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 | ||
c44af4eb | 42 | struct rust_op; |
3232fabd | 43 | typedef std::vector<const struct rust_op *> rust_op_vector; |
c44af4eb TT |
44 | |
45 | /* A typed integer constant. */ | |
46 | ||
47 | struct typed_val_int | |
48 | { | |
49 | LONGEST val; | |
50 | struct type *type; | |
51 | }; | |
52 | ||
53 | /* A typed floating point constant. */ | |
54 | ||
55 | struct typed_val_float | |
56 | { | |
edd079d9 | 57 | gdb_byte val[16]; |
c44af4eb TT |
58 | struct type *type; |
59 | }; | |
60 | ||
61 | /* An identifier and an expression. This is used to represent one | |
62 | element of a struct initializer. */ | |
63 | ||
64 | struct set_field | |
65 | { | |
66 | struct stoken name; | |
67 | const struct rust_op *init; | |
68 | }; | |
69 | ||
3232fabd | 70 | typedef std::vector<set_field> rust_set_vector; |
c44af4eb TT |
71 | |
72 | static int rustyylex (void); | |
73 | static void rust_push_back (char c); | |
74 | static const char *rust_copy_name (const char *, int); | |
75 | static struct stoken rust_concat3 (const char *, const char *, const char *); | |
76 | static struct stoken make_stoken (const char *); | |
77 | static struct block_symbol rust_lookup_symbol (const char *name, | |
78 | const struct block *block, | |
79 | const domain_enum domain); | |
80 | static struct type *rust_lookup_type (const char *name, | |
81 | const struct block *block); | |
82 | static struct type *rust_type (const char *name); | |
83 | ||
84 | static const struct rust_op *crate_name (const struct rust_op *name); | |
85 | static const struct rust_op *super_name (const struct rust_op *name, | |
86 | unsigned int n_supers); | |
87 | ||
88 | static const struct rust_op *ast_operation (enum exp_opcode opcode, | |
89 | const struct rust_op *left, | |
90 | const struct rust_op *right); | |
91 | static const struct rust_op *ast_compound_assignment | |
92 | (enum exp_opcode opcode, const struct rust_op *left, | |
93 | const struct rust_op *rust_op); | |
94 | static const struct rust_op *ast_literal (struct typed_val_int val); | |
95 | static const struct rust_op *ast_dliteral (struct typed_val_float val); | |
96 | static const struct rust_op *ast_structop (const struct rust_op *left, | |
97 | const char *name, | |
98 | int completing); | |
99 | static const struct rust_op *ast_structop_anonymous | |
100 | (const struct rust_op *left, struct typed_val_int number); | |
101 | static const struct rust_op *ast_unary (enum exp_opcode opcode, | |
102 | const struct rust_op *expr); | |
103 | static const struct rust_op *ast_cast (const struct rust_op *expr, | |
104 | const struct rust_op *type); | |
105 | static const struct rust_op *ast_call_ish (enum exp_opcode opcode, | |
106 | const struct rust_op *expr, | |
3232fabd | 107 | rust_op_vector *params); |
c44af4eb | 108 | static const struct rust_op *ast_path (struct stoken name, |
3232fabd | 109 | rust_op_vector *params); |
c44af4eb TT |
110 | static const struct rust_op *ast_string (struct stoken str); |
111 | static const struct rust_op *ast_struct (const struct rust_op *name, | |
3232fabd | 112 | rust_set_vector *fields); |
c44af4eb TT |
113 | static const struct rust_op *ast_range (const struct rust_op *lhs, |
114 | const struct rust_op *rhs); | |
115 | static const struct rust_op *ast_array_type (const struct rust_op *lhs, | |
116 | struct typed_val_int val); | |
117 | static const struct rust_op *ast_slice_type (const struct rust_op *type); | |
118 | static const struct rust_op *ast_reference_type (const struct rust_op *type); | |
119 | static const struct rust_op *ast_pointer_type (const struct rust_op *type, | |
120 | int is_mut); | |
121 | static const struct rust_op *ast_function_type (const struct rust_op *result, | |
3232fabd TT |
122 | rust_op_vector *params); |
123 | static const struct rust_op *ast_tuple_type (rust_op_vector *params); | |
c44af4eb | 124 | |
3232fabd | 125 | /* The current rust parser. */ |
c44af4eb | 126 | |
3232fabd TT |
127 | struct rust_parser; |
128 | static rust_parser *current_parser; | |
c44af4eb TT |
129 | |
130 | /* A regular expression for matching Rust numbers. This is split up | |
131 | since it is very long and this gives us a way to comment the | |
132 | sections. */ | |
133 | ||
134 | static const char *number_regex_text = | |
135 | /* subexpression 1: allows use of alternation, otherwise uninteresting */ | |
136 | "^(" | |
137 | /* First comes floating point. */ | |
138 | /* Recognize number after the decimal point, with optional | |
139 | exponent and optional type suffix. | |
140 | subexpression 2: allows "?", otherwise uninteresting | |
141 | subexpression 3: if present, type suffix | |
142 | */ | |
143 | "[0-9][0-9_]*\\.[0-9][0-9_]*([eE][-+]?[0-9][0-9_]*)?(f32|f64)?" | |
144 | #define FLOAT_TYPE1 3 | |
145 | "|" | |
146 | /* Recognize exponent without decimal point, with optional type | |
147 | suffix. | |
148 | subexpression 4: if present, type suffix | |
149 | */ | |
150 | #define FLOAT_TYPE2 4 | |
151 | "[0-9][0-9_]*[eE][-+]?[0-9][0-9_]*(f32|f64)?" | |
152 | "|" | |
153 | /* "23." is a valid floating point number, but "23.e5" and | |
154 | "23.f32" are not. So, handle the trailing-. case | |
155 | separately. */ | |
156 | "[0-9][0-9_]*\\." | |
157 | "|" | |
158 | /* Finally come integers. | |
159 | subexpression 5: text of integer | |
160 | subexpression 6: if present, type suffix | |
161 | subexpression 7: allows use of alternation, otherwise uninteresting | |
162 | */ | |
163 | #define INT_TEXT 5 | |
164 | #define INT_TYPE 6 | |
165 | "(0x[a-fA-F0-9_]+|0o[0-7_]+|0b[01_]+|[0-9][0-9_]*)" | |
166 | "([iu](size|8|16|32|64))?" | |
167 | ")"; | |
168 | /* The number of subexpressions to allocate space for, including the | |
169 | "0th" whole match subexpression. */ | |
170 | #define NUM_SUBEXPRESSIONS 8 | |
171 | ||
172 | /* The compiled number-matching regex. */ | |
173 | ||
174 | static regex_t number_regex; | |
175 | ||
3232fabd TT |
176 | /* Obstack for data temporarily allocated during parsing. Points to |
177 | the obstack in the rust_parser, or to a temporary obstack during | |
178 | unit testing. */ | |
179 | ||
180 | static auto_obstack *work_obstack; | |
181 | ||
182 | /* An instance of this is created before parsing, and destroyed when | |
183 | parsing is finished. */ | |
184 | ||
185 | struct rust_parser | |
186 | { | |
187 | rust_parser (struct parser_state *state) | |
188 | : rust_ast (nullptr), | |
189 | pstate (state) | |
190 | { | |
191 | gdb_assert (current_parser == nullptr); | |
192 | current_parser = this; | |
193 | work_obstack = &obstack; | |
194 | } | |
195 | ||
196 | ~rust_parser () | |
197 | { | |
198 | /* Clean up the globals we set. */ | |
199 | current_parser = nullptr; | |
200 | work_obstack = nullptr; | |
201 | } | |
202 | ||
203 | /* Create a new rust_set_vector. The storage for the new vector is | |
204 | managed by this class. */ | |
205 | rust_set_vector *new_set_vector () | |
206 | { | |
207 | rust_set_vector *result = new rust_set_vector; | |
208 | set_vectors.push_back (std::unique_ptr<rust_set_vector> (result)); | |
209 | return result; | |
210 | } | |
211 | ||
212 | /* Create a new rust_ops_vector. The storage for the new vector is | |
213 | managed by this class. */ | |
214 | rust_op_vector *new_op_vector () | |
215 | { | |
216 | rust_op_vector *result = new rust_op_vector; | |
217 | op_vectors.push_back (std::unique_ptr<rust_op_vector> (result)); | |
218 | return result; | |
219 | } | |
220 | ||
221 | /* Return the parser's language. */ | |
222 | const struct language_defn *language () const | |
223 | { | |
224 | return parse_language (pstate); | |
225 | } | |
226 | ||
227 | /* Return the parser's gdbarch. */ | |
228 | struct gdbarch *arch () const | |
229 | { | |
230 | return parse_gdbarch (pstate); | |
231 | } | |
232 | ||
233 | /* A pointer to this is installed globally. */ | |
234 | auto_obstack obstack; | |
235 | ||
236 | /* Result of parsing. Points into obstack. */ | |
237 | const struct rust_op *rust_ast; | |
238 | ||
239 | /* This keeps track of the various vectors we allocate. */ | |
240 | std::vector<std::unique_ptr<rust_set_vector>> set_vectors; | |
241 | std::vector<std::unique_ptr<rust_op_vector>> op_vectors; | |
242 | ||
243 | /* The parser state gdb gave us. */ | |
244 | struct parser_state *pstate; | |
245 | }; | |
c44af4eb TT |
246 | |
247 | %} | |
248 | ||
249 | %union | |
250 | { | |
251 | /* A typed integer constant. */ | |
252 | struct typed_val_int typed_val_int; | |
253 | ||
254 | /* A typed floating point constant. */ | |
255 | struct typed_val_float typed_val_float; | |
256 | ||
257 | /* An identifier or string. */ | |
258 | struct stoken sval; | |
259 | ||
260 | /* A token representing an opcode, like "==". */ | |
261 | enum exp_opcode opcode; | |
262 | ||
263 | /* A list of expressions; for example, the arguments to a function | |
264 | call. */ | |
3232fabd | 265 | rust_op_vector *params; |
c44af4eb TT |
266 | |
267 | /* A list of field initializers. */ | |
3232fabd | 268 | rust_set_vector *field_inits; |
c44af4eb TT |
269 | |
270 | /* A single field initializer. */ | |
271 | struct set_field one_field_init; | |
272 | ||
273 | /* An expression. */ | |
274 | const struct rust_op *op; | |
275 | ||
276 | /* A plain integer, for example used to count the number of | |
277 | "super::" prefixes on a path. */ | |
278 | unsigned int depth; | |
279 | } | |
280 | ||
281 | %{ | |
282 | ||
283 | /* Rust AST operations. We build a tree of these; then lower them | |
284 | to gdb expressions when parsing has completed. */ | |
285 | ||
286 | struct rust_op | |
287 | { | |
288 | /* The opcode. */ | |
289 | enum exp_opcode opcode; | |
290 | /* If OPCODE is OP_TYPE, then this holds information about what type | |
291 | is described by this node. */ | |
292 | enum type_code typecode; | |
293 | /* Indicates whether OPCODE actually represents a compound | |
294 | assignment. For example, if OPCODE is GTGT and this is false, | |
295 | then this rust_op represents an ordinary ">>"; but if this is | |
296 | true, then this rust_op represents ">>=". Unused in other | |
297 | cases. */ | |
298 | unsigned int compound_assignment : 1; | |
299 | /* Only used by a field expression; if set, indicates that the field | |
300 | name occurred at the end of the expression and is eligible for | |
301 | completion. */ | |
302 | unsigned int completing : 1; | |
303 | /* Operands of expression. Which one is used and how depends on the | |
304 | particular opcode. */ | |
305 | RUSTSTYPE left; | |
306 | RUSTSTYPE right; | |
307 | }; | |
308 | ||
309 | %} | |
310 | ||
311 | %token <sval> GDBVAR | |
312 | %token <sval> IDENT | |
313 | %token <sval> COMPLETE | |
314 | %token <typed_val_int> INTEGER | |
315 | %token <typed_val_int> DECIMAL_INTEGER | |
316 | %token <sval> STRING | |
317 | %token <sval> BYTESTRING | |
318 | %token <typed_val_float> FLOAT | |
319 | %token <opcode> COMPOUND_ASSIGN | |
320 | ||
321 | /* Keyword tokens. */ | |
322 | %token <voidval> KW_AS | |
323 | %token <voidval> KW_IF | |
324 | %token <voidval> KW_TRUE | |
325 | %token <voidval> KW_FALSE | |
326 | %token <voidval> KW_SUPER | |
327 | %token <voidval> KW_SELF | |
328 | %token <voidval> KW_MUT | |
329 | %token <voidval> KW_EXTERN | |
330 | %token <voidval> KW_CONST | |
331 | %token <voidval> KW_FN | |
cdf5a07c | 332 | %token <voidval> KW_SIZEOF |
c44af4eb TT |
333 | |
334 | /* Operator tokens. */ | |
335 | %token <voidval> DOTDOT | |
336 | %token <voidval> OROR | |
337 | %token <voidval> ANDAND | |
338 | %token <voidval> EQEQ | |
339 | %token <voidval> NOTEQ | |
340 | %token <voidval> LTEQ | |
341 | %token <voidval> GTEQ | |
342 | %token <voidval> LSH RSH | |
343 | %token <voidval> COLONCOLON | |
344 | %token <voidval> ARROW | |
345 | ||
