Overture — per-row licensing (live)¶

Overture aggregates many sources with different licenses. The backend surfaces a per-row license_id and warns when share-alike ODbL-1.0 (OpenStreetMap-derived) rows are present — critical for commercial redistribution.

Setup¶

Pull in the libraries used throughout: EarthLens is the unified entry point, LicenseWarning and row_license come from the Overture backend, and geopandas reads the written parquet.

In [ ]:

import warnings

import geopandas as gpd

from earthlens.earthlens import EarthLens
from earthlens.overture import LicenseWarning, row_license

import warnings import geopandas as gpd from earthlens.earthlens import EarthLens from earthlens.overture import LicenseWarning, row_license

1 · Define the area of interest¶

A single Times Square block, given as south/north latitudes and west/east longitudes, plus the output directory the buildings parquet will be written to.

In [ ]:

LAT_LIM = [40.757, 40.759]  # [south, north]
LON_LIM = [-73.987, -73.984]  # [west, east] — a Times Square block
OUT = "_overture_out"

LAT_LIM = [40.757, 40.759] # [south, north] LON_LIM = [-73.987, -73.984] # [west, east] — a Times Square block OUT = "_overture_out"

2 · Download the buildings layer¶

Build the Overture request first, then call download() on it. We run the download inside a warnings.catch_warnings(record=True) block so any LicenseWarning is captured rather than printed inline, letting us inspect it afterwards.

In [ ]:

overture = EarthLens(
    data_source="overture",
    dataset="buildings",
    variables=[],
    aoi=[LON_LIM[0], LAT_LIM[0], LON_LIM[1], LAT_LIM[1]],
    path=OUT,
)

with warnings.catch_warnings(record=True) as caught:
    warnings.simplefilter("always")
    paths = overture.download()

overture = EarthLens( data_source="overture", dataset="buildings", variables=[], aoi=[LON_LIM[0], LAT_LIM[0], LON_LIM[1], LAT_LIM[1]], path=OUT, ) with warnings.catch_warnings(record=True) as caught: warnings.simplefilter("always") paths = overture.download()

License warnings¶

Filter the captured warnings down to LicenseWarning instances and print them — this is how the backend flags the presence of share-alike ODbL-1.0 rows.

In [ ]:

for w in caught:
    if issubclass(w.category, LicenseWarning):
        print("LicenseWarning:", w.message)

for w in caught: if issubclass(w.category, LicenseWarning): print("LicenseWarning:", w.message)

3 · Per-row license_id¶

Read the written parquet back into a GeoDataFrame and tally the license_id column — every feature carries its own license, so the distribution is visible at a glance.

In [ ]:

gdf = gpd.read_parquet(paths[0])
gdf["license_id"].value_counts()

gdf = gpd.read_parquet(paths[0]) gdf["license_id"].value_counts()

4 · The derivation rule¶

The rule behind license_id is exposed directly as row_license for ad-hoc analysis of a sources cell — ODbL-1.0 wins whenever an OpenStreetMap source is present, even if it is not listed first.

In [ ]:

print(
    row_license(
        [
            {"dataset": "Overture", "license": "CDLA-Permissive-2.0"},
            {"dataset": "OpenStreetMap", "license": "ODbL-1.0"},
        ]
    )
)
print(row_license([{"dataset": "Foursquare", "license": "Apache-2.0"}]))

print( row_license( [ {"dataset": "Overture", "license": "CDLA-Permissive-2.0"}, {"dataset": "OpenStreetMap", "license": "ODbL-1.0"}, ] ) ) print(row_license([{"dataset": "Foursquare", "license": "Apache-2.0"}]))