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.
57 #include "expression.h"
60 #include "parser-defs.h"
63 /* Remap normal yacc parser interface names (yyparse, yylex, yyerror, etc),
64 as well as gratuitiously global symbol names, so we can have multiple
65 yacc generated parsers in gdb. Note that these are only the variables
66 produced by yacc. If other parser generators (bison, byacc, etc) produce
67 additional global names that conflict at link time, then those parser
68 generators need to be fixed instead of adding those names to this list. */
70 #define yymaxdepth chill_maxdepth
71 #define yyparse chill_parse
72 #define yylex chill_lex
73 #define yyerror chill_error
74 #define yylval chill_lval
75 #define yychar chill_char
76 #define yydebug chill_debug
77 #define yypact chill_pact
80 #define yydef chill_def
81 #define yychk chill_chk
82 #define yypgo chill_pgo
83 #define yyact chill_act
84 #define yyexca chill_exca
85 #define yyerrflag chill_errflag
86 #define yynerrs chill_nerrs
90 #define yy_yys chill_yys
91 #define yystate chill_state
92 #define yytmp chill_tmp
94 #define yy_yyv chill_yyv
95 #define yyval chill_val
96 #define yylloc chill_lloc
97 #define yyreds chill_reds /* With YYDEBUG defined */
98 #define yytoks chill_toks /* With YYDEBUG defined */
101 #define YYDEBUG 0 /* Default to no yydebug support */
105 yyparse PARAMS ((void));
108 yylex PARAMS ((void));
111 yyerror PARAMS ((char *));
115 /* Although the yacc "value" of an expression is not used,
116 since the result is stored in the structure being created,
117 other node types do have values. */
122 unsigned LONGEST ulval;
132 struct symtoken ssym;
135 enum exp_opcode opcode;
136 struct internalvar *ivar;
142 %token <voidval> FIXME_01
143 %token <voidval> FIXME_02
144 %token <voidval> FIXME_03
145 %token <voidval> FIXME_04
146 %token <voidval> FIXME_05
147 %token <voidval> FIXME_06
148 %token <voidval> FIXME_07
149 %token <voidval> FIXME_08
150 %token <voidval> FIXME_09
151 %token <voidval> FIXME_10
152 %token <voidval> FIXME_11
153 %token <voidval> FIXME_12
154 %token <voidval> FIXME_13
155 %token <voidval> FIXME_14
156 %token <voidval> FIXME_15
157 %token <voidval> FIXME_16
158 %token <voidval> FIXME_17
159 %token <voidval> FIXME_18
160 %token <voidval> FIXME_19
161 %token <voidval> FIXME_20
162 %token <voidval> FIXME_21
163 %token <voidval> FIXME_22
164 %token <voidval> FIXME_23
165 %token <voidval> FIXME_24
166 %token <voidval> FIXME_25
167 %token <voidval> FIXME_26
168 %token <voidval> FIXME_27
169 %token <voidval> FIXME_28
170 %token <voidval> FIXME_29
171 %token <voidval> FIXME_30
173 %token <typed_val> INTEGER_LITERAL
174 %token <ulval> BOOLEAN_LITERAL
175 %token <typed_val> CHARACTER_LITERAL
176 %token <dval> FLOAT_LITERAL
177 %token <ssym> GENERAL_PROCEDURE_NAME
178 %token <ssym> LOCATION_NAME
179 %token <voidval> SET_LITERAL
180 %token <voidval> EMPTINESS_LITERAL
181 %token <sval> CHARACTER_STRING_LITERAL
182 %token <sval> BIT_STRING_LITERAL
187 %token <voidval> CASE
189 %token <voidval> ESAC
190 %token <voidval> LOGIOR
191 %token <voidval> ORIF
192 %token <voidval> LOGXOR
193 %token <voidval> LOGAND
194 %token <voidval> ANDIF
196 %token <voidval> NOTEQUAL
206 %token <voidval> SLASH_SLASH
210 %token <voidval> POINTER
211 %token <voidval> RECEIVE
218 %token <voidval> THEN
219 %token <voidval> ELSE
221 %token <voidval> ELSIF
222 %token <voidval> ILLEGAL_TOKEN
224 %token <voidval> PRED
225 %token <voidval> SUCC
227 %token <voidval> CARD
230 %token <voidval> SIZE
231 %token <voidval> UPPER
232 %token <voidval> LOWER
233 %token <voidval> LENGTH
235 /* Tokens which are not Chill tokens used in expressions, but rather GDB
236 specific things that we recognize in the same context as Chill tokens
237 (register names for example). */
239 %token <lval> GDB_REGNAME /* Machine register name */
240 %token <lval> GDB_LAST /* Value history */
241 %token <ivar> GDB_VARIABLE /* Convenience variable */
242 %token <voidval> GDB_ASSIGNMENT /* Assign value to somewhere */
244 %type <voidval> location
245 %type <voidval> access_name
246 %type <voidval> primitive_value
