Improve data

pipeline/download/gias.py

@@ -192,6 +192,10 @@ def _read_csv_from_zip(zip_bytes: bytes) -> pl.DataFrame:
         infer_schema_length=20000,
         null_values=_NULL_VALUES,
         truncate_ragged_lines=True,
+        # Force the phone number to stay a string: schema inference reads it as
+        # an integer and strips the leading 0 (e.g. 020 8427 7222 -> 2084277222),
+        # making nearly every school phone number un-diallable.
+        schema_overrides={"TelephoneNum": pl.String},
     )