1 /* YACC grammar for Chill expressions, for GDB.
2 Copyright 1992, 1993, 1994 Free Software Foundation, Inc.
4 This file is part of GDB.
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 2 of the License, or
9 (at your option) any later version.
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.
16 You should have received a copy of the GNU General Public License
17 along with this program; if not, write to the Free Software
18 Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. */
20 /* Parse a Chill expression from text in a string,
21 and return the result as a struct expression pointer.
22 That structure contains arithmetic operations in reverse polish,
23 with constants represented by operations that are followed by special data.
24 See expression.h for the details of the format.
25 What is important here is that it can be built up sequentially
26 during the process of parsing; the lower levels of the tree always
27 come first in the result.
29 Note that malloc's and realloc's in this file are transformed to
30 xmalloc and xrealloc respectively by the same sed command in the
31 makefile that remaps any other malloc/realloc inserted by the parser
32 generator. Doing this with #defines and trying to control the interaction
33 with include files (<malloc.h> and <stdlib.h> for example) just became
34 too messy, particularly when such includes can be inserted at random
35 times by the parser generator.
37 Also note that the language accepted by this parser is more liberal
38 than the one accepted by an actual Chill compiler. For example, the
39 language rule that a simple name string can not be one of the reserved
40 simple name strings is not enforced (e.g "case" is not treated as a
41 reserved name). Another example is that Chill is a strongly typed
42 language, and certain expressions that violate the type constraints
43 may still be evaluated if gdb can do so in a meaningful manner, while
44 such expressions would be rejected by the compiler. The reason for
45 this more liberal behavior is the philosophy that the debugger
46 is intended to be a tool that is used by the programmer when things
47 go wrong, and as such, it should provide as few artificial barriers
48 to it's use as possible. If it can do something meaningful, even
49 something that violates language contraints that are enforced by the
50 compiler, it should do so without complaint.
59 #include "expression.h"
62 #include "parser-defs.h"
64 #include "bfd.h" /* Required by objfiles.h. */
65 #include "symfile.h" /* Required by objfiles.h. */
66 #include "objfiles.h" /* For have_full_symbols and have_partial_symbols */
68 /* Remap normal yacc parser interface names (yyparse, yylex, yyerror, etc),
69 as well as gratuitiously global symbol names, so we can have multiple
70 yacc generated parsers in gdb. Note that these are only the variables
71 produced by yacc. If other parser generators (bison, byacc, etc) produce
72 additional global names that conflict at link time, then those parser
73 generators need to be fixed instead of adding those names to this list. */
75 #define yymaxdepth chill_maxdepth
76 #define yyparse chill_parse
77 #define yylex chill_lex
78 #define yyerror chill_error
79 #define yylval chill_lval
80 #define yychar chill_char
81 #define yydebug chill_debug
82 #define yypact chill_pact
85 #define yydef chill_def
86 #define yychk chill_chk
87 #define yypgo chill_pgo
88 #define yyact chill_act
89 #define yyexca chill_exca
90 #define yyerrflag chill_errflag
91 #define yynerrs chill_nerrs
95 #define yy_yys chill_yys
96 #define yystate chill_state
97 #define yytmp chill_tmp
99 #define yy_yyv chill_yyv
100 #define yyval chill_val
101 #define yylloc chill_lloc
102 #define yyreds chill_reds /* With YYDEBUG defined */
103 #define yytoks chill_toks /* With YYDEBUG defined */
106 #define YYDEBUG 0 /* Default to no yydebug support */
110 yyparse PARAMS ((void));
113 yylex PARAMS ((void));
116 yyerror PARAMS ((char *));
120 /* Although the yacc "value" of an expression is not used,
121 since the result is stored in the structure being created,
122 other node types do have values. */
127 unsigned LONGEST ulval;
137 struct symtoken ssym;
