1 /* YACC grammar for Chill expressions, for GDB.
2 Copyright 1992, 1993, 1994 Free Software Foundation, Inc.
4 This file is part of GDB.
6 This program is free software; you can redistribute it and/or modify
7 it under the terms of the GNU General Public License as published by
8 the Free Software Foundation; either version 2 of the License, or
9 (at your option) any later version.
11 This program is distributed in the hope that it will be useful,
12 but WITHOUT ANY WARRANTY; without even the implied warranty of
13 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
14 GNU General Public License for more details.
16 You should have received a copy of the GNU General Public License
17 along with this program; if not, write to the Free Software
18 Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. */
20 /* Parse a Chill expression from text in a string,
21 and return the result as a struct expression pointer.
22 That structure contains arithmetic operations in reverse polish,
23 with constants represented by operations that are followed by special data.
24 See expression.h for the details of the format.
25 What is important here is that it can be built up sequentially
26 during the process of parsing; the lower levels of the tree always
27 come first in the result.
29 Note that malloc's and realloc's in this file are transformed to
30 xmalloc and xrealloc respectively by the same sed command in the
31 makefile that remaps any other malloc/realloc inserted by the parser
32 generator. Doing this with #defines and trying to control the interaction
33 with include files (<malloc.h> and <stdlib.h> for example) just became
34 too messy, particularly when such includes can be inserted at random
35 times by the parser generator.
37 Also note that the language accepted by this parser is more liberal
38 than the one accepted by an actual Chill compiler. For example, the
39 language rule that a simple name string can not be one of the reserved
40 simple name strings is not enforced (e.g "case" is not treated as a
41 reserved name). Another example is that Chill is a strongly typed
42 language, and certain expressions that violate the type constraints
43 may still be evaluated if gdb can do so in a meaningful manner, while
44 such expressions would be rejected by the compiler. The reason for
45 this more liberal behavior is the philosophy that the debugger
46 is intended to be a tool that is used by the programmer when things
47 go wrong, and as such, it should provide as few artificial barriers
48 to it's use as possible. If it can do something meaningful, even
49 something that violates language contraints that are enforced by the
50 compiler, it should do so without complaint.
59 #include "expression.h"
62 #include "parser-defs.h"
64 #include "bfd.h" /* Required by objfiles.h. */
65 #include "symfile.h" /* Required by objfiles.h. */
66 #include "objfiles.h" /* For have_full_symbols and have_partial_symbols */
68 /* Remap normal yacc parser interface names (yyparse, yylex, yyerror, etc),
69 as well as gratuitiously global symbol names, so we can have multiple
70 yacc generated parsers in gdb. Note that these are only the variables
71 produced by yacc. If other parser generators (bison, byacc, etc) produce
72 additional global names that conflict at link time, then those parser
73 generators need to be fixed instead of adding those names to this list. */
75 #define yymaxdepth chill_maxdepth
76 #define yyparse chill_parse
77 #define yylex chill_lex
78 #define yyerror chill_error
79 #define yylval chill_lval
80 #define yychar chill_char
81 #define yydebug chill_debug
82 #define yypact chill_pact
85 #define yydef chill_def
86 #define yychk chill_chk
87 #define yypgo chill_pgo
88 #define yyact chill_act
89 #define yyexca chill_exca
90 #define yyerrflag chill_errflag
91 #define yynerrs chill_nerrs
95 #define yy_yys chill_yys
96 #define yystate chill_state
97 #define yytmp chill_tmp
99 #define yy_yyv chill_yyv
100 #define yyval chill_val
101 #define yylloc chill_lloc
102 #define yyreds chill_reds /* With YYDEBUG defined */
103 #define yytoks chill_toks /* With YYDEBUG defined */
