1 /* YACC grammar for Chill expressions, for GDB.
2 Copyright 1992, 1993 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.
58 #include "expression.h"
61 #include "parser-defs.h"
64 /* Remap normal yacc parser interface names (yyparse, yylex, yyerror, etc),
65 as well as gratuitiously global symbol names, so we can have multiple
66 yacc generated parsers in gdb. Note that these are only the variables
67 produced by yacc. If other parser generators (bison, byacc, etc) produce
68 additional global names that conflict at link time, then those parser
69 generators need to be fixed instead of adding those names to this list. */
71 #define yymaxdepth chill_maxdepth
72 #define yyparse chill_parse
73 #define yylex chill_lex
74 #define yyerror chill_error
75 #define yylval chill_lval
76 #define yychar chill_char
77 #define yycheck chill_yycheck
78 #define yydebug chill_debug
79 #define yydefred chill_yydefred
80 #define yydgoto chill_yydgoto
81 #define yygindex chill_yygindex
82 #define yylen chill_yylen
83 #define yylhs chill_yylhs
84 #define yyrindex chill_yyrindex
85 #define yysccsid chill_yysccsid
86 #define yysindex chill_yysindex
87 #define yypact chill_pact
90 #define yydef chill_def
91 #define yychk chill_chk
92 #define yypgo chill_pgo
93 #define yyact chill_act
94 #define yyexca chill_exca
95 #define yyerrflag chill_errflag
96 #define yynerrs chill_nerrs
100 #define yyss chill_yyss
101 #define yyssp chill_yyssp
102 #define yy_yys chill_yys
103 #define yystate chill_state
104 #define yytable chill_yytable
105 #define yytmp chill_tmp
107 #define yyvs chill_vs
108 #define yyvss chill_vss
109 #define yy_yyv chill_yyv
110 #define yyval chill_val
111 #define yylloc chill_lloc
112 #define yyreds chill_reds /* With YYDEBUG defined */
113 #define yytoks chill_toks /* With YYDEBUG defined */
116 #define YYDEBUG 0 /* Default to no yydebug support */
120 yyparse PARAMS ((void));
123 yylex PARAMS ((void));
126 yyerror PARAMS ((char *));
130 /* Although the yacc "value" of an expression is not used,
131 since the result is stored in the structure being created,
132 other node types do have values. */
137 unsigned LONGEST ulval;
147 struct symtoken ssym;
150 enum exp_opcode opcode;
151 struct internalvar *ivar;
157 %token <voidval> FIXME_01
158 %token <voidval> FIXME_02
159 %token <voidval> FIXME_03
160 %token <voidval> FIXME_04
161 %token <voidval> FIXME_05
162 %token <voidval> FIXME_06
163 %token <voidval> FIXME_07
164 %token <voidval> FIXME_08
165 %token <voidval> FIXME_09
166 %token <voidval> FIXME_10
167 %token <voidval> FIXME_11
168 %token <voidval> FIXME_12
169 %token <voidval> FIXME_13
170 %token <voidval> FIXME_14
171 %token <voidval> FIXME_15
172 %token <voidval> FIXME_16
173 %token <voidval> FIXME_17
174 %token <voidval> FIXME_18
175 %token <voidval> FIXME_19
176 %token <voidval> FIXME_20
177 %token <voidval> FIXME_21
178 %token <voidval> FIXME_22
179 %token <voidval> FIXME_24
180 %token <voidval> FIXME_25
181 %token <voidval> FIXME_26
182 %token <voidval> FIXME_27
183 %token <voidval> FIXME_28
184 %token <voidval> FIXME_29
185 %token <voidval> FIXME_30
187 %token <typed_val> INTEGER_LITERAL
188 %token <ulval> BOOLEAN_LITERAL
189 %token <typed_val> CHARACTER_LITERAL
190 %token <dval> FLOAT_LITERAL
191 %token <ssym> GENERAL_PROCEDURE_NAME
192 %token <ssym> LOCATION_NAME
193 %token <voidval> SET_LITERAL
194 %token <voidval> EMPTINESS_LITERAL
195 %token <sval> CHARACTER_STRING_LITERAL
196 %token <sval> BIT_STRING_LITERAL
197 %token <tsym> TYPENAME
198 %token <sval> FIELD_NAME
203 %token <voidval> CASE
205 %token <voidval> ESAC
206 %token <voidval> LOGIOR
207 %token <voidval> ORIF
208 %token <voidval> LOGXOR
209 %token <voidval> LOGAND
210 %token <voidval> ANDIF
212 %token <voidval> NOTEQUAL
222 %token <voidval> SLASH_SLASH
226 %token <voidval> POINTER
227 %token <voidval> RECEIVE
234 %token <voidval> THEN
235 %token <voidval> ELSE
