fix email detection for user(at)domain.ch and add override for ssl expiry

Author: davidhuser

overrides.json

@@ -75,6 +75,7 @@
   "6195": {"domain": "loetschen.ch", "reason": "Shared Lötschental admin"},
   "6197": {"domain": "loetschen.ch", "reason": "Shared Lötschental admin"},
   "6205": {"domain": "gemeinde-bettmeralp.ch", "reason": "as on website"},
+  "6285": {"domain": "graechen.ch", "reason": "Wikidata URL incorrect (gemeinde-graechen.ch has SSL issues); actual website at gemeinde.graechen.ch, email on graechen.ch (MS365)"},
   "6404": {"domain": "ne.ch", "reason": "Using cantonal mail"},
   "6408": {"domain": "ne.ch", "reason": "Using cantonal mail"},
   "6413": {"domain": "ne.ch", "reason": "Using cantonal mail"},

src/mail_sovereignty/resolve.py

@@ -403,6 +403,13 @@ def extract_email_domains(html: str) -> set[str]:
                     domains.add(domain)
                 break
 
+    for match in re.findall(r'[\w.-]+\s*[\[(]at[\])]\s*[\w.-]+\.\w+', html, re.IGNORECASE):
+        normalized = re.sub(r'\s*[\[(]at[\])]\s*', '@', match, flags=re.IGNORECASE)
+        if "@" in normalized:
+            domain = normalized.split("@")[1].lower()
+            if domain not in SKIP_DOMAINS:
+                domains.add(domain)
+
     return {d for d in domains if _is_valid_domain(d)}
 
 

tests/test_resolve.py

@@ -483,6 +483,26 @@ def test_mailto_trailing_slash(self):
         domains = extract_email_domains(html)
         assert "example.org" in domains
 
+    def test_bracket_at_obfuscation(self):
+        html = "gemeinde[at]graechen.ch"
+        assert "graechen.ch" in extract_email_domains(html)
+
+    def test_paren_at_obfuscation(self):
+        html = "info(at)gemeinde.ch"
+        assert "gemeinde.ch" in extract_email_domains(html)
+
+    def test_bracket_at_with_spaces(self):
+        html = "info [at] town.ch"
+        assert "town.ch" in extract_email_domains(html)
+
+    def test_bracket_at_uppercase(self):
+        html = "admin[AT]village.ch"
+        assert "village.ch" in extract_email_domains(html)
+
+    def test_bracket_at_skip_domain(self):
+        html = "user[at]example.com"
+        assert extract_email_domains(html) == set()
+
     def test_domain_label_too_long(self):
         """Domains with labels > 63 chars should be filtered out."""
         long_label = "a" * 64