]>
Commit | Line | Data |
---|---|---|
a766d390 DE |
1 | /* YACC parser for Go expressions, for GDB. |
2 | ||
213516ef | 3 | Copyright (C) 2012-2023 Free Software Foundation, Inc. |
a766d390 DE |
4 | |
5 | This file is part of GDB. | |
6 | ||
7 | This program is free software; you can redistribute it and/or modify | |
8 | it under the terms of the GNU General Public License as published by | |
9 | the Free Software Foundation; either version 3 of the License, or | |
10 | (at your option) any later version. | |
11 | ||
12 | This program is distributed in the hope that it will be useful, | |
13 | but WITHOUT ANY WARRANTY; without even the implied warranty of | |
14 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
15 | GNU General Public License for more details. | |
16 | ||
17 | You should have received a copy of the GNU General Public License | |
18 | along with this program. If not, see <http://www.gnu.org/licenses/>. */ | |
19 | ||
20 | /* This file is derived from c-exp.y, p-exp.y. */ | |
21 | ||
22 | /* Parse a Go expression from text in a string, | |
23 | and return the result as a struct expression pointer. | |
24 | That structure contains arithmetic operations in reverse polish, | |
25 | with constants represented by operations that are followed by special data. | |
26 | See expression.h for the details of the format. | |
27 | What is important here is that it can be built up sequentially | |
28 | during the process of parsing; the lower levels of the tree always | |
29 | come first in the result. | |
30 | ||
31 | Note that malloc's and realloc's in this file are transformed to | |
32 | xmalloc and xrealloc respectively by the same sed command in the | |
33 | makefile that remaps any other malloc/realloc inserted by the parser | |
34 | generator. Doing this with #defines and trying to control the interaction | |
35 | with include files (<malloc.h> and <stdlib.h> for example) just became | |
36 | too messy, particularly when such includes can be inserted at random | |
37 | times by the parser generator. */ | |
38 | ||
39 | /* Known bugs or limitations: | |
40 | ||
41 | - Unicode | |
42 | - &^ | |
43 | - '_' (blank identifier) | |
44 | - automatic deref of pointers | |
45 | - method expressions | |
46 | - interfaces, channels, etc. | |
47 | ||
48 | And lots of other things. | |
49 | I'm sure there's some cleanup to do. | |
50 | */ | |
51 | ||
52 | %{ | |
53 | ||
54 | #include "defs.h" | |
a766d390 DE |
55 | #include <ctype.h> |
56 | #include "expression.h" | |
57 | #include "value.h" | |
58 | #include "parser-defs.h" | |
59 | #include "language.h" | |
60 | #include "c-lang.h" | |
61 | #include "go-lang.h" | |
a766d390 DE |
62 | #include "charset.h" |
63 | #include "block.h" | |
bb4e0574 | 64 | #include "expop.h" |
a766d390 | 65 | |
fa9f5be6 | 66 | #define parse_type(ps) builtin_type (ps->gdbarch ()) |
a766d390 | 67 | |
b3f11165 PA |
68 | /* Remap normal yacc parser interface names (yyparse, yylex, yyerror, |
69 | etc). */ | |
70 | #define GDB_YY_REMAP_PREFIX go_ | |
71 | #include "yy-remap.h" | |
a766d390 | 72 | |
410a0ff2 SDJ |
73 | /* The state of the parser, used internally when we are parsing the |
74 | expression. */ | |
75 | ||
76 | static struct parser_state *pstate = NULL; | |
77 | ||
a766d390 DE |
78 | int yyparse (void); |
79 | ||
80 | static int yylex (void); | |
81 | ||
69d340c6 | 82 | static void yyerror (const char *); |
a766d390 DE |
83 | |
84 | %} | |
85 | ||
86 | /* Although the yacc "value" of an expression is not used, | |
87 | since the result is stored in the structure being created, | |
88 | other node types do have values. */ | |
89 | ||
90 | %union | |
91 | { | |
92 | LONGEST lval; | |
93 | struct { | |
94 | LONGEST val; | |
95 | struct type *type; | |
96 | } typed_val_int; | |
97 | struct { | |
edd079d9 | 98 | gdb_byte val[16]; |
a766d390 DE |
99 | struct type *type; |
100 | } typed_val_float; | |
101 | struct stoken sval; | |
102 | struct symtoken ssym; | |
103 | struct type *tval; | |
104 | struct typed_stoken tsval; | |
105 | struct ttype tsym; | |
106 | int voidval; | |
107 | enum exp_opcode opcode; | |
108 | struct internalvar *ivar; | |
109 | struct stoken_vector svec; | |
110 | } | |
111 | ||
112 | %{ | |
113 | /* YYSTYPE gets defined by %union. */ | |
410a0ff2 SDJ |
114 | static int parse_number (struct parser_state *, |
115 | const char *, int, int, YYSTYPE *); | |
bb4e0574 TT |
116 | |
117 | using namespace expr; | |
a766d390 DE |
118 | %} |
119 | ||
120 | %type <voidval> exp exp1 type_exp start variable lcurly | |
121 | %type <lval> rcurly | |
122 | %type <tval> type | |
123 | ||
124 | %token <typed_val_int> INT | |
125 | %token <typed_val_float> FLOAT | |
126 | ||
127 | /* Both NAME and TYPENAME tokens represent symbols in the input, | |
128 | and both convey their data as strings. | |
129 | But a TYPENAME is a string that happens to be defined as a type | |
130 | or builtin type name (such as int or char) | |
131 | and a NAME is any other symbol. | |
132 | Contexts where this distinction is not important can use the | |
133 | nonterminal "name", which matches either NAME or TYPENAME. */ | |
134 | ||
135 | %token <tsval> RAW_STRING | |
136 | %token <tsval> STRING | |
137 | %token <tsval> CHAR | |
138 | %token <ssym> NAME | |
139 | %token <tsym> TYPENAME /* Not TYPE_NAME cus already taken. */ | |
140 | %token <voidval> COMPLETE | |
141 | /*%type <sval> name*/ | |
142 | %type <svec> string_exp | |
143 | %type <ssym> name_not_typename | |
144 | ||
145 | /* A NAME_OR_INT is a symbol which is not known in the symbol table, | |
146 | but which would parse as a valid number in the current input radix. | |
147 | E.g. "c" when input_radix==16. Depending on the parse, it will be | |
148 | turned into a name or into a number. */ | |
149 | %token <ssym> NAME_OR_INT | |
150 | ||
151 | %token <lval> TRUE_KEYWORD FALSE_KEYWORD | |
152 | %token STRUCT_KEYWORD INTERFACE_KEYWORD TYPE_KEYWORD CHAN_KEYWORD | |
153 | %token SIZEOF_KEYWORD | |
154 | %token LEN_KEYWORD CAP_KEYWORD | |
155 | %token NEW_KEYWORD | |
156 | %token IOTA_KEYWORD NIL_KEYWORD | |
157 | %token CONST_KEYWORD | |
158 | %token DOTDOTDOT | |
159 | %token ENTRY | |
160 | %token ERROR | |
161 | ||
162 | /* Special type cases. */ | |
163 | %token BYTE_KEYWORD /* An alias of uint8. */ | |
164 | ||
165 | %token <sval> DOLLAR_VARIABLE | |
166 | ||
167 | %token <opcode> ASSIGN_MODIFY | |
168 | ||
169 | %left ',' | |
170 | %left ABOVE_COMMA | |
171 | %right '=' ASSIGN_MODIFY | |
172 | %right '?' | |
173 | %left OROR | |
174 | %left ANDAND | |
175 | %left '|' | |
176 | %left '^' | |
177 | %left '&' | |
178 | %left ANDNOT | |
179 | %left EQUAL NOTEQUAL | |
180 | %left '<' '>' LEQ GEQ | |
181 | %left LSH RSH | |
182 | %left '@' | |
183 | %left '+' '-' | |
184 | %left '*' '/' '%' | |
185 | %right UNARY INCREMENT DECREMENT | |
186 | %right LEFT_ARROW '.' '[' '(' | |
187 | ||
188 | \f | |
189 | %% | |
190 | ||
191 | start : exp1 | |
192 | | type_exp | |
193 | ; | |
194 | ||
195 | type_exp: type | |
bb4e0574 | 196 | { pstate->push_new<type_operation> ($1); } |
a766d390 DE |
197 | ; |
198 | ||
199 | /* Expressions, including the comma operator. */ | |
200 | exp1 : exp | |
201 | | exp1 ',' exp | |
bb4e0574 | 202 | { pstate->wrap2<comma_operation> (); } |
a766d390 DE |
203 | ; |
204 | ||
205 | /* Expressions, not including the comma operator. */ | |
206 | exp : '*' exp %prec UNARY | |
bb4e0574 | 207 | { pstate->wrap<unop_ind_operation> (); } |
a766d390 DE |
208 | ; |
209 | ||
210 | exp : '&' exp %prec UNARY | |
bb4e0574 | 211 | { pstate->wrap<unop_addr_operation> (); } |
a766d390 DE |
212 | ; |
213 | ||
214 | exp : '-' exp %prec UNARY | |
bb4e0574 | 215 | { pstate->wrap<unary_neg_operation> (); } |
a766d390 DE |
216 | ; |
217 | ||
218 | exp : '+' exp %prec UNARY | |
bb4e0574 | 219 | { pstate->wrap<unary_plus_operation> (); } |
a766d390 DE |
220 | ; |
221 | ||
222 | exp : '!' exp %prec UNARY | |
bb4e0574 | 223 | { pstate->wrap<unary_logical_not_operation> (); } |
a766d390 DE |
224 | ; |
225 | ||
226 | exp : '^' exp %prec UNARY | |
bb4e0574 | 227 | { pstate->wrap<unary_complement_operation> (); } |
a766d390 DE |
228 | ; |
229 | ||
230 | exp : exp INCREMENT %prec UNARY | |
bb4e0574 | 231 | { pstate->wrap<postinc_operation> (); } |
a766d390 DE |
232 | ; |
233 | ||
234 | exp : exp DECREMENT %prec UNARY | |
bb4e0574 | 235 | { pstate->wrap<postdec_operation> (); } |
a766d390 DE |
236 | ; |
237 | ||
238 | /* foo->bar is not in Go. May want as a gdb extension. Later. */ | |
239 | ||
240 | exp : exp '.' name_not_typename | |
bb4e0574 TT |
241 | { |
242 | pstate->push_new<structop_operation> | |
243 | (pstate->pop (), copy_name ($3.stoken)); | |
244 | } | |
a766d390 DE |
245 | ; |
246 | ||
247 | exp : exp '.' name_not_typename COMPLETE | |
bb4e0574 TT |
248 | { |
249 | structop_base_operation *op | |
250 | = new structop_operation (pstate->pop (), | |
251 | copy_name ($3.stoken)); | |
252 | pstate->mark_struct_expression (op); | |
253 | pstate->push (operation_up (op)); | |
254 | } | |
a766d390 DE |
255 | ; |
256 | ||
257 | exp : exp '.' COMPLETE | |
bb4e0574 TT |
258 | { |
259 | structop_base_operation *op | |
260 | = new structop_operation (pstate->pop (), ""); | |
261 | pstate->mark_struct_expression (op); | |
262 | pstate->push (operation_up (op)); | |
263 | } | |
a766d390 DE |
264 | ; |
265 | ||
266 | exp : exp '[' exp1 ']' | |
bb4e0574 | 267 | { pstate->wrap2<subscript_operation> (); } |
a766d390 DE |
268 | ; |
269 | ||
270 | exp : exp '(' | |
271 | /* This is to save the value of arglist_len | |
272 | being accumulated by an outer function call. */ | |
43476f0b | 273 | { pstate->start_arglist (); } |
a766d390 | 274 | arglist ')' %prec LEFT_ARROW |
bb4e0574 TT |
275 | { |
276 | std::vector<operation_up> args | |
277 | = pstate->pop_vector (pstate->end_arglist ()); | |
278 | pstate->push_new<funcall_operation> | |
279 | (pstate->pop (), std::move (args)); | |
280 | } | |
a766d390 DE |
281 | ; |
282 | ||
283 | lcurly : '{' | |
43476f0b | 284 | { pstate->start_arglist (); } |
a766d390 DE |
285 | ; |
286 | ||
287 | arglist : | |
288 | ; | |
289 | ||
290 | arglist : exp | |
43476f0b | 291 | { pstate->arglist_len = 1; } |
a766d390 DE |
292 | ; |
293 | ||
294 | arglist : arglist ',' exp %prec ABOVE_COMMA | |
43476f0b | 295 | { pstate->arglist_len++; } |
a766d390 DE |
296 | ; |
297 | ||
298 | rcurly : '}' | |
43476f0b | 299 | { $$ = pstate->end_arglist () - 1; } |
a766d390 DE |
300 | ; |
301 | ||
302 | exp : lcurly type rcurly exp %prec UNARY | |
bb4e0574 TT |
303 | { |
304 | pstate->push_new<unop_memval_operation> | |
305 | (pstate->pop (), $2); | |
306 | } | |
a766d390 DE |
307 | ; |
308 | ||
309 | exp : type '(' exp ')' %prec UNARY | |
bb4e0574 TT |
310 | { |
311 | pstate->push_new<unop_cast_operation> | |
312 | (pstate->pop (), $1); | |
313 | } | |
a766d390 DE |
314 | ; |
315 | ||
316 | exp : '(' exp1 ')' | |
317 | { } | |
318 | ; | |
319 | ||
320 | /* Binary operators in order of decreasing precedence. */ | |
321 | ||
322 | exp : exp '@' exp | |
bb4e0574 | 323 | { pstate->wrap2<repeat_operation> (); } |
a766d390 DE |
324 | ; |
325 | ||
326 | exp : exp '*' exp | |
bb4e0574 | 327 | { pstate->wrap2<mul_operation> (); } |
a766d390 DE |
328 | ; |
329 | ||
330 | exp : exp '/' exp | |
bb4e0574 | 331 | { pstate->wrap2<div_operation> (); } |
a766d390 DE |
332 | ; |
333 | ||
334 | exp : exp '%' exp | |
bb4e0574 | 335 | { pstate->wrap2<rem_operation> (); } |
a766d390 DE |
336 | ; |
337 | ||
338 | exp : exp '+' exp | |
bb4e0574 | 339 | { pstate->wrap2<add_operation> (); } |
a766d390 DE |
340 | ; |
341 | ||
342 | exp : exp '-' exp | |
bb4e0574 | 343 | { pstate->wrap2<sub_operation> (); } |
a766d390 DE |
344 | ; |
345 | ||
346 | exp : exp LSH exp | |
bb4e0574 | 347 | { pstate->wrap2<lsh_operation> (); } |
a766d390 DE |
348 | ; |
349 | ||
350 | exp : exp RSH exp | |
bb4e0574 | 351 | { pstate->wrap2<rsh_operation> (); } |
a766d390 DE |
352 | ; |
353 | ||
354 | exp : exp EQUAL exp | |
bb4e0574 | 355 | { pstate->wrap2<equal_operation> (); } |
a766d390 DE |
356 | ; |
357 | ||
358 | exp : exp NOTEQUAL exp | |
bb4e0574 | 359 | { pstate->wrap2<notequal_operation> (); } |
a766d390 DE |
360 | ; |
361 | ||
362 | exp : exp LEQ exp | |
bb4e0574 | 363 | { pstate->wrap2<leq_operation> (); } |
a766d390 DE |
364 | ; |
365 | ||
366 | exp : exp GEQ exp | |
bb4e0574 | 367 | { pstate->wrap2<geq_operation> (); } |
a766d390 DE |
368 | ; |
369 | ||
370 | exp : exp '<' exp | |
bb4e0574 | 371 | { pstate->wrap2<less_operation> (); } |
a766d390 DE |
372 | ; |
373 | ||
374 | exp : exp '>' exp | |
bb4e0574 | 375 | { pstate->wrap2<gtr_operation> (); } |
a766d390 DE |
376 | ; |
377 | ||
378 | exp : exp '&' exp | |
bb4e0574 | 379 | { pstate->wrap2<bitwise_and_operation> (); } |
a766d390 DE |
380 | ; |
381 | ||
382 | exp : exp '^' exp | |
bb4e0574 | 383 | { pstate->wrap2<bitwise_xor_operation> (); } |
a766d390 DE |
384 | ; |
385 | ||
386 | exp : exp '|' exp | |
bb4e0574 | 387 | { pstate->wrap2<bitwise_ior_operation> (); } |
a766d390 DE |
388 | ; |
389 | ||
390 | exp : exp ANDAND exp | |
bb4e0574 | 391 | { pstate->wrap2<logical_and_operation> (); } |
a766d390 DE |
392 | ; |
393 | ||
394 | exp : exp OROR exp | |
bb4e0574 | 395 | { pstate->wrap2<logical_or_operation> (); } |
a766d390 DE |
396 | ; |
397 | ||
398 | exp : exp '?' exp ':' exp %prec '?' | |
bb4e0574 TT |
399 | { |
400 | operation_up last = pstate->pop (); | |
401 | operation_up mid = pstate->pop (); | |
402 | operation_up first = pstate->pop (); | |
403 | pstate->push_new<ternop_cond_operation> | |
404 | (std::move (first), std::move (mid), | |
