1 /* YACC grammar for Chill expressions, for GDB.
2 Copyright (C) 1992 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 yydebug chill_debug
78 #define yypact chill_pact
81 #define yydef chill_def
82 #define yychk chill_chk
83 #define yypgo chill_pgo
84 #define yyact chill_act
85 #define yyexca chill_exca
86 #define yyerrflag chill_errflag
87 #define yynerrs chill_nerrs
91 #define yy_yys chill_yys
92 #define yystate chill_state
93 #define yytmp chill_tmp
95 #define yy_yyv chill_yyv
96 #define yyval chill_val
97 #define yylloc chill_lloc
98 #define yyreds chill_reds /* With YYDEBUG defined */
99 #define yytoks chill_toks /* With YYDEBUG defined */
102 #define YYDEBUG 0 /* Default to no yydebug support */
106 yyparse PARAMS ((void));
109 yylex PARAMS ((void));
112 yyerror PARAMS ((char *));
116 /* Although the yacc "value" of an expression is not used,
117 since the result is stored in the structure being created,
118 other node types do have values. */
123 unsigned LONGEST ulval;
133 struct symtoken ssym;
136 enum exp_opcode opcode;
137 struct internalvar *ivar;
143 %token <voidval> FIXME_01
144 %token <voidval> FIXME_02
145 %token <voidval> FIXME_03
146 %token <voidval> FIXME_04
147 %token <voidval> FIXME_05
148 %token <voidval> FIXME_06
149 %token <voidval> FIXME_07
150 %token <voidval> FIXME_08
151 %token <voidval> FIXME_09
152 %token <voidval> FIXME_10
153 %token <voidval> FIXME_11
154 %token <voidval> FIXME_12
155 %token <voidval> FIXME_13
156 %token <voidval> FIXME_14
157 %token <voidval> FIXME_15
158 %token <voidval> FIXME_16
159 %token <voidval> FIXME_17
160 %token <voidval> FIXME_18
161 %token <voidval> FIXME_19
162 %token <voidval> FIXME_20
163 %token <voidval> FIXME_21
164 %token <voidval> FIXME_22
165 %token <voidval> FIXME_23
166 %token <voidval> FIXME_24
167 %token <voidval> FIXME_25
168 %token <voidval> FIXME_26
169 %token <voidval> FIXME_27
170 %token <voidval> FIXME_28
171 %token <voidval> FIXME_29
172 %token <voidval> FIXME_30
174 %token <typed_val> INTEGER_LITERAL
175 %token <ulval> BOOLEAN_LITERAL
176 %token <typed_val> CHARACTER_LITERAL
177 %token <dval> FLOAT_LITERAL
178 %token <ssym> GENERAL_PROCEDURE_NAME
179 %token <ssym> LOCATION_NAME
180 %token <voidval> SET_LITERAL
181 %token <voidval> EMPTINESS_LITERAL
182 %token <sval> CHARACTER_STRING_LITERAL
183 %token <sval> BIT_STRING_LITERAL
184 %token <tsym> TYPENAME
185 %token <sval> FIELD_NAME
190 %token <voidval> CASE
192 %token <voidval> ESAC
193 %token <voidval> LOGIOR
194 %token <voidval> ORIF
195 %token <voidval> LOGXOR
196 %token <voidval> LOGAND
197 %token <voidval> ANDIF
199 %token <voidval> NOTEQUAL
209 %token <voidval> SLASH_SLASH
213 %token <voidval> POINTER
214 %token <voidval> RECEIVE
221 %token <voidval> THEN
222 %token <voidval> ELSE
224 %token <voidval> ELSIF
225 %token <voidval> ILLEGAL_TOKEN
227 %token <voidval> PRED
228 %token <voidval> SUCC
230 %token <voidval> CARD
233 %token <voidval> SIZE
234 %token <voidval> UPPER
235 %token <voidval> LOWER
236 %token <voidval> LENGTH
238 /* Tokens which are not Chill tokens used in expressions, but rather GDB
239 specific things that we recognize in the same context as Chill tokens
240 (register names for example). */
242 %token <lval> GDB_REGNAME /* Machine register name */
243 %token <lval> GDB_LAST /* Value history */
244 %token <ivar> GDB_VARIABLE /* Convenience variable */
245 %token <voidval> GDB_ASSIGNMENT /* Assign value to somewhere */
247 %type <voidval> location
248 %type <voidval> access_name
