From b0a1359401f7b34ab46045b88ee6974e0f395652 Mon Sep 17 00:00:00 2001
From: Isaac Riley <isaac.r.riley@gmail.com>
Date: Wed, 29 Jan 2020 18:19:38 +0100
Subject: [PATCH] added small cleaning step

---
 aligner/bitext_align.py | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/aligner/bitext_align.py b/aligner/bitext_align.py
index 091c5c1..c9a086a 100644
--- a/aligner/bitext_align.py
+++ b/aligner/bitext_align.py
@@ -20,6 +20,8 @@ translate_client = translate.Client()
 def master_align(text0, text1, lang0, lang1): 
     """ Takes two equivalent texts (original and trnslation) and returns 
         aligned texts. """
+    text0 = re.sub(' ?. . . ?| … ?| ?... ?', '… ', text0)
+    text1 = re.sub(' ?. . . ?| … ?| ?... ?', '… ', text1)
     df0 = frame_from_text(text0, lang0, lang1)
     # print('A')
     df1 = frame_from_text(text1, lang1, lang0, is1=True)