106 #define YYDEBUG 0 /* Default to no yydebug support */
110 yyparse PARAMS ((void));
113 yylex PARAMS ((void));
116 yyerror PARAMS ((char *));
120 /* Although the yacc "value" of an expression is not used,
121 since the result is stored in the structure being created,
122 other node types do have values. */
127 unsigned LONGEST ulval;
137 struct symtoken ssym;
140 enum exp_opcode opcode;
141 struct internalvar *ivar;
147 %token <voidval> FIXME_01
148 %token <voidval> FIXME_02
149 %token <voidval> FIXME_03
150 %token <voidval> FIXME_04
151 %token <voidval> FIXME_05
152 %token <voidval> FIXME_06
153 %token <voidval> FIXME_07
154 %token <voidval> FIXME_08
155 %token <voidval> FIXME_09
156 %token <voidval> FIXME_10
157 %token <voidval> FIXME_11
158 %token <voidval> FIXME_12
159 %token <voidval> FIXME_13
160 %token <voidval> FIXME_14
161 %token <voidval> FIXME_15
162 %token <voidval> FIXME_16
163 %token <voidval> FIXME_17
164 %token <voidval> FIXME_18
165 %token <voidval> FIXME_19
166 %token <voidval> FIXME_20
167 %token <voidval> FIXME_21
168 %token <voidval> FIXME_22
169 %token <voidval> FIXME_24
170 %token <voidval> FIXME_25
171 %token <voidval> FIXME_26
172 %token <voidval> FIXME_27
173 %token <voidval> FIXME_28
174 %token <voidval> FIXME_29
175 %token <voidval> FIXME_30
177 %token <typed_val> INTEGER_LITERAL
178 %token <ulval> BOOLEAN_LITERAL
179 %token <typed_val> CHARACTER_LITERAL
180 %token <dval> FLOAT_LITERAL
181 %token <ssym> GENERAL_PROCEDURE_NAME
182 %token <ssym> LOCATION_NAME
183 %token <voidval> SET_LITERAL
184 %token <voidval> EMPTINESS_LITERAL
185 %token <sval> CHARACTER_STRING_LITERAL
186 %token <sval> BIT_STRING_LITERAL
187 %token <tsym> TYPENAME
188 %token <sval> FIELD_NAME
193 %token <voidval> CASE
195 %token <voidval> ESAC
196 %token <voidval> LOGIOR
197 %token <voidval> ORIF
198 %token <voidval> LOGXOR
199 %token <voidval> LOGAND
200 %token <voidval> ANDIF
202 %token <voidval> NOTEQUAL
212 %token <voidval> SLASH_SLASH
216 %token <voidval> POINTER
217 %token <voidval> RECEIVE
224 %token <voidval> THEN
225 %token <voidval> ELSE
227 %token <voidval> ELSIF
228 %token <voidval> ILLEGAL_TOKEN
230 %token <voidval> PRED
231 %token <voidval> SUCC
233 %token <voidval> CARD
234 %token <voidval> MAX_TOKEN
235 %token <voidval> MIN_TOKEN
236 %token <voidval> SIZE
237 %token <voidval> UPPER
238 %token <voidval> LOWER
239 %token <voidval> LENGTH
241 /* Tokens which are not Chill tokens used in expressions, but rather GDB
242 specific things that we recognize in the same context as Chill tokens
243 (register names for example). */
245 %token <lval> GDB_REGNAME /* Machine register name */
246 %token <lval> GDB_LAST /* Value history */
247 %token <ivar> GDB_VARIABLE /* Convenience variable */
248 %token <voidval> GDB_ASSIGNMENT /* Assign value to somewhere */
250 %type <voidval> access_name
251 %type <voidval> primitive_value
252 %type <voidval> value_name
253 %type <voidval> literal
254 %type <voidval> tuple
255 %type <voidval> value_string_element
256 %type <voidval> value_string_slice
257 %type <voidval> value_array_slice
258 %type <voidval> expression_conversion
259 %type <voidval> value_procedure_call
260 %type <voidval> value_built_in_routine_call
261 %type <voidval> chill_value_built_in_routine_call
262 %type <voidval> start_expression
263 %type <voidval> zero_adic_operator
264 %type <voidval> parenthesised_expression
265 %type <voidval> value
266 %type <voidval> undefined_value
267 %type <voidval> expression
268 %type <voidval> conditional_expression
269 %type <voidval> then_alternative
270 %type <voidval> else_alternative
271 %type <voidval> sub_expression
272 %type <voidval> value_case_alternative
273 %type <voidval> operand_0
274 %type <voidval> operand_1
275 %type <voidval> operand_2
276 %type <voidval> operand_3
277 %type <voidval> operand_4
278 %type <voidval> operand_5
279 %type <voidval> operand_6
280 %type <voidval> synonym_name
281 %type <voidval> value_enumeration_name
282 %type <voidval> value_do_with_name
283 %type <voidval> value_receive_name
284 %type <voidval> string_primitive_value
285 %type <voidval> start_element
286 %type <voidval> left_element
287 %type <voidval> right_element
288 %type <voidval> slice_size
289 %type <voidval> expression_list
290 %type <voidval> lower_element
291 %type <voidval> upper_element
292 %type <voidval> first_element
293 %type <tval> mode_argument
294 %type <voidval> upper_lower_argument
295 %type <voidval> length_argument
296 %type <voidval> array_mode_name
297 %type <voidval> string_mode_name
298 %type <voidval> variant_structure_mode_name
299 %type <voidval> boolean_expression
300 %type <voidval> case_selector_list
301 %type <voidval> subexpression
302 %type <voidval> case_label_specification
303 %type <voidval> buffer_location
304 %type <voidval> single_assignment_action
305 %type <tsym> mode_name
313 { write_exp_elt_opcode(OP_TYPE);
314 write_exp_elt_type($1.type);
315 write_exp_elt_opcode(OP_TYPE);}
328 undefined_value : FIXME_01
336 access_name : LOCATION_NAME
338 write_exp_elt_opcode (OP_VAR_VALUE);
339 write_exp_elt_block (NULL);
340 write_exp_elt_sym ($1.sym);
341 write_exp_elt_opcode (OP_VAR_VALUE);
343 | GDB_LAST /* gdb specific */
345 write_exp_elt_opcode (OP_LAST);
346 write_exp_elt_longcst ($1);
347 write_exp_elt_opcode (OP_LAST);
349 | GDB_REGNAME /* gdb specific */
351 write_exp_elt_opcode (OP_REGISTER);
352 write_exp_elt_longcst ($1);
353 write_exp_elt_opcode (OP_REGISTER);
355 | GDB_VARIABLE /* gdb specific */
357 write_exp_elt_opcode (OP_INTERNALVAR);
358 write_exp_elt_intern ($1);
359 write_exp_elt_opcode (OP_INTERNALVAR);
369 expression_list : expression
373 | expression_list ',' expression
384 | primitive_value '('
385 /* This is to save the value of arglist_len
386 being accumulated for each dimension. */
387 { start_arglist (); }
390 write_exp_elt_opcode (MULTI_SUBSCRIPT);
391 write_exp_elt_longcst ((LONGEST) end_arglist ());
392 write_exp_elt_opcode (MULTI_SUBSCRIPT);
394 | primitive_value FIELD_NAME
395 { write_exp_elt_opcode (STRUCTOP_STRUCT);
396 write_exp_string ($2);
397 write_exp_elt_opcode (STRUCTOP_STRUCT);
399 | primitive_value POINTER
401 write_exp_elt_opcode (UNOP_IND);
415 | value_string_element
427 | expression_conversion
431 | value_procedure_call
435 | value_built_in_routine_call
447 | parenthesised_expression
455 value_name : synonym_name
459 | value_enumeration_name
471 | GENERAL_PROCEDURE_NAME
473 write_exp_elt_opcode (OP_VAR_VALUE);
474 write_exp_elt_block (NULL);
475 write_exp_elt_sym ($1.sym);
476 write_exp_elt_opcode (OP_VAR_VALUE);
482 literal : INTEGER_LITERAL
484 write_exp_elt_opcode (OP_LONG);
485 write_exp_elt_type ($1.type);
486 write_exp_elt_longcst ((LONGEST) ($1.val));
487 write_exp_elt_opcode (OP_LONG);
491 write_exp_elt_opcode (OP_BOOL);
492 write_exp_elt_longcst ((LONGEST) $1);
493 write_exp_elt_opcode (OP_BOOL);
497 write_exp_elt_opcode (OP_LONG);
498 write_exp_elt_type ($1.type);
499 write_exp_elt_longcst ((LONGEST) ($1.val));
500 write_exp_elt_opcode (OP_LONG);
504 write_exp_elt_opcode (OP_DOUBLE);
505 write_exp_elt_type (builtin_type_double);
506 write_exp_elt_dblcst ($1);
507 write_exp_elt_opcode (OP_DOUBLE);
515 struct type *void_ptr_type
516 = lookup_pointer_type (builtin_type_void);
517 write_exp_elt_opcode (OP_LONG);
518 write_exp_elt_type (void_ptr_type);
519 write_exp_elt_longcst (0);
520 write_exp_elt_opcode (OP_LONG);
522 | CHARACTER_STRING_LITERAL
524 write_exp_elt_opcode (OP_STRING);
525 write_exp_string ($1);
526 write_exp_elt_opcode (OP_STRING);
530 write_exp_elt_opcode (OP_BITSTRING);
531 write_exp_bitstring ($1);
532 write_exp_elt_opcode (OP_BITSTRING);
538 tuple_element : expression
539 | named_record_element
542 named_record_element: FIELD_NAME ',' named_record_element
543 { write_exp_elt_opcode (OP_LABELED);
544 write_exp_string ($1);
545 write_exp_elt_opcode (OP_LABELED);
547 | FIELD_NAME ':' expression
548 { write_exp_elt_opcode (OP_LABELED);
549 write_exp_string ($1);
550 write_exp_elt_opcode (OP_LABELED);
554 tuple_elements : tuple_element
558 | tuple_elements ',' tuple_element
565 { start_arglist (); }
568 write_exp_elt_opcode (OP_ARRAY);
569 write_exp_elt_longcst ((LONGEST) 0);
570 write_exp_elt_longcst ((LONGEST) end_arglist () - 1);
571 write_exp_elt_opcode (OP_ARRAY);
575 { start_arglist (); }
