1 /* YACC grammar for Modula-2 expressions, for GDB.
2 Copyright (C) 1986-2019 Free Software Foundation, Inc.
3 Generated from expread.y (now c-exp.y) and contributed by the Department
4 of Computer Science at the State University of New York at Buffalo, 1991.
6 This file is part of GDB.
8 This program is free software; you can redistribute it and/or modify
9 it under the terms of the GNU General Public License as published by
10 the Free Software Foundation; either version 3 of the License, or
11 (at your option) any later version.
13 This program is distributed in the hope that it will be useful,
14 but WITHOUT ANY WARRANTY; without even the implied warranty of
15 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
16 GNU General Public License for more details.
18 You should have received a copy of the GNU General Public License
19 along with this program. If not, see <http://www.gnu.org/licenses/>. */
21 /* Parse a Modula-2 expression from text in a string,
22 and return the result as a struct expression pointer.
23 That structure contains arithmetic operations in reverse polish,
24 with constants represented by operations that are followed by special data.
25 See expression.h for the details of the format.
26 What is important here is that it can be built up sequentially
27 during the process of parsing; the lower levels of the tree always
28 come first in the result.
30 Note that malloc's and realloc's in this file are transformed to
31 xmalloc and xrealloc respectively by the same sed command in the
32 makefile that remaps any other malloc/realloc inserted by the parser
33 generator. Doing this with #defines and trying to control the interaction
34 with include files (<malloc.h> and <stdlib.h> for example) just became
35 too messy, particularly when such includes can be inserted at random
36 times by the parser generator. */
41 #include "expression.h"
44 #include "parser-defs.h"
46 #include "bfd.h" /* Required by objfiles.h. */
47 #include "symfile.h" /* Required by objfiles.h. */
48 #include "objfiles.h" /* For have_full_symbols and have_partial_symbols */
51 #define parse_type(ps) builtin_type (ps->gdbarch ())
52 #define parse_m2_type(ps) builtin_m2_type (ps->gdbarch ())
54 /* Remap normal yacc parser interface names (yyparse, yylex, yyerror,
56 #define GDB_YY_REMAP_PREFIX m2_
59 /* The state of the parser, used internally when we are parsing the
62 static struct parser_state *pstate = NULL;
66 static int yylex (void);
68 static void yyerror (const char *);
70 static int parse_number (int);
72 /* The sign of the number being parsed. */
73 static int number_sign = 1;
77 /* Although the yacc "value" of an expression is not used,
78 since the result is stored in the structure being created,
79 other node types do have values. */
90 const struct block *bval;
91 enum exp_opcode opcode;
92 struct internalvar *ivar;
98 %type <voidval> exp type_exp start set
99 %type <voidval> variable
104 %token <lval> INT HEX ERROR
105 %token <ulval> UINT M2_TRUE M2_FALSE CHAR
108 /* Both NAME and TYPENAME tokens represent symbols in the input,
109 and both convey their data as strings.
110 But a TYPENAME is a string that happens to be defined as a typedef
111 or builtin type name (such as int or char)
112 and a NAME is any other symbol.
114 Contexts where this distinction is not important can use the
115 nonterminal "name", which matches either NAME or TYPENAME. */
118 %token <sval> NAME BLOCKNAME IDENT VARNAME
119 %token <sval> TYPENAME
121 %token SIZE CAP ORD HIGH ABS MIN_FUNC MAX_FUNC FLOAT_FUNC VAL CHR ODD TRUNC
123 %token INC DEC INCL EXCL
125 /* The GDB scope operator */
128 %token <voidval> DOLLAR_VARIABLE
134 %left '<' '>' LEQ GEQ '=' NOTEQUAL '#' IN
136 %left LOGICAL_AND '&'
139 %left '*' '/' DIV MOD
141 %right '^' DOT '[' '('
144 /* This is not an actual token ; it is used for precedence.
