Lev Walkin | f15320b | 2004-06-03 03:38:44 +0000 | [diff] [blame] | 1 | %{ |
| 2 | |
| 3 | #include <string.h> |
| 4 | #include <errno.h> |
| 5 | #include <assert.h> |
| 6 | |
| 7 | #include "asn1parser.h" |
| 8 | #include "asn1p_y.h" |
| 9 | |
| 10 | int asn1p_lex(void); |
| 11 | void asn1p_lexer_hack_push_opaque_state(void); /* Used in .y */ |
| 12 | void asn1p_lexer_hack_enable_with_syntax(void); /* Used in .y */ |
| 13 | |
| 14 | #define YY_FATAL_ERROR(msg) do { \ |
| 15 | fprintf(stderr, \ |
| 16 | "lexer error at line %d, " \ |
| 17 | "text \"%s\"\n", \ |
| 18 | yylineno, yytext); \ |
| 19 | exit(1); \ |
| 20 | } while(0) |
| 21 | |
| 22 | int asn1p_lexer_pedantic_1990 = 0; |
| 23 | int asn1p_lexer_types_year = 0; |
| 24 | int asn1p_lexer_constructs_year = 0; |
| 25 | static int _check_dashes(char *ptr); |
| 26 | static asn1_integer_t asn1p_atoi(char *ptr); /* errno is either 0 or ERANGE */ |
| 27 | |
| 28 | /* |
| 29 | * Check that the type is defined in the year of the standard choosen. |
| 30 | */ |
| 31 | #define TYPE_LIFETIME(fyr, lyr) \ |
| 32 | (!asn1p_lexer_types_year \ |
| 33 | || (fyr && fyr <= asn1p_lexer_types_year) \ |
| 34 | || (lyr && lyr > asn1p_lexer_types_year)) |
| 35 | |
| 36 | /* |
| 37 | * Check the the construction (or concept, i.e. CLASS) is defined in |
| 38 | * a given year. |
| 39 | */ |
| 40 | #define CONSTRUCT_LIFETIME(fyr, lyr) \ |
| 41 | (!asn1p_lexer_constructs_year \ |
| 42 | || (fyr && fyr <= asn1p_lexer_constructs_year) \ |
| 43 | || (lyr && lyr > asn1p_lexer_constructs_year)) |
| 44 | |
| 45 | /* |
| 46 | * Make sure that the label is compliant with the naming rules. |
| 47 | */ |
| 48 | #define CHECK_DASHES do { \ |
| 49 | if(_check_dashes(yytext)) { \ |
| 50 | fprintf(stderr, \ |
| 51 | "%s: Identifier format invalid: " \ |
| 52 | "Improper dash location\n", yytext); \ |
| 53 | return -1; \ |
| 54 | } } while(0) |
| 55 | |
| 56 | /* |
| 57 | * Append quoted string. |
| 58 | */ |
| 59 | #define QAPPEND(text, tlen) do { \ |
| 60 | char *prev_text = asn1p_lval.tv_opaque.buf; \ |
| 61 | int prev_len = asn1p_lval.tv_opaque.len; \ |
| 62 | char *p; \ |
| 63 | \ |
| 64 | p = malloc((tlen) + prev_len + 1); \ |
| 65 | if(p == NULL) return -1; \ |
| 66 | \ |
| 67 | if(prev_text) memcpy(p, prev_text, prev_len); \ |
| 68 | memcpy(p + prev_len, text, tlen); \ |
| 69 | p[prev_len + (tlen)] = '\0'; \ |
| 70 | \ |
| 71 | free(asn1p_lval.tv_opaque.buf); \ |
| 72 | asn1p_lval.tv_opaque.buf = p; \ |
| 73 | asn1p_lval.tv_opaque.len = (tlen) + prev_len; \ |
| 74 | } while(0) |
| 75 | |
| 76 | %} |
| 77 | |
| 78 | %option never-interactive |
| 79 | %option noinput nounput |
| 80 | %option noyywrap stack |
| 81 | /* Performance penalty is OK */ |
| 82 | %option yylineno |
| 83 | /* Controlled from within application */ |
| 84 | %option debug |