249 %type <voidval> primitive_value
250 %type <voidval> location_contents
251 %type <voidval> value_name
252 %type <voidval> literal
253 %type <voidval> tuple
254 %type <voidval> value_string_element
255 %type <voidval> value_string_slice
256 %type <voidval> value_array_element
257 %type <voidval> value_array_slice
258 %type <voidval> value_structure_field
259 %type <voidval> expression_conversion
260 %type <voidval> value_procedure_call
261 %type <voidval> value_built_in_routine_call
262 %type <voidval> chill_value_built_in_routine_call
263 %type <voidval> start_expression
264 %type <voidval> zero_adic_operator
265 %type <voidval> parenthesised_expression
266 %type <voidval> value
267 %type <voidval> undefined_value
268 %type <voidval> expression
269 %type <voidval> conditional_expression
270 %type <voidval> then_alternative
271 %type <voidval> else_alternative
272 %type <voidval> sub_expression
273 %type <voidval> value_case_alternative
274 %type <voidval> operand_0
275 %type <voidval> operand_1
276 %type <voidval> operand_2
277 %type <voidval> operand_3
278 %type <voidval> operand_4
279 %type <voidval> operand_5
280 %type <voidval> operand_6
281 %type <voidval> synonym_name
282 %type <voidval> value_enumeration_name
283 %type <voidval> value_do_with_name
284 %type <voidval> value_receive_name
285 %type <voidval> string_primitive_value
286 %type <voidval> start_element
287 %type <voidval> left_element
288 %type <voidval> right_element
289 %type <voidval> slice_size
290 %type <voidval> array_primitive_value
291 %type <voidval> expression_list
292 %type <voidval> lower_element
293 %type <voidval> upper_element
294 %type <voidval> first_element
295 %type <voidval> structure_primitive_value
296 %type <voidval> mode_argument
297 %type <voidval> upper_lower_argument
298 %type <voidval> length_argument
299 %type <voidval> array_mode_name
300 %type <voidval> string_mode_name
301 %type <voidval> variant_structure_mode_name
302 %type <voidval> boolean_expression
303 %type <voidval> case_selector_list
304 %type <voidval> subexpression
305 %type <voidval> case_label_specification
306 %type <voidval> buffer_location
307 %type <voidval> single_assignment_action
308 %type <tsym> mode_name
316 { write_exp_elt_opcode(OP_TYPE);
317 write_exp_elt_type($1.type);
318 write_exp_elt_opcode(OP_TYPE);}
331 undefined_value : FIXME_01
339 location : access_name
343 | primitive_value POINTER
345 write_exp_elt_opcode (UNOP_IND);
351 access_name : LOCATION_NAME
353 write_exp_elt_opcode (OP_VAR_VALUE);
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_sym ($1.sym);
486 write_exp_elt_opcode (OP_VAR_VALUE);
492 literal : INTEGER_LITERAL
494 write_exp_elt_opcode (OP_LONG);
495 write_exp_elt_type ($1.type);
496 write_exp_elt_longcst ((LONGEST) ($1.val));
497 write_exp_elt_opcode (OP_LONG);
501 write_exp_elt_opcode (OP_BOOL);
502 write_exp_elt_longcst ((LONGEST) $1);
503 write_exp_elt_opcode (OP_BOOL);
507 write_exp_elt_opcode (OP_LONG);
508 write_exp_elt_type ($1.type);
509 write_exp_elt_longcst ((LONGEST) ($1.val));
510 write_exp_elt_opcode (OP_LONG);
514 write_exp_elt_opcode (OP_DOUBLE);
515 write_exp_elt_type (builtin_type_double);
516 write_exp_elt_dblcst ($1);
517 write_exp_elt_opcode (OP_DOUBLE);
527 | CHARACTER_STRING_LITERAL
529 write_exp_elt_opcode (OP_STRING);
530 write_exp_string ($1);
531 write_exp_elt_opcode (OP_STRING);
535 write_exp_elt_opcode (OP_BITSTRING);
536 write_exp_bitstring ($1);
537 write_exp_elt_opcode (OP_BITSTRING);
