grep -w: forward to next possible position after rejected match

grep -w accepts matches between non-word characters, only.  If a match
from regexec() doesn't meet this criteria, grep continues its search
after the first character of that match.

We can be a bit smarter here and skip all positions that follow a word
character first, as they can't match our criteria.  This way we can
consume characters quite cheaply and don't need to special-case the
handling of the beginning of a line.

Here's a contrived example command on msysgit (best of five runs):

	$ time git grep -w ...... v1.6.1 >/dev/null

	real    0m1.611s
	user    0m0.000s
	sys     0m0.015s

With the patch it's quite a bit faster:

	$ time git grep -w ...... v1.6.1 >/dev/null

	real    0m1.179s
	user    0m0.000s
	sys     0m0.015s

More common search patterns will gain a lot less, but it's a nice clean
up anyway.

Signed-off-by: Rene Scharfe <rene.scharfe@lsrfire.ath.cx>
Signed-off-by: Junio C Hamano <gitster@pobox.com>

diff --git a/grep.c b/grep.c
index 49e93199658e6b9214839232b5b464b7d9e76a7f..22a56b5d543d7924d4bd76d3d220b026b23ed3a9 100644 (file)
--- a/grep.c
+++ b/grep.c
@@ -294,7 +294,6 @@ static struct {
 static int match_one_pattern(struct grep_opt *opt, struct grep_pat *p, char *bol, char *eol, enum grep_context ctx)
 {
        int hit = 0;
-       int at_true_bol = 1;
        int saved_ch = 0;
        regmatch_t pmatch[10];
 
@@ -337,7 +336,7 @@ static int match_one_pattern(struct grep_opt *opt, struct grep_pat *p, char *bol
                 * either end of the line, or at word boundary
                 * (i.e. the next char must not be a word char).
                 */
-               if ( ((pmatch[0].rm_so == 0 && at_true_bol) ||
+               if ( ((pmatch[0].rm_so == 0) ||
                      !word_char(bol[pmatch[0].rm_so-1])) &&
                     ((pmatch[0].rm_eo == (eol-bol)) ||
                      !word_char(bol[pmatch[0].rm_eo])) )
@@ -349,10 +348,14 @@ static int match_one_pattern(struct grep_opt *opt, struct grep_pat *p, char *bol
                        /* There could be more than one match on the
                         * line, and the first match might not be
                         * strict word match.  But later ones could be!
+                        * Forward to the next possible start, i.e. the
+                        * next position following a non-word char.
                         */
                        bol = pmatch[0].rm_so + bol + 1;
-                       at_true_bol = 0;
-                       goto again;
+                       while (word_char(bol[-1]) && bol < eol)
+                               bol++;
+                       if (bol < eol)
+                               goto again;
                }
        }
        if (p->token == GREP_PATTERN_HEAD && saved_ch)