gdb/ada-lex.l

   1 /* FLEX lexer for Ada expressions, for GDB.
   2    Copyright (C) 1994-2014 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
  27 DIG     [0-9]
  28 NUM10   ({DIG}({DIG}|_)*)
  29 HEXDIG  [0-9a-f]
  30 NUM16   ({HEXDIG}({HEXDIG}|_)*)
  31 OCTDIG  [0-7]
  32 LETTER  [a-z_]
  33 ID      ({LETTER}({LETTER}|{DIG})*|"<"{LETTER}({LETTER}|{DIG})*">")
  34 WHITE   [ \t\n]
  35 TICK    ("'"{WHITE}*)
  36 GRAPHIC [a-z0-9 #&'()*+,-./:;<>=_|!$%?@\[\]\\^`{}~]
  37 OPER    ([-+*/=<>&]|"<="|">="|"**"|"/="|"and"|"or"|"xor"|"not"|"mod"|"rem"|"abs")
  38
  39 EXP     (e[+-]{NUM10})
  40 POSEXP  (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.  */
  48 static char numbuf[NUMERAL_WIDTH];
  49  static void canonicalizeNumeral (char *s1, const char *);
  50 static struct stoken processString (const char*, int);
  51 static int processInt (const char *, const char *, const char *);
  52 static int processReal (const char *);
  53 static struct stoken processId (const char *, int);
  54 static int processAttribute (const char *);
  55 static int find_dot_all (const char *);
  56 static void rewind_to_char (int);
  57
  58 #undef YY_DECL
  59 #define YY_DECL static int yylex ( void )
  60
  61 /* Flex generates a static function "input" which is not used.
  62    Defining YY_NO_INPUT comments it out.  */
  63 #define YY_NO_INPUT
  64
  65 #undef YY_INPUT
  66 #define YY_INPUT(BUF, RESULT, MAX_SIZE) \
  67     if ( *lexptr == '\000' ) \
  68       (RESULT) = YY_NULL; \
  69     else \
  70       { \
  71         *(BUF) = *lexptr; \
  72         (RESULT) = 1; \
  73         lexptr += 1; \
  74       }
  75
  76 static int find_dot_all (const char *);
  77
  78 %}
  79
  80 %option case-insensitive interactive nodefault
  81
  82 %s BEFORE_QUAL_QUOTE
  83
  84 %%
  85
  86 {WHITE}          { }
  87
  88 "--".*           { yyterminate(); }
  89
  90 {NUM10}{POSEXP}  {
  91                    canonicalizeNumeral (numbuf, yytext);
  92                    return processInt (NULL, numbuf, strrchr(numbuf, 'e')+1);
  93                  }
  94
  95 {NUM10}          {
  96                    canonicalizeNumeral (numbuf, yytext);
  97                    return processInt (NULL, numbuf, NULL);
  98                  }
  99
 100 {NUM10}"#"{HEXDIG}({HEXDIG}|_)*"#"{POSEXP} {
 101                    canonicalizeNumeral (numbuf, yytext);
 102                    return processInt (numbuf,
 103                                       strchr (numbuf, '#') + 1,
 104                                       strrchr(numbuf, '#') + 1);
 105                  }
 106
 107 {NUM10}"#"{HEXDIG}({HEXDIG}|_)*"#" {
 108                    canonicalizeNumeral (numbuf, yytext);
 109                    return processInt (numbuf, strchr (numbuf, '#') + 1, NULL);
 110                  }
 111
 112 "0x"{HEXDIG}+   {
 113                   canonicalizeNumeral (numbuf, yytext+2);
 114                   return processInt ("16#", numbuf, NULL);
 115                 }
 116
 117
 118 {NUM10}"."{NUM10}{EXP} {
 119                    canonicalizeNumeral (numbuf, yytext);
 120                    return processReal (numbuf);
 121                 }
 122
 123 {NUM10}"."{NUM10} {
 124                    canonicalizeNumeral (numbuf, yytext);
 125                    return processReal (numbuf);
 126                 }
 127
 128 {NUM10}"#"{NUM16}"."{NUM16}"#"{EXP} {
 129                    error (_("Based real literals not implemented yet."));
 130                 }
 131
 132 {NUM10}"#"{NUM16}"."{NUM16}"#" {
 133                    error (_("Based real literals not implemented yet."));
 134                 }
 135
 136 <INITIAL>"'"({GRAPHIC}|\")"'" {
 137                    yylval.typed_val.type = type_char ();
 138                    yylval.typed_val.val = yytext[1];
 139                    return CHARLIT;
 140                 }
 141