552 value_string_element: string_primitive_value '(' start_element ')'
560 value_string_slice: string_primitive_value '(' left_element ':' right_element ')'
564 | string_primitive_value '(' start_element UP slice_size ')'
572 value_array_element: array_primitive_value '('
573 /* This is to save the value of arglist_len
574 being accumulated for each dimension. */
575 { start_arglist (); }
578 write_exp_elt_opcode (MULTI_SUBSCRIPT);
579 write_exp_elt_longcst ((LONGEST) end_arglist ());
580 write_exp_elt_opcode (MULTI_SUBSCRIPT);
586 value_array_slice: array_primitive_value '(' lower_element ':' upper_element ')'
590 | array_primitive_value '(' first_element UP slice_size ')'
598 value_structure_field: primitive_value FIELD_NAME
599 { write_exp_elt_opcode (STRUCTOP_STRUCT);
600 write_exp_string ($2);
601 write_exp_elt_opcode (STRUCTOP_STRUCT);
607 expression_conversion: mode_name parenthesised_expression
609 write_exp_elt_opcode (UNOP_CAST);
610 write_exp_elt_type ($1.type);
611 write_exp_elt_opcode (UNOP_CAST);
617 value_procedure_call: FIXME_05
625 value_built_in_routine_call: chill_value_built_in_routine_call
633 start_expression: FIXME_06
636 } /* Not in GNU-Chill */
641 zero_adic_operator: FIXME_07
649 parenthesised_expression: '(' expression ')'
657 expression : operand_0
661 | single_assignment_action
665 | conditional_expression
671 conditional_expression : IF boolean_expression then_alternative else_alternative FI
675 | CASE case_selector_list OF value_case_alternative '[' ELSE sub_expression ']' ESAC
681 then_alternative: THEN subexpression
687 else_alternative: ELSE subexpression
691 | ELSIF boolean_expression then_alternative else_alternative
697 sub_expression : expression
703 value_case_alternative: case_label_specification ':' sub_expression ';'
711 operand_0 : operand_1
715 | operand_0 LOGIOR operand_1
717 write_exp_elt_opcode (BINOP_BITWISE_IOR);
719 | operand_0 ORIF operand_1
723 | operand_0 LOGXOR operand_1
725 write_exp_elt_opcode (BINOP_BITWISE_XOR);
731 operand_1 : operand_2
735 | operand_1 LOGAND operand_2
737 write_exp_elt_opcode (BINOP_BITWISE_AND);
739 | operand_1 ANDIF operand_2
747 operand_2 : operand_3
751 | operand_2 '=' operand_3
753 write_exp_elt_opcode (BINOP_EQUAL);
755 | operand_2 NOTEQUAL operand_3
757 write_exp_elt_opcode (BINOP_NOTEQUAL);
759 | operand_2 '>' operand_3
761 write_exp_elt_opcode (BINOP_GTR);
763 | operand_2 GTR operand_3
765 write_exp_elt_opcode (BINOP_GEQ);
767 | operand_2 '<' operand_3
769 write_exp_elt_opcode (BINOP_LESS);
771 | operand_2 LEQ operand_3
773 write_exp_elt_opcode (BINOP_LEQ);
775 | operand_2 IN operand_3
784 operand_3 : operand_4
788 | operand_3 '+' operand_4
790 write_exp_elt_opcode (BINOP_ADD);
792 | operand_3 '-' operand_4
794 write_exp_elt_opcode (BINOP_SUB);
796 | operand_3 SLASH_SLASH operand_4
798 write_exp_elt_opcode (BINOP_CONCAT);
804 operand_4 : operand_5
808 | operand_4 '*' operand_5
810 write_exp_elt_opcode (BINOP_MUL);
812 | operand_4 '/' operand_5
814 write_exp_elt_opcode (BINOP_DIV);
816 | operand_4 MOD operand_5
818 write_exp_elt_opcode (BINOP_MOD);
820 | operand_4 REM operand_5
822 write_exp_elt_opcode (BINOP_REM);
828 operand_5 : operand_6
834 write_exp_elt_opcode (UNOP_NEG);
838 write_exp_elt_opcode (UNOP_LOGICAL_NOT);
840 | parenthesised_expression CHARACTER_STRING_LITERAL
841 /* We require the string operand to be a literal, to avoid some
842 nasty parsing ambiguities. */
844 write_exp_elt_opcode (BINOP_CONCAT);
850 operand_6 : POINTER location
852 write_exp_elt_opcode (UNOP_ADDR);