247 %type <voidval> location_contents
248 %type <voidval> value_name
249 %type <voidval> literal
250 %type <voidval> tuple
251 %type <voidval> value_string_element
252 %type <voidval> value_string_slice
253 %type <voidval> value_array_element
254 %type <voidval> value_array_slice
255 %type <voidval> value_structure_field
256 %type <voidval> expression_conversion
257 %type <voidval> value_procedure_call
258 %type <voidval> value_built_in_routine_call
259 %type <voidval> chill_value_built_in_routine_call
260 %type <voidval> start_expression
261 %type <voidval> zero_adic_operator
262 %type <voidval> parenthesised_expression
263 %type <voidval> value
264 %type <voidval> undefined_value
265 %type <voidval> expression
266 %type <voidval> conditional_expression
267 %type <voidval> then_alternative
268 %type <voidval> else_alternative
269 %type <voidval> sub_expression
270 %type <voidval> value_case_alternative
271 %type <voidval> operand_0
272 %type <voidval> operand_1
273 %type <voidval> operand_2
274 %type <voidval> operand_3
275 %type <voidval> operand_4
276 %type <voidval> operand_5
277 %type <voidval> operand_6
278 %type <voidval> synonym_name
279 %type <voidval> value_enumeration_name
280 %type <voidval> value_do_with_name
281 %type <voidval> value_receive_name
282 %type <voidval> string_primitive_value
283 %type <voidval> start_element
284 %type <voidval> left_element
285 %type <voidval> right_element
286 %type <voidval> slice_size
287 %type <voidval> array_primitive_value
288 %type <voidval> expression_list
289 %type <voidval> lower_element
290 %type <voidval> upper_element
291 %type <voidval> first_element
292 %type <voidval> structure_primitive_value
293 %type <voidval> field_name
294 %type <voidval> mode_argument
295 %type <voidval> upper_lower_argument
296 %type <voidval> length_argument
297 %type <voidval> mode_name
298 %type <voidval> array_mode_name
299 %type <voidval> string_mode_name
300 %type <voidval> variant_structure_mode_name
301 %type <voidval> boolean_expression
302 %type <voidval> case_selector_list
303 %type <voidval> subexpression
304 %type <voidval> case_label_specification
305 %type <voidval> buffer_location
307 %type <voidval> single_assignment_action
323 undefined_value : FIXME_01
331 location : access_name
343 access_name : LOCATION_NAME
345 write_exp_elt_opcode (OP_VAR_VALUE);
346 write_exp_elt_sym ($1.sym);
347 write_exp_elt_opcode (OP_VAR_VALUE);
349 | GDB_LAST /* gdb specific */
351 write_exp_elt_opcode (OP_LAST);
352 write_exp_elt_longcst ($1);
353 write_exp_elt_opcode (OP_LAST);
355 | GDB_REGNAME /* gdb specific */
357 write_exp_elt_opcode (OP_REGISTER);
358 write_exp_elt_longcst ($1);
359 write_exp_elt_opcode (OP_REGISTER);
361 | GDB_VARIABLE /* gdb specific */
363 write_exp_elt_opcode (OP_INTERNALVAR);
364 write_exp_elt_intern ($1);
365 write_exp_elt_opcode (OP_INTERNALVAR);
375 expression_list : expression
379 | expression_list ',' expression
386 primitive_value : location_contents
402 | value_string_element
410 | value_array_element
418 | value_structure_field
422 | expression_conversion
426 | value_procedure_call
430 | value_built_in_routine_call
442 | parenthesised_expression
450 location_contents: location
458 value_name : synonym_name
462 | value_enumeration_name
474 | GENERAL_PROCEDURE_NAME
476 write_exp_elt_opcode (OP_VAR_VALUE);
477 write_exp_elt_sym ($1.sym);
478 write_exp_elt_opcode (OP_VAR_VALUE);
484 literal : INTEGER_LITERAL
486 write_exp_elt_opcode (OP_LONG);
487 write_exp_elt_type ($1.type);
488 write_exp_elt_longcst ((LONGEST) ($1.val));
489 write_exp_elt_opcode (OP_LONG);
493 write_exp_elt_opcode (OP_BOOL);
494 write_exp_elt_longcst ((LONGEST) $1);
495 write_exp_elt_opcode (OP_BOOL);
499 write_exp_elt_opcode (OP_LONG);
500 write_exp_elt_type ($1.type);
501 write_exp_elt_longcst ((LONGEST) ($1.val));
502 write_exp_elt_opcode (OP_LONG);
506 write_exp_elt_opcode (OP_DOUBLE);
507 write_exp_elt_type (builtin_type_double);
508 write_exp_elt_dblcst ($1);
509 write_exp_elt_opcode (OP_DOUBLE);
519 | CHARACTER_STRING_LITERAL
521 write_exp_elt_opcode (OP_STRING);
522 write_exp_string ($1);
