* language.h (language_format_info): New structure to bundle
[binutils-gdb.git] / gdb / ch-exp.y
1 /* YACC grammar for Chill expressions, for GDB.
2 Copyright (C) 1992 Free Software Foundation, Inc.
3
4 This file is part of GDB.
5
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.
10
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.
15
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. */
19
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.
28
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.
36
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.
51
52 */
53
54 %{
55
56 #include <stdio.h>
57 #include <string.h>
58 #include "defs.h"
59 #include "symtab.h"
60 #include "gdbtypes.h"
61 #include "frame.h"
62 #include "expression.h"
63 #include "language.h"
64 #include "value.h"
65 #include "parser-defs.h"
66 #include "bfd.h"
67 #include "symfile.h"
68 #include "objfiles.h"
69
70 /* These MUST be included in any grammar file!!!! Please choose unique names!
71 Note that this are a combined list of variables that can be produced
72 by any one of bison, byacc, or yacc. */
73 #define yymaxdepth chill_maxdepth
74 #define yyparse chill_parse
75 #define yylex chill_lex
76 #define yyerror chill_error
77 #define yylval chill_lval
78 #define yychar chill_char
79 #define yydebug chill_debug
80 #define yypact chill_pact
81 #define yyr1 chill_r1
82 #define yyr2 chill_r2
83 #define yydef chill_def
84 #define yychk chill_chk
85 #define yypgo chill_pgo
86 #define yyact chill_act
87 #define yyexca chill_exca
88 #define yyerrflag chill_errflag
89 #define yynerrs chill_nerrs
90 #define yyps chill_ps
91 #define yypv chill_pv
92 #define yys chill_s
93 #define yy_yys chill_yys
94 #define yystate chill_state
95 #define yytmp chill_tmp
96 #define yyv chill_v
97 #define yy_yyv chill_yyv
98 #define yyval chill_val
99 #define yylloc chill_lloc
100 #define yyss chill_yyss /* byacc */
101 #define yyssp chill_yysp /* byacc */
102 #define yyvs chill_yyvs /* byacc */
103 #define yyvsp chill_yyvsp /* byacc */
104
105 static int
106 yylex PARAMS ((void));
107
108 static void
109 yyerror PARAMS ((char *));
110
111 int
112 yyparse PARAMS ((void));
113
114 /* #define YYDEBUG 1 */
115
116 %}
117
118 /* Although the yacc "value" of an expression is not used,
119 since the result is stored in the structure being created,
120 other node types do have values. */
121
122 %union
123 {
124 LONGEST lval;
125 unsigned LONGEST ulval;
126 struct {
127 LONGEST val;
128 struct type *type;
129 } typed_val;
130 double dval;
131 struct symbol *sym;
132 struct type *tval;
133 struct stoken sval;
134 struct ttype tsym;
135 struct symtoken ssym;
136 int voidval;
137 struct block *bval;
138 enum exp_opcode opcode;
139 struct internalvar *ivar;
140
141 struct type **tvec;
142 int *ivec;
143 }
144
145 %{
146 static int parse_number PARAMS ((void));
147 %}
148
149 %token <voidval> FIXME
150
151 %token <typed_val> INTEGER_LITERAL
152 %token <ulval> BOOLEAN_LITERAL
153 %token <typed_val> CHARACTER_LITERAL
154 %token <voidval> SET_LITERAL
155 %token <voidval> EMPTINESS_LITERAL
156 %token <voidval> CHARACTER_STRING_LITERAL
157 %token <voidval> BIT_STRING_LITERAL
158
159 %token <voidval> STRING
160 %token <voidval> CONSTANT
161 %token <voidval> '.'
