fix: change message-id field size and filter larger values as spam (#…

…3653)

backend/mlarchive/archive/inspectors.py

@@ -131,4 +131,11 @@ def handle_file(self):
         pass
 
     def raise_error(self):
-        raise NoArchiveMessage('X-No-Archive  Message-ID: {}'.format(self.message_wrapper.msgid))
+        raise NoArchiveMessage('X-No-Archive  Message-ID: {}'.format(self.message_wrapper.msgid))
+
+
+class LongMessageIDSpamInspector(SpamInspector):
+    '''Checks if the Message-ID header exceeds max length'''
+    def has_condition(self):
+        msgid = self.message_wrapper.email_message.get('Message-ID')
+        return len(msgid) > 998

backend/mlarchive/archive/migrations/0002_alter_message_msgid.py

@@ -0,0 +1,18 @@
+# Generated by Django 4.2.7 on 2023-11-09 12:41
+
+from django.db import migrations, models
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ("archive", "0001_initial"),
+    ]
+
+    operations = [
+        migrations.AlterField(
+            model_name="message",
+            name="msgid",
+            field=models.CharField(db_index=True, max_length=998),
+        ),
+    ]

backend/mlarchive/archive/models.py

@@ -162,7 +162,7 @@ class Message(models.Model):
     in_reply_to_value = models.TextField(blank=True, default='')
     # mapping to MHonArc message number
     legacy_number = models.IntegerField(blank=True, null=True, db_index=True)
-    msgid = models.CharField(max_length=240, db_index=True)
+    msgid = models.CharField(max_length=998, db_index=True) # See RFC 2822
     references = models.TextField(blank=True, default='')
     spam_score = models.IntegerField(default=0)             # > 0 = spam
     subject = models.CharField(max_length=512, blank=True)

backend/mlarchive/settings/base.py

@@ -271,6 +271,7 @@
     'ListIdExistsSpamInspector': {'includes': ['httpbisa']},
     'SpamLevelSpamInspector': {'includes': ['rfc-dist', 'rfc-interest', 'httpbisa', 'ipp', 'krb-wg', 'ietf-dkim']},
     'NoArchiveInspector': {},
+    'LongMessageIDSpamInspector': {},
 }
 
 # AUTH

backend/mlarchive/tests/archive/inspectors.py

@@ -3,7 +3,8 @@
 import pytest
 
 from mlarchive.archive.inspectors import (ListIdSpamInspector, SpamMessage,
-    SpamLevelSpamInspector, NoArchiveInspector, NoArchiveMessage)
+    SpamLevelSpamInspector, NoArchiveInspector, NoArchiveMessage,
+    LongMessageIDSpamInspector)
 from mlarchive.archive.mail import MessageWrapper
 
 
@@ -72,3 +73,25 @@ def test_NoArchiveInspector(client, settings):
     with pytest.raises(NoArchiveMessage) as excinfo:
         inspector.inspect()
     assert 'X-No-Archive' in str(excinfo.value)
+
+
+@pytest.mark.django_db(transaction=True)
+def test_LongMessageIDSpamInspector(client, settings):
+    settings.INSPECTORS = {'LongMessageIDSpamInspector': {}}
+    # regular message
+    path = os.path.join(settings.BASE_DIR, 'tests', 'data', 'mail_normal.1')
+    with open(path) as f:
+        message = email.message_from_file(f)
+    mw = MessageWrapper.from_message(message, 'acme')
+    inspector = LongMessageIDSpamInspector(mw)
+    inspector.inspect()
+    # spam message
+    path = os.path.join(settings.BASE_DIR, 'tests', 'data', 'mail_long_messageid.1')
+    with open(path) as f:
+        message = email.message_from_file(f)
+    mw = MessageWrapper.from_message(message, 'acme')
+    inspector = LongMessageIDSpamInspector(mw)
+    with pytest.raises(SpamMessage) as excinfo:
+        inspector.inspect()
+    print(excinfo)
+    assert 'Spam' in str(excinfo.value)

backend/mlarchive/tests/data/mail_long_messageid.1

@@ -0,0 +1,10 @@
+From: User <[email protected]>
+To: User <[email protected]>
+Date: Thu, 7 Nov 2013 17:54:55 +0000
+Message-ID: <0000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000@amsl.com>
+Content-Type: text/plain; charset="us-ascii"
+Subject: This is a test
+
+Hello,
+
+This is a test email.