523 write_exp_elt_opcode (OP_STRING);
527 write_exp_elt_opcode (OP_BITSTRING);
528 write_exp_bitstring ($1);
529 write_exp_elt_opcode (OP_BITSTRING);
544 value_string_element: string_primitive_value '(' start_element ')'
552 value_string_slice: string_primitive_value '(' left_element ':' right_element ')'
556 | string_primitive_value '(' start_element UP slice_size ')'
564 value_array_element: array_primitive_value '('
565 /* This is to save the value of arglist_len
566 being accumulated for each dimension. */
567 { start_arglist (); }
570 write_exp_elt_opcode (MULTI_SUBSCRIPT);
571 write_exp_elt_longcst ((LONGEST) end_arglist ());
572 write_exp_elt_opcode (MULTI_SUBSCRIPT);
578 value_array_slice: array_primitive_value '(' lower_element ':' upper_element ')'
582 | array_primitive_value '(' first_element UP slice_size ')'
590 value_structure_field: structure_primitive_value '.' field_name
598 expression_conversion: mode_name parenthesised_expression
606 value_procedure_call: FIXME_05
614 value_built_in_routine_call: chill_value_built_in_routine_call
622 start_expression: FIXME_06
625 } /* Not in GNU-Chill */
630 zero_adic_operator: FIXME_07
638 parenthesised_expression: '(' expression ')'
646 expression : operand_0
650 | conditional_expression
656 conditional_expression : IF boolean_expression then_alternative else_alternative FI
660 | CASE case_selector_list OF value_case_alternative '[' ELSE sub_expression ']' ESAC
666 then_alternative: THEN subexpression
672 else_alternative: ELSE subexpression
676 | ELSIF boolean_expression then_alternative else_alternative
682 sub_expression : expression
688 value_case_alternative: case_label_specification ':' sub_expression ';'
696 operand_0 : operand_1
700 | operand_0 LOGIOR operand_1
702 write_exp_elt_opcode (BINOP_BITWISE_IOR);
704 | operand_0 ORIF operand_1
708 | operand_0 LOGXOR operand_1
710 write_exp_elt_opcode (BINOP_BITWISE_XOR);
712 | single_assignment_action
720 operand_1 : operand_2
724 | operand_1 LOGAND operand_2
726 write_exp_elt_opcode (BINOP_BITWISE_AND);
728 | operand_1 ANDIF operand_2
736 operand_2 : operand_3
740 | operand_2 '=' operand_3
742 write_exp_elt_opcode (BINOP_EQUAL);
744 | operand_2 NOTEQUAL operand_3
746 write_exp_elt_opcode (BINOP_NOTEQUAL);
748 | operand_2 '>' operand_3
750 write_exp_elt_opcode (BINOP_GTR);
752 | operand_2 GTR operand_3
754 write_exp_elt_opcode (BINOP_GEQ);
756 | operand_2 '<' operand_3
758 write_exp_elt_opcode (BINOP_LESS);
760 | operand_2 LEQ operand_3
762 write_exp_elt_opcode (BINOP_LEQ);
764 | operand_2 IN operand_3
773 operand_3 : operand_4
777 | operand_3 '+' operand_4
779 write_exp_elt_opcode (BINOP_ADD);
781 | operand_3 '-' operand_4
783 write_exp_elt_opcode (BINOP_SUB);
785 | operand_3 SLASH_SLASH operand_4
787 write_exp_elt_opcode (BINOP_CONCAT);
793 operand_4 : operand_5
797 | operand_4 '*' operand_5
799 write_exp_elt_opcode (BINOP_MUL);
801 | operand_4 '/' operand_5
803 write_exp_elt_opcode (BINOP_DIV);
805 | operand_4 MOD operand_5
807 write_exp_elt_opcode (BINOP_MOD);
809 | operand_4 REM operand_5
811 write_exp_elt_opcode (BINOP_REM);
816 /* Note that we accept any expression for BINOP_CONCAT, not just
817 integer literal expressions. (FIXME?) */
819 operand_5 : operand_6
825 write_exp_elt_opcode (UNOP_NEG);
829 write_exp_elt_opcode (UNOP_LOGICAL_NOT);
831 | parenthesised_expression operand_6
833 write_exp_elt_opcode (BINOP_CONCAT);
839 operand_6 : POINTER location
843 | RECEIVE buffer_location
856 single_assignment_action :
857 location GDB_ASSIGNMENT value
859 write_exp_elt_opcode (BINOP_ASSIGN);
865 chill_value_built_in_routine_call :
866 NUM '(' expression ')'
870 | PRED '(' expression ')'
874 | SUCC '(' expression ')'
878 | ABS '(' expression ')'
882 | CARD '(' expression ')'
886 | MAX '(' expression ')'
890 | MIN '(' expression ')'
894 | SIZE '(' location ')'
898 | SIZE '(' mode_argument ')'
902 | UPPER '(' upper_lower_argument ')'
906 | LOWER '(' upper_lower_argument ')'
910 | LENGTH '(' length_argument ')'
916 mode_argument : mode_name
920 | array_mode_name '(' expression ')'
924 | string_mode_name '(' expression ')'
928 | variant_structure_mode_name '(' expression_list ')'
934 upper_lower_argument : location
948 length_argument : location
960 array_primitive_value : primitive_value
967 /* Things which still need productions... */
969 array_mode_name : FIXME_08 { $$ = 0; }
970 string_mode_name : FIXME_09 { $$ = 0; }
971 variant_structure_mode_name: FIXME_10 { $$ = 0; }
972 synonym_name : FIXME_11 { $$ = 0; }
973 value_enumeration_name : FIXME_12 { $$ = 0; }
974 value_do_with_name : FIXME_13 { $$ = 0; }
975 value_receive_name : FIXME_14 { $$ = 0; }
976 string_primitive_value : FIXME_15 { $$ = 0; }
977 start_element : FIXME_16 { $$ = 0; }
978 left_element : FIXME_17 { $$ = 0; }
979 right_element : FIXME_18 { $$ = 0; }
980 slice_size : FIXME_19 { $$ = 0; }
981 lower_element : FIXME_20 { $$ = 0; }
982 upper_element : FIXME_21 { $$ = 0; }
983 first_element : FIXME_22 { $$ = 0; }
984 structure_primitive_value: FIXME_23 { $$ = 0; }
985 field_name : FIXME_24 { $$ = 0; }
986 mode_name : FIXME_25 { $$ = 0; }
987 boolean_expression : FIXME_26 { $$ = 0; }
988 case_selector_list : FIXME_27 { $$ = 0; }
989 subexpression : FIXME_28 { $$ = 0; }
990 case_label_specification: FIXME_29 { $$ = 0; }
991 buffer_location : FIXME_30 { $$ = 0; }
995 /* Implementation of a dynamically expandable buffer for processing input
996 characters acquired through lexptr and building a value to return in
999 static char *tempbuf; /* Current buffer contents */
1000 static int tempbufsize; /* Size of allocated buffer */
1001 static int tempbufindex; /* Current index into buffer */
1003 #define GROWBY_MIN_SIZE 64 /* Minimum amount to grow buffer by */
1005 #define CHECKBUF(size) \
1007 if (tempbufindex + (size) >= tempbufsize) \
1009 growbuf_by_size (size); \
1013 /* Grow the static temp buffer if necessary, including allocating the first one
1017 growbuf_by_size (count)
1022 growby = max (count, GROWBY_MIN_SIZE);
1023 tempbufsize += growby;
1024 if (tempbuf == NULL)
1026 tempbuf = (char *) malloc (tempbufsize);
1030 tempbuf = (char *) realloc (tempbuf, tempbufsize);
1034 /* Try to consume a simple name string token. If successful, returns
1035 a pointer to a nullbyte terminated copy of the name that can be used
1036 in symbol table lookups. If not successful, returns NULL. */
1039 match_simple_name_string ()
1041 char *tokptr = lexptr;
1043 if (isalpha (*tokptr))
1047 } while (isalpha (*tokptr) || isdigit (*tokptr) || (*tokptr == '_'));
1048 yylval.sval.ptr = lexptr;
1049 yylval.sval.length = tokptr - lexptr;
1051 return (copy_name (yylval.sval));
1056 /* Start looking for a value composed of valid digits as set by the base
1057 in use. Note that '_' characters are valid anywhere, in any quantity,
1058 and are simply ignored. Since we must find at least one valid digit,
1059 or reject this token as an integer literal, we keep track of how many
1060 digits we have encountered. */
1063 decode_integer_value (base, tokptrptr, ivalptr)
1068 char *tokptr = *tokptrptr;
1072 while (*tokptr != '\0')
1074 temp = tolower (*tokptr);
1080 case '0': case '1': case '2': case '3': case '4':
1081 case '5': case '6': case '7': case '8': case '9':
1084 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
1100 /* Found something not in domain for current base. */
1101 tokptr--; /* Unconsume what gave us indigestion. */
1106 /* If we didn't find any digits, then we don't have a valid integer
1107 value, so reject the entire token. Otherwise, update the lexical
1108 scan pointer, and return non-zero for success. */
1116 *tokptrptr = tokptr;
1122 decode_integer_literal (valptr, tokptrptr)
1126 char *tokptr = *tokptrptr;
1129 int explicit_base = 0;
1131 /* Look for an explicit base specifier, which is optional. */
1164 /* If we found an explicit base ensure that the character after the
1165 explicit base is a single quote. */
1167 if (explicit_base && (*tokptr++ != '\''))
1172 /* Attempt to decode whatever follows as an integer value in the
1173 indicated base, updating the token pointer in the process and
1174 computing the value into ival. Also, if we have an explicit
1175 base, then the next character must not be a single quote, or we
1176 have a bitstring literal, so reject the entire token in this case.
1177 Otherwise, update the lexical scan pointer, and return non-zero
1180 if (!decode_integer_value (base, &tokptr, &ival))
1184 else if (explicit_base && (*tokptr == '\''))
1191 *tokptrptr = tokptr;
1196 /* If it wasn't for the fact that floating point values can contain '_'
1197 characters, we could just let strtod do all the hard work by letting it
1198 try to consume as much of the current token buffer as possible and
1199 find a legal conversion. Unfortunately we need to filter out the '_'
1200 characters before calling strtod, which we do by copying the other
1201 legal chars to a local buffer to be converted. However since we also
1202 need to keep track of where the last unconsumed character in the input
1203 buffer is, we have transfer only as many characters as may compose a
1204 legal floating point value. */
1207 match_float_literal ()
1209 char *tokptr = lexptr;
1214 extern double strtod ();
1216 /* Make local buffer in which to build the string to convert. This is
1217 required because underscores are valid in chill floating point numbers
1218 but not in the string passed to strtod to convert. The string will be
1219 no longer than our input string. */
1221 copy = buf = (char *) alloca (strlen (tokptr) + 1);
1223 /* Transfer all leading digits to the conversion buffer, discarding any
1226 while (isdigit (*tokptr) || *tokptr == '_')
1235 /* Now accept either a '.', or one of [eEdD]. Dot is legal regardless
1236 of whether we found any leading digits, and we simply accept it and
1237 continue on to look for the fractional part and/or exponent. One of
1238 [eEdD] is legal only if we have seen digits, and means that there
1239 is no fractional part. If we find neither of these, then this is
1240 not a floating point number, so return failure. */
1245 /* Accept and then look for fractional part and/or exponent. */
1258 goto collect_exponent;
1266 /* We found a '.', copy any fractional digits to the conversion buffer, up
1267 to the first nondigit, non-underscore character. */
1269 while (isdigit (*tokptr) || *tokptr == '_')
1278 /* Look for an exponent, which must start with one of [eEdD]. If none
1279 is found, jump directly to trying to convert what we have collected
1296 /* Accept an optional '-' or '+' following one of [eEdD]. */
1299 if (*tokptr == '+' || *tokptr == '-')
1301 *copy++ = *tokptr++;
1304 /* Now copy an exponent into the conversion buffer. Note that at the
1305 moment underscores are *not* allowed in exponents. */
1307 while (isdigit (*tokptr))
1309 *copy++ = *tokptr++;
1312 /* If we transfered any chars to the conversion buffer, try to interpret its
1313 contents as a floating point value. If any characters remain, then we
1314 must not have a valid floating point string. */
1320 dval = strtod (buf, ©);
1325 return (FLOAT_LITERAL);
1331 /* Recognize a string literal. A string literal is a nonzero sequence
1332 of characters enclosed in matching single or double quotes, except that
1333 a single character inside single quotes is a character literal, which
1334 we reject as a string literal. To embed the terminator character inside
1335 a string, it is simply doubled (I.E. "this""is""one""string") */
1338 match_string_literal ()
1340 char *tokptr = lexptr;
1342 for (tempbufindex = 0, tokptr++; *tokptr != '\0'; tokptr++)
1345 if (*tokptr == *lexptr)
1347 if (*(tokptr + 1) == *lexptr)
1356 tempbuf[tempbufindex++] = *tokptr;
1358 if (*tokptr == '\0' /* no terminator */
1359 || tempbufindex == 0 /* no string */
1360 || (tempbufindex == 1 && *tokptr == '\'')) /* char literal */
1366 tempbuf[tempbufindex] = '\0';
1367 yylval.sval.ptr = tempbuf;
1368 yylval.sval.length = tempbufindex;
1370 return (CHARACTER_STRING_LITERAL);
1374 /* Recognize a character literal. A character literal is single character
1375 or a control sequence, enclosed in single quotes. A control sequence
1376 is a comma separated list of one or more integer literals, enclosed
1377 in parenthesis and introduced with a circumflex character.
1379 EX: 'a' '^(7)' '^(7,8)'
1381 As a GNU chill extension, the syntax C'xx' is also recognized as a
1382 character literal, where xx is a hex value for the character.
1384 Note that more than a single character, enclosed in single quotes, is
1387 Also note that the control sequence form is not in GNU Chill since it
1388 is ambiguous with the string literal form using single quotes. I.E.
1389 is '^(7)' a character literal or a string literal. In theory it it
1390 possible to tell by context, but GNU Chill doesn't accept the control
1391 sequence form, so neither do we (for now the code is disabled).
1393 Returns CHARACTER_LITERAL if a match is found.
1397 match_character_literal ()
1399 char *tokptr = lexptr;
1402 if ((tolower (*tokptr) == 'c') && (*(tokptr + 1) == '\''))
1404 /* We have a GNU chill extension form, so skip the leading "C'",
1405 decode the hex value, and then ensure that we have a trailing
1406 single quote character. */
1408 if (!decode_integer_value (16, &tokptr, &ival) || (*tokptr != '\''))
1414 else if (*tokptr == '\'')
1418 /* Determine which form we have, either a control sequence or the
1419 single character form. */
1421 if ((*tokptr == '^') && (*(tokptr + 1) == '('))
1423 #if 0 /* Disable, see note above. -fnf */
1424 /* Match and decode a control sequence. Return zero if we don't
1425 find a valid integer literal, or if the next unconsumed character
1426 after the integer literal is not the trailing ')'.
1427 FIXME: We currently don't handle the multiple integer literal
1430 if (!decode_integer_literal (&ival, &tokptr) || (*tokptr++ != ')'))
1443 /* The trailing quote has not yet been consumed. If we don't find
1444 it, then we have no match. */
1446 if (*tokptr++ != '\'')
1453 /* Not a character literal. */
1456 yylval.typed_val.val = ival;
1457 yylval.typed_val.type = builtin_type_chill_char;
1459 return (CHARACTER_LITERAL);
1462 /* Recognize an integer literal, as specified in Z.200 sec 5.2.4.2.
1463 Note that according to 5.2.4.2, a single "_" is also a valid integer
1464 literal, however GNU-chill requires there to be at least one "digit"
1465 in any integer literal. */
1468 match_integer_literal ()
1470 char *tokptr = lexptr;
1473 if (!decode_integer_literal (&ival, &tokptr))
1479 yylval.typed_val.val = ival;
1480 yylval.typed_val.type = builtin_type_int;
1482 return (INTEGER_LITERAL);
1486 /* Recognize a bit-string literal, as specified in Z.200 sec 5.2.4.8
1487 Note that according to 5.2.4.8, a single "_" is also a valid bit-string
1488 literal, however GNU-chill requires there to be at least one "digit"
1489 in any bit-string literal. */
1492 match_bitstring_literal ()
1494 char *tokptr = lexptr;
1503 /* Look for the required explicit base specifier. */
1524 /* Ensure that the character after the explicit base is a single quote. */
1526 if (*tokptr++ != '\'')
1531 while (*tokptr != '\0' && *tokptr != '\'')
1533 digit = tolower (*tokptr);
1539 case '0': case '1': case '2': case '3': case '4':
1540 case '5': case '6': case '7': case '8': case '9':
1543 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
1553 /* Found something not in domain for current base. */
1558 /* Extract bits from digit, starting with the msbit appropriate for
1559 the current base, and packing them into the bitstring byte,
1560 starting at the lsbit. */
1561 for (mask = (base >> 1); mask > 0; mask >>= 1)
1567 tempbuf[tempbufindex] |= (1 << bitoffset);
1570 if (bitoffset == HOST_CHAR_BIT)
1579 /* Verify that we consumed everything up to the trailing single quote,
1580 and that we found some bits (IE not just underbars). */
1582 if (*tokptr++ != '\'')
1588 yylval.sval.ptr = tempbuf;
1589 yylval.sval.length = bitcount;
1591 return (BIT_STRING_LITERAL);
1595 /* Recognize tokens that start with '$'. These include:
1597 $regname A native register name or a "standard
1599 Return token GDB_REGNAME.
1601 $variable A convenience variable with a name chosen
1603 Return token GDB_VARIABLE.
1605 $digits Value history with index <digits>, starting
1606 from the first value which has index 1.
1609 $$digits Value history with index <digits> relative
1610 to the last value. I.E. $$0 is the last
1611 value, $$1 is the one previous to that, $$2
1612 is the one previous to $$1, etc.
1613 Return token GDB_LAST.
1615 $ | $0 | $$0 The last value in the value history.
1616 Return token GDB_LAST.
1618 $$ An abbreviation for the second to the last
1619 value in the value history, I.E. $$1
1620 Return token GDB_LAST.
1622 Note that we currently assume that register names and convenience
1623 variables follow the convention of starting with a letter or '_'.
1628 match_dollar_tokens ()
1636 /* We will always have a successful match, even if it is just for
1637 a single '$', the abbreviation for $$0. So advance lexptr. */
1641 if (*tokptr == '_' || isalpha (*tokptr))
1643 /* Look for a match with a native register name, usually something
1644 like "r0" for example. */
1646 for (regno = 0; regno < NUM_REGS; regno++)
1648 namelength = strlen (reg_names[regno]);
1649 if (STREQN (tokptr, reg_names[regno], namelength)
1650 && !isalnum (tokptr[namelength]))
1652 yylval.lval = regno;
1653 lexptr += namelength + 1;
1654 return (GDB_REGNAME);
1658 /* Look for a match with a standard register name, usually something
1659 like "pc", which gdb always recognizes as the program counter
1660 regardless of what the native register name is. */
1662 for (regno = 0; regno < num_std_regs; regno++)
1664 namelength = strlen (std_regs[regno].name);
1665 if (STREQN (tokptr, std_regs[regno].name, namelength)
1666 && !isalnum (tokptr[namelength]))
1668 yylval.lval = std_regs[regno].regnum;
1669 lexptr += namelength;
1670 return (GDB_REGNAME);
1674 /* Attempt to match against a convenience variable. Note that
1675 this will always succeed, because if no variable of that name
1676 already exists, the lookup_internalvar will create one for us.
1677 Also note that both lexptr and tokptr currently point to the
1678 start of the input string we are trying to match, and that we
1679 have already tested the first character for non-numeric, so we
1680 don't have to treat it specially. */
1682 while (*tokptr == '_' || isalnum (*tokptr))
1686 yylval.sval.ptr = lexptr;
1687 yylval.sval.length = tokptr - lexptr;
1688 yylval.ivar = lookup_internalvar (copy_name (yylval.sval));
1690 return (GDB_VARIABLE);
1693 /* Since we didn't match against a register name or convenience
1694 variable, our only choice left is a history value. */
1708 /* Attempt to decode more characters as an integer value giving
1709 the index in the history list. If successful, the value will
1710 overwrite ival (currently 0 or 1), and if not, ival will be
1711 left alone, which is good since it is currently correct for
1712 the '$' or '$$' case. */
1714 decode_integer_literal (&ival, &tokptr);
1715 yylval.lval = negate ? -ival : ival;
1726 static const struct token tokentab6[] =
1728 { "LENGTH", LENGTH }
1731 static const struct token tokentab5[] =
1738 static const struct token tokentab4[] =
1747 static const struct token tokentab3[] =
1760 static const struct token tokentab2[] =
1762 { ":=", GDB_ASSIGNMENT },
1763 { "//", SLASH_SLASH },
1771 /* Read one token, getting characters through lexptr. */
1772 /* This is where we will check to make sure that the language and the
1773 operators used are compatible. */
1783 /* Skip over any leading whitespace. */
1784 while (isspace (*lexptr))
1788 /* Look for special single character cases which can't be the first
1789 character of some other multicharacter token. */
1807 /* Look for characters which start a particular kind of multicharacter
1808 token, such as a character literal, register name, convenience