| 85 | |
| 86 | %pointer |
| 87 | |
| 88 | %x dash_comment |
| 89 | %x cpp_comment |
| 90 | %x quoted |
| 91 | %x opaque |
| 92 | %x with_syntax |
| 93 | |
| 94 | /* Newline */ |
| 95 | NL [\r\v\f\n] |
| 96 | /* White-space */ |
| 97 | WSP [\t\r\v\f\n ] |
| 98 | |
| 99 | %% |
| 100 | |
| 101 | "--" yy_push_state(dash_comment); |
| 102 | <dash_comment>{ |
| 103 | |
| 104 | {NL} yy_pop_state(); |
| 105 | |
| 106 | -- yy_pop_state(); /* End of comment */ |
| 107 | - /* Eat single dash */ |
| 108 | [^\r\v\f\n-]+ /* Eat */ |
| 109 | |
| 110 | } |
| 111 | <INITIAL,cpp_comment>"/*" yy_push_state(cpp_comment); |
| 112 | <cpp_comment>{ |
| 113 | [^*/] /* Eat */ |
| 114 | "*/" yy_pop_state(); |
| 115 | . /* Eat */ |
| 116 | } |
| 117 | |
| 118 | |
| 119 | /* |
| 120 | * This is state is being set from corresponding .y module when |
| 121 | * higher-level data is necessary to make proper parsing of the |
| 122 | * underlying data. Thus, we enter the <opaque> state and save |
| 123 | * everything for later processing. |
| 124 | */ |
| 125 | <opaque>{ |
| 126 | |
| 127 | "{" { |
| 128 | yy_push_state(opaque); |
| 129 | asn1p_lval.tv_opaque.buf = strdup(yytext); |
| 130 | asn1p_lval.tv_opaque.len = yyleng; |
| 131 | return TOK_opaque; |
| 132 | } |
| 133 | |
| 134 | "}" { |
| 135 | yy_pop_state(); |
| 136 | asn1p_lval.tv_opaque.buf = strdup(yytext); |
| 137 | asn1p_lval.tv_opaque.len = yyleng; |
| 138 | return TOK_opaque; |
| 139 | } |
| 140 | |
| 141 | [^{}:=]+ { |
| 142 | asn1p_lval.tv_opaque.buf = strdup(yytext); |
| 143 | asn1p_lval.tv_opaque.len = yyleng; |
| 144 | return TOK_opaque; |
| 145 | } |
| 146 | |
| 147 | "::=" { |
| 148 | fprintf(stderr, |
| 149 | "ASN.1 Parser syncronization failure: " |
| 150 | "\"%s\" at line %d must not appear " |
| 151 | "inside value definition\n", |
| 152 | yytext, yylineno); |
| 153 | return -1; |
| 154 | } |
| 155 | |
| 156 | [:=] { |
| 157 | asn1p_lval.tv_opaque.buf = strdup(yytext); |
| 158 | asn1p_lval.tv_opaque.len = yyleng; |
| 159 | return TOK_opaque; |
| 160 | } |
| 161 | |
| 162 | } |
| 163 | |
| 164 | \"[^\"]* { |
| 165 | asn1p_lval.tv_opaque.buf = 0; |
| 166 | asn1p_lval.tv_opaque.len = 0; |
| 167 | QAPPEND(yytext+1, yyleng-1); |
| 168 | yy_push_state(quoted); |
| 169 | } |
| 170 | <quoted>{ |
| 171 | |
| 172 | \"\" { QAPPEND(yytext, yyleng-1); } /* Add a single quote */ |
| 173 | [^\"]+ { QAPPEND(yytext, yyleng); } |
| 174 | |
| 175 | \" { |
| 176 | yy_pop_state(); |
| 177 | /* Do not append last quote: |
| 178 | // QAPPEND(yytext, yyleng); */ |
| 179 | |
| 180 | if(asn1p_lexer_pedantic_1990 |
| 181 | && strchr(yytext, '\n')) { |
| 182 | fprintf(stderr, "%s: " |
| 183 | "Newlines are prohibited by ASN.1:1990\n", |
| 184 | asn1p_lval.tv_opaque.buf); |
| 185 | return -1; |
| 186 | } |
| 187 | |
| 188 | return TOK_cstring; |
