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> location
251 %type <voidval> access_name
252 %type <voidval> primitive_value
253 %type <voidval> location_contents
254 %type <voidval> value_name
255 %type <voidval> literal
256 %type <voidval> tuple
257 %type <voidval> value_string_element
258 %type <voidval> value_string_slice
259 %type <voidval> value_array_element
260 %type <voidval> value_array_slice
261 %type <voidval> value_structure_field
262 %type <voidval> expression_conversion
263 %type <voidval> value_procedure_call
264 %type <voidval> value_built_in_routine_call
265 %type <voidval> chill_value_built_in_routine_call
266 %type <voidval> start_expression
267 %type <voidval> zero_adic_operator
268 %type <voidval> parenthesised_expression
269 %type <voidval> value
270 %type <voidval> undefined_value
271 %type <voidval> expression
272 %type <voidval> conditional_expression
273 %type <voidval> then_alternative
274 %type <voidval> else_alternative
275 %type <voidval> sub_expression
276 %type <voidval> value_case_alternative
277 %type <voidval> operand_0
278 %type <voidval> operand_1
279 %type <voidval> operand_2
280 %type <voidval> operand_3
281 %type <voidval> operand_4
282 %type <voidval> operand_5
283 %type <voidval> operand_6
284 %type <voidval> synonym_name
285 %type <voidval> value_enumeration_name
286 %type <voidval> value_do_with_name
287 %type <voidval> value_receive_name
288 %type <voidval> string_primitive_value
289 %type <voidval> start_element
290 %type <voidval> left_element
291 %type <voidval> right_element
292 %type <voidval> slice_size
293 %type <voidval> array_primitive_value
294 %type <voidval> expression_list
295 %type <voidval> lower_element
296 %type <voidval> upper_element
297 %type <voidval> first_element
298 %type <voidval> mode_argument
299 %type <voidval> upper_lower_argument
300 %type <voidval> length_argument
301 %type <voidval> array_mode_name
302 %type <voidval> string_mode_name
303 %type <voidval> variant_structure_mode_name
304 %type <voidval> boolean_expression
305 %type <voidval> case_selector_list
306 %type <voidval> subexpression
307 %type <voidval> case_label_specification
308 %type <voidval> buffer_location
309 %type <voidval> single_assignment_action
310 %type <tsym> mode_name
318 { write_exp_elt_opcode(OP_TYPE);
319 write_exp_elt_type($1.type);
320 write_exp_elt_opcode(OP_TYPE);}
333 undefined_value : FIXME_01
341 location : access_name
342 | primitive_value POINTER
344 write_exp_elt_opcode (UNOP_IND);
350 access_name : LOCATION_NAME
352 write_exp_elt_opcode (OP_VAR_VALUE);
353 write_exp_elt_block (NULL);
354 write_exp_elt_sym ($1.sym);
355 write_exp_elt_opcode (OP_VAR_VALUE);
357 | GDB_LAST /* gdb specific */
359 write_exp_elt_opcode (OP_LAST);
360 write_exp_elt_longcst ($1);
361 write_exp_elt_opcode (OP_LAST);
363 | GDB_REGNAME /* gdb specific */
365 write_exp_elt_opcode (OP_REGISTER);
366 write_exp_elt_longcst ($1);
367 write_exp_elt_opcode (OP_REGISTER);
369 | GDB_VARIABLE /* gdb specific */
371 write_exp_elt_opcode (OP_INTERNALVAR);
372 write_exp_elt_intern ($1);
373 write_exp_elt_opcode (OP_INTERNALVAR);
383 expression_list : expression
387 | expression_list ',' expression
394 primitive_value : location_contents
