| 1 |
#include "EXTERN.h" |
| 2 |
#include "perl.h" |
| 3 |
#include "XSUB.h" |
| 4 |
|
| 5 |
#include <assert.h> |
| 6 |
#include <string.h> |
| 7 |
#include <stdlib.h> |
| 8 |
#include <stdio.h> |
| 9 |
#include <limits.h> |
| 10 |
#include <float.h> |
| 11 |
|
| 12 |
#define ECB_NO_THREADS 1 |
| 13 |
#include "ecb.h" |
| 14 |
|
| 15 |
// compatibility with perl <5.18 |
| 16 |
#ifndef HvNAMELEN_get |
| 17 |
# define HvNAMELEN_get(hv) strlen (HvNAME (hv)) |
| 18 |
#endif |
| 19 |
#ifndef HvNAMELEN |
| 20 |
# define HvNAMELEN(hv) HvNAMELEN_get (hv) |
| 21 |
#endif |
| 22 |
#ifndef HvNAMEUTF8 |
| 23 |
# define HvNAMEUTF8(hv) 0 |
| 24 |
#endif |
| 25 |
#ifndef SvREFCNT_dec_NN |
| 26 |
# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv) |
| 27 |
#endif |
| 28 |
|
| 29 |
// known major and minor types |
| 30 |
enum cbor_type |
| 31 |
{ |
| 32 |
MAJOR_SHIFT = 5, |
| 33 |
MINOR_MASK = 0x1f, |
| 34 |
|
| 35 |
MAJOR_POS_INT = 0 << MAJOR_SHIFT, |
| 36 |
MAJOR_NEG_INT = 1 << MAJOR_SHIFT, |
| 37 |
MAJOR_BYTES = 2 << MAJOR_SHIFT, |
| 38 |
MAJOR_TEXT = 3 << MAJOR_SHIFT, |
| 39 |
MAJOR_ARRAY = 4 << MAJOR_SHIFT, |
| 40 |
MAJOR_MAP = 5 << MAJOR_SHIFT, |
| 41 |
MAJOR_TAG = 6 << MAJOR_SHIFT, |
| 42 |
MAJOR_MISC = 7 << MAJOR_SHIFT, |
| 43 |
|
| 44 |
// INT/STRING/ARRAY/MAP subtypes |
| 45 |
LENGTH_EXT1 = 24, |
| 46 |
LENGTH_EXT2 = 25, |
| 47 |
LENGTH_EXT4 = 26, |
| 48 |
LENGTH_EXT8 = 27, |
| 49 |
|
| 50 |
// SIMPLE types (effectively MISC subtypes) |
| 51 |
SIMPLE_FALSE = 20, |
| 52 |
SIMPLE_TRUE = 21, |
| 53 |
SIMPLE_NULL = 22, |
| 54 |
SIMPLE_UNDEF = 23, |
| 55 |
|
| 56 |
// MISC subtype (unused) |
| 57 |
MISC_EXT1 = 24, |
| 58 |
MISC_FLOAT16 = 25, |
| 59 |
MISC_FLOAT32 = 26, |
| 60 |
MISC_FLOAT64 = 27, |
| 61 |
|
| 62 |
// BYTES/TEXT/ARRAY/MAP |
| 63 |
MINOR_INDEF = 31, |
| 64 |
}; |
| 65 |
|
| 66 |
// known tags |
| 67 |
enum cbor_tag |
| 68 |
{ |
| 69 |
// extensions |
| 70 |
CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref |
| 71 |
CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object |
| 72 |
CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object |
| 73 |
CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing |
| 74 |
CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing |
| 75 |
CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref |
| 76 |
CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection |
| 77 |
|
| 78 |
// rfc7049 |
| 79 |
CBOR_TAG_DATETIME = 0, // rfc4287, utf-8 |
| 80 |
CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any |
| 81 |
CBOR_TAG_POS_BIGNUM = 2, // byte string |
| 82 |
CBOR_TAG_NEG_BIGNUM = 3, // byte string |
| 83 |
CBOR_TAG_DECIMAL = 4, // decimal fraction, array |
| 84 |
CBOR_TAG_BIGFLOAT = 5, // array |
| 85 |
|
| 86 |
CBOR_TAG_CONV_B64U = 21, // base64url, any |
| 87 |
CBOR_TAG_CONV_B64 = 22, // base64, any |
| 88 |
CBOR_TAG_CONV_HEX = 23, // base16, any |
| 89 |
CBOR_TAG_CBOR = 24, // embedded cbor, byte string |
| 90 |
|
| 91 |
CBOR_TAG_URI = 32, // URI rfc3986, utf-8 |
| 92 |
CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8 |
| 93 |
CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8 |
| 94 |
CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8 |
| 95 |
CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8 |
| 96 |
|
| 97 |
CBOR_TAG_MAGIC = 55799, // self-describe cbor |
| 98 |
}; |
| 99 |
|
| 100 |
#define F_SHRINK 0x00000001UL |
| 101 |
#define F_ALLOW_UNKNOWN 0x00000002UL |
| 102 |
#define F_ALLOW_SHARING 0x00000004UL |
| 103 |
#define F_ALLOW_CYCLES 0x00000008UL |
| 104 |
#define F_PACK_STRINGS 0x00000010UL |
| 105 |
#define F_UTF8_STRINGS 0x00000020UL |
| 106 |
#define F_VALIDATE_UTF8 0x00000040UL |
| 107 |
|
| 108 |
#define INIT_SIZE 32 // initial scalar size to be allocated |
| 109 |
|
| 110 |
#define SB do { |
| 111 |
#define SE } while (0) |
| 112 |
|
| 113 |
#define IN_RANGE_INC(type,val,beg,end) \ |
| 114 |
((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \ |
| 115 |
<= (unsigned type)((unsigned type)(end) - (unsigned type)(beg))) |
| 116 |
|
| 117 |
#define ERR_NESTING_EXCEEDED "cbor text or perl structure exceeds maximum nesting level (max_depth set too low?)" |
| 118 |
|
| 119 |
#ifdef USE_ITHREADS |
| 120 |
# define CBOR_SLOW 1 |
| 121 |
# define CBOR_STASH (cbor_stash ? cbor_stash : gv_stashpv ("CBOR::XS", 1)) |
| 122 |
#else |
| 123 |
# define CBOR_SLOW 0 |
| 124 |
# define CBOR_STASH cbor_stash |
| 125 |
#endif |
| 126 |
|
| 127 |
static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS:: |
| 128 |
static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter; |
| 129 |
|
| 130 |
typedef struct { |
| 131 |
U32 flags; |
| 132 |
U32 max_depth; |
| 133 |
STRLEN max_size; |
| 134 |
SV *filter; |
| 135 |
|
| 136 |
// for the incremental parser |
| 137 |
STRLEN incr_pos; // the current offset into the text |
| 138 |
STRLEN incr_need; // minimum bytes needed to decode |
| 139 |
AV *incr_count; // for every nesting level, the number of outstanding values, or -1 for indef. |
| 140 |
} CBOR; |
| 141 |
|
| 142 |
ecb_inline void |
| 143 |
cbor_init (CBOR *cbor) |
| 144 |
{ |
| 145 |
Zero (cbor, 1, CBOR); |
| 146 |
cbor->max_depth = 512; |
| 147 |
} |
| 148 |
|
| 149 |
ecb_inline void |
| 150 |
cbor_free (CBOR *cbor) |
| 151 |
{ |
| 152 |
SvREFCNT_dec (cbor->filter); |
| 153 |
SvREFCNT_dec (cbor->incr_count); |
| 154 |
} |
| 155 |
|
| 156 |
///////////////////////////////////////////////////////////////////////////// |
| 157 |
// utility functions |
| 158 |
|
| 159 |
ecb_inline SV * |
| 160 |
get_bool (const char *name) |
| 161 |
{ |
| 162 |
SV *sv = get_sv (name, 1); |
| 163 |
|
| 164 |
SvREADONLY_on (sv); |
| 165 |
SvREADONLY_on (SvRV (sv)); |
| 166 |
|
| 167 |
return sv; |
| 168 |
} |
| 169 |
|
| 170 |
ecb_inline void |
| 171 |
shrink (SV *sv) |
| 172 |
{ |
| 173 |
sv_utf8_downgrade (sv, 1); |
| 174 |