| 189 | } |
| 190 | |
| 191 | } |
| 192 | |
| 193 | |
| 194 | '[0-9A-F \t\r\v\f\n]+'H { |
| 195 | /* " \t\r\n" weren't allowed in ASN.1:1990. */ |
| 196 | asn1p_lval.tv_str = yytext; |
| 197 | return TOK_hstring; |
| 198 | } |
| 199 | |
| 200 | '[01 \t\r\v\f\n]+'B { |
| 201 | /* " \t\r\n" weren't allowed in ASN.1:1990. */ |
| 202 | asn1p_lval.tv_str = strdup(yytext); |
| 203 | return TOK_bstring; |
| 204 | } |
| 205 | |
| 206 | |
| 207 | -[1-9][0-9]* { |
| 208 | asn1p_lval.a_int = asn1p_atoi(yytext); |
| 209 | if(errno == ERANGE) |
| 210 | return -1; |
| 211 | return TOK_number_negative; |
| 212 | } |
| 213 | |
| 214 | [1-9][0-9]* { |
| 215 | asn1p_lval.a_int = asn1p_atoi(yytext); |
| 216 | if(errno == ERANGE) |
| 217 | return -1; |
| 218 | return TOK_number; |
| 219 | } |
| 220 | |
| 221 | "0" { |
| 222 | asn1p_lval.a_int = asn1p_atoi(yytext); |
| 223 | if(errno == ERANGE) |
| 224 | return -1; |
| 225 | return TOK_number; |
| 226 | } |
| 227 | |
| 228 | /* |
| 229 | * Tags |
| 230 | */ |
| 231 | \[(UNIVERSAL[ \t\r\v\f\n]+|APPLICATION[ \t\r\v\f\n]+|PRIVATE[ \t\r\v\f\n]+)?[0-9]+\] { |
| 232 | char *p; |
| 233 | memset(&asn1p_lval.a_tag, 0, sizeof(asn1p_lval.a_tag)); |
| 234 | switch(yytext[1]) { |
| 235 | case 'U': |
| 236 | asn1p_lval.a_tag.tag_class = TC_UNIVERSAL; |
| 237 | p = yytext + sizeof("UNIVERSAL") + 1; |
| 238 | break; |
| 239 | case 'A': |
| 240 | asn1p_lval.a_tag.tag_class = TC_APPLICATION; |
| 241 | p = yytext + sizeof("APPLICATION") + 1; |
| 242 | break; |
| 243 | case 'P': |
| 244 | asn1p_lval.a_tag.tag_class = TC_PRIVATE; |
| 245 | p = yytext + sizeof("PRIVATE") + 1; |
| 246 | break; |
| 247 | default: |
| 248 | assert(yytext[1] >= '0' && yytext[1] <= '9'); |
| 249 | asn1p_lval.a_tag.tag_class = TC_CONTEXT_SPECIFIC; |
| 250 | p = yytext + 1; |
| 251 | break; |
| 252 | } |
| 253 | asn1p_lval.a_tag.tag_value = asn1p_atoi(p); |
| 254 | if(*p == '0' && asn1p_lval.a_tag.tag_value) { |
| 255 | fprintf(stderr, |
| 256 | "Tag value at line %d " |
| 257 | "cannot start with zero " |
| 258 | "and have multiple digits: \"%s\"\n", |
| 259 | yylineno, yytext); |
| 260 | return -1; |
| 261 | } |
| 262 | return TOK_tag; |
| 263 | } |
| 264 | |
| 265 | \[[A-Z]+[ \t\r\v\f\n]+[0-9]+\] { |
| 266 | fprintf(stderr, |
| 267 | "Unsupported tag syntax at line %d: \"%s\"\n", |
| 268 | yylineno, yytext); |
| 269 | return -1; |
| 270 | } |
| 271 | |
| 272 | ABSENT return TOK_ABSENT; |
| 273 | ABSTRACT-SYNTAX return TOK_ABSTRACT_SYNTAX; |
| 274 | ALL return TOK_ALL; |
| 275 | ANY { |
| 276 | /* Appeared in 1990, removed in 1997 */ |
| 277 | if(TYPE_LIFETIME(1990, 1997)) |
| 278 | return TOK_ANY; |
| 279 | fprintf(stderr, "Keyword \"%s\" at line %d " |
| 280 | "is obsolete\n", yytext, yylineno); |
| 281 | REJECT; |
| 282 | } |
| 283 | APPLICATION return TOK_APPLICATION; |
