Update clients.py

clients.py

@@ -10,6 +10,35 @@ import streamlit as st
 import pandas as pd
 import requests
 import os
+import pandas as pd
+
+def add_https_to_urls(df, column_name):
+    """
+    Adds 'https://' to URLs in the specified column of a DataFrame if they don't already start with a valid protocol.
+    Corrects URLs starting with 'http:/' or 'https:/'.
+    Handles missing values, trims whitespace, and is case-insensitive.
+
+    Parameters:
+    df (pandas.DataFrame): The DataFrame containing the URLs.
+    column_name (str): The name of the column with URLs.
+    """
+    # Define a helper function to add or correct protocols
+    def correct_protocol(url):
+        if pd.isna(url) or url.strip() == '':
+            return url  # Return as is if the URL is NaN or empty
+        url = url.strip()  # Trim whitespace
+        lower_url = url.lower()
+        if lower_url.startswith('http:/') and not lower_url.startswith('http://'):
+            return 'http://' + url[6:]
+        elif lower_url.startswith('https:/') and not lower_url.startswith('https://'):
+            return 'https://' + url[7:]
+        elif not lower_url.startswith(('http://', 'https://')):
+            return 'https://' + url
+        return url
+
+    # Apply the helper function to the specified column
+    df[column_name] = df[column_name].apply(correct_protocol)
+    return df
 
 
 host=os.getenv("backend")
@@ -92,6 +121,7 @@ def CompanySpecificClient(email_receiver):
             df = df.dropna().loc[~(df == '').all(axis=1)]
             
         df = df.dropna().loc[~(df == '').all(axis=1)]  
+        df=add_https_to_urls(df, 'Website')
         st.write(df)    
         # Convert DataFrame to CSV for transmission
         csv = df.to_csv(index=False)