🐍 Python (GeoPandas / PyArrow)

Python is ideal for quickly streaming data directly into memory for exploration or small-scale analysis.

1. Using GeoPandas

GeoPandas uses fsspec under the hood to handle S3 paths. Use storage_options to point to the EUBUCCO endpoint.

Download single region

import geopandas as gpd

storage_opts = {
    "anon": True,
    "client_kwargs": {"endpoint_url": "https://dev-s3.eubucco.com"}
}

# Example: Download Liguria (ITC3)
path = "s3://eubucco/v0.2/buildings/parquet/nuts_id=ITC3/ITC3.parquet"
gdf = gpd.read_parquet(path, storage_options=storage_opts)

2. Using PyArrow

PyArrow is significantly more efficient for large data chunks because it only loads the specific partitions or rows you filter for (predicate pushdown).

Download single regionDownload larger chunks

import fsspec
import pyarrow.dataset as ds

fs = fsspec.filesystem(
    protocol="s3",
    anon=True,
    client_kwargs={"endpoint_url": "https://dev-s3.eubucco.com"}
)

# Example: Download Liguria (ITC3)
dataset = ds.dataset(
    source="eubucco/v0.2/buildings/parquet/nuts_id=ITC3",
    filesystem=fs,
    format="parquet",
)
table = dataset.to_table()

import fsspec
import pyarrow.dataset as ds

fs = fsspec.filesystem(
    protocol="s3",
    anon=True,
    client_kwargs={"endpoint_url": "https://dev-s3.eubucco.com"}
)

dataset = ds.dataset(
    source="eubucco/v0.2/buildings/parquet/",
    filesystem=fs,
    format="parquet"
)

# Example: Get all data for Italy (regions starting with "IT")
table = dataset.to_table(filter=ds.field("nuts_id").starts_with("IT"))

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