346 | %type <op> type | |
347 | %type <op> path_for_expr | |
348 | %type <op> identifier_path_for_expr | |
349 | %type <op> path_for_type | |
350 | %type <op> identifier_path_for_type | |
351 | %type <op> just_identifiers_for_type | |
352 | ||
353 | %type <params> maybe_type_list | |
354 | %type <params> type_list | |
355 | ||
356 | %type <depth> super_path | |
357 | ||
358 | %type <op> literal | |
359 | %type <op> expr | |
360 | %type <op> field_expr | |
361 | %type <op> idx_expr | |
362 | %type <op> unop_expr | |
363 | %type <op> binop_expr | |
364 | %type <op> binop_expr_expr | |
365 | %type <op> type_cast_expr | |
366 | %type <op> assignment_expr | |
367 | %type <op> compound_assignment_expr | |
368 | %type <op> paren_expr | |
369 | %type <op> call_expr | |
370 | %type <op> path_expr | |
371 | %type <op> tuple_expr | |
372 | %type <op> unit_expr | |
373 | %type <op> struct_expr | |
374 | %type <op> array_expr | |
375 | %type <op> range_expr | |
376 | ||
377 | %type <params> expr_list | |
378 | %type <params> maybe_expr_list | |
379 | %type <params> paren_expr_list | |
380 | ||
381 | %type <field_inits> struct_expr_list | |
382 | %type <one_field_init> struct_expr_tail | |
383 | ||
384 | /* Precedence. */ | |
385 | %nonassoc DOTDOT | |
386 | %right '=' COMPOUND_ASSIGN | |
387 | %left OROR | |
388 | %left ANDAND | |
389 | %nonassoc EQEQ NOTEQ '<' '>' LTEQ GTEQ | |
390 | %left '|' | |
391 | %left '^' | |
392 | %left '&' | |
393 | %left LSH RSH | |
394 | %left '@' | |
395 | %left '+' '-' | |
396 | %left '*' '/' '%' | |
397 | /* These could be %precedence in Bison, but that isn't a yacc | |
398 | feature. */ | |
399 | %left KW_AS | |
400 | %left UNARY | |
401 | %left '[' '.' '(' | |
402 | ||
403 | %% | |
404 | ||
405 | start: | |
406 | expr | |
407 | { | |
408 | /* If we are completing and see a valid parse, | |
409 | rust_ast will already have been set. */ | |
3232fabd TT |
410 | if (current_parser->rust_ast == NULL) |
411 | current_parser->rust_ast = $1; | |
c44af4eb TT |
412 | } |
413 | ; | |
414 | ||
415 | /* Note that the Rust grammar includes a method_call_expr, but we | |
416 | handle this differently, to avoid a shift/reduce conflict with | |
417 | call_expr. */ | |
418 | expr: | |
419 | literal | |
420 | | path_expr | |
421 | | tuple_expr | |
422 | | unit_expr | |
423 | | struct_expr | |
424 | | field_expr | |
425 | | array_expr | |
426 | | idx_expr | |
427 | | range_expr | |
cdf5a07c | 428 | | unop_expr /* Must precede call_expr because of ambiguity with sizeof. */ |
c44af4eb TT |
429 | | binop_expr |
430 | | paren_expr | |
431 | | call_expr | |
432 | ; | |
433 | ||
434 | tuple_expr: | |
435 | '(' expr ',' maybe_expr_list ')' | |
436 | { | |
3232fabd | 437 | $4->push_back ($2); |
c44af4eb TT |
438 | error (_("Tuple expressions not supported yet")); |
439 | } | |
440 | ; | |
441 | ||
442 | unit_expr: | |
443 | '(' ')' | |
444 | { | |
445 | struct typed_val_int val; | |
446 | ||
447 | val.type | |
3232fabd TT |
448 | = language_lookup_primitive_type (current_parser->language (), |
449 | current_parser->arch (), | |
c44af4eb TT |
450 | "()"); |
451 | val.val = 0; | |
452 | $$ = ast_literal (val); | |
453 | } | |
454 | ; | |
455 | ||
456 | /* To avoid a shift/reduce conflict with call_expr, we don't handle | |
457 | tuple struct expressions here, but instead when examining the | |
458 | AST. */ | |
459 | struct_expr: | |
460 | path_for_expr '{' struct_expr_list '}' | |
461 | { $$ = ast_struct ($1, $3); } | |
462 | ; | |
463 | ||
464 | struct_expr_tail: | |
465 | DOTDOT expr | |
466 | { | |
467 | struct set_field sf; | |
468 | ||
469 | sf.name.ptr = NULL; | |
470 | sf.name.length = 0; | |
471 | sf.init = $2; | |
472 | ||
473 | $$ = sf; | |
474 | } | |
475 | | IDENT ':' expr | |
476 | { | |
477 | struct set_field sf; | |
478 | ||
479 | sf.name = $1; | |
480 | sf.init = $3; | |
481 | $$ = sf; | |
482 | } | |
483 | ; | |
484 | ||
c44af4eb | 485 | struct_expr_list: |
12df5c00 TT |
486 | /* %empty */ |
487 | { | |
3232fabd | 488 | $$ = current_parser->new_set_vector (); |
12df5c00 TT |
489 | } |
490 | | struct_expr_tail | |
c44af4eb | 491 | { |
3232fabd TT |
492 | rust_set_vector *result = current_parser->new_set_vector (); |
493 | result->push_back ($1); | |
c44af4eb TT |
494 | $$ = result; |
495 | } | |
496 | | IDENT ':' expr ',' struct_expr_list | |
497 | { | |
498 | struct set_field sf; | |
499 | ||
500 | sf.name = $1; | |
501 | sf.init = $3; | |
3232fabd | 502 | $5->push_back (sf); |
c44af4eb TT |
503 | $$ = $5; |
504 | } | |
505 | ; | |
506 | ||
507 | array_expr: | |
508 | '[' KW_MUT expr_list ']' | |
509 | { $$ = ast_call_ish (OP_ARRAY, NULL, $3); } | |
510 | | '[' expr_list ']' | |
511 | { $$ = ast_call_ish (OP_ARRAY, NULL, $2); } | |
512 | | '[' KW_MUT expr ';' expr ']' | |
513 | { $$ = ast_operation (OP_RUST_ARRAY, $3, $5); } | |
514 | | '[' expr ';' expr ']' | |
515 | { $$ = ast_operation (OP_RUST_ARRAY, $2, $4); } | |
516 | ; | |
517 | ||
518 | range_expr: | |
519 | expr DOTDOT | |
520 | { $$ = ast_range ($1, NULL); } | |
521 | | expr DOTDOT expr | |
522 | { $$ = ast_range ($1, $3); } | |
523 | | DOTDOT expr | |
524 | { $$ = ast_range (NULL, $2); } | |
525 | | DOTDOT | |
526 | { $$ = ast_range (NULL, NULL); } | |
527 | ; | |
528 | ||
529 | literal: | |
530 | INTEGER | |
531 | { $$ = ast_literal ($1); } | |
532 | | DECIMAL_INTEGER | |
533 | { $$ = ast_literal ($1); } | |
534 | | FLOAT | |
535 | { $$ = ast_dliteral ($1); } | |
536 | | STRING | |
537 | { | |
538 | const struct rust_op *str = ast_string ($1); | |
c44af4eb TT |
539 | struct set_field field; |
540 | struct typed_val_int val; | |
541 | struct stoken token; | |
542 | ||
3232fabd | 543 | rust_set_vector *fields = current_parser->new_set_vector (); |
c44af4eb TT |
544 | |
545 | /* Wrap the raw string in the &str struct. */ | |
546 | field.name.ptr = "data_ptr"; | |
547 | field.name.length = strlen (field.name.ptr); | |
548 | field.init = ast_unary (UNOP_ADDR, ast_string ($1)); | |
3232fabd | 549 | fields->push_back (field); |
c44af4eb TT |
550 | |
551 | val.type = rust_type ("usize"); | |
552 | val.val = $1.length; | |
553 | ||
554 | field.name.ptr = "length"; | |
555 | field.name.length = strlen (field.name.ptr); | |
556 | field.init = ast_literal (val); | |
3232fabd | 557 | fields->push_back (field); |
c44af4eb TT |
558 | |
559 | token.ptr = "&str"; | |
560 | token.length = strlen (token.ptr); | |
561 | $$ = ast_struct (ast_path (token, NULL), fields); | |
562 | } | |
563 | | BYTESTRING | |
564 | { $$ = ast_string ($1); } | |
565 | | KW_TRUE | |
566 | { | |
567 | struct typed_val_int val; | |
568 | ||
3232fabd TT |
569 | val.type = language_bool_type (current_parser->language (), |
570 | current_parser->arch ()); | |
c44af4eb TT |
571 | val.val = 1; |
572 | $$ = ast_literal (val); | |
573 | } | |
574 | | KW_FALSE | |
575 | { | |
576 | struct typed_val_int val; | |
577 | ||
3232fabd TT |
578 | val.type = language_bool_type (current_parser->language (), |
579 | current_parser->arch ()); | |
c44af4eb TT |
580 | val.val = 0; |
581 | $$ = ast_literal (val); | |
582 | } | |
583 | ; | |
584 | ||
585 | field_expr: | |
586 | expr '.' IDENT | |
587 | { $$ = ast_structop ($1, $3.ptr, 0); } | |
588 | | expr '.' COMPLETE | |
589 | { | |
590 | $$ = ast_structop ($1, $3.ptr, 1); | |
3232fabd | 591 | current_parser->rust_ast = $$; |
c44af4eb TT |
592 | } |
593 | | expr '.' DECIMAL_INTEGER | |
594 | { $$ = ast_structop_anonymous ($1, $3); } | |
595 | ; | |
596 | ||
597 | idx_expr: | |
598 | expr '[' expr ']' | |
599 | { $$ = ast_operation (BINOP_SUBSCRIPT, $1, $3); } | |
600 | ; | |
601 | ||
602 | unop_expr: | |
603 | '+' expr %prec UNARY | |
604 | { $$ = ast_unary (UNOP_PLUS, $2); } | |
605 | ||
606 | | '-' expr %prec UNARY | |
607 | { $$ = ast_unary (UNOP_NEG, $2); } | |
608 | ||
609 | | '!' expr %prec UNARY | |
610 | { | |
611 | /* Note that we provide a Rust-specific evaluator | |
612 | override for UNOP_COMPLEMENT, so it can do the | |
613 | right thing for both bool and integral | |
614 | values. */ | |
615 | $$ = ast_unary (UNOP_COMPLEMENT, $2); | |
616 | } | |
617 | ||
618 | | '*' expr %prec UNARY | |
619 | { $$ = ast_unary (UNOP_IND, $2); } | |
620 | ||
621 | | '&' expr %prec UNARY | |
622 | { $$ = ast_unary (UNOP_ADDR, $2); } | |
623 | ||
624 | | '&' KW_MUT expr %prec UNARY | |
625 | { $$ = ast_unary (UNOP_ADDR, $3); } | |
cdf5a07c MG |
626 | | KW_SIZEOF '(' expr ')' %prec UNARY |
627 | { $$ = ast_unary (UNOP_SIZEOF, $3); } | |
c44af4eb TT |
628 | ; |
629 | ||
630 | binop_expr: | |
631 | binop_expr_expr | |
632 | | type_cast_expr | |
633 | | assignment_expr | |
634 | | compound_assignment_expr | |
635 | ; | |
636 | ||
637 | binop_expr_expr: | |
638 | expr '*' expr | |
639 | { $$ = ast_operation (BINOP_MUL, $1, $3); } | |
640 | ||
641 | | expr '@' expr | |
642 | { $$ = ast_operation (BINOP_REPEAT, $1, $3); } | |
643 | ||
644 | | expr '/' expr | |
645 | { $$ = ast_operation (BINOP_DIV, $1, $3); } | |
646 | ||
647 | | expr '%' expr | |
648 | { $$ = ast_operation (BINOP_REM, $1, $3); } | |
649 | ||
650 | | expr '<' expr | |
651 | { $$ = ast_operation (BINOP_LESS, $1, $3); } | |
652 | ||
653 | | expr '>' expr | |
654 | { $$ = ast_operation (BINOP_GTR, $1, $3); } | |
655 | ||
656 | | expr '&' expr | |
657 | { $$ = ast_operation (BINOP_BITWISE_AND, $1, $3); } | |
658 | ||
659 | | expr '|' expr | |
660 | { $$ = ast_operation (BINOP_BITWISE_IOR, $1, $3); } | |
661 | ||
662 | | expr '^' expr | |
663 | { $$ = ast_operation (BINOP_BITWISE_XOR, $1, $3); } | |
664 | ||
665 | | expr '+' expr | |
666 | { $$ = ast_operation (BINOP_ADD, $1, $3); } | |
667 | ||
668 | | expr '-' expr | |
669 | { $$ = ast_operation (BINOP_SUB, $1, $3); } | |
670 | ||
671 | | expr OROR expr | |
672 | { $$ = ast_operation (BINOP_LOGICAL_OR, $1, $3); } | |
673 | ||
674 | | expr ANDAND expr | |
675 | { $$ = ast_operation (BINOP_LOGICAL_AND, $1, $3); } | |
676 | ||
677 | | expr EQEQ expr | |
678 | { $$ = ast_operation (BINOP_EQUAL, $1, $3); } | |
679 | ||
680 | | expr NOTEQ expr | |
681 | { $$ = ast_operation (BINOP_NOTEQUAL, $1, $3); } | |
682 | ||
683 | | expr LTEQ expr | |
684 | { $$ = ast_operation (BINOP_LEQ, $1, $3); } | |
685 | ||
686 | | expr GTEQ expr | |
687 | { $$ = ast_operation (BINOP_GEQ, $1, $3); } | |
688 | ||
689 | | expr LSH expr | |
690 | { $$ = ast_operation (BINOP_LSH, $1, $3); } | |
691 | ||
692 | | expr RSH expr | |
693 | { $$ = ast_operation (BINOP_RSH, $1, $3); } | |
694 | ; | |
695 | ||
696 | type_cast_expr: | |
697 | expr KW_AS type | |
698 | { $$ = ast_cast ($1, $3); } | |
699 | ; | |
700 | ||
701 | assignment_expr: | |
702 | expr '=' expr | |
703 | { $$ = ast_operation (BINOP_ASSIGN, $1, $3); } | |
704 | ; | |
705 | ||
706 | compound_assignment_expr: | |
707 | expr COMPOUND_ASSIGN expr | |
708 | { $$ = ast_compound_assignment ($2, $1, $3); } | |
709 | ||
710 | ; | |
711 | ||
712 | paren_expr: | |
713 | '(' expr ')' | |
714 | { $$ = $2; } | |
715 | ; | |
716 | ||
717 | expr_list: | |
718 | expr | |
719 | { | |
3232fabd TT |
720 | $$ = current_parser->new_op_vector (); |
721 | $$->push_back ($1); | |
c44af4eb TT |
722 | } |
723 | | expr_list ',' expr | |
724 | { | |
3232fabd | 725 | $1->push_back ($3); |
c44af4eb TT |
726 | $$ = $1; |
727 | } | |
728 | ; | |
729 | ||
730 | maybe_expr_list: | |
731 | /* %empty */ | |
732 | { | |
733 | /* The result can't be NULL. */ | |
3232fabd | 734 | $$ = current_parser->new_op_vector (); |
c44af4eb TT |
735 | } |
736 | | expr_list | |
737 | { $$ = $1; } | |
738 | ; | |
739 | ||
740 | paren_expr_list: | |
741 | '(' | |
742 | maybe_expr_list | |
743 | ')' | |
744 | { $$ = $2; } | |
745 | ; | |
746 | ||
747 | call_expr: | |
748 | expr paren_expr_list | |
749 | { $$ = ast_call_ish (OP_FUNCALL, $1, $2); } | |
750 | ; | |
751 | ||
752 | maybe_self_path: | |
753 | /* %empty */ | |