237 %token <voidval> ELSIF
238 %token <voidval> ILLEGAL_TOKEN
240 %token <voidval> PRED
241 %token <voidval> SUCC
243 %token <voidval> CARD
244 %token <voidval> MAX_TOKEN
245 %token <voidval> MIN_TOKEN
246 %token <voidval> SIZE
247 %token <voidval> UPPER
248 %token <voidval> LOWER
249 %token <voidval> LENGTH
251 /* Tokens which are not Chill tokens used in expressions, but rather GDB
252 specific things that we recognize in the same context as Chill tokens
253 (register names for example). */
255 %token <lval> GDB_REGNAME /* Machine register name */
256 %token <lval> GDB_LAST /* Value history */
257 %token <ivar> GDB_VARIABLE /* Convenience variable */
258 %token <voidval> GDB_ASSIGNMENT /* Assign value to somewhere */
260 %type <voidval> location
261 %type <voidval> access_name
262 %type <voidval> primitive_value
263 %type <voidval> location_contents
264 %type <voidval> value_name
265 %type <voidval> literal
266 %type <voidval> tuple
267 %type <voidval> value_string_element
268 %type <voidval> value_string_slice
269 %type <voidval> value_array_element
270 %type <voidval> value_array_slice
271 %type <voidval> value_structure_field
272 %type <voidval> expression_conversion
273 %type <voidval> value_procedure_call
274 %type <voidval> value_built_in_routine_call
275 %type <voidval> chill_value_built_in_routine_call
276 %type <voidval> start_expression
277 %type <voidval> zero_adic_operator
278 %type <voidval> parenthesised_expression
279 %type <voidval> value
280 %type <voidval> undefined_value
281 %type <voidval> expression
282 %type <voidval> conditional_expression
283 %type <voidval> then_alternative
284 %type <voidval> else_alternative
285 %type <voidval> sub_expression
286 %type <voidval> value_case_alternative
287 %type <voidval> operand_0
288 %type <voidval> operand_1
289 %type <voidval> operand_2
290 %type <voidval> operand_3
291 %type <voidval> operand_4
292 %type <voidval> operand_5
293 %type <voidval> operand_6
294 %type <voidval> synonym_name
295 %type <voidval> value_enumeration_name
296 %type <voidval> value_do_with_name
297 %type <voidval> value_receive_name
298 %type <voidval> string_primitive_value
299 %type <voidval> start_element
300 %type <voidval> left_element
301 %type <voidval> right_element
302 %type <voidval> slice_size
303 %type <voidval> array_primitive_value
304 %type <voidval> expression_list
305 %type <voidval> lower_element
306 %type <voidval> upper_element
307 %type <voidval> first_element
308 %type <voidval> mode_argument
309 %type <voidval> upper_lower_argument
310 %type <voidval> length_argument
311 %type <voidval> array_mode_name
312 %type <voidval> string_mode_name
313 %type <voidval> variant_structure_mode_name
314 %type <voidval> boolean_expression
315 %type <voidval> case_selector_list
316 %type <voidval> subexpression
317 %type <voidval> case_label_specification
318 %type <voidval> buffer_location
319 %type <voidval> single_assignment_action
320 %type <tsym> mode_name
328 { write_exp_elt_opcode(OP_TYPE);
329 write_exp_elt_type($1.type);
330 write_exp_elt_opcode(OP_TYPE);}
343 undefined_value : FIXME_01
351 location : access_name
352 | primitive_value POINTER
354 write_exp_elt_opcode (UNOP_IND);
360 access_name : LOCATION_NAME
362 write_exp_elt_opcode (OP_VAR_VALUE);
363 write_exp_elt_sym ($1.sym);
364 write_exp_elt_opcode (OP_VAR_VALUE);
366 | GDB_LAST /* gdb specific */
368 write_exp_elt_opcode (OP_LAST);
369 write_exp_elt_longcst ($1);
370 write_exp_elt_opcode (OP_LAST);
372 | GDB_REGNAME /* gdb specific */
374 write_exp_elt_opcode (OP_REGISTER);
375 write_exp_elt_longcst ($1);
376 write_exp_elt_opcode (OP_REGISTER);
378 | GDB_VARIABLE /* gdb specific */
380 write_exp_elt_opcode (OP_INTERNALVAR);
381 write_exp_elt_intern ($1);
382 write_exp_elt_opcode (OP_INTERNALVAR);
392 expression_list : expression
396 | expression_list ',' expression
403 primitive_value : location_contents
419 | value_string_element
