg) 520 if (ecb_expect_false (mg))
233 encode_ch (enc, 0xe0 | 31); 521 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
234 522
235 --enc->depth; 523 --enc->depth;
236} 524}
237 525
238// encode objects, arrays and special \0=false and \1=true values. 526// encode objects, arrays and special \0=false and \1=true values.
239static void 527static void
240encode_rv (enc_t *enc, SV *sv) 528encode_rv (enc_t *enc, SV *sv)
241{ 529{
242 svtype svt;
243
244 SvGETMAGIC (sv); 530 SvGETMAGIC (sv);
531
245 svt = SvTYPE (sv); 532 svtype svt = SvTYPE (sv);
246 533
247 if (expect_false (SvOBJECT (sv))) 534 if (ecb_expect_false (SvOBJECT (sv)))
248 { 535 {
249 HV *stash = !CBOR_SLOW || cbor_boolean_stash 536 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
250 ? cbor_boolean_stash 537 ? types_boolean_stash
538 : gv_stashpv ("Types::Serialiser::Boolean", 1);
539 HV *error_stash = !CBOR_SLOW || types_error_stash
540 ? types_error_stash
541 : gv_stashpv ("Types::Serialiser::Error", 1);
542 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
543 ? cbor_tagged_stash
251 : gv_stashpv ("CBOR::XS::Boolean", 1); 544 : gv_stashpv ("CBOR::XS::Tagged" , 1);
252 545
253 if (SvSTASH (sv) == stash) 546 HV *stash = SvSTASH (sv);
254 encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20); 547
548 if (stash == boolean_stash)
549 {
550 encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE);
551 return;
552 }
553 else if (stash == error_stash)
554 {
555 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
556 return;
557 }
558 else if (stash == tagged_stash)
559 {
560 if (svt != SVt_PVAV)
561 croak ("encountered CBOR::XS::Tagged object that isn't an array");
562
563 switch (av_len ((AV *)sv))
564 {
565 case 2-1:
566 // actually a tagged value
567 encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1)));
568 encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
569 break;
570
571 case 3-1:
572 // a forced type [value, type, undef]
573 encode_forced (enc, SvUV (*av_fetch ((AV *)sv, 1, 1)), *av_fetch ((AV *)sv, 0, 1));
574 break;
575
576 default:
577 croak ("encountered malformed CBOR::XS::Tagged object");
578 }
579
580 return;
581 }
582 }
583
584 if (ecb_expect_false (SvREFCNT (sv) > 1)
585 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
586 {
587 if (ecb_expect_false (!enc->shareable))
588 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
589
590 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
591
592 if (SvOK (*svp))
593 {
594 encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
595 encode_uint (enc, MAJOR_POS_INT, SvUV (*svp));
596 return;
597 }
255 else 598 else
256 { 599 {
257#if 0 //TODO 600 sv_setuv (*svp, enc->shareable_idx);
258 if (enc->cbor.flags & F_CONV_BLESSED) 601 ++enc->shareable_idx;
602 encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
603 }
604 }
605
606 if (ecb_expect_false (SvOBJECT (sv)))
607 {
608 HV *stash = SvSTASH (sv);
609 GV *method;
610
611 if (enc->cbor.flags & F_FORBID_OBJECTS)
612 croak ("encountered object '%s', but forbid_objects is enabled",
613 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
614 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
259 { 615 {
616 dSP;
617
618 ENTER; SAVETMPS;
619 PUSHMARK (SP);
260 // we re-bless the reference to get overload and other niceties right 620 // 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))); 621 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
269 622
270 // calling with G_SCALAR ensures that we always get a 1 return value
271 PUTBACK; 623 PUTBACK;
624 // G_SCALAR ensures that return value is 1
272 call_sv ((SV *)GvCV (to_cbor), G_SCALAR); 625 call_sv ((SV *)GvCV (method), G_SCALAR);
273 SPAGAIN; 626 SPAGAIN;
274 627
275 // catch this surprisingly common error 628 // catch this surprisingly common error
276 if (SvROK (TOPs) && SvRV (TOPs) == sv) 629 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))); 630 croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (stash));
278 631
279 sv = POPs;
280 PUTBACK;
281
282 encode_sv (enc, sv); 632 encode_sv (enc, POPs);
283 633
634 PUTBACK;
635
284 FREETMPS; LEAVE; 636 FREETMPS; LEAVE;
285 } 637 }
286 else if (enc->cbor.flags & F_ALLOW_BLESSED) 638 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
287 encode_str (enc, "null", 4, 0); 639 {
288 else 640 dSP;
289 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_CBOR method available on it", 641
290 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 642 ENTER; SAVETMPS;
643 PUSHMARK (SP);
644 EXTEND (SP, 2);
645 // we re-bless the reference to get overload and other niceties right
646 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
647 PUSHs (sv_cbor);
648
649 PUTBACK;
650 int count = call_sv ((SV *)GvCV (method), G_ARRAY);
651 SPAGAIN;
652
653 // catch this surprisingly common error
654 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
655 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
656
657 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
658 encode_uint (enc, MAJOR_ARRAY, count + 1);
659 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
660
661 {
662 int i;
663
664 for (i = 0; i < count; ++i)
665 encode_sv (enc, SP[i + 1 - count]);
666
667 SP -= count;
291 } 668 }
292 else if (enc->cbor.flags & F_ALLOW_BLESSED) 669
293 encode_str (enc, "null", 4, 0); 670 PUTBACK;
671
672 FREETMPS; LEAVE;
673 }
294 else 674 else
295 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled", 675 croak ("encountered object '%s', but no TO_CBOR or FREEZE methods available on it",
296 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 676 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
297#endif
298 }
299 } 677 }
300 else if (svt == SVt_PVHV) 678 else if (svt == SVt_PVHV)
301 encode_hv (enc, (HV *)sv); 679 encode_hv (enc, (HV *)sv);
302 else if (svt == SVt_PVAV) 680 else if (svt == SVt_PVAV)
303 encode_av (enc, (AV *)sv); 681 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 682 else
322 croak ("encountered %s, but CBOR can only represent references to arrays or hashes", 683 {
323 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 684 encode_tag (enc, CBOR_TAG_INDIRECTION);
685 encode_sv (enc, sv);
686 }
324} 687}
325 688
326static void 689static void
327encode_nv (enc_t *enc, SV *sv) 690encode_nv (enc_t *enc, SV *sv)
328{ 691{
329 double nv = SvNVX (sv); 692 double nv = SvNVX (sv);
330 693
331 need (enc, 9); 694 need (enc, 9);
332 695
333 if (expect_false (nv == (U32)nv)) 696 if (ecb_expect_false (nv == (NV)(U32)nv))
334 encode_uint (enc, 0x00, (U32)nv); 697 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
335 //TODO: maybe I32? 698 //TODO: maybe I32?
