add script to extract top mails

read_topmails.py

@@ -0,0 +1,50 @@
+#!/usr/bin/python3
+import mailbox
+import jinja2
+import email.header
+import email.utils
+import yaml
+import datetime
+import sys, os
+import pypandoc
+import argparse
+import quopri
+import pytz
+from pprint import pprint
+from dateutil import parser
+
+MAILDIR = "/home/yannik/mail/INBOX.Mailinglisten.FSMI/"
+CONFIG_FILE = "generator.conf"
+
+def decode_header(header):
+    decoded_header = email.header.decode_header(header)[0]
+    encoding = decoded_header[1] or "ascii"
+    if encoding == "unknown-8bit":
+        encoding = "ascii"
+    return decoded_header[0].decode(encoding, errors="replace") if isinstance(decoded_header[0], bytes) else decoded_header[0]
+
+if __name__ == "__main__":
+    config = yaml.full_load(open(CONFIG_FILE))
+    in_mbox = mailbox.Maildir(MAILDIR)
+    out_mbox = mailbox.mbox(config["top_mbox_file"])
+
+    out_mbox.clear()
+
+    buffer = []
+
+    timezone = pytz.timezone("Europe/Berlin")
+    last_fsr_date = datetime.date.fromisoformat(open("data/last_date").read().strip())
+    last_fsr_datetime = datetime.datetime.combine(last_fsr_date, datetime.time(17, 30), timezone)
+
+    for message in in_mbox:
+        if message["List-Id"]:
+            if "top.fsmi.uni-karlsruhe.de" in decode_header(message["List-Id"]).strip():
+                date = parser.parse(message["Date"])
+                if date > last_fsr_datetime:
+                    buffer.append(message)
+
+    for message in sorted(buffer, key=lambda x: parser.parse(x["Date"])):
+        out_mbox.add(message)
+    
+    out_mbox.close()
+