]>
Commit | Line | Data |
---|---|---|
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 "defs.h" | |
57 | #include <ctype.h> | |
58 | #include "expression.h" | |
59 | #include "language.h" | |
60 | #include "value.h" | |
61 | #include "parser-defs.h" | |
62 | #include "ch-lang.h" | |
63 | ||
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. */ | |
70 | ||
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 | |
79 | #define yyr1 chill_r1 | |
80 | #define yyr2 chill_r2 | |
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 | |
88 | #define yyps chill_ps | |
89 | #define yypv chill_pv | |
90 | #define yys chill_s | |
91 | #define yy_yys chill_yys | |
92 | #define yystate chill_state | |
93 | #define yytmp chill_tmp | |
94 | #define yyv chill_v | |
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 */ | |
100 | ||
101 | #ifndef YYDEBUG | |
102 | #define YYDEBUG 0 /* Default to no yydebug support */ | |
103 | #endif | |
104 | ||
105 | int | |
106 | yyparse PARAMS ((void)); | |
107 | ||
108 | static int | |
109 | yylex PARAMS ((void)); | |
110 | ||
111 | void | |
112 | yyerror PARAMS ((char *)); | |
113 | ||
114 | %} | |
115 | ||
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. */ | |
119 | ||
120 | %union | |
121 | { | |
122 | LONGEST lval; | |
123 | unsigned LONGEST ulval; | |
124 | struct { | |
125 | LONGEST val; | |
126 | struct type *type; | |
127 | } typed_val; | |
128 | double dval; | |
129 | struct symbol *sym; | |
130 | struct type *tval; | |
131 | struct stoken sval; | |
132 | struct ttype tsym; | |
133 | struct symtoken ssym; | |
134 | int voidval; | |
135 | struct block *bval; | |
136 | enum exp_opcode opcode; | |
137 | struct internalvar *ivar; | |
138 | ||
139 | struct type **tvec; | |
140 | int *ivec; | |
141 | } | |
142 | ||
143 | %token <voidval> FIXME_01 | |
144 | %token <voidval> FIXME_02 | |
145 | %token <voidval> FIXME_03 | |
146 | %token <voidval> FIXME_04 | |
147 | %token <voidval> FIXME_05 | |
148 | %token <voidval> FIXME_06 | |
149 | %token <voidval> FIXME_07 | |
150 | %token <voidval> FIXME_08 | |
151 | %token <voidval> FIXME_09 | |
152 | %token <voidval> FIXME_10 | |
153 | %token <voidval> FIXME_11 | |
154 | %token <voidval> FIXME_12 | |
155 | %token <voidval> FIXME_13 | |
156 | %token <voidval> FIXME_14 | |
157 | %token <voidval> FIXME_15 | |
158 | %token <voidval> FIXME_16 | |
159 | %token <voidval> FIXME_17 | |
160 | %token <voidval> FIXME_18 | |
161 | %token <voidval> FIXME_19 | |
162 | %token <voidval> FIXME_20 | |
163 | %token <voidval> FIXME_21 | |
164 | %token <voidval> FIXME_22 | |
165 | %token <voidval> FIXME_24 | |
166 | %token <voidval> FIXME_25 | |
167 | %token <voidval> FIXME_26 | |
168 | %token <voidval> FIXME_27 | |
169 | %token <voidval> FIXME_28 | |
170 | %token <voidval> FIXME_29 | |
171 | %token <voidval> FIXME_30 | |
172 | ||
173 | %token <typed_val> INTEGER_LITERAL | |
174 | %token <ulval> BOOLEAN_LITERAL | |
175 | %token <typed_val> CHARACTER_LITERAL | |
176 | %token <dval> FLOAT_LITERAL | |
177 | %token <ssym> GENERAL_PROCEDURE_NAME | |
178 | %token <ssym> LOCATION_NAME | |
179 | %token <voidval> SET_LITERAL | |
180 | %token <voidval> EMPTINESS_LITERAL | |
181 | %token <sval> CHARACTER_STRING_LITERAL | |
182 | %token <sval> BIT_STRING_LITERAL | |
183 | %token <tsym> TYPENAME | |
184 | %token <sval> FIELD_NAME | |
185 | ||
186 | %token <voidval> '.' | |
187 | %token <voidval> ';' | |
188 | %token <voidval> ':' | |
189 | %token <voidval> CASE | |
190 | %token <voidval> OF | |
191 | %token <voidval> ESAC | |
192 | %token <voidval> LOGIOR | |
193 | %token <voidval> ORIF | |
194 | %token <voidval> LOGXOR | |
195 | %token <voidval> LOGAND | |
196 | %token <voidval> ANDIF | |
197 | %token <voidval> '=' | |
198 | %token <voidval> NOTEQUAL | |
199 | %token <voidval> '>' | |
200 | %token <voidval> GTR | |
201 | %token <voidval> '<' | |
202 | %token <voidval> LEQ | |
203 | %token <voidval> IN | |
204 | %token <voidval> '+' | |
205 | %token <voidval> '-' | |
206 | %token <voidval> '*' | |
207 | %token <voidval> '/' | |
208 | %token <voidval> SLASH_SLASH | |
209 | %token <voidval> MOD | |
210 | %token <voidval> REM | |
211 | %token <voidval> NOT | |
212 | %token <voidval> POINTER | |
213 | %token <voidval> RECEIVE | |
214 | %token <voidval> '[' | |
215 | %token <voidval> ']' | |
216 | %token <voidval> '(' | |
217 | %token <voidval> ')' | |
218 | %token <voidval> UP | |
219 | %token <voidval> IF | |
220 | %token <voidval> THEN | |
221 | %token <voidval> ELSE | |
222 | %token <voidval> FI | |
223 | %token <voidval> ELSIF | |
224 | %token <voidval> ILLEGAL_TOKEN | |
225 | %token <voidval> NUM | |
226 | %token <voidval> PRED | |
227 | %token <voidval> SUCC | |
228 | %token <voidval> ABS | |
229 | %token <voidval> CARD | |
230 | %token <voidval> MAX | |
231 | %token <voidval> MIN | |
232 | %token <voidval> SIZE | |
233 | %token <voidval> UPPER | |
234 | %token <voidval> LOWER | |
235 | %token <voidval> LENGTH | |
236 | ||
237 | /* Tokens which are not Chill tokens used in expressions, but rather GDB | |
238 | specific things that we recognize in the same context as Chill tokens | |
239 | (register names for example). */ | |
240 | ||
241 | %token <lval> GDB_REGNAME /* Machine register name */ | |
242 | %token <lval> GDB_LAST /* Value history */ | |
243 | %token <ivar> GDB_VARIABLE /* Convenience variable */ | |
244 | %token <voidval> GDB_ASSIGNMENT /* Assign value to somewhere */ | |
245 | ||
246 | %type <voidval> location | |
247 | %type <voidval> access_name | |
248 | %type <voidval> primitive_value | |
249 | %type <voidval> location_contents | |
250 | %type <voidval> value_name | |
251 | %type <voidval> literal | |
252 | %type <voidval> tuple | |
253 | %type <voidval> value_string_element | |
254 | %type <voidval> value_string_slice | |
255 | %type <voidval> value_array_element | |
256 | %type <voidval> value_array_slice | |
257 | %type <voidval> value_structure_field | |
258 | %type <voidval> expression_conversion | |
259 | %type <voidval> value_procedure_call | |
260 | %type <voidval> value_built_in_routine_call | |
261 | %type <voidval> chill_value_built_in_routine_call | |
262 | %type <voidval> start_expression | |
263 | %type <voidval> zero_adic_operator | |
264 | %type <voidval> parenthesised_expression | |
265 | %type <voidval> value | |
266 | %type <voidval> undefined_value | |
267 | %type <voidval> expression | |
268 | %type <voidval> conditional_expression | |
269 | %type <voidval> then_alternative | |
270 | %type <voidval> else_alternative | |
271 | %type <voidval> sub_expression | |
272 | %type <voidval> value_case_alternative | |
273 | %type <voidval> operand_0 | |
274 | %type <voidval> operand_1 | |
275 | %type <voidval> operand_2 | |
276 | %type <voidval> operand_3 | |
277 | %type <voidval> operand_4 | |
278 | %type <voidval> operand_5 | |
279 | %type <voidval> operand_6 | |
280 | %type <voidval> synonym_name | |
281 | %type <voidval> value_enumeration_name | |
282 | %type <voidval> value_do_with_name | |
283 | %type <voidval> value_receive_name | |
284 | %type <voidval> string_primitive_value | |
