From e0e8b45b51673a4d9da3e83b8deac045e3aed04c Mon Sep 17 00:00:00 2001
From: Eliran Wong <hkwebtech@gmail.com>
Date: Wed, 22 Aug 2018 00:04:03 +0100
Subject: [PATCH] Add files via upload

---
 Script/removeGreekAccents.py | 40 ++++++++++++++++++++++++++++++++++++
 1 file changed, 40 insertions(+)
 create mode 100644 Script/removeGreekAccents.py

diff --git a/Script/removeGreekAccents.py b/Script/removeGreekAccents.py
new file mode 100644
index 0000000..35bbd86
--- /dev/null
+++ b/Script/removeGreekAccents.py
@@ -0,0 +1,40 @@
+import re
+
+inputFile = 'berean_accented.csv'
+outputFile = 'berean_unaccented.csv'
+
+f = open(inputFile,'r')
+newData = f.read()
+f.close()
+
+# Greek unicode characters
+
+newData = re.sub('[ἀἄᾄἂἆἁἅᾅἃάᾴὰᾶᾷᾳ]', 'α', newData)
+newData = re.sub('[ἈἌἎἉἍἋ]', 'Α', newData)
+
+newData = re.sub('[ἐἔἑἕἓέὲ]', 'ε', newData)
+newData = re.sub('[ἘἜἙἝἛ]', 'Ε', newData)
+
+newData = re.sub('[ἠἤᾔἢἦᾖᾐἡἥἣἧᾗᾑήῄὴῆῇῃ]', 'η', newData)
+newData = re.sub('[ἨἬἪἮἩἭἫ]', 'Η', newData)
+
+newData = re.sub('[ἰἴἶἱἵἳἷίὶῖϊΐῒ]', 'ι', newData)
+newData = re.sub('[ἸἼἹἽ]', 'Ι', newData)
+
+newData = re.sub('[ὀὄὂὁὅὃόὸ]', 'ο', newData)
+newData = re.sub('[ὈὌὉὍὋ]', 'Ο', newData)
+
+newData = re.sub('[ῥ]', 'ρ', newData)
+newData = re.sub('[Ῥ]', 'Ρ', newData)
+
+newData = re.sub('[ὐὔὒὖὑὕὓὗύὺῦϋΰῢ]', 'υ', newData)
+newData = re.sub('[ὙὝὟ]', 'Υ', newData)
+
+newData = re.sub('[ὠὤὢὦᾠὡὥὧᾧώῴὼῶῷῳ]', 'ω', newData)
+newData = re.sub('[ὨὬὪὮὩὭὯ]', 'Ω', newData)
+
+newData = re.sub("[\-\—\,\;\:\\\?\.\·\·\'\‘\’\‹\›\“\”\«\»\(\)\[\]\{\}\⧼\⧽\〈\〉\*\‿\᾽\⇔\¦]", "", newData)
+
+f = open(outputFile,'w')
+f.write(newData)
+f.close()
\ No newline at end of file