336 else if (expect_false (nv == (float)nv)) 699 else if (ecb_expect_false (nv == (float)nv))
337 { 700 encode_float32 (enc, nv);
338 uint32_t fp = ecb_float_to_binary32 (nv);
339
340 *enc->cur++ = 0xe0 | 26;
341
342 if (!ecb_big_endian ())
343 fp = ecb_bswap32 (fp);
344
345 memcpy (enc->cur, &fp, 4);
346 enc->cur += 4;
347 }
348 else 701 else
349 { 702 encode_float64 (enc, nv);
350 uint64_t fp = ecb_double_to_binary64 (nv);
351
352 *enc->cur++ = 0xe0 | 27;
353
354 if (!ecb_big_endian ())
355 fp = ecb_bswap64 (fp);
356
357 memcpy (enc->cur, &fp, 8);
358 enc->cur += 8;
359 }
360} 703}
361 704
362static void 705static void
363encode_sv (enc_t *enc, SV *sv) 706encode_sv (enc_t *enc, SV *sv)
364{ 707{
366 709
367 if (SvPOKp (sv)) 710 if (SvPOKp (sv))
368 { 711 {
369 STRLEN len; 712 STRLEN len;
370 char *str = SvPV (sv, len); 713 char *str = SvPV (sv, len);
371 encode_str (enc, SvUTF8 (sv), str, len); 714 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
372 } 715 }
373 else if (SvNOKp (sv)) 716 else if (SvNOKp (sv))
374 encode_nv (enc, sv); 717 encode_nv (enc, sv);
375 else if (SvIOKp (sv)) 718 else if (SvIOKp (sv))
376 { 719 encode_int (e,v sv);
377 if (SvIsUV (sv))
378 encode_uint (enc, 0x00, SvUVX (sv));
379 else if (SvIVX (sv) >= 0)
380 encode_uint (enc, 0x00, SvIVX (sv));
381 else
382 encode_uint (enc, 0x20, -(SvIVX (sv) + 1));
383 }
384 else if (SvROK (sv)) 720 else if (SvROK (sv))
385 encode_rv (enc, SvRV (sv)); 721 encode_rv (enc, SvRV (sv));
386 else if (!SvOK (sv)) 722 else if (!SvOK (sv))
387 encode_ch (enc, 0xe0 | 22); 723 encode_ch (enc, MAJOR_MISC | SIMPLE_NULL);
388 else if (enc->cbor.flags & F_ALLOW_UNKNOWN) 724 else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
389 encode_ch (enc, 0xe0 | 23); 725 encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF);
390 else 726 else
391 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", 727 croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
392 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 728 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
393} 729}
394 730
395static SV * 731static SV *
396encode_cbor (SV *scalar, CBOR *cbor) 732encode_cbor (SV *scalar, CBOR *cbor)
397{ 733{
398 enc_t enc; 734 enc_t enc = { 0 };
399 735
400 enc.cbor = *cbor; 736 enc.cbor = *cbor;
401 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 737 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
402 enc.cur = SvPVX (enc.sv); 738 enc.cur = SvPVX (enc.sv);
403 enc.end = SvEND (enc.sv); 739 enc.end = SvEND (enc.sv);
404 enc.depth = 0;
405 740
406 SvPOK_only (enc.sv); 741 SvPOK_only (enc.sv);
742
743 if (cbor->flags & F_PACK_STRINGS)
744 {
745 encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
746 enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
747 enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
748 }
749
407 encode_sv (&enc, scalar); 750 encode_sv (&enc, scalar);
408 751
409 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 752 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
410 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 753 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
411 754
425 U8 *end; // end of input string 768 U8 *end; // end of input string
426 const char *err; // parse error, if != 0 769 const char *err; // parse error, if != 0
427 CBOR cbor; 770 CBOR cbor;
428 U32 depth; // recursion depth 771 U32 depth; // recursion depth
429 U32 maxdepth; // recursion depth limit 772 U32 maxdepth; // recursion depth limit
773 AV *shareable;
774 AV *stringref;
775 SV *decode_tagged;
776 SV *err_sv; // optional sv for error, needs to be freed
430} dec_t; 777} dec_t;
431 778
432#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 779// set dec->err to ERRSV
780ecb_cold static void
781err_errsv (dec_t *dec)
782{
783 if (!dec->err)
784 {
785 dec->err_sv = newSVsv (ERRSV);
433 786
434#define WANT(len) if (expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data"); 787 // chop off the trailing \n
788 SvCUR_set (dec->err_sv, SvCUR (dec->err_sv) - 1);
789 *SvEND (dec->err_sv) = 0;
435 790
791 dec->err = SvPVutf8_nolen (dec->err_sv);
792 }
793}
794
795// the following functions are used to reduce code size and help the compiler to optimise
796ecb_cold static void
797err_set (dec_t *dec, const char *reason)
798{
799 if (!dec->err)
800 dec->err = reason;
801}
802
803ecb_cold static void
804err_unexpected_end (dec_t *dec)
805{
806 err_set (dec, "unexpected end of CBOR data");
807}
808
809#define ERR_DO(do) SB do; goto fail; SE
810#define ERR(reason) ERR_DO (err_set (dec, reason))
811#define ERR_ERRSV ERR_DO (err_errsv (dec))
812
813#define WANT(len) if (ecb_expect_false ((uintptr_t)(dec->end - dec->cur) < (STRLEN)len)) ERR_DO (err_unexpected_end (dec))
814