140 enum exp_opcode opcode;
141 struct internalvar *ivar;
147 %token <voidval> FIXME_01
148 %token <voidval> FIXME_02
149 %token <voidval> FIXME_03
150 %token <voidval> FIXME_04
151 %token <voidval> FIXME_05
152 %token <voidval> FIXME_06
153 %token <voidval> FIXME_07
154 %token <voidval> FIXME_08
155 %token <voidval> FIXME_09
156 %token <voidval> FIXME_10
157 %token <voidval> FIXME_11
158 %token <voidval> FIXME_12
159 %token <voidval> FIXME_13
160 %token <voidval> FIXME_14
161 %token <voidval> FIXME_15
162 %token <voidval> FIXME_16
163 %token <voidval> FIXME_17
164 %token <voidval> FIXME_18
165 %token <voidval> FIXME_19
166 %token <voidval> FIXME_20
167 %token <voidval> FIXME_21
168 %token <voidval> FIXME_22
169 %token <voidval> FIXME_24
170 %token <voidval> FIXME_25
171 %token <voidval> FIXME_26
172 %token <voidval> FIXME_27
173 %token <voidval> FIXME_28
174 %token <voidval> FIXME_29
175 %token <voidval> FIXME_30
177 %token <typed_val> INTEGER_LITERAL
178 %token <ulval> BOOLEAN_LITERAL
179 %token <typed_val> CHARACTER_LITERAL
180 %token <dval> FLOAT_LITERAL
181 %token <ssym> GENERAL_PROCEDURE_NAME
182 %token <ssym> LOCATION_NAME
183 %token <voidval> SET_LITERAL
184 %token <voidval> EMPTINESS_LITERAL
185 %token <sval> CHARACTER_STRING_LITERAL
186 %token <sval> BIT_STRING_LITERAL
187 %token <tsym> TYPENAME
188 %token <sval> FIELD_NAME
193 %token <voidval> CASE
195 %token <voidval> ESAC
196 %token <voidval> LOGIOR
197 %token <voidval> ORIF
198 %token <voidval> LOGXOR
199 %token <voidval> LOGAND
200 %token <voidval> ANDIF
202 %token <voidval> NOTEQUAL
212 %token <voidval> SLASH_SLASH
216 %token <voidval> POINTER
217 %token <voidval> RECEIVE
224 %token <voidval> THEN
225 %token <voidval> ELSE
227 %token <voidval> ELSIF
228 %token <voidval> ILLEGAL_TOKEN
230 %token <voidval> PRED
231 %token <voidval> SUCC
233 %token <voidval> CARD
234 %token <voidval> MAX_TOKEN
235 %token <voidval> MIN_TOKEN
236 %token <voidval> SIZE
237 %token <voidval> UPPER
238 %token <voidval> LOWER
239 %token <voidval> LENGTH
241 /* Tokens which are not Chill tokens used in expressions, but rather GDB
242 specific things that we recognize in the same context as Chill tokens
243 (register names for example). */
245 %token <lval> GDB_REGNAME /* Machine register name */
246 %token <lval> GDB_LAST /* Value history */
247 %token <ivar> GDB_VARIABLE /* Convenience variable */
248 %token <voidval> GDB_ASSIGNMENT /* Assign value to somewhere */
250 %type <voidval> access_name
251 %type <voidval> primitive_value
252 %type <voidval> value_name
253 %type <voidval> literal
254 %type <voidval> tuple
255 %type <voidval> value_string_element
256 %type <voidval> value_string_slice
257 %type <voidval> value_array_slice
258 %type <voidval> expression_conversion
259 %type <voidval> value_procedure_call
260 %type <voidval> value_built_in_routine_call
261 %type <voidval> chill_value_built_in_routine_call
262 %type <voidval> start_expression
263 %type <voidval> zero_adic_operator
264 %type <voidval> parenthesised_expression
265 %type <voidval> value
266 %type <voidval> undefined_value
267 %type <voidval> expression
268 %type <voidval> conditional_expression
269 %type <voidval> then_alternative
270 %type <voidval> else_alternative
271 %type <voidval> sub_expression
272 %type <voidval> value_case_alternative
273 %type <voidval> operand_0
274 %type <voidval> operand_1
275 %type <voidval> operand_2
276 %type <voidval> operand_3
277 %type <voidval> operand_4
278 %type <voidval> operand_5
279 %type <voidval> operand_6
280 %type <voidval> synonym_name
281 %type <voidval> value_enumeration_name
282 %type <voidval> value_do_with_name
283 %type <voidval> value_receive_name
284 %type <voidval> string_primitive_value
285 %type <voidval> start_element
286 %type <voidval> left_element
287 %type <voidval> right_element
288 %type <voidval> slice_size
289 %type <voidval> expression_list
290 %type <voidval> lower_element
291 %type <voidval> upper_element
292 %type <voidval> first_element
293 %type <tval> mode_argument
294 %type <voidval> upper_lower_argument
295 %type <voidval> length_argument
296 %type <voidval> array_mode_name
297 %type <voidval> string_mode_name