427 | value_array_element
435 | value_structure_field
439 | expression_conversion
443 | value_procedure_call
447 | value_built_in_routine_call
459 | parenthesised_expression
467 location_contents: location
475 value_name : synonym_name
479 | value_enumeration_name
491 | GENERAL_PROCEDURE_NAME
493 write_exp_elt_opcode (OP_VAR_VALUE);
494 write_exp_elt_sym ($1.sym);
495 write_exp_elt_opcode (OP_VAR_VALUE);
501 literal : INTEGER_LITERAL
503 write_exp_elt_opcode (OP_LONG);
504 write_exp_elt_type ($1.type);
505 write_exp_elt_longcst ((LONGEST) ($1.val));
506 write_exp_elt_opcode (OP_LONG);
510 write_exp_elt_opcode (OP_BOOL);
511 write_exp_elt_longcst ((LONGEST) $1);
512 write_exp_elt_opcode (OP_BOOL);
516 write_exp_elt_opcode (OP_LONG);
517 write_exp_elt_type ($1.type);
518 write_exp_elt_longcst ((LONGEST) ($1.val));
519 write_exp_elt_opcode (OP_LONG);
523 write_exp_elt_opcode (OP_DOUBLE);
524 write_exp_elt_type (builtin_type_double);
525 write_exp_elt_dblcst ($1);
526 write_exp_elt_opcode (OP_DOUBLE);
536 | CHARACTER_STRING_LITERAL
538 write_exp_elt_opcode (OP_STRING);
539 write_exp_string ($1);
540 write_exp_elt_opcode (OP_STRING);
544 write_exp_elt_opcode (OP_BITSTRING);
545 write_exp_bitstring ($1);
546 write_exp_elt_opcode (OP_BITSTRING);
561 value_string_element: string_primitive_value '(' start_element ')'
569 value_string_slice: string_primitive_value '(' left_element ':' right_element ')'
573 | string_primitive_value '(' start_element UP slice_size ')'
581 value_array_element: array_primitive_value '('
582 /* This is to save the value of arglist_len
583 being accumulated for each dimension. */
584 { start_arglist (); }
587 write_exp_elt_opcode (MULTI_SUBSCRIPT);
588 write_exp_elt_longcst ((LONGEST) end_arglist ());
589 write_exp_elt_opcode (MULTI_SUBSCRIPT);
595 value_array_slice: array_primitive_value '(' lower_element ':' upper_element ')'
599 | array_primitive_value '(' first_element UP slice_size ')'
607 value_structure_field: primitive_value FIELD_NAME
608 { write_exp_elt_opcode (STRUCTOP_STRUCT);
609 write_exp_string ($2);
610 write_exp_elt_opcode (STRUCTOP_STRUCT);
616 expression_conversion: mode_name parenthesised_expression
618 write_exp_elt_opcode (UNOP_CAST);
619 write_exp_elt_type ($1.type);
620 write_exp_elt_opcode (UNOP_CAST);
626 value_procedure_call: FIXME_05
634 value_built_in_routine_call: chill_value_built_in_routine_call
642 start_expression: FIXME_06
645 } /* Not in GNU-Chill */
650 zero_adic_operator: FIXME_07
658 parenthesised_expression: '(' expression ')'
666 expression : operand_0
670 | single_assignment_action
674 | conditional_expression
680 conditional_expression : IF boolean_expression then_alternative else_alternative FI
684 | CASE case_selector_list OF value_case_alternative '[' ELSE sub_expression ']' ESAC
690 then_alternative: THEN subexpression
696 else_alternative: ELSE subexpression
700 | ELSIF boolean_expression then_alternative else_alternative
706 sub_expression : expression
712 value_case_alternative: case_label_specification ':' sub_expression ';'
720 operand_0 : operand_1
724 | operand_0 LOGIOR operand_1
726 write_exp_elt_opcode (BINOP_BITWISE_IOR);
728 | operand_0 ORIF operand_1
732 | operand_0 LOGXOR operand_1
734 write_exp_elt_opcode (BINOP_BITWISE_XOR);
740 operand_1 : operand_2
744 | operand_1 LOGAND operand_2
746 write_exp_elt_opcode (BINOP_BITWISE_AND);
748 | operand_1 ANDIF operand_2
756 operand_2 : operand_3
760 | operand_2 '=' operand_3
762 write_exp_elt_opcode (BINOP_EQUAL);
764 | operand_2 NOTEQUAL operand_3
766 write_exp_elt_opcode (BINOP_NOTEQUAL);
768 | operand_2 '>' operand_3
770 write_exp_elt_opcode (BINOP_GTR);
772 | operand_2 GTR operand_3
774 write_exp_elt_opcode (BINOP_GEQ);
776 | operand_2 '<' operand_3
778 write_exp_elt_opcode (BINOP_LESS);
780 | operand_2 LEQ operand_3
782 write_exp_elt_opcode (BINOP_LEQ);
784 | operand_2 IN operand_3
793 operand_3 : operand_4
