feat: validate Anwarpy publication metadata

automation/anwarpy/test_sanitize.py

@@ -0,0 +1,44 @@
+import unittest
+
+from sanitize import assert_publishable, sanitize_text
+
+
+class SanitizeTests(unittest.TestCase):
+    def test_redacts_private_identifiers_and_credentials(self):
+        raw = """Connect to 10.1.2.3 or fd00::12 as admin@example.test.
+SSH root@private.internal and use password=FakeSecret123.
+Read /Users/example/private/config and call https://hidden.internal/api.
+Phone +1 555 123 4567; token: abcdefghijklmnop.
+"""
+        cleaned = sanitize_text(raw, allowed_hosts={"anwarpy.ai"})
+        for secret in (
+            "10.1.2.3",
+            "fd00::12",
+            "admin@example.test",
+            "root@private.internal",
+            "FakeSecret123",
+            "/Users/example/private/config",
+            "hidden.internal",
+            "+1 555 123 4567",
+            "abcdefghijklmnop",
+        ):
+            self.assertNotIn(secret, cleaned)
+        assert_publishable(cleaned, allowed_hosts={"anwarpy.ai"})
+
+    def test_preserves_approved_public_urls(self):
+        text = "راجع https://anwarpy.ai/posts/example للمزيد."
+        self.assertEqual(sanitize_text(text, allowed_hosts={"anwarpy.ai"}), text)
+        assert_publishable(text, allowed_hosts={"anwarpy.ai"})
+
+    def test_redacts_unapproved_public_hosts(self):
+        cleaned = sanitize_text("Source: https://unapproved.example/article", allowed_hosts={"anwarpy.ai"})
+        self.assertNotIn("unapproved.example", cleaned)
+        assert_publishable(cleaned, allowed_hosts={"anwarpy.ai"})
+
+    def test_assert_publishable_raises_without_using_assert_statement(self):
+        with self.assertRaises(ValueError):
+            assert_publishable("internal address 192.168.1.10", allowed_hosts={"anwarpy.ai"})
+
+
+if __name__ == "__main__":
+    unittest.main()