1809 variable name, string literal, etc. */
1814 /* First try to match a string literal, which is any nonzero
1815 sequence of characters enclosed in matching single or double
1816 quotes, except that a single character inside single quotes
1817 is a character literal, so we have to catch that case also. */
1818 token = match_string_literal ();
1823 if (*lexptr == '\'')
1825 token = match_character_literal ();
1834 token = match_character_literal ();
1841 token = match_dollar_tokens ();
1848 /* See if it is a special token of length 6. */
1849 for (i = 0; i < sizeof (tokentab6) / sizeof (tokentab6[0]); i++)
1851 if (STREQN (lexptr, tokentab6[i].operator, 6))
1854 return (tokentab6[i].token);
1857 /* See if it is a special token of length 5. */
1858 for (i = 0; i < sizeof (tokentab5) / sizeof (tokentab5[0]); i++)
1860 if (STREQN (lexptr, tokentab5[i].operator, 5))
1863 return (tokentab5[i].token);
1866 /* See if it is a special token of length 4. */
1867 for (i = 0; i < sizeof (tokentab4) / sizeof (tokentab4[0]); i++)
1869 if (STREQN (lexptr, tokentab4[i].operator, 4))
1872 return (tokentab4[i].token);
1875 /* See if it is a special token of length 3. */
1876 for (i = 0; i < sizeof (tokentab3) / sizeof (tokentab3[0]); i++)
1878 if (STREQN (lexptr, tokentab3[i].operator, 3))
1881 return (tokentab3[i].token);
1884 /* See if it is a special token of length 2. */
1885 for (i = 0; i < sizeof (tokentab2) / sizeof (tokentab2[0]); i++)
1887 if (STREQN (lexptr, tokentab2[i].operator, 2))
1890 return (tokentab2[i].token);
1893 /* Look for single character cases which which could be the first
1894 character of some other multicharacter token, but aren't, or we
1895 would already have found it. */
1904 /* Look for other special tokens. */
1905 if (STREQN (lexptr, "TRUE", 4)) /* FIXME: What about lowercase? */
1909 return (BOOLEAN_LITERAL);
1911 if (STREQN (lexptr, "FALSE", 5)) /* FIXME: What about lowercase? */
1915 return (BOOLEAN_LITERAL);
1917 /* Look for a float literal before looking for an integer literal, so
1918 we match as much of the input stream as possible. */
1919 token = match_float_literal ();
1924 token = match_bitstring_literal ();
1929 token = match_integer_literal ();
1935 /* Try to match a simple name string, and if a match is found, then
1936 further classify what sort of name it is and return an appropriate
1937 token. Note that attempting to match a simple name string consumes
1938 the token from lexptr, so we can't back out if we later find that
1939 we can't classify what sort of name it is. */
1941 simplename = match_simple_name_string ();
1942 if (simplename != NULL)
1944 sym = lookup_symbol (simplename, expression_context_block,
1945 VAR_NAMESPACE, (int *) NULL,
1946 (struct symtab **) NULL);
1949 yylval.ssym.stoken.ptr = NULL;
1950 yylval.ssym.stoken.length = 0;
1951 yylval.ssym.sym = sym;
1952 yylval.ssym.is_a_field_of_this = 0; /* FIXME, C++'ism */
1953 switch (SYMBOL_CLASS (sym))
1956 /* Found a procedure name. */
1957 return (GENERAL_PROCEDURE_NAME);
1959 /* Found a global or local static variable. */
1960 return (LOCATION_NAME);
1967 if (innermost_block == NULL
1968 || contained_in (block_found, innermost_block))
1970 innermost_block = block_found;
1972 return (LOCATION_NAME);
1976 return (LOCATION_NAME);
1980 case LOC_CONST_BYTES:
1981 error ("Symbol \"%s\" names no location.", simplename);
1985 else if (!have_full_symbols () && !have_partial_symbols ())
1987 error ("No symbol table is loaded. Use the \"file\" command.");
1991 error ("No symbol \"%s\" in current context.", simplename);
1995 /* Catch single character tokens which are not part of some
2000 case '.': /* Not float for example. */
2004 return (ILLEGAL_TOKEN);
2009 char *msg; /* unused */
2011 printf ("Parsing: %s\n", lexptr);
2014 error ("Invalid syntax in expression near character '%c'.", yychar);
2018 error ("Invalid syntax in expression");