797 | operand_3 '+' operand_4
799 write_exp_elt_opcode (BINOP_ADD);
801 | operand_3 '-' operand_4
803 write_exp_elt_opcode (BINOP_SUB);
805 | operand_3 SLASH_SLASH operand_4
807 write_exp_elt_opcode (BINOP_CONCAT);
813 operand_4 : operand_5
817 | operand_4 '*' operand_5
819 write_exp_elt_opcode (BINOP_MUL);
821 | operand_4 '/' operand_5
823 write_exp_elt_opcode (BINOP_DIV);
825 | operand_4 MOD operand_5
827 write_exp_elt_opcode (BINOP_MOD);
829 | operand_4 REM operand_5
831 write_exp_elt_opcode (BINOP_REM);
837 operand_5 : operand_6
843 write_exp_elt_opcode (UNOP_NEG);
847 write_exp_elt_opcode (UNOP_LOGICAL_NOT);
849 | parenthesised_expression literal
850 /* We require the string operand to be a literal, to avoid some
851 nasty parsing ambiguities. */
853 write_exp_elt_opcode (BINOP_CONCAT);
859 operand_6 : POINTER location
861 write_exp_elt_opcode (UNOP_ADDR);
863 | RECEIVE buffer_location
876 single_assignment_action :
877 location GDB_ASSIGNMENT value
879 write_exp_elt_opcode (BINOP_ASSIGN);
885 chill_value_built_in_routine_call :
886 NUM '(' expression ')'
890 | PRED '(' expression ')'
894 | SUCC '(' expression ')'
898 | ABS '(' expression ')'
902 | CARD '(' expression ')'
906 | MAX_TOKEN '(' expression ')'
910 | MIN_TOKEN '(' expression ')'
914 | SIZE '(' location ')'
918 | SIZE '(' mode_argument ')'
922 | UPPER '(' upper_lower_argument ')'
926 | LOWER '(' upper_lower_argument ')'
930 | LENGTH '(' length_argument ')'
936 mode_argument : mode_name
940 | array_mode_name '(' expression ')'
944 | string_mode_name '(' expression ')'
948 | variant_structure_mode_name '(' expression_list ')'
957 upper_lower_argument : expression
967 length_argument : expression
975 array_primitive_value : primitive_value
982 /* Things which still need productions... */
984 array_mode_name : FIXME_08 { $$ = 0; }
985 string_mode_name : FIXME_09 { $$ = 0; }
986 variant_structure_mode_name: FIXME_10 { $$ = 0; }
987 synonym_name : FIXME_11 { $$ = 0; }
988 value_enumeration_name : FIXME_12 { $$ = 0; }
989 value_do_with_name : FIXME_13 { $$ = 0; }
990 value_receive_name : FIXME_14 { $$ = 0; }
991 string_primitive_value : FIXME_15 { $$ = 0; }
992 start_element : FIXME_16 { $$ = 0; }
993 left_element : FIXME_17 { $$ = 0; }
994 right_element : FIXME_18 { $$ = 0; }
995 slice_size : FIXME_19 { $$ = 0; }
996 lower_element : FIXME_20 { $$ = 0; }
997 upper_element : FIXME_21 { $$ = 0; }
998 first_element : FIXME_22 { $$ = 0; }
999 boolean_expression : FIXME_26 { $$ = 0; }
1000 case_selector_list : FIXME_27 { $$ = 0; }
1001 subexpression : FIXME_28 { $$ = 0; }
1002 case_label_specification: FIXME_29 { $$ = 0; }
1003 buffer_location : FIXME_30 { $$ = 0; }
1007 /* Implementation of a dynamically expandable buffer for processing input
1008 characters acquired through lexptr and building a value to return in
1011 static char *tempbuf; /* Current buffer contents */
1012 static int tempbufsize; /* Size of allocated buffer */
1013 static int tempbufindex; /* Current index into buffer */
1015 #define GROWBY_MIN_SIZE 64 /* Minimum amount to grow buffer by */
1017 #define CHECKBUF(size) \
1019 if (tempbufindex + (size) >= tempbufsize) \
1021 growbuf_by_size (size); \
1025 /* Grow the static temp buffer if necessary, including allocating the first one
1029 growbuf_by_size (count)
1034 growby = max (count, GROWBY_MIN_SIZE);
1035 tempbufsize += growby;
1036 if (tempbuf == NULL)
1038 tempbuf = (char *) malloc (tempbufsize);
1042 tempbuf = (char *) realloc (tempbuf, tempbufsize);
1046 /* Try to consume a simple name string token. If successful, returns
1047 a pointer to a nullbyte terminated copy of the name that can be used
1048 in symbol table lookups. If not successful, returns NULL. */
1051 match_simple_name_string ()
1053 char *tokptr = lexptr;
1055 if (isalpha (*tokptr))
1060 } while (isalnum (*tokptr) || (*tokptr == '_'));
1061 yylval.sval.ptr = lexptr;