| 284 | AUTOMATIC return TOK_AUTOMATIC; |
| 285 | BEGIN return TOK_BEGIN; |
| 286 | BIT return TOK_BIT; |
| 287 | BMPString { |
| 288 | if(TYPE_LIFETIME(1994, 0)) |
| 289 | return TOK_BMPString; |
| 290 | REJECT; |
| 291 | } |
| 292 | BOOLEAN return TOK_BOOLEAN; |
| 293 | BY return TOK_BY; |
| 294 | CHARACTER return TOK_CHARACTER; |
| 295 | CHOICE return TOK_CHOICE; |
| 296 | CLASS return TOK_CLASS; |
| 297 | COMPONENT return TOK_COMPONENT; |
| 298 | COMPONENTS return TOK_COMPONENTS; |
| 299 | CONSRAINED return TOK_CONSTRAINED; |
| 300 | CONTAINING return TOK_CONTAINING; |
| 301 | DEFAULT return TOK_DEFAULT; |
| 302 | DEFINED { |
| 303 | /* Appeared in 1990, removed in 1997 */ |
| 304 | if(TYPE_LIFETIME(1990, 1997)) |
| 305 | return TOK_DEFINED; |
| 306 | fprintf(stderr, "Keyword \"%s\" at line %d " |
| 307 | "is obsolete\n", yytext, yylineno); |
| 308 | /* Deprecated since */ |
| 309 | REJECT; |
| 310 | } |
| 311 | DEFINITIONS return TOK_DEFINITIONS; |
| 312 | EMBEDDED return TOK_EMBEDDED; |
| 313 | ENCODED return TOK_ENCODED; |
| 314 | END return TOK_END; |
| 315 | ENUMERATED return TOK_ENUMERATED; |
| 316 | EXCEPT return TOK_EXCEPT; |
| 317 | EXPLICIT return TOK_EXPLICIT; |
| 318 | EXPORTS return TOK_EXPORTS; |
| 319 | EXTENSIBILITY return TOK_EXTENSIBILITY; |
| 320 | EXTERNAL return TOK_EXTERNAL; |
| 321 | FALSE return TOK_FALSE; |
| 322 | FROM return TOK_FROM; |
| 323 | GeneralizedTime return TOK_GeneralizedTime; |
| 324 | GeneralString return TOK_GeneralString; |
| 325 | GraphicString return TOK_GraphicString; |
| 326 | IA5String return TOK_IA5String; |
| 327 | IDENTIFIER return TOK_IDENTIFIER; |
| 328 | IMPLICIT return TOK_IMPLICIT; |
| 329 | IMPLIED return TOK_IMPLIED; |
| 330 | IMPORTS return TOK_IMPORTS; |
| 331 | INCLUDES return TOK_INCLUDES; |
| 332 | INSTANCE return TOK_INSTANCE; |
| 333 | INTEGER return TOK_INTEGER; |
| 334 | INTERSECTION return TOK_INTERSECTION; |
| 335 | ISO646String return TOK_ISO646String; |
| 336 | MAX return TOK_MAX; |
| 337 | MIN return TOK_MIN; |
| 338 | MINUS-INFINITY return TOK_MINUS_INFINITY; |
| 339 | NULL return TOK_NULL; |
| 340 | NumericString return TOK_NumericString; |
| 341 | OBJECT return TOK_OBJECT; |
| 342 | ObjectDescriptor return TOK_ObjectDescriptor; |
| 343 | OCTET return TOK_OCTET; |
| 344 | OF return TOK_OF; |
| 345 | OPTIONAL return TOK_OPTIONAL; |
| 346 | PATTERN return TOK_PATTERN; |
| 347 | PDV return TOK_PDV; |
| 348 | PLUS-INFINITY return TOK_PLUS_INFINITY; |
| 349 | PRESENT return TOK_PRESENT; |
| 350 | PrintableString return TOK_PrintableString; |
| 351 | PRIVATE return TOK_PRIVATE; |
| 352 | REAL return TOK_REAL; |
| 353 | RELATIVE-OID return TOK_RELATIVE_OID; |
| 354 | SEQUENCE return TOK_SEQUENCE; |
| 355 | SET return TOK_SET; |
| 356 | SIZE return TOK_SIZE; |
| 357 | STRING return TOK_STRING; |