 142 <INITIAL>"'[\""{HEXDIG}{2}"\"]'"   {
 143                    int v;
 144                    yylval.typed_val.type = type_char ();
 145                    sscanf (yytext+3, "%2x", &v);
 146                    yylval.typed_val.val = v;
 147                    return CHARLIT;
 148                 }
 149
 150 \"({GRAPHIC}|"[\""({HEXDIG}{2}|\")"\"]")*\"   {
 151                    yylval.sval = processString (yytext+1, yyleng-2);
 152                    return STRING;
 153                 }
 154
 155 \"              {
 156                    error (_("ill-formed or non-terminated string literal"));
 157                 }
 158
 159
 160 if              {
 161                   rewind_to_char ('i');
 162                   return 0;
 163                 }
 164
 165 task            {
 166                   rewind_to_char ('t');
 167                   return 0;
 168                 }
 169
 170 thread{WHITE}+{DIG} {
 171                   /* This keyword signals the end of the expression and
 172                      will be processed separately.  */
 173                   rewind_to_char ('t');
 174                   return 0;
 175                 }
 176
 177         /* ADA KEYWORDS */
 178
 179 abs             { return ABS; }
 180 and             { return _AND_; }
 181 else            { return ELSE; }
 182 in              { return IN; }
 183 mod             { return MOD; }
 184 new             { return NEW; }
 185 not             { return NOT; }
 186 null            { return NULL_PTR; }
 187 or              { return OR; }
 188 others          { return OTHERS; }
 189 rem             { return REM; }
 190 then            { return THEN; }
 191 xor             { return XOR; }
 192
 193         /* BOOLEAN "KEYWORDS" */
 194
 195  /* True and False are not keywords in Ada, but rather enumeration constants.
 196     However, the boolean type is no longer represented as an enum, so True
 197     and False are no longer defined in symbol tables.  We compromise by
 198     making them keywords (when bare). */
 199
 200 true            { return TRUEKEYWORD; }
 201 false           { return FALSEKEYWORD; }
 202
 203         /* ATTRIBUTES */
 204
 205 {TICK}[a-zA-Z][a-zA-Z]+ { return processAttribute (yytext+1); }
 206
 207         /* PUNCTUATION */
 208
 209 "=>"            { return ARROW; }
 210 ".."            { return DOTDOT; }
 211 "**"            { return STARSTAR; }
 212 ":="            { return ASSIGN; }
 213 "/="            { return NOTEQUAL; }
 214 "<="            { return LEQ; }
 215 ">="            { return GEQ; }
 216
 217 <BEFORE_QUAL_QUOTE>"'" { BEGIN INITIAL; return '\''; }
 218
 219 [-&*+./:<>=|;\[\]] { return yytext[0]; }
 220
 221 ","             { if (paren_depth == 0 && comma_terminates)
 222                     {
 223                       rewind_to_char (',');
 224                       return 0;
 225                     }
 226                   else
 227                     return ',';
 228                 }
 229
 230 "("             { paren_depth += 1; return '('; }
 231 ")"             { if (paren_depth == 0)
 232                     {
 233                       rewind_to_char (')');
 234                       return 0;
 235                     }
 236                   else
 237                     {
 238                       paren_depth -= 1;
 239                       return ')';
 240                     }
 241                 }
 242
 243 "."{WHITE}*all  { return DOT_ALL; }
 244
 245 "."{WHITE}*{ID} {
 246                   yylval.sval = processId (yytext+1, yyleng-1);
 247                   return DOT_ID;
 248                 }
 249
 250 {ID}({WHITE}*"."{WHITE}*({ID}|\"{OPER}\"))*(" "*"'")?  {
 251                   int all_posn = find_dot_all (yytext);
 252
 253                   if (all_posn == -1 && yytext[yyleng-1] == '\'')
 254                     {
 255                       BEGIN BEFORE_QUAL_QUOTE;
 256                       yyless (yyleng-1);
 257                     }
 258                   else if (all_posn >= 0)
 259                     yyless (all_posn);
 260                   yylval.sval = processId (yytext, yyleng);
 261                   return NAME;
 262                }
 263
 264
 265         /* GDB EXPRESSION CONSTRUCTS  */
 266
 267 "'"[^']+"'"{WHITE}*:: {
 268                   yyless (yyleng - 2);
 269                   yylval.sval = processId (yytext, yyleng);
 270                   return NAME;
 271                 }
 272
 273 "::"            { return COLONCOLON; }
 274
 275 [{}@]           { return yytext[0]; }
 276
 277         /* REGISTERS AND GDB CONVENIENCE VARIABLES */
 278
 279 "$"({LETTER}|{DIG}|"$")*  {
 280                   yylval.sval.ptr = yytext;
 281                   yylval.sval.length = yyleng;
 282                   return SPECIAL_VARIABLE;
 283                 }
 284
 285         /* CATCH-ALL ERROR CASE */
 286
 287 .               { error (_("Invalid character '%s' in expression."), yytext); }
 288 %%
 289
 290 #include <ctype.h>
 291 #include <string.h>
 292
 293 /* Initialize the lexer for processing new expression. */
 294
 295 static void
 296 lexer_init (FILE *inp)
 297 {
 298   BEGIN INITIAL;
 299   yyrestart (inp);
 300 }
 301
 302
 303 /* Copy S2 to S1, removing all underscores, and downcasing all letters.  */
 304
 305 static void
 306 canonicalizeNumeral (char *s1, const char *s2)
 307 {
 308   for (; *s2 != '\000'; s2 += 1)
 309     {
 310       if (*s2 != '_')
 311         {
 312           *s1 = tolower(*s2);
 313           s1 += 1;
 314         }
 315     }
 316   s1[0] = '\000';
 317 }
 318
 319 /* Interprets the prefix of NUM that consists of digits of the given BASE
 320    as an integer of that BASE, with the string EXP as an exponent.
 321    Puts value in yylval, and returns INT, if the string is valid.  Causes
 322    an error if the number is improperly formated.   BASE, if NULL, defaults
 323    to "10", and EXP to "1".  The EXP does not contain a leading 'e' or 'E'.
 324  */
 325
 326 static int
 327 processInt (const char *base0, const char *num0, const char *exp0)
 328 {
 329   ULONGEST result;
 330   long exp;
 331   int base;
 332   const char *trailer;
 333
 334   if (base0 == NULL)
 335     base = 10;
 336   else
 337     {
 338       base = strtol (base0, (char **) NULL, 10);
 339       if (base < 2 || base > 16)
 340         error (_("Invalid base: %d."), base);
 341     }
 342
 343   if (exp0 == NULL)
 344     exp = 0;
 345   else
 346     exp = strtol(exp0, (char **) NULL, 10);
 347
 348   errno = 0;
 349   result = strtoulst (num0, &trailer, base);
 350   if (errno == ERANGE)
 351     error (_("Integer literal out of range"));
 352   if (isxdigit(*trailer))
 353     error (_("Invalid digit `%c' in based literal"), *trailer);
 354
 355   while (exp > 0)
 356     {
 357       if (result > (ULONG_MAX / base))
 358         error (_("Integer literal out of range"));
 359       result *= base;
 360       exp -= 1;
 361     }
 362
 363   if ((result >> (gdbarch_int_bit (parse_gdbarch)-1)) == 0)
 364     yylval.typed_val.type = type_int ();
 365   else if ((result >> (gdbarch_long_bit (parse_gdbarch)-1)) == 0)
 366     yylval.typed_val.type = type_long ();
 367   else if (((result >> (gdbarch_long_bit (parse_gdbarch)-1)) >> 1) == 0)
 368     {
 369       /* We have a number representable as an unsigned integer quantity.
 370          For consistency with the C treatment, we will treat it as an
 371          anonymous modular (unsigned) quantity.  Alas, the types are such
 372          that we need to store .val as a signed quantity.  Sorry
 373          for the mess, but C doesn't officially guarantee that a simple
 374          assignment does the trick (no, it doesn't; read the reference manual).
