r.mask.status: Check mask status through a tool #2390
Conversation
Instead of using low-level test of file existence with hardcoded raster path and name, use a module to retrieve status of the raster mask. The new module r.mask.status reports presence or absence of the 2D raster mask and provides additional details about the mask. The PR requires additional changes in the library which are not yet included.
|
@wenzeslaus Please provide details on what is missing to review this PR? |
github-actions
bot
added
raster
Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
![github-actions[bot]](https://avatars.githubusercontent.com/in/15368?s=60&v=4){kind=small}
Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
|
This is now ready for first round of reviews. I'm leaving it as a draft because I've left a lot of comments in the code considering other ways to implement this (I can move them to GitHub comments if desired). There is one usage of this now and that's the one for a shell prompt. The prompt no longer relies on testing the file presence with the test program, but uses a GRASS tool to find out. The code goes out of its way to report both mask name (always MASK) and the underlying raster name if it is a reclassified (without rewriting the current C API). This is to mimic the existing C functions which are returning the underlying raster if MASK is a reclass. The tool and the new C API function return both. This is written with #2392 in mind trying to remove some of the details about 2D raster mask implementation, namely removing the name MASK from the API. This does not go as far as to resolve all the issues such as addressing the need for a simple mask presence test in #2390 or interactions with mask in Python tools which themselves need to use their own mask. |
On many places (more than covered here), 2D raster mask is called MASK conflating the concept (mask) and the implementation (MASK raster). Users using the r.mask tool may not interact with the underlying raster directly, so there is no reason to use MASK over mask. This is leaving many places as they are with MASK. Some will be better revisited with or after OSGeo#2390 and OSGeo#2392 when a more comprehensive solution is available. This fixes and keeps in sync wording r.null and r.external, and moves r.circle comment documenting the interface to a flag description.
…ainer'). The format option cannot be required otherwise running without parameters through the Python API does not work.
On many places (more than covered here), 2D raster mask is called MASK conflating the concept (mask) and the implementation (MASK raster). Users using the r.mask tool may not interact with the underlying raster directly, so there is no reason to use MASK over mask. This is changing MASK to mask and modifies related wording. However, this is also leaving many places as they are with MASK. Some will be better revisited with or after #2390 and #2392 when a more comprehensive solution is available. This fixes and keeps in sync wording in r.null and r.external, and moves r.circle comment documenting the interface to a flag description.
|
This is now ready for review. Example: > r.mask.status format=json
{
"present": true,
"full_name": "MASK@tmp_8ed35ec319de47fcb50dfef2245fef08",
"is_reclass_of": null
}
> r.mask -r
Raster MASK removed
> r.mask.status
Mask is not present |
Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
wenzeslaus
changed the title
Instead of using low-level test of file existence with hardcoded raster path and name, this offers a new tool to retrieve status of the raster mask.
The new r.mask.status tool reports presence or absence of the 2D raster mask and provides additional details about the mask.
There is one usage of this now and that's the one for a shell prompt. The prompt no longer relies on testing the file presence with the test program, but uses a GRASS tool to find out.
The code goes out of its way to report both mask name (currently always MASK) and the underlying raster name if it is a reclassified (without rewriting the current C API). This is to mimic the existing C functions which are returning the underlying raster if MASK is a reclass. The tool and the new C API function return both preparing a way for using an arbitrary name for the mask while having the option to look at the underlying reclassified raster map.