Add POI pipeline

download_pois.py

@@ -0,0 +1,54 @@
+"""Download POI data for the UK from Overture Maps."""
+
+from pathlib import Path
+
+import overturemaps
+import pyarrow as pa
+import pyarrow.parquet as pq
+from tqdm import tqdm
+
+# UK bounding box (west, south, east, north)
+UK_BBOX = (-8.65, 49.86, 1.77, 60.86)
+
+OUTPUT_DIR = Path("data_sources")
+OUTPUT_FILE = OUTPUT_DIR / "uk_pois.parquet"
+
+
+def main():
+    OUTPUT_DIR.mkdir(parents=True, exist_ok=True)
+
+    if OUTPUT_FILE.exists():
+        print(f"POI file already exists: {OUTPUT_FILE}")
+        print("Delete it manually to re-download.")
+        return
+
+    print("Downloading UK POI data from Overture Maps...")
+    print(f"Bounding box: {UK_BBOX}")
+    print("This may take several minutes...")
+
+    reader = overturemaps.record_batch_reader("place", bbox=UK_BBOX)
+
+    # Read all batches
+    batches = []
+    with tqdm(desc="Downloading batches", unit=" batches") as pbar:
+        for batch in reader:
+            batches.append(batch)
+            pbar.update(1)
+            pbar.set_postfix(rows=sum(b.num_rows for b in batches))
+
+    if not batches:
+        print("No data found in bounding box!")
+        return
+
+    # Combine batches into a table and write
+    table = pa.Table.from_batches(batches, schema=reader.schema)
+
+    print(f"\nWriting {table.num_rows:,} POIs to {OUTPUT_FILE}...")
+    pq.write_table(table, OUTPUT_FILE)
+
+    print(f"Download complete: {OUTPUT_FILE}")
+    print(f"File size: {OUTPUT_FILE.stat().st_size / 1024 / 1024:.1f} MB")
+
+
+if __name__ == "__main__":
+    main()