From 0073fbd33299a7a3d154827464bf0b9f22ae3a67 Mon Sep 17 00:00:00 2001
From: Atin Bainada <61903527+atin@users.noreply.github.com>
Date: Sat, 24 Apr 2021 17:43:41 +0530
Subject: [PATCH] chore(Python): cleaning string before word count (#256)

---
 algorithms/Python/strings/word_count.py | 19 +++++++++++++++++--
 1 file changed, 17 insertions(+), 2 deletions(-)

diff --git a/algorithms/Python/strings/word_count.py b/algorithms/Python/strings/word_count.py
index b4b32613..6119fd70 100644
--- a/algorithms/Python/strings/word_count.py
+++ b/algorithms/Python/strings/word_count.py
@@ -1,10 +1,25 @@
+import re
+
+test_string = '"the" is the most used word in the English language'
+
+
 def word_count(s: str) -> int:
+    """
+    >>> word_count(test_string)
+    10
+    """
+    s = re.sub('[^A-Za-z0-9 ]+', '', s)
     return len(s.lower().split())
 
 
 def unique_word_count(s: str) -> int:
+    """
+    >>> unique_word_count(test_string)
+    8
+    """
+    s = re.sub('[^A-Za-z0-9 ]+', '', s)
     return len(set(s.lower().split()))
-  
 
-for s in ("The Matrix", "To Be or Not to Be", "Kiss Kiss Bang Bang"):
+
+for s in ("The Matrix", "To Be or Not to Be", "Kiss Kiss Bang Bang", test_string):
     print(s, word_count(s), unique_word_count(s))