Commit | Line | Data |
---|---|---|
c44af4eb TT |
1 | /* Bison parser for Rust expressions, for GDB. |
2 | Copyright (C) 2016 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 3 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, see <http://www.gnu.org/licenses/>. */ | |
18 | ||
19 | /* Removing the last conflict seems difficult. */ | |
20 | %expect 1 | |
21 | ||
22 | %{ | |
23 | ||
24 | #include "defs.h" | |
25 | ||
26 | #include "block.h" | |
27 | #include "charset.h" | |
28 | #include "cp-support.h" | |
c44af4eb TT |
29 | #include "gdb_obstack.h" |
30 | #include "gdb_regex.h" | |
31 | #include "rust-lang.h" | |
32 | #include "parser-defs.h" | |
33 | #include "selftest.h" | |
34 | #include "value.h" | |
35 | #include "vec.h" | |
36 | ||
37 | #define GDB_YY_REMAP_PREFIX rust | |
38 | #include "yy-remap.h" | |
39 | ||
40 | #define RUSTSTYPE YYSTYPE | |
41 | ||
42 | extern initialize_file_ftype _initialize_rust_exp; | |
43 | ||
44 | struct rust_op; | |
45 | typedef const struct rust_op *rust_op_ptr; | |
46 | DEF_VEC_P (rust_op_ptr); | |
47 | ||
48 | /* A typed integer constant. */ | |
49 | ||
50 | struct typed_val_int | |
51 | { | |
52 | LONGEST val; | |
53 | struct type *type; | |
54 | }; | |
55 | ||
56 | /* A typed floating point constant. */ | |
57 | ||
58 | struct typed_val_float | |
59 | { | |
60 | DOUBLEST dval; | |
61 | struct type *type; | |
62 | }; | |
63 | ||
64 | /* An identifier and an expression. This is used to represent one | |
65 | element of a struct initializer. */ | |
66 | ||
67 | struct set_field | |
68 | { | |
69 | struct stoken name; | |
70 | const struct rust_op *init; | |
71 | }; | |
72 | ||
73 | typedef struct set_field set_field; | |
74 | ||
75 | DEF_VEC_O (set_field); | |
76 | ||
77 | ||
78 | static int rustyylex (void); | |
79 | static void rust_push_back (char c); | |
80 | static const char *rust_copy_name (const char *, int); | |
81 | static struct stoken rust_concat3 (const char *, const char *, const char *); | |
82 | static struct stoken make_stoken (const char *); | |
83 | static struct block_symbol rust_lookup_symbol (const char *name, | |
84 | const struct block *block, | |
85 | const domain_enum domain); | |
86 | static struct type *rust_lookup_type (const char *name, | |
87 | const struct block *block); | |
88 | static struct type *rust_type (const char *name); | |
89 | ||
90 | static const struct rust_op *crate_name (const struct rust_op *name); | |
91 | static const struct rust_op *super_name (const struct rust_op *name, | |
92 | unsigned int n_supers); | |
93 | ||
94 | static const struct rust_op *ast_operation (enum exp_opcode opcode, | |
95 | const struct rust_op *left, | |
96 | const struct rust_op *right); | |
97 | static const struct rust_op *ast_compound_assignment | |
98 | (enum exp_opcode opcode, const struct rust_op *left, | |
99 | const struct rust_op *rust_op); | |
100 | static const struct rust_op *ast_literal (struct typed_val_int val); | |
101 | static const struct rust_op *ast_dliteral (struct typed_val_float val); | |
102 | static const struct rust_op *ast_structop (const struct rust_op *left, | |
103 | const char *name, | |
104 | int completing); | |
105 | static const struct rust_op *ast_structop_anonymous | |
106 | (const struct rust_op *left, struct typed_val_int number); | |
107 | static const struct rust_op *ast_unary (enum exp_opcode opcode, | |
108 | const struct rust_op *expr); | |
109 | static const struct rust_op *ast_cast (const struct rust_op *expr, | |
110 | const struct rust_op *type); | |
111 | static const struct rust_op *ast_call_ish (enum exp_opcode opcode, | |
112 | const struct rust_op *expr, | |
113 | VEC (rust_op_ptr) **params); | |
114 | static const struct rust_op *ast_path (struct stoken name, | |
115 | VEC (rust_op_ptr) **params); | |
116 | static const struct rust_op *ast_string (struct stoken str); | |
117 | static const struct rust_op *ast_struct (const struct rust_op *name, | |
118 | VEC (set_field) **fields); | |
119 | static const struct rust_op *ast_range (const struct rust_op *lhs, | |
120 | const struct rust_op *rhs); | |
121 | static const struct rust_op *ast_array_type (const struct rust_op *lhs, | |
122 | struct typed_val_int val); | |
123 | static const struct rust_op *ast_slice_type (const struct rust_op *type); | |
124 | static const struct rust_op *ast_reference_type (const struct rust_op *type); | |
125 | static const struct rust_op *ast_pointer_type (const struct rust_op *type, | |
126 | int is_mut); | |
127 | static const struct rust_op *ast_function_type (const struct rust_op *result, | |
128 | VEC (rust_op_ptr) **params); | |
129 | static const struct rust_op *ast_tuple_type (VEC (rust_op_ptr) **params); | |
130 | ||
131 | /* The state of the parser, used internally when we are parsing the | |
132 | expression. */ | |
133 | ||
134 | static struct parser_state *pstate = NULL; | |
135 | ||
136 | /* A regular expression for matching Rust numbers. This is split up | |
137 | since it is very long and this gives us a way to comment the | |
138 | sections. */ | |
139 | ||
140 | static const char *number_regex_text = | |
141 | /* subexpression 1: allows use of alternation, otherwise uninteresting */ | |
142 | "^(" | |
143 | /* First comes floating point. */ | |
144 | /* Recognize number after the decimal point, with optional | |
145 | exponent and optional type suffix. | |
146 | subexpression 2: allows "?", otherwise uninteresting | |
147 | subexpression 3: if present, type suffix | |
148 | */ | |
149 | "[0-9][0-9_]*\\.[0-9][0-9_]*([eE][-+]?[0-9][0-9_]*)?(f32|f64)?" | |
150 | #define FLOAT_TYPE1 3 | |
151 | "|" | |
152 | /* Recognize exponent without decimal point, with optional type | |
153 | suffix. | |
154 | subexpression 4: if present, type suffix | |
155 | */ | |
156 | #define FLOAT_TYPE2 4 | |
157 | "[0-9][0-9_]*[eE][-+]?[0-9][0-9_]*(f32|f64)?" | |
158 | "|" | |
159 | /* "23." is a valid floating point number, but "23.e5" and | |
160 | "23.f32" are not. So, handle the trailing-. case | |
161 | separately. */ | |
162 | "[0-9][0-9_]*\\." | |
163 | "|" | |
164 | /* Finally come integers. | |
165 | subexpression 5: text of integer | |
166 | subexpression 6: if present, type suffix | |
167 | subexpression 7: allows use of alternation, otherwise uninteresting | |
168 | */ | |
169 | #define INT_TEXT 5 | |
170 | #define INT_TYPE 6 | |
171 | "(0x[a-fA-F0-9_]+|0o[0-7_]+|0b[01_]+|[0-9][0-9_]*)" | |
172 | "([iu](size|8|16|32|64))?" | |
173 | ")"; | |
174 | /* The number of subexpressions to allocate space for, including the | |
175 | "0th" whole match subexpression. */ | |
176 | #define NUM_SUBEXPRESSIONS 8 | |
177 | ||
178 | /* The compiled number-matching regex. */ | |
179 | ||
180 | static regex_t number_regex; | |
181 | ||
182 | /* True if we're running unit tests. */ | |
183 | ||
184 | static int unit_testing; | |
185 | ||
186 | /* Obstack for data temporarily allocated during parsing. */ | |
187 | ||
188 | static struct obstack work_obstack; | |
189 | ||
190 | /* Result of parsing. Points into work_obstack. */ | |
191 | ||
192 | static const struct rust_op *rust_ast; | |
193 | ||
194 | %} | |
195 | ||
196 | %union | |
197 | { | |
198 | /* A typed integer constant. */ | |
199 | struct typed_val_int typed_val_int; | |
200 | ||
201 | /* A typed floating point constant. */ | |
202 | struct typed_val_float typed_val_float; | |
203 | ||
204 | /* An identifier or string. */ | |
205 | struct stoken sval; | |
206 | ||
207 | /* A token representing an opcode, like "==". */ | |
208 | enum exp_opcode opcode; | |
209 | ||
210 | /* A list of expressions; for example, the arguments to a function | |
211 | call. */ | |
212 | VEC (rust_op_ptr) **params; | |
213 | ||
214 | /* A list of field initializers. */ | |
215 | VEC (set_field) **field_inits; | |
216 | ||
217 | /* A single field initializer. */ | |
218 | struct set_field one_field_init; | |
219 | ||
220 | /* An expression. */ | |
221 | const struct rust_op *op; | |
222 | ||
223 | /* A plain integer, for example used to count the number of | |
224 | "super::" prefixes on a path. */ | |
225 | unsigned int depth; | |
226 | } | |
227 | ||
228 | %{ | |
229 | ||
230 | /* Rust AST operations. We build a tree of these; then lower them | |
231 | to gdb expressions when parsing has completed. */ | |
232 | ||
233 | struct rust_op | |
234 | { | |
235 | /* The opcode. */ | |
236 | enum exp_opcode opcode; | |
237 | /* If OPCODE is OP_TYPE, then this holds information about what type | |
238 | is described by this node. */ | |
239 | enum type_code typecode; | |
240 | /* Indicates whether OPCODE actually represents a compound | |
241 | assignment. For example, if OPCODE is GTGT and this is false, | |
242 | then this rust_op represents an ordinary ">>"; but if this is | |
243 | true, then this rust_op represents ">>=". Unused in other | |
244 | cases. */ | |
245 | unsigned int compound_assignment : 1; | |
246 | /* Only used by a field expression; if set, indicates that the field | |
247 | name occurred at the end of the expression and is eligible for | |
248 | completion. */ | |
249 | unsigned int completing : 1; | |
250 | /* Operands of expression. Which one is used and how depends on the | |
251 | particular opcode. */ | |
252 | RUSTSTYPE left; | |
253 | RUSTSTYPE right; | |
254 | }; | |
255 | ||
256 | %} | |
257 | ||
258 | %token <sval> GDBVAR | |
259 | %token <sval> IDENT | |
260 | %token <sval> COMPLETE | |
261 | %token <typed_val_int> INTEGER | |
262 | %token <typed_val_int> DECIMAL_INTEGER | |
263 | %token <sval> STRING | |
264 | %token <sval> BYTESTRING | |
265 | %token <typed_val_float> FLOAT | |
266 | %token <opcode> COMPOUND_ASSIGN | |
267 | ||
268 | /* Keyword tokens. */ | |
269 | %token <voidval> KW_AS | |
270 | %token <voidval> KW_IF | |
271 | %token <voidval> KW_TRUE | |
272 | %token <voidval> KW_FALSE | |
273 | %token <voidval> KW_SUPER | |
274 | %token <voidval> KW_SELF | |
275 | %token <voidval> KW_MUT | |
276 | %token <voidval> KW_EXTERN | |
277 | %token <voidval> KW_CONST | |
278 | %token <voidval> KW_FN | |
279 | ||
280 | /* Operator tokens. */ | |
281 | %token <voidval> DOTDOT | |
282 | %token <voidval> OROR | |
283 | %token <voidval> ANDAND | |
284 | %token <voidval> EQEQ | |
285 | %token <voidval> NOTEQ | |
286 | %token <voidval> LTEQ | |
287 | %token <voidval> GTEQ | |
288 | %token <voidval> LSH RSH | |
289 | %token <voidval> COLONCOLON | |
290 | %token <voidval> ARROW | |
291 | ||
292 | %type <op> type | |
293 | %type <op> path_for_expr | |
294 | %type <op> identifier_path_for_expr | |
295 | %type <op> path_for_type | |
296 | %type <op> identifier_path_for_type | |
297 | %type <op> just_identifiers_for_type | |
298 | ||
299 | %type <params> maybe_type_list | |
300 | %type <params> type_list | |
301 | ||
302 | %type <depth> super_path | |
303 | ||
304 | %type <op> literal | |
305 | %type <op> expr | |
306 | %type <op> field_expr | |
307 | %type <op> idx_expr | |
308 | %type <op> unop_expr | |
309 | %type <op> binop_expr | |
310 | %type <op> binop_expr_expr | |
311 | %type <op> type_cast_expr | |
312 | %type <op> assignment_expr | |
313 | %type <op> compound_assignment_expr | |
314 | %type <op> paren_expr | |
315 | %type <op> call_expr | |
316 | %type <op> path_expr | |
317 | %type <op> tuple_expr | |
318 | %type <op> unit_expr | |
319 | %type <op> struct_expr | |
320 | %type <op> array_expr | |
321 | %type <op> range_expr | |
322 | ||
323 | %type <params> expr_list | |
324 | %type <params> maybe_expr_list | |
325 | %type <params> paren_expr_list | |
326 | ||
327 | %type <field_inits> struct_expr_list | |
328 | %type <one_field_init> struct_expr_tail | |
329 | ||
330 | /* Precedence. */ | |
331 | %nonassoc DOTDOT | |
332 | %right '=' COMPOUND_ASSIGN | |
333 | %left OROR | |
334 | %left ANDAND | |
335 | %nonassoc EQEQ NOTEQ '<' '>' LTEQ GTEQ | |
336 | %left '|' | |
337 | %left '^' | |
338 | %left '&' | |
339 | %left LSH RSH | |
340 | %left '@' | |
341 | %left '+' '-' | |
342 | %left '*' '/' '%' | |
343 | /* These could be %precedence in Bison, but that isn't a yacc | |
344 | feature. */ | |
345 | %left KW_AS | |
346 | %left UNARY | |
347 | %left '[' '.' '(' | |
348 | ||
349 | %% | |
350 | ||
351 | start: | |
352 | expr | |
353 | { | |
354 | /* If we are completing and see a valid parse, | |
355 | rust_ast will already have been set. */ | |
356 | if (rust_ast == NULL) | |
357 | rust_ast = $1; | |
358 | } | |
359 | ; | |
360 | ||
361 | /* Note that the Rust grammar includes a method_call_expr, but we | |
362 | handle this differently, to avoid a shift/reduce conflict with | |
363 | call_expr. */ | |
364 | expr: | |
365 | literal | |
366 | | path_expr | |
367 | | tuple_expr | |
368 | | unit_expr | |
369 | | struct_expr | |
370 | | field_expr | |
371 | | array_expr | |
372 | | idx_expr | |
373 | | range_expr | |
374 | | unop_expr | |
375 | | binop_expr | |
376 | | paren_expr | |
377 | | call_expr | |
378 | ; | |
379 | ||
380 | tuple_expr: | |
381 | '(' expr ',' maybe_expr_list ')' | |
382 | { | |
383 | VEC_safe_insert (rust_op_ptr, *$4, 0, $2); | |
384 | error (_("Tuple expressions not supported yet")); | |
385 | } | |
386 | ; | |
387 | ||
388 | unit_expr: | |
389 | '(' ')' | |
390 | { | |
391 | struct typed_val_int val; | |
392 | ||
393 | val.type | |
394 | = language_lookup_primitive_type (parse_language (pstate), | |
395 | parse_gdbarch (pstate), | |
396 | "()"); | |
397 | val.val = 0; | |
398 | $$ = ast_literal (val); | |
399 | } | |
400 | ; | |
401 | ||
402 | /* To avoid a shift/reduce conflict with call_expr, we don't handle | |
403 | tuple struct expressions here, but instead when examining the | |
404 | AST. */ | |
405 | struct_expr: | |
406 | path_for_expr '{' struct_expr_list '}' | |
407 | { $$ = ast_struct ($1, $3); } | |
408 | ; | |
409 | ||
410 | struct_expr_tail: | |
411 | DOTDOT expr | |
412 | { | |
413 | struct set_field sf; | |
414 | ||
415 | sf.name.ptr = NULL; | |
416 | sf.name.length = 0; | |
417 | sf.init = $2; | |
418 | ||
419 | $$ = sf; | |
420 | } | |
421 | | IDENT ':' expr | |
422 | { | |
423 | struct set_field sf; | |
424 | ||
425 | sf.name = $1; | |
426 | sf.init = $3; | |
427 | $$ = sf; | |
428 | } | |
429 | ; | |
430 | ||
c44af4eb | 431 | struct_expr_list: |
12df5c00 TT |
432 | /* %empty */ |
433 | { | |
434 | VEC (set_field) **result | |
435 | = OBSTACK_ZALLOC (&work_obstack, VEC (set_field) *); | |
436 | $$ = result; | |
437 | } | |
438 | | struct_expr_tail | |
c44af4eb TT |
439 | { |
440 | VEC (set_field) **result | |
441 | = OBSTACK_ZALLOC (&work_obstack, VEC (set_field) *); | |
442 | ||
443 | make_cleanup (VEC_cleanup (set_field), result); | |
444 | VEC_safe_push (set_field, *result, &$1); | |
445 | ||
446 | $$ = result; | |
447 | } | |
448 | | IDENT ':' expr ',' struct_expr_list | |
449 | { | |
450 | struct set_field sf; | |
451 | ||
452 | sf.name = $1; | |
453 | sf.init = $3; | |
454 | VEC_safe_push (set_field, *$5, &sf); | |
455 | $$ = $5; | |
456 | } | |
457 | ; | |
458 | ||
459 | array_expr: | |
460 | '[' KW_MUT expr_list ']' | |
461 | { $$ = ast_call_ish (OP_ARRAY, NULL, $3); } | |
462 | | '[' expr_list ']' | |
463 | { $$ = ast_call_ish (OP_ARRAY, NULL, $2); } | |
464 | | '[' KW_MUT expr ';' expr ']' | |
465 | { $$ = ast_operation (OP_RUST_ARRAY, $3, $5); } | |
466 | | '[' expr ';' expr ']' | |
467 | { $$ = ast_operation (OP_RUST_ARRAY, $2, $4); } | |
468 | ; | |
469 | ||
470 | range_expr: | |
471 | expr DOTDOT | |
472 | { $$ = ast_range ($1, NULL); } | |
473 | | expr DOTDOT expr | |
474 | { $$ = ast_range ($1, $3); } | |
475 | | DOTDOT expr | |
476 | { $$ = ast_range (NULL, $2); } | |
477 | | DOTDOT | |
478 | { $$ = ast_range (NULL, NULL); } | |
479 | ; | |
480 | ||
481 | literal: | |
482 | INTEGER | |
483 | { $$ = ast_literal ($1); } | |
484 | | DECIMAL_INTEGER | |
485 | { $$ = ast_literal ($1); } | |
486 | | FLOAT | |
487 | { $$ = ast_dliteral ($1); } | |
488 | | STRING | |
489 | { | |
490 | const struct rust_op *str = ast_string ($1); | |
491 | VEC (set_field) **fields; | |
492 | struct set_field field; | |
493 | struct typed_val_int val; | |
494 | struct stoken token; | |
495 | ||
496 | fields = OBSTACK_ZALLOC (&work_obstack, VEC (set_field) *); | |
497 | make_cleanup (VEC_cleanup (set_field), fields); | |
498 | ||
499 | /* Wrap the raw string in the &str struct. */ | |
500 | field.name.ptr = "data_ptr"; | |
501 | field.name.length = strlen (field.name.ptr); | |
502 | field.init = ast_unary (UNOP_ADDR, ast_string ($1)); | |
503 | VEC_safe_push (set_field, *fields, &field); | |
504 | ||
505 | val.type = rust_type ("usize"); | |
506 | val.val = $1.length; | |
507 | ||
508 | field.name.ptr = "length"; | |
509 | field.name.length = strlen (field.name.ptr); | |
510 | field.init = ast_literal (val); | |
511 | VEC_safe_push (set_field, *fields, &field); | |
512 | ||
513 | token.ptr = "&str"; | |
514 | token.length = strlen (token.ptr); | |
515 | $$ = ast_struct (ast_path (token, NULL), fields); | |
516 | } | |
517 | | BYTESTRING | |
518 | { $$ = ast_string ($1); } | |
519 | | KW_TRUE | |
520 | { | |
521 | struct typed_val_int val; | |
522 | ||
523 | val.type = language_bool_type (parse_language (pstate), | |
524 | parse_gdbarch (pstate)); | |
525 | val.val = 1; | |
526 | $$ = ast_literal (val); | |
527 | } | |
528 | | KW_FALSE | |
529 | { | |
530 | struct typed_val_int val; | |
531 | ||
532 | val.type = language_bool_type (parse_language (pstate), | |
533 | parse_gdbarch (pstate)); | |
534 | val.val = 0; | |
535 | $$ = ast_literal (val); | |
536 | } | |
537 | ; | |
538 | ||
539 | field_expr: | |
540 | expr '.' IDENT | |
541 | { $$ = ast_structop ($1, $3.ptr, 0); } | |
542 | | expr '.' COMPLETE | |
543 | { | |
544 | $$ = ast_structop ($1, $3.ptr, 1); | |
545 | rust_ast = $$; | |
546 | } | |
547 | | expr '.' DECIMAL_INTEGER | |
548 | { $$ = ast_structop_anonymous ($1, $3); } | |
549 | ; | |
550 | ||
551 | idx_expr: | |
552 | expr '[' expr ']' | |
553 | { $$ = ast_operation (BINOP_SUBSCRIPT, $1, $3); } | |
554 | ; | |
555 | ||
556 | unop_expr: | |
557 | '+' expr %prec UNARY | |
558 | { $$ = ast_unary (UNOP_PLUS, $2); } | |
559 | ||
560 | | '-' expr %prec UNARY | |
561 | { $$ = ast_unary (UNOP_NEG, $2); } | |
562 | ||
563 | | '!' expr %prec UNARY | |
564 | { | |
565 | /* Note that we provide a Rust-specific evaluator | |
566 | override for UNOP_COMPLEMENT, so it can do the | |
567 | right thing for both bool and integral | |
568 | values. */ | |
569 | $$ = ast_unary (UNOP_COMPLEMENT, $2); | |
570 | } | |
571 | ||
572 | | '*' expr %prec UNARY | |
573 | { $$ = ast_unary (UNOP_IND, $2); } | |
574 | ||
575 | | '&' expr %prec UNARY | |
576 | { $$ = ast_unary (UNOP_ADDR, $2); } | |
577 | ||
578 | | '&' KW_MUT expr %prec UNARY | |
579 | { $$ = ast_unary (UNOP_ADDR, $3); } | |
580 | ||
581 | ; | |
582 | ||
583 | binop_expr: | |
584 | binop_expr_expr | |
585 | | type_cast_expr | |
586 | | assignment_expr | |
587 | | compound_assignment_expr | |
588 | ; | |
589 | ||
590 | binop_expr_expr: | |
591 | expr '*' expr | |
592 | { $$ = ast_operation (BINOP_MUL, $1, $3); } | |
593 | ||
594 | | expr '@' expr | |
595 | { $$ = ast_operation (BINOP_REPEAT, $1, $3); } | |
596 | ||
597 | | expr '/' expr | |
598 | { $$ = ast_operation (BINOP_DIV, $1, $3); } | |
599 | ||
600 | | expr '%' expr | |
601 | { $$ = ast_operation (BINOP_REM, $1, $3); } | |
602 | ||
603 | | expr '<' expr | |
604 | { $$ = ast_operation (BINOP_LESS, $1, $3); } | |
605 | ||
606 | | expr '>' expr | |
607 | { $$ = ast_operation (BINOP_GTR, $1, $3); } | |
608 | ||
609 | | expr '&' expr | |
610 | { $$ = ast_operation (BINOP_BITWISE_AND, $1, $3); } | |
611 | ||
612 | | expr '|' expr | |
613 | { $$ = ast_operation (BINOP_BITWISE_IOR, $1, $3); } | |
614 | ||
615 | | expr '^' expr | |
616 | { $$ = ast_operation (BINOP_BITWISE_XOR, $1, $3); } | |
617 | ||
618 | | expr '+' expr | |
619 | { $$ = ast_operation (BINOP_ADD, $1, $3); } | |
620 | ||
621 | | expr '-' expr | |
622 | { $$ = ast_operation (BINOP_SUB, $1, $3); } | |
623 | ||
624 | | expr OROR expr | |
625 | { $$ = ast_operation (BINOP_LOGICAL_OR, $1, $3); } | |
626 | ||
627 | | expr ANDAND expr | |
628 | { $$ = ast_operation (BINOP_LOGICAL_AND, $1, $3); } | |
629 | ||
630 | | expr EQEQ expr | |
631 | { $$ = ast_operation (BINOP_EQUAL, $1, $3); } | |
632 | ||
633 | | expr NOTEQ expr | |
634 | { $$ = ast_operation (BINOP_NOTEQUAL, $1, $3); } | |
635 | ||
636 | | expr LTEQ expr | |
637 | { $$ = ast_operation (BINOP_LEQ, $1, $3); } | |
638 | ||
639 | | expr GTEQ expr | |
640 | { $$ = ast_operation (BINOP_GEQ, $1, $3); } | |
641 | ||
642 | | expr LSH expr | |
643 | { $$ = ast_operation (BINOP_LSH, $1, $3); } | |
644 | ||
645 | | expr RSH expr | |
646 | { $$ = ast_operation (BINOP_RSH, $1, $3); } | |
647 | ; | |
648 | ||
649 | type_cast_expr: | |
650 | expr KW_AS type | |
651 | { $$ = ast_cast ($1, $3); } | |
652 | ; | |
653 | ||
654 | assignment_expr: | |
655 | expr '=' expr | |
656 | { $$ = ast_operation (BINOP_ASSIGN, $1, $3); } | |
657 | ; | |
658 | ||
659 | compound_assignment_expr: | |
660 | expr COMPOUND_ASSIGN expr | |
661 | { $$ = ast_compound_assignment ($2, $1, $3); } | |
662 | ||
663 | ; | |
664 | ||
665 | paren_expr: | |
666 | '(' expr ')' | |
667 | { $$ = $2; } | |
668 | ; | |
669 | ||
670 | expr_list: | |
671 | expr | |
672 | { | |
673 | $$ = OBSTACK_ZALLOC (&work_obstack, VEC (rust_op_ptr) *); | |
674 | make_cleanup (VEC_cleanup (rust_op_ptr), $$); | |
675 | VEC_safe_push (rust_op_ptr, *$$, $1); | |
676 | } | |
677 | | expr_list ',' expr | |
678 | { | |
679 | VEC_safe_push (rust_op_ptr, *$1, $3); | |
680 | $$ = $1; | |
681 | } | |
682 | ; | |
683 | ||
684 | maybe_expr_list: | |
685 | /* %empty */ | |
686 | { | |
687 | /* The result can't be NULL. */ | |
688 | $$ = OBSTACK_ZALLOC (&work_obstack, VEC (rust_op_ptr) *); | |
689 | make_cleanup (VEC_cleanup (rust_op_ptr), $$); | |
690 | } | |
691 | | expr_list | |
692 | { $$ = $1; } | |
693 | ; | |
694 | ||
695 | paren_expr_list: | |
696 | '(' | |
697 | maybe_expr_list | |
698 | ')' | |
699 | { $$ = $2; } | |
700 | ; | |
701 | ||
702 | call_expr: | |
703 | expr paren_expr_list | |
704 | { $$ = ast_call_ish (OP_FUNCALL, $1, $2); } | |
705 | ; | |
706 | ||
707 | maybe_self_path: | |
708 | /* %empty */ | |
709 | | KW_SELF COLONCOLON | |
710 | ; | |
711 | ||
712 | super_path: | |
713 | KW_SUPER COLONCOLON | |
714 | { $$ = 1; } | |
715 | | super_path KW_SUPER COLONCOLON | |
716 | { $$ = $1 + 1; } | |
717 | ; | |
718 | ||
719 | path_expr: | |
720 | path_for_expr | |
721 | { $$ = $1; } | |
722 | | GDBVAR | |
723 | { $$ = ast_path ($1, NULL); } | |
724 | | KW_SELF | |
725 | { $$ = ast_path (make_stoken ("self"), NULL); } | |
726 | ; | |
727 | ||
728 | path_for_expr: | |
729 | identifier_path_for_expr | |
730 | | KW_SELF COLONCOLON identifier_path_for_expr | |
731 | { $$ = super_name ($3, 0); } | |
732 | | maybe_self_path super_path identifier_path_for_expr | |
733 | { $$ = super_name ($3, $2); } | |
734 | | COLONCOLON identifier_path_for_expr | |
735 | { $$ = crate_name ($2); } | |
736 | | KW_EXTERN identifier_path_for_expr | |
737 | { | |
738 | /* This is a gdb extension to make it possible to | |
739 | refer to items in other crates. It just bypasses | |
740 | adding the current crate to the front of the | |
741 | name. */ | |
742 | $$ = ast_path (rust_concat3 ("::", $2->left.sval.ptr, NULL), | |
743 | $2->right.params); | |
744 | } | |
745 | ; | |
746 | ||
747 | identifier_path_for_expr: | |
748 | IDENT | |
749 | { $$ = ast_path ($1, NULL); } | |
750 | | identifier_path_for_expr COLONCOLON IDENT | |
751 | { | |
752 | $$ = ast_path (rust_concat3 ($1->left.sval.ptr, "::", | |
753 | $3.ptr), | |
754 | NULL); | |
755 | } | |
756 | | identifier_path_for_expr COLONCOLON '<' type_list '>' | |
757 | { $$ = ast_path ($1->left.sval, $4); } | |
758 | | identifier_path_for_expr COLONCOLON '<' type_list RSH | |
759 | { | |
760 | $$ = ast_path ($1->left.sval, $4); | |
761 | rust_push_back ('>'); | |
762 | } | |
763 | ; | |
764 | ||
765 | path_for_type: | |
766 | identifier_path_for_type | |
767 | | KW_SELF COLONCOLON identifier_path_for_type | |
768 | { $$ = super_name ($3, 0); } | |
769 | | maybe_self_path super_path identifier_path_for_type | |
770 | { $$ = super_name ($3, $2); } | |
771 | | COLONCOLON identifier_path_for_type | |
772 | { $$ = crate_name ($2); } | |
773 | | KW_EXTERN identifier_path_for_type | |
774 | { | |
775 | /* This is a gdb extension to make it possible to | |
776 | refer to items in other crates. It just bypasses | |
777 | adding the current crate to the front of the | |
778 | name. */ | |
779 | $$ = ast_path (rust_concat3 ("::", $2->left.sval.ptr, NULL), | |
780 | $2->right.params); | |
781 | } | |
782 | ; | |
783 | ||
784 | just_identifiers_for_type: | |
785 | IDENT | |
786 | { $$ = ast_path ($1, NULL); } | |
787 | | just_identifiers_for_type COLONCOLON IDENT | |
788 | { | |
789 | $$ = ast_path (rust_concat3 ($1->left.sval.ptr, "::", | |
790 | $3.ptr), | |
791 | NULL); | |
792 | } | |
793 | ; | |
794 | ||
795 | identifier_path_for_type: | |
796 | just_identifiers_for_type | |
797 | | just_identifiers_for_type '<' type_list '>' | |
798 | { $$ = ast_path ($1->left.sval, $3); } | |
799 | | just_identifiers_for_type '<' type_list RSH | |
800 | { | |
801 | $$ = ast_path ($1->left.sval, $3); | |
802 | rust_push_back ('>'); | |
803 | } | |
804 | ; | |
805 | ||
806 | type: | |
807 | path_for_type | |
808 | | '[' type ';' INTEGER ']' | |
809 | { $$ = ast_array_type ($2, $4); } | |
810 | | '[' type ';' DECIMAL_INTEGER ']' | |
811 | { $$ = ast_array_type ($2, $4); } | |
812 | | '&' '[' type ']' | |
813 | { $$ = ast_slice_type ($3); } | |
814 | | '&' type | |
815 | { $$ = ast_reference_type ($2); } | |
816 | | '*' KW_MUT type | |
817 | { $$ = ast_pointer_type ($3, 1); } | |
818 | | '*' KW_CONST type | |
819 | { $$ = ast_pointer_type ($3, 0); } | |
820 | | KW_FN '(' maybe_type_list ')' ARROW type | |
821 | { $$ = ast_function_type ($6, $3); } | |
822 | | '(' maybe_type_list ')' | |
823 | { $$ = ast_tuple_type ($2); } | |
824 | ; | |
825 | ||
826 | maybe_type_list: | |
827 | /* %empty */ | |
828 | { $$ = NULL; } | |
829 | | type_list | |
830 | { $$ = $1; } | |
831 | ; | |
832 | ||
833 | type_list: | |
834 | type | |
835 | { | |
836 | VEC (rust_op_ptr) **result | |
837 | = OBSTACK_ZALLOC (&work_obstack, VEC (rust_op_ptr) *); | |
838 | ||
839 | make_cleanup (VEC_cleanup (rust_op_ptr), result); | |
840 | VEC_safe_push (rust_op_ptr, *result, $1); | |
841 | $$ = result; | |
842 | } | |
843 | | type_list ',' type | |
844 | { | |
845 | VEC_safe_push (rust_op_ptr, *$1, $3); | |
846 | $$ = $1; | |
847 | } | |
848 | ; | |
849 | ||
850 | %% | |
851 | ||
852 | /* A struct of this type is used to describe a token. */ | |
853 | ||
854 | struct token_info | |
855 | { | |
856 | const char *name; | |
857 | int value; | |
858 | enum exp_opcode opcode; | |
859 | }; | |
860 | ||
861 | /* Identifier tokens. */ | |
862 | ||
863 | static const struct token_info identifier_tokens[] = | |
864 | { | |
865 | { "as", KW_AS, OP_NULL }, | |
866 | { "false", KW_FALSE, OP_NULL }, | |
867 | { "if", 0, OP_NULL }, | |
868 | { "mut", KW_MUT, OP_NULL }, | |
869 | { "const", KW_CONST, OP_NULL }, | |
870 | { "self", KW_SELF, OP_NULL }, | |
871 | { "super", KW_SUPER, OP_NULL }, | |
872 | { "true", KW_TRUE, OP_NULL }, | |
873 | { "extern", KW_EXTERN, OP_NULL }, | |
874 | { "fn", KW_FN, OP_NULL }, | |
875 | }; | |
876 | ||
877 | /* Operator tokens, sorted longest first. */ | |
878 | ||
879 | static const struct token_info operator_tokens[] = | |
880 | { | |
881 | { ">>=", COMPOUND_ASSIGN, BINOP_RSH }, | |
882 | { "<<=", COMPOUND_ASSIGN, BINOP_LSH }, | |
883 | ||
884 | { "<<", LSH, OP_NULL }, | |
885 | { ">>", RSH, OP_NULL }, | |
886 | { "&&", ANDAND, OP_NULL }, | |
887 | { "||", OROR, OP_NULL }, | |
888 | { "==", EQEQ, OP_NULL }, | |
889 | { "!=", NOTEQ, OP_NULL }, | |
890 | { "<=", LTEQ, OP_NULL }, | |
891 | { ">=", GTEQ, OP_NULL }, | |
892 | { "+=", COMPOUND_ASSIGN, BINOP_ADD }, | |
893 | { "-=", COMPOUND_ASSIGN, BINOP_SUB }, | |
894 | { "*=", COMPOUND_ASSIGN, BINOP_MUL }, | |
895 | { "/=", COMPOUND_ASSIGN, BINOP_DIV }, | |
896 | { "%=", COMPOUND_ASSIGN, BINOP_REM }, | |
897 | { "&=", COMPOUND_ASSIGN, BINOP_BITWISE_AND }, | |
898 | { "|=", COMPOUND_ASSIGN, BINOP_BITWISE_IOR }, | |
899 | { "^=", COMPOUND_ASSIGN, BINOP_BITWISE_XOR }, | |
900 | ||
901 | { "::", COLONCOLON, OP_NULL }, | |
902 | { "..", DOTDOT, OP_NULL }, | |
903 | { "->", ARROW, OP_NULL } | |
904 | }; | |
905 | ||
906 | /* Helper function to copy to the name obstack. */ | |
907 | ||
908 | static const char * | |
909 | rust_copy_name (const char *name, int len) | |
910 | { | |
911 | return (const char *) obstack_copy0 (&work_obstack, name, len); | |
912 | } | |
913 | ||
914 | /* Helper function to make an stoken from a C string. */ | |
915 | ||
916 | static struct stoken | |
917 | make_stoken (const char *p) | |
918 | { | |
919 | struct stoken result; | |
920 | ||
921 | result.ptr = p; | |
922 | result.length = strlen (result.ptr); | |
923 | return result; | |
924 | } | |
925 | ||
926 | /* Helper function to concatenate three strings on the name | |
927 | obstack. */ | |
928 | ||
929 | static struct stoken | |
930 | rust_concat3 (const char *s1, const char *s2, const char *s3) | |
931 | { | |
932 | return make_stoken (obconcat (&work_obstack, s1, s2, s3, (char *) NULL)); | |
933 | } | |
934 | ||
935 | /* Return an AST node referring to NAME, but relative to the crate's | |
936 | name. */ | |
937 | ||
938 | static const struct rust_op * | |
939 | crate_name (const struct rust_op *name) | |
940 | { | |
941 | char *crate = rust_crate_for_block (expression_context_block); | |
942 | struct stoken result; | |
943 | ||
944 | gdb_assert (name->opcode == OP_VAR_VALUE); | |
945 | ||
946 | if (crate == NULL) | |
947 | error (_("Could not find crate for current location")); | |
948 | result = make_stoken (obconcat (&work_obstack, "::", crate, "::", | |
949 | name->left.sval.ptr, (char *) NULL)); | |
950 | xfree (crate); | |
951 | ||
952 | return ast_path (result, name->right.params); | |
953 | } | |
954 | ||
955 | /* Create an AST node referring to a "super::" qualified name. IDENT | |
956 | is the base name and N_SUPERS is how many "super::"s were | |
957 | provided. N_SUPERS can be zero. */ | |
958 | ||
959 | static const struct rust_op * | |
960 | super_name (const struct rust_op *ident, unsigned int n_supers) | |
961 | { | |
962 | const char *scope = block_scope (expression_context_block); | |
963 | int offset; | |
964 | ||
965 | gdb_assert (ident->opcode == OP_VAR_VALUE); | |
966 | ||
967 | if (scope[0] == '\0') | |
968 | error (_("Couldn't find namespace scope for self::")); | |
969 | ||
970 | if (n_supers > 0) | |
971 | { | |
972 | int i; | |
973 | int len; | |
974 | VEC (int) *offsets = NULL; | |
78cc6c2d | 975 | unsigned int current_len; |
c44af4eb TT |
976 | struct cleanup *cleanup; |
977 | ||
978 | cleanup = make_cleanup (VEC_cleanup (int), &offsets); | |
979 | current_len = cp_find_first_component (scope); | |
c44af4eb TT |
980 | while (scope[current_len] != '\0') |
981 | { | |
982 | VEC_safe_push (int, offsets, current_len); | |
983 | gdb_assert (scope[current_len] == ':'); | |
c44af4eb TT |
984 | /* The "::". */ |
985 | current_len += 2; | |
986 | current_len += cp_find_first_component (scope | |
987 | + current_len); | |
988 | } | |
989 | ||
990 | len = VEC_length (int, offsets); | |
991 | if (n_supers >= len) | |
992 | error (_("Too many super:: uses from '%s'"), scope); | |
993 | ||
994 | offset = VEC_index (int, offsets, len - n_supers); | |
995 | ||
996 | do_cleanups (cleanup); | |
997 | } | |
998 | else | |
999 | offset = strlen (scope); | |
1000 | ||
1001 | obstack_grow (&work_obstack, "::", 2); | |
1002 | obstack_grow (&work_obstack, scope, offset); | |
1003 | obstack_grow (&work_obstack, "::", 2); | |
1004 | obstack_grow0 (&work_obstack, ident->left.sval.ptr, ident->left.sval.length); | |
1005 | ||
1006 | return ast_path (make_stoken ((const char *) obstack_finish (&work_obstack)), | |
1007 | ident->right.params); | |
1008 | } | |
1009 | ||
1010 | /* A helper that updates innermost_block as appropriate. */ | |
1011 | ||
1012 | static void | |
1013 | update_innermost_block (struct block_symbol sym) | |
1014 | { | |
1015 | if (symbol_read_needs_frame (sym.symbol) | |
1016 | && (innermost_block == NULL | |
1017 | || contained_in (sym.block, innermost_block))) | |
1018 | innermost_block = sym.block; | |
1019 | } | |
1020 | ||
1021 | /* A helper to look up a Rust type, or fail. This only works for | |
1022 | types defined by rust_language_arch_info. */ | |
1023 | ||
1024 | static struct type * | |
1025 | rust_type (const char *name) | |
1026 | { | |
1027 | struct type *type; | |
1028 | ||
1029 | /* When unit testing, we don't bother checking the types, so avoid a | |
1030 | possibly-failing lookup here. */ | |
1031 | if (unit_testing) | |
1032 | return NULL; | |
1033 | ||
1034 | type = language_lookup_primitive_type (parse_language (pstate), | |
1035 | parse_gdbarch (pstate), | |
1036 | name); | |
1037 | if (type == NULL) | |
1038 | error (_("Could not find Rust type %s"), name); | |
1039 | return type; | |
1040 | } | |
1041 | ||
1042 | /* Lex a hex number with at least MIN digits and at most MAX | |
1043 | digits. */ | |
1044 | ||
1045 | static uint32_t | |
1046 | lex_hex (int min, int max) | |
1047 | { | |
1048 | uint32_t result = 0; | |
1049 | int len = 0; | |
1050 | /* We only want to stop at MAX if we're lexing a byte escape. */ | |
1051 | int check_max = min == max; | |
1052 | ||
1053 | while ((check_max ? len <= max : 1) | |
1054 | && ((lexptr[0] >= 'a' && lexptr[0] <= 'f') | |
1055 | || (lexptr[0] >= 'A' && lexptr[0] <= 'F') | |
1056 | || (lexptr[0] >= '0' && lexptr[0] <= '9'))) | |
1057 | { | |
1058 | result *= 16; | |
1059 | if (lexptr[0] >= 'a' && lexptr[0] <= 'f') | |
1060 | result = result + 10 + lexptr[0] - 'a'; | |
1061 | else if (lexptr[0] >= 'A' && lexptr[0] <= 'F') | |
1062 | result = result + 10 + lexptr[0] - 'A'; | |
1063 | else | |
1064 | result = result + lexptr[0] - '0'; | |
1065 | ++lexptr; | |
1066 | ++len; | |
1067 | } | |
1068 | ||
1069 | if (len < min) | |
1070 | error (_("Not enough hex digits seen")); | |
1071 | if (len > max) | |
1072 | { | |
1073 | gdb_assert (min != max); | |
1074 | error (_("Overlong hex escape")); | |
1075 | } | |
1076 | ||
1077 | return result; | |
1078 | } | |
1079 | ||
1080 | /* Lex an escape. IS_BYTE is true if we're lexing a byte escape; | |
1081 | otherwise we're lexing a character escape. */ | |
1082 | ||
1083 | static uint32_t | |
1084 | lex_escape (int is_byte) | |
1085 | { | |
1086 | uint32_t result; | |
1087 | ||
1088 | gdb_assert (lexptr[0] == '\\'); | |
1089 | ++lexptr; | |
1090 | switch (lexptr[0]) | |
1091 | { | |
1092 | case 'x': | |
1093 | ++lexptr; | |
1094 | result = lex_hex (2, 2); | |
1095 | break; | |
1096 | ||
1097 | case 'u': | |
1098 | if (is_byte) | |
1099 | error (_("Unicode escape in byte literal")); | |
1100 | ++lexptr; | |
1101 | if (lexptr[0] != '{') | |
1102 | error (_("Missing '{' in Unicode escape")); | |
1103 | ++lexptr; | |
1104 | result = lex_hex (1, 6); | |
1105 | /* Could do range checks here. */ | |
1106 | if (lexptr[0] != '}') | |
1107 | error (_("Missing '}' in Unicode escape")); | |
1108 | ++lexptr; | |
1109 | break; | |
1110 | ||
1111 | case 'n': | |
1112 | result = '\n'; | |
1113 | ++lexptr; | |
1114 | break; | |
1115 | case 'r': | |
1116 | result = '\r'; | |
1117 | ++lexptr; | |
1118 | break; | |
1119 | case 't': | |
1120 | result = '\t'; | |
1121 | ++lexptr; | |
1122 | break; | |
1123 | case '\\': | |
1124 | result = '\\'; | |
1125 | ++lexptr; | |
1126 | break; | |
1127 | case '0': | |
1128 | result = '\0'; | |
1129 | ++lexptr; | |
1130 | break; | |
1131 | case '\'': | |
1132 | result = '\''; | |
1133 | ++lexptr; | |
1134 | break; | |
1135 | case '"': | |
1136 | result = '"'; | |
1137 | ++lexptr; | |
1138 | break; | |
1139 | ||
1140 | default: | |
1141 | error (_("Invalid escape \\%c in literal"), lexptr[0]); | |
1142 | } | |
1143 | ||
1144 | return result; | |
1145 | } | |
1146 | ||
1147 | /* Lex a character constant. */ | |
1148 | ||
1149 | static int | |
1150 | lex_character (void) | |
1151 | { | |
1152 | int is_byte = 0; | |
1153 | uint32_t value; | |
1154 | ||
1155 | if (lexptr[0] == 'b') | |
1156 | { | |
1157 | is_byte = 1; | |
1158 | ++lexptr; | |
1159 | } | |
1160 | gdb_assert (lexptr[0] == '\''); | |
1161 | ++lexptr; | |
1162 | /* This should handle UTF-8 here. */ | |
1163 | if (lexptr[0] == '\\') | |
1164 | value = lex_escape (is_byte); | |
1165 | else | |
1166 | { | |
1167 | value = lexptr[0] & 0xff; | |
1168 | ++lexptr; | |
1169 | } | |
1170 | ||
1171 | if (lexptr[0] != '\'') | |
1172 | error (_("Unterminated character literal")); | |
1173 | ++lexptr; | |
1174 | ||
1175 | rustyylval.typed_val_int.val = value; | |
1176 | rustyylval.typed_val_int.type = rust_type (is_byte ? "u8" : "char"); | |
1177 | ||
1178 | return INTEGER; | |
1179 | } | |
1180 | ||
1181 | /* Return the offset of the double quote if STR looks like the start | |
1182 | of a raw string, or 0 if STR does not start a raw string. */ | |
1183 | ||
1184 | static int | |
1185 | starts_raw_string (const char *str) | |
1186 | { | |
1187 | const char *save = str; | |
1188 | ||
1189 | if (str[0] != 'r') | |
1190 | return 0; | |
1191 | ++str; | |
1192 | while (str[0] == '#') | |
1193 | ++str; | |
1194 | if (str[0] == '"') | |
1195 | return str - save; | |
1196 | return 0; | |
1197 | } | |
1198 | ||
1199 | /* Return true if STR looks like the end of a raw string that had N | |
1200 | hashes at the start. */ | |
1201 | ||
1202 | static int | |
1203 | ends_raw_string (const char *str, int n) | |
1204 | { | |
1205 | int i; | |
1206 | ||
1207 | gdb_assert (str[0] == '"'); | |
1208 | for (i = 0; i < n; ++i) | |
1209 | if (str[i + 1] != '#') | |
1210 | return 0; | |
1211 | return 1; | |
1212 | } | |
1213 | ||
1214 | /* Lex a string constant. */ | |
1215 | ||
1216 | static int | |
1217 | lex_string (void) | |
1218 | { | |
1219 | int is_byte = lexptr[0] == 'b'; | |
1220 | int raw_length; | |
1221 | int len_in_chars = 0; | |
1222 | ||
1223 | if (is_byte) | |
1224 | ++lexptr; | |
1225 | raw_length = starts_raw_string (lexptr); | |
1226 | lexptr += raw_length; | |
1227 | gdb_assert (lexptr[0] == '"'); | |
1228 | ++lexptr; | |
1229 | ||
1230 | while (1) | |
1231 | { | |
1232 | uint32_t value; | |
1233 | ||
1234 | if (raw_length > 0) | |
1235 | { | |
1236 | if (lexptr[0] == '"' && ends_raw_string (lexptr, raw_length - 1)) | |
1237 | { | |
1238 | /* Exit with lexptr pointing after the final "#". */ | |
1239 | lexptr += raw_length; | |
1240 | break; | |
1241 | } | |
1242 | else if (lexptr[0] == '\0') | |
1243 | error (_("Unexpected EOF in string")); | |
1244 | ||
1245 | value = lexptr[0] & 0xff; | |
1246 | if (is_byte && value > 127) | |
1247 | error (_("Non-ASCII value in raw byte string")); | |
1248 | obstack_1grow (&work_obstack, value); | |
1249 | ||
1250 | ++lexptr; | |
1251 | } | |
1252 | else if (lexptr[0] == '"') | |
1253 | { | |
1254 | /* Make sure to skip the quote. */ | |
1255 | ++lexptr; | |
1256 | break; | |
1257 | } | |
1258 | else if (lexptr[0] == '\\') | |
1259 | { | |
1260 | value = lex_escape (is_byte); | |
1261 | ||
1262 | if (is_byte) | |
1263 | obstack_1grow (&work_obstack, value); | |
1264 | else | |
1265 | convert_between_encodings ("UTF-32", "UTF-8", (gdb_byte *) &value, | |
1266 | sizeof (value), sizeof (value), | |
1267 | &work_obstack, translit_none); | |
1268 | } | |
1269 | else if (lexptr[0] == '\0') | |
1270 | error (_("Unexpected EOF in string")); | |
1271 | else | |
1272 | { | |
1273 | value = lexptr[0] & 0xff; | |
1274 | if (is_byte && value > 127) | |
1275 | error (_("Non-ASCII value in byte string")); | |
1276 | obstack_1grow (&work_obstack, value); | |
1277 | ++lexptr; | |
1278 | } | |
1279 | } | |
1280 | ||
1281 | rustyylval.sval.length = obstack_object_size (&work_obstack); | |
1282 | rustyylval.sval.ptr = (const char *) obstack_finish (&work_obstack); | |
1283 | return is_byte ? BYTESTRING : STRING; | |
1284 | } | |
1285 | ||
1286 | /* Return true if STRING starts with whitespace followed by a digit. */ | |
1287 | ||
1288 | static int | |
1289 | space_then_number (const char *string) | |
1290 | { | |
1291 | const char *p = string; | |
1292 | ||
1293 | while (p[0] == ' ' || p[0] == '\t') | |
1294 | ++p; | |
1295 | if (p == string) | |
1296 | return 0; | |
1297 | ||
1298 | return *p >= '0' && *p <= '9'; | |
1299 | } | |
1300 | ||
1301 | /* Return true if C can start an identifier. */ | |
1302 | ||
1303 | static int | |
1304 | rust_identifier_start_p (char c) | |
1305 | { | |
1306 | return ((c >= 'a' && c <= 'z') | |
1307 | || (c >= 'A' && c <= 'Z') | |
1308 | || c == '_' | |
1309 | || c == '$'); | |
1310 | } | |
1311 | ||
1312 | /* Lex an identifier. */ | |
1313 | ||
1314 | static int | |
1315 | lex_identifier (void) | |
1316 | { | |
1317 | const char *start = lexptr; | |
1318 | unsigned int length; | |
1319 | const struct token_info *token; | |
1320 | int i; | |
1321 | int is_gdb_var = lexptr[0] == '$'; | |
1322 | ||
1323 | gdb_assert (rust_identifier_start_p (lexptr[0])); | |
1324 | ||
1325 | ++lexptr; | |
1326 | ||
1327 | /* For the time being this doesn't handle Unicode rules. Non-ASCII | |
1328 | identifiers are gated anyway. */ | |
1329 | while ((lexptr[0] >= 'a' && lexptr[0] <= 'z') | |
1330 | || (lexptr[0] >= 'A' && lexptr[0] <= 'Z') | |
1331 | || lexptr[0] == '_' | |
1332 | || (is_gdb_var && lexptr[0] == '$') | |
1333 | || (lexptr[0] >= '0' && lexptr[0] <= '9')) | |
1334 | ++lexptr; | |
1335 | ||
1336 | ||
1337 | length = lexptr - start; | |
1338 | token = NULL; | |
1339 | for (i = 0; i < ARRAY_SIZE (identifier_tokens); ++i) | |
1340 | { | |
1341 | if (length == strlen (identifier_tokens[i].name) | |
1342 | && strncmp (identifier_tokens[i].name, start, length) == 0) | |
1343 | { | |
1344 | token = &identifier_tokens[i]; | |
1345 | break; | |
1346 | } | |
1347 | } | |
1348 | ||
1349 | if (token != NULL) | |
1350 | { | |
1351 | if (token->value == 0) | |
1352 | { | |
1353 | /* Leave the terminating token alone. */ | |
1354 | lexptr = start; | |
1355 | return 0; | |
1356 | } | |
1357 | } | |
1358 | else if (token == NULL | |
1359 | && (strncmp (start, "thread", length) == 0 | |
1360 | || strncmp (start, "task", length) == 0) | |
1361 | && space_then_number (lexptr)) | |
1362 | { | |
1363 | /* "task" or "thread" followed by a number terminates the | |
1364 | parse, per gdb rules. */ | |
1365 | lexptr = start; | |
1366 | return 0; | |
1367 | } | |
1368 | ||
1369 | if (token == NULL || (parse_completion && lexptr[0] == '\0')) | |
1370 | rustyylval.sval = make_stoken (rust_copy_name (start, length)); | |
1371 | ||
1372 | if (parse_completion && lexptr[0] == '\0') | |
1373 | { | |
1374 | /* Prevent rustyylex from returning two COMPLETE tokens. */ | |
1375 | prev_lexptr = lexptr; | |
1376 | return COMPLETE; | |
1377 | } | |
1378 | ||
1379 | if (token != NULL) | |
1380 | return token->value; | |
1381 | if (is_gdb_var) | |
1382 | return GDBVAR; | |
1383 | return IDENT; | |
1384 | } | |
1385 | ||
1386 | /* Lex an operator. */ | |
1387 | ||
1388 | static int | |
1389 | lex_operator (void) | |
1390 | { | |
1391 | const struct token_info *token = NULL; | |
1392 | int i; | |
1393 | ||
1394 | for (i = 0; i < ARRAY_SIZE (operator_tokens); ++i) | |
1395 | { | |
1396 | if (strncmp (operator_tokens[i].name, lexptr, | |
1397 | strlen (operator_tokens[i].name)) == 0) | |
1398 | { | |
1399 | lexptr += strlen (operator_tokens[i].name); | |
1400 | token = &operator_tokens[i]; | |
1401 | break; | |
1402 | } | |
1403 | } | |
1404 | ||
1405 | if (token != NULL) | |
1406 | { | |
1407 | rustyylval.opcode = token->opcode; | |
1408 | return token->value; | |
1409 | } | |
1410 | ||
1411 | return *lexptr++; | |
1412 | } | |
1413 | ||
1414 | /* Lex a number. */ | |
1415 | ||
1416 | static int | |
1417 | lex_number (void) | |
1418 | { | |
1419 | regmatch_t subexps[NUM_SUBEXPRESSIONS]; | |
1420 | int match; | |
1421 | int is_integer = 0; | |
1422 | int could_be_decimal = 1; | |
347dc102 | 1423 | int implicit_i32 = 0; |
c44af4eb TT |
1424 | char *type_name = NULL; |
1425 | struct type *type; | |
1426 | int end_index; | |
1427 | int type_index = -1; | |
1428 | int i, out; | |
1429 | char *number; | |
1430 | struct cleanup *cleanup = make_cleanup (null_cleanup, NULL); | |
1431 | ||
1432 | match = regexec (&number_regex, lexptr, ARRAY_SIZE (subexps), subexps, 0); | |
1433 | /* Failure means the regexp is broken. */ | |
1434 | gdb_assert (match == 0); | |
1435 | ||
1436 | if (subexps[INT_TEXT].rm_so != -1) | |
1437 | { | |
1438 | /* Integer part matched. */ | |
1439 | is_integer = 1; | |
1440 | end_index = subexps[INT_TEXT].rm_eo; | |
1441 | if (subexps[INT_TYPE].rm_so == -1) | |
347dc102 TT |
1442 | { |
1443 | type_name = "i32"; | |
1444 | implicit_i32 = 1; | |
1445 | } | |
c44af4eb TT |
1446 | else |
1447 | { | |
1448 | type_index = INT_TYPE; | |
1449 | could_be_decimal = 0; | |
1450 | } | |
1451 | } | |
1452 | else if (subexps[FLOAT_TYPE1].rm_so != -1) | |
1453 | { | |
1454 | /* Found floating point type suffix. */ | |
1455 | end_index = subexps[FLOAT_TYPE1].rm_so; | |
1456 | type_index = FLOAT_TYPE1; | |
1457 | } | |
1458 | else if (subexps[FLOAT_TYPE2].rm_so != -1) | |
1459 | { | |
1460 | /* Found floating point type suffix. */ | |
1461 | end_index = subexps[FLOAT_TYPE2].rm_so; | |
1462 | type_index = FLOAT_TYPE2; | |
1463 | } | |
1464 | else | |
1465 | { | |
1466 | /* Any other floating point match. */ | |
1467 | end_index = subexps[0].rm_eo; | |
1468 | type_name = "f64"; | |
1469 | } | |
1470 | ||
1471 | /* We need a special case if the final character is ".". In this | |
1472 | case we might need to parse an integer. For example, "23.f()" is | |
1473 | a request for a trait method call, not a syntax error involving | |
1474 | the floating point number "23.". */ | |
1475 | gdb_assert (subexps[0].rm_eo > 0); | |
1476 | if (lexptr[subexps[0].rm_eo - 1] == '.') | |
1477 | { | |
1478 | const char *next = skip_spaces_const (&lexptr[subexps[0].rm_eo]); | |
1479 | ||
1480 | if (rust_identifier_start_p (*next) || *next == '.') | |
1481 | { | |
1482 | --subexps[0].rm_eo; | |
1483 | is_integer = 1; | |
1484 | end_index = subexps[0].rm_eo; | |
1485 | type_name = "i32"; | |
1486 | could_be_decimal = 1; | |
347dc102 | 1487 | implicit_i32 = 1; |
c44af4eb TT |
1488 | } |
1489 | } | |
1490 | ||
1491 | /* Compute the type name if we haven't already. */ | |
1492 | if (type_name == NULL) | |
1493 | { | |
1494 | gdb_assert (type_index != -1); | |
1495 | type_name = xstrndup (lexptr + subexps[type_index].rm_so, | |
1496 | (subexps[type_index].rm_eo | |
1497 | - subexps[type_index].rm_so)); | |
1498 | make_cleanup (xfree, type_name); | |
1499 | } | |
1500 | ||
1501 | /* Look up the type. */ | |
1502 | type = rust_type (type_name); | |
1503 | ||
1504 | /* Copy the text of the number and remove the "_"s. */ | |
1505 | number = xstrndup (lexptr, end_index); | |
1506 | make_cleanup (xfree, number); | |
1507 | for (i = out = 0; number[i]; ++i) | |
1508 | { | |
1509 | if (number[i] == '_') | |
1510 | could_be_decimal = 0; | |
1511 | else | |
1512 | number[out++] = number[i]; | |
1513 | } | |
1514 | number[out] = '\0'; | |
1515 | ||
1516 | /* Advance past the match. */ | |
1517 | lexptr += subexps[0].rm_eo; | |
1518 | ||
1519 | /* Parse the number. */ | |
1520 | if (is_integer) | |
1521 | { | |
347dc102 | 1522 | uint64_t value; |
c44af4eb TT |
1523 | int radix = 10; |
1524 | if (number[0] == '0') | |
1525 | { | |
1526 | if (number[1] == 'x') | |
1527 | radix = 16; | |
1528 | else if (number[1] == 'o') | |
1529 | radix = 8; | |
1530 | else if (number[1] == 'b') | |
1531 | radix = 2; | |
1532 | if (radix != 10) | |
1533 | { | |
1534 | number += 2; | |
1535 | could_be_decimal = 0; | |
1536 | } | |
1537 | } | |
347dc102 TT |
1538 | |
1539 | value = strtoul (number, NULL, radix); | |
1540 | if (implicit_i32 && value >= ((uint64_t) 1) << 31) | |
1541 | type = rust_type ("i64"); | |
1542 | ||
1543 | rustyylval.typed_val_int.val = value; | |
c44af4eb TT |
1544 | rustyylval.typed_val_int.type = type; |
1545 | } | |
1546 | else | |
1547 | { | |
1548 | rustyylval.typed_val_float.dval = strtod (number, NULL); | |
1549 | rustyylval.typed_val_float.type = type; | |
1550 | } | |
1551 | ||
1552 | do_cleanups (cleanup); | |
1553 | return is_integer ? (could_be_decimal ? DECIMAL_INTEGER : INTEGER) : FLOAT; | |
1554 | } | |
1555 | ||
1556 | /* The lexer. */ | |
1557 | ||
1558 | static int | |
1559 | rustyylex (void) | |
1560 | { | |
1561 | /* Skip all leading whitespace. */ | |
1562 | while (lexptr[0] == ' ' || lexptr[0] == '\t' || lexptr[0] == '\r' | |
1563 | || lexptr[0] == '\n') | |
1564 | ++lexptr; | |
1565 | ||
1566 | /* If we hit EOF and we're completing, then return COMPLETE -- maybe | |
1567 | we're completing an empty string at the end of a field_expr. | |
1568 | But, we don't want to return two COMPLETE tokens in a row. */ | |
1569 | if (lexptr[0] == '\0' && lexptr == prev_lexptr) | |
1570 | return 0; | |
1571 | prev_lexptr = lexptr; | |
1572 | if (lexptr[0] == '\0') | |
1573 | { | |
1574 | if (parse_completion) | |
1575 | { | |
1576 | rustyylval.sval = make_stoken (""); | |
1577 | return COMPLETE; | |
1578 | } | |
1579 | return 0; | |
1580 | } | |
1581 | ||
1582 | if (lexptr[0] >= '0' && lexptr[0] <= '9') | |
1583 | return lex_number (); | |
1584 | else if (lexptr[0] == 'b' && lexptr[1] == '\'') | |
1585 | return lex_character (); | |
1586 | else if (lexptr[0] == 'b' && lexptr[1] == '"') | |
1587 | return lex_string (); | |
1588 | else if (lexptr[0] == 'b' && starts_raw_string (lexptr + 1)) | |
1589 | return lex_string (); | |
1590 | else if (starts_raw_string (lexptr)) | |
1591 | return lex_string (); | |
1592 | else if (rust_identifier_start_p (lexptr[0])) | |
1593 | return lex_identifier (); | |
1594 | else if (lexptr[0] == '"') | |
1595 | return lex_string (); | |
1596 | else if (lexptr[0] == '\'') | |
1597 | return lex_character (); | |
1598 | else if (lexptr[0] == '}' || lexptr[0] == ']') | |
1599 | { | |
1600 | /* Falls through to lex_operator. */ | |
1601 | --paren_depth; | |
1602 | } | |
1603 | else if (lexptr[0] == '(' || lexptr[0] == '{') | |
1604 | { | |
1605 | /* Falls through to lex_operator. */ | |
1606 | ++paren_depth; | |
1607 | } | |
1608 | else if (lexptr[0] == ',' && comma_terminates && paren_depth == 0) | |
1609 | return 0; | |
1610 | ||
1611 | return lex_operator (); | |
1612 | } | |
1613 | ||
1614 | /* Push back a single character to be re-lexed. */ | |
1615 | ||
1616 | static void | |
1617 | rust_push_back (char c) | |
1618 | { | |
1619 | /* Can't be called before any lexing. */ | |
1620 | gdb_assert (prev_lexptr != NULL); | |
1621 | ||
1622 | --lexptr; | |
1623 | gdb_assert (*lexptr == c); | |
1624 | } | |
1625 | ||
1626 | \f | |
1627 | ||
1628 | /* Make an arbitrary operation and fill in the fields. */ | |
1629 | ||
1630 | static const struct rust_op * | |
1631 | ast_operation (enum exp_opcode opcode, const struct rust_op *left, | |
1632 | const struct rust_op *right) | |
1633 | { | |
1634 | struct rust_op *result = OBSTACK_ZALLOC (&work_obstack, struct rust_op); | |
1635 | ||
1636 | result->opcode = opcode; | |
1637 | result->left.op = left; | |
1638 | result->right.op = right; | |
1639 | ||
1640 | return result; | |
1641 | } | |
1642 | ||
1643 | /* Make a compound assignment operation. */ | |
1644 | ||
1645 | static const struct rust_op * | |
1646 | ast_compound_assignment (enum exp_opcode opcode, const struct rust_op *left, | |
1647 | const struct rust_op *right) | |
1648 | { | |
1649 | struct rust_op *result = OBSTACK_ZALLOC (&work_obstack, struct rust_op); | |
1650 | ||
1651 | result->opcode = opcode; | |
1652 | result->compound_assignment = 1; | |
1653 | result->left.op = left; | |
1654 | result->right.op = right; | |
1655 | ||
1656 | return result; | |
1657 | } | |
1658 | ||
1659 | /* Make a typed integer literal operation. */ | |
1660 | ||
1661 | static const struct rust_op * | |
1662 | ast_literal (struct typed_val_int val) | |
1663 | { | |
1664 | struct rust_op *result = OBSTACK_ZALLOC (&work_obstack, struct rust_op); | |
1665 | ||
1666 | result->opcode = OP_LONG; | |
1667 | result->left.typed_val_int = val; | |
1668 | ||
1669 | return result; | |
1670 | } | |
1671 | ||
1672 | /* Make a typed floating point literal operation. */ | |
1673 | ||
1674 | static const struct rust_op * | |
1675 | ast_dliteral (struct typed_val_float val) | |
1676 | { | |
1677 | struct rust_op *result = OBSTACK_ZALLOC (&work_obstack, struct rust_op); | |
1678 | ||
1679 | result->opcode = OP_DOUBLE; | |
1680 | result->left.typed_val_float = val; | |
1681 | ||
1682 | return result; | |
1683 | } | |
1684 | ||
1685 | /* Make a unary operation. */ | |
1686 | ||
1687 | static const struct rust_op * | |
1688 | ast_unary (enum exp_opcode opcode, const struct rust_op *expr) | |
1689 | { | |
1690 | return ast_operation (opcode, expr, NULL); | |
1691 | } | |
1692 | ||
1693 | /* Make a cast operation. */ | |
1694 | ||
1695 | static const struct rust_op * | |
1696 | ast_cast (const struct rust_op *expr, const struct rust_op *type) | |
1697 | { | |
1698 | struct rust_op *result = OBSTACK_ZALLOC (&work_obstack, struct rust_op); | |
1699 | ||
1700 | result->opcode = UNOP_CAST; | |
1701 | result->left.op = expr; | |
1702 | result->right.op = type; | |
1703 | ||
1704 | return result; | |
1705 | } | |
1706 | ||
1707 | /* Make a call-like operation. This is nominally a function call, but | |
1708 | when lowering we may discover that it actually represents the | |
1709 | creation of a tuple struct. */ | |
1710 | ||
1711 | static const struct rust_op * | |
1712 | ast_call_ish (enum exp_opcode opcode, const struct rust_op *expr, | |
1713 | VEC (rust_op_ptr) **params) | |
1714 | { | |
1715 | struct rust_op *result = OBSTACK_ZALLOC (&work_obstack, struct rust_op); | |
1716 | ||
1717 | result->opcode = opcode; | |
1718 | result->left.op = expr; | |
1719 | result->right.params = params; | |
1720 | ||
1721 | return result; | |
1722 | } | |
1723 | ||
1724 | /* Make a structure creation operation. */ | |
1725 | ||
1726 | static const struct rust_op * | |
1727 | ast_struct (const struct rust_op *name, VEC (set_field) **fields) | |
1728 | { | |
1729 | struct rust_op *result = OBSTACK_ZALLOC (&work_obstack, struct rust_op); | |
1730 | ||
1731 | result->opcode = OP_AGGREGATE; | |
1732 | result->left.op = name; | |
1733 | result->right.field_inits = fields; | |
1734 | ||
1735 | return result; | |
1736 | } | |
1737 | ||
1738 | /* Make an identifier path. */ | |
1739 | ||
1740 | static const struct rust_op * | |
1741 | ast_path (struct stoken path, VEC (rust_op_ptr) **params) | |
1742 | { | |
1743 | struct rust_op *result = OBSTACK_ZALLOC (&work_obstack, struct rust_op); | |
1744 | ||
1745 | result->opcode = OP_VAR_VALUE; | |
1746 | result->left.sval = path; | |
1747 | result->right.params = params; | |
1748 | ||
1749 | return result; | |
1750 | } | |
1751 | ||
1752 | /* Make a string constant operation. */ | |
1753 | ||
1754 | static const struct rust_op * | |
1755 | ast_string (struct stoken str) | |
1756 | { | |
1757 | struct rust_op *result = OBSTACK_ZALLOC (&work_obstack, struct rust_op); | |
1758 | ||
1759 | result->opcode = OP_STRING; | |
1760 | result->left.sval = str; | |
1761 | ||
1762 | return result; | |
1763 | } | |
1764 | ||
1765 | /* Make a field expression. */ | |
1766 | ||
1767 | static const struct rust_op * | |
1768 | ast_structop (const struct rust_op *left, const char *name, int completing) | |
1769 | { | |
1770 | struct rust_op *result = OBSTACK_ZALLOC (&work_obstack, struct rust_op); | |
1771 | ||
1772 | result->opcode = STRUCTOP_STRUCT; | |
1773 | result->completing = completing; | |
1774 | result->left.op = left; | |
1775 | result->right.sval = make_stoken (name); | |
1776 | ||
1777 | return result; | |
1778 | } | |
1779 | ||
1780 | /* Make an anonymous struct operation, like 'x.0'. */ | |
1781 | ||
1782 | static const struct rust_op * | |
1783 | ast_structop_anonymous (const struct rust_op *left, | |
1784 | struct typed_val_int number) | |
1785 | { | |
1786 | struct rust_op *result = OBSTACK_ZALLOC (&work_obstack, struct rust_op); | |
1787 | ||
1788 | result->opcode = STRUCTOP_ANONYMOUS; | |
1789 | result->left.op = left; | |
1790 | result->right.typed_val_int = number; | |
1791 | ||
1792 | return result; | |
1793 | } | |
1794 | ||
1795 | /* Make a range operation. */ | |
1796 | ||
1797 | static const struct rust_op * | |
1798 | ast_range (const struct rust_op *lhs, const struct rust_op *rhs) | |
1799 | { | |
1800 | struct rust_op *result = OBSTACK_ZALLOC (&work_obstack, struct rust_op); | |
1801 | ||
01739a3b | 1802 | result->opcode = OP_RANGE; |
c44af4eb TT |
1803 | result->left.op = lhs; |
1804 | result->right.op = rhs; | |
1805 | ||
1806 | return result; | |
1807 | } | |
1808 | ||
1809 | /* A helper function to make a type-related AST node. */ | |
1810 | ||
1811 | static struct rust_op * | |
1812 | ast_basic_type (enum type_code typecode) | |
1813 | { | |
1814 | struct rust_op *result = OBSTACK_ZALLOC (&work_obstack, struct rust_op); | |
1815 | ||
1816 | result->opcode = OP_TYPE; | |
1817 | result->typecode = typecode; | |
1818 | return result; | |
1819 | } | |
1820 | ||
1821 | /* Create an AST node describing an array type. */ | |
1822 | ||
1823 | static const struct rust_op * | |
1824 | ast_array_type (const struct rust_op *lhs, struct typed_val_int val) | |
1825 | { | |
1826 | struct rust_op *result = ast_basic_type (TYPE_CODE_ARRAY); | |
1827 | ||
1828 | result->left.op = lhs; | |
1829 | result->right.typed_val_int = val; | |
1830 | return result; | |
1831 | } | |
1832 | ||
1833 | /* Create an AST node describing a reference type. */ | |
1834 | ||
1835 | static const struct rust_op * | |
1836 | ast_slice_type (const struct rust_op *type) | |
1837 | { | |
1838 | /* Use TYPE_CODE_COMPLEX just because it is handy. */ | |
1839 | struct rust_op *result = ast_basic_type (TYPE_CODE_COMPLEX); | |
1840 | ||
1841 | result->left.op = type; | |
1842 | return result; | |
1843 | } | |
1844 | ||
1845 | /* Create an AST node describing a reference type. */ | |
1846 | ||
1847 | static const struct rust_op * | |
1848 | ast_reference_type (const struct rust_op *type) | |
1849 | { | |
1850 | struct rust_op *result = ast_basic_type (TYPE_CODE_REF); | |
1851 | ||
1852 | result->left.op = type; | |
1853 | return result; | |
1854 | } | |
1855 | ||
1856 | /* Create an AST node describing a pointer type. */ | |
1857 | ||
1858 | static const struct rust_op * | |
1859 | ast_pointer_type (const struct rust_op *type, int is_mut) | |
1860 | { | |
1861 | struct rust_op *result = ast_basic_type (TYPE_CODE_PTR); | |
1862 | ||
1863 | result->left.op = type; | |
1864 | /* For the time being we ignore is_mut. */ | |
1865 | return result; | |
1866 | } | |
1867 | ||
1868 | /* Create an AST node describing a function type. */ | |
1869 | ||
1870 | static const struct rust_op * | |
1871 | ast_function_type (const struct rust_op *rtype, VEC (rust_op_ptr) **params) | |
1872 | { | |
1873 | struct rust_op *result = ast_basic_type (TYPE_CODE_FUNC); | |
1874 | ||
1875 | result->left.op = rtype; | |
1876 | result->right.params = params; | |
1877 | return result; | |
1878 | } | |
1879 | ||
1880 | /* Create an AST node describing a tuple type. */ | |
1881 | ||
1882 | static const struct rust_op * | |
1883 | ast_tuple_type (VEC (rust_op_ptr) **params) | |
1884 | { | |
1885 | struct rust_op *result = ast_basic_type (TYPE_CODE_STRUCT); | |
1886 | ||
1887 | result->left.params = params; | |
1888 | return result; | |
1889 | } | |
1890 | ||
1891 | /* A helper to appropriately munge NAME and BLOCK depending on the | |
1892 | presence of a leading "::". */ | |
1893 | ||
1894 | static void | |
1895 | munge_name_and_block (const char **name, const struct block **block) | |
1896 | { | |
1897 | /* If it is a global reference, skip the current block in favor of | |
1898 | the static block. */ | |
1899 | if (strncmp (*name, "::", 2) == 0) | |
1900 | { | |
1901 | *name += 2; | |
1902 | *block = block_static_block (*block); | |
1903 | } | |
1904 | } | |
1905 | ||
1906 | /* Like lookup_symbol, but handles Rust namespace conventions, and | |
1907 | doesn't require field_of_this_result. */ | |
1908 | ||
1909 | static struct block_symbol | |
1910 | rust_lookup_symbol (const char *name, const struct block *block, | |
1911 | const domain_enum domain) | |
1912 | { | |
1913 | struct block_symbol result; | |
1914 | ||
1915 | munge_name_and_block (&name, &block); | |
1916 | ||
1917 | result = lookup_symbol (name, block, domain, NULL); | |
1918 | if (result.symbol != NULL) | |
1919 | update_innermost_block (result); | |
1920 | return result; | |
1921 | } | |
1922 | ||
1923 | /* Look up a type, following Rust namespace conventions. */ | |
1924 | ||
1925 | static struct type * | |
1926 | rust_lookup_type (const char *name, const struct block *block) | |
1927 | { | |
1928 | struct block_symbol result; | |
1929 | struct type *type; | |
1930 | ||
1931 | munge_name_and_block (&name, &block); | |
1932 | ||
1933 | result = lookup_symbol (name, block, STRUCT_DOMAIN, NULL); | |
1934 | if (result.symbol != NULL) | |
1935 | { | |
1936 | update_innermost_block (result); | |
1937 | return SYMBOL_TYPE (result.symbol); | |
1938 | } | |
1939 | ||
1940 | type = lookup_typename (parse_language (pstate), parse_gdbarch (pstate), | |
1941 | name, NULL, 1); | |
1942 | if (type != NULL) | |
1943 | return type; | |
1944 | ||
1945 | /* Last chance, try a built-in type. */ | |
1946 | return language_lookup_primitive_type (parse_language (pstate), | |
1947 | parse_gdbarch (pstate), | |
1948 | name); | |
1949 | } | |
1950 | ||
1951 | static struct type *convert_ast_to_type (struct parser_state *state, | |
1952 | const struct rust_op *operation); | |
1953 | static const char *convert_name (struct parser_state *state, | |
1954 | const struct rust_op *operation); | |
1955 | ||
1956 | /* Convert a vector of rust_ops representing types to a vector of | |
1957 | types. */ | |
1958 | ||
1959 | static VEC (type_ptr) * | |
1960 | convert_params_to_types (struct parser_state *state, VEC (rust_op_ptr) *params) | |
1961 | { | |
1962 | int i; | |
1963 | const struct rust_op *op; | |
1964 | VEC (type_ptr) *result = NULL; | |
1965 | struct cleanup *cleanup = make_cleanup (VEC_cleanup (type_ptr), &result); | |
1966 | ||
1967 | for (i = 0; VEC_iterate (rust_op_ptr, params, i, op); ++i) | |
1968 | VEC_safe_push (type_ptr, result, convert_ast_to_type (state, op)); | |
1969 | ||
1970 | discard_cleanups (cleanup); | |
1971 | return result; | |
1972 | } | |
1973 | ||
1974 | /* Convert a rust_op representing a type to a struct type *. */ | |
1975 | ||
1976 | static struct type * | |
1977 | convert_ast_to_type (struct parser_state *state, | |
1978 | const struct rust_op *operation) | |
1979 | { | |
1980 | struct type *type, *result = NULL; | |
1981 | ||
1982 | if (operation->opcode == OP_VAR_VALUE) | |
1983 | { | |
1984 | const char *varname = convert_name (state, operation); | |
1985 | ||
1986 | result = rust_lookup_type (varname, expression_context_block); | |
1987 | if (result == NULL) | |
1988 | error (_("No typed name '%s' in current context"), varname); | |
1989 | return result; | |
1990 | } | |
1991 | ||
1992 | gdb_assert (operation->opcode == OP_TYPE); | |
1993 | ||
1994 | switch (operation->typecode) | |
1995 | { | |
1996 | case TYPE_CODE_ARRAY: | |
1997 | type = convert_ast_to_type (state, operation->left.op); | |
1998 | if (operation->right.typed_val_int.val < 0) | |
1999 | error (_("Negative array length")); | |
2000 | result = lookup_array_range_type (type, 0, | |
2001 | operation->right.typed_val_int.val - 1); | |
2002 | break; | |
2003 | ||
2004 | case TYPE_CODE_COMPLEX: | |
2005 | { | |
2006 | struct type *usize = rust_type ("usize"); | |
2007 | ||
2008 | type = convert_ast_to_type (state, operation->left.op); | |
2009 | result = rust_slice_type ("&[*gdb*]", type, usize); | |
2010 | } | |
2011 | break; | |
2012 | ||
2013 | case TYPE_CODE_REF: | |
2014 | case TYPE_CODE_PTR: | |
2015 | /* For now we treat &x and *x identically. */ | |
2016 | type = convert_ast_to_type (state, operation->left.op); | |
2017 | result = lookup_pointer_type (type); | |
2018 | break; | |
2019 | ||
2020 | case TYPE_CODE_FUNC: | |
2021 | { | |
2022 | VEC (type_ptr) *args | |
2023 | = convert_params_to_types (state, *operation->right.params); | |
2024 | struct cleanup *cleanup | |
2025 | = make_cleanup (VEC_cleanup (type_ptr), &args); | |
2026 | struct type **argtypes = NULL; | |
2027 | ||
2028 | type = convert_ast_to_type (state, operation->left.op); | |
2029 | if (!VEC_empty (type_ptr, args)) | |
2030 | argtypes = VEC_address (type_ptr, args); | |
2031 | ||
2032 | result | |
2033 | = lookup_function_type_with_arguments (type, | |
2034 | VEC_length (type_ptr, args), | |
2035 | argtypes); | |
2036 | result = lookup_pointer_type (result); | |
2037 | ||
2038 | do_cleanups (cleanup); | |
2039 | } | |
2040 | break; | |
2041 | ||
2042 | case TYPE_CODE_STRUCT: | |
2043 | { | |
2044 | VEC (type_ptr) *args | |
2045 | = convert_params_to_types (state, *operation->left.params); | |
2046 | struct cleanup *cleanup | |
2047 | = make_cleanup (VEC_cleanup (type_ptr), &args); | |
2048 | int i; | |
2049 | struct type *type; | |
2050 | const char *name; | |
2051 | ||
2052 | obstack_1grow (&work_obstack, '('); | |
2053 | for (i = 0; VEC_iterate (type_ptr, args, i, type); ++i) | |
2054 | { | |
2055 | char *type_name = type_to_string (type); | |
2056 | ||
2057 | if (i > 0) | |
2058 | obstack_1grow (&work_obstack, ','); | |
2059 | obstack_grow_str (&work_obstack, type_name); | |
2060 | ||
2061 | xfree (type_name); | |
2062 | } | |
2063 | ||
2064 | obstack_grow_str0 (&work_obstack, ")"); | |
2065 | name = (const char *) obstack_finish (&work_obstack); | |
2066 | ||
2067 | /* We don't allow creating new tuple types (yet), but we do | |
2068 | allow looking up existing tuple types. */ | |
2069 | result = rust_lookup_type (name, expression_context_block); | |
2070 | if (result == NULL) | |
2071 | error (_("could not find tuple type '%s'"), name); | |
2072 | ||
2073 | do_cleanups (cleanup); | |
2074 | } | |
2075 | break; | |
2076 | ||
2077 | default: | |
2078 | gdb_assert_not_reached ("unhandled opcode in convert_ast_to_type"); | |
2079 | } | |
2080 | ||
2081 | gdb_assert (result != NULL); | |
2082 | return result; | |
2083 | } | |
2084 | ||
2085 | /* A helper function to turn a rust_op representing a name into a full | |
2086 | name. This applies generic arguments as needed. The returned name | |
2087 | is allocated on the work obstack. */ | |
2088 | ||
2089 | static const char * | |
2090 | convert_name (struct parser_state *state, const struct rust_op *operation) | |
2091 | { | |
2092 | VEC (type_ptr) *types; | |
2093 | struct cleanup *cleanup; | |
2094 | int i; | |
2095 | struct type *type; | |
2096 | ||
2097 | gdb_assert (operation->opcode == OP_VAR_VALUE); | |
2098 | ||
2099 | if (operation->right.params == NULL) | |
2100 | return operation->left.sval.ptr; | |
2101 | ||
2102 | types = convert_params_to_types (state, *operation->right.params); | |
2103 | cleanup = make_cleanup (VEC_cleanup (type_ptr), &types); | |
2104 | ||
2105 | obstack_grow_str (&work_obstack, operation->left.sval.ptr); | |
2106 | obstack_1grow (&work_obstack, '<'); | |
2107 | for (i = 0; VEC_iterate (type_ptr, types, i, type); ++i) | |
2108 | { | |
2109 | char *type_name = type_to_string (type); | |
2110 | ||
2111 | if (i > 0) | |
2112 | obstack_1grow (&work_obstack, ','); | |
2113 | ||
2114 | obstack_grow_str (&work_obstack, type_name); | |
2115 | xfree (type_name); | |
2116 | } | |
2117 | obstack_grow_str0 (&work_obstack, ">"); | |
2118 | ||
2119 | do_cleanups (cleanup); | |
2120 | ||
2121 | return (const char *) obstack_finish (&work_obstack); | |
2122 | } | |
2123 | ||
2124 | static void convert_ast_to_expression (struct parser_state *state, | |
2125 | const struct rust_op *operation, | |
2126 | const struct rust_op *top); | |
2127 | ||
2128 | /* A helper function that converts a vec of rust_ops to a gdb | |
2129 | expression. */ | |
2130 | ||
2131 | static void | |
2132 | convert_params_to_expression (struct parser_state *state, | |
2133 | VEC (rust_op_ptr) *params, | |
2134 | const struct rust_op *top) | |
2135 | { | |
2136 | int i; | |
2137 | rust_op_ptr elem; | |
2138 | ||
2139 | for (i = 0; VEC_iterate (rust_op_ptr, params, i, elem); ++i) | |
2140 | convert_ast_to_expression (state, elem, top); | |
2141 | } | |
2142 | ||
2143 | /* Lower a rust_op to a gdb expression. STATE is the parser state. | |
2144 | OPERATION is the operation to lower. TOP is a pointer to the | |
2145 | top-most operation; it is used to handle the special case where the | |
2146 | top-most expression is an identifier and can be optionally lowered | |
2147 | to OP_TYPE. */ | |
2148 | ||
2149 | static void | |
2150 | convert_ast_to_expression (struct parser_state *state, | |
2151 | const struct rust_op *operation, | |
2152 | const struct rust_op *top) | |
2153 | { | |
2154 | switch (operation->opcode) | |
2155 | { | |
2156 | case OP_LONG: | |
2157 | write_exp_elt_opcode (state, OP_LONG); | |
2158 | write_exp_elt_type (state, operation->left.typed_val_int.type); | |
2159 | write_exp_elt_longcst (state, operation->left.typed_val_int.val); | |
2160 | write_exp_elt_opcode (state, OP_LONG); | |
2161 | break; | |
2162 | ||
2163 | case OP_DOUBLE: | |
2164 | write_exp_elt_opcode (state, OP_DOUBLE); | |
2165 | write_exp_elt_type (state, operation->left.typed_val_float.type); | |
2166 | write_exp_elt_dblcst (state, operation->left.typed_val_float.dval); | |
2167 | write_exp_elt_opcode (state, OP_DOUBLE); | |
2168 | break; | |
2169 | ||
2170 | case STRUCTOP_STRUCT: | |
2171 | { | |
2172 | convert_ast_to_expression (state, operation->left.op, top); | |
2173 | ||
2174 | if (operation->completing) | |
2175 | mark_struct_expression (state); | |
2176 | write_exp_elt_opcode (state, STRUCTOP_STRUCT); | |
2177 | write_exp_string (state, operation->right.sval); | |
2178 | write_exp_elt_opcode (state, STRUCTOP_STRUCT); | |
2179 | } | |
2180 | break; | |
2181 | ||
2182 | case STRUCTOP_ANONYMOUS: | |
2183 | { | |
2184 | convert_ast_to_expression (state, operation->left.op, top); | |
2185 | ||
2186 | write_exp_elt_opcode (state, STRUCTOP_ANONYMOUS); | |
2187 | write_exp_elt_longcst (state, operation->right.typed_val_int.val); | |
2188 | write_exp_elt_opcode (state, STRUCTOP_ANONYMOUS); | |
2189 | } | |
2190 | break; | |
2191 | ||
2192 | case UNOP_PLUS: | |
2193 | case UNOP_NEG: | |
2194 | case UNOP_COMPLEMENT: | |
2195 | case UNOP_IND: | |
2196 | case UNOP_ADDR: | |
2197 | convert_ast_to_expression (state, operation->left.op, top); | |
2198 | write_exp_elt_opcode (state, operation->opcode); | |
2199 | break; | |
2200 | ||
2201 | case BINOP_SUBSCRIPT: | |
2202 | case BINOP_MUL: | |
2203 | case BINOP_REPEAT: | |
2204 | case BINOP_DIV: | |
2205 | case BINOP_REM: | |
2206 | case BINOP_LESS: | |
2207 | case BINOP_GTR: | |
2208 | case BINOP_BITWISE_AND: | |
2209 | case BINOP_BITWISE_IOR: | |
2210 | case BINOP_BITWISE_XOR: | |
2211 | case BINOP_ADD: | |
2212 | case BINOP_SUB: | |
2213 | case BINOP_LOGICAL_OR: | |
2214 | case BINOP_LOGICAL_AND: | |
2215 | case BINOP_EQUAL: | |
2216 | case BINOP_NOTEQUAL: | |
2217 | case BINOP_LEQ: | |
2218 | case BINOP_GEQ: | |
2219 | case BINOP_LSH: | |
2220 | case BINOP_RSH: | |
2221 | case BINOP_ASSIGN: | |
2222 | case OP_RUST_ARRAY: | |
2223 | convert_ast_to_expression (state, operation->left.op, top); | |
2224 | convert_ast_to_expression (state, operation->right.op, top); | |
2225 | if (operation->compound_assignment) | |
2226 | { | |
2227 | write_exp_elt_opcode (state, BINOP_ASSIGN_MODIFY); | |
2228 | write_exp_elt_opcode (state, operation->opcode); | |
2229 | write_exp_elt_opcode (state, BINOP_ASSIGN_MODIFY); | |
2230 | } | |
2231 | else | |
2232 | write_exp_elt_opcode (state, operation->opcode); | |
2233 | ||
2234 | if (operation->compound_assignment | |
2235 | || operation->opcode == BINOP_ASSIGN) | |
2236 | { | |
2237 | struct type *type; | |
2238 | ||
2239 | type = language_lookup_primitive_type (parse_language (state), | |
2240 | parse_gdbarch (state), | |
2241 | "()"); | |
2242 | ||
2243 | write_exp_elt_opcode (state, OP_LONG); | |
2244 | write_exp_elt_type (state, type); | |
2245 | write_exp_elt_longcst (state, 0); | |
2246 | write_exp_elt_opcode (state, OP_LONG); | |
2247 | ||
2248 | write_exp_elt_opcode (state, BINOP_COMMA); | |
2249 | } | |
2250 | break; | |
2251 | ||
2252 | case UNOP_CAST: | |
2253 | { | |
2254 | struct type *type = convert_ast_to_type (state, operation->right.op); | |
2255 | ||
2256 | convert_ast_to_expression (state, operation->left.op, top); | |
2257 | write_exp_elt_opcode (state, UNOP_CAST); | |
2258 | write_exp_elt_type (state, type); | |
2259 | write_exp_elt_opcode (state, UNOP_CAST); | |
2260 | } | |
2261 | break; | |
2262 | ||
2263 | case OP_FUNCALL: | |
2264 | { | |
2265 | if (operation->left.op->opcode == OP_VAR_VALUE) | |
2266 | { | |
2267 | struct type *type; | |
2268 | const char *varname = convert_name (state, operation->left.op); | |
2269 | ||
2270 | type = rust_lookup_type (varname, expression_context_block); | |
2271 | if (type != NULL) | |
2272 | { | |
2273 | /* This is actually a tuple struct expression, not a | |
2274 | call expression. */ | |
2275 | rust_op_ptr elem; | |
2276 | int i; | |
2277 | VEC (rust_op_ptr) *params = *operation->right.params; | |
2278 | ||
2279 | if (TYPE_CODE (type) != TYPE_CODE_NAMESPACE) | |
2280 | { | |
2281 | if (!rust_tuple_struct_type_p (type)) | |
2282 | error (_("Type %s is not a tuple struct"), varname); | |
2283 | ||
2284 | for (i = 0; | |
2285 | VEC_iterate (rust_op_ptr, params, i, elem); | |
2286 | ++i) | |
2287 | { | |
2288 | char *cell = get_print_cell (); | |
2289 | ||
2290 | xsnprintf (cell, PRINT_CELL_SIZE, "__%d", i); | |
2291 | write_exp_elt_opcode (state, OP_NAME); | |
2292 | write_exp_string (state, make_stoken (cell)); | |
2293 | write_exp_elt_opcode (state, OP_NAME); | |
2294 | ||
2295 | convert_ast_to_expression (state, elem, top); | |
2296 | } | |
2297 | ||
2298 | write_exp_elt_opcode (state, OP_AGGREGATE); | |
2299 | write_exp_elt_type (state, type); | |
2300 | write_exp_elt_longcst (state, | |
2301 | 2 * VEC_length (rust_op_ptr, | |
2302 | params)); | |
2303 | write_exp_elt_opcode (state, OP_AGGREGATE); | |
2304 | break; | |
2305 | } | |
2306 | } | |
2307 | } | |
2308 | convert_ast_to_expression (state, operation->left.op, top); | |
2309 | convert_params_to_expression (state, *operation->right.params, top); | |
2310 | write_exp_elt_opcode (state, OP_FUNCALL); | |
2311 | write_exp_elt_longcst (state, VEC_length (rust_op_ptr, | |
2312 | *operation->right.params)); | |
2313 | write_exp_elt_longcst (state, OP_FUNCALL); | |
2314 | } | |
2315 | break; | |
2316 | ||
2317 | case OP_ARRAY: | |
2318 | gdb_assert (operation->left.op == NULL); | |
2319 | convert_params_to_expression (state, *operation->right.params, top); | |
2320 | write_exp_elt_opcode (state, OP_ARRAY); | |
2321 | write_exp_elt_longcst (state, 0); | |
2322 | write_exp_elt_longcst (state, VEC_length (rust_op_ptr, | |
2323 | *operation->right.params) - 1); | |
2324 | write_exp_elt_longcst (state, OP_ARRAY); | |
2325 | break; | |
2326 | ||
2327 | case OP_VAR_VALUE: | |
2328 | { | |
2329 | struct block_symbol sym; | |
2330 | const char *varname; | |
2331 | ||
2332 | if (operation->left.sval.ptr[0] == '$') | |
2333 | { | |
2334 | write_dollar_variable (state, operation->left.sval); | |
2335 | break; | |
2336 | } | |
2337 | ||
2338 | varname = convert_name (state, operation); | |
2339 | sym = rust_lookup_symbol (varname, expression_context_block, | |
2340 | VAR_DOMAIN); | |
2341 | if (sym.symbol != NULL) | |
2342 | { | |
2343 | write_exp_elt_opcode (state, OP_VAR_VALUE); | |
2344 | write_exp_elt_block (state, sym.block); | |
2345 | write_exp_elt_sym (state, sym.symbol); | |
2346 | write_exp_elt_opcode (state, OP_VAR_VALUE); | |
2347 | } | |
2348 | else | |
2349 | { | |
2350 | struct type *type; | |
2351 | ||
2352 | type = rust_lookup_type (varname, expression_context_block); | |
2353 | if (type == NULL) | |
2354 | error (_("No symbol '%s' in current context"), varname); | |
2355 | ||
2356 | if (TYPE_CODE (type) == TYPE_CODE_STRUCT | |
2357 | && TYPE_NFIELDS (type) == 0) | |
2358 | { | |
2359 | /* A unit-like struct. */ | |
2360 | write_exp_elt_opcode (state, OP_AGGREGATE); | |
2361 | write_exp_elt_type (state, type); | |
2362 | write_exp_elt_longcst (state, 0); | |
2363 | write_exp_elt_opcode (state, OP_AGGREGATE); | |
2364 | } | |
2365 | else if (operation == top) | |
2366 | { | |
2367 | write_exp_elt_opcode (state, OP_TYPE); | |
2368 | write_exp_elt_type (state, type); | |
2369 | write_exp_elt_opcode (state, OP_TYPE); | |
2370 | break; | |
2371 | } | |
2372 | } | |
2373 | } | |
2374 | break; | |
2375 | ||
2376 | case OP_AGGREGATE: | |
2377 | { | |
2378 | int i; | |
2379 | int length; | |
2380 | struct set_field *init; | |
2381 | VEC (set_field) *fields = *operation->right.field_inits; | |
2382 | struct type *type; | |
2383 | const char *name; | |
2384 | ||
2385 | length = 0; | |
2386 | for (i = 0; VEC_iterate (set_field, fields, i, init); ++i) | |
2387 | { | |
2388 | if (init->name.ptr != NULL) | |
2389 | { | |
2390 | write_exp_elt_opcode (state, OP_NAME); | |
2391 | write_exp_string (state, init->name); | |
2392 | write_exp_elt_opcode (state, OP_NAME); | |
2393 | ++length; | |
2394 | } | |
2395 | ||
2396 | convert_ast_to_expression (state, init->init, top); | |
2397 | ++length; | |
2398 | ||
2399 | if (init->name.ptr == NULL) | |
2400 | { | |
2401 | /* This is handled differently from Ada in our | |
2402 | evaluator. */ | |
2403 | write_exp_elt_opcode (state, OP_OTHERS); | |
2404 | } | |
2405 | } | |
2406 | ||
2407 | name = convert_name (state, operation->left.op); | |
2408 | type = rust_lookup_type (name, expression_context_block); | |
2409 | if (type == NULL) | |
2410 | error (_("Could not find type '%s'"), operation->left.sval.ptr); | |
2411 | ||
2412 | if (TYPE_CODE (type) != TYPE_CODE_STRUCT | |
2413 | || rust_tuple_type_p (type) | |
2414 | || rust_tuple_struct_type_p (type)) | |
2415 | error (_("Struct expression applied to non-struct type")); | |
2416 | ||
2417 | write_exp_elt_opcode (state, OP_AGGREGATE); | |
2418 | write_exp_elt_type (state, type); | |
2419 | write_exp_elt_longcst (state, length); | |
2420 | write_exp_elt_opcode (state, OP_AGGREGATE); | |
2421 | } | |
2422 | break; | |
2423 | ||
2424 | case OP_STRING: | |
2425 | { | |
2426 | write_exp_elt_opcode (state, OP_STRING); | |
2427 | write_exp_string (state, operation->left.sval); | |
2428 | write_exp_elt_opcode (state, OP_STRING); | |
2429 | } | |
2430 | break; | |
2431 | ||
01739a3b | 2432 | case OP_RANGE: |
c44af4eb | 2433 | { |
01739a3b | 2434 | enum range_type kind = BOTH_BOUND_DEFAULT; |
c44af4eb TT |
2435 | |
2436 | if (operation->left.op != NULL) | |
2437 | { | |
2438 | convert_ast_to_expression (state, operation->left.op, top); | |
2439 | kind = HIGH_BOUND_DEFAULT; | |
2440 | } | |
2441 | if (operation->right.op != NULL) | |
2442 | { | |
2443 | convert_ast_to_expression (state, operation->right.op, top); | |
2444 | if (kind == BOTH_BOUND_DEFAULT) | |
2445 | kind = LOW_BOUND_DEFAULT; | |
2446 | else | |
2447 | { | |
2448 | gdb_assert (kind == HIGH_BOUND_DEFAULT); | |
2449 | kind = NONE_BOUND_DEFAULT; | |
2450 | } | |
2451 | } | |
01739a3b | 2452 | write_exp_elt_opcode (state, OP_RANGE); |
c44af4eb | 2453 | write_exp_elt_longcst (state, kind); |
01739a3b | 2454 | write_exp_elt_opcode (state, OP_RANGE); |
c44af4eb TT |
2455 | } |
2456 | break; | |
2457 | ||
2458 | default: | |
2459 | gdb_assert_not_reached ("unhandled opcode in convert_ast_to_expression"); | |
2460 | } | |
2461 | } | |
2462 | ||
2463 | \f | |
2464 | ||
2465 | /* The parser as exposed to gdb. */ | |
2466 | ||
2467 | int | |
2468 | rust_parse (struct parser_state *state) | |
2469 | { | |
2470 | int result; | |
2471 | struct cleanup *cleanup; | |
2472 | ||
2473 | obstack_init (&work_obstack); | |
2474 | cleanup = make_cleanup_obstack_free (&work_obstack); | |
2475 | rust_ast = NULL; | |
2476 | ||
2477 | pstate = state; | |
2478 | result = rustyyparse (); | |
2479 | ||
2480 | if (!result || (parse_completion && rust_ast != NULL)) | |
2481 | { | |
2482 | const struct rust_op *ast = rust_ast; | |
2483 | ||
2484 | rust_ast = NULL; | |
2485 | gdb_assert (ast != NULL); | |
2486 | convert_ast_to_expression (state, ast, ast); | |
2487 | } | |
2488 | ||
2489 | do_cleanups (cleanup); | |
2490 | return result; | |
2491 | } | |
2492 | ||
2493 | /* The parser error handler. */ | |
2494 | ||
2495 | void | |
2496 | rustyyerror (char *msg) | |
2497 | { | |
2498 | const char *where = prev_lexptr ? prev_lexptr : lexptr; | |
2499 | error (_("%s in expression, near `%s'."), (msg ? msg : "Error"), where); | |
2500 | } | |
2501 | ||
2502 | \f | |
2503 | ||
2504 | #if GDB_SELF_TEST | |
2505 | ||
2506 | /* Initialize the lexer for testing. */ | |
2507 | ||
2508 | static void | |
2509 | rust_lex_test_init (const char *input) | |
2510 | { | |
2511 | prev_lexptr = NULL; | |
2512 | lexptr = input; | |
2513 | paren_depth = 0; | |
2514 | } | |
2515 | ||
2516 | /* A test helper that lexes a string, expecting a single token. It | |
2517 | returns the lexer data for this token. */ | |
2518 | ||
2519 | static RUSTSTYPE | |
2520 | rust_lex_test_one (const char *input, int expected) | |
2521 | { | |
2522 | int token; | |
2523 | RUSTSTYPE result; | |
2524 | ||
2525 | rust_lex_test_init (input); | |
2526 | ||
2527 | token = rustyylex (); | |
2528 | SELF_CHECK (token == expected); | |
2529 | result = rustyylval; | |
2530 | ||
2531 | if (token) | |
2532 | { | |
2533 | token = rustyylex (); | |
2534 | SELF_CHECK (token == 0); | |
2535 | } | |
2536 | ||
2537 | return result; | |
2538 | } | |
2539 | ||
2540 | /* Test that INPUT lexes as the integer VALUE. */ | |
2541 | ||
2542 | static void | |
2543 | rust_lex_int_test (const char *input, int value, int kind) | |
2544 | { | |
2545 | RUSTSTYPE result = rust_lex_test_one (input, kind); | |
2546 | SELF_CHECK (result.typed_val_int.val == value); | |
2547 | } | |
2548 | ||
2549 | /* Test that INPUT throws an exception with text ERR. */ | |
2550 | ||
2551 | static void | |
2552 | rust_lex_exception_test (const char *input, const char *err) | |
2553 | { | |
2554 | TRY | |
2555 | { | |
2556 | /* The "kind" doesn't matter. */ | |
2557 | rust_lex_test_one (input, DECIMAL_INTEGER); | |
2558 | SELF_CHECK (0); | |
2559 | } | |
2560 | CATCH (except, RETURN_MASK_ERROR) | |
2561 | { | |
2562 | SELF_CHECK (strcmp (except.message, err) == 0); | |
2563 | } | |
2564 | END_CATCH | |
2565 | } | |
2566 | ||
2567 | /* Test that INPUT lexes as the identifier, string, or byte-string | |
2568 | VALUE. KIND holds the expected token kind. */ | |
2569 | ||
2570 | static void | |
2571 | rust_lex_stringish_test (const char *input, const char *value, int kind) | |
2572 | { | |
2573 | RUSTSTYPE result = rust_lex_test_one (input, kind); | |
2574 | SELF_CHECK (result.sval.length == strlen (value)); | |
2575 | SELF_CHECK (strncmp (result.sval.ptr, value, result.sval.length) == 0); | |
2576 | } | |
2577 | ||
2578 | /* Helper to test that a string parses as a given token sequence. */ | |
2579 | ||
2580 | static void | |
2581 | rust_lex_test_sequence (const char *input, int len, const int expected[]) | |
2582 | { | |
2583 | int i; | |
2584 | ||
2585 | lexptr = input; | |
2586 | paren_depth = 0; | |
2587 | ||
2588 | for (i = 0; i < len; ++i) | |
2589 | { | |
2590 | int token = rustyylex (); | |
2591 | ||
2592 | SELF_CHECK (token == expected[i]); | |
2593 | } | |
2594 | } | |
2595 | ||
2596 | /* Tests for an integer-parsing corner case. */ | |
2597 | ||
2598 | static void | |
2599 | rust_lex_test_trailing_dot (void) | |
2600 | { | |
2601 | const int expected1[] = { DECIMAL_INTEGER, '.', IDENT, '(', ')', 0 }; | |
2602 | const int expected2[] = { INTEGER, '.', IDENT, '(', ')', 0 }; | |
2603 | const int expected3[] = { FLOAT, EQEQ, '(', ')', 0 }; | |
2604 | const int expected4[] = { DECIMAL_INTEGER, DOTDOT, DECIMAL_INTEGER, 0 }; | |
2605 | ||
2606 | rust_lex_test_sequence ("23.g()", ARRAY_SIZE (expected1), expected1); | |
2607 | rust_lex_test_sequence ("23_0.g()", ARRAY_SIZE (expected2), expected2); | |
2608 | rust_lex_test_sequence ("23.==()", ARRAY_SIZE (expected3), expected3); | |
2609 | rust_lex_test_sequence ("23..25", ARRAY_SIZE (expected4), expected4); | |
2610 | } | |
2611 | ||
2612 | /* Tests of completion. */ | |
2613 | ||
2614 | static void | |
2615 | rust_lex_test_completion (void) | |
2616 | { | |
2617 | const int expected[] = { IDENT, '.', COMPLETE, 0 }; | |
2618 | ||
2619 | parse_completion = 1; | |
2620 | ||
2621 | rust_lex_test_sequence ("something.wha", ARRAY_SIZE (expected), expected); | |
2622 | rust_lex_test_sequence ("something.", ARRAY_SIZE (expected), expected); | |
2623 | ||
2624 | parse_completion = 0; | |
2625 | } | |
2626 | ||
2627 | /* Test pushback. */ | |
2628 | ||
2629 | static void | |
2630 | rust_lex_test_push_back (void) | |
2631 | { | |
2632 | int token; | |
2633 | ||
2634 | rust_lex_test_init (">>="); | |
2635 | ||
2636 | token = rustyylex (); | |
2637 | SELF_CHECK (token == COMPOUND_ASSIGN); | |
2638 | SELF_CHECK (rustyylval.opcode == BINOP_RSH); | |
2639 | ||
2640 | rust_push_back ('='); | |
2641 | ||
2642 | token = rustyylex (); | |
2643 | SELF_CHECK (token == '='); | |
2644 | ||
2645 | token = rustyylex (); | |
2646 | SELF_CHECK (token == 0); | |
2647 | } | |
2648 | ||
2649 | /* Unit test the lexer. */ | |
2650 | ||
2651 | static void | |
2652 | rust_lex_tests (void) | |
2653 | { | |
2654 | int i; | |
2655 | ||
2656 | obstack_init (&work_obstack); | |
2657 | unit_testing = 1; | |
2658 | ||
2659 | rust_lex_test_one ("", 0); | |
2660 | rust_lex_test_one (" \t \n \r ", 0); | |
2661 | rust_lex_test_one ("thread 23", 0); | |
2662 | rust_lex_test_one ("task 23", 0); | |
2663 | rust_lex_test_one ("th 104", 0); | |
2664 | rust_lex_test_one ("ta 97", 0); | |
2665 | ||
2666 | rust_lex_int_test ("'z'", 'z', INTEGER); | |
2667 | rust_lex_int_test ("'\\xff'", 0xff, INTEGER); | |
2668 | rust_lex_int_test ("'\\u{1016f}'", 0x1016f, INTEGER); | |
2669 | rust_lex_int_test ("b'z'", 'z', INTEGER); | |
2670 | rust_lex_int_test ("b'\\xfe'", 0xfe, INTEGER); | |
2671 | rust_lex_int_test ("b'\\xFE'", 0xfe, INTEGER); | |
2672 | rust_lex_int_test ("b'\\xfE'", 0xfe, INTEGER); | |
2673 | ||
2674 | /* Test all escapes in both modes. */ | |
2675 | rust_lex_int_test ("'\\n'", '\n', INTEGER); | |
2676 | rust_lex_int_test ("'\\r'", '\r', INTEGER); | |
2677 | rust_lex_int_test ("'\\t'", '\t', INTEGER); | |
2678 | rust_lex_int_test ("'\\\\'", '\\', INTEGER); | |
2679 | rust_lex_int_test ("'\\0'", '\0', INTEGER); | |
2680 | rust_lex_int_test ("'\\''", '\'', INTEGER); | |
2681 | rust_lex_int_test ("'\\\"'", '"', INTEGER); | |
2682 | ||
2683 | rust_lex_int_test ("b'\\n'", '\n', INTEGER); | |
2684 | rust_lex_int_test ("b'\\r'", '\r', INTEGER); | |
2685 | rust_lex_int_test ("b'\\t'", '\t', INTEGER); | |
2686 | rust_lex_int_test ("b'\\\\'", '\\', INTEGER); | |
2687 | rust_lex_int_test ("b'\\0'", '\0', INTEGER); | |
2688 | rust_lex_int_test ("b'\\''", '\'', INTEGER); | |
2689 | rust_lex_int_test ("b'\\\"'", '"', INTEGER); | |
2690 | ||
2691 | rust_lex_exception_test ("'z", "Unterminated character literal"); | |
2692 | rust_lex_exception_test ("b'\\x0'", "Not enough hex digits seen"); | |
2693 | rust_lex_exception_test ("b'\\u{0}'", "Unicode escape in byte literal"); | |
2694 | rust_lex_exception_test ("'\\x0'", "Not enough hex digits seen"); | |
2695 | rust_lex_exception_test ("'\\u0'", "Missing '{' in Unicode escape"); | |
2696 | rust_lex_exception_test ("'\\u{0", "Missing '}' in Unicode escape"); | |
2697 | rust_lex_exception_test ("'\\u{0000007}", "Overlong hex escape"); | |
2698 | rust_lex_exception_test ("'\\u{}", "Not enough hex digits seen"); | |
2699 | rust_lex_exception_test ("'\\Q'", "Invalid escape \\Q in literal"); | |
2700 | rust_lex_exception_test ("b'\\Q'", "Invalid escape \\Q in literal"); | |
2701 | ||
2702 | rust_lex_int_test ("23", 23, DECIMAL_INTEGER); | |
2703 | rust_lex_int_test ("2_344__29", 234429, INTEGER); | |
2704 | rust_lex_int_test ("0x1f", 0x1f, INTEGER); | |
2705 | rust_lex_int_test ("23usize", 23, INTEGER); | |
2706 | rust_lex_int_test ("23i32", 23, INTEGER); | |
2707 | rust_lex_int_test ("0x1_f", 0x1f, INTEGER); | |
2708 | rust_lex_int_test ("0b1_101011__", 0x6b, INTEGER); | |
2709 | rust_lex_int_test ("0o001177i64", 639, INTEGER); | |
2710 | ||
2711 | rust_lex_test_trailing_dot (); | |
2712 | ||
2713 | rust_lex_test_one ("23.", FLOAT); | |
2714 | rust_lex_test_one ("23.99f32", FLOAT); | |
2715 | rust_lex_test_one ("23e7", FLOAT); | |
2716 | rust_lex_test_one ("23E-7", FLOAT); | |
2717 | rust_lex_test_one ("23e+7", FLOAT); | |
2718 | rust_lex_test_one ("23.99e+7f64", FLOAT); | |
2719 | rust_lex_test_one ("23.82f32", FLOAT); | |
2720 | ||
2721 | rust_lex_stringish_test ("hibob", "hibob", IDENT); | |
2722 | rust_lex_stringish_test ("hibob__93", "hibob__93", IDENT); | |
2723 | rust_lex_stringish_test ("thread", "thread", IDENT); | |
2724 | ||
2725 | rust_lex_stringish_test ("\"string\"", "string", STRING); | |
2726 | rust_lex_stringish_test ("\"str\\ting\"", "str\ting", STRING); | |
2727 | rust_lex_stringish_test ("\"str\\\"ing\"", "str\"ing", STRING); | |
2728 | rust_lex_stringish_test ("r\"str\\ing\"", "str\\ing", STRING); | |
2729 | rust_lex_stringish_test ("r#\"str\\ting\"#", "str\\ting", STRING); | |
2730 | rust_lex_stringish_test ("r###\"str\\\"ing\"###", "str\\\"ing", STRING); | |
2731 | ||
2732 | rust_lex_stringish_test ("b\"string\"", "string", BYTESTRING); | |
2733 | rust_lex_stringish_test ("b\"\x73tring\"", "string", BYTESTRING); | |
2734 | rust_lex_stringish_test ("b\"str\\\"ing\"", "str\"ing", BYTESTRING); | |
2735 | rust_lex_stringish_test ("br####\"\\x73tring\"####", "\\x73tring", | |
2736 | BYTESTRING); | |
2737 | ||
2738 | for (i = 0; i < ARRAY_SIZE (identifier_tokens); ++i) | |
2739 | rust_lex_test_one (identifier_tokens[i].name, identifier_tokens[i].value); | |
2740 | ||
2741 | for (i = 0; i < ARRAY_SIZE (operator_tokens); ++i) | |
2742 | rust_lex_test_one (operator_tokens[i].name, operator_tokens[i].value); | |
2743 | ||
2744 | rust_lex_test_completion (); | |
2745 | rust_lex_test_push_back (); | |
2746 | ||
2747 | obstack_free (&work_obstack, NULL); | |
2748 | unit_testing = 0; | |
2749 | } | |
2750 | ||
2751 | #endif /* GDB_SELF_TEST */ | |
2752 | ||
2753 | void | |
2754 | _initialize_rust_exp (void) | |
2755 | { | |
2756 | int code = regcomp (&number_regex, number_regex_text, REG_EXTENDED); | |
2757 | /* If the regular expression was incorrect, it was a programming | |
2758 | error. */ | |
2759 | gdb_assert (code == 0); | |
2760 | ||
2761 | #if GDB_SELF_TEST | |
2762 | register_self_test (rust_lex_tests); | |
2763 | #endif | |
2764 | } |