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
188 %token <voidval> CASE
190 %token <voidval> ESAC
191 %token <voidval> LOGIOR
192 %token <voidval> ORIF
193 %token <voidval> LOGXOR
194 %token <voidval> LOGAND
195 %token <voidval> ANDIF
197 %token <voidval> NOTEQUAL
207 %token <voidval> SLASH_SLASH
211 %token <voidval> POINTER
212 %token <voidval> RECEIVE
219 %token <voidval> THEN
220 %token <voidval> ELSE
222 %token <voidval> ELSIF
223 %token <voidval> ILLEGAL_TOKEN
225 %token <voidval> PRED
226 %token <voidval> SUCC
228 %token <voidval> CARD
231 %token <voidval> SIZE
232 %token <voidval> UPPER
233 %token <voidval> LOWER
234 %token <voidval> LENGTH
236 /* Tokens which are not Chill tokens used in expressions, but rather GDB
237 specific things that we recognize in the same context as Chill tokens
238 (register names for example). */
240 %token <lval> GDB_REGNAME /* Machine register name */
241 %token <lval> GDB_LAST /* Value history */
242 %token <ivar> GDB_VARIABLE /* Convenience variable */
243 %token <voidval> GDB_ASSIGNMENT /* Assign value to somewhere */
245 %type <voidval> location
246 %type <voidval> access_name
247 %type <voidval> primitive_value
248 %type <voidval> location_contents
249 %type <voidval> value_name
250 %type <voidval> literal
251 %type <voidval> tuple
252 %type <voidval> value_string_element
253 %type <voidval> value_string_slice
254 %type <voidval> value_array_element
255 %type <voidval> value_array_slice
256 %type <voidval> value_structure_field
257 %type <voidval> expression_conversion
258 %type <voidval> value_procedure_call
259 %type <voidval> value_built_in_routine_call
260 %type <voidval> chill_value_built_in_routine_call
261 %type <voidval> start_expression
262 %type <voidval> zero_adic_operator
263 %type <voidval> parenthesised_expression
264 %type <voidval> value
265 %type <voidval> undefined_value
266 %type <voidval> expression
267 %type <voidval> conditional_expression
268 %type <voidval> then_alternative
269 %type <voidval> else_alternative
270 %type <voidval> sub_expression
271 %type <voidval> value_case_alternative
272 %type <voidval> operand_0
273 %type <voidval> operand_1
274 %type <voidval> operand_2
275 %type <voidval> operand_3
276 %type <voidval> operand_4
277 %type <voidval> operand_5
278 %type <voidval> operand_6
279 %type <voidval> synonym_name
280 %type <voidval> value_enumeration_name
281 %type <voidval> value_do_with_name
282 %type <voidval> value_receive_name
283 %type <voidval> string_primitive_value
284 %type <voidval> start_element
285 %type <voidval> left_element
286 %type <voidval> right_element
287 %type <voidval> slice_size
288 %type <voidval> array_primitive_value
289 %type <voidval> expression_list
290 %type <voidval> lower_element
291 %type <voidval> upper_element
292 %type <voidval> first_element
293 %type <voidval> structure_primitive_value
294 %type <voidval> field_name
295 %type <voidval> mode_argument
296 %type <voidval> upper_lower_argument
297 %type <voidval> length_argument
298 %type <voidval> mode_name
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
308 %type <voidval> single_assignment_action
324 undefined_value : FIXME_01
332 location : access_name
344 access_name : LOCATION_NAME
346 write_exp_elt_opcode (OP_VAR_VALUE);
347 write_exp_elt_sym ($1.sym);
348 write_exp_elt_opcode (OP_VAR_VALUE);
350 | GDB_LAST /* gdb specific */
352 write_exp_elt_opcode (OP_LAST);
353 write_exp_elt_longcst ($1);
354 write_exp_elt_opcode (OP_LAST);
356 | GDB_REGNAME /* gdb specific */
358 write_exp_elt_opcode (OP_REGISTER);
359 write_exp_elt_longcst ($1);
360 write_exp_elt_opcode (OP_REGISTER);
362 | GDB_VARIABLE /* gdb specific */
364 write_exp_elt_opcode (OP_INTERNALVAR);
365 write_exp_elt_intern ($1);
366 write_exp_elt_opcode (OP_INTERNALVAR);
376 expression_list : expression
380 | expression_list ',' expression
387 primitive_value : location_contents
403 | value_string_element
411 | value_array_element
419 | value_structure_field
423 | expression_conversion
427 | value_procedure_call
431 | value_built_in_routine_call
443 | parenthesised_expression
451 location_contents: location
459 value_name : synonym_name
463 | value_enumeration_name
475 | GENERAL_PROCEDURE_NAME
477 write_exp_elt_opcode (OP_VAR_VALUE);
478 write_exp_elt_sym ($1.sym);
479 write_exp_elt_opcode (OP_VAR_VALUE);
485 literal : INTEGER_LITERAL
487 write_exp_elt_opcode (OP_LONG);
488 write_exp_elt_type ($1.type);
489 write_exp_elt_longcst ((LONGEST) ($1.val));
490 write_exp_elt_opcode (OP_LONG);
494 write_exp_elt_opcode (OP_BOOL);
495 write_exp_elt_longcst ((LONGEST) $1);
496 write_exp_elt_opcode (OP_BOOL);
500 write_exp_elt_opcode (OP_LONG);
501 write_exp_elt_type ($1.type);
502 write_exp_elt_longcst ((LONGEST) ($1.val));
503 write_exp_elt_opcode (OP_LONG);
507 write_exp_elt_opcode (OP_DOUBLE);
508 write_exp_elt_type (builtin_type_double);
509 write_exp_elt_dblcst ($1);
510 write_exp_elt_opcode (OP_DOUBLE);
520 | CHARACTER_STRING_LITERAL
522 write_exp_elt_opcode (OP_STRING);
523 write_exp_string ($1);
524 write_exp_elt_opcode (OP_STRING);
528 write_exp_elt_opcode (OP_BITSTRING);
529 write_exp_bitstring ($1);
530 write_exp_elt_opcode (OP_BITSTRING);
545 value_string_element: string_primitive_value '(' start_element ')'
553 value_string_slice: string_primitive_value '(' left_element ':' right_element ')'
557 | string_primitive_value '(' start_element UP slice_size ')'
565 value_array_element: array_primitive_value '('
566 /* This is to save the value of arglist_len
567 being accumulated for each dimension. */
568 { start_arglist (); }
571 write_exp_elt_opcode (MULTI_SUBSCRIPT);
572 write_exp_elt_longcst ((LONGEST) end_arglist ());
573 write_exp_elt_opcode (MULTI_SUBSCRIPT);
579 value_array_slice: array_primitive_value '(' lower_element ':' upper_element ')'
583 | array_primitive_value '(' first_element UP slice_size ')'
591 value_structure_field: structure_primitive_value '.' field_name
599 expression_conversion: mode_name parenthesised_expression
607 value_procedure_call: FIXME_05
615 value_built_in_routine_call: chill_value_built_in_routine_call
623 start_expression: FIXME_06
626 } /* Not in GNU-Chill */
631 zero_adic_operator: FIXME_07
639 parenthesised_expression: '(' expression ')'
647 expression : operand_0
651 | conditional_expression
657 conditional_expression : IF boolean_expression then_alternative else_alternative FI
661 | CASE case_selector_list OF value_case_alternative '[' ELSE sub_expression ']' ESAC
667 then_alternative: THEN subexpression
673 else_alternative: ELSE subexpression
677 | ELSIF boolean_expression then_alternative else_alternative
683 sub_expression : expression
689 value_case_alternative: case_label_specification ':' sub_expression ';'
697 operand_0 : operand_1
701 | operand_0 LOGIOR operand_1
703 write_exp_elt_opcode (BINOP_BITWISE_IOR);
705 | operand_0 ORIF operand_1
709 | operand_0 LOGXOR operand_1
711 write_exp_elt_opcode (BINOP_BITWISE_XOR);
713 | single_assignment_action
721 operand_1 : operand_2
725 | operand_1 LOGAND operand_2
727 write_exp_elt_opcode (BINOP_BITWISE_AND);
729 | operand_1 ANDIF operand_2
737 operand_2 : operand_3
741 | operand_2 '=' operand_3
743 write_exp_elt_opcode (BINOP_EQUAL);
745 | operand_2 NOTEQUAL operand_3
747 write_exp_elt_opcode (BINOP_NOTEQUAL);
749 | operand_2 '>' operand_3
751 write_exp_elt_opcode (BINOP_GTR);
753 | operand_2 GTR operand_3
755 write_exp_elt_opcode (BINOP_GEQ);
757 | operand_2 '<' operand_3
759 write_exp_elt_opcode (BINOP_LESS);
761 | operand_2 LEQ operand_3
763 write_exp_elt_opcode (BINOP_LEQ);
765 | operand_2 IN operand_3
