1 /* Bison parser for Rust expressions, for GDB.
2 Copyright (C) 2016-2018 Free Software Foundation, Inc.
4 This file is part of GDB.
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.
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.
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/>. */
19 /* Removing the last conflict seems difficult. */
28 #include "cp-support.h"
29 #include "gdb_obstack.h"
30 #include "gdb_regex.h"
31 #include "rust-lang.h"
32 #include "parser-defs.h"
37 #define GDB_YY_REMAP_PREFIX rust
40 #define RUSTSTYPE YYSTYPE
43 typedef std::vector<const struct rust_op *> rust_op_vector;
45 /* A typed integer constant. */
53 /* A typed floating point constant. */
55 struct typed_val_float
61 /* An identifier and an expression. This is used to represent one
62 element of a struct initializer. */
67 const struct rust_op *init;
70 typedef std::vector<set_field> rust_set_vector;
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);
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);
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,
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,
107 rust_op_vector *params);
108 static const struct rust_op *ast_path (struct stoken name,
109 rust_op_vector *params);
110 static const struct rust_op *ast_string (struct stoken str);
111 static const struct rust_op *ast_struct (const struct rust_op *name,
112 rust_set_vector *fields);
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,
121 static const struct rust_op *ast_function_type (const struct rust_op *result,
122 rust_op_vector *params);
123 static const struct rust_op *ast_tuple_type (rust_op_vector *params);
125 /* The current rust parser. */
128 static rust_parser *current_parser;
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
134 static const char *number_regex_text =
135 /* subexpression 1: allows use of alternation, otherwise uninteresting */
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
143 "[0-9][0-9_]*\\.[0-9][0-9_]*([eE][-+]?[0-9][0-9_]*)?(f32|f64)?"
144 #define FLOAT_TYPE1 3
146 /* Recognize exponent without decimal point, with optional type
148 subexpression 4: if present, type suffix
150 #define FLOAT_TYPE2 4
151 "[0-9][0-9_]*[eE][-+]?[0-9][0-9_]*(f32|f64)?"
153 /* "23." is a valid floating point number, but "23.e5" and
154 "23.f32" are not. So, handle the trailing-. case
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
165 "(0x[a-fA-F0-9_]+|0o[0-7_]+|0b[01_]+|[0-9][0-9_]*)"
166 "([iu](size|8|16|32|64))?"
168 /* The number of subexpressions to allocate space for, including the
169 "0th" whole match subexpression. */
170 #define NUM_SUBEXPRESSIONS 8
172 /* The compiled number-matching regex. */
174 static regex_t number_regex;
176 /* Obstack for data temporarily allocated during parsing. Points to
177 the obstack in the rust_parser, or to a temporary obstack during
180 static auto_obstack *work_obstack;
182 /* An instance of this is created before parsing, and destroyed when
183 parsing is finished. */
187 rust_parser (struct parser_state *state)
188 : rust_ast (nullptr),
191 gdb_assert (current_parser == nullptr);
192 current_parser = this;
193 work_obstack = &obstack;
198 /* Clean up the globals we set. */
199 current_parser = nullptr;
200 work_obstack = nullptr;
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 ()
207 rust_set_vector *result = new rust_set_vector;
208 set_vectors.push_back (std::unique_ptr<rust_set_vector> (result));
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 ()
216 rust_op_vector *result = new rust_op_vector;
217 op_vectors.push_back (std::unique_ptr<rust_op_vector> (result));
221 /* Return the parser's language. */
222 const struct language_defn *language () const
224 return parse_language (pstate);
227 /* Return the parser's gdbarch. */
228 struct gdbarch *arch () const
230 return parse_gdbarch (pstate);
233 /* A pointer to this is installed globally. */
234 auto_obstack obstack;
236 /* Result of parsing. Points into obstack. */
237 const struct rust_op *rust_ast;
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;
243 /* The parser state gdb gave us. */
244 struct parser_state *pstate;
251 /* A typed integer constant. */
252 struct typed_val_int typed_val_int;
254 /* A typed floating point constant. */
255 struct typed_val_float typed_val_float;
257 /* An identifier or string. */
260 /* A token representing an opcode, like "==". */
261 enum exp_opcode opcode;
263 /* A list of expressions; for example, the arguments to a function
265 rust_op_vector *params;
267 /* A list of field initializers. */
268 rust_set_vector *field_inits;
270 /* A single field initializer. */
271 struct set_field one_field_init;
274 const struct rust_op *op;
276 /* A plain integer, for example used to count the number of
277 "super::" prefixes on a path. */
283 /* Rust AST operations. We build a tree of these; then lower them
284 to gdb expressions when parsing has completed. */
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
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
302 unsigned int completing : 1;
303 /* Operands of expression. Which one is used and how depends on the
304 particular opcode. */
313 %token <sval> COMPLETE
314 %token <typed_val_int> INTEGER
315 %token <typed_val_int> DECIMAL_INTEGER
317 %token <sval> BYTESTRING
318 %token <typed_val_float> FLOAT
319 %token <opcode> COMPOUND_ASSIGN
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
332 %token <voidval> KW_SIZEOF
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
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
353 %type <params> maybe_type_list
354 %type <params> type_list
356 %type <depth> super_path
360 %type <op> field_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
371 %type <op> tuple_expr
373 %type <op> struct_expr
374 %type <op> array_expr
375 %type <op> range_expr
377 %type <params> expr_list
378 %type <params> maybe_expr_list
379 %type <params> paren_expr_list
381 %type <field_inits> struct_expr_list
382 %type <one_field_init> struct_expr_tail
386 %right '=' COMPOUND_ASSIGN
389 %nonassoc EQEQ NOTEQ '<' '>' LTEQ GTEQ
397 /* These could be %precedence in Bison, but that isn't a yacc
408 /* If we are completing and see a valid parse,
409 rust_ast will already have been set. */
410 if (current_parser->rust_ast == NULL)
411 current_parser->rust_ast = $1;
415 /* Note that the Rust grammar includes a method_call_expr, but we
416 handle this differently, to avoid a shift/reduce conflict with
428 | unop_expr /* Must precede call_expr because of ambiguity with
436 '(' expr ',' maybe_expr_list ')'
439 error (_("Tuple expressions not supported yet"));
446 struct typed_val_int val;
449 = (language_lookup_primitive_type
