1 /* YACC grammar for Chill expressions, for GDB.
2 Copyright 1992, 1993 Free Software Foundation, Inc.
4 This file is part of GDB.
6 This program is free software; you can redistribute it and/or modify
7 it under the terms of the GNU General Public License as published by
8 the Free Software Foundation; either version 2 of the License, or
9 (at your option) any later version.
11 This program is distributed in the hope that it will be useful,
12 but WITHOUT ANY WARRANTY; without even the implied warranty of
13 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
14 GNU General Public License for more details.
16 You should have received a copy of the GNU General Public License
17 along with this program; if not, write to the Free Software
18 Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. */
20 /* Parse a Chill expression from text in a string,
21 and return the result as a struct expression pointer.
22 That structure contains arithmetic operations in reverse polish,
23 with constants represented by operations that are followed by special data.
24 See expression.h for the details of the format.
25 What is important here is that it can be built up sequentially
26 during the process of parsing; the lower levels of the tree always
27 come first in the result.
29 Note that malloc's and realloc's in this file are transformed to
30 xmalloc and xrealloc respectively by the same sed command in the
31 makefile that remaps any other malloc/realloc inserted by the parser
32 generator. Doing this with #defines and trying to control the interaction
33 with include files (<malloc.h> and <stdlib.h> for example) just became
34 too messy, particularly when such includes can be inserted at random
35 times by the parser generator.
37 Also note that the language accepted by this parser is more liberal
38 than the one accepted by an actual Chill compiler. For example, the
39 language rule that a simple name string can not be one of the reserved
40 simple name strings is not enforced (e.g "case" is not treated as a
41 reserved name). Another example is that Chill is a strongly typed
42 language, and certain expressions that violate the type constraints
43 may still be evaluated if gdb can do so in a meaningful manner, while
44 such expressions would be rejected by the compiler. The reason for
45 this more liberal behavior is the philosophy that the debugger
46 is intended to be a tool that is used by the programmer when things
47 go wrong, and as such, it should provide as few artificial barriers
48 to it's use as possible. If it can do something meaningful, even
49 something that violates language contraints that are enforced by the
50 compiler, it should do so without complaint.
58 #include "expression.h"
61 #include "parser-defs.h"
64 /* Remap normal yacc parser interface names (yyparse, yylex, yyerror, etc),
65 as well as gratuitiously global symbol names, so we can have multiple
66 yacc generated parsers in gdb. Note that these are only the variables
67 produced by yacc. If other parser generators (bison, byacc, etc) produce
68 additional global names that conflict at link time, then those parser
69 generators need to be fixed instead of adding those names to this list. */
71 #define yymaxdepth chill_maxdepth
72 #define yyparse chill_parse
73 #define yylex chill_lex
74 #define yyerror chill_error
75 #define yylval chill_lval
76 #define yychar chill_char
77 #define 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_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
183 %token <tsym> TYPENAME
184 %token <sval> FIELD_NAME
189 %token <voidval> CASE
191 %token <voidval> ESAC
192 %token <voidval> LOGIOR
193 %token <voidval> ORIF
194 %token <voidval> LOGXOR
195 %token <voidval> LOGAND
196 %token <voidval> ANDIF
198 %token <voidval> NOTEQUAL
208 %token <voidval> SLASH_SLASH
212 %token <voidval> POINTER
213 %token <voidval> RECEIVE
220 %token <voidval> THEN
221 %token <voidval> ELSE
223 %token <voidval> ELSIF
224 %token <voidval> ILLEGAL_TOKEN
226 %token <voidval> PRED
227 %token <voidval> SUCC
