bsgrep: cosmetics, docs.

1 files changed, 10 insertions, 7 deletions

diff --git a/bsgrep b/bsgrep
index 65d138c..bae517d 100755
--- a/bsgrep
+++ b/bsgrep
@@ -130,15 +130,13 @@ while(<>) {
 		}
 	}
 	if(/$cont\s+$/) {
-		warn "$self: $ARGV, line $.: whitespace after continuation, malformed input?\n" unless $opt{s};
+		warn "$self: $ARGV:$.: whitespace after continuation, malformed input?\n" unless $opt{s};
 	}
 	s/^\s+// if $out && $opt{w};
 	$start_line = $. unless defined $out;
 	$out .= $_;
 	if(/$cont$/) {
-		if($opt{k}) {
-			$out .= "\n";
-		} else {
+		if(!$opt{k}) {
 			$out =~ s/$cont$//;
 		}
 	} else {
@@ -149,7 +147,7 @@ while(<>) {
 	# reset $. on each new file (perldoc -f eof)
 	if(eof) {
 		if($out) {
-			warn "$self: $ARGV: last line ends with continuation\n" unless $opt{s};
+			warn "$self: $ARGV:$.: last line ends with continuation\n" unless $opt{s};
 			handle_line($out);
 			undef $out;
 		}
@@ -189,6 +187,9 @@ options, and it uses Perl regular expressions rather than POSIX.
 Input is read from one or more files, or standard input if no files
 are given. Output goes to standard output.
 
+The search is done after lines are joined together, so the regex can
+match text split across continuation lines.
+
 If B<bsgrep> is run as B<bsjoin> (via symbolic or hard link, or just
 copying the executable), it will simply join together continued lines
 without searching for anything. In this mode, only the B<-k>, B<-n>,
@@ -206,12 +207,14 @@ Use I<char> as the continuation character, rather than a backslash.
 
 =item -k
 
-Keep the backslashes and newlines when joining continued lines together.
+Keep the continuation characters when joining continued lines together.
 This option does not exist in B<grep>.
 
 =item -n
 
-Prefix output lines with line numbers (same as B<grep>).
+Prefix output lines with line numbers (same as B<grep>). For lines
+that are split with continuation characters, the line number will be
+that of the first line in the set.
 
 =item -w