436#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED) 815#define DEC_INC_DEPTH if (ecb_expect_false (++dec->depth > dec->cbor.max_depth)) ERR (ERR_NESTING_EXCEEDED)
437#define DEC_DEC_DEPTH --dec->depth 816#define DEC_DEC_DEPTH --dec->depth
438 817
439static UV 818static UV
440decode_uint (dec_t *dec) 819decode_uint (dec_t *dec)
441{ 820{
442 switch (*dec->cur & 31) 821 U8 m = *dec->cur & MINOR_MASK;
443 { 822 ++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;
448 823
449 case 24: 824 if (ecb_expect_true (m < LENGTH_EXT1))
825 return m;
826 else if (ecb_expect_true (m == LENGTH_EXT1))
827 {
450 WANT (2); 828 WANT (1);
451 dec->cur += 2; 829 dec->cur += 1;
452 return dec->cur[-1]; 830 return dec->cur[-1];
453 831 }
454 case 25: 832 else if (ecb_expect_true (m == LENGTH_EXT2))
833 {
455 WANT (3); 834 WANT (2);
456 dec->cur += 3; 835 dec->cur += 2;
457 return (((UV)dec->cur[-2]) << 8) 836 return (((UV)dec->cur[-2]) << 8)
458 | ((UV)dec->cur[-1]); 837 | ((UV)dec->cur[-1]);
459 838 }
460 case 26: 839 else if (ecb_expect_true (m == LENGTH_EXT4))
840 {
461 WANT (5); 841 WANT (4);
462 dec->cur += 5; 842 dec->cur += 4;
463 return (((UV)dec->cur[-4]) << 24) 843 return (((UV)dec->cur[-4]) << 24)
464 | (((UV)dec->cur[-3]) << 16) 844 | (((UV)dec->cur[-3]) << 16)
465 | (((UV)dec->cur[-2]) << 8) 845 | (((UV)dec->cur[-2]) << 8)
466 | ((UV)dec->cur[-1]); 846 | ((UV)dec->cur[-1]);
467 847 }
468 case 27: 848 else if (ecb_expect_true (m == LENGTH_EXT8))
849 {
469 WANT (9); 850 WANT (8);
470 dec->cur += 9; 851 dec->cur += 8;
852
853 return
854#if UVSIZE < 8
855 0
856#else
471 return (((UV)dec->cur[-8]) << 56) 857 (((UV)dec->cur[-8]) << 56)
472 | (((UV)dec->cur[-7]) << 48) 858 | (((UV)dec->cur[-7]) << 48)
473 | (((UV)dec->cur[-6]) << 40) 859 | (((UV)dec->cur[-6]) << 40)
474 | (((UV)dec->cur[-5]) << 32) 860 | (((UV)dec->cur[-5]) << 32)
861#endif
475 | (((UV)dec->cur[-4]) << 24) 862 | (((UV)dec->cur[-4]) << 24)
476 | (((UV)dec->cur[-3]) << 16) 863 | (((UV)dec->cur[-3]) << 16)
477 | (((UV)dec->cur[-2]) << 8) 864 | (((UV)dec->cur[-2]) << 8)
478 | ((UV)dec->cur[-1]); 865 | ((UV)dec->cur[-1]);
479 866 }
480 default: 867 else
481 ERR ("corrupted CBOR data (unsupported integer minor encoding)"); 868 ERR ("corrupted CBOR data (unsupported integer minor encoding)");
482 }
483 869
484fail: 870fail:
485 return 0; 871 return 0;
486} 872}
487 873
492{ 878{
493 AV *av = newAV (); 879 AV *av = newAV ();
494 880
495 DEC_INC_DEPTH; 881 DEC_INC_DEPTH;
496 882
497 if ((*dec->cur & 31) == 31) 883 if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF))
498 { 884 {
499 ++dec->cur; 885 ++dec->cur;
500 886
501 for (;;) 887 for (;;)
502 { 888 {
503 WANT (1); 889 WANT (1);
504 890
505 if (*dec->cur == (0xe0 | 31)) 891 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
506 { 892 {
507 ++dec->cur; 893 ++dec->cur;
508 break; 894 break;
509 } 895 }
510 896
511 av_push (av, decode_sv (dec)); 897 av_push (av, decode_sv (dec));
512 } 898 }
513 } 899 }
514 else 900 else
515 { 901 {
516 int i, len = decode_uint (dec); 902 UV i, len = decode_uint (dec);
517 903
904 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); 905 av_fill (av, len - 1);
519 906
520 for (i = 0; i < len; ++i) 907 for (i = 0; i < len; ++i)
521 AvARRAY (av)[i] = decode_sv (dec); 908 AvARRAY (av)[i] = decode_sv (dec);
522 } 909 }
523 910
524 DEC_DEC_DEPTH; 911 DEC_DEC_DEPTH;
525 return newRV_noinc ((SV *)av); 912 return newRV_noinc ((SV *)av);
526 913
527fail: 914fail:
528 SvREFCNT_dec (av); 915 SvREFCNT_dec_NN (av);
529 DEC_DEC_DEPTH; 916 DEC_DEC_DEPTH;
530 return &PL_sv_undef; 917 return &PL_sv_undef;
531} 918}
532 919
920static void
921decode_he (dec_t *dec, HV *hv)
922{
923 // for speed reasons, we specialcase single-string
924 // byte or utf-8 strings as keys, but only when !stringref
925
926 if (ecb_expect_true (!dec->stringref))
927 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
928 {
929 STRLEN len = decode_uint (dec);
930 char *key = (char *)dec->cur;
931
932 WANT (len);
933 dec->cur += len;
934
935 hv_store (hv, key, len, decode_sv (dec), 0);
936
937 return;
938 }
939 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
940 {
941 STRLEN len = decode_uint (dec);
942 char *key = (char *)dec->cur;
943
944 WANT (len);
945 dec->cur += len;
946
947 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
948 if (!is_utf8_string (key, len))
949 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
950
951 hv_store (hv, key, -len, decode_sv (dec), 0);
952
953 return;
954 }
955
956 SV *k = decode_sv (dec);
957 SV *v = decode_sv (dec);
958
959 // we leak memory if uncaught exceptions are thrown by random magical
960 // methods, and this is hopefully the only place where it can happen,
961 // so if there is a chance of an exception, take the very slow path.