405 | std::move (last)); | |
406 | } | |
a766d390 DE |
407 | ; |
408 | ||
409 | exp : exp '=' exp | |
bb4e0574 | 410 | { pstate->wrap2<assign_operation> (); } |
a766d390 DE |
411 | ; |
412 | ||
413 | exp : exp ASSIGN_MODIFY exp | |
bb4e0574 TT |
414 | { |
415 | operation_up rhs = pstate->pop (); | |
416 | operation_up lhs = pstate->pop (); | |
417 | pstate->push_new<assign_modify_operation> | |
418 | ($2, std::move (lhs), std::move (rhs)); | |
419 | } | |
a766d390 DE |
420 | ; |
421 | ||
422 | exp : INT | |
bb4e0574 TT |
423 | { |
424 | pstate->push_new<long_const_operation> | |
425 | ($1.type, $1.val); | |
426 | } | |
a766d390 DE |
427 | ; |
428 | ||
429 | exp : CHAR | |
430 | { | |
431 | struct stoken_vector vec; | |
432 | vec.len = 1; | |
433 | vec.tokens = &$1; | |
bb4e0574 | 434 | pstate->push_c_string ($1.type, &vec); |
a766d390 DE |
435 | } |
436 | ; | |
437 | ||
438 | exp : NAME_OR_INT | |
439 | { YYSTYPE val; | |
410a0ff2 SDJ |
440 | parse_number (pstate, $1.stoken.ptr, |
441 | $1.stoken.length, 0, &val); | |
bb4e0574 TT |
442 | pstate->push_new<long_const_operation> |
443 | (val.typed_val_int.type, | |
444 | val.typed_val_int.val); | |
a766d390 DE |
445 | } |
446 | ; | |
447 | ||
448 | ||
449 | exp : FLOAT | |
bb4e0574 TT |
450 | { |
451 | float_data data; | |
452 | std::copy (std::begin ($1.val), std::end ($1.val), | |
453 | std::begin (data)); | |
454 | pstate->push_new<float_const_operation> ($1.type, data); | |
455 | } | |
a766d390 DE |
456 | ; |
457 | ||
458 | exp : variable | |
459 | ; | |
460 | ||
461 | exp : DOLLAR_VARIABLE | |
462 | { | |
bb4e0574 | 463 | pstate->push_dollar ($1); |
a766d390 DE |
464 | } |
465 | ; | |
466 | ||
467 | exp : SIZEOF_KEYWORD '(' type ')' %prec UNARY | |
468 | { | |
469 | /* TODO(dje): Go objects in structs. */ | |
a766d390 | 470 | /* TODO(dje): What's the right type here? */ |
bb4e0574 TT |
471 | struct type *size_type |
472 | = parse_type (pstate)->builtin_unsigned_int; | |
f168693b | 473 | $3 = check_typedef ($3); |
bb4e0574 | 474 | pstate->push_new<long_const_operation> |
df86565b | 475 | (size_type, (LONGEST) $3->length ()); |
a766d390 DE |
476 | } |
477 | ; | |
478 | ||
479 | exp : SIZEOF_KEYWORD '(' exp ')' %prec UNARY | |
480 | { | |
481 | /* TODO(dje): Go objects in structs. */ | |
bb4e0574 | 482 | pstate->wrap<unop_sizeof_operation> (); |
a766d390 DE |
483 | } |
484 | ||
485 | string_exp: | |
486 | STRING | |
487 | { | |
488 | /* We copy the string here, and not in the | |
489 | lexer, to guarantee that we do not leak a | |
490 | string. */ | |
491 | /* Note that we NUL-terminate here, but just | |
492 | for convenience. */ | |
493 | struct typed_stoken *vec = XNEW (struct typed_stoken); | |
494 | $$.len = 1; | |
495 | $$.tokens = vec; | |
496 | ||
497 | vec->type = $1.type; | |
498 | vec->length = $1.length; | |
224c3ddb | 499 | vec->ptr = (char *) malloc ($1.length + 1); |
a766d390 DE |
500 | memcpy (vec->ptr, $1.ptr, $1.length + 1); |
501 | } | |
502 | ||
503 | | string_exp '+' STRING | |
504 | { | |
505 | /* Note that we NUL-terminate here, but just | |
506 | for convenience. */ | |
507 | char *p; | |
508 | ++$$.len; | |
224c3ddb SM |
509 | $$.tokens = XRESIZEVEC (struct typed_stoken, |
510 | $$.tokens, $$.len); | |
a766d390 | 511 | |
224c3ddb | 512 | p = (char *) malloc ($3.length + 1); |
a766d390 DE |
513 | memcpy (p, $3.ptr, $3.length + 1); |
514 | ||
515 | $$.tokens[$$.len - 1].type = $3.type; | |
516 | $$.tokens[$$.len - 1].length = $3.length; | |
517 | $$.tokens[$$.len - 1].ptr = p; | |
518 | } | |
519 | ; | |
520 | ||
521 | exp : string_exp %prec ABOVE_COMMA | |
522 | { | |
523 | int i; | |
524 | ||
bb4e0574 TT |
525 | /* Always utf8. */ |
526 | pstate->push_c_string (0, &$1); | |
a766d390 DE |
527 | for (i = 0; i < $1.len; ++i) |
528 | free ($1.tokens[i].ptr); | |
529 | free ($1.tokens); | |
530 | } | |
531 | ; | |
532 | ||
533 | exp : TRUE_KEYWORD | |
bb4e0574 | 534 | { pstate->push_new<bool_operation> ($1); } |
a766d390 DE |
535 | ; |
536 | ||
537 | exp : FALSE_KEYWORD | |
bb4e0574 | 538 | { pstate->push_new<bool_operation> ($1); } |
a766d390 DE |
539 | ; |
540 | ||
541 | variable: name_not_typename ENTRY | |
d12307c1 | 542 | { struct symbol *sym = $1.sym.symbol; |
a766d390 DE |
543 | |
544 | if (sym == NULL | |
d9743061 | 545 | || !sym->is_argument () |
a766d390 DE |
546 | || !symbol_read_needs_frame (sym)) |
547 | error (_("@entry can be used only for function " | |
548 | "parameters, not for \"%s\""), | |
61f4b350 | 549 | copy_name ($1.stoken).c_str ()); |
a766d390 | 550 | |
bb4e0574 | 551 | pstate->push_new<var_entry_value_operation> (sym); |
a766d390 DE |
552 | } |
553 | ; | |
554 | ||
555 | variable: name_not_typename | |
d12307c1 | 556 | { struct block_symbol sym = $1.sym; |
a766d390 | 557 | |
d12307c1 | 558 | if (sym.symbol) |
a766d390 | 559 | { |
d12307c1 | 560 | if (symbol_read_needs_frame (sym.symbol)) |
699bd4cf | 561 | pstate->block_tracker->update (sym); |
a766d390 | 562 | |
9e5e03df | 563 | pstate->push_new<var_value_operation> (sym); |
a766d390 DE |
564 | } |
565 | else if ($1.is_a_field_of_this) | |
566 | { | |
567 | /* TODO(dje): Can we get here? | |
568 | E.g., via a mix of c++ and go? */ | |
569 | gdb_assert_not_reached ("go with `this' field"); | |
570 | } | |
571 | else | |
572 | { | |
7c7b6655 | 573 | struct bound_minimal_symbol msymbol; |
61f4b350 | 574 | std::string arg = copy_name ($1.stoken); |
a766d390 DE |
575 | |
576 | msymbol = | |
61f4b350 | 577 | lookup_bound_minimal_symbol (arg.c_str ()); |
7c7b6655 | 578 | if (msymbol.minsym != NULL) |
bb4e0574 | 579 | pstate->push_new<var_msym_value_operation> |
9c79936b | 580 | (msymbol); |
a766d390 DE |
581 | else if (!have_full_symbols () |
582 | && !have_partial_symbols ()) | |
583 | error (_("No symbol table is loaded. " | |
584 | "Use the \"file\" command.")); | |
585 | else | |
586 | error (_("No symbol \"%s\" in current context."), | |
61f4b350 | 587 | arg.c_str ()); |
a766d390 DE |
588 | } |
589 | } | |
590 | ; | |
591 | ||
592 | /* TODO | |
593 | method_exp: PACKAGENAME '.' name '.' name | |
594 | { | |
595 | } | |
596 | ; | |
597 | */ | |
598 | ||
599 | type /* Implements (approximately): [*] type-specifier */ | |
600 | : '*' type | |
601 | { $$ = lookup_pointer_type ($2); } | |
602 | | TYPENAME | |
603 | { $$ = $1.type; } | |
604 | /* | |
605 | | STRUCT_KEYWORD name | |
606 | { $$ = lookup_struct (copy_name ($2), | |
607 | expression_context_block); } | |
608 | */ | |
609 | | BYTE_KEYWORD | |
fa9f5be6 | 610 | { $$ = builtin_go_type (pstate->gdbarch ()) |
a766d390 DE |
611 | ->builtin_uint8; } |
612 | ; | |
613 | ||
614 | /* TODO | |
615 | name : NAME { $$ = $1.stoken; } | |
616 | | TYPENAME { $$ = $1.stoken; } | |
617 | | NAME_OR_INT { $$ = $1.stoken; } | |
618 | ; | |
619 | */ | |
620 | ||
621 | name_not_typename | |
622 | : NAME | |
623 | /* These would be useful if name_not_typename was useful, but it is just | |
624 | a fake for "variable", so these cause reduce/reduce conflicts because | |
625 | the parser can't tell whether NAME_OR_INT is a name_not_typename (=variable, | |
626 | =exp) or just an exp. If name_not_typename was ever used in an lvalue | |
627 | context where only a name could occur, this might be useful. | |
628 | | NAME_OR_INT | |
629 | */ | |
630 | ; | |
631 | ||
632 | %% | |
633 | ||
a766d390 DE |
634 | /* Take care of parsing a number (anything that starts with a digit). |
635 | Set yylval and return the token type; update lexptr. | |
636 | LEN is the number of characters in it. */ | |
637 | ||
638 | /* FIXME: Needs some error checking for the float case. */ | |
639 | /* FIXME(dje): IWBN to use c-exp.y's parse_number if we could. | |
640 | That will require moving the guts into a function that we both call | |
641 | as our YYSTYPE is different than c-exp.y's */ | |
642 | ||
643 | static int | |
410a0ff2 SDJ |
644 | parse_number (struct parser_state *par_state, |
645 | const char *p, int len, int parsed_float, YYSTYPE *putithere) | |
a766d390 | 646 | { |
01772c54 PA |
647 | ULONGEST n = 0; |
648 | ULONGEST prevn = 0; | |
a766d390 DE |
649 | |
650 | int i = 0; | |
651 | int c; | |
652 | int base = input_radix; | |
653 | int unsigned_p = 0; | |
654 | ||
655 | /* Number of "L" suffixes encountered. */ | |
656 | int long_p = 0; | |
657 | ||
658 | /* We have found a "L" or "U" suffix. */ | |
659 | int found_suffix = 0; | |
660 | ||
a766d390 DE |
661 | if (parsed_float) |
662 | { | |
edd079d9 | 663 | const struct builtin_go_type *builtin_go_types |
fa9f5be6 | 664 | = builtin_go_type (par_state->gdbarch ()); |
edd079d9 UW |
665 | |
666 | /* Handle suffixes: 'f' for float32, 'l' for long double. | |
667 | FIXME: This appears to be an extension -- do we want this? */ | |
668 | if (len >= 1 && tolower (p[len - 1]) == 'f') | |
669 | { | |
670 | putithere->typed_val_float.type | |
671 | = builtin_go_types->builtin_float32; | |
672 | len--; | |
673 | } | |
674 | else if (len >= 1 && tolower (p[len - 1]) == 'l') | |
675 | { | |
676 | putithere->typed_val_float.type | |
677 | = parse_type (par_state)->builtin_long_double; | |
678 | len--; | |
679 | } | |
680 | /* Default type for floating-point literals is float64. */ | |
681 | else | |
dda83cd7 | 682 | { |
edd079d9 UW |
683 | putithere->typed_val_float.type |
684 | = builtin_go_types->builtin_float64; | |
dda83cd7 | 685 | } |
edd079d9 UW |
686 | |
687 | if (!parse_float (p, len, | |
688 | putithere->typed_val_float.type, | |
689 | putithere->typed_val_float.val)) | |
dda83cd7 | 690 | return ERROR; |
a766d390 DE |
691 | return FLOAT; |
692 | } | |
693 | ||
694 | /* Handle base-switching prefixes 0x, 0t, 0d, 0. */ | |
01772c54 | 695 | if (p[0] == '0' && len > 1) |
a766d390 DE |
696 | switch (p[1]) |
697 | { | |
698 | case 'x': | |
699 | case 'X': | |
700 | if (len >= 3) | |
701 | { | |
702 | p += 2; | |
703 | base = 16; | |
704 | len -= 2; | |
705 | } | |
706 | break; | |
707 | ||
708 | case 'b': | |
709 | case 'B': | |
710 | if (len >= 3) | |
711 | { | |
712 | p += 2; | |
713 | base = 2; | |
714 | len -= 2; | |
715 | } | |
716 | break; | |
717 | ||
718 | case 't': | |
719 | case 'T': | |
720 | case 'd': | |
721 | case 'D': | |
722 | if (len >= 3) | |
723 | { | |
724 | p += 2; | |
725 | base = 10; | |
726 | len -= 2; | |
727 | } | |
728 | break; | |
729 | ||
730 | default: | |
731 | base = 8; | |
732 | break; | |
733 | } | |
734 | ||
735 | while (len-- > 0) | |
736 | { | |
737 | c = *p++; | |
738 | if (c >= 'A' && c <= 'Z') | |
739 | c += 'a' - 'A'; | |
740 | if (c != 'l' && c != 'u') | |
741 | n *= base; | |
742 | if (c >= '0' && c <= '9') | |
743 | { | |
744 | if (found_suffix) | |
745 | return ERROR; | |
746 | n += i = c - '0'; | |
747 | } | |
748 | else | |
749 | { | |
750 | if (base > 10 && c >= 'a' && c <= 'f') | |
751 | { | |
752 | if (found_suffix) | |
753 | return ERROR; | |
754 | n += i = c - 'a' + 10; | |
755 | } | |
756 | else if (c == 'l') | |
757 | { | |
758 | ++long_p; | |
759 | found_suffix = 1; | |
760 | } | |
761 | else if (c == 'u') | |
762 | { | |
763 | unsigned_p = 1; | |
764 | found_suffix = 1; | |
765 | } | |
766 | else | |
767 | return ERROR; /* Char not a digit */ | |
768 | } | |
769 | if (i >= base) | |
770 | return ERROR; /* Invalid digit in this base. */ | |
771 | ||
4c4d769a | 772 | if (c != 'l' && c != 'u') |
a766d390 | 773 | { |
4c4d769a TV |
774 | /* Test for overflow. */ |
775 | if (n == 0 && prevn == 0) | |
776 | ; | |
777 | else if (prevn >= n) | |
a766d390 DE |
778 | error (_("Numeric constant too large.")); |
779 | } | |
780 | prevn = n; | |
781 | } | |
782 | ||
783 | /* An integer constant is an int, a long, or a long long. An L | |
784 | suffix forces it to be long; an LL suffix forces it to be long | |
785 | long. If not forced to a larger size, it gets the first type of | |
786 | the above that it fits in. To figure out whether it fits, we | |
787 | shift it right and see whether anything remains. Note that we | |
788 | can't shift sizeof (LONGEST) * HOST_CHAR_BIT bits or more in one | |
789 | operation, because many compilers will warn about such a shift | |
790 | (which always produces a zero result). Sometimes gdbarch_int_bit | |
791 | or gdbarch_long_bit will be that big, sometimes not. To deal with | |
792 | the case where it is we just always shift the value more than | |
793 | once, with fewer bits each time. */ | |
794 | ||
4c4d769a TV |
795 | int int_bits = gdbarch_int_bit (par_state->gdbarch ()); |
796 | int long_bits = gdbarch_long_bit (par_state->gdbarch ()); | |
797 | int long_long_bits = gdbarch_long_long_bit (par_state->gdbarch ()); | |
798 | bool have_signed = !unsigned_p; | |
799 | bool have_int = long_p == 0; | |
800 | bool have_long = long_p <= 1; | |
801 | if (have_int && have_signed && fits_in_type (1, n, int_bits, true)) | |
802 | putithere->typed_val_int.type = parse_type (par_state)->builtin_int; | |
803 | else if (have_int && fits_in_type (1, n, int_bits, false)) | |
804 | putithere->typed_val_int.type | |
805 | = parse_type (par_state)->builtin_unsigned_int; | |
806 | else if (have_long && have_signed && fits_in_type (1, n, long_bits, true)) | |
807 | putithere->typed_val_int.type = parse_type (par_state)->builtin_long; | |
808 | else if (have_long && fits_in_type (1, n, long_bits, false)) | |
809 | putithere->typed_val_int.type | |
810 | = parse_type (par_state)->builtin_unsigned_long; | |
811 | else if (have_signed && fits_in_type (1, n, long_long_bits, true)) | |
812 | putithere->typed_val_int.type | |
813 | = parse_type (par_state)->builtin_long_long; | |
814 | else if (fits_in_type (1, n, long_long_bits, false)) | |
815 | putithere->typed_val_int.type | |
816 | = parse_type (par_state)->builtin_unsigned_long_long; | |
a766d390 | 817 | else |
4c4d769a TV |
818 | error (_("Numeric constant too large.")); |
819 | putithere->typed_val_int.val = n; | |
a766d390 DE |
820 | |
821 | return INT; | |
822 | } | |
823 | ||
824 | /* Temporary obstack used for holding strings. */ | |
825 | static struct obstack tempbuf; | |
826 | static int tempbuf_init; | |
827 | ||
828 | /* Parse a string or character literal from TOKPTR. The string or | |
829 | character may be wide or unicode. *OUTPTR is set to just after the | |
830 | end of the literal in the input string. The resulting token is | |
831 | stored in VALUE. This returns a token value, either STRING or | |
832 | CHAR, depending on what was parsed. *HOST_CHARS is set to the | |
833 | number of host characters in the literal. */ | |
834 | ||
835 | static int | |
d7561cbb KS |
836 | parse_string_or_char (const char *tokptr, const char **outptr, |
837 | struct typed_stoken *value, int *host_chars) | |
a766d390 DE |
838 | { |
839 | int quote; | |
840 | ||
841 | /* Build the gdb internal form of the input string in tempbuf. Note | |
842 | that the buffer is null byte terminated *only* for the | |
843 | convenience of debugging gdb itself and printing the buffer | |
844 | contents when the buffer contains no embedded nulls. Gdb does | |
845 | not depend upon the buffer being null byte terminated, it uses | |
846 | the length string instead. This allows gdb to handle C strings | |
847 | (as well as strings in other languages) with embedded null | |
848 | bytes */ | |
849 | ||
850 | if (!tempbuf_init) | |
851 | tempbuf_init = 1; | |
852 | else | |
853 | obstack_free (&tempbuf, NULL); | |
854 | obstack_init (&tempbuf); | |
855 | ||
856 | /* Skip the quote. */ | |
857 | quote = *tokptr; | |
858 | ++tokptr; | |
859 | ||
860 | *host_chars = 0; | |
861 | ||
862 | while (*tokptr) | |
863 | { | |
864 | char c = *tokptr; | |
865 | if (c == '\\') | |
866 | { | |
867 | ++tokptr; | |
868 | *host_chars += c_parse_escape (&tokptr, &tempbuf); | |
869 | } | |
870 | else if (c == quote) | |
871 | break; | |
872 | else | |
873 | { | |
874 | obstack_1grow (&tempbuf, c); | |
875 | ++tokptr; | |
876 | /* FIXME: this does the wrong thing with multi-byte host | |
877 | characters. We could use mbrlen here, but that would | |
878 | make "set host-charset" a bit less useful. */ | |
879 | ++*host_chars; | |
880 | } | |
881 | } | |
882 | ||
883 | if (*tokptr != quote) | |
884 | { | |
885 | if (quote == '"') | |
886 | error (_("Unterminated string in expression.")); | |
887 | else | |
888 | error (_("Unmatched single quote.")); | |
889 | } | |
890 | ++tokptr; | |
891 | ||
04902b09 | 892 | value->type = (int) C_STRING | (quote == '\'' ? C_CHAR : 0); /*FIXME*/ |
79f33898 | 893 | value->ptr = (char *) obstack_base (&tempbuf); |
a766d390 DE |
894 | value->length = obstack_object_size (&tempbuf); |
895 | ||
896 | *outptr = tokptr; | |
897 | ||
898 | return quote == '\'' ? CHAR : STRING; | |
899 | } | |
900 | ||
e72b937d | 901 | struct go_token |
a766d390 | 902 | { |
a121b7c1 | 903 | const char *oper; |
a766d390 DE |
904 | int token; |
905 | enum exp_opcode opcode; | |
906 | }; | |
907 | ||
e72b937d | 908 | static const struct go_token tokentab3[] = |
a766d390 DE |
909 | { |
910 | {">>=", ASSIGN_MODIFY, BINOP_RSH}, | |
911 | {"<<=", ASSIGN_MODIFY, BINOP_LSH}, | |
912 | /*{"&^=", ASSIGN_MODIFY, BINOP_BITWISE_ANDNOT}, TODO */ | |
913 | {"...", DOTDOTDOT, OP_NULL}, | |
914 | }; | |
915 | ||
e72b937d | 916 | static const struct go_token tokentab2[] = |
a766d390 DE |
917 | { |
918 | {"+=", ASSIGN_MODIFY, BINOP_ADD}, | |
919 | {"-=", ASSIGN_MODIFY, BINOP_SUB}, | |
920 | {"*=", ASSIGN_MODIFY, BINOP_MUL}, | |
921 | {"/=", ASSIGN_MODIFY, BINOP_DIV}, | |
922 | {"%=", ASSIGN_MODIFY, BINOP_REM}, | |
923 | {"|=", ASSIGN_MODIFY, BINOP_BITWISE_IOR}, | |
924 | {"&=", ASSIGN_MODIFY, BINOP_BITWISE_AND}, | |
925 | {"^=", ASSIGN_MODIFY, BINOP_BITWISE_XOR}, | |
79ab486e TT |
926 | {"++", INCREMENT, OP_NULL}, |
927 | {"--", DECREMENT, OP_NULL}, | |
928 | /*{"->", RIGHT_ARROW, OP_NULL}, Doesn't exist in Go. */ | |
929 | {"<-", LEFT_ARROW, OP_NULL}, | |
930 | {"&&", ANDAND, OP_NULL}, | |
931 | {"||", OROR, OP_NULL}, | |
932 | {"<<", LSH, OP_NULL}, | |
933 | {">>", RSH, OP_NULL}, | |
934 | {"==", EQUAL, OP_NULL}, | |
935 | {"!=", NOTEQUAL, OP_NULL}, | |
936 | {"<=", LEQ, OP_NULL}, | |
937 | {">=", GEQ, OP_NULL}, | |
938 | /*{"&^", ANDNOT, OP_NULL}, TODO */ | |
a766d390 DE |
939 | }; |
940 | ||
941 | /* Identifier-like tokens. */ | |
e72b937d | 942 | static const struct go_token ident_tokens[] = |
a766d390 DE |
943 | { |
944 | {"true", TRUE_KEYWORD, OP_NULL}, | |
945 | {"false", FALSE_KEYWORD, OP_NULL}, | |
946 | {"nil", NIL_KEYWORD, OP_NULL}, | |
947 | {"const", CONST_KEYWORD, OP_NULL}, | |
948 | {"struct", STRUCT_KEYWORD, OP_NULL}, | |
949 | {"type", TYPE_KEYWORD, OP_NULL}, | |
950 | {"interface", INTERFACE_KEYWORD, OP_NULL}, | |
951 | {"chan", CHAN_KEYWORD, OP_NULL}, | |
952 | {"byte", BYTE_KEYWORD, OP_NULL}, /* An alias of uint8. */ | |
953 | {"len", LEN_KEYWORD, OP_NULL}, | |
954 | {"cap", CAP_KEYWORD, OP_NULL}, | |
955 | {"new", NEW_KEYWORD, OP_NULL}, | |
956 | {"iota", IOTA_KEYWORD, OP_NULL}, | |
957 | }; | |
958 | ||
959 | /* This is set if a NAME token appeared at the very end of the input | |
960 | string, with no whitespace separating the name from the EOF. This | |
961 | is used only when parsing to do field name completion. */ | |
962 | static int saw_name_at_eof; | |
963 | ||
964 | /* This is set if the previously-returned token was a structure | |
965 | operator -- either '.' or ARROW. This is used only when parsing to | |
966 | do field name completion. */ | |
967 | static int last_was_structop; | |
968 | ||
28aaf3fd TT |
969 | /* Depth of parentheses. */ |
970 | static int paren_depth; | |
971 | ||
a766d390 DE |
972 | /* Read one token, getting characters through lexptr. */ |
973 | ||
974 | static int | |
410a0ff2 | 975 | lex_one_token (struct parser_state *par_state) |
a766d390 DE |
976 | { |
977 | int c; | |
978 | int namelen; | |
d7561cbb | 979 | const char *tokstart; |
a766d390 | 980 | int saw_structop = last_was_structop; |
a766d390 DE |
981 | |
982 | last_was_structop = 0; | |
983 | ||
984 | retry: | |
985 | ||
5776fca3 | 986 | par_state->prev_lexptr = par_state->lexptr; |
a766d390 | 987 | |
5776fca3 | 988 | tokstart = par_state->lexptr; |
a766d390 | 989 | /* See if it is a special token of length 3. */ |
696d6f4d TT |
990 | for (const auto &token : tokentab3) |
991 | if (strncmp (tokstart, token.oper, 3) == 0) | |
a766d390 | 992 | { |
5776fca3 | 993 | par_state->lexptr += 3; |
696d6f4d TT |
994 | yylval.opcode = token.opcode; |
995 | return token.token; | |
a766d390 DE |
996 | } |
997 | ||
998 | /* See if it is a special token of length 2. */ | |
696d6f4d TT |
999 | for (const auto &token : tokentab2) |
1000 | if (strncmp (tokstart, token.oper, 2) == 0) | |
a766d390 | 1001 | { |
5776fca3 | 1002 | par_state->lexptr += 2; |
696d6f4d | 1003 | yylval.opcode = token.opcode; |
a766d390 DE |
1004 | /* NOTE: -> doesn't exist in Go, so we don't need to watch for |
1005 | setting last_was_structop here. */ | |
696d6f4d | 1006 | return token.token; |
a766d390 DE |
1007 | } |
1008 | ||
1009 | switch (c = *tokstart) | |
1010 | { | |
1011 | case 0: | |
1012 | if (saw_name_at_eof) | |
1013 | { | |
1014 | saw_name_at_eof = 0; | |
1015 | return COMPLETE; | |
1016 | } | |
1017 | else if (saw_structop) | |
1018 | return COMPLETE; | |
1019 | else | |
dda83cd7 | 1020 | return 0; |
a766d390 DE |
1021 | |
1022 | case ' ': | |
1023 | case '\t': | |
1024 | case '\n': | |
5776fca3 | 1025 | par_state->lexptr++; |
a766d390 DE |
1026 | goto retry; |
1027 | ||
1028 | case '[': | |
1029 | case '(': | |
1030 | paren_depth++; | |
5776fca3 | 1031 | par_state->lexptr++; |
a766d390 DE |
1032 | return c; |
1033 | ||
1034 | case ']': | |
1035 | case ')': | |
1036 | if (paren_depth == 0) | |
1037 | return 0; | |
1038 | paren_depth--; | |
5776fca3 | 1039 | par_state->lexptr++; |
a766d390 DE |
1040 | return c; |
1041 | ||
1042 | case ',': | |
8621b685 | 1043 | if (pstate->comma_terminates |
dda83cd7 | 1044 | && paren_depth == 0) |
a766d390 | 1045 | return 0; |
5776fca3 | 1046 | par_state->lexptr++; |
a766d390 DE |
1047 | return c; |
1048 | ||
1049 | case '.': | |
1050 | /* Might be a floating point number. */ | |
5776fca3 | 1051 | if (par_state->lexptr[1] < '0' || par_state->lexptr[1] > '9') |
a766d390 | 1052 | { |
2a612529 | 1053 | if (pstate->parse_completion) |
a766d390 DE |
1054 | last_was_structop = 1; |
1055 | goto symbol; /* Nope, must be a symbol. */ | |
1056 | } | |
d182e398 | 1057 | [[fallthrough]]; |
a766d390 DE |
1058 | |
1059 | case '0': | |
1060 | case '1': | |
1061 | case '2': | |
1062 | case '3': | |
1063 | case '4': | |
1064 | case '5': | |
1065 | case '6': | |
1066 | case '7': | |
1067 | case '8': | |
1068 | case '9': | |
1069 | { | |
1070 | /* It's a number. */ | |
1071 | int got_dot = 0, got_e = 0, toktype; | |
d7561cbb | 1072 | const char *p = tokstart; |
a766d390 DE |
1073 | int hex = input_radix > 10; |
1074 | ||
1075 | if (c == '0' && (p[1] == 'x' || p[1] == 'X')) | |
1076 | { | |
1077 | p += 2; | |
1078 | hex = 1; | |
1079 | } | |
1080 | ||
1081 | for (;; ++p) | |
1082 | { | |
1083 | /* This test includes !hex because 'e' is a valid hex digit | |
1084 | and thus does not indicate a floating point number when | |
1085 | the radix is hex. */ | |
1086 | if (!hex && !got_e && (*p == 'e' || *p == 'E')) | |
1087 | got_dot = got_e = 1; | |
1088 | /* This test does not include !hex, because a '.' always indicates | |
1089 | a decimal floating point number regardless of the radix. */ | |
1090 | else if (!got_dot && *p == '.') | |
1091 | got_dot = 1; | |
1092 | else if (got_e && (p[-1] == 'e' || p[-1] == 'E') | |
1093 | && (*p == '-' || *p == '+')) | |
1094 | /* This is the sign of the exponent, not the end of the | |
1095 | number. */ | |
1096 | continue; | |
1097 | /* We will take any letters or digits. parse_number will | |
1098 | complain if past the radix, or if L or U are not final. */ | |
1099 | else if ((*p < '0' || *p > '9') | |
1100 | && ((*p < 'a' || *p > 'z') | |
1101 | && (*p < 'A' || *p > 'Z'))) | |
1102 | break; | |
1103 | } | |
410a0ff2 SDJ |
1104 | toktype = parse_number (par_state, tokstart, p - tokstart, |
1105 | got_dot|got_e, &yylval); | |
dda83cd7 | 1106 | if (toktype == ERROR) |
a766d390 DE |
1107 | { |
1108 | char *err_copy = (char *) alloca (p - tokstart + 1); | |
1109 | ||
1110 | memcpy (err_copy, tokstart, p - tokstart); | |
1111 | err_copy[p - tokstart] = 0; | |
1112 | error (_("Invalid number \"%s\"."), err_copy); | |
1113 | } | |
5776fca3 | 1114 | par_state->lexptr = p; |
a766d390 DE |
1115 | return toktype; |
1116 | } | |
1117 | ||
1118 | case '@': | |
1119 | { | |
d7561cbb | 1120 | const char *p = &tokstart[1]; |
a766d390 DE |
1121 | size_t len = strlen ("entry"); |
1122 | ||
1123 | while (isspace (*p)) | |
1124 | p++; | |
1125 | if (strncmp (p, "entry", len) == 0 && !isalnum (p[len]) | |
1126 | && p[len] != '_') | |
1127 | { | |
5776fca3 | 1128 | par_state->lexptr = &p[len]; |
a766d390 DE |
1129 | return ENTRY; |
1130 | } | |
1131 | } | |
d182e398 | 1132 | [[fallthrough]]; |
a766d390 DE |
1133 | case '+': |
1134 | case '-': | |
1135 | case '*': | |
1136 | case '/': | |
1137 | case '%': | |
1138 | case '|': | |
1139 | case '&': | |
1140 | case '^': | |
1141 | case '~': | |
1142 | case '!': | |
1143 | case '<': | |
1144 | case '>': | |
1145 | case '?': | |
1146 | case ':': | |
1147 | case '=': | |
1148 | case '{': | |
1149 | case '}': | |
1150 | symbol: | |
5776fca3 | 1151 | par_state->lexptr++; |
a766d390 DE |
1152 | return c; |
1153 | ||
1154 | case '\'': | |
1155 | case '"': | |
1156 | case '`': | |
1157 | { | |
1158 | int host_len; | |
5776fca3 TT |
1159 | int result = parse_string_or_char (tokstart, &par_state->lexptr, |
1160 | &yylval.tsval, &host_len); | |
a766d390 DE |
1161 | if (result == CHAR) |
1162 | { | |
1163 | if (host_len == 0) | |
1164 | error (_("Empty character constant.")); | |
1165 | else if (host_len > 2 && c == '\'') | |
1166 | { | |
1167 | ++tokstart; | |
5776fca3 | 1168 | namelen = par_state->lexptr - tokstart - 1; |
a766d390 DE |
1169 | goto tryname; |
1170 | } | |
1171 | else if (host_len > 1) | |
1172 | error (_("Invalid character constant.")); | |
1173 | } | |
1174 | return result; | |
1175 | } | |
1176 | } | |
1177 | ||
1178 | if (!(c == '_' || c == '$' | |
1179 | || (c >= 'a' && c <= 'z') || (c >= 'A' && c <= 'Z'))) | |
1180 | /* We must have come across a bad character (e.g. ';'). */ | |
1181 | error (_("Invalid character '%c' in expression."), c); | |
1182 | ||
1183 | /* It's a name. See how long it is. */ | |
1184 | namelen = 0; | |
1185 | for (c = tokstart[namelen]; | |
1186 | (c == '_' || c == '$' || (c >= '0' && c <= '9') | |
1187 | || (c >= 'a' && c <= 'z') || (c >= 'A' && c <= 'Z'));) | |
1188 | { | |
1189 | c = tokstart[++namelen]; | |
1190 | } | |
1191 | ||
1192 | /* The token "if" terminates the expression and is NOT removed from | |
1193 | the input stream. It doesn't count if it appears in the | |
1194 | expansion of a macro. */ | |
1195 | if (namelen == 2 | |
1196 | && tokstart[0] == 'i' | |
1197 | && tokstart[1] == 'f') | |
1198 | { | |
1199 | return 0; | |
1200 | } | |
1201 | ||
1202 | /* For the same reason (breakpoint conditions), "thread N" | |
1203 | terminates the expression. "thread" could be an identifier, but | |
1204 | an identifier is never followed by a number without intervening | |
1205 | punctuation. | |
1206 | Handle abbreviations of these, similarly to | |
1207 | breakpoint.c:find_condition_and_thread. | |
1208 | TODO: Watch for "goroutine" here? */ | |
1209 | if (namelen >= 1 | |
1210 | && strncmp (tokstart, "thread", namelen) == 0 | |
1211 | && (tokstart[namelen] == ' ' || tokstart[namelen] == '\t')) | |
1212 | { | |
d7561cbb KS |
1213 | const char *p = tokstart + namelen + 1; |
1214 | ||
a766d390 DE |
1215 | while (*p == ' ' || *p == '\t') |
1216 | p++; | |
1217 | if (*p >= '0' && *p <= '9') | |
1218 | return 0; | |
1219 | } | |
1220 | ||
5776fca3 | 1221 | par_state->lexptr += namelen; |
a766d390 DE |
1222 | |
1223 | tryname: | |
1224 | ||
1225 | yylval.sval.ptr = tokstart; | |
1226 | yylval.sval.length = namelen; | |
1227 | ||
1228 | /* Catch specific keywords. */ | |
61f4b350 | 1229 | std::string copy = copy_name (yylval.sval); |
696d6f4d TT |
1230 | for (const auto &token : ident_tokens) |
1231 | if (copy == token.oper) | |
a766d390 DE |
1232 | { |
1233 | /* It is ok to always set this, even though we don't always | |
1234 | strictly need to. */ | |
696d6f4d TT |
1235 | yylval.opcode = token.opcode; |
1236 | return token.token; | |
a766d390 DE |
1237 | } |
1238 | ||
1239 | if (*tokstart == '$') | |
1240 | return DOLLAR_VARIABLE; | |
1241 | ||
2a612529 | 1242 | if (pstate->parse_completion && *par_state->lexptr == '\0') |
a766d390 DE |
1243 | saw_name_at_eof = 1; |
1244 | return NAME; | |
1245 | } | |
1246 | ||
1247 | /* An object of this type is pushed on a FIFO by the "outer" lexer. */ | |
9972aac2 | 1248 | struct go_token_and_value |
a766d390 DE |
1249 | { |
1250 | int token; | |
1251 | YYSTYPE value; | |
5fe3f3e4 | 1252 | }; |
a766d390 DE |
1253 | |
1254 | /* A FIFO of tokens that have been read but not yet returned to the | |
1255 | parser. */ | |
9972aac2 | 1256 | static std::vector<go_token_and_value> token_fifo; |
a766d390 DE |
1257 | |
1258 | /* Non-zero if the lexer should return tokens from the FIFO. */ | |
1259 | static int popping; | |
1260 | ||
1261 | /* Temporary storage for yylex; this holds symbol names as they are | |
1262 | built up. */ | |
8268c778 | 1263 | static auto_obstack name_obstack; |
a766d390 DE |
1264 | |
1265 | /* Build "package.name" in name_obstack. | |
1266 | For convenience of the caller, the name is NUL-terminated, | |
1267 | but the NUL is not included in the recorded length. */ | |
1268 | ||
1269 | static struct stoken | |
1270 | build_packaged_name (const char *package, int package_len, | |
1271 | const char *name, int name_len) | |
1272 | { | |
1273 | struct stoken result; | |
1274 | ||
8268c778 | 1275 | name_obstack.clear (); |
a766d390 DE |
1276 | obstack_grow (&name_obstack, package, package_len); |
1277 | obstack_grow_str (&name_obstack, "."); | |
1278 | obstack_grow (&name_obstack, name, name_len); | |
1279 | obstack_grow (&name_obstack, "", 1); | |
79f33898 | 1280 | result.ptr = (char *) obstack_base (&name_obstack); |
a766d390 DE |
1281 | result.length = obstack_object_size (&name_obstack) - 1; |
1282 | ||
1283 | return result; | |
1284 | } | |
1285 | ||
1286 | /* Return non-zero if NAME is a package name. | |
1287 | BLOCK is the scope in which to interpret NAME; this can be NULL | |
1288 | to mean the global scope. */ | |
1289 | ||
1290 | static int | |
270140bd | 1291 | package_name_p (const char *name, const struct block *block) |
a766d390 DE |
1292 | { |
1293 | struct symbol *sym; | |
1993b719 | 1294 | struct field_of_this_result is_a_field_of_this; |
a766d390 | 1295 | |
d12307c1 | 1296 | sym = lookup_symbol (name, block, STRUCT_DOMAIN, &is_a_field_of_this).symbol; |
a766d390 DE |
1297 | |
1298 | if (sym | |
66d7f48f | 1299 | && sym->aclass () == LOC_TYPEDEF |
5f9c5a63 | 1300 | && sym->type ()->code () == TYPE_CODE_MODULE) |
a766d390 DE |
1301 | return 1; |
1302 | ||
1303 | return 0; | |
1304 | } | |
1305 | ||
1306 | /* Classify a (potential) function in the "unsafe" package. | |
1307 | We fold these into "keywords" to keep things simple, at least until | |
1308 | something more complex is warranted. */ | |
1309 | ||
1310 | static int | |
1311 | classify_unsafe_function (struct stoken function_name) | |
1312 | { | |
61f4b350 | 1313 | std::string copy = copy_name (function_name); |
a766d390 | 1314 | |
61f4b350 | 1315 | if (copy == "Sizeof") |
a766d390 DE |
1316 | { |
1317 | yylval.sval = function_name; | |
1318 | return SIZEOF_KEYWORD; | |
1319 | } | |
1320 | ||
61f4b350 | 1321 | error (_("Unknown function in `unsafe' package: %s"), copy.c_str ()); |
a766d390 DE |
1322 | } |
1323 | ||
1324 | /* Classify token(s) "name1.name2" where name1 is known to be a package. | |
1325 | The contents of the token are in `yylval'. | |
1326 | Updates yylval and returns the new token type. | |
1327 | ||
1328 | The result is one of NAME, NAME_OR_INT, or TYPENAME. */ | |
1329 | ||
1330 | static int | |
270140bd | 1331 | classify_packaged_name (const struct block *block) |
a766d390 | 1332 | { |
d12307c1 | 1333 | struct block_symbol sym; |
1993b719 | 1334 | struct field_of_this_result is_a_field_of_this; |
a766d390 | 1335 | |
61f4b350 | 1336 | std::string copy = copy_name (yylval.sval); |
a766d390 | 1337 | |
61f4b350 | 1338 | sym = lookup_symbol (copy.c_str (), block, VAR_DOMAIN, &is_a_field_of_this); |
a766d390 | 1339 | |
d12307c1 | 1340 | if (sym.symbol) |
a766d390 DE |
1341 | { |
1342 | yylval.ssym.sym = sym; | |
1993b719 | 1343 | yylval.ssym.is_a_field_of_this = is_a_field_of_this.type != NULL; |
a766d390 DE |
1344 | } |
1345 | ||
1346 | return NAME; | |
1347 | } | |
1348 | ||
1349 | /* Classify a NAME token. | |
1350 | The contents of the token are in `yylval'. | |
1351 | Updates yylval and returns the new token type. | |
1352 | BLOCK is the block in which lookups start; this can be NULL | |
1353 | to mean the global scope. | |
1354 | ||
1355 | The result is one of NAME, NAME_OR_INT, or TYPENAME. */ | |
1356 | ||
1357 | static int | |
410a0ff2 | 1358 | classify_name (struct parser_state *par_state, const struct block *block) |
a766d390 DE |
1359 | { |
1360 | struct type *type; | |
d12307c1 | 1361 | struct block_symbol sym; |
1993b719 | 1362 | struct field_of_this_result is_a_field_of_this; |
a766d390 | 1363 | |
61f4b350 | 1364 | std::string copy = copy_name (yylval.sval); |
a766d390 DE |
1365 | |
1366 | /* Try primitive types first so they win over bad/weird debug info. */ | |
73923d7e | 1367 | type = language_lookup_primitive_type (par_state->language (), |
fa9f5be6 | 1368 | par_state->gdbarch (), |
61f4b350 | 1369 | copy.c_str ()); |
a766d390 DE |
1370 | if (type != NULL) |
1371 | { | |
1372 | /* NOTE: We take advantage of the fact that yylval coming in was a | |
1373 | NAME, and that struct ttype is a compatible extension of struct | |
1374 | stoken, so yylval.tsym.stoken is already filled in. */ | |
1375 | yylval.tsym.type = type; | |
1376 | return TYPENAME; | |
1377 | } | |
1378 | ||
1379 | /* TODO: What about other types? */ | |
1380 | ||
61f4b350 | 1381 | sym = lookup_symbol (copy.c_str (), block, VAR_DOMAIN, &is_a_field_of_this); |
a766d390 | 1382 | |
d12307c1 | 1383 | if (sym.symbol) |
a766d390 DE |
1384 | { |
1385 | yylval.ssym.sym = sym; | |
1993b719 | 1386 | yylval.ssym.is_a_field_of_this = is_a_field_of_this.type != NULL; |
a766d390 DE |
1387 | return NAME; |
1388 | } | |
1389 | ||
1390 | /* If we didn't find a symbol, look again in the current package. | |
1391 | This is to, e.g., make "p global_var" work without having to specify | |
1392 | the package name. We intentionally only looks for objects in the | |
1393 | current package. */ | |
1394 | ||
1395 | { | |
be643e07 TT |
1396 | gdb::unique_xmalloc_ptr<char> current_package_name |
1397 | = go_block_package_name (block); | |
a766d390 DE |
1398 | |
1399 | if (current_package_name != NULL) | |
1400 | { | |
1401 | struct stoken sval = | |
be643e07 TT |
1402 | build_packaged_name (current_package_name.get (), |
1403 | strlen (current_package_name.get ()), | |
61f4b350 | 1404 | copy.c_str (), copy.size ()); |
a766d390 | 1405 | |
a766d390 DE |
1406 | sym = lookup_symbol (sval.ptr, block, VAR_DOMAIN, |
1407 | &is_a_field_of_this); | |
d12307c1 | 1408 | if (sym.symbol) |
a766d390 | 1409 | { |
3929b321 | 1410 | yylval.ssym.stoken = sval; |
a766d390 | 1411 | yylval.ssym.sym = sym; |
1993b719 | 1412 | yylval.ssym.is_a_field_of_this = is_a_field_of_this.type != NULL; |
a766d390 DE |
1413 | return NAME; |
1414 | } | |
1415 | } | |
1416 | } | |
1417 | ||
1418 | /* Input names that aren't symbols but ARE valid hex numbers, when | |
1419 | the input radix permits them, can be names or numbers depending | |
1420 | on the parse. Note we support radixes > 16 here. */ | |
1421 | if ((copy[0] >= 'a' && copy[0] < 'a' + input_radix - 10) | |
1422 | || (copy[0] >= 'A' && copy[0] < 'A' + input_radix - 10)) | |
1423 | { | |
1424 | YYSTYPE newlval; /* Its value is ignored. */ | |
61f4b350 TT |
1425 | int hextype = parse_number (par_state, copy.c_str (), |
1426 | yylval.sval.length, 0, &newlval); | |
a766d390 | 1427 | if (hextype == INT) |
3929b321 | 1428 | { |
d12307c1 PMR |
1429 | yylval.ssym.sym.symbol = NULL; |
1430 | yylval.ssym.sym.block = NULL; | |
3929b321 DE |
1431 | yylval.ssym.is_a_field_of_this = 0; |
1432 | return NAME_OR_INT; | |
1433 | } | |
a766d390 DE |
1434 | } |
1435 | ||
d12307c1 PMR |
1436 | yylval.ssym.sym.symbol = NULL; |
1437 | yylval.ssym.sym.block = NULL; | |
3929b321 | 1438 | yylval.ssym.is_a_field_of_this = 0; |
a766d390 DE |
1439 | return NAME; |
1440 | } | |
1441 | ||
1442 | /* This is taken from c-exp.y mostly to get something working. | |
1443 | The basic structure has been kept because we may yet need some of it. */ | |
1444 | ||
1445 | static int | |
1446 | yylex (void) | |
1447 | { | |
9972aac2 | 1448 | go_token_and_value current, next; |
a766d390 | 1449 | |
5fe3f3e4 | 1450 | if (popping && !token_fifo.empty ()) |
a766d390 | 1451 | { |
9972aac2 | 1452 | go_token_and_value tv = token_fifo[0]; |
5fe3f3e4 | 1453 | token_fifo.erase (token_fifo.begin ()); |
a766d390 DE |
1454 | yylval = tv.value; |
1455 | /* There's no need to fall through to handle package.name | |
1456 | as that can never happen here. In theory. */ | |
1457 | return tv.token; | |
1458 | } | |
1459 | popping = 0; | |
1460 | ||
410a0ff2 | 1461 | current.token = lex_one_token (pstate); |
a766d390 DE |
1462 | |
1463 | /* TODO: Need a way to force specifying name1 as a package. | |
1464 | .name1.name2 ? */ | |
1465 | ||
1466 | if (current.token != NAME) | |
1467 | return current.token; | |
1468 | ||
1469 | /* See if we have "name1 . name2". */ | |
1470 | ||
1471 | current.value = yylval; | |
410a0ff2 | 1472 | next.token = lex_one_token (pstate); |
a766d390 DE |
1473 | next.value = yylval; |
1474 | ||
1475 | if (next.token == '.') | |
1476 | { | |
9972aac2 | 1477 | go_token_and_value name2; |
a766d390 | 1478 | |
410a0ff2 | 1479 | name2.token = lex_one_token (pstate); |
a766d390 DE |
1480 | name2.value = yylval; |
1481 | ||
1482 | if (name2.token == NAME) | |
1483 | { | |
1484 | /* Ok, we have "name1 . name2". */ | |
61f4b350 | 1485 | std::string copy = copy_name (current.value.sval); |
a766d390 | 1486 | |
61f4b350 | 1487 | if (copy == "unsafe") |
a766d390 DE |
1488 | { |
1489 | popping = 1; | |
1490 | return classify_unsafe_function (name2.value.sval); | |
1491 | } | |
1492 | ||
61f4b350 | 1493 | if (package_name_p (copy.c_str (), pstate->expression_context_block)) |
a766d390 DE |
1494 | { |
1495 | popping = 1; | |
1496 | yylval.sval = build_packaged_name (current.value.sval.ptr, | |
1497 | current.value.sval.length, | |
1498 | name2.value.sval.ptr, | |
1499 | name2.value.sval.length); | |
1e58a4a4 | 1500 | return classify_packaged_name (pstate->expression_context_block); |
a766d390 DE |
1501 | } |
1502 | } | |
1503 | ||
5fe3f3e4 TT |
1504 | token_fifo.push_back (next); |
1505 | token_fifo.push_back (name2); | |
a766d390 DE |
1506 | } |
1507 | else | |
5fe3f3e4 | 1508 | token_fifo.push_back (next); |
a766d390 DE |
1509 | |
1510 | /* If we arrive here we don't have a package-qualified name. */ | |
1511 | ||
1512 | popping = 1; | |
1513 | yylval = current.value; | |
1e58a4a4 | 1514 | return classify_name (pstate, pstate->expression_context_block); |
a766d390 DE |
1515 | } |
1516 | ||
82fc57fd AB |
1517 | /* See language.h. */ |
1518 | ||
a766d390 | 1519 | int |
82fc57fd | 1520 | go_language::parser (struct parser_state *par_state) const |
a766d390 | 1521 | { |
410a0ff2 | 1522 | /* Setting up the parser state. */ |
eae49211 | 1523 | scoped_restore pstate_restore = make_scoped_restore (&pstate); |
410a0ff2 SDJ |
1524 | gdb_assert (par_state != NULL); |
1525 | pstate = par_state; | |
1526 | ||
156d9eab | 1527 | scoped_restore restore_yydebug = make_scoped_restore (&yydebug, |
e360af5a | 1528 | par_state->debug); |
a766d390 DE |
1529 | |
1530 | /* Initialize some state used by the lexer. */ | |
1531 | last_was_structop = 0; | |
1532 | saw_name_at_eof = 0; | |
28aaf3fd | 1533 | paren_depth = 0; |
a766d390 | 1534 | |
5fe3f3e4 | 1535 | token_fifo.clear (); |
a766d390 | 1536 | popping = 0; |
8268c778 | 1537 | name_obstack.clear (); |
a766d390 | 1538 | |
bb4e0574 TT |
1539 | int result = yyparse (); |
1540 | if (!result) | |
1541 | pstate->set_operation (pstate->pop ()); | |
1542 | return result; | |
a766d390 DE |
1543 | } |
1544 | ||
69d340c6 | 1545 | static void |
a121b7c1 | 1546 | yyerror (const char *msg) |
a766d390 | 1547 | { |
5776fca3 TT |
1548 | if (pstate->prev_lexptr) |
1549 | pstate->lexptr = pstate->prev_lexptr; | |
a766d390 | 1550 | |
5776fca3 | 1551 | error (_("A %s in expression, near `%s'."), msg, pstate->lexptr); |
a766d390 | 1552 | } |