229 %token <voidval> CARD
230 %token <voidval> MAX_TOKEN
231 %token <voidval> MIN_TOKEN
232 %token <voidval> SIZE
233 %token <voidval> UPPER
234 %token <voidval> LOWER
235 %token <voidval> LENGTH
237 /* Tokens which are not Chill tokens used in expressions, but rather GDB
238 specific things that we recognize in the same context as Chill tokens
239 (register names for example). */
241 %token <lval> GDB_REGNAME /* Machine register name */
242 %token <lval> GDB_LAST /* Value history */
243 %token <ivar> GDB_VARIABLE /* Convenience variable */
244 %token <voidval> GDB_ASSIGNMENT /* Assign value to somewhere */
246 %type <voidval> location
247 %type <voidval> access_name
248 %type <voidval> primitive_value
249 %type <voidval> location_contents
250 %type <voidval> value_name
251 %type <voidval> literal
252 %type <voidval> tuple
253 %type <voidval> value_string_element
254 %type <voidval> value_string_slice
255 %type <voidval> value_array_element
256 %type <voidval> value_array_slice
257 %type <voidval> value_structure_field
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> array_primitive_value
290 %type <voidval> expression_list
291 %type <voidval> lower_element
292 %type <voidval> upper_element
293 %type <voidval> first_element
294 %type <voidval> mode_argument
295 %type <voidval> upper_lower_argument
296 %type <voidval> length_argument
297 %type <voidval> array_mode_name
298 %type <voidval> string_mode_name
299 %type <voidval> variant_structure_mode_name
300 %type <voidval> boolean_expression
301 %type <voidval> case_selector_list
302 %type <voidval> subexpression
303 %type <voidval> case_label_specification
304 %type <voidval> buffer_location
305 %type <voidval> single_assignment_action
306 %type <tsym> mode_name
314 { write_exp_elt_opcode(OP_TYPE);
315 write_exp_elt_type($1.type);
316 write_exp_elt_opcode(OP_TYPE);}
329 undefined_value : FIXME_01
337 location : access_name
338 | primitive_value POINTER
340 write_exp_elt_opcode (UNOP_IND);
346 access_name : LOCATION_NAME
348 write_exp_elt_opcode (OP_VAR_VALUE);
349 write_exp_elt_sym ($1.sym);
350 write_exp_elt_opcode (OP_VAR_VALUE);
352 | GDB_LAST /* gdb specific */
354 write_exp_elt_opcode (OP_LAST);
355 write_exp_elt_longcst ($1);
356 write_exp_elt_opcode (OP_LAST);
358 | GDB_REGNAME /* gdb specific */
360 write_exp_elt_opcode (OP_REGISTER);
361 write_exp_elt_longcst ($1);
362 write_exp_elt_opcode (OP_REGISTER);
364 | GDB_VARIABLE /* gdb specific */
366 write_exp_elt_opcode (OP_INTERNALVAR);
367 write_exp_elt_intern ($1);
368 write_exp_elt_opcode (OP_INTERNALVAR);
378 expression_list : expression
382 | expression_list ',' expression
389 primitive_value : location_contents
405 | value_string_element
413 | value_array_element
421 | value_structure_field
425 | expression_conversion
429 | value_procedure_call
433 | value_built_in_routine_call
445 | parenthesised_expression
453 location_contents: location
461 value_name : synonym_name
465 | value_enumeration_name
477 | GENERAL_PROCEDURE_NAME
479 write_exp_elt_opcode (OP_VAR_VALUE);
480 write_exp_elt_sym ($1.sym);
481 write_exp_elt_opcode (OP_VAR_VALUE);
487 literal : INTEGER_LITERAL
489 write_exp_elt_opcode (OP_LONG);
490 write_exp_elt_type ($1.type);
491 write_exp_elt_longcst ((LONGEST) ($1.val));
492 write_exp_elt_opcode (OP_LONG);
496 write_exp_elt_opcode (OP_BOOL);
497 write_exp_elt_longcst ((LONGEST) $1);
498 write_exp_elt_opcode (OP_BOOL);
502 write_exp_elt_opcode (OP_LONG);
503 write_exp_elt_type ($1.type);
504 write_exp_elt_longcst ((LONGEST) ($1.val));
505 write_exp_elt_opcode (OP_LONG);
509 write_exp_elt_opcode (OP_DOUBLE);
510 write_exp_elt_type (builtin_type_double);
511 write_exp_elt_dblcst ($1);
512 write_exp_elt_opcode (OP_DOUBLE);
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);
547 value_string_element: string_primitive_value '(' start_element ')'
555 value_string_slice: string_primitive_value '(' left_element ':' right_element ')'
559 | string_primitive_value '(' start_element UP slice_size ')'
567 value_array_element: array_primitive_value '('