450 (current_parser->language (), current_parser->arch (),
453 $$ = ast_literal (val);
457 /* To avoid a shift/reduce conflict with call_expr, we don't handle
458 tuple struct expressions here, but instead when examining the
461 path_for_expr '{' struct_expr_list '}'
462 { $$ = ast_struct ($1, $3); }
489 sf.init = ast_path ($1, NULL);
497 $$ = current_parser->new_set_vector ();
501 rust_set_vector *result = current_parser->new_set_vector ();
502 result->push_back ($1);
505 | IDENT ':' expr ',' struct_expr_list
514 | IDENT ',' struct_expr_list
519 sf.init = ast_path ($1, NULL);
526 '[' KW_MUT expr_list ']'
527 { $$ = ast_call_ish (OP_ARRAY, NULL, $3); }
529 { $$ = ast_call_ish (OP_ARRAY, NULL, $2); }
530 | '[' KW_MUT expr ';' expr ']'
531 { $$ = ast_operation (OP_RUST_ARRAY, $3, $5); }
532 | '[' expr ';' expr ']'
533 { $$ = ast_operation (OP_RUST_ARRAY, $2, $4); }
538 { $$ = ast_range ($1, NULL); }
540 { $$ = ast_range ($1, $3); }
542 { $$ = ast_range (NULL, $2); }
544 { $$ = ast_range (NULL, NULL); }
549 { $$ = ast_literal ($1); }
551 { $$ = ast_literal ($1); }
553 { $$ = ast_dliteral ($1); }
556 const struct rust_op *str = ast_string ($1);
557 struct set_field field;
558 struct typed_val_int val;
561 rust_set_vector *fields = current_parser->new_set_vector ();
563 /* Wrap the raw string in the &str struct. */
564 field.name.ptr = "data_ptr";
565 field.name.length = strlen (field.name.ptr);
566 field.init = ast_unary (UNOP_ADDR, ast_string ($1));
567 fields->push_back (field);
569 val.type = rust_type ("usize");
572 field.name.ptr = "length";
573 field.name.length = strlen (field.name.ptr);
574 field.init = ast_literal (val);
575 fields->push_back (field);
578 token.length = strlen (token.ptr);
579 $$ = ast_struct (ast_path (token, NULL), fields);
582 { $$ = ast_string ($1); }
585 struct typed_val_int val;
587 val.type = language_bool_type (current_parser->language (),
588 current_parser->arch ());
590 $$ = ast_literal (val);
594 struct typed_val_int val;
596 val.type = language_bool_type (current_parser->language (),
597 current_parser->arch ());
599 $$ = ast_literal (val);
605 { $$ = ast_structop ($1, $3.ptr, 0); }
608 $$ = ast_structop ($1, $3.ptr, 1);
609 current_parser->rust_ast = $$;
611 | expr '.' DECIMAL_INTEGER
612 { $$ = ast_structop_anonymous ($1, $3); }
617 { $$ = ast_operation (BINOP_SUBSCRIPT, $1, $3); }
622 { $$ = ast_unary (UNOP_PLUS, $2); }
624 | '-' expr %prec UNARY
625 { $$ = ast_unary (UNOP_NEG, $2); }
627 | '!' expr %prec UNARY
629 /* Note that we provide a Rust-specific evaluator
630 override for UNOP_COMPLEMENT, so it can do the
631 right thing for both bool and integral
633 $$ = ast_unary (UNOP_COMPLEMENT, $2);
636 | '*' expr %prec UNARY
637 { $$ = ast_unary (UNOP_IND, $2); }
639 | '&' expr %prec UNARY
640 { $$ = ast_unary (UNOP_ADDR, $2); }
642 | '&' KW_MUT expr %prec UNARY
643 { $$ = ast_unary (UNOP_ADDR, $3); }
644 | KW_SIZEOF '(' expr ')' %prec UNARY
645 { $$ = ast_unary (UNOP_SIZEOF, $3); }
652 | compound_assignment_expr
657 { $$ = ast_operation (BINOP_MUL, $1, $3); }
660 { $$ = ast_operation (BINOP_REPEAT, $1, $3); }
663 { $$ = ast_operation (BINOP_DIV, $1, $3); }
666 { $$ = ast_operation (BINOP_REM, $1, $3); }
669 { $$ = ast_operation (BINOP_LESS, $1, $3); }
672 { $$ = ast_operation (BINOP_GTR, $1, $3); }
675 { $$ = ast_operation (BINOP_BITWISE_AND, $1, $3); }
678 { $$ = ast_operation (BINOP_BITWISE_IOR, $1, $3); }
681 { $$ = ast_operation (BINOP_BITWISE_XOR, $1, $3); }
684 { $$ = ast_operation (BINOP_ADD, $1, $3); }
687 { $$ = ast_operation (BINOP_SUB, $1, $3); }
690 { $$ = ast_operation (BINOP_LOGICAL_OR, $1, $3); }
693 { $$ = ast_operation (BINOP_LOGICAL_AND, $1, $3); }
696 { $$ = ast_operation (BINOP_EQUAL, $1, $3); }
699 { $$ = ast_operation (BINOP_NOTEQUAL, $1, $3); }
702 { $$ = ast_operation (BINOP_LEQ, $1, $3); }
705 { $$ = ast_operation (BINOP_GEQ, $1, $3); }
708 { $$ = ast_operation (BINOP_LSH, $1, $3); }
711 { $$ = ast_operation (BINOP_RSH, $1, $3); }
716 { $$ = ast_cast ($1, $3); }
721 { $$ = ast_operation (BINOP_ASSIGN, $1, $3); }
724 compound_assignment_expr:
725 expr COMPOUND_ASSIGN expr
726 { $$ = ast_compound_assignment ($2, $1, $3); }
738 $$ = current_parser->new_op_vector ();
751 /* The result can't be NULL. */
752 $$ = current_parser->new_op_vector ();
759 '(' maybe_expr_list ')'
765 { $$ = ast_call_ish (OP_FUNCALL, $1, $2); }
776 | super_path KW_SUPER COLONCOLON
784 { $$ = ast_path ($1, NULL); }
786 { $$ = ast_path (make_stoken ("self"), NULL); }
790 identifier_path_for_expr
791 | KW_SELF COLONCOLON identifier_path_for_expr
792 { $$ = super_name ($3, 0); }
793 | maybe_self_path super_path identifier_path_for_expr
794 { $$ = super_name ($3, $2); }
795 | COLONCOLON identifier_path_for_expr
796 { $$ = crate_name ($2); }
797 | KW_EXTERN identifier_path_for_expr
799 /* This is a gdb extension to make it possible to
800 refer to items in other crates. It just bypasses
801 adding the current crate to the front of the
803 $$ = ast_path (rust_concat3 ("::", $2->left.sval.ptr, NULL),
808 identifier_path_for_expr:
810 { $$ = ast_path ($1, NULL); }
811 | identifier_path_for_expr COLONCOLON IDENT
813 $$ = ast_path (rust_concat3 ($1->left.sval.ptr, "::",
817 | identifier_path_for_expr COLONCOLON '<' type_list '>'
818 { $$ = ast_path ($1->left.sval, $4); }
819 | identifier_path_for_expr COLONCOLON '<' type_list RSH
821 $$ = ast_path ($1->left.sval, $4);
822 rust_push_back ('>');
827 identifier_path_for_type
828 | KW_SELF COLONCOLON identifier_path_for_type
829 { $$ = super_name ($3, 0); }
830 | maybe_self_path super_path identifier_path_for_type
831 { $$ = super_name ($3, $2); }
832 | COLONCOLON identifier_path_for_type
833 { $$ = crate_name ($2); }
834 | KW_EXTERN identifier_path_for_type
836 /* This is a gdb extension to make it possible to
837 refer to items in other crates. It just bypasses
838 adding the current crate to the front of the
840 $$ = ast_path (rust_concat3 ("::", $2->left.sval.ptr, NULL),
845 just_identifiers_for_type:
847 { $$ = ast_path ($1, NULL); }
848 | just_identifiers_for_type COLONCOLON IDENT
850 $$ = ast_path (rust_concat3 ($1->left.sval.ptr, "::",
856 identifier_path_for_type:
857 just_identifiers_for_type
858 | just_identifiers_for_type '<' type_list '>'
859 { $$ = ast_path ($1->left.sval, $3); }
860 | just_identifiers_for_type '<' type_list RSH
862 $$ = ast_path ($1->left.sval, $3);
863 rust_push_back ('>');
869 | '[' type ';' INTEGER ']'
870 { $$ = ast_array_type ($2, $4); }
871 | '[' type ';' DECIMAL_INTEGER ']'
872 { $$ = ast_array_type ($2, $4); }
874 { $$ = ast_slice_type ($3); }
876 { $$ = ast_reference_type ($2); }
878 { $$ = ast_pointer_type ($3, 1); }
880 { $$ = ast_pointer_type ($3, 0); }
881 | KW_FN '(' maybe_type_list ')' ARROW type
882 { $$ = ast_function_type ($6, $3); }
883 | '(' maybe_type_list ')'
884 { $$ = ast_tuple_type ($2); }
897 rust_op_vector *result = current_parser->new_op_vector ();
898 result->push_back ($1);
910 /* A struct of this type is used to describe a token. */
916 enum exp_opcode opcode;
919 /* Identifier tokens. */
921 static const struct token_info identifier_tokens[] =
923 { "as", KW_AS, OP_NULL },
924 { "false", KW_FALSE, OP_NULL },
925 { "if", 0, OP_NULL },
926 { "mut", KW_MUT, OP_NULL },
927 { "const", KW_CONST, OP_NULL },
928 { "self", KW_SELF, OP_NULL },
929 { "super", KW_SUPER, OP_NULL },
930 { "true", KW_TRUE, OP_NULL },
931 { "extern", KW_EXTERN, OP_NULL },
932 { "fn", KW_FN, OP_NULL },
933 { "sizeof", KW_SIZEOF, OP_NULL },
936 /* Operator tokens, sorted longest first. */
938 static const struct token_info operator_tokens[] =
940 { ">>=", COMPOUND_ASSIGN, BINOP_RSH },
941 { "<<=", COMPOUND_ASSIGN, BINOP_LSH },
943 { "<<", LSH, OP_NULL },
944 { ">>", RSH, OP_NULL },
945 { "&&", ANDAND, OP_NULL },
946 { "||", OROR, OP_NULL },
947 { "==", EQEQ, OP_NULL },
948 { "!=", NOTEQ, OP_NULL },
949 { "<=", LTEQ, OP_NULL },
950 { ">=", GTEQ, OP_NULL },
951 { "+=", COMPOUND_ASSIGN, BINOP_ADD },
952 { "-=", COMPOUND_ASSIGN, BINOP_SUB },
953 { "*=", COMPOUND_ASSIGN, BINOP_MUL },
954 { "/=", COMPOUND_ASSIGN, BINOP_DIV },
955 { "%=", COMPOUND_ASSIGN, BINOP_REM },
956 { "&=", COMPOUND_ASSIGN, BINOP_BITWISE_AND },
957 { "|=", COMPOUND_ASSIGN, BINOP_BITWISE_IOR },
958 { "^=", COMPOUND_ASSIGN, BINOP_BITWISE_XOR },
960 { "::", COLONCOLON, OP_NULL },
961 { "..", DOTDOT, OP_NULL },
962 { "->", ARROW, OP_NULL }
965 /* Helper function to copy to the name obstack. */
968 rust_copy_name (const char *name, int len)
970 return (const char *) obstack_copy0 (work_obstack, name, len);
973 /* Helper function to make an stoken from a C string. */
976 make_stoken (const char *p)
978 struct stoken result;
981 result.length = strlen (result.ptr);
985 /* Helper function to concatenate three strings on the name
989 rust_concat3 (const char *s1, const char *s2, const char *s3)
991 return make_stoken (obconcat (work_obstack, s1, s2, s3, (char *) NULL));
994 /* Return an AST node referring to NAME, but relative to the crate's
997 static const struct rust_op *
998 crate_name (const struct rust_op *name)
1000 std::string crate = rust_crate_for_block (expression_context_block);
1001 struct stoken result;
1003 gdb_assert (name->opcode == OP_VAR_VALUE);
1006 error (_("Could not find crate for current location"));
1007 result = make_stoken (obconcat (work_obstack, "::", crate.c_str (), "::",
1008 name->left.sval.ptr, (char *) NULL));
1010 return ast_path (result, name->right.params);
1013 /* Create an AST node referring to a "super::" qualified name. IDENT
1014 is the base name and N_SUPERS is how many "super::"s were
1015 provided. N_SUPERS can be zero. */
1017 static const struct rust_op *
1018 super_name (const struct rust_op *ident, unsigned int n_supers)
1020 const char *scope = block_scope (expression_context_block);
1023 gdb_assert (ident->opcode == OP_VAR_VALUE);
1025 if (scope[0] == '\0')
1026 error (_("Couldn't find namespace scope for self::"));
1031 std::vector<int> offsets;
1032 unsigned int current_len;
1034 current_len = cp_find_first_component (scope);
1035 while (scope[current_len] != '\0')
1037 offsets.push_back (current_len);
1038 gdb_assert (scope[current_len] == ':');
1041 current_len += cp_find_first_component (scope
1045 len = offsets.size ();
1046 if (n_supers >= len)
1047 error (_("Too many super:: uses from '%s'"), scope);
1049 offset = offsets[len - n_supers];
1052 offset = strlen (scope);
1054 obstack_grow (work_obstack, "::", 2);
1055 obstack_grow (work_obstack, scope, offset);
1056 obstack_grow (work_obstack, "::", 2);
1057 obstack_grow0 (work_obstack, ident->left.sval.ptr, ident->left.sval.length);
1059 return ast_path (make_stoken ((const char *) obstack_finish (work_obstack)),
1060 ident->right.params);
1063 /* A helper that updates the innermost block as appropriate. */
1066 update_innermost_block (struct block_symbol sym)
1068 if (symbol_read_needs_frame (sym.symbol))
1069 innermost_block.update (sym);
1072 /* A helper to look up a Rust type, or fail. This only works for
1073 types defined by rust_language_arch_info. */
1075 static struct type *
1076 rust_type (const char *name)
1080 type = language_lookup_primitive_type (current_parser->language (),
1081 current_parser->arch (),
1084 error (_("Could not find Rust type %s"), name);
1088 /* Lex a hex number with at least MIN digits and at most MAX
1092 lex_hex (int min, int max)
1094 uint32_t result = 0;
1096 /* We only want to stop at MAX if we're lexing a byte escape. */
1097 int check_max = min == max;
1099 while ((check_max ? len <= max : 1)
1100 && ((lexptr[0] >= 'a' && lexptr[0] <= 'f')
1101 || (lexptr[0] >= 'A' && lexptr[0] <= 'F')
1102 || (lexptr[0] >= '0' && lexptr[0] <= '9')))
1105 if (lexptr[0] >= 'a' && lexptr[0] <= 'f')
1106 result = result + 10 + lexptr[0] - 'a';
1107 else if (lexptr[0] >= 'A' && lexptr[0] <= 'F')
1108 result = result + 10 + lexptr[0] - 'A';
1110 result = result + lexptr[0] - '0';
1116 error (_("Not enough hex digits seen"));
1119 gdb_assert (min != max);
1120 error (_("Overlong hex escape"));
1126 /* Lex an escape. IS_BYTE is true if we're lexing a byte escape;
1127 otherwise we're lexing a character escape. */
1130 lex_escape (int is_byte)
1134 gdb_assert (lexptr[0] == '\\');
1140 result = lex_hex (2, 2);
1145 error (_("Unicode escape in byte literal"));
1147 if (lexptr[0] != '{')
1148 error (_("Missing '{' in Unicode escape"));
1150 result = lex_hex (1, 6);
1151 /* Could do range checks here. */
1152 if (lexptr[0] != '}')
1153 error (_("Missing '}' in Unicode escape"));
1187 error (_("Invalid escape \\%c in literal"), lexptr[0]);
1193 /* Lex a character constant. */
1196 lex_character (void)
1201 if (lexptr[0] == 'b')
1206 gdb_assert (lexptr[0] == '\'');
1208 /* This should handle UTF-8 here. */
1209 if (lexptr[0] == '\\')
1210 value = lex_escape (is_byte);
1213 value = lexptr[0] & 0xff;
1217 if (lexptr[0] != '\'')
1218 error (_("Unterminated character literal"));
1221 rustyylval.typed_val_int.val = value;
1222 rustyylval.typed_val_int.type = rust_type (is_byte ? "u8" : "char");
1227 /* Return the offset of the double quote if STR looks like the start
1228 of a raw string, or 0 if STR does not start a raw string. */
1231 starts_raw_string (const char *str)
1233 const char *save = str;
1238 while (str[0] == '#')
1245 /* Return true if STR looks like the end of a raw string that had N
1246 hashes at the start. */
1249 ends_raw_string (const char *str, int n)
1253 gdb_assert (str[0] == '"');
1254 for (i = 0; i < n; ++i)
1255 if (str[i + 1] != '#')
1260 /* Lex a string constant. */
1265 int is_byte = lexptr[0] == 'b';
1270 raw_length = starts_raw_string (lexptr);
1271 lexptr += raw_length;
1272 gdb_assert (lexptr[0] == '"');
1281 if (lexptr[0] == '"' && ends_raw_string (lexptr, raw_length - 1))
1283 /* Exit with lexptr pointing after the final "#". */
1284 lexptr += raw_length;
1287 else if (lexptr[0] == '\0')
1288 error (_("Unexpected EOF in string"));
1290 value = lexptr[0] & 0xff;
1291 if (is_byte && value > 127)
1292 error (_("Non-ASCII value in raw byte string"));
1293 obstack_1grow (work_obstack, value);
1297 else if (lexptr[0] == '"')
1299 /* Make sure to skip the quote. */
1303 else if (lexptr[0] == '\\')
1305 value = lex_escape (is_byte);
1308 obstack_1grow (work_obstack, value);
1310 convert_between_encodings ("UTF-32", "UTF-8", (gdb_byte *) &value,
1311 sizeof (value), sizeof (value),
1312 work_obstack, translit_none);
1314 else if (lexptr[0] == '\0')
1315 error (_("Unexpected EOF in string"));
1318 value = lexptr[0] & 0xff;
1319 if (is_byte && value > 127)
1320 error (_("Non-ASCII value in byte string"));
1321 obstack_1grow (work_obstack, value);
1326 rustyylval.sval.length = obstack_object_size (work_obstack);
1327 rustyylval.sval.ptr = (const char *) obstack_finish (work_obstack);
1328 return is_byte ? BYTESTRING : STRING;
1331 /* Return true if STRING starts with whitespace followed by a digit. */
1334 space_then_number (const char *string)
1336 const char *p = string;
1338 while (p[0] == ' ' || p[0] == '\t')
1343 return *p >= '0' && *p <= '9';
1346 /* Return true if C can start an identifier. */
1349 rust_identifier_start_p (char c)
1351 return ((c >= 'a' && c <= 'z')
1352 || (c >= 'A' && c <= 'Z')
1357 /* Lex an identifier. */
1360 lex_identifier (void)
1362 const char *start = lexptr;
1363 unsigned int length;
1364 const struct token_info *token;
1366 int is_gdb_var = lexptr[0] == '$';
1368 gdb_assert (rust_identifier_start_p (lexptr[0]));
1372 /* For the time being this doesn't handle Unicode rules. Non-ASCII
1373 identifiers are gated anyway. */
1374 while ((lexptr[0] >= 'a' && lexptr[0] <= 'z')
1375 || (lexptr[0] >= 'A' && lexptr[0] <= 'Z')
1377 || (is_gdb_var && lexptr[0] == '$')
1378 || (lexptr[0] >= '0' && lexptr[0] <= '9'))
1382 length = lexptr - start;
1384 for (i = 0; i < ARRAY_SIZE (identifier_tokens); ++i)
1386 if (length == strlen (identifier_tokens[i].name)
1387 && strncmp (identifier_tokens[i].name, start, length) == 0)
1389 token = &identifier_tokens[i];
1396 if (token->value == 0)
1398 /* Leave the terminating token alone. */
1403 else if (token == NULL
1404 && (strncmp (start, "thread", length) == 0
1405 || strncmp (start, "task", length) == 0)
1406 && space_then_number (lexptr))
1408 /* "task" or "thread" followed by a number terminates the
1409 parse, per gdb rules. */
1414 if (token == NULL || (parse_completion && lexptr[0] == '\0'))
1415 rustyylval.sval = make_stoken (rust_copy_name (start, length));
1417 if (parse_completion && lexptr[0] == '\0')
1419 /* Prevent rustyylex from returning two COMPLETE tokens. */
1420 prev_lexptr = lexptr;
1425 return token->value;
1431 /* Lex an operator. */
1436 const struct token_info *token = NULL;
1439 for (i = 0; i < ARRAY_SIZE (operator_tokens); ++i)
1441 if (strncmp (operator_tokens[i].name, lexptr,
1442 strlen (operator_tokens[i].name)) == 0)
1444 lexptr += strlen (operator_tokens[i].name);
1445 token = &operator_tokens[i];
1452 rustyylval.opcode = token->opcode;
1453 return token->value;
1464 regmatch_t subexps[NUM_SUBEXPRESSIONS];
1467 int could_be_decimal = 1;
1468 int implicit_i32 = 0;
1469 const char *type_name = NULL;
1472 int type_index = -1;
1475 match = regexec (&number_regex, lexptr, ARRAY_SIZE (subexps), subexps, 0);
1476 /* Failure means the regexp is broken. */
1477 gdb_assert (match == 0);
1479 if (subexps[INT_TEXT].rm_so != -1)
1481 /* Integer part matched. */
1483 end_index = subexps[INT_TEXT].rm_eo;
1484 if (subexps[INT_TYPE].rm_so == -1)
1491 type_index = INT_TYPE;
1492 could_be_decimal = 0;
1495 else if (subexps[FLOAT_TYPE1].rm_so != -1)
1497 /* Found floating point type suffix. */
1498 end_index = subexps[FLOAT_TYPE1].rm_so;
1499 type_index = FLOAT_TYPE1;
1501 else if (subexps[FLOAT_TYPE2].rm_so != -1)
1503 /* Found floating point type suffix. */
1504 end_index = subexps[FLOAT_TYPE2].rm_so;
1505 type_index = FLOAT_TYPE2;
1509 /* Any other floating point match. */
1510 end_index = subexps[0].rm_eo;
1514 /* We need a special case if the final character is ".". In this
1515 case we might need to parse an integer. For example, "23.f()" is
1516 a request for a trait method call, not a syntax error involving
1517 the floating point number "23.". */
1518 gdb_assert (subexps[0].rm_eo > 0);
1519 if (lexptr[subexps[0].rm_eo - 1] == '.')
1521 const char *next = skip_spaces (&lexptr[subexps[0].rm_eo]);
1523 if (rust_identifier_start_p (*next) || *next == '.')
1527 end_index = subexps[0].rm_eo;
1529 could_be_decimal = 1;
1534 /* Compute the type name if we haven't already. */
1535 std::string type_name_holder;
1536 if (type_name == NULL)
1538 gdb_assert (type_index != -1);
1539 type_name_holder = std::string (lexptr + subexps[type_index].rm_so,
1540 (subexps[type_index].rm_eo
1541 - subexps[type_index].rm_so));
1542 type_name = type_name_holder.c_str ();
1545 /* Look up the type. */
1546 type = rust_type (type_name);
1548 /* Copy the text of the number and remove the "_"s. */
1550 for (i = 0; i < end_index && lexptr[i]; ++i)
1552 if (lexptr[i] == '_')
1553 could_be_decimal = 0;
1555 number.push_back (lexptr[i]);
1558 /* Advance past the match. */
1559 lexptr += subexps[0].rm_eo;
1561 /* Parse the number. */
1568 if (number[0] == '0')
1570 if (number[1] == 'x')
1572 else if (number[1] == 'o')
1574 else if (number[1] == 'b')
1579 could_be_decimal = 0;
1583 value = strtoul (number.c_str () + offset, NULL, radix);
1584 if (implicit_i32 && value >= ((uint64_t) 1) << 31)
1585 type = rust_type ("i64");
1587 rustyylval.typed_val_int.val = value;
1588 rustyylval.typed_val_int.type = type;
1592 rustyylval.typed_val_float.type = type;
1593 bool parsed = parse_float (number.c_str (), number.length (),
1594 rustyylval.typed_val_float.type,
1595 rustyylval.typed_val_float.val);
1596 gdb_assert (parsed);
1599 return is_integer ? (could_be_decimal ? DECIMAL_INTEGER : INTEGER) : FLOAT;
1607 /* Skip all leading whitespace. */
1608 while (lexptr[0] == ' ' || lexptr[0] == '\t' || lexptr[0] == '\r'
1609 || lexptr[0] == '\n')
1612 /* If we hit EOF and we're completing, then return COMPLETE -- maybe
1613 we're completing an empty string at the end of a field_expr.
1614 But, we don't want to return two COMPLETE tokens in a row. */
1615 if (lexptr[0] == '\0' && lexptr == prev_lexptr)
1617 prev_lexptr = lexptr;
1618 if (lexptr[0] == '\0')
1620 if (parse_completion)
1622 rustyylval.sval = make_stoken ("");
1628 if (lexptr[0] >= '0' && lexptr[0] <= '9')
1629 return lex_number ();
1630 else if (lexptr[0] == 'b' && lexptr[1] == '\'')
1631 return lex_character ();
1632 else if (lexptr[0] == 'b' && lexptr[1] == '"')
1633 return lex_string ();
1634 else if (lexptr[0] == 'b' && starts_raw_string (lexptr + 1))
1635 return lex_string ();
1636 else if (starts_raw_string (lexptr))
1637 return lex_string ();
1638 else if (rust_identifier_start_p (lexptr[0]))
1639 return lex_identifier ();
1640 else if (lexptr[0] == '"')
1641 return lex_string ();
1642 else if (lexptr[0] == '\'')
1643 return lex_character ();
1644 else if (lexptr[0] == '}' || lexptr[0] == ']')
1646 /* Falls through to lex_operator. */
1649 else if (lexptr[0] == '(' || lexptr[0] == '{')
1651 /* Falls through to lex_operator. */
1654 else if (lexptr[0] == ',' && comma_terminates && paren_depth == 0)
1657 return lex_operator ();
1660 /* Push back a single character to be re-lexed. */
1663 rust_push_back (char c)
1665 /* Can't be called before any lexing. */
1666 gdb_assert (prev_lexptr != NULL);
1669 gdb_assert (*lexptr == c);
1674 /* Make an arbitrary operation and fill in the fields. */
1676 static const struct rust_op *
1677 ast_operation (enum exp_opcode opcode, const struct rust_op *left,
1678 const struct rust_op *right)
1680 struct rust_op *result = OBSTACK_ZALLOC (work_obstack, struct rust_op);
1682 result->opcode = opcode;
1683 result->left.op = left;
1684 result->right.op = right;
1689 /* Make a compound assignment operation. */
1691 static const struct rust_op *
1692 ast_compound_assignment (enum exp_opcode opcode, const struct rust_op *left,
1693 const struct rust_op *right)
1695 struct rust_op *result = OBSTACK_ZALLOC (work_obstack, struct rust_op);
1697 result->opcode = opcode;
1698 result->compound_assignment = 1;
1699 result->left.op = left;
1700 result->right.op = right;
1705 /* Make a typed integer literal operation. */
1707 static const struct rust_op *
1708 ast_literal (struct typed_val_int val)
1710 struct rust_op *result = OBSTACK_ZALLOC (work_obstack, struct rust_op);
1712 result->opcode = OP_LONG;
1713 result->left.typed_val_int = val;
1718 /* Make a typed floating point literal operation. */
1720 static const struct rust_op *
1721 ast_dliteral (struct typed_val_float val)
1723 struct rust_op *result = OBSTACK_ZALLOC (work_obstack, struct rust_op);
1725 result->opcode = OP_FLOAT;
1726 result->left.typed_val_float = val;
1731 /* Make a unary operation. */
1733 static const struct rust_op *
1734 ast_unary (enum exp_opcode opcode, const struct rust_op *expr)
1736 return ast_operation (opcode, expr, NULL);
1739 /* Make a cast operation. */
1741 static const struct rust_op *
1742 ast_cast (const struct rust_op *expr, const struct rust_op *type)
1744 struct rust_op *result = OBSTACK_ZALLOC (work_obstack, struct rust_op);
1746 result->opcode = UNOP_CAST;
1747 result->left.op = expr;
1748 result->right.op = type;
1753 /* Make a call-like operation. This is nominally a function call, but
1754 when lowering we may discover that it actually represents the
1755 creation of a tuple struct. */
1757 static const struct rust_op *
1758 ast_call_ish (enum exp_opcode opcode, const struct rust_op *expr,
1759 rust_op_vector *params)
1761 struct rust_op *result = OBSTACK_ZALLOC (work_obstack, struct rust_op);
1763 result->opcode = opcode;
1764 result->left.op = expr;
1765 result->right.params = params;
1770 /* Make a structure creation operation. */
1772 static const struct rust_op *
1773 ast_struct (const struct rust_op *name, rust_set_vector *fields)
1775 struct rust_op *result = OBSTACK_ZALLOC (work_obstack, struct rust_op);
1777 result->opcode = OP_AGGREGATE;
1778 result->left.op = name;
1779 result->right.field_inits = fields;
1784 /* Make an identifier path. */
1786 static const struct rust_op *
1787 ast_path (struct stoken path, rust_op_vector *params)
1789 struct rust_op *result = OBSTACK_ZALLOC (work_obstack, struct rust_op);
1791 result->opcode = OP_VAR_VALUE;
1792 result->left.sval = path;
1793 result->right.params = params;
1798 /* Make a string constant operation. */
1800 static const struct rust_op *
1801 ast_string (struct stoken str)
1803 struct rust_op *result = OBSTACK_ZALLOC (work_obstack, struct rust_op);
1805 result->opcode = OP_STRING;
1806 result->left.sval = str;
1811 /* Make a field expression. */
1813 static const struct rust_op *
1814 ast_structop (const struct rust_op *left, const char *name, int completing)
1816 struct rust_op *result = OBSTACK_ZALLOC (work_obstack, struct rust_op);
1818 result->opcode = STRUCTOP_STRUCT;
1819 result->completing = completing;
1820 result->left.op = left;
1821 result->right.sval = make_stoken (name);
1826 /* Make an anonymous struct operation, like 'x.0'. */
1828 static const struct rust_op *
1829 ast_structop_anonymous (const struct rust_op *left,
1830 struct typed_val_int number)
1832 struct rust_op *result = OBSTACK_ZALLOC (work_obstack, struct rust_op);
1834 result->opcode = STRUCTOP_ANONYMOUS;
1835 result->left.op = left;
1836 result->right.typed_val_int = number;
1841 /* Make a range operation. */
1843 static const struct rust_op *
1844 ast_range (const struct rust_op *lhs, const struct rust_op *rhs)
1846 struct rust_op *result = OBSTACK_ZALLOC (work_obstack, struct rust_op);
1848 result->opcode = OP_RANGE;
1849 result->left.op = lhs;
1850 result->right.op = rhs;
1855 /* A helper function to make a type-related AST node. */
1857 static struct rust_op *
1858 ast_basic_type (enum type_code typecode)
1860 struct rust_op *result = OBSTACK_ZALLOC (work_obstack, struct rust_op);
1862 result->opcode = OP_TYPE;
1863 result->typecode = typecode;
1867 /* Create an AST node describing an array type. */
1869 static const struct rust_op *
1870 ast_array_type (const struct rust_op *lhs, struct typed_val_int val)
1872 struct rust_op *result = ast_basic_type (TYPE_CODE_ARRAY);
1874 result->left.op = lhs;
1875 result->right.typed_val_int = val;
1879 /* Create an AST node describing a reference type. */
1881 static const struct rust_op *
1882 ast_slice_type (const struct rust_op *type)
1884 /* Use TYPE_CODE_COMPLEX just because it is handy. */
1885 struct rust_op *result = ast_basic_type (TYPE_CODE_COMPLEX);
1887 result->left.op = type;
1891 /* Create an AST node describing a reference type. */
1893 static const struct rust_op *
1894 ast_reference_type (const struct rust_op *type)
1896 struct rust_op *result = ast_basic_type (TYPE_CODE_REF);
1898 result->left.op = type;
1902 /* Create an AST node describing a pointer type. */
1904 static const struct rust_op *
1905 ast_pointer_type (const struct rust_op *type, int is_mut)
1907 struct rust_op *result = ast_basic_type (TYPE_CODE_PTR);
1909 result->left.op = type;
1910 /* For the time being we ignore is_mut. */
1914 /* Create an AST node describing a function type. */
1916 static const struct rust_op *
1917 ast_function_type (const struct rust_op *rtype, rust_op_vector *params)
1919 struct rust_op *result = ast_basic_type (TYPE_CODE_FUNC);
1921 result->left.op = rtype;
1922 result->right.params = params;
1926 /* Create an AST node describing a tuple type. */
1928 static const struct rust_op *
1929 ast_tuple_type (rust_op_vector *params)
1931 struct rust_op *result = ast_basic_type (TYPE_CODE_STRUCT);
1933 result->left.params = params;
1937 /* A helper to appropriately munge NAME and BLOCK depending on the
1938 presence of a leading "::". */
1941 munge_name_and_block (const char **name, const struct block **block)
1943 /* If it is a global reference, skip the current block in favor of
1944 the static block. */
1945 if (strncmp (*name, "::", 2) == 0)
1948 *block = block_static_block (*block);
1952 /* Like lookup_symbol, but handles Rust namespace conventions, and
1953 doesn't require field_of_this_result. */
1955 static struct block_symbol
1956 rust_lookup_symbol (const char *name, const struct block *block,
1957 const domain_enum domain)
1959 struct block_symbol result;
1961 munge_name_and_block (&name, &block);
1963 result = lookup_symbol (name, block, domain, NULL);
1964 if (result.symbol != NULL)
1965 update_innermost_block (result);
1969 /* Look up a type, following Rust namespace conventions. */
1971 static struct type *
1972 rust_lookup_type (const char *name, const struct block *block)
1974 struct block_symbol result;
1977 munge_name_and_block (&name, &block);
1979 result = lookup_symbol (name, block, STRUCT_DOMAIN, NULL);
1980 if (result.symbol != NULL)
1982 update_innermost_block (result);
1983 return SYMBOL_TYPE (result.symbol);
1986 type = lookup_typename (current_parser->language (), current_parser->arch (),
1991 /* Last chance, try a built-in type. */
1992 return language_lookup_primitive_type (current_parser->language (),
1993 current_parser->arch (),
1997 static struct type *convert_ast_to_type (struct parser_state *state,
1998 const struct rust_op *operation);
1999 static const char *convert_name (struct parser_state *state,
2000 const struct rust_op *operation);
2002 /* Convert a vector of rust_ops representing types to a vector of
2005 static std::vector<struct type *>
2006 convert_params_to_types (struct parser_state *state, rust_op_vector *params)
2008 std::vector<struct type *> result;
2010 for (const rust_op *op : *params)
2011 result.push_back (convert_ast_to_type (state, op));
2016 /* Convert a rust_op representing a type to a struct type *. */
2018 static struct type *
2019 convert_ast_to_type (struct parser_state *state,
2020 const struct rust_op *operation)
2022 struct type *type, *result = NULL;
2024 if (operation->opcode == OP_VAR_VALUE)
2026 const char *varname = convert_name (state, operation);
2028 result = rust_lookup_type (varname, expression_context_block);
2030 error (_("No typed name '%s' in current context"), varname);
2034 gdb_assert (operation->opcode == OP_TYPE);
2036 switch (operation->typecode)
2038 case TYPE_CODE_ARRAY:
2039 type = convert_ast_to_type (state, operation->left.op);
2040 if (operation->right.typed_val_int.val < 0)
2041 error (_("Negative array length"));
2042 result = lookup_array_range_type (type, 0,
2043 operation->right.typed_val_int.val - 1);
2046 case TYPE_CODE_COMPLEX:
2048 struct type *usize = rust_type ("usize");
2050 type = convert_ast_to_type (state, operation->left.op);
2051 result = rust_slice_type ("&[*gdb*]", type, usize);
2057 /* For now we treat &x and *x identically. */
2058 type = convert_ast_to_type (state, operation->left.op);
2059 result = lookup_pointer_type (type);
2062 case TYPE_CODE_FUNC:
2064 std::vector<struct type *> args
2065 (convert_params_to_types (state, operation->right.params));
2066 struct type **argtypes = NULL;
2068 type = convert_ast_to_type (state, operation->left.op);
2070 argtypes = args.data ();
2073 = lookup_function_type_with_arguments (type, args.size (),
2075 result = lookup_pointer_type (result);
2079 case TYPE_CODE_STRUCT:
2081 std::vector<struct type *> args
2082 (convert_params_to_types (state, operation->left.params));
2086 obstack_1grow (work_obstack, '(');
2087 for (i = 0; i < args.size (); ++i)
2089 std::string type_name = type_to_string (args[i]);
2092 obstack_1grow (work_obstack, ',');
2093 obstack_grow_str (work_obstack, type_name.c_str ());
2096 obstack_grow_str0 (work_obstack, ")");
2097 name = (const char *) obstack_finish (work_obstack);
2099 /* We don't allow creating new tuple types (yet), but we do
2100 allow looking up existing tuple types. */
2101 result = rust_lookup_type (name, expression_context_block);
2103 error (_("could not find tuple type '%s'"), name);
2108 gdb_assert_not_reached ("unhandled opcode in convert_ast_to_type");
2111 gdb_assert (result != NULL);
2115 /* A helper function to turn a rust_op representing a name into a full
2116 name. This applies generic arguments as needed. The returned name
2117 is allocated on the work obstack. */
2120 convert_name (struct parser_state *state, const struct rust_op *operation)
2124 gdb_assert (operation->opcode == OP_VAR_VALUE);
2126 if (operation->right.params == NULL)
2127 return operation->left.sval.ptr;
2129 std::vector<struct type *> types
2130 (convert_params_to_types (state, operation->right.params));
2132 obstack_grow_str (work_obstack, operation->left.sval.ptr);
2133 obstack_1grow (work_obstack, '<');
2134 for (i = 0; i < types.size (); ++i)
2136 std::string type_name = type_to_string (types[i]);
2139 obstack_1grow (work_obstack, ',');
2141 obstack_grow_str (work_obstack, type_name.c_str ());
2143 obstack_grow_str0 (work_obstack, ">");
2145 return (const char *) obstack_finish (work_obstack);
2148 static void convert_ast_to_expression (struct parser_state *state,
2149 const struct rust_op *operation,
2150 const struct rust_op *top,
2151 bool want_type = false);
2153 /* A helper function that converts a vec of rust_ops to a gdb
2157 convert_params_to_expression (struct parser_state *state,
2158 rust_op_vector *params,
2159 const struct rust_op *top)
2161 for (const rust_op *elem : *params)
2162 convert_ast_to_expression (state, elem, top);
2165 /* Lower a rust_op to a gdb expression. STATE is the parser state.
2166 OPERATION is the operation to lower. TOP is a pointer to the
2167 top-most operation; it is used to handle the special case where the
2168 top-most expression is an identifier and can be optionally lowered
2169 to OP_TYPE. WANT_TYPE is a flag indicating that, if the expression
2170 is the name of a type, then emit an OP_TYPE for it (rather than
2171 erroring). If WANT_TYPE is set, then the similar TOP handling is
2175 convert_ast_to_expression (struct parser_state *state,
2176 const struct rust_op *operation,
2177 const struct rust_op *top,
2180 switch (operation->opcode)
2183 write_exp_elt_opcode (state, OP_LONG);
2184 write_exp_elt_type (state, operation->left.typed_val_int.type);
2185 write_exp_elt_longcst (state, operation->left.typed_val_int.val);
2186 write_exp_elt_opcode (state, OP_LONG);
2190 write_exp_elt_opcode (state, OP_FLOAT);
2191 write_exp_elt_type (state, operation->left.typed_val_float.type);
2192 write_exp_elt_floatcst (state, operation->left.typed_val_float.val);
2193 write_exp_elt_opcode (state, OP_FLOAT);
2196 case STRUCTOP_STRUCT:
2198 convert_ast_to_expression (state, operation->left.op, top);
2200 if (operation->completing)
2201 mark_struct_expression (state);
2202 write_exp_elt_opcode (state, STRUCTOP_STRUCT);
2203 write_exp_string (state, operation->right.sval);
2204 write_exp_elt_opcode (state, STRUCTOP_STRUCT);
2208 case STRUCTOP_ANONYMOUS:
2210 convert_ast_to_expression (state, operation->left.op, top);
2212 write_exp_elt_opcode (state, STRUCTOP_ANONYMOUS);
2213 write_exp_elt_longcst (state, operation->right.typed_val_int.val);
2214 write_exp_elt_opcode (state, STRUCTOP_ANONYMOUS);
2219 convert_ast_to_expression (state, operation->left.op, top, true);
2220 write_exp_elt_opcode (state, UNOP_SIZEOF);
2225 case UNOP_COMPLEMENT:
2228 convert_ast_to_expression (state, operation->left.op, top);
2229 write_exp_elt_opcode (state, operation->opcode);
2232 case BINOP_SUBSCRIPT:
2239 case BINOP_BITWISE_AND:
2240 case BINOP_BITWISE_IOR:
2241 case BINOP_BITWISE_XOR:
2244 case BINOP_LOGICAL_OR:
2245 case BINOP_LOGICAL_AND:
2247 case BINOP_NOTEQUAL:
2254 convert_ast_to_expression (state, operation->left.op, top);
2255 convert_ast_to_expression (state, operation->right.op, top);
2256 if (operation->compound_assignment)
2258 write_exp_elt_opcode (state, BINOP_ASSIGN_MODIFY);
2259 write_exp_elt_opcode (state, operation->opcode);
2260 write_exp_elt_opcode (state, BINOP_ASSIGN_MODIFY);
2263 write_exp_elt_opcode (state, operation->opcode);
2265 if (operation->compound_assignment
2266 || operation->opcode == BINOP_ASSIGN)
2270 type = language_lookup_primitive_type (parse_language (state),
2271 parse_gdbarch (state),
2274 write_exp_elt_opcode (state, OP_LONG);
2275 write_exp_elt_type (state, type);
2276 write_exp_elt_longcst (state, 0);
2277 write_exp_elt_opcode (state, OP_LONG);
2279 write_exp_elt_opcode (state, BINOP_COMMA);
2285 struct type *type = convert_ast_to_type (state, operation->right.op);
2287 convert_ast_to_expression (state, operation->left.op, top);
2288 write_exp_elt_opcode (state, UNOP_CAST);
2289 write_exp_elt_type (state, type);
2290 write_exp_elt_opcode (state, UNOP_CAST);
2296 if (operation->left.op->opcode == OP_VAR_VALUE)
2299 const char *varname = convert_name (state, operation->left.op);
2301 type = rust_lookup_type (varname, expression_context_block);
2304 /* This is actually a tuple struct expression, not a
2306 rust_op_vector *params = operation->right.params;
2308 if (TYPE_CODE (type) != TYPE_CODE_NAMESPACE)
2310 if (!rust_tuple_struct_type_p (type))
2311 error (_("Type %s is not a tuple struct"), varname);
2313 for (int i = 0; i < params->size (); ++i)
2315 char *cell = get_print_cell ();
2317 xsnprintf (cell, PRINT_CELL_SIZE, "__%d", i);
2318 write_exp_elt_opcode (state, OP_NAME);
2319 write_exp_string (state, make_stoken (cell));
2320 write_exp_elt_opcode (state, OP_NAME);
2322 convert_ast_to_expression (state, (*params)[i], top);
2325 write_exp_elt_opcode (state, OP_AGGREGATE);
2326 write_exp_elt_type (state, type);
2327 write_exp_elt_longcst (state, 2 * params->size ());
2328 write_exp_elt_opcode (state, OP_AGGREGATE);
2333 convert_ast_to_expression (state, operation->left.op, top);
2334 convert_params_to_expression (state, operation->right.params, top);
2335 write_exp_elt_opcode (state, OP_FUNCALL);
2336 write_exp_elt_longcst (state, operation->right.params->size ());
2337 write_exp_elt_longcst (state, OP_FUNCALL);
2342 gdb_assert (operation->left.op == NULL);
2343 convert_params_to_expression (state, operation->right.params, top);
2344 write_exp_elt_opcode (state, OP_ARRAY);
2345 write_exp_elt_longcst (state, 0);
2346 write_exp_elt_longcst (state, operation->right.params->size () - 1);
2347 write_exp_elt_longcst (state, OP_ARRAY);
2352 struct block_symbol sym;
2353 const char *varname;
2355 if (operation->left.sval.ptr[0] == '$')
2357 write_dollar_variable (state, operation->left.sval);
2361 varname = convert_name (state, operation);
2362 sym = rust_lookup_symbol (varname, expression_context_block,
2364 if (sym.symbol != NULL && SYMBOL_CLASS (sym.symbol) != LOC_TYPEDEF)
2366 write_exp_elt_opcode (state, OP_VAR_VALUE);
2367 write_exp_elt_block (state, sym.block);
2368 write_exp_elt_sym (state, sym.symbol);
2369 write_exp_elt_opcode (state, OP_VAR_VALUE);
2373 struct type *type = NULL;
2375 if (sym.symbol != NULL)
2377 gdb_assert (SYMBOL_CLASS (sym.symbol) == LOC_TYPEDEF);
2378 type = SYMBOL_TYPE (sym.symbol);
2381 type = rust_lookup_type (varname, expression_context_block);
2383 error (_("No symbol '%s' in current context"), varname);
2386 && TYPE_CODE (type) == TYPE_CODE_STRUCT
2387 && TYPE_NFIELDS (type) == 0)
2389 /* A unit-like struct. */
2390 write_exp_elt_opcode (state, OP_AGGREGATE);
2391 write_exp_elt_type (state, type);
2392 write_exp_elt_longcst (state, 0);
2393 write_exp_elt_opcode (state, OP_AGGREGATE);
2395 else if (want_type || operation == top)
2397 write_exp_elt_opcode (state, OP_TYPE);
2398 write_exp_elt_type (state, type);
2399 write_exp_elt_opcode (state, OP_TYPE);
2402 error (_("Found type '%s', which can't be "
2403 "evaluated in this context"),
2412 rust_set_vector *fields = operation->right.field_inits;
2417 for (const set_field &init : *fields)
2419 if (init.name.ptr != NULL)
2421 write_exp_elt_opcode (state, OP_NAME);
2422 write_exp_string (state, init.name);
2423 write_exp_elt_opcode (state, OP_NAME);
2427 convert_ast_to_expression (state, init.init, top);
2430 if (init.name.ptr == NULL)
2432 /* This is handled differently from Ada in our
2434 write_exp_elt_opcode (state, OP_OTHERS);
2438 name = convert_name (state, operation->left.op);
2439 type = rust_lookup_type (name, expression_context_block);
2441 error (_("Could not find type '%s'"), operation->left.sval.ptr);
2443 if (TYPE_CODE (type) != TYPE_CODE_STRUCT
2444 || rust_tuple_type_p (type)
2445 || rust_tuple_struct_type_p (type))
2446 error (_("Struct expression applied to non-struct type"));
2448 write_exp_elt_opcode (state, OP_AGGREGATE);
2449 write_exp_elt_type (state, type);
2450 write_exp_elt_longcst (state, length);
2451 write_exp_elt_opcode (state, OP_AGGREGATE);
2457 write_exp_elt_opcode (state, OP_STRING);
2458 write_exp_string (state, operation->left.sval);
2459 write_exp_elt_opcode (state, OP_STRING);
2465 enum range_type kind = BOTH_BOUND_DEFAULT;
2467 if (operation->left.op != NULL)
2469 convert_ast_to_expression (state, operation->left.op, top);
2470 kind = HIGH_BOUND_DEFAULT;
2472 if (operation->right.op != NULL)
2474 convert_ast_to_expression (state, operation->right.op, top);
2475 if (kind == BOTH_BOUND_DEFAULT)
2476 kind = LOW_BOUND_DEFAULT;
2479 gdb_assert (kind == HIGH_BOUND_DEFAULT);
2480 kind = NONE_BOUND_DEFAULT;
2483 write_exp_elt_opcode (state, OP_RANGE);
2484 write_exp_elt_longcst (state, kind);
2485 write_exp_elt_opcode (state, OP_RANGE);
2490 gdb_assert_not_reached ("unhandled opcode in convert_ast_to_expression");
2496 /* The parser as exposed to gdb. */
2499 rust_parse (struct parser_state *state)
2503 /* This sets various globals and also clears them on
2505 rust_parser parser (state);
2507 result = rustyyparse ();
2509 if (!result || (parse_completion && parser.rust_ast != NULL))
2510 convert_ast_to_expression (state, parser.rust_ast, parser.rust_ast);
2515 /* The parser error handler. */
2518 rustyyerror (const char *msg)
2520 const char *where = prev_lexptr ? prev_lexptr : lexptr;
2521 error (_("%s in expression, near `%s'."), (msg ? msg : "Error"), where);
2528 /* Initialize the lexer for testing. */
2531 rust_lex_test_init (const char *input)
2538 /* A test helper that lexes a string, expecting a single token. It
2539 returns the lexer data for this token. */
2542 rust_lex_test_one (const char *input, int expected)
2547 rust_lex_test_init (input);
2549 token = rustyylex ();
2550 SELF_CHECK (token == expected);
2551 result = rustyylval;
2555 token = rustyylex ();
2556 SELF_CHECK (token == 0);
2562 /* Test that INPUT lexes as the integer VALUE. */
2565 rust_lex_int_test (const char *input, int value, int kind)
2567 RUSTSTYPE result = rust_lex_test_one (input, kind);
2568 SELF_CHECK (result.typed_val_int.val == value);
2571 /* Test that INPUT throws an exception with text ERR. */
2574 rust_lex_exception_test (const char *input, const char *err)
2578 /* The "kind" doesn't matter. */
2579 rust_lex_test_one (input, DECIMAL_INTEGER);
2582 CATCH (except, RETURN_MASK_ERROR)
2584 SELF_CHECK (strcmp (except.message, err) == 0);
2589 /* Test that INPUT lexes as the identifier, string, or byte-string
2590 VALUE. KIND holds the expected token kind. */
2593 rust_lex_stringish_test (const char *input, const char *value, int kind)
2595 RUSTSTYPE result = rust_lex_test_one (input, kind);
2596 SELF_CHECK (result.sval.length == strlen (value));
2597 SELF_CHECK (strncmp (result.sval.ptr, value, result.sval.length) == 0);
2600 /* Helper to test that a string parses as a given token sequence. */
2603 rust_lex_test_sequence (const char *input, int len, const int expected[])
2610 for (i = 0; i < len; ++i)
2612 int token = rustyylex ();
2614 SELF_CHECK (token == expected[i]);
2618 /* Tests for an integer-parsing corner case. */
2621 rust_lex_test_trailing_dot (void)
2623 const int expected1[] = { DECIMAL_INTEGER, '.', IDENT, '(', ')', 0 };
2624 const int expected2[] = { INTEGER, '.', IDENT, '(', ')', 0 };
2625 const int expected3[] = { FLOAT, EQEQ, '(', ')', 0 };
2626 const int expected4[] = { DECIMAL_INTEGER, DOTDOT, DECIMAL_INTEGER, 0 };
2628 rust_lex_test_sequence ("23.g()", ARRAY_SIZE (expected1), expected1);
2629 rust_lex_test_sequence ("23_0.g()", ARRAY_SIZE (expected2), expected2);
2630 rust_lex_test_sequence ("23.==()", ARRAY_SIZE (expected3), expected3);
2631 rust_lex_test_sequence ("23..25", ARRAY_SIZE (expected4), expected4);
2634 /* Tests of completion. */
2637 rust_lex_test_completion (void)
2639 const int expected[] = { IDENT, '.', COMPLETE, 0 };
2641 parse_completion = 1;
2643 rust_lex_test_sequence ("something.wha", ARRAY_SIZE (expected), expected);
2644 rust_lex_test_sequence ("something.", ARRAY_SIZE (expected), expected);
2646 parse_completion = 0;
2649 /* Test pushback. */
2652 rust_lex_test_push_back (void)
2656 rust_lex_test_init (">>=");
2658 token = rustyylex ();
2659 SELF_CHECK (token == COMPOUND_ASSIGN);
2660 SELF_CHECK (rustyylval.opcode == BINOP_RSH);
2662 rust_push_back ('=');
2664 token = rustyylex ();
2665 SELF_CHECK (token == '=');
2667 token = rustyylex ();
2668 SELF_CHECK (token == 0);
2671 /* Unit test the lexer. */
2674 rust_lex_tests (void)
2678 auto_obstack test_obstack;
2679 scoped_restore obstack_holder = make_scoped_restore (&work_obstack,
2682 // Set up dummy "parser", so that rust_type works.
2683 struct parser_state ps (0, &rust_language_defn, target_gdbarch ());
2684 rust_parser parser (&ps);
2686 rust_lex_test_one ("", 0);
2687 rust_lex_test_one (" \t \n \r ", 0);
2688 rust_lex_test_one ("thread 23", 0);
2689 rust_lex_test_one ("task 23", 0);
2690 rust_lex_test_one ("th 104", 0);
2691 rust_lex_test_one ("ta 97", 0);
2693 rust_lex_int_test ("'z'", 'z', INTEGER);
2694 rust_lex_int_test ("'\\xff'", 0xff, INTEGER);
2695 rust_lex_int_test ("'\\u{1016f}'", 0x1016f, INTEGER);
2696 rust_lex_int_test ("b'z'", 'z', INTEGER);
2697 rust_lex_int_test ("b'\\xfe'", 0xfe, INTEGER);
2698 rust_lex_int_test ("b'\\xFE'", 0xfe, INTEGER);
2699 rust_lex_int_test ("b'\\xfE'", 0xfe, INTEGER);
2701 /* Test all escapes in both modes. */
2702 rust_lex_int_test ("'\\n'", '\n', INTEGER);
2703 rust_lex_int_test ("'\\r'", '\r', INTEGER);
2704 rust_lex_int_test ("'\\t'", '\t', INTEGER);
2705 rust_lex_int_test ("'\\\\'", '\\', INTEGER);
2706 rust_lex_int_test ("'\\0'", '\0', INTEGER);
2707 rust_lex_int_test ("'\\''", '\'', INTEGER);
2708 rust_lex_int_test ("'\\\"'", '"', INTEGER);
2710 rust_lex_int_test ("b'\\n'", '\n', INTEGER);
2711 rust_lex_int_test ("b'\\r'", '\r', INTEGER);
2712 rust_lex_int_test ("b'\\t'", '\t', INTEGER);
2713 rust_lex_int_test ("b'\\\\'", '\\', INTEGER);
2714 rust_lex_int_test ("b'\\0'", '\0', INTEGER);
2715 rust_lex_int_test ("b'\\''", '\'', INTEGER);
2716 rust_lex_int_test ("b'\\\"'", '"', INTEGER);
2718 rust_lex_exception_test ("'z", "Unterminated character literal");
2719 rust_lex_exception_test ("b'\\x0'", "Not enough hex digits seen");
2720 rust_lex_exception_test ("b'\\u{0}'", "Unicode escape in byte literal");
2721 rust_lex_exception_test ("'\\x0'", "Not enough hex digits seen");
2722 rust_lex_exception_test ("'\\u0'", "Missing '{' in Unicode escape");
2723 rust_lex_exception_test ("'\\u{0", "Missing '}' in Unicode escape");
2724 rust_lex_exception_test ("'\\u{0000007}", "Overlong hex escape");
2725 rust_lex_exception_test ("'\\u{}", "Not enough hex digits seen");
2726 rust_lex_exception_test ("'\\Q'", "Invalid escape \\Q in literal");
2727 rust_lex_exception_test ("b'\\Q'", "Invalid escape \\Q in literal");
2729 rust_lex_int_test ("23", 23, DECIMAL_INTEGER);
2730 rust_lex_int_test ("2_344__29", 234429, INTEGER);
2731 rust_lex_int_test ("0x1f", 0x1f, INTEGER);
2732 rust_lex_int_test ("23usize", 23, INTEGER);
2733 rust_lex_int_test ("23i32", 23, INTEGER);
2734 rust_lex_int_test ("0x1_f", 0x1f, INTEGER);
2735 rust_lex_int_test ("0b1_101011__", 0x6b, INTEGER);
2736 rust_lex_int_test ("0o001177i64", 639, INTEGER);
2738 rust_lex_test_trailing_dot ();
2740 rust_lex_test_one ("23.", FLOAT);
2741 rust_lex_test_one ("23.99f32", FLOAT);
2742 rust_lex_test_one ("23e7", FLOAT);
2743 rust_lex_test_one ("23E-7", FLOAT);
2744 rust_lex_test_one ("23e+7", FLOAT);
2745 rust_lex_test_one ("23.99e+7f64", FLOAT);
2746 rust_lex_test_one ("23.82f32", FLOAT);
2748 rust_lex_stringish_test ("hibob", "hibob", IDENT);
2749 rust_lex_stringish_test ("hibob__93", "hibob__93", IDENT);
2750 rust_lex_stringish_test ("thread", "thread", IDENT);
2752 rust_lex_stringish_test ("\"string\"", "string", STRING);
2753 rust_lex_stringish_test ("\"str\\ting\"", "str\ting", STRING);
2754 rust_lex_stringish_test ("\"str\\\"ing\"", "str\"ing", STRING);
2755 rust_lex_stringish_test ("r\"str\\ing\"", "str\\ing", STRING);
2756 rust_lex_stringish_test ("r#\"str\\ting\"#", "str\\ting", STRING);
2757 rust_lex_stringish_test ("r###\"str\\\"ing\"###", "str\\\"ing", STRING);
2759 rust_lex_stringish_test ("b\"string\"", "string", BYTESTRING);
2760 rust_lex_stringish_test ("b\"\x73tring\"", "string", BYTESTRING);
2761 rust_lex_stringish_test ("b\"str\\\"ing\"", "str\"ing", BYTESTRING);
2762 rust_lex_stringish_test ("br####\"\\x73tring\"####", "\\x73tring",
2765 for (i = 0; i < ARRAY_SIZE (identifier_tokens); ++i)
2766 rust_lex_test_one (identifier_tokens[i].name, identifier_tokens[i].value);
2768 for (i = 0; i < ARRAY_SIZE (operator_tokens); ++i)
2769 rust_lex_test_one (operator_tokens[i].name, operator_tokens[i].value);
2771 rust_lex_test_completion ();
2772 rust_lex_test_push_back ();
2775 #endif /* GDB_SELF_TEST */
2778 _initialize_rust_exp (void)
2780 int code = regcomp (&number_regex, number_regex_text, REG_EXTENDED);
2781 /* If the regular expression was incorrect, it was a programming
2783 gdb_assert (code == 0);
2786 selftests::register_test ("rust-lex", rust_lex_tests);