578 write_exp_elt_opcode (OP_ARRAY);
579 write_exp_elt_longcst ((LONGEST) 0);
580 write_exp_elt_longcst ((LONGEST) end_arglist () - 1);
581 write_exp_elt_opcode (OP_ARRAY);
583 write_exp_elt_opcode (UNOP_CAST);
584 write_exp_elt_type ($1.type);
585 write_exp_elt_opcode (UNOP_CAST);
592 value_string_element: string_primitive_value '(' start_element ')'
600 value_string_slice: string_primitive_value '(' left_element ':' right_element ')'
604 | string_primitive_value '(' start_element UP slice_size ')'
612 value_array_slice: primitive_value '(' lower_element ':' upper_element ')'
616 | primitive_value '(' first_element UP slice_size ')'
624 expression_conversion: mode_name parenthesised_expression
626 write_exp_elt_opcode (UNOP_CAST);
627 write_exp_elt_type ($1.type);
628 write_exp_elt_opcode (UNOP_CAST);
634 value_procedure_call: FIXME_05
642 value_built_in_routine_call: chill_value_built_in_routine_call
650 start_expression: FIXME_06
653 } /* Not in GNU-Chill */
658 zero_adic_operator: FIXME_07
666 parenthesised_expression: '(' expression ')'
674 expression : operand_0
678 | single_assignment_action
682 | conditional_expression
688 conditional_expression : IF boolean_expression then_alternative else_alternative FI
692 | CASE case_selector_list OF value_case_alternative ELSE sub_expression ESAC
698 then_alternative: THEN subexpression
704 else_alternative: ELSE subexpression
708 | ELSIF boolean_expression then_alternative else_alternative
714 sub_expression : expression
720 value_case_alternative: case_label_specification ':' sub_expression ';'
728 operand_0 : operand_1
732 | operand_0 LOGIOR operand_1
734 write_exp_elt_opcode (BINOP_BITWISE_IOR);
736 | operand_0 ORIF operand_1
740 | operand_0 LOGXOR operand_1
742 write_exp_elt_opcode (BINOP_BITWISE_XOR);
748 operand_1 : operand_2
752 | operand_1 LOGAND operand_2
754 write_exp_elt_opcode (BINOP_BITWISE_AND);
756 | operand_1 ANDIF operand_2
764 operand_2 : operand_3
768 | operand_2 '=' operand_3
770 write_exp_elt_opcode (BINOP_EQUAL);
772 | operand_2 NOTEQUAL operand_3
774 write_exp_elt_opcode (BINOP_NOTEQUAL);
776 | operand_2 '>' operand_3
778 write_exp_elt_opcode (BINOP_GTR);
780 | operand_2 GTR operand_3
782 write_exp_elt_opcode (BINOP_GEQ);
784 | operand_2 '<' operand_3
786 write_exp_elt_opcode (BINOP_LESS);
788 | operand_2 LEQ operand_3
790 write_exp_elt_opcode (BINOP_LEQ);
792 | operand_2 IN operand_3
794 write_exp_elt_opcode (BINOP_IN);
801 operand_3 : operand_4
805 | operand_3 '+' operand_4
807 write_exp_elt_opcode (BINOP_ADD);
809 | operand_3 '-' operand_4
811 write_exp_elt_opcode (BINOP_SUB);
813 | operand_3 SLASH_SLASH operand_4
815 write_exp_elt_opcode (BINOP_CONCAT);
821 operand_4 : operand_5
825 | operand_4 '*' operand_5
827 write_exp_elt_opcode (BINOP_MUL);
829 | operand_4 '/' operand_5
831 write_exp_elt_opcode (BINOP_DIV);
833 | operand_4 MOD operand_5
835 write_exp_elt_opcode (BINOP_MOD);
837 | operand_4 REM operand_5
839 write_exp_elt_opcode (BINOP_REM);
845 operand_5 : operand_6
851 write_exp_elt_opcode (UNOP_NEG);
855 write_exp_elt_opcode (UNOP_LOGICAL_NOT);
857 | parenthesised_expression literal
858 /* We require the string operand to be a literal, to avoid some
859 nasty parsing ambiguities. */
861 write_exp_elt_opcode (BINOP_CONCAT);
867 operand_6 : POINTER primitive_value
869 write_exp_elt_opcode (UNOP_ADDR);
871 | RECEIVE buffer_location
884 single_assignment_action :
885 primitive_value GDB_ASSIGNMENT value
887 write_exp_elt_opcode (BINOP_ASSIGN);
893 chill_value_built_in_routine_call :
894 NUM '(' expression ')'
898 | PRED '(' expression ')'
902 | SUCC '(' expression ')'
906 | ABS '(' expression ')'
910 | CARD '(' expression ')'
914 | MAX_TOKEN '(' expression ')'
918 | MIN_TOKEN '(' expression ')'
922 | SIZE '(' expression ')'
923 { write_exp_elt_opcode (UNOP_SIZEOF); }
924 | SIZE '(' mode_argument ')'
925 { write_exp_elt_opcode (OP_LONG);
926 write_exp_elt_type (builtin_type_int);
927 write_exp_elt_longcst ((LONGEST) TYPE_LENGTH ($3));
928 write_exp_elt_opcode (OP_LONG); }
929 | UPPER '(' upper_lower_argument ')'
933 | LOWER '(' upper_lower_argument ')'
937 | LENGTH '(' length_argument ')'