568 /* This is to save the value of arglist_len
569 being accumulated for each dimension. */
570 { start_arglist (); }
573 write_exp_elt_opcode (MULTI_SUBSCRIPT);
574 write_exp_elt_longcst ((LONGEST) end_arglist ());
575 write_exp_elt_opcode (MULTI_SUBSCRIPT);
581 value_array_slice: array_primitive_value '(' lower_element ':' upper_element ')'
585 | array_primitive_value '(' first_element UP slice_size ')'
593 value_structure_field: primitive_value FIELD_NAME
594 { write_exp_elt_opcode (STRUCTOP_STRUCT);
595 write_exp_string ($2);
596 write_exp_elt_opcode (STRUCTOP_STRUCT);
602 expression_conversion: mode_name parenthesised_expression
604 write_exp_elt_opcode (UNOP_CAST);
605 write_exp_elt_type ($1.type);
606 write_exp_elt_opcode (UNOP_CAST);
612 value_procedure_call: FIXME_05
620 value_built_in_routine_call: chill_value_built_in_routine_call
628 start_expression: FIXME_06
631 } /* Not in GNU-Chill */
636 zero_adic_operator: FIXME_07
644 parenthesised_expression: '(' expression ')'
652 expression : operand_0
656 | single_assignment_action
660 | conditional_expression
666 conditional_expression : IF boolean_expression then_alternative else_alternative FI
670 | CASE case_selector_list OF value_case_alternative '[' ELSE sub_expression ']' ESAC
676 then_alternative: THEN subexpression
682 else_alternative: ELSE subexpression
686 | ELSIF boolean_expression then_alternative else_alternative
692 sub_expression : expression
698 value_case_alternative: case_label_specification ':' sub_expression ';'
706 operand_0 : operand_1
710 | operand_0 LOGIOR operand_1
712 write_exp_elt_opcode (BINOP_BITWISE_IOR);
714 | operand_0 ORIF operand_1
718 | operand_0 LOGXOR operand_1
720 write_exp_elt_opcode (BINOP_BITWISE_XOR);
726 operand_1 : operand_2
730 | operand_1 LOGAND operand_2
732 write_exp_elt_opcode (BINOP_BITWISE_AND);
734 | operand_1 ANDIF operand_2
742 operand_2 : operand_3
746 | operand_2 '=' operand_3
748 write_exp_elt_opcode (BINOP_EQUAL);
750 | operand_2 NOTEQUAL operand_3
752 write_exp_elt_opcode (BINOP_NOTEQUAL);
754 | operand_2 '>' operand_3
756 write_exp_elt_opcode (BINOP_GTR);
758 | operand_2 GTR operand_3
760 write_exp_elt_opcode (BINOP_GEQ);
762 | operand_2 '<' operand_3
764 write_exp_elt_opcode (BINOP_LESS);
766 | operand_2 LEQ operand_3
768 write_exp_elt_opcode (BINOP_LEQ);
770 | operand_2 IN operand_3
779 operand_3 : operand_4
783 | operand_3 '+' operand_4
785 write_exp_elt_opcode (BINOP_ADD);
787 | operand_3 '-' operand_4
789 write_exp_elt_opcode (BINOP_SUB);
791 | operand_3 SLASH_SLASH operand_4
793 write_exp_elt_opcode (BINOP_CONCAT);
799 operand_4 : operand_5
803 | operand_4 '*' operand_5
805 write_exp_elt_opcode (BINOP_MUL);
807 | operand_4 '/' operand_5
809 write_exp_elt_opcode (BINOP_DIV);
811 | operand_4 MOD operand_5
813 write_exp_elt_opcode (BINOP_MOD);
815 | operand_4 REM operand_5
817 write_exp_elt_opcode (BINOP_REM);
823 operand_5 : operand_6
829 write_exp_elt_opcode (UNOP_NEG);
833 write_exp_elt_opcode (UNOP_LOGICAL_NOT);
835 | parenthesised_expression literal
836 /* We require the string operand to be a literal, to avoid some
837 nasty parsing ambiguities. */
839 write_exp_elt_opcode (BINOP_CONCAT);
845 operand_6 : POINTER location
847 write_exp_elt_opcode (UNOP_ADDR);
849 | RECEIVE buffer_location
862 single_assignment_action :
863 location GDB_ASSIGNMENT value
865 write_exp_elt_opcode (BINOP_ASSIGN);
871 chill_value_built_in_routine_call :
872 NUM '(' expression ')'
876 | PRED '(' expression ')'
880 | SUCC '(' expression ')'
884 | ABS '(' expression ')'
888 | CARD '(' expression ')'
892 | MAX_TOKEN '(' expression ')'
896 | MIN_TOKEN '(' expression ')'
900 | SIZE '(' location ')'
904 | SIZE '(' mode_argument ')'
908 | UPPER '(' upper_lower_argument ')'
912 | LOWER '(' upper_lower_argument ')'
916 | LENGTH '(' length_argument ')'
922 mode_argument : mode_name
926 | array_mode_name '(' expression ')'
930 | string_mode_name '(' expression ')'
934 | variant_structure_mode_name '(' expression_list ')'
943 upper_lower_argument : expression
953 length_argument : expression
