From: Michael Tremer <michael.tremer@ipfire.org>
Date: Sun, 28 Dec 2025 13:42:57 +0000 (+0000)
Subject: sources: Normalize any international domain names
X-Git-Url: http://git.ipfire.org/cgi-bin/gitweb.cgi?a=commitdiff_plain;h=d3676d72e4fe1d46fedff19785dcf7ac6411b467;p=dbl.git

sources: Normalize any international domain names

We don't want these to hit the database in Unicode, but only in ASCII
format.

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>
---

diff --git a/src/dnsbl/sources.py b/src/dnsbl/sources.py
index 236a542..e9754c7 100644
--- a/src/dnsbl/sources.py
+++ b/src/dnsbl/sources.py
@@ -23,6 +23,7 @@ import email.utils
 import enum
 import gzip
 import httpx
+import idna
 import io
 import itertools
 import logging
@@ -236,6 +237,14 @@ class Source(sqlmodel.SQLModel, database.BackendMixin, table=True):
 							# Remove any trailing dots
 							domain = domain.removesuffix(".")
 
+							# Normalize any international domain names
+							try:
+								domain = idna.encode(domain).decode("ascii")
+
+							except idna.IDNAError as e:
+								log.warning(_("Skipping invalid domain: %s") % domain)
+								continue
+
 							# Skip any invalid domain names
 							if not util.is_fqdn(domain):
 								# Silently skip any IP addresses