|
| 175 |
if (SvLEN (sv) > SvCUR (sv) + 1) |
| 176 |
{ |
| 177 |
#ifdef SvPV_shrink_to_cur |
| 178 |
SvPV_shrink_to_cur (sv); |
| 179 |
#elif defined (SvPV_renew) |
| 180 |
SvPV_renew (sv, SvCUR (sv) + 1); |
| 181 |
#endif |
| 182 |
} |
| 183 |
} |
| 184 |
|
| 185 |
// minimum length of a string to be registered for stringref |
| 186 |
ecb_inline int |
| 187 |
minimum_string_length (UV idx) |
| 188 |
{ |
| 189 |
return idx > 23 |
| 190 |
? idx > 0xffU |
| 191 |
? idx > 0xffffU |
| 192 |
? idx > 0xffffffffU |
| 193 |
? 11 |
| 194 |
: 7 |
| 195 |
: 5 |
| 196 |
: 4 |
| 197 |
: 3; |
| 198 |
} |
| 199 |
|
| 200 |
///////////////////////////////////////////////////////////////////////////// |
| 201 |
// encoder |
| 202 |
|
| 203 |
// structure used for encoding CBOR |
| 204 |
typedef struct |
| 205 |
{ |
| 206 |
char *cur; // SvPVX (sv) + current output position |
| 207 |
char *end; // SvEND (sv) |
| 208 |
SV *sv; // result scalar |
| 209 |
CBOR cbor; |
| 210 |
U32 depth; // recursion level |
| 211 |
HV *stringref[2]; // string => index, or 0 ([0] = bytes, [1] = utf-8) |
| 212 |
UV stringref_idx; |
| 213 |
HV *shareable; // ptr => index, or 0 |
| 214 |
UV shareable_idx; |
| 215 |
} enc_t; |
| 216 |
|
| 217 |
ecb_inline void |
| 218 |
need (enc_t *enc, STRLEN len) |
| 219 |
{ |
| 220 |
if (ecb_expect_false (enc->cur + len >= enc->end)) |
| 221 |
{ |
| 222 |
STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); |
| 223 |
SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); |
| 224 |
enc->cur = SvPVX (enc->sv) + cur; |
| 225 |
enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; |
| 226 |
} |
| 227 |
} |
| 228 |
|
| 229 |
ecb_inline void |
| 230 |
encode_ch (enc_t *enc, char ch) |
| 231 |
{ |
| 232 |
need (enc, 1); |
| 233 |
*enc->cur++ = ch; |
| 234 |
} |
| 235 |
|
| 236 |
static void |
| 237 |
encode_uint (enc_t *enc, int major, UV len) |
| 238 |
{ |
| 239 |
need (enc, 9); |
| 240 |
|
| 241 |
if (ecb_expect_true (len < LENGTH_EXT1)) |
| 242 |
*enc->cur++ = major | len; |
| 243 |
else if (ecb_expect_true (len <= 0xffU)) |
| 244 |
{ |
| 245 |
*enc->cur++ = major | LENGTH_EXT1; |
| 246 |
*enc->cur++ = len; |
| 247 |
} |
| 248 |
else if (len <= 0xffffU) |
| 249 |
{ |
| 250 |
*enc->cur++ = major | LENGTH_EXT2; |
| 251 |
*enc->cur++ = len >> 8; |
| 252 |
*enc->cur++ = len; |
| 253 |
} |
| 254 |
else if (len <= 0xffffffffU) |
| 255 |
{ |
| 256 |
*enc->cur++ = major | LENGTH_EXT4; |
| 257 |
*enc->cur++ = len >> 24; |
| 258 |
*enc->cur++ = len >> 16; |
| 259 |
*enc->cur++ = len >> 8; |
| 260 |
*enc->cur++ = len; |
| 261 |
} |
| 262 |
else |
| 263 |
{ |
| 264 |
*enc->cur++ = major | LENGTH_EXT8; |
| 265 |
*enc->cur++ = len >> 56; |
| 266 |
*enc->cur++ = len >> 48; |
| 267 |
*enc->cur++ = len >> 40; |
| 268 |
*enc->cur++ = len >> 32; |
| 269 |
*enc->cur++ = len >> 24; |
| 270 |
*enc->cur++ = len >> 16; |
| 271 |
*enc->cur++ = len >> 8; |
| 272 |
*enc->cur++ = len; |
| 273 |
} |
| 274 |
} |
| 275 |
|
| 276 |
ecb_inline void |
| 277 |
encode_tag (enc_t *enc, UV tag) |
| 278 |
{ |
| 279 |
encode_uint (enc, MAJOR_TAG, tag); |
| 280 |
} |
| 281 |
|
| 282 |
ecb_inline void |
| 283 |
encode_str (enc_t *enc, int utf8, char *str, STRLEN len) |
| 284 |
{ |
| 285 |
if (ecb_expect_false (enc->cbor.flags & F_UTF8_STRINGS)) |
| 286 |
if (!utf8) |
| 287 |
{ |
| 288 |
// exceptional path for bytze strings that need to be utf8-encoded |
| 289 |
STRLEN ulen = len; |
| 290 |
U8 *p, *pend = (U8 *)str + len; |
| 291 |
|
| 292 |
for (p = (U8 *)str; p < pend; ++p) |
| 293 |
ulen += *p >> 7; // count set high bits |
| 294 |
|
| 295 |
encode_uint (enc, MAJOR_TEXT, ulen); |
| 296 |
|
| 297 |
need (enc, ulen); |
| 298 |
for (p = (U8 *)str; p < pend; ++p) |
| 299 |
if (*p < 0x80) |
| 300 |
*enc->cur++ = *p; |
| 301 |
else |
| 302 |
{ |
| 303 |
*enc->cur++ = 0xc0 + (*p >> 6); |
| 304 |
*enc->cur++ = 0x80 + (*p & 63); |
| 305 |
} |
| 306 |
|
| 307 |
return; |
| 308 |
} |
| 309 |
|
| 310 |
encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len); |
| 311 |
need (enc, len); |
| 312 |
memcpy (enc->cur, str, len); |
| 313 |
enc->cur += len; |
| 314 |
} |
| 315 |
|
| 316 |
static void |
| 317 |
encode_strref (enc_t *enc, int utf8, char *str, STRLEN len) |
| 318 |
{ |
| 319 |
if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS)) |
| 320 |
{ |
| 321 |
SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1); |
| 322 |
|
| 323 |
if (SvOK (*svp)) |
| 324 |
{ |
| 325 |
// already registered, use stringref |
| 326 |
encode_tag (enc, CBOR_TAG_STRINGREF); |
| 327 |
encode_uint (enc, MAJOR_POS_INT, SvUV (*svp)); |
| 328 |
return; |
| 329 |
} |
| 330 |
else if (len >= minimum_string_length (enc->stringref_idx)) |
| 331 |
{ |
| 332 |
// register only |
| 333 |
sv_setuv (*svp, enc->stringref_idx); |
| 334 |
++enc->stringref_idx; |
| 335 |
} |
| 336 |
} |
| 337 |
|
| 338 |
encode_str (enc, utf8, str, len); |
| 339 |
} |
| 340 |
|
| 341 |
static void encode_sv (enc_t *enc, SV *sv); |
| 342 |
|
| 343 |
static void |
| 344 |
encode_av (enc_t *enc, AV *av) |
| 345 |
{ |
| 346 |
int i, len = av_len (av); |
| 347 |
|
| 348 |
if (enc->depth >= enc->cbor.max_depth) |
| 349 |
croak (ERR_NESTING_EXCEEDED); |
| 350 |
|
| 351 |
++enc->depth; |
| 352 |
|
| 353 |
encode_uint (enc, MAJOR_ARRAY, len + 1); |
| 354 |
|
| 355 |
if (SvMAGICAL (av)) |
| 356 |
for (i = 0; i <= len; ++i) |
| 357 |
{ |
| 358 |
SV **svp = av_fetch (av, i, 0); |
| 359 |
encode_sv (enc, svp ? *svp : &PL_sv_undef); |
| 360 |
} |
| 361 |
else |
| 362 |
for (i = 0; i <= len; ++i) |
| 363 |
{ |
| 364 |
SV *sv = AvARRAY (av)[i]; |
| 365 |
encode_sv (enc, sv ? sv : &PL_sv_undef); |
| 366 |
} |
| 367 |
|
| 368 |
--enc->depth; |
| 369 |
} |
| 370 |
|
| 371 |
static void |
| 372 |
encode_hv (enc_t *enc, HV *hv) |
| 373 |
{ |
| 374 |
HE *he; |
| 375 |
|
| 376 |
if (enc->depth >= enc->cbor.max_depth) |
