1 /* YACC grammar for Chill expressions, for GDB.
2 Copyright (C) 1992 Free Software Foundation, Inc.
4 This file is part of GDB.
6 This program is free software; you can redistribute it and/or modify
7 it under the terms of the GNU General Public License as published by
8 the Free Software Foundation; either version 2 of the License, or
9 (at your option) any later version.
11 This program is distributed in the hope that it will be useful,
12 but WITHOUT ANY WARRANTY; without even the implied warranty of
13 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
14 GNU General Public License for more details.
16 You should have received a copy of the GNU General Public License
17 along with this program; if not, write to the Free Software
18 Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. */
20 /* Parse a Chill expression from text in a string,
21 and return the result as a struct expression pointer.
22 That structure contains arithmetic operations in reverse polish,
23 with constants represented by operations that are followed by special data.
24 See expression.h for the details of the format.
25 What is important here is that it can be built up sequentially
26 during the process of parsing; the lower levels of the tree always
27 come first in the result.
29 Note that malloc's and realloc's in this file are transformed to
30 xmalloc and xrealloc respectively by the same sed command in the
31 makefile that remaps any other malloc/realloc inserted by the parser
32 generator. Doing this with #defines and trying to control the interaction
33 with include files (<malloc.h> and <stdlib.h> for example) just became
34 too messy, particularly when such includes can be inserted at random
35 times by the parser generator.
37 Also note that the language accepted by this parser is more liberal
38 than the one accepted by an actual Chill compiler. For example, the
39 language rule that a simple name string can not be one of the reserved
40 simple name strings is not enforced (e.g "case" is not treated as a
41 reserved name). Another example is that Chill is a strongly typed
42 language, and certain expressions that violate the type constraints
43 may still be evaluated if gdb can do so in a meaningful manner, while
44 such expressions would be rejected by the compiler. The reason for
45 this more liberal behavior is the philosophy that the debugger
46 is intended to be a tool that is used by the programmer when things
47 go wrong, and as such, it should provide as few artificial barriers
48 to it's use as possible. If it can do something meaningful, even
49 something that violates language contraints that are enforced by the
50 compiler, it should do so without complaint.
58 #include "expression.h"
61 #include "parser-defs.h"
64 /* Remap normal yacc parser interface names (yyparse, yylex, yyerror, etc),
65 as well as gratuitiously global symbol names, so we can have multiple
66 yacc generated parsers in gdb. Note that these are only the variables
67 produced by yacc. If other parser generators (bison, byacc, etc) produce
68 additional global names that conflict at link time, then those parser
69 generators need to be fixed instead of adding those names to this list. */
71 #define yymaxdepth chill_maxdepth
72 #define yyparse chill_parse
73 #define yylex chill_lex
74 #define yyerror chill_error
75 #define yylval chill_lval
76 #define yychar chill_char
77 #define yydebug chill_debug
78 #define yypact chill_pact
81 #define yydef chill_def
82 #define yychk chill_chk
83 #define yypgo chill_pgo
84 #define yyact chill_act
85 #define yyexca chill_exca
86 #define yyerrflag chill_errflag
87 #define yynerrs chill_nerrs
91 #define yy_yys chill_yys
92 #define yystate chill_state
93 #define yytmp chill_tmp
95 #define yy_yyv chill_yyv
96 #define yyval chill_val
97 #define yylloc chill_lloc
98 #define yyreds chill_reds /* With YYDEBUG defined */
99 #define yytoks chill_toks /* With YYDEBUG defined */