961 array_primitive_value : primitive_value
968 /* Things which still need productions... */
970 array_mode_name : FIXME_08 { $$ = 0; }
971 string_mode_name : FIXME_09 { $$ = 0; }
972 variant_structure_mode_name: FIXME_10 { $$ = 0; }
973 synonym_name : FIXME_11 { $$ = 0; }
974 value_enumeration_name : FIXME_12 { $$ = 0; }
975 value_do_with_name : FIXME_13 { $$ = 0; }
976 value_receive_name : FIXME_14 { $$ = 0; }
977 string_primitive_value : FIXME_15 { $$ = 0; }
978 start_element : FIXME_16 { $$ = 0; }
979 left_element : FIXME_17 { $$ = 0; }
980 right_element : FIXME_18 { $$ = 0; }
981 slice_size : FIXME_19 { $$ = 0; }
982 lower_element : FIXME_20 { $$ = 0; }
983 upper_element : FIXME_21 { $$ = 0; }
984 first_element : FIXME_22 { $$ = 0; }
985 boolean_expression : FIXME_26 { $$ = 0; }
986 case_selector_list : FIXME_27 { $$ = 0; }
987 subexpression : FIXME_28 { $$ = 0; }
988 case_label_specification: FIXME_29 { $$ = 0; }
989 buffer_location : FIXME_30 { $$ = 0; }
993 /* Implementation of a dynamically expandable buffer for processing input
994 characters acquired through lexptr and building a value to return in
997 static char *tempbuf; /* Current buffer contents */
998 static int tempbufsize; /* Size of allocated buffer */
999 static int tempbufindex; /* Current index into buffer */
1001 #define GROWBY_MIN_SIZE 64 /* Minimum amount to grow buffer by */
1003 #define CHECKBUF(size) \
1005 if (tempbufindex + (size) >= tempbufsize) \
1007 growbuf_by_size (size); \
1011 /* Grow the static temp buffer if necessary, including allocating the first one
1015 growbuf_by_size (count)
1020 growby = max (count, GROWBY_MIN_SIZE);
1021 tempbufsize += growby;
1022 if (tempbuf == NULL)
1024 tempbuf = (char *) malloc (tempbufsize);
1028 tempbuf = (char *) realloc (tempbuf, tempbufsize);
1032 /* Try to consume a simple name string token. If successful, returns
1033 a pointer to a nullbyte terminated copy of the name that can be used
1034 in symbol table lookups. If not successful, returns NULL. */
1037 match_simple_name_string ()
1039 char *tokptr = lexptr;
1041 if (isalpha (*tokptr))
1046 } while (isalnum (*tokptr) || (*tokptr == '_'));
1047 yylval.sval.ptr = lexptr;
1048 yylval.sval.length = tokptr - lexptr;
1050 result = copy_name (yylval.sval);
1051 for (tokptr = result; *tokptr; tokptr++)
1052 if (isupper (*tokptr))
1053 *tokptr = tolower(*tokptr);
1059 /* Start looking for a value composed of valid digits as set by the base
1060 in use. Note that '_' characters are valid anywhere, in any quantity,
1061 and are simply ignored. Since we must find at least one valid digit,
1062 or reject this token as an integer literal, we keep track of how many
1063 digits we have encountered. */
1066 decode_integer_value (base, tokptrptr, ivalptr)
1071 char *tokptr = *tokptrptr;
1075 while (*tokptr != '\0')
1077 temp = tolower (*tokptr);
1083 case '0': case '1': case '2': case '3': case '4':
1084 case '5': case '6': case '7': case '8': case '9':
1087 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
1103 /* Found something not in domain for current base. */
1104 tokptr--; /* Unconsume what gave us indigestion. */
1109 /* If we didn't find any digits, then we don't have a valid integer
1110 value, so reject the entire token. Otherwise, update the lexical
1111 scan pointer, and return non-zero for success. */
1119 *tokptrptr = tokptr;
1125 decode_integer_literal (valptr, tokptrptr)
1129 char *tokptr = *tokptrptr;
1132 int explicit_base = 0;
1134 /* Look for an explicit base specifier, which is optional. */
1167 /* If we found an explicit base ensure that the character after the
1168 explicit base is a single quote. */
1170 if (explicit_base && (*tokptr++ != '\''))
1175 /* Attempt to decode whatever follows as an integer value in the
1176 indicated base, updating the token pointer in the process and
1177 computing the value into ival. Also, if we have an explicit
1178 base, then the next character must not be a single quote, or we
1179 have a bitstring literal, so reject the entire token in this case.