962 // since catching exceptions is "undocumented/internal/forbidden" by
963 // the new p5p powers, we need to call out to a perl function :/
964 if (ecb_expect_false (SvAMAGIC (k)))
965 {
966 dSP;
967
968 ENTER; SAVETMPS;
969 PUSHMARK (SP);
970 EXTEND (SP, 3);
971 PUSHs (sv_2mortal (newRV_inc ((SV *)hv)));
972 PUSHs (sv_2mortal (k));
973 PUSHs (sv_2mortal (v));
974
975 PUTBACK;
976 call_pv ("CBOR::XS::_hv_store", G_VOID | G_DISCARD | G_EVAL);
977 SPAGAIN;
978
979 FREETMPS; LEAVE;
980
981 if (SvTRUE (ERRSV))
982 ERR_ERRSV;
983
984 return;
985 }
986
987 hv_store_ent (hv, k, v, 0);
988 SvREFCNT_dec_NN (k);
989
990fail:
991 ;
992}
993
533static SV * 994static SV *
534decode_hv (dec_t *dec) 995decode_hv (dec_t *dec)
535{ 996{
536 HV *hv = newHV (); 997 HV *hv = newHV ();
537 998
538 DEC_INC_DEPTH; 999 DEC_INC_DEPTH;
539 1000
540 if ((*dec->cur & 31) == 31) 1001 if (*dec->cur == (MAJOR_MAP | MINOR_INDEF))
541 { 1002 {
542 ++dec->cur; 1003 ++dec->cur;
543 1004
544 for (;;) 1005 for (;;)
545 { 1006 {
546 WANT (1); 1007 WANT (1);
547 1008
548 if (*dec->cur == (0xe0 | 31)) 1009 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
549 { 1010 {
550 ++dec->cur; 1011 ++dec->cur;
551 break; 1012 break;
552 } 1013 }
553 1014
554 SV *k = decode_sv (dec); 1015 decode_he (dec, hv);
555 SV *v = decode_sv (dec);
556
557 hv_store_ent (hv, k, v, 0);
558 } 1016 }
559 } 1017 }
560 else 1018 else
561 { 1019 {
562 int len = decode_uint (dec); 1020 UV pairs = decode_uint (dec);
563 1021
1022 WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes
1023
564 while (len--) 1024 while (pairs--)
565 { 1025 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 } 1026 }
572 1027
573 DEC_DEC_DEPTH; 1028 DEC_DEC_DEPTH;
574 return newRV_noinc ((SV *)hv); 1029 return newRV_noinc ((SV *)hv);
575 1030
576#if 0 1031fail:
577 SV *sv; 1032 SvREFCNT_dec_NN (hv);
578 HV *hv = newHV ();
579
580 DEC_INC_DEPTH; 1033 DEC_DEC_DEPTH;
581 decode_ws (dec); 1034 return &PL_sv_undef;
1035}
582 1036
583 for (;;) 1037static SV *
584 { 1038decode_str (dec_t *dec, int utf8)
585 // heuristic: assume that 1039{
586 // a) decode_str + hv_store_ent are abysmally slow. 1040 SV *sv = 0;
587 // b) most hash keys are short, simple ascii text. 1041
588 // => try to "fast-match" such strings to avoid 1042 if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF))
589 // the overhead of decode_str + hv_store_ent.
590 { 1043 {
591 SV *value; 1044 // indefinite length strings
592 char *p = dec->cur; 1045 ++dec->cur;
593 char *e = p + 24; // only try up to 24 bytes
594 1046
1047 U8 major = *dec->cur & MAJOR_MISC;
1048
1049 sv = newSVpvn ("", 0);
1050
595 for (;;) 1051 for (;;)
596 { 1052 {
597 // the >= 0x80 is false on most architectures 1053 WANT (1);
598 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\') 1054
1055 if ((*dec->cur - major) > LENGTH_EXT8)
1056 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF))
599 { 1057 {
600 // slow path, back up and use decode_str 1058 ++dec->cur;
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; 1059 break;
619 } 1060 }
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 { 1061 else
685 dSP; 1062 ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
686 int count;
687 1063
688 ENTER; SAVETMPS; PUSHMARK (SP); 1064 STRLEN len = decode_uint (dec);
689 XPUSHs (HeVAL (he));
690 sv_2mortal (sv);
691 1065
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
731fail:
732 SvREFCNT_dec (hv);
733 DEC_DEC_DEPTH;
734 return &PL_sv_undef;
735}
736
737static SV *
738decode_str (dec_t *dec, int utf8)
739{
740 SV *sv;
741
742 if ((*dec->cur & 31) == 31)
743 {
744 ++dec->cur;
745
746 sv = newSVpvn ("", 0);
747
748 // not very fast, and certainly not robust against illegal input
749 for (;;)
750 {
751 WANT (1); 1066 WANT (len);
752 1067 sv_catpvn (sv, dec->cur, len);
753 if (*dec->cur == (0xe0 | 31)) 1068 dec->cur += len;
754 {
755 ++dec->cur;
756 break;
757 }
758
759 SV *sv2 = decode_sv (dec);
760 sv_catsv (sv, sv2);
761 } 1069 }
762 } 1070 }
763 else 1071 else
764 { 1072 {
765 STRLEN len = decode_uint (dec); 1073 STRLEN len = decode_uint (dec);
766 1074
767 WANT (len); 1075 WANT (len);
768 sv = newSVpvn (dec->cur, len); 1076 sv = newSVpvn (dec->cur, len);
769 dec->cur += len; 1077 dec->cur += len;
1078
1079 if (ecb_expect_false (dec->stringref)
1080 && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
1081 av_push (dec->stringref, SvREFCNT_inc_NN (sv));
770 } 1082 }
771 1083
772 if (utf8) 1084 if (utf8)
1085 {
1086 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
1087 if (!is_utf8_string (SvPVX (sv), SvCUR (sv)))
1088 ERR ("corrupted CBOR data (invalid UTF-8 in text string)");
1089
773 SvUTF8_on (sv); 1090 SvUTF8_on (sv);
1091 }
774 1092
775 return sv; 1093 return sv;
776 1094
777fail: 1095fail:
1096 SvREFCNT_dec (sv);