102 #define YYDEBUG 0 /* Default to no yydebug support */
106 yyparse PARAMS ((void));
109 yylex PARAMS ((void));
112 yyerror PARAMS ((char *));
116 /* Although the yacc "value" of an expression is not used,
117 since the result is stored in the structure being created,
118 other node types do have values. */
123 unsigned LONGEST ulval;
133 struct symtoken ssym;
136 enum exp_opcode opcode;
137 struct internalvar *ivar;
143 %token <voidval> FIXME_01
144 %token <voidval> FIXME_02
145 %token <voidval> FIXME_03
146 %token <voidval> FIXME_04
147 %token <voidval> FIXME_05
148 %token <voidval> FIXME_06
149 %token <voidval> FIXME_07
150 %token <voidval> FIXME_08
151 %token <voidval> FIXME_09
152 %token <voidval> FIXME_10
153 %token <voidval> FIXME_11
154 %token <voidval> FIXME_12
155 %token <voidval> FIXME_13
156 %token <voidval> FIXME_14
157 %token <voidval> FIXME_15
158 %token <voidval> FIXME_16
159 %token <voidval> FIXME_17
160 %token <voidval> FIXME_18
161 %token <voidval> FIXME_19
162 %token <voidval> FIXME_20
163 %token <voidval> FIXME_21
164 %token <voidval> FIXME_22
165 %token <voidval> FIXME_23
166 %token <voidval> FIXME_24
167 %token <voidval> FIXME_25
168 %token <voidval> FIXME_26
169 %token <voidval> FIXME_27
170 %token <voidval> FIXME_28
171 %token <voidval> FIXME_29
172 %token <voidval> FIXME_30
174 %token <typed_val> INTEGER_LITERAL
175 %token <ulval> BOOLEAN_LITERAL
176 %token <typed_val> CHARACTER_LITERAL
177 %token <dval> FLOAT_LITERAL
178 %token <ssym> GENERAL_PROCEDURE_NAME
179 %token <ssym> LOCATION_NAME
180 %token <voidval> SET_LITERAL
181 %token <voidval> EMPTINESS_LITERAL
182 %token <sval> CHARACTER_STRING_LITERAL
183 %token <sval> BIT_STRING_LITERAL
184 %token <tsym> TYPENAME
185 %token <sval> FIELD_NAME
190 %token <voidval> CASE
192 %token <voidval> ESAC
193 %token <voidval> LOGIOR
194 %token <voidval> ORIF
195 %token <voidval> LOGXOR
196 %token <voidval> LOGAND
197 %token <voidval> ANDIF
199 %token <voidval> NOTEQUAL
209 %token <voidval> SLASH_SLASH
213 %token <voidval> POINTER
214 %token <voidval> RECEIVE
221 %token <voidval> THEN
222 %token <voidval> ELSE
224 %token <voidval> ELSIF
225 %token <voidval> ILLEGAL_TOKEN
227 %token <voidval> PRED
228 %token <voidval> SUCC
230 %token <voidval> CARD
233 %token <voidval> SIZE
234 %token <voidval> UPPER
235 %token <voidval> LOWER
236 %token <voidval> LENGTH
238 /* Tokens which are not Chill tokens used in expressions, but rather GDB
239 specific things that we recognize in the same context as Chill tokens
240 (register names for example). */
242 %token <lval> GDB_REGNAME /* Machine register name */
243 %token <lval> GDB_LAST /* Value history */
244 %token <ivar> GDB_VARIABLE /* Convenience variable */
245 %token <voidval> GDB_ASSIGNMENT /* Assign value to somewhere */
247 %type <voidval> location
248 %type <voidval> access_name
249 %type <voidval> primitive_value
250 %type <voidval> location_contents
251 %type <voidval> value_name
252 %type <voidval> literal
253 %type <voidval> tuple
254 %type <voidval> value_string_element
255 %type <voidval> value_string_slice
256 %type <voidval> value_array_element
257 %type <voidval> value_array_slice
258 %type <voidval> value_structure_field
259 %type <voidval> expression_conversion
260 %type <voidval> value_procedure_call
261 %type <voidval> value_built_in_routine_call
262 %type <voidval> chill_value_built_in_routine_call
263 %type <voidval> start_expression
264 %type <voidval> zero_adic_operator
265 %type <voidval> parenthesised_expression
266 %type <voidval> value
267 %type <voidval> undefined_value
268 %type <voidval> expression
269 %type <voidval> conditional_expression
270 %type <voidval> then_alternative
271 %type <voidval> else_alternative
272 %type <voidval> sub_expression
273 %type <voidval> value_case_alternative
274 %type <voidval> operand_0
275 %type <voidval> operand_1
276 %type <voidval> operand_2
277 %type <voidval> operand_3
278 %type <voidval> operand_4
279 %type <voidval> operand_5
280 %type <voidval> operand_6
281 %type <voidval> synonym_name
282 %type <voidval> value_enumeration_name
283 %type <voidval> value_do_with_name
284 %type <voidval> value_receive_name
285 %type <voidval> string_primitive_value
286 %type <voidval> start_element
287 %type <voidval> left_element
288 %type <voidval> right_element
289 %type <voidval> slice_size
290 %type <voidval> array_primitive_value
291 %type <voidval> expression_list
292 %type <voidval> lower_element
293 %type <voidval> upper_element
294 %type <voidval> first_element
295 %type <voidval> structure_primitive_value
296 %type <voidval> mode_argument
297 %type <voidval> upper_lower_argument
298 %type <voidval> length_argument
299 %type <voidval> array_mode_name
300 %type <voidval> string_mode_name
301 %type <voidval> variant_structure_mode_name
302 %type <voidval> boolean_expression
303 %type <voidval> case_selector_list
304 %type <voidval> subexpression
305 %type <voidval> case_label_specification
306 %type <voidval> buffer_location
307 %type <voidval> single_assignment_action
308 %type <tsym> mode_name
316 { write_exp_elt_opcode(OP_TYPE);
317 write_exp_elt_type($1.type);
318 write_exp_elt_opcode(OP_TYPE);}
331 undefined_value : FIXME_01
339 location : access_name
340 | primitive_value POINTER
342 write_exp_elt_opcode (UNOP_IND);
348 access_name : LOCATION_NAME
350 write_exp_elt_opcode (OP_VAR_VALUE);
351 write_exp_elt_sym ($1.sym);
352 write_exp_elt_opcode (OP_VAR_VALUE);
354 | GDB_LAST /* gdb specific */
356 write_exp_elt_opcode (OP_LAST);
357 write_exp_elt_longcst ($1);
358 write_exp_elt_opcode (OP_LAST);
360 | GDB_REGNAME /* gdb specific */
362 write_exp_elt_opcode (OP_REGISTER);
363 write_exp_elt_longcst ($1);
364 write_exp_elt_opcode (OP_REGISTER);
366 | GDB_VARIABLE /* gdb specific */
368 write_exp_elt_opcode (OP_INTERNALVAR);
369 write_exp_elt_intern ($1);
370 write_exp_elt_opcode (OP_INTERNALVAR);
380 expression_list : expression
384 | expression_list ',' expression
391 primitive_value : location_contents
407 | value_string_element
415 | value_array_element
423 | value_structure_field
427 | expression_conversion
431 | value_procedure_call
435 | value_built_in_routine_call
447 | parenthesised_expression
455 location_contents: location
463 value_name : synonym_name
467 | value_enumeration_name
479 | GENERAL_PROCEDURE_NAME
481 write_exp_elt_opcode (OP_VAR_VALUE);
482 write_exp_elt_sym ($1.sym);
483 write_exp_elt_opcode (OP_VAR_VALUE);
489 literal : INTEGER_LITERAL
491 write_exp_elt_opcode (OP_LONG);
492 write_exp_elt_type ($1.type);
493 write_exp_elt_longcst ((LONGEST) ($1.val));
494 write_exp_elt_opcode (OP_LONG);
498 write_exp_elt_opcode (OP_BOOL);
499 write_exp_elt_longcst ((LONGEST) $1);
500 write_exp_elt_opcode (OP_BOOL);
504 write_exp_elt_opcode (OP_LONG);
505 write_exp_elt_type ($1.type);
506 write_exp_elt_longcst ((LONGEST) ($1.val));
507 write_exp_elt_opcode (OP_LONG);
511 write_exp_elt_opcode (OP_DOUBLE);
512 write_exp_elt_type (builtin_type_double);
513 write_exp_elt_dblcst ($1);
514 write_exp_elt_opcode (OP_DOUBLE);
524 | CHARACTER_STRING_LITERAL
526 write_exp_elt_opcode (OP_STRING);
527 write_exp_string ($1);
528 write_exp_elt_opcode (OP_STRING);
532 write_exp_elt_opcode (OP_BITSTRING);
533 write_exp_bitstring ($1);
534 write_exp_elt_opcode (OP_BITSTRING);
549 value_string_element: string_primitive_value '(' start_element ')'
557 value_string_slice: string_primitive_value '(' left_element ':' right_element ')'
561 | string_primitive_value '(' start_element UP slice_size ')'
569 value_array_element: array_primitive_value '('
570 /* This is to save the value of arglist_len
571 being accumulated for each dimension. */
572 { start_arglist (); }
575 write_exp_elt_opcode (MULTI_SUBSCRIPT);
576 write_exp_elt_longcst ((LONGEST) end_arglist ());
577 write_exp_elt_opcode (MULTI_SUBSCRIPT);