156 { write_exp_elt_opcode (pstate, OP_TYPE);
157 write_exp_elt_type (pstate, $1);
158 write_exp_elt_opcode (pstate, OP_TYPE);
164 exp : exp '^' %prec UNARY
165 { write_exp_elt_opcode (pstate, UNOP_IND); }
169 { number_sign = -1; }
172 write_exp_elt_opcode (pstate, UNOP_NEG); }
175 exp : '+' exp %prec UNARY
176 { write_exp_elt_opcode (pstate, UNOP_PLUS); }
179 exp : not_exp exp %prec UNARY
180 { write_exp_elt_opcode (pstate, UNOP_LOGICAL_NOT); }
187 exp : CAP '(' exp ')'
188 { write_exp_elt_opcode (pstate, UNOP_CAP); }
191 exp : ORD '(' exp ')'
192 { write_exp_elt_opcode (pstate, UNOP_ORD); }
195 exp : ABS '(' exp ')'
196 { write_exp_elt_opcode (pstate, UNOP_ABS); }
199 exp : HIGH '(' exp ')'
200 { write_exp_elt_opcode (pstate, UNOP_HIGH); }
203 exp : MIN_FUNC '(' type ')'
204 { write_exp_elt_opcode (pstate, UNOP_MIN);
205 write_exp_elt_type (pstate, $3);
206 write_exp_elt_opcode (pstate, UNOP_MIN); }
209 exp : MAX_FUNC '(' type ')'
210 { write_exp_elt_opcode (pstate, UNOP_MAX);
211 write_exp_elt_type (pstate, $3);
212 write_exp_elt_opcode (pstate, UNOP_MAX); }
215 exp : FLOAT_FUNC '(' exp ')'
216 { write_exp_elt_opcode (pstate, UNOP_FLOAT); }
219 exp : VAL '(' type ',' exp ')'
220 { write_exp_elt_opcode (pstate, BINOP_VAL);
221 write_exp_elt_type (pstate, $3);
222 write_exp_elt_opcode (pstate, BINOP_VAL); }
225 exp : CHR '(' exp ')'
226 { write_exp_elt_opcode (pstate, UNOP_CHR); }
229 exp : ODD '(' exp ')'
230 { write_exp_elt_opcode (pstate, UNOP_ODD); }
233 exp : TRUNC '(' exp ')'
234 { write_exp_elt_opcode (pstate, UNOP_TRUNC); }
237 exp : TSIZE '(' exp ')'
238 { write_exp_elt_opcode (pstate, UNOP_SIZEOF); }
241 exp : SIZE exp %prec UNARY
242 { write_exp_elt_opcode (pstate, UNOP_SIZEOF); }
246 exp : INC '(' exp ')'
247 { write_exp_elt_opcode (pstate, UNOP_PREINCREMENT); }
250 exp : INC '(' exp ',' exp ')'
251 { write_exp_elt_opcode (pstate, BINOP_ASSIGN_MODIFY);
252 write_exp_elt_opcode (pstate, BINOP_ADD);
253 write_exp_elt_opcode (pstate,
254 BINOP_ASSIGN_MODIFY); }
257 exp : DEC '(' exp ')'
258 { write_exp_elt_opcode (pstate, UNOP_PREDECREMENT);}
261 exp : DEC '(' exp ',' exp ')'
262 { write_exp_elt_opcode (pstate, BINOP_ASSIGN_MODIFY);
263 write_exp_elt_opcode (pstate, BINOP_SUB);
264 write_exp_elt_opcode (pstate,
265 BINOP_ASSIGN_MODIFY); }
269 { write_exp_elt_opcode (pstate, STRUCTOP_STRUCT);
270 write_exp_string (pstate, $3);
271 write_exp_elt_opcode (pstate, STRUCTOP_STRUCT); }
278 { error (_("Sets are not implemented."));}
281 exp : INCL '(' exp ',' exp ')'
282 { error (_("Sets are not implemented."));}
285 exp : EXCL '(' exp ',' exp ')'
286 { error (_("Sets are not implemented."));}
289 set : '{' arglist '}'
290 { error (_("Sets are not implemented."));}
291 | type '{' arglist '}'
292 { error (_("Sets are not implemented."));}
296 /* Modula-2 array subscript notation [a,b,c...] */
298 /* This function just saves the number of arguments
299 that follow in the list. It is *not* specific to
301 { pstate->start_arglist(); }
302 non_empty_arglist ']' %prec DOT