754 | | KW_SELF COLONCOLON | |
755 | ; | |
756 | ||
757 | super_path: | |
758 | KW_SUPER COLONCOLON | |
759 | { $$ = 1; } | |
760 | | super_path KW_SUPER COLONCOLON | |
761 | { $$ = $1 + 1; } | |
762 | ; | |
763 | ||
764 | path_expr: | |
765 | path_for_expr | |
766 | { $$ = $1; } | |
767 | | GDBVAR | |
768 | { $$ = ast_path ($1, NULL); } | |
769 | | KW_SELF | |
770 | { $$ = ast_path (make_stoken ("self"), NULL); } | |
771 | ; | |
772 | ||
773 | path_for_expr: | |
774 | identifier_path_for_expr | |
775 | | KW_SELF COLONCOLON identifier_path_for_expr | |
776 | { $$ = super_name ($3, 0); } | |
777 | | maybe_self_path super_path identifier_path_for_expr | |
778 | { $$ = super_name ($3, $2); } | |
779 | | COLONCOLON identifier_path_for_expr | |
780 | { $$ = crate_name ($2); } | |
781 | | KW_EXTERN identifier_path_for_expr | |
782 | { | |
783 | /* This is a gdb extension to make it possible to | |
784 | refer to items in other crates. It just bypasses | |
785 | adding the current crate to the front of the | |
786 | name. */ | |
787 | $$ = ast_path (rust_concat3 ("::", $2->left.sval.ptr, NULL), | |
788 | $2->right.params); | |
789 | } | |
790 | ; | |
791 | ||
792 | identifier_path_for_expr: | |
793 | IDENT | |
794 | { $$ = ast_path ($1, NULL); } | |
795 | | identifier_path_for_expr COLONCOLON IDENT | |
796 | { | |
797 | $$ = ast_path (rust_concat3 ($1->left.sval.ptr, "::", | |
798 | $3.ptr), | |
799 | NULL); | |
800 | } | |
801 | | identifier_path_for_expr COLONCOLON '<' type_list '>' | |
802 | { $$ = ast_path ($1->left.sval, $4); } | |
803 | | identifier_path_for_expr COLONCOLON '<' type_list RSH | |
804 | { | |
805 | $$ = ast_path ($1->left.sval, $4); | |
806 | rust_push_back ('>'); | |
807 | } | |
808 | ; | |
809 | ||
810 | path_for_type: | |
811 | identifier_path_for_type | |
812 | | KW_SELF COLONCOLON identifier_path_for_type | |
813 | { $$ = super_name ($3, 0); } | |
814 | | maybe_self_path super_path identifier_path_for_type | |
815 | { $$ = super_name ($3, $2); } | |
816 | | COLONCOLON identifier_path_for_type | |
817 | { $$ = crate_name ($2); } | |
818 | | KW_EXTERN identifier_path_for_type | |
819 | { | |
820 | /* This is a gdb extension to make it possible to | |
821 | refer to items in other crates. It just bypasses | |
822 | adding the current crate to the front of the | |
823 | name. */ | |
824 | $$ = ast_path (rust_concat3 ("::", $2->left.sval.ptr, NULL), | |
825 | $2->right.params); | |
826 | } | |
827 | ; | |
828 | ||
829 | just_identifiers_for_type: | |
830 | IDENT | |
831 | { $$ = ast_path ($1, NULL); } | |
832 | | just_identifiers_for_type COLONCOLON IDENT | |
833 | { | |
834 | $$ = ast_path (rust_concat3 ($1->left.sval.ptr, "::", | |
835 | $3.ptr), | |
836 | NULL); | |
837 | } | |
838 | ; | |
839 | ||
840 | identifier_path_for_type: | |
841 | just_identifiers_for_type | |
842 | | just_identifiers_for_type '<' type_list '>' | |
843 | { $$ = ast_path ($1->left.sval, $3); } | |
844 | | just_identifiers_for_type '<' type_list RSH | |
845 | { | |
846 | $$ = ast_path ($1->left.sval, $3); | |
847 | rust_push_back ('>'); | |
848 | } | |
849 | ; | |
850 | ||
851 | type: | |
852 | path_for_type | |
853 | | '[' type ';' INTEGER ']' | |
854 | { $$ = ast_array_type ($2, $4); } | |
855 | | '[' type ';' DECIMAL_INTEGER ']' | |
856 | { $$ = ast_array_type ($2, $4); } | |
857 | | '&' '[' type ']' | |
858 | { $$ = ast_slice_type ($3); } | |
859 | | '&' type | |
860 | { $$ = ast_reference_type ($2); } | |
861 | | '*' KW_MUT type | |
862 | { $$ = ast_pointer_type ($3, 1); } | |
863 | | '*' KW_CONST type | |
864 | { $$ = ast_pointer_type ($3, 0); } | |
865 | | KW_FN '(' maybe_type_list ')' ARROW type | |
866 | { $$ = ast_function_type ($6, $3); } | |
867 | | '(' maybe_type_list ')' | |
868 | { $$ = ast_tuple_type ($2); } | |
869 | ; | |
870 | ||
871 | maybe_type_list: | |
872 | /* %empty */ | |
873 | { $$ = NULL; } | |
874 | | type_list | |
875 | { $$ = $1; } | |
876 | ; | |
877 | ||
878 | type_list: | |
879 | type | |
880 | { | |
3232fabd TT |
881 | rust_op_vector *result = current_parser->new_op_vector (); |
882 | result->push_back ($1); | |
c44af4eb TT |
883 | $$ = result; |
884 | } | |
885 | | type_list ',' type | |
886 | { | |
3232fabd | 887 | $1->push_back ($3); |
c44af4eb TT |
888 | $$ = $1; |
889 | } | |
890 | ; | |
891 | ||
892 | %% | |
893 | ||
894 | /* A struct of this type is used to describe a token. */ | |
895 | ||
896 | struct token_info | |
897 | { | |
898 | const char *name; | |
899 | int value; | |
900 | enum exp_opcode opcode; | |
901 | }; | |
902 | ||
903 | /* Identifier tokens. */ | |
904 | ||
905 | static const struct token_info identifier_tokens[] = | |
906 | { | |
907 | { "as", KW_AS, OP_NULL }, | |
908 | { "false", KW_FALSE, OP_NULL }, | |
909 | { "if", 0, OP_NULL }, | |
910 | { "mut", KW_MUT, OP_NULL }, | |
911 | { "const", KW_CONST, OP_NULL }, | |
912 | { "self", KW_SELF, OP_NULL }, | |
913 | { "super", KW_SUPER, OP_NULL }, | |
914 | { "true", KW_TRUE, OP_NULL }, | |
915 | { "extern", KW_EXTERN, OP_NULL }, | |
916 | { "fn", KW_FN, OP_NULL }, | |
cdf5a07c | 917 | { "sizeof", KW_SIZEOF, OP_NULL }, |
c44af4eb TT |
918 | }; |
919 | ||
920 | /* Operator tokens, sorted longest first. */ | |
921 | ||
922 | static const struct token_info operator_tokens[] = | |
923 | { | |
924 | { ">>=", COMPOUND_ASSIGN, BINOP_RSH }, | |
925 | { "<<=", COMPOUND_ASSIGN, BINOP_LSH }, | |
926 | ||
927 | { "<<", LSH, OP_NULL }, | |
928 | { ">>", RSH, OP_NULL }, | |
929 | { "&&", ANDAND, OP_NULL }, | |
930 | { "||", OROR, OP_NULL }, | |
931 | { "==", EQEQ, OP_NULL }, | |
932 | { "!=", NOTEQ, OP_NULL }, | |
933 | { "<=", LTEQ, OP_NULL }, | |
934 | { ">=", GTEQ, OP_NULL }, | |
935 | { "+=", COMPOUND_ASSIGN, BINOP_ADD }, | |
936 | { "-=", COMPOUND_ASSIGN, BINOP_SUB }, | |
937 | { "*=", COMPOUND_ASSIGN, BINOP_MUL }, | |
938 | { "/=", COMPOUND_ASSIGN, BINOP_DIV }, | |
939 | { "%=", COMPOUND_ASSIGN, BINOP_REM }, | |
940 | { "&=", COMPOUND_ASSIGN, BINOP_BITWISE_AND }, | |
941 | { "|=", COMPOUND_ASSIGN, BINOP_BITWISE_IOR }, | |
942 | { "^=", COMPOUND_ASSIGN, BINOP_BITWISE_XOR }, | |
943 | ||
944 | { "::", COLONCOLON, OP_NULL }, | |
945 | { "..", DOTDOT, OP_NULL }, | |
946 | { "->", ARROW, OP_NULL } | |
947 | }; | |
948 | ||
949 | /* Helper function to copy to the name obstack. */ | |
950 | ||
951 | static const char * | |
952 | rust_copy_name (const char *name, int len) | |
953 | { | |
3232fabd | 954 | return (const char *) obstack_copy0 (work_obstack, name, len); |
c44af4eb TT |
955 | } |
956 | ||
957 | /* Helper function to make an stoken from a C string. */ | |
958 | ||
959 | static struct stoken | |
960 | make_stoken (const char *p) | |
961 | { | |
962 | struct stoken result; | |
963 | ||
964 | result.ptr = p; | |
965 | result.length = strlen (result.ptr); | |
966 | return result; | |
967 | } | |
968 | ||
969 | /* Helper function to concatenate three strings on the name | |
970 | obstack. */ | |
971 | ||
972 | static struct stoken | |
973 | rust_concat3 (const char *s1, const char *s2, const char *s3) | |
974 | { | |
3232fabd | 975 | return make_stoken (obconcat (work_obstack, s1, s2, s3, (char *) NULL)); |
c44af4eb TT |
976 | } |
977 | ||
978 | /* Return an AST node referring to NAME, but relative to the crate's | |
979 | name. */ | |
980 | ||
981 | static const struct rust_op * | |
982 | crate_name (const struct rust_op *name) | |
983 | { | |
03c85b11 | 984 | std::string crate = rust_crate_for_block (expression_context_block); |
c44af4eb TT |
985 | struct stoken result; |
986 | ||
987 | gdb_assert (name->opcode == OP_VAR_VALUE); | |
988 | ||
03c85b11 | 989 | if (crate.empty ()) |
c44af4eb | 990 | error (_("Could not find crate for current location")); |
3232fabd | 991 | result = make_stoken (obconcat (work_obstack, "::", crate.c_str (), "::", |
c44af4eb | 992 | name->left.sval.ptr, (char *) NULL)); |
c44af4eb TT |
993 | |
994 | return ast_path (result, name->right.params); | |
995 | } | |
996 | ||
997 | /* Create an AST node referring to a "super::" qualified name. IDENT | |
998 | is the base name and N_SUPERS is how many "super::"s were | |
999 | provided. N_SUPERS can be zero. */ | |
1000 | ||
1001 | static const struct rust_op * | |
1002 | super_name (const struct rust_op *ident, unsigned int n_supers) | |
1003 | { | |
1004 | const char *scope = block_scope (expression_context_block); | |
1005 | int offset; | |
1006 | ||
1007 | gdb_assert (ident->opcode == OP_VAR_VALUE); | |
1008 | ||
1009 | if (scope[0] == '\0') | |
1010 | error (_("Couldn't find namespace scope for self::")); | |
1011 | ||
1012 | if (n_supers > 0) | |
1013 | { | |
c44af4eb | 1014 | int len; |
8001f118 | 1015 | std::vector<int> offsets; |
78cc6c2d | 1016 | unsigned int current_len; |
c44af4eb | 1017 | |
c44af4eb | 1018 | current_len = cp_find_first_component (scope); |
c44af4eb TT |
1019 | while (scope[current_len] != '\0') |
1020 | { | |
8001f118 | 1021 | offsets.push_back (current_len); |
c44af4eb | 1022 | gdb_assert (scope[current_len] == ':'); |
c44af4eb TT |
1023 | /* The "::". */ |
1024 | current_len += 2; | |
1025 | current_len += cp_find_first_component (scope | |
1026 | + current_len); | |
1027 | } | |
1028 | ||
8001f118 | 1029 | len = offsets.size (); |
c44af4eb TT |
1030 | if (n_supers >= len) |
1031 | error (_("Too many super:: uses from '%s'"), scope); | |
1032 | ||
8001f118 | 1033 | offset = offsets[len - n_supers]; |
c44af4eb TT |
1034 | } |
1035 | else | |
1036 | offset = strlen (scope); | |
1037 | ||
3232fabd TT |
1038 | obstack_grow (work_obstack, "::", 2); |
1039 | obstack_grow (work_obstack, scope, offset); | |
1040 | obstack_grow (work_obstack, "::", 2); | |
1041 | obstack_grow0 (work_obstack, ident->left.sval.ptr, ident->left.sval.length); | |
c44af4eb | 1042 | |
3232fabd | 1043 | return ast_path (make_stoken ((const char *) obstack_finish (work_obstack)), |
c44af4eb TT |
1044 | ident->right.params); |
1045 | } | |
1046 | ||
1047 | /* A helper that updates innermost_block as appropriate. */ | |
1048 | ||
1049 | static void | |
1050 | update_innermost_block (struct block_symbol sym) | |
1051 | { | |
1052 | if (symbol_read_needs_frame (sym.symbol) | |
1053 | && (innermost_block == NULL | |
1054 | || contained_in (sym.block, innermost_block))) | |
1055 | innermost_block = sym.block; | |
1056 | } | |
1057 | ||
1058 | /* A helper to look up a Rust type, or fail. This only works for | |
1059 | types defined by rust_language_arch_info. */ | |
1060 | ||
1061 | static struct type * | |
1062 | rust_type (const char *name) | |
1063 | { | |
1064 | struct type *type; | |
1065 | ||
3232fabd TT |
1066 | type = language_lookup_primitive_type (current_parser->language (), |
1067 | current_parser->arch (), | |
c44af4eb TT |
1068 | name); |
1069 | if (type == NULL) | |
1070 | error (_("Could not find Rust type %s"), name); | |
1071 | return type; | |
1072 | } | |
1073 | ||
1074 | /* Lex a hex number with at least MIN digits and at most MAX | |
1075 | digits. */ | |
1076 | ||
1077 | static uint32_t | |
1078 | lex_hex (int min, int max) | |
1079 | { | |
1080 | uint32_t result = 0; | |
1081 | int len = 0; | |
1082 | /* We only want to stop at MAX if we're lexing a byte escape. */ | |
1083 | int check_max = min == max; | |
1084 | ||
1085 | while ((check_max ? len <= max : 1) | |
1086 | && ((lexptr[0] >= 'a' && lexptr[0] <= 'f') | |
1087 | || (lexptr[0] >= 'A' && lexptr[0] <= 'F') | |
1088 | || (lexptr[0] >= '0' && lexptr[0] <= '9'))) | |
1089 | { | |
1090 | result *= 16; | |
1091 | if (lexptr[0] >= 'a' && lexptr[0] <= 'f') | |
1092 | result = result + 10 + lexptr[0] - 'a'; | |
1093 | else if (lexptr[0] >= 'A' && lexptr[0] <= 'F') | |
1094 | result = result + 10 + lexptr[0] - 'A'; | |
1095 | else | |
1096 | result = result + lexptr[0] - '0'; | |
1097 | ++lexptr; | |
1098 | ++len; | |
1099 | } | |
1100 | ||
1101 | if (len < min) | |
1102 | error (_("Not enough hex digits seen")); | |
1103 | if (len > max) | |
1104 | { | |
1105 | gdb_assert (min != max); | |
1106 | error (_("Overlong hex escape")); | |
1107 | } | |
1108 | ||
1109 | return result; | |
1110 | } | |
1111 | ||
1112 | /* Lex an escape. IS_BYTE is true if we're lexing a byte escape; | |
1113 | otherwise we're lexing a character escape. */ | |
1114 | ||
1115 | static uint32_t | |
1116 | lex_escape (int is_byte) | |
1117 | { | |
1118 | uint32_t result; | |
1119 | ||
1120 | gdb_assert (lexptr[0] == '\\'); | |
1121 | ++lexptr; | |
1122 | switch (lexptr[0]) | |
1123 | { | |
1124 | case 'x': | |
1125 | ++lexptr; | |
1126 | result = lex_hex (2, 2); | |
1127 | break; | |
1128 | ||
1129 | case 'u': | |
1130 | if (is_byte) | |
1131 | error (_("Unicode escape in byte literal")); | |
1132 | ++lexptr; | |
1133 | if (lexptr[0] != '{') | |
1134 | error (_("Missing '{' in Unicode escape")); | |
1135 | ++lexptr; | |
1136 | result = lex_hex (1, 6); | |
1137 | /* Could do range checks here. */ | |
1138 | if (lexptr[0] != '}') | |
1139 | error (_("Missing '}' in Unicode escape")); | |
1140 | ++lexptr; | |
1141 | break; | |
1142 | ||
1143 | case 'n': | |
1144 | result = '\n'; | |
1145 | ++lexptr; | |
1146 | break; | |
1147 | case 'r': | |
1148 | result = '\r'; | |
1149 | ++lexptr; | |
1150 | break; | |
1151 | case 't': | |
1152 | result = '\t'; | |
1153 | ++lexptr; | |
1154 | break; | |
1155 | case '\\': | |
1156 | result = '\\'; | |
1157 | ++lexptr; | |
1158 | break; | |
1159 | case '0': | |
1160 | result = '\0'; | |
1161 | ++lexptr; | |
1162 | break; | |
1163 | case '\'': | |
1164 | result = '\''; | |
1165 | ++lexptr; | |
1166 | break; | |
1167 | case '"': | |
1168 | result = '"'; | |
1169 | ++lexptr; | |
1170 | break; | |
1171 | ||
1172 | default: | |
1173 | error (_("Invalid escape \\%c in literal"), lexptr[0]); | |
1174 | } | |
1175 | ||
1176 | return result; | |
1177 | } | |
1178 | ||
1179 | /* Lex a character constant. */ | |
1180 | ||
1181 | static int | |
1182 | lex_character (void) | |
1183 | { | |
1184 | int is_byte = 0; | |
1185 | uint32_t value; | |
1186 | ||
1187 | if (lexptr[0] == 'b') | |
1188 | { | |
1189 | is_byte = 1; | |
1190 | ++lexptr; | |
1191 | } | |
1192 | gdb_assert (lexptr[0] == '\''); | |
1193 | ++lexptr; | |
1194 | /* This should handle UTF-8 here. */ | |
1195 | if (lexptr[0] == '\\') | |
1196 | value = lex_escape (is_byte); | |
1197 | else | |
1198 | { | |
1199 | value = lexptr[0] & 0xff; | |
1200 | ++lexptr; | |
1201 | } | |
1202 | ||
1203 | if (lexptr[0] != '\'') | |
1204 | error (_("Unterminated character literal")); | |
1205 | ++lexptr; | |
1206 | ||
1207 | rustyylval.typed_val_int.val = value; | |
1208 | rustyylval.typed_val_int.type = rust_type (is_byte ? "u8" : "char"); | |
1209 | ||
1210 | return INTEGER; | |
1211 | } | |
1212 | ||
1213 | /* Return the offset of the double quote if STR looks like the start | |
1214 | of a raw string, or 0 if STR does not start a raw string. */ | |
1215 | ||
1216 | static int | |
1217 | starts_raw_string (const char *str) | |
1218 | { | |
1219 | const char *save = str; | |
1220 | ||
1221 | if (str[0] != 'r') | |
1222 | return 0; | |
1223 | ++str; | |
1224 | while (str[0] == '#') | |
1225 | ++str; | |
1226 | if (str[0] == '"') | |
1227 | return str - save; | |
1228 | return 0; | |
1229 | } | |
1230 | ||
1231 | /* Return true if STR looks like the end of a raw string that had N | |
1232 | hashes at the start. */ | |
1233 | ||
65c40c95 | 1234 | static bool |
c44af4eb TT |
1235 | ends_raw_string (const char *str, int n) |
1236 | { | |
1237 | int i; | |
1238 | ||
1239 | gdb_assert (str[0] == '"'); | |
1240 | for (i = 0; i < n; ++i) | |
1241 | if (str[i + 1] != '#') | |
65c40c95 TT |
1242 | return false; |
1243 | return true; | |
c44af4eb TT |
1244 | } |
1245 | ||
1246 | /* Lex a string constant. */ | |
1247 | ||
1248 | static int | |
1249 | lex_string (void) | |
1250 | { | |
1251 | int is_byte = lexptr[0] == 'b'; | |
1252 | int raw_length; | |
c44af4eb TT |
1253 | |
1254 | if (is_byte) | |
1255 | ++lexptr; | |
1256 | raw_length = starts_raw_string (lexptr); | |
1257 | lexptr += raw_length; | |
1258 | gdb_assert (lexptr[0] == '"'); | |
1259 | ++lexptr; | |
1260 | ||
1261 | while (1) | |
1262 | { | |
1263 | uint32_t value; | |
1264 | ||
1265 | if (raw_length > 0) | |
1266 | { | |
1267 | if (lexptr[0] == '"' && ends_raw_string (lexptr, raw_length - 1)) | |
1268 | { | |
1269 | /* Exit with lexptr pointing after the final "#". */ | |
1270 | lexptr += raw_length; | |
1271 | break; | |
1272 | } | |
1273 | else if (lexptr[0] == '\0') | |
1274 | error (_("Unexpected EOF in string")); | |
1275 | ||
1276 | value = lexptr[0] & 0xff; | |
1277 | if (is_byte && value > 127) | |
1278 | error (_("Non-ASCII value in raw byte string")); | |
3232fabd | 1279 | obstack_1grow (work_obstack, value); |
c44af4eb TT |
1280 | |
1281 | ++lexptr; | |
1282 | } | |
1283 | else if (lexptr[0] == '"') | |
1284 | { | |
1285 | /* Make sure to skip the quote. */ | |
1286 | ++lexptr; | |
1287 | break; | |
1288 | } | |
1289 | else if (lexptr[0] == '\\') | |
1290 | { | |
1291 | value = lex_escape (is_byte); | |
1292 | ||
1293 | if (is_byte) | |
3232fabd | 1294 | obstack_1grow (work_obstack, value); |
c44af4eb TT |
1295 | else |
1296 | convert_between_encodings ("UTF-32", "UTF-8", (gdb_byte *) &value, | |
1297 | sizeof (value), sizeof (value), | |
3232fabd | 1298 | work_obstack, translit_none); |
c44af4eb TT |
1299 | } |
1300 | else if (lexptr[0] == '\0') | |
1301 | error (_("Unexpected EOF in string")); | |
1302 | else | |
1303 | { | |
1304 | value = lexptr[0] & 0xff; | |
1305 | if (is_byte && value > 127) | |
1306 | error (_("Non-ASCII value in byte string")); | |
3232fabd | 1307 | obstack_1grow (work_obstack, value); |
c44af4eb TT |
1308 | ++lexptr; |
1309 | } | |
1310 | } | |
1311 | ||
3232fabd TT |
1312 | rustyylval.sval.length = obstack_object_size (work_obstack); |
1313 | rustyylval.sval.ptr = (const char *) obstack_finish (work_obstack); | |
c44af4eb TT |
1314 | return is_byte ? BYTESTRING : STRING; |
1315 | } | |
1316 | ||
1317 | /* Return true if STRING starts with whitespace followed by a digit. */ | |
1318 | ||
65c40c95 | 1319 | static bool |
c44af4eb TT |
1320 | space_then_number (const char *string) |
1321 | { | |
1322 | const char *p = string; | |
1323 | ||
1324 | while (p[0] == ' ' || p[0] == '\t') | |
1325 | ++p; | |
1326 | if (p == string) | |
65c40c95 | 1327 | return false; |
c44af4eb TT |
1328 | |
1329 | return *p >= '0' && *p <= '9'; | |
1330 | } | |
1331 | ||
1332 | /* Return true if C can start an identifier. */ | |
1333 | ||
65c40c95 | 1334 | static bool |
c44af4eb TT |
1335 | rust_identifier_start_p (char c) |
1336 | { | |
1337 | return ((c >= 'a' && c <= 'z') | |
1338 | || (c >= 'A' && c <= 'Z') | |
1339 | || c == '_' | |
1340 | || c == '$'); | |
1341 | } | |
1342 | ||
1343 | /* Lex an identifier. */ | |
1344 | ||
1345 | static int | |
1346 | lex_identifier (void) | |
1347 | { | |
1348 | const char *start = lexptr; | |
1349 | unsigned int length; | |
1350 | const struct token_info *token; | |
1351 | int i; | |
1352 | int is_gdb_var = lexptr[0] == '$'; | |
1353 | ||
1354 | gdb_assert (rust_identifier_start_p (lexptr[0])); | |
1355 | ||
1356 | ++lexptr; | |
1357 | ||
1358 | /* For the time being this doesn't handle Unicode rules. Non-ASCII | |
1359 | identifiers are gated anyway. */ | |
1360 | while ((lexptr[0] >= 'a' && lexptr[0] <= 'z') | |
1361 | || (lexptr[0] >= 'A' && lexptr[0] <= 'Z') | |
1362 | || lexptr[0] == '_' | |
1363 | || (is_gdb_var && lexptr[0] == '$') | |
1364 | || (lexptr[0] >= '0' && lexptr[0] <= '9')) | |
1365 | ++lexptr; | |
1366 | ||
1367 | ||
1368 | length = lexptr - start; | |
1369 | token = NULL; | |
1370 | for (i = 0; i < ARRAY_SIZE (identifier_tokens); ++i) | |
1371 | { | |
1372 | if (length == strlen (identifier_tokens[i].name) | |
1373 | && strncmp (identifier_tokens[i].name, start, length) == 0) | |
1374 | { | |
1375 | token = &identifier_tokens[i]; | |
1376 | break; | |
1377 | } | |
1378 | } | |
1379 | ||
1380 | if (token != NULL) | |
1381 | { | |
1382 | if (token->value == 0) | |
1383 | { | |
1384 | /* Leave the terminating token alone. */ | |
1385 | lexptr = start; | |
1386 | return 0; | |
1387 | } | |
1388 | } | |
1389 | else if (token == NULL | |
1390 | && (strncmp (start, "thread", length) == 0 | |
1391 | || strncmp (start, "task", length) == 0) | |
1392 | && space_then_number (lexptr)) | |
1393 | { | |
1394 | /* "task" or "thread" followed by a number terminates the | |
1395 | parse, per gdb rules. */ | |
1396 | lexptr = start; | |
1397 | return 0; | |
1398 | } | |
1399 | ||
1400 | if (token == NULL || (parse_completion && lexptr[0] == '\0')) | |
1401 | rustyylval.sval = make_stoken (rust_copy_name (start, length)); | |
1402 | ||
1403 | if (parse_completion && lexptr[0] == '\0') | |
1404 | { | |
1405 | /* Prevent rustyylex from returning two COMPLETE tokens. */ | |
1406 | prev_lexptr = lexptr; | |
1407 | return COMPLETE; | |
1408 | } | |
1409 | ||
1410 | if (token != NULL) | |
1411 | return token->value; | |
1412 | if (is_gdb_var) | |
1413 | return GDBVAR; | |
1414 | return IDENT; | |
1415 | } | |
1416 | ||
1417 | /* Lex an operator. */ | |
1418 | ||
1419 | static int | |
1420 | lex_operator (void) | |
1421 | { | |
1422 | const struct token_info *token = NULL; | |
1423 | int i; | |
1424 | ||
1425 | for (i = 0; i < ARRAY_SIZE (operator_tokens); ++i) | |
1426 | { | |
1427 | if (strncmp (operator_tokens[i].name, lexptr, | |
1428 | strlen (operator_tokens[i].name)) == 0) | |
1429 | { | |
1430 | lexptr += strlen (operator_tokens[i].name); | |
1431 | token = &operator_tokens[i]; | |
1432 | break; | |
1433 | } | |
1434 | } | |
1435 | ||
1436 | if (token != NULL) | |
1437 | { | |
1438 | rustyylval.opcode = token->opcode; | |
1439 | return token->value; | |
1440 | } | |
1441 | ||
1442 | return *lexptr++; | |
1443 | } | |
1444 | ||
1445 | /* Lex a number. */ | |
1446 | ||
1447 | static int | |
1448 | lex_number (void) | |
1449 | { | |
1450 | regmatch_t subexps[NUM_SUBEXPRESSIONS]; | |
1451 | int match; | |
1452 | int is_integer = 0; | |
1453 | int could_be_decimal = 1; | |
347dc102 | 1454 | int implicit_i32 = 0; |
8001f118 | 1455 | const char *type_name = NULL; |
c44af4eb TT |
1456 | struct type *type; |
1457 | int end_index; | |
1458 | int type_index = -1; | |
8001f118 | 1459 | int i; |
c44af4eb TT |
1460 | |
1461 | match = regexec (&number_regex, lexptr, ARRAY_SIZE (subexps), subexps, 0); | |
1462 | /* Failure means the regexp is broken. */ | |
1463 | gdb_assert (match == 0); | |
1464 | ||
1465 | if (subexps[INT_TEXT].rm_so != -1) | |
1466 | { | |
1467 | /* Integer part matched. */ | |
1468 | is_integer = 1; | |
1469 | end_index = subexps[INT_TEXT].rm_eo; | |
1470 | if (subexps[INT_TYPE].rm_so == -1) | |
347dc102 TT |
1471 | { |
1472 | type_name = "i32"; | |
1473 | implicit_i32 = 1; | |
1474 | } | |
c44af4eb TT |
1475 | else |
1476 | { | |
1477 | type_index = INT_TYPE; | |
1478 | could_be_decimal = 0; | |
1479 | } | |
1480 | } | |
1481 | else if (subexps[FLOAT_TYPE1].rm_so != -1) | |
1482 | { | |
1483 | /* Found floating point type suffix. */ | |
1484 | end_index = subexps[FLOAT_TYPE1].rm_so; | |
1485 | type_index = FLOAT_TYPE1; | |
1486 | } | |
1487 | else if (subexps[FLOAT_TYPE2].rm_so != -1) | |
1488 | { | |
1489 | /* Found floating point type suffix. */ | |
1490 | end_index = subexps[FLOAT_TYPE2].rm_so; | |
1491 | type_index = FLOAT_TYPE2; | |
1492 | } | |
1493 | else | |
1494 | { | |
1495 | /* Any other floating point match. */ | |
1496 | end_index = subexps[0].rm_eo; | |
1497 | type_name = "f64"; | |
1498 | } | |
1499 | ||
1500 | /* We need a special case if the final character is ".". In this | |
1501 | case we might need to parse an integer. For example, "23.f()" is | |
1502 | a request for a trait method call, not a syntax error involving | |
1503 | the floating point number "23.". */ | |
1504 | gdb_assert (subexps[0].rm_eo > 0); | |
1505 | if (lexptr[subexps[0].rm_eo - 1] == '.') | |
1506 | { | |
f1735a53 | 1507 | const char *next = skip_spaces (&lexptr[subexps[0].rm_eo]); |
c44af4eb TT |
1508 | |
1509 | if (rust_identifier_start_p (*next) || *next == '.') | |
1510 | { | |
1511 | --subexps[0].rm_eo; | |
1512 | is_integer = 1; | |
1513 | end_index = subexps[0].rm_eo; | |
1514 | type_name = "i32"; | |
1515 | could_be_decimal = 1; | |
347dc102 | 1516 | implicit_i32 = 1; |
c44af4eb TT |
1517 | } |
1518 | } | |
1519 | ||
1520 | /* Compute the type name if we haven't already. */ | |
8001f118 | 1521 | std::string type_name_holder; |
c44af4eb TT |
1522 | if (type_name == NULL) |
1523 | { | |
1524 | gdb_assert (type_index != -1); | |
8001f118 TT |
1525 | type_name_holder = std::string (lexptr + subexps[type_index].rm_so, |
1526 | (subexps[type_index].rm_eo | |
1527 | - subexps[type_index].rm_so)); | |
1528 | type_name = type_name_holder.c_str (); | |
c44af4eb TT |
1529 | } |
1530 | ||
1531 | /* Look up the type. */ | |
1532 | type = rust_type (type_name); | |
1533 | ||
1534 | /* Copy the text of the number and remove the "_"s. */ | |
8001f118 TT |
1535 | std::string number; |
1536 | for (i = 0; i < end_index && lexptr[i]; ++i) | |
c44af4eb | 1537 | { |
8001f118 | 1538 | if (lexptr[i] == '_') |
c44af4eb TT |
1539 | could_be_decimal = 0; |
1540 | else | |
8001f118 | 1541 | number.push_back (lexptr[i]); |
c44af4eb | 1542 | } |
c44af4eb TT |
1543 | |
1544 | /* Advance past the match. */ | |
1545 | lexptr += subexps[0].rm_eo; | |
1546 | ||
1547 | /* Parse the number. */ | |
1548 | if (is_integer) | |
1549 | { | |
347dc102 | 1550 | uint64_t value; |
c44af4eb | 1551 | int radix = 10; |
8001f118 TT |
1552 | int offset = 0; |
1553 | ||
c44af4eb TT |
1554 | if (number[0] == '0') |
1555 | { | |
1556 | if (number[1] == 'x') | |
1557 | radix = 16; | |
1558 | else if (number[1] == 'o') | |
1559 | radix = 8; | |
1560 | else if (number[1] == 'b') | |
1561 | radix = 2; | |
1562 | if (radix != 10) | |
1563 | { | |
8001f118 | 1564 | offset = 2; |
c44af4eb TT |
1565 | could_be_decimal = 0; |
1566 | } | |
1567 | } | |
347dc102 | 1568 | |
8001f118 | 1569 | value = strtoul (number.c_str () + offset, NULL, radix); |
347dc102 TT |
1570 | if (implicit_i32 && value >= ((uint64_t) 1) << 31) |
1571 | type = rust_type ("i64"); | |
1572 | ||
1573 | rustyylval.typed_val_int.val = value; | |
c44af4eb TT |
1574 | rustyylval.typed_val_int.type = type; |
1575 | } | |
1576 | else | |
1577 | { | |
c44af4eb | 1578 | rustyylval.typed_val_float.type = type; |
edd079d9 UW |
1579 | bool parsed = parse_float (number.c_str (), number.length (), |
1580 | rustyylval.typed_val_float.type, | |
1581 | rustyylval.typed_val_float.val); | |
1582 | gdb_assert (parsed); | |
c44af4eb TT |
1583 | } |
1584 | ||
c44af4eb TT |
1585 | return is_integer ? (could_be_decimal ? DECIMAL_INTEGER : INTEGER) : FLOAT; |
1586 | } | |
1587 | ||
1588 | /* The lexer. */ | |
1589 | ||
1590 | static int | |
1591 | rustyylex (void) | |
1592 | { | |
1593 | /* Skip all leading whitespace. */ | |
1594 | while (lexptr[0] == ' ' || lexptr[0] == '\t' || lexptr[0] == '\r' | |
1595 | || lexptr[0] == '\n') | |
1596 | ++lexptr; | |
1597 | ||
1598 | /* If we hit EOF and we're completing, then return COMPLETE -- maybe | |
1599 | we're completing an empty string at the end of a field_expr. | |
1600 | But, we don't want to return two COMPLETE tokens in a row. */ | |
1601 | if (lexptr[0] == '\0' && lexptr == prev_lexptr) | |
1602 | return 0; | |
1603 | prev_lexptr = lexptr; | |
1604 | if (lexptr[0] == '\0') | |
1605 | { | |
1606 | if (parse_completion) | |
1607 | { | |
1608 | rustyylval.sval = make_stoken (""); | |
1609 | return COMPLETE; | |
1610 | } | |
1611 | return 0; | |
1612 | } | |
1613 | ||
1614 | if (lexptr[0] >= '0' && lexptr[0] <= '9') | |
1615 | return lex_number (); | |
1616 | else if (lexptr[0] == 'b' && lexptr[1] == '\'') | |
1617 | return lex_character (); | |
1618 | else if (lexptr[0] == 'b' && lexptr[1] == '"') | |
1619 | return lex_string (); | |
1620 | else if (lexptr[0] == 'b' && starts_raw_string (lexptr + 1)) | |
1621 | return lex_string (); | |
1622 | else if (starts_raw_string (lexptr)) | |
1623 | return lex_string (); | |
1624 | else if (rust_identifier_start_p (lexptr[0])) | |
1625 | return lex_identifier (); | |
1626 | else if (lexptr[0] == '"') | |
1627 | return lex_string (); | |
1628 | else if (lexptr[0] == '\'') | |
1629 | return lex_character (); | |
1630 | else if (lexptr[0] == '}' || lexptr[0] == ']') | |
1631 | { | |
1632 | /* Falls through to lex_operator. */ | |
1633 | --paren_depth; | |
1634 | } | |
1635 | else if (lexptr[0] == '(' || lexptr[0] == '{') | |
1636 | { | |
1637 | /* Falls through to lex_operator. */ | |
1638 | ++paren_depth; | |
1639 | } | |
1640 | else if (lexptr[0] == ',' && comma_terminates && paren_depth == 0) | |
1641 | return 0; | |
1642 | ||
1643 | return lex_operator (); | |
1644 | } | |
1645 | ||
1646 | /* Push back a single character to be re-lexed. */ | |
1647 | ||
1648 | static void | |
1649 | rust_push_back (char c) | |
1650 | { | |
1651 | /* Can't be called before any lexing. */ | |
1652 | gdb_assert (prev_lexptr != NULL); | |
1653 | ||
1654 | --lexptr; | |
1655 | gdb_assert (*lexptr == c); | |
1656 | } | |
1657 | ||
1658 | \f | |
1659 | ||
1660 | /* Make an arbitrary operation and fill in the fields. */ | |
1661 | ||
1662 | static const struct rust_op * | |
1663 | ast_operation (enum exp_opcode opcode, const struct rust_op *left, | |
1664 | const struct rust_op *right) | |
1665 | { | |
3232fabd | 1666 | struct rust_op *result = OBSTACK_ZALLOC (work_obstack, struct rust_op); |
c44af4eb TT |
1667 | |
1668 | result->opcode = opcode; | |
1669 | result->left.op = left; | |
1670 | result->right.op = right; | |
1671 | ||
1672 | return result; | |
1673 | } | |
1674 | ||
1675 | /* Make a compound assignment operation. */ | |
1676 | ||
1677 | static const struct rust_op * | |
1678 | ast_compound_assignment (enum exp_opcode opcode, const struct rust_op *left, | |
1679 | const struct rust_op *right) | |
1680 | { | |
3232fabd | 1681 | struct rust_op *result = OBSTACK_ZALLOC (work_obstack, struct rust_op); |
c44af4eb TT |
1682 | |
1683 | result->opcode = opcode; | |
1684 | result->compound_assignment = 1; | |
1685 | result->left.op = left; | |
1686 | result->right.op = right; | |
1687 | ||
1688 | return result; | |
1689 | } | |
1690 | ||
1691 | /* Make a typed integer literal operation. */ | |
1692 | ||
1693 | static const struct rust_op * | |
1694 | ast_literal (struct typed_val_int val) | |
1695 | { | |
3232fabd | 1696 | struct rust_op *result = OBSTACK_ZALLOC (work_obstack, struct rust_op); |
c44af4eb TT |
1697 | |
1698 | result->opcode = OP_LONG; | |
1699 | result->left.typed_val_int = val; | |
1700 | ||
1701 | return result; | |
1702 | } | |
1703 | ||
1704 | /* Make a typed floating point literal operation. */ | |
1705 | ||
1706 | static const struct rust_op * | |
1707 | ast_dliteral (struct typed_val_float val) | |
1708 | { | |
3232fabd | 1709 | struct rust_op *result = OBSTACK_ZALLOC (work_obstack, struct rust_op); |
c44af4eb | 1710 | |
edd079d9 | 1711 | result->opcode = OP_FLOAT; |
c44af4eb TT |
1712 | result->left.typed_val_float = val; |
1713 | ||
1714 | return result; | |
1715 | } | |
1716 | ||
1717 | /* Make a unary operation. */ | |
1718 | ||
1719 | static const struct rust_op * | |
1720 | ast_unary (enum exp_opcode opcode, const struct rust_op *expr) | |
1721 | { | |
1722 | return ast_operation (opcode, expr, NULL); | |
1723 | } | |
1724 | ||
1725 | /* Make a cast operation. */ | |
1726 | ||
1727 | static const struct rust_op * | |
1728 | ast_cast (const struct rust_op *expr, const struct rust_op *type) | |
1729 | { | |
3232fabd | 1730 | struct rust_op *result = OBSTACK_ZALLOC (work_obstack, struct rust_op); |
c44af4eb TT |
1731 | |
1732 | result->opcode = UNOP_CAST; | |
1733 | result->left.op = expr; | |
1734 | result->right.op = type; | |
1735 | ||
1736 | return result; | |
1737 | } | |
1738 | ||
1739 | /* Make a call-like operation. This is nominally a function call, but | |
1740 | when lowering we may discover that it actually represents the | |
1741 | creation of a tuple struct. */ | |
1742 | ||
1743 | static const struct rust_op * | |
1744 | ast_call_ish (enum exp_opcode opcode, const struct rust_op *expr, | |
3232fabd | 1745 | rust_op_vector *params) |
c44af4eb | 1746 | { |
3232fabd | 1747 | struct rust_op *result = OBSTACK_ZALLOC (work_obstack, struct rust_op); |
c44af4eb TT |
1748 | |
1749 | result->opcode = opcode; | |
1750 | result->left.op = expr; | |
1751 | result->right.params = params; | |
1752 | ||
1753 | return result; | |
1754 | } | |
1755 | ||
1756 | /* Make a structure creation operation. */ | |
1757 | ||
1758 | static const struct rust_op * | |
3232fabd | 1759 | ast_struct (const struct rust_op *name, rust_set_vector *fields) |
c44af4eb | 1760 | { |
3232fabd | 1761 | struct rust_op *result = OBSTACK_ZALLOC (work_obstack, struct rust_op); |
c44af4eb TT |
1762 | |
1763 | result->opcode = OP_AGGREGATE; | |
1764 | result->left.op = name; | |
1765 | result->right.field_inits = fields; | |
1766 | ||
1767 | return result; | |
1768 | } | |
1769 | ||
1770 | /* Make an identifier path. */ | |
1771 | ||
1772 | static const struct rust_op * | |
3232fabd | 1773 | ast_path (struct stoken path, rust_op_vector *params) |
c44af4eb | 1774 | { |
3232fabd | 1775 | struct rust_op *result = OBSTACK_ZALLOC (work_obstack, struct rust_op); |
c44af4eb TT |
1776 | |
1777 | result->opcode = OP_VAR_VALUE; | |
1778 | result->left.sval = path; | |
1779 | result->right.params = params; | |
1780 | ||
1781 | return result; | |
1782 | } | |
1783 | ||
1784 | /* Make a string constant operation. */ | |
1785 | ||
1786 | static const struct rust_op * | |
1787 | ast_string (struct stoken str) | |
1788 | { | |
3232fabd | 1789 | struct rust_op *result = OBSTACK_ZALLOC (work_obstack, struct rust_op); |
c44af4eb TT |
1790 | |
1791 | result->opcode = OP_STRING; | |
1792 | result->left.sval = str; | |
1793 | ||
1794 | return result; | |
1795 | } | |
1796 | ||
1797 | /* Make a field expression. */ | |
1798 | ||
1799 | static const struct rust_op * | |
1800 | ast_structop (const struct rust_op *left, const char *name, int completing) | |
1801 | { | |
3232fabd | 1802 | struct rust_op *result = OBSTACK_ZALLOC (work_obstack, struct rust_op); |
c44af4eb TT |
1803 | |
1804 | result->opcode = STRUCTOP_STRUCT; | |
1805 | result->completing = completing; | |
1806 | result->left.op = left; | |
1807 | result->right.sval = make_stoken (name); | |
1808 | ||
1809 | return result; | |
1810 | } | |
1811 | ||
1812 | /* Make an anonymous struct operation, like 'x.0'. */ | |
1813 | ||
1814 | static const struct rust_op * | |
1815 | ast_structop_anonymous (const struct rust_op *left, | |
1816 | struct typed_val_int number) | |
1817 | { | |
3232fabd | 1818 | struct rust_op *result = OBSTACK_ZALLOC (work_obstack, struct rust_op); |
c44af4eb TT |
1819 | |
1820 | result->opcode = STRUCTOP_ANONYMOUS; | |
1821 | result->left.op = left; | |
1822 | result->right.typed_val_int = number; | |
1823 | ||
1824 | return result; | |
1825 | } | |
1826 | ||
1827 | /* Make a range operation. */ | |
1828 | ||
1829 | static const struct rust_op * | |
1830 | ast_range (const struct rust_op *lhs, const struct rust_op *rhs) | |
1831 | { | |
3232fabd | 1832 | struct rust_op *result = OBSTACK_ZALLOC (work_obstack, struct rust_op); |
c44af4eb | 1833 | |
01739a3b | 1834 | result->opcode = OP_RANGE; |
c44af4eb TT |
1835 | result->left.op = lhs; |
1836 | result->right.op = rhs; | |
1837 | ||
1838 | return result; | |
1839 | } | |
1840 | ||
1841 | /* A helper function to make a type-related AST node. */ | |
1842 | ||
1843 | static struct rust_op * | |
1844 | ast_basic_type (enum type_code typecode) | |
1845 | { | |
3232fabd | 1846 | struct rust_op *result = OBSTACK_ZALLOC (work_obstack, struct rust_op); |
c44af4eb TT |
1847 | |
1848 | result->opcode = OP_TYPE; | |
1849 | result->typecode = typecode; | |
1850 | return result; | |
1851 | } | |
1852 | ||
1853 | /* Create an AST node describing an array type. */ | |
1854 | ||
1855 | static const struct rust_op * | |
1856 | ast_array_type (const struct rust_op *lhs, struct typed_val_int val) | |
1857 | { | |
1858 | struct rust_op *result = ast_basic_type (TYPE_CODE_ARRAY); | |
1859 | ||
1860 | result->left.op = lhs; | |
1861 | result->right.typed_val_int = val; | |
1862 | return result; | |
1863 | } | |
1864 | ||
1865 | /* Create an AST node describing a reference type. */ | |
1866 | ||
1867 | static const struct rust_op * | |
1868 | ast_slice_type (const struct rust_op *type) | |
1869 | { | |
1870 | /* Use TYPE_CODE_COMPLEX just because it is handy. */ | |
1871 | struct rust_op *result = ast_basic_type (TYPE_CODE_COMPLEX); | |
1872 | ||
1873 | result->left.op = type; | |
1874 | return result; | |
1875 | } | |
1876 | ||
1877 | /* Create an AST node describing a reference type. */ | |
1878 | ||
1879 | static const struct rust_op * | |
1880 | ast_reference_type (const struct rust_op *type) | |
1881 | { | |
1882 | struct rust_op *result = ast_basic_type (TYPE_CODE_REF); | |
1883 | ||
1884 | result->left.op = type; | |
1885 | return result; | |
1886 | } | |
1887 | ||
1888 | /* Create an AST node describing a pointer type. */ | |
1889 | ||
1890 | static const struct rust_op * | |
1891 | ast_pointer_type (const struct rust_op *type, int is_mut) | |
1892 | { | |
1893 | struct rust_op *result = ast_basic_type (TYPE_CODE_PTR); | |
1894 | ||
1895 | result->left.op = type; | |
1896 | /* For the time being we ignore is_mut. */ | |
1897 | return result; | |
1898 | } | |
1899 | ||
1900 | /* Create an AST node describing a function type. */ | |
1901 | ||
1902 | static const struct rust_op * | |
3232fabd | 1903 | ast_function_type (const struct rust_op *rtype, rust_op_vector *params) |
c44af4eb TT |
1904 | { |
1905 | struct rust_op *result = ast_basic_type (TYPE_CODE_FUNC); | |
1906 | ||
1907 | result->left.op = rtype; | |
1908 | result->right.params = params; | |
1909 | return result; | |
1910 | } | |
1911 | ||
1912 | /* Create an AST node describing a tuple type. */ | |
1913 | ||
1914 | static const struct rust_op * | |
3232fabd | 1915 | ast_tuple_type (rust_op_vector *params) |
c44af4eb TT |
1916 | { |
1917 | struct rust_op *result = ast_basic_type (TYPE_CODE_STRUCT); | |
1918 | ||
1919 | result->left.params = params; | |
1920 | return result; | |
1921 | } | |
1922 | ||
1923 | /* A helper to appropriately munge NAME and BLOCK depending on the | |
1924 | presence of a leading "::". */ | |
1925 | ||
1926 | static void | |
1927 | munge_name_and_block (const char **name, const struct block **block) | |
1928 | { | |
1929 | /* If it is a global reference, skip the current block in favor of | |
1930 | the static block. */ | |
1931 | if (strncmp (*name, "::", 2) == 0) | |
1932 | { | |
1933 | *name += 2; | |
1934 | *block = block_static_block (*block); | |
1935 | } | |
1936 | } | |
1937 | ||
1938 | /* Like lookup_symbol, but handles Rust namespace conventions, and | |
1939 | doesn't require field_of_this_result. */ | |
1940 | ||
1941 | static struct block_symbol | |
1942 | rust_lookup_symbol (const char *name, const struct block *block, | |
1943 | const domain_enum domain) | |
1944 | { | |
1945 | struct block_symbol result; | |
1946 | ||
1947 | munge_name_and_block (&name, &block); | |
1948 | ||
1949 | result = lookup_symbol (name, block, domain, NULL); | |
1950 | if (result.symbol != NULL) | |
1951 | update_innermost_block (result); | |
1952 | return result; | |
1953 | } | |
1954 | ||
1955 | /* Look up a type, following Rust namespace conventions. */ | |
1956 | ||
1957 | static struct type * | |
1958 | rust_lookup_type (const char *name, const struct block *block) | |
1959 | { | |
1960 | struct block_symbol result; | |
1961 | struct type *type; | |
1962 | ||
1963 | munge_name_and_block (&name, &block); | |
1964 | ||
1965 | result = lookup_symbol (name, block, STRUCT_DOMAIN, NULL); | |
1966 | if (result.symbol != NULL) | |
1967 | { | |
1968 | update_innermost_block (result); | |
1969 | return SYMBOL_TYPE (result.symbol); | |
1970 | } | |
1971 | ||
3232fabd | 1972 | type = lookup_typename (current_parser->language (), current_parser->arch (), |
c44af4eb TT |
1973 | name, NULL, 1); |
1974 | if (type != NULL) | |
1975 | return type; | |
1976 | ||
1977 | /* Last chance, try a built-in type. */ | |
3232fabd TT |
1978 | return language_lookup_primitive_type (current_parser->language (), |
1979 | current_parser->arch (), | |
c44af4eb TT |
1980 | name); |
1981 | } | |
1982 | ||
1983 | static struct type *convert_ast_to_type (struct parser_state *state, | |
1984 | const struct rust_op *operation); | |
1985 | static const char *convert_name (struct parser_state *state, | |
1986 | const struct rust_op *operation); | |
1987 | ||
1988 | /* Convert a vector of rust_ops representing types to a vector of | |
1989 | types. */ | |
1990 | ||
8001f118 | 1991 | static std::vector<struct type *> |
3232fabd | 1992 | convert_params_to_types (struct parser_state *state, rust_op_vector *params) |
c44af4eb | 1993 | { |
8001f118 | 1994 | std::vector<struct type *> result; |
c44af4eb | 1995 | |
3232fabd | 1996 | for (const rust_op *op : *params) |
8001f118 | 1997 | result.push_back (convert_ast_to_type (state, op)); |
c44af4eb | 1998 | |
c44af4eb TT |
1999 | return result; |
2000 | } | |
2001 | ||
2002 | /* Convert a rust_op representing a type to a struct type *. */ | |
2003 | ||
2004 | static struct type * | |
2005 | convert_ast_to_type (struct parser_state *state, | |
2006 | const struct rust_op *operation) | |
2007 | { | |
2008 | struct type *type, *result = NULL; | |
2009 | ||
2010 | if (operation->opcode == OP_VAR_VALUE) | |
2011 | { | |
2012 | const char *varname = convert_name (state, operation); | |
2013 | ||
2014 | result = rust_lookup_type (varname, expression_context_block); | |
2015 | if (result == NULL) | |
2016 | error (_("No typed name '%s' in current context"), varname); | |
2017 | return result; | |
2018 | } | |
2019 | ||
2020 | gdb_assert (operation->opcode == OP_TYPE); | |
2021 | ||
2022 | switch (operation->typecode) | |
2023 | { | |
2024 | case TYPE_CODE_ARRAY: | |
2025 | type = convert_ast_to_type (state, operation->left.op); | |
2026 | if (operation->right.typed_val_int.val < 0) | |
2027 | error (_("Negative array length")); | |
2028 | result = lookup_array_range_type (type, 0, | |
2029 | operation->right.typed_val_int.val - 1); | |
2030 | break; | |
2031 | ||
2032 | case TYPE_CODE_COMPLEX: | |
2033 | { | |
2034 | struct type *usize = rust_type ("usize"); | |
2035 | ||
2036 | type = convert_ast_to_type (state, operation->left.op); | |
2037 | result = rust_slice_type ("&[*gdb*]", type, usize); | |
2038 | } | |
2039 | break; | |
2040 | ||
2041 | case TYPE_CODE_REF: | |
2042 | case TYPE_CODE_PTR: | |
2043 | /* For now we treat &x and *x identically. */ | |
2044 | type = convert_ast_to_type (state, operation->left.op); | |
2045 | result = lookup_pointer_type (type); | |
2046 | break; | |
2047 | ||
2048 | case TYPE_CODE_FUNC: | |
2049 | { | |
8001f118 | 2050 | std::vector<struct type *> args |
3232fabd | 2051 | (convert_params_to_types (state, operation->right.params)); |
c44af4eb TT |
2052 | struct type **argtypes = NULL; |
2053 | ||
2054 | type = convert_ast_to_type (state, operation->left.op); | |
8001f118 TT |
2055 | if (!args.empty ()) |
2056 | argtypes = args.data (); | |
c44af4eb TT |
2057 | |
2058 | result | |
8001f118 | 2059 | = lookup_function_type_with_arguments (type, args.size (), |
c44af4eb TT |
2060 | argtypes); |
2061 | result = lookup_pointer_type (result); | |
c44af4eb TT |
2062 | } |
2063 | break; | |
2064 | ||
2065 | case TYPE_CODE_STRUCT: | |
2066 | { | |
8001f118 | 2067 | std::vector<struct type *> args |
3232fabd | 2068 | (convert_params_to_types (state, operation->left.params)); |
c44af4eb | 2069 | int i; |
c44af4eb TT |
2070 | const char *name; |
2071 | ||
3232fabd | 2072 | obstack_1grow (work_obstack, '('); |
8001f118 | 2073 | for (i = 0; i < args.size (); ++i) |
c44af4eb | 2074 | { |
8001f118 | 2075 | std::string type_name = type_to_string (args[i]); |
c44af4eb TT |
2076 | |
2077 | if (i > 0) | |
3232fabd TT |
2078 | obstack_1grow (work_obstack, ','); |
2079 | obstack_grow_str (work_obstack, type_name.c_str ()); | |
c44af4eb TT |
2080 | } |
2081 | ||
3232fabd TT |
2082 | obstack_grow_str0 (work_obstack, ")"); |
2083 | name = (const char *) obstack_finish (work_obstack); | |
c44af4eb TT |
2084 | |
2085 | /* We don't allow creating new tuple types (yet), but we do | |
2086 | allow looking up existing tuple types. */ | |
2087 | result = rust_lookup_type (name, expression_context_block); | |
2088 | if (result == NULL) | |
2089 | error (_("could not find tuple type '%s'"), name); | |
c44af4eb TT |
2090 | } |
2091 | break; | |
2092 | ||
2093 | default: | |
2094 | gdb_assert_not_reached ("unhandled opcode in convert_ast_to_type"); | |
2095 | } | |
2096 | ||
2097 | gdb_assert (result != NULL); | |
2098 | return result; | |
2099 | } | |
2100 | ||
2101 | /* A helper function to turn a rust_op representing a name into a full | |
2102 | name. This applies generic arguments as needed. The returned name | |
2103 | is allocated on the work obstack. */ | |
2104 | ||
2105 | static const char * | |
2106 | convert_name (struct parser_state *state, const struct rust_op *operation) | |
2107 | { | |
c44af4eb | 2108 | int i; |
c44af4eb TT |
2109 | |
2110 | gdb_assert (operation->opcode == OP_VAR_VALUE); | |
2111 | ||
2112 | if (operation->right.params == NULL) | |
2113 | return operation->left.sval.ptr; | |
2114 | ||
8001f118 | 2115 | std::vector<struct type *> types |
3232fabd | 2116 | (convert_params_to_types (state, operation->right.params)); |
c44af4eb | 2117 | |
3232fabd TT |
2118 | obstack_grow_str (work_obstack, operation->left.sval.ptr); |
2119 | obstack_1grow (work_obstack, '<'); | |
8001f118 | 2120 | for (i = 0; i < types.size (); ++i) |
c44af4eb | 2121 | { |
8001f118 | 2122 | std::string type_name = type_to_string (types[i]); |
c44af4eb TT |
2123 | |
2124 | if (i > 0) | |
3232fabd | 2125 | obstack_1grow (work_obstack, ','); |
c44af4eb | 2126 | |
3232fabd | 2127 | obstack_grow_str (work_obstack, type_name.c_str ()); |
c44af4eb | 2128 | } |
3232fabd | 2129 | obstack_grow_str0 (work_obstack, ">"); |
c44af4eb | 2130 | |
3232fabd | 2131 | return (const char *) obstack_finish (work_obstack); |
c44af4eb TT |
2132 | } |
2133 | ||
2134 | static void convert_ast_to_expression (struct parser_state *state, | |
2135 | const struct rust_op *operation, | |
8880f2a9 TT |
2136 | const struct rust_op *top, |
2137 | bool want_type = false); | |
c44af4eb TT |
2138 | |
2139 | /* A helper function that converts a vec of rust_ops to a gdb | |
2140 | expression. */ | |
2141 | ||
2142 | static void | |
2143 | convert_params_to_expression (struct parser_state *state, | |
3232fabd | 2144 | rust_op_vector *params, |
c44af4eb TT |
2145 | const struct rust_op *top) |
2146 | { | |
3232fabd | 2147 | for (const rust_op *elem : *params) |
c44af4eb TT |
2148 | convert_ast_to_expression (state, elem, top); |
2149 | } | |
2150 | ||
2151 | /* Lower a rust_op to a gdb expression. STATE is the parser state. | |
2152 | OPERATION is the operation to lower. TOP is a pointer to the | |
2153 | top-most operation; it is used to handle the special case where the | |
2154 | top-most expression is an identifier and can be optionally lowered | |
8880f2a9 TT |
2155 | to OP_TYPE. WANT_TYPE is a flag indicating that, if the expression |
2156 | is the name of a type, then emit an OP_TYPE for it (rather than | |
2157 | erroring). If WANT_TYPE is set, then the similar TOP handling is | |
2158 | not done. */ | |
c44af4eb TT |
2159 | |
2160 | static void | |
2161 | convert_ast_to_expression (struct parser_state *state, | |
2162 | const struct rust_op *operation, | |
8880f2a9 TT |
2163 | const struct rust_op *top, |
2164 | bool want_type) | |
c44af4eb TT |
2165 | { |
2166 | switch (operation->opcode) | |
2167 | { | |
2168 | case OP_LONG: | |
2169 | write_exp_elt_opcode (state, OP_LONG); | |
2170 | write_exp_elt_type (state, operation->left.typed_val_int.type); | |
2171 | write_exp_elt_longcst (state, operation->left.typed_val_int.val); | |
2172 | write_exp_elt_opcode (state, OP_LONG); | |
2173 | break; | |
2174 | ||
edd079d9 UW |
2175 | case OP_FLOAT: |
2176 | write_exp_elt_opcode (state, OP_FLOAT); | |
c44af4eb | 2177 | write_exp_elt_type (state, operation->left.typed_val_float.type); |
edd079d9 UW |
2178 | write_exp_elt_floatcst (state, operation->left.typed_val_float.val); |
2179 | write_exp_elt_opcode (state, OP_FLOAT); | |
c44af4eb TT |
2180 | break; |
2181 | ||
2182 | case STRUCTOP_STRUCT: | |
2183 | { | |
2184 | convert_ast_to_expression (state, operation->left.op, top); | |
2185 | ||
2186 | if (operation->completing) | |
2187 | mark_struct_expression (state); | |
2188 | write_exp_elt_opcode (state, STRUCTOP_STRUCT); | |
2189 | write_exp_string (state, operation->right.sval); | |
2190 | write_exp_elt_opcode (state, STRUCTOP_STRUCT); | |
2191 | } | |
2192 | break; | |
2193 | ||
2194 | case STRUCTOP_ANONYMOUS: | |
2195 | { | |
2196 | convert_ast_to_expression (state, operation->left.op, top); | |
2197 | ||
2198 | write_exp_elt_opcode (state, STRUCTOP_ANONYMOUS); | |
2199 | write_exp_elt_longcst (state, operation->right.typed_val_int.val); | |
2200 | write_exp_elt_opcode (state, STRUCTOP_ANONYMOUS); | |
2201 | } | |
2202 | break; | |
2203 | ||
8880f2a9 TT |
2204 | case UNOP_SIZEOF: |
2205 | convert_ast_to_expression (state, operation->left.op, top, true); | |
2206 | write_exp_elt_opcode (state, UNOP_SIZEOF); | |
2207 | break; | |
2208 | ||
c44af4eb TT |
2209 | case UNOP_PLUS: |
2210 | case UNOP_NEG: | |
2211 | case UNOP_COMPLEMENT: | |
2212 | case UNOP_IND: | |
2213 | case UNOP_ADDR: | |
2214 | convert_ast_to_expression (state, operation->left.op, top); | |
2215 | write_exp_elt_opcode (state, operation->opcode); | |
2216 | break; | |
2217 | ||
2218 | case BINOP_SUBSCRIPT: | |
2219 | case BINOP_MUL: | |
2220 | case BINOP_REPEAT: | |
2221 | case BINOP_DIV: | |
2222 | case BINOP_REM: | |
2223 | case BINOP_LESS: | |
2224 | case BINOP_GTR: | |
2225 | case BINOP_BITWISE_AND: | |
2226 | case BINOP_BITWISE_IOR: | |
2227 | case BINOP_BITWISE_XOR: | |
2228 | case BINOP_ADD: | |
2229 | case BINOP_SUB: | |
2230 | case BINOP_LOGICAL_OR: | |
2231 | case BINOP_LOGICAL_AND: | |
2232 | case BINOP_EQUAL: | |
2233 | case BINOP_NOTEQUAL: | |
2234 | case BINOP_LEQ: | |
2235 | case BINOP_GEQ: | |
2236 | case BINOP_LSH: | |
2237 | case BINOP_RSH: | |
2238 | case BINOP_ASSIGN: | |
2239 | case OP_RUST_ARRAY: | |
2240 | convert_ast_to_expression (state, operation->left.op, top); | |
2241 | convert_ast_to_expression (state, operation->right.op, top); | |
2242 | if (operation->compound_assignment) | |
2243 | { | |
2244 | write_exp_elt_opcode (state, BINOP_ASSIGN_MODIFY); | |
2245 | write_exp_elt_opcode (state, operation->opcode); | |
2246 | write_exp_elt_opcode (state, BINOP_ASSIGN_MODIFY); | |
2247 | } | |
2248 | else | |
2249 | write_exp_elt_opcode (state, operation->opcode); | |
2250 | ||
2251 | if (operation->compound_assignment | |
2252 | || operation->opcode == BINOP_ASSIGN) | |
2253 | { | |
2254 | struct type *type; | |
2255 | ||
2256 | type = language_lookup_primitive_type (parse_language (state), | |
2257 | parse_gdbarch (state), | |
2258 | "()"); | |
2259 | ||
2260 | write_exp_elt_opcode (state, OP_LONG); | |
2261 | write_exp_elt_type (state, type); | |
2262 | write_exp_elt_longcst (state, 0); | |
2263 | write_exp_elt_opcode (state, OP_LONG); | |
2264 | ||
2265 | write_exp_elt_opcode (state, BINOP_COMMA); | |
2266 | } | |
2267 | break; | |
2268 | ||
2269 | case UNOP_CAST: | |
2270 | { | |
2271 | struct type *type = convert_ast_to_type (state, operation->right.op); | |
2272 | ||
2273 | convert_ast_to_expression (state, operation->left.op, top); | |
2274 | write_exp_elt_opcode (state, UNOP_CAST); | |
2275 | write_exp_elt_type (state, type); | |
2276 | write_exp_elt_opcode (state, UNOP_CAST); | |
2277 | } | |
2278 | break; | |
2279 | ||
2280 | case OP_FUNCALL: | |
2281 | { | |
2282 | if (operation->left.op->opcode == OP_VAR_VALUE) | |
2283 | { | |
2284 | struct type *type; | |
2285 | const char *varname = convert_name (state, operation->left.op); | |
2286 | ||
2287 | type = rust_lookup_type (varname, expression_context_block); | |
2288 | if (type != NULL) | |
2289 | { | |
2290 | /* This is actually a tuple struct expression, not a | |
2291 | call expression. */ | |
3232fabd | 2292 | rust_op_vector *params = operation->right.params; |
c44af4eb TT |
2293 | |
2294 | if (TYPE_CODE (type) != TYPE_CODE_NAMESPACE) | |
2295 | { | |
2296 | if (!rust_tuple_struct_type_p (type)) | |
2297 | error (_("Type %s is not a tuple struct"), varname); | |
2298 | ||
3232fabd | 2299 | for (int i = 0; i < params->size (); ++i) |
c44af4eb TT |
2300 | { |
2301 | char *cell = get_print_cell (); | |
2302 | ||
2303 | xsnprintf (cell, PRINT_CELL_SIZE, "__%d", i); | |
2304 | write_exp_elt_opcode (state, OP_NAME); | |
2305 | write_exp_string (state, make_stoken (cell)); | |
2306 | write_exp_elt_opcode (state, OP_NAME); | |
2307 | ||
3232fabd | 2308 | convert_ast_to_expression (state, (*params)[i], top); |
c44af4eb TT |
2309 | } |
2310 | ||
2311 | write_exp_elt_opcode (state, OP_AGGREGATE); | |
2312 | write_exp_elt_type (state, type); | |
3232fabd | 2313 | write_exp_elt_longcst (state, 2 * params->size ()); |
c44af4eb TT |
2314 | write_exp_elt_opcode (state, OP_AGGREGATE); |
2315 | break; | |
2316 | } | |
2317 | } | |
2318 | } | |
2319 | convert_ast_to_expression (state, operation->left.op, top); | |
3232fabd | 2320 | convert_params_to_expression (state, operation->right.params, top); |
c44af4eb | 2321 | write_exp_elt_opcode (state, OP_FUNCALL); |
3232fabd | 2322 | write_exp_elt_longcst (state, operation->right.params->size ()); |
c44af4eb TT |
2323 | write_exp_elt_longcst (state, OP_FUNCALL); |
2324 | } | |
2325 | break; | |
2326 | ||
2327 | case OP_ARRAY: | |
2328 | gdb_assert (operation->left.op == NULL); | |
3232fabd | 2329 | convert_params_to_expression (state, operation->right.params, top); |
c44af4eb TT |
2330 | write_exp_elt_opcode (state, OP_ARRAY); |
2331 | write_exp_elt_longcst (state, 0); | |
3232fabd | 2332 | write_exp_elt_longcst (state, operation->right.params->size () - 1); |
c44af4eb TT |
2333 | write_exp_elt_longcst (state, OP_ARRAY); |
2334 | break; | |
2335 | ||
2336 | case OP_VAR_VALUE: | |
2337 | { | |
2338 | struct block_symbol sym; | |
2339 | const char *varname; | |
2340 | ||
2341 | if (operation->left.sval.ptr[0] == '$') | |
2342 | { | |
2343 | write_dollar_variable (state, operation->left.sval); | |
2344 | break; | |
2345 | } | |
2346 | ||
2347 | varname = convert_name (state, operation); | |
2348 | sym = rust_lookup_symbol (varname, expression_context_block, | |
2349 | VAR_DOMAIN); | |
65547233 | 2350 | if (sym.symbol != NULL && SYMBOL_CLASS (sym.symbol) != LOC_TYPEDEF) |
c44af4eb TT |
2351 | { |
2352 | write_exp_elt_opcode (state, OP_VAR_VALUE); | |
2353 | write_exp_elt_block (state, sym.block); | |
2354 | write_exp_elt_sym (state, sym.symbol); | |
2355 | write_exp_elt_opcode (state, OP_VAR_VALUE); | |
2356 | } | |
2357 | else | |
2358 | { | |
65547233 | 2359 | struct type *type = NULL; |
c44af4eb | 2360 | |
65547233 TT |
2361 | if (sym.symbol != NULL) |
2362 | { | |
2363 | gdb_assert (SYMBOL_CLASS (sym.symbol) == LOC_TYPEDEF); | |
2364 | type = SYMBOL_TYPE (sym.symbol); | |
2365 | } | |
2366 | if (type == NULL) | |
2367 | type = rust_lookup_type (varname, expression_context_block); | |
c44af4eb TT |
2368 | if (type == NULL) |
2369 | error (_("No symbol '%s' in current context"), varname); | |
2370 | ||
8880f2a9 TT |
2371 | if (!want_type |
2372 | && TYPE_CODE (type) == TYPE_CODE_STRUCT | |
c44af4eb TT |
2373 | && TYPE_NFIELDS (type) == 0) |
2374 | { | |
2375 | /* A unit-like struct. */ | |
2376 | write_exp_elt_opcode (state, OP_AGGREGATE); | |
2377 | write_exp_elt_type (state, type); | |
2378 | write_exp_elt_longcst (state, 0); | |
2379 | write_exp_elt_opcode (state, OP_AGGREGATE); | |
2380 | } | |
8880f2a9 | 2381 | else if (want_type || operation == top) |
c44af4eb TT |
2382 | { |
2383 | write_exp_elt_opcode (state, OP_TYPE); | |
2384 | write_exp_elt_type (state, type); | |
2385 | write_exp_elt_opcode (state, OP_TYPE); | |
c44af4eb | 2386 | } |
8880f2a9 TT |
2387 | else |
2388 | error (_("Found type '%s', which can't be " | |
2389 | "evaluated in this context"), | |
2390 | varname); | |
c44af4eb TT |
2391 | } |
2392 | } | |
2393 | break; | |
2394 | ||
2395 | case OP_AGGREGATE: | |
2396 | { | |
c44af4eb | 2397 | int length; |
3232fabd | 2398 | rust_set_vector *fields = operation->right.field_inits; |
c44af4eb TT |
2399 | struct type *type; |
2400 | const char *name; | |
2401 | ||
2402 | length = 0; | |
3232fabd | 2403 | for (const set_field &init : *fields) |
c44af4eb | 2404 | { |
3232fabd | 2405 | if (init.name.ptr != NULL) |
c44af4eb TT |
2406 | { |
2407 | write_exp_elt_opcode (state, OP_NAME); | |
3232fabd | 2408 | write_exp_string (state, init.name); |
c44af4eb TT |
2409 | write_exp_elt_opcode (state, OP_NAME); |
2410 | ++length; | |
2411 | } | |
2412 | ||
3232fabd | 2413 | convert_ast_to_expression (state, init.init, top); |
c44af4eb TT |
2414 | ++length; |
2415 | ||
3232fabd | 2416 | if (init.name.ptr == NULL) |
c44af4eb TT |
2417 | { |
2418 | /* This is handled differently from Ada in our | |
2419 | evaluator. */ | |
2420 | write_exp_elt_opcode (state, OP_OTHERS); | |
2421 | } | |
2422 | } | |
2423 | ||
2424 | name = convert_name (state, operation->left.op); | |
2425 | type = rust_lookup_type (name, expression_context_block); | |
2426 | if (type == NULL) | |
2427 | error (_("Could not find type '%s'"), operation->left.sval.ptr); | |
2428 | ||
2429 | if (TYPE_CODE (type) != TYPE_CODE_STRUCT | |
2430 | || rust_tuple_type_p (type) | |
2431 | || rust_tuple_struct_type_p (type)) | |
2432 | error (_("Struct expression applied to non-struct type")); | |
2433 | ||
2434 | write_exp_elt_opcode (state, OP_AGGREGATE); | |
2435 | write_exp_elt_type (state, type); | |
2436 | write_exp_elt_longcst (state, length); | |
2437 | write_exp_elt_opcode (state, OP_AGGREGATE); | |
2438 | } | |
2439 | break; | |
2440 | ||
2441 | case OP_STRING: | |
2442 | { | |
2443 | write_exp_elt_opcode (state, OP_STRING); | |
2444 | write_exp_string (state, operation->left.sval); | |
2445 | write_exp_elt_opcode (state, OP_STRING); | |
2446 | } | |
2447 | break; | |
2448 | ||
01739a3b | 2449 | case OP_RANGE: |
c44af4eb | 2450 | { |
01739a3b | 2451 | enum range_type kind = BOTH_BOUND_DEFAULT; |
c44af4eb TT |
2452 | |
2453 | if (operation->left.op != NULL) | |
2454 | { | |
2455 | convert_ast_to_expression (state, operation->left.op, top); | |
2456 | kind = HIGH_BOUND_DEFAULT; | |
2457 | } | |
2458 | if (operation->right.op != NULL) | |
2459 | { | |
2460 | convert_ast_to_expression (state, operation->right.op, top); | |
2461 | if (kind == BOTH_BOUND_DEFAULT) | |
2462 | kind = LOW_BOUND_DEFAULT; | |
2463 | else | |
2464 | { | |
2465 | gdb_assert (kind == HIGH_BOUND_DEFAULT); | |
2466 | kind = NONE_BOUND_DEFAULT; | |
2467 | } | |
2468 | } | |
01739a3b | 2469 | write_exp_elt_opcode (state, OP_RANGE); |
c44af4eb | 2470 | write_exp_elt_longcst (state, kind); |
01739a3b | 2471 | write_exp_elt_opcode (state, OP_RANGE); |
c44af4eb TT |
2472 | } |
2473 | break; | |
2474 | ||
2475 | default: | |
2476 | gdb_assert_not_reached ("unhandled opcode in convert_ast_to_expression"); | |
2477 | } | |
2478 | } | |
2479 | ||
2480 | \f | |
2481 | ||
2482 | /* The parser as exposed to gdb. */ | |
2483 | ||
2484 | int | |
2485 | rust_parse (struct parser_state *state) | |
2486 | { | |
2487 | int result; | |
c44af4eb | 2488 | |
3232fabd TT |
2489 | /* This sets various globals and also clears them on |
2490 | destruction. */ | |
2491 | rust_parser parser (state); | |
8268c778 | 2492 | |
c44af4eb TT |
2493 | result = rustyyparse (); |
2494 | ||
3232fabd TT |
2495 | if (!result || (parse_completion && parser.rust_ast != NULL)) |
2496 | convert_ast_to_expression (state, parser.rust_ast, parser.rust_ast); | |
c44af4eb | 2497 | |
c44af4eb TT |
2498 | return result; |
2499 | } | |
2500 | ||
2501 | /* The parser error handler. */ | |
2502 | ||
2503 | void | |
a121b7c1 | 2504 | rustyyerror (const char *msg) |
c44af4eb TT |
2505 | { |
2506 | const char *where = prev_lexptr ? prev_lexptr : lexptr; | |
2507 | error (_("%s in expression, near `%s'."), (msg ? msg : "Error"), where); | |
2508 | } | |
2509 | ||
2510 | \f | |
2511 | ||
2512 | #if GDB_SELF_TEST | |
2513 | ||
2514 | /* Initialize the lexer for testing. */ | |
2515 | ||
2516 | static void | |
2517 | rust_lex_test_init (const char *input) | |
2518 | { | |
2519 | prev_lexptr = NULL; | |
2520 | lexptr = input; | |
2521 | paren_depth = 0; | |
2522 | } | |
2523 | ||
2524 | /* A test helper that lexes a string, expecting a single token. It | |
2525 | returns the lexer data for this token. */ | |
2526 | ||
2527 | static RUSTSTYPE | |
2528 | rust_lex_test_one (const char *input, int expected) | |
2529 | { | |
2530 | int token; | |
2531 | RUSTSTYPE result; | |
2532 | ||
2533 | rust_lex_test_init (input); | |
2534 | ||
2535 | token = rustyylex (); | |
2536 | SELF_CHECK (token == expected); | |
2537 | result = rustyylval; | |
2538 | ||
2539 | if (token) | |
2540 | { | |
2541 | token = rustyylex (); | |
2542 | SELF_CHECK (token == 0); | |
2543 | } | |
2544 | ||
2545 | return result; | |
2546 | } | |
2547 | ||
2548 | /* Test that INPUT lexes as the integer VALUE. */ | |
2549 | ||
2550 | static void | |
2551 | rust_lex_int_test (const char *input, int value, int kind) | |
2552 | { | |
2553 | RUSTSTYPE result = rust_lex_test_one (input, kind); | |
2554 | SELF_CHECK (result.typed_val_int.val == value); | |
2555 | } | |
2556 | ||
2557 | /* Test that INPUT throws an exception with text ERR. */ | |
2558 | ||
2559 | static void | |
2560 | rust_lex_exception_test (const char *input, const char *err) | |
2561 | { | |
2562 | TRY | |
2563 | { | |
2564 | /* The "kind" doesn't matter. */ | |
2565 | rust_lex_test_one (input, DECIMAL_INTEGER); | |
2566 | SELF_CHECK (0); | |
2567 | } | |
2568 | CATCH (except, RETURN_MASK_ERROR) | |
2569 | { | |
2570 | SELF_CHECK (strcmp (except.message, err) == 0); | |
2571 | } | |
2572 | END_CATCH | |
2573 | } | |
2574 | ||
2575 | /* Test that INPUT lexes as the identifier, string, or byte-string | |
2576 | VALUE. KIND holds the expected token kind. */ | |
2577 | ||
2578 | static void | |
2579 | rust_lex_stringish_test (const char *input, const char *value, int kind) | |
2580 | { | |
2581 | RUSTSTYPE result = rust_lex_test_one (input, kind); | |
2582 | SELF_CHECK (result.sval.length == strlen (value)); | |
2583 | SELF_CHECK (strncmp (result.sval.ptr, value, result.sval.length) == 0); | |
2584 | } | |
2585 | ||
2586 | /* Helper to test that a string parses as a given token sequence. */ | |
2587 | ||
2588 | static void | |
2589 | rust_lex_test_sequence (const char *input, int len, const int expected[]) | |
2590 | { | |
2591 | int i; | |
2592 | ||
2593 | lexptr = input; | |
2594 | paren_depth = 0; | |
2595 | ||
2596 | for (i = 0; i < len; ++i) | |
2597 | { | |
2598 | int token = rustyylex (); | |
2599 | ||
2600 | SELF_CHECK (token == expected[i]); | |
2601 | } | |
2602 | } | |
2603 | ||
2604 | /* Tests for an integer-parsing corner case. */ | |
2605 | ||
2606 | static void | |
2607 | rust_lex_test_trailing_dot (void) | |
2608 | { | |
2609 | const int expected1[] = { DECIMAL_INTEGER, '.', IDENT, '(', ')', 0 }; | |
2610 | const int expected2[] = { INTEGER, '.', IDENT, '(', ')', 0 }; | |
2611 | const int expected3[] = { FLOAT, EQEQ, '(', ')', 0 }; | |
2612 | const int expected4[] = { DECIMAL_INTEGER, DOTDOT, DECIMAL_INTEGER, 0 }; | |
2613 | ||
2614 | rust_lex_test_sequence ("23.g()", ARRAY_SIZE (expected1), expected1); | |
2615 | rust_lex_test_sequence ("23_0.g()", ARRAY_SIZE (expected2), expected2); | |
2616 | rust_lex_test_sequence ("23.==()", ARRAY_SIZE (expected3), expected3); | |
2617 | rust_lex_test_sequence ("23..25", ARRAY_SIZE (expected4), expected4); | |
2618 | } | |
2619 | ||
2620 | /* Tests of completion. */ | |
2621 | ||
2622 | static void | |
2623 | rust_lex_test_completion (void) | |
2624 | { | |
2625 | const int expected[] = { IDENT, '.', COMPLETE, 0 }; | |
2626 | ||
2627 | parse_completion = 1; | |
2628 | ||
2629 | rust_lex_test_sequence ("something.wha", ARRAY_SIZE (expected), expected); | |
2630 | rust_lex_test_sequence ("something.", ARRAY_SIZE (expected), expected); | |
2631 | ||
2632 | parse_completion = 0; | |
2633 | } | |
2634 | ||
2635 | /* Test pushback. */ | |
2636 | ||
2637 | static void | |
2638 | rust_lex_test_push_back (void) | |
2639 | { | |
2640 | int token; | |
2641 | ||
2642 | rust_lex_test_init (">>="); | |
2643 | ||
2644 | token = rustyylex (); | |
2645 | SELF_CHECK (token == COMPOUND_ASSIGN); | |
2646 | SELF_CHECK (rustyylval.opcode == BINOP_RSH); | |
2647 | ||
2648 | rust_push_back ('='); | |
2649 | ||
2650 | token = rustyylex (); | |
2651 | SELF_CHECK (token == '='); | |
2652 | ||
2653 | token = rustyylex (); | |
2654 | SELF_CHECK (token == 0); | |
2655 | } | |
2656 | ||
2657 | /* Unit test the lexer. */ | |
2658 | ||
2659 | static void | |
2660 | rust_lex_tests (void) | |
2661 | { | |
2662 | int i; | |
2663 | ||
3232fabd TT |
2664 | auto_obstack test_obstack; |
2665 | scoped_restore obstack_holder = make_scoped_restore (&work_obstack, | |
2666 | &test_obstack); | |
2667 | ||
edd079d9 | 2668 | // Set up dummy "parser", so that rust_type works. |
e9d9f57e | 2669 | struct parser_state ps (0, &rust_language_defn, target_gdbarch ()); |
edd079d9 | 2670 | rust_parser parser (&ps); |
c44af4eb TT |
2671 | |
2672 | rust_lex_test_one ("", 0); | |
2673 | rust_lex_test_one (" \t \n \r ", 0); | |
2674 | rust_lex_test_one ("thread 23", 0); | |
2675 | rust_lex_test_one ("task 23", 0); | |
2676 | rust_lex_test_one ("th 104", 0); | |
2677 | rust_lex_test_one ("ta 97", 0); | |
2678 | ||
2679 | rust_lex_int_test ("'z'", 'z', INTEGER); | |
2680 | rust_lex_int_test ("'\\xff'", 0xff, INTEGER); | |
2681 | rust_lex_int_test ("'\\u{1016f}'", 0x1016f, INTEGER); | |
2682 | rust_lex_int_test ("b'z'", 'z', INTEGER); | |
2683 | rust_lex_int_test ("b'\\xfe'", 0xfe, INTEGER); | |
2684 | rust_lex_int_test ("b'\\xFE'", 0xfe, INTEGER); | |
2685 | rust_lex_int_test ("b'\\xfE'", 0xfe, INTEGER); | |
2686 | ||
2687 | /* Test all escapes in both modes. */ | |
2688 | rust_lex_int_test ("'\\n'", '\n', INTEGER); | |
2689 | rust_lex_int_test ("'\\r'", '\r', INTEGER); | |
2690 | rust_lex_int_test ("'\\t'", '\t', INTEGER); | |
2691 | rust_lex_int_test ("'\\\\'", '\\', INTEGER); | |
2692 | rust_lex_int_test ("'\\0'", '\0', INTEGER); | |
2693 | rust_lex_int_test ("'\\''", '\'', INTEGER); | |
2694 | rust_lex_int_test ("'\\\"'", '"', INTEGER); | |
2695 | ||
2696 | rust_lex_int_test ("b'\\n'", '\n', INTEGER); | |
2697 | rust_lex_int_test ("b'\\r'", '\r', INTEGER); | |
2698 | rust_lex_int_test ("b'\\t'", '\t', INTEGER); | |
2699 | rust_lex_int_test ("b'\\\\'", '\\', INTEGER); | |
2700 | rust_lex_int_test ("b'\\0'", '\0', INTEGER); | |
2701 | rust_lex_int_test ("b'\\''", '\'', INTEGER); | |
2702 | rust_lex_int_test ("b'\\\"'", '"', INTEGER); | |
2703 | ||
2704 | rust_lex_exception_test ("'z", "Unterminated character literal"); | |
2705 | rust_lex_exception_test ("b'\\x0'", "Not enough hex digits seen"); | |
2706 | rust_lex_exception_test ("b'\\u{0}'", "Unicode escape in byte literal"); | |
2707 | rust_lex_exception_test ("'\\x0'", "Not enough hex digits seen"); | |
2708 | rust_lex_exception_test ("'\\u0'", "Missing '{' in Unicode escape"); | |
2709 | rust_lex_exception_test ("'\\u{0", "Missing '}' in Unicode escape"); | |
2710 | rust_lex_exception_test ("'\\u{0000007}", "Overlong hex escape"); | |
2711 | rust_lex_exception_test ("'\\u{}", "Not enough hex digits seen"); | |
2712 | rust_lex_exception_test ("'\\Q'", "Invalid escape \\Q in literal"); | |
2713 | rust_lex_exception_test ("b'\\Q'", "Invalid escape \\Q in literal"); | |
2714 | ||
2715 | rust_lex_int_test ("23", 23, DECIMAL_INTEGER); | |
2716 | rust_lex_int_test ("2_344__29", 234429, INTEGER); | |
2717 | rust_lex_int_test ("0x1f", 0x1f, INTEGER); | |
2718 | rust_lex_int_test ("23usize", 23, INTEGER); | |
2719 | rust_lex_int_test ("23i32", 23, INTEGER); | |
2720 | rust_lex_int_test ("0x1_f", 0x1f, INTEGER); | |
2721 | rust_lex_int_test ("0b1_101011__", 0x6b, INTEGER); | |
2722 | rust_lex_int_test ("0o001177i64", 639, INTEGER); | |
2723 | ||
2724 | rust_lex_test_trailing_dot (); | |
2725 | ||
2726 | rust_lex_test_one ("23.", FLOAT); | |
2727 | rust_lex_test_one ("23.99f32", FLOAT); | |
2728 | rust_lex_test_one ("23e7", FLOAT); | |
2729 | rust_lex_test_one ("23E-7", FLOAT); | |
2730 | rust_lex_test_one ("23e+7", FLOAT); | |
2731 | rust_lex_test_one ("23.99e+7f64", FLOAT); | |
2732 | rust_lex_test_one ("23.82f32", FLOAT); | |
2733 | ||
2734 | rust_lex_stringish_test ("hibob", "hibob", IDENT); | |
2735 | rust_lex_stringish_test ("hibob__93", "hibob__93", IDENT); | |
2736 | rust_lex_stringish_test ("thread", "thread", IDENT); | |
2737 | ||
2738 | rust_lex_stringish_test ("\"string\"", "string", STRING); | |
2739 | rust_lex_stringish_test ("\"str\\ting\"", "str\ting", STRING); | |
2740 | rust_lex_stringish_test ("\"str\\\"ing\"", "str\"ing", STRING); | |
2741 | rust_lex_stringish_test ("r\"str\\ing\"", "str\\ing", STRING); | |
2742 | rust_lex_stringish_test ("r#\"str\\ting\"#", "str\\ting", STRING); | |
2743 | rust_lex_stringish_test ("r###\"str\\\"ing\"###", "str\\\"ing", STRING); | |
2744 | ||
2745 | rust_lex_stringish_test ("b\"string\"", "string", BYTESTRING); | |
2746 | rust_lex_stringish_test ("b\"\x73tring\"", "string", BYTESTRING); | |
2747 | rust_lex_stringish_test ("b\"str\\\"ing\"", "str\"ing", BYTESTRING); | |
2748 | rust_lex_stringish_test ("br####\"\\x73tring\"####", "\\x73tring", | |
2749 | BYTESTRING); | |
2750 | ||
2751 | for (i = 0; i < ARRAY_SIZE (identifier_tokens); ++i) | |
2752 | rust_lex_test_one (identifier_tokens[i].name, identifier_tokens[i].value); | |
2753 | ||
2754 | for (i = 0; i < ARRAY_SIZE (operator_tokens); ++i) | |
2755 | rust_lex_test_one (operator_tokens[i].name, operator_tokens[i].value); | |
2756 | ||
2757 | rust_lex_test_completion (); | |
2758 | rust_lex_test_push_back (); | |
c44af4eb TT |
2759 | } |
2760 | ||
2761 | #endif /* GDB_SELF_TEST */ | |
2762 | ||
2763 | void | |
2764 | _initialize_rust_exp (void) | |
2765 | { | |
2766 | int code = regcomp (&number_regex, number_regex_text, REG_EXTENDED); | |
2767 | /* If the regular expression was incorrect, it was a programming | |
2768 | error. */ | |
2769 | gdb_assert (code == 0); | |
2770 | ||
2771 | #if GDB_SELF_TEST | |
1526853e | 2772 | selftests::register_test ("rust-lex", rust_lex_tests); |
c44af4eb TT |
2773 | #endif |
2774 | } |