854 | RECEIVE buffer_location
867 single_assignment_action :
868 location GDB_ASSIGNMENT value
870 write_exp_elt_opcode (BINOP_ASSIGN);
876 chill_value_built_in_routine_call :
877 NUM '(' expression ')'
881 | PRED '(' expression ')'
885 | SUCC '(' expression ')'
889 | ABS '(' expression ')'
893 | CARD '(' expression ')'
897 | MAX '(' expression ')'
901 | MIN '(' expression ')'
905 | SIZE '(' location ')'
909 | SIZE '(' mode_argument ')'
913 | UPPER '(' upper_lower_argument ')'
917 | LOWER '(' upper_lower_argument ')'
921 | LENGTH '(' length_argument ')'
927 mode_argument : mode_name
931 | array_mode_name '(' expression ')'
935 | string_mode_name '(' expression ')'
939 | variant_structure_mode_name '(' expression_list ')'
948 upper_lower_argument : expression
958 length_argument : expression
966 array_primitive_value : primitive_value
973 /* Things which still need productions... */
975 array_mode_name : FIXME_08 { $$ = 0; }
976 string_mode_name : FIXME_09 { $$ = 0; }
977 variant_structure_mode_name: FIXME_10 { $$ = 0; }
978 synonym_name : FIXME_11 { $$ = 0; }
979 value_enumeration_name : FIXME_12 { $$ = 0; }
980 value_do_with_name : FIXME_13 { $$ = 0; }
981 value_receive_name : FIXME_14 { $$ = 0; }
982 string_primitive_value : FIXME_15 { $$ = 0; }
983 start_element : FIXME_16 { $$ = 0; }
984 left_element : FIXME_17 { $$ = 0; }
985 right_element : FIXME_18 { $$ = 0; }
986 slice_size : FIXME_19 { $$ = 0; }
987 lower_element : FIXME_20 { $$ = 0; }
988 upper_element : FIXME_21 { $$ = 0; }
989 first_element : FIXME_22 { $$ = 0; }
990 structure_primitive_value: FIXME_23 { $$ = 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))
1051 } while (isalpha (*tokptr) || isdigit (*tokptr) || (*tokptr == '_'));
1052 yylval.sval.ptr = lexptr;
1053 yylval.sval.length = tokptr - lexptr;
1055 return (copy_name (yylval.sval));
1060 /* Start looking for a value composed of valid digits as set by the base
1061 in use. Note that '_' characters are valid anywhere, in any quantity,
1062 and are simply ignored. Since we must find at least one valid digit,
1063 or reject this token as an integer literal, we keep track of how many
1064 digits we have encountered. */
1067 decode_integer_value (base, tokptrptr, ivalptr)
1072 char *tokptr = *tokptrptr;
1076 while (*tokptr != '\0')
1078 temp = tolower (*tokptr);
1084 case '0': case '1': case '2': case '3': case '4':
1085 case '5': case '6': case '7': case '8': case '9':
1088 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
1104 /* Found something not in domain for current base. */
1105 tokptr--; /* Unconsume what gave us indigestion. */
1110 /* If we didn't find any digits, then we don't have a valid integer
1111 value, so reject the entire token. Otherwise, update the lexical
1112 scan pointer, and return non-zero for success. */
1120 *tokptrptr = tokptr;
1126 decode_integer_literal (valptr, tokptrptr)
1130 char *tokptr = *tokptrptr;
1133 int explicit_base = 0;
1135 /* Look for an explicit base specifier, which is optional. */
1168 /* If we found an explicit base ensure that the character after the
1169 explicit base is a single quote. */
1171 if (explicit_base && (*tokptr++ != '\''))
1176 /* Attempt to decode whatever follows as an integer value in the
1177 indicated base, updating the token pointer in the process and
1178 computing the value into ival. Also, if we have an explicit
1179 base, then the next character must not be a single quote, or we
1180 have a bitstring literal, so reject the entire token in this case.