410 | value_string_element
418 | value_array_element
426 | value_structure_field
430 | expression_conversion
434 | value_procedure_call
438 | value_built_in_routine_call
450 | parenthesised_expression
458 location_contents: location
466 value_name : synonym_name
470 | value_enumeration_name
482 | GENERAL_PROCEDURE_NAME
484 write_exp_elt_opcode (OP_VAR_VALUE);
485 write_exp_elt_block (NULL);
486 write_exp_elt_sym ($1.sym);
487 write_exp_elt_opcode (OP_VAR_VALUE);
493 literal : INTEGER_LITERAL
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_BOOL);
503 write_exp_elt_longcst ((LONGEST) $1);
504 write_exp_elt_opcode (OP_BOOL);
508 write_exp_elt_opcode (OP_LONG);
509 write_exp_elt_type ($1.type);
510 write_exp_elt_longcst ((LONGEST) ($1.val));
511 write_exp_elt_opcode (OP_LONG);
515 write_exp_elt_opcode (OP_DOUBLE);
516 write_exp_elt_type (builtin_type_double);
517 write_exp_elt_dblcst ($1);
518 write_exp_elt_opcode (OP_DOUBLE);
528 | CHARACTER_STRING_LITERAL
530 write_exp_elt_opcode (OP_STRING);
531 write_exp_string ($1);
532 write_exp_elt_opcode (OP_STRING);
536 write_exp_elt_opcode (OP_BITSTRING);
537 write_exp_bitstring ($1);
538 write_exp_elt_opcode (OP_BITSTRING);
553 value_string_element: string_primitive_value '(' start_element ')'
561 value_string_slice: string_primitive_value '(' left_element ':' right_element ')'
565 | string_primitive_value '(' start_element UP slice_size ')'
573 value_array_element: array_primitive_value '('
574 /* This is to save the value of arglist_len
575 being accumulated for each dimension. */
576 { start_arglist (); }
579 write_exp_elt_opcode (MULTI_SUBSCRIPT);
580 write_exp_elt_longcst ((LONGEST) end_arglist ());
581 write_exp_elt_opcode (MULTI_SUBSCRIPT);
587 value_array_slice: array_primitive_value '(' lower_element ':' upper_element ')'
591 | array_primitive_value '(' first_element UP slice_size ')'
599 value_structure_field: primitive_value FIELD_NAME
600 { write_exp_elt_opcode (STRUCTOP_STRUCT);
601 write_exp_string ($2);
602 write_exp_elt_opcode (STRUCTOP_STRUCT);
608 expression_conversion: mode_name parenthesised_expression
610 write_exp_elt_opcode (UNOP_CAST);
611 write_exp_elt_type ($1.type);
612 write_exp_elt_opcode (UNOP_CAST);
618 value_procedure_call: FIXME_05
626 value_built_in_routine_call: chill_value_built_in_routine_call
634 start_expression: FIXME_06
637 } /* Not in GNU-Chill */
642 zero_adic_operator: FIXME_07
650 parenthesised_expression: '(' expression ')'
658 expression : operand_0
662 | single_assignment_action
666 | conditional_expression
672 conditional_expression : IF boolean_expression then_alternative else_alternative FI
676 | CASE case_selector_list OF value_case_alternative '[' ELSE sub_expression ']' ESAC
682 then_alternative: THEN subexpression
688 else_alternative: ELSE subexpression
692 | ELSIF boolean_expression then_alternative else_alternative
698 sub_expression : expression
704 value_case_alternative: case_label_specification ':' sub_expression ';'
712 operand_0 : operand_1
716 | operand_0 LOGIOR operand_1
718 write_exp_elt_opcode (BINOP_BITWISE_IOR);
720 | operand_0 ORIF operand_1
724 | operand_0 LOGXOR operand_1
726 write_exp_elt_opcode (BINOP_BITWISE_XOR);