583 value_array_slice: array_primitive_value '(' lower_element ':' upper_element ')'
587 | array_primitive_value '(' first_element UP slice_size ')'
595 value_structure_field: primitive_value FIELD_NAME
596 { write_exp_elt_opcode (STRUCTOP_STRUCT);
597 write_exp_string ($2);
598 write_exp_elt_opcode (STRUCTOP_STRUCT);
604 expression_conversion: mode_name parenthesised_expression
606 write_exp_elt_opcode (UNOP_CAST);
607 write_exp_elt_type ($1.type);
608 write_exp_elt_opcode (UNOP_CAST);
614 value_procedure_call: FIXME_05
622 value_built_in_routine_call: chill_value_built_in_routine_call
630 start_expression: FIXME_06
633 } /* Not in GNU-Chill */
638 zero_adic_operator: FIXME_07
646 parenthesised_expression: '(' expression ')'
654 expression : operand_0
658 | single_assignment_action
662 | conditional_expression
668 conditional_expression : IF boolean_expression then_alternative else_alternative FI
672 | CASE case_selector_list OF value_case_alternative '[' ELSE sub_expression ']' ESAC
678 then_alternative: THEN subexpression
684 else_alternative: ELSE subexpression
688 | ELSIF boolean_expression then_alternative else_alternative
694 sub_expression : expression
700 value_case_alternative: case_label_specification ':' sub_expression ';'
708 operand_0 : operand_1
712 | operand_0 LOGIOR operand_1
714 write_exp_elt_opcode (BINOP_BITWISE_IOR);
716 | operand_0 ORIF operand_1
720 | operand_0 LOGXOR operand_1
722 write_exp_elt_opcode (BINOP_BITWISE_XOR);
728 operand_1 : operand_2
732 | operand_1 LOGAND operand_2
734 write_exp_elt_opcode (BINOP_BITWISE_AND);
736 | operand_1 ANDIF operand_2
744 operand_2 : operand_3
748 | operand_2 '=' operand_3
750 write_exp_elt_opcode (BINOP_EQUAL);
752 | operand_2 NOTEQUAL operand_3
754 write_exp_elt_opcode (BINOP_NOTEQUAL);
756 | operand_2 '>' operand_3
758 write_exp_elt_opcode (BINOP_GTR);
760 | operand_2 GTR operand_3
762 write_exp_elt_opcode (BINOP_GEQ);
764 | operand_2 '<' operand_3
766 write_exp_elt_opcode (BINOP_LESS);
768 | operand_2 LEQ operand_3
770 write_exp_elt_opcode (BINOP_LEQ);
772 | operand_2 IN operand_3
781 operand_3 : operand_4
785 | operand_3 '+' operand_4
787 write_exp_elt_opcode (BINOP_ADD);
789 | operand_3 '-' operand_4
791 write_exp_elt_opcode (BINOP_SUB);
793 | operand_3 SLASH_SLASH operand_4
795 write_exp_elt_opcode (BINOP_CONCAT);
801 operand_4 : operand_5
805 | operand_4 '*' operand_5
807 write_exp_elt_opcode (BINOP_MUL);
809 | operand_4 '/' operand_5
811 write_exp_elt_opcode (BINOP_DIV);
813 | operand_4 MOD operand_5
815 write_exp_elt_opcode (BINOP_MOD);
817 | operand_4 REM operand_5
819 write_exp_elt_opcode (BINOP_REM);
825 operand_5 : operand_6
831 write_exp_elt_opcode (UNOP_NEG);
835 write_exp_elt_opcode (UNOP_LOGICAL_NOT);
837 | parenthesised_expression literal
838 /* We require the string operand to be a literal, to avoid some
839 nasty parsing ambiguities. */
841 write_exp_elt_opcode (BINOP_CONCAT);
847 operand_6 : POINTER location
849 write_exp_elt_opcode (UNOP_ADDR);
851 | RECEIVE buffer_location
864 single_assignment_action :
865 location GDB_ASSIGNMENT value
867 write_exp_elt_opcode (BINOP_ASSIGN);
873 chill_value_built_in_routine_call :
874 NUM '(' expression ')'
878 | PRED '(' expression ')'
882 | SUCC '(' expression ')'
886 | ABS '(' expression ')'
890 | CARD '(' expression ')'
894 | MAX '(' expression ')'
898 | MIN '(' expression ')'
902 | SIZE '(' location ')'
906 | SIZE '(' mode_argument ')'
910 | UPPER '(' upper_lower_argument ')'
914 | LOWER '(' upper_lower_argument ')'
918 | LENGTH '(' length_argument ')'
924 mode_argument : mode_name
928 | array_mode_name '(' expression ')'
932 | string_mode_name '(' expression ')'
936 | variant_structure_mode_name '(' expression_list ')'
945 upper_lower_argument : expression
955 length_argument : expression
963 array_primitive_value : primitive_value
970 /* Things which still need productions... */
972 array_mode_name : FIXME_08 { $$ = 0; }
973 string_mode_name : FIXME_09 { $$ = 0; }