1062 yylval.sval.length = tokptr - lexptr;
1064 result = copy_name (yylval.sval);
1065 for (tokptr = result; *tokptr; tokptr++)
1066 if (isupper (*tokptr))
1067 *tokptr = tolower(*tokptr);
1073 /* Start looking for a value composed of valid digits as set by the base
1074 in use. Note that '_' characters are valid anywhere, in any quantity,
1075 and are simply ignored. Since we must find at least one valid digit,
1076 or reject this token as an integer literal, we keep track of how many
1077 digits we have encountered. */
1080 decode_integer_value (base, tokptrptr, ivalptr)
1085 char *tokptr = *tokptrptr;
1089 while (*tokptr != '\0')
1091 temp = tolower (*tokptr);
1097 case '0': case '1': case '2': case '3': case '4':
1098 case '5': case '6': case '7': case '8': case '9':
1101 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
1117 /* Found something not in domain for current base. */
1118 tokptr--; /* Unconsume what gave us indigestion. */
1123 /* If we didn't find any digits, then we don't have a valid integer
1124 value, so reject the entire token. Otherwise, update the lexical
1125 scan pointer, and return non-zero for success. */
1133 *tokptrptr = tokptr;
1139 decode_integer_literal (valptr, tokptrptr)
1143 char *tokptr = *tokptrptr;
1146 int explicit_base = 0;
1148 /* Look for an explicit base specifier, which is optional. */
1181 /* If we found an explicit base ensure that the character after the
1182 explicit base is a single quote. */
1184 if (explicit_base && (*tokptr++ != '\''))
1189 /* Attempt to decode whatever follows as an integer value in the
1190 indicated base, updating the token pointer in the process and
1191 computing the value into ival. Also, if we have an explicit
1192 base, then the next character must not be a single quote, or we
1193 have a bitstring literal, so reject the entire token in this case.
1194 Otherwise, update the lexical scan pointer, and return non-zero
1197 if (!decode_integer_value (base, &tokptr, &ival))
1201 else if (explicit_base && (*tokptr == '\''))
1208 *tokptrptr = tokptr;
1213 /* If it wasn't for the fact that floating point values can contain '_'
1214 characters, we could just let strtod do all the hard work by letting it
1215 try to consume as much of the current token buffer as possible and
1216 find a legal conversion. Unfortunately we need to filter out the '_'
1217 characters before calling strtod, which we do by copying the other
1218 legal chars to a local buffer to be converted. However since we also
1219 need to keep track of where the last unconsumed character in the input
1220 buffer is, we have transfer only as many characters as may compose a
1221 legal floating point value. */
1224 match_float_literal ()
1226 char *tokptr = lexptr;
1231 extern double strtod ();
1233 /* Make local buffer in which to build the string to convert. This is
1234 required because underscores are valid in chill floating point numbers
1235 but not in the string passed to strtod to convert. The string will be
1236 no longer than our input string. */
1238 copy = buf = (char *) alloca (strlen (tokptr) + 1);
1240 /* Transfer all leading digits to the conversion buffer, discarding any
1243 while (isdigit (*tokptr) || *tokptr == '_')
1252 /* Now accept either a '.', or one of [eEdD]. Dot is legal regardless
1253 of whether we found any leading digits, and we simply accept it and
1254 continue on to look for the fractional part and/or exponent. One of
1255 [eEdD] is legal only if we have seen digits, and means that there
1256 is no fractional part. If we find neither of these, then this is
1257 not a floating point number, so return failure. */
1262 /* Accept and then look for fractional part and/or exponent. */
1275 goto collect_exponent;
1283 /* We found a '.', copy any fractional digits to the conversion buffer, up
1284 to the first nondigit, non-underscore character. */
1286 while (isdigit (*tokptr) || *tokptr == '_')