732 operand_1 : operand_2
736 | operand_1 LOGAND operand_2
738 write_exp_elt_opcode (BINOP_BITWISE_AND);
740 | operand_1 ANDIF operand_2
748 operand_2 : operand_3
752 | operand_2 '=' operand_3
754 write_exp_elt_opcode (BINOP_EQUAL);
756 | operand_2 NOTEQUAL operand_3
758 write_exp_elt_opcode (BINOP_NOTEQUAL);
760 | operand_2 '>' operand_3
762 write_exp_elt_opcode (BINOP_GTR);
764 | operand_2 GTR operand_3
766 write_exp_elt_opcode (BINOP_GEQ);
768 | operand_2 '<' operand_3
770 write_exp_elt_opcode (BINOP_LESS);
772 | operand_2 LEQ operand_3
774 write_exp_elt_opcode (BINOP_LEQ);
776 | operand_2 IN operand_3
778 write_exp_elt_opcode (BINOP_IN);
785 operand_3 : operand_4
789 | operand_3 '+' operand_4
791 write_exp_elt_opcode (BINOP_ADD);
793 | operand_3 '-' operand_4
795 write_exp_elt_opcode (BINOP_SUB);
797 | operand_3 SLASH_SLASH operand_4
799 write_exp_elt_opcode (BINOP_CONCAT);
805 operand_4 : operand_5
809 | operand_4 '*' operand_5
811 write_exp_elt_opcode (BINOP_MUL);
813 | operand_4 '/' operand_5
815 write_exp_elt_opcode (BINOP_DIV);
817 | operand_4 MOD operand_5
819 write_exp_elt_opcode (BINOP_MOD);
821 | operand_4 REM operand_5
823 write_exp_elt_opcode (BINOP_REM);
829 operand_5 : operand_6
835 write_exp_elt_opcode (UNOP_NEG);
839 write_exp_elt_opcode (UNOP_LOGICAL_NOT);
841 | parenthesised_expression literal
842 /* We require the string operand to be a literal, to avoid some
843 nasty parsing ambiguities. */
845 write_exp_elt_opcode (BINOP_CONCAT);
851 operand_6 : POINTER location
853 write_exp_elt_opcode (UNOP_ADDR);
855 | RECEIVE buffer_location
868 single_assignment_action :
869 location GDB_ASSIGNMENT value
871 write_exp_elt_opcode (BINOP_ASSIGN);
877 chill_value_built_in_routine_call :
878 NUM '(' expression ')'
882 | PRED '(' expression ')'
886 | SUCC '(' expression ')'
890 | ABS '(' expression ')'
894 | CARD '(' expression ')'
898 | MAX_TOKEN '(' expression ')'
902 | MIN_TOKEN '(' expression ')'
906 | SIZE '(' location ')'
910 | SIZE '(' mode_argument ')'
914 | UPPER '(' upper_lower_argument ')'
918 | LOWER '(' upper_lower_argument ')'
922 | LENGTH '(' length_argument ')'
928 mode_argument : mode_name
932 | array_mode_name '(' expression ')'
936 | string_mode_name '(' expression ')'
940 | variant_structure_mode_name '(' expression_list ')'
949 upper_lower_argument : expression
959 length_argument : expression
967 array_primitive_value : primitive_value
974 /* Things which still need productions... */
976 array_mode_name : FIXME_08 { $$ = 0; }
977 string_mode_name : FIXME_09 { $$ = 0; }
978 variant_structure_mode_name: FIXME_10 { $$ = 0; }
979 synonym_name : FIXME_11 { $$ = 0; }
980 value_enumeration_name : FIXME_12 { $$ = 0; }
981 value_do_with_name : FIXME_13 { $$ = 0; }
982 value_receive_name : FIXME_14 { $$ = 0; }
983 string_primitive_value : FIXME_15 { $$ = 0; }
984 start_element : FIXME_16 { $$ = 0; }
985 left_element : FIXME_17 { $$ = 0; }
986 right_element : FIXME_18 { $$ = 0; }
987 slice_size : FIXME_19 { $$ = 0; }
988 lower_element : FIXME_20 { $$ = 0; }
989 upper_element : FIXME_21 { $$ = 0; }
990 first_element : FIXME_22 { $$ = 0; }
991 boolean_expression : FIXME_26 { $$ = 0; }
992 case_selector_list : FIXME_27 { $$ = 0; }
993 subexpression : FIXME_28 { $$ = 0; }
994 case_label_specification: FIXME_29 { $$ = 0; }
995 buffer_location : FIXME_30 { $$ = 0; }
999 /* Implementation of a dynamically expandable buffer for processing input
1000 characters acquired through lexptr and building a value to return in
1003 static char *tempbuf; /* Current buffer contents */
1004 static int tempbufsize; /* Size of allocated buffer */
1005 static int tempbufindex; /* Current index into buffer */
1007 #define GROWBY_MIN_SIZE 64 /* Minimum amount to grow buffer by */
1009 #define CHECKBUF(size) \
1011 if (tempbufindex + (size) >= tempbufsize) \
1013 growbuf_by_size (size); \
1017 /* Grow the static temp buffer if necessary, including allocating the first one
1021 growbuf_by_size (count)
1026 growby = max (count, GROWBY_MIN_SIZE);
1027 tempbufsize += growby;
1028 if (tempbuf == NULL)
1030 tempbuf = (char *) malloc (tempbufsize);
1034 tempbuf = (char *) realloc (tempbuf, tempbufsize);
1038 /* Try to consume a simple name string token. If successful, returns
1039 a pointer to a nullbyte terminated copy of the name that can be used
1040 in symbol table lookups. If not successful, returns NULL. */
1043 match_simple_name_string ()
1045 char *tokptr = lexptr;
1047 if (isalpha (*tokptr) || *tokptr == '_')
1052 } while (isalnum (*tokptr) || (*tokptr == '_'));
1053 yylval.sval.ptr = lexptr;
1054 yylval.sval.length = tokptr - lexptr;
1056 result = copy_name (yylval.sval);
1057 for (tokptr = result; *tokptr; tokptr++)
1058 if (isupper (*tokptr))
1059 *tokptr = tolower(*tokptr);
1065 /* Start looking for a value composed of valid digits as set by the base
1066 in use. Note that '_' characters are valid anywhere, in any quantity,
1067 and are simply ignored. Since we must find at least one valid digit,
1068 or reject this token as an integer literal, we keep track of how many
1069 digits we have encountered. */
1072 decode_integer_value (base, tokptrptr, ivalptr)
1077 char *tokptr = *tokptrptr;
1081 while (*tokptr != '\0')
1083 temp = tolower (*tokptr);
1089 case '0': case '1': case '2': case '3': case '4':
1090 case '5': case '6': case '7': case '8': case '9':
1093 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
1109 /* Found something not in domain for current base. */
1110 tokptr--; /* Unconsume what gave us indigestion. */
1115 /* If we didn't find any digits, then we don't have a valid integer
1116 value, so reject the entire token. Otherwise, update the lexical
1117 scan pointer, and return non-zero for success. */
1125 *tokptrptr = tokptr;
1131 decode_integer_literal (valptr, tokptrptr)
1135 char *tokptr = *tokptrptr;
1138 int explicit_base = 0;
1140 /* Look for an explicit base specifier, which is optional. */
1173 /* If we found an explicit base ensure that the character after the
1174 explicit base is a single quote. */
1176 if (explicit_base && (*tokptr++ != '\''))
1181 /* Attempt to decode whatever follows as an integer value in the
1182 indicated base, updating the token pointer in the process and
1183 computing the value into ival. Also, if we have an explicit
1184 base, then the next character must not be a single quote, or we
1185 have a bitstring literal, so reject the entire token in this case.