303 { write_exp_elt_opcode (pstate, MULTI_SUBSCRIPT);
304 write_exp_elt_longcst (pstate,
305 pstate->end_arglist());
306 write_exp_elt_opcode (pstate, MULTI_SUBSCRIPT); }
309 exp : exp '[' exp ']'
310 { write_exp_elt_opcode (pstate, BINOP_SUBSCRIPT); }
314 /* This is to save the value of arglist_len
315 being accumulated by an outer function call. */
316 { pstate->start_arglist (); }
317 arglist ')' %prec DOT
318 { write_exp_elt_opcode (pstate, OP_FUNCALL);
319 write_exp_elt_longcst (pstate,
320 pstate->end_arglist ());
321 write_exp_elt_opcode (pstate, OP_FUNCALL); }
328 { pstate->arglist_len = 1; }
331 arglist : arglist ',' exp %prec ABOVE_COMMA
332 { pstate->arglist_len++; }
337 { pstate->arglist_len = 1; }
341 : non_empty_arglist ',' exp %prec ABOVE_COMMA
342 { pstate->arglist_len++; }
346 exp : '{' type '}' exp %prec UNARY
347 { write_exp_elt_opcode (pstate, UNOP_MEMVAL);
348 write_exp_elt_type (pstate, $2);
349 write_exp_elt_opcode (pstate, UNOP_MEMVAL); }
352 exp : type '(' exp ')' %prec UNARY
353 { write_exp_elt_opcode (pstate, UNOP_CAST);
354 write_exp_elt_type (pstate, $1);
355 write_exp_elt_opcode (pstate, UNOP_CAST); }
362 /* Binary operators in order of decreasing precedence. Note that some
363 of these operators are overloaded! (ie. sets) */
367 { write_exp_elt_opcode (pstate, BINOP_REPEAT); }
371 { write_exp_elt_opcode (pstate, BINOP_MUL); }
375 { write_exp_elt_opcode (pstate, BINOP_DIV); }
379 { write_exp_elt_opcode (pstate, BINOP_INTDIV); }
383 { write_exp_elt_opcode (pstate, BINOP_REM); }
387 { write_exp_elt_opcode (pstate, BINOP_ADD); }
391 { write_exp_elt_opcode (pstate, BINOP_SUB); }
395 { write_exp_elt_opcode (pstate, BINOP_EQUAL); }
398 exp : exp NOTEQUAL exp
399 { write_exp_elt_opcode (pstate, BINOP_NOTEQUAL); }
401 { write_exp_elt_opcode (pstate, BINOP_NOTEQUAL); }
405 { write_exp_elt_opcode (pstate, BINOP_LEQ); }
409 { write_exp_elt_opcode (pstate, BINOP_GEQ); }
413 { write_exp_elt_opcode (pstate, BINOP_LESS); }
417 { write_exp_elt_opcode (pstate, BINOP_GTR); }
420 exp : exp LOGICAL_AND exp
421 { write_exp_elt_opcode (pstate, BINOP_LOGICAL_AND); }
425 { write_exp_elt_opcode (pstate, BINOP_LOGICAL_OR); }
429 { write_exp_elt_opcode (pstate, BINOP_ASSIGN); }
436 { write_exp_elt_opcode (pstate, OP_BOOL);
437 write_exp_elt_longcst (pstate, (LONGEST) $1);
438 write_exp_elt_opcode (pstate, OP_BOOL); }
442 { write_exp_elt_opcode (pstate, OP_BOOL);
443 write_exp_elt_longcst (pstate, (LONGEST) $1);
444 write_exp_elt_opcode (pstate, OP_BOOL); }
448 { write_exp_elt_opcode (pstate, OP_LONG);
449 write_exp_elt_type (pstate,
450 parse_m2_type (pstate)->builtin_int);
451 write_exp_elt_longcst (pstate, (LONGEST) $1);
452 write_exp_elt_opcode (pstate, OP_LONG); }
457 write_exp_elt_opcode (pstate, OP_LONG);
458 write_exp_elt_type (pstate,
459 parse_m2_type (pstate)
461 write_exp_elt_longcst (pstate, (LONGEST) $1);
462 write_exp_elt_opcode (pstate, OP_LONG);
467 { write_exp_elt_opcode (pstate, OP_LONG);
468 write_exp_elt_type (pstate,
469 parse_m2_type (pstate)
471 write_exp_elt_longcst (pstate, (LONGEST) $1);