| 377 |
croak (ERR_NESTING_EXCEEDED); |
| 378 |
|
| 379 |
++enc->depth; |
| 380 |
|
| 381 |
int pairs = hv_iterinit (hv); |
| 382 |
int mg = SvMAGICAL (hv); |
| 383 |
|
| 384 |
if (mg) |
| 385 |
encode_ch (enc, MAJOR_MAP | MINOR_INDEF); |
| 386 |
else |
| 387 |
encode_uint (enc, MAJOR_MAP, pairs); |
| 388 |
|
| 389 |
while ((he = hv_iternext (hv))) |
| 390 |
{ |
| 391 |
if (HeKLEN (he) == HEf_SVKEY) |
| 392 |
encode_sv (enc, HeSVKEY (he)); |
| 393 |
else |
| 394 |
encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); |
| 395 |
|
| 396 |
encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); |
| 397 |
} |
| 398 |
|
| 399 |
if (mg) |
| 400 |
encode_ch (enc, MAJOR_MISC | MINOR_INDEF); |
| 401 |
|
| 402 |
--enc->depth; |
| 403 |
} |
| 404 |
|
| 405 |
// encode objects, arrays and special \0=false and \1=true values. |
| 406 |
static void |
| 407 |
encode_rv (enc_t *enc, SV *sv) |
| 408 |
{ |
| 409 |
SvGETMAGIC (sv); |
| 410 |
|
| 411 |
svtype svt = SvTYPE (sv); |
| 412 |
|
| 413 |
if (ecb_expect_false (SvOBJECT (sv))) |
| 414 |
{ |
| 415 |
HV *boolean_stash = !CBOR_SLOW || types_boolean_stash |
| 416 |
? types_boolean_stash |
| 417 |
: gv_stashpv ("Types::Serialiser::Boolean", 1); |
| 418 |
HV *error_stash = !CBOR_SLOW || types_error_stash |
| 419 |
? types_error_stash |
| 420 |
: gv_stashpv ("Types::Serialiser::Error", 1); |
| 421 |
HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash |
| 422 |
? cbor_tagged_stash |
| 423 |
: gv_stashpv ("CBOR::XS::Tagged" , 1); |
| 424 |
|
| 425 |
HV *stash = SvSTASH (sv); |
| 426 |
|
| 427 |
if (stash == boolean_stash) |
| 428 |
{ |
| 429 |
encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE); |
| 430 |
return; |
| 431 |
} |
| 432 |
else if (stash == error_stash) |
| 433 |
{ |
| 434 |
encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF); |
| 435 |
return; |
| 436 |
} |
| 437 |
else if (stash == tagged_stash) |
| 438 |
{ |
| 439 |
if (svt != SVt_PVAV) |
| 440 |
croak ("encountered CBOR::XS::Tagged object that isn't an array"); |
| 441 |
|
| 442 |
encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1))); |
| 443 |
encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); |
| 444 |
|
| 445 |
return; |
| 446 |
} |
| 447 |
} |
| 448 |
|
| 449 |
if (ecb_expect_false (SvREFCNT (sv) > 1) |
| 450 |
&& ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)) |
| 451 |
{ |
| 452 |
if (!enc->shareable) |
| 453 |
enc->shareable = (HV *)sv_2mortal ((SV *)newHV ()); |
| 454 |
|
| 455 |
SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1); |
| 456 |
|
| 457 |
if (SvOK (*svp)) |
| 458 |
{ |
| 459 |
encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF); |
| 460 |
encode_uint (enc, MAJOR_POS_INT, SvUV (*svp)); |
| 461 |
return; |
| 462 |
} |
| 463 |
else |
| 464 |
{ |
| 465 |
sv_setuv (*svp, enc->shareable_idx); |
| 466 |
++enc->shareable_idx; |
| 467 |
encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE); |
| 468 |
} |
| 469 |
} |
| 470 |
|
| 471 |
if (ecb_expect_false (SvOBJECT (sv))) |
| 472 |
{ |
| 473 |
HV *stash = SvSTASH (sv); |
| 474 |
GV *method; |
| 475 |
|
| 476 |
if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) |
| 477 |
{ |
| 478 |
dSP; |
| 479 |
|
| 480 |
ENTER; SAVETMPS; |
| 481 |
PUSHMARK (SP); |
| 482 |
// we re-bless the reference to get overload and other niceties right |
| 483 |
XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); |
| 484 |
|
| 485 |
PUTBACK; |
| 486 |
// G_SCALAR ensures that return value is 1 |
| 487 |
call_sv ((SV *)GvCV (method), G_SCALAR); |
| 488 |
SPAGAIN; |
| 489 |
|
| 490 |
// catch this surprisingly common error |
| 491 |
if (SvROK (TOPs) && SvRV (TOPs) == sv) |
| 492 |
croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (stash)); |
| 493 |
|
| 494 |
encode_sv (enc, POPs); |
| 495 |
|
| 496 |
PUTBACK; |
| 497 |
|
| 498 |
FREETMPS; LEAVE; |
| 499 |
} |
| 500 |
else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) |
| 501 |
{ |
| 502 |
dSP; |
| 503 |
|
| 504 |
ENTER; SAVETMPS; |
| 505 |
SAVESTACK_POS (); |
| 506 |
PUSHMARK (SP); |
| 507 |
EXTEND (SP, 2); |
| 508 |
// we re-bless the reference to get overload and other niceties right |
| 509 |
PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); |
| 510 |
PUSHs (sv_cbor); |
| 511 |
|
| 512 |
PUTBACK; |
| 513 |
int count = call_sv ((SV *)GvCV (method), G_ARRAY); |
| 514 |
SPAGAIN; |
| 515 |
|
| 516 |
// catch this surprisingly common error |
| 517 |
if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) |
| 518 |
croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); |
| 519 |
|
| 520 |
encode_tag (enc, CBOR_TAG_PERL_OBJECT); |
| 521 |
encode_uint (enc, MAJOR_ARRAY, count + 1); |
| 522 |
encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); |
| 523 |
|
| 524 |
while (count) |
| 525 |
encode_sv (enc, SP[1 - count--]); |
| 526 |
|
| 527 |
PUTBACK; |
| 528 |
|
| 529 |
FREETMPS; LEAVE; |
| 530 |
} |
| 531 |
else |
| 532 |
croak ("encountered object '%s', but no TO_CBOR or FREEZE methods available on it", |
| 533 |
SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
| 534 |
} |
| 535 |
else if (svt == SVt_PVHV) |
| 536 |
encode_hv (enc, (HV *)sv); |
| 537 |
else if (svt == SVt_PVAV) |
| 538 |
encode_av (enc, (AV *)sv); |
| 539 |
else |
| 540 |
{ |
| 541 |
encode_tag (enc, CBOR_TAG_INDIRECTION); |
| 542 |
encode_sv (enc, sv); |
| 543 |
} |
| 544 |
} |
| 545 |
|
| 546 |
static void |
| 547 |
encode_nv (enc_t *enc, SV *sv) |
| 548 |
{ |
| 549 |
double nv = SvNVX (sv); |
| 550 |
|
| 551 |
need (enc, 9); |
| 552 |
|
| 553 |
if (ecb_expect_false (nv == (NV)(U32)nv)) |
| 554 |
encode_uint (enc, MAJOR_POS_INT, (U32)nv); |
| 555 |
//TODO: maybe I32? |
| 556 |
else if (ecb_expect_false (nv == (float)nv)) |
| 557 |
{ |
| 558 |
uint32_t fp = ecb_float_to_binary32 (nv); |
| 559 |
|
| 560 |
*enc->cur++ = MAJOR_MISC | MISC_FLOAT32; |
| 561 |
|
| 562 |
if (!ecb_big_endian ()) |
| 563 |
fp = ecb_bswap32 (fp); |
| 564 |
|
| 565 |
memcpy (enc->cur, &fp, 4); |
| 566 |
enc->cur += 4; |
| 567 |
} |
| 568 |
else |
| 569 |
{ |
| 570 |
uint64_t fp = ecb_double_to_binary64 (nv); |
| 571 |
|
| 572 |
*enc->cur++ = MAJOR_MISC | MISC_FLOAT64; |
| 573 |
|
| 574 |
if (!ecb_big_endian ()) |
| 575 |
fp = ecb_bswap64 (fp); |
| 576 |
|
| 577 |
memcpy (enc->cur, &fp, 8); |
| 578 |
enc->cur += 8; |
| 579 |
} |
| 580 |
} |
| 581 |
|
| 582 |
static void |
| 583 |
encode_sv (enc_t *enc, SV *sv) |
| 584 |
{ |
| 585 |
SvGETMAGIC (sv); |
| 586 |
|
| 587 |
if (SvPOKp (sv)) |
| 588 |
{ |
| 589 |
STRLEN len; |
| 590 |
char *str = SvPV (sv, len); |
| 591 |
encode_strref (enc, SvUTF8 (sv), str, len); |
| 592 |
} |
| 593 |
else if (SvNOKp (sv)) |
| 594 |
encode_nv (enc, sv); |
| 595 |
else if (SvIOKp (sv)) |
| 596 |
{ |
| 597 |
if (SvIsUV (sv)) |
| 598 |
encode_uint (enc, MAJOR_POS_INT, SvUVX (sv)); |
| 599 |
else if (SvIVX (sv) >= 0) |
| 600 |
encode_uint (enc, MAJOR_POS_INT, SvIVX (sv)); |
| 601 |
else |
| 602 |
encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1)); |
| 603 |
} |
| 604 |
else if (SvROK (sv)) |
| 605 |
encode_rv (enc, SvRV (sv)); |
| 606 |
else if (!SvOK (sv)) |
| 607 |
encode_ch (enc, MAJOR_MISC | SIMPLE_NULL); |
| 608 |
else if (enc->cbor.flags & F_ALLOW_UNKNOWN) |
| 609 |
encode_ch (enc, MAJOR_MISC | SIMPLE_UNDEF); |
| 610 |
else |
| 611 |
croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data", |
| 612 |
SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); |
| 613 |
} |
| 614 |
|
| 615 |
static SV * |
| 616 |
encode_cbor (SV *scalar, CBOR *cbor) |
| 617 |
{ |
| 618 |
enc_t enc = { 0 }; |
| 619 |
|
| 620 |
enc.cbor = *cbor; |
| 621 |
enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
| 622 |
enc.cur = SvPVX (enc.sv); |
| 623 |
enc.end = SvEND (enc.sv); |
| 624 |
|
| 625 |
SvPOK_only (enc.sv); |
| 626 |
|
| 627 |
if (cbor->flags & F_PACK_STRINGS) |
| 628 |
{ |
| 629 |
encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE); |
| 630 |
enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ()); |
| 631 |
enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ()); |
| 632 |
} |
| 633 |
|
| 634 |
encode_sv (&enc, scalar); |
| 635 |
|
| 636 |
SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); |
| 637 |
*SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings |
| 638 |
|
| 639 |
if (enc.cbor.flags & F_SHRINK) |
| 640 |
shrink (enc.sv); |
| 641 |
|
| 642 |
return enc.sv; |
| 643 |
} |
| 644 |
|
| 645 |
///////////////////////////////////////////////////////////////////////////// |
| 646 |
// decoder |
| 647 |
|
| 648 |
// structure used for decoding CBOR |
| 649 |
typedef struct |
| 650 |
{ |
| 651 |
U8 *cur; // current parser pointer |
| 652 |
U8 *end; // end of input string |
| 653 |
const char *err; // parse error, if != 0 |
| 654 |
CBOR cbor; |
| 655 |
U32 depth; // recursion depth |
| 656 |
U32 maxdepth; // recursion depth limit |
| 657 |
AV *shareable; |
| 658 |
AV *stringref; |
| 659 |
SV *decode_tagged; |
| 660 |
} dec_t; |
| 661 |
|
| 662 |
#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE |
| 663 |
|
| 664 |
#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") |
| 665 |
|
| 666 |
#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED) |
| 667 |
#define DEC_DEC_DEPTH --dec->depth |
| 668 |
|
| 669 |
static UV |
| 670 |
decode_uint (dec_t *dec) |
| 671 |
{ |
| 672 |
U8 m = *dec->cur & MINOR_MASK; |
| 673 |
++dec->cur; |
| 674 |
|
| 675 |
if (ecb_expect_true (m < LENGTH_EXT1)) |
| 676 |
return m; |
| 677 |
else if (ecb_expect_true (m == LENGTH_EXT1)) |
| 678 |
{ |
| 679 |
WANT (1); |
| 680 |
dec->cur += 1; |
| 681 |
return dec->cur[-1]; |
| 682 |
} |
| 683 |
else if (ecb_expect_true (m == LENGTH_EXT2)) |
| 684 |
{ |
| 685 |
WANT (2); |
| 686 |
dec->cur += 2; |
| 687 |
return (((UV)dec->cur[-2]) << 8) |
| 688 |
| ((UV)dec->cur[-1]); |
| 689 |
} |
| 690 |
else if (ecb_expect_true (m == LENGTH_EXT4)) |
| 691 |
{ |
| 692 |
WANT (4); |
| 693 |
dec->cur += 4; |
| 694 |
return (((UV)dec->cur[-4]) << 24) |
| 695 |
| (((UV)dec->cur[-3]) << 16) |
| 696 |
| (((UV)dec->cur[-2]) << 8) |
| 697 |
| ((UV)dec->cur[-1]); |
| 698 |
} |
| 699 |
else if (ecb_expect_true (m == LENGTH_EXT8)) |
| 700 |
{ |
| 701 |
WANT (8); |
| 702 |
dec->cur += 8; |
| 703 |
|
| 704 |
return |
| 705 |
#if UVSIZE < 8 |
| 706 |
0 |
| 707 |
#else |
| 708 |
(((UV)dec->cur[-8]) << 56) |
| 709 |
| (((UV)dec->cur[-7]) << 48) |
| 710 |
| (((UV)dec->cur[-6]) << 40) |
| 711 |
| (((UV)dec->cur[-5]) << 32) |
| 712 |
#endif |
| 713 |
| (((UV)dec->cur[-4]) << 24) |
| 714 |
| (((UV)dec->cur[-3]) << 16) |
| 715 |
| (((UV)dec->cur[-2]) << 8) |
| 716 |
| ((UV)dec->cur[-1]); |
| 717 |
} |
| 718 |
else |
| 719 |
ERR ("corrupted CBOR data (unsupported integer minor encoding)"); |
| 720 |
|
| 721 |
fail: |
| 722 |
return 0; |
| 723 |
} |
| 724 |
|
| 725 |
static SV *decode_sv (dec_t *dec); |
| 726 |
|
| 727 |
static SV * |
| 728 |
decode_av (dec_t *dec) |
| 729 |
{ |
| 730 |
AV *av = newAV (); |
| 731 |
|
| 732 |
DEC_INC_DEPTH; |
| 733 |
|
| 734 |
if (*dec->cur == (MAJOR_ARRAY | MINOR_INDEF)) |
| 735 |
{ |
| 736 |
++dec->cur; |
| 737 |
|
| 738 |
for (;;) |
| 739 |
{ |
| 740 |
WANT (1); |
| 741 |
|
| 742 |
if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) |
| 743 |
{ |
| 744 |
++dec->cur; |
| 745 |
break; |
| 746 |
} |
| 747 |
|
| 748 |
av_push (av, decode_sv (dec)); |
| 749 |
} |
| 750 |
} |
| 751 |
else |
| 752 |
{ |
| 753 |
int i, len = decode_uint (dec); |
| 754 |
|
| 755 |
WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays |
| 756 |
av_fill (av, len - 1); |
| 757 |
|
| 758 |
for (i = 0; i < len; ++i) |
| 759 |
AvARRAY (av)[i] = decode_sv (dec); |
| 760 |
} |
| 761 |
|
| 762 |
DEC_DEC_DEPTH; |
| 763 |
return newRV_noinc ((SV *)av); |
| 764 |
|
| 765 |
fail: |
| 766 |
SvREFCNT_dec (av); |
| 767 |
DEC_DEC_DEPTH; |
| 768 |
return &PL_sv_undef; |
| 769 |
} |
| 770 |
|
| 771 |
static void |
| 772 |
decode_he (dec_t *dec, HV *hv) |
| 773 |
{ |
| 774 |
// for speed reasons, we specialcase single-string |
| 775 |
// byte or utf-8 strings as keys, but only when !stringref |
| 776 |
|
| 777 |
if (ecb_expect_true (!dec->stringref)) |
| 778 |
if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8)) |
| 779 |
{ |
| 780 |
I32 len = decode_uint (dec); |
| 781 |
char *key = (char *)dec->cur; |
| 782 |
|
| 783 |
WANT (len); |
| 784 |
dec->cur += len; |
| 785 |
|
| 786 |
hv_store (hv, key, len, decode_sv (dec), 0); |
| 787 |
|
| 788 |
return; |
| 789 |
} |
| 790 |
else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8)) |
| 791 |
{ |
| 792 |
I32 len = decode_uint (dec); |
| 793 |
char *key = (char *)dec->cur; |
| 794 |
|
| 795 |
WANT (len); |
| 796 |
dec->cur += len; |
| 797 |
|
| 798 |
if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8)) |
| 799 |
if (!is_utf8_string (key, len)) |
| 800 |
ERR ("corrupted CBOR data (invalid UTF-8 in map key)"); |
| 801 |
|
| 802 |
hv_store (hv, key, -len, decode_sv (dec), 0); |
| 803 |
|
| 804 |
return; |
| 805 |
} |
| 806 |
|
| 807 |
SV *k = decode_sv (dec); |
| 808 |
SV *v = decode_sv (dec); |
| 809 |
|
| 810 |
hv_store_ent (hv, k, v, 0); |
| 811 |
SvREFCNT_dec (k); |
| 812 |
|
| 813 |
fail: |
| 814 |
; |
| 815 |
} |
| 816 |
|
| 817 |
static SV * |
| 818 |
decode_hv (dec_t *dec) |
| 819 |
{ |
| 820 |
HV *hv = newHV (); |
| 821 |
|
| 822 |
DEC_INC_DEPTH; |
| 823 |
|
| 824 |
if (*dec->cur == (MAJOR_MAP | MINOR_INDEF)) |
| 825 |
{ |
| 826 |
++dec->cur; |
| 827 |
|
| 828 |
for (;;) |
| 829 |
{ |
| 830 |
WANT (1); |
| 831 |
|
| 832 |
if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) |
| 833 |
{ |
| 834 |
++dec->cur; |
| 835 |
break; |
| 836 |
} |
| 837 |
|
| 838 |
decode_he (dec, hv); |
| 839 |
} |
| 840 |
} |
| 841 |
else |
| 842 |
{ |
| 843 |
int pairs = decode_uint (dec); |
| 844 |
|
| 845 |
while (pairs--) |
| 846 |
decode_he (dec, hv); |
| 847 |
} |
| 848 |
|
| 849 |
DEC_DEC_DEPTH; |
| 850 |
return newRV_noinc ((SV *)hv); |
| 851 |
|
| 852 |
fail: |
| 853 |
SvREFCNT_dec (hv); |
| 854 |
DEC_DEC_DEPTH; |
| 855 |
return &PL_sv_undef; |
| 856 |
} |
| 857 |
|
| 858 |
static SV * |
| 859 |
decode_str (dec_t *dec, int utf8) |
| 860 |
{ |
| 861 |
SV *sv = 0; |
| 862 |
|
| 863 |
if ((*dec->cur & MINOR_MASK) == MINOR_INDEF) |
| 864 |
{ |
| 865 |
// indefinite length strings |
| 866 |
++dec->cur; |
| 867 |
|
| 868 |
U8 major = *dec->cur & MAJOR_MISC; |
| 869 |
|
| 870 |
sv = newSVpvn ("", 0); |
| 871 |
|
| 872 |
for (;;) |
| 873 |
{ |
| 874 |
WANT (1); |
| 875 |
|
| 876 |
if ((*dec->cur - major) > LENGTH_EXT8) |
| 877 |
if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) |
| 878 |
{ |
| 879 |
++dec->cur; |
| 880 |
break; |
| 881 |
} |
| 882 |
else |
| 883 |
ERR ("corrupted CBOR data (invalid chunks in indefinite length string)"); |
| 884 |
|
| 885 |
STRLEN len = decode_uint (dec); |
| 886 |
|
| 887 |
WANT (len); |
| 888 |
sv_catpvn (sv, dec->cur, len); |
| 889 |
dec->cur += len; |
| 890 |
} |
| 891 |
} |
| 892 |
else |
| 893 |
{ |
| 894 |
STRLEN len = decode_uint (dec); |
| 895 |
|
| 896 |
WANT (len); |
| 897 |
sv = newSVpvn (dec->cur, len); |
| 898 |
dec->cur += len; |
| 899 |
|
| 900 |
if (ecb_expect_false (dec->stringref) |
| 901 |
&& SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1)) |
| 902 |
av_push (dec->stringref, SvREFCNT_inc_NN (sv)); |
| 903 |
} |
| 904 |
|
| 905 |
if (utf8) |
| 906 |
{ |
| 907 |
if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8)) |
| 908 |
if (!is_utf8_string (SvPVX (sv), SvCUR (sv))) |
| 909 |
ERR ("corrupted CBOR data (invalid UTF-8 in text string)"); |
| 910 |
|
| 911 |
SvUTF8_on (sv); |
| 912 |
} |
| 913 |
|
| 914 |
return sv; |
| 915 |
|
| 916 |
fail: |
| 917 |
SvREFCNT_dec (sv); |
| 918 |
return &PL_sv_undef; |
| 919 |
} |
| 920 |
|
| 921 |
static SV * |
| 922 |
decode_tagged (dec_t *dec) |
| 923 |
{ |
| 924 |
SV *sv = 0; |
| 925 |
UV tag = decode_uint (dec); |
| 926 |
|
| 927 |
WANT (1); |
| 928 |
|
| 929 |
switch (tag) |
| 930 |
{ |
| 931 |
case CBOR_TAG_MAGIC: |
| 932 |
sv = decode_sv (dec); |
| 933 |
break; |
| 934 |
|
| 935 |
case CBOR_TAG_INDIRECTION: |
| 936 |
sv = newRV_noinc (decode_sv (dec)); |
| 937 |
break; |
| 938 |
|
| 939 |
case CBOR_TAG_STRINGREF_NAMESPACE: |
| 940 |
{ |
| 941 |
ENTER; SAVETMPS; |
| 942 |
|
| 943 |
SAVESPTR (dec->stringref); |
| 944 |
dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); |
| 945 |
|
| 946 |
sv = decode_sv (dec); |
| 947 |
|
| 948 |
FREETMPS; LEAVE; |
| 949 |
} |
| 950 |
break; |
| 951 |
|
| 952 |
case CBOR_TAG_STRINGREF: |
| 953 |
{ |
| 954 |
if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) |
| 955 |
ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); |
| 956 |
|
| 957 |
UV idx = decode_uint (dec); |
| 958 |
|
| 959 |
if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) |
| 960 |
ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)"); |
| 961 |
|
| 962 |
sv = newSVsv (AvARRAY (dec->stringref)[idx]); |
| 963 |
} |
| 964 |
break; |
| 965 |
|
| 966 |
case CBOR_TAG_VALUE_SHAREABLE: |
| 967 |
{ |
| 968 |
if (ecb_expect_false (!dec->shareable)) |
| 969 |
dec->shareable = (AV *)sv_2mortal ((SV *)newAV ()); |
| 970 |
|
| 971 |
if (dec->cbor.flags & F_ALLOW_CYCLES) |
| 972 |
{ |
| 973 |
sv = newSV (0); |
| 974 |
av_push (dec->shareable, SvREFCNT_inc_NN (sv)); |
| 975 |
|
| 976 |
SV *osv = decode_sv (dec); |
| 977 |
sv_setsv (sv, osv); |
| 978 |
SvREFCNT_dec_NN (osv); |
| 979 |
} |
| 980 |
else |
| 981 |
{ |
| 982 |
av_push (dec->shareable, &PL_sv_undef); |
| 983 |
int idx = AvFILLp (dec->shareable); |
| 984 |
sv = decode_sv (dec); |
| 985 |
av_store (dec->shareable, idx, SvREFCNT_inc_NN (sv)); |
| 986 |