1186 Otherwise, update the lexical scan pointer, and return non-zero
1189 if (!decode_integer_value (base, &tokptr, &ival))
1193 else if (explicit_base && (*tokptr == '\''))
1200 *tokptrptr = tokptr;
1205 /* If it wasn't for the fact that floating point values can contain '_'
1206 characters, we could just let strtod do all the hard work by letting it
1207 try to consume as much of the current token buffer as possible and
1208 find a legal conversion. Unfortunately we need to filter out the '_'
1209 characters before calling strtod, which we do by copying the other
1210 legal chars to a local buffer to be converted. However since we also
1211 need to keep track of where the last unconsumed character in the input
1212 buffer is, we have transfer only as many characters as may compose a
1213 legal floating point value. */
1216 match_float_literal ()
1218 char *tokptr = lexptr;
1222 extern double strtod ();
1224 /* Make local buffer in which to build the string to convert. This is
1225 required because underscores are valid in chill floating point numbers
1226 but not in the string passed to strtod to convert. The string will be
1227 no longer than our input string. */
1229 copy = buf = (char *) alloca (strlen (tokptr) + 1);
1231 /* Transfer all leading digits to the conversion buffer, discarding any
1234 while (isdigit (*tokptr) || *tokptr == '_')
1243 /* Now accept either a '.', or one of [eEdD]. Dot is legal regardless
1244 of whether we found any leading digits, and we simply accept it and
1245 continue on to look for the fractional part and/or exponent. One of
1246 [eEdD] is legal only if we have seen digits, and means that there
1247 is no fractional part. If we find neither of these, then this is
1248 not a floating point number, so return failure. */
1253 /* Accept and then look for fractional part and/or exponent. */
1266 goto collect_exponent;
1274 /* We found a '.', copy any fractional digits to the conversion buffer, up
1275 to the first nondigit, non-underscore character. */
1277 while (isdigit (*tokptr) || *tokptr == '_')
1286 /* Look for an exponent, which must start with one of [eEdD]. If none
1287 is found, jump directly to trying to convert what we have collected
1304 /* Accept an optional '-' or '+' following one of [eEdD]. */
1307 if (*tokptr == '+' || *tokptr == '-')
1309 *copy++ = *tokptr++;
1312 /* Now copy an exponent into the conversion buffer. Note that at the
1313 moment underscores are *not* allowed in exponents. */
1315 while (isdigit (*tokptr))
1317 *copy++ = *tokptr++;
1320 /* If we transfered any chars to the conversion buffer, try to interpret its
1321 contents as a floating point value. If any characters remain, then we
1322 must not have a valid floating point string. */
1328 dval = strtod (buf, ©);
1333 return (FLOAT_LITERAL);
1339 /* Recognize a string literal. A string literal is a sequence
1340 of characters enclosed in matching single or double quotes, except that
1341 a single character inside single quotes is a character literal, which
1342 we reject as a string literal. To embed the terminator character inside
1343 a string, it is simply doubled (I.E. "this""is""one""string") */
1346 match_string_literal ()
1348 char *tokptr = lexptr;
1350 for (tempbufindex = 0, tokptr++; *tokptr != '\0'; tokptr++)
1353 if (*tokptr == *lexptr)
1355 if (*(tokptr + 1) == *lexptr)
1364 tempbuf[tempbufindex++] = *tokptr;
1366 if (*tokptr == '\0' /* no terminator */
1367 || (tempbufindex == 1 && *tokptr == '\'')) /* char literal */
1373 tempbuf[tempbufindex] = '\0';
1374 yylval.sval.ptr = tempbuf;
1375 yylval.sval.length = tempbufindex;
1377 return (CHARACTER_STRING_LITERAL);
1381 /* Recognize a character literal. A character literal is single character
1382 or a control sequence, enclosed in single quotes. A control sequence
1383 is a comma separated list of one or more integer literals, enclosed
1384 in parenthesis and introduced with a circumflex character.
1386 EX: 'a' '^(7)' '^(7,8)'
1388 As a GNU chill extension, the syntax C'xx' is also recognized as a
1389 character literal, where xx is a hex value for the character.