285 | %type <voidval> start_element | |
286 | %type <voidval> left_element | |
287 | %type <voidval> right_element | |
288 | %type <voidval> slice_size | |
289 | %type <voidval> array_primitive_value | |
290 | %type <voidval> expression_list | |
291 | %type <voidval> lower_element | |
292 | %type <voidval> upper_element | |
293 | %type <voidval> first_element | |
294 | %type <voidval> mode_argument | |
295 | %type <voidval> upper_lower_argument | |
296 | %type <voidval> length_argument | |
297 | %type <voidval> array_mode_name | |
298 | %type <voidval> string_mode_name | |
299 | %type <voidval> variant_structure_mode_name | |
300 | %type <voidval> boolean_expression | |
301 | %type <voidval> case_selector_list | |
302 | %type <voidval> subexpression | |
303 | %type <voidval> case_label_specification | |
304 | %type <voidval> buffer_location | |
305 | %type <voidval> single_assignment_action | |
306 | %type <tsym> mode_name | |
307 | ||
308 | %% | |
309 | ||
310 | /* Z.200, 5.3.1 */ | |
311 | ||
312 | start : value | |
313 | | mode_name | |
314 | { write_exp_elt_opcode(OP_TYPE); | |
315 | write_exp_elt_type($1.type); | |
316 | write_exp_elt_opcode(OP_TYPE);} | |
317 | ; | |
318 | ||
319 | value : expression | |
320 | { | |
321 | $$ = 0; /* FIXME */ | |
322 | } | |
323 | | undefined_value | |
324 | { | |
325 | $$ = 0; /* FIXME */ | |
326 | } | |
327 | ; | |
328 | ||
329 | undefined_value : FIXME_01 | |
330 | { | |
331 | $$ = 0; /* FIXME */ | |
332 | } | |
333 | ; | |
334 | ||
335 | /* Z.200, 4.2.1 */ | |
336 | ||
337 | location : access_name | |
338 | | primitive_value POINTER | |
339 | { | |
340 | write_exp_elt_opcode (UNOP_IND); | |
341 | } | |
342 | ; | |
343 | ||
344 | /* Z.200, 4.2.2 */ | |
345 | ||
346 | access_name : LOCATION_NAME | |
347 | { | |
348 | write_exp_elt_opcode (OP_VAR_VALUE); | |
349 | write_exp_elt_sym ($1.sym); | |
350 | write_exp_elt_opcode (OP_VAR_VALUE); | |
351 | } | |
352 | | GDB_LAST /* gdb specific */ | |
353 | { | |
354 | write_exp_elt_opcode (OP_LAST); | |
355 | write_exp_elt_longcst ($1); | |
356 | write_exp_elt_opcode (OP_LAST); | |
357 | } | |
358 | | GDB_REGNAME /* gdb specific */ | |
359 | { | |
360 | write_exp_elt_opcode (OP_REGISTER); | |
361 | write_exp_elt_longcst ($1); | |
362 | write_exp_elt_opcode (OP_REGISTER); | |
363 | } | |
364 | | GDB_VARIABLE /* gdb specific */ | |
365 | { | |
366 | write_exp_elt_opcode (OP_INTERNALVAR); | |
367 | write_exp_elt_intern ($1); | |
368 | write_exp_elt_opcode (OP_INTERNALVAR); | |
369 | } | |
370 | | FIXME_03 | |
371 | { | |
372 | $$ = 0; /* FIXME */ | |
373 | } | |
374 | ; | |
375 | ||
376 | /* Z.200, 4.2.8 */ | |
377 | ||
378 | expression_list : expression | |
379 | { | |
380 | arglist_len = 1; | |
381 | } | |
382 | | expression_list ',' expression | |
383 | { | |
384 | arglist_len++; | |
385 | } | |
386 | ||
387 | /* Z.200, 5.2.1 */ | |
388 | ||
389 | primitive_value : location_contents | |
390 | { | |
391 | $$ = 0; /* FIXME */ | |
392 | } | |
393 | | value_name | |
394 | { | |
395 | $$ = 0; /* FIXME */ | |
396 | } | |
397 | | literal | |
398 | { | |
399 | $$ = 0; /* FIXME */ | |
400 | } | |
401 | | tuple | |
402 | { | |
403 | $$ = 0; /* FIXME */ | |
404 | } | |
405 | | value_string_element | |
406 | { | |
407 | $$ = 0; /* FIXME */ | |
408 | } | |
409 | | value_string_slice | |
410 | { | |
411 | $$ = 0; /* FIXME */ | |
412 | } | |
413 | | value_array_element | |
414 | { | |
415 | $$ = 0; /* FIXME */ | |
416 | } | |
417 | | value_array_slice | |
418 | { | |
419 | $$ = 0; /* FIXME */ | |
420 | } | |
421 | | value_structure_field | |
422 | { | |
423 | $$ = 0; /* FIXME */ | |
424 | } | |
425 | | expression_conversion | |
426 | { | |
427 | $$ = 0; /* FIXME */ | |
428 | } | |
429 | | value_procedure_call | |
430 | { | |
431 | $$ = 0; /* FIXME */ | |
432 | } | |
433 | | value_built_in_routine_call | |
434 | { | |
435 | $$ = 0; /* FIXME */ | |
436 | } | |
437 | | start_expression | |
438 | { | |
439 | $$ = 0; /* FIXME */ | |
440 | } | |
441 | | zero_adic_operator | |
442 | { | |
443 | $$ = 0; /* FIXME */ | |
444 | } | |
445 | | parenthesised_expression | |
446 | { | |
447 | $$ = 0; /* FIXME */ | |
448 | } | |
449 | ; | |
450 | ||
451 | /* Z.200, 5.2.2 */ | |
452 | ||
453 | location_contents: location | |
454 | { | |
455 | $$ = 0; /* FIXME */ | |
456 | } | |
457 | ; | |
458 | ||
459 | /* Z.200, 5.2.3 */ | |
460 | ||
461 | value_name : synonym_name | |
462 | { | |
463 | $$ = 0; /* FIXME */ | |
464 | } | |
465 | | value_enumeration_name | |
466 | { | |
467 | $$ = 0; /* FIXME */ | |
468 | } | |
469 | | value_do_with_name | |
470 | { | |
471 | $$ = 0; /* FIXME */ | |
472 | } | |
473 | | value_receive_name | |
474 | { | |
475 | $$ = 0; /* FIXME */ | |
476 | } | |
477 | | GENERAL_PROCEDURE_NAME | |
478 | { | |
479 | write_exp_elt_opcode (OP_VAR_VALUE); | |
480 | write_exp_elt_sym ($1.sym); | |
481 | write_exp_elt_opcode (OP_VAR_VALUE); | |
482 | } | |
483 | ; | |
484 | ||
485 | /* Z.200, 5.2.4.1 */ | |
486 | ||
487 | literal : INTEGER_LITERAL | |
488 | { | |
489 | write_exp_elt_opcode (OP_LONG); | |
490 | write_exp_elt_type ($1.type); | |
491 | write_exp_elt_longcst ((LONGEST) ($1.val)); | |
492 | write_exp_elt_opcode (OP_LONG); | |
493 | } | |
494 | | BOOLEAN_LITERAL | |
495 | { | |
496 | write_exp_elt_opcode (OP_BOOL); | |
497 | write_exp_elt_longcst ((LONGEST) $1); | |
498 | write_exp_elt_opcode (OP_BOOL); | |
499 | } | |
500 | | CHARACTER_LITERAL | |
501 | { | |
502 | write_exp_elt_opcode (OP_LONG); | |
503 | write_exp_elt_type ($1.type); | |
504 | write_exp_elt_longcst ((LONGEST) ($1.val)); | |
505 | write_exp_elt_opcode (OP_LONG); | |
506 | } | |
507 | | FLOAT_LITERAL | |
508 | { | |
509 | write_exp_elt_opcode (OP_DOUBLE); | |
510 | write_exp_elt_type (builtin_type_double); | |
511 | write_exp_elt_dblcst ($1); | |
512 | write_exp_elt_opcode (OP_DOUBLE); | |
513 | } | |
514 | | SET_LITERAL | |
515 | { | |
516 | $$ = 0; /* FIXME */ | |
517 | } | |
518 | | EMPTINESS_LITERAL | |
519 | { | |
520 | $$ = 0; /* FIXME */ | |
521 | } | |
522 | | CHARACTER_STRING_LITERAL | |
523 | { | |
524 | write_exp_elt_opcode (OP_STRING); | |
525 | write_exp_string ($1); | |
526 | write_exp_elt_opcode (OP_STRING); | |
527 | } | |
528 | | BIT_STRING_LITERAL | |
529 | { | |
530 | write_exp_elt_opcode (OP_BITSTRING); | |
531 | write_exp_bitstring ($1); | |
532 | write_exp_elt_opcode (OP_BITSTRING); | |
533 | } | |
534 | ; | |
535 | ||
536 | /* Z.200, 5.2.5 */ | |
537 | ||
538 | tuple : FIXME_04 | |
539 | { | |
540 | $$ = 0; /* FIXME */ | |
541 | } | |
542 | ; | |
543 | ||
544 | ||
545 | /* Z.200, 5.2.6 */ | |
546 | ||
547 | value_string_element: string_primitive_value '(' start_element ')' | |
548 | { | |
549 | $$ = 0; /* FIXME */ | |
550 | } | |
551 | ; | |
552 | ||
553 | /* Z.200, 5.2.7 */ | |
554 | ||
555 | value_string_slice: string_primitive_value '(' left_element ':' right_element ')' | |
556 | { | |
557 | $$ = 0; /* FIXME */ | |
558 | } | |
559 | | string_primitive_value '(' start_element UP slice_size ')' | |
560 | { | |
561 | $$ = 0; /* FIXME */ | |
562 | } | |
563 | ; | |
