Update app.py

app.py

@@ -4,6 +4,70 @@ import matplotlib.pyplot as plt
 import seaborn as sns
 import plotly.express as px
 
+num_rows = 20000  
+df = pd.read_csv('emails.csv', on_bad_lines='skip', nrows=num_rows)
+
+def get_message(Series: pd.Series):
+    result = pd.Series(index=Series.index)
+    for row, message in enumerate(Series):
+        message_words = message.split('\n')
+        del message_words[:15]
+        result.iloc[row] = ''.join(message_words).strip()
+    return result
+
+def get_date(Series: pd.Series):
+    result = pd.Series(index=Series.index)
+    for row, message in enumerate(Series):
+        message_words = message.split('\n')
+        del message_words[0]
+        del message_words[1:]
+        result.iloc[row] = ''.join(message_words).strip()
+        result.iloc[row] = result.iloc[row].replace('Date: ', '')
+    print('Done parsing, converting to datetime format..')
+    return pd.to_datetime(result)
+
+def get_sender_and_receiver(Series: pd.Series):
+    sender = pd.Series(index = Series.index)
+    recipient1 = pd.Series(index = Series.index)
+    recipient2 = pd.Series(index = Series.index)
+    recipient3 = pd.Series(index = Series.index)
+
+    for row,message in enumerate(Series):
+        message_words = message.split('\n')
+        sender[row] = message_words[2].replace('From: ', '')
+        recipient1[row] = message_words[3].replace('To: ', '')
+        recipient2[row] = message_words[10].replace('X-cc: ', '')
+        recipient3[row] = message_words[11].replace('X-bcc: ', '')
+
+    return sender, recipient1, recipient2, recipient3
+
+def get_subject(Series: pd.Series):
+    result = pd.Series(index = Series.index)
+
+    for row, message in enumerate(Series):
+        message_words = message.split('\n')
+        message_words = message_words[4]
+        result[row] = message_words.replace('Subject: ', '')
+    return result
+
+def get_folder(Series: pd.Series):
+    result = pd.Series(index = Series.index)
+
+    for row, message in enumerate(Series):
+        message_words = message.split('\n')
+        message_words = message_words[12]
+        result[row] = message_words.replace('X-Folder: ', '')
+    return result
+
+df['text'] = get_message(df.message)
+df['sender'], df['recipient1'], df['recipient2'], df['recipient3'] = get_sender_and_receiver(df.message)
+df['Subject'] = get_subject(df.message)
+df['folder'] = get_folder(df.message)
+df['date'] = get_date(df.message)
+
+df = df.drop(['message', 'file'], axis = 1)
+
+
 import chromadb
 chroma_client = chromadb.Client()