1391 Note that more than a single character, enclosed in single quotes, is
1394 Also note that the control sequence form is not in GNU Chill since it
1395 is ambiguous with the string literal form using single quotes. I.E.
1396 is '^(7)' a character literal or a string literal. In theory it it
1397 possible to tell by context, but GNU Chill doesn't accept the control
1398 sequence form, so neither do we (for now the code is disabled).
1400 Returns CHARACTER_LITERAL if a match is found.
1404 match_character_literal ()
1406 char *tokptr = lexptr;
1409 if ((tolower (*tokptr) == 'c') && (*(tokptr + 1) == '\''))
1411 /* We have a GNU chill extension form, so skip the leading "C'",
1412 decode the hex value, and then ensure that we have a trailing
1413 single quote character. */
1415 if (!decode_integer_value (16, &tokptr, &ival) || (*tokptr != '\''))
1421 else if (*tokptr == '\'')
1425 /* Determine which form we have, either a control sequence or the
1426 single character form. */
1428 if ((*tokptr == '^') && (*(tokptr + 1) == '('))
1430 #if 0 /* Disable, see note above. -fnf */
1431 /* Match and decode a control sequence. Return zero if we don't
1432 find a valid integer literal, or if the next unconsumed character
1433 after the integer literal is not the trailing ')'.
1434 FIXME: We currently don't handle the multiple integer literal
1437 if (!decode_integer_literal (&ival, &tokptr) || (*tokptr++ != ')'))
1450 /* The trailing quote has not yet been consumed. If we don't find
1451 it, then we have no match. */
1453 if (*tokptr++ != '\'')
1460 /* Not a character literal. */
1463 yylval.typed_val.val = ival;
1464 yylval.typed_val.type = builtin_type_chill_char;
1466 return (CHARACTER_LITERAL);
1469 /* Recognize an integer literal, as specified in Z.200 sec 5.2.4.2.
1470 Note that according to 5.2.4.2, a single "_" is also a valid integer
1471 literal, however GNU-chill requires there to be at least one "digit"
1472 in any integer literal. */
1475 match_integer_literal ()
1477 char *tokptr = lexptr;
1480 if (!decode_integer_literal (&ival, &tokptr))
1486 yylval.typed_val.val = ival;
1487 yylval.typed_val.type = builtin_type_int;
1489 return (INTEGER_LITERAL);
1493 /* Recognize a bit-string literal, as specified in Z.200 sec 5.2.4.8
1494 Note that according to 5.2.4.8, a single "_" is also a valid bit-string
1495 literal, however GNU-chill requires there to be at least one "digit"
1496 in any bit-string literal. */
1499 match_bitstring_literal ()
1501 char *tokptr = lexptr;
1510 /* Look for the required explicit base specifier. */
1531 /* Ensure that the character after the explicit base is a single quote. */
1533 if (*tokptr++ != '\'')
1538 while (*tokptr != '\0' && *tokptr != '\'')
1540 digit = tolower (*tokptr);
1546 case '0': case '1': case '2': case '3': case '4':
1547 case '5': case '6': case '7': case '8': case '9':
1550 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
1560 /* Found something not in domain for current base. */
1565 /* Extract bits from digit, starting with the msbit appropriate for
1566 the current base, and packing them into the bitstring byte,
1567 starting at the lsbit. */
1568 for (mask = (base >> 1); mask > 0; mask >>= 1)
1574 tempbuf[tempbufindex] |= (1 << bitoffset);
1577 if (bitoffset == HOST_CHAR_BIT)
1586 /* Verify that we consumed everything up to the trailing single quote,
1587 and that we found some bits (IE not just underbars). */
1589 if (*tokptr++ != '\'')
1595 yylval.sval.ptr = tempbuf;
1596 yylval.sval.length = bitcount;
1598 return (BIT_STRING_LITERAL);
1602 /* Recognize tokens that start with '$'. These include:
1604 $regname A native register name or a "standard
1606 Return token GDB_REGNAME.
1608 $variable A convenience variable with a name chosen
1610 Return token GDB_VARIABLE.
1612 $digits Value history with index <digits>, starting
1613 from the first value which has index 1.