778 return &PL_sv_undef; 1097 return &PL_sv_undef;
779} 1098}
780 1099
781static SV * 1100static SV *
782decode_tagged (dec_t *dec) 1101decode_tagged (dec_t *dec)
783{ 1102{
1103 SV *sv = 0;
784 UV tag = decode_uint (dec); 1104 UV tag = decode_uint (dec);
1105
1106 WANT (1);
1107
1108 switch (tag)
1109 {
1110 case CBOR_TAG_MAGIC:
785 SV *sv = decode_sv (dec); 1111 sv = decode_sv (dec);
1112 break;
786 1113
787 if (tag == 55799) // 2.4.5 Self-Describe CBOR 1114 case CBOR_TAG_INDIRECTION:
1115 sv = newRV_noinc (decode_sv (dec));
1116 break;
1117
1118 case CBOR_TAG_STRINGREF_NAMESPACE:
1119 {
1120 // do not use SAVETMPS/FREETMPS, as these will
1121 // erase mortalised caches, e.g. "shareable"
1122 ENTER;
1123
1124 SAVESPTR (dec->stringref);
1125 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
1126
1127 sv = decode_sv (dec);
1128
1129 LEAVE;
1130 }
1131 break;
1132
1133 case CBOR_TAG_STRINGREF:
1134 {
1135 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
1136 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
1137
1138 UV idx = decode_uint (dec);
1139
1140 if (!dec->stringref || idx >= (UV)(1 + AvFILLp (dec->stringref)))
1141 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
1142
1143 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
1144 }
1145 break;
1146
1147 case CBOR_TAG_VALUE_SHAREABLE:
1148 {
1149 if (ecb_expect_false (!dec->shareable))
1150 dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
1151
1152 if (dec->cbor.flags & F_ALLOW_CYCLES)
1153 {
1154 sv = newSV (0);
1155 av_push (dec->shareable, SvREFCNT_inc_NN (sv));
1156
1157 SV *osv = decode_sv (dec);
1158 sv_setsv (sv, osv);
1159 SvREFCNT_dec_NN (osv);
1160 }
1161 else
1162 {
1163 av_push (dec->shareable, &PL_sv_undef);
1164 int idx = AvFILLp (dec->shareable);
1165 sv = decode_sv (dec);
1166 av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv));
1167 }
1168 }
1169 break;
1170
1171 case CBOR_TAG_VALUE_SHAREDREF:
1172 {
1173 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
1174 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
1175
1176 UV idx = decode_uint (dec);
1177
1178 if (!dec->shareable || idx >= (UV)(1 + AvFILLp (dec->shareable)))
1179 ERR ("corrupted CBOR data (sharedref index out of bounds)");
1180
1181 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
1182
1183 if (sv == &PL_sv_undef)
1184 ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled");
1185 }
1186 break;
1187
1188 case CBOR_TAG_PERL_OBJECT:
1189 {
1190 if (dec->cbor.flags & F_FORBID_OBJECTS)
1191 goto filter;
1192
1193 sv = decode_sv (dec);
1194
1195 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
1196 ERR ("corrupted CBOR data (non-array perl object)");
1197
1198 AV *av = (AV *)SvRV (sv);
1199 int len = av_len (av) + 1;
1200 HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
1201
1202 if (!stash)
1203 ERR ("cannot decode perl-object (package does not exist)");
1204
1205 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
1206
1207 if (!method)
1208 ERR ("cannot decode perl-object (package does not have a THAW method)");
1209
1210 dSP;
1211
1212 ENTER; SAVETMPS;
1213 PUSHMARK (SP);
1214 EXTEND (SP, len + 1);
1215 // we re-bless the reference to get overload and other niceties right
1216 PUSHs (*av_fetch (av, 0, 1));
1217 PUSHs (sv_cbor);
1218
1219 int i;
1220
1221 for (i = 1; i < len; ++i)
1222 PUSHs (*av_fetch (av, i, 1));
1223
1224 PUTBACK;
1225 call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
1226 SPAGAIN;
1227
1228 if (SvTRUE (ERRSV))
1229 {
1230 FREETMPS; LEAVE;
1231 ERR_ERRSV;
1232 }
1233
1234 SvREFCNT_dec_NN (sv);
1235 sv = SvREFCNT_inc (POPs);
1236
1237 PUTBACK;
1238
1239 FREETMPS; LEAVE;
1240 }
1241 break;
1242
1243 default:
1244 filter:
1245 {
1246 SV *tag_sv = newSVuv (tag);
1247
1248 sv = decode_sv (dec);
1249
1250 dSP;
1251 ENTER; SAVETMPS;
1252 PUSHMARK (SP);
1253 EXTEND (SP, 2);
1254 PUSHs (tag_sv);
1255 PUSHs (sv);
1256
1257 PUTBACK;
1258 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1259 SPAGAIN;
1260
1261 if (SvTRUE (ERRSV))
1262 {
1263 SvREFCNT_dec_NN (tag_sv);
1264 FREETMPS; LEAVE;
1265 ERR_ERRSV;
1266 }
1267
1268 if (count)
1269 {
1270 SvREFCNT_dec_NN (tag_sv);
1271 SvREFCNT_dec_NN (sv);
1272 sv = SvREFCNT_inc_NN (TOPs);
1273 SP -= count;
1274 }
1275 else
1276 {
1277 AV *av = newAV ();
1278 av_push (av, tag_sv);
1279 av_push (av, sv);
1280
1281 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
1282 ? cbor_tagged_stash
1283 : gv_stashpv ("CBOR::XS::Tagged" , 1);
1284 sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
1285 }
1286
1287 PUTBACK;
1288
1289 FREETMPS; LEAVE;
1290 }
1291 break;
1292 }
1293
788 return sv; 1294 return sv;
789 1295
790 AV *av = newAV (); 1296fail:
791 av_push (av, newSVuv (tag)); 1297 SvREFCNT_dec (sv);
792 av_push (av, sv); 1298 return &PL_sv_undef;