} |
| 987 |
} |
| 988 |
break; |
| 989 |
|
| 990 |
case CBOR_TAG_VALUE_SHAREDREF: |
| 991 |
{ |
| 992 |
if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) |
| 993 |
ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); |
| 994 |
|
| 995 |
UV idx = decode_uint (dec); |
| 996 |
|
| 997 |
if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) |
| 998 |
ERR ("corrupted CBOR data (sharedref index out of bounds)"); |
| 999 |
|
| 1000 |
sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); |
| 1001 |
|
| 1002 |
if (sv == &PL_sv_undef) |
| 1003 |
ERR ("cyclic CBOR data structure found, but allow_cycles is not enabled"); |
| 1004 |
} |
| 1005 |
break; |
| 1006 |
|
| 1007 |
case CBOR_TAG_PERL_OBJECT: |
| 1008 |
{ |
| 1009 |
sv = decode_sv (dec); |
| 1010 |
|
| 1011 |
if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) |
| 1012 |
ERR ("corrupted CBOR data (non-array perl object)"); |
| 1013 |
|
| 1014 |
AV *av = (AV *)SvRV (sv); |
| 1015 |
int len = av_len (av) + 1; |
| 1016 |
HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0); |
| 1017 |
|
| 1018 |
if (!stash) |
| 1019 |
ERR ("cannot decode perl-object (package does not exist)"); |
| 1020 |
|
| 1021 |
GV *method = gv_fetchmethod_autoload (stash, "THAW", 0); |
| 1022 |
|
| 1023 |
if (!method) |
| 1024 |
ERR ("cannot decode perl-object (package does not have a THAW method)"); |
| 1025 |
|
| 1026 |
dSP; |
| 1027 |
|
| 1028 |
ENTER; SAVETMPS; |
| 1029 |
PUSHMARK (SP); |
| 1030 |
EXTEND (SP, len + 1); |
| 1031 |
// we re-bless the reference to get overload and other niceties right |
| 1032 |
PUSHs (*av_fetch (av, 0, 1)); |
| 1033 |
PUSHs (sv_cbor); |
| 1034 |
|
| 1035 |
int i; |
| 1036 |
|
| 1037 |
for (i = 1; i < len; ++i) |
| 1038 |
PUSHs (*av_fetch (av, i, 1)); |
| 1039 |
|
| 1040 |
PUTBACK; |
| 1041 |
call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL); |
| 1042 |
SPAGAIN; |
| 1043 |
|
| 1044 |
if (SvTRUE (ERRSV)) |
| 1045 |
{ |
| 1046 |
FREETMPS; LEAVE; |
| 1047 |
ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); |
| 1048 |
} |
| 1049 |
|
| 1050 |
SvREFCNT_dec (sv); |
| 1051 |
sv = SvREFCNT_inc (POPs); |
| 1052 |
|
| 1053 |
PUTBACK; |
| 1054 |
|
| 1055 |
FREETMPS; LEAVE; |
| 1056 |
} |
| 1057 |
break; |
| 1058 |
|
| 1059 |
default: |
| 1060 |
{ |
| 1061 |
sv = decode_sv (dec); |
| 1062 |
|
| 1063 |
dSP; |
| 1064 |
ENTER; SAVETMPS; |
| 1065 |
SAVESTACK_POS (); |
| 1066 |
PUSHMARK (SP); |
| 1067 |
EXTEND (SP, 2); |
| 1068 |
PUSHs (newSVuv (tag)); |
| 1069 |
PUSHs (sv); |
| 1070 |
|
| 1071 |
PUTBACK; |
| 1072 |
int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL); |
| 1073 |
SPAGAIN; |
| 1074 |
|
| 1075 |
if (SvTRUE (ERRSV)) |
| 1076 |
{ |
| 1077 |
FREETMPS; LEAVE; |
| 1078 |
ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); |
| 1079 |
} |
| 1080 |
|
| 1081 |
if (count) |
| 1082 |
{ |
| 1083 |
SvREFCNT_dec (sv); |
| 1084 |
sv = SvREFCNT_inc (POPs); |
| 1085 |
} |
| 1086 |
else |
| 1087 |
{ |
| 1088 |
AV *av = newAV (); |
| 1089 |
av_push (av, newSVuv (tag)); |
| 1090 |
av_push (av, sv); |
| 1091 |
|
| 1092 |
HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash |
| 1093 |
? cbor_tagged_stash |
| 1094 |
: gv_stashpv ("CBOR::XS::Tagged" , 1); |
| 1095 |
sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash); |
| 1096 |
} |
| 1097 |
|
| 1098 |
PUTBACK; |
| 1099 |
|
| 1100 |
FREETMPS; LEAVE; |
| 1101 |
} |
| 1102 |
break; |
| 1103 |
} |
| 1104 |
|
| 1105 |
return sv; |
| 1106 |
|
| 1107 |
fail: |
| 1108 |
SvREFCNT_dec (sv); |
| 1109 |
return &PL_sv_undef; |
| 1110 |
} |
| 1111 |
|
| 1112 |
static SV * |
| 1113 |
decode_sv (dec_t *dec) |
| 1114 |
{ |
| 1115 |
WANT (1); |
| 1116 |
|
| 1117 |
switch (*dec->cur >> MAJOR_SHIFT) |
| 1118 |
{ |
| 1119 |
case MAJOR_POS_INT >> MAJOR_SHIFT: return newSVuv (decode_uint (dec)); |
| 1120 |
case MAJOR_NEG_INT >> MAJOR_SHIFT: return newSViv (-1 - (IV)decode_uint (dec)); |
| 1121 |
case MAJOR_BYTES >> MAJOR_SHIFT: return decode_str (dec, 0); |
| 1122 |
case MAJOR_TEXT >> MAJOR_SHIFT: return decode_str (dec, 1); |
| 1123 |
case MAJOR_ARRAY >> MAJOR_SHIFT: return decode_av (dec); |
| 1124 |
case MAJOR_MAP >> MAJOR_SHIFT: return decode_hv (dec); |
| 1125 |
case MAJOR_TAG >> MAJOR_SHIFT: return decode_tagged (dec); |
| 1126 |
|
| 1127 |
case MAJOR_MISC >> MAJOR_SHIFT: |
| 1128 |
switch (*dec->cur++ & MINOR_MASK) |
| 1129 |
{ |
| 1130 |
case SIMPLE_FALSE: |
| 1131 |
#if CBOR_SLOW |
| 1132 |
types_false = get_bool ("Types::Serialiser::false"); |
| 1133 |
#endif |
| 1134 |
return newSVsv (types_false); |
| 1135 |
case SIMPLE_TRUE: |
| 1136 |
#if CBOR_SLOW |
| 1137 |
types_true = get_bool ("Types::Serialiser::true"); |
| 1138 |
#endif |
| 1139 |
return newSVsv (types_true); |
| 1140 |
case SIMPLE_NULL: |
| 1141 |
return newSVsv (&PL_sv_undef); |
| 1142 |
case SIMPLE_UNDEF: |
| 1143 |
#if CBOR_SLOW |
| 1144 |
types_error = get_bool ("Types::Serialiser::error"); |
| 1145 |
#endif |
| 1146 |
return newSVsv (types_error); |
| 1147 |
|
| 1148 |
case MISC_FLOAT16: |
| 1149 |
{ |
| 1150 |
WANT (2); |
| 1151 |
|
| 1152 |
uint16_t fp = (dec->cur[0] << 8) | dec->cur[1]; |
| 1153 |
dec->cur += 2; |
| 1154 |
|
| 1155 |
return newSVnv (ecb_binary16_to_float (fp)); |
| 1156 |
} |
| 1157 |
|
| 1158 |
case MISC_FLOAT32: |
| 1159 |
{ |
| 1160 |
uint32_t fp; |
| 1161 |
WANT (4); |
| 1162 |
memcpy (&fp, dec->cur, 4); |
| 1163 |
dec->cur += 4; |
| 1164 |
|
| 1165 |
if (!ecb_big_endian ()) |
| 1166 |
fp = ecb_bswap32 (fp); |
| 1167 |
|
| 1168 |
return newSVnv (ecb_binary32_to_float (fp)); |
| 1169 |
} |
| 1170 |
|
| 1171 |
case MISC_FLOAT64: |
| 1172 |
{ |
| 1173 |
uint64_t fp; |
| 1174 |
WANT (8); |
| 1175 |
memcpy (&fp, dec->cur, 8); |
| 1176 |
dec->cur += 8; |
| 1177 |
|
| 1178 |
if (!ecb_big_endian ()) |
| 1179 |
fp = ecb_bswap64 (fp); |
| 1180 |
|
| 1181 |
return newSVnv (ecb_binary64_to_double (fp)); |