1616 $$digits Value history with index <digits> relative
1617 to the last value. I.E. $$0 is the last
1618 value, $$1 is the one previous to that, $$2
1619 is the one previous to $$1, etc.
1620 Return token GDB_LAST.
1622 $ | $0 | $$0 The last value in the value history.
1623 Return token GDB_LAST.
1625 $$ An abbreviation for the second to the last
1626 value in the value history, I.E. $$1
1627 Return token GDB_LAST.
1629 Note that we currently assume that register names and convenience
1630 variables follow the convention of starting with a letter or '_'.
1635 match_dollar_tokens ()
1643 /* We will always have a successful match, even if it is just for
1644 a single '$', the abbreviation for $$0. So advance lexptr. */
1648 if (*tokptr == '_' || isalpha (*tokptr))
1650 /* Look for a match with a native register name, usually something
1651 like "r0" for example. */
1653 for (regno = 0; regno < NUM_REGS; regno++)
1655 namelength = strlen (reg_names[regno]);
1656 if (STREQN (tokptr, reg_names[regno], namelength)
1657 && !isalnum (tokptr[namelength]))
1659 yylval.lval = regno;
1660 lexptr += namelength;
1661 return (GDB_REGNAME);
1665 /* Look for a match with a standard register name, usually something
1666 like "pc", which gdb always recognizes as the program counter
1667 regardless of what the native register name is. */
1669 for (regno = 0; regno < num_std_regs; regno++)
1671 namelength = strlen (std_regs[regno].name);
1672 if (STREQN (tokptr, std_regs[regno].name, namelength)
1673 && !isalnum (tokptr[namelength]))
1675 yylval.lval = std_regs[regno].regnum;
1676 lexptr += namelength;
1677 return (GDB_REGNAME);
1681 /* Attempt to match against a convenience variable. Note that
1682 this will always succeed, because if no variable of that name
1683 already exists, the lookup_internalvar will create one for us.
1684 Also note that both lexptr and tokptr currently point to the
1685 start of the input string we are trying to match, and that we
1686 have already tested the first character for non-numeric, so we
1687 don't have to treat it specially. */
1689 while (*tokptr == '_' || isalnum (*tokptr))
1693 yylval.sval.ptr = lexptr;
1694 yylval.sval.length = tokptr - lexptr;
1695 yylval.ivar = lookup_internalvar (copy_name (yylval.sval));
1697 return (GDB_VARIABLE);
1700 /* Since we didn't match against a register name or convenience
1701 variable, our only choice left is a history value. */
1715 /* Attempt to decode more characters as an integer value giving
1716 the index in the history list. If successful, the value will
1717 overwrite ival (currently 0 or 1), and if not, ival will be
1718 left alone, which is good since it is currently correct for
1719 the '$' or '$$' case. */
1721 decode_integer_literal (&ival, &tokptr);
1722 yylval.lval = negate ? -ival : ival;
1733 static const struct token idtokentab[] =
1735 { "length", LENGTH },
1746 { "max", MAX_TOKEN },
1747 { "min", MIN_TOKEN },
1757 static const struct token tokentab2[] =
1759 { ":=", GDB_ASSIGNMENT },
1760 { "//", SLASH_SLASH },
1767 /* Read one token, getting characters through lexptr. */
1768 /* This is where we will check to make sure that the language and the
1769 operators used are compatible. */
1779 /* Skip over any leading whitespace. */
1780 while (isspace (*lexptr))
1784 /* Look for special single character cases which can't be the first
1785 character of some other multicharacter token. */
1802 /* Look for characters which start a particular kind of multicharacter