298 %type <voidval> variant_structure_mode_name
299 %type <voidval> boolean_expression
300 %type <voidval> case_selector_list
301 %type <voidval> subexpression
302 %type <voidval> case_label_specification
303 %type <voidval> buffer_location
304 %type <voidval> single_assignment_action
305 %type <tsym> mode_name
313 { write_exp_elt_opcode(OP_TYPE);
314 write_exp_elt_type($1.type);
315 write_exp_elt_opcode(OP_TYPE);}
328 undefined_value : FIXME_01
336 access_name : LOCATION_NAME
338 write_exp_elt_opcode (OP_VAR_VALUE);
339 write_exp_elt_block (NULL);
340 write_exp_elt_sym ($1.sym);
341 write_exp_elt_opcode (OP_VAR_VALUE);
343 | GDB_LAST /* gdb specific */
345 write_exp_elt_opcode (OP_LAST);
346 write_exp_elt_longcst ($1);
347 write_exp_elt_opcode (OP_LAST);
349 | GDB_REGNAME /* gdb specific */
351 write_exp_elt_opcode (OP_REGISTER);
352 write_exp_elt_longcst ($1);
353 write_exp_elt_opcode (OP_REGISTER);
355 | GDB_VARIABLE /* gdb specific */
357 write_exp_elt_opcode (OP_INTERNALVAR);
358 write_exp_elt_intern ($1);
359 write_exp_elt_opcode (OP_INTERNALVAR);
369 expression_list : expression
373 | expression_list ',' expression
382 | primitive_value '('
383 /* This is to save the value of arglist_len
384 being accumulated for each dimension. */
385 { start_arglist (); }
388 write_exp_elt_opcode (MULTI_SUBSCRIPT);
389 write_exp_elt_longcst ((LONGEST) end_arglist ());
390 write_exp_elt_opcode (MULTI_SUBSCRIPT);
392 | primitive_value FIELD_NAME
393 { write_exp_elt_opcode (STRUCTOP_STRUCT);
394 write_exp_string ($2);
395 write_exp_elt_opcode (STRUCTOP_STRUCT);
397 | primitive_value POINTER
399 write_exp_elt_opcode (UNOP_IND);
413 | value_string_element
425 | expression_conversion
429 | value_procedure_call
433 | value_built_in_routine_call
445 | parenthesised_expression
453 value_name : synonym_name
457 | value_enumeration_name
469 | GENERAL_PROCEDURE_NAME
471 write_exp_elt_opcode (OP_VAR_VALUE);
472 write_exp_elt_block (NULL);
473 write_exp_elt_sym ($1.sym);
474 write_exp_elt_opcode (OP_VAR_VALUE);
480 literal : INTEGER_LITERAL
482 write_exp_elt_opcode (OP_LONG);
483 write_exp_elt_type ($1.type);
484 write_exp_elt_longcst ((LONGEST) ($1.val));
485 write_exp_elt_opcode (OP_LONG);
489 write_exp_elt_opcode (OP_BOOL);
490 write_exp_elt_longcst ((LONGEST) $1);
491 write_exp_elt_opcode (OP_BOOL);
495 write_exp_elt_opcode (OP_LONG);
496 write_exp_elt_type ($1.type);
497 write_exp_elt_longcst ((LONGEST) ($1.val));
498 write_exp_elt_opcode (OP_LONG);
502 write_exp_elt_opcode (OP_DOUBLE);
503 write_exp_elt_type (builtin_type_double);
504 write_exp_elt_dblcst ($1);
505 write_exp_elt_opcode (OP_DOUBLE);
515 | CHARACTER_STRING_LITERAL
517 write_exp_elt_opcode (OP_STRING);
518 write_exp_string ($1);
519 write_exp_elt_opcode (OP_STRING);
523 write_exp_elt_opcode (OP_BITSTRING);
524 write_exp_bitstring ($1);
525 write_exp_elt_opcode (OP_BITSTRING);
540 value_string_element: string_primitive_value '(' start_element ')'
548 value_string_slice: string_primitive_value '(' left_element ':' right_element ')'
552 | string_primitive_value '(' start_element UP slice_size ')'
560 value_array_slice: primitive_value '(' lower_element ':' upper_element ')'
564 | primitive_value '(' first_element UP slice_size ')'
572 expression_conversion: mode_name parenthesised_expression
574 write_exp_elt_opcode (UNOP_CAST);
575 write_exp_elt_type ($1.type);
576 write_exp_elt_opcode (UNOP_CAST);
582 value_procedure_call: FIXME_05
590 value_built_in_routine_call: chill_value_built_in_routine_call
598 start_expression: FIXME_06
601 } /* Not in GNU-Chill */
606 zero_adic_operator: FIXME_07
614 parenthesised_expression: '(' expression ')'
622 expression : operand_0
626 | single_assignment_action
630 | conditional_expression
636 conditional_expression : IF boolean_expression then_alternative else_alternative FI
640 | CASE case_selector_list OF value_case_alternative '[' ELSE sub_expression ']' ESAC
646 then_alternative: THEN subexpression
652 else_alternative: ELSE subexpression
656 | ELSIF boolean_expression then_alternative else_alternative
662 sub_expression : expression
668 value_case_alternative: case_label_specification ':' sub_expression ';'
676 operand_0 : operand_1
