1 /* Bison parser for Rust expressions, for GDB.
2 Copyright (C) 2016-2017 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 /* True if we're running unit tests. */
178 static int unit_testing;
180 /* Obstack for data temporarily allocated during parsing. Points to
181 the obstack in the rust_parser, or to a temporary obstack during
184 static auto_obstack *work_obstack;
186 /* An instance of this is created before parsing, and destroyed when
187 parsing is finished. */
191 rust_parser (struct parser_state *state)
192 : rust_ast (nullptr),
195 gdb_assert (current_parser == nullptr);
196 current_parser = this;
197 work_obstack = &obstack;
202 /* Clean up the globals we set. */
203 current_parser = nullptr;
204 work_obstack = nullptr;
207 /* Create a new rust_set_vector. The storage for the new vector is
208 managed by this class. */
209 rust_set_vector *new_set_vector ()
211 rust_set_vector *result = new rust_set_vector;
212 set_vectors.push_back (std::unique_ptr<rust_set_vector> (result));
216 /* Create a new rust_ops_vector. The storage for the new vector is
217 managed by this class. */
218 rust_op_vector *new_op_vector ()
220 rust_op_vector *result = new rust_op_vector;
221 op_vectors.push_back (std::unique_ptr<rust_op_vector> (result));
225 /* Return the parser's language. */
226 const struct language_defn *language () const
228 return parse_language (pstate);
231 /* Return the parser's gdbarch. */
232 struct gdbarch *arch () const
234 return parse_gdbarch (pstate);
237 /* A pointer to this is installed globally. */
238 auto_obstack obstack;
240 /* Result of parsing. Points into obstack. */
241 const struct rust_op *rust_ast;
243 /* This keeps track of the various vectors we allocate. */
244 std::vector<std::unique_ptr<rust_set_vector>> set_vectors;
245 std::vector<std::unique_ptr<rust_op_vector>> op_vectors;
247 /* The parser state gdb gave us. */
248 struct parser_state *pstate;
255 /* A typed integer constant. */
256 struct typed_val_int typed_val_int;
258 /* A typed floating point constant. */
259 struct typed_val_float typed_val_float;
261 /* An identifier or string. */
264 /* A token representing an opcode, like "==". */
265 enum exp_opcode opcode;
267 /* A list of expressions; for example, the arguments to a function
269 rust_op_vector *params;
271 /* A list of field initializers. */
272 rust_set_vector *field_inits;
274 /* A single field initializer. */
275 struct set_field one_field_init;
278 const struct rust_op *op;
280 /* A plain integer, for example used to count the number of
281 "super::" prefixes on a path. */
287 /* Rust AST operations. We build a tree of these; then lower them
288 to gdb expressions when parsing has completed. */
293 enum exp_opcode opcode;
294 /* If OPCODE is OP_TYPE, then this holds information about what type
295 is described by this node. */
296 enum type_code typecode;
297 /* Indicates whether OPCODE actually represents a compound
298 assignment. For example, if OPCODE is GTGT and this is false,
299 then this rust_op represents an ordinary ">>"; but if this is
300 true, then this rust_op represents ">>=". Unused in other
302 unsigned int compound_assignment : 1;
303 /* Only used by a field expression; if set, indicates that the field
304 name occurred at the end of the expression and is eligible for
306 unsigned int completing : 1;
307 /* Operands of expression. Which one is used and how depends on the
308 particular opcode. */
317 %token <sval> COMPLETE
318 %token <typed_val_int> INTEGER
319 %token <typed_val_int> DECIMAL_INTEGER
321 %token <sval> BYTESTRING
322 %token <typed_val_float> FLOAT
323 %token <opcode> COMPOUND_ASSIGN
325 /* Keyword tokens. */
326 %token <voidval> KW_AS
327 %token <voidval> KW_IF
328 %token <voidval> KW_TRUE
329 %token <voidval> KW_FALSE
330 %token <voidval> KW_SUPER
331 %token <voidval> KW_SELF
332 %token <voidval> KW_MUT
333 %token <voidval> KW_EXTERN
334 %token <voidval> KW_CONST
335 %token <voidval> KW_FN
336 %token <voidval> KW_SIZEOF
338 /* Operator tokens. */
339 %token <voidval> DOTDOT
340 %token <voidval> OROR
341 %token <voidval> ANDAND
342 %token <voidval> EQEQ
343 %token <voidval> NOTEQ
344 %token <voidval> LTEQ
345 %token <voidval> GTEQ
346 %token <voidval> LSH RSH
347 %token <voidval> COLONCOLON
348 %token <voidval> ARROW
351 %type <op> path_for_expr
352 %type <op> identifier_path_for_expr
353 %type <op> path_for_type
354 %type <op> identifier_path_for_type
355 %type <op> just_identifiers_for_type
357 %type <params> maybe_type_list
358 %type <params> type_list
360 %type <depth> super_path
364 %type <op> field_expr
367 %type <op> binop_expr
368 %type <op> binop_expr_expr
369 %type <op> type_cast_expr
370 %type <op> assignment_expr
371 %type <op> compound_assignment_expr
372 %type <op> paren_expr
375 %type <op> tuple_expr
377 %type <op> struct_expr
378 %type <op> array_expr
379 %type <op> range_expr
381 %type <params> expr_list
382 %type <params> maybe_expr_list
383 %type <params> paren_expr_list
385 %type <field_inits> struct_expr_list
386 %type <one_field_init> struct_expr_tail
390 %right '=' COMPOUND_ASSIGN
393 %nonassoc EQEQ NOTEQ '<' '>' LTEQ GTEQ
401 /* These could be %precedence in Bison, but that isn't a yacc
412 /* If we are completing and see a valid parse,
413 rust_ast will already have been set. */
414 if (current_parser->rust_ast == NULL)
415 current_parser->rust_ast = $1;
419 /* Note that the Rust grammar includes a method_call_expr, but we
420 handle this differently, to avoid a shift/reduce conflict with
432 | unop_expr /* Must precede call_expr because of ambiguity with sizeof. */
439 '(' expr ',' maybe_expr_list ')'
442 error (_("Tuple expressions not supported yet"));
449 struct typed_val_int val;
452 = language_lookup_primitive_type (current_parser->language (),
453 current_parser->arch (),
456 $$ = ast_literal (val);
460 /* To avoid a shift/reduce conflict with call_expr, we don't handle
461 tuple struct expressions here, but instead when examining the
464 path_for_expr '{' struct_expr_list '}'
465 { $$ = ast_struct ($1, $3); }
492 $$ = current_parser->new_set_vector ();
496 rust_set_vector *result = current_parser->new_set_vector ();
497 result->push_back ($1);
500 | IDENT ':' expr ',' struct_expr_list
512 '[' KW_MUT expr_list ']'
513 { $$ = ast_call_ish (OP_ARRAY, NULL, $3); }
515 { $$ = ast_call_ish (OP_ARRAY, NULL, $2); }
516 | '[' KW_MUT expr ';' expr ']'
517 { $$ = ast_operation (OP_RUST_ARRAY, $3, $5); }
518 | '[' expr ';' expr ']'
519 { $$ = ast_operation (OP_RUST_ARRAY, $2, $4); }
524 { $$ = ast_range ($1, NULL); }
526 { $$ = ast_range ($1, $3); }
528 { $$ = ast_range (NULL, $2); }
530 { $$ = ast_range (NULL, NULL); }
535 { $$ = ast_literal ($1); }
537 { $$ = ast_literal ($1); }
539 { $$ = ast_dliteral ($1); }
542 const struct rust_op *str = ast_string ($1);
543 struct set_field field;
544 struct typed_val_int val;
547 rust_set_vector *fields = current_parser->new_set_vector ();
549 /* Wrap the raw string in the &str struct. */
550 field.name.ptr = "data_ptr";
551 field.name.length = strlen (field.name.ptr);
552 field.init = ast_unary (UNOP_ADDR, ast_string ($1));
553 fields->push_back (field);
555 val.type = rust_type ("usize");
558 field.name.ptr = "length";
559 field.name.length = strlen (field.name.ptr);
560 field.init = ast_literal (val);
561 fields->push_back (field);
564 token.length = strlen (token.ptr);
565 $$ = ast_struct (ast_path (token, NULL), fields);
568 { $$ = ast_string ($1); }
571 struct typed_val_int val;
573 val.type = language_bool_type (current_parser->language (),
574 current_parser->arch ());
576 $$ = ast_literal (val);
580 struct typed_val_int val;
582 val.type = language_bool_type (current_parser->language (),
583 current_parser->arch ());
585 $$ = ast_literal (val);
591 { $$ = ast_structop ($1, $3.ptr, 0); }
594 $$ = ast_structop ($1, $3.ptr, 1);
595 current_parser->rust_ast = $$;
597 | expr '.' DECIMAL_INTEGER
598 { $$ = ast_structop_anonymous ($1, $3); }
603 { $$ = ast_operation (BINOP_SUBSCRIPT, $1, $3); }
608 { $$ = ast_unary (UNOP_PLUS, $2); }
610 | '-' expr %prec UNARY
611 { $$ = ast_unary (UNOP_NEG, $2); }
613 | '!' expr %prec UNARY
615 /* Note that we provide a Rust-specific evaluator
616 override for UNOP_COMPLEMENT, so it can do the
617 right thing for both bool and integral
619 $$ = ast_unary (UNOP_COMPLEMENT, $2);
622 | '*' expr %prec UNARY
623 { $$ = ast_unary (UNOP_IND, $2); }
625 | '&' expr %prec UNARY
626 { $$ = ast_unary (UNOP_ADDR, $2); }
628 | '&' KW_MUT expr %prec UNARY
629 { $$ = ast_unary (UNOP_ADDR, $3); }
630 | KW_SIZEOF '(' expr ')' %prec UNARY
631 { $$ = ast_unary (UNOP_SIZEOF, $3); }
638 | compound_assignment_expr
643 { $$ = ast_operation (BINOP_MUL, $1, $3); }
646 { $$ = ast_operation (BINOP_REPEAT, $1, $3); }
649 { $$ = ast_operation (BINOP_DIV, $1, $3); }
652 { $$ = ast_operation (BINOP_REM, $1, $3); }
655 { $$ = ast_operation (BINOP_LESS, $1, $3); }
658 { $$ = ast_operation (BINOP_GTR, $1, $3); }
661 { $$ = ast_operation (BINOP_BITWISE_AND, $1, $3); }
664 { $$ = ast_operation (BINOP_BITWISE_IOR, $1, $3); }
667 { $$ = ast_operation (BINOP_BITWISE_XOR, $1, $3); }
670 { $$ = ast_operation (BINOP_ADD, $1, $3); }
673 { $$ = ast_operation (BINOP_SUB, $1, $3); }
676 { $$ = ast_operation (BINOP_LOGICAL_OR, $1, $3); }
679 { $$ = ast_operation (BINOP_LOGICAL_AND, $1, $3); }
682 { $$ = ast_operation (BINOP_EQUAL, $1, $3); }
685 { $$ = ast_operation (BINOP_NOTEQUAL, $1, $3); }
688 { $$ = ast_operation (BINOP_LEQ, $1, $3); }
691 { $$ = ast_operation (BINOP_GEQ, $1, $3); }
694 { $$ = ast_operation (BINOP_LSH, $1, $3); }
697 { $$ = ast_operation (BINOP_RSH, $1, $3); }
702 { $$ = ast_cast ($1, $3); }
707 { $$ = ast_operation (BINOP_ASSIGN, $1, $3); }
710 compound_assignment_expr:
711 expr COMPOUND_ASSIGN expr
712 { $$ = ast_compound_assignment ($2, $1, $3); }
724 $$ = current_parser->new_op_vector ();
737 /* The result can't be NULL. */
738 $$ = current_parser->new_op_vector ();
753 { $$ = ast_call_ish (OP_FUNCALL, $1, $2); }
764 | super_path KW_SUPER COLONCOLON
772 { $$ = ast_path ($1, NULL); }
774 { $$ = ast_path (make_stoken ("self"), NULL); }
778 identifier_path_for_expr
779 | KW_SELF COLONCOLON identifier_path_for_expr
780 { $$ = super_name ($3, 0); }
781 | maybe_self_path super_path identifier_path_for_expr
782 { $$ = super_name ($3, $2); }
783 | COLONCOLON identifier_path_for_expr
784 { $$ = crate_name ($2); }
785 | KW_EXTERN identifier_path_for_expr
787 /* This is a gdb extension to make it possible to
788 refer to items in other crates. It just bypasses
789 adding the current crate to the front of the
791 $$ = ast_path (rust_concat3 ("::", $2->left.sval.ptr, NULL),
796 identifier_path_for_expr:
798 { $$ = ast_path ($1, NULL); }
799 | identifier_path_for_expr COLONCOLON IDENT
801 $$ = ast_path (rust_concat3 ($1->left.sval.ptr, "::",
805 | identifier_path_for_expr COLONCOLON '<' type_list '>'
806 { $$ = ast_path ($1->left.sval, $4); }
807 | identifier_path_for_expr COLONCOLON '<' type_list RSH
809 $$ = ast_path ($1->left.sval, $4);
810 rust_push_back ('>');
815 identifier_path_for_type
816 | KW_SELF COLONCOLON identifier_path_for_type
817 { $$ = super_name ($3, 0); }
818 | maybe_self_path super_path identifier_path_for_type
819 { $$ = super_name ($3, $2); }
820 | COLONCOLON identifier_path_for_type
821 { $$ = crate_name ($2); }
822 | KW_EXTERN identifier_path_for_type
824 /* This is a gdb extension to make it possible to
825 refer to items in other crates. It just bypasses
826 adding the current crate to the front of the
828 $$ = ast_path (rust_concat3 ("::", $2->left.sval.ptr, NULL),
833 just_identifiers_for_type:
835 { $$ = ast_path ($1, NULL); }
836 | just_identifiers_for_type COLONCOLON IDENT
838 $$ = ast_path (rust_concat3 ($1->left.sval.ptr, "::",
844 identifier_path_for_type:
845 just_identifiers_for_type
846 | just_identifiers_for_type '<' type_list '>'
847 { $$ = ast_path ($1->left.sval, $3); }
848 | just_identifiers_for_type '<' type_list RSH
850 $$ = ast_path ($1->left.sval, $3);
851 rust_push_back ('>');
857 | '[' type ';' INTEGER ']'
858 { $$ = ast_array_type ($2, $4); }
859 | '[' type ';' DECIMAL_INTEGER ']'
860 { $$ = ast_array_type ($2, $4); }
862 { $$ = ast_slice_type ($3); }
864 { $$ = ast_reference_type ($2); }
866 { $$ = ast_pointer_type ($3, 1); }
868 { $$ = ast_pointer_type ($3, 0); }
869 | KW_FN '(' maybe_type_list ')' ARROW type
870 { $$ = ast_function_type ($6, $3); }
871 | '(' maybe_type_list ')'
872 { $$ = ast_tuple_type ($2); }
885 rust_op_vector *result = current_parser->new_op_vector ();
886 result->push_back ($1);
898 /* A struct of this type is used to describe a token. */
904 enum exp_opcode opcode;
907 /* Identifier tokens. */
909 static const struct token_info identifier_tokens[] =
911 { "as", KW_AS, OP_NULL },
912 { "false", KW_FALSE, OP_NULL },
913 { "if", 0, OP_NULL },
914 { "mut", KW_MUT, OP_NULL },
915 { "const", KW_CONST, OP_NULL },
916 { "self", KW_SELF, OP_NULL },
917 { "super", KW_SUPER, OP_NULL },
918 { "true", KW_TRUE, OP_NULL },
919 { "extern", KW_EXTERN, OP_NULL },
920 { "fn", KW_FN, OP_NULL },
921 { "sizeof", KW_SIZEOF, OP_NULL },
924 /* Operator tokens, sorted longest first. */
926 static const struct token_info operator_tokens[] =
928 { ">>=", COMPOUND_ASSIGN, BINOP_RSH },
929 { "<<=", COMPOUND_ASSIGN, BINOP_LSH },
931 { "<<", LSH, OP_NULL },
932 { ">>", RSH, OP_NULL },
933 { "&&", ANDAND, OP_NULL },
934 { "||", OROR, OP_NULL },
935 { "==", EQEQ, OP_NULL },
936 { "!=", NOTEQ, OP_NULL },
937 { "<=", LTEQ, OP_NULL },
938 { ">=", GTEQ, OP_NULL },
939 { "+=", COMPOUND_ASSIGN, BINOP_ADD },
940 { "-=", COMPOUND_ASSIGN, BINOP_SUB },
941 { "*=", COMPOUND_ASSIGN, BINOP_MUL },
942 { "/=", COMPOUND_ASSIGN, BINOP_DIV },
943 { "%=", COMPOUND_ASSIGN, BINOP_REM },
944 { "&=", COMPOUND_ASSIGN, BINOP_BITWISE_AND },
945 { "|=", COMPOUND_ASSIGN, BINOP_BITWISE_IOR },
946 { "^=", COMPOUND_ASSIGN, BINOP_BITWISE_XOR },
948 { "::", COLONCOLON, OP_NULL },
949 { "..", DOTDOT, OP_NULL },
950 { "->", ARROW, OP_NULL }
953 /* Helper function to copy to the name obstack. */
956 rust_copy_name (const char *name, int len)
958 return (const char *) obstack_copy0 (work_obstack, name, len);
961 /* Helper function to make an stoken from a C string. */
964 make_stoken (const char *p)
966 struct stoken result;
969 result.length = strlen (result.ptr);
973 /* Helper function to concatenate three strings on the name
977 rust_concat3 (const char *s1, const char *s2, const char *s3)
979 return make_stoken (obconcat (work_obstack, s1, s2, s3, (char *) NULL));
982 /* Return an AST node referring to NAME, but relative to the crate's
985 static const struct rust_op *
986 crate_name (const struct rust_op *name)
988 std::string crate = rust_crate_for_block (expression_context_block);
989 struct stoken result;
991 gdb_assert (name->opcode == OP_VAR_VALUE);
994 error (_("Could not find crate for current location"));
995 result = make_stoken (obconcat (work_obstack, "::", crate.c_str (), "::",
996 name->left.sval.ptr, (char *) NULL));
998 return ast_path (result, name->right.params);
1001 /* Create an AST node referring to a "super::" qualified name. IDENT
1002 is the base name and N_SUPERS is how many "super::"s were
1003 provided. N_SUPERS can be zero. */
1005 static const struct rust_op *
1006 super_name (const struct rust_op *ident, unsigned int n_supers)
1008 const char *scope = block_scope (expression_context_block);
1011 gdb_assert (ident->opcode == OP_VAR_VALUE);
1013 if (scope[0] == '\0')
1014 error (_("Couldn't find namespace scope for self::"));
1020 std::vector<int> offsets;
1021 unsigned int current_len;
1023 current_len = cp_find_first_component (scope);
1024 while (scope[current_len] != '\0')
1026 offsets.push_back (current_len);
1027 gdb_assert (scope[current_len] == ':');
1030 current_len += cp_find_first_component (scope
1034 len = offsets.size ();
1035 if (n_supers >= len)
1036 error (_("Too many super:: uses from '%s'"), scope);
1038 offset = offsets[len - n_supers];
1041 offset = strlen (scope);
1043 obstack_grow (work_obstack, "::", 2);
1044 obstack_grow (work_obstack, scope, offset);
1045 obstack_grow (work_obstack, "::", 2);
1046 obstack_grow0 (work_obstack, ident->left.sval.ptr, ident->left.sval.length);
1048 return ast_path (make_stoken ((const char *) obstack_finish (work_obstack)),
1049 ident->right.params);
1052 /* A helper that updates innermost_block as appropriate. */
1055 update_innermost_block (struct block_symbol sym)
1057 if (symbol_read_needs_frame (sym.symbol)
1058 && (innermost_block == NULL
1059 || contained_in (sym.block, innermost_block)))
1060 innermost_block = sym.block;
1063 /* A helper to look up a Rust type, or fail. This only works for
1064 types defined by rust_language_arch_info. */
1066 static struct type *
1067 rust_type (const char *name)
1071 /* When unit testing, we don't bother checking the types, so avoid a
1072 possibly-failing lookup here. */
1076 type = language_lookup_primitive_type (current_parser->language (),
1077 current_parser->arch (),
1080 error (_("Could not find Rust type %s"), name);
1084 /* Lex a hex number with at least MIN digits and at most MAX
1088 lex_hex (int min, int max)
1090 uint32_t result = 0;
1092 /* We only want to stop at MAX if we're lexing a byte escape. */
1093 int check_max = min == max;
1095 while ((check_max ? len <= max : 1)
1096 && ((lexptr[0] >= 'a' && lexptr[0] <= 'f')
1097 || (lexptr[0] >= 'A' && lexptr[0] <= 'F')
1098 || (lexptr[0] >= '0' && lexptr[0] <= '9')))
1101 if (lexptr[0] >= 'a' && lexptr[0] <= 'f')
1102 result = result + 10 + lexptr[0] - 'a';
1103 else if (lexptr[0] >= 'A' && lexptr[0] <= 'F')
1104 result = result + 10 + lexptr[0] - 'A';
1106 result = result + lexptr[0] - '0';
1112 error (_("Not enough hex digits seen"));
1115 gdb_assert (min != max);
1116 error (_("Overlong hex escape"));
1122 /* Lex an escape. IS_BYTE is true if we're lexing a byte escape;
1123 otherwise we're lexing a character escape. */
1126 lex_escape (int is_byte)
1130 gdb_assert (lexptr[0] == '\\');
1136 result = lex_hex (2, 2);
1141 error (_("Unicode escape in byte literal"));
1143 if (lexptr[0] != '{')
1144 error (_("Missing '{' in Unicode escape"));
1146 result = lex_hex (1, 6);
1147 /* Could do range checks here. */
1148 if (lexptr[0] != '}')
1149 error (_("Missing '}' in Unicode escape"));
1183 error (_("Invalid escape \\%c in literal"), lexptr[0]);
1189 /* Lex a character constant. */
1192 lex_character (void)
1197 if (lexptr[0] == 'b')
1202 gdb_assert (lexptr[0] == '\'');
1204 /* This should handle UTF-8 here. */
1205 if (lexptr[0] == '\\')
1206 value = lex_escape (is_byte);
1209 value = lexptr[0] & 0xff;
1213 if (lexptr[0] != '\'')
1214 error (_("Unterminated character literal"));
1217 rustyylval.typed_val_int.val = value;
1218 rustyylval.typed_val_int.type = rust_type (is_byte ? "u8" : "char");
1223 /* Return the offset of the double quote if STR looks like the start
1224 of a raw string, or 0 if STR does not start a raw string. */
1227 starts_raw_string (const char *str)
1229 const char *save = str;
1234 while (str[0] == '#')
1241 /* Return true if STR looks like the end of a raw string that had N
1242 hashes at the start. */
1245 ends_raw_string (const char *str, int n)
1249 gdb_assert (str[0] == '"');
1250 for (i = 0; i < n; ++i)
1251 if (str[i + 1] != '#')
1256 /* Lex a string constant. */
1261 int is_byte = lexptr[0] == 'b';
1263 int len_in_chars = 0;
1267 raw_length = starts_raw_string (lexptr);
1268 lexptr += raw_length;
1269 gdb_assert (lexptr[0] == '"');
1278 if (lexptr[0] == '"' && ends_raw_string (lexptr, raw_length - 1))
1280 /* Exit with lexptr pointing after the final "#". */
1281 lexptr += raw_length;
1284 else if (lexptr[0] == '\0')
1285 error (_("Unexpected EOF in string"));
1287 value = lexptr[0] & 0xff;
1288 if (is_byte && value > 127)
1289 error (_("Non-ASCII value in raw byte string"));
1290 obstack_1grow (work_obstack, value);
1294 else if (lexptr[0] == '"')
1296 /* Make sure to skip the quote. */
1300 else if (lexptr[0] == '\\')
1302 value = lex_escape (is_byte);
1305 obstack_1grow (work_obstack, value);
1307 convert_between_encodings ("UTF-32", "UTF-8", (gdb_byte *) &value,
1308 sizeof (value), sizeof (value),
1309 work_obstack, translit_none);
1311 else if (lexptr[0] == '\0')
1312 error (_("Unexpected EOF in string"));
1315 value = lexptr[0] & 0xff;
1316 if (is_byte && value > 127)
1317 error (_("Non-ASCII value in byte string"));
1318 obstack_1grow (work_obstack, value);
1323 rustyylval.sval.length = obstack_object_size (work_obstack);
1324 rustyylval.sval.ptr = (const char *) obstack_finish (work_obstack);
1325 return is_byte ? BYTESTRING : STRING;
1328 /* Return true if STRING starts with whitespace followed by a digit. */
1331 space_then_number (const char *string)
1333 const char *p = string;
1335 while (p[0] == ' ' || p[0] == '\t')
1340 return *p >= '0' && *p <= '9';
1343 /* Return true if C can start an identifier. */
1346 rust_identifier_start_p (char c)
1348 return ((c >= 'a' && c <= 'z')
1349 || (c >= 'A' && c <= 'Z')
1354 /* Lex an identifier. */
1357 lex_identifier (void)
1359 const char *start = lexptr;
1360 unsigned int length;
1361 const struct token_info *token;
1363 int is_gdb_var = lexptr[0] == '$';
1365 gdb_assert (rust_identifier_start_p (lexptr[0]));
1369 /* For the time being this doesn't handle Unicode rules. Non-ASCII
1370 identifiers are gated anyway. */
1371 while ((lexptr[0] >= 'a' && lexptr[0] <= 'z')
1372 || (lexptr[0] >= 'A' && lexptr[0] <= 'Z')
1374 || (is_gdb_var && lexptr[0] == '$')
1375 || (lexptr[0] >= '0' && lexptr[0] <= '9'))
1379 length = lexptr - start;
1381 for (i = 0; i < ARRAY_SIZE (identifier_tokens); ++i)
1383 if (length == strlen (identifier_tokens[i].name)
1384 && strncmp (identifier_tokens[i].name, start, length) == 0)
1386 token = &identifier_tokens[i];
1393 if (token->value == 0)
1395 /* Leave the terminating token alone. */
1400 else if (token == NULL
1401 && (strncmp (start, "thread", length) == 0
1402 || strncmp (start, "task", length) == 0)
1403 && space_then_number (lexptr))
1405 /* "task" or "thread" followed by a number terminates the
1406 parse, per gdb rules. */
1411 if (token == NULL || (parse_completion && lexptr[0] == '\0'))
1412 rustyylval.sval = make_stoken (rust_copy_name (start, length));
1414 if (parse_completion && lexptr[0] == '\0')
1416 /* Prevent rustyylex from returning two COMPLETE tokens. */
1417 prev_lexptr = lexptr;
1422 return token->value;
1428 /* Lex an operator. */
1433 const struct token_info *token = NULL;
1436 for (i = 0; i < ARRAY_SIZE (operator_tokens); ++i)
1438 if (strncmp (operator_tokens[i].name, lexptr,
1439 strlen (operator_tokens[i].name)) == 0)
1441 lexptr += strlen (operator_tokens[i].name);
1442 token = &operator_tokens[i];
1449 rustyylval.opcode = token->opcode;
1450 return token->value;
1461 regmatch_t subexps[NUM_SUBEXPRESSIONS];
1464 int could_be_decimal = 1;
1465 int implicit_i32 = 0;
1466 const char *type_name = NULL;
1469 int type_index = -1;
1472 match = regexec (&number_regex, lexptr, ARRAY_SIZE (subexps), subexps, 0);
1473 /* Failure means the regexp is broken. */
1474 gdb_assert (match == 0);
1476 if (subexps[INT_TEXT].rm_so != -1)
1478 /* Integer part matched. */
1480 end_index = subexps[INT_TEXT].rm_eo;
1481 if (subexps[INT_TYPE].rm_so == -1)
1488 type_index = INT_TYPE;
1489 could_be_decimal = 0;
1492 else if (subexps[FLOAT_TYPE1].rm_so != -1)
1494 /* Found floating point type suffix. */
1495 end_index = subexps[FLOAT_TYPE1].rm_so;
1496 type_index = FLOAT_TYPE1;
1498 else if (subexps[FLOAT_TYPE2].rm_so != -1)
1500 /* Found floating point type suffix. */
1501 end_index = subexps[FLOAT_TYPE2].rm_so;
1502 type_index = FLOAT_TYPE2;
1506 /* Any other floating point match. */
1507 end_index = subexps[0].rm_eo;
1511 /* We need a special case if the final character is ".". In this
1512 case we might need to parse an integer. For example, "23.f()" is
1513 a request for a trait method call, not a syntax error involving
1514 the floating point number "23.". */
1515 gdb_assert (subexps[0].rm_eo > 0);
1516 if (lexptr[subexps[0].rm_eo - 1] == '.')
1518 const char *next = skip_spaces (&lexptr[subexps[0].rm_eo]);
1520 if (rust_identifier_start_p (*next) || *next == '.')
1524 end_index = subexps[0].rm_eo;
1526 could_be_decimal = 1;
1531 /* Compute the type name if we haven't already. */
1532 std::string type_name_holder;
1533 if (type_name == NULL)
1535 gdb_assert (type_index != -1);
1536 type_name_holder = std::string (lexptr + subexps[type_index].rm_so,
1537 (subexps[type_index].rm_eo
1538 - subexps[type_index].rm_so));
1539 type_name = type_name_holder.c_str ();
1542 /* Look up the type. */
1543 type = rust_type (type_name);
1545 /* Copy the text of the number and remove the "_"s. */
1547 for (i = 0; i < end_index && lexptr[i]; ++i)
1549 if (lexptr[i] == '_')
1550 could_be_decimal = 0;
1552 number.push_back (lexptr[i]);
1555 /* Advance past the match. */
1556 lexptr += subexps[0].rm_eo;
1558 /* Parse the number. */
1565 if (number[0] == '0')
1567 if (number[1] == 'x')
1569 else if (number[1] == 'o')
1571 else if (number[1] == 'b')
1576 could_be_decimal = 0;
1580 value = strtoul (number.c_str () + offset, NULL, radix);
1581 if (implicit_i32 && value >= ((uint64_t) 1) << 31)
1582 type = rust_type ("i64");
1584 rustyylval.typed_val_int.val = value;
1585 rustyylval.typed_val_int.type = type;
1589 rustyylval.typed_val_float.dval = strtod (number.c_str (), NULL);
1590 rustyylval.typed_val_float.type = type;
1593 return is_integer ? (could_be_decimal ? DECIMAL_INTEGER : INTEGER) : FLOAT;
1601 /* Skip all leading whitespace. */
1602 while (lexptr[0] == ' ' || lexptr[0] == '\t' || lexptr[0] == '\r'
1603 || lexptr[0] == '\n')
1606 /* If we hit EOF and we're completing, then return COMPLETE -- maybe
1607 we're completing an empty string at the end of a field_expr.
1608 But, we don't want to return two COMPLETE tokens in a row. */
1609 if (lexptr[0] == '\0' && lexptr == prev_lexptr)
1611 prev_lexptr = lexptr;
1612 if (lexptr[0] == '\0')
1614 if (parse_completion)
1616 rustyylval.sval = make_stoken ("");
1622 if (lexptr[0] >= '0' && lexptr[0] <= '9')
1623 return lex_number ();
1624 else if (lexptr[0] == 'b' && lexptr[1] == '\'')
1625 return lex_character ();
1626 else if (lexptr[0] == 'b' && lexptr[1] == '"')
1627 return lex_string ();
1628 else if (lexptr[0] == 'b' && starts_raw_string (lexptr + 1))
1629 return lex_string ();
1630 else if (starts_raw_string (lexptr))
1631 return lex_string ();
1632 else if (rust_identifier_start_p (lexptr[0]))
1633 return lex_identifier ();
1634 else if (lexptr[0] == '"')
1635 return lex_string ();
1636 else if (lexptr[0] == '\'')
1637 return lex_character ();
1638 else if (lexptr[0] == '}' || lexptr[0] == ']')
1640 /* Falls through to lex_operator. */
1643 else if (lexptr[0] == '(' || lexptr[0] == '{')
1645 /* Falls through to lex_operator. */
1648 else if (lexptr[0] == ',' && comma_terminates && paren_depth == 0)
1651 return lex_operator ();
1654 /* Push back a single character to be re-lexed. */
1657 rust_push_back (char c)
1659 /* Can't be called before any lexing. */
1660 gdb_assert (prev_lexptr != NULL);
1663 gdb_assert (*lexptr == c);
1668 /* Make an arbitrary operation and fill in the fields. */
1670 static const struct rust_op *
1671 ast_operation (enum exp_opcode opcode, const struct rust_op *left,
1672 const struct rust_op *right)
1674 struct rust_op *result = OBSTACK_ZALLOC (work_obstack, struct rust_op);
1676 result->opcode = opcode;
1677 result->left.op = left;
1678 result->right.op = right;
1683 /* Make a compound assignment operation. */
1685 static const struct rust_op *
1686 ast_compound_assignment (enum exp_opcode opcode, const struct rust_op *left,
1687 const struct rust_op *right)
1689 struct rust_op *result = OBSTACK_ZALLOC (work_obstack, struct rust_op);
1691 result->opcode = opcode;
1692 result->compound_assignment = 1;
1693 result->left.op = left;
1694 result->right.op = right;
1699 /* Make a typed integer literal operation. */
1701 static const struct rust_op *
1702 ast_literal (struct typed_val_int val)
1704 struct rust_op *result = OBSTACK_ZALLOC (work_obstack, struct rust_op);
1706 result->opcode = OP_LONG;
1707 result->left.typed_val_int = val;
1712 /* Make a typed floating point literal operation. */
1714 static const struct rust_op *
1715 ast_dliteral (struct typed_val_float val)
1717 struct rust_op *result = OBSTACK_ZALLOC (work_obstack, struct rust_op);
1719 result->opcode = OP_DOUBLE;
1720 result->left.typed_val_float = val;
1725 /* Make a unary operation. */
1727 static const struct rust_op *
1728 ast_unary (enum exp_opcode opcode, const struct rust_op *expr)
1730 return ast_operation (opcode, expr, NULL);
1733 /* Make a cast operation. */
1735 static const struct rust_op *
1736 ast_cast (const struct rust_op *expr, const struct rust_op *type)
1738 struct rust_op *result = OBSTACK_ZALLOC (work_obstack, struct rust_op);
1740 result->opcode = UNOP_CAST;
1741 result->left.op = expr;
1742 result->right.op = type;
1747 /* Make a call-like operation. This is nominally a function call, but
1748 when lowering we may discover that it actually represents the
1749 creation of a tuple struct. */
1751 static const struct rust_op *
1752 ast_call_ish (enum exp_opcode opcode, const struct rust_op *expr,
1753 rust_op_vector *params)
1755 struct rust_op *result = OBSTACK_ZALLOC (work_obstack, struct rust_op);
1757 result->opcode = opcode;
1758 result->left.op = expr;
1759 result->right.params = params;
1764 /* Make a structure creation operation. */
1766 static const struct rust_op *
1767 ast_struct (const struct rust_op *name, rust_set_vector *fields)
1769 struct rust_op *result = OBSTACK_ZALLOC (work_obstack, struct rust_op);
1771 result->opcode = OP_AGGREGATE;
1772 result->left.op = name;
1773 result->right.field_inits = fields;
1778 /* Make an identifier path. */
1780 static const struct rust_op *
1781 ast_path (struct stoken path, rust_op_vector *params)
1783 struct rust_op *result = OBSTACK_ZALLOC (work_obstack, struct rust_op);
1785 result->opcode = OP_VAR_VALUE;
1786 result->left.sval = path;
1787 result->right.params = params;
1792 /* Make a string constant operation. */
1794 static const struct rust_op *
1795 ast_string (struct stoken str)
1797 struct rust_op *result = OBSTACK_ZALLOC (work_obstack, struct rust_op);
1799 result->opcode = OP_STRING;
1800 result->left.sval = str;
1805 /* Make a field expression. */
1807 static const struct rust_op *
1808 ast_structop (const struct rust_op *left, const char *name, int completing)
1810 struct rust_op *result = OBSTACK_ZALLOC (work_obstack, struct rust_op);
1812 result->opcode = STRUCTOP_STRUCT;
1813 result->completing = completing;
1814 result->left.op = left;
1815 result->right.sval = make_stoken (name);
1820 /* Make an anonymous struct operation, like 'x.0'. */
1822 static const struct rust_op *
1823 ast_structop_anonymous (const struct rust_op *left,
1824 struct typed_val_int number)
1826 struct rust_op *result = OBSTACK_ZALLOC (work_obstack, struct rust_op);
1828 result->opcode = STRUCTOP_ANONYMOUS;
1829 result->left.op = left;
1830 result->right.typed_val_int = number;
1835 /* Make a range operation. */
1837 static const struct rust_op *
1838 ast_range (const struct rust_op *lhs, const struct rust_op *rhs)
1840 struct rust_op *result = OBSTACK_ZALLOC (work_obstack, struct rust_op);
1842 result->opcode = OP_RANGE;
1843 result->left.op = lhs;
1844 result->right.op = rhs;
1849 /* A helper function to make a type-related AST node. */
1851 static struct rust_op *
1852 ast_basic_type (enum type_code typecode)
1854 struct rust_op *result = OBSTACK_ZALLOC (work_obstack, struct rust_op);
1856 result->opcode = OP_TYPE;
1857 result->typecode = typecode;
1861 /* Create an AST node describing an array type. */
1863 static const struct rust_op *
1864 ast_array_type (const struct rust_op *lhs, struct typed_val_int val)
1866 struct rust_op *result = ast_basic_type (TYPE_CODE_ARRAY);
1868 result->left.op = lhs;
1869 result->right.typed_val_int = val;
1873 /* Create an AST node describing a reference type. */
1875 static const struct rust_op *
1876 ast_slice_type (const struct rust_op *type)
1878 /* Use TYPE_CODE_COMPLEX just because it is handy. */
1879 struct rust_op *result = ast_basic_type (TYPE_CODE_COMPLEX);
1881 result->left.op = type;
1885 /* Create an AST node describing a reference type. */
1887 static const struct rust_op *
1888 ast_reference_type (const struct rust_op *type)
1890 struct rust_op *result = ast_basic_type (TYPE_CODE_REF);
1892 result->left.op = type;
1896 /* Create an AST node describing a pointer type. */
1898 static const struct rust_op *
1899 ast_pointer_type (const struct rust_op *type, int is_mut)
1901 struct rust_op *result = ast_basic_type (TYPE_CODE_PTR);
1903 result->left.op = type;
1904 /* For the time being we ignore is_mut. */
1908 /* Create an AST node describing a function type. */
1910 static const struct rust_op *
1911 ast_function_type (const struct rust_op *rtype, rust_op_vector *params)
1913 struct rust_op *result = ast_basic_type (TYPE_CODE_FUNC);
1915 result->left.op = rtype;
1916 result->right.params = params;
1920 /* Create an AST node describing a tuple type. */
1922 static const struct rust_op *
1923 ast_tuple_type (rust_op_vector *params)
1925 struct rust_op *result = ast_basic_type (TYPE_CODE_STRUCT);
1927 result->left.params = params;
1931 /* A helper to appropriately munge NAME and BLOCK depending on the
1932 presence of a leading "::". */
1935 munge_name_and_block (const char **name, const struct block **block)
1937 /* If it is a global reference, skip the current block in favor of
1938 the static block. */
1939 if (strncmp (*name, "::", 2) == 0)
1942 *block = block_static_block (*block);
1946 /* Like lookup_symbol, but handles Rust namespace conventions, and
1947 doesn't require field_of_this_result. */
1949 static struct block_symbol
1950 rust_lookup_symbol (const char *name, const struct block *block,
1951 const domain_enum domain)
1953 struct block_symbol result;
1955 munge_name_and_block (&name, &block);
1957 result = lookup_symbol (name, block, domain, NULL);
1958 if (result.symbol != NULL)
1959 update_innermost_block (result);
1963 /* Look up a type, following Rust namespace conventions. */
1965 static struct type *
1966 rust_lookup_type (const char *name, const struct block *block)
1968 struct block_symbol result;
1971 munge_name_and_block (&name, &block);
1973 result = lookup_symbol (name, block, STRUCT_DOMAIN, NULL);
1974 if (result.symbol != NULL)
1976 update_innermost_block (result);
1977 return SYMBOL_TYPE (result.symbol);
1980 type = lookup_typename (current_parser->language (), current_parser->arch (),
1985 /* Last chance, try a built-in type. */
1986 return language_lookup_primitive_type (current_parser->language (),
1987 current_parser->arch (),
1991 static struct type *convert_ast_to_type (struct parser_state *state,
1992 const struct rust_op *operation);
1993 static const char *convert_name (struct parser_state *state,
1994 const struct rust_op *operation);
1996 /* Convert a vector of rust_ops representing types to a vector of
1999 static std::vector<struct type *>
2000 convert_params_to_types (struct parser_state *state, rust_op_vector *params)
2002 std::vector<struct type *> result;
2004 for (const rust_op *op : *params)
2005 result.push_back (convert_ast_to_type (state, op));
2010 /* Convert a rust_op representing a type to a struct type *. */
2012 static struct type *
2013 convert_ast_to_type (struct parser_state *state,
2014 const struct rust_op *operation)
2016 struct type *type, *result = NULL;
2018 if (operation->opcode == OP_VAR_VALUE)
2020 const char *varname = convert_name (state, operation);
2022 result = rust_lookup_type (varname, expression_context_block);
2024 error (_("No typed name '%s' in current context"), varname);
2028 gdb_assert (operation->opcode == OP_TYPE);
2030 switch (operation->typecode)
2032 case TYPE_CODE_ARRAY:
2033 type = convert_ast_to_type (state, operation->left.op);
2034 if (operation->right.typed_val_int.val < 0)
2035 error (_("Negative array length"));
2036 result = lookup_array_range_type (type, 0,
2037 operation->right.typed_val_int.val - 1);
2040 case TYPE_CODE_COMPLEX:
2042 struct type *usize = rust_type ("usize");
2044 type = convert_ast_to_type (state, operation->left.op);
2045 result = rust_slice_type ("&[*gdb*]", type, usize);
2051 /* For now we treat &x and *x identically. */
2052 type = convert_ast_to_type (state, operation->left.op);
2053 result = lookup_pointer_type (type);
2056 case TYPE_CODE_FUNC:
2058 std::vector<struct type *> args
2059 (convert_params_to_types (state, operation->right.params));
2060 struct type **argtypes = NULL;
2062 type = convert_ast_to_type (state, operation->left.op);
2064 argtypes = args.data ();
2067 = lookup_function_type_with_arguments (type, args.size (),
2069 result = lookup_pointer_type (result);
2073 case TYPE_CODE_STRUCT:
2075 std::vector<struct type *> args
2076 (convert_params_to_types (state, operation->left.params));
2081 obstack_1grow (work_obstack, '(');
2082 for (i = 0; i < args.size (); ++i)
2084 std::string type_name = type_to_string (args[i]);
2087 obstack_1grow (work_obstack, ',');
2088 obstack_grow_str (work_obstack, type_name.c_str ());
2091 obstack_grow_str0 (work_obstack, ")");
2092 name = (const char *) obstack_finish (work_obstack);
2094 /* We don't allow creating new tuple types (yet), but we do
2095 allow looking up existing tuple types. */
2096 result = rust_lookup_type (name, expression_context_block);
2098 error (_("could not find tuple type '%s'"), name);
2103 gdb_assert_not_reached ("unhandled opcode in convert_ast_to_type");
2106 gdb_assert (result != NULL);
2110 /* A helper function to turn a rust_op representing a name into a full
2111 name. This applies generic arguments as needed. The returned name
2112 is allocated on the work obstack. */
2115 convert_name (struct parser_state *state, const struct rust_op *operation)
2119 gdb_assert (operation->opcode == OP_VAR_VALUE);
2121 if (operation->right.params == NULL)
2122 return operation->left.sval.ptr;
2124 std::vector<struct type *> types
2125 (convert_params_to_types (state, operation->right.params));
2127 obstack_grow_str (work_obstack, operation->left.sval.ptr);
2128 obstack_1grow (work_obstack, '<');
2129 for (i = 0; i < types.size (); ++i)
2131 std::string type_name = type_to_string (types[i]);
2134 obstack_1grow (work_obstack, ',');
2136 obstack_grow_str (work_obstack, type_name.c_str ());
2138 obstack_grow_str0 (work_obstack, ">");
2140 return (const char *) obstack_finish (work_obstack);
2143 static void convert_ast_to_expression (struct parser_state *state,
2144 const struct rust_op *operation,
2145 const struct rust_op *top,
2146 bool want_type = false);
2148 /* A helper function that converts a vec of rust_ops to a gdb
2152 convert_params_to_expression (struct parser_state *state,
2153 rust_op_vector *params,
2154 const struct rust_op *top)
2156 for (const rust_op *elem : *params)
2157 convert_ast_to_expression (state, elem, top);
2160 /* Lower a rust_op to a gdb expression. STATE is the parser state.
2161 OPERATION is the operation to lower. TOP is a pointer to the
2162 top-most operation; it is used to handle the special case where the
2163 top-most expression is an identifier and can be optionally lowered
2164 to OP_TYPE. WANT_TYPE is a flag indicating that, if the expression
2165 is the name of a type, then emit an OP_TYPE for it (rather than
2166 erroring). If WANT_TYPE is set, then the similar TOP handling is
2170 convert_ast_to_expression (struct parser_state *state,
2171 const struct rust_op *operation,
2172 const struct rust_op *top,
2175 switch (operation->opcode)
2178 write_exp_elt_opcode (state, OP_LONG);
2179 write_exp_elt_type (state, operation->left.typed_val_int.type);
2180 write_exp_elt_longcst (state, operation->left.typed_val_int.val);
2181 write_exp_elt_opcode (state, OP_LONG);
2185 write_exp_elt_opcode (state, OP_DOUBLE);
2186 write_exp_elt_type (state, operation->left.typed_val_float.type);
2187 write_exp_elt_dblcst (state, operation->left.typed_val_float.dval);
2188 write_exp_elt_opcode (state, OP_DOUBLE);
2191 case STRUCTOP_STRUCT:
2193 convert_ast_to_expression (state, operation->left.op, top);
2195 if (operation->completing)
2196 mark_struct_expression (state);
2197 write_exp_elt_opcode (state, STRUCTOP_STRUCT);
2198 write_exp_string (state, operation->right.sval);
2199 write_exp_elt_opcode (state, STRUCTOP_STRUCT);
2203 case STRUCTOP_ANONYMOUS:
2205 convert_ast_to_expression (state, operation->left.op, top);
2207 write_exp_elt_opcode (state, STRUCTOP_ANONYMOUS);
2208 write_exp_elt_longcst (state, operation->right.typed_val_int.val);
2209 write_exp_elt_opcode (state, STRUCTOP_ANONYMOUS);
2214 convert_ast_to_expression (state, operation->left.op, top, true);
2215 write_exp_elt_opcode (state, UNOP_SIZEOF);
2220 case UNOP_COMPLEMENT:
2223 convert_ast_to_expression (state, operation->left.op, top);
2224 write_exp_elt_opcode (state, operation->opcode);
2227 case BINOP_SUBSCRIPT:
2234 case BINOP_BITWISE_AND:
2235 case BINOP_BITWISE_IOR:
2236 case BINOP_BITWISE_XOR:
2239 case BINOP_LOGICAL_OR:
2240 case BINOP_LOGICAL_AND:
2242 case BINOP_NOTEQUAL:
2249 convert_ast_to_expression (state, operation->left.op, top);
2250 convert_ast_to_expression (state, operation->right.op, top);
2251 if (operation->compound_assignment)
2253 write_exp_elt_opcode (state, BINOP_ASSIGN_MODIFY);
2254 write_exp_elt_opcode (state, operation->opcode);
2255 write_exp_elt_opcode (state, BINOP_ASSIGN_MODIFY);
2258 write_exp_elt_opcode (state, operation->opcode);
2260 if (operation->compound_assignment
2261 || operation->opcode == BINOP_ASSIGN)
2265 type = language_lookup_primitive_type (parse_language (state),
2266 parse_gdbarch (state),
2269 write_exp_elt_opcode (state, OP_LONG);
2270 write_exp_elt_type (state, type);
2271 write_exp_elt_longcst (state, 0);
2272 write_exp_elt_opcode (state, OP_LONG);
2274 write_exp_elt_opcode (state, BINOP_COMMA);
2280 struct type *type = convert_ast_to_type (state, operation->right.op);
2282 convert_ast_to_expression (state, operation->left.op, top);
2283 write_exp_elt_opcode (state, UNOP_CAST);
2284 write_exp_elt_type (state, type);
2285 write_exp_elt_opcode (state, UNOP_CAST);
2291 if (operation->left.op->opcode == OP_VAR_VALUE)
2294 const char *varname = convert_name (state, operation->left.op);
2296 type = rust_lookup_type (varname, expression_context_block);
2299 /* This is actually a tuple struct expression, not a
2301 rust_op_vector *params = operation->right.params;
2303 if (TYPE_CODE (type) != TYPE_CODE_NAMESPACE)
2305 if (!rust_tuple_struct_type_p (type))
2306 error (_("Type %s is not a tuple struct"), varname);
2308 for (int i = 0; i < params->size (); ++i)
2310 char *cell = get_print_cell ();
2312 xsnprintf (cell, PRINT_CELL_SIZE, "__%d", i);
2313 write_exp_elt_opcode (state, OP_NAME);
2314 write_exp_string (state, make_stoken (cell));
2315 write_exp_elt_opcode (state, OP_NAME);
2317 convert_ast_to_expression (state, (*params)[i], top);
2320 write_exp_elt_opcode (state, OP_AGGREGATE);
2321 write_exp_elt_type (state, type);
2322 write_exp_elt_longcst (state, 2 * params->size ());
2323 write_exp_elt_opcode (state, OP_AGGREGATE);
2328 convert_ast_to_expression (state, operation->left.op, top);
2329 convert_params_to_expression (state, operation->right.params, top);
2330 write_exp_elt_opcode (state, OP_FUNCALL);
2331 write_exp_elt_longcst (state, operation->right.params->size ());
2332 write_exp_elt_longcst (state, OP_FUNCALL);
2337 gdb_assert (operation->left.op == NULL);
2338 convert_params_to_expression (state, operation->right.params, top);
2339 write_exp_elt_opcode (state, OP_ARRAY);
2340 write_exp_elt_longcst (state, 0);
2341 write_exp_elt_longcst (state, operation->right.params->size () - 1);
2342 write_exp_elt_longcst (state, OP_ARRAY);
2347 struct block_symbol sym;
2348 const char *varname;
2350 if (operation->left.sval.ptr[0] == '$')
2352 write_dollar_variable (state, operation->left.sval);
2356 varname = convert_name (state, operation);
2357 sym = rust_lookup_symbol (varname, expression_context_block,
2359 if (sym.symbol != NULL && SYMBOL_CLASS (sym.symbol) != LOC_TYPEDEF)
2361 write_exp_elt_opcode (state, OP_VAR_VALUE);
2362 write_exp_elt_block (state, sym.block);
2363 write_exp_elt_sym (state, sym.symbol);
2364 write_exp_elt_opcode (state, OP_VAR_VALUE);
2368 struct type *type = NULL;
2370 if (sym.symbol != NULL)
2372 gdb_assert (SYMBOL_CLASS (sym.symbol) == LOC_TYPEDEF);
2373 type = SYMBOL_TYPE (sym.symbol);
2376 type = rust_lookup_type (varname, expression_context_block);
2378 error (_("No symbol '%s' in current context"), varname);
2381 && TYPE_CODE (type) == TYPE_CODE_STRUCT
2382 && TYPE_NFIELDS (type) == 0)
2384 /* A unit-like struct. */
2385 write_exp_elt_opcode (state, OP_AGGREGATE);
2386 write_exp_elt_type (state, type);
2387 write_exp_elt_longcst (state, 0);
2388 write_exp_elt_opcode (state, OP_AGGREGATE);
2390 else if (want_type || operation == top)
2392 write_exp_elt_opcode (state, OP_TYPE);
2393 write_exp_elt_type (state, type);
2394 write_exp_elt_opcode (state, OP_TYPE);
2397 error (_("Found type '%s', which can't be "
2398 "evaluated in this context"),
2408 rust_set_vector *fields = operation->right.field_inits;
2413 for (const set_field &init : *fields)
2415 if (init.name.ptr != NULL)
2417 write_exp_elt_opcode (state, OP_NAME);
2418 write_exp_string (state, init.name);
2419 write_exp_elt_opcode (state, OP_NAME);
2423 convert_ast_to_expression (state, init.init, top);
2426 if (init.name.ptr == NULL)
2428 /* This is handled differently from Ada in our
2430 write_exp_elt_opcode (state, OP_OTHERS);
2434 name = convert_name (state, operation->left.op);
2435 type = rust_lookup_type (name, expression_context_block);
2437 error (_("Could not find type '%s'"), operation->left.sval.ptr);
2439 if (TYPE_CODE (type) != TYPE_CODE_STRUCT
2440 || rust_tuple_type_p (type)
2441 || rust_tuple_struct_type_p (type))
2442 error (_("Struct expression applied to non-struct type"));
2444 write_exp_elt_opcode (state, OP_AGGREGATE);
2445 write_exp_elt_type (state, type);
2446 write_exp_elt_longcst (state, length);
2447 write_exp_elt_opcode (state, OP_AGGREGATE);
2453 write_exp_elt_opcode (state, OP_STRING);
2454 write_exp_string (state, operation->left.sval);
2455 write_exp_elt_opcode (state, OP_STRING);
2461 enum range_type kind = BOTH_BOUND_DEFAULT;
2463 if (operation->left.op != NULL)
2465 convert_ast_to_expression (state, operation->left.op, top);
2466 kind = HIGH_BOUND_DEFAULT;
2468 if (operation->right.op != NULL)
2470 convert_ast_to_expression (state, operation->right.op, top);
2471 if (kind == BOTH_BOUND_DEFAULT)
2472 kind = LOW_BOUND_DEFAULT;
2475 gdb_assert (kind == HIGH_BOUND_DEFAULT);
2476 kind = NONE_BOUND_DEFAULT;
2479 write_exp_elt_opcode (state, OP_RANGE);
2480 write_exp_elt_longcst (state, kind);
2481 write_exp_elt_opcode (state, OP_RANGE);
2486 gdb_assert_not_reached ("unhandled opcode in convert_ast_to_expression");
2492 /* The parser as exposed to gdb. */
2495 rust_parse (struct parser_state *state)
2499 /* This sets various globals and also clears them on
2501 rust_parser parser (state);
2503 result = rustyyparse ();
2505 if (!result || (parse_completion && parser.rust_ast != NULL))
2506 convert_ast_to_expression (state, parser.rust_ast, parser.rust_ast);
2511 /* The parser error handler. */
2514 rustyyerror (const char *msg)
2516 const char *where = prev_lexptr ? prev_lexptr : lexptr;
2517 error (_("%s in expression, near `%s'."), (msg ? msg : "Error"), where);
2524 /* Initialize the lexer for testing. */
2527 rust_lex_test_init (const char *input)
2534 /* A test helper that lexes a string, expecting a single token. It
2535 returns the lexer data for this token. */
2538 rust_lex_test_one (const char *input, int expected)
2543 rust_lex_test_init (input);
2545 token = rustyylex ();
2546 SELF_CHECK (token == expected);
2547 result = rustyylval;
2551 token = rustyylex ();
2552 SELF_CHECK (token == 0);
2558 /* Test that INPUT lexes as the integer VALUE. */
2561 rust_lex_int_test (const char *input, int value, int kind)
2563 RUSTSTYPE result = rust_lex_test_one (input, kind);
2564 SELF_CHECK (result.typed_val_int.val == value);
2567 /* Test that INPUT throws an exception with text ERR. */
2570 rust_lex_exception_test (const char *input, const char *err)
2574 /* The "kind" doesn't matter. */
2575 rust_lex_test_one (input, DECIMAL_INTEGER);
2578 CATCH (except, RETURN_MASK_ERROR)
2580 SELF_CHECK (strcmp (except.message, err) == 0);
2585 /* Test that INPUT lexes as the identifier, string, or byte-string
2586 VALUE. KIND holds the expected token kind. */
2589 rust_lex_stringish_test (const char *input, const char *value, int kind)
2591 RUSTSTYPE result = rust_lex_test_one (input, kind);
2592 SELF_CHECK (result.sval.length == strlen (value));
2593 SELF_CHECK (strncmp (result.sval.ptr, value, result.sval.length) == 0);
2596 /* Helper to test that a string parses as a given token sequence. */
2599 rust_lex_test_sequence (const char *input, int len, const int expected[])
2606 for (i = 0; i < len; ++i)
2608 int token = rustyylex ();
2610 SELF_CHECK (token == expected[i]);
2614 /* Tests for an integer-parsing corner case. */
2617 rust_lex_test_trailing_dot (void)
2619 const int expected1[] = { DECIMAL_INTEGER, '.', IDENT, '(', ')', 0 };
2620 const int expected2[] = { INTEGER, '.', IDENT, '(', ')', 0 };
2621 const int expected3[] = { FLOAT, EQEQ, '(', ')', 0 };
2622 const int expected4[] = { DECIMAL_INTEGER, DOTDOT, DECIMAL_INTEGER, 0 };
2624 rust_lex_test_sequence ("23.g()", ARRAY_SIZE (expected1), expected1);
2625 rust_lex_test_sequence ("23_0.g()", ARRAY_SIZE (expected2), expected2);
2626 rust_lex_test_sequence ("23.==()", ARRAY_SIZE (expected3), expected3);
2627 rust_lex_test_sequence ("23..25", ARRAY_SIZE (expected4), expected4);
2630 /* Tests of completion. */
2633 rust_lex_test_completion (void)
2635 const int expected[] = { IDENT, '.', COMPLETE, 0 };
2637 parse_completion = 1;
2639 rust_lex_test_sequence ("something.wha", ARRAY_SIZE (expected), expected);
2640 rust_lex_test_sequence ("something.", ARRAY_SIZE (expected), expected);
2642 parse_completion = 0;
2645 /* Test pushback. */
2648 rust_lex_test_push_back (void)
2652 rust_lex_test_init (">>=");
2654 token = rustyylex ();
2655 SELF_CHECK (token == COMPOUND_ASSIGN);
2656 SELF_CHECK (rustyylval.opcode == BINOP_RSH);
2658 rust_push_back ('=');
2660 token = rustyylex ();
2661 SELF_CHECK (token == '=');
2663 token = rustyylex ();
2664 SELF_CHECK (token == 0);
2667 /* Unit test the lexer. */
2670 rust_lex_tests (void)
2674 auto_obstack test_obstack;
2675 scoped_restore obstack_holder = make_scoped_restore (&work_obstack,
2680 rust_lex_test_one ("", 0);
2681 rust_lex_test_one (" \t \n \r ", 0);
2682 rust_lex_test_one ("thread 23", 0);
2683 rust_lex_test_one ("task 23", 0);
2684 rust_lex_test_one ("th 104", 0);
2685 rust_lex_test_one ("ta 97", 0);
2687 rust_lex_int_test ("'z'", 'z', INTEGER);
2688 rust_lex_int_test ("'\\xff'", 0xff, INTEGER);
2689 rust_lex_int_test ("'\\u{1016f}'", 0x1016f, INTEGER);
2690 rust_lex_int_test ("b'z'", 'z', INTEGER);
2691 rust_lex_int_test ("b'\\xfe'", 0xfe, INTEGER);
2692 rust_lex_int_test ("b'\\xFE'", 0xfe, INTEGER);
2693 rust_lex_int_test ("b'\\xfE'", 0xfe, INTEGER);
2695 /* Test all escapes in both modes. */
2696 rust_lex_int_test ("'\\n'", '\n', INTEGER);
2697 rust_lex_int_test ("'\\r'", '\r', INTEGER);
2698 rust_lex_int_test ("'\\t'", '\t', INTEGER);
2699 rust_lex_int_test ("'\\\\'", '\\', INTEGER);
2700 rust_lex_int_test ("'\\0'", '\0', INTEGER);
2701 rust_lex_int_test ("'\\''", '\'', INTEGER);
2702 rust_lex_int_test ("'\\\"'", '"', INTEGER);
2704 rust_lex_int_test ("b'\\n'", '\n', INTEGER);
2705 rust_lex_int_test ("b'\\r'", '\r', INTEGER);
2706 rust_lex_int_test ("b'\\t'", '\t', INTEGER);
2707 rust_lex_int_test ("b'\\\\'", '\\', INTEGER);
2708 rust_lex_int_test ("b'\\0'", '\0', INTEGER);
2709 rust_lex_int_test ("b'\\''", '\'', INTEGER);
2710 rust_lex_int_test ("b'\\\"'", '"', INTEGER);
2712 rust_lex_exception_test ("'z", "Unterminated character literal");
2713 rust_lex_exception_test ("b'\\x0'", "Not enough hex digits seen");
2714 rust_lex_exception_test ("b'\\u{0}'", "Unicode escape in byte literal");
2715 rust_lex_exception_test ("'\\x0'", "Not enough hex digits seen");
2716 rust_lex_exception_test ("'\\u0'", "Missing '{' in Unicode escape");
2717 rust_lex_exception_test ("'\\u{0", "Missing '}' in Unicode escape");
2718 rust_lex_exception_test ("'\\u{0000007}", "Overlong hex escape");
2719 rust_lex_exception_test ("'\\u{}", "Not enough hex digits seen");
2720 rust_lex_exception_test ("'\\Q'", "Invalid escape \\Q in literal");
2721 rust_lex_exception_test ("b'\\Q'", "Invalid escape \\Q in literal");
2723 rust_lex_int_test ("23", 23, DECIMAL_INTEGER);
2724 rust_lex_int_test ("2_344__29", 234429, INTEGER);
2725 rust_lex_int_test ("0x1f", 0x1f, INTEGER);
2726 rust_lex_int_test ("23usize", 23, INTEGER);
2727 rust_lex_int_test ("23i32", 23, INTEGER);
2728 rust_lex_int_test ("0x1_f", 0x1f, INTEGER);
2729 rust_lex_int_test ("0b1_101011__", 0x6b, INTEGER);
2730 rust_lex_int_test ("0o001177i64", 639, INTEGER);
2732 rust_lex_test_trailing_dot ();
2734 rust_lex_test_one ("23.", FLOAT);
2735 rust_lex_test_one ("23.99f32", FLOAT);
2736 rust_lex_test_one ("23e7", FLOAT);
2737 rust_lex_test_one ("23E-7", FLOAT);
2738 rust_lex_test_one ("23e+7", FLOAT);
2739 rust_lex_test_one ("23.99e+7f64", FLOAT);
2740 rust_lex_test_one ("23.82f32", FLOAT);
2742 rust_lex_stringish_test ("hibob", "hibob", IDENT);
2743 rust_lex_stringish_test ("hibob__93", "hibob__93", IDENT);
2744 rust_lex_stringish_test ("thread", "thread", IDENT);
2746 rust_lex_stringish_test ("\"string\"", "string", STRING);
2747 rust_lex_stringish_test ("\"str\\ting\"", "str\ting", STRING);
2748 rust_lex_stringish_test ("\"str\\\"ing\"", "str\"ing", STRING);
2749 rust_lex_stringish_test ("r\"str\\ing\"", "str\\ing", STRING);
2750 rust_lex_stringish_test ("r#\"str\\ting\"#", "str\\ting", STRING);
2751 rust_lex_stringish_test ("r###\"str\\\"ing\"###", "str\\\"ing", STRING);
2753 rust_lex_stringish_test ("b\"string\"", "string", BYTESTRING);
2754 rust_lex_stringish_test ("b\"\x73tring\"", "string", BYTESTRING);
2755 rust_lex_stringish_test ("b\"str\\\"ing\"", "str\"ing", BYTESTRING);
2756 rust_lex_stringish_test ("br####\"\\x73tring\"####", "\\x73tring",
2759 for (i = 0; i < ARRAY_SIZE (identifier_tokens); ++i)
2760 rust_lex_test_one (identifier_tokens[i].name, identifier_tokens[i].value);
2762 for (i = 0; i < ARRAY_SIZE (operator_tokens); ++i)
2763 rust_lex_test_one (operator_tokens[i].name, operator_tokens[i].value);
2765 rust_lex_test_completion ();
2766 rust_lex_test_push_back ();
2771 #endif /* GDB_SELF_TEST */
2774 _initialize_rust_exp (void)
2776 int code = regcomp (&number_regex, number_regex_text, REG_EXTENDED);
2777 /* If the regular expression was incorrect, it was a programming
2779 gdb_assert (code == 0);
2782 selftests::register_test ("rust-lex", rust_lex_tests);