From 8bc9e5d9cfe634191c83bd735af4571525133cfe Mon Sep 17 00:00:00 2001
From: Chris Dyer <cdyer@allegro.clab.cs.cmu.edu>
Date: Wed, 5 Dec 2012 20:27:30 -0500
Subject: remove logging, you should be using pv

---
 corpus/support/quote-norm.pl | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)

(limited to 'corpus/support/quote-norm.pl')

diff --git a/corpus/support/quote-norm.pl b/corpus/support/quote-norm.pl
index 0c5b9c26..72b0064d 100755
--- a/corpus/support/quote-norm.pl
+++ b/corpus/support/quote-norm.pl
@@ -18,13 +18,18 @@ while(<STDIN>) {
   s/(\W)(euro?)(\d*\.\d+|\d+)/$1EUR $3/gi;
   s/&\s*#45\s*;\s*&\s*#45\s*;/--/g;
   s/&\s*#45\s*;/--/g;
+  s/�c/--/g;
   s/ ,,/ "/g;
   s/``/"/g;
   s/''/"/g;
+  s/[「」]/"/g;
   s/〃/"/g;
   s/¨/"/g;
   s/¡/ ¡ /g;
   s/¿/ ¿ /g;
+  # â<U+0080><U+0099>
+  s/â(\x{80}\x{99}|\x{80}\x{98})/'/g;
+  s/â(\x{80}\x{9c}|\x{80}\x{9d})/"/g;
   s/ˇ/'/g;
   s/´/'/g;
   s/`/'/g;
@@ -39,7 +44,7 @@ while(<STDIN>) {
   s/»/"/g;
   tr/！-～/!-~/;
   s/、/,/g;
-  s/。/./g;
+  # s/。/./g;
   s/…/.../g;
   s/―/--/g;
   s/–/--/g;
-- 
cgit v1.2.3