prevent duplicate comments by keeping a cache

1 files changed, 10 insertions, 0 deletions

diff --git a/email_helpers.py b/email_helpers.py
index 8f5b2ac..eeb2168 100644
--- a/email_helpers.py
+++ b/email_helpers.py
@@ -121,6 +121,7 @@ def extract_response(content):
     return reply
 
 if __name__ == "__main__":
+    import pickle
     import re
     import sqlite3
 
@@ -132,8 +133,15 @@ if __name__ == "__main__":
     sql_str = "INSERT INTO comments" \
               "(date, news_id, user_id, content, content_cache) " \
               "VALUES(?, ?, ?, ?, ?)"
+    try:
+        with open('.pickle') as fh:
+            already_seen = pickle.load(fh)
+    except FileNotFoundError:
+        already_seen = set()
 
     for msg_id in ListMessagesWithLabels(GmailMessage._service, 'me', 'INBOX'):
+        if msg_id in already_seen:
+            continue
         mail = GmailMessage.from_id(msg_id['id'])
         m = regex.match(mail['To'])
         if m:
@@ -160,3 +168,5 @@ if __name__ == "__main__":
                 continue
             db.commit()
             print(msg_id['id'], news_id, email_addr, date, comment)
+        already_seen.add(msg_id['id'])
+    pickle.dump(already_seen, '.pickle')