1 /* Bison parser for Rust expressions, for GDB.
2 Copyright (C) 2016 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
42 extern initialize_file_ftype _initialize_rust_exp;
45 typedef const struct rust_op *rust_op_ptr;
46 DEF_VEC_P (rust_op_ptr);
48 /* A typed integer constant. */
56 /* A typed floating point constant. */
58 struct typed_val_float
64 /* An identifier and an expression. This is used to represent one
65 element of a struct initializer. */
70 const struct rust_op *init;
73 typedef struct set_field set_field;
75 DEF_VEC_O (set_field);
78 static int rustyylex (void);
79 static void rust_push_back (char c);
80 static const char *rust_copy_name (const char *, int);
81 static struct stoken rust_concat3 (const char *, const char *, const char *);
82 static struct stoken make_stoken (const char *);
83 static struct block_symbol rust_lookup_symbol (const char *name,
84 const struct block *block,
85 const domain_enum domain);
86 static struct type *rust_lookup_type (const char *name,
87 const struct block *block);
88 static struct type *rust_type (const char *name);
90 static const struct rust_op *crate_name (const struct rust_op *name);
91 static const struct rust_op *super_name (const struct rust_op *name,
92 unsigned int n_supers);
94 static const struct rust_op *ast_operation (enum exp_opcode opcode,
95 const struct rust_op *left,
96 const struct rust_op *right);
97 static const struct rust_op *ast_compound_assignment
98 (enum exp_opcode opcode, const struct rust_op *left,
99 const struct rust_op *rust_op);
100 static const struct rust_op *ast_literal (struct typed_val_int val);
101 static const struct rust_op *ast_dliteral (struct typed_val_float val);
102 static const struct rust_op *ast_structop (const struct rust_op *left,
105 static const struct rust_op *ast_structop_anonymous
106 (const struct rust_op *left, struct typed_val_int number);
107 static const struct rust_op *ast_unary (enum exp_opcode opcode,
108 const struct rust_op *expr);
109 static const struct rust_op *ast_cast (const struct rust_op *expr,
110 const struct rust_op *type);
111 static const struct rust_op *ast_call_ish (enum exp_opcode opcode,
112 const struct rust_op *expr,
113 VEC (rust_op_ptr) **params);
114 static const struct rust_op *ast_path (struct stoken name,
115 VEC (rust_op_ptr) **params);
116 static const struct rust_op *ast_string (struct stoken str);
117 static const struct rust_op *ast_struct (const struct rust_op *name,
118 VEC (set_field) **fields);
119 static const struct rust_op *ast_range (const struct rust_op *lhs,
120 const struct rust_op *rhs);
121 static const struct rust_op *ast_array_type (const struct rust_op *lhs,
122 struct typed_val_int val);
123 static const struct rust_op *ast_slice_type (const struct rust_op *type);
124 static const struct rust_op *ast_reference_type (const struct rust_op *type);
125 static const struct rust_op *ast_pointer_type (const struct rust_op *type,
127 static const struct rust_op *ast_function_type (const struct rust_op *result,
128 VEC (rust_op_ptr) **params);
129 static const struct rust_op *ast_tuple_type (VEC (rust_op_ptr) **params);
131 /* The state of the parser, used internally when we are parsing the
134 static struct parser_state *pstate = NULL;
136 /* A regular expression for matching Rust numbers. This is split up
137 since it is very long and this gives us a way to comment the
140 static const char *number_regex_text =
141 /* subexpression 1: allows use of alternation, otherwise uninteresting */
143 /* First comes floating point. */
144 /* Recognize number after the decimal point, with optional
145 exponent and optional type suffix.
146 subexpression 2: allows "?", otherwise uninteresting
147 subexpression 3: if present, type suffix
149 "[0-9][0-9_]*\\.[0-9][0-9_]*([eE][-+]?[0-9][0-9_]*)?(f32|f64)?"
150 #define FLOAT_TYPE1 3
152 /* Recognize exponent without decimal point, with optional type
154 subexpression 4: if present, type suffix
156 #define FLOAT_TYPE2 4
157 "[0-9][0-9_]*[eE][-+]?[0-9][0-9_]*(f32|f64)?"
159 /* "23." is a valid floating point number, but "23.e5" and
160 "23.f32" are not. So, handle the trailing-. case
164 /* Finally come integers.
165 subexpression 5: text of integer
166 subexpression 6: if present, type suffix
167 subexpression 7: allows use of alternation, otherwise uninteresting
171 "(0x[a-fA-F0-9_]+|0o[0-7_]+|0b[01_]+|[0-9][0-9_]*)"
172 "([iu](size|8|16|32|64))?"
174 /* The number of subexpressions to allocate space for, including the
175 "0th" whole match subexpression. */
176 #define NUM_SUBEXPRESSIONS 8
178 /* The compiled number-matching regex. */
180 static regex_t number_regex;
182 /* True if we're running unit tests. */
184 static int unit_testing;
186 /* Obstack for data temporarily allocated during parsing. */
188 static struct obstack work_obstack;
190 /* Result of parsing. Points into work_obstack. */
192 static const struct rust_op *rust_ast;
198 /* A typed integer constant. */
199 struct typed_val_int typed_val_int;
201 /* A typed floating point constant. */
202 struct typed_val_float typed_val_float;
204 /* An identifier or string. */
207 /* A token representing an opcode, like "==". */
208 enum exp_opcode opcode;
210 /* A list of expressions; for example, the arguments to a function
212 VEC (rust_op_ptr) **params;
214 /* A list of field initializers. */
215 VEC (set_field) **field_inits;
217 /* A single field initializer. */
218 struct set_field one_field_init;
221 const struct rust_op *op;
223 /* A plain integer, for example used to count the number of
224 "super::" prefixes on a path. */
230 /* Rust AST operations. We build a tree of these; then lower them
231 to gdb expressions when parsing has completed. */
236 enum exp_opcode opcode;
237 /* If OPCODE is OP_TYPE, then this holds information about what type
238 is described by this node. */
239 enum type_code typecode;
240 /* Indicates whether OPCODE actually represents a compound
241 assignment. For example, if OPCODE is GTGT and this is false,
242 then this rust_op represents an ordinary ">>"; but if this is
243 true, then this rust_op represents ">>=". Unused in other
245 unsigned int compound_assignment : 1;
246 /* Only used by a field expression; if set, indicates that the field
247 name occurred at the end of the expression and is eligible for
249 unsigned int completing : 1;
250 /* Operands of expression. Which one is used and how depends on the
251 particular opcode. */
260 %token <sval> COMPLETE
261 %token <typed_val_int> INTEGER
262 %token <typed_val_int> DECIMAL_INTEGER
264 %token <sval> BYTESTRING
265 %token <typed_val_float> FLOAT
266 %token <opcode> COMPOUND_ASSIGN
268 /* Keyword tokens. */
269 %token <voidval> KW_AS
270 %token <voidval> KW_IF
271 %token <voidval> KW_TRUE
272 %token <voidval> KW_FALSE
273 %token <voidval> KW_SUPER
274 %token <voidval> KW_SELF
275 %token <voidval> KW_MUT
276 %token <voidval> KW_EXTERN
277 %token <voidval> KW_CONST
278 %token <voidval> KW_FN
280 /* Operator tokens. */
281 %token <voidval> DOTDOT
282 %token <voidval> OROR
283 %token <voidval> ANDAND
284 %token <voidval> EQEQ
285 %token <voidval> NOTEQ
286 %token <voidval> LTEQ
287 %token <voidval> GTEQ
288 %token <voidval> LSH RSH
289 %token <voidval> COLONCOLON
290 %token <voidval> ARROW
293 %type <op> path_for_expr
294 %type <op> identifier_path_for_expr
295 %type <op> path_for_type
296 %type <op> identifier_path_for_type
297 %type <op> just_identifiers_for_type
299 %type <params> maybe_type_list
300 %type <params> type_list
302 %type <depth> super_path
306 %type <op> field_expr
309 %type <op> binop_expr
310 %type <op> binop_expr_expr
311 %type <op> type_cast_expr
312 %type <op> assignment_expr
313 %type <op> compound_assignment_expr
314 %type <op> paren_expr
317 %type <op> tuple_expr
319 %type <op> struct_expr
320 %type <op> array_expr
321 %type <op> range_expr
323 %type <params> expr_list
324 %type <params> maybe_expr_list
325 %type <params> paren_expr_list
327 %type <field_inits> struct_expr_list
328 %type <one_field_init> struct_expr_tail
332 %right '=' COMPOUND_ASSIGN
335 %nonassoc EQEQ NOTEQ '<' '>' LTEQ GTEQ
343 /* These could be %precedence in Bison, but that isn't a yacc
354 /* If we are completing and see a valid parse,
355 rust_ast will already have been set. */
356 if (rust_ast == NULL)
361 /* Note that the Rust grammar includes a method_call_expr, but we
362 handle this differently, to avoid a shift/reduce conflict with
381 '(' expr ',' maybe_expr_list ')'
383 VEC_safe_insert (rust_op_ptr, *$4, 0, $2);
384 error (_("Tuple expressions not supported yet"));
391 struct typed_val_int val;
394 = language_lookup_primitive_type (parse_language (pstate),
395 parse_gdbarch (pstate),
398 $$ = ast_literal (val);
402 /* To avoid a shift/reduce conflict with call_expr, we don't handle
403 tuple struct expressions here, but instead when examining the
406 path_for_expr '{' struct_expr_list '}'
407 { $$ = ast_struct ($1, $3); }
431 /* S{} is documented as valid but seems to be an unstable feature, so
432 it is left out here. */
436 VEC (set_field) **result
437 = OBSTACK_ZALLOC (&work_obstack, VEC (set_field) *);
439 make_cleanup (VEC_cleanup (set_field), result);
440 VEC_safe_push (set_field, *result, &$1);
444 | IDENT ':' expr ',' struct_expr_list
450 VEC_safe_push (set_field, *$5, &sf);
456 '[' KW_MUT expr_list ']'
457 { $$ = ast_call_ish (OP_ARRAY, NULL, $3); }
459 { $$ = ast_call_ish (OP_ARRAY, NULL, $2); }
460 | '[' KW_MUT expr ';' expr ']'
461 { $$ = ast_operation (OP_RUST_ARRAY, $3, $5); }
462 | '[' expr ';' expr ']'
463 { $$ = ast_operation (OP_RUST_ARRAY, $2, $4); }
468 { $$ = ast_range ($1, NULL); }
470 { $$ = ast_range ($1, $3); }
472 { $$ = ast_range (NULL, $2); }
474 { $$ = ast_range (NULL, NULL); }
479 { $$ = ast_literal ($1); }
481 { $$ = ast_literal ($1); }
483 { $$ = ast_dliteral ($1); }
486 const struct rust_op *str = ast_string ($1);
487 VEC (set_field) **fields;
488 struct set_field field;
489 struct typed_val_int val;
492 fields = OBSTACK_ZALLOC (&work_obstack, VEC (set_field) *);
493 make_cleanup (VEC_cleanup (set_field), fields);
495 /* Wrap the raw string in the &str struct. */
496 field.name.ptr = "data_ptr";
497 field.name.length = strlen (field.name.ptr);
498 field.init = ast_unary (UNOP_ADDR, ast_string ($1));
499 VEC_safe_push (set_field, *fields, &field);
501 val.type = rust_type ("usize");
504 field.name.ptr = "length";
505 field.name.length = strlen (field.name.ptr);
506 field.init = ast_literal (val);
507 VEC_safe_push (set_field, *fields, &field);
510 token.length = strlen (token.ptr);
511 $$ = ast_struct (ast_path (token, NULL), fields);
514 { $$ = ast_string ($1); }
517 struct typed_val_int val;
519 val.type = language_bool_type (parse_language (pstate),
520 parse_gdbarch (pstate));
522 $$ = ast_literal (val);
526 struct typed_val_int val;
528 val.type = language_bool_type (parse_language (pstate),
529 parse_gdbarch (pstate));
531 $$ = ast_literal (val);
537 { $$ = ast_structop ($1, $3.ptr, 0); }
540 $$ = ast_structop ($1, $3.ptr, 1);
543 | expr '.' DECIMAL_INTEGER
544 { $$ = ast_structop_anonymous ($1, $3); }
549 { $$ = ast_operation (BINOP_SUBSCRIPT, $1, $3); }
554 { $$ = ast_unary (UNOP_PLUS, $2); }
556 | '-' expr %prec UNARY
557 { $$ = ast_unary (UNOP_NEG, $2); }
559 | '!' expr %prec UNARY
561 /* Note that we provide a Rust-specific evaluator
562 override for UNOP_COMPLEMENT, so it can do the
563 right thing for both bool and integral
565 $$ = ast_unary (UNOP_COMPLEMENT, $2);
568 | '*' expr %prec UNARY
569 { $$ = ast_unary (UNOP_IND, $2); }
571 | '&' expr %prec UNARY
572 { $$ = ast_unary (UNOP_ADDR, $2); }
574 | '&' KW_MUT expr %prec UNARY
575 { $$ = ast_unary (UNOP_ADDR, $3); }
583 | compound_assignment_expr
588 { $$ = ast_operation (BINOP_MUL, $1, $3); }
591 { $$ = ast_operation (BINOP_REPEAT, $1, $3); }
594 { $$ = ast_operation (BINOP_DIV, $1, $3); }
597 { $$ = ast_operation (BINOP_REM, $1, $3); }
600 { $$ = ast_operation (BINOP_LESS, $1, $3); }
603 { $$ = ast_operation (BINOP_GTR, $1, $3); }
606 { $$ = ast_operation (BINOP_BITWISE_AND, $1, $3); }
609 { $$ = ast_operation (BINOP_BITWISE_IOR, $1, $3); }
612 { $$ = ast_operation (BINOP_BITWISE_XOR, $1, $3); }
615 { $$ = ast_operation (BINOP_ADD, $1, $3); }
618 { $$ = ast_operation (BINOP_SUB, $1, $3); }
621 { $$ = ast_operation (BINOP_LOGICAL_OR, $1, $3); }
624 { $$ = ast_operation (BINOP_LOGICAL_AND, $1, $3); }
627 { $$ = ast_operation (BINOP_EQUAL, $1, $3); }
630 { $$ = ast_operation (BINOP_NOTEQUAL, $1, $3); }
633 { $$ = ast_operation (BINOP_LEQ, $1, $3); }
636 { $$ = ast_operation (BINOP_GEQ, $1, $3); }
639 { $$ = ast_operation (BINOP_LSH, $1, $3); }
642 { $$ = ast_operation (BINOP_RSH, $1, $3); }
647 { $$ = ast_cast ($1, $3); }
652 { $$ = ast_operation (BINOP_ASSIGN, $1, $3); }
655 compound_assignment_expr:
656 expr COMPOUND_ASSIGN expr
657 { $$ = ast_compound_assignment ($2, $1, $3); }
669 $$ = OBSTACK_ZALLOC (&work_obstack, VEC (rust_op_ptr) *);
670 make_cleanup (VEC_cleanup (rust_op_ptr), $$);
671 VEC_safe_push (rust_op_ptr, *$$, $1);
675 VEC_safe_push (rust_op_ptr, *$1, $3);
683 /* The result can't be NULL. */
684 $$ = OBSTACK_ZALLOC (&work_obstack, VEC (rust_op_ptr) *);
685 make_cleanup (VEC_cleanup (rust_op_ptr), $$);
700 { $$ = ast_call_ish (OP_FUNCALL, $1, $2); }
711 | super_path KW_SUPER COLONCOLON
719 { $$ = ast_path ($1, NULL); }
721 { $$ = ast_path (make_stoken ("self"), NULL); }
725 identifier_path_for_expr
726 | KW_SELF COLONCOLON identifier_path_for_expr
727 { $$ = super_name ($3, 0); }
728 | maybe_self_path super_path identifier_path_for_expr
729 { $$ = super_name ($3, $2); }
730 | COLONCOLON identifier_path_for_expr
731 { $$ = crate_name ($2); }
732 | KW_EXTERN identifier_path_for_expr
734 /* This is a gdb extension to make it possible to
735 refer to items in other crates. It just bypasses
736 adding the current crate to the front of the
738 $$ = ast_path (rust_concat3 ("::", $2->left.sval.ptr, NULL),
743 identifier_path_for_expr:
745 { $$ = ast_path ($1, NULL); }
746 | identifier_path_for_expr COLONCOLON IDENT
748 $$ = ast_path (rust_concat3 ($1->left.sval.ptr, "::",
752 | identifier_path_for_expr COLONCOLON '<' type_list '>'
753 { $$ = ast_path ($1->left.sval, $4); }
754 | identifier_path_for_expr COLONCOLON '<' type_list RSH
756 $$ = ast_path ($1->left.sval, $4);
757 rust_push_back ('>');
762 identifier_path_for_type
763 | KW_SELF COLONCOLON identifier_path_for_type
764 { $$ = super_name ($3, 0); }
765 | maybe_self_path super_path identifier_path_for_type
766 { $$ = super_name ($3, $2); }
767 | COLONCOLON identifier_path_for_type
768 { $$ = crate_name ($2); }
769 | KW_EXTERN identifier_path_for_type
771 /* This is a gdb extension to make it possible to
772 refer to items in other crates. It just bypasses
773 adding the current crate to the front of the
775 $$ = ast_path (rust_concat3 ("::", $2->left.sval.ptr, NULL),
780 just_identifiers_for_type:
782 { $$ = ast_path ($1, NULL); }
783 | just_identifiers_for_type COLONCOLON IDENT
785 $$ = ast_path (rust_concat3 ($1->left.sval.ptr, "::",
791 identifier_path_for_type:
792 just_identifiers_for_type
793 | just_identifiers_for_type '<' type_list '>'
794 { $$ = ast_path ($1->left.sval, $3); }
795 | just_identifiers_for_type '<' type_list RSH
797 $$ = ast_path ($1->left.sval, $3);
798 rust_push_back ('>');
804 | '[' type ';' INTEGER ']'
805 { $$ = ast_array_type ($2, $4); }
806 | '[' type ';' DECIMAL_INTEGER ']'
807 { $$ = ast_array_type ($2, $4); }
809 { $$ = ast_slice_type ($3); }
811 { $$ = ast_reference_type ($2); }
813 { $$ = ast_pointer_type ($3, 1); }
815 { $$ = ast_pointer_type ($3, 0); }
816 | KW_FN '(' maybe_type_list ')' ARROW type
817 { $$ = ast_function_type ($6, $3); }
818 | '(' maybe_type_list ')'
819 { $$ = ast_tuple_type ($2); }
832 VEC (rust_op_ptr) **result
833 = OBSTACK_ZALLOC (&work_obstack, VEC (rust_op_ptr) *);
835 make_cleanup (VEC_cleanup (rust_op_ptr), result);
836 VEC_safe_push (rust_op_ptr, *result, $1);
841 VEC_safe_push (rust_op_ptr, *$1, $3);
848 /* A struct of this type is used to describe a token. */
854 enum exp_opcode opcode;
857 /* Identifier tokens. */
859 static const struct token_info identifier_tokens[] =
861 { "as", KW_AS, OP_NULL },
862 { "false", KW_FALSE, OP_NULL },
863 { "if", 0, OP_NULL },
864 { "mut", KW_MUT, OP_NULL },
865 { "const", KW_CONST, OP_NULL },
866 { "self", KW_SELF, OP_NULL },
867 { "super", KW_SUPER, OP_NULL },
868 { "true", KW_TRUE, OP_NULL },
869 { "extern", KW_EXTERN, OP_NULL },
870 { "fn", KW_FN, OP_NULL },
873 /* Operator tokens, sorted longest first. */
875 static const struct token_info operator_tokens[] =
877 { ">>=", COMPOUND_ASSIGN, BINOP_RSH },
878 { "<<=", COMPOUND_ASSIGN, BINOP_LSH },
880 { "<<", LSH, OP_NULL },
881 { ">>", RSH, OP_NULL },
882 { "&&", ANDAND, OP_NULL },
883 { "||", OROR, OP_NULL },
884 { "==", EQEQ, OP_NULL },
885 { "!=", NOTEQ, OP_NULL },
886 { "<=", LTEQ, OP_NULL },
887 { ">=", GTEQ, OP_NULL },
888 { "+=", COMPOUND_ASSIGN, BINOP_ADD },
889 { "-=", COMPOUND_ASSIGN, BINOP_SUB },
890 { "*=", COMPOUND_ASSIGN, BINOP_MUL },
891 { "/=", COMPOUND_ASSIGN, BINOP_DIV },
892 { "%=", COMPOUND_ASSIGN, BINOP_REM },
893 { "&=", COMPOUND_ASSIGN, BINOP_BITWISE_AND },
894 { "|=", COMPOUND_ASSIGN, BINOP_BITWISE_IOR },
895 { "^=", COMPOUND_ASSIGN, BINOP_BITWISE_XOR },
897 { "::", COLONCOLON, OP_NULL },
898 { "..", DOTDOT, OP_NULL },
899 { "->", ARROW, OP_NULL }
902 /* Helper function to copy to the name obstack. */
905 rust_copy_name (const char *name, int len)
907 return (const char *) obstack_copy0 (&work_obstack, name, len);
910 /* Helper function to make an stoken from a C string. */
913 make_stoken (const char *p)
915 struct stoken result;
918 result.length = strlen (result.ptr);
922 /* Helper function to concatenate three strings on the name
926 rust_concat3 (const char *s1, const char *s2, const char *s3)
928 return make_stoken (obconcat (&work_obstack, s1, s2, s3, (char *) NULL));
931 /* Return an AST node referring to NAME, but relative to the crate's
934 static const struct rust_op *
935 crate_name (const struct rust_op *name)
937 char *crate = rust_crate_for_block (expression_context_block);
938 struct stoken result;
940 gdb_assert (name->opcode == OP_VAR_VALUE);
943 error (_("Could not find crate for current location"));
944 result = make_stoken (obconcat (&work_obstack, "::", crate, "::",
945 name->left.sval.ptr, (char *) NULL));
948 return ast_path (result, name->right.params);
951 /* Create an AST node referring to a "super::" qualified name. IDENT
952 is the base name and N_SUPERS is how many "super::"s were
953 provided. N_SUPERS can be zero. */
955 static const struct rust_op *
956 super_name (const struct rust_op *ident, unsigned int n_supers)
958 const char *scope = block_scope (expression_context_block);
961 gdb_assert (ident->opcode == OP_VAR_VALUE);
963 if (scope[0] == '\0')
964 error (_("Couldn't find namespace scope for self::"));
970 VEC (int) *offsets = NULL;
971 unsigned int current_len;
972 struct cleanup *cleanup;
974 cleanup = make_cleanup (VEC_cleanup (int), &offsets);
975 current_len = cp_find_first_component (scope);
976 while (scope[current_len] != '\0')
978 VEC_safe_push (int, offsets, current_len);
979 gdb_assert (scope[current_len] == ':');
982 current_len += cp_find_first_component (scope
986 len = VEC_length (int, offsets);
988 error (_("Too many super:: uses from '%s'"), scope);
990 offset = VEC_index (int, offsets, len - n_supers);
992 do_cleanups (cleanup);
995 offset = strlen (scope);
997 obstack_grow (&work_obstack, "::", 2);
998 obstack_grow (&work_obstack, scope, offset);
999 obstack_grow (&work_obstack, "::", 2);
1000 obstack_grow0 (&work_obstack, ident->left.sval.ptr, ident->left.sval.length);
1002 return ast_path (make_stoken ((const char *) obstack_finish (&work_obstack)),
1003 ident->right.params);
1006 /* A helper that updates innermost_block as appropriate. */
1009 update_innermost_block (struct block_symbol sym)
1011 if (symbol_read_needs_frame (sym.symbol)
1012 && (innermost_block == NULL
1013 || contained_in (sym.block, innermost_block)))
1014 innermost_block = sym.block;
1017 /* A helper to look up a Rust type, or fail. This only works for
1018 types defined by rust_language_arch_info. */
1020 static struct type *
1021 rust_type (const char *name)
1025 /* When unit testing, we don't bother checking the types, so avoid a
1026 possibly-failing lookup here. */
1030 type = language_lookup_primitive_type (parse_language (pstate),
1031 parse_gdbarch (pstate),
1034 error (_("Could not find Rust type %s"), name);
1038 /* Lex a hex number with at least MIN digits and at most MAX
1042 lex_hex (int min, int max)
1044 uint32_t result = 0;
1046 /* We only want to stop at MAX if we're lexing a byte escape. */
1047 int check_max = min == max;
1049 while ((check_max ? len <= max : 1)
1050 && ((lexptr[0] >= 'a' && lexptr[0] <= 'f')
1051 || (lexptr[0] >= 'A' && lexptr[0] <= 'F')
1052 || (lexptr[0] >= '0' && lexptr[0] <= '9')))
1055 if (lexptr[0] >= 'a' && lexptr[0] <= 'f')
1056 result = result + 10 + lexptr[0] - 'a';
1057 else if (lexptr[0] >= 'A' && lexptr[0] <= 'F')
1058 result = result + 10 + lexptr[0] - 'A';
1060 result = result + lexptr[0] - '0';
1066 error (_("Not enough hex digits seen"));
1069 gdb_assert (min != max);
1070 error (_("Overlong hex escape"));
1076 /* Lex an escape. IS_BYTE is true if we're lexing a byte escape;
1077 otherwise we're lexing a character escape. */
1080 lex_escape (int is_byte)
1084 gdb_assert (lexptr[0] == '\\');
1090 result = lex_hex (2, 2);
1095 error (_("Unicode escape in byte literal"));
1097 if (lexptr[0] != '{')
1098 error (_("Missing '{' in Unicode escape"));
1100 result = lex_hex (1, 6);
1101 /* Could do range checks here. */
1102 if (lexptr[0] != '}')
1103 error (_("Missing '}' in Unicode escape"));
1137 error (_("Invalid escape \\%c in literal"), lexptr[0]);
1143 /* Lex a character constant. */
1146 lex_character (void)
1151 if (lexptr[0] == 'b')
1156 gdb_assert (lexptr[0] == '\'');
1158 /* This should handle UTF-8 here. */
1159 if (lexptr[0] == '\\')
1160 value = lex_escape (is_byte);
1163 value = lexptr[0] & 0xff;
1167 if (lexptr[0] != '\'')
1168 error (_("Unterminated character literal"));
1171 rustyylval.typed_val_int.val = value;
1172 rustyylval.typed_val_int.type = rust_type (is_byte ? "u8" : "char");
1177 /* Return the offset of the double quote if STR looks like the start
1178 of a raw string, or 0 if STR does not start a raw string. */
1181 starts_raw_string (const char *str)
1183 const char *save = str;
1188 while (str[0] == '#')
1195 /* Return true if STR looks like the end of a raw string that had N
1196 hashes at the start. */
1199 ends_raw_string (const char *str, int n)
1203 gdb_assert (str[0] == '"');
1204 for (i = 0; i < n; ++i)
1205 if (str[i + 1] != '#')
1210 /* Lex a string constant. */
1215 int is_byte = lexptr[0] == 'b';
1217 int len_in_chars = 0;
1221 raw_length = starts_raw_string (lexptr);
1222 lexptr += raw_length;
1223 gdb_assert (lexptr[0] == '"');
1232 if (lexptr[0] == '"' && ends_raw_string (lexptr, raw_length - 1))
1234 /* Exit with lexptr pointing after the final "#". */
1235 lexptr += raw_length;
1238 else if (lexptr[0] == '\0')
1239 error (_("Unexpected EOF in string"));
1241 value = lexptr[0] & 0xff;
1242 if (is_byte && value > 127)
1243 error (_("Non-ASCII value in raw byte string"));
1244 obstack_1grow (&work_obstack, value);
1248 else if (lexptr[0] == '"')
1250 /* Make sure to skip the quote. */
1254 else if (lexptr[0] == '\\')
1256 value = lex_escape (is_byte);
1259 obstack_1grow (&work_obstack, value);
1261 convert_between_encodings ("UTF-32", "UTF-8", (gdb_byte *) &value,
1262 sizeof (value), sizeof (value),
1263 &work_obstack, translit_none);
1265 else if (lexptr[0] == '\0')
1266 error (_("Unexpected EOF in string"));
1269 value = lexptr[0] & 0xff;
1270 if (is_byte && value > 127)
1271 error (_("Non-ASCII value in byte string"));
1272 obstack_1grow (&work_obstack, value);
1277 rustyylval.sval.length = obstack_object_size (&work_obstack);
1278 rustyylval.sval.ptr = (const char *) obstack_finish (&work_obstack);
1279 return is_byte ? BYTESTRING : STRING;
1282 /* Return true if STRING starts with whitespace followed by a digit. */
1285 space_then_number (const char *string)
1287 const char *p = string;
1289 while (p[0] == ' ' || p[0] == '\t')
1294 return *p >= '0' && *p <= '9';
1297 /* Return true if C can start an identifier. */
1300 rust_identifier_start_p (char c)
1302 return ((c >= 'a' && c <= 'z')
1303 || (c >= 'A' && c <= 'Z')
1308 /* Lex an identifier. */
1311 lex_identifier (void)
1313 const char *start = lexptr;
1314 unsigned int length;
1315 const struct token_info *token;
1317 int is_gdb_var = lexptr[0] == '$';
1319 gdb_assert (rust_identifier_start_p (lexptr[0]));
1323 /* For the time being this doesn't handle Unicode rules. Non-ASCII
1324 identifiers are gated anyway. */
1325 while ((lexptr[0] >= 'a' && lexptr[0] <= 'z')
1326 || (lexptr[0] >= 'A' && lexptr[0] <= 'Z')
1328 || (is_gdb_var && lexptr[0] == '$')
1329 || (lexptr[0] >= '0' && lexptr[0] <= '9'))
1333 length = lexptr - start;
1335 for (i = 0; i < ARRAY_SIZE (identifier_tokens); ++i)
1337 if (length == strlen (identifier_tokens[i].name)
1338 && strncmp (identifier_tokens[i].name, start, length) == 0)
1340 token = &identifier_tokens[i];
1347 if (token->value == 0)
1349 /* Leave the terminating token alone. */
1354 else if (token == NULL
1355 && (strncmp (start, "thread", length) == 0
1356 || strncmp (start, "task", length) == 0)
1357 && space_then_number (lexptr))
1359 /* "task" or "thread" followed by a number terminates the
1360 parse, per gdb rules. */
1365 if (token == NULL || (parse_completion && lexptr[0] == '\0'))
1366 rustyylval.sval = make_stoken (rust_copy_name (start, length));
1368 if (parse_completion && lexptr[0] == '\0')
1370 /* Prevent rustyylex from returning two COMPLETE tokens. */
1371 prev_lexptr = lexptr;
1376 return token->value;
1382 /* Lex an operator. */
1387 const struct token_info *token = NULL;
1390 for (i = 0; i < ARRAY_SIZE (operator_tokens); ++i)
1392 if (strncmp (operator_tokens[i].name, lexptr,
1393 strlen (operator_tokens[i].name)) == 0)
1395 lexptr += strlen (operator_tokens[i].name);
1396 token = &operator_tokens[i];
1403 rustyylval.opcode = token->opcode;
1404 return token->value;
1415 regmatch_t subexps[NUM_SUBEXPRESSIONS];
1418 int could_be_decimal = 1;
1419 int implicit_i32 = 0;
1420 char *type_name = NULL;
1423 int type_index = -1;
1426 struct cleanup *cleanup = make_cleanup (null_cleanup, NULL);
1428 match = regexec (&number_regex, lexptr, ARRAY_SIZE (subexps), subexps, 0);
1429 /* Failure means the regexp is broken. */
1430 gdb_assert (match == 0);
1432 if (subexps[INT_TEXT].rm_so != -1)
1434 /* Integer part matched. */
1436 end_index = subexps[INT_TEXT].rm_eo;
1437 if (subexps[INT_TYPE].rm_so == -1)
1444 type_index = INT_TYPE;
1445 could_be_decimal = 0;
1448 else if (subexps[FLOAT_TYPE1].rm_so != -1)
1450 /* Found floating point type suffix. */
1451 end_index = subexps[FLOAT_TYPE1].rm_so;
1452 type_index = FLOAT_TYPE1;
1454 else if (subexps[FLOAT_TYPE2].rm_so != -1)
1456 /* Found floating point type suffix. */
1457 end_index = subexps[FLOAT_TYPE2].rm_so;
1458 type_index = FLOAT_TYPE2;
1462 /* Any other floating point match. */
1463 end_index = subexps[0].rm_eo;
1467 /* We need a special case if the final character is ".". In this
1468 case we might need to parse an integer. For example, "23.f()" is
1469 a request for a trait method call, not a syntax error involving
1470 the floating point number "23.". */
1471 gdb_assert (subexps[0].rm_eo > 0);
1472 if (lexptr[subexps[0].rm_eo - 1] == '.')
1474 const char *next = skip_spaces_const (&lexptr[subexps[0].rm_eo]);
1476 if (rust_identifier_start_p (*next) || *next == '.')
1480 end_index = subexps[0].rm_eo;
1482 could_be_decimal = 1;
1487 /* Compute the type name if we haven't already. */
1488 if (type_name == NULL)
1490 gdb_assert (type_index != -1);
1491 type_name = xstrndup (lexptr + subexps[type_index].rm_so,
1492 (subexps[type_index].rm_eo
1493 - subexps[type_index].rm_so));
1494 make_cleanup (xfree, type_name);
1497 /* Look up the type. */
1498 type = rust_type (type_name);
1500 /* Copy the text of the number and remove the "_"s. */
1501 number = xstrndup (lexptr, end_index);
1502 make_cleanup (xfree, number);
1503 for (i = out = 0; number[i]; ++i)
1505 if (number[i] == '_')
1506 could_be_decimal = 0;
1508 number[out++] = number[i];
1512 /* Advance past the match. */
1513 lexptr += subexps[0].rm_eo;
1515 /* Parse the number. */
1520 if (number[0] == '0')
1522 if (number[1] == 'x')
1524 else if (number[1] == 'o')
1526 else if (number[1] == 'b')
1531 could_be_decimal = 0;
1535 value = strtoul (number, NULL, radix);
1536 if (implicit_i32 && value >= ((uint64_t) 1) << 31)
1537 type = rust_type ("i64");
1539 rustyylval.typed_val_int.val = value;
1540 rustyylval.typed_val_int.type = type;
1544 rustyylval.typed_val_float.dval = strtod (number, NULL);
1545 rustyylval.typed_val_float.type = type;
1548 do_cleanups (cleanup);
1549 return is_integer ? (could_be_decimal ? DECIMAL_INTEGER : INTEGER) : FLOAT;
1557 /* Skip all leading whitespace. */
1558 while (lexptr[0] == ' ' || lexptr[0] == '\t' || lexptr[0] == '\r'
1559 || lexptr[0] == '\n')
1562 /* If we hit EOF and we're completing, then return COMPLETE -- maybe
1563 we're completing an empty string at the end of a field_expr.
1564 But, we don't want to return two COMPLETE tokens in a row. */
1565 if (lexptr[0] == '\0' && lexptr == prev_lexptr)
1567 prev_lexptr = lexptr;
1568 if (lexptr[0] == '\0')
1570 if (parse_completion)
1572 rustyylval.sval = make_stoken ("");
1578 if (lexptr[0] >= '0' && lexptr[0] <= '9')
1579 return lex_number ();
1580 else if (lexptr[0] == 'b' && lexptr[1] == '\'')
1581 return lex_character ();
1582 else if (lexptr[0] == 'b' && lexptr[1] == '"')
1583 return lex_string ();
1584 else if (lexptr[0] == 'b' && starts_raw_string (lexptr + 1))
1585 return lex_string ();
1586 else if (starts_raw_string (lexptr))
1587 return lex_string ();
1588 else if (rust_identifier_start_p (lexptr[0]))
1589 return lex_identifier ();
1590 else if (lexptr[0] == '"')
1591 return lex_string ();
1592 else if (lexptr[0] == '\'')
1593 return lex_character ();
1594 else if (lexptr[0] == '}' || lexptr[0] == ']')
1596 /* Falls through to lex_operator. */
1599 else if (lexptr[0] == '(' || lexptr[0] == '{')
1601 /* Falls through to lex_operator. */
1604 else if (lexptr[0] == ',' && comma_terminates && paren_depth == 0)
1607 return lex_operator ();
1610 /* Push back a single character to be re-lexed. */
1613 rust_push_back (char c)
1615 /* Can't be called before any lexing. */
1616 gdb_assert (prev_lexptr != NULL);
1619 gdb_assert (*lexptr == c);
1624 /* Make an arbitrary operation and fill in the fields. */
1626 static const struct rust_op *
1627 ast_operation (enum exp_opcode opcode, const struct rust_op *left,
1628 const struct rust_op *right)
1630 struct rust_op *result = OBSTACK_ZALLOC (&work_obstack, struct rust_op);
1632 result->opcode = opcode;
1633 result->left.op = left;
1634 result->right.op = right;
1639 /* Make a compound assignment operation. */
1641 static const struct rust_op *
1642 ast_compound_assignment (enum exp_opcode opcode, const struct rust_op *left,
1643 const struct rust_op *right)
1645 struct rust_op *result = OBSTACK_ZALLOC (&work_obstack, struct rust_op);
1647 result->opcode = opcode;
1648 result->compound_assignment = 1;
1649 result->left.op = left;
1650 result->right.op = right;
1655 /* Make a typed integer literal operation. */
1657 static const struct rust_op *
1658 ast_literal (struct typed_val_int val)
1660 struct rust_op *result = OBSTACK_ZALLOC (&work_obstack, struct rust_op);
1662 result->opcode = OP_LONG;
1663 result->left.typed_val_int = val;
1668 /* Make a typed floating point literal operation. */
1670 static const struct rust_op *
1671 ast_dliteral (struct typed_val_float val)
1673 struct rust_op *result = OBSTACK_ZALLOC (&work_obstack, struct rust_op);
1675 result->opcode = OP_DOUBLE;
1676 result->left.typed_val_float = val;
1681 /* Make a unary operation. */
1683 static const struct rust_op *
1684 ast_unary (enum exp_opcode opcode, const struct rust_op *expr)
1686 return ast_operation (opcode, expr, NULL);
1689 /* Make a cast operation. */
1691 static const struct rust_op *
1692 ast_cast (const struct rust_op *expr, const struct rust_op *type)
1694 struct rust_op *result = OBSTACK_ZALLOC (&work_obstack, struct rust_op);
1696 result->opcode = UNOP_CAST;
1697 result->left.op = expr;
1698 result->right.op = type;
1703 /* Make a call-like operation. This is nominally a function call, but
1704 when lowering we may discover that it actually represents the
1705 creation of a tuple struct. */
1707 static const struct rust_op *
1708 ast_call_ish (enum exp_opcode opcode, const struct rust_op *expr,
1709 VEC (rust_op_ptr) **params)
1711 struct rust_op *result = OBSTACK_ZALLOC (&work_obstack, struct rust_op);
1713 result->opcode = opcode;
1714 result->left.op = expr;
1715 result->right.params = params;
1720 /* Make a structure creation operation. */
1722 static const struct rust_op *
1723 ast_struct (const struct rust_op *name, VEC (set_field) **fields)
1725 struct rust_op *result = OBSTACK_ZALLOC (&work_obstack, struct rust_op);
1727 result->opcode = OP_AGGREGATE;
1728 result->left.op = name;
1729 result->right.field_inits = fields;
1734 /* Make an identifier path. */
1736 static const struct rust_op *
1737 ast_path (struct stoken path, VEC (rust_op_ptr) **params)
1739 struct rust_op *result = OBSTACK_ZALLOC (&work_obstack, struct rust_op);
1741 result->opcode = OP_VAR_VALUE;
1742 result->left.sval = path;
1743 result->right.params = params;
1748 /* Make a string constant operation. */
1750 static const struct rust_op *
1751 ast_string (struct stoken str)
1753 struct rust_op *result = OBSTACK_ZALLOC (&work_obstack, struct rust_op);
1755 result->opcode = OP_STRING;
1756 result->left.sval = str;
1761 /* Make a field expression. */
1763 static const struct rust_op *
1764 ast_structop (const struct rust_op *left, const char *name, int completing)
1766 struct rust_op *result = OBSTACK_ZALLOC (&work_obstack, struct rust_op);
1768 result->opcode = STRUCTOP_STRUCT;
1769 result->completing = completing;
1770 result->left.op = left;
1771 result->right.sval = make_stoken (name);
1776 /* Make an anonymous struct operation, like 'x.0'. */
1778 static const struct rust_op *
1779 ast_structop_anonymous (const struct rust_op *left,
1780 struct typed_val_int number)
1782 struct rust_op *result = OBSTACK_ZALLOC (&work_obstack, struct rust_op);
1784 result->opcode = STRUCTOP_ANONYMOUS;
1785 result->left.op = left;
1786 result->right.typed_val_int = number;
1791 /* Make a range operation. */
1793 static const struct rust_op *
1794 ast_range (const struct rust_op *lhs, const struct rust_op *rhs)
1796 struct rust_op *result = OBSTACK_ZALLOC (&work_obstack, struct rust_op);
1798 result->opcode = OP_RANGE;
1799 result->left.op = lhs;
1800 result->right.op = rhs;
1805 /* A helper function to make a type-related AST node. */
1807 static struct rust_op *
1808 ast_basic_type (enum type_code typecode)
1810 struct rust_op *result = OBSTACK_ZALLOC (&work_obstack, struct rust_op);
1812 result->opcode = OP_TYPE;
1813 result->typecode = typecode;
1817 /* Create an AST node describing an array type. */
1819 static const struct rust_op *
1820 ast_array_type (const struct rust_op *lhs, struct typed_val_int val)
1822 struct rust_op *result = ast_basic_type (TYPE_CODE_ARRAY);
1824 result->left.op = lhs;
1825 result->right.typed_val_int = val;
1829 /* Create an AST node describing a reference type. */
1831 static const struct rust_op *
1832 ast_slice_type (const struct rust_op *type)
1834 /* Use TYPE_CODE_COMPLEX just because it is handy. */
1835 struct rust_op *result = ast_basic_type (TYPE_CODE_COMPLEX);
1837 result->left.op = type;
1841 /* Create an AST node describing a reference type. */
1843 static const struct rust_op *
1844 ast_reference_type (const struct rust_op *type)
1846 struct rust_op *result = ast_basic_type (TYPE_CODE_REF);
1848 result->left.op = type;
1852 /* Create an AST node describing a pointer type. */
1854 static const struct rust_op *
1855 ast_pointer_type (const struct rust_op *type, int is_mut)
1857 struct rust_op *result = ast_basic_type (TYPE_CODE_PTR);
1859 result->left.op = type;
1860 /* For the time being we ignore is_mut. */
1864 /* Create an AST node describing a function type. */
1866 static const struct rust_op *
1867 ast_function_type (const struct rust_op *rtype, VEC (rust_op_ptr) **params)
1869 struct rust_op *result = ast_basic_type (TYPE_CODE_FUNC);
1871 result->left.op = rtype;
1872 result->right.params = params;
1876 /* Create an AST node describing a tuple type. */
1878 static const struct rust_op *
1879 ast_tuple_type (VEC (rust_op_ptr) **params)
1881 struct rust_op *result = ast_basic_type (TYPE_CODE_STRUCT);
1883 result->left.params = params;
1887 /* A helper to appropriately munge NAME and BLOCK depending on the
1888 presence of a leading "::". */
1891 munge_name_and_block (const char **name, const struct block **block)
1893 /* If it is a global reference, skip the current block in favor of
1894 the static block. */
1895 if (strncmp (*name, "::", 2) == 0)
1898 *block = block_static_block (*block);
1902 /* Like lookup_symbol, but handles Rust namespace conventions, and
1903 doesn't require field_of_this_result. */
1905 static struct block_symbol
1906 rust_lookup_symbol (const char *name, const struct block *block,
1907 const domain_enum domain)
1909 struct block_symbol result;
1911 munge_name_and_block (&name, &block);
1913 result = lookup_symbol (name, block, domain, NULL);
1914 if (result.symbol != NULL)
1915 update_innermost_block (result);
1919 /* Look up a type, following Rust namespace conventions. */
1921 static struct type *
1922 rust_lookup_type (const char *name, const struct block *block)
1924 struct block_symbol result;
1927 munge_name_and_block (&name, &block);
1929 result = lookup_symbol (name, block, STRUCT_DOMAIN, NULL);
1930 if (result.symbol != NULL)
1932 update_innermost_block (result);
1933 return SYMBOL_TYPE (result.symbol);
1936 type = lookup_typename (parse_language (pstate), parse_gdbarch (pstate),
1941 /* Last chance, try a built-in type. */
1942 return language_lookup_primitive_type (parse_language (pstate),
1943 parse_gdbarch (pstate),
1947 static struct type *convert_ast_to_type (struct parser_state *state,
1948 const struct rust_op *operation);
1949 static const char *convert_name (struct parser_state *state,
1950 const struct rust_op *operation);
1952 /* Convert a vector of rust_ops representing types to a vector of
1955 static VEC (type_ptr) *
1956 convert_params_to_types (struct parser_state *state, VEC (rust_op_ptr) *params)
1959 const struct rust_op *op;
1960 VEC (type_ptr) *result = NULL;
1961 struct cleanup *cleanup = make_cleanup (VEC_cleanup (type_ptr), &result);
1963 for (i = 0; VEC_iterate (rust_op_ptr, params, i, op); ++i)
1964 VEC_safe_push (type_ptr, result, convert_ast_to_type (state, op));
1966 discard_cleanups (cleanup);
1970 /* Convert a rust_op representing a type to a struct type *. */
1972 static struct type *
1973 convert_ast_to_type (struct parser_state *state,
1974 const struct rust_op *operation)
1976 struct type *type, *result = NULL;
1978 if (operation->opcode == OP_VAR_VALUE)
1980 const char *varname = convert_name (state, operation);
1982 result = rust_lookup_type (varname, expression_context_block);
1984 error (_("No typed name '%s' in current context"), varname);
1988 gdb_assert (operation->opcode == OP_TYPE);
1990 switch (operation->typecode)
1992 case TYPE_CODE_ARRAY:
1993 type = convert_ast_to_type (state, operation->left.op);
1994 if (operation->right.typed_val_int.val < 0)
1995 error (_("Negative array length"));
1996 result = lookup_array_range_type (type, 0,
1997 operation->right.typed_val_int.val - 1);
2000 case TYPE_CODE_COMPLEX:
2002 struct type *usize = rust_type ("usize");
2004 type = convert_ast_to_type (state, operation->left.op);
2005 result = rust_slice_type ("&[*gdb*]", type, usize);
2011 /* For now we treat &x and *x identically. */
2012 type = convert_ast_to_type (state, operation->left.op);
2013 result = lookup_pointer_type (type);
2016 case TYPE_CODE_FUNC:
2018 VEC (type_ptr) *args
2019 = convert_params_to_types (state, *operation->right.params);
2020 struct cleanup *cleanup
2021 = make_cleanup (VEC_cleanup (type_ptr), &args);
2022 struct type **argtypes = NULL;
2024 type = convert_ast_to_type (state, operation->left.op);
2025 if (!VEC_empty (type_ptr, args))
2026 argtypes = VEC_address (type_ptr, args);
2029 = lookup_function_type_with_arguments (type,
2030 VEC_length (type_ptr, args),
2032 result = lookup_pointer_type (result);
2034 do_cleanups (cleanup);
2038 case TYPE_CODE_STRUCT:
2040 VEC (type_ptr) *args
2041 = convert_params_to_types (state, *operation->left.params);
2042 struct cleanup *cleanup
2043 = make_cleanup (VEC_cleanup (type_ptr), &args);
2048 obstack_1grow (&work_obstack, '(');
2049 for (i = 0; VEC_iterate (type_ptr, args, i, type); ++i)
2051 char *type_name = type_to_string (type);
2054 obstack_1grow (&work_obstack, ',');
2055 obstack_grow_str (&work_obstack, type_name);
2060 obstack_grow_str0 (&work_obstack, ")");
2061 name = (const char *) obstack_finish (&work_obstack);
2063 /* We don't allow creating new tuple types (yet), but we do
2064 allow looking up existing tuple types. */
2065 result = rust_lookup_type (name, expression_context_block);
2067 error (_("could not find tuple type '%s'"), name);
2069 do_cleanups (cleanup);
2074 gdb_assert_not_reached ("unhandled opcode in convert_ast_to_type");
2077 gdb_assert (result != NULL);
2081 /* A helper function to turn a rust_op representing a name into a full
2082 name. This applies generic arguments as needed. The returned name
2083 is allocated on the work obstack. */
2086 convert_name (struct parser_state *state, const struct rust_op *operation)
2088 VEC (type_ptr) *types;
2089 struct cleanup *cleanup;
2093 gdb_assert (operation->opcode == OP_VAR_VALUE);
2095 if (operation->right.params == NULL)
2096 return operation->left.sval.ptr;
2098 types = convert_params_to_types (state, *operation->right.params);
2099 cleanup = make_cleanup (VEC_cleanup (type_ptr), &types);
2101 obstack_grow_str (&work_obstack, operation->left.sval.ptr);
2102 obstack_1grow (&work_obstack, '<');
2103 for (i = 0; VEC_iterate (type_ptr, types, i, type); ++i)
2105 char *type_name = type_to_string (type);
2108 obstack_1grow (&work_obstack, ',');
2110 obstack_grow_str (&work_obstack, type_name);
2113 obstack_grow_str0 (&work_obstack, ">");
2115 do_cleanups (cleanup);
2117 return (const char *) obstack_finish (&work_obstack);
2120 static void convert_ast_to_expression (struct parser_state *state,
2121 const struct rust_op *operation,
2122 const struct rust_op *top);
2124 /* A helper function that converts a vec of rust_ops to a gdb
2128 convert_params_to_expression (struct parser_state *state,
2129 VEC (rust_op_ptr) *params,
2130 const struct rust_op *top)
2135 for (i = 0; VEC_iterate (rust_op_ptr, params, i, elem); ++i)
2136 convert_ast_to_expression (state, elem, top);
2139 /* Lower a rust_op to a gdb expression. STATE is the parser state.
2140 OPERATION is the operation to lower. TOP is a pointer to the
2141 top-most operation; it is used to handle the special case where the
2142 top-most expression is an identifier and can be optionally lowered
2146 convert_ast_to_expression (struct parser_state *state,
2147 const struct rust_op *operation,
2148 const struct rust_op *top)
2150 switch (operation->opcode)
2153 write_exp_elt_opcode (state, OP_LONG);
2154 write_exp_elt_type (state, operation->left.typed_val_int.type);
2155 write_exp_elt_longcst (state, operation->left.typed_val_int.val);
2156 write_exp_elt_opcode (state, OP_LONG);
2160 write_exp_elt_opcode (state, OP_DOUBLE);
2161 write_exp_elt_type (state, operation->left.typed_val_float.type);
2162 write_exp_elt_dblcst (state, operation->left.typed_val_float.dval);
2163 write_exp_elt_opcode (state, OP_DOUBLE);
2166 case STRUCTOP_STRUCT:
2168 convert_ast_to_expression (state, operation->left.op, top);
2170 if (operation->completing)
2171 mark_struct_expression (state);
2172 write_exp_elt_opcode (state, STRUCTOP_STRUCT);
2173 write_exp_string (state, operation->right.sval);
2174 write_exp_elt_opcode (state, STRUCTOP_STRUCT);
2178 case STRUCTOP_ANONYMOUS:
2180 convert_ast_to_expression (state, operation->left.op, top);
2182 write_exp_elt_opcode (state, STRUCTOP_ANONYMOUS);
2183 write_exp_elt_longcst (state, operation->right.typed_val_int.val);
2184 write_exp_elt_opcode (state, STRUCTOP_ANONYMOUS);
2190 case UNOP_COMPLEMENT:
2193 convert_ast_to_expression (state, operation->left.op, top);
2194 write_exp_elt_opcode (state, operation->opcode);
2197 case BINOP_SUBSCRIPT:
2204 case BINOP_BITWISE_AND:
2205 case BINOP_BITWISE_IOR:
2206 case BINOP_BITWISE_XOR:
2209 case BINOP_LOGICAL_OR:
2210 case BINOP_LOGICAL_AND:
2212 case BINOP_NOTEQUAL:
2219 convert_ast_to_expression (state, operation->left.op, top);
2220 convert_ast_to_expression (state, operation->right.op, top);
2221 if (operation->compound_assignment)
2223 write_exp_elt_opcode (state, BINOP_ASSIGN_MODIFY);
2224 write_exp_elt_opcode (state, operation->opcode);
2225 write_exp_elt_opcode (state, BINOP_ASSIGN_MODIFY);
2228 write_exp_elt_opcode (state, operation->opcode);
2230 if (operation->compound_assignment
2231 || operation->opcode == BINOP_ASSIGN)
2235 type = language_lookup_primitive_type (parse_language (state),
2236 parse_gdbarch (state),
2239 write_exp_elt_opcode (state, OP_LONG);
2240 write_exp_elt_type (state, type);
2241 write_exp_elt_longcst (state, 0);
2242 write_exp_elt_opcode (state, OP_LONG);
2244 write_exp_elt_opcode (state, BINOP_COMMA);
2250 struct type *type = convert_ast_to_type (state, operation->right.op);
2252 convert_ast_to_expression (state, operation->left.op, top);
2253 write_exp_elt_opcode (state, UNOP_CAST);
2254 write_exp_elt_type (state, type);
2255 write_exp_elt_opcode (state, UNOP_CAST);
2261 if (operation->left.op->opcode == OP_VAR_VALUE)
2264 const char *varname = convert_name (state, operation->left.op);
2266 type = rust_lookup_type (varname, expression_context_block);
2269 /* This is actually a tuple struct expression, not a
2273 VEC (rust_op_ptr) *params = *operation->right.params;
2275 if (TYPE_CODE (type) != TYPE_CODE_NAMESPACE)
2277 if (!rust_tuple_struct_type_p (type))
2278 error (_("Type %s is not a tuple struct"), varname);
2281 VEC_iterate (rust_op_ptr, params, i, elem);
2284 char *cell = get_print_cell ();
2286 xsnprintf (cell, PRINT_CELL_SIZE, "__%d", i);
2287 write_exp_elt_opcode (state, OP_NAME);
2288 write_exp_string (state, make_stoken (cell));
2289 write_exp_elt_opcode (state, OP_NAME);
2291 convert_ast_to_expression (state, elem, top);
2294 write_exp_elt_opcode (state, OP_AGGREGATE);
2295 write_exp_elt_type (state, type);
2296 write_exp_elt_longcst (state,
2297 2 * VEC_length (rust_op_ptr,
2299 write_exp_elt_opcode (state, OP_AGGREGATE);
2304 convert_ast_to_expression (state, operation->left.op, top);
2305 convert_params_to_expression (state, *operation->right.params, top);
2306 write_exp_elt_opcode (state, OP_FUNCALL);
2307 write_exp_elt_longcst (state, VEC_length (rust_op_ptr,
2308 *operation->right.params));
2309 write_exp_elt_longcst (state, OP_FUNCALL);
2314 gdb_assert (operation->left.op == NULL);
2315 convert_params_to_expression (state, *operation->right.params, top);
2316 write_exp_elt_opcode (state, OP_ARRAY);
2317 write_exp_elt_longcst (state, 0);
2318 write_exp_elt_longcst (state, VEC_length (rust_op_ptr,
2319 *operation->right.params) - 1);
2320 write_exp_elt_longcst (state, OP_ARRAY);
2325 struct block_symbol sym;
2326 const char *varname;
2328 if (operation->left.sval.ptr[0] == '$')
2330 write_dollar_variable (state, operation->left.sval);
2334 varname = convert_name (state, operation);
2335 sym = rust_lookup_symbol (varname, expression_context_block,
2337 if (sym.symbol != NULL)
2339 write_exp_elt_opcode (state, OP_VAR_VALUE);
2340 write_exp_elt_block (state, sym.block);
2341 write_exp_elt_sym (state, sym.symbol);
2342 write_exp_elt_opcode (state, OP_VAR_VALUE);
2348 type = rust_lookup_type (varname, expression_context_block);
2350 error (_("No symbol '%s' in current context"), varname);
2352 if (TYPE_CODE (type) == TYPE_CODE_STRUCT
2353 && TYPE_NFIELDS (type) == 0)
2355 /* A unit-like struct. */
2356 write_exp_elt_opcode (state, OP_AGGREGATE);
2357 write_exp_elt_type (state, type);
2358 write_exp_elt_longcst (state, 0);
2359 write_exp_elt_opcode (state, OP_AGGREGATE);
2361 else if (operation == top)
2363 write_exp_elt_opcode (state, OP_TYPE);
2364 write_exp_elt_type (state, type);
2365 write_exp_elt_opcode (state, OP_TYPE);
2376 struct set_field *init;
2377 VEC (set_field) *fields = *operation->right.field_inits;
2382 for (i = 0; VEC_iterate (set_field, fields, i, init); ++i)
2384 if (init->name.ptr != NULL)
2386 write_exp_elt_opcode (state, OP_NAME);
2387 write_exp_string (state, init->name);
2388 write_exp_elt_opcode (state, OP_NAME);
2392 convert_ast_to_expression (state, init->init, top);
2395 if (init->name.ptr == NULL)
2397 /* This is handled differently from Ada in our
2399 write_exp_elt_opcode (state, OP_OTHERS);
2403 name = convert_name (state, operation->left.op);
2404 type = rust_lookup_type (name, expression_context_block);
2406 error (_("Could not find type '%s'"), operation->left.sval.ptr);
2408 if (TYPE_CODE (type) != TYPE_CODE_STRUCT
2409 || rust_tuple_type_p (type)
2410 || rust_tuple_struct_type_p (type))
2411 error (_("Struct expression applied to non-struct type"));
2413 write_exp_elt_opcode (state, OP_AGGREGATE);
2414 write_exp_elt_type (state, type);
2415 write_exp_elt_longcst (state, length);
2416 write_exp_elt_opcode (state, OP_AGGREGATE);
2422 write_exp_elt_opcode (state, OP_STRING);
2423 write_exp_string (state, operation->left.sval);
2424 write_exp_elt_opcode (state, OP_STRING);
2430 enum range_type kind = BOTH_BOUND_DEFAULT;
2432 if (operation->left.op != NULL)
2434 convert_ast_to_expression (state, operation->left.op, top);
2435 kind = HIGH_BOUND_DEFAULT;
2437 if (operation->right.op != NULL)
2439 convert_ast_to_expression (state, operation->right.op, top);
2440 if (kind == BOTH_BOUND_DEFAULT)
2441 kind = LOW_BOUND_DEFAULT;
2444 gdb_assert (kind == HIGH_BOUND_DEFAULT);
2445 kind = NONE_BOUND_DEFAULT;
2448 write_exp_elt_opcode (state, OP_RANGE);
2449 write_exp_elt_longcst (state, kind);
2450 write_exp_elt_opcode (state, OP_RANGE);
2455 gdb_assert_not_reached ("unhandled opcode in convert_ast_to_expression");
2461 /* The parser as exposed to gdb. */
2464 rust_parse (struct parser_state *state)
2467 struct cleanup *cleanup;
2469 obstack_init (&work_obstack);
2470 cleanup = make_cleanup_obstack_free (&work_obstack);
2474 result = rustyyparse ();
2476 if (!result || (parse_completion && rust_ast != NULL))
2478 const struct rust_op *ast = rust_ast;
2481 gdb_assert (ast != NULL);
2482 convert_ast_to_expression (state, ast, ast);
2485 do_cleanups (cleanup);
2489 /* The parser error handler. */
2492 rustyyerror (char *msg)
2494 const char *where = prev_lexptr ? prev_lexptr : lexptr;
2495 error (_("%s in expression, near `%s'."), (msg ? msg : "Error"), where);
2502 /* Initialize the lexer for testing. */
2505 rust_lex_test_init (const char *input)
2512 /* A test helper that lexes a string, expecting a single token. It
2513 returns the lexer data for this token. */
2516 rust_lex_test_one (const char *input, int expected)
2521 rust_lex_test_init (input);
2523 token = rustyylex ();
2524 SELF_CHECK (token == expected);
2525 result = rustyylval;
2529 token = rustyylex ();
2530 SELF_CHECK (token == 0);
2536 /* Test that INPUT lexes as the integer VALUE. */
2539 rust_lex_int_test (const char *input, int value, int kind)
2541 RUSTSTYPE result = rust_lex_test_one (input, kind);
2542 SELF_CHECK (result.typed_val_int.val == value);
2545 /* Test that INPUT throws an exception with text ERR. */
2548 rust_lex_exception_test (const char *input, const char *err)
2552 /* The "kind" doesn't matter. */
2553 rust_lex_test_one (input, DECIMAL_INTEGER);
2556 CATCH (except, RETURN_MASK_ERROR)
2558 SELF_CHECK (strcmp (except.message, err) == 0);
2563 /* Test that INPUT lexes as the identifier, string, or byte-string
2564 VALUE. KIND holds the expected token kind. */
2567 rust_lex_stringish_test (const char *input, const char *value, int kind)
2569 RUSTSTYPE result = rust_lex_test_one (input, kind);
2570 SELF_CHECK (result.sval.length == strlen (value));
2571 SELF_CHECK (strncmp (result.sval.ptr, value, result.sval.length) == 0);
2574 /* Helper to test that a string parses as a given token sequence. */
2577 rust_lex_test_sequence (const char *input, int len, const int expected[])
2584 for (i = 0; i < len; ++i)
2586 int token = rustyylex ();
2588 SELF_CHECK (token == expected[i]);
2592 /* Tests for an integer-parsing corner case. */
2595 rust_lex_test_trailing_dot (void)
2597 const int expected1[] = { DECIMAL_INTEGER, '.', IDENT, '(', ')', 0 };
2598 const int expected2[] = { INTEGER, '.', IDENT, '(', ')', 0 };
2599 const int expected3[] = { FLOAT, EQEQ, '(', ')', 0 };
2600 const int expected4[] = { DECIMAL_INTEGER, DOTDOT, DECIMAL_INTEGER, 0 };
2602 rust_lex_test_sequence ("23.g()", ARRAY_SIZE (expected1), expected1);
2603 rust_lex_test_sequence ("23_0.g()", ARRAY_SIZE (expected2), expected2);
2604 rust_lex_test_sequence ("23.==()", ARRAY_SIZE (expected3), expected3);
2605 rust_lex_test_sequence ("23..25", ARRAY_SIZE (expected4), expected4);
2608 /* Tests of completion. */
2611 rust_lex_test_completion (void)
2613 const int expected[] = { IDENT, '.', COMPLETE, 0 };
2615 parse_completion = 1;
2617 rust_lex_test_sequence ("something.wha", ARRAY_SIZE (expected), expected);
2618 rust_lex_test_sequence ("something.", ARRAY_SIZE (expected), expected);
2620 parse_completion = 0;
2623 /* Test pushback. */
2626 rust_lex_test_push_back (void)
2630 rust_lex_test_init (">>=");
2632 token = rustyylex ();
2633 SELF_CHECK (token == COMPOUND_ASSIGN);
2634 SELF_CHECK (rustyylval.opcode == BINOP_RSH);
2636 rust_push_back ('=');
2638 token = rustyylex ();
2639 SELF_CHECK (token == '=');
2641 token = rustyylex ();
2642 SELF_CHECK (token == 0);
2645 /* Unit test the lexer. */
2648 rust_lex_tests (void)
2652 obstack_init (&work_obstack);
2655 rust_lex_test_one ("", 0);
2656 rust_lex_test_one (" \t \n \r ", 0);
2657 rust_lex_test_one ("thread 23", 0);
2658 rust_lex_test_one ("task 23", 0);
2659 rust_lex_test_one ("th 104", 0);
2660 rust_lex_test_one ("ta 97", 0);
2662 rust_lex_int_test ("'z'", 'z', INTEGER);
2663 rust_lex_int_test ("'\\xff'", 0xff, INTEGER);
2664 rust_lex_int_test ("'\\u{1016f}'", 0x1016f, INTEGER);
2665 rust_lex_int_test ("b'z'", 'z', INTEGER);
2666 rust_lex_int_test ("b'\\xfe'", 0xfe, INTEGER);
2667 rust_lex_int_test ("b'\\xFE'", 0xfe, INTEGER);
2668 rust_lex_int_test ("b'\\xfE'", 0xfe, INTEGER);
2670 /* Test all escapes in both modes. */
2671 rust_lex_int_test ("'\\n'", '\n', INTEGER);
2672 rust_lex_int_test ("'\\r'", '\r', INTEGER);
2673 rust_lex_int_test ("'\\t'", '\t', INTEGER);
2674 rust_lex_int_test ("'\\\\'", '\\', INTEGER);
2675 rust_lex_int_test ("'\\0'", '\0', INTEGER);
2676 rust_lex_int_test ("'\\''", '\'', INTEGER);
2677 rust_lex_int_test ("'\\\"'", '"', INTEGER);
2679 rust_lex_int_test ("b'\\n'", '\n', INTEGER);
2680 rust_lex_int_test ("b'\\r'", '\r', INTEGER);
2681 rust_lex_int_test ("b'\\t'", '\t', INTEGER);
2682 rust_lex_int_test ("b'\\\\'", '\\', INTEGER);
2683 rust_lex_int_test ("b'\\0'", '\0', INTEGER);
2684 rust_lex_int_test ("b'\\''", '\'', INTEGER);
2685 rust_lex_int_test ("b'\\\"'", '"', INTEGER);
2687 rust_lex_exception_test ("'z", "Unterminated character literal");
2688 rust_lex_exception_test ("b'\\x0'", "Not enough hex digits seen");
2689 rust_lex_exception_test ("b'\\u{0}'", "Unicode escape in byte literal");
2690 rust_lex_exception_test ("'\\x0'", "Not enough hex digits seen");
2691 rust_lex_exception_test ("'\\u0'", "Missing '{' in Unicode escape");
2692 rust_lex_exception_test ("'\\u{0", "Missing '}' in Unicode escape");
2693 rust_lex_exception_test ("'\\u{0000007}", "Overlong hex escape");
2694 rust_lex_exception_test ("'\\u{}", "Not enough hex digits seen");
2695 rust_lex_exception_test ("'\\Q'", "Invalid escape \\Q in literal");
2696 rust_lex_exception_test ("b'\\Q'", "Invalid escape \\Q in literal");
2698 rust_lex_int_test ("23", 23, DECIMAL_INTEGER);
2699 rust_lex_int_test ("2_344__29", 234429, INTEGER);
2700 rust_lex_int_test ("0x1f", 0x1f, INTEGER);
2701 rust_lex_int_test ("23usize", 23, INTEGER);
2702 rust_lex_int_test ("23i32", 23, INTEGER);
2703 rust_lex_int_test ("0x1_f", 0x1f, INTEGER);
2704 rust_lex_int_test ("0b1_101011__", 0x6b, INTEGER);
2705 rust_lex_int_test ("0o001177i64", 639, INTEGER);
2707 rust_lex_test_trailing_dot ();
2709 rust_lex_test_one ("23.", FLOAT);
2710 rust_lex_test_one ("23.99f32", FLOAT);
2711 rust_lex_test_one ("23e7", FLOAT);
2712 rust_lex_test_one ("23E-7", FLOAT);
2713 rust_lex_test_one ("23e+7", FLOAT);
2714 rust_lex_test_one ("23.99e+7f64", FLOAT);
2715 rust_lex_test_one ("23.82f32", FLOAT);
2717 rust_lex_stringish_test ("hibob", "hibob", IDENT);
2718 rust_lex_stringish_test ("hibob__93", "hibob__93", IDENT);
2719 rust_lex_stringish_test ("thread", "thread", IDENT);
2721 rust_lex_stringish_test ("\"string\"", "string", STRING);
2722 rust_lex_stringish_test ("\"str\\ting\"", "str\ting", STRING);
2723 rust_lex_stringish_test ("\"str\\\"ing\"", "str\"ing", STRING);
2724 rust_lex_stringish_test ("r\"str\\ing\"", "str\\ing", STRING);
2725 rust_lex_stringish_test ("r#\"str\\ting\"#", "str\\ting", STRING);
2726 rust_lex_stringish_test ("r###\"str\\\"ing\"###", "str\\\"ing", STRING);
2728 rust_lex_stringish_test ("b\"string\"", "string", BYTESTRING);
2729 rust_lex_stringish_test ("b\"\x73tring\"", "string", BYTESTRING);
2730 rust_lex_stringish_test ("b\"str\\\"ing\"", "str\"ing", BYTESTRING);
2731 rust_lex_stringish_test ("br####\"\\x73tring\"####", "\\x73tring",
2734 for (i = 0; i < ARRAY_SIZE (identifier_tokens); ++i)
2735 rust_lex_test_one (identifier_tokens[i].name, identifier_tokens[i].value);
2737 for (i = 0; i < ARRAY_SIZE (operator_tokens); ++i)
2738 rust_lex_test_one (operator_tokens[i].name, operator_tokens[i].value);
2740 rust_lex_test_completion ();
2741 rust_lex_test_push_back ();
2743 obstack_free (&work_obstack, NULL);
2747 #endif /* GDB_SELF_TEST */
2750 _initialize_rust_exp (void)
2752 int code = regcomp (&number_regex, number_regex_text, REG_EXTENDED);
2753 /* If the regular expression was incorrect, it was a programming
2755 gdb_assert (code == 0);
2758 register_self_test (rust_lex_tests);