Clean up code base, add pre-commit, add dependencies #7
Conversation
Zeitsperre
added
enhancement
| @@ -2,6 +2,9 @@ | |||
| from pathlib import Path | |||
|
|
|||
| import xarray as xr | |||
|
|
|||
| # FIXME: This is a circular dependency, as clisops will be listing roocs-grids as a dependency | |||
There was a problem hiding this comment.
The scripts directory is basically a directory with auxiliary scripts to prepare the grid files. None of its contents are imported by the roocs_grids module. I don't know if there is a proper approach for something like this. Maybe one could rename it to aux and move it one directory up, next to tests, roocs_grids so it won't get accidentally imported?
| @@ -30,6 +33,7 @@ | |||
| # Create clore.Grid object | |||
| g = clore.Grid(ds) | |||
| # Drop variables and global attributes | |||
| # FIXME: This is broken in newer versions of xarray | |||
There was a problem hiding this comment.
I will need to have a look in the clisops regridding PR when time allows, to solve this.
| [[ ! -e "$targetpath"/grid_T63_lsm_binary.nc ]] && { | ||
| mv "$targetpath"/sftlf_fx_MPI-ESM1-2-LR_1pctCO2_r1i1p1f1_gn.nc "$targetpath"/grid_T63_lsm_binary.nc | ||
| ncks -A -h -v areacella $targetpath/areacella_fx_MPI-ESM1-2-LR_1pctCO2_r1i1p1f1_gn.nc "$targetpath"/grid_T63_lsm_binary.nc | ||
| nccopy -d 9 -s "$targetpath"/grid_T63_lsm_binary.nc "$targetpath"/grid_T63_lsm_binary_def.nc |
There was a problem hiding this comment.
I will need to extend the compression to all downloaded grid files and remove unnecessary attributes, variables and dimensions. By just applying nccopy -d 9 -s to all the grid files, the volume could be shrinked from 45M to 2.4M, which would be fine? The volume was one of the reasons, roocs_grids was a separate repository and not part of the clisops regridding PR.
I will follow up your PR with another one dealing with the compression and further cleaning of the netcdf files.
|
Your comments sound good. I'll merge this now and leave you to work on the comments raised here. All the best! |
Addresses issues raised in #2, but does not close it.
Ping @agstephens @sol1105
Changes
Work needed
Proposal
Would this library better serve as a module of CLISOPS instead? Are there instances where the tools here would be used independently of CLISOPS? Given the tight focus and coupling with CLISOPS, this seems more like a specialized helper library. If so, we can very easily push this directly into CLISOPS.