 375        */
 376       yylval.typed_val.type
 377         = builtin_type (parse_gdbarch)->builtin_unsigned_long;
 378       if (result & LONGEST_SIGN)
 379         yylval.typed_val.val =
 380           (LONGEST) (result & ~LONGEST_SIGN)
 381           - (LONGEST_SIGN>>1) - (LONGEST_SIGN>>1);
 382       else
 383         yylval.typed_val.val = (LONGEST) result;
 384       return INT;
 385     }
 386   else
 387     yylval.typed_val.type = type_long_long ();
 388
 389   yylval.typed_val.val = (LONGEST) result;
 390   return INT;
 391 }
 392
 393 static int
 394 processReal (const char *num0)
 395 {
 396   sscanf (num0, "%" DOUBLEST_SCAN_FORMAT, &yylval.typed_val_float.dval);
 397
 398   yylval.typed_val_float.type = type_float ();
 399   if (sizeof(DOUBLEST) >= gdbarch_double_bit (parse_gdbarch)
 400                             / TARGET_CHAR_BIT)
 401     yylval.typed_val_float.type = type_double ();
 402   if (sizeof(DOUBLEST) >= gdbarch_long_double_bit (parse_gdbarch)
 403                             / TARGET_CHAR_BIT)
 404     yylval.typed_val_float.type = type_long_double ();
 405
 406   return FLOAT;
 407 }
 408
 409
 410 /* Store a canonicalized version of NAME0[0..LEN-1] in yylval.ssym.  The
 411    resulting string is valid until the next call to ada_parse.  If
 412    NAME0 contains the substring "___", it is assumed to be already
 413    encoded and the resulting name is equal to it.  Otherwise, it differs
 414    from NAME0 in that:
 415     + Characters between '...' or <...> are transfered verbatim to
 416       yylval.ssym.
 417     + <, >, and trailing "'" characters in quoted sequences are removed
 418       (a leading quote is preserved to indicate that the name is not to be
 419       GNAT-encoded).
 420     + Unquoted whitespace is removed.
 421     + Unquoted alphabetic characters are mapped to lower case.
 422    Result is returned as a struct stoken, but for convenience, the string
 423    is also null-terminated.  Result string valid until the next call of
 424    ada_parse.
 425  */
 426 static struct stoken
 427 processId (const char *name0, int len)
 428 {
 429   char *name = obstack_alloc (&temp_parse_space, len + 11);
 430   int i0, i;
 431   struct stoken result;
 432
 433   result.ptr = name;
 434   while (len > 0 && isspace (name0[len-1]))
 435     len -= 1;
 436
 437   if (strstr (name0, "___") != NULL)
 438     {
 439       strncpy (name, name0, len);
 440       name[len] = '\000';
 441       result.length = len;
 442       return result;
 443     }
 444
 445   i = i0 = 0;
 446   while (i0 < len)
 447     {
 448       if (isalnum (name0[i0]))
 449         {
 450           name[i] = tolower (name0[i0]);
 451           i += 1; i0 += 1;
 452         }
 453       else switch (name0[i0])
 454         {
 455         default:
 456           name[i] = name0[i0];
 457           i += 1; i0 += 1;
 458           break;
 459         case ' ': case '\t':
 460           i0 += 1;
 461           break;
 462         case '\'':
 463           do
 464             {
 465               name[i] = name0[i0];
 466               i += 1; i0 += 1;
 467             }
 468           while (i0 < len && name0[i0] != '\'');
 469           i0 += 1;
 470           break;
 471         case '<':
 472           i0 += 1;
 473           while (i0 < len && name0[i0] != '>')
 474             {
 475               name[i] = name0[i0];
 476               i += 1; i0 += 1;
 477             }
 478           i0 += 1;
 479           break;
 480         }
 481     }
 482   name[i] = '\000';
 483
 484   result.length = i;
 485   return result;
 486 }
 487
 488 /* Return TEXT[0..LEN-1], a string literal without surrounding quotes,
 489    with special hex character notations replaced with characters.
 490    Result valid until the next call to ada_parse.  */
 491
 492 static struct stoken
 493 processString (const char *text, int len)
 494 {
 495   const char *p;
 496   char *q;
 497   const char *lim = text + len;
 498   struct stoken result;
 499
 500   q = obstack_alloc (&temp_parse_space, len);
 501   result.ptr = q;
 502   p = text;
 503   while (p < lim)
 504     {
 505       if (p[0] == '[' && p[1] == '"' && p+2 < lim)
 506          {
 507            if (p[2] == '"')  /* "...["""]... */
 508              {
 509                *q = '"';
 510                p += 4;
 511              }
 512            else
 513              {
 514                int chr;
 515                sscanf (p+2, "%2x", &chr);
 516                *q = (char) chr;
 517                p += 5;
 518              }
 519          }
 520        else
 521          *q = *p;
 522        q += 1;
 523        p += 1;
 524      }
 525   result.length = q - result.ptr;
 526   return result;
 527 }
 528
 529 /* Returns the position within STR of the '.' in a
 530    '.{WHITE}*all' component of a dotted name, or -1 if there is none.