680 | operand_0 LOGIOR operand_1
682 write_exp_elt_opcode (BINOP_BITWISE_IOR);
684 | operand_0 ORIF operand_1
688 | operand_0 LOGXOR operand_1
690 write_exp_elt_opcode (BINOP_BITWISE_XOR);
696 operand_1 : operand_2
700 | operand_1 LOGAND operand_2
702 write_exp_elt_opcode (BINOP_BITWISE_AND);
704 | operand_1 ANDIF operand_2
712 operand_2 : operand_3
716 | operand_2 '=' operand_3
718 write_exp_elt_opcode (BINOP_EQUAL);
720 | operand_2 NOTEQUAL operand_3
722 write_exp_elt_opcode (BINOP_NOTEQUAL);
724 | operand_2 '>' operand_3
726 write_exp_elt_opcode (BINOP_GTR);
728 | operand_2 GTR operand_3
730 write_exp_elt_opcode (BINOP_GEQ);
732 | operand_2 '<' operand_3
734 write_exp_elt_opcode (BINOP_LESS);
736 | operand_2 LEQ operand_3
738 write_exp_elt_opcode (BINOP_LEQ);
740 | operand_2 IN operand_3
742 write_exp_elt_opcode (BINOP_IN);
749 operand_3 : operand_4
753 | operand_3 '+' operand_4
755 write_exp_elt_opcode (BINOP_ADD);
757 | operand_3 '-' operand_4
759 write_exp_elt_opcode (BINOP_SUB);
761 | operand_3 SLASH_SLASH operand_4
763 write_exp_elt_opcode (BINOP_CONCAT);
769 operand_4 : operand_5
773 | operand_4 '*' operand_5
775 write_exp_elt_opcode (BINOP_MUL);
777 | operand_4 '/' operand_5
779 write_exp_elt_opcode (BINOP_DIV);
781 | operand_4 MOD operand_5
783 write_exp_elt_opcode (BINOP_MOD);
785 | operand_4 REM operand_5
787 write_exp_elt_opcode (BINOP_REM);
793 operand_5 : operand_6
799 write_exp_elt_opcode (UNOP_NEG);
803 write_exp_elt_opcode (UNOP_LOGICAL_NOT);
805 | parenthesised_expression literal
806 /* We require the string operand to be a literal, to avoid some
807 nasty parsing ambiguities. */
809 write_exp_elt_opcode (BINOP_CONCAT);
815 operand_6 : POINTER primitive_value
817 write_exp_elt_opcode (UNOP_ADDR);
819 | RECEIVE buffer_location
832 single_assignment_action :
833 primitive_value GDB_ASSIGNMENT value
835 write_exp_elt_opcode (BINOP_ASSIGN);
841 chill_value_built_in_routine_call :
842 NUM '(' expression ')'
846 | PRED '(' expression ')'
850 | SUCC '(' expression ')'
854 | ABS '(' expression ')'
858 | CARD '(' expression ')'
862 | MAX_TOKEN '(' expression ')'
866 | MIN_TOKEN '(' expression ')'
870 | SIZE '(' expression ')'
871 { write_exp_elt_opcode (UNOP_SIZEOF); }
872 | SIZE '(' mode_argument ')'
873 { write_exp_elt_opcode (OP_LONG);
874 write_exp_elt_type (builtin_type_int);
875 write_exp_elt_longcst ((LONGEST) TYPE_LENGTH ($3));
876 write_exp_elt_opcode (OP_LONG); }
877 | UPPER '(' upper_lower_argument ')'
881 | LOWER '(' upper_lower_argument ')'
885 | LENGTH '(' length_argument ')'
891 mode_argument : mode_name
895 | array_mode_name '(' expression ')'
899 | string_mode_name '(' expression ')'
903 | variant_structure_mode_name '(' expression_list ')'
912 upper_lower_argument : expression
922 length_argument : expression
928 /* Things which still need productions... */
930 array_mode_name : FIXME_08 { $$ = 0; }
931 string_mode_name : FIXME_09 { $$ = 0; }
932 variant_structure_mode_name: FIXME_10 { $$ = 0; }
933 synonym_name : FIXME_11 { $$ = 0; }
934 value_enumeration_name : FIXME_12 { $$ = 0; }
935 value_do_with_name : FIXME_13 { $$ = 0; }
936 value_receive_name : FIXME_14 { $$ = 0; }
937 string_primitive_value : FIXME_15 { $$ = 0; }
938 start_element : FIXME_16 { $$ = 0; }
939 left_element : FIXME_17 { $$ = 0; }
940 right_element : FIXME_18 { $$ = 0; }
941 slice_size : FIXME_19 { $$ = 0; }
942 lower_element : FIXME_20 { $$ = 0; }
943 upper_element : FIXME_21 { $$ = 0; }
944 first_element : FIXME_22 { $$ = 0; }
945 boolean_expression : FIXME_26 { $$ = 0; }
946 case_selector_list : FIXME_27 { $$ = 0; }
947 subexpression : FIXME_28 { $$ = 0; }
948 case_label_specification: FIXME_29 { $$ = 0; }
949 buffer_location : FIXME_30 { $$ = 0; }
953 /* Implementation of a dynamically expandable buffer for processing input
954 characters acquired through lexptr and building a value to return in
957 static char *tempbuf; /* Current buffer contents */
958 static int tempbufsize; /* Size of allocated buffer */
959 static int tempbufindex; /* Current index into buffer */
961 #define GROWBY_MIN_SIZE 64 /* Minimum amount to grow buffer by */
963 #define CHECKBUF(size) \
965 if (tempbufindex + (size) >= tempbufsize) \