974 variant_structure_mode_name: FIXME_10 { $$ = 0; }
975 synonym_name : FIXME_11 { $$ = 0; }
976 value_enumeration_name : FIXME_12 { $$ = 0; }
977 value_do_with_name : FIXME_13 { $$ = 0; }
978 value_receive_name : FIXME_14 { $$ = 0; }
979 string_primitive_value : FIXME_15 { $$ = 0; }
980 start_element : FIXME_16 { $$ = 0; }
981 left_element : FIXME_17 { $$ = 0; }
982 right_element : FIXME_18 { $$ = 0; }
983 slice_size : FIXME_19 { $$ = 0; }
984 lower_element : FIXME_20 { $$ = 0; }
985 upper_element : FIXME_21 { $$ = 0; }
986 first_element : FIXME_22 { $$ = 0; }
987 structure_primitive_value: FIXME_23 { $$ = 0; }
988 boolean_expression : FIXME_26 { $$ = 0; }
989 case_selector_list : FIXME_27 { $$ = 0; }
990 subexpression : FIXME_28 { $$ = 0; }
991 case_label_specification: FIXME_29 { $$ = 0; }
992 buffer_location : FIXME_30 { $$ = 0; }
996 /* Implementation of a dynamically expandable buffer for processing input
997 characters acquired through lexptr and building a value to return in
1000 static char *tempbuf; /* Current buffer contents */
1001 static int tempbufsize; /* Size of allocated buffer */
1002 static int tempbufindex; /* Current index into buffer */
1004 #define GROWBY_MIN_SIZE 64 /* Minimum amount to grow buffer by */
1006 #define CHECKBUF(size) \
1008 if (tempbufindex + (size) >= tempbufsize) \
1010 growbuf_by_size (size); \
1014 /* Grow the static temp buffer if necessary, including allocating the first one
1018 growbuf_by_size (count)
1023 growby = max (count, GROWBY_MIN_SIZE);
1024 tempbufsize += growby;
1025 if (tempbuf == NULL)
1027 tempbuf = (char *) malloc (tempbufsize);
1031 tempbuf = (char *) realloc (tempbuf, tempbufsize);
1035 /* Try to consume a simple name string token. If successful, returns
1036 a pointer to a nullbyte terminated copy of the name that can be used
1037 in symbol table lookups. If not successful, returns NULL. */
1040 match_simple_name_string ()
1042 char *tokptr = lexptr;
1044 if (isalpha (*tokptr))
1049 } while (isalnum (*tokptr) || (*tokptr == '_'));
1050 yylval.sval.ptr = lexptr;
1051 yylval.sval.length = tokptr - lexptr;
1053 result = copy_name (yylval.sval);
1054 for (tokptr = result; *tokptr; tokptr++)
1055 if (isupper (*tokptr))
1056 *tokptr = tolower(*tokptr);
1062 /* Start looking for a value composed of valid digits as set by the base
1063 in use. Note that '_' characters are valid anywhere, in any quantity,
1064 and are simply ignored. Since we must find at least one valid digit,
1065 or reject this token as an integer literal, we keep track of how many
1066 digits we have encountered. */
1069 decode_integer_value (base, tokptrptr, ivalptr)
1074 char *tokptr = *tokptrptr;
1078 while (*tokptr != '\0')
1080 temp = tolower (*tokptr);
1086 case '0': case '1': case '2': case '3': case '4':
1087 case '5': case '6': case '7': case '8': case '9':
1090 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
1106 /* Found something not in domain for current base. */
1107 tokptr--; /* Unconsume what gave us indigestion. */
1112 /* If we didn't find any digits, then we don't have a valid integer
1113 value, so reject the entire token. Otherwise, update the lexical
1114 scan pointer, and return non-zero for success. */
1122 *tokptrptr = tokptr;
1128 decode_integer_literal (valptr, tokptrptr)
1132 char *tokptr = *tokptrptr;
1135 int explicit_base = 0;
1137 /* Look for an explicit base specifier, which is optional. */
1170 /* If we found an explicit base ensure that the character after the
1171 explicit base is a single quote. */
1173 if (explicit_base && (*tokptr++ != '\''))
1178 /* Attempt to decode whatever follows as an integer value in the
1179 indicated base, updating the token pointer in the process and
1180 computing the value into ival. Also, if we have an explicit
1181 base, then the next character must not be a single quote, or we
1182 have a bitstring literal, so reject the entire token in this case.
