[gdal-dev] Options of gdalwarp not all exposed in the GDAL Python gdal.Warp ?

[Even Rouault]

Thomas,

part of the unexposed options are indeed due to:

- sometimes desynchronization between the C++ and Python API

- sometimes not finding quickly a good enough Python name, and the 
option being of marginal use

- or because the handling of the switch must be done when 
opening/creating the input/output dataset and, as gdal.Warp() can accept 
dataset objects in addition to filenames, the option would be without 
effect. If you look at apps/gdalwarp_lib.cpp in the 
GDALWarpAppOptionsNew() function, all switches that have only the effect 
of setting a field of the psOptionsForBinary structure aren't good 
candidates to be mapped as kwargsoptions. -overwrite is in this category 
(if you pass an output dataset *object*)

Regarding -rb, -rc, -rcs they are historical shortcurs for -r bilinear, 
-r cubic, -r cubicspline that have been undocumented for phasing them 
out as we have now a bunch of methods compared to the few ones of older 
versions.

Even

Le 31/05/2021 à 22:50, Thomas Gratier a écrit :
> Hello all,
>
> Due to question at 
> https://gis.stackexchange.com/questions/398223/how-to-remap-gdal-options-from-string-version-to-a-dictionary-version 
> <https://gis.stackexchange.com/questions/398223/how-to-remap-gdal-options-from-string-version-to-a-dictionary-version>
> I've taken a look at code
>
> When trying to use gdal.Warp from Python, most options are supported 
> but some seems to be missing
> Although an approach with only `gdal.Warp(dest, origin, 
> options='-overwrite')` can work without code change,
> the dict approach like below seems to be simpler from a Python perspective
>
> kwargsoptions = {"overwrite": True} # particular key does not exist, 
> just for the explanation
> gdal.Warp(dest, origin, **kwargsoptions)
>
> Below are the command options I only found in gdalwarp command line 
> but absent in the WarpOptions (from doc 
> https://gdal.org/programs/gdalwarp.html#cmdoption-gdalwarp-r 
> <https://gdal.org/programs/gdalwarp.html#cmdoption-gdalwarp-r>)
>
> [-ct string]
> [-novshiftgrid]
> [-refine_gcps tolerance [minimum_gcps]]
> [-ovr level|AUTO|AUTO-n|NONE]
> [-srcalpha|-nosrcalpha]
> [-q]
> [-if format]*
> [-overwrite]
> [-oo NAME=VALUE]*
> [-doo NAME=VALUE]*
>
> -srcalpha is present but not it counterpart -nosrcalpha
>
> In gdal.py code, there are references to resampling option not in the 
> doc e.g below excerpt
> (from 
> https://github.com/OSGeo/gdal/blob/master/gdal/swig/python/osgeo/gdal.py#L588 
> <https://github.com/OSGeo/gdal/blob/master/gdal/swig/python/osgeo/gdal.py#L588>) 
> .
>
> elif resampleAlg == GRIORA_Bilinear:
>     new_options += ['-rb']
> elif resampleAlg == GRIORA_Cubic:
>     new_options += ['-rc']
> elif resampleAlg == GRIORA_CubicSpline:
>     new_options += ['-rcs']
>
> It may to need to be removed if looking at 
> https://gdal.org/programs/gdalwarp.html#cmdoption-gdalwarp-r 
> <https://gdal.org/programs/gdalwarp.html#cmdoption-gdalwarp-r>
> e.g (no -rcs, -rb or -rc options) except maybe for compatibility reason
>
> I will try to make a PR. The main issue is more about naming for each 
> option.
> I also don't know if command line and Python API "unsync" is due to 
> evolution, to intended API choices or other reasons.
>
> I also only discuss here the gdal.Warp options but I suppose there are 
> maybe other Python tools ported from RFC 59
> https://gdal.org/development/rfc/rfc59_utilities_as_a_library.html 
> <https://gdal.org/development/rfc/rfc59_utilities_as_a_library.html> 
> that may have the same "unsync" issue.
> Did not look at the moment
>
>
> Regards
>
> Thomas Gratier
>
> _______________________________________________
> gdal-dev mailing list
> gdal-dev at lists.osgeo.org
> https://lists.osgeo.org/mailman/listinfo/gdal-dev

-- 
http://www.spatialys.com
My software is free, but my time generally not.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/gdal-dev/attachments/20210531/27931d2a/attachment.html>