1181 Otherwise, update the lexical scan pointer, and return non-zero
1184 if (!decode_integer_value (base, &tokptr, &ival))
1188 else if (explicit_base && (*tokptr == '\''))
1195 *tokptrptr = tokptr;
1200 /* If it wasn't for the fact that floating point values can contain '_'
1201 characters, we could just let strtod do all the hard work by letting it
1202 try to consume as much of the current token buffer as possible and
1203 find a legal conversion. Unfortunately we need to filter out the '_'
1204 characters before calling strtod, which we do by copying the other
1205 legal chars to a local buffer to be converted. However since we also
1206 need to keep track of where the last unconsumed character in the input
1207 buffer is, we have transfer only as many characters as may compose a
1208 legal floating point value. */
1211 match_float_literal ()
1213 char *tokptr = lexptr;
1218 extern double strtod ();
1220 /* Make local buffer in which to build the string to convert. This is
1221 required because underscores are valid in chill floating point numbers
1222 but not in the string passed to strtod to convert. The string will be
1223 no longer than our input string. */
1225 copy = buf = (char *) alloca (strlen (tokptr) + 1);
1227 /* Transfer all leading digits to the conversion buffer, discarding any
1230 while (isdigit (*tokptr) || *tokptr == '_')
1239 /* Now accept either a '.', or one of [eEdD]. Dot is legal regardless
1240 of whether we found any leading digits, and we simply accept it and
1241 continue on to look for the fractional part and/or exponent. One of
1242 [eEdD] is legal only if we have seen digits, and means that there
1243 is no fractional part. If we find neither of these, then this is
1244 not a floating point number, so return failure. */
1249 /* Accept and then look for fractional part and/or exponent. */
1262 goto collect_exponent;
1270 /* We found a '.', copy any fractional digits to the conversion buffer, up
1271 to the first nondigit, non-underscore character. */
1273 while (isdigit (*tokptr) || *tokptr == '_')
1282 /* Look for an exponent, which must start with one of [eEdD]. If none
1283 is found, jump directly to trying to convert what we have collected
1300 /* Accept an optional '-' or '+' following one of [eEdD]. */
1303 if (*tokptr == '+' || *tokptr == '-')
1305 *copy++ = *tokptr++;
1308 /* Now copy an exponent into the conversion buffer. Note that at the
1309 moment underscores are *not* allowed in exponents. */
1311 while (isdigit (*tokptr))
1313 *copy++ = *tokptr++;
1316 /* If we transfered any chars to the conversion buffer, try to interpret its
1317 contents as a floating point value. If any characters remain, then we
1318 must not have a valid floating point string. */
1324 dval = strtod (buf, ©);
1329 return (FLOAT_LITERAL);
1335 /* Recognize a string literal. A string literal is a nonzero sequence
1336 of characters enclosed in matching single or double quotes, except that
1337 a single character inside single quotes is a character literal, which
1338 we reject as a string literal. To embed the terminator character inside
1339 a string, it is simply doubled (I.E. "this""is""one""string") */
1342 match_string_literal ()
1344 char *tokptr = lexptr;
1346 for (tempbufindex = 0, tokptr++; *tokptr != '\0'; tokptr++)
1349 if (*tokptr == *lexptr)
1351 if (*(tokptr + 1) == *lexptr)
1360 tempbuf[tempbufindex++] = *tokptr;
1362 if (*tokptr == '\0' /* no terminator */
1363 || tempbufindex == 0 /* no string */
1364 || (tempbufindex == 1 && *tokptr == '\'')) /* char literal */
1370 tempbuf[tempbufindex] = '\0';
1371 yylval.sval.ptr = tempbuf;
1372 yylval.sval.length = tempbufindex;
1374 return (CHARACTER_STRING_LITERAL);
1378 /* Recognize a character literal. A character literal is single character
1379 or a control sequence, enclosed in single quotes. A control sequence
1380 is a comma separated list of one or more integer literals, enclosed
1381 in parenthesis and introduced with a circumflex character.
