From 5e2ea81a6c264c718e9fdc7b6fae4a0a7174a624 Mon Sep 17 00:00:00 2001
From: Son <nguyenkims@users.noreply.github.com>
Date: Tue, 4 Jan 2022 18:06:08 +0100
Subject: [PATCH] do not consider out-of-office as bounce

---
 app/email/headers.py        |  4 ++++
 email_handler.py            | 26 ++++++++++++++++++++++++--
 tests/test_email_handler.py | 23 ++++++++++++++++++++++-
 3 files changed, 50 insertions(+), 3 deletions(-)

diff --git a/app/email/headers.py b/app/email/headers.py
index 0196a54e..bb13d3d3 100644
--- a/app/email/headers.py
+++ b/app/email/headers.py
@@ -42,3 +42,7 @@ MIME_HEADERS = [
 ]
 # convert to lowercase to facilitate header look up
 MIME_HEADERS = [h.lower() for h in MIME_HEADERS]
+
+# if any of these headers are present, that means automatic out of office email
+AUTO_REPLY1 = "X-Autoreply"
+AUTO_REPLY2 = "Auto-Submitted"
diff --git a/email_handler.py b/email_handler.py
index 1f99ec70..a3c8c9c6 100644
--- a/email_handler.py
+++ b/email_handler.py
@@ -1751,6 +1751,22 @@ def handle_spam(
         )
 
 
+def is_automatic_out_of_office(msg: Message) -> bool:
+    if msg[headers.AUTO_REPLY1] is not None:
+        LOG.d(
+            "out-of-office email %s:%s", headers.AUTO_REPLY1, msg[headers.AUTO_REPLY1]
+        )
+        return True
+
+    if msg[headers.AUTO_REPLY2] is not None:
+        LOG.d(
+            "out-of-office email %s:%s", headers.AUTO_REPLY2, msg[headers.AUTO_REPLY2]
+        )
+        return True
+
+    return False
+
+
 def handle_unsubscribe(envelope: Envelope, msg: Message) -> str:
     """return the SMTP status"""
     # format: alias_id:
@@ -2060,13 +2076,19 @@ def handle(envelope: Envelope) -> str:
         len(rcpt_tos) == 1
         and rcpt_tos[0].startswith(BOUNCE_PREFIX)
         and rcpt_tos[0].endswith(BOUNCE_SUFFIX)
+        # out of office is sent to the mail_from
+        # more info on https://support.google.com/mail/thread/21246740/my-auto-reply-filter-isn-t-replying-to-original-sender-address?hl=en&msgid=21261237
+        and not is_automatic_out_of_office(msg)
     ):
+
         email_log_id = parse_id_from_bounce(rcpt_tos[0])
         email_log = EmailLog.get(email_log_id)
         return handle_bounce(envelope, email_log, msg)
 
-    if len(rcpt_tos) == 1 and rcpt_tos[0].startswith(
-        f"{BOUNCE_PREFIX_FOR_REPLY_PHASE}+"
+    if (
+        len(rcpt_tos) == 1
+        and rcpt_tos[0].startswith(f"{BOUNCE_PREFIX_FOR_REPLY_PHASE}+")
+        and not is_automatic_out_of_office(msg)
     ):
         email_log_id = parse_id_from_bounce(rcpt_tos[0])
         email_log = EmailLog.get(email_log_id)
diff --git a/tests/test_email_handler.py b/tests/test_email_handler.py
index 885510e8..f6bf8af0 100644
--- a/tests/test_email_handler.py
+++ b/tests/test_email_handler.py
@@ -1,5 +1,12 @@
+from email.message import EmailMessage
+
+from app.email import headers
 from app.models import User, Alias, AuthorizedAddress, IgnoredEmail
-from email_handler import get_mailbox_from_mail_from, should_ignore
+from email_handler import (
+    get_mailbox_from_mail_from,
+    should_ignore,
+    is_automatic_out_of_office,
+)
 
 
 def test_get_mailbox_from_mail_from(flask_client):
@@ -40,3 +47,17 @@ def test_should_ignore(flask_client):
     assert not should_ignore("mail_from", ["rcpt_to"])
     IgnoredEmail.create(mail_from="mail_from", rcpt_to="rcpt_to", commit=True)
     assert should_ignore("mail_from", ["rcpt_to"])
+
+
+def test_is_automatic_out_of_office():
+    msg = EmailMessage()
+    assert not is_automatic_out_of_office(msg)
+
+    msg[headers.AUTO_REPLY1] = "yes"
+    assert is_automatic_out_of_office(msg)
+
+    del msg[headers.AUTO_REPLY1]
+    assert not is_automatic_out_of_office(msg)
+
+    msg[headers.AUTO_REPLY2] = "auto-replied"
+    assert is_automatic_out_of_office(msg)