564 | ||
565 | /* Z.200, 5.2.8 */ | |
566 | ||
567 | value_array_element: array_primitive_value '(' | |
568 | /* This is to save the value of arglist_len | |
569 | being accumulated for each dimension. */ | |
570 | { start_arglist (); } | |
571 | expression_list ')' | |
572 | { | |
573 | write_exp_elt_opcode (MULTI_SUBSCRIPT); | |
574 | write_exp_elt_longcst ((LONGEST) end_arglist ()); | |
575 | write_exp_elt_opcode (MULTI_SUBSCRIPT); | |
576 | } | |
577 | ; | |
578 | ||
579 | /* Z.200, 5.2.9 */ | |
580 | ||
581 | value_array_slice: array_primitive_value '(' lower_element ':' upper_element ')' | |
582 | { | |
583 | $$ = 0; /* FIXME */ | |
584 | } | |
585 | | array_primitive_value '(' first_element UP slice_size ')' | |
586 | { | |
587 | $$ = 0; /* FIXME */ | |
588 | } | |
589 | ; | |
590 | ||
591 | /* Z.200, 5.2.10 */ | |
592 | ||
593 | value_structure_field: primitive_value FIELD_NAME | |
594 | { write_exp_elt_opcode (STRUCTOP_STRUCT); | |
595 | write_exp_string ($2); | |
596 | write_exp_elt_opcode (STRUCTOP_STRUCT); | |
597 | } | |
598 | ; | |
599 | ||
600 | /* Z.200, 5.2.11 */ | |
601 | ||
602 | expression_conversion: mode_name parenthesised_expression | |
603 | { | |
604 | write_exp_elt_opcode (UNOP_CAST); | |
605 | write_exp_elt_type ($1.type); | |
606 | write_exp_elt_opcode (UNOP_CAST); | |
607 | } | |
608 | ; | |
609 | ||
610 | /* Z.200, 5.2.12 */ | |
611 | ||
612 | value_procedure_call: FIXME_05 | |
613 | { | |
614 | $$ = 0; /* FIXME */ | |
615 | } | |
616 | ; | |
617 | ||
618 | /* Z.200, 5.2.13 */ | |
619 | ||
620 | value_built_in_routine_call: chill_value_built_in_routine_call | |
621 | { | |
622 | $$ = 0; /* FIXME */ | |
623 | } | |
624 | ; | |
625 | ||
626 | /* Z.200, 5.2.14 */ | |
627 | ||
628 | start_expression: FIXME_06 | |
629 | { | |
630 | $$ = 0; /* FIXME */ | |
631 | } /* Not in GNU-Chill */ | |
632 | ; | |
633 | ||
634 | /* Z.200, 5.2.15 */ | |
635 | ||
636 | zero_adic_operator: FIXME_07 | |
637 | { | |
638 | $$ = 0; /* FIXME */ | |
639 | } | |
640 | ; | |
641 | ||
642 | /* Z.200, 5.2.16 */ | |
643 | ||
644 | parenthesised_expression: '(' expression ')' | |
645 | { | |
646 | $$ = 0; /* FIXME */ | |
647 | } | |
648 | ; | |
649 | ||
650 | /* Z.200, 5.3.2 */ | |
651 | ||
652 | expression : operand_0 | |
653 | { | |
654 | $$ = 0; /* FIXME */ | |
655 | } | |
656 | | single_assignment_action | |
657 | { | |
658 | $$ = 0; /* FIXME */ | |
659 | } | |
660 | | conditional_expression | |
661 | { | |
662 | $$ = 0; /* FIXME */ | |
663 | } | |
664 | ; | |
665 | ||
666 | conditional_expression : IF boolean_expression then_alternative else_alternative FI | |
667 | { | |
668 | $$ = 0; /* FIXME */ | |
669 | } | |
670 | | CASE case_selector_list OF value_case_alternative '[' ELSE sub_expression ']' ESAC | |
671 | { | |
672 | $$ = 0; /* FIXME */ | |
673 | } | |
674 | ; | |
675 | ||
676 | then_alternative: THEN subexpression | |
677 | { | |
678 | $$ = 0; /* FIXME */ | |
679 | } | |
680 | ; | |
681 | ||
682 | else_alternative: ELSE subexpression | |
683 | { | |
684 | $$ = 0; /* FIXME */ | |
685 | } | |
686 | | ELSIF boolean_expression then_alternative else_alternative | |
687 | { | |
688 | $$ = 0; /* FIXME */ | |
689 | } | |
690 | ; | |
691 | ||
692 | sub_expression : expression | |
693 | { | |
694 | $$ = 0; /* FIXME */ | |
695 | } | |
696 | ; | |
697 | ||
698 | value_case_alternative: case_label_specification ':' sub_expression ';' | |
699 | { | |
700 | $$ = 0; /* FIXME */ | |
701 | } | |
702 | ; | |
703 | ||
704 | /* Z.200, 5.3.3 */ | |
705 | ||
706 | operand_0 : operand_1 | |
707 | { | |
708 | $$ = 0; /* FIXME */ | |
709 | } | |
710 | | operand_0 LOGIOR operand_1 | |
711 | { | |
712 | write_exp_elt_opcode (BINOP_BITWISE_IOR); | |
713 | } | |
714 | | operand_0 ORIF operand_1 | |
715 | { | |
716 | $$ = 0; /* FIXME */ | |
717 | } | |
718 | | operand_0 LOGXOR operand_1 | |
719 | { | |
720 | write_exp_elt_opcode (BINOP_BITWISE_XOR); | |
721 | } | |
722 | ; | |
723 | ||
724 | /* Z.200, 5.3.4 */ | |
725 | ||
726 | operand_1 : operand_2 | |
727 | { | |
728 | $$ = 0; /* FIXME */ | |
729 | } | |
730 | | operand_1 LOGAND operand_2 | |
731 | { | |
732 | write_exp_elt_opcode (BINOP_BITWISE_AND); | |
733 | } | |
734 | | operand_1 ANDIF operand_2 | |
735 | { | |
736 | $$ = 0; /* FIXME */ | |
737 | } | |
738 | ; | |
739 | ||
740 | /* Z.200, 5.3.5 */ | |
741 | ||
742 | operand_2 : operand_3 | |
743 | { | |
744 | $$ = 0; /* FIXME */ | |
745 | } | |
746 | | operand_2 '=' operand_3 | |
747 | { | |
748 | write_exp_elt_opcode (BINOP_EQUAL); | |
749 | } | |
750 | | operand_2 NOTEQUAL operand_3 | |
751 | { | |
752 | write_exp_elt_opcode (BINOP_NOTEQUAL); | |
753 | } | |
754 | | operand_2 '>' operand_3 | |
755 | { | |
756 | write_exp_elt_opcode (BINOP_GTR); | |
757 | } | |
758 | | operand_2 GTR operand_3 | |
759 | { | |
760 | write_exp_elt_opcode (BINOP_GEQ); | |
761 | } | |
762 | | operand_2 '<' operand_3 | |
763 | { | |
764 | write_exp_elt_opcode (BINOP_LESS); | |
765 | } | |
766 | | operand_2 LEQ operand_3 | |
767 | { | |
768 | write_exp_elt_opcode (BINOP_LEQ); | |
769 | } | |
770 | | operand_2 IN operand_3 | |
771 | { | |
772 | $$ = 0; /* FIXME */ | |
773 | } | |
774 | ; | |
775 | ||
776 | ||
777 | /* Z.200, 5.3.6 */ | |
778 | ||
779 | operand_3 : operand_4 | |
780 | { | |
781 | $$ = 0; /* FIXME */ | |
782 | } | |
783 | | operand_3 '+' operand_4 | |
784 | { | |
785 | write_exp_elt_opcode (BINOP_ADD); | |
786 | } | |
787 | | operand_3 '-' operand_4 | |
788 | { | |
789 | write_exp_elt_opcode (BINOP_SUB); | |
790 | } | |
791 | | operand_3 SLASH_SLASH operand_4 | |
792 | { | |
793 | write_exp_elt_opcode (BINOP_CONCAT); | |
794 | } | |
795 | ; | |
796 | ||
797 | /* Z.200, 5.3.7 */ | |
798 | ||
799 | operand_4 : operand_5 | |
800 | { | |
801 | $$ = 0; /* FIXME */ | |
802 | } | |
803 | | operand_4 '*' operand_5 | |
804 | { | |
805 | write_exp_elt_opcode (BINOP_MUL); | |
806 | } | |
807 | | operand_4 '/' operand_5 | |
808 | { | |
809 | write_exp_elt_opcode (BINOP_DIV); | |
810 | } | |
811 | | operand_4 MOD operand_5 | |
812 | { | |
813 | write_exp_elt_opcode (BINOP_MOD); | |
814 | } | |
815 | | operand_4 REM operand_5 | |
816 | { | |
817 | write_exp_elt_opcode (BINOP_REM); | |
818 | } | |
819 | ; | |
820 | ||
821 | /* Z.200, 5.3.8 */ | |
822 | ||
823 | operand_5 : operand_6 | |
824 | { | |
825 | $$ = 0; /* FIXME */ | |
826 | } | |
827 | | '-' operand_6 | |
828 | { | |
829 | write_exp_elt_opcode (UNOP_NEG); | |
830 | } | |
831 | | NOT operand_6 | |
832 | { | |
833 | write_exp_elt_opcode (UNOP_LOGICAL_NOT); | |
834 | } | |
835 | | parenthesised_expression literal | |
836 | /* We require the string operand to be a literal, to avoid some | |
837 | nasty parsing ambiguities. */ | |
838 | { | |
839 | write_exp_elt_opcode (BINOP_CONCAT); | |
840 | } | |
841 | ; | |
842 | ||
843 | /* Z.200, 5.3.9 */ | |
844 | ||
845 | operand_6 : POINTER location | |
846 | { | |
847 | write_exp_elt_opcode (UNOP_ADDR); | |
848 | } | |
849 | | RECEIVE buffer_location | |
850 | { | |
851 | $$ = 0; /* FIXME */ | |
852 | } | |
853 | | primitive_value | |
854 | { | |
855 | $$ = 0; /* FIXME */ | |
856 | } | |
