Commit | Line | Data |
---|---|---|
c44af4eb | 1 | /* Bison parser for Rust expressions, for GDB. |
61baf725 | 2 | Copyright (C) 2016-2017 Free Software Foundation, Inc. |
c44af4eb TT |
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; | |
3232fabd | 45 | typedef std::vector<const struct rust_op *> rust_op_vector; |
c44af4eb TT |
46 | |
47 | /* A typed integer constant. */ | |
48 | ||
49 | struct typed_val_int | |
50 | { | |
51 | LONGEST val; | |
52 | struct type *type; | |
53 | }; | |
54 | ||
55 | /* A typed floating point constant. */ | |
56 | ||
57 | struct typed_val_float | |
58 | { | |
59 | DOUBLEST dval; | |
60 | struct type *type; | |
61 | }; | |
62 | ||
63 | /* An identifier and an expression. This is used to represent one | |
64 | element of a struct initializer. */ | |
65 | ||
66 | struct set_field | |
67 | { | |
68 | struct stoken name; | |
69 | const struct rust_op *init; | |
70 | }; | |
71 | ||
3232fabd | 72 | typedef std::vector<set_field> rust_set_vector; |
c44af4eb TT |
73 | |
74 | static int rustyylex (void); | |
75 | static void rust_push_back (char c); | |
76 | static const char *rust_copy_name (const char *, int); | |
77 | static struct stoken rust_concat3 (const char *, const char *, const char *); | |
78 | static struct stoken make_stoken (const char *); | |
79 | static struct block_symbol rust_lookup_symbol (const char *name, | |
80 | const struct block *block, | |
81 | const domain_enum domain); | |
82 | static struct type *rust_lookup_type (const char *name, | |
83 | const struct block *block); | |
84 | static struct type *rust_type (const char *name); | |
85 | ||
86 | static const struct rust_op *crate_name (const struct rust_op *name); | |
87 | static const struct rust_op *super_name (const struct rust_op *name, | |
88 | unsigned int n_supers); | |
89 | ||
90 | static const struct rust_op *ast_operation (enum exp_opcode opcode, | |
91 | const struct rust_op *left, | |
92 | const struct rust_op *right); | |
93 | static const struct rust_op *ast_compound_assignment | |
94 | (enum exp_opcode opcode, const struct rust_op *left, | |
95 | const struct rust_op *rust_op); | |
96 | static const struct rust_op *ast_literal (struct typed_val_int val); | |
97 | static const struct rust_op *ast_dliteral (struct typed_val_float val); | |
98 | static const struct rust_op *ast_structop (const struct rust_op *left, | |
99 | const char *name, | |
100 | int completing); | |
101 | static const struct rust_op *ast_structop_anonymous | |
102 | (const struct rust_op *left, struct typed_val_int number); | |
103 | static const struct rust_op *ast_unary (enum exp_opcode opcode, | |
104 | const struct rust_op *expr); | |
105 | static const struct rust_op *ast_cast (const struct rust_op *expr, | |
106 | const struct rust_op *type); | |
107 | static const struct rust_op *ast_call_ish (enum exp_opcode opcode, | |
108 | const struct rust_op *expr, | |
3232fabd | 109 | rust_op_vector *params); |
c44af4eb | 110 | static const struct rust_op *ast_path (struct stoken name, |
3232fabd | 111 | rust_op_vector *params); |
c44af4eb TT |
112 | static const struct rust_op *ast_string (struct stoken str); |
113 | static const struct rust_op *ast_struct (const struct rust_op *name, | |
3232fabd | 114 | rust_set_vector *fields); |
c44af4eb TT |
115 | static const struct rust_op *ast_range (const struct rust_op *lhs, |
116 | const struct rust_op *rhs); | |
117 | static const struct rust_op *ast_array_type (const struct rust_op *lhs, | |
118 | struct typed_val_int val); | |
119 | static const struct rust_op *ast_slice_type (const struct rust_op *type); | |
120 | static const struct rust_op *ast_reference_type (const struct rust_op *type); | |
121 | static const struct rust_op *ast_pointer_type (const struct rust_op *type, | |
122 | int is_mut); | |
123 | static const struct rust_op *ast_function_type (const struct rust_op *result, | |
3232fabd TT |
124 | rust_op_vector *params); |
125 | static const struct rust_op *ast_tuple_type (rust_op_vector *params); | |
c44af4eb | 126 | |
3232fabd | 127 | /* The current rust parser. */ |
c44af4eb | 128 | |
3232fabd TT |
129 | struct rust_parser; |
130 | static rust_parser *current_parser; | |
c44af4eb TT |
131 | |
132 | /* A regular expression for matching Rust numbers. This is split up | |
133 | since it is very long and this gives us a way to comment the | |
134 | sections. */ | |
135 | ||
136 | static const char *number_regex_text = | |
137 | /* subexpression 1: allows use of alternation, otherwise uninteresting */ | |
138 | "^(" | |
139 | /* First comes floating point. */ | |
140 | /* Recognize number after the decimal point, with optional | |
141 | exponent and optional type suffix. | |
142 | subexpression 2: allows "?", otherwise uninteresting | |
143 | subexpression 3: if present, type suffix | |
144 | */ | |
145 | "[0-9][0-9_]*\\.[0-9][0-9_]*([eE][-+]?[0-9][0-9_]*)?(f32|f64)?" | |
146 | #define FLOAT_TYPE1 3 | |
147 | "|" | |
148 | /* Recognize exponent without decimal point, with optional type | |
149 | suffix. | |
150 | subexpression 4: if present, type suffix | |
151 | */ | |
152 | #define FLOAT_TYPE2 4 | |
153 | "[0-9][0-9_]*[eE][-+]?[0-9][0-9_]*(f32|f64)?" | |
154 | "|" | |
155 | /* "23." is a valid floating point number, but "23.e5" and | |
156 | "23.f32" are not. So, handle the trailing-. case | |
157 | separately. */ | |
158 | "[0-9][0-9_]*\\." | |
159 | "|" | |
160 | /* Finally come integers. | |
161 | subexpression 5: text of integer | |
162 | subexpression 6: if present, type suffix | |
163 | subexpression 7: allows use of alternation, otherwise uninteresting | |
164 | */ | |
165 | #define INT_TEXT 5 | |
166 | #define INT_TYPE 6 | |
167 | "(0x[a-fA-F0-9_]+|0o[0-7_]+|0b[01_]+|[0-9][0-9_]*)" | |
168 | "([iu](size|8|16|32|64))?" | |
169 | ")"; | |
170 | /* The number of subexpressions to allocate space for, including the | |
171 | "0th" whole match subexpression. */ | |
172 | #define NUM_SUBEXPRESSIONS 8 | |
173 | ||
174 | /* The compiled number-matching regex. */ | |
175 | ||
176 | static regex_t number_regex; | |
177 | ||
178 | /* True if we're running unit tests. */ | |
179 | ||
180 | static int unit_testing; | |
181 | ||
3232fabd TT |
182 | /* Obstack for data temporarily allocated during parsing. Points to |
183 | the obstack in the rust_parser, or to a temporary obstack during | |
184 | unit testing. */ | |
185 | ||
186 | static auto_obstack *work_obstack; | |
187 | ||
188 | /* An instance of this is created before parsing, and destroyed when | |
189 | parsing is finished. */ | |
190 | ||
191 | struct rust_parser | |
192 | { | |
193 | rust_parser (struct parser_state *state) | |
194 | : rust_ast (nullptr), | |
195 | pstate (state) | |
196 | { | |
197 | gdb_assert (current_parser == nullptr); | |
198 | current_parser = this; | |
199 | work_obstack = &obstack; | |
200 | } | |
201 | ||
202 | ~rust_parser () | |
203 | { | |
204 | /* Clean up the globals we set. */ | |
205 | current_parser = nullptr; | |
206 | work_obstack = nullptr; | |
207 | } | |
208 | ||
209 | /* Create a new rust_set_vector. The storage for the new vector is | |
210 | managed by this class. */ | |
211 | rust_set_vector *new_set_vector () | |
212 | { | |
213 | rust_set_vector *result = new rust_set_vector; | |
214 | set_vectors.push_back (std::unique_ptr<rust_set_vector> (result)); | |
215 | return result; | |
216 | } | |
217 | ||
218 | /* Create a new rust_ops_vector. The storage for the new vector is | |
219 | managed by this class. */ | |
220 | rust_op_vector *new_op_vector () | |
221 | { | |
222 | rust_op_vector *result = new rust_op_vector; | |
223 | op_vectors.push_back (std::unique_ptr<rust_op_vector> (result)); | |
224 | return result; | |
225 | } | |
226 | ||
227 | /* Return the parser's language. */ | |
228 | const struct language_defn *language () const | |
229 | { | |
230 | return parse_language (pstate); | |
231 | } | |
232 | ||
233 | /* Return the parser's gdbarch. */ | |
234 | struct gdbarch *arch () const | |
235 | { | |
236 | return parse_gdbarch (pstate); | |
237 | } | |
238 | ||
239 | /* A pointer to this is installed globally. */ | |
240 | auto_obstack obstack; | |
241 | ||
242 | /* Result of parsing. Points into obstack. */ | |
243 | const struct rust_op *rust_ast; | |
244 | ||
245 | /* This keeps track of the various vectors we allocate. */ | |
246 | std::vector<std::unique_ptr<rust_set_vector>> set_vectors; | |
247 | std::vector<std::unique_ptr<rust_op_vector>> op_vectors; | |
248 | ||
249 | /* The parser state gdb gave us. */ | |
250 | struct parser_state *pstate; | |
251 | }; | |
c44af4eb TT |
252 | |
253 | %} | |
254 | ||
255 | %union | |
256 | { | |
257 | /* A typed integer constant. */ | |
258 | struct typed_val_int typed_val_int; | |
259 | ||
260 | /* A typed floating point constant. */ | |
261 | struct typed_val_float typed_val_float; | |
262 | ||
263 | /* An identifier or string. */ | |
264 | struct stoken sval; | |
265 | ||
266 | /* A token representing an opcode, like "==". */ | |
267 | enum exp_opcode opcode; | |
268 | ||
269 | /* A list of expressions; for example, the arguments to a function | |
270 | call. */ | |
3232fabd | 271 | rust_op_vector *params; |
c44af4eb TT |
272 | |
273 | /* A list of field initializers. */ | |
3232fabd | 274 | rust_set_vector *field_inits; |
c44af4eb TT |
275 | |
276 | /* A single field initializer. */ | |
277 | struct set_field one_field_init; | |
278 | ||
279 | /* An expression. */ | |
280 | const struct rust_op *op; | |
281 | ||
282 | /* A plain integer, for example used to count the number of | |
283 | "super::" prefixes on a path. */ | |
284 | unsigned int depth; | |
285 | } | |
286 | ||
287 | %{ | |
288 | ||
289 | /* Rust AST operations. We build a tree of these; then lower them | |
290 | to gdb expressions when parsing has completed. */ | |
291 | ||
292 | struct rust_op | |
293 | { | |
294 | /* The opcode. */ | |
295 | enum exp_opcode opcode; | |
296 | /* If OPCODE is OP_TYPE, then this holds information about what type | |
297 | is described by this node. */ | |
298 | enum type_code typecode; | |
299 | /* Indicates whether OPCODE actually represents a compound | |
300 | assignment. For example, if OPCODE is GTGT and this is false, | |
301 | then this rust_op represents an ordinary ">>"; but if this is | |
302 | true, then this rust_op represents ">>=". Unused in other | |
303 | cases. */ | |
304 | unsigned int compound_assignment : 1; | |
305 | /* Only used by a field expression; if set, indicates that the field | |
306 | name occurred at the end of the expression and is eligible for | |
307 | completion. */ | |
308 | unsigned int completing : 1; | |
309 | /* Operands of expression. Which one is used and how depends on the | |
310 | particular opcode. */ | |
311 | RUSTSTYPE left; | |
312 | RUSTSTYPE right; | |
313 | }; | |
314 | ||
315 | %} | |
316 | ||
317 | %token <sval> GDBVAR | |
318 | %token <sval> IDENT | |
319 | %token <sval> COMPLETE | |
320 | %token <typed_val_int> INTEGER | |
321 | %token <typed_val_int> DECIMAL_INTEGER | |
322 | %token <sval> STRING | |
323 | %token <sval> BYTESTRING | |
324 | %token <typed_val_float> FLOAT | |
325 | %token <opcode> COMPOUND_ASSIGN | |
326 | ||
327 | /* Keyword tokens. */ | |
328 | %token <voidval> KW_AS | |
329 | %token <voidval> KW_IF | |
330 | %token <voidval> KW_TRUE | |
331 | %token <voidval> KW_FALSE | |
332 | %token <voidval> KW_SUPER | |
333 | %token <voidval> KW_SELF | |
334 | %token <voidval> KW_MUT | |
335 | %token <voidval> KW_EXTERN | |
336 | %token <voidval> KW_CONST | |
337 | %token <voidval> KW_FN | |
cdf5a07c | 338 | %token <voidval> KW_SIZEOF |
c44af4eb TT |
339 | |
340 | /* Operator tokens. */ | |
341 | %token <voidval> DOTDOT | |
342 | %token <voidval> OROR | |
343 | %token <voidval> ANDAND | |
344 | %token <voidval> EQEQ | |
345 | %token <voidval> NOTEQ | |
346 | %token <voidval> LTEQ | |
347 | %token <voidval> GTEQ | |
348 | %token <voidval> LSH RSH | |
349 | %token <voidval> COLONCOLON | |
350 | %token <voidval> ARROW | |
351 | ||
352 | %type <op> type | |
353 | %type <op> path_for_expr | |
354 | %type <op> identifier_path_for_expr | |
355 | %type <op> path_for_type | |
356 | %type <op> identifier_path_for_type | |
357 | %type <op> just_identifiers_for_type | |
358 | ||
359 | %type <params> maybe_type_list | |
360 | %type <params> type_list | |
361 | ||
362 | %type <depth> super_path | |
363 | ||
364 | %type <op> literal | |
365 | %type <op> expr | |
366 | %type <op> field_expr | |
367 | %type <op> idx_expr | |
368 | %type <op> unop_expr | |
369 | %type <op> binop_expr | |
370 | %type <op> binop_expr_expr | |
371 | %type <op> type_cast_expr | |
372 | %type <op> assignment_expr | |
373 | %type <op> compound_assignment_expr | |
374 | %type <op> paren_expr | |
375 | %type <op> call_expr | |
376 | %type <op> path_expr | |
377 | %type <op> tuple_expr | |
378 | %type <op> unit_expr | |
379 | %type <op> struct_expr | |
380 | %type <op> array_expr | |
381 | %type <op> range_expr | |
382 | ||
383 | %type <params> expr_list | |
384 | %type <params> maybe_expr_list | |
385 | %type <params> paren_expr_list | |
386 | ||
387 | %type <field_inits> struct_expr_list | |
388 | %type <one_field_init> struct_expr_tail | |
389 | ||
390 | /* Precedence. */ | |
391 | %nonassoc DOTDOT | |
392 | %right '=' COMPOUND_ASSIGN | |
393 | %left OROR | |
394 | %left ANDAND | |
395 | %nonassoc EQEQ NOTEQ '<' '>' LTEQ GTEQ | |
396 | %left '|' | |
397 | %left '^' | |
398 | %left '&' | |
399 | %left LSH RSH | |
400 | %left '@' | |
401 | %left '+' '-' | |
402 | %left '*' '/' '%' | |
403 | /* These could be %precedence in Bison, but that isn't a yacc | |
404 | feature. */ | |
405 | %left KW_AS | |
406 | %left UNARY | |
407 | %left '[' '.' '(' | |
408 | ||
409 | %% | |
410 | ||
411 | start: | |
412 | expr | |
413 | { | |
414 | /* If we are completing and see a valid parse, | |
415 | rust_ast will already have been set. */ | |
3232fabd TT |
416 | if (current_parser->rust_ast == NULL) |
417 | current_parser->rust_ast = $1; | |
c44af4eb TT |
418 | } |
419 | ; | |
420 | ||
421 | /* Note that the Rust grammar includes a method_call_expr, but we | |
422 | handle this differently, to avoid a shift/reduce conflict with | |
423 | call_expr. */ | |
424 | expr: | |
425 | literal | |
426 | | path_expr | |
427 | | tuple_expr | |
428 | | unit_expr | |
429 | | struct_expr | |
430 | | field_expr | |
431 | | array_expr | |
432 | | idx_expr | |
433 | | range_expr | |
cdf5a07c | 434 | | unop_expr /* Must precede call_expr because of ambiguity with sizeof. */ |
c44af4eb TT |
435 | | binop_expr |
436 | | paren_expr | |
437 | | call_expr | |
438 | ; | |
439 | ||
440 | tuple_expr: | |
441 | '(' expr ',' maybe_expr_list ')' | |
442 | { | |
3232fabd | 443 | $4->push_back ($2); |
c44af4eb TT |
444 | error (_("Tuple expressions not supported yet")); |
445 | } | |
446 | ; | |
447 | ||
448 | unit_expr: | |
449 | '(' ')' | |
450 | { | |
451 | struct typed_val_int val; | |
452 | ||
453 | val.type | |
3232fabd TT |
454 | = language_lookup_primitive_type (current_parser->language (), |
455 | current_parser->arch (), | |
c44af4eb TT |
456 | "()"); |
457 | val.val = 0; | |
458 | $$ = ast_literal (val); | |
459 | } | |
460 | ; | |
461 | ||
462 | /* To avoid a shift/reduce conflict with call_expr, we don't handle | |
463 | tuple struct expressions here, but instead when examining the | |
464 | AST. */ | |
465 | struct_expr: | |
466 | path_for_expr '{' struct_expr_list '}' | |
467 | { $$ = ast_struct ($1, $3); } | |
468 | ; | |
469 | ||
470 | struct_expr_tail: | |
471 | DOTDOT expr | |
472 | { | |
473 | struct set_field sf; | |
474 | ||
475 | sf.name.ptr = NULL; | |
476 | sf.name.length = 0; | |
477 | sf.init = $2; | |
478 | ||
479 | $$ = sf; | |
480 | } | |
481 | | IDENT ':' expr | |
482 | { | |
483 | struct set_field sf; | |
484 | ||
485 | sf.name = $1; | |
486 | sf.init = $3; | |
487 | $$ = sf; | |
488 | } | |
489 | ; | |
490 | ||
c44af4eb | 491 | struct_expr_list: |
12df5c00 TT |
492 | /* %empty */ |
493 | { | |
3232fabd | 494 | $$ = current_parser->new_set_vector (); |
12df5c00 TT |
495 | } |
496 | | struct_expr_tail | |
c44af4eb | 497 | { |
3232fabd TT |
498 | rust_set_vector *result = current_parser->new_set_vector (); |
499 | result->push_back ($1); | |
c44af4eb TT |
500 | $$ = result; |
501 | } | |
502 | | IDENT ':' expr ',' struct_expr_list | |
503 | { | |
504 | struct set_field sf; | |
505 | ||
506 | sf.name = $1; | |
507 | sf.init = $3; | |
3232fabd | 508 | $5->push_back (sf); |
c44af4eb TT |
509 | $$ = $5; |
510 | } | |
511 | ; | |
512 | ||
513 | array_expr: | |
514 | '[' KW_MUT expr_list ']' | |
515 | { $$ = ast_call_ish (OP_ARRAY, NULL, $3); } | |
516 | | '[' expr_list ']' | |
517 | { $$ = ast_call_ish (OP_ARRAY, NULL, $2); } | |
518 | | '[' KW_MUT expr ';' expr ']' | |
519 | { $$ = ast_operation (OP_RUST_ARRAY, $3, $5); } | |
520 | | '[' expr ';' expr ']' | |
521 | { $$ = ast_operation (OP_RUST_ARRAY, $2, $4); } | |
522 | ; | |
523 | ||
524 | range_expr: | |
525 | expr DOTDOT | |
526 | { $$ = ast_range ($1, NULL); } | |
527 | | expr DOTDOT expr | |
528 | { $$ = ast_range ($1, $3); } | |
529 | | DOTDOT expr | |
530 | { $$ = ast_range (NULL, $2); } | |
531 | | DOTDOT | |
532 | { $$ = ast_range (NULL, NULL); } | |
533 | ; | |
534 | ||
535 | literal: | |
536 | INTEGER | |
537 | { $$ = ast_literal ($1); } | |
538 | | DECIMAL_INTEGER | |
539 | { $$ = ast_literal ($1); } | |
540 | | FLOAT | |
541 | { $$ = ast_dliteral ($1); } | |
542 | | STRING | |
543 | { | |
544 | const struct rust_op *str = ast_string ($1); | |
c44af4eb TT |
545 | struct set_field field; |
546 | struct typed_val_int val; | |
547 | struct stoken token; | |
548 | ||
3232fabd | 549 | rust_set_vector *fields = current_parser->new_set_vector (); |
c44af4eb TT |
550 | |
551 | /* Wrap the raw string in the &str struct. */ | |
552 | field.name.ptr = "data_ptr"; | |
553 | field.name.length = strlen (field.name.ptr); | |
554 | field.init = ast_unary (UNOP_ADDR, ast_string ($1)); | |
3232fabd | 555 | fields->push_back (field); |
c44af4eb TT |
556 | |
557 | val.type = rust_type ("usize"); | |
558 | val.val = $1.length; | |
559 | ||
560 | field.name.ptr = "length"; | |
561 | field.name.length = strlen (field.name.ptr); | |
562 | field.init = ast_literal (val); | |
3232fabd | 563 | fields->push_back (field); |
c44af4eb TT |
564 | |
565 | token.ptr = "&str"; | |
566 | token.length = strlen (token.ptr); | |
567 | $$ = ast_struct (ast_path (token, NULL), fields); | |
568 | } | |
569 | | BYTESTRING | |
570 | { $$ = ast_string ($1); } | |
571 | | KW_TRUE | |
572 | { | |
573 | struct typed_val_int val; | |
574 | ||
3232fabd TT |
575 | val.type = language_bool_type (current_parser->language (), |
576 | current_parser->arch ()); | |
c44af4eb TT |
577 | val.val = 1; |
578 | $$ = ast_literal (val); | |
579 | } | |
580 | | KW_FALSE | |
581 | { | |
582 | struct typed_val_int val; | |
583 | ||
3232fabd TT |
584 | val.type = language_bool_type (current_parser->language (), |
585 | current_parser->arch ()); | |
c44af4eb TT |
586 | val.val = 0; |
587 | $$ = ast_literal (val); | |
588 | } | |
589 | ; | |
590 | ||
591 | field_expr: | |
592 | expr '.' IDENT | |
593 | { $$ = ast_structop ($1, $3.ptr, 0); } | |
594 | | expr '.' COMPLETE | |
595 | { | |
596 | $$ = ast_structop ($1, $3.ptr, 1); | |
3232fabd | 597 | current_parser->rust_ast = $$; |
c44af4eb TT |
598 | } |
599 | | expr '.' DECIMAL_INTEGER | |
600 | { $$ = ast_structop_anonymous ($1, $3); } | |
601 | ; | |
602 | ||
603 | idx_expr: | |
604 | expr '[' expr ']' | |
605 | { $$ = ast_operation (BINOP_SUBSCRIPT, $1, $3); } | |
606 | ; | |
607 | ||
608 | unop_expr: | |
609 | '+' expr %prec UNARY | |
610 | { $$ = ast_unary (UNOP_PLUS, $2); } | |
611 | ||
612 | | '-' expr %prec UNARY | |
613 | { $$ = ast_unary (UNOP_NEG, $2); } | |
614 | ||
615 | | '!' expr %prec UNARY | |
616 | { | |
617 | /* Note that we provide a Rust-specific evaluator | |
618 | override for UNOP_COMPLEMENT, so it can do the | |
619 | right thing for both bool and integral | |
620 | values. */ | |
621 | $$ = ast_unary (UNOP_COMPLEMENT, $2); | |
622 | } | |
623 | ||
624 | | '*' expr %prec UNARY | |
625 | { $$ = ast_unary (UNOP_IND, $2); } | |
626 | ||
627 | | '&' expr %prec UNARY | |
628 | { $$ = ast_unary (UNOP_ADDR, $2); } | |
629 | ||
630 | | '&' KW_MUT expr %prec UNARY | |
631 | { $$ = ast_unary (UNOP_ADDR, $3); } | |
cdf5a07c MG |
632 | | KW_SIZEOF '(' expr ')' %prec UNARY |
633 | { $$ = ast_unary (UNOP_SIZEOF, $3); } | |
c44af4eb TT |
634 | ; |
635 | ||
636 | binop_expr: | |
637 | binop_expr_expr | |
638 | | type_cast_expr | |
639 | | assignment_expr | |
640 | | compound_assignment_expr | |
641 | ; | |
642 | ||
643 | binop_expr_expr: | |
644 | expr '*' expr | |
645 | { $$ = ast_operation (BINOP_MUL, $1, $3); } | |
646 | ||
647 | | expr '@' expr | |
648 | { $$ = ast_operation (BINOP_REPEAT, $1, $3); } | |
649 | ||
650 | | expr '/' expr | |
651 | { $$ = ast_operation (BINOP_DIV, $1, $3); } | |
652 | ||
653 | | expr '%' expr | |
654 | { $$ = ast_operation (BINOP_REM, $1, $3); } | |
655 | ||
656 | | expr '<' expr | |
657 | { $$ = ast_operation (BINOP_LESS, $1, $3); } | |
658 | ||
659 | | expr '>' expr | |
660 | { $$ = ast_operation (BINOP_GTR, $1, $3); } | |
661 | ||
662 | | expr '&' expr | |
663 | { $$ = ast_operation (BINOP_BITWISE_AND, $1, $3); } | |
664 | ||
665 | | expr '|' expr | |
666 | { $$ = ast_operation (BINOP_BITWISE_IOR, $1, $3); } | |
667 | ||
668 | | expr '^' expr | |
669 | { $$ = ast_operation (BINOP_BITWISE_XOR, $1, $3); } | |
670 | ||
671 | | expr '+' expr | |
672 | { $$ = ast_operation (BINOP_ADD, $1, $3); } | |
673 | ||
674 | | expr '-' expr | |
675 | { $$ = ast_operation (BINOP_SUB, $1, $3); } | |
676 | ||
677 | | expr OROR expr | |
678 | { $$ = ast_operation (BINOP_LOGICAL_OR, $1, $3); } | |
679 | ||
680 | | expr ANDAND expr | |
681 | { $$ = ast_operation (BINOP_LOGICAL_AND, $1, $3); } | |
682 | ||
683 | | expr EQEQ expr | |
684 | { $$ = ast_operation (BINOP_EQUAL, $1, $3); } | |
685 | ||
686 | | expr NOTEQ expr | |
687 | { $$ = ast_operation (BINOP_NOTEQUAL, $1, $3); } | |
688 | ||
689 | | expr LTEQ expr | |
690 | { $$ = ast_operation (BINOP_LEQ, $1, $3); } | |
691 | ||
692 | | expr GTEQ expr | |
693 | { $$ = ast_operation (BINOP_GEQ, $1, $3); } | |
694 | ||
695 | | expr LSH expr | |
696 | { $$ = ast_operation (BINOP_LSH, $1, $3); } | |
697 | ||
698 | | expr RSH expr | |
699 | { $$ = ast_operation (BINOP_RSH, $1, $3); } | |
700 | ; | |
701 | ||
702 | type_cast_expr: | |
703 | expr KW_AS type | |
704 | { $$ = ast_cast ($1, $3); } | |
705 | ; | |
706 | ||
707 | assignment_expr: | |
708 | expr '=' expr | |
709 | { $$ = ast_operation (BINOP_ASSIGN, $1, $3); } | |
710 | ; | |
711 | ||
712 | compound_assignment_expr: | |
713 | expr COMPOUND_ASSIGN expr | |
714 | { $$ = ast_compound_assignment ($2, $1, $3); } | |
715 | ||
716 | ; | |
717 | ||
718 | paren_expr: | |
719 | '(' expr ')' | |
720 | { $$ = $2; } | |
721 | ; | |
722 | ||
723 | expr_list: | |
724 | expr | |
725 | { | |
3232fabd TT |
726 | $$ = current_parser->new_op_vector (); |
727 | $$->push_back ($1); | |
c44af4eb TT |
728 | } |
729 | | expr_list ',' expr | |
730 | { | |
3232fabd | 731 | $1->push_back ($3); |
c44af4eb TT |
732 | $$ = $1; |
733 | } | |
734 | ; | |
735 | ||
736 | maybe_expr_list: | |
737 | /* %empty */ | |
738 | { | |
739 | /* The result can't be NULL. */ | |
3232fabd | 740 | $$ = current_parser->new_op_vector (); |
c44af4eb TT |
741 | } |
742 | | expr_list | |
743 | { $$ = $1; } | |
744 | ; | |
745 | ||
746 | paren_expr_list: | |
747 | '(' | |
748 | maybe_expr_list | |
749 | ')' | |
750 | { $$ = $2; } | |
751 | ; | |
752 | ||
753 | call_expr: | |
754 | expr paren_expr_list | |
755 | { $$ = ast_call_ish (OP_FUNCALL, $1, $2); } | |
756 | ; | |
757 | ||
758 | maybe_self_path: | |
759 | /* %empty */ | |
760 | | KW_SELF COLONCOLON | |
761 | ; | |
762 | ||
763 | super_path: | |
764 | KW_SUPER COLONCOLON | |
765 | { $$ = 1; } | |
766 | | super_path KW_SUPER COLONCOLON | |
767 | { $$ = $1 + 1; } | |
768 | ; | |
769 | ||
770 | path_expr: | |
771 | path_for_expr | |
772 | { $$ = $1; } | |
773 | | GDBVAR | |
774 | { $$ = ast_path ($1, NULL); } | |
775 | | KW_SELF | |
776 | { $$ = ast_path (make_stoken ("self"), NULL); } | |
777 | ; | |
778 | ||
779 | path_for_expr: | |
780 | identifier_path_for_expr | |
781 | | KW_SELF COLONCOLON identifier_path_for_expr | |
782 | { $$ = super_name ($3, 0); } | |
783 | | maybe_self_path super_path identifier_path_for_expr | |
784 | { $$ = super_name ($3, $2); } | |
785 | | COLONCOLON identifier_path_for_expr | |
786 | { $$ = crate_name ($2); } | |
787 | | KW_EXTERN identifier_path_for_expr | |
788 | { | |
789 | /* This is a gdb extension to make it possible to | |
790 | refer to items in other crates. It just bypasses | |
791 | adding the current crate to the front of the | |
792 | name. */ | |
793 | $$ = ast_path (rust_concat3 ("::", $2->left.sval.ptr, NULL), | |
794 | $2->right.params); | |
795 | } | |
796 | ; | |
797 | ||
798 | identifier_path_for_expr: | |
799 | IDENT | |
800 | { $$ = ast_path ($1, NULL); } | |
801 | | identifier_path_for_expr COLONCOLON IDENT | |
802 | { | |
803 | $$ = ast_path (rust_concat3 ($1->left.sval.ptr, "::", | |
804 | $3.ptr), | |
805 | NULL); | |
806 | } | |
807 | | identifier_path_for_expr COLONCOLON '<' type_list '>' | |
808 | { $$ = ast_path ($1->left.sval, $4); } | |
809 | | identifier_path_for_expr COLONCOLON '<' type_list RSH | |
810 | { | |
811 | $$ = ast_path ($1->left.sval, $4); | |
812 | rust_push_back ('>'); | |
813 | } | |
814 | ; | |
815 | ||
816 | path_for_type: | |
817 | identifier_path_for_type | |
818 | | KW_SELF COLONCOLON identifier_path_for_type | |
819 | { $$ = super_name ($3, 0); } | |
820 | | maybe_self_path super_path identifier_path_for_type | |
821 | { $$ = super_name ($3, $2); } | |
822 | | COLONCOLON identifier_path_for_type | |
823 | { $$ = crate_name ($2); } | |
824 | | KW_EXTERN identifier_path_for_type | |
825 | { | |
826 | /* This is a gdb extension to make it possible to | |
827 | refer to items in other crates. It just bypasses | |
828 | adding the current crate to the front of the | |
829 | name. */ | |
830 | $$ = ast_path (rust_concat3 ("::", $2->left.sval.ptr, NULL), | |
831 | $2->right.params); | |
832 | } | |
833 | ; | |
834 | ||
835 | just_identifiers_for_type: | |
836 | IDENT | |
837 | { $$ = ast_path ($1, NULL); } | |
838 | | just_identifiers_for_type COLONCOLON IDENT | |
839 | { | |
840 | $$ = ast_path (rust_concat3 ($1->left.sval.ptr, "::", | |
841 | $3.ptr), | |
842 | NULL); | |
843 | } | |
844 | ; | |
845 | ||
846 | identifier_path_for_type: | |
847 | just_identifiers_for_type | |
848 | | just_identifiers_for_type '<' type_list '>' | |
849 | { $$ = ast_path ($1->left.sval, $3); } | |
850 | | just_identifiers_for_type '<' type_list RSH | |
851 | { | |
852 | $$ = ast_path ($1->left.sval, $3); | |
853 | rust_push_back ('>'); | |
854 | } | |
855 | ; | |
856 | ||
857 | type: | |
858 | path_for_type | |
859 | | '[' type ';' INTEGER ']' | |
860 | { $$ = ast_array_type ($2, $4); } | |
861 | | '[' type ';' DECIMAL_INTEGER ']' | |
862 | { $$ = ast_array_type ($2, $4); } | |
863 | | '&' '[' type ']' | |
864 | { $$ = ast_slice_type ($3); } | |
865 | | '&' type | |
866 | { $$ = ast_reference_type ($2); } | |
867 | | '*' KW_MUT type | |
868 | { $$ = ast_pointer_type ($3, 1); } | |
869 | | '*' KW_CONST type | |
870 | { $$ = ast_pointer_type ($3, 0); } | |
871 | | KW_FN '(' maybe_type_list ')' ARROW type | |
872 | { $$ = ast_function_type ($6, $3); } | |
873 | | '(' maybe_type_list ')' | |
874 | { $$ = ast_tuple_type ($2); } | |
875 | ; | |
876 | ||
877 | maybe_type_list: | |
878 | /* %empty */ | |
879 | { $$ = NULL; } | |
880 | | type_list | |
881 | { $$ = $1; } | |
882 | ; | |
883 | ||
884 | type_list: | |
885 | type | |
886 | { | |
3232fabd TT |
887 | rust_op_vector *result = current_parser->new_op_vector (); |
888 | result->push_back ($1); | |
c44af4eb TT |
889 | $$ = result; |
890 | } | |
891 | | type_list ',' type | |
892 | { | |
3232fabd | 893 | $1->push_back ($3); |
c44af4eb TT |
894 | $$ = $1; |
895 | } | |
896 | ; | |
897 | ||
898 | %% | |
899 | ||
900 | /* A struct of this type is used to describe a token. */ | |
901 | ||
902 | struct token_info | |
903 | { | |
904 | const char *name; | |
905 | int value; | |
906 | enum exp_opcode opcode; | |
907 | }; | |
908 | ||
909 | /* Identifier tokens. */ | |
910 | ||
911 | static const struct token_info identifier_tokens[] = | |
912 | { | |
913 | { "as", KW_AS, OP_NULL }, | |
914 | { "false", KW_FALSE, OP_NULL }, | |
915 | { "if", 0, OP_NULL }, | |
916 | { "mut", KW_MUT, OP_NULL }, | |
917 | { "const", KW_CONST, OP_NULL }, | |
918 | { "self", KW_SELF, OP_NULL }, | |
919 | { "super", KW_SUPER, OP_NULL }, | |
920 | { "true", KW_TRUE, OP_NULL }, | |
921 | { "extern", KW_EXTERN, OP_NULL }, | |
922 | { "fn", KW_FN, OP_NULL }, | |
cdf5a07c | 923 | { "sizeof", KW_SIZEOF, OP_NULL }, |
c44af4eb TT |
924 | }; |
925 | ||
926 | /* Operator tokens, sorted longest first. */ | |
927 | ||
928 | static const struct token_info operator_tokens[] = | |
929 | { | |
930 | { ">>=", COMPOUND_ASSIGN, BINOP_RSH }, | |
931 | { "<<=", COMPOUND_ASSIGN, BINOP_LSH }, | |
932 | ||
933 | { "<<", LSH, OP_NULL }, | |
934 | { ">>", RSH, OP_NULL }, | |
935 | { "&&", ANDAND, OP_NULL }, | |
936 | { "||", OROR, OP_NULL }, | |
937 | { "==", EQEQ, OP_NULL }, | |
938 | { "!=", NOTEQ, OP_NULL }, | |
939 | { "<=", LTEQ, OP_NULL }, | |
940 | { ">=", GTEQ, OP_NULL }, | |
941 | { "+=", COMPOUND_ASSIGN, BINOP_ADD }, | |
942 | { "-=", COMPOUND_ASSIGN, BINOP_SUB }, | |
943 | { "*=", COMPOUND_ASSIGN, BINOP_MUL }, | |
944 | { "/=", COMPOUND_ASSIGN, BINOP_DIV }, | |
945 | { "%=", COMPOUND_ASSIGN, BINOP_REM }, | |
946 | { "&=", COMPOUND_ASSIGN, BINOP_BITWISE_AND }, | |
947 | { "|=", COMPOUND_ASSIGN, BINOP_BITWISE_IOR }, | |
948 | { "^=", COMPOUND_ASSIGN, BINOP_BITWISE_XOR }, | |
949 | ||
950 | { "::", COLONCOLON, OP_NULL }, | |
951 | { "..", DOTDOT, OP_NULL }, | |
952 | { "->", ARROW, OP_NULL } | |
953 | }; | |
954 | ||
955 | /* Helper function to copy to the name obstack. */ | |
956 | ||
957 | static const char * | |
958 | rust_copy_name (const char *name, int len) | |
959 | { | |
3232fabd | 960 | return (const char *) obstack_copy0 (work_obstack, name, len); |
c44af4eb TT |
961 | } |
962 | ||
963 | /* Helper function to make an stoken from a C string. */ | |
964 | ||
965 | static struct stoken | |
966 | make_stoken (const char *p) | |
967 | { | |
968 | struct stoken result; | |
969 | ||
970 | result.ptr = p; | |
971 | result.length = strlen (result.ptr); | |
972 | return result; | |
973 | } | |
974 | ||
975 | /* Helper function to concatenate three strings on the name | |
976 | obstack. */ | |
977 | ||
978 | static struct stoken | |
979 | rust_concat3 (const char *s1, const char *s2, const char *s3) | |
980 | { | |
3232fabd | 981 | return make_stoken (obconcat (work_obstack, s1, s2, s3, (char *) NULL)); |
c44af4eb TT |
982 | } |
983 | ||
984 | /* Return an AST node referring to NAME, but relative to the crate's | |
985 | name. */ | |
986 | ||
987 | static const struct rust_op * | |
988 | crate_name (const struct rust_op *name) | |
989 | { | |
03c85b11 | 990 | std::string crate = rust_crate_for_block (expression_context_block); |
c44af4eb TT |
991 | struct stoken result; |
992 | ||
993 | gdb_assert (name->opcode == OP_VAR_VALUE); | |
994 | ||
03c85b11 | 995 | if (crate.empty ()) |
c44af4eb | 996 | error (_("Could not find crate for current location")); |
3232fabd | 997 | result = make_stoken (obconcat (work_obstack, "::", crate.c_str (), "::", |
c44af4eb | 998 | name->left.sval.ptr, (char *) NULL)); |
c44af4eb TT |
999 | |
1000 | return ast_path (result, name->right.params); | |
1001 | } | |
1002 | ||
1003 | /* Create an AST node referring to a "super::" qualified name. IDENT | |
1004 | is the base name and N_SUPERS is how many "super::"s were | |
1005 | provided. N_SUPERS can be zero. */ | |
1006 | ||
1007 | static const struct rust_op * | |
1008 | super_name (const struct rust_op *ident, unsigned int n_supers) | |
1009 | { | |
1010 | const char *scope = block_scope (expression_context_block); | |
1011 | int offset; | |
1012 | ||
1013 | gdb_assert (ident->opcode == OP_VAR_VALUE); | |
1014 | ||
1015 | if (scope[0] == '\0') | |
1016 | error (_("Couldn't find namespace scope for self::")); | |
1017 | ||
1018 | if (n_supers > 0) | |
1019 | { | |
1020 | int i; | |
1021 | int len; | |
8001f118 | 1022 | std::vector<int> offsets; |
78cc6c2d | 1023 | unsigned int current_len; |
c44af4eb | 1024 | |
c44af4eb | 1025 | current_len = cp_find_first_component (scope); |
c44af4eb TT |
1026 | while (scope[current_len] != '\0') |
1027 | { | |
8001f118 | 1028 | offsets.push_back (current_len); |
c44af4eb | 1029 | gdb_assert (scope[current_len] == ':'); |
c44af4eb TT |
1030 | /* The "::". */ |
1031 | current_len += 2; | |
1032 | current_len += cp_find_first_component (scope | |
1033 | + current_len); | |
1034 | } | |
1035 | ||
8001f118 | 1036 | len = offsets.size (); |
c44af4eb TT |
1037 | if (n_supers >= len) |
1038 | error (_("Too many super:: uses from '%s'"), scope); | |
1039 | ||
8001f118 | 1040 | offset = offsets[len - n_supers]; |
c44af4eb TT |
1041 | } |
1042 | else | |
1043 | offset = strlen (scope); | |
1044 | ||
3232fabd TT |
1045 | obstack_grow (work_obstack, "::", 2); |
1046 | obstack_grow (work_obstack, scope, offset); | |
1047 | obstack_grow (work_obstack, "::", 2); | |
1048 | obstack_grow0 (work_obstack, ident->left.sval.ptr, ident->left.sval.length); | |
c44af4eb | 1049 | |
3232fabd | 1050 | return ast_path (make_stoken ((const char *) obstack_finish (work_obstack)), |
c44af4eb TT |
1051 | ident->right.params); |
1052 | } | |
1053 | ||
1054 | /* A helper that updates innermost_block as appropriate. */ | |
1055 | ||
1056 | static void | |
1057 | update_innermost_block (struct block_symbol sym) | |
1058 | { | |
1059 | if (symbol_read_needs_frame (sym.symbol) | |
1060 | && (innermost_block == NULL | |
1061 | || contained_in (sym.block, innermost_block))) | |
1062 | innermost_block = sym.block; | |
1063 | } | |
1064 | ||
1065 | /* A helper to look up a Rust type, or fail. This only works for | |
1066 | types defined by rust_language_arch_info. */ | |
1067 | ||
1068 | static struct type * | |
1069 | rust_type (const char *name) | |
1070 | { | |
1071 | struct type *type; | |
1072 | ||
1073 | /* When unit testing, we don't bother checking the types, so avoid a | |
1074 | possibly-failing lookup here. */ | |
1075 | if (unit_testing) | |
1076 | return NULL; | |
1077 | ||
3232fabd TT |
1078 | type = language_lookup_primitive_type (current_parser->language (), |
1079 | current_parser->arch (), | |
c44af4eb TT |
1080 | name); |
1081 | if (type == NULL) | |
1082 | error (_("Could not find Rust type %s"), name); | |
1083 | return type; | |
1084 | } | |
1085 | ||
1086 | /* Lex a hex number with at least MIN digits and at most MAX | |
1087 | digits. */ | |
1088 | ||
1089 | static uint32_t | |
1090 | lex_hex (int min, int max) | |
1091 | { | |
1092 | uint32_t result = 0; | |
1093 | int len = 0; | |
1094 | /* We only want to stop at MAX if we're lexing a byte escape. */ | |
1095 | int check_max = min == max; | |
1096 | ||
1097 | while ((check_max ? len <= max : 1) | |
1098 | && ((lexptr[0] >= 'a' && lexptr[0] <= 'f') | |
1099 | || (lexptr[0] >= 'A' && lexptr[0] <= 'F') | |
1100 | || (lexptr[0] >= '0' && lexptr[0] <= '9'))) | |
1101 | { | |
1102 | result *= 16; | |
1103 | if (lexptr[0] >= 'a' && lexptr[0] <= 'f') | |
1104 | result = result + 10 + lexptr[0] - 'a'; | |
1105 | else if (lexptr[0] >= 'A' && lexptr[0] <= 'F') | |
1106 | result = result + 10 + lexptr[0] - 'A'; | |
1107 | else | |
1108 | result = result + lexptr[0] - '0'; | |
1109 | ++lexptr; | |
1110 | ++len; | |
1111 | } | |
1112 | ||
1113 | if (len < min) | |
1114 | error (_("Not enough hex digits seen")); | |
1115 | if (len > max) | |
1116 | { | |
1117 | gdb_assert (min != max); | |
1118 | error (_("Overlong hex escape")); | |
1119 | } | |
1120 | ||
1121 | return result; | |
1122 | } | |
1123 | ||
1124 | /* Lex an escape. IS_BYTE is true if we're lexing a byte escape; | |
1125 | otherwise we're lexing a character escape. */ | |
1126 | ||
1127 | static uint32_t | |
1128 | lex_escape (int is_byte) | |
1129 | { | |
1130 | uint32_t result; | |
1131 | ||
1132 | gdb_assert (lexptr[0] == '\\'); | |
1133 | ++lexptr; | |
1134 | switch (lexptr[0]) | |
1135 | { | |
1136 | case 'x': | |
1137 | ++lexptr; | |
1138 | result = lex_hex (2, 2); | |
1139 | break; | |
1140 | ||
1141 | case 'u': | |
1142 | if (is_byte) | |
1143 | error (_("Unicode escape in byte literal")); | |
1144 | ++lexptr; | |
1145 | if (lexptr[0] != '{') | |
1146 | error (_("Missing '{' in Unicode escape")); | |
1147 | ++lexptr; | |
1148 | result = lex_hex (1, 6); | |
1149 | /* Could do range checks here. */ | |
1150 | if (lexptr[0] != '}') | |
1151 | error (_("Missing '}' in Unicode escape")); | |
1152 | ++lexptr; | |
1153 | break; | |
1154 | ||
1155 | case 'n': | |
1156 | result = '\n'; | |
1157 | ++lexptr; | |
1158 | break; | |
1159 | case 'r': | |
1160 | result = '\r'; | |
1161 | ++lexptr; | |
1162 | break; | |
1163 | case 't': | |
1164 | result = '\t'; | |
1165 | ++lexptr; | |
1166 | break; | |
1167 | case '\\': | |
1168 | result = '\\'; | |
1169 | ++lexptr; | |
1170 | break; | |
1171 | case '0': | |
1172 | result = '\0'; | |
1173 | ++lexptr; | |
1174 | break; | |
1175 | case '\'': | |
1176 | result = '\''; | |
1177 | ++lexptr; | |
1178 | break; | |
1179 | case '"': | |
1180 | result = '"'; | |
1181 | ++lexptr; | |
1182 | break; | |
1183 | ||
1184 | default: | |
1185 | error (_("Invalid escape \\%c in literal"), lexptr[0]); | |
1186 | } | |
1187 | ||
1188 | return result; | |
1189 | } | |
1190 | ||
1191 | /* Lex a character constant. */ | |
1192 | ||
1193 | static int | |
1194 | lex_character (void) | |
1195 | { | |
1196 | int is_byte = 0; | |
1197 | uint32_t value; | |
1198 | ||
1199 | if (lexptr[0] == 'b') | |
1200 | { | |
1201 | is_byte = 1; | |
1202 | ++lexptr; | |
1203 | } | |
1204 | gdb_assert (lexptr[0] == '\''); | |
1205 | ++lexptr; | |
1206 | /* This should handle UTF-8 here. */ | |
1207 | if (lexptr[0] == '\\') | |
1208 | value = lex_escape (is_byte); | |
1209 | else | |
1210 | { | |
1211 | value = lexptr[0] & 0xff; | |
1212 | ++lexptr; | |
1213 | } | |
1214 | ||
1215 | if (lexptr[0] != '\'') | |
1216 | error (_("Unterminated character literal")); | |
1217 | ++lexptr; | |
1218 | ||
1219 | rustyylval.typed_val_int.val = value; | |
1220 | rustyylval.typed_val_int.type = rust_type (is_byte ? "u8" : "char"); | |
1221 | ||
1222 | return INTEGER; | |
1223 | } | |
1224 | ||
1225 | /* Return the offset of the double quote if STR looks like the start | |
1226 | of a raw string, or 0 if STR does not start a raw string. */ | |
1227 | ||
1228 | static int | |
1229 | starts_raw_string (const char *str) | |
1230 | { | |
1231 | const char *save = str; | |
1232 | ||
1233 | if (str[0] != 'r') | |
1234 | return 0; | |
1235 | ++str; | |
1236 | while (str[0] == '#') | |
1237 | ++str; | |
1238 | if (str[0] == '"') | |
1239 | return str - save; | |
1240 | return 0; | |
1241 | } | |
1242 | ||
1243 | /* Return true if STR looks like the end of a raw string that had N | |
1244 | hashes at the start. */ | |
1245 | ||
65c40c95 | 1246 | static bool |
c44af4eb TT |
1247 | ends_raw_string (const char *str, int n) |
1248 | { | |
1249 | int i; | |
1250 | ||
1251 | gdb_assert (str[0] == '"'); | |
1252 | for (i = 0; i < n; ++i) | |
1253 | if (str[i + 1] != '#') | |
65c40c95 TT |
1254 | return false; |
1255 | return true; | |
c44af4eb TT |
1256 | } |
1257 | ||
1258 | /* Lex a string constant. */ | |
1259 | ||
1260 | static int | |
1261 | lex_string (void) | |
1262 | { | |
1263 | int is_byte = lexptr[0] == 'b'; | |
1264 | int raw_length; | |
1265 | int len_in_chars = 0; | |
1266 | ||
1267 | if (is_byte) | |
1268 | ++lexptr; | |
1269 | raw_length = starts_raw_string (lexptr); | |
1270 | lexptr += raw_length; | |
1271 | gdb_assert (lexptr[0] == '"'); | |
1272 | ++lexptr; | |
1273 | ||
1274 | while (1) | |
1275 | { | |
1276 | uint32_t value; | |
1277 | ||
1278 | if (raw_length > 0) | |
1279 | { | |
1280 | if (lexptr[0] == '"' && ends_raw_string (lexptr, raw_length - 1)) | |
1281 | { | |
1282 | /* Exit with lexptr pointing after the final "#". */ | |
1283 | lexptr += raw_length; | |
1284 | break; | |
1285 | } | |
1286 | else if (lexptr[0] == '\0') | |
1287 | error (_("Unexpected EOF in string")); | |
1288 | ||
1289 | value = lexptr[0] & 0xff; | |
1290 | if (is_byte && value > 127) | |
1291 | error (_("Non-ASCII value in raw byte string")); | |
3232fabd | 1292 | obstack_1grow (work_obstack, value); |
c44af4eb TT |
1293 | |
1294 | ++lexptr; | |
1295 | } | |
1296 | else if (lexptr[0] == '"') | |
1297 | { | |
1298 | /* Make sure to skip the quote. */ | |
1299 | ++lexptr; | |
1300 | break; | |
1301 | } | |
1302 | else if (lexptr[0] == '\\') | |
1303 | { | |
1304 | value = lex_escape (is_byte); | |
1305 | ||
1306 | if (is_byte) | |
3232fabd | 1307 | obstack_1grow (work_obstack, value); |
c44af4eb TT |
1308 | else |
1309 | convert_between_encodings ("UTF-32", "UTF-8", (gdb_byte *) &value, | |
1310 | sizeof (value), sizeof (value), | |
3232fabd | 1311 | work_obstack, translit_none); |
c44af4eb TT |
1312 | } |
1313 | else if (lexptr[0] == '\0') | |
1314 | error (_("Unexpected EOF in string")); | |
1315 | else | |
1316 | { | |
1317 | value = lexptr[0] & 0xff; | |
1318 | if (is_byte && value > 127) | |
1319 | error (_("Non-ASCII value in byte string")); | |
3232fabd | 1320 | obstack_1grow (work_obstack, value); |
c44af4eb TT |
1321 | ++lexptr; |
1322 | } | |
1323 | } | |
1324 | ||
3232fabd TT |
1325 | rustyylval.sval.length = obstack_object_size (work_obstack); |
1326 | rustyylval.sval.ptr = (const char *) obstack_finish (work_obstack); | |
c44af4eb TT |
1327 | return is_byte ? BYTESTRING : STRING; |
1328 | } | |
1329 | ||
1330 | /* Return true if STRING starts with whitespace followed by a digit. */ | |
1331 | ||
65c40c95 | 1332 | static bool |
c44af4eb TT |
1333 | space_then_number (const char *string) |
1334 | { | |
1335 | const char *p = string; | |
1336 | ||
1337 | while (p[0] == ' ' || p[0] == '\t') | |
1338 | ++p; | |
1339 | if (p == string) | |
65c40c95 | 1340 | return false; |
c44af4eb TT |
1341 | |
1342 | return *p >= '0' && *p <= '9'; | |
1343 | } | |
1344 | ||
1345 | /* Return true if C can start an identifier. */ | |
1346 | ||
65c40c95 | 1347 | static bool |
c44af4eb TT |
1348 | rust_identifier_start_p (char c) |
1349 | { | |
1350 | return ((c >= 'a' && c <= 'z') | |
1351 | || (c >= 'A' && c <= 'Z') | |
1352 | || c == '_' | |
1353 | || c == '$'); | |
1354 | } | |
1355 | ||
1356 | /* Lex an identifier. */ | |
1357 | ||
1358 | static int | |
1359 | lex_identifier (void) | |
1360 | { | |
1361 | const char *start = lexptr; | |
1362 | unsigned int length; | |
1363 | const struct token_info *token; | |
1364 | int i; | |
1365 | int is_gdb_var = lexptr[0] == '$'; | |
1366 | ||
1367 | gdb_assert (rust_identifier_start_p (lexptr[0])); | |
1368 | ||
1369 | ++lexptr; | |
1370 | ||
1371 | /* For the time being this doesn't handle Unicode rules. Non-ASCII | |
1372 | identifiers are gated anyway. */ | |
1373 | while ((lexptr[0] >= 'a' && lexptr[0] <= 'z') | |
1374 | || (lexptr[0] >= 'A' && lexptr[0] <= 'Z') | |
1375 | || lexptr[0] == '_' | |
1376 | || (is_gdb_var && lexptr[0] == '$') | |
1377 | || (lexptr[0] >= '0' && lexptr[0] <= '9')) | |
1378 | ++lexptr; | |
1379 | ||
1380 | ||
1381 | length = lexptr - start; | |
1382 | token = NULL; | |
1383 | for (i = 0; i < ARRAY_SIZE (identifier_tokens); ++i) | |
1384 | { | |
1385 | if (length == strlen (identifier_tokens[i].name) | |
1386 | && strncmp (identifier_tokens[i].name, start, length) == 0) | |
1387 | { | |
1388 | token = &identifier_tokens[i]; | |
1389 | break; | |
1390 | } | |
1391 | } | |
1392 | ||
1393 | if (token != NULL) | |
1394 | { | |
1395 | if (token->value == 0) | |
1396 | { | |
1397 | /* Leave the terminating token alone. */ | |
1398 | lexptr = start; | |
1399 | return 0; | |
1400 | } | |
1401 | } | |
1402 | else if (token == NULL | |
1403 | && (strncmp (start, "thread", length) == 0 | |
1404 | || strncmp (start, "task", length) == 0) | |
1405 | && space_then_number (lexptr)) | |
1406 | { | |
1407 | /* "task" or "thread" followed by a number terminates the | |
1408 | parse, per gdb rules. */ | |
1409 | lexptr = start; | |
1410 | return 0; | |
1411 | } | |
1412 | ||
1413 | if (token == NULL || (parse_completion && lexptr[0] == '\0')) | |
1414 | rustyylval.sval = make_stoken (rust_copy_name (start, length)); | |
1415 | ||
1416 | if (parse_completion && lexptr[0] == '\0') | |
1417 | { | |
1418 | /* Prevent rustyylex from returning two COMPLETE tokens. */ | |
1419 | prev_lexptr = lexptr; | |
1420 | return COMPLETE; | |
1421 | } | |
1422 | ||
1423 | if (token != NULL) | |
1424 | return token->value; | |
1425 | if (is_gdb_var) | |
1426 | return GDBVAR; | |
1427 | return IDENT; | |
1428 | } | |
1429 | ||
1430 | /* Lex an operator. */ | |
1431 | ||
1432 | static int | |
1433 | lex_operator (void) | |
1434 | { | |
1435 | const struct token_info *token = NULL; | |
1436 | int i; | |
1437 | ||
1438 | for (i = 0; i < ARRAY_SIZE (operator_tokens); ++i) | |
1439 | { | |
1440 | if (strncmp (operator_tokens[i].name, lexptr, | |
1441 | strlen (operator_tokens[i].name)) == 0) | |
1442 | { | |
1443 | lexptr += strlen (operator_tokens[i].name); | |
1444 | token = &operator_tokens[i]; | |
1445 | break; | |
1446 | } | |
1447 | } | |
1448 | ||
1449 | if (token != NULL) | |
1450 | { | |
1451 | rustyylval.opcode = token->opcode; | |
1452 | return token->value; | |
1453 | } | |
1454 | ||
1455 | return *lexptr++; | |
1456 | } | |
1457 | ||
1458 | /* Lex a number. */ | |
1459 | ||
1460 | static int | |
1461 | lex_number (void) | |
1462 | { | |
1463 | regmatch_t subexps[NUM_SUBEXPRESSIONS]; | |
1464 | int match; | |
1465 | int is_integer = 0; | |
1466 | int could_be_decimal = 1; | |
347dc102 | 1467 | int implicit_i32 = 0; |
8001f118 | 1468 | const char *type_name = NULL; |
c44af4eb TT |
1469 | struct type *type; |
1470 | int end_index; | |
1471 | int type_index = -1; | |
8001f118 | 1472 | int i; |
c44af4eb TT |
1473 | |
1474 | match = regexec (&number_regex, lexptr, ARRAY_SIZE (subexps), subexps, 0); | |
1475 | /* Failure means the regexp is broken. */ | |
1476 | gdb_assert (match == 0); | |
1477 | ||
1478 | if (subexps[INT_TEXT].rm_so != -1) | |
1479 | { | |
1480 | /* Integer part matched. */ | |
1481 | is_integer = 1; | |
1482 | end_index = subexps[INT_TEXT].rm_eo; | |
1483 | if (subexps[INT_TYPE].rm_so == -1) | |
347dc102 TT |
1484 | { |
1485 | type_name = "i32"; | |
1486 | implicit_i32 = 1; | |
1487 | } | |
c44af4eb TT |
1488 | else |
1489 | { | |
1490 | type_index = INT_TYPE; | |
1491 | could_be_decimal = 0; | |
1492 | } | |
1493 | } | |
1494 | else if (subexps[FLOAT_TYPE1].rm_so != -1) | |
1495 | { | |
1496 | /* Found floating point type suffix. */ | |
1497 | end_index = subexps[FLOAT_TYPE1].rm_so; | |
1498 | type_index = FLOAT_TYPE1; | |
1499 | } | |
1500 | else if (subexps[FLOAT_TYPE2].rm_so != -1) | |
1501 | { | |
1502 | /* Found floating point type suffix. */ | |
1503 | end_index = subexps[FLOAT_TYPE2].rm_so; | |
1504 | type_index = FLOAT_TYPE2; | |
1505 | } | |
1506 | else | |
1507 | { | |
1508 | /* Any other floating point match. */ | |
1509 | end_index = subexps[0].rm_eo; | |
1510 | type_name = "f64"; | |
1511 | } | |
1512 | ||
1513 | /* We need a special case if the final character is ".". In this | |
1514 | case we might need to parse an integer. For example, "23.f()" is | |
1515 | a request for a trait method call, not a syntax error involving | |
1516 | the floating point number "23.". */ | |
1517 | gdb_assert (subexps[0].rm_eo > 0); | |
1518 | if (lexptr[subexps[0].rm_eo - 1] == '.') | |
1519 | { | |
1520 | const char *next = skip_spaces_const (&lexptr[subexps[0].rm_eo]); | |
1521 | ||
1522 | if (rust_identifier_start_p (*next) || *next == '.') | |
1523 | { | |
1524 | --subexps[0].rm_eo; | |
1525 | is_integer = 1; | |
1526 | end_index = subexps[0].rm_eo; | |
1527 | type_name = "i32"; | |
1528 | could_be_decimal = 1; | |
347dc102 | 1529 | implicit_i32 = 1; |
c44af4eb TT |
1530 | } |
1531 | } | |
1532 | ||
1533 | /* Compute the type name if we haven't already. */ | |
8001f118 | 1534 | std::string type_name_holder; |
c44af4eb TT |
1535 | if (type_name == NULL) |
1536 | { | |
1537 | gdb_assert (type_index != -1); | |
8001f118 TT |
1538 | type_name_holder = std::string (lexptr + subexps[type_index].rm_so, |
1539 | (subexps[type_index].rm_eo | |
1540 | - subexps[type_index].rm_so)); | |
1541 | type_name = type_name_holder.c_str (); | |
c44af4eb TT |
1542 | } |
1543 | ||
1544 | /* Look up the type. */ | |
1545 | type = rust_type (type_name); | |
1546 | ||
1547 | /* Copy the text of the number and remove the "_"s. */ | |
8001f118 TT |
1548 | std::string number; |
1549 | for (i = 0; i < end_index && lexptr[i]; ++i) | |
c44af4eb | 1550 | { |
8001f118 | 1551 | if (lexptr[i] == '_') |
c44af4eb TT |
1552 | could_be_decimal = 0; |
1553 | else | |
8001f118 | 1554 | number.push_back (lexptr[i]); |
c44af4eb | 1555 | } |
c44af4eb TT |
1556 | |
1557 | /* Advance past the match. */ | |
1558 | lexptr += subexps[0].rm_eo; | |
1559 | ||
1560 | /* Parse the number. */ | |
1561 | if (is_integer) | |
1562 | { | |
347dc102 | 1563 | uint64_t value; |
c44af4eb | 1564 | int radix = 10; |
8001f118 TT |
1565 | int offset = 0; |
1566 | ||
c44af4eb TT |
1567 | if (number[0] == '0') |
1568 | { | |
1569 | if (number[1] == 'x') | |
1570 | radix = 16; | |
1571 | else if (number[1] == 'o') | |
1572 | radix = 8; | |
1573 | else if (number[1] == 'b') | |
1574 | radix = 2; | |
1575 | if (radix != 10) | |
1576 | { | |
8001f118 | 1577 | offset = 2; |
c44af4eb TT |
1578 | could_be_decimal = 0; |
1579 | } | |
1580 | } | |
347dc102 | 1581 | |
8001f118 | 1582 | value = strtoul (number.c_str () + offset, NULL, radix); |
347dc102 TT |
1583 | if (implicit_i32 && value >= ((uint64_t) 1) << 31) |
1584 | type = rust_type ("i64"); | |
1585 | ||
1586 | rustyylval.typed_val_int.val = value; | |
c44af4eb TT |
1587 | rustyylval.typed_val_int.type = type; |
1588 | } | |
1589 | else | |
1590 | { | |
8001f118 | 1591 | rustyylval.typed_val_float.dval = strtod (number.c_str (), NULL); |
c44af4eb TT |
1592 | rustyylval.typed_val_float.type = type; |
1593 | } | |
1594 | ||
c44af4eb TT |
1595 | return is_integer ? (could_be_decimal ? DECIMAL_INTEGER : INTEGER) : FLOAT; |
1596 | } | |
1597 | ||
1598 | /* The lexer. */ | |
1599 | ||
1600 | static int | |
1601 | rustyylex (void) | |
1602 | { | |
1603 | /* Skip all leading whitespace. */ | |
1604 | while (lexptr[0] == ' ' || lexptr[0] == '\t' || lexptr[0] == '\r' | |
1605 | || lexptr[0] == '\n') | |
1606 | ++lexptr; | |
1607 | ||
1608 | /* If we hit EOF and we're completing, then return COMPLETE -- maybe | |
1609 | we're completing an empty string at the end of a field_expr. | |
1610 | But, we don't want to return two COMPLETE tokens in a row. */ | |
1611 | if (lexptr[0] == '\0' && lexptr == prev_lexptr) | |
1612 | return 0; | |
1613 | prev_lexptr = lexptr; | |
1614 | if (lexptr[0] == '\0') | |
1615 | { | |
1616 | if (parse_completion) | |
1617 | { | |
1618 | rustyylval.sval = make_stoken (""); | |
1619 | return COMPLETE; | |
1620 | } | |
1621 | return 0; | |
1622 | } | |
1623 | ||
1624 | if (lexptr[0] >= '0' && lexptr[0] <= '9') | |
1625 | return lex_number (); | |
1626 | else if (lexptr[0] == 'b' && lexptr[1] == '\'') | |
1627 | return lex_character (); | |
1628 | else if (lexptr[0] == 'b' && lexptr[1] == '"') | |
1629 | return lex_string (); | |
1630 | else if (lexptr[0] == 'b' && starts_raw_string (lexptr + 1)) | |
1631 | return lex_string (); | |
1632 | else if (starts_raw_string (lexptr)) | |
1633 | return lex_string (); | |
1634 | else if (rust_identifier_start_p (lexptr[0])) | |
1635 | return lex_identifier (); | |
1636 | else if (lexptr[0] == '"') | |
1637 | return lex_string (); | |
1638 | else if (lexptr[0] == '\'') | |
1639 | return lex_character (); | |
1640 | else if (lexptr[0] == '}' || lexptr[0] == ']') | |
1641 | { | |
1642 | /* Falls through to lex_operator. */ | |
1643 | --paren_depth; | |
1644 | } | |
1645 | else if (lexptr[0] == '(' || lexptr[0] == '{') | |
1646 | { | |
1647 | /* Falls through to lex_operator. */ | |
1648 | ++paren_depth; | |
1649 | } | |
1650 | else if (lexptr[0] == ',' && comma_terminates && paren_depth == 0) | |
1651 | return 0; | |
1652 | ||
1653 | return lex_operator (); | |
1654 | } | |
1655 | ||
1656 | /* Push back a single character to be re-lexed. */ | |
1657 | ||
1658 | static void | |
1659 | rust_push_back (char c) | |
1660 | { | |
1661 | /* Can't be called before any lexing. */ | |
1662 | gdb_assert (prev_lexptr != NULL); | |
1663 | ||
1664 | --lexptr; | |
1665 | gdb_assert (*lexptr == c); | |
1666 | } | |
1667 | ||
1668 | \f | |
1669 | ||
1670 | /* Make an arbitrary operation and fill in the fields. */ | |
1671 | ||
1672 | static const struct rust_op * | |
1673 | ast_operation (enum exp_opcode opcode, const struct rust_op *left, | |
1674 | const struct rust_op *right) | |
1675 | { | |
3232fabd | 1676 | struct rust_op *result = OBSTACK_ZALLOC (work_obstack, struct rust_op); |
c44af4eb TT |
1677 | |
1678 | result->opcode = opcode; | |
1679 | result->left.op = left; | |
1680 | result->right.op = right; | |
1681 | ||
1682 | return result; | |
1683 | } | |
1684 | ||
1685 | /* Make a compound assignment operation. */ | |
1686 | ||
1687 | static const struct rust_op * | |
1688 | ast_compound_assignment (enum exp_opcode opcode, const struct rust_op *left, | |
1689 | const struct rust_op *right) | |
1690 | { | |
3232fabd | 1691 | struct rust_op *result = OBSTACK_ZALLOC (work_obstack, struct rust_op); |
c44af4eb TT |
1692 | |
1693 | result->opcode = opcode; | |
1694 | result->compound_assignment = 1; | |
1695 | result->left.op = left; | |
1696 | result->right.op = right; | |
1697 | ||
1698 | return result; | |
1699 | } | |
1700 | ||
1701 | /* Make a typed integer literal operation. */ | |
1702 | ||
1703 | static const struct rust_op * | |
1704 | ast_literal (struct typed_val_int val) | |
1705 | { | |
3232fabd | 1706 | struct rust_op *result = OBSTACK_ZALLOC (work_obstack, struct rust_op); |
c44af4eb TT |
1707 | |
1708 | result->opcode = OP_LONG; | |
1709 | result->left.typed_val_int = val; | |
1710 | ||
1711 | return result; | |
1712 | } | |
1713 | ||
1714 | /* Make a typed floating point literal operation. */ | |
1715 | ||
1716 | static const struct rust_op * | |
1717 | ast_dliteral (struct typed_val_float val) | |
1718 | { | |
3232fabd | 1719 | struct rust_op *result = OBSTACK_ZALLOC (work_obstack, struct rust_op); |
c44af4eb TT |
1720 | |
1721 | result->opcode = OP_DOUBLE; | |
1722 | result->left.typed_val_float = val; | |
1723 | ||
1724 | return result; | |
1725 | } | |
1726 | ||
1727 | /* Make a unary operation. */ | |
1728 | ||
1729 | static const struct rust_op * | |
1730 | ast_unary (enum exp_opcode opcode, const struct rust_op *expr) | |
1731 | { | |
1732 | return ast_operation (opcode, expr, NULL); | |
1733 | } | |
1734 | ||
1735 | /* Make a cast operation. */ | |
1736 | ||
1737 | static const struct rust_op * | |
1738 | ast_cast (const struct rust_op *expr, const struct rust_op *type) | |
1739 | { | |
3232fabd | 1740 | struct rust_op *result = OBSTACK_ZALLOC (work_obstack, struct rust_op); |
c44af4eb TT |
1741 | |
1742 | result->opcode = UNOP_CAST; | |
1743 | result->left.op = expr; | |
1744 | result->right.op = type; | |
1745 | ||
1746 | return result; | |
1747 | } | |
1748 | ||
1749 | /* Make a call-like operation. This is nominally a function call, but | |
1750 | when lowering we may discover that it actually represents the | |
1751 | creation of a tuple struct. */ | |
1752 | ||
1753 | static const struct rust_op * | |
1754 | ast_call_ish (enum exp_opcode opcode, const struct rust_op *expr, | |
3232fabd | 1755 | rust_op_vector *params) |
c44af4eb | 1756 | { |
3232fabd | 1757 | struct rust_op *result = OBSTACK_ZALLOC (work_obstack, struct rust_op); |
c44af4eb TT |
1758 | |
1759 | result->opcode = opcode; | |
1760 | result->left.op = expr; | |
1761 | result->right.params = params; | |
1762 | ||
1763 | return result; | |
1764 | } | |
1765 | ||
1766 | /* Make a structure creation operation. */ | |
1767 | ||
1768 | static const struct rust_op * | |
3232fabd | 1769 | ast_struct (const struct rust_op *name, rust_set_vector *fields) |
c44af4eb | 1770 | { |
3232fabd | 1771 | struct rust_op *result = OBSTACK_ZALLOC (work_obstack, struct rust_op); |
c44af4eb TT |
1772 | |
1773 | result->opcode = OP_AGGREGATE; | |
1774 | result->left.op = name; | |
1775 | result->right.field_inits = fields; | |
1776 | ||
1777 | return result; | |
1778 | } | |
1779 | ||
1780 | /* Make an identifier path. */ | |
1781 | ||
1782 | static const struct rust_op * | |
3232fabd | 1783 | ast_path (struct stoken path, rust_op_vector *params) |
c44af4eb | 1784 | { |
3232fabd | 1785 | struct rust_op *result = OBSTACK_ZALLOC (work_obstack, struct rust_op); |
c44af4eb TT |
1786 | |
1787 | result->opcode = OP_VAR_VALUE; | |
1788 | result->left.sval = path; | |
1789 | result->right.params = params; | |
1790 | ||
1791 | return result; | |
1792 | } | |
1793 | ||
1794 | /* Make a string constant operation. */ | |
1795 | ||
1796 | static const struct rust_op * | |
1797 | ast_string (struct stoken str) | |
1798 | { | |
3232fabd | 1799 | struct rust_op *result = OBSTACK_ZALLOC (work_obstack, struct rust_op); |
c44af4eb TT |
1800 | |
1801 | result->opcode = OP_STRING; | |
1802 | result->left.sval = str; | |
1803 | ||
1804 | return result; | |
1805 | } | |
1806 | ||
1807 | /* Make a field expression. */ | |
1808 | ||
1809 | static const struct rust_op * | |
1810 | ast_structop (const struct rust_op *left, const char *name, int completing) | |
1811 | { | |
3232fabd | 1812 | struct rust_op *result = OBSTACK_ZALLOC (work_obstack, struct rust_op); |
c44af4eb TT |
1813 | |
1814 | result->opcode = STRUCTOP_STRUCT; | |
1815 | result->completing = completing; | |
1816 | result->left.op = left; | |
1817 | result->right.sval = make_stoken (name); | |
1818 | ||
1819 | return result; | |
1820 | } | |
1821 | ||
1822 | /* Make an anonymous struct operation, like 'x.0'. */ | |
1823 | ||
1824 | static const struct rust_op * | |
1825 | ast_structop_anonymous (const struct rust_op *left, | |
1826 | struct typed_val_int number) | |
1827 | { | |
3232fabd | 1828 | struct rust_op *result = OBSTACK_ZALLOC (work_obstack, struct rust_op); |
c44af4eb TT |
1829 | |
1830 | result->opcode = STRUCTOP_ANONYMOUS; | |
1831 | result->left.op = left; | |
1832 | result->right.typed_val_int = number; | |
1833 | ||
1834 | return result; | |
1835 | } | |
1836 | ||
1837 | /* Make a range operation. */ | |
1838 | ||
1839 | static const struct rust_op * | |
1840 | ast_range (const struct rust_op *lhs, const struct rust_op *rhs) | |
1841 | { | |
3232fabd | 1842 | struct rust_op *result = OBSTACK_ZALLOC (work_obstack, struct rust_op); |
c44af4eb | 1843 | |
01739a3b | 1844 | result->opcode = OP_RANGE; |
c44af4eb TT |
1845 | result->left.op = lhs; |
1846 | result->right.op = rhs; | |
1847 | ||
1848 | return result; | |
1849 | } | |
1850 | ||
1851 | /* A helper function to make a type-related AST node. */ | |
1852 | ||
1853 | static struct rust_op * | |
1854 | ast_basic_type (enum type_code typecode) | |
1855 | { | |
3232fabd | 1856 | struct rust_op *result = OBSTACK_ZALLOC (work_obstack, struct rust_op); |
c44af4eb TT |
1857 | |
1858 | result->opcode = OP_TYPE; | |
1859 | result->typecode = typecode; | |
1860 | return result; | |
1861 | } | |
1862 | ||
1863 | /* Create an AST node describing an array type. */ | |
1864 | ||
1865 | static const struct rust_op * | |
1866 | ast_array_type (const struct rust_op *lhs, struct typed_val_int val) | |
1867 | { | |
1868 | struct rust_op *result = ast_basic_type (TYPE_CODE_ARRAY); | |
1869 | ||
1870 | result->left.op = lhs; | |
1871 | result->right.typed_val_int = val; | |
1872 | return result; | |
1873 | } | |
1874 | ||
1875 | /* Create an AST node describing a reference type. */ | |
1876 | ||
1877 | static const struct rust_op * | |
1878 | ast_slice_type (const struct rust_op *type) | |
1879 | { | |
1880 | /* Use TYPE_CODE_COMPLEX just because it is handy. */ | |
1881 | struct rust_op *result = ast_basic_type (TYPE_CODE_COMPLEX); | |
1882 | ||
1883 | result->left.op = type; | |
1884 | return result; | |
1885 | } | |
1886 | ||
1887 | /* Create an AST node describing a reference type. */ | |
1888 | ||
1889 | static const struct rust_op * | |
1890 | ast_reference_type (const struct rust_op *type) | |
1891 | { | |
1892 | struct rust_op *result = ast_basic_type (TYPE_CODE_REF); | |
1893 | ||
1894 | result->left.op = type; | |
1895 | return result; | |
1896 | } | |
1897 | ||
1898 | /* Create an AST node describing a pointer type. */ | |
1899 | ||
1900 | static const struct rust_op * | |
1901 | ast_pointer_type (const struct rust_op *type, int is_mut) | |
1902 | { | |
1903 | struct rust_op *result = ast_basic_type (TYPE_CODE_PTR); | |
1904 | ||
1905 | result->left.op = type; | |
1906 | /* For the time being we ignore is_mut. */ | |
1907 | return result; | |
1908 | } | |
1909 | ||
1910 | /* Create an AST node describing a function type. */ | |
1911 | ||
1912 | static const struct rust_op * | |
3232fabd | 1913 | ast_function_type (const struct rust_op *rtype, rust_op_vector *params) |
c44af4eb TT |
1914 | { |
1915 | struct rust_op *result = ast_basic_type (TYPE_CODE_FUNC); | |
1916 | ||
1917 | result->left.op = rtype; | |
1918 | result->right.params = params; | |
1919 | return result; | |
1920 | } | |
1921 | ||
1922 | /* Create an AST node describing a tuple type. */ | |
1923 | ||
1924 | static const struct rust_op * | |
3232fabd | 1925 | ast_tuple_type (rust_op_vector *params) |
c44af4eb TT |
1926 | { |
1927 | struct rust_op *result = ast_basic_type (TYPE_CODE_STRUCT); | |
1928 | ||
1929 | result->left.params = params; | |
1930 | return result; | |
1931 | } | |
1932 | ||
1933 | /* A helper to appropriately munge NAME and BLOCK depending on the | |
1934 | presence of a leading "::". */ | |
1935 | ||
1936 | static void | |
1937 | munge_name_and_block (const char **name, const struct block **block) | |
1938 | { | |
1939 | /* If it is a global reference, skip the current block in favor of | |
1940 | the static block. */ | |
1941 | if (strncmp (*name, "::", 2) == 0) | |
1942 | { | |
1943 | *name += 2; | |
1944 | *block = block_static_block (*block); | |
1945 | } | |
1946 | } | |
1947 | ||
1948 | /* Like lookup_symbol, but handles Rust namespace conventions, and | |
1949 | doesn't require field_of_this_result. */ | |
1950 | ||
1951 | static struct block_symbol | |
1952 | rust_lookup_symbol (const char *name, const struct block *block, | |
1953 | const domain_enum domain) | |
1954 | { | |
1955 | struct block_symbol result; | |
1956 | ||
1957 | munge_name_and_block (&name, &block); | |
1958 | ||
1959 | result = lookup_symbol (name, block, domain, NULL); | |
1960 | if (result.symbol != NULL) | |
1961 | update_innermost_block (result); | |
1962 | return result; | |
1963 | } | |
1964 | ||
1965 | /* Look up a type, following Rust namespace conventions. */ | |
1966 | ||
1967 | static struct type * | |
1968 | rust_lookup_type (const char *name, const struct block *block) | |
1969 | { | |
1970 | struct block_symbol result; | |
1971 | struct type *type; | |
1972 | ||
1973 | munge_name_and_block (&name, &block); | |
1974 | ||
1975 | result = lookup_symbol (name, block, STRUCT_DOMAIN, NULL); | |
1976 | if (result.symbol != NULL) | |
1977 | { | |
1978 | update_innermost_block (result); | |
1979 | return SYMBOL_TYPE (result.symbol); | |
1980 | } | |
1981 | ||
3232fabd | 1982 | type = lookup_typename (current_parser->language (), current_parser->arch (), |
c44af4eb TT |
1983 | name, NULL, 1); |
1984 | if (type != NULL) | |
1985 | return type; | |
1986 | ||
1987 | /* Last chance, try a built-in type. */ | |
3232fabd TT |
1988 | return language_lookup_primitive_type (current_parser->language (), |
1989 | current_parser->arch (), | |
c44af4eb TT |
1990 | name); |
1991 | } | |
1992 | ||
1993 | static struct type *convert_ast_to_type (struct parser_state *state, | |
1994 | const struct rust_op *operation); | |
1995 | static const char *convert_name (struct parser_state *state, | |
1996 | const struct rust_op *operation); | |
1997 | ||
1998 | /* Convert a vector of rust_ops representing types to a vector of | |
1999 | types. */ | |
2000 | ||
8001f118 | 2001 | static std::vector<struct type *> |
3232fabd | 2002 | convert_params_to_types (struct parser_state *state, rust_op_vector *params) |
c44af4eb | 2003 | { |
8001f118 | 2004 | std::vector<struct type *> result; |
c44af4eb | 2005 | |
3232fabd | 2006 | for (const rust_op *op : *params) |
8001f118 | 2007 | result.push_back (convert_ast_to_type (state, op)); |
c44af4eb | 2008 | |
c44af4eb TT |
2009 | return result; |
2010 | } | |
2011 | ||
2012 | /* Convert a rust_op representing a type to a struct type *. */ | |
2013 | ||
2014 | static struct type * | |
2015 | convert_ast_to_type (struct parser_state *state, | |
2016 | const struct rust_op *operation) | |
2017 | { | |
2018 | struct type *type, *result = NULL; | |
2019 | ||
2020 | if (operation->opcode == OP_VAR_VALUE) | |
2021 | { | |
2022 | const char *varname = convert_name (state, operation); | |
2023 | ||
2024 | result = rust_lookup_type (varname, expression_context_block); | |
2025 | if (result == NULL) | |
2026 | error (_("No typed name '%s' in current context"), varname); | |
2027 | return result; | |
2028 | } | |
2029 | ||
2030 | gdb_assert (operation->opcode == OP_TYPE); | |
2031 | ||
2032 | switch (operation->typecode) | |
2033 | { | |
2034 | case TYPE_CODE_ARRAY: | |
2035 | type = convert_ast_to_type (state, operation->left.op); | |
2036 | if (operation->right.typed_val_int.val < 0) | |
2037 | error (_("Negative array length")); | |
2038 | result = lookup_array_range_type (type, 0, | |
2039 | operation->right.typed_val_int.val - 1); | |
2040 | break; | |
2041 | ||
2042 | case TYPE_CODE_COMPLEX: | |
2043 | { | |
2044 | struct type *usize = rust_type ("usize"); | |
2045 | ||
2046 | type = convert_ast_to_type (state, operation->left.op); | |
2047 | result = rust_slice_type ("&[*gdb*]", type, usize); | |
2048 | } | |
2049 | break; | |
2050 | ||
2051 | case TYPE_CODE_REF: | |
2052 | case TYPE_CODE_PTR: | |
2053 | /* For now we treat &x and *x identically. */ | |
2054 | type = convert_ast_to_type (state, operation->left.op); | |
2055 | result = lookup_pointer_type (type); | |
2056 | break; | |
2057 | ||
2058 | case TYPE_CODE_FUNC: | |
2059 | { | |
8001f118 | 2060 | std::vector<struct type *> args |
3232fabd | 2061 | (convert_params_to_types (state, operation->right.params)); |
c44af4eb TT |
2062 | struct type **argtypes = NULL; |
2063 | ||
2064 | type = convert_ast_to_type (state, operation->left.op); | |
8001f118 TT |
2065 | if (!args.empty ()) |
2066 | argtypes = args.data (); | |
c44af4eb TT |
2067 | |
2068 | result | |
8001f118 | 2069 | = lookup_function_type_with_arguments (type, args.size (), |
c44af4eb TT |
2070 | argtypes); |
2071 | result = lookup_pointer_type (result); | |
c44af4eb TT |
2072 | } |
2073 | break; | |
2074 | ||
2075 | case TYPE_CODE_STRUCT: | |
2076 | { | |
8001f118 | 2077 | std::vector<struct type *> args |
3232fabd | 2078 | (convert_params_to_types (state, operation->left.params)); |
c44af4eb TT |
2079 | int i; |
2080 | struct type *type; | |
2081 | const char *name; | |
2082 | ||
3232fabd | 2083 | obstack_1grow (work_obstack, '('); |
8001f118 | 2084 | for (i = 0; i < args.size (); ++i) |
c44af4eb | 2085 | { |
8001f118 | 2086 | std::string type_name = type_to_string (args[i]); |
c44af4eb TT |
2087 | |
2088 | if (i > 0) | |
3232fabd TT |
2089 | obstack_1grow (work_obstack, ','); |
2090 | obstack_grow_str (work_obstack, type_name.c_str ()); | |
c44af4eb TT |
2091 | } |
2092 | ||
3232fabd TT |
2093 | obstack_grow_str0 (work_obstack, ")"); |
2094 | name = (const char *) obstack_finish (work_obstack); | |
c44af4eb TT |
2095 | |
2096 | /* We don't allow creating new tuple types (yet), but we do | |
2097 | allow looking up existing tuple types. */ | |
2098 | result = rust_lookup_type (name, expression_context_block); | |
2099 | if (result == NULL) | |
2100 | error (_("could not find tuple type '%s'"), name); | |
c44af4eb TT |
2101 | } |
2102 | break; | |
2103 | ||
2104 | default: | |
2105 | gdb_assert_not_reached ("unhandled opcode in convert_ast_to_type"); | |
2106 | } | |
2107 | ||
2108 | gdb_assert (result != NULL); | |
2109 | return result; | |
2110 | } | |
2111 | ||
2112 | /* A helper function to turn a rust_op representing a name into a full | |
2113 | name. This applies generic arguments as needed. The returned name | |
2114 | is allocated on the work obstack. */ | |
2115 | ||
2116 | static const char * | |
2117 | convert_name (struct parser_state *state, const struct rust_op *operation) | |
2118 | { | |
c44af4eb | 2119 | int i; |
c44af4eb TT |
2120 | |
2121 | gdb_assert (operation->opcode == OP_VAR_VALUE); | |
2122 | ||
2123 | if (operation->right.params == NULL) | |
2124 | return operation->left.sval.ptr; | |
2125 | ||
8001f118 | 2126 | std::vector<struct type *> types |
3232fabd | 2127 | (convert_params_to_types (state, operation->right.params)); |
c44af4eb | 2128 | |
3232fabd TT |
2129 | obstack_grow_str (work_obstack, operation->left.sval.ptr); |
2130 | obstack_1grow (work_obstack, '<'); | |
8001f118 | 2131 | for (i = 0; i < types.size (); ++i) |
c44af4eb | 2132 | { |
8001f118 | 2133 | std::string type_name = type_to_string (types[i]); |
c44af4eb TT |
2134 | |
2135 | if (i > 0) | |
3232fabd | 2136 | obstack_1grow (work_obstack, ','); |
c44af4eb | 2137 | |
3232fabd | 2138 | obstack_grow_str (work_obstack, type_name.c_str ()); |
c44af4eb | 2139 | } |
3232fabd | 2140 | obstack_grow_str0 (work_obstack, ">"); |
c44af4eb | 2141 | |
3232fabd | 2142 | return (const char *) obstack_finish (work_obstack); |
c44af4eb TT |
2143 | } |
2144 | ||
2145 | static void convert_ast_to_expression (struct parser_state *state, | |
2146 | const struct rust_op *operation, | |
8880f2a9 TT |
2147 | const struct rust_op *top, |
2148 | bool want_type = false); | |
c44af4eb TT |
2149 | |
2150 | /* A helper function that converts a vec of rust_ops to a gdb | |
2151 | expression. */ | |
2152 | ||
2153 | static void | |
2154 | convert_params_to_expression (struct parser_state *state, | |
3232fabd | 2155 | rust_op_vector *params, |
c44af4eb TT |
2156 | const struct rust_op *top) |
2157 | { | |
3232fabd | 2158 | for (const rust_op *elem : *params) |
c44af4eb TT |
2159 | convert_ast_to_expression (state, elem, top); |
2160 | } | |
2161 | ||
2162 | /* Lower a rust_op to a gdb expression. STATE is the parser state. | |
2163 | OPERATION is the operation to lower. TOP is a pointer to the | |
2164 | top-most operation; it is used to handle the special case where the | |
2165 | top-most expression is an identifier and can be optionally lowered | |
8880f2a9 TT |
2166 | to OP_TYPE. WANT_TYPE is a flag indicating that, if the expression |
2167 | is the name of a type, then emit an OP_TYPE for it (rather than | |
2168 | erroring). If WANT_TYPE is set, then the similar TOP handling is | |
2169 | not done. */ | |
c44af4eb TT |
2170 | |
2171 | static void | |
2172 | convert_ast_to_expression (struct parser_state *state, | |
2173 | const struct rust_op *operation, | |
8880f2a9 TT |
2174 | const struct rust_op *top, |
2175 | bool want_type) | |
c44af4eb TT |
2176 | { |
2177 | switch (operation->opcode) | |
2178 | { | |
2179 | case OP_LONG: | |
2180 | write_exp_elt_opcode (state, OP_LONG); | |
2181 | write_exp_elt_type (state, operation->left.typed_val_int.type); | |
2182 | write_exp_elt_longcst (state, operation->left.typed_val_int.val); | |
2183 | write_exp_elt_opcode (state, OP_LONG); | |
2184 | break; | |
2185 | ||
2186 | case OP_DOUBLE: | |
2187 | write_exp_elt_opcode (state, OP_DOUBLE); | |
2188 | write_exp_elt_type (state, operation->left.typed_val_float.type); | |
2189 | write_exp_elt_dblcst (state, operation->left.typed_val_float.dval); | |
2190 | write_exp_elt_opcode (state, OP_DOUBLE); | |
2191 | break; | |
2192 | ||
2193 | case STRUCTOP_STRUCT: | |
2194 | { | |
2195 | convert_ast_to_expression (state, operation->left.op, top); | |
2196 | ||
2197 | if (operation->completing) | |
2198 | mark_struct_expression (state); | |
2199 | write_exp_elt_opcode (state, STRUCTOP_STRUCT); | |
2200 | write_exp_string (state, operation->right.sval); | |
2201 | write_exp_elt_opcode (state, STRUCTOP_STRUCT); | |
2202 | } | |
2203 | break; | |
2204 | ||
2205 | case STRUCTOP_ANONYMOUS: | |
2206 | { | |
2207 | convert_ast_to_expression (state, operation->left.op, top); | |
2208 | ||
2209 | write_exp_elt_opcode (state, STRUCTOP_ANONYMOUS); | |
2210 | write_exp_elt_longcst (state, operation->right.typed_val_int.val); | |
2211 | write_exp_elt_opcode (state, STRUCTOP_ANONYMOUS); | |
2212 | } | |
2213 | break; | |
2214 | ||
8880f2a9 TT |
2215 | case UNOP_SIZEOF: |
2216 | convert_ast_to_expression (state, operation->left.op, top, true); | |
2217 | write_exp_elt_opcode (state, UNOP_SIZEOF); | |
2218 | break; | |
2219 | ||
c44af4eb TT |
2220 | case UNOP_PLUS: |
2221 | case UNOP_NEG: | |
2222 | case UNOP_COMPLEMENT: | |
2223 | case UNOP_IND: | |
2224 | case UNOP_ADDR: | |
2225 | convert_ast_to_expression (state, operation->left.op, top); | |
2226 | write_exp_elt_opcode (state, operation->opcode); | |
2227 | break; | |
2228 | ||
2229 | case BINOP_SUBSCRIPT: | |
2230 | case BINOP_MUL: | |
2231 | case BINOP_REPEAT: | |
2232 | case BINOP_DIV: | |
2233 | case BINOP_REM: | |
2234 | case BINOP_LESS: | |
2235 | case BINOP_GTR: | |
2236 | case BINOP_BITWISE_AND: | |
2237 | case BINOP_BITWISE_IOR: | |
2238 | case BINOP_BITWISE_XOR: | |
2239 | case BINOP_ADD: | |
2240 | case BINOP_SUB: | |
2241 | case BINOP_LOGICAL_OR: | |
2242 | case BINOP_LOGICAL_AND: | |
2243 | case BINOP_EQUAL: | |
2244 | case BINOP_NOTEQUAL: | |
2245 | case BINOP_LEQ: | |
2246 | case BINOP_GEQ: | |
2247 | case BINOP_LSH: | |
2248 | case BINOP_RSH: | |
2249 | case BINOP_ASSIGN: | |
2250 | case OP_RUST_ARRAY: | |
2251 | convert_ast_to_expression (state, operation->left.op, top); | |
2252 | convert_ast_to_expression (state, operation->right.op, top); | |
2253 | if (operation->compound_assignment) | |
2254 | { | |
2255 | write_exp_elt_opcode (state, BINOP_ASSIGN_MODIFY); | |
2256 | write_exp_elt_opcode (state, operation->opcode); | |
2257 | write_exp_elt_opcode (state, BINOP_ASSIGN_MODIFY); | |
2258 | } | |
2259 | else | |
2260 | write_exp_elt_opcode (state, operation->opcode); | |
2261 | ||
2262 | if (operation->compound_assignment | |
2263 | || operation->opcode == BINOP_ASSIGN) | |
2264 | { | |
2265 | struct type *type; | |
2266 | ||
2267 | type = language_lookup_primitive_type (parse_language (state), | |
2268 | parse_gdbarch (state), | |
2269 | "()"); | |
2270 | ||
2271 | write_exp_elt_opcode (state, OP_LONG); | |
2272 | write_exp_elt_type (state, type); | |
2273 | write_exp_elt_longcst (state, 0); | |
2274 | write_exp_elt_opcode (state, OP_LONG); | |
2275 | ||
2276 | write_exp_elt_opcode (state, BINOP_COMMA); | |
2277 | } | |
2278 | break; | |
2279 | ||
2280 | case UNOP_CAST: | |
2281 | { | |
2282 | struct type *type = convert_ast_to_type (state, operation->right.op); | |
2283 | ||
2284 | convert_ast_to_expression (state, operation->left.op, top); | |
2285 | write_exp_elt_opcode (state, UNOP_CAST); | |
2286 | write_exp_elt_type (state, type); | |
2287 | write_exp_elt_opcode (state, UNOP_CAST); | |
2288 | } | |
2289 | break; | |
2290 | ||
2291 | case OP_FUNCALL: | |
2292 | { | |
2293 | if (operation->left.op->opcode == OP_VAR_VALUE) | |
2294 | { | |
2295 | struct type *type; | |
2296 | const char *varname = convert_name (state, operation->left.op); | |
2297 | ||
2298 | type = rust_lookup_type (varname, expression_context_block); | |
2299 | if (type != NULL) | |
2300 | { | |
2301 | /* This is actually a tuple struct expression, not a | |
2302 | call expression. */ | |
3232fabd | 2303 | rust_op_vector *params = operation->right.params; |
c44af4eb TT |
2304 | |
2305 | if (TYPE_CODE (type) != TYPE_CODE_NAMESPACE) | |
2306 | { | |
2307 | if (!rust_tuple_struct_type_p (type)) | |
2308 | error (_("Type %s is not a tuple struct"), varname); | |
2309 | ||
3232fabd | 2310 | for (int i = 0; i < params->size (); ++i) |
c44af4eb TT |
2311 | { |
2312 | char *cell = get_print_cell (); | |
2313 | ||
2314 | xsnprintf (cell, PRINT_CELL_SIZE, "__%d", i); | |
2315 | write_exp_elt_opcode (state, OP_NAME); | |
2316 | write_exp_string (state, make_stoken (cell)); | |
2317 | write_exp_elt_opcode (state, OP_NAME); | |
2318 | ||
3232fabd | 2319 | convert_ast_to_expression (state, (*params)[i], top); |
c44af4eb TT |
2320 | } |
2321 | ||
2322 | write_exp_elt_opcode (state, OP_AGGREGATE); | |
2323 | write_exp_elt_type (state, type); | |
3232fabd | 2324 | write_exp_elt_longcst (state, 2 * params->size ()); |
c44af4eb TT |
2325 | write_exp_elt_opcode (state, OP_AGGREGATE); |
2326 | break; | |
2327 | } | |
2328 | } | |
2329 | } | |
2330 | convert_ast_to_expression (state, operation->left.op, top); | |
3232fabd | 2331 | convert_params_to_expression (state, operation->right.params, top); |
c44af4eb | 2332 | write_exp_elt_opcode (state, OP_FUNCALL); |
3232fabd | 2333 | write_exp_elt_longcst (state, operation->right.params->size ()); |
c44af4eb TT |
2334 | write_exp_elt_longcst (state, OP_FUNCALL); |
2335 | } | |
2336 | break; | |
2337 | ||
2338 | case OP_ARRAY: | |
2339 | gdb_assert (operation->left.op == NULL); | |
3232fabd | 2340 | convert_params_to_expression (state, operation->right.params, top); |
c44af4eb TT |
2341 | write_exp_elt_opcode (state, OP_ARRAY); |
2342 | write_exp_elt_longcst (state, 0); | |
3232fabd | 2343 | write_exp_elt_longcst (state, operation->right.params->size () - 1); |
c44af4eb TT |
2344 | write_exp_elt_longcst (state, OP_ARRAY); |
2345 | break; | |
2346 | ||
2347 | case OP_VAR_VALUE: | |
2348 | { | |
2349 | struct block_symbol sym; | |
2350 | const char *varname; | |
2351 | ||
2352 | if (operation->left.sval.ptr[0] == '$') | |
2353 | { | |
2354 | write_dollar_variable (state, operation->left.sval); | |
2355 | break; | |
2356 | } | |
2357 | ||
2358 | varname = convert_name (state, operation); | |
2359 | sym = rust_lookup_symbol (varname, expression_context_block, | |
2360 | VAR_DOMAIN); | |
65547233 | 2361 | if (sym.symbol != NULL && SYMBOL_CLASS (sym.symbol) != LOC_TYPEDEF) |
c44af4eb TT |
2362 | { |
2363 | write_exp_elt_opcode (state, OP_VAR_VALUE); | |
2364 | write_exp_elt_block (state, sym.block); | |
2365 | write_exp_elt_sym (state, sym.symbol); | |
2366 | write_exp_elt_opcode (state, OP_VAR_VALUE); | |
2367 | } | |
2368 | else | |
2369 | { | |
65547233 | 2370 | struct type *type = NULL; |
c44af4eb | 2371 | |
65547233 TT |
2372 | if (sym.symbol != NULL) |
2373 | { | |
2374 | gdb_assert (SYMBOL_CLASS (sym.symbol) == LOC_TYPEDEF); | |
2375 | type = SYMBOL_TYPE (sym.symbol); | |
2376 | } | |
2377 | if (type == NULL) | |
2378 | type = rust_lookup_type (varname, expression_context_block); | |
c44af4eb TT |
2379 | if (type == NULL) |
2380 | error (_("No symbol '%s' in current context"), varname); | |
2381 | ||
8880f2a9 TT |
2382 | if (!want_type |
2383 | && TYPE_CODE (type) == TYPE_CODE_STRUCT | |
c44af4eb TT |
2384 | && TYPE_NFIELDS (type) == 0) |
2385 | { | |
2386 | /* A unit-like struct. */ | |
2387 | write_exp_elt_opcode (state, OP_AGGREGATE); | |
2388 | write_exp_elt_type (state, type); | |
2389 | write_exp_elt_longcst (state, 0); | |
2390 | write_exp_elt_opcode (state, OP_AGGREGATE); | |
2391 | } | |
8880f2a9 | 2392 | else if (want_type || operation == top) |
c44af4eb TT |
2393 | { |
2394 | write_exp_elt_opcode (state, OP_TYPE); | |
2395 | write_exp_elt_type (state, type); | |
2396 | write_exp_elt_opcode (state, OP_TYPE); | |
c44af4eb | 2397 | } |
8880f2a9 TT |
2398 | else |
2399 | error (_("Found type '%s', which can't be " | |
2400 | "evaluated in this context"), | |
2401 | varname); | |
c44af4eb TT |
2402 | } |
2403 | } | |
2404 | break; | |
2405 | ||
2406 | case OP_AGGREGATE: | |
2407 | { | |
2408 | int i; | |
2409 | int length; | |
3232fabd | 2410 | rust_set_vector *fields = operation->right.field_inits; |
c44af4eb TT |
2411 | struct type *type; |
2412 | const char *name; | |
2413 | ||
2414 | length = 0; | |
3232fabd | 2415 | for (const set_field &init : *fields) |
c44af4eb | 2416 | { |
3232fabd | 2417 | if (init.name.ptr != NULL) |
c44af4eb TT |
2418 | { |
2419 | write_exp_elt_opcode (state, OP_NAME); | |
3232fabd | 2420 | write_exp_string (state, init.name); |
c44af4eb TT |
2421 | write_exp_elt_opcode (state, OP_NAME); |
2422 | ++length; | |
2423 | } | |
2424 | ||
3232fabd | 2425 | convert_ast_to_expression (state, init.init, top); |
c44af4eb TT |
2426 | ++length; |
2427 | ||
3232fabd | 2428 | if (init.name.ptr == NULL) |
c44af4eb TT |
2429 | { |
2430 | /* This is handled differently from Ada in our | |
2431 | evaluator. */ | |
2432 | write_exp_elt_opcode (state, OP_OTHERS); | |
2433 | } | |
2434 | } | |
2435 | ||
2436 | name = convert_name (state, operation->left.op); | |
2437 | type = rust_lookup_type (name, expression_context_block); | |
2438 | if (type == NULL) | |
2439 | error (_("Could not find type '%s'"), operation->left.sval.ptr); | |
2440 | ||
2441 | if (TYPE_CODE (type) != TYPE_CODE_STRUCT | |
2442 | || rust_tuple_type_p (type) | |
2443 | || rust_tuple_struct_type_p (type)) | |
2444 | error (_("Struct expression applied to non-struct type")); | |
2445 | ||
2446 | write_exp_elt_opcode (state, OP_AGGREGATE); | |
2447 | write_exp_elt_type (state, type); | |
2448 | write_exp_elt_longcst (state, length); | |
2449 | write_exp_elt_opcode (state, OP_AGGREGATE); | |
2450 | } | |
2451 | break; | |
2452 | ||
2453 | case OP_STRING: | |
2454 | { | |
2455 | write_exp_elt_opcode (state, OP_STRING); | |
2456 | write_exp_string (state, operation->left.sval); | |
2457 | write_exp_elt_opcode (state, OP_STRING); | |
2458 | } | |
2459 | break; | |
2460 | ||
01739a3b | 2461 | case OP_RANGE: |
c44af4eb | 2462 | { |
01739a3b | 2463 | enum range_type kind = BOTH_BOUND_DEFAULT; |
c44af4eb TT |
2464 | |
2465 | if (operation->left.op != NULL) | |
2466 | { | |
2467 | convert_ast_to_expression (state, operation->left.op, top); | |
2468 | kind = HIGH_BOUND_DEFAULT; | |
2469 | } | |
2470 | if (operation->right.op != NULL) | |
2471 | { | |
2472 | convert_ast_to_expression (state, operation->right.op, top); | |
2473 | if (kind == BOTH_BOUND_DEFAULT) | |
2474 | kind = LOW_BOUND_DEFAULT; | |
2475 | else | |
2476 | { | |
2477 | gdb_assert (kind == HIGH_BOUND_DEFAULT); | |
2478 | kind = NONE_BOUND_DEFAULT; | |
2479 | } | |
2480 | } | |
01739a3b | 2481 | write_exp_elt_opcode (state, OP_RANGE); |
c44af4eb | 2482 | write_exp_elt_longcst (state, kind); |
01739a3b | 2483 | write_exp_elt_opcode (state, OP_RANGE); |
c44af4eb TT |
2484 | } |
2485 | break; | |
2486 | ||
2487 | default: | |
2488 | gdb_assert_not_reached ("unhandled opcode in convert_ast_to_expression"); | |
2489 | } | |
2490 | } | |
2491 | ||
2492 | \f | |
2493 | ||
2494 | /* The parser as exposed to gdb. */ | |
2495 | ||
2496 | int | |
2497 | rust_parse (struct parser_state *state) | |
2498 | { | |
2499 | int result; | |
c44af4eb | 2500 | |
3232fabd TT |
2501 | /* This sets various globals and also clears them on |
2502 | destruction. */ | |
2503 | rust_parser parser (state); | |
8268c778 | 2504 | |
c44af4eb TT |
2505 | result = rustyyparse (); |
2506 | ||
3232fabd TT |
2507 | if (!result || (parse_completion && parser.rust_ast != NULL)) |
2508 | convert_ast_to_expression (state, parser.rust_ast, parser.rust_ast); | |
c44af4eb | 2509 | |
c44af4eb TT |
2510 | return result; |
2511 | } | |
2512 | ||
2513 | /* The parser error handler. */ | |
2514 | ||
2515 | void | |
a121b7c1 | 2516 | rustyyerror (const char *msg) |
c44af4eb TT |
2517 | { |
2518 | const char *where = prev_lexptr ? prev_lexptr : lexptr; | |
2519 | error (_("%s in expression, near `%s'."), (msg ? msg : "Error"), where); | |
2520 | } | |
2521 | ||
2522 | \f | |
2523 | ||
2524 | #if GDB_SELF_TEST | |
2525 | ||
2526 | /* Initialize the lexer for testing. */ | |
2527 | ||
2528 | static void | |
2529 | rust_lex_test_init (const char *input) | |
2530 | { | |
2531 | prev_lexptr = NULL; | |
2532 | lexptr = input; | |
2533 | paren_depth = 0; | |
2534 | } | |
2535 | ||
2536 | /* A test helper that lexes a string, expecting a single token. It | |
2537 | returns the lexer data for this token. */ | |
2538 | ||
2539 | static RUSTSTYPE | |
2540 | rust_lex_test_one (const char *input, int expected) | |
2541 | { | |
2542 | int token; | |
2543 | RUSTSTYPE result; | |
2544 | ||
2545 | rust_lex_test_init (input); | |
2546 | ||
2547 | token = rustyylex (); | |
2548 | SELF_CHECK (token == expected); | |
2549 | result = rustyylval; | |
2550 | ||
2551 | if (token) | |
2552 | { | |
2553 | token = rustyylex (); | |
2554 | SELF_CHECK (token == 0); | |
2555 | } | |
2556 | ||
2557 | return result; | |
2558 | } | |
2559 | ||
2560 | /* Test that INPUT lexes as the integer VALUE. */ | |
2561 | ||
2562 | static void | |
2563 | rust_lex_int_test (const char *input, int value, int kind) | |
2564 | { | |
2565 | RUSTSTYPE result = rust_lex_test_one (input, kind); | |
2566 | SELF_CHECK (result.typed_val_int.val == value); | |
2567 | } | |
2568 | ||
2569 | /* Test that INPUT throws an exception with text ERR. */ | |
2570 | ||
2571 | static void | |
2572 | rust_lex_exception_test (const char *input, const char *err) | |
2573 | { | |
2574 | TRY | |
2575 | { | |
2576 | /* The "kind" doesn't matter. */ | |
2577 | rust_lex_test_one (input, DECIMAL_INTEGER); | |
2578 | SELF_CHECK (0); | |
2579 | } | |
2580 | CATCH (except, RETURN_MASK_ERROR) | |
2581 | { | |
2582 | SELF_CHECK (strcmp (except.message, err) == 0); | |
2583 | } | |
2584 | END_CATCH | |
2585 | } | |
2586 | ||
2587 | /* Test that INPUT lexes as the identifier, string, or byte-string | |
2588 | VALUE. KIND holds the expected token kind. */ | |
2589 | ||
2590 | static void | |
2591 | rust_lex_stringish_test (const char *input, const char *value, int kind) | |
2592 | { | |
2593 | RUSTSTYPE result = rust_lex_test_one (input, kind); | |
2594 | SELF_CHECK (result.sval.length == strlen (value)); | |
2595 | SELF_CHECK (strncmp (result.sval.ptr, value, result.sval.length) == 0); | |
2596 | } | |
2597 | ||
2598 | /* Helper to test that a string parses as a given token sequence. */ | |
2599 | ||
2600 | static void | |
2601 | rust_lex_test_sequence (const char *input, int len, const int expected[]) | |
2602 | { | |
2603 | int i; | |
2604 | ||
2605 | lexptr = input; | |
2606 | paren_depth = 0; | |
2607 | ||
2608 | for (i = 0; i < len; ++i) | |
2609 | { | |
2610 | int token = rustyylex (); | |
2611 | ||
2612 | SELF_CHECK (token == expected[i]); | |
2613 | } | |
2614 | } | |
2615 | ||
2616 | /* Tests for an integer-parsing corner case. */ | |
2617 | ||
2618 | static void | |
2619 | rust_lex_test_trailing_dot (void) | |
2620 | { | |
2621 | const int expected1[] = { DECIMAL_INTEGER, '.', IDENT, '(', ')', 0 }; | |
2622 | const int expected2[] = { INTEGER, '.', IDENT, '(', ')', 0 }; | |
2623 | const int expected3[] = { FLOAT, EQEQ, '(', ')', 0 }; | |
2624 | const int expected4[] = { DECIMAL_INTEGER, DOTDOT, DECIMAL_INTEGER, 0 }; | |
2625 | ||
2626 | rust_lex_test_sequence ("23.g()", ARRAY_SIZE (expected1), expected1); | |
2627 | rust_lex_test_sequence ("23_0.g()", ARRAY_SIZE (expected2), expected2); | |
2628 | rust_lex_test_sequence ("23.==()", ARRAY_SIZE (expected3), expected3); | |
2629 | rust_lex_test_sequence ("23..25", ARRAY_SIZE (expected4), expected4); | |
2630 | } | |
2631 | ||
2632 | /* Tests of completion. */ | |
2633 | ||
2634 | static void | |
2635 | rust_lex_test_completion (void) | |
2636 | { | |
2637 | const int expected[] = { IDENT, '.', COMPLETE, 0 }; | |
2638 | ||
2639 | parse_completion = 1; | |
2640 | ||
2641 | rust_lex_test_sequence ("something.wha", ARRAY_SIZE (expected), expected); | |
2642 | rust_lex_test_sequence ("something.", ARRAY_SIZE (expected), expected); | |
2643 | ||
2644 | parse_completion = 0; | |
2645 | } | |
2646 | ||
2647 | /* Test pushback. */ | |
2648 | ||
2649 | static void | |
2650 | rust_lex_test_push_back (void) | |
2651 | { | |
2652 | int token; | |
2653 | ||
2654 | rust_lex_test_init (">>="); | |
2655 | ||
2656 | token = rustyylex (); | |
2657 | SELF_CHECK (token == COMPOUND_ASSIGN); | |
2658 | SELF_CHECK (rustyylval.opcode == BINOP_RSH); | |
2659 | ||
2660 | rust_push_back ('='); | |
2661 | ||
2662 | token = rustyylex (); | |
2663 | SELF_CHECK (token == '='); | |
2664 | ||
2665 | token = rustyylex (); | |
2666 | SELF_CHECK (token == 0); | |
2667 | } | |
2668 | ||
2669 | /* Unit test the lexer. */ | |
2670 | ||
2671 | static void | |
2672 | rust_lex_tests (void) | |
2673 | { | |
2674 | int i; | |
2675 | ||
3232fabd TT |
2676 | auto_obstack test_obstack; |
2677 | scoped_restore obstack_holder = make_scoped_restore (&work_obstack, | |
2678 | &test_obstack); | |
2679 | ||
c44af4eb TT |
2680 | unit_testing = 1; |
2681 | ||
2682 | rust_lex_test_one ("", 0); | |
2683 | rust_lex_test_one (" \t \n \r ", 0); | |
2684 | rust_lex_test_one ("thread 23", 0); | |
2685 | rust_lex_test_one ("task 23", 0); | |
2686 | rust_lex_test_one ("th 104", 0); | |
2687 | rust_lex_test_one ("ta 97", 0); | |
2688 | ||
2689 | rust_lex_int_test ("'z'", 'z', INTEGER); | |
2690 | rust_lex_int_test ("'\\xff'", 0xff, INTEGER); | |
2691 | rust_lex_int_test ("'\\u{1016f}'", 0x1016f, INTEGER); | |
2692 | rust_lex_int_test ("b'z'", 'z', INTEGER); | |
2693 | rust_lex_int_test ("b'\\xfe'", 0xfe, INTEGER); | |
2694 | rust_lex_int_test ("b'\\xFE'", 0xfe, INTEGER); | |
2695 | rust_lex_int_test ("b'\\xfE'", 0xfe, INTEGER); | |
2696 | ||
2697 | /* Test all escapes in both modes. */ | |
2698 | rust_lex_int_test ("'\\n'", '\n', INTEGER); | |
2699 | rust_lex_int_test ("'\\r'", '\r', INTEGER); | |
2700 | rust_lex_int_test ("'\\t'", '\t', INTEGER); | |
2701 | rust_lex_int_test ("'\\\\'", '\\', INTEGER); | |
2702 | rust_lex_int_test ("'\\0'", '\0', INTEGER); | |
2703 | rust_lex_int_test ("'\\''", '\'', INTEGER); | |
2704 | rust_lex_int_test ("'\\\"'", '"', INTEGER); | |
2705 | ||
2706 | rust_lex_int_test ("b'\\n'", '\n', INTEGER); | |
2707 | rust_lex_int_test ("b'\\r'", '\r', INTEGER); | |
2708 | rust_lex_int_test ("b'\\t'", '\t', INTEGER); | |
2709 | rust_lex_int_test ("b'\\\\'", '\\', INTEGER); | |
2710 | rust_lex_int_test ("b'\\0'", '\0', INTEGER); | |
2711 | rust_lex_int_test ("b'\\''", '\'', INTEGER); | |
2712 | rust_lex_int_test ("b'\\\"'", '"', INTEGER); | |
2713 | ||
2714 | rust_lex_exception_test ("'z", "Unterminated character literal"); | |
2715 | rust_lex_exception_test ("b'\\x0'", "Not enough hex digits seen"); | |
2716 | rust_lex_exception_test ("b'\\u{0}'", "Unicode escape in byte literal"); | |
2717 | rust_lex_exception_test ("'\\x0'", "Not enough hex digits seen"); | |
2718 | rust_lex_exception_test ("'\\u0'", "Missing '{' in Unicode escape"); | |
2719 | rust_lex_exception_test ("'\\u{0", "Missing '}' in Unicode escape"); | |
2720 | rust_lex_exception_test ("'\\u{0000007}", "Overlong hex escape"); | |
2721 | rust_lex_exception_test ("'\\u{}", "Not enough hex digits seen"); | |
2722 | rust_lex_exception_test ("'\\Q'", "Invalid escape \\Q in literal"); | |
2723 | rust_lex_exception_test ("b'\\Q'", "Invalid escape \\Q in literal"); | |
2724 | ||
2725 | rust_lex_int_test ("23", 23, DECIMAL_INTEGER); | |
2726 | rust_lex_int_test ("2_344__29", 234429, INTEGER); | |
2727 | rust_lex_int_test ("0x1f", 0x1f, INTEGER); | |
2728 | rust_lex_int_test ("23usize", 23, INTEGER); | |
2729 | rust_lex_int_test ("23i32", 23, INTEGER); | |
2730 | rust_lex_int_test ("0x1_f", 0x1f, INTEGER); | |
2731 | rust_lex_int_test ("0b1_101011__", 0x6b, INTEGER); | |
2732 | rust_lex_int_test ("0o001177i64", 639, INTEGER); | |
2733 | ||
2734 | rust_lex_test_trailing_dot (); | |
2735 | ||
2736 | rust_lex_test_one ("23.", FLOAT); | |
2737 | rust_lex_test_one ("23.99f32", FLOAT); | |
2738 | rust_lex_test_one ("23e7", FLOAT); | |
2739 | rust_lex_test_one ("23E-7", FLOAT); | |
2740 | rust_lex_test_one ("23e+7", FLOAT); | |
2741 | rust_lex_test_one ("23.99e+7f64", FLOAT); | |
2742 | rust_lex_test_one ("23.82f32", FLOAT); | |
2743 | ||
2744 | rust_lex_stringish_test ("hibob", "hibob", IDENT); | |
2745 | rust_lex_stringish_test ("hibob__93", "hibob__93", IDENT); | |
2746 | rust_lex_stringish_test ("thread", "thread", IDENT); | |
2747 | ||
2748 | rust_lex_stringish_test ("\"string\"", "string", STRING); | |
2749 | rust_lex_stringish_test ("\"str\\ting\"", "str\ting", STRING); | |
2750 | rust_lex_stringish_test ("\"str\\\"ing\"", "str\"ing", STRING); | |
2751 | rust_lex_stringish_test ("r\"str\\ing\"", "str\\ing", STRING); | |
2752 | rust_lex_stringish_test ("r#\"str\\ting\"#", "str\\ting", STRING); | |
2753 | rust_lex_stringish_test ("r###\"str\\\"ing\"###", "str\\\"ing", STRING); | |
2754 | ||
2755 | rust_lex_stringish_test ("b\"string\"", "string", BYTESTRING); | |
2756 | rust_lex_stringish_test ("b\"\x73tring\"", "string", BYTESTRING); | |
2757 | rust_lex_stringish_test ("b\"str\\\"ing\"", "str\"ing", BYTESTRING); | |
2758 | rust_lex_stringish_test ("br####\"\\x73tring\"####", "\\x73tring", | |
2759 | BYTESTRING); | |
2760 | ||
2761 | for (i = 0; i < ARRAY_SIZE (identifier_tokens); ++i) | |
2762 | rust_lex_test_one (identifier_tokens[i].name, identifier_tokens[i].value); | |
2763 | ||
2764 | for (i = 0; i < ARRAY_SIZE (operator_tokens); ++i) | |
2765 | rust_lex_test_one (operator_tokens[i].name, operator_tokens[i].value); | |
2766 | ||
2767 | rust_lex_test_completion (); | |
2768 | rust_lex_test_push_back (); | |
2769 | ||
c44af4eb TT |
2770 | unit_testing = 0; |
2771 | } | |
2772 | ||
2773 | #endif /* GDB_SELF_TEST */ | |
2774 | ||
2775 | void | |
2776 | _initialize_rust_exp (void) | |
2777 | { | |
2778 | int code = regcomp (&number_regex, number_regex_text, REG_EXTENDED); | |
2779 | /* If the regular expression was incorrect, it was a programming | |
2780 | error. */ | |
2781 | gdb_assert (code == 0); | |
2782 | ||
2783 | #if GDB_SELF_TEST | |
2784 | register_self_test (rust_lex_tests); | |
2785 | #endif | |
2786 | } |