| 358 | SYNTAX return TOK_SYNTAX; |
| 359 | T61String return TOK_T61String; |
| 360 | TAGS return TOK_TAGS; |
| 361 | TeletexString return TOK_TeletexString; |
| 362 | TRUE return TOK_TRUE; |
| 363 | TYPE-IDENTIFIER return TOK_TYPE_IDENTIFIER; |
| 364 | UNION return TOK_UNION; |
| 365 | UNIQUE return TOK_UNIQUE; |
| 366 | UNIVERSAL return TOK_UNIVERSAL; |
| 367 | UniversalString { |
| 368 | if(TYPE_LIFETIME(1994, 0)) |
| 369 | return TOK_UniversalString; |
| 370 | REJECT; |
| 371 | } |
| 372 | UTCTime return TOK_UTCTime; |
| 373 | UTF8String { |
| 374 | if(TYPE_LIFETIME(1994, 0)) |
| 375 | return TOK_UTF8String; |
| 376 | REJECT; |
| 377 | } |
| 378 | VideotexString return TOK_VideotexString; |
| 379 | VisibleString return TOK_VisibleString; |
| 380 | WITH return TOK_WITH; |
| 381 | |
| 382 | |
| 383 | <INITIAL,with_syntax>&[A-Z][A-Za-z0-9-]* { |
| 384 | CHECK_DASHES; |
| 385 | asn1p_lval.tv_str = strdup(yytext); |
| 386 | return TOK_typefieldreference; |
| 387 | } |
| 388 | |
| 389 | <INITIAL,with_syntax>&[a-z][a-zA-Z0-9-]* { |
| 390 | CHECK_DASHES; |
| 391 | asn1p_lval.tv_str = strdup(yytext); |
| 392 | return TOK_valuefieldreference; |
| 393 | } |
| 394 | |
| 395 | |
| 396 | [a-z][a-zA-Z0-9-]* { |
| 397 | CHECK_DASHES; |
| 398 | asn1p_lval.tv_str = strdup(yytext); |
| 399 | return TOK_identifier; |
| 400 | } |
| 401 | |
| 402 | /* |
| 403 | * objectclassreference |
| 404 | */ |
| 405 | [A-Z][A-Z0-9-]* { |
| 406 | CHECK_DASHES; |
| 407 | asn1p_lval.tv_str = strdup(yytext); |
| 408 | return TOK_objectclassreference; |
| 409 | } |
| 410 | |
| 411 | /* |
| 412 | * typereference, modulereference |
| 413 | * NOTE: TOK_objectclassreference must be combined |
| 414 | * with this token to produce true typereference. |
| 415 | */ |
| 416 | [A-Z][A-Za-z0-9-]* { |
| 417 | CHECK_DASHES; |
| 418 | asn1p_lval.tv_str = strdup(yytext); |
| 419 | return TOK_typereference; |
| 420 | } |
| 421 | |
| 422 | "::=" return TOK_PPEQ; |
| 423 | |
| 424 | "..." return TOK_ThreeDots; |
| 425 | ".." return TOK_TwoDots; |
| 426 | |
| 427 | [(){},;:|!.&@\[\]] return yytext[0]; |
| 428 | |
| 429 | {WSP}+ /* Ignore whitespace */ |
| 430 | |
| 431 | [^A-Za-z0-9:=,{}<.@()[]'\"|&^*;!-] { |
| 432 | if(TYPE_LIFETIME(1994, 0)) |
| 433 | fprintf(stderr, "ERROR: "); |
| 434 | fprintf(stderr, |
| 435 | "Symbol '%c' at line %d is prohibited " |
| 436 | "by ASN.1:1994 and ASN.1:1997\n", |
| 437 | yytext[0], yylineno); |
| 438 | if(TYPE_LIFETIME(1994, 0)) |
| 439 | return -1; |
| 440 | } |
| 441 | |
| 442 | <with_syntax>{ |
| 443 | |
| 444 | [^&{} \t\r\v\f\n]+ { |
| 445 | asn1p_lval.tv_opaque.buf = strdup(yytext); |
| 446 | asn1p_lval.tv_opaque.len = yyleng; |
| 447 | return TOK_opaque; |
| 448 | } |
| 449 | |
| 450 | {WSP}+ { |
| 451 | asn1p_lval.tv_opaque.buf = strdup(yytext); |
| 452 | asn1p_lval.tv_opaque.len = yyleng; |
