diff options
author | Ulrich Drepper <drepper@redhat.com> | 2003-11-21 23:42:18 +0000 |
---|---|---|
committer | Ulrich Drepper <drepper@redhat.com> | 2003-11-21 23:42:18 +0000 |
commit | f0d77aa8af0a5c13bb69385c27a137199b748b7a (patch) | |
tree | 8df29f1f0b04cc8d9afbe2b698ce73f5515fd4a2 /posix/regcomp.c | |
parent | 73163a09732e60a6c16fcf73635091acd307bef4 (diff) | |
download | glibc-f0d77aa8af0a5c13bb69385c27a137199b748b7a.zip glibc-f0d77aa8af0a5c13bb69385c27a137199b748b7a.tar.gz glibc-f0d77aa8af0a5c13bb69385c27a137199b748b7a.tar.bz2 |
Update.
2003-11-21 Ulrich Drepper <drepper@redhat.com>
* posix/regcomp.c (fetch_token): Change interface to match
peek_token. This avoid some copying and reduces code size.
Diffstat (limited to 'posix/regcomp.c')
-rw-r--r-- | posix/regcomp.c | 36 |
1 files changed, 17 insertions, 19 deletions
diff --git a/posix/regcomp.c b/posix/regcomp.c index c94a440..7c43359 100644 --- a/posix/regcomp.c +++ b/posix/regcomp.c @@ -51,7 +51,8 @@ static reg_errcode_t calc_eclosure_iter (re_node_set *new_set, re_dfa_t *dfa, static void calc_inveclosure (re_dfa_t *dfa); static int fetch_number (re_string_t *input, re_token_t *token, reg_syntax_t syntax); -static re_token_t fetch_token (re_string_t *input, reg_syntax_t syntax); +static void fetch_token (re_token_t *result, re_string_t *input, + reg_syntax_t syntax); static int peek_token (re_token_t *token, re_string_t *input, reg_syntax_t syntax); static int peek_token_bracket (re_token_t *token, re_string_t *input, @@ -1588,16 +1589,13 @@ calc_eclosure_iter (new_set, dfa, node, root) /* Fetch a token from INPUT. We must not use this function inside bracket expressions. */ -static re_token_t -fetch_token (input, syntax) +static void +fetch_token (result, input, syntax) + re_token_t *result; re_string_t *input; reg_syntax_t syntax; { - re_token_t token; - int consumed_byte; - consumed_byte = peek_token (&token, input, syntax); - re_string_skip_bytes (input, consumed_byte); - return token; + re_string_skip_bytes (input, peek_token (result, input, syntax)); } /* Peek a token from INPUT, and return the length of the token. @@ -1927,7 +1925,7 @@ parse (regexp, preg, syntax, err) re_dfa_t *dfa = (re_dfa_t *) preg->buffer; bin_tree_t *tree, *eor, *root; re_token_t current_token; - current_token = fetch_token (regexp, syntax | RE_CARET_ANCHORS_HERE); + fetch_token (¤t_token, regexp, syntax | RE_CARET_ANCHORS_HERE); tree = parse_reg_exp (regexp, preg, ¤t_token, syntax, 0, err); if (BE (*err != REG_NOERROR && tree == NULL, 0)) return NULL; @@ -1971,7 +1969,7 @@ parse_reg_exp (regexp, preg, token, syntax, nest, err) while (token->type == OP_ALT) { re_token_t alt_token = *token; - *token = fetch_token (regexp, syntax | RE_CARET_ANCHORS_HERE); + fetch_token (token, regexp, syntax | RE_CARET_ANCHORS_HERE); if (token->type != OP_ALT && token->type != END_OF_RE && (nest == 0 || token->type != OP_CLOSE_SUBEXP)) { @@ -2073,7 +2071,7 @@ parse_expression (regexp, preg, token, syntax, nest, err) && !re_string_first_byte (regexp, re_string_cur_idx (regexp))) { bin_tree_t *mbc_remain; - *token = fetch_token (regexp, syntax); + fetch_token (token, regexp, syntax); mbc_remain = re_dfa_add_tree_node (dfa, NULL, NULL, *token); tree = create_tree (dfa, tree, mbc_remain, CONCAT, 0); if (BE (mbc_remain == NULL || tree == NULL, 0)) @@ -2129,7 +2127,7 @@ parse_expression (regexp, preg, token, syntax, nest, err) } else if (syntax & RE_CONTEXT_INDEP_OPS) { - *token = fetch_token (regexp, syntax); + fetch_token (token, regexp, syntax); return parse_expression (regexp, preg, token, syntax, nest, err); } /* else fall through */ @@ -2188,7 +2186,7 @@ parse_expression (regexp, preg, token, syntax, nest, err) by repetition operators. eg. RE"^*" is invalid or "<ANCHOR(^)><CHAR(*)>", it must not be "<ANCHOR(^)><REPEAT(*)>". */ - *token = fetch_token (regexp, syntax); + fetch_token (token, regexp, syntax); return tree; case OP_PERIOD: tree = re_dfa_add_tree_node (dfa, NULL, NULL, *token); @@ -2233,7 +2231,7 @@ parse_expression (regexp, preg, token, syntax, nest, err) #endif return NULL; } - *token = fetch_token (regexp, syntax); + fetch_token (token, regexp, syntax); while (token->type == OP_DUP_ASTERISK || token->type == OP_DUP_PLUS || token->type == OP_DUP_QUESTION || token->type == OP_OPEN_DUP_NUM) @@ -2298,7 +2296,7 @@ parse_sub_exp (regexp, preg, token, syntax, nest, err) return NULL; } dfa->nodes[left_par->node_idx].opr.idx = cur_nsub; - *token = fetch_token (regexp, syntax | RE_CARET_ANCHORS_HERE); + fetch_token (token, regexp, syntax | RE_CARET_ANCHORS_HERE); /* The subexpression may be a null string. */ if (token->type == OP_CLOSE_SUBEXP) @@ -2378,7 +2376,7 @@ parse_dup_op (dup_elem, regexp, dfa, token, syntax, err) if (BE (start == 0 && end == 0, 0)) { /* We treat "<re>{0}" and "<re>{0,0}" as null string. */ - *token = fetch_token (regexp, syntax); + fetch_token (token, regexp, syntax); return NULL; } @@ -2457,7 +2455,7 @@ parse_dup_op (dup_elem, regexp, dfa, token, syntax, err) return NULL; } } - *token = fetch_token (regexp, syntax); + fetch_token (token, regexp, syntax); return tree; parse_dup_op_espace: @@ -3591,7 +3589,7 @@ fetch_number (input, token, syntax) unsigned char c; while (1) { - *token = fetch_token (input, syntax); + fetch_token (token, input, syntax); c = token->opr.c; if (BE (token->type == END_OF_RE, 0)) return -2; @@ -3669,7 +3667,7 @@ re_dfa_add_tree_node (dfa, left, right, token) bin_tree_t *left; bin_tree_t *right; re_token_t token; -{ +{ int new_idx = re_dfa_add_node (dfa, token, 0); if (new_idx == -1) |