| 1182 |
} |
| 1183 |
|
| 1184 |
// 0..19 unassigned simple |
| 1185 |
// 24 reserved + unassigned simple (reserved values are not encodable) |
| 1186 |
// 28-30 unassigned misc |
| 1187 |
// 31 break code |
| 1188 |
default: |
| 1189 |
ERR ("corrupted CBOR data (reserved/unassigned/unexpected major 7 value)"); |
| 1190 |
} |
| 1191 |
|
| 1192 |
break; |
| 1193 |
} |
| 1194 |
|
| 1195 |
fail: |
| 1196 |
return &PL_sv_undef; |
| 1197 |
} |
| 1198 |
|
| 1199 |
static SV * |
| 1200 |
decode_cbor (SV *string, CBOR *cbor, char **offset_return) |
| 1201 |
{ |
| 1202 |
dec_t dec = { 0 }; |
| 1203 |
SV *sv; |
| 1204 |
STRLEN len; |
| 1205 |
char *data = SvPVbyte (string, len); |
| 1206 |
|
| 1207 |
if (len > cbor->max_size && cbor->max_size) |
| 1208 |
croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu", |
| 1209 |
(unsigned long)len, (unsigned long)cbor->max_size); |
| 1210 |
|
| 1211 |
dec.cbor = *cbor; |
| 1212 |
dec.cur = (U8 *)data; |
| 1213 |
dec.end = (U8 *)data + len; |
| 1214 |
|
| 1215 |
sv = decode_sv (&dec); |
| 1216 |
|
| 1217 |
if (offset_return) |
| 1218 |
*offset_return = dec.cur; |
| 1219 |
|
| 1220 |
if (!(offset_return || !sv)) |
| 1221 |
if (dec.cur != dec.end && !dec.err) |
| 1222 |
dec.err = "garbage after CBOR object"; |
| 1223 |
|
| 1224 |
if (dec.err) |
| 1225 |
{ |
| 1226 |
if (dec.shareable) |
| 1227 |
{ |
| 1228 |
// need to break cyclic links, which whould all be in shareable |
| 1229 |
int i; |
| 1230 |
SV **svp; |
| 1231 |
|
| 1232 |
for (i = av_len (dec.shareable) + 1; i--; ) |
| 1233 |
if ((svp = av_fetch (dec.shareable, i, 0))) |
| 1234 |
sv_setsv (*svp, &PL_sv_undef); |
| 1235 |
} |
| 1236 |
|
| 1237 |
SvREFCNT_dec (sv); |
| 1238 |
croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); |
| 1239 |
} |
| 1240 |
|
| 1241 |
sv = sv_2mortal (sv); |
| 1242 |
|
| 1243 |
return sv; |
| 1244 |
} |
| 1245 |
|
| 1246 |
///////////////////////////////////////////////////////////////////////////// |
| 1247 |
// incremental parser |
| 1248 |
|
| 1249 |
#define INCR_DONE(cbor) (AvFILLp (cbor->incr_count) < 0) |
| 1250 |
|
| 1251 |
// returns 0 for notyet, 1 for success or error |
| 1252 |
static int |
| 1253 |
incr_parse (CBOR *self, SV *cborstr) |
| 1254 |
{ |
| 1255 |
STRLEN cur; |
| 1256 |
SvPV (cborstr, cur); |
| 1257 |
|
| 1258 |
while (ecb_expect_true (self->incr_need <= cur)) |
| 1259 |
{ |
| 1260 |
// table of integer count bytes |
| 1261 |
static I8 incr_len[MINOR_MASK + 1] = { |
| 1262 |
0, 0, 0, 0, 0, 0, 0, 0, |
| 1263 |
0, 0, 0, 0, 0, 0, 0, 0, |
| 1264 |
0, 0, 0, 0, 0, 0, 0, 0, |
| 1265 |
1, 2, 4, 8,-1,-1,-1,-2 |
| 1266 |
}; |
| 1267 |
|
| 1268 |
const U8 *p = SvPVX (cborstr) + self->incr_pos; |
| 1269 |
U8 m = *p & MINOR_MASK; |
| 1270 |
IV count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]); |
| 1271 |
I8 ilen = incr_len[m]; |
| 1272 |
|
| 1273 |
self->incr_need = self->incr_pos + 1; |
| 1274 |
|
| 1275 |
if (ecb_expect_false (ilen < 0)) |
| 1276 |
{ |
| 1277 |
if (m != MINOR_INDEF) |
| 1278 |
return 1; // error |
| 1279 |
|
| 1280 |
if (*p == (MAJOR_MISC | MINOR_INDEF)) |
| 1281 |
{ |
| 1282 |
if (count >= 0) |
| 1283 |
return 1; // error |
| 1284 |
|
| 1285 |
count = 1; |
| 1286 |
} |
| 1287 |
else |
| 1288 |
{ |
| 1289 |
av_push (self->incr_count, newSViv (-1)); //TODO: nest |
| 1290 |
count = -1; |
| 1291 |
} |
| 1292 |
} |
| 1293 |
else |
| 1294 |
{ |
| 1295 |
self->incr_need += ilen; |
| 1296 |
if (ecb_expect_false (self->incr_need > cur)) |
| 1297 |
return 0; |
| 1298 |
|
| 1299 |
int major = *p >> MAJOR_SHIFT; |
| 1300 |
|
| 1301 |
switch (major) |
| 1302 |
{ |
| 1303 |
case MAJOR_TAG >> MAJOR_SHIFT: |
| 1304 |
++count; // tags merely prefix another value |
| 1305 |
break; |
| 1306 |
|
| 1307 |
case MAJOR_BYTES >> MAJOR_SHIFT: |
| 1308 |
case MAJOR_TEXT >> MAJOR_SHIFT: |
| 1309 |
case MAJOR_ARRAY >> MAJOR_SHIFT: |
| 1310 |
case MAJOR_MAP >> MAJOR_SHIFT: |
| 1311 |
{ |
| 1312 |
UV len; |
| 1313 |
|
| 1314 |
if (ecb_expect_false (ilen)) |
| 1315 |
{ |
| 1316 |
len = 0; |
| 1317 |
|
| 1318 |
do { |
| 1319 |
len = (len << 8) | *++p; |
| 1320 |
} while (--ilen); |
| 1321 |
} |
| 1322 |
else |
| 1323 |
len = m; |
| 1324 |
|
| 1325 |
switch (major) |
| 1326 |
{ |
| 1327 |
case MAJOR_BYTES >> MAJOR_SHIFT: |
| 1328 |
case MAJOR_TEXT >> MAJOR_SHIFT: |
| 1329 |
self->incr_need += len; |
| 1330 |
if (ecb_expect_false (self->incr_need > cur)) |
| 1331 |
return 0; |
| 1332 |
|
| 1333 |
break; |
| 1334 |
|
| 1335 |
case MAJOR_MAP >> MAJOR_SHIFT: |
| 1336 |
len <<= 1; |
| 1337 |
case MAJOR_ARRAY >> MAJOR_SHIFT: |
| 1338 |
if (len) |
| 1339 |
{ |
| 1340 |
av_push (self->incr_count, newSViv (len + 1)); //TODO: nest |
| 1341 |
count = len + 1; |
| 1342 |
} |
| 1343 |
break; |
| 1344 |
} |
| 1345 |
} |
| 1346 |
} |
| 1347 |
} |
| 1348 |
|
| 1349 |
self->incr_pos = self->incr_need; |
| 1350 |
|
| 1351 |
if (count > 0) |
| 1352 |
{ |
| 1353 |
while (!--count) |
| 1354 |
{ |
| 1355 |
if (!AvFILLp (self->incr_count)) |
| 1356 |
return 1; // done |
| 1357 |
|
| 1358 |
SvREFCNT_dec_NN (av_pop (self->incr_count)); |
| 1359 |
count = SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]); |
| 1360 |
} |
| 1361 |
|
| 1362 |
SvIVX (AvARRAY (self->incr_count)[AvFILLp (self->incr_count)]) = count; |
| 1363 |
} |
| 1364 |
} |
| 1365 |
|
| 1366 |
return 0; |
| 1367 |
} |
| 1368 |
|
| 1369 |
|
| 1370 |
///////////////////////////////////////////////////////////////////////////// |
| 1371 |
// XS interface functions |
| 1372 |
|
| 1373 |
MODULE = CBOR::XS PACKAGE = CBOR::XS |
| 1374 |
|
| 1375 |
BOOT: |
| 1376 |
{ |
| 1377 |
cbor_stash = gv_stashpv ("CBOR::XS" , 1); |
| 1378 |
cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1); |
| 1379 |
|
| 1380 |
types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1); |
| 1381 |
types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1); |