1183 Otherwise, update the lexical scan pointer, and return non-zero
1186 if (!decode_integer_value (base, &tokptr, &ival))
1190 else if (explicit_base && (*tokptr == '\''))
1197 *tokptrptr = tokptr;
1202 /* If it wasn't for the fact that floating point values can contain '_'
1203 characters, we could just let strtod do all the hard work by letting it
1204 try to consume as much of the current token buffer as possible and
1205 find a legal conversion. Unfortunately we need to filter out the '_'
1206 characters before calling strtod, which we do by copying the other
1207 legal chars to a local buffer to be converted. However since we also
1208 need to keep track of where the last unconsumed character in the input
1209 buffer is, we have transfer only as many characters as may compose a
1210 legal floating point value. */
1213 match_float_literal ()
1215 char *tokptr = lexptr;
1220 extern double strtod ();
1222 /* Make local buffer in which to build the string to convert. This is
1223 required because underscores are valid in chill floating point numbers
1224 but not in the string passed to strtod to convert. The string will be
1225 no longer than our input string. */
1227 copy = buf = (char *) alloca (strlen (tokptr) + 1);
1229 /* Transfer all leading digits to the conversion buffer, discarding any
1232 while (isdigit (*tokptr) || *tokptr == '_')
1241 /* Now accept either a '.', or one of [eEdD]. Dot is legal regardless
1242 of whether we found any leading digits, and we simply accept it and
1243 continue on to look for the fractional part and/or exponent. One of
1244 [eEdD] is legal only if we have seen digits, and means that there
1245 is no fractional part. If we find neither of these, then this is
1246 not a floating point number, so return failure. */
1251 /* Accept and then look for fractional part and/or exponent. */
1264 goto collect_exponent;
1272 /* We found a '.', copy any fractional digits to the conversion buffer, up
1273 to the first nondigit, non-underscore character. */
1275 while (isdigit (*tokptr) || *tokptr == '_')
1284 /* Look for an exponent, which must start with one of [eEdD]. If none
1285 is found, jump directly to trying to convert what we have collected
1302 /* Accept an optional '-' or '+' following one of [eEdD]. */
1305 if (*tokptr == '+' || *tokptr == '-')
1307 *copy++ = *tokptr++;
1310 /* Now copy an exponent into the conversion buffer. Note that at the
1311 moment underscores are *not* allowed in exponents. */
1313 while (isdigit (*tokptr))
1315 *copy++ = *tokptr++;
1318 /* If we transfered any chars to the conversion buffer, try to interpret its
1319 contents as a floating point value. If any characters remain, then we
1320 must not have a valid floating point string. */
1326 dval = strtod (buf, ©);
1331 return (FLOAT_LITERAL);
1337 /* Recognize a string literal. A string literal is a nonzero sequence
1338 of characters enclosed in matching single or double quotes, except that
1339 a single character inside single quotes is a character literal, which
1340 we reject as a string literal. To embed the terminator character inside
1341 a string, it is simply doubled (I.E. "this""is""one""string") */
1344 match_string_literal ()
1346 char *tokptr = lexptr;
1348 for (tempbufindex = 0, tokptr++; *tokptr != '\0'; tokptr++)
1351 if (*tokptr == *lexptr)
1353 if (*(tokptr + 1) == *lexptr)
1362 tempbuf[tempbufindex++] = *tokptr;
1364 if (*tokptr == '\0' /* no terminator */
1365 || tempbufindex == 0 /* no string */
1366 || (tempbufindex == 1 && *tokptr == '\'')) /* char literal */
1372 tempbuf[tempbufindex] = '\0';
1373 yylval.sval.ptr = tempbuf;
1374 yylval.sval.length = tempbufindex;
1376 return (CHARACTER_STRING_LITERAL);
1380 /* Recognize a character literal. A character literal is single character
1381 or a control sequence, enclosed in single quotes. A control sequence
1382 is a comma separated list of one or more integer literals, enclosed
1383 in parenthesis and introduced with a circumflex character.
1385 EX: 'a' '^(7)' '^(7,8)'
1387 As a GNU chill extension, the syntax C'xx' is also recognized as a
1388 character literal, where xx is a hex value for the character.
1390 Note that more than a single character, enclosed in single quotes, is
1393 Also note that the control sequence form is not in GNU Chill since it
1394 is ambiguous with the string literal form using single quotes. I.E.
1395 is '^(7)' a character literal or a string literal. In theory it it
1396 possible to tell by context, but GNU Chill doesn't accept the control
1397 sequence form, so neither do we (for now the code is disabled).
1399 Returns CHARACTER_LITERAL if a match is found.
1403 match_character_literal ()
1405 char *tokptr = lexptr;
1408 if ((tolower (*tokptr) == 'c') && (*(tokptr + 1) == '\''))
1410 /* We have a GNU chill extension form, so skip the leading "C'",
1411 decode the hex value, and then ensure that we have a trailing
1412 single quote character. */
1414 if (!decode_integer_value (16, &tokptr, &ival) || (*tokptr != '\''))
1420 else if (*tokptr == '\'')
1424 /* Determine which form we have, either a control sequence or the
1425 single character form. */
1427 if ((*tokptr == '^') && (*(tokptr + 1) == '('))
1429 #if 0 /* Disable, see note above. -fnf */
1430 /* Match and decode a control sequence. Return zero if we don't
1431 find a valid integer literal, or if the next unconsumed character
1432 after the integer literal is not the trailing ')'.
1433 FIXME: We currently don't handle the multiple integer literal
1436 if (!decode_integer_literal (&ival, &tokptr) || (*tokptr++ != ')'))
1449 /* The trailing quote has not yet been consumed. If we don't find
1450 it, then we have no match. */
1452 if (*tokptr++ != '\'')
1459 /* Not a character literal. */
1462 yylval.typed_val.val = ival;
1463 yylval.typed_val.type = builtin_type_chill_char;
1465 return (CHARACTER_LITERAL);
1468 /* Recognize an integer literal, as specified in Z.200 sec 5.2.4.2.
