Created a script to update the csv imports that don't have the date

extract_leads.py

@@ -30,6 +30,7 @@ class Lead:
     anrede: Optional[str] = None  # Salutation
     land: Optional[str] = None  # Country
     privacy: Optional[bool] = None
+    received_date: Optional[str] = None
 
 
 def parse_mbox_file(filepath: str) -> List[Lead]:
@@ -64,12 +65,48 @@ def parse_mbox_file(filepath: str) -> List[Lead]:
         # Extract lead data from body
         lead = parse_email_body(body)
 
+        # Extract received date from headers
+        try:
+            lead.received_date = extract_received_date(headers)
+        except ValueError as e:
+            print(f"WARNING: {e}")
+            raise
+
         if lead.name or lead.mail:  # Only add if we have some data
             leads.append(lead)
 
     return leads
 
 
+def extract_received_date(headers: str) -> Optional[str]:
+    """
+    Extract the Date header from email headers and convert to ISO format.
+
+    Args:
+        headers: Email headers section
+
+    Returns:
+        ISO format date string from the Date header, or None if not found
+
+    Raises:
+        ValueError: If Date header cannot be parsed to ISO format
+    """
+    from email.utils import parsedate_to_datetime
+
+    for line in headers.split('\n'):
+        if line.startswith('Date:'):
+            # Extract everything after "Date: "
+            date_value = line[6:].strip()
+            try:
+                # Parse the RFC 2822 date format and convert to ISO format
+                dt = parsedate_to_datetime(date_value)
+                return dt.isoformat()
+            except (TypeError, ValueError) as e:
+                # Raise exception so parsing failures are caught and reported
+                raise ValueError(f"Failed to parse date '{date_value}': {e}")
+    return None
+
+
 def parse_email_body(body: str) -> Lead:
     """
     Parse the body of an email to extract lead information.
@@ -172,7 +209,8 @@ def export_to_csv(leads: List[Lead], output_file: str) -> None:
         'device',
         'anrede',
         'land',
-        'privacy'
+        'privacy',
+        'received_date'
     ]
 
     with open(output_file, 'w', newline='', encoding='utf-8') as f: