ada-lex.l revision 1.6
1/* FLEX lexer for Ada expressions, for GDB. 2 Copyright (C) 1994-2016 Free Software Foundation, Inc. 3 4 This file is part of GDB. 5 6 This program is free software; you can redistribute it and/or modify 7 it under the terms of the GNU General Public License as published by 8 the Free Software Foundation; either version 3 of the License, or 9 (at your option) any later version. 10 11 This program is distributed in the hope that it will be useful, 12 but WITHOUT ANY WARRANTY; without even the implied warranty of 13 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the 14 GNU General Public License for more details. 15 16 You should have received a copy of the GNU General Public License 17 along with this program. If not, see <http://www.gnu.org/licenses/>. */ 18 19/*----------------------------------------------------------------------*/ 20 21/* The converted version of this file is to be included in ada-exp.y, */ 22/* the Ada parser for gdb. The function yylex obtains characters from */ 23/* the global pointer lexptr. It returns a syntactic category for */ 24/* each successive token and places a semantic value into yylval */ 25/* (ada-lval), defined by the parser. */ 26 27DIG [0-9] 28NUM10 ({DIG}({DIG}|_)*) 29HEXDIG [0-9a-f] 30NUM16 ({HEXDIG}({HEXDIG}|_)*) 31OCTDIG [0-7] 32LETTER [a-z_] 33ID ({LETTER}({LETTER}|{DIG})*|"<"{LETTER}({LETTER}|{DIG})*">") 34WHITE [ \t\n] 35TICK ("'"{WHITE}*) 36GRAPHIC [a-z0-9 #&'()*+,-./:;<>=_|!$%?@\[\]\\^`{}~] 37OPER ([-+*/=<>&]|"<="|">="|"**"|"/="|"and"|"or"|"xor"|"not"|"mod"|"rem"|"abs") 38 39EXP (e[+-]{NUM10}) 40POSEXP (e"+"?{NUM10}) 41 42%{ 43 44#define NUMERAL_WIDTH 256 45#define LONGEST_SIGN ((ULONGEST) 1 << (sizeof(LONGEST) * HOST_CHAR_BIT - 1)) 46 47/* Temporary staging for numeric literals. */ 48static char numbuf[NUMERAL_WIDTH]; 49 static void canonicalizeNumeral (char *s1, const char *); 50static struct stoken processString (const char*, int); 51static int processInt (struct parser_state *, const char *, const char *, 52 const char *); 53static int processReal (struct parser_state *, const char *); 54static struct stoken processId (const char *, int); 55static int processAttribute (const char *); 56static int find_dot_all (const char *); 57static void rewind_to_char (int); 58 59#undef YY_DECL 60#define YY_DECL static int yylex ( void ) 61 62/* Flex generates a static function "input" which is not used. 63 Defining YY_NO_INPUT comments it out. */ 64#define YY_NO_INPUT 65 66#undef YY_INPUT 67#define YY_INPUT(BUF, RESULT, MAX_SIZE) \ 68 if ( *lexptr == '\000' ) \ 69 (RESULT) = YY_NULL; \ 70 else \ 71 { \ 72 *(BUF) = *lexptr; \ 73 (RESULT) = 1; \ 74 lexptr += 1; \ 75 } 76 77static int find_dot_all (const char *); 78 79%} 80 81%option case-insensitive interactive nodefault 82 83%s BEFORE_QUAL_QUOTE 84 85%% 86 87{WHITE} { } 88 89"--".* { yyterminate(); } 90 91{NUM10}{POSEXP} { 92 canonicalizeNumeral (numbuf, yytext); 93 return processInt (pstate, NULL, numbuf, 94 strrchr (numbuf, 'e') + 1); 95 } 96 97{NUM10} { 98 canonicalizeNumeral (numbuf, yytext); 99 return processInt (pstate, NULL, numbuf, NULL); 100 } 101 102{NUM10}"#"{HEXDIG}({HEXDIG}|_)*"#"{POSEXP} { 103 canonicalizeNumeral (numbuf, yytext); 104 return processInt (pstate, numbuf, 105 strchr (numbuf, '#') + 1, 106 strrchr(numbuf, '#') + 1); 107 } 108 109{NUM10}"#"{HEXDIG}({HEXDIG}|_)*"#" { 110 canonicalizeNumeral (numbuf, yytext); 111 return processInt (pstate, numbuf, strchr (numbuf, '#') + 1, 112 NULL); 113 } 114 115"0x"{HEXDIG}+ { 116 canonicalizeNumeral (numbuf, yytext+2); 117 return processInt (pstate, "16#", numbuf, NULL); 118 } 119 120 121{NUM10}"."{NUM10}{EXP} { 122 canonicalizeNumeral (numbuf, yytext); 123 return processReal (pstate, numbuf); 124 } 125 126{NUM10}"."{NUM10} { 127 canonicalizeNumeral (numbuf, yytext); 128 return processReal (pstate, numbuf); 129 } 130 131{NUM10}"#"{NUM16}"."{NUM16}"#"{EXP} { 132 error (_("Based real literals not implemented yet.")); 133 } 134 135{NUM10}"#"{NUM16}"."{NUM16}"#" { 136 error (_("Based real literals not implemented yet.")); 137 } 138 139<INITIAL>"'"({GRAPHIC}|\")"'" { 140 yylval.typed_val.type = type_char (pstate); 141 yylval.typed_val.val = yytext[1]; 142 return CHARLIT; 143 } 144 145<INITIAL>"'[\""{HEXDIG}{2}"\"]'" { 146 int v; 147 yylval.typed_val.type = type_char (pstate); 148 sscanf (yytext+3, "%2x", &v); 149 yylval.typed_val.val = v; 150 return CHARLIT; 151 } 152 153\"({GRAPHIC}|"[\""({HEXDIG}{2}|\")"\"]")*\" { 154 yylval.sval = processString (yytext+1, yyleng-2); 155 return STRING; 156 } 157 158\" { 159 error (_("ill-formed or non-terminated string literal")); 160 } 161 162 163if { 164 rewind_to_char ('i'); 165 return 0; 166 } 167 168task { 169 rewind_to_char ('t'); 170 return 0; 171 } 172 173thread{WHITE}+{DIG} { 174 /* This keyword signals the end of the expression and 175 will be processed separately. */ 176 rewind_to_char ('t'); 177 return 0; 178 } 179 180 /* ADA KEYWORDS */ 181 182abs { return ABS; } 183and { return _AND_; } 184else { return ELSE; } 185in { return IN; } 186mod { return MOD; } 187new { return NEW; } 188not { return NOT; } 189null { return NULL_PTR; } 190or { return OR; } 191others { return OTHERS; } 192rem { return REM; } 193then { return THEN; } 194xor { return XOR; } 195 196 /* BOOLEAN "KEYWORDS" */ 197 198 /* True and False are not keywords in Ada, but rather enumeration constants. 199 However, the boolean type is no longer represented as an enum, so True 200 and False are no longer defined in symbol tables. We compromise by 201 making them keywords (when bare). */ 202 203true { return TRUEKEYWORD; } 204false { return FALSEKEYWORD; } 205 206 /* ATTRIBUTES */ 207 208{TICK}[a-zA-Z][a-zA-Z]+ { BEGIN INITIAL; return processAttribute (yytext+1); } 209 210 /* PUNCTUATION */ 211 212"=>" { return ARROW; } 213".." { return DOTDOT; } 214"**" { return STARSTAR; } 215":=" { return ASSIGN; } 216"/=" { return NOTEQUAL; } 217"<=" { return LEQ; } 218">=" { return GEQ; } 219 220<BEFORE_QUAL_QUOTE>"'" { BEGIN INITIAL; return '\''; } 221 222[-&*+./:<>=|;\[\]] { return yytext[0]; } 223 224"," { if (paren_depth == 0 && comma_terminates) 225 { 226 rewind_to_char (','); 227 return 0; 228 } 229 else 230 return ','; 231 } 232 233"(" { paren_depth += 1; return '('; } 234")" { if (paren_depth == 0) 235 { 236 rewind_to_char (')'); 237 return 0; 238 } 239 else 240 { 241 paren_depth -= 1; 242 return ')'; 243 } 244 } 245 246"."{WHITE}*all { return DOT_ALL; } 247 248"."{WHITE}*{ID} { 249 yylval.sval = processId (yytext+1, yyleng-1); 250 return DOT_ID; 251 } 252 253{ID}({WHITE}*"."