774 operand_3 : operand_4
778 | operand_3 '+' operand_4
780 write_exp_elt_opcode (BINOP_ADD);
782 | operand_3 '-' operand_4
784 write_exp_elt_opcode (BINOP_SUB);
786 | operand_3 SLASH_SLASH operand_4
788 write_exp_elt_opcode (BINOP_CONCAT);
794 operand_4 : operand_5
798 | operand_4 '*' operand_5
800 write_exp_elt_opcode (BINOP_MUL);
802 | operand_4 '/' operand_5
804 write_exp_elt_opcode (BINOP_DIV);
806 | operand_4 MOD operand_5
808 write_exp_elt_opcode (BINOP_MOD);
810 | operand_4 REM operand_5
812 write_exp_elt_opcode (BINOP_REM);
817 /* Note that we accept any expression for BINOP_CONCAT, not just
818 integer literal expressions. (FIXME?) */
820 operand_5 : operand_6
826 write_exp_elt_opcode (UNOP_NEG);
830 write_exp_elt_opcode (UNOP_LOGICAL_NOT);
832 | parenthesised_expression operand_6
834 write_exp_elt_opcode (BINOP_CONCAT);
840 operand_6 : POINTER location
844 | RECEIVE buffer_location
857 single_assignment_action :
858 location GDB_ASSIGNMENT value
860 write_exp_elt_opcode (BINOP_ASSIGN);
866 chill_value_built_in_routine_call :
867 NUM '(' expression ')'
871 | PRED '(' expression ')'
875 | SUCC '(' expression ')'
879 | ABS '(' expression ')'
883 | CARD '(' expression ')'
887 | MAX '(' expression ')'
891 | MIN '(' expression ')'
895 | SIZE '(' location ')'
899 | SIZE '(' mode_argument ')'
903 | UPPER '(' upper_lower_argument ')'
907 | LOWER '(' upper_lower_argument ')'
911 | LENGTH '(' length_argument ')'
917 mode_argument : mode_name
921 | array_mode_name '(' expression ')'
925 | string_mode_name '(' expression ')'
929 | variant_structure_mode_name '(' expression_list ')'
935 upper_lower_argument : location
949 length_argument : location
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 { $$ = 0; }
985 structure_primitive_value: FIXME_23 { $$ = 0; }
986 field_name : FIXME_24 { $$ = 0; }
987 mode_name : FIXME_25 { $$ = 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))
1048 } while (isalpha (*tokptr) || isdigit (*tokptr) || (*tokptr == '_'));
1049 yylval.sval.ptr = lexptr;
1050 yylval.sval.length = tokptr - lexptr;
1052 return (copy_name (yylval.sval));
1057 /* Start looking for a value composed of valid digits as set by the base
1058 in use. Note that '_' characters are valid anywhere, in any quantity,
1059 and are simply ignored. Since we must find at least one valid digit,
1060 or reject this token as an integer literal, we keep track of how many
1061 digits we have encountered. */
1064 decode_integer_value (base, tokptrptr, ivalptr)
1069 char *tokptr = *tokptrptr;
1073 while (*tokptr != '\0')
1075 temp = tolower (*tokptr);
1081 case '0': case '1': case '2': case '3': case '4':
1082 case '5': case '6': case '7': case '8': case '9':
1085 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
1101 /* Found something not in domain for current base. */
1102 tokptr--; /* Unconsume what gave us indigestion. */
1107 /* If we didn't find any digits, then we don't have a valid integer
1108 value, so reject the entire token. Otherwise, update the lexical
1109 scan pointer, and return non-zero for success. */
1117 *tokptrptr = tokptr;
1123 decode_integer_literal (valptr, tokptrptr)
1127 char *tokptr = *tokptrptr;
1130 int explicit_base = 0;
1132 /* Look for an explicit base specifier, which is optional. */
1165 /* If we found an explicit base ensure that the character after the
1166 explicit base is a single quote. */
1168 if (explicit_base && (*tokptr++ != '\''))
1173 /* Attempt to decode whatever follows as an integer value in the
1174 indicated base, updating the token pointer in the process and
1175 computing the value into ival. Also, if we have an explicit
1176 base, then the next character must not be a single quote, or we
1177 have a bitstring literal, so reject the entire token in this case.
1178 Otherwise, update the lexical scan pointer, and return non-zero
1181 if (!decode_integer_value (base, &tokptr, &ival))
1185 else if (explicit_base && (*tokptr == '\''))
1192 *tokptrptr = tokptr;
1197 /* If it wasn't for the fact that floating point values can contain '_'
1198 characters, we could just let strtod do all the hard work by letting it
1199 try to consume as much of the current token buffer as possible and
1200 find a legal conversion. Unfortunately we need to filter out the '_'
1201 characters before calling strtod, which we do by copying the other
1202 legal chars to a local buffer to be converted. However since we also
1203 need to keep track of where the last unconsumed character in the input
1204 buffer is, we have transfer only as many characters as may compose a
1205 legal floating point value. */
1208 match_float_literal ()
1210 char *tokptr = lexptr;
1215 extern double strtod ();
1217 /* Make local buffer in which to build the string to convert. This is
1218 required because underscores are valid in chill floating point numbers
1219 but not in the string passed to strtod to convert. The string will be
1220 no longer than our input string. */
1222 copy = buf = (char *) alloca (strlen (tokptr) + 1);
1224 /* Transfer all leading digits to the conversion buffer, discarding any
1227 while (isdigit (*tokptr) || *tokptr == '_')
1236 /* Now accept either a '.', or one of [eEdD]. Dot is legal regardless
1237 of whether we found any leading digits, and we simply accept it and
1238 continue on to look for the fractional part and/or exponent. One of
1239 [eEdD] is legal only if we have seen digits, and means that there
1240 is no fractional part. If we find neither of these, then this is
1241 not a floating point number, so return failure. */
1246 /* Accept and then look for fractional part and/or exponent. */
1259 goto collect_exponent;
1267 /* We found a '.', copy any fractional digits to the conversion buffer, up
1268 to the first nondigit, non-underscore character. */
1270 while (isdigit (*tokptr) || *tokptr == '_')
1279 /* Look for an exponent, which must start with one of [eEdD]. If none
1280 is found, jump directly to trying to convert what we have collected
1297 /* Accept an optional '-' or '+' following one of [eEdD]. */
1300 if (*tokptr == '+' || *tokptr == '-')
1302 *copy++ = *tokptr++;
1305 /* Now copy an exponent into the conversion buffer. Note that at the
1306 moment underscores are *not* allowed in exponents. */
1308 while (isdigit (*tokptr))
1310 *copy++ = *tokptr++;
1313 /* If we transfered any chars to the conversion buffer, try to interpret its
1314 contents as a floating point value. If any characters remain, then we
1315 must not have a valid floating point string. */
1321 dval = strtod (buf, ©);
1326 return (FLOAT_LITERAL);
1332 /* Recognize a string literal. A string literal is a nonzero sequence
1333 of characters enclosed in matching single or double quotes, except that
1334 a single character inside single quotes is a character literal, which
1335 we reject as a string literal. To embed the terminator character inside
1336 a string, it is simply doubled (I.E. "this""is""one""string") */
1339 match_string_literal ()
1341 char *tokptr = lexptr;
1343 for (tempbufindex = 0, tokptr++; *tokptr != '\0'; tokptr++)
1346 if (*tokptr == *lexptr)
1348 if (*(tokptr + 1) == *lexptr)
1357 tempbuf[tempbufindex++] = *tokptr;
1359 if (*tokptr == '\0' /* no terminator */
1360 || tempbufindex == 0 /* no string */
1361 || (tempbufindex == 1 && *tokptr == '\'')) /* char literal */
1367 tempbuf[tempbufindex] = '\0';
1368 yylval.sval.ptr = tempbuf;
1369 yylval.sval.length = tempbufindex;
1371 return (CHARACTER_STRING_LITERAL);
1375 /* Recognize a character literal. A character literal is single character
1376 or a control sequence, enclosed in single quotes. A control sequence
1377 is a comma separated list of one or more integer literals, enclosed
1378 in parenthesis and introduced with a circumflex character.