| 1382 |
|
| 1383 |
types_true = get_bool ("Types::Serialiser::true" ); |
| 1384 |
types_false = get_bool ("Types::Serialiser::false"); |
| 1385 |
types_error = get_bool ("Types::Serialiser::error"); |
| 1386 |
|
| 1387 |
default_filter = newSVpv ("CBOR::XS::default_filter", 0); |
| 1388 |
|
| 1389 |
sv_cbor = newSVpv ("CBOR", 0); |
| 1390 |
SvREADONLY_on (sv_cbor); |
| 1391 |
} |
| 1392 |
|
| 1393 |
PROTOTYPES: DISABLE |
| 1394 |
|
| 1395 |
void CLONE (...) |
| 1396 |
CODE: |
| 1397 |
cbor_stash = 0; |
| 1398 |
cbor_tagged_stash = 0; |
| 1399 |
types_error_stash = 0; |
| 1400 |
types_boolean_stash = 0; |
| 1401 |
|
| 1402 |
void new (char *klass) |
| 1403 |
PPCODE: |
| 1404 |
{ |
| 1405 |
SV *pv = NEWSV (0, sizeof (CBOR)); |
| 1406 |
SvPOK_only (pv); |
| 1407 |
cbor_init ((CBOR *)SvPVX (pv)); |
| 1408 |
XPUSHs (sv_2mortal (sv_bless ( |
| 1409 |
newRV_noinc (pv), |
| 1410 |
strEQ (klass, "CBOR::XS") ? CBOR_STASH : gv_stashpv (klass, 1) |
| 1411 |
))); |
| 1412 |
} |
| 1413 |
|
| 1414 |
void shrink (CBOR *self, int enable = 1) |
| 1415 |
ALIAS: |
| 1416 |
shrink = F_SHRINK |
| 1417 |
allow_unknown = F_ALLOW_UNKNOWN |
| 1418 |
allow_sharing = F_ALLOW_SHARING |
| 1419 |
allow_cycles = F_ALLOW_CYCLES |
| 1420 |
pack_strings = F_PACK_STRINGS |
| 1421 |
utf8_strings = F_UTF8_STRINGS |
| 1422 |
validate_utf8 = F_VALIDATE_UTF8 |
| 1423 |
PPCODE: |
| 1424 |
{ |
| 1425 |
if (enable) |
| 1426 |
self->flags |= ix; |
| 1427 |
else |
| 1428 |
self->flags &= ~ix; |
| 1429 |
|
| 1430 |
XPUSHs (ST (0)); |
| 1431 |
} |
| 1432 |
|
| 1433 |
void get_shrink (CBOR *self) |
| 1434 |
ALIAS: |
| 1435 |
get_shrink = F_SHRINK |
| 1436 |
get_allow_unknown = F_ALLOW_UNKNOWN |
| 1437 |
get_allow_sharing = F_ALLOW_SHARING |
| 1438 |
get_allow_cycles = F_ALLOW_CYCLES |
| 1439 |
get_pack_strings = F_PACK_STRINGS |
| 1440 |
get_validate_utf8 = F_VALIDATE_UTF8 |
| 1441 |
PPCODE: |
| 1442 |
XPUSHs (boolSV (self->flags & ix)); |
| 1443 |
|
| 1444 |
void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) |
| 1445 |
PPCODE: |
| 1446 |
self->max_depth = max_depth; |
| 1447 |
XPUSHs (ST (0)); |
| 1448 |
|
| 1449 |
U32 get_max_depth (CBOR *self) |
| 1450 |
CODE: |
| 1451 |
RETVAL = self->max_depth; |
| 1452 |
OUTPUT: |
| 1453 |
RETVAL |
| 1454 |
|
| 1455 |
void max_size (CBOR *self, U32 max_size = 0) |
| 1456 |
PPCODE: |
| 1457 |
self->max_size = max_size; |
| 1458 |
XPUSHs (ST (0)); |
| 1459 |
|
| 1460 |
int get_max_size (CBOR *self) |
| 1461 |
CODE: |
| 1462 |
RETVAL = self->max_size; |
| 1463 |
OUTPUT: |
| 1464 |
RETVAL |
| 1465 |
|
| 1466 |
void filter (CBOR *self, SV *filter = 0) |
| 1467 |
PPCODE: |
| 1468 |
SvREFCNT_dec (self->filter); |
| 1469 |
self->filter = filter ? newSVsv (filter) : filter; |
| 1470 |
XPUSHs (ST (0)); |
| 1471 |
|
| 1472 |
SV *get_filter (CBOR *self) |
| 1473 |
CODE: |
| 1474 |
RETVAL = self->filter ? self->filter : NEWSV (0, 0); |
| 1475 |
OUTPUT: |
| 1476 |
RETVAL |
| 1477 |
|
| 1478 |
void encode (CBOR *self, SV *scalar) |
| 1479 |
PPCODE: |
| 1480 |
PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN; |
| 1481 |
XPUSHs (scalar); |
| 1482 |
|
| 1483 |
void decode (CBOR *self, SV *cborstr) |
| 1484 |
PPCODE: |
| 1485 |
PUTBACK; cborstr = decode_cbor (cborstr, self, 0); SPAGAIN; |
| 1486 |
XPUSHs (cborstr); |
| 1487 |
|
| 1488 |
void decode_prefix (CBOR *self, SV *cborstr) |
| 1489 |
PPCODE: |
| 1490 |
{ |
| 1491 |
SV *sv; |
| 1492 |
char *offset; |
| 1493 |
PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN; |
| 1494 |
EXTEND (SP, 2); |
| 1495 |
PUSHs (sv); |
| 1496 |
PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); |
| 1497 |
} |
| 1498 |
|
| 1499 |
void incr_parse (CBOR *self, SV *cborstr) |
| 1500 |
ALIAS: |
| 1501 |
incr_parse_multiple = 1 |
| 1502 |
PPCODE: |
| 1503 |
{ |
| 1504 |
if (SvUTF8 (cborstr)) |
| 1505 |
sv_utf8_downgrade (cborstr, 0); |
| 1506 |
|
| 1507 |
if (!self->incr_count) |
| 1508 |
{ |
| 1509 |
self->incr_count = newAV (); |
| 1510 |
self->incr_pos = 0; |
| 1511 |
self->incr_need = 1; |
| 1512 |
|
| 1513 |
av_push (self->incr_count, newSViv (1)); |
| 1514 |
} |
| 1515 |
|
| 1516 |
do |
| 1517 |
{ |
| 1518 |
if (!incr_parse (self, cborstr)) |
| 1519 |
{ |
| 1520 |
if (self->incr_need > self->max_size && self->max_size) |
| 1521 |
croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu", |
| 1522 |
(unsigned long)self->incr_need, (unsigned long)self->max_size); |
| 1523 |
|
| 1524 |
break; |
| 1525 |
} |
| 1526 |
|
| 1527 |
SV *sv; |
| 1528 |
char *offset; |
| 1529 |
|
| 1530 |
PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN; |
| 1531 |
XPUSHs (sv); |
| 1532 |
|
| 1533 |
sv_chop (cborstr, offset); |
| 1534 |
|
| 1535 |
av_clear (self->incr_count); |
| 1536 |
av_push (self->incr_count, newSViv (1)); |
| 1537 |
|
| 1538 |
self->incr_pos = 0; |
| 1539 |
self->incr_need = self->incr_pos + 1; |
| 1540 |
} |
| 1541 |
while (ix); |
| 1542 |
} |
| 1543 |
|
| 1544 |
void incr_reset (CBOR *self) |
| 1545 |
CODE: |
| 1546 |
{ |
| 1547 |
SvREFCNT_dec (self->incr_count); |
| 1548 |
self->incr_count = 0; |
| 1549 |
} |
| 1550 |
|
| 1551 |
void DESTROY (CBOR *self) |
| 1552 |
PPCODE: |
| 1553 |
cbor_free (self); |
| 1554 |
|
| 1555 |
PROTOTYPES: ENABLE |
| 1556 |
|
| 1557 |
void encode_cbor (SV *scalar) |
| 1558 |
ALIAS: |
| 1559 |
encode_cbor = 0 |
| 1560 |
encode_cbor_sharing = F_ALLOW_SHARING |
| 1561 |
PPCODE: |
| 1562 |
{ |
| 1563 |
CBOR cbor; |
| 1564 |
cbor_init (&cbor); |
| 1565 |
cbor.flags |= ix; |
| 1566 |
PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN; |
| 1567 |
XPUSHs (scalar); |
| 1568 |
} |
| 1569 |
|
| 1570 |
void decode_cbor (SV *cborstr) |
| 1571 |
PPCODE: |
| 1572 |
{ |
| 1573 |
CBOR cbor; |
| 1574 |
cbor_init (&cbor); |
| 1575 |
PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN; |
| 1576 |
XPUSHs (cborstr); |
| 1577 |
} |
| 1578 |
|