{WHITE}*({ID}|\"{OPER}\"))*(" "*"'")? { 254 int all_posn = find_dot_all (yytext); 255 256 if (all_posn == -1 && yytext[yyleng-1] == '\'') 257 { 258 BEGIN BEFORE_QUAL_QUOTE; 259 yyless (yyleng-1); 260 } 261 else if (all_posn >= 0) 262 yyless (all_posn); 263 yylval.sval = processId (yytext, yyleng); 264 return NAME; 265 } 266 267 268 /* GDB EXPRESSION CONSTRUCTS */ 269 270"'"[^']+"'"{WHITE}*:: { 271 yyless (yyleng - 2); 272 yylval.sval = processId (yytext, yyleng); 273 return NAME; 274 } 275 276"::" { return COLONCOLON; } 277 278[{}@] { return yytext[0]; } 279 280 /* REGISTERS AND GDB CONVENIENCE VARIABLES */ 281 282"$"({LETTER}|{DIG}|"$")* { 283 yylval.sval.ptr = yytext; 284 yylval.sval.length = yyleng; 285 return SPECIAL_VARIABLE; 286 } 287 288 /* CATCH-ALL ERROR CASE */ 289 290. { error (_("Invalid character '%s' in expression."), yytext); } 291%% 292 293#include <ctype.h> 294/* Initialize the lexer for processing new expression. */ 295 296static void 297lexer_init (FILE *inp) 298{ 299 BEGIN INITIAL; 300 yyrestart (inp); 301} 302 303 304/* Copy S2 to S1, removing all underscores, and downcasing all letters. */ 305 306static void 307canonicalizeNumeral (char *s1, const char *s2) 308{ 309 for (; *s2 != '\000'; s2 += 1) 310 { 311 if (*s2 != '_') 312 { 313 *s1 = tolower(*s2); 314 s1 += 1; 315 } 316 } 317 s1[0] = '\000'; 318} 319 320/* Interprets the prefix of NUM that consists of digits of the given BASE 321 as an integer of that BASE, with the string EXP as an exponent. 322 Puts value in yylval, and returns INT, if the string is valid. Causes 323 an error if the number is improperly formated. BASE, if NULL, defaults 324 to "10", and EXP to "1". The EXP does not contain a leading 'e' or 'E'. 325 */ 326 327static int 328processInt (struct parser_state *par_state, const char *base0, 329 const char *num0, const char *exp0) 330{ 331 ULONGEST result; 332 long exp; 333 int base; 334 const char *trailer; 335 336 if (base0 == NULL) 337 base = 10; 338 else 339 { 340 base = strtol (base0, (char **) NULL, 10); 341 if (base < 2 || base > 16) 342 error (_("Invalid base: %d."), base); 343 } 344 345 if (exp0 == NULL) 346 exp = 0; 347 else 348 exp = strtol(exp0, (char **) NULL, 10); 349 350 errno = 0; 351 result = strtoulst (num0, &trailer, base); 352 if (errno == ERANGE) 353 error (_("Integer literal out of range")); 354 if (isxdigit(*trailer)) 355 error (_("Invalid digit `%c' in based literal"), *trailer); 356 357 while (exp > 0) 358 { 359 if (result > (ULONG_MAX / base)) 360 error (_("Integer literal out of range")); 361 result *= base; 362 exp -= 1; 363 } 364 365 if ((result >> (gdbarch_int_bit (parse_gdbarch (par_state))-1)) == 0) 366 yylval.typed_val.type = type_int (par_state); 367 else if ((result >> (gdbarch_long_bit (parse_gdbarch (par_state))-1)) == 0) 368 yylval.typed_val.type = type_long (par_state); 369 else if (((result >> (gdbarch_long_bit (parse_gdbarch (par_state))-1)) >> 1) == 0) 370 { 371 /* We have a number representable as an unsigned integer quantity. 372 For consistency with the C treatment, we will treat it as an 373 anonymous modular (unsigned) quantity. Alas, the types are such 374 that we need to store .val as a signed quantity. Sorry 375 for the mess, but C doesn't officially guarantee that a simple 376 assignment does the trick (no, it doesn't; read the reference manual). 377 */ 378 yylval.typed_val.type 379 = builtin_type (parse_gdbarch (par_state))->builtin_unsigned_long; 380 if (result & LONGEST_SIGN) 381 yylval.typed_val.val = 382 (LONGEST) (result & ~LONGEST_SIGN) 383 - (LONGEST_SIGN>>1) - (LONGEST_SIGN>>1); 384 else 385 yylval.typed_val.val = (LONGEST) result; 386 return INT; 387 } 388 else 389 yylval.typed_val.type = type_long_long (par_state); 390 391 yylval.typed_val.val = (LONGEST) result; 392 return INT; 393} 394 395static int 396processReal (struct parser_state *par_state, const char *num0) 397{ 398 sscanf (num0, "%" DOUBLEST_SCAN_FORMAT, &yylval.typed_val_float.dval); 399 400 yylval.typed_val_float.type = type_float (par_state); 401 if (sizeof(DOUBLEST) >= gdbarch_double_bit (parse_gdbarch (par_state)) 402 / TARGET_CHAR_BIT) 403 yylval.typed_val_float.type = type_double (par_state); 404 if (sizeof(DOUBLEST) >= gdbarch_long_double_bit (parse_gdbarch (par_state)) 405 / TARGET_CHAR_BIT) 406 yylval.typed_val_float.type = type_long_double (par_state); 407 408 return FLOAT; 409} 410 411 412/* Store a canonicalized version of NAME0[0..LEN-1] in yylval.ssym. The 413 resulting string is valid until the next call to ada_parse. If 414 NAME0 contains the substring "___", it is assumed to be already 415 encoded and the resulting name is equal to it. Otherwise, it differs 416 from NAME0 in that: 417 + Characters between '...' or <...> are transfered verbatim to 418 yylval.ssym. 419 + <, >, and trailing "'" characters in quoted sequences are removed 420 (a leading quote is preserved to indicate that the name is not to be 421 GNAT-encoded). 422 + Unquoted whitespace is removed. 423 + Unquoted alphabetic characters are mapped to lower case. 424 Result is returned as a struct stoken, but for convenience, the string 425 is also null-terminated. Result string valid until the next call of 426 ada_parse. 427 */ 428static struct stoken 429processId (const char *name0, int len) 430{ 431 char *name = (char *) obstack_alloc (&temp_parse_space, len + 11); 432 int i0, i; 433 struct stoken result; 434 435 result.ptr = name; 436 while (len > 0 && isspace (name0[len-1])) 437 len -= 1; 438 439 if (strstr (name0, "___") != NULL) 440 { 441 strncpy (name, name0, len); 442 name[len] = '\000'; 443 result.length = len; 444 return result; 445 } 446 447 i = i0 = 0; 448 while (i0 < len) 449 { 450 if (isalnum (name0[i0])) 451 { 452 name[i] = tolower (name0[i0]); 453 i += 1; i0 += 1; 454 } 455 else switch (name0[i0]) 456 { 457 default: 458 name[i] = name0[i0]; 459 i += 1; i0 += 1; 460 break; 461 case ' ': case '\t': 462 i0 += 1; 463 break; 464 case '\'': 465 do 466 { 467 name[i] = name0[i0]; 468 i += 1; i0 += 1; 469 } 470 while (i0 < len && name0[i0] != '\''); 471 i0 += 1; 472 break; 473 case '<': 474 i0 += 1; 475 while (i0 < len && name0[i0] != '>') 476 { 477 name[i] = name0[i0]; 478 i += 1; i0 += 1; 479 } 480 i0 += 1; 481 break; 482 } 483 } 484 name[i] = '\000'; 485 486 result.length = i; 487 return result; 488} 489 490/* Return TEXT[0..LEN-1], a string literal without surrounding quotes, 491 with special hex character notations replaced with characters. 492 Result valid until the next call to ada_parse. */ 493 494static struct stoken 495processString (const char *text, int len) 496{ 497 const char *p; 498 char *q; 499 const char *lim = text + len; 500 struct stoken result; 501 502 q = (char *) obstack_alloc (&temp_parse_space, len); 503 result.ptr = q; 504 p = text; 505 while (p < lim) 506 { 507 if (p[0] == '[' && p[1] == '"' && p+2 < lim) 508 { 509 if (p[2] == '"') /* "...