1380 EX: 'a' '^(7)' '^(7,8)'
1382 As a GNU chill extension, the syntax C'xx' is also recognized as a
1383 character literal, where xx is a hex value for the character.
1385 Note that more than a single character, enclosed in single quotes, is
1388 Also note that the control sequence form is not in GNU Chill since it
1389 is ambiguous with the string literal form using single quotes. I.E.
1390 is '^(7)' a character literal or a string literal. In theory it it
1391 possible to tell by context, but GNU Chill doesn't accept the control
1392 sequence form, so neither do we (for now the code is disabled).
1394 Returns CHARACTER_LITERAL if a match is found.
1398 match_character_literal ()
1400 char *tokptr = lexptr;
1403 if ((tolower (*tokptr) == 'c') && (*(tokptr + 1) == '\''))
1405 /* We have a GNU chill extension form, so skip the leading "C'",
1406 decode the hex value, and then ensure that we have a trailing
1407 single quote character. */
1409 if (!decode_integer_value (16, &tokptr, &ival) || (*tokptr != '\''))
1415 else if (*tokptr == '\'')
1419 /* Determine which form we have, either a control sequence or the
1420 single character form. */
1422 if ((*tokptr == '^') && (*(tokptr + 1) == '('))
1424 #if 0 /* Disable, see note above. -fnf */
1425 /* Match and decode a control sequence. Return zero if we don't
1426 find a valid integer literal, or if the next unconsumed character
1427 after the integer literal is not the trailing ')'.
1428 FIXME: We currently don't handle the multiple integer literal
1431 if (!decode_integer_literal (&ival, &tokptr) || (*tokptr++ != ')'))
1444 /* The trailing quote has not yet been consumed. If we don't find
1445 it, then we have no match. */
1447 if (*tokptr++ != '\'')
1454 /* Not a character literal. */
1457 yylval.typed_val.val = ival;
1458 yylval.typed_val.type = builtin_type_chill_char;
1460 return (CHARACTER_LITERAL);
1463 /* Recognize an integer literal, as specified in Z.200 sec 5.2.4.2.
1464 Note that according to 5.2.4.2, a single "_" is also a valid integer
1465 literal, however GNU-chill requires there to be at least one "digit"
1466 in any integer literal. */
1469 match_integer_literal ()
1471 char *tokptr = lexptr;
1474 if (!decode_integer_literal (&ival, &tokptr))
1480 yylval.typed_val.val = ival;
1481 yylval.typed_val.type = builtin_type_int;
1483 return (INTEGER_LITERAL);
1487 /* Recognize a bit-string literal, as specified in Z.200 sec 5.2.4.8
1488 Note that according to 5.2.4.8, a single "_" is also a valid bit-string
1489 literal, however GNU-chill requires there to be at least one "digit"
1490 in any bit-string literal. */
1493 match_bitstring_literal ()
1495 char *tokptr = lexptr;
1504 /* Look for the required explicit base specifier. */
1525 /* Ensure that the character after the explicit base is a single quote. */
1527 if (*tokptr++ != '\'')
1532 while (*tokptr != '\0' && *tokptr != '\'')
1534 digit = tolower (*tokptr);
1540 case '0': case '1': case '2': case '3': case '4':
1541 case '5': case '6': case '7': case '8': case '9':
1544 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
1554 /* Found something not in domain for current base. */
1559 /* Extract bits from digit, starting with the msbit appropriate for
1560 the current base, and packing them into the bitstring byte,
1561 starting at the lsbit. */
1562 for (mask = (base >> 1); mask > 0; mask >>= 1)
1568 tempbuf[tempbufindex] |= (1 << bitoffset);
1571 if (bitoffset == HOST_CHAR_BIT)
1580 /* Verify that we consumed everything up to the trailing single quote,
1581 and that we found some bits (IE not just underbars). */
1583 if (*tokptr++ != '\'')
1589 yylval.sval.ptr = tempbuf;
1590 yylval.sval.length = bitcount;
1592 return (BIT_STRING_LITERAL);
1596 /* Recognize tokens that start with '$'. These include:
1598 $regname A native register name or a "standard
1600 Return token GDB_REGNAME.
1602 $variable A convenience variable with a name chosen
1604 Return token GDB_VARIABLE.
1606 $digits Value history with index <digits>, starting
1607 from the first value which has index 1.