793 return newRV_noinc ((SV *)av);
794} 1299}
795 1300
796static SV * 1301static SV *
797decode_sv (dec_t *dec) 1302decode_sv (dec_t *dec)
798{ 1303{
799 WANT (1); 1304 WANT (1);
800 1305
801 switch (*dec->cur >> 5) 1306 switch (*dec->cur >> MAJOR_SHIFT)
802 { 1307 {
803 case 0: // unsigned int 1308 case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec));
804 //TODO: 64 bit values on 3 2bit perls 1309 case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec));
805 return newSVuv (decode_uint (dec)); 1310 case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0);
806 case 1: // negative int 1311 case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1);
807 return newSViv (-1 - (IV)decode_uint (dec)); 1312 case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec);
808 case 2: // octet string 1313 case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec);
809 return decode_str (dec, 0); 1314 case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec);
810 case 3: // utf-8 string 1315
811 return decode_str (dec, 1); 1316 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) 1317 switch (*dec->cur++ & MINOR_MASK)
820 { 1318 {
821 case 20: 1319 case SIMPLE_FALSE:
822#if CBOR_SLOW 1320#if CBOR_SLOW
823 cbor_false = get_bool ("CBOR::XS::false"); 1321 types_false = get_bool ("Types::Serialiser::false");
824#endif 1322#endif
825 return newSVsv (cbor_false); 1323 return newSVsv (types_false);
826 case 21: 1324 case SIMPLE_TRUE:
827#if CBOR_SLOW 1325#if CBOR_SLOW
828 cbor_true = get_bool ("CBOR::XS::true"); 1326 types_true = get_bool ("Types::Serialiser::true");
829#endif 1327#endif
830 return newSVsv (cbor_true); 1328 return newSVsv (types_true);
831 case 22: 1329 case SIMPLE_NULL:
832 return newSVsv (&PL_sv_undef); 1330 return newSVsv (&PL_sv_undef);
1331 case SIMPLE_UNDEF:
1332#if CBOR_SLOW
1333 types_error = get_bool ("Types::Serialiser::error");
1334#endif
1335 return newSVsv (types_error);
833 1336
834 case 25: 1337 case MISC_FLOAT16:
835 { 1338 {
836 WANT (2); 1339 WANT (2);
837 1340
838 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; 1341 uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
839 dec->cur += 2; 1342 dec->cur += 2;
840 1343
841 return newSVnv (ecb_binary16_to_float (fp)); 1344 return newSVnv (ecb_binary16_to_float (fp));
842 } 1345 }
843 1346
844 case 26: 1347 case MISC_FLOAT32:
845 { 1348 {
846 uint32_t fp; 1349 uint32_t fp;
847 WANT (4); 1350 WANT (4);
848 memcpy (&fp, dec->cur, 4); 1351 memcpy (&fp, dec->cur, 4);
849 dec->cur += 4; 1352 dec->cur += 4;
852 fp = ecb_bswap32 (fp); 1355 fp = ecb_bswap32 (fp);
853 1356
854 return newSVnv (ecb_binary32_to_float (fp)); 1357 return newSVnv (ecb_binary32_to_float (fp));
855 } 1358 }
856 1359
857 case 27: 1360 case MISC_FLOAT64:
858 { 1361 {
859 uint64_t fp; 1362 uint64_t fp;
860 WANT (8); 1363 WANT (8);
861 memcpy (&fp, dec->cur, 8); 1364 memcpy (&fp, dec->cur, 8);
862 dec->cur += 8; 1365 dec->cur += 8;
865 fp = ecb_bswap64 (fp); 1368 fp = ecb_bswap64 (fp);
866 1369
867 return newSVnv (ecb_binary64_to_double (fp)); 1370 return newSVnv (ecb_binary64_to_double (fp));
868 } 1371 }
869 1372
870 // 0..19 unassigned 1373 // 0..19 unassigned simple
871 // 24 reserved + unassigned (reserved values are not encodable) 1374 // 24 reserved + unassigned simple (reserved values are not encodable)
1375 // 28-30 unassigned misc
1376 // 31 break code
872 default: 1377 default:
873 ERR ("corrupted CBOR data (reserved/unassigned major 7 value)"); 1378 ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)");
874 } 1379 }
875 1380
876 break; 1381 break;
877 } 1382 }
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 1383
935fail: 1384fail:
936 return &PL_sv_undef; 1385 return &PL_sv_undef;
937} 1386}
938 1387
939static SV * 1388static SV *
940decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1389decode_cbor (SV *string, CBOR *cbor, char **offset_return)
941{ 1390{
942 dec_t dec; 1391 dec_t dec = { 0 };
943 SV *sv; 1392 SV *sv;
1393 STRLEN len;
1394 char *data = SvPVbyte (string, len);
944 1395
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) 1396 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", 1397 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); 1398 (unsigned long)len, (unsigned long)cbor->max_size);
976 }
977
978 sv_utf8_downgrade (string, 0);
979 1399
980 dec.cbor = *cbor; 1400 dec.cbor = *cbor;
981 dec.cur = (U8 *)SvPVX (string); 1401 dec.cur = (U8 *)data;
982 dec.end = (U8 *)SvEND (string); 1402 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 1403
989 sv = decode_sv (&dec); 1404 sv = decode_sv (&dec);
990 1405
991 if (offset_return) 1406 if (offset_return)
992 *offset_return = dec.cur; 1407 *offset_return = dec.cur;
995 if (dec.cur != dec.end && !dec.err) 1410 if (dec.cur != dec.end && !dec.err)