943 mode_argument : mode_name
947 | array_mode_name '(' expression ')'
951 | string_mode_name '(' expression ')'
955 | variant_structure_mode_name '(' expression_list ')'
964 upper_lower_argument : expression
974 length_argument : expression
980 /* Things which still need productions... */
982 array_mode_name : FIXME_08 { $$ = 0; }
983 string_mode_name : FIXME_09 { $$ = 0; }
984 variant_structure_mode_name: FIXME_10 { $$ = 0; }
985 synonym_name : FIXME_11 { $$ = 0; }
986 value_enumeration_name : FIXME_12 { $$ = 0; }
987 value_do_with_name : FIXME_13 { $$ = 0; }
988 value_receive_name : FIXME_14 { $$ = 0; }
989 string_primitive_value : FIXME_15 { $$ = 0; }
990 start_element : FIXME_16 { $$ = 0; }
991 left_element : FIXME_17 { $$ = 0; }
992 right_element : FIXME_18 { $$ = 0; }
993 slice_size : FIXME_19 { $$ = 0; }
994 lower_element : FIXME_20 { $$ = 0; }
995 upper_element : FIXME_21 { $$ = 0; }
996 first_element : FIXME_22 { $$ = 0; }
997 boolean_expression : FIXME_26 { $$ = 0; }
998 case_selector_list : FIXME_27 { $$ = 0; }
999 subexpression : FIXME_28 { $$ = 0; }
1000 case_label_specification: FIXME_29 { $$ = 0; }
1001 buffer_location : FIXME_30 { $$ = 0; }
1005 /* Implementation of a dynamically expandable buffer for processing input
1006 characters acquired through lexptr and building a value to return in
1009 static char *tempbuf; /* Current buffer contents */
1010 static int tempbufsize; /* Size of allocated buffer */
1011 static int tempbufindex; /* Current index into buffer */
1013 #define GROWBY_MIN_SIZE 64 /* Minimum amount to grow buffer by */
1015 #define CHECKBUF(size) \
1017 if (tempbufindex + (size) >= tempbufsize) \
1019 growbuf_by_size (size); \
1023 /* Grow the static temp buffer if necessary, including allocating the first one
1027 growbuf_by_size (count)
1032 growby = max (count, GROWBY_MIN_SIZE);
1033 tempbufsize += growby;
1034 if (tempbuf == NULL)
1036 tempbuf = (char *) malloc (tempbufsize);
1040 tempbuf = (char *) realloc (tempbuf, tempbufsize);
1044 /* Try to consume a simple name string token. If successful, returns
1045 a pointer to a nullbyte terminated copy of the name that can be used
1046 in symbol table lookups. If not successful, returns NULL. */
1049 match_simple_name_string ()
1051 char *tokptr = lexptr;
1053 if (isalpha (*tokptr) || *tokptr == '_')
1058 } while (isalnum (*tokptr) || (*tokptr == '_'));
1059 yylval.sval.ptr = lexptr;
1060 yylval.sval.length = tokptr - lexptr;
1062 result = copy_name (yylval.sval);
1068 /* Start looking for a value composed of valid digits as set by the base
1069 in use. Note that '_' characters are valid anywhere, in any quantity,
1070 and are simply ignored. Since we must find at least one valid digit,
1071 or reject this token as an integer literal, we keep track of how many
1072 digits we have encountered. */
1075 decode_integer_value (base, tokptrptr, ivalptr)
1080 char *tokptr = *tokptrptr;
1084 while (*tokptr != '\0')
1088 temp = tolower (temp);
1094 case '0': case '1': case '2': case '3': case '4':
1095 case '5': case '6': case '7': case '8': case '9':
1098 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
1114 /* Found something not in domain for current base. */
1115 tokptr--; /* Unconsume what gave us indigestion. */
1120 /* If we didn't find any digits, then we don't have a valid integer
1121 value, so reject the entire token. Otherwise, update the lexical
1122 scan pointer, and return non-zero for success. */
1130 *tokptrptr = tokptr;
1136 decode_integer_literal (valptr, tokptrptr)
1140 char *tokptr = *tokptrptr;
1143 int explicit_base = 0;
1145 /* Look for an explicit base specifier, which is optional. */
1178 /* If we found an explicit base ensure that the character after the
1179 explicit base is a single quote. */
1181 if (explicit_base && (*tokptr++ != '\''))
1186 /* Attempt to decode whatever follows as an integer value in the
1187 indicated base, updating the token pointer in the process and
1188 computing the value into ival. Also, if we have an explicit
1189 base, then the next character must not be a single quote, or we
1190 have a bitstring literal, so reject the entire token in this case.
