[echoix]

This is a first start of fixing issues that I noted in comments in Weblate, for python files.

Description

Some messages were built from separately translated fragments, or from a runtime string passed to gettext, so translators cannot adapt word order or punctuation. These were flagged by comments on Weblate. This PR replaces them with complete format strings in the Python tools and the wxGUI; the C code is in a separate PR.

• r.reclass.area: complete messages, one per combination of thresholds.
• r.in.wms, v.what.strds, v.rast.stats, r.in.srtm, m.proj: error and status messages are single strings (no trailing-space fragments).
• Temporal framework (core.py, temporal_algebra.py): the mapset error and the wrong-input error were built before being passed to gettext, so they were never translated.
• wxGUI: render status bar, scatterplot title and axis labels, database table tab labels, GCP overwrite question, toolbar "Set map canvas for ... / Zoom to map" tooltips, parameter-dialog status text, mapset/type selector labels, model condition, layer group, psmap labels and distance measuring log are complete strings.
• wxGUI RLi sample area title was _("Select sample area 1 of " + str(n)); it now uses the same msgid as the update just above it.
• wxGUI tplot: the CSV error message ended in "or " with nothing after it. It now reads "Please change name of output CSV file or enable overwriting." Please check the wording matches the intent.
• wxGUI preferences: removed the trailing space in "Random colors according to category number".

Motivation and context

Translators on Weblate reported that these messages cannot be translated correctly with a different word order.

How has this been tested?

ruff format and ruff check report no changes; the modified files compile.

Types of changes

• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause existing
  functionality to not work as before)

Checklist

• PR title provides summary of the changes and starts with one of the
  pre-defined prefixes
• My code follows the code style
  of this project.
• My change requires a change to the documentation.
• I have updated the documentation accordingly.
• I have added tests to cover my changes.

AI assistance (Claude) was used to find the affected strings and to draft these changes.

[echoix]

Sorry for the noise. On a fresh mind, I relooked at the PR. I fine tuned some little extra messages that weren't in the initial PR, but that had similar messages than the ones that were changed in this PR, such as all of the similar messages are fixed in the same way.

Ai reviewed the different branches of r.reclass.area, where now the message isn't built by parts. See the table in the full output of the last review I made it do, even if it's worded for me, the table is still fine and cross-verifies the work:

Details

CI is still running on the new head. I'm checking the r.reclass.area message branches for each threshold combination.

I updated my local copy to your new head, f16b010; your latest commit there names the placeholder {pixel_count}. Then I reviewed the whole PR again against main (9c7baf5): 11 commits, 29 files. I found no correctness problems.

What changed since my last review is only in python/grass/temporal/core.py and scripts/r.reclass.area/r.reclass.area.py. The other 27 files are the same as when I first reviewed them.

• core.py: all six Unable to … {reason} messages are now complete strings. _create_mapset_error_message() is translated too, so the whole error can be translated. No call site still builds a msgid at runtime.

• r.reclass.area: I ran the threshold branching from the file under seven input combinations. In each case the message matches the range actually applied:

  lower, upper	Message	Range applied
  1, 5	≥ 1.0 and ≤ 5.0 hectares	1 to 5
  1, none	≥ 1.0 hectares	1 to ∞
  none, 5	≤ 5.0 hectares	−∞ to 5
  0, none	"without area size limits"	−∞ to ∞
  0, 5	≤ 5.0 hectares	−∞ to 5
  2, 0	≥ 2.0 hectares	2 to ∞
  0.00001, none	≥ 1e-05 hectares	0.00001 to ∞

  The "None" case is gone, and upper=0 correctly counts as no upper limit.

• Both files parse and pass ruff check and ruff format --check.

I didn't build GRASS, so I didn't run the tool or its testsuite/. CI on f16b010 is still running: two checks have passed and 17 are queued or in progress.

Nits (all optional):

1. Very small thresholds print in scientific notation. A value like 0.00001 hectares now prints as 1e-05; the old code printed 0.000010. Values that small are unlikely, so I'd leave it.
2. Commit history. This only matters if the PR isn't squash-merged:
   • Four commits have no category prefix: e24da30, 5e1fdee, 53a4b7c and f16b010.
   • 1f55ab1 still has a Co-Authored-By: Claude trailer, which AGENTS.md asks to avoid.
   • e24da30 lists you as your own co-author. GitHub adds that trailer when you commit a suggestion.
   • Five commits are unsigned: 397ec0a, 4067a2c, 5d5fbba, 1f55ab1 and 9705e3b. Your d265a24 and f16b010 are also unsigned.
3. PR description. "How has this been tested?" still says only ruff and compiling. You could mention that the r.reclass.area messages were checked against each threshold combination. The core.py bullet could also say that the mapset-access reason is now translated.

I haven't posted any of this to the PR.

[echoix]

@ninsbl can you take a look here? Just to validate that I didn't miss something big (and AI either), I think it's fine

[ninsbl]

Looks good to me.