162 %token <voidval> ';'
163 %token <voidval> ':'
164 %token <voidval> CASE
165 %token <voidval> OF
166 %token <voidval> ESAC
167 %token <voidval> LOGIOR
168 %token <voidval> ORIF
169 %token <voidval> LOGXOR
170 %token <voidval> LOGAND
171 %token <voidval> ANDIF
172 %token <voidval> '='
173 %token <voidval> NOTEQUAL
174 %token <voidval> '>'
175 %token <voidval> GTR
176 %token <voidval> '<'
177 %token <voidval> LEQ
178 %token <voidval> IN
179 %token <voidval> '+'
180 %token <voidval> '-'
181 %token <voidval> '*'
182 %token <voidval> '/'
183 %token <voidval> SLASH_SLASH
184 %token <voidval> MOD
185 %token <voidval> REM
186 %token <voidval> NOT
187 %token <voidval> POINTER
188 %token <voidval> RECEIVE
189 %token <voidval> SC
190 %token <voidval> '['
191 %token <voidval> ']'
192 %token <voidval> '('
193 %token <voidval> ')'
194 %token <voidval> UP
195 %token <voidval> IF
196 %token <voidval> THEN
197 %token <voidval> ELSE
198 %token <voidval> FI
199 %token <voidval> ELSIF
200 %token <voidval> ILLEGAL_TOKEN
201
202 %type <voidval> location
203 %type <voidval> primitive_value
204 %type <voidval> location_contents
205 %type <voidval> value_name
206 %type <voidval> literal
207 %type <voidval> tuple
208 %type <voidval> value_string_element
209 %type <voidval> value_string_slice
210 %type <voidval> value_array_element
211 %type <voidval> value_array_slice
212 %type <voidval> value_structure_field
213 %type <voidval> expression_conversion
214 %type <voidval> value_procedure_call
215 %type <voidval> value_built_in_routine_call
216 %type <voidval> start_expression
217 %type <voidval> zero_adic_operator
218 %type <voidval> parenthesised_expression
219 %type <voidval> value
220 %type <voidval> undefined_value
221 %type <voidval> expression
222 %type <voidval> conditional_expression
223 %type <voidval> then_alternative
224 %type <voidval> else_alternative
225 %type <voidval> sub_expression
226 %type <voidval> value_case_alternative
227 %type <voidval> operand_0
228 %type <voidval> operand_1
229 %type <voidval> operand_2
230 %type <voidval> operand_3
231 %type <voidval> operand_4
232 %type <voidval> operand_5
233 %type <voidval> operand_6
234 %type <voidval> integer_literal_expression
235 %type <voidval> synonym_name
236 %type <voidval> value_enumeration_name
237 %type <voidval> value_do_with_name
238 %type <voidval> value_receive_name
239 %type <voidval> general_procedure_name
240 %type <voidval> string_primitive_value
241 %type <voidval> start_element
242 %type <voidval> left_element
243 %type <voidval> right_element
244 %type <voidval> slice_size
245 %type <voidval> array_primitive_value
246 %type <voidval> expression_list
247 %type <voidval> lower_element
248 %type <voidval> upper_element
249 %type <voidval> first_element
250 %type <voidval> structure_primitive_value
251 %type <voidval> field_name
252 %type <voidval> mode_name
253 %type <voidval> boolean_expression
254 %type <voidval> case_selector_list
255 %type <voidval> subexpression
256 %type <voidval> case_label_specification
257 %type <voidval> buffer_location
258
259 %%
260
261 /* Z.200, 5.3.1 */
262
263 value : expression
264 {
265 $$ = 0; /* FIXME */
266 }
267 | undefined_value
268 {
269 $$ = 0; /* FIXME */
270 }
271 ;
272
273 undefined_value : FIXME
274 {
275 $$ = 0; /* FIXME */
276 }
277 ;
278
279 /* Z.200, 4.2.1 */
280
281 location : FIXME
282 {
283 $$ = 0; /* FIXME */
284 }
285 ;
286
287 /* Z.200, 5.2.1 */
288
289 primitive_value : location_contents
290 {
291 $$ = 0; /* FIXME */
292 }
293 | value_name
294 {
295 $$ = 0; /* FIXME */
296 }
297 | literal
298 {
299 $$ = 0; /* FIXME */
300 }
301 | tuple
302 {
303 $$ = 0; /* FIXME */
304 }
305 | value_string_element
306 {
307 $$ = 0; /* FIXME */
308 }
309 | value_string_slice
310 {
311 $$ = 0; /* FIXME */
312 }
313 | value_array_element
314 {
315 $$ = 0; /* FIXME */
316 }
317 | value_array_slice
318 {
319 $$ = 0; /* FIXME */