 531    Note: we actually don't need this routine, since 'all' can never be an
 532    Ada identifier.  Thus, looking up foo.all or foo.all.x as a name
 533    must fail, and will eventually be interpreted as (foo).all or
 534    (foo).all.x.  However, this does avoid an extraneous lookup. */
 535
 536 static int
 537 find_dot_all (const char *str)
 538 {
 539   int i;
 540
 541   for (i = 0; str[i] != '\000'; i++)
 542     if (str[i] == '.')
 543       {
 544         int i0 = i;
 545
 546         do
 547           i += 1;
 548         while (isspace (str[i]));
 549
 550         if (strncasecmp (str + i, "all", 3) == 0
 551             && !isalnum (str[i + 3]) && str[i + 3] != '_')
 552           return i0;
 553       }
 554   return -1;
 555 }
 556
 557 /* Returns non-zero iff string SUBSEQ matches a subsequence of STR, ignoring
 558    case.  */
 559
 560 static int
 561 subseqMatch (const char *subseq, const char *str)
 562 {
 563   if (subseq[0] == '\0')
 564     return 1;
 565   else if (str[0] == '\0')
 566     return 0;
 567   else if (tolower (subseq[0]) == tolower (str[0]))
 568     return subseqMatch (subseq+1, str+1) || subseqMatch (subseq, str+1);
 569   else
 570     return subseqMatch (subseq, str+1);
 571 }
 572
 573
 574 static struct { const char *name; int code; }
 575 attributes[] = {
 576   { "address", TICK_ADDRESS },
 577   { "unchecked_access", TICK_ACCESS },
 578   { "unrestricted_access", TICK_ACCESS },
 579   { "access", TICK_ACCESS },
 580   { "first", TICK_FIRST },
 581   { "last", TICK_LAST },
 582   { "length", TICK_LENGTH },
 583   { "max", TICK_MAX },
 584   { "min", TICK_MIN },
 585   { "modulus", TICK_MODULUS },
 586   { "pos", TICK_POS },
 587   { "range", TICK_RANGE },
 588   { "size", TICK_SIZE },
 589   { "tag", TICK_TAG },
 590   { "val", TICK_VAL },
 591   { NULL, -1 }
 592 };
 593
 594 /* Return the syntactic code corresponding to the attribute name or
 595    abbreviation STR.  */
 596
 597 static int
 598 processAttribute (const char *str)
 599 {
 600   int i, k;
 601
 602   for (i = 0; attributes[i].code != -1; i += 1)
 603     if (strcasecmp (str, attributes[i].name) == 0)
 604       return attributes[i].code;
 605
 606   for (i = 0, k = -1; attributes[i].code != -1; i += 1)
 607     if (subseqMatch (str, attributes[i].name))
 608       {
 609         if (k == -1)
 610           k = i;
 611         else
 612           error (_("ambiguous attribute name: `%s'"), str);
 613       }
 614   if (k == -1)
 615     error (_("unrecognized attribute: `%s'"), str);
 616
 617   return attributes[k].code;
 618 }
 619
 620 /* Back up lexptr by yyleng and then to the rightmost occurrence of
 621    character CH, case-folded (there must be one).  WARNING: since
 622    lexptr points to the next input character that Flex has not yet
 623    transferred to its internal buffer, the use of this function
 624    depends on the assumption that Flex calls YY_INPUT only when it is
 625    logically necessary to do so (thus, there is no reading ahead
 626    farther than needed to identify the next token.)  */
 627
 628 static void
 629 rewind_to_char (int ch)
 630 {
 631   lexptr -= yyleng;
 632   while (toupper (*lexptr) != toupper (ch))
 633     lexptr -= 1;
 634   yyrestart (NULL);
 635 }
 636
 637 int
 638 yywrap(void)
 639 {
 640   return 1;
 641 }
 642
 643 /* Dummy definition to suppress warnings about unused static definitions. */
 644 typedef void (*dummy_function) ();
 645 dummy_function ada_flex_use[] =
 646 {
 647   (dummy_function) yyunput
 648 };