[BUG] UnicodeDecodeError when save_to_file=True and CalledProcessError for building types

[felipedantas-pi]

Contributing guidelines

• I understand the contributing guidelines

Documentation

• My proposal is not addressed by the documentation or examples

Existing issues

• Nothing similar appears in an existing issue

Describe the bug

I am experiencing two distinct issues when using load_overture_data:

A UnicodeDecodeError occurs whenever save_to_file=True (specifically during the gpd.read_file step).

A CalledProcessError occurs when attempting to download building data, regardless of the save_to_file setting.

Issue 1: UnicodeDecodeError

When setting save_to_file=True, the data is downloaded correctly, but the library fails to read it back into a GeoDataFrame.

Error Traceback:

UnicodeDecodeError: 'utf-8' codec can't decode byte 0xea in position 24: invalid continuation byte
...
File ~\...\city2graph\data.py:513, in _download_and_process_type
--> 513     gdf = gpd.read_file(output_path)

Interestingly, the user reports that if certain parameters are changed (like setting save_to_file=False in some contexts), it might work, but for building, it consistently crashes.

Issue 2: CalledProcessError with 'building' type

The download for the building layer fails when executed via subprocess.

Error Traceback:

CalledProcessError: Command '['overturemaps', 'download', '--bbox=...', '-f', 'geojson', '--type=building', '-r', '2026-04-15.0']' returned non-zero exit status 1.

Update: I tested the command generated by the library directly in the terminal, and it works perfectly.

overturemaps download --bbox=-42.8902293715,-5.2530579218,-42.6397663876,-4.9296567813 -f geojson --type=building -r 2026-04-15.0

It seems the Overture Maps CLI is returning an error that city2graph is not catching gracefully.

To Reproduce

import city2graph as c2g

subdatasets_types = ["segment", "connector"]
dados = c2g.load_overture_data(
    area=zonaUrbana_3km, # GeoDataFrame/Polygon
    types=subdatasets_types,
    output_dir=RAW_DATA_PATH,
    save_to_file=True,
    return_data=True,
    release='2026-04-15.0'
)

subdataset_building = ['building']
dados_building = c2g.load_overture_data(
    area=zonaUrbana_3km,
    types=subdataset_building,
    save_to_file=False, # or True
    use_stac=True,
    release='2026-04-15.0'
)

Expected behavior

The load_overture_data function should successfully download the requested datasets (including building types), save them to the specified output_dir as GeoJSON files, and return a GeoDataFrame without encountering encoding errors.

Specifically:

When save_to_file=True, the library should handle Windows encoding defaults (likely by enforcing UTF-8) to avoid UnicodeDecodeError during gpd.read_file.

The building subdataset should be downloaded via subprocess successfully, as the command is valid and works when executed directly in the terminal.

Environment (please complete the following information):

• OS: Windows 11
• CPU: Intel Xeon E5-2696-V3
• GPU: NVIDIA GTX 1060 3GB]
• Python version: [e.g., 3.12.12]
• city2graph version: 0.3.1

How did you install city2graph?

via PyPI (e.g., pip)

Additional context

No response

[yu-ta-sato]

@felipedantas-pi, thank you so much for the report! Could you let me know the content of zonaUrbana_3km for reproducing the issue?

[felipedantas-pi]

@yu-ta-sato

link: https://github.com/felipedantas-pi/mapeprof-accessibility-ubs/blob/main/dados/terceiros/zona_urbana/zonaUrbana_3k.geojson

[yu-ta-sato]

Thanks, I assume you loaded the GeoJSON as a gdf object. In case, could you update the replication code from the beginning including it?

[felipedantas-pi]

“Thank you for your response. I don’t have much experience with GitHub, so I apologize if I’m not being clear enough. In response to your request, here is the complete code:”

from pathlib import Path
import city2graph as c2g
import geopandas as gpd

# Diretório para os dados brutos baixados da Overture Maps
RAW_DATA_PATH = Path("../dados/brutos")
RAW_DATA_PATH.mkdir(parents=True, exist_ok=True)

print(f"city2graph version: {c2g.__version__ if hasattr(c2g, '__version__') else 'development'}")


zonaUrbana = gpd.read_file("../dados/terceiros/zona_urbana/zu.geojson")
zonaUrbana_3km = (
    zonaUrbana
    .to_crs(epsg=31983)
    .buffer(distance=3000) 
    .to_crs(epsg=4326) 
)


subdatasets_types = ["segment", "connector"]
dados = c2g.load_overture_data(
    area=zonaUrbana_3km,
    types=subdatasets_types,
    output_dir=RAW_DATA_PATH,
    prefix="zonaUrbana_3k_",
    save_to_file=False,
    return_data=True,
    release='2026-04-15.0',
    connect_timeout=None,
    request_timeout=None,
    use_stac=False,
)


subdataset_building = ['building']
dados_building1 = c2g.load_overture_data(
    area=zonaUrbana_3km,
    types=subdataset_building,
    output_dir=RAW_DATA_PATH / 'building',
    prefix="zonaUrbana_3km_",
    save_to_file=False,
    return_data=True,
    release='2026-04-15.0',
    connect_timeout=None,
    request_timeout=None,
    use_stac=True
)