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