geo-l1-tokyo-busstop-buffer

Rerun

import geopandas as gpd

def main():
    # Load the input file
    input_file = 'tokyo_connectors.geojson'
    output_file = 'tokyo_stop_catchments.geoparquet'
    
    try:
        gdf = gpd.read_file(input_file)
    except Exception as e:
        print(f"Error reading {input_file}: {e}")
        return

    print(f"Loaded {len(gdf)} connectors.")
    print(f"CRS: {gdf.crs}")

    # Ensure we have a projection suitable for buffering in meters
    # Most likely EPSG:4326 (WGS84). To buffer in meters, we need a metric CRS.
    # For Tokyo, EPSG:3857 (Web Mercator) is common but can have distortion.
    # Better: EPSG:6677 (JGD2011 / Japan Plane Rectangular CS IX) or similar.
    # Let's try to project to a metric CRS.
    
    # We'll project to a local metric CRS if it's not already. 
    # EPSG:3857 is often used for simple buffers, but let's check for a better one.
    # For Tokyo, UTM Zone 54N (EPSG:32654) or Japan Plane Rectangular CS is better.
    
    if gdf.crs is None:
        print("CRS not found. Assuming EPSG:4326.")
        gdf.set_crs(epsg=4326, inplace=True)
    
    # Transform to a metric CRS (UTM Zone 54N is good for Tokyo)
    # We can also use a more general one like EPSG:3857 if we want to be quick,
    # but let's use a more accurate one if possible.
    # Let's use EPSG:32654 (WGS 84 / UTM zone 54N)
    gdf_metric = gdf.to_crs(epsg=32654)
    
    # Apply the 400m buffer
    print("Applying 400m buffer...")
    gdf_metric['geometry'] = gdf_metric.buffer(400)
    
    # Select only the required columns
    # Requirement: "Please keep connector_id on every row with the original value preserved"
    # We'll keep 'connector_id' and 'geometry'
    if 'connector_id' not in gdf_metric.columns:
        print(f"Warning: 'connector_id' not found in columns. Available: {gdf_metric.columns.tolist()}")
        # Let's check if it's case sensitive or slightly different
        potential_ids = [col for col in gdf_metric.columns if 'id' in col.lower()]
        if potential_ids:
            print(f"Found potential ID columns: {potential_ids}. Using {potential_ids[0]}")
            gdf_metric = gdf_metric.rename(columns={potential_ids[0]: 'connector_id'})
        else:
            print("Could not find an ID column.")
            return

    output_gdf = gdf_metric[['connector_id', 'geometry']]
    
    # Convert back to the original CRS (or WGS84) if necessary, 
    # though GeoParquet can store any CRS. Usually, it's good to return to WGS84 for compatibility.
    output_gdf = output_gdf.to_crs(gdf.crs)
    
    # Save to GeoParquet
    print(f"Saving to {output_file}...")
    output_gdf.to_parquet(output_file)
    print("Done.")

if __name__ == "__main__":
    main()

import geopandas as gpd

def main():
    input_file = 'tokyo_connectors.geojson'
    output_file = 'tokyo_stop_catchments.geoparquet'
    
    gdf = gpd.read_file(input_file)
    print(f"Input columns: {gdf.columns.tolist()}")

    # Buffer in metric CRS
    gdf_metric = gdf.to_crs(epsg=32654)
    gdf_metric['geometry'] = gdf_metric.buffer(400)
    
    # Check for connector_id
    if 'connector_id' not in gdf_metric.columns:
        raise KeyError(f"connector_id not found in input. Columns are: {gdf_metric.columns.tolist()}")

    # Keep connector_id and geometry, back to WGS84
    output_gdf = gdf_metric[['connector_id', 'geometry']].to_crs(epsg=4326)
    
    # Ensure one row per input connector (buffer preserves row count)
    # Save to GeoParquet
    output_gdf.to_parquet(output_file)
    print(f"Successfully saved {len(output_gdf)} rows to {output_file}")

if __name__ == "__main__":
    main()