summaryrefslogtreecommitdiff
path: root/corpus/paste-files.pl
diff options
context:
space:
mode:
Diffstat (limited to 'corpus/paste-files.pl')
-rwxr-xr-xcorpus/paste-files.pl12
1 files changed, 11 insertions, 1 deletions
diff --git a/corpus/paste-files.pl b/corpus/paste-files.pl
index 24c70599..0b788386 100755
--- a/corpus/paste-files.pl
+++ b/corpus/paste-files.pl
@@ -17,6 +17,7 @@ for my $file (@ARGV) {
binmode(STDOUT,":utf8");
binmode(STDERR,":utf8");
+my $bad = 0;
my $lc = 0;
my $done = 0;
my $fl = 0;
@@ -34,7 +35,15 @@ while(1) {
last;
}
chomp $r;
- die "$ARGV[$anum]:$lc contains a ||| symbol - please remove.\n" if $r =~ /\|\|\|/;
+ if ($r =~ /\|\|\|/) {
+ $r = '';
+ $bad++;
+ }
+ warn "$ARGV[$anum]:$lc contains a ||| symbol - please remove.\n" if $r =~ /\|\|\|/;
+ $r =~ s/\|\|\|/ /g;
+ $r =~ s/ +//g;
+ $r =~ s/^ //;
+ $r =~ s/ $//;
$anum++;
push @line, $r;
}
@@ -47,4 +56,5 @@ for (my $i = 1; $i < scalar @fhs; $i++) {
my $r = <$fh>;
die "Mismatched number of lines.\n" if defined $r;
}
+print STDERR "Bad lines containing ||| were $bad\n";