1295 /* Look for an exponent, which must start with one of [eEdD]. If none
1296 is found, jump directly to trying to convert what we have collected
1313 /* Accept an optional '-' or '+' following one of [eEdD]. */
1316 if (*tokptr == '+' || *tokptr == '-')
1318 *copy++ = *tokptr++;
1321 /* Now copy an exponent into the conversion buffer. Note that at the
1322 moment underscores are *not* allowed in exponents. */
1324 while (isdigit (*tokptr))
1326 *copy++ = *tokptr++;
1329 /* If we transfered any chars to the conversion buffer, try to interpret its
1330 contents as a floating point value. If any characters remain, then we
1331 must not have a valid floating point string. */
1337 dval = strtod (buf, ©);
1342 return (FLOAT_LITERAL);
1348 /* Recognize a string literal. A string literal is a nonzero sequence
1349 of characters enclosed in matching single or double quotes, except that
1350 a single character inside single quotes is a character literal, which
1351 we reject as a string literal. To embed the terminator character inside
1352 a string, it is simply doubled (I.E. "this""is""one""string") */
1355 match_string_literal ()
1357 char *tokptr = lexptr;
1359 for (tempbufindex = 0, tokptr++; *tokptr != '\0'; tokptr++)
1362 if (*tokptr == *lexptr)
1364 if (*(tokptr + 1) == *lexptr)
1373 tempbuf[tempbufindex++] = *tokptr;
1375 if (*tokptr == '\0' /* no terminator */
1376 || tempbufindex == 0 /* no string */
1377 || (tempbufindex == 1 && *tokptr == '\'')) /* char literal */
1383 tempbuf[tempbufindex] = '\0';
1384 yylval.sval.ptr = tempbuf;
1385 yylval.sval.length = tempbufindex;
1387 return (CHARACTER_STRING_LITERAL);
1391 /* Recognize a character literal. A character literal is single character
1392 or a control sequence, enclosed in single quotes. A control sequence
1393 is a comma separated list of one or more integer literals, enclosed
1394 in parenthesis and introduced with a circumflex character.
1396 EX: 'a' '^(7)' '^(7,8)'
1398 As a GNU chill extension, the syntax C'xx' is also recognized as a
1399 character literal, where xx is a hex value for the character.
1401 Note that more than a single character, enclosed in single quotes, is
1404 Also note that the control sequence form is not in GNU Chill since it
1405 is ambiguous with the string literal form using single quotes. I.E.
1406 is '^(7)' a character literal or a string literal. In theory it it
1407 possible to tell by context, but GNU Chill doesn't accept the control
1408 sequence form, so neither do we (for now the code is disabled).
1410 Returns CHARACTER_LITERAL if a match is found.
1414 match_character_literal ()
1416 char *tokptr = lexptr;
1419 if ((tolower (*tokptr) == 'c') && (*(tokptr + 1) == '\''))
1421 /* We have a GNU chill extension form, so skip the leading "C'",
1422 decode the hex value, and then ensure that we have a trailing
1423 single quote character. */
1425 if (!decode_integer_value (16, &tokptr, &ival) || (*tokptr != '\''))
1431 else if (*tokptr == '\'')
1435 /* Determine which form we have, either a control sequence or the
1436 single character form. */
1438 if ((*tokptr == '^') && (*(tokptr + 1) == '('))
1440 #if 0 /* Disable, see note above. -fnf */
1441 /* Match and decode a control sequence. Return zero if we don't
1442 find a valid integer literal, or if the next unconsumed character
1443 after the integer literal is not the trailing ')'.
1444 FIXME: We currently don't handle the multiple integer literal
1447 if (!decode_integer_literal (&ival, &tokptr) || (*tokptr++ != ')'))
1460 /* The trailing quote has not yet been consumed. If we don't find
1461 it, then we have no match. */
1463 if (*tokptr++ != '\'')
1470 /* Not a character literal. */
1473 yylval.typed_val.val = ival;
1474 yylval.typed_val.type = builtin_type_chill_char;
1476 return (CHARACTER_LITERAL);
1479 /* Recognize an integer literal, as specified in Z.200 sec 5.2.4.2.
