From 4088d67207ca76a4c8e847045fa13f348bf92c57 Mon Sep 17 00:00:00 2001
From: Karol Idaszak <karida@st.amu.edu.pl>
Date: Wed, 1 Jun 2022 10:59:15 +0200
Subject: [PATCH] Add preprocessing for replacing polish characters

---
 system_mockup/analizator_jezyka_naturalnego.py | 12 ++++++++++++
 1 file changed, 12 insertions(+)

diff --git a/system_mockup/analizator_jezyka_naturalnego.py b/system_mockup/analizator_jezyka_naturalnego.py
index e012297..ce8643a 100644
--- a/system_mockup/analizator_jezyka_naturalnego.py
+++ b/system_mockup/analizator_jezyka_naturalnego.py
@@ -46,4 +46,16 @@ def analizator_jezyka_naturalnego(text):
 
 def text_preprocess(text):
     text = text.lower()
+    text = text.replace("ą", "a")
+    text = text.replace("ć", "c")
+    text = text.replace("ę", "e")
+    text = text.replace("ł", "l")
+    text = text.replace("ń", "n")
+    text = text.replace("ó", "o")
+    text = text.replace("ś", "s")
+    text = text.replace("ź", "z")
+    text = text.replace("ż", "z")
+    text = text.replace("\n", " ")
+    text = text.replace("\t", " ")
+    text = text.replace("  ", " ")
     return text