1469 Note that according to 5.2.4.2, a single "_" is also a valid integer
1470 literal, however GNU-chill requires there to be at least one "digit"
1471 in any integer literal. */
1474 match_integer_literal ()
1476 char *tokptr = lexptr;
1479 if (!decode_integer_literal (&ival, &tokptr))
1485 yylval.typed_val.val = ival;
1486 yylval.typed_val.type = builtin_type_int;
1488 return (INTEGER_LITERAL);
1492 /* Recognize a bit-string literal, as specified in Z.200 sec 5.2.4.8
1493 Note that according to 5.2.4.8, a single "_" is also a valid bit-string
1494 literal, however GNU-chill requires there to be at least one "digit"
1495 in any bit-string literal. */
1498 match_bitstring_literal ()
1500 char *tokptr = lexptr;
1509 /* Look for the required explicit base specifier. */
1530 /* Ensure that the character after the explicit base is a single quote. */
1532 if (*tokptr++ != '\'')
1537 while (*tokptr != '\0' && *tokptr != '\'')
1539 digit = tolower (*tokptr);
1545 case '0': case '1': case '2': case '3': case '4':
1546 case '5': case '6': case '7': case '8': case '9':
1549 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
1559 /* Found something not in domain for current base. */
1564 /* Extract bits from digit, starting with the msbit appropriate for
1565 the current base, and packing them into the bitstring byte,
1566 starting at the lsbit. */
1567 for (mask = (base >> 1); mask > 0; mask >>= 1)
1573 tempbuf[tempbufindex] |= (1 << bitoffset);
1576 if (bitoffset == HOST_CHAR_BIT)
1585 /* Verify that we consumed everything up to the trailing single quote,
1586 and that we found some bits (IE not just underbars). */
1588 if (*tokptr++ != '\'')
1594 yylval.sval.ptr = tempbuf;
1595 yylval.sval.length = bitcount;
1597 return (BIT_STRING_LITERAL);
1601 /* Recognize tokens that start with '$'. These include:
1603 $regname A native register name or a "standard
1605 Return token GDB_REGNAME.
1607 $variable A convenience variable with a name chosen
1609 Return token GDB_VARIABLE.
1611 $digits Value history with index <digits>, starting
1612 from the first value which has index 1.
1615 $$digits Value history with index <digits> relative
1616 to the last value. I.E. $$0 is the last
1617 value, $$1 is the one previous to that, $$2
1618 is the one previous to $$1, etc.
1619 Return token GDB_LAST.
1621 $ | $0 | $$0 The last value in the value history.
1622 Return token GDB_LAST.
1624 $$ An abbreviation for the second to the last
1625 value in the value history, I.E. $$1
1626 Return token GDB_LAST.
1628 Note that we currently assume that register names and convenience
1629 variables follow the convention of starting with a letter or '_'.
1634 match_dollar_tokens ()
1642 /* We will always have a successful match, even if it is just for
1643 a single '$', the abbreviation for $$0. So advance lexptr. */
1647 if (*tokptr == '_' || isalpha (*tokptr))
1649 /* Look for a match with a native register name, usually something
1650 like "r0" for example. */
1652 for (regno = 0; regno < NUM_REGS; regno++)
1654 namelength = strlen (reg_names[regno]);
1655 if (STREQN (tokptr, reg_names[regno], namelength)
1656 && !isalnum (tokptr[namelength]))
1658 yylval.lval = regno;
1659 lexptr += namelength + 1;
1660 return (GDB_REGNAME);
1664 /* Look for a match with a standard register name, usually something
1665 like "pc", which gdb always recognizes as the program counter
1666 regardless of what the native register name is. */
1668 for (regno = 0; regno < num_std_regs; regno++)
1670 namelength = strlen (std_regs[regno].name);
1671 if (STREQN (tokptr, std_regs[regno].name, namelength)
1672 && !isalnum (tokptr[namelength]))
1674 yylval.lval = std_regs[regno].regnum;
1675 lexptr += namelength;
1676 return (GDB_REGNAME);
1680 /* Attempt to match against a convenience variable. Note that
1681 this will always succeed, because if no variable of that name
1682 already exists, the lookup_internalvar will create one for us.
