🦆 DuckDB (SQL)

DuckDB offers the most powerful way to query EUBUCCO. It treats the S3 bucket like a database, letting you filter by geography before any data is actually downloaded.

1. Using SQL

Configuration

Install spatial extension:

INSTALL httpfs; LOAD httpfs;
INSTALL spatial; LOAD spatial

Configure EUBUCCO's MinIO endpoint:

SET s3_endpoint='dev-s3.eubucco.com';
SET s3_url_style='path';

Data access

Download single regionDownload larger chunks

SELECT * FROM 's3://eubucco/v0.2/buildings/parquet/nuts_id=ITC3/ITC3.parquet' 
LIMIT 10

By using hive_partitioning, DuckDB understands the folder structure (e.g., nuts_id=...) and can filter hundreds of files instantly.

-- Example: Query all of France (FR) for buildings higher than 50 meters
SELECT id, height, ST_AsText(geometry) 
FROM read_parquet(
    's3://eubucco/v0.2/buildings/parquet/**/*.parquet',
    hive_partitioning = true
)
WHERE nuts_id LIKE 'FR%' 
AND height > 50;

2. Using DuckDB's Python library

pip install duckdb

import duckdb

con = duckdb.connect()
con.execute("INSTALL httpfs; LOAD httpfs;")
con.execute("INSTALL spatial; LOAD spatial;")

con.execute("SET s3_endpoint='dev-s3.eubucco.com';") 
con.execute("SET s3_url_style='path';")

query = """
SELECT * EXCLUDE geometry, ST_AsWKB(geometry) AS geometry FROM 's3://eubucco/v0.2/buildings/parquet/nuts_id=ITC3/ITC3.parquet' 
LIMIT 10
"""
# Fetch data with WKB-encoded geometries
df = con.execute(query).arrow().to_pandas()

# Optionally convert to GeoDataFrame
gdf = gpd.GeoDataFrame(
    df,
    geometry=gpd.GeoSeries.from_wkb(df["geometry"]),
    crs="EPSG:3035",
)