1383 EX: 'a' '^(7)' '^(7,8)'
1385 As a GNU chill extension, the syntax C'xx' is also recognized as a
1386 character literal, where xx is a hex value for the character.
1388 Note that more than a single character, enclosed in single quotes, is
1391 Also note that the control sequence form is not in GNU Chill since it
1392 is ambiguous with the string literal form using single quotes. I.E.
1393 is '^(7)' a character literal or a string literal. In theory it it
1394 possible to tell by context, but GNU Chill doesn't accept the control
1395 sequence form, so neither do we (for now the code is disabled).
1397 Returns CHARACTER_LITERAL if a match is found.
1401 match_character_literal ()
1403 char *tokptr = lexptr;
1406 if ((tolower (*tokptr) == 'c') && (*(tokptr + 1) == '\''))
1408 /* We have a GNU chill extension form, so skip the leading "C'",
1409 decode the hex value, and then ensure that we have a trailing
1410 single quote character. */
1412 if (!decode_integer_value (16, &tokptr, &ival) || (*tokptr != '\''))
1418 else if (*tokptr == '\'')
1422 /* Determine which form we have, either a control sequence or the
1423 single character form. */
1425 if ((*tokptr == '^') && (*(tokptr + 1) == '('))
1427 #if 0 /* Disable, see note above. -fnf */
1428 /* Match and decode a control sequence. Return zero if we don't
1429 find a valid integer literal, or if the next unconsumed character
1430 after the integer literal is not the trailing ')'.
1431 FIXME: We currently don't handle the multiple integer literal
1434 if (!decode_integer_literal (&ival, &tokptr) || (*tokptr++ != ')'))
1447 /* The trailing quote has not yet been consumed. If we don't find
1448 it, then we have no match. */
1450 if (*tokptr++ != '\'')
1457 /* Not a character literal. */
1460 yylval.typed_val.val = ival;
1461 yylval.typed_val.type = builtin_type_chill_char;
1463 return (CHARACTER_LITERAL);
1466 /* Recognize an integer literal, as specified in Z.200 sec 5.2.4.2.
1467 Note that according to 5.2.4.2, a single "_" is also a valid integer
1468 literal, however GNU-chill requires there to be at least one "digit"
1469 in any integer literal. */
1472 match_integer_literal ()
1474 char *tokptr = lexptr;
1477 if (!decode_integer_literal (&ival, &tokptr))
1483 yylval.typed_val.val = ival;
1484 yylval.typed_val.type = builtin_type_int;
1486 return (INTEGER_LITERAL);
1490 /* Recognize a bit-string literal, as specified in Z.200 sec 5.2.4.8
1491 Note that according to 5.2.4.8, a single "_" is also a valid bit-string
1492 literal, however GNU-chill requires there to be at least one "digit"
1493 in any bit-string literal. */
1496 match_bitstring_literal ()
1498 char *tokptr = lexptr;
1507 /* Look for the required explicit base specifier. */
1528 /* Ensure that the character after the explicit base is a single quote. */
1530 if (*tokptr++ != '\'')
1535 while (*tokptr != '\0' && *tokptr != '\'')
1537 digit = tolower (*tokptr);
1543 case '0': case '1': case '2': case '3': case '4':
1544 case '5': case '6': case '7': case '8': case '9':
1547 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
1557 /* Found something not in domain for current base. */
1562 /* Extract bits from digit, starting with the msbit appropriate for
1563 the current base, and packing them into the bitstring byte,
1564 starting at the lsbit. */
1565 for (mask = (base >> 1); mask > 0; mask >>= 1)
1571 tempbuf[tempbufindex] |= (1 << bitoffset);
1574 if (bitoffset == HOST_CHAR_BIT)
1583 /* Verify that we consumed everything up to the trailing single quote,
1584 and that we found some bits (IE not just underbars). */
1586 if (*tokptr++ != '\'')
1592 yylval.sval.ptr = tempbuf;
1593 yylval.sval.length = bitcount;
1595 return (BIT_STRING_LITERAL);
1599 /* Recognize tokens that start with '$'. These include:
1601 $regname A native register name or a "standard
1603 Return token GDB_REGNAME.
1605 $variable A convenience variable with a name chosen
1607 Return token GDB_VARIABLE.
1609 $digits Value history with index <digits>, starting
1610 from the first value which has index 1.
