Open file and immediately save it makes exception

[culebron]

I open ASTER DEM file with such simple code:
ds = gr.from_file(input_file)
ds.to_tiff('/tmp/test')

And it makes exception:

  File "store_gridfs.py", line 13, in main
ds.to_tiff('/tmp/test')
  File ".../georasters.py", line 478, in to_tiff
self.raster.data[self.raster.mask] = self.nodata_value
TypeError: int() argument must be a string, a bytes-like object or a number, not 'NoneType'

I tried shielding it with an if closure, but then it made another exception:

  File ".../georasters.py", line 122, in create_geotiff
    if isinstance(datatype, np.int) == False:
  File "/usr/local/lib/python3.6/dist-packages/numpy/ma/core.py", line 3353, in __setitem__
    _data[indx] = dval

You usually expect open and save methods to work seamlessly.

[culebron]

In #33 , I added checks of masked data length and converting ndv to double. But check if it works in earlier python versions.

[ozak]

I wonder if this is related to #28? What do you get when requesting ds.nodata_value? I thought about forcing a NDV on the raster fi missing, but that may cause issues so perhaps it would be best to just add an error message that is more informative.

[ozak]

@culebron were you able to test this (#33)? Or can you share the file you were using?

[ozak]

Fixed in commit acfa8f1 (PR #73, merged to master) — same root cause as #47.

Root cause: nodata_value is None for files that don't set a nodata value in GDAL. to_tiff() then crashes trying to assign None into the masked array data.

Fix: to_tiff() and create_geotiff() now guard all NDV usages against None. from_file() also now falls back to a type-appropriate default and warns when no NDV is present in the file.