1610 $$digits Value history with index <digits> relative
1611 to the last value. I.E. $$0 is the last
1612 value, $$1 is the one previous to that, $$2
1613 is the one previous to $$1, etc.
1614 Return token GDB_LAST.
1616 $ | $0 | $$0 The last value in the value history.
1617 Return token GDB_LAST.
1619 $$ An abbreviation for the second to the last
1620 value in the value history, I.E. $$1
1621 Return token GDB_LAST.
1623 Note that we currently assume that register names and convenience
1624 variables follow the convention of starting with a letter or '_'.
1629 match_dollar_tokens ()
1637 /* We will always have a successful match, even if it is just for
1638 a single '$', the abbreviation for $$0. So advance lexptr. */
1642 if (*tokptr == '_' || isalpha (*tokptr))
1644 /* Look for a match with a native register name, usually something
1645 like "r0" for example. */
1647 for (regno = 0; regno < NUM_REGS; regno++)
1649 namelength = strlen (reg_names[regno]);
1650 if (STREQN (tokptr, reg_names[regno], namelength)
1651 && !isalnum (tokptr[namelength]))
1653 yylval.lval = regno;
1654 lexptr += namelength + 1;
1655 return (GDB_REGNAME);
1659 /* Look for a match with a standard register name, usually something
1660 like "pc", which gdb always recognizes as the program counter
1661 regardless of what the native register name is. */
1663 for (regno = 0; regno < num_std_regs; regno++)
1665 namelength = strlen (std_regs[regno].name);
1666 if (STREQN (tokptr, std_regs[regno].name, namelength)
1667 && !isalnum (tokptr[namelength]))
1669 yylval.lval = std_regs[regno].regnum;
1670 lexptr += namelength;
1671 return (GDB_REGNAME);
1675 /* Attempt to match against a convenience variable. Note that
1676 this will always succeed, because if no variable of that name
1677 already exists, the lookup_internalvar will create one for us.
1678 Also note that both lexptr and tokptr currently point to the
1679 start of the input string we are trying to match, and that we
1680 have already tested the first character for non-numeric, so we
1681 don't have to treat it specially. */
1683 while (*tokptr == '_' || isalnum (*tokptr))
1687 yylval.sval.ptr = lexptr;
1688 yylval.sval.length = tokptr - lexptr;
1689 yylval.ivar = lookup_internalvar (copy_name (yylval.sval));
1691 return (GDB_VARIABLE);
1694 /* Since we didn't match against a register name or convenience
1695 variable, our only choice left is a history value. */
1709 /* Attempt to decode more characters as an integer value giving
1710 the index in the history list. If successful, the value will
1711 overwrite ival (currently 0 or 1), and if not, ival will be
1712 left alone, which is good since it is currently correct for
1713 the '$' or '$$' case. */
1715 decode_integer_literal (&ival, &tokptr);
1716 yylval.lval = negate ? -ival : ival;
1727 static const struct token tokentab6[] =
1729 { "LENGTH", LENGTH }
1732 static const struct token tokentab5[] =
1739 static const struct token tokentab4[] =
1748 static const struct token tokentab3[] =
1761 static const struct token tokentab2[] =
1763 { ":=", GDB_ASSIGNMENT },
1764 { "//", SLASH_SLASH },
1772 /* Read one token, getting characters through lexptr. */
1773 /* This is where we will check to make sure that the language and the
1774 operators used are compatible. */
1784 /* Skip over any leading whitespace. */
1785 while (isspace (*lexptr))
1789 /* Look for special single character cases which can't be the first
1790 character of some other multicharacter token. */
1808 /* Look for characters which start a particular kind of multicharacter
1809 token, such as a character literal, register name, convenience
1810 variable name, string literal, etc. */
1815 /* First try to match a string literal, which is any nonzero
1816 sequence of characters enclosed in matching single or double
1817 quotes, except that a single character inside single quotes
1818 is a character literal, so we have to catch that case also. */
1819 token = match_string_literal ();
1824 if (*lexptr == '\'')
1826 token = match_character_literal ();
1835 token = match_character_literal ();
1842 token = match_dollar_tokens ();
1849 /* See if it is a special token of length 6. */
1850 for (i = 0; i < sizeof (tokentab6) / sizeof (tokentab6[0]); i++)
1852 if (STREQN (lexptr, tokentab6[i].operator, 6))
1855 return (tokentab6[i].token);
1858 /* See if it is a special token of length 5. */
1859 for (i = 0; i < sizeof (tokentab5) / sizeof (tokentab5[0]); i++)
1861 if (STREQN (lexptr, tokentab5[i].operator, 5))
1864 return (tokentab5[i].token);
1867 /* See if it is a special token of length 4. */
1868 for (i = 0; i < sizeof (tokentab4) / sizeof (tokentab4[0]); i++)
1870 if (STREQN (lexptr, tokentab4[i].operator, 4))
1873 return (tokentab4[i].token);
1876 /* See if it is a special token of length 3. */
1877 for (i = 0; i < sizeof (tokentab3) / sizeof (tokentab3[0]); i++)
1879 if (STREQN (lexptr, tokentab3[i].operator, 3))
1882 return (tokentab3[i].token);
1885 /* See if it is a special token of length 2. */
1886 for (i = 0; i < sizeof (tokentab2) / sizeof (tokentab2[0]); i++)
1888 if (STREQN (lexptr, tokentab2[i].operator, 2))
1891 return (tokentab2[i].token);
1894 /* Look for single character cases which which could be the first
1895 character of some other multicharacter token, but aren't, or we
1896 would already have found it. */
1905 /* Look for other special tokens. */
1906 if (STREQN (lexptr, "TRUE", 4)) /* FIXME: What about lowercase? */
1910 return (BOOLEAN_LITERAL);
1912 if (STREQN (lexptr, "FALSE", 5)) /* FIXME: What about lowercase? */
1916 return (BOOLEAN_LITERAL);
1918 /* Look for a float literal before looking for an integer literal, so
1919 we match as much of the input stream as possible. */
1920 token = match_float_literal ();
1925 token = match_bitstring_literal ();
1930 token = match_integer_literal ();
1936 /* Try to match a simple name string, and if a match is found, then
1937 further classify what sort of name it is and return an appropriate
1938 token. Note that attempting to match a simple name string consumes
1939 the token from lexptr, so we can't back out if we later find that
1940 we can't classify what sort of name it is. */
1942 simplename = match_simple_name_string ();
1943 if (simplename != NULL)
1945 sym = lookup_symbol (simplename, expression_context_block,
1946 VAR_NAMESPACE, (int *) NULL,
1947 (struct symtab **) NULL);
1950 yylval.ssym.stoken.ptr = NULL;
1951 yylval.ssym.stoken.length = 0;
1952 yylval.ssym.sym = sym;
1953 yylval.ssym.is_a_field_of_this = 0; /* FIXME, C++'ism */
1954 switch (SYMBOL_CLASS (sym))
1957 /* Found a procedure name. */
1958 return (GENERAL_PROCEDURE_NAME);
1960 /* Found a global or local static variable. */
1961 return (LOCATION_NAME);
1968 if (innermost_block == NULL
1969 || contained_in (block_found, innermost_block))
1971 innermost_block = block_found;
1973 return (LOCATION_NAME);
1977 return (LOCATION_NAME);
1981 case LOC_CONST_BYTES:
1982 error ("Symbol \"%s\" names no location.", simplename);
1986 else if (!have_full_symbols () && !have_partial_symbols ())
1988 error ("No symbol table is loaded. Use the \"file\" command.");
1992 error ("No symbol \"%s\" in current context.", simplename);
1996 /* Catch single character tokens which are not part of some
2001 case '.': /* Not float for example. */
2005 return (ILLEGAL_TOKEN);
2010 char *msg; /* unused */
2012 printf ("Parsing: %s\n", lexptr);
2015 error ("Invalid syntax in expression near character '%c'.", yychar);
2019 error ("Invalid syntax in expression");