1180 Otherwise, update the lexical scan pointer, and return non-zero
1183 if (!decode_integer_value (base, &tokptr, &ival))
1187 else if (explicit_base && (*tokptr == '\''))
1194 *tokptrptr = tokptr;
1199 /* If it wasn't for the fact that floating point values can contain '_'
1200 characters, we could just let strtod do all the hard work by letting it
1201 try to consume as much of the current token buffer as possible and
1202 find a legal conversion. Unfortunately we need to filter out the '_'
1203 characters before calling strtod, which we do by copying the other
1204 legal chars to a local buffer to be converted. However since we also
1205 need to keep track of where the last unconsumed character in the input
1206 buffer is, we have transfer only as many characters as may compose a
1207 legal floating point value. */
1210 match_float_literal ()
1212 char *tokptr = lexptr;
1217 extern double strtod ();
1219 /* Make local buffer in which to build the string to convert. This is
1220 required because underscores are valid in chill floating point numbers
1221 but not in the string passed to strtod to convert. The string will be
1222 no longer than our input string. */
1224 copy = buf = (char *) alloca (strlen (tokptr) + 1);
1226 /* Transfer all leading digits to the conversion buffer, discarding any
1229 while (isdigit (*tokptr) || *tokptr == '_')
1238 /* Now accept either a '.', or one of [eEdD]. Dot is legal regardless
1239 of whether we found any leading digits, and we simply accept it and
1240 continue on to look for the fractional part and/or exponent. One of
1241 [eEdD] is legal only if we have seen digits, and means that there
1242 is no fractional part. If we find neither of these, then this is
1243 not a floating point number, so return failure. */
1248 /* Accept and then look for fractional part and/or exponent. */
1261 goto collect_exponent;
1269 /* We found a '.', copy any fractional digits to the conversion buffer, up
1270 to the first nondigit, non-underscore character. */
1272 while (isdigit (*tokptr) || *tokptr == '_')
1281 /* Look for an exponent, which must start with one of [eEdD]. If none
1282 is found, jump directly to trying to convert what we have collected
1299 /* Accept an optional '-' or '+' following one of [eEdD]. */
1302 if (*tokptr == '+' || *tokptr == '-')
1304 *copy++ = *tokptr++;
1307 /* Now copy an exponent into the conversion buffer. Note that at the
1308 moment underscores are *not* allowed in exponents. */
1310 while (isdigit (*tokptr))
1312 *copy++ = *tokptr++;
1315 /* If we transfered any chars to the conversion buffer, try to interpret its
1316 contents as a floating point value. If any characters remain, then we
1317 must not have a valid floating point string. */
1323 dval = strtod (buf, ©);
1328 return (FLOAT_LITERAL);
1334 /* Recognize a string literal. A string literal is a nonzero sequence
1335 of characters enclosed in matching single or double quotes, except that
1336 a single character inside single quotes is a character literal, which
1337 we reject as a string literal. To embed the terminator character inside
1338 a string, it is simply doubled (I.E. "this""is""one""string") */
1341 match_string_literal ()
1343 char *tokptr = lexptr;
1345 for (tempbufindex = 0, tokptr++; *tokptr != '\0'; tokptr++)
1348 if (*tokptr == *lexptr)
1350 if (*(tokptr + 1) == *lexptr)
1359 tempbuf[tempbufindex++] = *tokptr;
1361 if (*tokptr == '\0' /* no terminator */
1362 || tempbufindex == 0 /* no string */
1363 || (tempbufindex == 1 && *tokptr == '\'')) /* char literal */
1369 tempbuf[tempbufindex] = '\0';
1370 yylval.sval.ptr = tempbuf;
1371 yylval.sval.length = tempbufindex;
1373 return (CHARACTER_STRING_LITERAL);
1377 /* Recognize a character literal. A character literal is single character
1378 or a control sequence, enclosed in single quotes. A control sequence
1379 is a comma separated list of one or more integer literals, enclosed
1380 in parenthesis and introduced with a circumflex character.