1803 token, such as a character literal, register name, convenience
1804 variable name, string literal, etc. */
1809 /* First try to match a string literal, which is any
1810 sequence of characters enclosed in matching single or double
1811 quotes, except that a single character inside single quotes
1812 is a character literal, so we have to catch that case also. */
1813 token = match_string_literal ();
1818 if (*lexptr == '\'')
1820 token = match_character_literal ();
1829 token = match_character_literal ();
1836 token = match_dollar_tokens ();
1843 /* See if it is a special token of length 2. */
1844 for (i = 0; i < sizeof (tokentab2) / sizeof (tokentab2[0]); i++)
1846 if (STREQN (lexptr, tokentab2[i].operator, 2))
1849 return (tokentab2[i].token);
1852 /* Look for single character cases which which could be the first
1853 character of some other multicharacter token, but aren't, or we
1854 would already have found it. */
1864 /* Look for a float literal before looking for an integer literal, so
1865 we match as much of the input stream as possible. */
1866 token = match_float_literal ();
1871 token = match_bitstring_literal ();
1876 token = match_integer_literal ();
1882 /* Try to match a simple name string, and if a match is found, then
1883 further classify what sort of name it is and return an appropriate
1884 token. Note that attempting to match a simple name string consumes
1885 the token from lexptr, so we can't back out if we later find that
1886 we can't classify what sort of name it is. */
1888 simplename = match_simple_name_string ();
1890 if (simplename != NULL)
1892 /* See if it is a reserved identifier. */
1893 for (i = 0; i < sizeof (idtokentab) / sizeof (idtokentab[0]); i++)
1895 if (STREQ (simplename, idtokentab[i].operator))
1897 return (idtokentab[i].token);
1901 /* Look for other special tokens. */
1902 if (STREQ (simplename, "true"))
1905 return (BOOLEAN_LITERAL);
1907 if (STREQ (simplename, "false"))
1910 return (BOOLEAN_LITERAL);
1913 sym = lookup_symbol (simplename, expression_context_block,
1914 VAR_NAMESPACE, (int *) NULL,
1915 (struct symtab **) NULL);
1918 yylval.ssym.stoken.ptr = NULL;
1919 yylval.ssym.stoken.length = 0;
1920 yylval.ssym.sym = sym;
1921 yylval.ssym.is_a_field_of_this = 0; /* FIXME, C++'ism */
1922 switch (SYMBOL_CLASS (sym))
1925 /* Found a procedure name. */
1926 return (GENERAL_PROCEDURE_NAME);
1928 /* Found a global or local static variable. */
1929 return (LOCATION_NAME);
1934 case LOC_REGPARM_ADDR:
1938 case LOC_BASEREG_ARG:
1939 if (innermost_block == NULL
1940 || contained_in (block_found, innermost_block))
1942 innermost_block = block_found;
1944 return (LOCATION_NAME);
1948 return (LOCATION_NAME);
1951 yylval.tsym.type = SYMBOL_TYPE (sym);
1954 case LOC_CONST_BYTES:
1955 case LOC_OPTIMIZED_OUT:
1956 error ("Symbol \"%s\" names no location.", simplename);
1960 else if (!have_full_symbols () && !have_partial_symbols ())
1962 error ("No symbol table is loaded. Use the \"file\" command.");
1966 error ("No symbol \"%s\" in current context.", simplename);
1970 /* Catch single character tokens which are not part of some
1975 case '.': /* Not float for example. */
1977 while (isspace (*lexptr)) lexptr++;
1978 simplename = match_simple_name_string ();
1984 return (ILLEGAL_TOKEN);
1989 char *msg; /* unused */
1991 printf_unfiltered ("Parsing: %s\n", lexptr);
1994 error ("Invalid syntax in expression near character '%c'.", yychar);
1998 error ("Invalid syntax in expression");