1613 $$digits Value history with index <digits> relative
1614 to the last value. I.E. $$0 is the last
1615 value, $$1 is the one previous to that, $$2
1616 is the one previous to $$1, etc.
1617 Return token GDB_LAST.
1619 $ | $0 | $$0 The last value in the value history.
1620 Return token GDB_LAST.
1622 $$ An abbreviation for the second to the last
1623 value in the value history, I.E. $$1
1624 Return token GDB_LAST.
1626 Note that we currently assume that register names and convenience
1627 variables follow the convention of starting with a letter or '_'.
1632 match_dollar_tokens ()
1640 /* We will always have a successful match, even if it is just for
1641 a single '$', the abbreviation for $$0. So advance lexptr. */
1645 if (*tokptr == '_' || isalpha (*tokptr))
1647 /* Look for a match with a native register name, usually something
1648 like "r0" for example. */
1650 for (regno = 0; regno < NUM_REGS; regno++)
1652 namelength = strlen (reg_names[regno]);
1653 if (STREQN (tokptr, reg_names[regno], namelength)
1654 && !isalnum (tokptr[namelength]))
1656 yylval.lval = regno;
1657 lexptr += namelength + 1;
1658 return (GDB_REGNAME);
1662 /* Look for a match with a standard register name, usually something
1663 like "pc", which gdb always recognizes as the program counter
1664 regardless of what the native register name is. */
1666 for (regno = 0; regno < num_std_regs; regno++)
1668 namelength = strlen (std_regs[regno].name);
1669 if (STREQN (tokptr, std_regs[regno].name, namelength)
1670 && !isalnum (tokptr[namelength]))
1672 yylval.lval = std_regs[regno].regnum;
1673 lexptr += namelength;
1674 return (GDB_REGNAME);
1678 /* Attempt to match against a convenience variable. Note that
1679 this will always succeed, because if no variable of that name
1680 already exists, the lookup_internalvar will create one for us.
1681 Also note that both lexptr and tokptr currently point to the
1682 start of the input string we are trying to match, and that we
1683 have already tested the first character for non-numeric, so we
1684 don't have to treat it specially. */
1686 while (*tokptr == '_' || isalnum (*tokptr))
1690 yylval.sval.ptr = lexptr;
1691 yylval.sval.length = tokptr - lexptr;
1692 yylval.ivar = lookup_internalvar (copy_name (yylval.sval));
1694 return (GDB_VARIABLE);
1697 /* Since we didn't match against a register name or convenience
1698 variable, our only choice left is a history value. */
1712 /* Attempt to decode more characters as an integer value giving
1713 the index in the history list. If successful, the value will
1714 overwrite ival (currently 0 or 1), and if not, ival will be
1715 left alone, which is good since it is currently correct for
1716 the '$' or '$$' case. */
1718 decode_integer_literal (&ival, &tokptr);
1719 yylval.lval = negate ? -ival : ival;
1730 static const struct token tokentab6[] =
1732 { "LENGTH", LENGTH }
1735 static const struct token tokentab5[] =
1742 static const struct token tokentab4[] =
1751 static const struct token tokentab3[] =
1764 static const struct token tokentab2[] =
1766 { ":=", GDB_ASSIGNMENT },
1767 { "//", SLASH_SLASH },
1776 /* Read one token, getting characters through lexptr. */
1777 /* This is where we will check to make sure that the language and the
1778 operators used are compatible. */
1788 /* Skip over any leading whitespace. */
1789 while (isspace (*lexptr))
1793 /* Look for special single character cases which can't be the first
1794 character of some other multicharacter token. */
1811 /* Look for characters which start a particular kind of multicharacter
1812 token, such as a character literal, register name, convenience
1813 variable name, string literal, etc. */
1818 /* First try to match a string literal, which is any nonzero
1819 sequence of characters enclosed in matching single or double