["""]... */ 510 { 511 *q = '"'; 512 p += 4; 513 } 514 else 515 { 516 int chr; 517 sscanf (p+2, "%2x", &chr); 518 *q = (char) chr; 519 p += 5; 520 } 521 } 522 else 523 *q = *p; 524 q += 1; 525 p += 1; 526 } 527 result.length = q - result.ptr; 528 return result; 529} 530 531/* Returns the position within STR of the '.' in a 532 '.{WHITE}*all' component of a dotted name, or -1 if there is none. 533 Note: we actually don't need this routine, since 'all' can never be an 534 Ada identifier. Thus, looking up foo.all or foo.all.x as a name 535 must fail, and will eventually be interpreted as (foo).all or 536 (foo).all.x. However, this does avoid an extraneous lookup. */ 537 538static int 539find_dot_all (const char *str) 540{ 541 int i; 542 543 for (i = 0; str[i] != '\000'; i++) 544 if (str[i] == '.') 545 { 546 int i0 = i; 547 548 do 549 i += 1; 550 while (isspace (str[i])); 551 552 if (strncasecmp (str + i, "all", 3) == 0 553 && !isalnum (str[i + 3]) && str[i + 3] != '_') 554 return i0; 555 } 556 return -1; 557} 558 559/* Returns non-zero iff string SUBSEQ matches a subsequence of STR, ignoring 560 case. */ 561 562static int 563subseqMatch (const char *subseq, const char *str) 564{ 565 if (subseq[0] == '\0') 566 return 1; 567 else if (str[0] == '\0') 568 return 0; 569 else if (tolower (subseq[0]) == tolower (str[0])) 570 return subseqMatch (subseq+1, str+1) || subseqMatch (subseq, str+1); 571 else 572 return subseqMatch (subseq, str+1); 573} 574 575 576static struct { const char *name; int code; } 577attributes[] = { 578 { "address", TICK_ADDRESS }, 579 { "unchecked_access", TICK_ACCESS }, 580 { "unrestricted_access", TICK_ACCESS }, 581 { "access", TICK_ACCESS }, 582 { "first", TICK_FIRST }, 583 { "last", TICK_LAST }, 584 { "length", TICK_LENGTH }, 585 { "max", TICK_MAX }, 586 { "min", TICK_MIN }, 587 { "modulus", TICK_MODULUS }, 588 { "pos", TICK_POS }, 589 { "range", TICK_RANGE }, 590 { "size", TICK_SIZE }, 591 { "tag", TICK_TAG }, 592 { "val", TICK_VAL }, 593 { NULL, -1 } 594}; 595 596/* Return the syntactic code corresponding to the attribute name or 597 abbreviation STR. */ 598 599static int 600processAttribute (const char *str) 601{ 602 int i, k; 603 604 for (i = 0; attributes[i].code != -1; i += 1) 605 if (strcasecmp (str, attributes[i].name) == 0) 606 return attributes[i].code; 607 608 for (i = 0, k = -1; attributes[i].code != -1; i += 1) 609 if (subseqMatch (str, attributes[i].name)) 610 { 611 if (k == -1) 612 k = i; 613 else 614 error (_("ambiguous attribute name: `%s'"), str); 615 } 616 if (k == -1) 617 error (_("unrecognized attribute: `%s'"), str); 618 619 return attributes[k].code; 620} 621 622/* Back up lexptr by yyleng and then to the rightmost occurrence of 623 character CH, case-folded (there must be one). WARNING: since 624 lexptr points to the next input character that Flex has not yet 625 transferred to its internal buffer, the use of this function 626 depends on the assumption that Flex calls YY_INPUT only when it is 627 logically necessary to do so (thus, there is no reading ahead 628 farther than needed to identify the next token.) */ 629 630static void 631rewind_to_char (int ch) 632{ 633 lexptr -= yyleng; 634 while (toupper (*lexptr) != toupper (ch)) 635 lexptr -= 1; 636 yyrestart (NULL); 637} 638 639int 640yywrap(void) 641{ 642 return 1; 643} 644 645/* Dummy definition to suppress warnings about unused static definitions. */ 646typedef void (*dummy_function) (); 647dummy_function ada_flex_use[] = 648{ 649 (dummy_function) yyunput 650}; 651