1382 EX: 'a' '^(7)' '^(7,8)'
1384 As a GNU chill extension, the syntax C'xx' is also recognized as a
1385 character literal, where xx is a hex value for the character.
1387 Note that more than a single character, enclosed in single quotes, is
1390 Also note that the control sequence form is not in GNU Chill since it
1391 is ambiguous with the string literal form using single quotes. I.E.
1392 is '^(7)' a character literal or a string literal. In theory it it
1393 possible to tell by context, but GNU Chill doesn't accept the control
1394 sequence form, so neither do we (for now the code is disabled).
1396 Returns CHARACTER_LITERAL if a match is found.
1400 match_character_literal ()
1402 char *tokptr = lexptr;
1405 if ((tolower (*tokptr) == 'c') && (*(tokptr + 1) == '\''))
1407 /* We have a GNU chill extension form, so skip the leading "C'",
1408 decode the hex value, and then ensure that we have a trailing
1409 single quote character. */
1411 if (!decode_integer_value (16, &tokptr, &ival) || (*tokptr != '\''))
1417 else if (*tokptr == '\'')
1421 /* Determine which form we have, either a control sequence or the
1422 single character form. */
1424 if ((*tokptr == '^') && (*(tokptr + 1) == '('))
1426 #if 0 /* Disable, see note above. -fnf */
1427 /* Match and decode a control sequence. Return zero if we don't
1428 find a valid integer literal, or if the next unconsumed character
1429 after the integer literal is not the trailing ')'.
1430 FIXME: We currently don't handle the multiple integer literal
1433 if (!decode_integer_literal (&ival, &tokptr) || (*tokptr++ != ')'))
1446 /* The trailing quote has not yet been consumed. If we don't find
1447 it, then we have no match. */
1449 if (*tokptr++ != '\'')
1456 /* Not a character literal. */
1459 yylval.typed_val.val = ival;
1460 yylval.typed_val.type = builtin_type_chill_char;
1462 return (CHARACTER_LITERAL);
1465 /* Recognize an integer literal, as specified in Z.200 sec 5.2.4.2.
1466 Note that according to 5.2.4.2, a single "_" is also a valid integer
1467 literal, however GNU-chill requires there to be at least one "digit"
1468 in any integer literal. */
1471 match_integer_literal ()
1473 char *tokptr = lexptr;
1476 if (!decode_integer_literal (&ival, &tokptr))
1482 yylval.typed_val.val = ival;
1483 yylval.typed_val.type = builtin_type_int;
1485 return (INTEGER_LITERAL);
1489 /* Recognize a bit-string literal, as specified in Z.200 sec 5.2.4.8
1490 Note that according to 5.2.4.8, a single "_" is also a valid bit-string
1491 literal, however GNU-chill requires there to be at least one "digit"
1492 in any bit-string literal. */
1495 match_bitstring_literal ()
1497 char *tokptr = lexptr;
1506 /* Look for the required explicit base specifier. */
1527 /* Ensure that the character after the explicit base is a single quote. */
1529 if (*tokptr++ != '\'')
1534 while (*tokptr != '\0' && *tokptr != '\'')
1536 digit = tolower (*tokptr);
1542 case '0': case '1': case '2': case '3': case '4':
1543 case '5': case '6': case '7': case '8': case '9':
1546 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
1556 /* Found something not in domain for current base. */
1561 /* Extract bits from digit, starting with the msbit appropriate for
1562 the current base, and packing them into the bitstring byte,
1563 starting at the lsbit. */
1564 for (mask = (base >> 1); mask > 0; mask >>= 1)
1570 tempbuf[tempbufindex] |= (1 << bitoffset);
1573 if (bitoffset == HOST_CHAR_BIT)
1582 /* Verify that we consumed everything up to the trailing single quote,
1583 and that we found some bits (IE not just underbars). */
1585 if (*tokptr++ != '\'')
1591 yylval.sval.ptr = tempbuf;
1592 yylval.sval.length = bitcount;
1594 return (BIT_STRING_LITERAL);
1598 /* Recognize tokens that start with '$'. These include:
1600 $regname A native register name or a "standard
1602 Return token GDB_REGNAME.
1604 $variable A convenience variable with a name chosen
1606 Return token GDB_VARIABLE.
1608 $digits Value history with index <digits>, starting
1609 from the first value which has index 1.
1612 $$digits Value history with index <digits> relative
1613 to the last value. I.E. $$0 is the last
1614 value, $$1 is the one previous to that, $$2
1615 is the one previous to $$1, etc.
1616 Return token GDB_LAST.
1618 $ | $0 | $$0 The last value in the value history.
1619 Return token GDB_LAST.
1621 $$ An abbreviation for the second to the last
1622 value in the value history, I.E. $$1
1623 Return token GDB_LAST.
1625 Note that we currently assume that register names and convenience
1626 variables follow the convention of starting with a letter or '_'.
1631 match_dollar_tokens ()
1639 /* We will always have a successful match, even if it is just for
1640 a single '$', the abbreviation for $$0. So advance lexptr. */
1644 if (*tokptr == '_' || isalpha (*tokptr))
1646 /* Look for a match with a native register name, usually something
1647 like "r0" for example. */
1649 for (regno = 0; regno < NUM_REGS; regno++)
1651 namelength = strlen (reg_names[regno]);
1652 if (STREQN (tokptr, reg_names[regno], namelength)
1653 && !isalnum (tokptr[namelength]))
1655 yylval.lval = regno;
1656 lexptr += namelength + 1;
1657 return (GDB_REGNAME);
1661 /* Look for a match with a standard register name, usually something
1662 like "pc", which gdb always recognizes as the program counter
1663 regardless of what the native register name is. */
1665 for (regno = 0; regno < num_std_regs; regno++)
1667 namelength = strlen (std_regs[regno].name);
1668 if (STREQN (tokptr, std_regs[regno].name, namelength)
1669 && !isalnum (tokptr[namelength]))
1671 yylval.lval = std_regs[regno].regnum;
1672 lexptr += namelength;
1673 return (GDB_REGNAME);
1677 /* Attempt to match against a convenience variable. Note that
1678 this will always succeed, because if no variable of that name
1679 already exists, the lookup_internalvar will create one for us.