1191 Otherwise, update the lexical scan pointer, and return non-zero
1194 if (!decode_integer_value (base, &tokptr, &ival))
1198 else if (explicit_base && (*tokptr == '\''))
1205 *tokptrptr = tokptr;
1210 /* If it wasn't for the fact that floating point values can contain '_'
1211 characters, we could just let strtod do all the hard work by letting it
1212 try to consume as much of the current token buffer as possible and
1213 find a legal conversion. Unfortunately we need to filter out the '_'
1214 characters before calling strtod, which we do by copying the other
1215 legal chars to a local buffer to be converted. However since we also
1216 need to keep track of where the last unconsumed character in the input
1217 buffer is, we have transfer only as many characters as may compose a
1218 legal floating point value. */
1221 match_float_literal ()
1223 char *tokptr = lexptr;
1227 extern double strtod ();
1229 /* Make local buffer in which to build the string to convert. This is
1230 required because underscores are valid in chill floating point numbers
1231 but not in the string passed to strtod to convert. The string will be
1232 no longer than our input string. */
1234 copy = buf = (char *) alloca (strlen (tokptr) + 1);
1236 /* Transfer all leading digits to the conversion buffer, discarding any
1239 while (isdigit (*tokptr) || *tokptr == '_')
1248 /* Now accept either a '.', or one of [eEdD]. Dot is legal regardless
1249 of whether we found any leading digits, and we simply accept it and
1250 continue on to look for the fractional part and/or exponent. One of
1251 [eEdD] is legal only if we have seen digits, and means that there
1252 is no fractional part. If we find neither of these, then this is
1253 not a floating point number, so return failure. */
1258 /* Accept and then look for fractional part and/or exponent. */
1271 goto collect_exponent;
1279 /* We found a '.', copy any fractional digits to the conversion buffer, up
1280 to the first nondigit, non-underscore character. */
1282 while (isdigit (*tokptr) || *tokptr == '_')
1291 /* Look for an exponent, which must start with one of [eEdD]. If none
1292 is found, jump directly to trying to convert what we have collected
1309 /* Accept an optional '-' or '+' following one of [eEdD]. */
1312 if (*tokptr == '+' || *tokptr == '-')
1314 *copy++ = *tokptr++;
1317 /* Now copy an exponent into the conversion buffer. Note that at the
1318 moment underscores are *not* allowed in exponents. */
1320 while (isdigit (*tokptr))
1322 *copy++ = *tokptr++;
1325 /* If we transfered any chars to the conversion buffer, try to interpret its
1326 contents as a floating point value. If any characters remain, then we
1327 must not have a valid floating point string. */
1333 dval = strtod (buf, ©);
1338 return (FLOAT_LITERAL);
1344 /* Recognize a string literal. A string literal is a sequence
1345 of characters enclosed in matching single or double quotes, except that
1346 a single character inside single quotes is a character literal, which
1347 we reject as a string literal. To embed the terminator character inside
1348 a string, it is simply doubled (I.E. "this""is""one""string") */
1351 match_string_literal ()
1353 char *tokptr = lexptr;
1355 for (tempbufindex = 0, tokptr++; *tokptr != '\0'; tokptr++)
1358 if (*tokptr == *lexptr)
1360 if (*(tokptr + 1) == *lexptr)
1369 tempbuf[tempbufindex++] = *tokptr;
1371 if (*tokptr == '\0' /* no terminator */
1372 || (tempbufindex == 1 && *tokptr == '\'')) /* char literal */
1378 tempbuf[tempbufindex] = '\0';
1379 yylval.sval.ptr = tempbuf;
1380 yylval.sval.length = tempbufindex;
1382 return (CHARACTER_STRING_LITERAL);
1386 /* Recognize a character literal. A character literal is single character
1387 or a control sequence, enclosed in single quotes. A control sequence
1388 is a comma separated list of one or more integer literals, enclosed
1389 in parenthesis and introduced with a circumflex character.
1391 EX: 'a' '^(7)' '^(7,8)'
1393 As a GNU chill extension, the syntax C'xx' is also recognized as a
1394 character literal, where xx is a hex value for the character.
1396 Note that more than a single character, enclosed in single quotes, is
1399 Also note that the control sequence form is not in GNU Chill since it
1400 is ambiguous with the string literal form using single quotes. I.E.
1401 is '^(7)' a character literal or a string literal. In theory it it
1402 possible to tell by context, but GNU Chill doesn't accept the control
1403 sequence form, so neither do we (for now the code is disabled).
1405 Returns CHARACTER_LITERAL if a match is found.
1409 match_character_literal ()
1411 char *tokptr = lexptr;
1414 if ((*tokptr == 'c' || *tokptr == 'C') && (*(tokptr + 1) == '\''))
1416 /* We have a GNU chill extension form, so skip the leading "C'",
1417 decode the hex value, and then ensure that we have a trailing
1418 single quote character. */
1420 if (!decode_integer_value (16, &tokptr, &ival) || (*tokptr != '\''))
1426 else if (*tokptr == '\'')
1430 /* Determine which form we have, either a control sequence or the
1431 single character form. */
1433 if ((*tokptr == '^') && (*(tokptr + 1) == '('))
1435 #if 0 /* Disable, see note above. -fnf */
1436 /* Match and decode a control sequence. Return zero if we don't
1437 find a valid integer literal, or if the next unconsumed character
1438 after the integer literal is not the trailing ')'.
1439 FIXME: We currently don't handle the multiple integer literal
1442 if (!decode_integer_literal (&ival, &tokptr) || (*tokptr++ != ')'))
1455 /* The trailing quote has not yet been consumed. If we don't find
1456 it, then we have no match. */
1458 if (*tokptr++ != '\'')
1465 /* Not a character literal. */
1468 yylval.typed_val.val = ival;
1469 yylval.typed_val.type = builtin_type_chill_char;
1471 return (CHARACTER_LITERAL);
1474 /* Recognize an integer literal, as specified in Z.200 sec 5.2.4.2.
