ENH: add "repair" and/or "salvage" option to `on_invalid` parameter in `read_dataframe` - Githubissues

geopandas / pyogrio

Vectorized vector I/O using OGR

https://pyogrio.readthedocs.io

MIT License

257 stars 21 forks source link

ENH: add "repair" and/or "salvage" option to `on_invalid` parameter in `read_dataframe` #423

Open theroggy opened 2 weeks ago

theroggy commented 2 weeks ago

The on_invalid functionality in read_dataframe always results in None geometries now.

It would be interesting to also have a "repair" and/or "salvage" option, so it becomes possible to recover data if possible.

In theory this should be possible, but as GEOS doesn't support this kind of thing, it is not trivial:

support for something like that could be added in GEOS.
Or via another library, but this most likely implies extra dependencies which might not be ideal for an edge case like this.
Or an option to get back the WKB's instead of trying to read them, so you can salvage yourself
...

xref https://github.com/geopandas/pyogrio/pull/422