967 growbuf_by_size (size); \
971 /* Grow the static temp buffer if necessary, including allocating the first one
975 growbuf_by_size (count)
980 growby = max (count, GROWBY_MIN_SIZE);
981 tempbufsize += growby;
984 tempbuf = (char *) malloc (tempbufsize);
988 tempbuf = (char *) realloc (tempbuf, tempbufsize);
992 /* Try to consume a simple name string token. If successful, returns
993 a pointer to a nullbyte terminated copy of the name that can be used
994 in symbol table lookups. If not successful, returns NULL. */
997 match_simple_name_string ()
999 char *tokptr = lexptr;
1001 if (isalpha (*tokptr) || *tokptr == '_')
1006 } while (isalnum (*tokptr) || (*tokptr == '_'));
1007 yylval.sval.ptr = lexptr;
1008 yylval.sval.length = tokptr - lexptr;
1010 result = copy_name (yylval.sval);
1016 /* Start looking for a value composed of valid digits as set by the base
1017 in use. Note that '_' characters are valid anywhere, in any quantity,
1018 and are simply ignored. Since we must find at least one valid digit,
1019 or reject this token as an integer literal, we keep track of how many
1020 digits we have encountered. */
1023 decode_integer_value (base, tokptrptr, ivalptr)
1028 char *tokptr = *tokptrptr;
1032 while (*tokptr != '\0')
1036 temp = tolower (temp);
1042 case '0': case '1': case '2': case '3': case '4':
1043 case '5': case '6': case '7': case '8': case '9':
1046 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
1062 /* Found something not in domain for current base. */
1063 tokptr--; /* Unconsume what gave us indigestion. */
1068 /* If we didn't find any digits, then we don't have a valid integer
1069 value, so reject the entire token. Otherwise, update the lexical
1070 scan pointer, and return non-zero for success. */
1078 *tokptrptr = tokptr;
1084 decode_integer_literal (valptr, tokptrptr)
1088 char *tokptr = *tokptrptr;
1091 int explicit_base = 0;
1093 /* Look for an explicit base specifier, which is optional. */
1126 /* If we found an explicit base ensure that the character after the
1127 explicit base is a single quote. */
1129 if (explicit_base && (*tokptr++ != '\''))
1134 /* Attempt to decode whatever follows as an integer value in the
1135 indicated base, updating the token pointer in the process and
1136 computing the value into ival. Also, if we have an explicit
1137 base, then the next character must not be a single quote, or we
1138 have a bitstring literal, so reject the entire token in this case.
1139 Otherwise, update the lexical scan pointer, and return non-zero
1142 if (!decode_integer_value (base, &tokptr, &ival))
1146 else if (explicit_base && (*tokptr == '\''))
1153 *tokptrptr = tokptr;
1158 /* If it wasn't for the fact that floating point values can contain '_'
1159 characters, we could just let strtod do all the hard work by letting it
1160 try to consume as much of the current token buffer as possible and
1161 find a legal conversion. Unfortunately we need to filter out the '_'
1162 characters before calling strtod, which we do by copying the other
1163 legal chars to a local buffer to be converted. However since we also
1164 need to keep track of where the last unconsumed character in the input
1165 buffer is, we have transfer only as many characters as may compose a
1166 legal floating point value. */
1169 match_float_literal ()
1171 char *tokptr = lexptr;
1175 extern double strtod ();
1177 /* Make local buffer in which to build the string to convert. This is
1178 required because underscores are valid in chill floating point numbers
1179 but not in the string passed to strtod to convert. The string will be
1180 no longer than our input string. */
1182 copy = buf = (char *) alloca (strlen (tokptr) + 1);
1184 /* Transfer all leading digits to the conversion buffer, discarding any
1187 while (isdigit (*tokptr) || *tokptr == '_')
1196 /* Now accept either a '.', or one of [eEdD]. Dot is legal regardless
1197 of whether we found any leading digits, and we simply accept it and
1198 continue on to look for the fractional part and/or exponent. One of
1199 [eEdD] is legal only if we have seen digits, and means that there
1200 is no fractional part. If we find neither of these, then this is
