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, previous_len;
972 struct cleanup *cleanup;
974 cleanup = make_cleanup (VEC_cleanup (int), &offsets);
975 current_len = cp_find_first_component (scope);
977 while (scope[current_len] != '\0')
979 VEC_safe_push (int, offsets, current_len);
980 gdb_assert (scope[current_len] == ':');
981 previous_len = current_len;
984 current_len += cp_find_first_component (scope
988 len = VEC_length (int, offsets);
990 error (_("Too many super:: uses from '%s'"), scope);
992 offset = VEC_index (int, offsets, len - n_supers);
994 do_cleanups (cleanup);
997 offset = strlen (scope);
999 obstack_grow (&work_obstack, "::", 2);
1000 obstack_grow (&work_obstack, scope, offset);
1001 obstack_grow (&work_obstack, "::", 2);
1002 obstack_grow0 (&work_obstack, ident->left.sval.ptr, ident->left.sval.length);
1004 return ast_path (make_stoken ((const char *) obstack_finish (&work_obstack)),
1005 ident->right.params);
1008 /* A helper that updates innermost_block as appropriate. */
1011 update_innermost_block (struct block_symbol sym)
1013 if (symbol_read_needs_frame (sym.symbol)
1014 && (innermost_block == NULL
1015 || contained_in (sym.block, innermost_block)))
1016 innermost_block = sym.block;
1019 /* A helper to look up a Rust type, or fail. This only works for
1020 types defined by rust_language_arch_info. */
1022 static struct type *
1023 rust_type (const char *name)
1027 /* When unit testing, we don't bother checking the types, so avoid a
1028 possibly-failing lookup here. */
1032 type = language_lookup_primitive_type (parse_language (pstate),
1033 parse_gdbarch (pstate),
1036 error (_("Could not find Rust type %s"), name);
1040 /* Lex a hex number with at least MIN digits and at most MAX
1044 lex_hex (int min, int max)
1046 uint32_t result = 0;
1048 /* We only want to stop at MAX if we're lexing a byte escape. */
1049 int check_max = min == max;
1051 while ((check_max ? len <= max : 1)
1052 && ((lexptr[0] >= 'a' && lexptr[0] <= 'f')
1053 || (lexptr[0] >= 'A' && lexptr[0] <= 'F')
1054 || (lexptr[0] >= '0' && lexptr[0] <= '9')))
1057 if (lexptr[0] >= 'a' && lexptr[0] <= 'f')
1058 result = result + 10 + lexptr[0] - 'a';
1059 else if (lexptr[0] >= 'A' && lexptr[0] <= 'F')
1060 result = result + 10 + lexptr[0] - 'A';
1062 result = result + lexptr[0] - '0';
1068 error (_("Not enough hex digits seen"));
1071 gdb_assert (min != max);
1072 error (_("Overlong hex escape"));
1078 /* Lex an escape. IS_BYTE is true if we're lexing a byte escape;
1079 otherwise we're lexing a character escape. */
1082 lex_escape (int is_byte)
1086 gdb_assert (lexptr[0] == '\\');
1092 result = lex_hex (2, 2);
1097 error (_("Unicode escape in byte literal"));
1099 if (lexptr[0] != '{')
1100 error (_("Missing '{' in Unicode escape"));
1102 result = lex_hex (1, 6);
1103 /* Could do range checks here. */
1104 if (lexptr[0] != '}')
1105 error (_("Missing '}' in Unicode escape"));
1139 error (_("Invalid escape \\%c in literal"), lexptr[0]);
1145 /* Lex a character constant. */
1148 lex_character (void)
1153 if (lexptr[0] == 'b')
1158 gdb_assert (lexptr[0] == '\'');
1160 /* This should handle UTF-8 here. */
1161 if (lexptr[0] == '\\')
1162 value = lex_escape (is_byte);
1165 value = lexptr[0] & 0xff;
1169 if (lexptr[0] != '\'')
1170 error (_("Unterminated character literal"));
1173 rustyylval.typed_val_int.val = value;
1174 rustyylval.typed_val_int.type = rust_type (is_byte ? "u8" : "char");
1179 /* Return the offset of the double quote if STR looks like the start
1180 of a raw string, or 0 if STR does not start a raw string. */
1183 starts_raw_string (const char *str)
1185 const char *save = str;
1190 while (str[0] == '#')
1197 /* Return true if STR looks like the end of a raw string that had N
1198 hashes at the start. */
1201 ends_raw_string (const char *str, int n)
1205 gdb_assert (str[0] == '"');
1206 for (i = 0; i < n; ++i)
1207 if (str[i + 1] != '#')
1212 /* Lex a string constant. */
1217 int is_byte = lexptr[0] == 'b';
1219 int len_in_chars = 0;
1223 raw_length = starts_raw_string (lexptr);
1224 lexptr += raw_length;
1225 gdb_assert (lexptr[0] == '"');
1234 if (lexptr[0] == '"' && ends_raw_string (lexptr, raw_length - 1))
1236 /* Exit with lexptr pointing after the final "#". */
1237 lexptr += raw_length;
1240 else if (lexptr[0] == '\0')
1241 error (_("Unexpected EOF in string"));
1243 value = lexptr[0] & 0xff;
1244 if (is_byte && value > 127)
1245 error (_("Non-ASCII value in raw byte string"));
1246 obstack_1grow (&work_obstack, value);
1250 else if (lexptr[0] == '"')
1252 /* Make sure to skip the quote. */
1256 else if (lexptr[0] == '\\')
1258 value = lex_escape (is_byte);
1261 obstack_1grow (&work_obstack, value);
1263 convert_between_encodings ("UTF-32", "UTF-8", (gdb_byte *) &value,
1264 sizeof (value), sizeof (value),
1265 &work_obstack, translit_none);
1267 else if (lexptr[0] == '\0')
1268 error (_("Unexpected EOF in string"));
1271 value = lexptr[0] & 0xff;
1272 if (is_byte && value > 127)
1273 error (_("Non-ASCII value in byte string"));
1274 obstack_1grow (&work_obstack, value);
1279 rustyylval.sval.length = obstack_object_size (&work_obstack);
1280 rustyylval.sval.ptr = (const char *) obstack_finish (&work_obstack);
1281 return is_byte ? BYTESTRING : STRING;
1284 /* Return true if STRING starts with whitespace followed by a digit. */
1287 space_then_number (const char *string)
1289 const char *p = string;
1291 while (p[0] == ' ' || p[0] == '\t')
1296 return *p >= '0' && *p <= '9';
1299 /* Return true if C can start an identifier. */
1302 rust_identifier_start_p (char c)
1304 return ((c >= 'a' && c <= 'z')
1305 || (c >= 'A' && c <= 'Z')
1310 /* Lex an identifier. */
1313 lex_identifier (void)
1315 const char *start = lexptr;
1316 unsigned int length;
1317 const struct token_info *token;
1319 int is_gdb_var = lexptr[0] == '$';
1321 gdb_assert (rust_identifier_start_p (lexptr[0]));
1325 /* For the time being this doesn't handle Unicode rules. Non-ASCII
1326 identifiers are gated anyway. */
1327 while ((lexptr[0] >= 'a' && lexptr[0] <= 'z')
1328 || (lexptr[0] >= 'A' && lexptr[0] <= 'Z')
1330 || (is_gdb_var && lexptr[0] == '$')
1331 || (lexptr[0] >= '0' && lexptr[0] <= '9'))
1335 length = lexptr - start;
1337 for (i = 0; i < ARRAY_SIZE (identifier_tokens); ++i)
1339 if (length == strlen (identifier_tokens[i].name)
1340 && strncmp (identifier_tokens[i].name, start, length) == 0)
1342 token = &identifier_tokens[i];
1349 if (token->value == 0)
1351 /* Leave the terminating token alone. */
1356 else if (token == NULL
1357 && (strncmp (start, "thread", length) == 0
1358 || strncmp (start, "task", length) == 0)
1359 && space_then_number (lexptr))
1361 /* "task" or "thread" followed by a number terminates the
1362 parse, per gdb rules. */
1367 if (token == NULL || (parse_completion && lexptr[0] == '\0'))
1368 rustyylval.sval = make_stoken (rust_copy_name (start, length));
1370 if (parse_completion && lexptr[0] == '\0')
1372 /* Prevent rustyylex from returning two COMPLETE tokens. */
1373 prev_lexptr = lexptr;
1378 return token->value;
1384 /* Lex an operator. */
1389 const struct token_info *token = NULL;
1392 for (i = 0; i < ARRAY_SIZE (operator_tokens); ++i)
1394 if (strncmp (operator_tokens[i].name, lexptr,
1395 strlen (operator_tokens[i].name)) == 0)
1397 lexptr += strlen (operator_tokens[i].name);
1398 token = &operator_tokens[i];
1405 rustyylval.opcode = token->opcode;
1406 return token->value;
1417 regmatch_t subexps[NUM_SUBEXPRESSIONS];
1420 int could_be_decimal = 1;
1421 char *type_name = NULL;
1424 int type_index = -1;
1427 struct cleanup *cleanup = make_cleanup (null_cleanup, NULL);
1429 match = regexec (&number_regex, lexptr, ARRAY_SIZE (subexps), subexps, 0);
1430 /* Failure means the regexp is broken. */
1431 gdb_assert (match == 0);
1433 if (subexps[INT_TEXT].rm_so != -1)
1435 /* Integer part matched. */
1437 end_index = subexps[INT_TEXT].rm_eo;
1438 if (subexps[INT_TYPE].rm_so == -1)
1442 type_index = INT_TYPE;
1443 could_be_decimal = 0;
1446 else if (subexps[FLOAT_TYPE1].rm_so != -1)
1448 /* Found floating point type suffix. */
1449 end_index = subexps[FLOAT_TYPE1].rm_so;
1450 type_index = FLOAT_TYPE1;
1452 else if (subexps[FLOAT_TYPE2].rm_so != -1)
1454 /* Found floating point type suffix. */
1455 end_index = subexps[FLOAT_TYPE2].rm_so;
1456 type_index = FLOAT_TYPE2;
1460 /* Any other floating point match. */
1461 end_index = subexps[0].rm_eo;
1465 /* We need a special case if the final character is ".". In this
1466 case we might need to parse an integer. For example, "23.f()" is
1467 a request for a trait method call, not a syntax error involving
1468 the floating point number "23.". */
1469 gdb_assert (subexps[0].rm_eo > 0);
1470 if (lexptr[subexps[0].rm_eo - 1] == '.')
1472 const char *next = skip_spaces_const (&lexptr[subexps[0].rm_eo]);
1474 if (rust_identifier_start_p (*next) || *next == '.')
1478 end_index = subexps[0].rm_eo;
1480 could_be_decimal = 1;
1484 /* Compute the type name if we haven't already. */
1485 if (type_name == NULL)
1487 gdb_assert (type_index != -1);
1488 type_name = xstrndup (lexptr + subexps[type_index].rm_so,
1489 (subexps[type_index].rm_eo
1490 - subexps[type_index].rm_so));
1491 make_cleanup (xfree, type_name);
1494 /* Look up the type. */
1495 type = rust_type (type_name);
1497 /* Copy the text of the number and remove the "_"s. */
1498 number = xstrndup (lexptr, end_index);
1499 make_cleanup (xfree, number);
1500 for (i = out = 0; number[i]; ++i)
1502 if (number[i] == '_')
1503 could_be_decimal = 0;
1505 number[out++] = number[i];
1509 /* Advance past the match. */
1510 lexptr += subexps[0].rm_eo;
1512 /* Parse the number. */
1516 if (number[0] == '0')
1518 if (number[1] == 'x')
1520 else if (number[1] == 'o')
1522 else if (number[1] == 'b')
1527 could_be_decimal = 0;
1530 rustyylval.typed_val_int.val = strtoul (number, NULL, radix);
1531 rustyylval.typed_val_int.type = type;
1535 rustyylval.typed_val_float.dval = strtod (number, NULL);
1536 rustyylval.typed_val_float.type = type;
1539 do_cleanups (cleanup);
1540 return is_integer ? (could_be_decimal ? DECIMAL_INTEGER : INTEGER) : FLOAT;
1548 /* Skip all leading whitespace. */
1549 while (lexptr[0] == ' ' || lexptr[0] == '\t' || lexptr[0] == '\r'
1550 || lexptr[0] == '\n')
1553 /* If we hit EOF and we're completing, then return COMPLETE -- maybe
1554 we're completing an empty string at the end of a field_expr.
1555 But, we don't want to return two COMPLETE tokens in a row. */
1556 if (lexptr[0] == '\0' && lexptr == prev_lexptr)
1558 prev_lexptr = lexptr;
1559 if (lexptr[0] == '\0')
1561 if (parse_completion)
1563 rustyylval.sval = make_stoken ("");
1569 if (lexptr[0] >= '0' && lexptr[0] <= '9')
1570 return lex_number ();
1571 else if (lexptr[0] == 'b' && lexptr[1] == '\'')
1572 return lex_character ();
1573 else if (lexptr[0] == 'b' && lexptr[1] == '"')
1574 return lex_string ();
1575 else if (lexptr[0] == 'b' && starts_raw_string (lexptr + 1))
1576 return lex_string ();
1577 else if (starts_raw_string (lexptr))
1578 return lex_string ();
1579 else if (rust_identifier_start_p (lexptr[0]))
1580 return lex_identifier ();
1581 else if (lexptr[0] == '"')
1582 return lex_string ();
1583 else if (lexptr[0] == '\'')
1584 return lex_character ();
1585 else if (lexptr[0] == '}' || lexptr[0] == ']')
1587 /* Falls through to lex_operator. */
1590 else if (lexptr[0] == '(' || lexptr[0] == '{')
1592 /* Falls through to lex_operator. */
1595 else if (lexptr[0] == ',' && comma_terminates && paren_depth == 0)
1598 return lex_operator ();
1601 /* Push back a single character to be re-lexed. */
1604 rust_push_back (char c)
1606 /* Can't be called before any lexing. */
1607 gdb_assert (prev_lexptr != NULL);
1610 gdb_assert (*lexptr == c);
1615 /* Make an arbitrary operation and fill in the fields. */
1617 static const struct rust_op *
1618 ast_operation (enum exp_opcode opcode, const struct rust_op *left,
1619 const struct rust_op *right)
1621 struct rust_op *result = OBSTACK_ZALLOC (&work_obstack, struct rust_op);
1623 result->opcode = opcode;
1624 result->left.op = left;
1625 result->right.op = right;
1630 /* Make a compound assignment operation. */
1632 static const struct rust_op *
1633 ast_compound_assignment (enum exp_opcode opcode, const struct rust_op *left,
1634 const struct rust_op *right)
1636 struct rust_op *result = OBSTACK_ZALLOC (&work_obstack, struct rust_op);
1638 result->opcode = opcode;
1639 result->compound_assignment = 1;
1640 result->left.op = left;
1641 result->right.op = right;
1646 /* Make a typed integer literal operation. */
1648 static const struct rust_op *
1649 ast_literal (struct typed_val_int val)
1651 struct rust_op *result = OBSTACK_ZALLOC (&work_obstack, struct rust_op);
1653 result->opcode = OP_LONG;
1654 result->left.typed_val_int = val;
1659 /* Make a typed floating point literal operation. */
1661 static const struct rust_op *
1662 ast_dliteral (struct typed_val_float val)
1664 struct rust_op *result = OBSTACK_ZALLOC (&work_obstack, struct rust_op);
1666 result->opcode = OP_DOUBLE;
1667 result->left.typed_val_float = val;
1672 /* Make a unary operation. */
1674 static const struct rust_op *
1675 ast_unary (enum exp_opcode opcode, const struct rust_op *expr)
1677 return ast_operation (opcode, expr, NULL);
1680 /* Make a cast operation. */
1682 static const struct rust_op *
1683 ast_cast (const struct rust_op *expr, const struct rust_op *type)
1685 struct rust_op *result = OBSTACK_ZALLOC (&work_obstack, struct rust_op);
1687 result->opcode = UNOP_CAST;
1688 result->left.op = expr;
1689 result->right.op = type;
1694 /* Make a call-like operation. This is nominally a function call, but
1695 when lowering we may discover that it actually represents the
1696 creation of a tuple struct. */
1698 static const struct rust_op *
1699 ast_call_ish (enum exp_opcode opcode, const struct rust_op *expr,
1700 VEC (rust_op_ptr) **params)
1702 struct rust_op *result = OBSTACK_ZALLOC (&work_obstack, struct rust_op);
1704 result->opcode = opcode;
1705 result->left.op = expr;
1706 result->right.params = params;
1711 /* Make a structure creation operation. */
1713 static const struct rust_op *
1714 ast_struct (const struct rust_op *name, VEC (set_field) **fields)
1716 struct rust_op *result = OBSTACK_ZALLOC (&work_obstack, struct rust_op);
1718 result->opcode = OP_AGGREGATE;
1719 result->left.op = name;
1720 result->right.field_inits = fields;
1725 /* Make an identifier path. */
1727 static const struct rust_op *
1728 ast_path (struct stoken path, VEC (rust_op_ptr) **params)
1730 struct rust_op *result = OBSTACK_ZALLOC (&work_obstack, struct rust_op);
1732 result->opcode = OP_VAR_VALUE;
1733 result->left.sval = path;
1734 result->right.params = params;
1739 /* Make a string constant operation. */
1741 static const struct rust_op *
1742 ast_string (struct stoken str)
1744 struct rust_op *result = OBSTACK_ZALLOC (&work_obstack, struct rust_op);
1746 result->opcode = OP_STRING;
1747 result->left.sval = str;
1752 /* Make a field expression. */
1754 static const struct rust_op *
1755 ast_structop (const struct rust_op *left, const char *name, int completing)
1757 struct rust_op *result = OBSTACK_ZALLOC (&work_obstack, struct rust_op);
1759 result->opcode = STRUCTOP_STRUCT;
1760 result->completing = completing;
1761 result->left.op = left;
1762 result->right.sval = make_stoken (name);
1767 /* Make an anonymous struct operation, like 'x.0'. */
1769 static const struct rust_op *
1770 ast_structop_anonymous (const struct rust_op *left,
1771 struct typed_val_int number)
1773 struct rust_op *result = OBSTACK_ZALLOC (&work_obstack, struct rust_op);
1775 result->opcode = STRUCTOP_ANONYMOUS;
1776 result->left.op = left;
1777 result->right.typed_val_int = number;
1782 /* Make a range operation. */
1784 static const struct rust_op *
1785 ast_range (const struct rust_op *lhs, const struct rust_op *rhs)
1787 struct rust_op *result = OBSTACK_ZALLOC (&work_obstack, struct rust_op);
1789 result->opcode = OP_RANGE;
1790 result->left.op = lhs;
1791 result->right.op = rhs;
1796 /* A helper function to make a type-related AST node. */
1798 static struct rust_op *
1799 ast_basic_type (enum type_code typecode)
1801 struct rust_op *result = OBSTACK_ZALLOC (&work_obstack, struct rust_op);
1803 result->opcode = OP_TYPE;
1804 result->typecode = typecode;
1808 /* Create an AST node describing an array type. */
1810 static const struct rust_op *
1811 ast_array_type (const struct rust_op *lhs, struct typed_val_int val)
1813 struct rust_op *result = ast_basic_type (TYPE_CODE_ARRAY);
1815 result->left.op = lhs;
1816 result->right.typed_val_int = val;
1820 /* Create an AST node describing a reference type. */
1822 static const struct rust_op *
1823 ast_slice_type (const struct rust_op *type)
1825 /* Use TYPE_CODE_COMPLEX just because it is handy. */
1826 struct rust_op *result = ast_basic_type (TYPE_CODE_COMPLEX);
1828 result->left.op = type;
1832 /* Create an AST node describing a reference type. */
1834 static const struct rust_op *
1835 ast_reference_type (const struct rust_op *type)
1837 struct rust_op *result = ast_basic_type (TYPE_CODE_REF);
1839 result->left.op = type;
1843 /* Create an AST node describing a pointer type. */
1845 static const struct rust_op *
1846 ast_pointer_type (const struct rust_op *type, int is_mut)
1848 struct rust_op *result = ast_basic_type (TYPE_CODE_PTR);
1850 result->left.op = type;
1851 /* For the time being we ignore is_mut. */
1855 /* Create an AST node describing a function type. */
1857 static const struct rust_op *
1858 ast_function_type (const struct rust_op *rtype, VEC (rust_op_ptr) **params)
1860 struct rust_op *result = ast_basic_type (TYPE_CODE_FUNC);
1862 result->left.op = rtype;
1863 result->right.params = params;
1867 /* Create an AST node describing a tuple type. */
1869 static const struct rust_op *
1870 ast_tuple_type (VEC (rust_op_ptr) **params)
1872 struct rust_op *result = ast_basic_type (TYPE_CODE_STRUCT);
1874 result->left.params = params;
1878 /* A helper to appropriately munge NAME and BLOCK depending on the
1879 presence of a leading "::". */
1882 munge_name_and_block (const char **name, const struct block **block)
1884 /* If it is a global reference, skip the current block in favor of
1885 the static block. */
1886 if (strncmp (*name, "::", 2) == 0)
1889 *block = block_static_block (*block);
1893 /* Like lookup_symbol, but handles Rust namespace conventions, and
1894 doesn't require field_of_this_result. */
1896 static struct block_symbol
1897 rust_lookup_symbol (const char *name, const struct block *block,
1898 const domain_enum domain)
1900 struct block_symbol result;
1902 munge_name_and_block (&name, &block);
1904 result = lookup_symbol (name, block, domain, NULL);
1905 if (result.symbol != NULL)
1906 update_innermost_block (result);
1910 /* Look up a type, following Rust namespace conventions. */
1912 static struct type *
1913 rust_lookup_type (const char *name, const struct block *block)
1915 struct block_symbol result;
1918 munge_name_and_block (&name, &block);
1920 result = lookup_symbol (name, block, STRUCT_DOMAIN, NULL);
1921 if (result.symbol != NULL)
1923 update_innermost_block (result);
1924 return SYMBOL_TYPE (result.symbol);
1927 type = lookup_typename (parse_language (pstate), parse_gdbarch (pstate),
1932 /* Last chance, try a built-in type. */
1933 return language_lookup_primitive_type (parse_language (pstate),
1934 parse_gdbarch (pstate),
1938 static struct type *convert_ast_to_type (struct parser_state *state,
1939 const struct rust_op *operation);
1940 static const char *convert_name (struct parser_state *state,
1941 const struct rust_op *operation);
1943 /* Convert a vector of rust_ops representing types to a vector of
1946 static VEC (type_ptr) *
1947 convert_params_to_types (struct parser_state *state, VEC (rust_op_ptr) *params)
1950 const struct rust_op *op;
1951 VEC (type_ptr) *result = NULL;
1952 struct cleanup *cleanup = make_cleanup (VEC_cleanup (type_ptr), &result);
1954 for (i = 0; VEC_iterate (rust_op_ptr, params, i, op); ++i)
1955 VEC_safe_push (type_ptr, result, convert_ast_to_type (state, op));
1957 discard_cleanups (cleanup);
1961 /* Convert a rust_op representing a type to a struct type *. */
1963 static struct type *
1964 convert_ast_to_type (struct parser_state *state,
1965 const struct rust_op *operation)
1967 struct type *type, *result = NULL;
1969 if (operation->opcode == OP_VAR_VALUE)
1971 const char *varname = convert_name (state, operation);
1973 result = rust_lookup_type (varname, expression_context_block);
1975 error (_("No typed name '%s' in current context"), varname);
1979 gdb_assert (operation->opcode == OP_TYPE);
1981 switch (operation->typecode)
1983 case TYPE_CODE_ARRAY:
1984 type = convert_ast_to_type (state, operation->left.op);
1985 if (operation->right.typed_val_int.val < 0)
1986 error (_("Negative array length"));
1987 result = lookup_array_range_type (type, 0,
1988 operation->right.typed_val_int.val - 1);
1991 case TYPE_CODE_COMPLEX:
1993 struct type *usize = rust_type ("usize");
1995 type = convert_ast_to_type (state, operation->left.op);
1996 result = rust_slice_type ("&[*gdb*]", type, usize);
2002 /* For now we treat &x and *x identically. */
2003 type = convert_ast_to_type (state, operation->left.op);
2004 result = lookup_pointer_type (type);
2007 case TYPE_CODE_FUNC:
2009 VEC (type_ptr) *args
2010 = convert_params_to_types (state, *operation->right.params);
2011 struct cleanup *cleanup
2012 = make_cleanup (VEC_cleanup (type_ptr), &args);
2013 struct type **argtypes = NULL;
2015 type = convert_ast_to_type (state, operation->left.op);
2016 if (!VEC_empty (type_ptr, args))
2017 argtypes = VEC_address (type_ptr, args);
2020 = lookup_function_type_with_arguments (type,
2021 VEC_length (type_ptr, args),
2023 result = lookup_pointer_type (result);
2025 do_cleanups (cleanup);
2029 case TYPE_CODE_STRUCT:
2031 VEC (type_ptr) *args
2032 = convert_params_to_types (state, *operation->left.params);
2033 struct cleanup *cleanup
2034 = make_cleanup (VEC_cleanup (type_ptr), &args);
2039 obstack_1grow (&work_obstack, '(');
2040 for (i = 0; VEC_iterate (type_ptr, args, i, type); ++i)
2042 char *type_name = type_to_string (type);
2045 obstack_1grow (&work_obstack, ',');
2046 obstack_grow_str (&work_obstack, type_name);
2051 obstack_grow_str0 (&work_obstack, ")");
2052 name = (const char *) obstack_finish (&work_obstack);
2054 /* We don't allow creating new tuple types (yet), but we do
2055 allow looking up existing tuple types. */
2056 result = rust_lookup_type (name, expression_context_block);
2058 error (_("could not find tuple type '%s'"), name);
2060 do_cleanups (cleanup);
2065 gdb_assert_not_reached ("unhandled opcode in convert_ast_to_type");
2068 gdb_assert (result != NULL);
2072 /* A helper function to turn a rust_op representing a name into a full
2073 name. This applies generic arguments as needed. The returned name
2074 is allocated on the work obstack. */
2077 convert_name (struct parser_state *state, const struct rust_op *operation)
2079 VEC (type_ptr) *types;
2080 struct cleanup *cleanup;
2084 gdb_assert (operation->opcode == OP_VAR_VALUE);
2086 if (operation->right.params == NULL)
2087 return operation->left.sval.ptr;
2089 types = convert_params_to_types (state, *operation->right.params);
2090 cleanup = make_cleanup (VEC_cleanup (type_ptr), &types);
2092 obstack_grow_str (&work_obstack, operation->left.sval.ptr);
2093 obstack_1grow (&work_obstack, '<');
2094 for (i = 0; VEC_iterate (type_ptr, types, i, type); ++i)
2096 char *type_name = type_to_string (type);
2099 obstack_1grow (&work_obstack, ',');
2101 obstack_grow_str (&work_obstack, type_name);
2104 obstack_grow_str0 (&work_obstack, ">");
2106 do_cleanups (cleanup);
2108 return (const char *) obstack_finish (&work_obstack);
2111 static void convert_ast_to_expression (struct parser_state *state,
2112 const struct rust_op *operation,
2113 const struct rust_op *top);
2115 /* A helper function that converts a vec of rust_ops to a gdb
2119 convert_params_to_expression (struct parser_state *state,
2120 VEC (rust_op_ptr) *params,
2121 const struct rust_op *top)
2126 for (i = 0; VEC_iterate (rust_op_ptr, params, i, elem); ++i)
2127 convert_ast_to_expression (state, elem, top);
2130 /* Lower a rust_op to a gdb expression. STATE is the parser state.
2131 OPERATION is the operation to lower. TOP is a pointer to the
2132 top-most operation; it is used to handle the special case where the
2133 top-most expression is an identifier and can be optionally lowered
2137 convert_ast_to_expression (struct parser_state *state,
2138 const struct rust_op *operation,
2139 const struct rust_op *top)
2141 switch (operation->opcode)
2144 write_exp_elt_opcode (state, OP_LONG);
2145 write_exp_elt_type (state, operation->left.typed_val_int.type);
2146 write_exp_elt_longcst (state, operation->left.typed_val_int.val);
2147 write_exp_elt_opcode (state, OP_LONG);
2151 write_exp_elt_opcode (state, OP_DOUBLE);
2152 write_exp_elt_type (state, operation->left.typed_val_float.type);
2153 write_exp_elt_dblcst (state, operation->left.typed_val_float.dval);
2154 write_exp_elt_opcode (state, OP_DOUBLE);
2157 case STRUCTOP_STRUCT:
2159 convert_ast_to_expression (state, operation->left.op, top);
2161 if (operation->completing)
2162 mark_struct_expression (state);
2163 write_exp_elt_opcode (state, STRUCTOP_STRUCT);
2164 write_exp_string (state, operation->right.sval);
2165 write_exp_elt_opcode (state, STRUCTOP_STRUCT);
2169 case STRUCTOP_ANONYMOUS:
2171 convert_ast_to_expression (state, operation->left.op, top);
2173 write_exp_elt_opcode (state, STRUCTOP_ANONYMOUS);
2174 write_exp_elt_longcst (state, operation->right.typed_val_int.val);
2175 write_exp_elt_opcode (state, STRUCTOP_ANONYMOUS);
2181 case UNOP_COMPLEMENT:
2184 convert_ast_to_expression (state, operation->left.op, top);
2185 write_exp_elt_opcode (state, operation->opcode);
2188 case BINOP_SUBSCRIPT:
2195 case BINOP_BITWISE_AND:
2196 case BINOP_BITWISE_IOR:
2197 case BINOP_BITWISE_XOR:
2200 case BINOP_LOGICAL_OR:
2201 case BINOP_LOGICAL_AND:
2203 case BINOP_NOTEQUAL:
2210 convert_ast_to_expression (state, operation->left.op, top);
2211 convert_ast_to_expression (state, operation->right.op, top);
2212 if (operation->compound_assignment)
2214 write_exp_elt_opcode (state, BINOP_ASSIGN_MODIFY);
2215 write_exp_elt_opcode (state, operation->opcode);
2216 write_exp_elt_opcode (state, BINOP_ASSIGN_MODIFY);
2219 write_exp_elt_opcode (state, operation->opcode);
2221 if (operation->compound_assignment
2222 || operation->opcode == BINOP_ASSIGN)
2226 type = language_lookup_primitive_type (parse_language (state),
2227 parse_gdbarch (state),
2230 write_exp_elt_opcode (state, OP_LONG);
2231 write_exp_elt_type (state, type);
2232 write_exp_elt_longcst (state, 0);
2233 write_exp_elt_opcode (state, OP_LONG);
2235 write_exp_elt_opcode (state, BINOP_COMMA);
2241 struct type *type = convert_ast_to_type (state, operation->right.op);
2243 convert_ast_to_expression (state, operation->left.op, top);
2244 write_exp_elt_opcode (state, UNOP_CAST);
2245 write_exp_elt_type (state, type);
2246 write_exp_elt_opcode (state, UNOP_CAST);
2252 if (operation->left.op->opcode == OP_VAR_VALUE)
2255 const char *varname = convert_name (state, operation->left.op);
2257 type = rust_lookup_type (varname, expression_context_block);
2260 /* This is actually a tuple struct expression, not a
2264 VEC (rust_op_ptr) *params = *operation->right.params;
2266 if (TYPE_CODE (type) != TYPE_CODE_NAMESPACE)
2268 if (!rust_tuple_struct_type_p (type))
2269 error (_("Type %s is not a tuple struct"), varname);
2272 VEC_iterate (rust_op_ptr, params, i, elem);
2275 char *cell = get_print_cell ();
2277 xsnprintf (cell, PRINT_CELL_SIZE, "__%d", i);
2278 write_exp_elt_opcode (state, OP_NAME);
2279 write_exp_string (state, make_stoken (cell));
2280 write_exp_elt_opcode (state, OP_NAME);
2282 convert_ast_to_expression (state, elem, top);
2285 write_exp_elt_opcode (state, OP_AGGREGATE);
2286 write_exp_elt_type (state, type);
2287 write_exp_elt_longcst (state,
2288 2 * VEC_length (rust_op_ptr,
2290 write_exp_elt_opcode (state, OP_AGGREGATE);
2295 convert_ast_to_expression (state, operation->left.op, top);
2296 convert_params_to_expression (state, *operation->right.params, top);
2297 write_exp_elt_opcode (state, OP_FUNCALL);
2298 write_exp_elt_longcst (state, VEC_length (rust_op_ptr,
2299 *operation->right.params));
2300 write_exp_elt_longcst (state, OP_FUNCALL);
2305 gdb_assert (operation->left.op == NULL);
2306 convert_params_to_expression (state, *operation->right.params, top);
2307 write_exp_elt_opcode (state, OP_ARRAY);
2308 write_exp_elt_longcst (state, 0);
2309 write_exp_elt_longcst (state, VEC_length (rust_op_ptr,
2310 *operation->right.params) - 1);
2311 write_exp_elt_longcst (state, OP_ARRAY);
2316 struct block_symbol sym;
2317 const char *varname;
2319 if (operation->left.sval.ptr[0] == '$')
2321 write_dollar_variable (state, operation->left.sval);
2325 varname = convert_name (state, operation);
2326 sym = rust_lookup_symbol (varname, expression_context_block,
2328 if (sym.symbol != NULL)
2330 write_exp_elt_opcode (state, OP_VAR_VALUE);
2331 write_exp_elt_block (state, sym.block);
2332 write_exp_elt_sym (state, sym.symbol);
2333 write_exp_elt_opcode (state, OP_VAR_VALUE);
2339 type = rust_lookup_type (varname, expression_context_block);
2341 error (_("No symbol '%s' in current context"), varname);
2343 if (TYPE_CODE (type) == TYPE_CODE_STRUCT
2344 && TYPE_NFIELDS (type) == 0)
2346 /* A unit-like struct. */
2347 write_exp_elt_opcode (state, OP_AGGREGATE);
2348 write_exp_elt_type (state, type);
2349 write_exp_elt_longcst (state, 0);
2350 write_exp_elt_opcode (state, OP_AGGREGATE);
2352 else if (operation == top)
2354 write_exp_elt_opcode (state, OP_TYPE);
2355 write_exp_elt_type (state, type);
2356 write_exp_elt_opcode (state, OP_TYPE);
2367 struct set_field *init;
2368 VEC (set_field) *fields = *operation->right.field_inits;
2373 for (i = 0; VEC_iterate (set_field, fields, i, init); ++i)
2375 if (init->name.ptr != NULL)
2377 write_exp_elt_opcode (state, OP_NAME);
2378 write_exp_string (state, init->name);
2379 write_exp_elt_opcode (state, OP_NAME);
2383 convert_ast_to_expression (state, init->init, top);
2386 if (init->name.ptr == NULL)
2388 /* This is handled differently from Ada in our
2390 write_exp_elt_opcode (state, OP_OTHERS);
2394 name = convert_name (state, operation->left.op);
2395 type = rust_lookup_type (name, expression_context_block);
2397 error (_("Could not find type '%s'"), operation->left.sval.ptr);
2399 if (TYPE_CODE (type) != TYPE_CODE_STRUCT
2400 || rust_tuple_type_p (type)
2401 || rust_tuple_struct_type_p (type))
2402 error (_("Struct expression applied to non-struct type"));
2404 write_exp_elt_opcode (state, OP_AGGREGATE);
2405 write_exp_elt_type (state, type);
2406 write_exp_elt_longcst (state, length);
2407 write_exp_elt_opcode (state, OP_AGGREGATE);
2413 write_exp_elt_opcode (state, OP_STRING);
2414 write_exp_string (state, operation->left.sval);
2415 write_exp_elt_opcode (state, OP_STRING);
2421 enum range_type kind = BOTH_BOUND_DEFAULT;
2423 if (operation->left.op != NULL)
2425 convert_ast_to_expression (state, operation->left.op, top);
2426 kind = HIGH_BOUND_DEFAULT;
2428 if (operation->right.op != NULL)
2430 convert_ast_to_expression (state, operation->right.op, top);
2431 if (kind == BOTH_BOUND_DEFAULT)
2432 kind = LOW_BOUND_DEFAULT;
2435 gdb_assert (kind == HIGH_BOUND_DEFAULT);
2436 kind = NONE_BOUND_DEFAULT;
2439 write_exp_elt_opcode (state, OP_RANGE);
2440 write_exp_elt_longcst (state, kind);
2441 write_exp_elt_opcode (state, OP_RANGE);
2446 gdb_assert_not_reached ("unhandled opcode in convert_ast_to_expression");
2452 /* The parser as exposed to gdb. */
2455 rust_parse (struct parser_state *state)
2458 struct cleanup *cleanup;
2460 obstack_init (&work_obstack);
2461 cleanup = make_cleanup_obstack_free (&work_obstack);
2465 result = rustyyparse ();
2467 if (!result || (parse_completion && rust_ast != NULL))
2469 const struct rust_op *ast = rust_ast;
2472 gdb_assert (ast != NULL);
2473 convert_ast_to_expression (state, ast, ast);
2476 do_cleanups (cleanup);
2480 /* The parser error handler. */
2483 rustyyerror (char *msg)
2485 const char *where = prev_lexptr ? prev_lexptr : lexptr;
2486 error (_("%s in expression, near `%s'."), (msg ? msg : "Error"), where);
2493 /* Initialize the lexer for testing. */
2496 rust_lex_test_init (const char *input)
2503 /* A test helper that lexes a string, expecting a single token. It
2504 returns the lexer data for this token. */
2507 rust_lex_test_one (const char *input, int expected)
2512 rust_lex_test_init (input);
2514 token = rustyylex ();
2515 SELF_CHECK (token == expected);
2516 result = rustyylval;
2520 token = rustyylex ();
2521 SELF_CHECK (token == 0);
2527 /* Test that INPUT lexes as the integer VALUE. */
2530 rust_lex_int_test (const char *input, int value, int kind)
2532 RUSTSTYPE result = rust_lex_test_one (input, kind);
2533 SELF_CHECK (result.typed_val_int.val == value);
2536 /* Test that INPUT throws an exception with text ERR. */
2539 rust_lex_exception_test (const char *input, const char *err)
2543 /* The "kind" doesn't matter. */
2544 rust_lex_test_one (input, DECIMAL_INTEGER);
2547 CATCH (except, RETURN_MASK_ERROR)
2549 SELF_CHECK (strcmp (except.message, err) == 0);
2554 /* Test that INPUT lexes as the identifier, string, or byte-string
2555 VALUE. KIND holds the expected token kind. */
2558 rust_lex_stringish_test (const char *input, const char *value, int kind)
2560 RUSTSTYPE result = rust_lex_test_one (input, kind);
2561 SELF_CHECK (result.sval.length == strlen (value));
2562 SELF_CHECK (strncmp (result.sval.ptr, value, result.sval.length) == 0);
2565 /* Helper to test that a string parses as a given token sequence. */
2568 rust_lex_test_sequence (const char *input, int len, const int expected[])
2575 for (i = 0; i < len; ++i)
2577 int token = rustyylex ();
2579 SELF_CHECK (token == expected[i]);
2583 /* Tests for an integer-parsing corner case. */
2586 rust_lex_test_trailing_dot (void)
2588 const int expected1[] = { DECIMAL_INTEGER, '.', IDENT, '(', ')', 0 };
2589 const int expected2[] = { INTEGER, '.', IDENT, '(', ')', 0 };
2590 const int expected3[] = { FLOAT, EQEQ, '(', ')', 0 };
2591 const int expected4[] = { DECIMAL_INTEGER, DOTDOT, DECIMAL_INTEGER, 0 };
2593 rust_lex_test_sequence ("23.g()", ARRAY_SIZE (expected1), expected1);
2594 rust_lex_test_sequence ("23_0.g()", ARRAY_SIZE (expected2), expected2);
2595 rust_lex_test_sequence ("23.==()", ARRAY_SIZE (expected3), expected3);
2596 rust_lex_test_sequence ("23..25", ARRAY_SIZE (expected4), expected4);
2599 /* Tests of completion. */
2602 rust_lex_test_completion (void)
2604 const int expected[] = { IDENT, '.', COMPLETE, 0 };
2606 parse_completion = 1;
2608 rust_lex_test_sequence ("something.wha", ARRAY_SIZE (expected), expected);
2609 rust_lex_test_sequence ("something.", ARRAY_SIZE (expected), expected);
2611 parse_completion = 0;
2614 /* Test pushback. */
2617 rust_lex_test_push_back (void)
2621 rust_lex_test_init (">>=");
2623 token = rustyylex ();
2624 SELF_CHECK (token == COMPOUND_ASSIGN);
2625 SELF_CHECK (rustyylval.opcode == BINOP_RSH);
2627 rust_push_back ('=');
2629 token = rustyylex ();
2630 SELF_CHECK (token == '=');
2632 token = rustyylex ();
2633 SELF_CHECK (token == 0);
2636 /* Unit test the lexer. */
2639 rust_lex_tests (void)
2643 obstack_init (&work_obstack);
2646 rust_lex_test_one ("", 0);
2647 rust_lex_test_one (" \t \n \r ", 0);
2648 rust_lex_test_one ("thread 23", 0);
2649 rust_lex_test_one ("task 23", 0);
2650 rust_lex_test_one ("th 104", 0);
2651 rust_lex_test_one ("ta 97", 0);
2653 rust_lex_int_test ("'z'", 'z', INTEGER);
2654 rust_lex_int_test ("'\\xff'", 0xff, INTEGER);
2655 rust_lex_int_test ("'\\u{1016f}'", 0x1016f, INTEGER);
2656 rust_lex_int_test ("b'z'", 'z', INTEGER);
2657 rust_lex_int_test ("b'\\xfe'", 0xfe, INTEGER);
2658 rust_lex_int_test ("b'\\xFE'", 0xfe, INTEGER);
2659 rust_lex_int_test ("b'\\xfE'", 0xfe, INTEGER);
2661 /* Test all escapes in both modes. */
2662 rust_lex_int_test ("'\\n'", '\n', INTEGER);
2663 rust_lex_int_test ("'\\r'", '\r', INTEGER);
2664 rust_lex_int_test ("'\\t'", '\t', INTEGER);
2665 rust_lex_int_test ("'\\\\'", '\\', INTEGER);
2666 rust_lex_int_test ("'\\0'", '\0', INTEGER);
2667 rust_lex_int_test ("'\\''", '\'', INTEGER);
2668 rust_lex_int_test ("'\\\"'", '"', INTEGER);
2670 rust_lex_int_test ("b'\\n'", '\n', INTEGER);
2671 rust_lex_int_test ("b'\\r'", '\r', INTEGER);
2672 rust_lex_int_test ("b'\\t'", '\t', INTEGER);
2673 rust_lex_int_test ("b'\\\\'", '\\', INTEGER);
2674 rust_lex_int_test ("b'\\0'", '\0', INTEGER);
2675 rust_lex_int_test ("b'\\''", '\'', INTEGER);
2676 rust_lex_int_test ("b'\\\"'", '"', INTEGER);
2678 rust_lex_exception_test ("'z", "Unterminated character literal");
2679 rust_lex_exception_test ("b'\\x0'", "Not enough hex digits seen");
2680 rust_lex_exception_test ("b'\\u{0}'", "Unicode escape in byte literal");
2681 rust_lex_exception_test ("'\\x0'", "Not enough hex digits seen");
2682 rust_lex_exception_test ("'\\u0'", "Missing '{' in Unicode escape");
2683 rust_lex_exception_test ("'\\u{0", "Missing '}' in Unicode escape");
2684 rust_lex_exception_test ("'\\u{0000007}", "Overlong hex escape");
2685 rust_lex_exception_test ("'\\u{}", "Not enough hex digits seen");
2686 rust_lex_exception_test ("'\\Q'", "Invalid escape \\Q in literal");
2687 rust_lex_exception_test ("b'\\Q'", "Invalid escape \\Q in literal");
2689 rust_lex_int_test ("23", 23, DECIMAL_INTEGER);
2690 rust_lex_int_test ("2_344__29", 234429, INTEGER);
2691 rust_lex_int_test ("0x1f", 0x1f, INTEGER);
2692 rust_lex_int_test ("23usize", 23, INTEGER);
2693 rust_lex_int_test ("23i32", 23, INTEGER);
2694 rust_lex_int_test ("0x1_f", 0x1f, INTEGER);
2695 rust_lex_int_test ("0b1_101011__", 0x6b, INTEGER);
2696 rust_lex_int_test ("0o001177i64", 639, INTEGER);
2698 rust_lex_test_trailing_dot ();
2700 rust_lex_test_one ("23.", FLOAT);
2701 rust_lex_test_one ("23.99f32", FLOAT);
2702 rust_lex_test_one ("23e7", FLOAT);
2703 rust_lex_test_one ("23E-7", FLOAT);
2704 rust_lex_test_one ("23e+7", FLOAT);
2705 rust_lex_test_one ("23.99e+7f64", FLOAT);
2706 rust_lex_test_one ("23.82f32", FLOAT);
2708 rust_lex_stringish_test ("hibob", "hibob", IDENT);
2709 rust_lex_stringish_test ("hibob__93", "hibob__93", IDENT);
2710 rust_lex_stringish_test ("thread", "thread", IDENT);
2712 rust_lex_stringish_test ("\"string\"", "string", STRING);
2713 rust_lex_stringish_test ("\"str\\ting\"", "str\ting", STRING);
2714 rust_lex_stringish_test ("\"str\\\"ing\"", "str\"ing", STRING);
2715 rust_lex_stringish_test ("r\"str\\ing\"", "str\\ing", STRING);
2716 rust_lex_stringish_test ("r#\"str\\ting\"#", "str\\ting", STRING);
2717 rust_lex_stringish_test ("r###\"str\\\"ing\"###", "str\\\"ing", STRING);
2719 rust_lex_stringish_test ("b\"string\"", "string", BYTESTRING);
2720 rust_lex_stringish_test ("b\"\x73tring\"", "string", BYTESTRING);
2721 rust_lex_stringish_test ("b\"str\\\"ing\"", "str\"ing", BYTESTRING);
2722 rust_lex_stringish_test ("br####\"\\x73tring\"####", "\\x73tring",
2725 for (i = 0; i < ARRAY_SIZE (identifier_tokens); ++i)
2726 rust_lex_test_one (identifier_tokens[i].name, identifier_tokens[i].value);
2728 for (i = 0; i < ARRAY_SIZE (operator_tokens); ++i)
2729 rust_lex_test_one (operator_tokens[i].name, operator_tokens[i].value);
2731 rust_lex_test_completion ();
2732 rust_lex_test_push_back ();
2734 obstack_free (&work_obstack, NULL);
2738 #endif /* GDB_SELF_TEST */
2741 _initialize_rust_exp (void)
2743 int code = regcomp (&number_regex, number_regex_text, REG_EXTENDED);
2744 /* If the regular expression was incorrect, it was a programming
2746 gdb_assert (code == 0);
2749 register_self_test (rust_lex_tests);