1820 quotes, except that a single character inside single quotes
1821 is a character literal, so we have to catch that case also. */
1822 token = match_string_literal ();
1827 if (*lexptr == '\'')
1829 token = match_character_literal ();
1838 token = match_character_literal ();
1845 token = match_dollar_tokens ();
1852 /* See if it is a special token of length 6. */
1853 for (i = 0; i < sizeof (tokentab6) / sizeof (tokentab6[0]); i++)
1855 if (STREQN (lexptr, tokentab6[i].operator, 6))
1858 return (tokentab6[i].token);
1861 /* See if it is a special token of length 5. */
1862 for (i = 0; i < sizeof (tokentab5) / sizeof (tokentab5[0]); i++)
1864 if (STREQN (lexptr, tokentab5[i].operator, 5))
1867 return (tokentab5[i].token);
1870 /* See if it is a special token of length 4. */
1871 for (i = 0; i < sizeof (tokentab4) / sizeof (tokentab4[0]); i++)
1873 if (STREQN (lexptr, tokentab4[i].operator, 4))
1876 return (tokentab4[i].token);
1879 /* See if it is a special token of length 3. */
1880 for (i = 0; i < sizeof (tokentab3) / sizeof (tokentab3[0]); i++)
1882 if (STREQN (lexptr, tokentab3[i].operator, 3))
1885 return (tokentab3[i].token);
1888 /* See if it is a special token of length 2. */
1889 for (i = 0; i < sizeof (tokentab2) / sizeof (tokentab2[0]); i++)
1891 if (STREQN (lexptr, tokentab2[i].operator, 2))
1894 return (tokentab2[i].token);
1897 /* Look for single character cases which which could be the first
1898 character of some other multicharacter token, but aren't, or we
1899 would already have found it. */
1909 /* Look for other special tokens. */
1910 if (STREQN (lexptr, "TRUE", 4)) /* FIXME: What about lowercase? */
1914 return (BOOLEAN_LITERAL);
1916 if (STREQN (lexptr, "FALSE", 5)) /* FIXME: What about lowercase? */
1920 return (BOOLEAN_LITERAL);
1922 /* Look for a float literal before looking for an integer literal, so
1923 we match as much of the input stream as possible. */
1924 token = match_float_literal ();
1929 token = match_bitstring_literal ();
1934 token = match_integer_literal ();
1940 /* Try to match a simple name string, and if a match is found, then
1941 further classify what sort of name it is and return an appropriate
1942 token. Note that attempting to match a simple name string consumes
1943 the token from lexptr, so we can't back out if we later find that
1944 we can't classify what sort of name it is. */
1946 simplename = match_simple_name_string ();
1947 if (simplename != NULL)
1949 sym = lookup_symbol (simplename, expression_context_block,
1950 VAR_NAMESPACE, (int *) NULL,
1951 (struct symtab **) NULL);
1954 yylval.ssym.stoken.ptr = NULL;
1955 yylval.ssym.stoken.length = 0;
1956 yylval.ssym.sym = sym;
1957 yylval.ssym.is_a_field_of_this = 0; /* FIXME, C++'ism */
1958 switch (SYMBOL_CLASS (sym))
1961 /* Found a procedure name. */
1962 return (GENERAL_PROCEDURE_NAME);
1964 /* Found a global or local static variable. */
1965 return (LOCATION_NAME);
1972 if (innermost_block == NULL
1973 || contained_in (block_found, innermost_block))
1975 innermost_block = block_found;
1977 return (LOCATION_NAME);
1981 return (LOCATION_NAME);
1984 yylval.tsym.type = SYMBOL_TYPE (sym);
1987 case LOC_CONST_BYTES:
1988 case LOC_OPTIMIZED_OUT:
1989 error ("Symbol \"%s\" names no location.", simplename);
1993 else if (!have_full_symbols () && !have_partial_symbols ())
1995 error ("No symbol table is loaded. Use the \"file\" command.");
1999 error ("No symbol \"%s\" in current context.", simplename);
2003 /* Catch single character tokens which are not part of some
2008 case '.': /* Not float for example. */
2010 while (isspace (*lexptr)) lexptr++;
2011 simplename = match_simple_name_string ();
2017 return (ILLEGAL_TOKEN);
2022 char *msg; /* unused */
2024 printf ("Parsing: %s\n", lexptr);
2027 error ("Invalid syntax in expression near character '%c'.", yychar);
2031 error ("Invalid syntax in expression");