472 write_exp_elt_opcode (pstate, OP_LONG); }
477 { write_exp_elt_opcode (pstate, OP_FLOAT);
478 write_exp_elt_type (pstate,
479 parse_m2_type (pstate)
481 write_exp_elt_floatcst (pstate, $1);
482 write_exp_elt_opcode (pstate, OP_FLOAT); }
488 exp : SIZE '(' type ')' %prec UNARY
489 { write_exp_elt_opcode (pstate, OP_LONG);
490 write_exp_elt_type (pstate,
491 parse_type (pstate)->builtin_int);
492 write_exp_elt_longcst (pstate,
493 (LONGEST) TYPE_LENGTH ($3));
494 write_exp_elt_opcode (pstate, OP_LONG); }
498 { write_exp_elt_opcode (pstate, OP_M2_STRING);
499 write_exp_string (pstate, $1);
500 write_exp_elt_opcode (pstate, OP_M2_STRING); }
503 /* This will be used for extensions later. Like adding modules. */
505 { $$ = SYMBOL_BLOCK_VALUE($1); }
510 = lookup_symbol (copy_name ($1).c_str (),
511 pstate->expression_context_block,
512 VAR_DOMAIN, 0).symbol;
517 /* GDB scope operator */
518 fblock : block COLONCOLON BLOCKNAME
520 = lookup_symbol (copy_name ($3).c_str (), $1,
521 VAR_DOMAIN, 0).symbol;
522 if (!tem || SYMBOL_CLASS (tem) != LOC_BLOCK)
523 error (_("No function \"%s\" in specified context."),
524 copy_name ($3).c_str ());
529 /* Useful for assigning to PROCEDURE variables */
531 { write_exp_elt_opcode (pstate, OP_VAR_VALUE);
532 write_exp_elt_block (pstate, NULL);
533 write_exp_elt_sym (pstate, $1);
534 write_exp_elt_opcode (pstate, OP_VAR_VALUE); }
537 /* GDB internal ($foo) variable */
538 variable: DOLLAR_VARIABLE
541 /* GDB scope operator */
542 variable: block COLONCOLON NAME
543 { struct block_symbol sym
544 = lookup_symbol (copy_name ($3).c_str (), $1,
548 error (_("No symbol \"%s\" in specified context."),
549 copy_name ($3).c_str ());
550 if (symbol_read_needs_frame (sym.symbol))
551 pstate->block_tracker->update (sym);
553 write_exp_elt_opcode (pstate, OP_VAR_VALUE);
554 write_exp_elt_block (pstate, sym.block);
555 write_exp_elt_sym (pstate, sym.symbol);
556 write_exp_elt_opcode (pstate, OP_VAR_VALUE); }
559 /* Base case for variables. */
561 { struct block_symbol sym;
562 struct field_of_this_result is_a_field_of_this;
565 = lookup_symbol (copy_name ($1).c_str (),
566 pstate->expression_context_block,
568 &is_a_field_of_this);
572 if (symbol_read_needs_frame (sym.symbol))
573 pstate->block_tracker->update (sym);
575 write_exp_elt_opcode (pstate, OP_VAR_VALUE);
576 write_exp_elt_block (pstate, sym.block);
577 write_exp_elt_sym (pstate, sym.symbol);
578 write_exp_elt_opcode (pstate, OP_VAR_VALUE);
582 struct bound_minimal_symbol msymbol;
583 std::string arg = copy_name ($1);
586 lookup_bound_minimal_symbol (arg.c_str ());
587 if (msymbol.minsym != NULL)
588 write_exp_msymbol (pstate, msymbol);
589 else if (!have_full_symbols () && !have_partial_symbols ())
590 error (_("No symbol table is loaded. Use the \"symbol-file\" command."));
592 error (_("No symbol \"%s\" in current context."),
601 = lookup_typename (pstate->language (),
602 copy_name ($1).c_str (),
603 pstate->expression_context_block,
611 /* Take care of parsing a number (anything that starts with a digit).
