Updated Combined Program

combined_email_finder.py

@@ -0,0 +1,47 @@
+# combined_email_finder.py
+
+from emailfinder.core import processing
+from io import StringIO
+import sys
+import requests
+import re
+
+from email_finder import find_emails
+from emailfinder_wrapper import run_emailfinder
+
+# FILTERING FUNCTION
+def filter_emails(email_list):
+    filtered_emails = [
+        email for email in email_list if not (
+            email.startswith(('22', 'u0027', 'jsmith', 'jdoe', 'jane.doe', 'First',
+                              'John.Doe', 'FLast', 'doe', 'johnsmith', 'janedoe',
+                              'Last', 'j-doe', 'LFirst', 'Jane.Doe', 'Doe',
+                              'John_Smith', 'JSmith', 'JDoe', 'j_doe', 'J.Smith',
+                              'JohnSmith', 'John_Doe', 'j.doe', 'Smith.John',
+                              'jane@', 'd_jane', 'd-jane', 'jane_doe', 'jane.d', 'john.d'))            
+        )
+    ]
+
+    cleaned_emails = []
+    for email in filtered_emails:
+        if email in filtered_emails:
+            if email.startswith('u003'):
+                email = email[4:]
+            elif email.startswith('x3e'):
+                email = email[3:]
+            cleaned_emails.append(email)
+        
+    cleaned_emails = [email for email in cleaned_emails if not re.match(r'^[a-zA-Z]@', email)]
+    return cleaned_emails
+
+# METHOD COMBINATION
+
+def get_combined_emails(domain):
+    """
+    Returns a merged, deduplicated list of emails from both methods
+    """
+    direct_emails = find_emails(domain)
+    finder_emails = run_emailfinder(domain)
+
+    combined = set(direct_emails) | set(finder_emails)
+    return list(combined) if combined else ["No emails found."]

emailfinder_wrapper.py

@@ -1,19 +1,24 @@
 # emailfinder_wrapper.py
+
 from emailfinder.core import processing
 from io import StringIO
 import sys
+import re
 
 def run_emailfinder(domain):
     # Redirect stdout temporarily
     old_stdout = sys.stdout
     sys.stdout = mystdout = StringIO()
+    results = []
 
     try:
         processing(domain, proxies=None)
         output = mystdout.getvalue()
+        results = re.findall(r"[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\.[a-z]{2,}", output)
     except Exception as e:
-        output = f"Error: {e}"
+        results = [f"EmailFinder error: {e}"]
+    
+    finally:
+        sys.stdout = old_stdout
 
-    # Restore stdout
-    sys.stdout = old_stdout
-    return output
+    return results