1680 Also note that both lexptr and tokptr currently point to the
1681 start of the input string we are trying to match, and that we
1682 have already tested the first character for non-numeric, so we
1683 don't have to treat it specially. */
1685 while (*tokptr == '_' || isalnum (*tokptr))
1689 yylval.sval.ptr = lexptr;
1690 yylval.sval.length = tokptr - lexptr;
1691 yylval.ivar = lookup_internalvar (copy_name (yylval.sval));
1693 return (GDB_VARIABLE);
1696 /* Since we didn't match against a register name or convenience
1697 variable, our only choice left is a history value. */
1711 /* Attempt to decode more characters as an integer value giving
1712 the index in the history list. If successful, the value will
1713 overwrite ival (currently 0 or 1), and if not, ival will be
1714 left alone, which is good since it is currently correct for
1715 the '$' or '$$' case. */
1717 decode_integer_literal (&ival, &tokptr);
1718 yylval.lval = negate ? -ival : ival;
1729 static const struct token idtokentab[] =
1731 { "length", LENGTH },
1742 { "max", MAX_TOKEN },
1743 { "min", MIN_TOKEN },
1753 static const struct token tokentab2[] =
1755 { ":=", GDB_ASSIGNMENT },
1756 { "//", SLASH_SLASH },
1763 /* Read one token, getting characters through lexptr. */
1764 /* This is where we will check to make sure that the language and the
1765 operators used are compatible. */
1775 /* Skip over any leading whitespace. */
1776 while (isspace (*lexptr))
1780 /* Look for special single character cases which can't be the first
1781 character of some other multicharacter token. */
1798 /* Look for characters which start a particular kind of multicharacter
1799 token, such as a character literal, register name, convenience
1800 variable name, string literal, etc. */
1805 /* First try to match a string literal, which is any nonzero
1806 sequence of characters enclosed in matching single or double
1807 quotes, except that a single character inside single quotes
1808 is a character literal, so we have to catch that case also. */
1809 token = match_string_literal ();
1814 if (*lexptr == '\'')
1816 token = match_character_literal ();
1825 token = match_character_literal ();
1832 token = match_dollar_tokens ();
1839 /* See if it is a special token of length 2. */
1840 for (i = 0; i < sizeof (tokentab2) / sizeof (tokentab2[0]); i++)
1842 if (STREQN (lexptr, tokentab2[i].operator, 2))
1845 return (tokentab2[i].token);
1848 /* Look for single character cases which which could be the first
1849 character of some other multicharacter token, but aren't, or we
1850 would already have found it. */
1860 /* Look for a float literal before looking for an integer literal, so
1861 we match as much of the input stream as possible. */
1862 token = match_float_literal ();
1867 token = match_bitstring_literal ();
1872 token = match_integer_literal ();
1878 /* Try to match a simple name string, and if a match is found, then
1879 further classify what sort of name it is and return an appropriate
1880 token. Note that attempting to match a simple name string consumes
1881 the token from lexptr, so we can't back out if we later find that
1882 we can't classify what sort of name it is. */
1884 simplename = match_simple_name_string ();
1886 if (simplename != NULL)
1888 /* See if it is a reserved identifier. */
1889 for (i = 0; i < sizeof (idtokentab) / sizeof (idtokentab[0]); i++)
1891 if (STREQ (simplename, idtokentab[i].operator))
1893 return (idtokentab[i].token);
1897 /* Look for other special tokens. */
1898 if (STREQ (simplename, "true"))
1901 return (BOOLEAN_LITERAL);
1903 if (STREQ (simplename, "false"))
1906 return (BOOLEAN_LITERAL);
1909 sym = lookup_symbol (simplename, expression_context_block,
1910 VAR_NAMESPACE, (int *) NULL,
1911 (struct symtab **) NULL);
1914 yylval.ssym.stoken.ptr = NULL;
1915 yylval.ssym.stoken.length = 0;
1916 yylval.ssym.sym = sym;
1917 yylval.ssym.is_a_field_of_this = 0; /* FIXME, C++'ism */
1918 switch (SYMBOL_CLASS (sym))
1921 /* Found a procedure name. */
1922 return (GENERAL_PROCEDURE_NAME);
1924 /* Found a global or local static variable. */
1925 return (LOCATION_NAME);
1932 if (innermost_block == NULL
1933 || contained_in (block_found, innermost_block))
1935 innermost_block = block_found;
1937 return (LOCATION_NAME);
1941 return (LOCATION_NAME);
1944 yylval.tsym.type = SYMBOL_TYPE (sym);
1947 case LOC_CONST_BYTES:
1948 case LOC_OPTIMIZED_OUT:
1949 error ("Symbol \"%s\" names no location.", simplename);
1953 else if (!have_full_symbols () && !have_partial_symbols ())
1955 error ("No symbol table is loaded. Use the \"file\" command.");
1959 error ("No symbol \"%s\" in current context.", simplename);
1963 /* Catch single character tokens which are not part of some
1968 case '.': /* Not float for example. */
1970 while (isspace (*lexptr)) lexptr++;
1971 simplename = match_simple_name_string ();
1977 return (ILLEGAL_TOKEN);
1982 char *msg; /* unused */
1984 printf ("Parsing: %s\n", lexptr);
1987 error ("Invalid syntax in expression near character '%c'.", yychar);
1991 error ("Invalid syntax in expression");