320 }
321 | value_structure_field
322 {
323 $$ = 0; /* FIXME */
324 }
325 | expression_conversion
326 {
327 $$ = 0; /* FIXME */
328 }
329 | value_procedure_call
330 {
331 $$ = 0; /* FIXME */
332 }
333 | value_built_in_routine_call
334 {
335 $$ = 0; /* FIXME */
336 }
337 | start_expression
338 {
339 $$ = 0; /* FIXME */
340 }
341 | zero_adic_operator
342 {
343 $$ = 0; /* FIXME */
344 }
345 | parenthesised_expression
346 {
347 $$ = 0; /* FIXME */
348 }
349 ;
350
351 /* Z.200, 5.2.2 */
352
353 location_contents: location
354 {
355 $$ = 0; /* FIXME */
356 }
357 ;
358
359 /* Z.200, 5.2.3 */
360
361 value_name : synonym_name
362 {
363 $$ = 0; /* FIXME */
364 }
365 | value_enumeration_name
366 {
367 $$ = 0; /* FIXME */
368 }
369 | value_do_with_name
370 {
371 $$ = 0; /* FIXME */
372 }
373 | value_receive_name
374 {
375 $$ = 0; /* FIXME */
376 }
377 | general_procedure_name
378 {
379 $$ = 0; /* FIXME */
380 }
381 ;
382
383 /* Z.200, 5.2.4.1 */
384
385 literal : INTEGER_LITERAL
386 {
387 write_exp_elt_opcode (OP_LONG);
388 write_exp_elt_type ($1.type);
389 write_exp_elt_longcst ((LONGEST) ($1.val));
390 write_exp_elt_opcode (OP_LONG);
391 }
392 | BOOLEAN_LITERAL
393 {
394 write_exp_elt_opcode (OP_BOOL);
395 write_exp_elt_longcst ((LONGEST) $1);
396 write_exp_elt_opcode (OP_BOOL);
397 }
398 | CHARACTER_LITERAL
399 {
400 write_exp_elt_opcode (OP_LONG);
401 write_exp_elt_type ($1.type);
402 write_exp_elt_longcst ((LONGEST) ($1.val));
403 write_exp_elt_opcode (OP_LONG);
404 }
405 | SET_LITERAL
406 {
407 $$ = 0; /* FIXME */
408 }
409 | EMPTINESS_LITERAL
410 {
411 $$ = 0; /* FIXME */
412 }
413 | CHARACTER_STRING_LITERAL
414 {
415 $$ = 0; /* FIXME */
416 }
417 | BIT_STRING_LITERAL
418 {
419 $$ = 0; /* FIXME */
420 }
421 ;
422
423 /* Z.200, 5.2.5 */
424
425 tuple : FIXME
426 {
427 $$ = 0; /* FIXME */
428 }
429 ;
430
431
432 /* Z.200, 5.2.6 */
433
434 value_string_element: string_primitive_value '(' start_element ')'
435 {
436 $$ = 0; /* FIXME */
437 }
438 ;
439
440 /* Z.200, 5.2.7 */
441
442 value_string_slice: string_primitive_value '(' left_element ':' right_element ')'
443 {
444 $$ = 0; /* FIXME */
445 }
446 | string_primitive_value '(' start_element UP slice_size ')'
447 {
448 $$ = 0; /* FIXME */
449 }
450 ;
451
452 /* Z.200, 5.2.8 */
453
454 value_array_element: array_primitive_value '(' expression_list ')'
455 {
456 $$ = 0; /* FIXME */
457 }
458 ;
459
460 /* Z.200, 5.2.9 */
461
462 value_array_slice: array_primitive_value '(' lower_element ':' upper_element ')'
463 {
464 $$ = 0; /* FIXME */
465 }
466 | array_primitive_value '(' first_element UP slice_size '('
467 {
468 $$ = 0; /* FIXME */
469 }
470 ;
471
472 /* Z.200, 5.2.10 */
473
474 value_structure_field: structure_primitive_value '.' field_name
475 {
476 $$ = 0; /* FIXME */
477 }
478 ;
479
480 /* Z.200, 5.2.11 */
481
482 expression_conversion: mode_name '(' expression ')'
483 {
484 $$ = 0; /* FIXME */
485 }
486 ;
487
488 /* Z.200, 5.2.12 */
489
490 value_procedure_call: FIXME
491 {
492 $$ = 0; /* FIXME */
493 }
494 ;
495
496 /* Z.200, 5.2.13 */
497
498 value_built_in_routine_call: FIXME
499 {
500 $$ = 0; /* FIXME */
501 }
502 ;
503
504 /* Z.200, 5.2.14 */
505
506 start_expression: FIXME
507 {
508 $$ = 0; /* FIXME */
509 } /* Not in GNU-Chill */
510 ;
511
512 /* Z.200, 5.2.15 */
513
514 zero_adic_operator: FIXME
515 {
516 $$ = 0; /* FIXME */
517 }
518 ;
519
520 /* Z.200, 5.2.16 */
521
522 parenthesised_expression: '(' expression ')'
523 {
524 $$ = 0; /* FIXME */
525 }
526 ;
527
528 /* Z.200, 5.3.2 */
529
530 expression : operand_0
531 {
532 $$ = 0; /* FIXME */
533 }
534 | conditional_expression
535 {
536 $$ = 0; /* FIXME */
537 }
538 ;
539
540 conditional_expression : IF boolean_expression then_alternative else_alternative FI
541 {
542 $$ = 0; /* FIXME */
543 }
544 | CASE case_selector_list OF value_case_alternative '[' ELSE sub_expression ']' ESAC
545 {
546 $$ = 0; /* FIXME */
547 }
548 ;
549
550 then_alternative: THEN subexpression
551 {
552 $$ = 0; /* FIXME */
553 }
554 ;
555
556 else_alternative: ELSE subexpression
557 {
558 $$ = 0; /* FIXME */
559 }
560 | ELSIF boolean_expression then_alternative else_alternative
561 {
562 $$ = 0; /* FIXME */
563 }
564 ;
565
566 sub_expression : expression
567 {
568 $$ = 0; /* FIXME */
569 }
570 ;
571
572 value_case_alternative: case_label_specification ':' sub_expression ';'
573 {
574 $$ = 0; /* FIXME */
575 }
576 ;
577
578 /* Z.200, 5.3.3 */
579
580 operand_0 : operand_1
581 {
582 $$ = 0; /* FIXME */
583 }
584 | operand_0 LOGIOR operand_1
585 {
586 write_exp_elt_opcode (BINOP_BITWISE_IOR);
587 }
588 | operand_0 ORIF operand_1
589 {
590 $$ = 0; /* FIXME */
591 }
592 | operand_0 LOGXOR operand_1
593 {
594 write_exp_elt_opcode (BINOP_BITWISE_XOR);
595 }
596 ;
597
598 /* Z.200, 5.3.4 */
599
600 operand_1 : operand_2
601 {
602 $$ = 0; /* FIXME */
603 }
604 | operand_1 LOGAND operand_2
605 {
606 write_exp_elt_opcode (BINOP_BITWISE_AND);
607 }
608 | operand_1 ANDIF operand_2
609 {
610 $$ = 0; /* FIXME */
611 }
612 ;
613
614 /* Z.200, 5.3.5 */
615
616 operand_2 : operand_3
617 {
618 $$ = 0; /* FIXME */
619 }
620 | operand_2 '=' operand_3
621 {
622 write_exp_elt_opcode (BINOP_EQUAL);
623 }
624 | operand_2 NOTEQUAL operand_3
625 {
626 write_exp_elt_opcode (BINOP_NOTEQUAL);
627 }
628 | operand_2 '>' operand_3
629 {
630 write_exp_elt_opcode (BINOP_GTR);
631 }
632 | operand_2 GTR operand_3
633 {
634 write_exp_elt_opcode (BINOP_GEQ);
635 }
636 | operand_2 '<' operand_3
637 {
638 write_exp_elt_opcode (BINOP_LESS);
639 }
640 | operand_2 LEQ operand_3
641 {
642 write_exp_elt_opcode (BINOP_LEQ);
643 }
644 | operand_2 IN operand_3
645 {
646 $$ = 0; /* FIXME */
647 }
648 ;
649
650
651 /* Z.200, 5.3.6 */
652
653 operand_3 : operand_4
654 {
655 $$ = 0; /* FIXME */
656 }
657 | operand_3 '+' operand_4
658 {
659 write_exp_elt_opcode (BINOP_ADD);
660 }
661 | operand_3 '-' operand_4
662 {
663 write_exp_elt_opcode (BINOP_SUB);
664 }
665 | operand_3 SLASH_SLASH operand_4
666 {
667 $$ = 0; /* FIXME */
668 }
669 ;
670
671 /* Z.200, 5.3.7 */
672
673 operand_4 : operand_5
674 {
675 $$ = 0; /* FIXME */
676 }
677 | operand_4 '*' operand_5
678 {
679 write_exp_elt_opcode (BINOP_MUL);
680 }
681 | operand_4 '/' operand_5
682 {
683 write_exp_elt_opcode (BINOP_DIV);
684 }
685 | operand_4 MOD operand_5
686 {
687 $$ = 0; /* FIXME */
688 }
689 | operand_4 REM operand_5
690 {
691 $$ = 0; /* FIXME */
692 }
693 ;
694
695 /* Z.200, 5.3.8 */
696
697 operand_5 : operand_6
698 {
699 $$ = 0; /* FIXME */
700 }
701 | '-' operand_6
702 {
703 write_exp_elt_opcode (UNOP_NEG);
704 }
705 | NOT operand_6
706 {
707 write_exp_elt_opcode (UNOP_LOGICAL_NOT);
708 }
709 | '(' integer_literal_expression ')' operand_6
710 {
711 $$ = 0; /* FIXME */
712 }
713 ;
714
715 /* Z.200, 5.3.9 */
716
717 operand_6 : POINTER location
718 {
719 $$ = 0; /* FIXME */
720 }
721 | RECEIVE buffer_location
722 {
723 $$ = 0; /* FIXME */
724 }
725 | primitive_value
726 {
727 $$ = 0; /* FIXME */
728 }
729 ;
730
731
732 /* Z.200, 12.4.3 */
733 /* FIXME: For now we just accept only a single integer literal. */
734
735 integer_literal_expression:
736 INTEGER_LITERAL
737 {
738 $$ = 0;
739 }
740
741 /* Things which still need productions... */
742 synonym_name : FIXME { $$ = 0; }
743 value_enumeration_name : FIXME { $$ = 0; }
744 value_do_with_name : FIXME { $$ = 0; }
745 value_receive_name : FIXME { $$ = 0; }
746 general_procedure_name : FIXME { $$ = 0; }
747 string_primitive_value : FIXME { $$ = 0; }
748 start_element : FIXME { $$ = 0; }
749 left_element : FIXME { $$ = 0; }
750 right_element : FIXME { $$ = 0; }
751 slice_size : FIXME { $$ = 0; }
752 array_primitive_value : FIXME { $$ = 0; }
753 expression_list : FIXME { $$ = 0; }
754 lower_element : FIXME { $$ = 0; }
755 upper_element : FIXME { $$ = 0; }
756 first_element : FIXME { $$ = 0; }
757 structure_primitive_value: FIXME { $$ = 0; }
758 field_name : FIXME { $$ = 0; }
759 mode_name : FIXME { $$ = 0; }
760 boolean_expression : FIXME { $$ = 0; }
761 case_selector_list : FIXME { $$ = 0; }
762 subexpression : FIXME { $$ = 0; }
763 case_label_specification: FIXME { $$ = 0; }
764 buffer_location : FIXME { $$ = 0; }
765
766 %%
767
768 static int
769 decode_integer_literal (valptr, tokptrptr)
770 int *valptr;
771 char **tokptrptr;
772 {
773 char *tokptr = *tokptrptr;
774 int base = 0;
775 int ival = 0;
776 int digits = 0;
777 int temp;
778 int explicit_base = 0;
779
780 /* Look for an explicit base specifier, which is optional. */
781
782 switch (*tokptr)
783 {
784 case 'd':
785 case 'D':
786 explicit_base++;
787 base = 10;
788 tokptr++;
789 break;
790 case 'b':
791 case 'B':
792 explicit_base++;
793 base = 2;
794 tokptr++;
795 break;
796 case 'h':
797 case 'H':
798 explicit_base++;
799 base = 16;
800 tokptr++;
801 break;
802 case 'o':
803 case 'O':
804 explicit_base++;
805 base = 8;
806 tokptr++;
807 break;
808 default:
809 base = 10;
810 break;
811 }
812
813 /* If we found an explicit base ensure that the character after the
814 explicit base is a single quote. */
815
816 if (explicit_base && (*tokptr++ != '\''))
817 {
818 return (0);
819 }
820
821 /* Start looking for a value composed of valid digits as set by the base
822 in use. Note that '_' characters are valid anywhere, in any quantity,
823 and are simply ignored. Since we must find at least one valid digit,
824 or reject this token as an integer literal, we keep track of how many
825 digits we have encountered. */
826
827 while (*tokptr != '\0')
828 {
829 temp = tolower (*tokptr);
830 tokptr++;
831 switch (temp)
832 {
833 case '_':
834 continue;
835 case '0': case '1': case '2': case '3': case '4':
836 case '5': case '6': case '7': case '8': case '9':
837 temp -= '0';
838 break;
839 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
840 temp -= 'a';
841 temp += 10;
842 break;
843 default:
844 temp = base;
845 break;
846 }
847 if (temp < base)
848 {
849 digits++;
850 ival *= base;
851 ival += temp;
852 }
853 else
854 {
855 /* Found something not in domain for current base. */
856 tokptr--; /* Unconsume what gave us indigestion. */
857 break;
858 }
859 }
860
861 /* If we didn't find any digits, then we don't have a valid integer
862 literal, so reject the entire token. Also, if we have an explicit
863 base, then the next character must not be a single quote, or we
864 have a bitstring literal, so reject the entire token in this case
865 as well. Otherwise, update the lexical scan pointer, and return
866 non-zero for success. */
867
868 if (digits == 0)
869 {
870 return (0);
871 }
872 else if (explicit_base && (*tokptr == '\''))
873 {
874 return (0);
875 }
876 else
877 {
878 *valptr = ival;
879 *tokptrptr = tokptr;
880 return (1);
881 }
882 }
883
884 /* Recognize a character literal. A character literal is single character
885 or a control sequence, enclosed in single quotes. A control sequence
886 is a comma separated list of one or more integer literals, enclosed
887 in parenthesis and introduced with a circumflex character.
888
889 EX: 'a' '^(7)' '^(7,8)'
890
891 Returns CHARACTER_LITERAL if a match is found.
892 */
893
894 static int
895 match_character_literal ()
896 {
897 char *tokptr = lexptr;
898 int ival = 0;
899
900 /* All character literals must start with a single quote. If we don't
901 find it, don't bother looking any further. */
902
903 if (*tokptr++ != '\'')
904 {
905 return (0);
906 }
907
908 /* Determine which form we have, either a control sequence or the
909 single character form. */
910
911 if ((*tokptr == '^') && (*(tokptr + 1) == '('))
912 {
913 /* Match and decode a control sequence. Return zero if we don't
914 find a valid integer literal, or if the next unconsumed character
915 after the integer literal is not the trailing ')'.
916 FIXME: We currently don't handle the multiple integer literal
917 form. */
918 tokptr += 2;
919 if (!decode_integer_literal (&ival, &tokptr) || (*tokptr++ != ')'))
920 {
921 return (0);
922 }
923 }
924 else
925 {
926 ival = *tokptr++;
927 }
928
929 /* The trailing quote has not yet been consumed. If we don't find
930 it, then we have no match. */
931
932 if (*tokptr++ != '\'')
933 {
934 return (0);
935 }
936
937 yylval.typed_val.val = ival;
938 yylval.typed_val.type = builtin_type_chill_char;
939 lexptr = tokptr;
940 return (CHARACTER_LITERAL);
941 }
942
943 /* Recognize an integer literal, as specified in Z.200 sec 5.2.4.2.
944 Note that according to 5.2.4.2, a single "_" is also a valid integer
945 literal, however GNU-chill requires there to be at least one "digit"
946 in any integer literal. */
947
948 static int
949 match_integer_literal ()
950 {
951 char *tokptr = lexptr;
952 int ival = 0;
953 int base = 0;
954 int digits = 0;
955 int temp;
956
957 /* Look for an explicit base specifier, which is optional. */
958
959 switch (*tokptr)
960 {
961 case 'd':
962 case 'D':
963 base = 10;
964 tokptr++;
965 break;
966 case 'b':
967 case 'B':
968 base = 2;
969 tokptr++;
970 break;
971 case 'h':
972 case 'H':
973 base = 16;
974 tokptr++;
975 break;
976 case 'o':
977 case 'O':
978 base = 8;
979 tokptr++;
980 break;
981 }
982
983 /* If we found no explicit base then default to 10, otherwise ensure
984 that the character after the explicit base is a single quote. */
985
986 if (base == 0)
987 {
988 base = 10;
989 }
990 else
991 {
992 if (*tokptr++ != '\'')
993 {
994 return (0);
995 }
996 }
997
998 /* Start looking for a value composed of valid digits as set by the base
999 in use. Note that '_' characters are valid anywhere, in any quantity,
1000 and are simply ignored. Since we must find at least one valid digit,
1001 or reject this token as an integer literal, we keep track of how many
1002 digits we have encountered. */
1003
1004 while (*tokptr != '\0')
1005 {
1006 temp = tolower (*tokptr);
1007 tokptr++;
1008 switch (temp)
1009 {
1010 case '_':
1011 continue;
1012 case '0': case '1': case '2': case '3': case '4':
1013 case '5': case '6': case '7': case '8': case '9':
1014 temp -= '0';
1015 break;
1016 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
1017 temp -= 'a';
1018 temp += 10;
1019 break;
1020 default:
1021 temp = base;
1022 break;
1023 }
1024 if (temp < base)
1025 {
1026 digits++;
1027 ival *= base;
1028 ival += temp;
1029 }
1030 else
1031 {
1032 /* Found something not in domain for current base. */
1033 tokptr--; /* Unconsume what gave us indigestion. */
1034 break;
1035 }
1036 }
1037
1038 /* If we didn't find any digits, then we don't have a valid integer
1039 literal, so reject the entire token. Otherwise, set up the parser
1040 variables, advance the current lexical scan pointer, and return the
1041 INTEGER_LITERAL token. */
1042
1043 if (digits == 0)
1044 {
1045 return (0);
1046 }
1047 else
1048 {
1049 yylval.typed_val.val = ival;
1050 yylval.typed_val.type = builtin_type_int;
1051 lexptr = tokptr;
1052 return (INTEGER_LITERAL);
1053 }
1054 }
1055
1056 static void convert_float ()
1057 {
1058 #if 0
1059 extern double strtod ();
1060 double d;
1061 char tmp[256];
1062 char *p = yytext, *p1 = tmp;
1063 char c;
1064
1065 while (c = *p++)
1066 {
1067 switch (c)
1068 {
1069 case '_':
1070 break;
1071 case 'E':
1072 case 'd':
1073 case 'D':
1074 *p1++ = 'e';
1075 break;
1076 default:
1077 *p1++ = c;
1078 break;
1079 }
1080 }
1081 *p1 = '\0';
1082 d = strtod (tmp, &p1);
1083 if (*p1)
1084 {
1085 /* add error handling here */
1086 ;
1087 }
1088 yylval.dval = d;
1089 #endif
1090 }
1091
1092 /* Take care of parsing a number (anything that starts with a digit).
1093 Set yylval and return the token type; update lexptr.
1094 LEN is the number of characters in it. */
1095
1096 /*** Needs some error checking for the float case ***/
1097
1098 static int
1099 parse_number ()
1100 {
1101 }
1102
1103 struct token
1104 {
1105 char *operator;
1106 int token;
1107 };
1108
1109 const static struct token tokentab5[] =
1110 {
1111 { "ANDIF", ANDIF }
1112 };
1113
1114 const static struct token tokentab4[] =
1115 {
1116 { "ORIF", ORIF }
1117 };
1118
1119 const static struct token tokentab3[] =
1120 {
1121 { "NOT", NOT },
1122 { "XOR", LOGXOR },
1123 { "AND", LOGAND }
1124 };
1125
1126 const static struct token tokentab2[] =
1127 {
1128 { "//", SLASH_SLASH },
1129 { "/=", NOTEQUAL },
1130 { "<=", LEQ },
1131 { ">=", GTR },
1132 { "IN", IN },
1133 { "OR", LOGIOR }
1134 };
1135
1136 /* Read one token, getting characters through lexptr. */
1137 /* This is where we will check to make sure that the language and the
1138 operators used are compatible. */
1139
1140 static int
1141 yylex ()
1142 {
1143 unsigned int i;
1144 int token;
1145
1146 /* Skip over any leading whitespace. */
1147 while (isspace (*lexptr))
1148 {
1149 lexptr++;
1150 }
1151 /* Look for special single character cases which can't be the first
1152 character of some other multicharacter token. */
1153 switch (*lexptr)
1154 {
1155 case '\0':
1156 return (0);
1157 case '.':
1158 case '=':
1159 case ':':
1160 case ';':
1161 case '!':
1162 case '+':
1163 case '-':
1164 case '*':
1165 case '/':
1166 case '(':
1167 case ')':
1168 case '[':
1169 case ']':
1170 return (*lexptr++);
1171 }
1172 /* Look for characters which start a particular kind of multicharacter
1173 token, such as a character literal. */
1174 switch (*lexptr)
1175 {
1176 case '^':
1177 case '\'':
1178 token = match_character_literal ();
1179 if (token != 0)
1180 {
1181 return (token);
1182 }
1183 break;
1184 }
1185 /* See if it is a special token of length 5. */
1186 for (i = 0; i < sizeof (tokentab5) / sizeof (tokentab5[0]); i++)
1187 {
1188 if (strncmp (lexptr, tokentab5[i].operator, 5) == 0)
1189 {
1190 lexptr += 5;
1191 return (tokentab5[i].token);
1192 }
1193 }
1194 /* See if it is a special token of length 4. */
1195 for (i = 0; i < sizeof (tokentab4) / sizeof (tokentab4[0]); i++)
1196 {
1197 if (strncmp (lexptr, tokentab4[i].operator, 4) == 0)
1198 {
1199 lexptr += 4;
1200 return (tokentab4[i].token);
1201 }
1202 }
1203 /* See if it is a special token of length 3. */
1204 for (i = 0; i < sizeof (tokentab3) / sizeof (tokentab3[0]); i++)
1205 {
1206 if (strncmp (lexptr, tokentab3[i].operator, 3) == 0)
1207 {
1208 lexptr += 3;
1209 return (tokentab3[i].token);
1210 }
1211 }
1212 /* See if it is a special token of length 2. */
1213 for (i = 0; i < sizeof (tokentab2) / sizeof (tokentab2[0]); i++)
1214 {
1215 if (strncmp (lexptr, tokentab2[i].operator, 2) == 0)
1216 {
1217 lexptr += 2;
1218 return (tokentab2[i].token);
1219 }
1220 }
1221 /* Look for single character cases which which could be the first
1222 character of some other multicharacter token, but aren't, or we
1223 would already have found it. */
1224 switch (*lexptr)
1225 {
1226 case '/':
1227 case '<':
1228 case '>':
1229 return (*lexptr++);
1230 }
1231 /* Look for other special tokens. */
1232 if (strncmp (lexptr, "TRUE", 4) == 0) /* FIXME: What about lowercase? */
1233 {
1234 yylval.ulval = 1;
1235 lexptr += 4;
1236 return (BOOLEAN_LITERAL);
1237 }
1238 if (strncmp (lexptr, "FALSE", 5) == 0) /* FIXME: What about lowercase? */
1239 {
1240 yylval.ulval = 0;
1241 lexptr += 5;
1242 return (BOOLEAN_LITERAL);
1243 }
1244 token = match_integer_literal ();
1245 if (token != 0);
1246 {
1247 return (token);
1248 }
1249 return (ILLEGAL_TOKEN);
1250 }
1251
1252 static void
1253 yyerror (msg)
1254 char *msg; /* unused */
1255 {
1256 printf ("Parsing: %s\n", lexptr);
1257 if (yychar < 256)
1258 {
1259 error ("Invalid syntax in expression near character '%c'.", yychar);
1260 }
1261 else
1262 {
1263 error ("Invalid syntax in expression");
1264 }
1265 }
1266
1267 \f
1268 /* Table of operators and their precedences for printing expressions. */
1269
1270 const static struct op_print chill_op_print_tab[] = {
1271 {"AND", BINOP_LOGICAL_AND, PREC_LOGICAL_AND, 0},
1272 {"OR", BINOP_LOGICAL_OR, PREC_LOGICAL_OR, 0},
1273 {"NOT", UNOP_LOGICAL_NOT, PREC_PREFIX, 0},
1274 {"MOD", BINOP_REM, PREC_MUL, 0},
1275 {":=", BINOP_ASSIGN, PREC_ASSIGN, 1},
1276 {"=", BINOP_EQUAL, PREC_EQUAL, 0},
1277 {"/=", BINOP_NOTEQUAL, PREC_EQUAL, 0},
1278 {"<=", BINOP_LEQ, PREC_ORDER, 0},
1279 {">=", BINOP_GEQ, PREC_ORDER, 0},
1280 {">", BINOP_GTR, PREC_ORDER, 0},
1281 {"<", BINOP_LESS, PREC_ORDER, 0},
1282 {"+", BINOP_ADD, PREC_ADD, 0},
1283 {"-", BINOP_SUB, PREC_ADD, 0},
1284 {"*", BINOP_MUL, PREC_MUL, 0},
1285 {"/", BINOP_DIV, PREC_MUL, 0},
1286 {"-", UNOP_NEG, PREC_PREFIX, 0},
1287 {NULL, 0, 0, 0}
1288 };
1289
1290 \f
1291 /* The built-in types of Chill. */
1292
1293 struct type *builtin_type_chill_bool;
1294 struct type *builtin_type_chill_char;
1295 struct type *builtin_type_chill_long;
1296 struct type *builtin_type_chill_ulong;
1297 struct type *builtin_type_chill_real;
1298
1299 struct type ** const (chill_builtin_types[]) =
1300 {
1301 &builtin_type_chill_bool,
1302 &builtin_type_chill_char,
1303 &builtin_type_chill_long,
1304 &builtin_type_chill_ulong,
1305 &builtin_type_chill_real,
1306 0
1307 };
1308
1309 const struct language_defn chill_language_defn = {
1310 "chill",
1311 language_chill,
1312 chill_builtin_types,
1313 range_check_on,
1314 type_check_on,
1315 chill_parse, /* parser */
1316 chill_error, /* parser error function */
1317 &BUILTIN_TYPE_LONGEST, /* longest signed integral type */
1318 &BUILTIN_TYPE_UNSIGNED_LONGEST,/* longest unsigned integral type */
1319 &builtin_type_chill_real, /* longest floating point type */
1320 {"", "B'", "", ""}, /* Binary format info */
1321 {"O'%o", "O'", "o", ""}, /* Octal format info */
1322 {"D'%d", "D'", "d", ""}, /* Decimal format info */
1323 {"H'%x", "H'", "x", ""}, /* Hex format info */
1324 chill_op_print_tab, /* expression operators for printing */
1325 LANG_MAGIC
1326 };
1327
1328 /* Initialization for Chill */
1329
1330 void
1331 _initialize_chill_exp ()
1332 {
1333 builtin_type_chill_bool =
1334 init_type (TYPE_CODE_BOOL, TARGET_INT_BIT / TARGET_CHAR_BIT,
1335 TYPE_FLAG_UNSIGNED,
1336 "BOOL", (struct objfile *) NULL);
1337 builtin_type_chill_char =
1338 init_type (TYPE_CODE_CHAR, TARGET_CHAR_BIT / TARGET_CHAR_BIT,
1339 TYPE_FLAG_UNSIGNED,
1340 "CHAR", (struct objfile *) NULL);
1341 builtin_type_chill_long =
1342 init_type (TYPE_CODE_INT, TARGET_LONG_BIT / TARGET_CHAR_BIT,
1343 0,
1344 "LONG", (struct objfile *) NULL);
1345 builtin_type_chill_ulong =
1346 init_type (TYPE_CODE_INT, TARGET_LONG_BIT / TARGET_CHAR_BIT,
1347 TYPE_FLAG_UNSIGNED,
1348 "ULONG", (struct objfile *) NULL);
1349 builtin_type_chill_real =
1350 init_type (TYPE_CODE_FLT, TARGET_DOUBLE_BIT / TARGET_CHAR_BIT,
1351 0,
1352 "LONG_REAL", (struct objfile *) NULL);
1353
1354 add_language (&chill_language_defn);
1355 }