996 dec.err = "garbage after CBOR object"; 1411 dec.err = "garbage after CBOR object";
997 1412
998 if (dec.err) 1413 if (dec.err)
999 { 1414 {
1415 if (dec.shareable)
1416 {
1417 // need to break cyclic links, which would all be in shareable
1418 int i;
1419 SV **svp;
1420
1421 for (i = av_len (dec.shareable) + 1; i--; )
1422 if ((svp = av_fetch (dec.shareable, i, 0)))
1423 sv_setsv (*svp, &PL_sv_undef);
1424 }
1425
1000 SvREFCNT_dec (sv); 1426 SvREFCNT_dec_NN (sv);
1427
1428 if (dec.err_sv)
1429 sv_2mortal (dec.err_sv);
1430
1001 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)SvPVX (string), (int)(uint8_t)*dec.cur); 1431 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
1002 } 1432 }
1003 1433
1004 sv = sv_2mortal (sv); 1434 sv = sv_2mortal (sv);
1005 1435
1006 return sv; 1436 return sv;
1007} 1437}
1008 1438
1439/////////////////////////////////////////////////////////////////////////////
1440// incremental parser
1441
1442#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0)
1443
1444// returns 0 for notyet, 1 for success or error
1445static int
1446incr_parse (CBOR *self, SV *cborstr)
1447{
1448 STRLEN cur;
1449 SvPV (cborstr, cur);
1450
1451 while (ecb_expect_true (self->incr_need <= cur))
1452 {
1453 // table of integer count bytes
1454 static I8 incr_len[MINOR_MASK + 1] = {
1455 0, 0, 0, 0, 0, 0, 0, 0,
1456 0, 0, 0, 0, 0, 0, 0, 0,
1457 0, 0, 0, 0, 0, 0, 0, 0,
1458 1, 2, 4, 8,-1,-1,-1,-2
1459 };
1460
1461 const U8 *p = SvPVX (cborstr) + self->incr_pos;
1462 U8 m = *p & MINOR_MASK;
1463 IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1464 I8 ilen = incr_len[m];
1465
1466 self->incr_need = self->incr_pos + 1;
1467
1468 if (ecb_expect_false (ilen < 0))
1469 {
1470 if (m != MINOR_INDEF)
1471 return 1; // error
1472
1473 if (*p == (MAJOR_MISC | MINOR_INDEF))
1474 {
1475 if (count >= 0)
1476 return 1; // error
1477
1478 count = 1;
1479 }
1480 else
1481 {
1482 av_push (self->incr_count, newSViv (-1)); //TODO: nest
1483 count = -1;
1484 }
1485 }
1486 else
1487 {
1488 self->incr_need += ilen;
1489 if (ecb_expect_false (self->incr_need > cur))
1490 return 0;
1491
1492 int major = *p >> MAJOR_SHIFT;
1493
1494 switch (major)
1495 {
1496 case MAJOR_TAG >> MAJOR_SHIFT:
1497 ++count; // tags merely prefix another value
1498 break;
1499
1500 case MAJOR_BYTES >> MAJOR_SHIFT:
1501 case MAJOR_TEXT >> MAJOR_SHIFT:
1502 case MAJOR_ARRAY >> MAJOR_SHIFT:
1503 case MAJOR_MAP >> MAJOR_SHIFT:
1504 {
1505 UV len;
1506
1507 if (ecb_expect_false (ilen))
1508 {
1509 len = 0;
1510
1511 do {
1512 len = (len << 8) | *++p;
1513 } while (--ilen);
1514 }
1515 else
1516 len = m;
1517
1518 switch (major)
1519 {
1520 case MAJOR_BYTES >> MAJOR_SHIFT:
1521 case MAJOR_TEXT >> MAJOR_SHIFT:
1522 self->incr_need += len;
1523 if (ecb_expect_false (self->incr_need > cur))
1524 return 0;
1525
1526 break;
1527
1528 case MAJOR_MAP >> MAJOR_SHIFT:
1529 len <<= 1;
1530 case MAJOR_ARRAY >> MAJOR_SHIFT:
1531 if (len)
1532 {
1533 av_push (self->incr_count, newSViv (len + 1)); //TODO: nest
1534 count = len + 1;
1535 }
1536 break;
1537 }
1538 }
1539 }
1540 }
1541
1542 self->incr_pos = self->incr_need;
1543
1544 if (count > 0)
1545 {
1546 while (!--count)
1547 {
1548 if (!AvFILLp (self->incr_count))
1549 return 1; // done
1550
1551 SvREFCNT_dec_NN (av_pop (self->incr_count));
1552 count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]);
1553 }
1554
1555 SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count;
1556 }
1557 }
1558
1559 return 0;
1560}
1561
1562
1009///////////////////////////////////////////////////////////////////////////// 1563/////////////////////////////////////////////////////////////////////////////
1010// XS interface functions 1564// XS interface functions
1011 1565
1012MODULE = CBOR::XS PACKAGE = CBOR::XS 1566MODULE = CBOR::XS PACKAGE = CBOR::XS
1013 1567
1014BOOT: 1568BOOT:
1015{ 1569{
1016 cbor_stash = gv_stashpv ("CBOR::XS" , 1); 1570 cbor_stash = gv_stashpv ("CBOR::XS" , 1);
1017 cbor_boolean_stash = gv_stashpv ("CBOR::XS::Boolean", 1); 1571 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1);
1018 1572
1019 cbor_true = get_bool ("CBOR::XS::true"); 1573 types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
1020 cbor_false = get_bool ("CBOR::XS::false"); 1574 types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
1575
1576 types_true = get_bool ("Types::Serialiser::true" );
1577 types_false = get_bool ("Types::Serialiser::false");
1578 types_error = get_bool ("Types::Serialiser::error");
1579
1580 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1581
1582 sv_cbor = newSVpv ("CBOR", 0);
1583 SvREADONLY_on (sv_cbor);
1584
1585 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1021} 1586}
1022 1587
1023PROTOTYPES: DISABLE 1588PROTOTYPES: DISABLE
1024 1589
1025void CLONE (...) 1590void CLONE (...)
1026 CODE: 1591 CODE:
1027 cbor_stash = 0; 1592 cbor_stash = 0;
1593 cbor_tagged_stash = 0;
1594 types_error_stash = 0;
1028 cbor_boolean_stash = 0; 1595 types_boolean_stash = 0;
1029 1596
1030void new (char *klass) 1597void new (char *klass)
1031 PPCODE: 1598 PPCODE:
1032{ 1599{
1033 SV *pv = NEWSV (0, sizeof (CBOR)); 1600 SV *pv = NEWSV (0, sizeof (CBOR));
1041 1608
1042void shrink (CBOR *self, int enable = 1) 1609void shrink (CBOR *self, int enable = 1)
1043 ALIAS: 1610 ALIAS:
1044 shrink = F_SHRINK 1611 shrink = F_SHRINK
1045 allow_unknown = F_ALLOW_UNKNOWN 1612 allow_unknown = F_ALLOW_UNKNOWN
1613 allow_sharing = F_ALLOW_SHARING
1614 allow_cycles = F_ALLOW_CYCLES
1615 forbid_objects = F_FORBID_OBJECTS
1616 pack_strings = F_PACK_STRINGS
1617 text_keys = F_TEXT_KEYS
1618 text_strings = F_TEXT_STRINGS
1619 validate_utf8 = F_VALIDATE_UTF8
1046 PPCODE: 1620 PPCODE:
1047{ 1621{
1048 if (enable) 1622 if (enable)
1049 self->flags |= ix; 1623 self->flags |= ix;
1050 else 1624 else
1055 1629
1056void get_shrink (CBOR *self) 1630void get_shrink (CBOR *self)
1057 ALIAS: 1631 ALIAS:
1058 get_shrink = F_SHRINK 1632 get_shrink = F_SHRINK
1059 get_allow_unknown = F_ALLOW_UNKNOWN 1633 get_allow_unknown = F_ALLOW_UNKNOWN
1634 get_allow_sharing = F_ALLOW_SHARING
1635 get_allow_cycles = F_ALLOW_CYCLES
1636 get_forbid_objects = F_FORBID_OBJECTS
1637 get_pack_strings = F_PACK_STRINGS
1638 get_text_keys = F_TEXT_KEYS
1639 get_text_strings = F_TEXT_STRINGS
1640 get_validate_utf8 = F_VALIDATE_UTF8
1060 PPCODE: 1641 PPCODE:
1061 XPUSHs (boolSV (self->flags & ix)); 1642 XPUSHs (boolSV (self->flags & ix));
1062 1643
1063void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1644void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1064 PPCODE: 1645 PPCODE:
1080 CODE: 1661 CODE:
1081 RETVAL = self->max_size; 1662 RETVAL = self->max_size;
1082 OUTPUT: 1663 OUTPUT:
1083 RETVAL 1664 RETVAL
1084 1665
1085#if 0 //TODO 1666void filter (CBOR *self, SV *filter = 0)
1086
1087void filter_cbor_object (CBOR *self, SV *cb = &PL_sv_undef)
1088 PPCODE: 1667 PPCODE:
1089{
1090 SvREFCNT_dec (self->cb_object); 1668 SvREFCNT_dec (self->filter);
1091 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; 1669 self->filter = filter ? newSVsv (filter) : filter;
1092
1093 XPUSHs (ST (0)); 1670 XPUSHs (ST (0));
1094}
1095 1671
1096void filter_cbor_single_key_object (CBOR *self, SV *key, SV *cb = &PL_sv_undef) 1672SV *get_filter (CBOR *self)
1097 PPCODE: 1673 CODE:
1098{ 1674 RETVAL = self->filter ? self->filter : NEWSV (0, 0);
1099 if (!self->cb_sk_object) 1675 OUTPUT:
1100 self->cb_sk_object = newHV (); 1676 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 1677
1120void encode (CBOR *self, SV *scalar) 1678void encode (CBOR *self, SV *scalar)
1121 PPCODE: 1679 PPCODE:
1122 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; 1680 PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
1123 XPUSHs (scalar); 1681 XPUSHs (scalar);
1136 EXTEND (SP, 2); 1694 EXTEND (SP, 2);
1137 PUSHs (sv); 1695 PUSHs (sv);
1138 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1696 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1139} 1697}
1140 1698
1699void incr_parse (CBOR *self, SV *cborstr)
1700 ALIAS:
1701 incr_parse_multiple = 1
1702 PPCODE:
1703{
1704 if (SvUTF8 (cborstr))
1705 sv_utf8_downgrade (cborstr, 0);
1706
1707 if (!self->incr_count)
1708 {
1709 self->incr_count = newAV ();
1710 self->incr_pos = 0;
1711 self->incr_need = 1;
1712
1713 av_push (self->incr_count, newSViv (1));
1714 }
1715
1716 do
1717 {
1718 if (!incr_parse (self, cborstr))
1719 {
1720 if (self->incr_need > self->max_size && self->max_size)
1721 croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1722 (unsigned long)self->incr_need, (unsigned long)self->max_size);
1723
1724 break;
1725 }
1726
1727 SV *sv;
1728 char *offset;
1729
1730 PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1731 XPUSHs (sv);
1732
1733 sv_chop (cborstr, offset);
1734
1735 av_clear (self->incr_count);
1736 av_push (self->incr_count, newSViv (1));
1737
1738 self->incr_pos = 0;
1739 self->incr_need = self->incr_pos + 1;
1740 }
1741 while (ix);
1742}
1743
1744void incr_reset (CBOR *self)
1745 CODE:
1746{
1747 SvREFCNT_dec (self->incr_count);
1748 self->incr_count = 0;
1749}
1750
1141void DESTROY (CBOR *self) 1751void DESTROY (CBOR *self)
1142 CODE: 1752 PPCODE:
1143 SvREFCNT_dec (self->cb_sk_object); 1753 cbor_free (self);
1144 SvREFCNT_dec (self->cb_object);
1145 1754
1146PROTOTYPES: ENABLE 1755PROTOTYPES: ENABLE
1147 1756
1148void encode_cbor (SV *scalar) 1757void encode_cbor (SV *scalar)
1758 ALIAS:
1759 encode_cbor = 0
1760 encode_cbor_sharing = F_ALLOW_SHARING
1149 PPCODE: 1761 PPCODE:
1150{ 1762{
1151 CBOR cbor; 1763 CBOR cbor;
1152 cbor_init (&cbor); 1764 cbor_init (&cbor);
1765 cbor.flags |= ix;
1153 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; 1766 PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1154 XPUSHs (scalar); 1767 XPUSHs (scalar);
1155} 1768}
1156 1769
1157void decode_cbor (SV *cborstr) 1770void decode_cbor (SV *cborstr)
1161 cbor_init (&cbor); 1774 cbor_init (&cbor);
1162 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN; 1775 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;
1163 XPUSHs (cborstr); 1776 XPUSHs (cborstr);
1164} 1777}
1165 1778
1779#ifdef __AFL_COMPILER
1780
1781void
1782afl_init ()
1783 CODE:
1784 __AFL_INIT ();
1785
1786int
1787afl_loop (unsigned int count = 10000)
1788 CODE:
1789 RETVAL = __AFL_LOOP (count);
1790 OUTPUT:
1791 RETVAL
1792
1793#endif
1794

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines