diff options
author | Tom Tromey <tom@tromey.com> | 2019-03-24 21:30:56 -0600 |
---|---|---|
committer | Tom Tromey <tom@tromey.com> | 2019-04-04 19:55:11 -0600 |
commit | 5776fca307b8af3d852525b77e9b917a9aa97370 (patch) | |
tree | ac923aa97349376a076996f5f8db12b44e728cb5 /gdb/rust-exp.y | |
parent | 8621b685bfdcb8773b8177fb2b89e45499902868 (diff) | |
download | gdb-5776fca307b8af3d852525b77e9b917a9aa97370.zip gdb-5776fca307b8af3d852525b77e9b917a9aa97370.tar.gz gdb-5776fca307b8af3d852525b77e9b917a9aa97370.tar.bz2 |
Move lexptr and prev_lexptr to parser_state
This removes the lexptr and prev_lexptr globals, in favor of members
of parser_state. prev_lexptr could be isolated to each parser, but
since every parser uses it, that did not seem necessary.
gdb/ChangeLog
2019-04-04 Tom Tromey <tom@tromey.com>
* rust-exp.y (struct rust_parser) <lex_hex, lex_escape,
lex_operator, push_back>: New methods.
Update all rules.
(rust_parser::lex_hex, lex_escape): Rename and update.
(rust_parser::lex_string, rust_parser::lex_identifier): Update.
(rust_parser::lex_operator): Rename and update.
(rust_parser::lex_number, rustyylex, rustyyerror)
(rust_lex_test_init, rust_lex_test_sequence)
(rust_lex_test_push_back, rust_lex_tests): Update.
* parser-defs.h (struct parser_state) <parser_state>: Add "input"
parameter.
<lexptr, prev_lexptr>: New members.
(lexptr, prev_lexptr): Don't declare.
* parse.c (lexptr, prev_lexptr): Remove globals.
(parse_exp_in_context): Update.
* p-exp.y (yylex, yyerror): Update.
* m2-exp.y (parse_number, yylex, yyerror): Update.
* go-exp.y (lex_one_token, yyerror): Update.
* f-exp.y (match_string_literal, yylex, yyerror): Update.
* d-exp.y (lex_one_token, yyerror): Update.
* c-exp.y (scan_macro_expansion, finished_macro_expansion)
(lex_one_token, yyerror): Update.
* ada-lex.l (YY_INPUT): Update.
(rewind_to_char): Update.
* ada-exp.y (yyerror): Update.
Diffstat (limited to 'gdb/rust-exp.y')
-rw-r--r-- | gdb/rust-exp.y | 248 |
1 files changed, 130 insertions, 118 deletions
diff --git a/gdb/rust-exp.y b/gdb/rust-exp.y index 2f2cc8f..782117e 100644 --- a/gdb/rust-exp.y +++ b/gdb/rust-exp.y @@ -117,7 +117,6 @@ struct rust_parser; static int rustyylex (YYSTYPE *, rust_parser *); static void rustyyerror (rust_parser *parser, const char *msg); -static void rust_push_back (char c); static struct stoken make_stoken (const char *); static struct block_symbol rust_lookup_symbol (const char *name, const struct block *block, @@ -237,6 +236,10 @@ struct rust_parser int lex_number (YYSTYPE *lvalp); int lex_string (YYSTYPE *lvalp); int lex_identifier (YYSTYPE *lvalp); + uint32_t lex_hex (int min, int max); + uint32_t lex_escape (int is_byte); + int lex_operator (YYSTYPE *lvalp); + void push_back (char c); struct type *rust_lookup_type (const char *name, const struct block *block); std::vector<struct type *> convert_params_to_types (rust_op_vector *params); @@ -855,7 +858,7 @@ identifier_path_for_expr: | identifier_path_for_expr COLONCOLON '<' type_list RSH { $$ = parser->ast_path ($1->left.sval, $4); - rust_push_back ('>'); + parser->push_back ('>'); } ; @@ -898,7 +901,7 @@ identifier_path_for_type: | just_identifiers_for_type '<' type_list RSH { $$ = parser->ast_path ($1->left.sval, $3); - rust_push_back ('>'); + parser->push_back ('>'); } ; @@ -1111,8 +1114,8 @@ update_innermost_block (struct block_symbol sym) /* Lex a hex number with at least MIN digits and at most MAX digits. */ -static uint32_t -lex_hex (int min, int max) +uint32_t +rust_parser::lex_hex (int min, int max) { uint32_t result = 0; int len = 0; @@ -1120,18 +1123,18 @@ lex_hex (int min, int max) int check_max = min == max; while ((check_max ? len <= max : 1) - && ((lexptr[0] >= 'a' && lexptr[0] <= 'f') - || (lexptr[0] >= 'A' && lexptr[0] <= 'F') - || (lexptr[0] >= '0' && lexptr[0] <= '9'))) + && ((pstate->lexptr[0] >= 'a' && pstate->lexptr[0] <= 'f') + || (pstate->lexptr[0] >= 'A' && pstate->lexptr[0] <= 'F') + || (pstate->lexptr[0] >= '0' && pstate->lexptr[0] <= '9'))) { result *= 16; - if (lexptr[0] >= 'a' && lexptr[0] <= 'f') - result = result + 10 + lexptr[0] - 'a'; - else if (lexptr[0] >= 'A' && lexptr[0] <= 'F') - result = result + 10 + lexptr[0] - 'A'; + if (pstate->lexptr[0] >= 'a' && pstate->lexptr[0] <= 'f') + result = result + 10 + pstate->lexptr[0] - 'a'; + else if (pstate->lexptr[0] >= 'A' && pstate->lexptr[0] <= 'F') + result = result + 10 + pstate->lexptr[0] - 'A'; else - result = result + lexptr[0] - '0'; - ++lexptr; + result = result + pstate->lexptr[0] - '0'; + ++pstate->lexptr; ++len; } @@ -1149,65 +1152,65 @@ lex_hex (int min, int max) /* Lex an escape. IS_BYTE is true if we're lexing a byte escape; otherwise we're lexing a character escape. */ -static uint32_t -lex_escape (int is_byte) +uint32_t +rust_parser::lex_escape (int is_byte) { uint32_t result; - gdb_assert (lexptr[0] == '\\'); - ++lexptr; - switch (lexptr[0]) + gdb_assert (pstate->lexptr[0] == '\\'); + ++pstate->lexptr; + switch (pstate->lexptr[0]) { case 'x': - ++lexptr; + ++pstate->lexptr; result = lex_hex (2, 2); break; case 'u': if (is_byte) error (_("Unicode escape in byte literal")); - ++lexptr; - if (lexptr[0] != '{') + ++pstate->lexptr; + if (pstate->lexptr[0] != '{') error (_("Missing '{' in Unicode escape")); - ++lexptr; + ++pstate->lexptr; result = lex_hex (1, 6); /* Could do range checks here. */ - if (lexptr[0] != '}') + if (pstate->lexptr[0] != '}') error (_("Missing '}' in Unicode escape")); - ++lexptr; + ++pstate->lexptr; break; case 'n': result = '\n'; - ++lexptr; + ++pstate->lexptr; break; case 'r': result = '\r'; - ++lexptr; + ++pstate->lexptr; break; case 't': result = '\t'; - ++lexptr; + ++pstate->lexptr; break; case '\\': result = '\\'; - ++lexptr; + ++pstate->lexptr; break; case '0': result = '\0'; - ++lexptr; + ++pstate->lexptr; break; case '\'': result = '\''; - ++lexptr; + ++pstate->lexptr; break; case '"': result = '"'; - ++lexptr; + ++pstate->lexptr; break; default: - error (_("Invalid escape \\%c in literal"), lexptr[0]); + error (_("Invalid escape \\%c in literal"), pstate->lexptr[0]); } return result; @@ -1221,25 +1224,25 @@ rust_parser::lex_character (YYSTYPE *lvalp) int is_byte = 0; uint32_t value; - if (lexptr[0] == 'b') + if (pstate->lexptr[0] == 'b') { is_byte = 1; - ++lexptr; + ++pstate->lexptr; } - gdb_assert (lexptr[0] == '\''); - ++lexptr; + gdb_assert (pstate->lexptr[0] == '\''); + ++pstate->lexptr; /* This should handle UTF-8 here. */ - if (lexptr[0] == '\\') + if (pstate->lexptr[0] == '\\') value = lex_escape (is_byte); else { - value = lexptr[0] & 0xff; - ++lexptr; + value = pstate->lexptr[0] & 0xff; + ++pstate->lexptr; } - if (lexptr[0] != '\'') + if (pstate->lexptr[0] != '\'') error (_("Unterminated character literal")); - ++lexptr; + ++pstate->lexptr; lvalp->typed_val_int.val = value; lvalp->typed_val_int.type = get_type (is_byte ? "u8" : "char"); @@ -1285,15 +1288,15 @@ ends_raw_string (const char *str, int n) int rust_parser::lex_string (YYSTYPE *lvalp) { - int is_byte = lexptr[0] == 'b'; + int is_byte = pstate->lexptr[0] == 'b'; int raw_length; if (is_byte) - ++lexptr; - raw_length = starts_raw_string (lexptr); - lexptr += raw_length; - gdb_assert (lexptr[0] == '"'); - ++lexptr; + ++pstate->lexptr; + raw_length = starts_raw_string (pstate->lexptr); + pstate->lexptr += raw_length; + gdb_assert (pstate->lexptr[0] == '"'); + ++pstate->lexptr; while (1) { @@ -1301,29 +1304,30 @@ rust_parser::lex_string (YYSTYPE *lvalp) if (raw_length > 0) { - if (lexptr[0] == '"' && ends_raw_string (lexptr, raw_length - 1)) + if (pstate->lexptr[0] == '"' && ends_raw_string (pstate->lexptr, + raw_length - 1)) { /* Exit with lexptr pointing after the final "#". */ - lexptr += raw_length; + pstate->lexptr += raw_length; break; } - else if (lexptr[0] == '\0') + else if (pstate->lexptr[0] == '\0') error (_("Unexpected EOF in string")); - value = lexptr[0] & 0xff; + value = pstate->lexptr[0] & 0xff; if (is_byte && value > 127) error (_("Non-ASCII value in raw byte string")); obstack_1grow (&obstack, value); - ++lexptr; + ++pstate->lexptr; } - else if (lexptr[0] == '"') + else if (pstate->lexptr[0] == '"') { /* Make sure to skip the quote. */ - ++lexptr; + ++pstate->lexptr; break; } - else if (lexptr[0] == '\\') + else if (pstate->lexptr[0] == '\\') { value = lex_escape (is_byte); @@ -1334,15 +1338,15 @@ rust_parser::lex_string (YYSTYPE *lvalp) sizeof (value), sizeof (value), &obstack, translit_none); } - else if (lexptr[0] == '\0') + else if (pstate->lexptr[0] == '\0') error (_("Unexpected EOF in string")); else { - value = lexptr[0] & 0xff; + value = pstate->lexptr[0] & 0xff; if (is_byte && value > 127) error (_("Non-ASCII value in byte string")); obstack_1grow (&obstack, value); - ++lexptr; + ++pstate->lexptr; } } @@ -1382,27 +1386,27 @@ rust_identifier_start_p (char c) int rust_parser::lex_identifier (YYSTYPE *lvalp) { - const char *start = lexptr; + const char *start = pstate->lexptr; unsigned int length; const struct token_info *token; int i; - int is_gdb_var = lexptr[0] == '$'; + int is_gdb_var = pstate->lexptr[0] == '$'; - gdb_assert (rust_identifier_start_p (lexptr[0])); + gdb_assert (rust_identifier_start_p (pstate->lexptr[0])); - ++lexptr; + ++pstate->lexptr; /* For the time being this doesn't handle Unicode rules. Non-ASCII identifiers are gated anyway. */ - while ((lexptr[0] >= 'a' && lexptr[0] <= 'z') - || (lexptr[0] >= 'A' && lexptr[0] <= 'Z') - || lexptr[0] == '_' - || (is_gdb_var && lexptr[0] == '$') - || (lexptr[0] >= '0' && lexptr[0] <= '9')) - ++lexptr; + while ((pstate->lexptr[0] >= 'a' && pstate->lexptr[0] <= 'z') + || (pstate->lexptr[0] >= 'A' && pstate->lexptr[0] <= 'Z') + || pstate->lexptr[0] == '_' + || (is_gdb_var && pstate->lexptr[0] == '$') + || (pstate->lexptr[0] >= '0' && pstate->lexptr[0] <= '9')) + ++pstate->lexptr; - length = lexptr - start; + length = pstate->lexptr - start; token = NULL; for (i = 0; i < ARRAY_SIZE (identifier_tokens); ++i) { @@ -1419,28 +1423,28 @@ rust_parser::lex_identifier (YYSTYPE *lvalp) if (token->value == 0) { /* Leave the terminating token alone. */ - lexptr = start; + pstate->lexptr = start; return 0; } } else if (token == NULL && (strncmp (start, "thread", length) == 0 || strncmp (start, "task", length) == 0) - && space_then_number (lexptr)) + && space_then_number (pstate->lexptr)) { /* "task" or "thread" followed by a number terminates the parse, per gdb rules. */ - lexptr = start; + pstate->lexptr = start; return 0; } - if (token == NULL || (parse_completion && lexptr[0] == '\0')) + if (token == NULL || (parse_completion && pstate->lexptr[0] == '\0')) lvalp->sval = make_stoken (copy_name (start, length)); - if (parse_completion && lexptr[0] == '\0') + if (parse_completion && pstate->lexptr[0] == '\0') { /* Prevent rustyylex from returning two COMPLETE tokens. */ - prev_lexptr = lexptr; + pstate->prev_lexptr = pstate->lexptr; return COMPLETE; } @@ -1453,18 +1457,18 @@ rust_parser::lex_identifier (YYSTYPE *lvalp) /* Lex an operator. */ -static int -lex_operator (YYSTYPE *lvalp) +int +rust_parser::lex_operator (YYSTYPE *lvalp) { const struct token_info *token = NULL; int i; for (i = 0; i < ARRAY_SIZE (operator_tokens); ++i) { - if (strncmp (operator_tokens[i].name, lexptr, + if (strncmp (operator_tokens[i].name, pstate->lexptr, strlen (operator_tokens[i].name)) == 0) { - lexptr += strlen (operator_tokens[i].name); + pstate->lexptr += strlen (operator_tokens[i].name); token = &operator_tokens[i]; break; } @@ -1476,7 +1480,7 @@ lex_operator (YYSTYPE *lvalp) return token->value; } - return *lexptr++; + return *pstate->lexptr++; } /* Lex a number. */ @@ -1495,7 +1499,8 @@ rust_parser::lex_number (YYSTYPE *lvalp) int type_index = -1; int i; - match = regexec (&number_regex, lexptr, ARRAY_SIZE (subexps), subexps, 0); + match = regexec (&number_regex, pstate->lexptr, ARRAY_SIZE (subexps), + subexps, 0); /* Failure means the regexp is broken. */ gdb_assert (match == 0); @@ -1539,9 +1544,9 @@ rust_parser::lex_number (YYSTYPE *lvalp) a request for a trait method call, not a syntax error involving the floating point number "23.". */ gdb_assert (subexps[0].rm_eo > 0); - if (lexptr[subexps[0].rm_eo - 1] == '.') + if (pstate->lexptr[subexps[0].rm_eo - 1] == '.') { - const char *next = skip_spaces (&lexptr[subexps[0].rm_eo]); + const char *next = skip_spaces (&pstate->lexptr[subexps[0].rm_eo]); if (rust_identifier_start_p (*next) || *next == '.') { @@ -1559,7 +1564,8 @@ rust_parser::lex_number (YYSTYPE *lvalp) if (type_name == NULL) { gdb_assert (type_index != -1); - type_name_holder = std::string (lexptr + subexps[type_index].rm_so, + type_name_holder = std::string ((pstate->lexptr + + subexps[type_index].rm_so), (subexps[type_index].rm_eo - subexps[type_index].rm_so)); type_name = type_name_holder.c_str (); @@ -1570,16 +1576,16 @@ rust_parser::lex_number (YYSTYPE *lvalp) /* Copy the text of the number and remove the "_"s. */ std::string number; - for (i = 0; i < end_index && lexptr[i]; ++i) + for (i = 0; i < end_index && pstate->lexptr[i]; ++i) { - if (lexptr[i] == '_') + if (pstate->lexptr[i] == '_') could_be_decimal = 0; else - number.push_back (lexptr[i]); + number.push_back (pstate->lexptr[i]); } /* Advance past the match. */ - lexptr += subexps[0].rm_eo; + pstate->lexptr += subexps[0].rm_eo; /* Parse the number. */ if (is_integer) @@ -1627,18 +1633,22 @@ rust_parser::lex_number (YYSTYPE *lvalp) static int rustyylex (YYSTYPE *lvalp, rust_parser *parser) { + struct parser_state *pstate = parser->pstate; + /* Skip all leading whitespace. */ - while (lexptr[0] == ' ' || lexptr[0] == '\t' || lexptr[0] == '\r' - || lexptr[0] == '\n') - ++lexptr; + while (pstate->lexptr[0] == ' ' + || pstate->lexptr[0] == '\t' + || pstate->lexptr[0] == '\r' + || pstate->lexptr[0] == '\n') + ++pstate->lexptr; /* If we hit EOF and we're completing, then return COMPLETE -- maybe we're completing an empty string at the end of a field_expr. But, we don't want to return two COMPLETE tokens in a row. */ - if (lexptr[0] == '\0' && lexptr == prev_lexptr) + if (pstate->lexptr[0] == '\0' && pstate->lexptr == pstate->prev_lexptr) return 0; - prev_lexptr = lexptr; - if (lexptr[0] == '\0') + pstate->prev_lexptr = pstate->lexptr; + if (pstate->lexptr[0] == '\0') { if (parse_completion) { @@ -1648,49 +1658,49 @@ rustyylex (YYSTYPE *lvalp, rust_parser *parser) return 0; } - if (lexptr[0] >= '0' && lexptr[0] <= '9') + if (pstate->lexptr[0] >= '0' && pstate->lexptr[0] <= '9') return parser->lex_number (lvalp); - else if (lexptr[0] == 'b' && lexptr[1] == '\'') + else if (pstate->lexptr[0] == 'b' && pstate->lexptr[1] == '\'') return parser->lex_character (lvalp); - else if (lexptr[0] == 'b' && lexptr[1] == '"') + else if (pstate->lexptr[0] == 'b' && pstate->lexptr[1] == '"') return parser->lex_string (lvalp); - else if (lexptr[0] == 'b' && starts_raw_string (lexptr + 1)) + else if (pstate->lexptr[0] == 'b' && starts_raw_string (pstate->lexptr + 1)) return parser->lex_string (lvalp); - else if (starts_raw_string (lexptr)) + else if (starts_raw_string (pstate->lexptr)) return parser->lex_string (lvalp); - else if (rust_identifier_start_p (lexptr[0])) + else if (rust_identifier_start_p (pstate->lexptr[0])) return parser->lex_identifier (lvalp); - else if (lexptr[0] == '"') + else if (pstate->lexptr[0] == '"') return parser->lex_string (lvalp); - else if (lexptr[0] == '\'') + else if (pstate->lexptr[0] == '\'') return parser->lex_character (lvalp); - else if (lexptr[0] == '}' || lexptr[0] == ']') + else if (pstate->lexptr[0] == '}' || pstate->lexptr[0] == ']') { /* Falls through to lex_operator. */ --parser->paren_depth; } - else if (lexptr[0] == '(' || lexptr[0] == '{') + else if (pstate->lexptr[0] == '(' || pstate->lexptr[0] == '{') { /* Falls through to lex_operator. */ ++parser->paren_depth; } - else if (lexptr[0] == ',' && parser->pstate->comma_terminates + else if (pstate->lexptr[0] == ',' && pstate->comma_terminates && parser->paren_depth == 0) return 0; - return lex_operator (lvalp); + return parser->lex_operator (lvalp); } /* Push back a single character to be re-lexed. */ -static void -rust_push_back (char c) +void +rust_parser::push_back (char c) { /* Can't be called before any lexing. */ - gdb_assert (prev_lexptr != NULL); + gdb_assert (pstate->prev_lexptr != NULL); - --lexptr; - gdb_assert (*lexptr == c); + --pstate->lexptr; + gdb_assert (*pstate->lexptr == c); } @@ -2545,7 +2555,9 @@ rust_parse (struct parser_state *state) static void rustyyerror (rust_parser *parser, const char *msg) { - const char *where = prev_lexptr ? prev_lexptr : lexptr; + const char *where = (parser->pstate->prev_lexptr + ? parser->pstate->prev_lexptr + : parser->pstate->lexptr); error (_("%s in expression, near `%s'."), msg, where); } @@ -2558,8 +2570,8 @@ rustyyerror (rust_parser *parser, const char *msg) static void rust_lex_test_init (rust_parser *parser, const char *input) { - prev_lexptr = NULL; - lexptr = input; + parser->pstate->prev_lexptr = NULL; + parser->pstate->lexptr = input; parser->paren_depth = 0; } @@ -2635,7 +2647,7 @@ rust_lex_test_sequence (rust_parser *parser, const char *input, int len, { int i; - lexptr = input; + parser->pstate->lexptr = input; parser->paren_depth = 0; for (i = 0; i < len; ++i) @@ -2696,7 +2708,7 @@ rust_lex_test_push_back (rust_parser *parser) SELF_CHECK (token == COMPOUND_ASSIGN); SELF_CHECK (lval.opcode == BINOP_RSH); - rust_push_back ('='); + parser->push_back ('='); token = rustyylex (&lval, parser); SELF_CHECK (token == '='); @@ -2714,7 +2726,7 @@ rust_lex_tests (void) // Set up dummy "parser", so that rust_type works. struct parser_state ps (&rust_language_defn, target_gdbarch (), - nullptr, 0, 0); + nullptr, 0, 0, nullptr); rust_parser parser (&ps); rust_lex_test_one (&parser, "", 0); |