612 Set yylval and return the token type; update lexptr.
613 LEN is the number of characters in it. */
615 /*** Needs some error checking for the float case ***/
618 parse_number (int olen)
620 const char *p = pstate->lexptr;
624 int base = input_radix;
626 int unsigned_p = number_sign == 1 ? 1 : 0;
633 else if(p[len-1] == 'C' || p[len-1] == 'B')
636 ischar = p[len-1] == 'C';
640 /* Scan the number */
641 for (c = 0; c < len; c++)
643 if (p[c] == '.' && base == 10)
645 /* It's a float since it contains a point. */
646 if (!parse_float (p, len,
647 parse_m2_type (pstate)->builtin_real,
651 pstate->lexptr += len;
654 if (p[c] == '.' && base != 10)
655 error (_("Floating point numbers must be base 10."));
656 if (base == 10 && (p[c] < '0' || p[c] > '9'))
657 error (_("Invalid digit \'%c\' in number."),p[c]);
664 if( base == 8 && (c == '8' || c == '9'))
665 error (_("Invalid digit \'%c\' in octal number."),c);
666 if (c >= '0' && c <= '9')
670 if (base == 16 && c >= 'A' && c <= 'F')
678 if(!unsigned_p && number_sign == 1 && (prevn >= n))
679 unsigned_p=1; /* Try something unsigned */
680 /* Don't do the range check if n==i and i==0, since that special
681 case will give an overflow error. */
682 if(RANGE_CHECK && n!=i && i)
684 if((unsigned_p && (unsigned)prevn >= (unsigned)n) ||
685 ((!unsigned_p && number_sign==-1) && -prevn <= -n))
686 range_error (_("Overflow on numeric constant."));
692 if(*p == 'B' || *p == 'C' || *p == 'H')
693 pstate->lexptr++; /* Advance past B,C or H */
700 else if ( unsigned_p && number_sign == 1)
705 else if((unsigned_p && (n<0))) {
706 range_error (_("Overflow on numeric constant -- number too large."));
707 /* But, this can return if range_check == range_warn. */
722 { {'<', '>'}, NOTEQUAL },
723 { {':', '='}, ASSIGN },
726 { {':', ':'}, COLONCOLON },
730 /* Some specific keywords */
737 static struct keyword keytab[] =
740 {"IN", IN },/* Note space after IN */
741 {"AND", LOGICAL_AND},
759 {"FLOAT", FLOAT_FUNC },
765 /* Depth of parentheses. */
766 static int paren_depth;
768 /* Read one token, getting characters through lexptr. */
770 /* This is where we will check to make sure that the language and the
771 operators used are compatible */
779 const char *tokstart;
784 pstate->prev_lexptr = pstate->lexptr;
786 tokstart = pstate->lexptr;
789 /* See if it is a special token of length 2 */
790 for( i = 0 ; i < (int) (sizeof tokentab2 / sizeof tokentab2[0]) ; i++)
791 if (strncmp (tokentab2[i].name, tokstart, 2) == 0)
794 return tokentab2[i].token;
797 switch (c = *tokstart)
814 if (paren_depth == 0)
821 if (pstate->comma_terminates && paren_depth == 0)
827 /* Might be a floating point number. */
828 if (pstate->lexptr[1] >= '0' && pstate->lexptr[1] <= '9')
829 break; /* Falls into number code. */
836 /* These are character tokens that appear as-is in the YACC grammar */
859 for (namelen = 1; (c = tokstart[namelen]) != quote && c != '\0'; namelen++)
862 c = tokstart[++namelen];
863 if (c >= '0' && c <= '9')
865 c = tokstart[++namelen];
866 if (c >= '0' && c <= '9')
867 c = tokstart[++namelen];
871 error (_("Unterminated string or character constant."));
872 yylval.sval.ptr = tokstart + 1;
873 yylval.sval.length = namelen - 1;
874 pstate->lexptr += namelen + 1;
876 if(namelen == 2) /* Single character */
878 yylval.ulval = tokstart[1];
885 /* Is it a number? */
886 /* Note: We have already dealt with the case of the token '.'.
887 See case '.' above. */
888 if ((c >= '0' && c <= '9'))
891 int got_dot = 0, got_e = 0;
892 const char *p = tokstart;
897 if (!got_e && (*p == 'e' || *p == 'E'))
899 else if (!got_dot && *p == '.')
901 else if (got_e && (p[-1] == 'e' || p[-1] == 'E')
902 && (*p == '-' || *p == '+'))
903 /* This is the sign of the exponent, not the end of the
906 else if ((*p < '0' || *p > '9') &&
907 (*p < 'A' || *p > 'F') &&
908 (*p != 'H')) /* Modula-2 hexadecimal number */
911 toktype = parse_number (p - tokstart);
912 if (toktype == ERROR)
914 char *err_copy = (char *) alloca (p - tokstart + 1);
916 memcpy (err_copy, tokstart, p - tokstart);
917 err_copy[p - tokstart] = 0;
918 error (_("Invalid number \"%s\"."), err_copy);
924 if (!(c == '_' || c == '$'
925 || (c >= 'a' && c <= 'z') || (c >= 'A' && c <= 'Z')))
926 /* We must have come across a bad character (e.g. ';'). */
927 error (_("Invalid character '%c' in expression."), c);
929 /* It's a name. See how long it is. */
931 for (c = tokstart[namelen];
932 (c == '_' || c == '$' || (c >= '0' && c <= '9')
933 || (c >= 'a' && c <= 'z') || (c >= 'A' && c <= 'Z'));
934 c = tokstart[++namelen])
937 /* The token "if" terminates the expression and is NOT
938 removed from the input stream. */
939 if (namelen == 2 && tokstart[0] == 'i' && tokstart[1] == 'f')
944 pstate->lexptr += namelen;
946 /* Lookup special keywords */
947 for(i = 0 ; i < (int) (sizeof(keytab) / sizeof(keytab[0])) ; i++)
948 if (namelen == strlen (keytab[i].keyw)
949 && strncmp (tokstart, keytab[i].keyw, namelen) == 0)
950 return keytab[i].token;
952 yylval.sval.ptr = tokstart;
953 yylval.sval.length = namelen;
955 if (*tokstart == '$')
957 write_dollar_variable (pstate, yylval.sval);
958 return DOLLAR_VARIABLE;
961 /* Use token-type BLOCKNAME for symbols that happen to be defined as
962 functions. If this is not so, then ...
963 Use token-type TYPENAME for symbols that happen to be defined
964 currently as names of types; NAME for other symbols.
965 The caller is not constrained to care about the distinction. */
967 std::string tmp = copy_name (yylval.sval);
970 if (lookup_symtab (tmp.c_str ()))
972 sym = lookup_symbol (tmp.c_str (), pstate->expression_context_block,
973 VAR_DOMAIN, 0).symbol;
974 if (sym && SYMBOL_CLASS (sym) == LOC_BLOCK)
976 if (lookup_typename (pstate->language (),
977 tmp.c_str (), pstate->expression_context_block, 1))
982 switch(SYMBOL_CLASS (sym))
988 case LOC_REGPARM_ADDR:
991 case LOC_CONST_BYTES:
992 case LOC_OPTIMIZED_OUT:
1003 error (_("internal: Undefined class in m2lex()"));
1006 case LOC_UNRESOLVED:
1007 error (_("internal: Unforseen case in m2lex()"));
1010 error (_("unhandled token in m2lex()"));
1016 /* Built-in BOOLEAN type. This is sort of a hack. */
1017 if (strncmp (tokstart, "TRUE", 4) == 0)
1022 else if (strncmp (tokstart, "FALSE", 5) == 0)
1029 /* Must be another type of name... */
1035 m2_parse (struct parser_state *par_state)
1037 /* Setting up the parser state. */
1038 scoped_restore pstate_restore = make_scoped_restore (&pstate);
1039 gdb_assert (par_state != NULL);
1047 yyerror (const char *msg)
1049 if (pstate->prev_lexptr)
1050 pstate->lexptr = pstate->prev_lexptr;
1052 error (_("A %s in expression, near `%s'."), msg, pstate->lexptr);