1683 Also note that both lexptr and tokptr currently point to the
1684 start of the input string we are trying to match, and that we
1685 have already tested the first character for non-numeric, so we
1686 don't have to treat it specially. */
1688 while (*tokptr == '_' || isalnum (*tokptr))
1692 yylval.sval.ptr = lexptr;
1693 yylval.sval.length = tokptr - lexptr;
1694 yylval.ivar = lookup_internalvar (copy_name (yylval.sval));
1696 return (GDB_VARIABLE);
1699 /* Since we didn't match against a register name or convenience
1700 variable, our only choice left is a history value. */
1714 /* Attempt to decode more characters as an integer value giving
1715 the index in the history list. If successful, the value will
1716 overwrite ival (currently 0 or 1), and if not, ival will be
1717 left alone, which is good since it is currently correct for
1718 the '$' or '$$' case. */
1720 decode_integer_literal (&ival, &tokptr);
1721 yylval.lval = negate ? -ival : ival;
1732 static const struct token idtokentab[] =
1734 { "length", LENGTH },
1756 static const struct token tokentab2[] =
1758 { ":=", GDB_ASSIGNMENT },
1759 { "//", SLASH_SLASH },
1766 /* Read one token, getting characters through lexptr. */
1767 /* This is where we will check to make sure that the language and the
1768 operators used are compatible. */
1778 /* Skip over any leading whitespace. */
1779 while (isspace (*lexptr))
1783 /* Look for special single character cases which can't be the first
1784 character of some other multicharacter token. */
1801 /* Look for characters which start a particular kind of multicharacter
1802 token, such as a character literal, register name, convenience
1803 variable name, string literal, etc. */
1808 /* First try to match a string literal, which is any nonzero
1809 sequence of characters enclosed in matching single or double
1810 quotes, except that a single character inside single quotes
1811 is a character literal, so we have to catch that case also. */
1812 token = match_string_literal ();
1817 if (*lexptr == '\'')
1819 token = match_character_literal ();
1828 token = match_character_literal ();
1835 token = match_dollar_tokens ();
1842 /* See if it is a special token of length 2. */
1843 for (i = 0; i < sizeof (tokentab2) / sizeof (tokentab2[0]); i++)
1845 if (STREQN (lexptr, tokentab2[i].operator, 2))
1848 return (tokentab2[i].token);
1851 /* Look for single character cases which which could be the first
1852 character of some other multicharacter token, but aren't, or we
1853 would already have found it. */
1863 /* Look for a float literal before looking for an integer literal, so
1864 we match as much of the input stream as possible. */
1865 token = match_float_literal ();
1870 token = match_bitstring_literal ();
1875 token = match_integer_literal ();
1881 /* Try to match a simple name string, and if a match is found, then
1882 further classify what sort of name it is and return an appropriate
1883 token. Note that attempting to match a simple name string consumes
1884 the token from lexptr, so we can't back out if we later find that
1885 we can't classify what sort of name it is. */
1887 simplename = match_simple_name_string ();
1889 /* See if it is a reserved identifier. */
1890 for (i = 0; i < sizeof (idtokentab) / sizeof (idtokentab[0]); i++)
1892 if (STREQ (simplename, idtokentab[i].operator))
1894 return (idtokentab[i].token);
1898 /* Look for other special tokens. */
1899 if (STREQ (simplename, "true"))
1902 return (BOOLEAN_LITERAL);
1904 if (STREQ (simplename, "false"))
1907 return (BOOLEAN_LITERAL);
1910 if (simplename != NULL)
1912 sym = lookup_symbol (simplename, expression_context_block,
1913 VAR_NAMESPACE, (int *) NULL,
1914 (struct symtab **) NULL);
1917 yylval.ssym.stoken.ptr = NULL;
1918 yylval.ssym.stoken.length = 0;
1919 yylval.ssym.sym = sym;
1920 yylval.ssym.is_a_field_of_this = 0; /* FIXME, C++'ism */
1921 switch (SYMBOL_CLASS (sym))
1924 /* Found a procedure name. */
1925 return (GENERAL_PROCEDURE_NAME);
1927 /* Found a global or local static variable. */
1928 return (LOCATION_NAME);
1935 if (innermost_block == NULL
1936 || contained_in (block_found, innermost_block))
1938 innermost_block = block_found;
1940 return (LOCATION_NAME);
1944 return (LOCATION_NAME);
1947 yylval.tsym.type = SYMBOL_TYPE (sym);
1950 case LOC_CONST_BYTES:
1951 case LOC_OPTIMIZED_OUT:
1952 error ("Symbol \"%s\" names no location.", simplename);
1956 else if (!have_full_symbols () && !have_partial_symbols ())
1958 error ("No symbol table is loaded. Use the \"file\" command.");
1962 error ("No symbol \"%s\" in current context.", simplename);
1966 /* Catch single character tokens which are not part of some
1971 case '.': /* Not float for example. */
1973 while (isspace (*lexptr)) lexptr++;
1974 simplename = match_simple_name_string ();
1980 return (ILLEGAL_TOKEN);
1985 char *msg; /* unused */
1987 printf ("Parsing: %s\n", lexptr);
1990 error ("Invalid syntax in expression near character '%c'.", yychar);
1994 error ("Invalid syntax in expression");