vis

commit d26ac2103119d7b809c15b980eb34f5974c5fac9
parent 821c113695540f79cbd908a5aec17207f0bc86f5
Author: Marc André Tanner <mat@brain-dump.org>
Date:   Wed, 13 Apr 2016 17:20:18 +0200

sam: improve ^ matching start of line

There are some nasty differences between the meaning of ^ in
Plan 9's regexp library and POSIX when using REG_NEWLINE.
The former only matches at the beginning of a line wheras the
latter matches the zero-length string immediately after a
newline character \n. As a result this also matches after the
very last newline at the end of the file.

This is undesired behavior for a command like :x/^/c/#/
Hence we try to filter out this last match.

Close #264

Diffstat:
M
sam.c
 | 
14
++++++++++----

1 file changed, 10 insertions(+), 4 deletions(-)
diff --git a/sam.c b/sam.c
@@ -833,7 +833,8 @@ static bool cmd_extract(Vis *vis, Win *win, Command *cmd, const char *argv[], Cu
 		RegexMatch match[1];
 		while (start < end) {
 			bool found = text_search_range_forward(txt, start,
-				end - start, cmd->regex, 1, match, 0) == 0;
+				end - start, cmd->regex, 1, match,
+				start > range->start ? REG_NOTBOL : 0) == 0;
 			Filerange r = text_range_empty();
 			if (found) {
 				if (argv[0][0] == 'x')
@@ -845,10 +846,15 @@ static bool cmd_extract(Vis *vis, Win *win, Command *cmd, const char *argv[], Cu
 						start++;
 						continue;
 					}
-					start = match[0].end+1;
-				} else {
-					start = match[0].end;
+					/* in Plan 9's regexp library ^ matches the beginning
+					 * of a line, however in POSIX with REG_NEWLINE ^
+					 * matches the zero-length string immediately after a
+					 * newline. Try filtering out the last such match at EOF.
+					 */
+					if (end == match[0].start && start > range->start)
+						break;
 				}
+				start = match[0].end;
 			} else {
 				if (argv[0][0] == 'y')
 					r = text_range_new(start, end);