857 | ; | |
858 | ||
859 | ||
860 | /* Z.200, 6.2 */ | |
861 | ||
862 | single_assignment_action : | |
863 | location GDB_ASSIGNMENT value | |
864 | { | |
865 | write_exp_elt_opcode (BINOP_ASSIGN); | |
866 | } | |
867 | ; | |
868 | ||
869 | /* Z.200, 6.20.3 */ | |
870 | ||
871 | chill_value_built_in_routine_call : | |
872 | NUM '(' expression ')' | |
873 | { | |
874 | $$ = 0; /* FIXME */ | |
875 | } | |
876 | | PRED '(' expression ')' | |
877 | { | |
878 | $$ = 0; /* FIXME */ | |
879 | } | |
880 | | SUCC '(' expression ')' | |
881 | { | |
882 | $$ = 0; /* FIXME */ | |
883 | } | |
884 | | ABS '(' expression ')' | |
885 | { | |
886 | $$ = 0; /* FIXME */ | |
887 | } | |
888 | | CARD '(' expression ')' | |
889 | { | |
890 | $$ = 0; /* FIXME */ | |
891 | } | |
892 | | MAX '(' expression ')' | |
893 | { | |
894 | $$ = 0; /* FIXME */ | |
895 | } | |
896 | | MIN '(' expression ')' | |
897 | { | |
898 | $$ = 0; /* FIXME */ | |
899 | } | |
900 | | SIZE '(' location ')' | |
901 | { | |
902 | $$ = 0; /* FIXME */ | |
903 | } | |
904 | | SIZE '(' mode_argument ')' | |
905 | { | |
906 | $$ = 0; /* FIXME */ | |
907 | } | |
908 | | UPPER '(' upper_lower_argument ')' | |
909 | { | |
910 | $$ = 0; /* FIXME */ | |
911 | } | |
912 | | LOWER '(' upper_lower_argument ')' | |
913 | { | |
914 | $$ = 0; /* FIXME */ | |
915 | } | |
916 | | LENGTH '(' length_argument ')' | |
917 | { | |
918 | $$ = 0; /* FIXME */ | |
919 | } | |
920 | ; | |
921 | ||
922 | mode_argument : mode_name | |
923 | { | |
924 | $$ = 0; /* FIXME */ | |
925 | } | |
926 | | array_mode_name '(' expression ')' | |
927 | { | |
928 | $$ = 0; /* FIXME */ | |
929 | } | |
930 | | string_mode_name '(' expression ')' | |
931 | { | |
932 | $$ = 0; /* FIXME */ | |
933 | } | |
934 | | variant_structure_mode_name '(' expression_list ')' | |
935 | { | |
936 | $$ = 0; /* FIXME */ | |
937 | } | |
938 | ; | |
939 | ||
940 | mode_name : TYPENAME | |
941 | ; | |
942 | ||
943 | upper_lower_argument : expression | |
944 | { | |
945 | $$ = 0; /* FIXME */ | |
946 | } | |
947 | | mode_name | |
948 | { | |
949 | $$ = 0; /* FIXME */ | |
950 | } | |
951 | ; | |
952 | ||
953 | length_argument : expression | |
954 | { | |
955 | $$ = 0; /* FIXME */ | |
956 | } | |
957 | ; | |
958 | ||
959 | /* Z.200, 12.4.3 */ | |
960 | ||
961 | array_primitive_value : primitive_value | |
962 | { | |
963 | $$ = 0; | |
964 | } | |
965 | ; | |
966 | ||
967 | ||
968 | /* Things which still need productions... */ | |
969 | ||
970 | array_mode_name : FIXME_08 { $$ = 0; } | |
971 | string_mode_name : FIXME_09 { $$ = 0; } | |
972 | variant_structure_mode_name: FIXME_10 { $$ = 0; } | |
973 | synonym_name : FIXME_11 { $$ = 0; } | |
974 | value_enumeration_name : FIXME_12 { $$ = 0; } | |
975 | value_do_with_name : FIXME_13 { $$ = 0; } | |
976 | value_receive_name : FIXME_14 { $$ = 0; } | |
977 | string_primitive_value : FIXME_15 { $$ = 0; } | |
978 | start_element : FIXME_16 { $$ = 0; } | |
979 | left_element : FIXME_17 { $$ = 0; } | |
980 | right_element : FIXME_18 { $$ = 0; } | |
981 | slice_size : FIXME_19 { $$ = 0; } | |
982 | lower_element : FIXME_20 { $$ = 0; } | |
983 | upper_element : FIXME_21 { $$ = 0; } | |
984 | first_element : FIXME_22 { $$ = 0; } | |
985 | boolean_expression : FIXME_26 { $$ = 0; } | |
986 | case_selector_list : FIXME_27 { $$ = 0; } | |
987 | subexpression : FIXME_28 { $$ = 0; } | |
988 | case_label_specification: FIXME_29 { $$ = 0; } | |
989 | buffer_location : FIXME_30 { $$ = 0; } | |
990 | ||
991 | %% | |
992 | ||
993 | /* Implementation of a dynamically expandable buffer for processing input | |
994 | characters acquired through lexptr and building a value to return in | |
995 | yylval. */ | |
996 | ||
997 | static char *tempbuf; /* Current buffer contents */ | |
998 | static int tempbufsize; /* Size of allocated buffer */ | |
999 | static int tempbufindex; /* Current index into buffer */ | |
1000 | ||
1001 | #define GROWBY_MIN_SIZE 64 /* Minimum amount to grow buffer by */ | |
1002 | ||
1003 | #define CHECKBUF(size) \ | |
1004 | do { \ | |
1005 | if (tempbufindex + (size) >= tempbufsize) \ | |
1006 | { \ | |
1007 | growbuf_by_size (size); \ | |
1008 | } \ | |
1009 | } while (0); | |
1010 | ||
1011 | /* Grow the static temp buffer if necessary, including allocating the first one | |
1012 | on demand. */ | |
1013 | ||
1014 | static void | |
1015 | growbuf_by_size (count) | |
1016 | int count; | |
1017 | { | |
1018 | int growby; | |
1019 | ||
1020 | growby = max (count, GROWBY_MIN_SIZE); | |
1021 | tempbufsize += growby; | |
1022 | if (tempbuf == NULL) | |
1023 | { | |
1024 | tempbuf = (char *) malloc (tempbufsize); | |
1025 | } | |
1026 | else | |
1027 | { | |
1028 | tempbuf = (char *) realloc (tempbuf, tempbufsize); | |
1029 | } | |
1030 | } | |
1031 | ||
1032 | /* Try to consume a simple name string token. If successful, returns | |
1033 | a pointer to a nullbyte terminated copy of the name that can be used | |
1034 | in symbol table lookups. If not successful, returns NULL. */ | |
1035 | ||
1036 | static char * | |
1037 | match_simple_name_string () | |
1038 | { | |
1039 | char *tokptr = lexptr; | |
1040 | ||
1041 | if (isalpha (*tokptr)) | |
1042 | { | |
1043 | char *result; | |
1044 | do { | |
1045 | tokptr++; | |
1046 | } while (isalnum (*tokptr) || (*tokptr == '_')); | |
1047 | yylval.sval.ptr = lexptr; | |
1048 | yylval.sval.length = tokptr - lexptr; | |
1049 | lexptr = tokptr; | |
1050 | result = copy_name (yylval.sval); | |
1051 | for (tokptr = result; *tokptr; tokptr++) | |
1052 | if (isupper (*tokptr)) | |
1053 | *tokptr = tolower(*tokptr); | |
1054 | return result; | |
1055 | } | |
1056 | return (NULL); | |
1057 | } | |
1058 | ||
1059 | /* Start looking for a value composed of valid digits as set by the base | |
1060 | in use. Note that '_' characters are valid anywhere, in any quantity, | |
1061 | and are simply ignored. Since we must find at least one valid digit, | |
1062 | or reject this token as an integer literal, we keep track of how many | |
1063 | digits we have encountered. */ | |
1064 | ||
1065 | static int | |
1066 | decode_integer_value (base, tokptrptr, ivalptr) | |
1067 | int base; | |
1068 | char **tokptrptr; | |
1069 | int *ivalptr; | |
1070 | { | |
1071 | char *tokptr = *tokptrptr; | |
1072 | int temp; | |
1073 | int digits = 0; | |
1074 | ||
1075 | while (*tokptr != '\0') | |
1076 | { | |
1077 | temp = tolower (*tokptr); | |
1078 | tokptr++; | |
1079 | switch (temp) | |
1080 | { | |
1081 | case '_': | |
1082 | continue; | |
1083 | case '0': case '1': case '2': case '3': case '4': | |
1084 | case '5': case '6': case '7': case '8': case '9': | |
1085 | temp -= '0'; | |
1086 | break; | |
1087 | case 'a': case 'b': case 'c': case 'd': case 'e': case 'f': | |
1088 | temp -= 'a'; | |
1089 | temp += 10; | |
1090 | break; | |
1091 | default: | |
1092 | temp = base; | |
1093 | break; | |
1094 | } | |
1095 | if (temp < base) | |
1096 | { | |
1097 | digits++; | |
1098 | *ivalptr *= base; | |
1099 | *ivalptr += temp; | |
1100 | } | |
1101 | else | |
1102 | { | |
1103 | /* Found something not in domain for current base. */ | |
1104 | tokptr--; /* Unconsume what gave us indigestion. */ | |
1105 | break; | |
1106 | } | |
1107 | } | |
1108 | ||
1109 | /* If we didn't find any digits, then we don't have a valid integer | |
1110 | value, so reject the entire token. Otherwise, update the lexical | |
1111 | scan pointer, and return non-zero for success. */ | |
1112 | ||
1113 | if (digits == 0) | |
1114 | { | |
1115 | return (0); | |
1116 | } | |
1117 | else | |
1118 | { | |
1119 | *tokptrptr = tokptr; | |
1120 | return (1); | |
1121 | } | |
1122 | } | |
1123 | ||
1124 | static int | |
1125 | decode_integer_literal (valptr, tokptrptr) | |
1126 | int *valptr; | |
1127 | char **tokptrptr; | |
1128 | { | |
1129 | char *tokptr = *tokptrptr; | |
1130 | int base = 0; | |
1131 | int ival = 0; | |
1132 | int explicit_base = 0; | |
1133 | ||
1134 | /* Look for an explicit base specifier, which is optional. */ | |
1135 | ||
1136 | switch (*tokptr) | |
1137 | { | |
1138 | case 'd': | |
1139 | case 'D': | |
1140 | explicit_base++; | |
1141 | base = 10; | |
1142 | tokptr++; | |
1143 | break; | |
1144 | case 'b': | |
1145 | case 'B': | |
1146 | explicit_base++; | |
1147 | base = 2; | |
1148 | tokptr++; | |
1149 | break; | |
1150 | case 'h': | |
1151 | case 'H': | |
1152 | explicit_base++; | |
1153 | base = 16; | |
1154 | tokptr++; | |
1155 | break; | |
1156 | case 'o': | |
1157 | case 'O': | |
1158 | explicit_base++; | |
1159 | base = 8; | |
1160 | tokptr++; | |
1161 | break; | |
1162 | default: | |
1163 | base = 10; | |
1164 | break; | |
1165 | } | |
1166 | ||
1167 | /* If we found an explicit base ensure that the character after the | |
1168 | explicit base is a single quote. */ | |
1169 | ||
1170 | if (explicit_base && (*tokptr++ != '\'')) | |
1171 | { | |
1172 | return (0); | |
1173 | } | |
1174 | ||
1175 | /* Attempt to decode whatever follows as an integer value in the | |
1176 | indicated base, updating the token pointer in the process and | |
1177 | computing the value into ival. Also, if we have an explicit | |
1178 | base, then the next character must not be a single quote, or we | |
1179 | have a bitstring literal, so reject the entire token in this case. | |
1180 | Otherwise, update the lexical scan pointer, and return non-zero | |
1181 | for success. */ | |
1182 | ||
1183 | if (!decode_integer_value (base, &tokptr, &ival)) | |
1184 | { | |
1185 | return (0); | |
1186 | } | |
1187 | else if (explicit_base && (*tokptr == '\'')) | |
1188 | { | |
1189 | return (0); | |
1190 | } | |
1191 | else | |
1192 | { | |
1193 | *valptr = ival; | |
1194 | *tokptrptr = tokptr; | |
1195 | return (1); | |
1196 | } | |
1197 | } | |
1198 | ||
1199 | /* If it wasn't for the fact that floating point values can contain '_' | |
1200 | characters, we could just let strtod do all the hard work by letting it | |
1201 | try to consume as much of the current token buffer as possible and | |
1202 | find a legal conversion. Unfortunately we need to filter out the '_' | |
1203 | characters before calling strtod, which we do by copying the other | |
1204 | legal chars to a local buffer to be converted. However since we also | |
1205 | need to keep track of where the last unconsumed character in the input | |
1206 | buffer is, we have transfer only as many characters as may compose a | |
1207 | legal floating point value. */ | |
1208 | ||
1209 | static int | |
1210 | match_float_literal () | |
1211 | { | |
1212 | char *tokptr = lexptr; | |
1213 | char *buf; | |
1214 | char *copy; | |
1215 | char ch; | |
1216 | double dval; | |
1217 | extern double strtod (); | |
1218 | ||
1219 | /* Make local buffer in which to build the string to convert. This is | |
1220 | required because underscores are valid in chill floating point numbers | |
1221 | but not in the string passed to strtod to convert. The string will be | |
1222 | no longer than our input string. */ | |
1223 | ||
1224 | copy = buf = (char *) alloca (strlen (tokptr) + 1); | |
1225 | ||
1226 | /* Transfer all leading digits to the conversion buffer, discarding any | |
1227 | underscores. */ | |
1228 | ||
1229 | while (isdigit (*tokptr) || *tokptr == '_') | |
1230 | { | |
1231 | if (*tokptr != '_') | |
1232 | { | |
1233 | *copy++ = *tokptr; | |
1234 | } | |
1235 | tokptr++; | |
1236 | } | |
1237 | ||
1238 | /* Now accept either a '.', or one of [eEdD]. Dot is legal regardless | |
1239 | of whether we found any leading digits, and we simply accept it and | |
1240 | continue on to look for the fractional part and/or exponent. One of | |
1241 | [eEdD] is legal only if we have seen digits, and means that there | |
1242 | is no fractional part. If we find neither of these, then this is | |
1243 | not a floating point number, so return failure. */ | |
1244 | ||
1245 | switch (*tokptr++) | |
1246 | { | |
1247 | case '.': | |
1248 | /* Accept and then look for fractional part and/or exponent. */ | |
1249 | *copy++ = '.'; | |
1250 | break; | |
1251 | ||
1252 | case 'e': | |
1253 | case 'E': | |
1254 | case 'd': | |
1255 | case 'D': | |
1256 | if (copy == buf) | |
1257 | { | |
1258 | return (0); | |
1259 | } | |
1260 | *copy++ = 'e'; | |
1261 | goto collect_exponent; | |
1262 | break; | |
1263 | ||
1264 | default: | |
1265 | return (0); | |
1266 | break; | |
1267 | } | |
1268 | ||
1269 | /* We found a '.', copy any fractional digits to the conversion buffer, up | |
1270 | to the first nondigit, non-underscore character. */ | |
1271 | ||
1272 | while (isdigit (*tokptr) || *tokptr == '_') | |
1273 | { | |
1274 | if (*tokptr != '_') | |
1275 | { | |
1276 | *copy++ = *tokptr; | |
1277 | } | |
1278 | tokptr++; | |
1279 | } | |
1280 | ||
1281 | /* Look for an exponent, which must start with one of [eEdD]. If none | |
1282 | is found, jump directly to trying to convert what we have collected | |
1283 | so far. */ | |
1284 | ||
1285 | switch (*tokptr) | |
1286 | { | |
1287 | case 'e': | |
1288 | case 'E': | |
1289 | case 'd': | |
1290 | case 'D': | |
1291 | *copy++ = 'e'; | |
1292 | tokptr++; | |
1293 | break; | |
1294 | default: | |
1295 | goto convert_float; | |
1296 | break; | |
1297 | } | |
1298 | ||
1299 | /* Accept an optional '-' or '+' following one of [eEdD]. */ | |
1300 | ||
1301 | collect_exponent: | |
1302 | if (*tokptr == '+' || *tokptr == '-') | |
1303 | { | |
1304 | *copy++ = *tokptr++; | |
1305 | } | |
1306 | ||
1307 | /* Now copy an exponent into the conversion buffer. Note that at the | |
1308 | moment underscores are *not* allowed in exponents. */ | |
1309 | ||
1310 | while (isdigit (*tokptr)) | |
1311 | { | |
1312 | *copy++ = *tokptr++; | |
1313 | } | |
1314 | ||
1315 | /* If we transfered any chars to the conversion buffer, try to interpret its | |
1316 | contents as a floating point value. If any characters remain, then we | |
1317 | must not have a valid floating point string. */ | |
1318 | ||
1319 | convert_float: | |
1320 | *copy = '\0'; | |
1321 | if (copy != buf) | |
1322 | { | |
1323 | dval = strtod (buf, ©); | |
1324 | if (*copy == '\0') | |
1325 | { | |
1326 | yylval.dval = dval; | |
1327 | lexptr = tokptr; | |
1328 | return (FLOAT_LITERAL); | |
1329 | } | |
1330 | } | |
1331 | return (0); | |
1332 | } | |
1333 | ||
1334 | /* Recognize a string literal. A string literal is a nonzero sequence | |
1335 | of characters enclosed in matching single or double quotes, except that | |
1336 | a single character inside single quotes is a character literal, which | |
1337 | we reject as a string literal. To embed the terminator character inside | |
1338 | a string, it is simply doubled (I.E. "this""is""one""string") */ | |
1339 | ||
1340 | static int | |
1341 | match_string_literal () | |
1342 | { | |
1343 | char *tokptr = lexptr; | |
1344 | ||
1345 | for (tempbufindex = 0, tokptr++; *tokptr != '\0'; tokptr++) | |
1346 | { | |
1347 | CHECKBUF (1); | |
1348 | if (*tokptr == *lexptr) | |
1349 | { | |
1350 | if (*(tokptr + 1) == *lexptr) | |
1351 | { | |
1352 | tokptr++; | |
1353 | } | |
1354 | else | |
1355 | { | |
1356 | break; | |
1357 | } | |
1358 | } | |
1359 | tempbuf[tempbufindex++] = *tokptr; | |
1360 | } | |
1361 | if (*tokptr == '\0' /* no terminator */ | |
1362 | || tempbufindex == 0 /* no string */ | |
1363 | || (tempbufindex == 1 && *tokptr == '\'')) /* char literal */ | |
1364 | { | |
1365 | return (0); | |
1366 | } | |
1367 | else | |
1368 | { | |
1369 | tempbuf[tempbufindex] = '\0'; | |
1370 | yylval.sval.ptr = tempbuf; | |
1371 | yylval.sval.length = tempbufindex; | |
1372 | lexptr = ++tokptr; | |
1373 | return (CHARACTER_STRING_LITERAL); | |
1374 | } | |
1375 | } | |
1376 | ||
1377 | /* Recognize a character literal. A character literal is single character | |
1378 | or a control sequence, enclosed in single quotes. A control sequence | |
1379 | is a comma separated list of one or more integer literals, enclosed | |
1380 | in parenthesis and introduced with a circumflex character. | |
1381 | ||
1382 | EX: 'a' '^(7)' '^(7,8)' | |
1383 | ||
1384 | As a GNU chill extension, the syntax C'xx' is also recognized as a | |
1385 | character literal, where xx is a hex value for the character. | |
1386 | ||
1387 | Note that more than a single character, enclosed in single quotes, is | |
1388 | a string literal. | |
1389 | ||
1390 | Also note that the control sequence form is not in GNU Chill since it | |
1391 | is ambiguous with the string literal form using single quotes. I.E. | |
1392 | is '^(7)' a character literal or a string literal. In theory it it | |
1393 | possible to tell by context, but GNU Chill doesn't accept the control | |
1394 | sequence form, so neither do we (for now the code is disabled). | |
1395 | ||
1396 | Returns CHARACTER_LITERAL if a match is found. | |
1397 | */ | |
1398 | ||
1399 | static int | |
1400 | match_character_literal () | |
1401 | { | |
1402 | char *tokptr = lexptr; | |
1403 | int ival = 0; | |
1404 | ||
1405 | if ((tolower (*tokptr) == 'c') && (*(tokptr + 1) == '\'')) | |
1406 | { | |
1407 | /* We have a GNU chill extension form, so skip the leading "C'", | |
1408 | decode the hex value, and then ensure that we have a trailing | |
1409 | single quote character. */ | |
1410 | tokptr += 2; | |
1411 | if (!decode_integer_value (16, &tokptr, &ival) || (*tokptr != '\'')) | |
1412 | { | |
1413 | return (0); | |
1414 | } | |
1415 | tokptr++; | |
1416 | } | |
1417 | else if (*tokptr == '\'') | |
1418 | { | |
1419 | tokptr++; | |
1420 | ||
1421 | /* Determine which form we have, either a control sequence or the | |
1422 | single character form. */ | |
1423 | ||
1424 | if ((*tokptr == '^') && (*(tokptr + 1) == '(')) | |
1425 | { | |
1426 | #if 0 /* Disable, see note above. -fnf */ | |
1427 | /* Match and decode a control sequence. Return zero if we don't | |
1428 | find a valid integer literal, or if the next unconsumed character | |
1429 | after the integer literal is not the trailing ')'. | |
1430 | FIXME: We currently don't handle the multiple integer literal | |
1431 | form. */ | |
1432 | tokptr += 2; | |
1433 | if (!decode_integer_literal (&ival, &tokptr) || (*tokptr++ != ')')) | |
1434 | { | |
1435 | return (0); | |
1436 | } | |
1437 | #else | |
1438 | return (0); | |
1439 | #endif | |
1440 | } | |
1441 | else | |
1442 | { | |
1443 | ival = *tokptr++; | |
1444 | } | |
1445 | ||
1446 | /* The trailing quote has not yet been consumed. If we don't find | |
1447 | it, then we have no match. */ | |
1448 | ||
1449 | if (*tokptr++ != '\'') | |
1450 | { | |
1451 | return (0); | |
1452 | } | |
1453 | } | |
1454 | else | |
1455 | { | |
1456 | /* Not a character literal. */ | |
1457 | return (0); | |
1458 | } | |
1459 | yylval.typed_val.val = ival; | |
1460 | yylval.typed_val.type = builtin_type_chill_char; | |
1461 | lexptr = tokptr; | |
1462 | return (CHARACTER_LITERAL); | |
1463 | } | |
1464 | ||
1465 | /* Recognize an integer literal, as specified in Z.200 sec 5.2.4.2. | |
1466 | Note that according to 5.2.4.2, a single "_" is also a valid integer | |
1467 | literal, however GNU-chill requires there to be at least one "digit" | |
1468 | in any integer literal. */ | |
1469 | ||
1470 | static int | |
1471 | match_integer_literal () | |
1472 | { | |
1473 | char *tokptr = lexptr; | |
1474 | int ival; | |
1475 | ||
1476 | if (!decode_integer_literal (&ival, &tokptr)) | |
1477 | { | |
1478 | return (0); | |
1479 | } | |
1480 | else | |
1481 | { | |
1482 | yylval.typed_val.val = ival; | |
1483 | yylval.typed_val.type = builtin_type_int; | |
1484 | lexptr = tokptr; | |
1485 | return (INTEGER_LITERAL); | |
1486 | } | |
1487 | } | |
1488 | ||
1489 | /* Recognize a bit-string literal, as specified in Z.200 sec 5.2.4.8 | |
1490 | Note that according to 5.2.4.8, a single "_" is also a valid bit-string | |
1491 | literal, however GNU-chill requires there to be at least one "digit" | |
1492 | in any bit-string literal. */ | |
1493 | ||
1494 | static int | |
1495 | match_bitstring_literal () | |
1496 | { | |
1497 | char *tokptr = lexptr; | |
1498 | int mask; | |
1499 | int bitoffset = 0; | |
1500 | int bitcount = 0; | |
1501 | int base; | |
1502 | int digit; | |
1503 | ||
1504 | tempbufindex = 0; | |
1505 | ||
1506 | /* Look for the required explicit base specifier. */ | |
1507 | ||
1508 | switch (*tokptr++) | |
1509 | { | |
1510 | case 'b': | |
1511 | case 'B': | |
1512 | base = 2; | |
1513 | break; | |
1514 | case 'o': | |
1515 | case 'O': | |
1516 | base = 8; | |
1517 | break; | |
1518 | case 'h': | |
1519 | case 'H': | |
1520 | base = 16; | |
1521 | break; | |
1522 | default: | |
1523 | return (0); | |
1524 | break; | |
1525 | } | |
1526 | ||
1527 | /* Ensure that the character after the explicit base is a single quote. */ | |
1528 | ||
1529 | if (*tokptr++ != '\'') | |
1530 | { | |
1531 | return (0); | |
1532 | } | |
1533 | ||
1534 | while (*tokptr != '\0' && *tokptr != '\'') | |
1535 | { | |
1536 | digit = tolower (*tokptr); | |
1537 | tokptr++; | |
1538 | switch (digit) | |
1539 | { | |
1540 | case '_': | |
1541 | continue; | |
1542 | case '0': case '1': case '2': case '3': case '4': | |
1543 | case '5': case '6': case '7': case '8': case '9': | |
1544 | digit -= '0'; | |
1545 | break; | |
1546 | case 'a': case 'b': case 'c': case 'd': case 'e': case 'f': | |
1547 | digit -= 'a'; | |
1548 | digit += 10; | |
1549 | break; | |
1550 | default: | |
1551 | return (0); | |
1552 | break; | |
1553 | } | |
1554 | if (digit >= base) | |
1555 | { | |
1556 | /* Found something not in domain for current base. */ | |
1557 | return (0); | |
1558 | } | |
1559 | else | |
1560 | { | |
1561 | /* Extract bits from digit, starting with the msbit appropriate for | |
1562 | the current base, and packing them into the bitstring byte, | |
1563 | starting at the lsbit. */ | |
1564 | for (mask = (base >> 1); mask > 0; mask >>= 1) | |
1565 | { | |
1566 | bitcount++; | |
1567 | CHECKBUF (1); | |
1568 | if (digit & mask) | |
1569 | { | |
1570 | tempbuf[tempbufindex] |= (1 << bitoffset); | |
1571 | } | |
1572 | bitoffset++; | |
1573 | if (bitoffset == HOST_CHAR_BIT) | |
1574 | { | |
1575 | bitoffset = 0; | |
1576 | tempbufindex++; | |
1577 | } | |
1578 | } | |
1579 | } | |
1580 | } | |
1581 | ||
1582 | /* Verify that we consumed everything up to the trailing single quote, | |
1583 | and that we found some bits (IE not just underbars). */ | |
1584 | ||
1585 | if (*tokptr++ != '\'') | |
1586 | { | |
1587 | return (0); | |
1588 | } | |
1589 | else | |
1590 | { | |
1591 | yylval.sval.ptr = tempbuf; | |
1592 | yylval.sval.length = bitcount; | |
1593 | lexptr = tokptr; | |
1594 | return (BIT_STRING_LITERAL); | |
1595 | } | |
1596 | } | |
1597 | ||
1598 | /* Recognize tokens that start with '$'. These include: | |
1599 | ||
1600 | $regname A native register name or a "standard | |
1601 | register name". | |
1602 | Return token GDB_REGNAME. | |
1603 | ||
1604 | $variable A convenience variable with a name chosen | |
1605 | by the user. | |
1606 | Return token GDB_VARIABLE. | |
1607 | ||
1608 | $digits Value history with index <digits>, starting | |
1609 | from the first value which has index 1. | |
1610 | Return GDB_LAST. | |
1611 | ||
1612 | $$digits Value history with index <digits> relative | |
1613 | to the last value. I.E. $$0 is the last | |
1614 | value, $$1 is the one previous to that, $$2 | |
1615 | is the one previous to $$1, etc. | |
1616 | Return token GDB_LAST. | |
1617 | ||
1618 | $ | $0 | $$0 The last value in the value history. | |
1619 | Return token GDB_LAST. | |
1620 | ||
1621 | $$ An abbreviation for the second to the last | |
1622 | value in the value history, I.E. $$1 | |
1623 | Return token GDB_LAST. | |
1624 | ||
1625 | Note that we currently assume that register names and convenience | |
1626 | variables follow the convention of starting with a letter or '_'. | |
1627 | ||
1628 | */ | |
1629 | ||
1630 | static int | |
1631 | match_dollar_tokens () | |
1632 | { | |
1633 | char *tokptr; | |
1634 | int regno; | |
1635 | int namelength; | |
1636 | int negate; | |
1637 | int ival; | |
1638 | ||
1639 | /* We will always have a successful match, even if it is just for | |
1640 | a single '$', the abbreviation for $$0. So advance lexptr. */ | |
1641 | ||
1642 | tokptr = ++lexptr; | |
1643 | ||
1644 | if (*tokptr == '_' || isalpha (*tokptr)) | |
1645 | { | |
1646 | /* Look for a match with a native register name, usually something | |
1647 | like "r0" for example. */ | |
1648 | ||
1649 | for (regno = 0; regno < NUM_REGS; regno++) | |
1650 | { | |
1651 | namelength = strlen (reg_names[regno]); | |
1652 | if (STREQN (tokptr, reg_names[regno], namelength) | |
1653 | && !isalnum (tokptr[namelength])) | |
1654 | { | |
1655 | yylval.lval = regno; | |
1656 | lexptr += namelength + 1; | |
1657 | return (GDB_REGNAME); | |
1658 | } | |
1659 | } | |
1660 | ||
1661 | /* Look for a match with a standard register name, usually something | |
1662 | like "pc", which gdb always recognizes as the program counter | |
1663 | regardless of what the native register name is. */ | |
1664 | ||
1665 | for (regno = 0; regno < num_std_regs; regno++) | |
1666 | { | |
1667 | namelength = strlen (std_regs[regno].name); | |
1668 | if (STREQN (tokptr, std_regs[regno].name, namelength) | |
1669 | && !isalnum (tokptr[namelength])) | |
1670 | { | |
1671 | yylval.lval = std_regs[regno].regnum; | |
1672 | lexptr += namelength; | |
1673 | return (GDB_REGNAME); | |
1674 | } | |
1675 | } | |
1676 | ||
1677 | /* Attempt to match against a convenience variable. Note that | |
1678 | this will always succeed, because if no variable of that name | |
1679 | already exists, the lookup_internalvar will create one for us. | |
1680 | Also note that both lexptr and tokptr currently point to the | |
1681 | start of the input string we are trying to match, and that we | |
1682 | have already tested the first character for non-numeric, so we | |
1683 | don't have to treat it specially. */ | |
1684 | ||
1685 | while (*tokptr == '_' || isalnum (*tokptr)) | |
1686 | { | |
1687 | tokptr++; | |
1688 | } | |
1689 | yylval.sval.ptr = lexptr; | |
1690 | yylval.sval.length = tokptr - lexptr; | |
1691 | yylval.ivar = lookup_internalvar (copy_name (yylval.sval)); | |
1692 | lexptr = tokptr; | |
1693 | return (GDB_VARIABLE); | |
1694 | } | |
1695 | ||
1696 | /* Since we didn't match against a register name or convenience | |
1697 | variable, our only choice left is a history value. */ | |
1698 | ||
1699 | if (*tokptr == '$') | |
1700 | { | |
1701 | negate = 1; | |
1702 | ival = 1; | |
1703 | tokptr++; | |
1704 | } | |
1705 | else | |
1706 | { | |
1707 | negate = 0; | |
1708 | ival = 0; | |
1709 | } | |
1710 | ||
1711 | /* Attempt to decode more characters as an integer value giving | |
1712 | the index in the history list. If successful, the value will | |
1713 | overwrite ival (currently 0 or 1), and if not, ival will be | |
1714 | left alone, which is good since it is currently correct for | |
1715 | the '$' or '$$' case. */ | |
1716 | ||
1717 | decode_integer_literal (&ival, &tokptr); | |
1718 | yylval.lval = negate ? -ival : ival; | |
1719 | lexptr = tokptr; | |
1720 | return (GDB_LAST); | |
1721 | } | |
1722 | ||
1723 | struct token | |
1724 | { | |
1725 | char *operator; | |
1726 | int token; | |
1727 | }; | |
1728 | ||
1729 | static const struct token idtokentab[] = | |
1730 | { | |
1731 | { "length", LENGTH }, | |
1732 | { "lower", LOWER }, | |
1733 | { "upper", UPPER }, | |
1734 | { "andif", ANDIF }, | |
1735 | { "pred", PRED }, | |
1736 | { "succ", SUCC }, | |
1737 | { "card", CARD }, | |
1738 | { "size", SIZE }, | |
1739 | { "orif", ORIF }, | |
1740 | { "num", NUM }, | |
1741 | { "abs", ABS }, | |
1742 | { "max", MAX }, | |
1743 | { "min", MIN }, | |
1744 | { "mod", MOD }, | |
1745 | { "rem", REM }, | |
1746 | { "not", NOT }, | |
1747 | { "xor", LOGXOR }, | |
1748 | { "and", LOGAND }, | |
1749 | { "in", IN }, | |
1750 | { "or", LOGIOR } | |
1751 | }; | |
1752 | ||
1753 | static const struct token tokentab2[] = | |
1754 | { | |
1755 | { ":=", GDB_ASSIGNMENT }, | |
1756 | { "//", SLASH_SLASH }, | |
1757 | { "->", POINTER }, | |
1758 | { "/=", NOTEQUAL }, | |
1759 | { "<=", LEQ }, | |
1760 | { ">=", GTR } | |
1761 | }; | |
1762 | ||
1763 | /* Read one token, getting characters through lexptr. */ | |
1764 | /* This is where we will check to make sure that the language and the | |
1765 | operators used are compatible. */ | |
1766 | ||
1767 | static int | |
1768 | yylex () | |
1769 | { | |
1770 | unsigned int i; | |
1771 | int token; | |
1772 | char *simplename; | |
1773 | struct symbol *sym; | |
1774 | ||
1775 | /* Skip over any leading whitespace. */ | |
1776 | while (isspace (*lexptr)) | |
1777 | { | |
1778 | lexptr++; | |
1779 | } | |
1780 | /* Look for special single character cases which can't be the first | |
1781 | character of some other multicharacter token. */ | |
1782 | switch (*lexptr) | |
1783 | { | |
1784 | case '\0': | |
1785 | return (0); | |
1786 | case ',': | |
1787 | case '=': | |
1788 | case ';': | |
1789 | case '!': | |
1790 | case '+': | |
1791 | case '*': | |
1792 | case '(': | |
1793 | case ')': | |
1794 | case '[': | |
1795 | case ']': | |
1796 | return (*lexptr++); | |
1797 | } | |
1798 | /* Look for characters which start a particular kind of multicharacter | |
1799 | token, such as a character literal, register name, convenience | |
1800 | variable name, string literal, etc. */ | |
1801 | switch (*lexptr) | |
1802 | { | |
1803 | case '\'': | |
1804 | case '\"': | |
1805 | /* First try to match a string literal, which is any nonzero | |
1806 | sequence of characters enclosed in matching single or double | |
1807 | quotes, except that a single character inside single quotes | |
1808 | is a character literal, so we have to catch that case also. */ | |
1809 | token = match_string_literal (); | |
1810 | if (token != 0) | |
1811 | { | |
1812 | return (token); | |
1813 | } | |
1814 | if (*lexptr == '\'') | |
1815 | { | |
1816 | token = match_character_literal (); | |
1817 | if (token != 0) | |
1818 | { | |
1819 | return (token); | |
1820 | } | |
1821 | } | |
1822 | break; | |
1823 | case 'C': | |
1824 | case 'c': | |
1825 | token = match_character_literal (); | |
1826 | if (token != 0) | |
1827 | { | |
1828 | return (token); | |
1829 | } | |
1830 | break; | |
1831 | case '$': | |
1832 | token = match_dollar_tokens (); | |
1833 | if (token != 0) | |
1834 | { | |
1835 | return (token); | |
1836 | } | |
1837 | break; | |
1838 | } | |
1839 | /* See if it is a special token of length 2. */ | |
1840 | for (i = 0; i < sizeof (tokentab2) / sizeof (tokentab2[0]); i++) | |
1841 | { | |
1842 | if (STREQN (lexptr, tokentab2[i].operator, 2)) | |
1843 | { | |
1844 | lexptr += 2; | |
1845 | return (tokentab2[i].token); | |
1846 | } | |
1847 | } | |
1848 | /* Look for single character cases which which could be the first | |
1849 | character of some other multicharacter token, but aren't, or we | |
1850 | would already have found it. */ | |
1851 | switch (*lexptr) | |
1852 | { | |
1853 | case '-': | |
1854 | case ':': | |
1855 | case '/': | |
1856 | case '<': | |
1857 | case '>': | |
1858 | return (*lexptr++); | |
1859 | } | |
1860 | /* Look for a float literal before looking for an integer literal, so | |
1861 | we match as much of the input stream as possible. */ | |
1862 | token = match_float_literal (); | |
1863 | if (token != 0) | |
1864 | { | |
1865 | return (token); | |
1866 | } | |
1867 | token = match_bitstring_literal (); | |
1868 | if (token != 0) | |
1869 | { | |
1870 | return (token); | |
1871 | } | |
1872 | token = match_integer_literal (); | |
1873 | if (token != 0) | |
1874 | { | |
1875 | return (token); | |
1876 | } | |
1877 | ||
1878 | /* Try to match a simple name string, and if a match is found, then | |
1879 | further classify what sort of name it is and return an appropriate | |
1880 | token. Note that attempting to match a simple name string consumes | |
1881 | the token from lexptr, so we can't back out if we later find that | |
1882 | we can't classify what sort of name it is. */ | |
1883 | ||
1884 | simplename = match_simple_name_string (); | |
1885 | ||
1886 | /* See if it is a reserved identifier. */ | |
1887 | for (i = 0; i < sizeof (idtokentab) / sizeof (idtokentab[0]); i++) | |
1888 | { | |
1889 | if (STREQ (simplename, idtokentab[i].operator)) | |
1890 | { | |
1891 | return (idtokentab[i].token); | |
1892 | } | |
1893 | } | |
1894 | ||
1895 | /* Look for other special tokens. */ | |
1896 | if (STREQ (simplename, "true")) | |
1897 | { | |
1898 | yylval.ulval = 1; | |
1899 | return (BOOLEAN_LITERAL); | |
1900 | } | |
1901 | if (STREQ (simplename, "false")) | |
1902 | { | |
1903 | yylval.ulval = 0; | |
1904 | return (BOOLEAN_LITERAL); | |
1905 | } | |
1906 | ||
1907 | if (simplename != NULL) | |
1908 | { | |
1909 | sym = lookup_symbol (simplename, expression_context_block, | |
1910 | VAR_NAMESPACE, (int *) NULL, | |
1911 | (struct symtab **) NULL); | |
1912 | if (sym != NULL) | |
1913 | { | |
1914 | yylval.ssym.stoken.ptr = NULL; | |
1915 | yylval.ssym.stoken.length = 0; | |
1916 | yylval.ssym.sym = sym; | |
1917 | yylval.ssym.is_a_field_of_this = 0; /* FIXME, C++'ism */ | |
1918 | switch (SYMBOL_CLASS (sym)) | |
1919 | { | |
1920 | case LOC_BLOCK: | |
1921 | /* Found a procedure name. */ | |
1922 | return (GENERAL_PROCEDURE_NAME); | |
1923 | case LOC_STATIC: | |
1924 | /* Found a global or local static variable. */ | |
1925 | return (LOCATION_NAME); | |
1926 | case LOC_REGISTER: | |
1927 | case LOC_ARG: | |
1928 | case LOC_REF_ARG: | |
1929 | case LOC_REGPARM: | |
1930 | case LOC_LOCAL: | |
1931 | case LOC_LOCAL_ARG: | |
1932 | if (innermost_block == NULL | |
1933 | || contained_in (block_found, innermost_block)) | |
1934 | { | |
1935 | innermost_block = block_found; | |
1936 | } | |
1937 | return (LOCATION_NAME); | |
1938 | break; | |
1939 | case LOC_CONST: | |
1940 | case LOC_LABEL: | |
1941 | return (LOCATION_NAME); | |
1942 | break; | |
1943 | case LOC_TYPEDEF: | |
1944 | yylval.tsym.type = SYMBOL_TYPE (sym); | |
1945 | return TYPENAME; | |
1946 | case LOC_UNDEF: | |
1947 | case LOC_CONST_BYTES: | |
1948 | case LOC_OPTIMIZED_OUT: | |
1949 | error ("Symbol \"%s\" names no location.", simplename); | |
1950 | break; | |
1951 | } | |
1952 | } | |
1953 | else if (!have_full_symbols () && !have_partial_symbols ()) | |
1954 | { | |
1955 | error ("No symbol table is loaded. Use the \"file\" command."); | |
1956 | } | |
1957 | else | |
1958 | { | |
1959 | error ("No symbol \"%s\" in current context.", simplename); | |
1960 | } | |
1961 | } | |
1962 | ||
1963 | /* Catch single character tokens which are not part of some | |
1964 | longer token. */ | |
1965 | ||
1966 | switch (*lexptr) | |
1967 | { | |
1968 | case '.': /* Not float for example. */ | |
1969 | lexptr++; | |
1970 | while (isspace (*lexptr)) lexptr++; | |
1971 | simplename = match_simple_name_string (); | |
1972 | if (!simplename) | |
1973 | return '.'; | |
1974 | return FIELD_NAME; | |
1975 | } | |
1976 | ||
1977 | return (ILLEGAL_TOKEN); | |
1978 | } | |
1979 | ||
1980 | void | |
1981 | yyerror (msg) | |
1982 | char *msg; /* unused */ | |
1983 | { | |
1984 | printf ("Parsing: %s\n", lexptr); | |
1985 | if (yychar < 256) | |
1986 | { | |
1987 | error ("Invalid syntax in expression near character '%c'.", yychar); | |
1988 | } | |
1989 | else | |
1990 | { | |
1991 | error ("Invalid syntax in expression"); | |
1992 | } | |
1993 | } |