1201 not a floating point number, so return failure. */
1206 /* Accept and then look for fractional part and/or exponent. */
1219 goto collect_exponent;
1227 /* We found a '.', copy any fractional digits to the conversion buffer, up
1228 to the first nondigit, non-underscore character. */
1230 while (isdigit (*tokptr) || *tokptr == '_')
1239 /* Look for an exponent, which must start with one of [eEdD]. If none
1240 is found, jump directly to trying to convert what we have collected
1257 /* Accept an optional '-' or '+' following one of [eEdD]. */
1260 if (*tokptr == '+' || *tokptr == '-')
1262 *copy++ = *tokptr++;
1265 /* Now copy an exponent into the conversion buffer. Note that at the
1266 moment underscores are *not* allowed in exponents. */
1268 while (isdigit (*tokptr))
1270 *copy++ = *tokptr++;
1273 /* If we transfered any chars to the conversion buffer, try to interpret its
1274 contents as a floating point value. If any characters remain, then we
1275 must not have a valid floating point string. */
1281 dval = strtod (buf, ©);
1286 return (FLOAT_LITERAL);
1292 /* Recognize a string literal. A string literal is a sequence
1293 of characters enclosed in matching single or double quotes, except that
1294 a single character inside single quotes is a character literal, which
1295 we reject as a string literal. To embed the terminator character inside
1296 a string, it is simply doubled (I.E. "this""is""one""string") */
1299 match_string_literal ()
1301 char *tokptr = lexptr;
1303 for (tempbufindex = 0, tokptr++; *tokptr != '\0'; tokptr++)
1306 if (*tokptr == *lexptr)
1308 if (*(tokptr + 1) == *lexptr)
1317 tempbuf[tempbufindex++] = *tokptr;
1319 if (*tokptr == '\0' /* no terminator */
1320 || (tempbufindex == 1 && *tokptr == '\'')) /* char literal */
1326 tempbuf[tempbufindex] = '\0';
1327 yylval.sval.ptr = tempbuf;
1328 yylval.sval.length = tempbufindex;
1330 return (CHARACTER_STRING_LITERAL);
1334 /* Recognize a character literal. A character literal is single character
1335 or a control sequence, enclosed in single quotes. A control sequence
1336 is a comma separated list of one or more integer literals, enclosed
1337 in parenthesis and introduced with a circumflex character.
1339 EX: 'a' '^(7)' '^(7,8)'
1341 As a GNU chill extension, the syntax C'xx' is also recognized as a
1342 character literal, where xx is a hex value for the character.
1344 Note that more than a single character, enclosed in single quotes, is
1347 Also note that the control sequence form is not in GNU Chill since it
1348 is ambiguous with the string literal form using single quotes. I.E.
1349 is '^(7)' a character literal or a string literal. In theory it it
1350 possible to tell by context, but GNU Chill doesn't accept the control
1351 sequence form, so neither do we (for now the code is disabled).
1353 Returns CHARACTER_LITERAL if a match is found.
1357 match_character_literal ()
1359 char *tokptr = lexptr;
1362 if ((*tokptr == 'c' || *tokptr == 'C') && (*(tokptr + 1) == '\''))
1364 /* We have a GNU chill extension form, so skip the leading "C'",
1365 decode the hex value, and then ensure that we have a trailing
1366 single quote character. */
1368 if (!decode_integer_value (16, &tokptr, &ival) || (*tokptr != '\''))
1374 else if (*tokptr == '\'')
1378 /* Determine which form we have, either a control sequence or the
1379 single character form. */
1381 if ((*tokptr == '^') && (*(tokptr + 1) == '('))
1383 #if 0 /* Disable, see note above. -fnf */
1384 /* Match and decode a control sequence. Return zero if we don't
1385 find a valid integer literal, or if the next unconsumed character
1386 after the integer literal is not the trailing ')'.
1387 FIXME: We currently don't handle the multiple integer literal
1390 if (!decode_integer_literal (&ival, &tokptr) || (*tokptr++ != ')'))
1403 /* The trailing quote has not yet been consumed. If we don't find
1404 it, then we have no match. */
1406 if (*tokptr++ != '\'')
1413 /* Not a character literal. */
1416 yylval.typed_val.val = ival;
1417 yylval.typed_val.type = builtin_type_chill_char;
1419 return (CHARACTER_LITERAL);
1422 /* Recognize an integer literal, as specified in Z.200 sec 5.2.4.2.
