Cache data sources

download_arcgis_data.py

@@ -101,6 +101,10 @@ def convert_to_parquet(data_path: Path, parquet_path: Path) -> None:
 
 
 def main() -> None:
+    if PARQUET_PATH.exists():
+        print(f"Parquet already exists at {PARQUET_PATH}, skipping")
+        return
+
     if not DOWNLOAD_PATH.exists():
         download_with_progress(URL, DOWNLOAD_PATH)
     else:

download_land_registry.py

@@ -95,6 +95,10 @@ def convert_to_parquet(csv_path: Path, parquet_path: Path) -> None:
 
 
 def main() -> None:
+    if PARQUET_PATH.exists():
+        print(f"Parquet already exists at {PARQUET_PATH}, skipping")
+        return
+
     if not CSV_PATH.exists():
         download_with_progress(URL, CSV_PATH)
     else:

generate_tfl_client.py

@@ -20,10 +20,10 @@ def main() -> None:
     if not OPENAPI_SPEC.exists():
         raise FileNotFoundError(f"OpenAPI spec not found: {OPENAPI_SPEC}")
 
-    # Remove existing client if present
+    # Skip if client already exists
     if OUTPUT_PATH.exists():
-        print(f"Removing existing client at {OUTPUT_PATH}")
-        shutil.rmtree(OUTPUT_PATH)
+        print(f"TfL client already exists at {OUTPUT_PATH}, skipping")
+        return
 
     # Generate the client
     print(f"Generating client from {OPENAPI_SPEC}")