Normalize structured German TTS input

2026-08-23 13:30:46 +02:00
parent a067b79ed2
commit d726afff70
4 changed files with 113 additions and 0 deletions
@@ -78,6 +78,27 @@ class LanguageSegmentationTests(unittest.TestCase):
        text = "Docker-Container laufen, die Health-Checks melden healthy."
        self.assertEqual(gateway.segment_languages(text), [("de", text)])

+    def test_weather_notation_is_spoken_naturally(self):
+        text = (
+            "Rastatt (76437), Sonntag 23.08.: 22° / 10°, "
+            "0 % Regen. Quellen: wetter.com und wetteronline.de"
+        )
+        spoken = gateway.clean_for_speech(text)
+        self.assertIn("Postleitzahl sieben sechs vier drei sieben", spoken)
+        self.assertIn("23. August", spoken)
+        self.assertIn("Höchstwert 22 Grad, Tiefstwert 10 Grad", spoken)
+        self.assertIn("0 Prozent Regen", spoken)
+        self.assertIn("wetter Punkt C O M", spoken)
+        self.assertIn("wetteronline Punkt D E", spoken)
+
+    def test_full_date_and_raw_url_are_normalized(self):
+        spoken = gateway.clean_for_speech(
+            "Am 23.08.2026 steht es auf https://www.example.org/path?q=1."
+        )
+        self.assertIn("23. August 2026", spoken)
+        self.assertIn("example Punkt O R G", spoken)
+        self.assertNotIn("https", spoken)
+

 class FallbackTests(unittest.TestCase):
    def setUp(self):