1423 Note that according to 5.2.4.2, a single "_" is also a valid integer
1424 literal, however GNU-chill requires there to be at least one "digit"
1425 in any integer literal. */
1428 match_integer_literal ()
1430 char *tokptr = lexptr;
1433 if (!decode_integer_literal (&ival, &tokptr))
1439 yylval.typed_val.val = ival;
1440 yylval.typed_val.type = builtin_type_int;
1442 return (INTEGER_LITERAL);
1446 /* Recognize a bit-string literal, as specified in Z.200 sec 5.2.4.8
1447 Note that according to 5.2.4.8, a single "_" is also a valid bit-string
1448 literal, however GNU-chill requires there to be at least one "digit"
1449 in any bit-string literal. */
1452 match_bitstring_literal ()
1454 char *tokptr = lexptr;
1463 /* Look for the required explicit base specifier. */
1484 /* Ensure that the character after the explicit base is a single quote. */
1486 if (*tokptr++ != '\'')
1491 while (*tokptr != '\0' && *tokptr != '\'')
1494 if (isupper (digit))
1495 digit = tolower (digit);
1501 case '0': case '1': case '2': case '3': case '4':
1502 case '5': case '6': case '7': case '8': case '9':
1505 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
1515 /* Found something not in domain for current base. */
1520 /* Extract bits from digit, starting with the msbit appropriate for
1521 the current base, and packing them into the bitstring byte,
1522 starting at the lsbit. */
1523 for (mask = (base >> 1); mask > 0; mask >>= 1)
1529 tempbuf[tempbufindex] |= (1 << bitoffset);
1532 if (bitoffset == HOST_CHAR_BIT)
1541 /* Verify that we consumed everything up to the trailing single quote,
1542 and that we found some bits (IE not just underbars). */
1544 if (*tokptr++ != '\'')
1550 yylval.sval.ptr = tempbuf;
1551 yylval.sval.length = bitcount;
1553 return (BIT_STRING_LITERAL);
1557 /* Recognize tokens that start with '$'. These include:
1559 $regname A native register name or a "standard
1561 Return token GDB_REGNAME.
1563 $variable A convenience variable with a name chosen
1565 Return token GDB_VARIABLE.
1567 $digits Value history with index <digits>, starting
1568 from the first value which has index 1.
1571 $$digits Value history with index <digits> relative
1572 to the last value. I.E. $$0 is the last
1573 value, $$1 is the one previous to that, $$2
1574 is the one previous to $$1, etc.
1575 Return token GDB_LAST.
1577 $ | $0 | $$0 The last value in the value history.
1578 Return token GDB_LAST.
1580 $$ An abbreviation for the second to the last
1581 value in the value history, I.E. $$1
1582 Return token GDB_LAST.
1584 Note that we currently assume that register names and convenience
1585 variables follow the convention of starting with a letter or '_'.
1590 match_dollar_tokens ()
1598 /* We will always have a successful match, even if it is just for
1599 a single '$', the abbreviation for $$0. So advance lexptr. */
1603 if (*tokptr == '_' || isalpha (*tokptr))
1605 /* Look for a match with a native register name, usually something
1606 like "r0" for example. */
1608 for (regno = 0; regno < NUM_REGS; regno++)
1610 namelength = strlen (reg_names[regno]);
1611 if (STREQN (tokptr, reg_names[regno], namelength)
1612 && !isalnum (tokptr[namelength]))
1614 yylval.lval = regno;
1615 lexptr += namelength;
1616 return (GDB_REGNAME);
1620 /* Look for a match with a standard register name, usually something
1621 like "pc", which gdb always recognizes as the program counter
1622 regardless of what the native register name is. */
1624 for (regno = 0; regno < num_std_regs; regno++)
1626 namelength = strlen (std_regs[regno].name);
1627 if (STREQN (tokptr, std_regs[regno].name, namelength)
1628 && !isalnum (tokptr[namelength]))
1630 yylval.lval = std_regs[regno].regnum;
1631 lexptr += namelength;
1632 return (GDB_REGNAME);
1636 /* Attempt to match against a convenience variable. Note that
1637 this will always succeed, because if no variable of that name
1638 already exists, the lookup_internalvar will create one for us.
1639 Also note that both lexptr and tokptr currently point to the
1640 start of the input string we are trying to match, and that we
1641 have already tested the first character for non-numeric, so we
1642 don't have to treat it specially. */
1644 while (*tokptr == '_' || isalnum (*tokptr))
1648 yylval.sval.ptr = lexptr;
1649 yylval.sval.length = tokptr - lexptr;
1650 yylval.ivar = lookup_internalvar (copy_name (yylval.sval));
1652 return (GDB_VARIABLE);
1655 /* Since we didn't match against a register name or convenience
1656 variable, our only choice left is a history value. */
1670 /* Attempt to decode more characters as an integer value giving
1671 the index in the history list. If successful, the value will
1672 overwrite ival (currently 0 or 1), and if not, ival will be
1673 left alone, which is good since it is currently correct for
1674 the '$' or '$$' case. */
1676 decode_integer_literal (&ival, &tokptr);
1677 yylval.lval = negate ? -ival : ival;
1688 static const struct token idtokentab[] =
1690 { "length", LENGTH },
1701 { "max", MAX_TOKEN },
1702 { "min", MIN_TOKEN },
1712 static const struct token tokentab2[] =
1714 { ":=", GDB_ASSIGNMENT },
1715 { "//", SLASH_SLASH },
1722 /* Read one token, getting characters through lexptr. */
1723 /* This is where we will check to make sure that the language and the
1724 operators used are compatible. */
1734 /* Skip over any leading whitespace. */
1735 while (isspace (*lexptr))
1739 /* Look for special single character cases which can't be the first
1740 character of some other multicharacter token. */
1757 /* Look for characters which start a particular kind of multicharacter
1758 token, such as a character literal, register name, convenience
1759 variable name, string literal, etc. */
1764 /* First try to match a string literal, which is any
1765 sequence of characters enclosed in matching single or double
1766 quotes, except that a single character inside single quotes
1767 is a character literal, so we have to catch that case also. */
1768 token = match_string_literal ();
1773 if (*lexptr == '\'')
1775 token = match_character_literal ();
1784 token = match_character_literal ();
1791 token = match_dollar_tokens ();
1798 /* See if it is a special token of length 2. */
1799 for (i = 0; i < sizeof (tokentab2) / sizeof (tokentab2[0]); i++)
1801 if (STREQN (lexptr, tokentab2[i].operator, 2))
1804 return (tokentab2[i].token);
1807 /* Look for single character cases which which could be the first
1808 character of some other multicharacter token, but aren't, or we
1809 would already have found it. */
1819 /* Look for a float literal before looking for an integer literal, so
1820 we match as much of the input stream as possible. */
1821 token = match_float_literal ();
1826 token = match_bitstring_literal ();
1831 token = match_integer_literal ();
1837 /* Try to match a simple name string, and if a match is found, then
1838 further classify what sort of name it is and return an appropriate
1839 token. Note that attempting to match a simple name string consumes
1840 the token from lexptr, so we can't back out if we later find that
1841 we can't classify what sort of name it is. */
1843 inputname = match_simple_name_string ();
1845 if (inputname != NULL)
1847 char *simplename = (char*) alloca (strlen (inputname));
1849 char *dptr = simplename, *sptr = inputname;
1850 for (; *sptr; sptr++)
1851 *dptr++ = isupper (*sptr) ? tolower(*sptr) : *sptr;
1853 /* See if it is a reserved identifier. */
1854 for (i = 0; i < sizeof (idtokentab) / sizeof (idtokentab[0]); i++)
1856 if (STREQ (simplename, idtokentab[i].operator))
1858 return (idtokentab[i].token);
1862 /* Look for other special tokens. */
1863 if (STREQ (simplename, "true"))
1866 return (BOOLEAN_LITERAL);
1868 if (STREQ (simplename, "false"))
1871 return (BOOLEAN_LITERAL);
1874 sym = lookup_symbol (inputname, expression_context_block,
1875 VAR_NAMESPACE, (int *) NULL,
1876 (struct symtab **) NULL);
1877 if (sym == NULL && strcmp (inputname, simplename) != 0)
1879 sym = lookup_symbol (simplename, expression_context_block,
1880 VAR_NAMESPACE, (int *) NULL,
1881 (struct symtab **) NULL);
1885 yylval.ssym.stoken.ptr = NULL;
1886 yylval.ssym.stoken.length = 0;
1887 yylval.ssym.sym = sym;
1888 yylval.ssym.is_a_field_of_this = 0; /* FIXME, C++'ism */
1889 switch (SYMBOL_CLASS (sym))
1892 /* Found a procedure name. */
1893 return (GENERAL_PROCEDURE_NAME);
1895 /* Found a global or local static variable. */
1896 return (LOCATION_NAME);
1901 case LOC_REGPARM_ADDR:
1905 case LOC_BASEREG_ARG:
1906 if (innermost_block == NULL
1907 || contained_in (block_found, innermost_block))
1909 innermost_block = block_found;
1911 return (LOCATION_NAME);
1915 return (LOCATION_NAME);
1918 yylval.tsym.type = SYMBOL_TYPE (sym);
1921 case LOC_CONST_BYTES:
1922 case LOC_OPTIMIZED_OUT:
1923 error ("Symbol \"%s\" names no location.", inputname);
1927 else if (!have_full_symbols () && !have_partial_symbols ())
1929 error ("No symbol table is loaded. Use the \"file\" command.");
1933 error ("No symbol \"%s\" in current context.", inputname);
1937 /* Catch single character tokens which are not part of some
1942 case '.': /* Not float for example. */
1944 while (isspace (*lexptr)) lexptr++;
1945 inputname = match_simple_name_string ();
1951 return (ILLEGAL_TOKEN);
1958 error ("A %s in expression, near `%s'.", (msg ? msg : "error"), lexptr);