1480 Note that according to 5.2.4.2, a single "_" is also a valid integer
1481 literal, however GNU-chill requires there to be at least one "digit"
1482 in any integer literal. */
1485 match_integer_literal ()
1487 char *tokptr = lexptr;
1490 if (!decode_integer_literal (&ival, &tokptr))
1496 yylval.typed_val.val = ival;
1497 yylval.typed_val.type = builtin_type_int;
1499 return (INTEGER_LITERAL);
1503 /* Recognize a bit-string literal, as specified in Z.200 sec 5.2.4.8
1504 Note that according to 5.2.4.8, a single "_" is also a valid bit-string
1505 literal, however GNU-chill requires there to be at least one "digit"
1506 in any bit-string literal. */
1509 match_bitstring_literal ()
1511 char *tokptr = lexptr;
1520 /* Look for the required explicit base specifier. */
1541 /* Ensure that the character after the explicit base is a single quote. */
1543 if (*tokptr++ != '\'')
1548 while (*tokptr != '\0' && *tokptr != '\'')
1550 digit = tolower (*tokptr);
1556 case '0': case '1': case '2': case '3': case '4':
1557 case '5': case '6': case '7': case '8': case '9':
1560 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
1570 /* Found something not in domain for current base. */
1575 /* Extract bits from digit, starting with the msbit appropriate for
1576 the current base, and packing them into the bitstring byte,
1577 starting at the lsbit. */
1578 for (mask = (base >> 1); mask > 0; mask >>= 1)
1584 tempbuf[tempbufindex] |= (1 << bitoffset);
1587 if (bitoffset == HOST_CHAR_BIT)
1596 /* Verify that we consumed everything up to the trailing single quote,
1597 and that we found some bits (IE not just underbars). */
1599 if (*tokptr++ != '\'')
1605 yylval.sval.ptr = tempbuf;
1606 yylval.sval.length = bitcount;
1608 return (BIT_STRING_LITERAL);
1612 /* Recognize tokens that start with '$'. These include:
1614 $regname A native register name or a "standard
1616 Return token GDB_REGNAME.
1618 $variable A convenience variable with a name chosen
1620 Return token GDB_VARIABLE.
1622 $digits Value history with index <digits>, starting
1623 from the first value which has index 1.
1626 $$digits Value history with index <digits> relative
1627 to the last value. I.E. $$0 is the last
1628 value, $$1 is the one previous to that, $$2
1629 is the one previous to $$1, etc.
1630 Return token GDB_LAST.
1632 $ | $0 | $$0 The last value in the value history.
1633 Return token GDB_LAST.
1635 $$ An abbreviation for the second to the last
1636 value in the value history, I.E. $$1
1637 Return token GDB_LAST.
1639 Note that we currently assume that register names and convenience
1640 variables follow the convention of starting with a letter or '_'.
1645 match_dollar_tokens ()
1653 /* We will always have a successful match, even if it is just for
1654 a single '$', the abbreviation for $$0. So advance lexptr. */
1658 if (*tokptr == '_' || isalpha (*tokptr))
1660 /* Look for a match with a native register name, usually something
1661 like "r0" for example. */
1663 for (regno = 0; regno < NUM_REGS; regno++)
1665 namelength = strlen (reg_names[regno]);
1666 if (STREQN (tokptr, reg_names[regno], namelength)
1667 && !isalnum (tokptr[namelength]))
1669 yylval.lval = regno;
1670 lexptr += namelength + 1;
1671 return (GDB_REGNAME);
1675 /* Look for a match with a standard register name, usually something
1676 like "pc", which gdb always recognizes as the program counter
1677 regardless of what the native register name is. */
1679 for (regno = 0; regno < num_std_regs; regno++)
1681 namelength = strlen (std_regs[regno].name);
1682 if (STREQN (tokptr, std_regs[regno].name, namelength)
1683 && !isalnum (tokptr[namelength]))
1685 yylval.lval = std_regs[regno].regnum;
1686 lexptr += namelength;
1687 return (GDB_REGNAME);
1691 /* Attempt to match against a convenience variable. Note that
1692 this will always succeed, because if no variable of that name
1693 already exists, the lookup_internalvar will create one for us.
1694 Also note that both lexptr and tokptr currently point to the
1695 start of the input string we are trying to match, and that we
1696 have already tested the first character for non-numeric, so we
1697 don't have to treat it specially. */
1699 while (*tokptr == '_' || isalnum (*tokptr))
1703 yylval.sval.ptr = lexptr;
1704 yylval.sval.length = tokptr - lexptr;
1705 yylval.ivar = lookup_internalvar (copy_name (yylval.sval));
1707 return (GDB_VARIABLE);
1710 /* Since we didn't match against a register name or convenience
1711 variable, our only choice left is a history value. */
1725 /* Attempt to decode more characters as an integer value giving
1726 the index in the history list. If successful, the value will
1727 overwrite ival (currently 0 or 1), and if not, ival will be
1728 left alone, which is good since it is currently correct for
1729 the '$' or '$$' case. */
1731 decode_integer_literal (&ival, &tokptr);
1732 yylval.lval = negate ? -ival : ival;
1743 static const struct token idtokentab[] =
1745 { "length", LENGTH },
1756 { "max", MAX_TOKEN },
1757 { "min", MIN_TOKEN },
1767 static const struct token tokentab2[] =
1769 { ":=", GDB_ASSIGNMENT },
1770 { "//", SLASH_SLASH },
1777 /* Read one token, getting characters through lexptr. */
1778 /* This is where we will check to make sure that the language and the
1779 operators used are compatible. */
1789 /* Skip over any leading whitespace. */
1790 while (isspace (*lexptr))
1794 /* Look for special single character cases which can't be the first
1795 character of some other multicharacter token. */
1812 /* Look for characters which start a particular kind of multicharacter
1813 token, such as a character literal, register name, convenience
1814 variable name, string literal, etc. */
1819 /* First try to match a string literal, which is any nonzero
1820 sequence of characters enclosed in matching single or double
1821 quotes, except that a single character inside single quotes
1822 is a character literal, so we have to catch that case also. */
1823 token = match_string_literal ();
1828 if (*lexptr == '\'')
1830 token = match_character_literal ();
1839 token = match_character_literal ();
1846 token = match_dollar_tokens ();
1853 /* See if it is a special token of length 2. */
1854 for (i = 0; i < sizeof (tokentab2) / sizeof (tokentab2[0]); i++)
1856 if (STREQN (lexptr, tokentab2[i].operator, 2))
1859 return (tokentab2[i].token);
1862 /* Look for single character cases which which could be the first
1863 character of some other multicharacter token, but aren't, or we
1864 would already have found it. */
1874 /* Look for a float literal before looking for an integer literal, so
1875 we match as much of the input stream as possible. */
1876 token = match_float_literal ();
1881 token = match_bitstring_literal ();
1886 token = match_integer_literal ();
1892 /* Try to match a simple name string, and if a match is found, then
1893 further classify what sort of name it is and return an appropriate
1894 token. Note that attempting to match a simple name string consumes
1895 the token from lexptr, so we can't back out if we later find that
1896 we can't classify what sort of name it is. */
1898 simplename = match_simple_name_string ();
1900 if (simplename != NULL)
1902 /* See if it is a reserved identifier. */
1903 for (i = 0; i < sizeof (idtokentab) / sizeof (idtokentab[0]); i++)
1905 if (STREQ (simplename, idtokentab[i].operator))
1907 return (idtokentab[i].token);
1911 /* Look for other special tokens. */
1912 if (STREQ (simplename, "true"))
1915 return (BOOLEAN_LITERAL);
1917 if (STREQ (simplename, "false"))
1920 return (BOOLEAN_LITERAL);
1923 sym = lookup_symbol (simplename, expression_context_block,
1924 VAR_NAMESPACE, (int *) NULL,
1925 (struct symtab **) NULL);
1928 yylval.ssym.stoken.ptr = NULL;
1929 yylval.ssym.stoken.length = 0;
1930 yylval.ssym.sym = sym;
1931 yylval.ssym.is_a_field_of_this = 0; /* FIXME, C++'ism */
1932 switch (SYMBOL_CLASS (sym))
1935 /* Found a procedure name. */
1936 return (GENERAL_PROCEDURE_NAME);
1938 /* Found a global or local static variable. */
1939 return (LOCATION_NAME);
1946 if (innermost_block == NULL
1947 || contained_in (block_found, innermost_block))
1949 innermost_block = block_found;
1951 return (LOCATION_NAME);
1955 return (LOCATION_NAME);
1958 yylval.tsym.type = SYMBOL_TYPE (sym);
1961 case LOC_CONST_BYTES:
1962 case LOC_OPTIMIZED_OUT:
1963 error ("Symbol \"%s\" names no location.", simplename);
1967 else if (!have_full_symbols () && !have_partial_symbols ())
1969 error ("No symbol table is loaded. Use the \"file\" command.");
1973 error ("No symbol \"%s\" in current context.", simplename);
1977 /* Catch single character tokens which are not part of some
1982 case '.': /* Not float for example. */
1984 while (isspace (*lexptr)) lexptr++;
1985 simplename = match_simple_name_string ();
1991 return (ILLEGAL_TOKEN);
1996 char *msg; /* unused */
1998 printf ("Parsing: %s\n", lexptr);
2001 error ("Invalid syntax in expression near character '%c'.", yychar);
2005 error ("Invalid syntax in expression");