Not possible to read from file handles without warnings

[vilim]

Reading files with the method:

with open(file, "rb") as f:
    pyogrio.read_dataframe(f)

always raises a warning like:

/home/vstih/.conda/envs/detektp/lib/python3.11/site-packages/pyogrio/raw.py:198: RuntimeWarning:

File [/vsimem/pyogrio_c99dcda56d63460497ec381f7b73f788](http://localhost:8810/vsimem/pyogrio_c99dcda56d63460497ec381f7b73f788) has GPKG application_id, but non conformant file extension

I am using pyogrio version 0.10.0
if a byte stream is passed to read_dataframe, I guess the warning should not be raised.

[brendan-ward]

The error comes from GDAL; it is trying to enforce file name conventions for GPKG files. The problem here is that we pass off a byte stream which contains the bytes of the GPKG and a randomly-generated filename without extension, but we don't currently detect that it is a GPKG file on our end in order to add the correct extension. I recall that with GPKG it is a little tricky; we can read the first few bytes to determine that it is an SQLite file, but I think GDAL does several more heuristics to determine that it is indeed a GPKG.

I guess the other thing we could do is specifically ignore this warning internally when we know we are using bytes instead of a file path.

[Juanouo]

I'm having the same issue. Is there any way I can silence the warning?

[theroggy]

You can filter it like any warning, e.g. like this:

import warnings

import pyogrio

file = "C:\Temp\polygon_parcel\polygon-parcel.gpkg"

with open(file, "rb") as f:
    with warnings.catch_warnings():
        warnings.filterwarnings("ignore", ".*has GPKG application_id.*", RuntimeWarning)
        pyogrio.read_dataframe(f)

[jorisvandenbossche]

I guess the other thing we could do is specifically ignore this warning internally when we know we are using bytes instead of a file path.

I think this would be a good (at least temporary / short-term) solution to avoid users seeing this warning

[sehoffmann]

Have the same issue. Feels weird to get issued a warning when using an API in the intended way. Weirdly enough, passing driver="GPKG" makes matters worse:

with resources.joinpath(f'datasets/africa_v0/{split}_locations.gpkg').open('rb') as f:
            gdf = gpd.read_file(f, driver='GPKG')


/python3.12/site-packages/pyogrio/raw.py:198: RuntimeWarning: driver GPKG does not support open option DRIVER
  return ogr_read(
/python3.12/site-packages/pyogrio/raw.py:198: RuntimeWarning: File /vsimem/pyogrio_b78c659510294d97b233b7c494d93468 has GPKG application_id, but non conformant file extension