From e1f71a6ce868d116f04082b697a8d61afcd625f1 Mon Sep 17 00:00:00 2001
From: Chris Dyer <cdyer@allegro.clab.cs.cmu.edu>
Date: Mon, 21 Jan 2013 16:53:05 -0500
Subject: a little bit of cleanup

---
 corpus/filter-length.pl | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

(limited to 'corpus/filter-length.pl')

diff --git a/corpus/filter-length.pl b/corpus/filter-length.pl
index 38837f14..2e257cda 100755
--- a/corpus/filter-length.pl
+++ b/corpus/filter-length.pl
@@ -44,11 +44,11 @@ while(<F>) {
   $lines++;
   if ($lines % 100000 == 0) { print STDERR " [$lines]\n"; }
   elsif ($lines % 2500 == 0) { print STDERR "."; }
-  my ($sf, $se, @d) = split / \|\|\| /;
+  my ($sf, $se, @d) = split /\s*\|\|\|\s*/;
   if (scalar @d != 0 or !defined $se) {
     $bad_format++;
     if ($bad_format > 100 && ($bad_format / $lines) > 0.02) {
-      die "Corpus appears to be incorretly formatted, example: $_";
+      die "$bad_format / $lines : Corpus appears to be incorretly formatted, example: $_";
     }
     next;
   }
-- 
cgit v1.2.3