| 453 | return TOK_opaque; |
| 454 | } |
| 455 | |
| 456 | "}" { |
| 457 | yy_pop_state(); |
| 458 | return '}'; |
| 459 | } |
| 460 | |
| 461 | } |
| 462 | |
| 463 | |
| 464 | <*>. { |
| 465 | fprintf(stderr, |
| 466 | "Unexpected token at line %d: \"%s\"\n", |
| 467 | yylineno, yytext); |
| 468 | while(YYSTATE != INITIAL) |
| 469 | yy_pop_state(); |
| 470 | yy_top_state(); /* Just to use this function. */ |
| 471 | yyterminate(); |
| 472 | yy_fatal_error("Unexpected token"); |
| 473 | return -1; |
| 474 | } |
| 475 | |
| 476 | <*><<EOF>> { |
| 477 | while(YYSTATE != INITIAL) |
| 478 | yy_pop_state(); |
| 479 | yyterminate(); |
| 480 | } |
| 481 | |
| 482 | |
| 483 | %% |
| 484 | |
| 485 | /* |
| 486 | * Very dirty but wonderful hack allowing to rule states from within .y file. |
| 487 | */ |
| 488 | void |
| 489 | asn1p_lexer_hack_push_opaque_state() { |
| 490 | yy_push_state(opaque); |
| 491 | } |
| 492 | |
| 493 | /* |
| 494 | * Another hack which disables recognizing some tokens when inside WITH SYNTAX. |
| 495 | */ |
| 496 | void |
| 497 | asn1p_lexer_hack_enable_with_syntax() { |
| 498 | yy_push_state(with_syntax); |
| 499 | } |
| 500 | |
| 501 | /* |
| 502 | * Check that a token does not end with dash and does not contain |
| 503 | * several dashes in succession. |
| 504 | * "Name", "Type-Id", "T-y-p-e-i-d" are OK |
| 505 | * "end-", "vustom--value" are INVALID |
| 506 | */ |
| 507 | static int |
| 508 | _check_dashes(char *ptr) { |
| 509 | int prev_dash = 0; |
| 510 | |
| 511 | assert(*ptr != '-'); |
| 512 | |
| 513 | for(;; ptr++) { |
| 514 | switch(*ptr) { |
| 515 | case '-': |
| 516 | if(prev_dash++) /* No double dashes */ |
| 517 | return -1; |
| 518 | continue; |
| 519 | case '\0': |
| 520 | if(prev_dash) /* No dashes at the end */ |
| 521 | return -1; |
| 522 | break; |
| 523 | default: |
| 524 | prev_dash = 0; |
| 525 | continue; |
| 526 | } |
| 527 | break; |
| 528 | } |
| 529 | |
| 530 | return 0; |
| 531 | } |
| 532 | |
| 533 | static asn1_integer_t |
| 534 | asn1p_atoi(char *ptr) { |
| 535 | asn1_integer_t value; |
| 536 | errno = 0; /* Clear the error code */ |
| 537 | |
| 538 | if(sizeof(value) <= sizeof(int)) { |
| 539 | value = strtol(ptr, 0, 10); |
| 540 | } else { |
| 541 | #ifdef HAVE_STRTOIMAX |
| 542 | value = strtoimax(ptr, 0, 10); |
| 543 | #elif HAVE_STRTOLL |
| 544 | value = strtoll(ptr, 0, 10); |
| 545 | #else |
| 546 | value = strtol(ptr, 0, 10); |
| 547 | #endif |
| 548 | } |
| 549 | |
| 550 | if(errno == ERANGE) { |
| 551 | fprintf(stderr, |
| 552 | "Value \"%s\" at line %d is too large " |
| 553 | "for this compiler! Please contact the vendor.", |
| 554 | ptr, yylineno); |
| 555 | errno = ERANGE; /* Restore potentially clobbered errno */ |
| 556 | } |
| 557 | |
| 558 | return value; |
| 559 | } |
| 560 | |