1475 Note that according to 5.2.4.2, a single "_" is also a valid integer
1476 literal, however GNU-chill requires there to be at least one "digit"
1477 in any integer literal. */
1480 match_integer_literal ()
1482 char *tokptr = lexptr;
1485 if (!decode_integer_literal (&ival, &tokptr))
1491 yylval.typed_val.val = ival;
1492 yylval.typed_val.type = builtin_type_int;
1494 return (INTEGER_LITERAL);
1498 /* Recognize a bit-string literal, as specified in Z.200 sec 5.2.4.8
1499 Note that according to 5.2.4.8, a single "_" is also a valid bit-string
1500 literal, however GNU-chill requires there to be at least one "digit"
1501 in any bit-string literal. */
1504 match_bitstring_literal ()
1506 char *tokptr = lexptr;
1515 /* Look for the required explicit base specifier. */
1536 /* Ensure that the character after the explicit base is a single quote. */
1538 if (*tokptr++ != '\'')
1543 while (*tokptr != '\0' && *tokptr != '\'')
1546 if (isupper (digit))
1547 digit = tolower (digit);
1553 case '0': case '1': case '2': case '3': case '4':
1554 case '5': case '6': case '7': case '8': case '9':
1557 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
1567 /* Found something not in domain for current base. */
1572 /* Extract bits from digit, starting with the msbit appropriate for
1573 the current base, and packing them into the bitstring byte,
1574 starting at the lsbit. */
1575 for (mask = (base >> 1); mask > 0; mask >>= 1)
1581 tempbuf[tempbufindex] |= (1 << bitoffset);
1584 if (bitoffset == HOST_CHAR_BIT)
1593 /* Verify that we consumed everything up to the trailing single quote,
1594 and that we found some bits (IE not just underbars). */
1596 if (*tokptr++ != '\'')
1602 yylval.sval.ptr = tempbuf;
1603 yylval.sval.length = bitcount;
1605 return (BIT_STRING_LITERAL);
1609 /* Recognize tokens that start with '$'. These include:
1611 $regname A native register name or a "standard
1613 Return token GDB_REGNAME.
1615 $variable A convenience variable with a name chosen
1617 Return token GDB_VARIABLE.
1619 $digits Value history with index <digits>, starting
1620 from the first value which has index 1.
1623 $$digits Value history with index <digits> relative
1624 to the last value. I.E. $$0 is the last
1625 value, $$1 is the one previous to that, $$2
1626 is the one previous to $$1, etc.
1627 Return token GDB_LAST.
1629 $ | $0 | $$0 The last value in the value history.
1630 Return token GDB_LAST.
1632 $$ An abbreviation for the second to the last
1633 value in the value history, I.E. $$1
1634 Return token GDB_LAST.
1636 Note that we currently assume that register names and convenience
1637 variables follow the convention of starting with a letter or '_'.
1642 match_dollar_tokens ()
1650 /* We will always have a successful match, even if it is just for
1651 a single '$', the abbreviation for $$0. So advance lexptr. */
1655 if (*tokptr == '_' || isalpha (*tokptr))
1657 /* Look for a match with a native register name, usually something
1658 like "r0" for example. */
1660 for (regno = 0; regno < NUM_REGS; regno++)
1662 namelength = strlen (reg_names[regno]);
1663 if (STREQN (tokptr, reg_names[regno], namelength)
1664 && !isalnum (tokptr[namelength]))
1666 yylval.lval = regno;
1667 lexptr += namelength;
1668 return (GDB_REGNAME);
1672 /* Look for a match with a standard register name, usually something
1673 like "pc", which gdb always recognizes as the program counter
1674 regardless of what the native register name is. */
1676 for (regno = 0; regno < num_std_regs; regno++)
1678 namelength = strlen (std_regs[regno].name);
1679 if (STREQN (tokptr, std_regs[regno].name, namelength)
1680 && !isalnum (tokptr[namelength]))
1682 yylval.lval = std_regs[regno].regnum;
1683 lexptr += namelength;
1684 return (GDB_REGNAME);
1688 /* Attempt to match against a convenience variable. Note that
1689 this will always succeed, because if no variable of that name
1690 already exists, the lookup_internalvar will create one for us.
1691 Also note that both lexptr and tokptr currently point to the
1692 start of the input string we are trying to match, and that we
1693 have already tested the first character for non-numeric, so we
1694 don't have to treat it specially. */
1696 while (*tokptr == '_' || isalnum (*tokptr))
1700 yylval.sval.ptr = lexptr;
1701 yylval.sval.length = tokptr - lexptr;
1702 yylval.ivar = lookup_internalvar (copy_name (yylval.sval));
1704 return (GDB_VARIABLE);
1707 /* Since we didn't match against a register name or convenience
1708 variable, our only choice left is a history value. */
1722 /* Attempt to decode more characters as an integer value giving
1723 the index in the history list. If successful, the value will
1724 overwrite ival (currently 0 or 1), and if not, ival will be
1725 left alone, which is good since it is currently correct for
1726 the '$' or '$$' case. */
1728 decode_integer_literal (&ival, &tokptr);
1729 yylval.lval = negate ? -ival : ival;
1740 static const struct token idtokentab[] =
1742 { "length", LENGTH },
1753 { "max", MAX_TOKEN },
1754 { "min", MIN_TOKEN },
1762 { "null", EMPTINESS_LITERAL }
1765 static const struct token tokentab2[] =
1767 { ":=", GDB_ASSIGNMENT },
1768 { "//", SLASH_SLASH },
1775 /* Read one token, getting characters through lexptr. */
1776 /* This is where we will check to make sure that the language and the
1777 operators used are compatible. */
1787 /* Skip over any leading whitespace. */
1788 while (isspace (*lexptr))
1792 /* Look for special single character cases which can't be the first
1793 character of some other multicharacter token. */
1810 /* Look for characters which start a particular kind of multicharacter
1811 token, such as a character literal, register name, convenience
1812 variable name, string literal, etc. */
1817 /* First try to match a string literal, which is any
1818 sequence of characters enclosed in matching single or double
1819 quotes, except that a single character inside single quotes
1820 is a character literal, so we have to catch that case also. */
1821 token = match_string_literal ();
1826 if (*lexptr == '\'')
1828 token = match_character_literal ();
1837 token = match_character_literal ();
1844 token = match_dollar_tokens ();
1851 /* See if it is a special token of length 2. */
1852 for (i = 0; i < sizeof (tokentab2) / sizeof (tokentab2[0]); i++)
1854 if (STREQN (lexptr, tokentab2[i].operator, 2))
1857 return (tokentab2[i].token);
1860 /* Look for single character cases which which could be the first
1861 character of some other multicharacter token, but aren't, or we
1862 would already have found it. */
1872 /* Look for a float literal before looking for an integer literal, so
1873 we match as much of the input stream as possible. */
1874 token = match_float_literal ();
1879 token = match_bitstring_literal ();
1884 token = match_integer_literal ();
1890 /* Try to match a simple name string, and if a match is found, then
1891 further classify what sort of name it is and return an appropriate
1892 token. Note that attempting to match a simple name string consumes
1893 the token from lexptr, so we can't back out if we later find that
1894 we can't classify what sort of name it is. */
1896 inputname = match_simple_name_string ();
1898 if (inputname != NULL)
1900 char *simplename = (char*) alloca (strlen (inputname) + 1);
1902 char *dptr = simplename, *sptr = inputname;
1903 for (; *sptr; sptr++)
1904 *dptr++ = isupper (*sptr) ? tolower(*sptr) : *sptr;
1907 /* See if it is a reserved identifier. */
1908 for (i = 0; i < sizeof (idtokentab) / sizeof (idtokentab[0]); i++)
1910 if (STREQ (simplename, idtokentab[i].operator))
1912 return (idtokentab[i].token);
1916 /* Look for other special tokens. */
1917 if (STREQ (simplename, "true"))
1920 return (BOOLEAN_LITERAL);
1922 if (STREQ (simplename, "false"))
1925 return (BOOLEAN_LITERAL);
1928 sym = lookup_symbol (inputname, expression_context_block,
1929 VAR_NAMESPACE, (int *) NULL,
1930 (struct symtab **) NULL);
1931 if (sym == NULL && strcmp (inputname, simplename) != 0)
1933 sym = lookup_symbol (simplename, expression_context_block,
1934 VAR_NAMESPACE, (int *) NULL,
1935 (struct symtab **) NULL);
1939 yylval.ssym.stoken.ptr = NULL;
1940 yylval.ssym.stoken.length = 0;
1941 yylval.ssym.sym = sym;
1942 yylval.ssym.is_a_field_of_this = 0; /* FIXME, C++'ism */
1943 switch (SYMBOL_CLASS (sym))
1946 /* Found a procedure name. */
1947 return (GENERAL_PROCEDURE_NAME);
1949 /* Found a global or local static variable. */
1950 return (LOCATION_NAME);
1955 case LOC_REGPARM_ADDR:
1959 case LOC_BASEREG_ARG:
1960 if (innermost_block == NULL
1961 || contained_in (block_found, innermost_block))
1963 innermost_block = block_found;
1965 return (LOCATION_NAME);
1969 return (LOCATION_NAME);
1972 yylval.tsym.type = SYMBOL_TYPE (sym);
1975 case LOC_CONST_BYTES:
1976 case LOC_OPTIMIZED_OUT:
1977 error ("Symbol \"%s\" names no location.", inputname);
1981 else if (!have_full_symbols () && !have_partial_symbols ())
1983 error ("No symbol table is loaded. Use the \"file\" command.");
1987 error ("No symbol \"%s\" in current context.", inputname);
1991 /* Catch single character tokens which are not part of some
1996 case '.': /* Not float for example. */
1998 while (isspace (*lexptr)) lexptr++;
1999 inputname = match_simple_name_string ();
2005 return (ILLEGAL_TOKEN);
2012 error ("A %s in expression, near `%s'.", (msg ? msg : "error"), lexptr);