[alteralto]

v.db.addtable replaced only $MAP/ in the database path, so a per-map database named after the map, such as $GISDBASE/$LOCATION_NAME/$MAPSET/sqlite/$MAP.db, reached db.tables with $MAP still in it and the tool failed with "Unable to describe table".

$MAP is now replaced wherever it appears, as Vect_subst_var() and pygrass do. $MAPSET is left for the driver to expand: the library substitutes it before $MAP, and the script now skips it explicitly, which is what the trailing slash used to guard against.

The new pytest test (scripts/v.db.addtable/tests/, using xy_empty_dataset_session) adds a table to a fresh vector for two connections, sqlite/$MAP.db and the existing vector/$MAP/sqlite.db, and checks the column and the database file.

Tested in the osgeo/grass-gis:main-alpine image (8.6.0dev): on main the sqlite/$MAP.db case fails with the db.tables error from the issue and the vector/$MAP/sqlite.db case passes; with this change both pass. The reproducer from the issue also works, and so does the default sqlite/sqlite.db connection.

Fixes #8106

[nilason]

Hmm, https://github.com/OSGeo/grass/blob/main/CONTRIBUTING.md !!

[alteralto]

I now checked this md doc again. Thanks for the pointer. When you looked, the description still said I was running the tests; they're done now, and I've reworked the test to follow the guide: it's pytest in scripts/v.db.addtable/tests/ (with xy_empty_dataset_session), since it needs no sample data, and the testsuite change is reverted. It covers the sqlite/$MAP.db form from the issue and the vector/$MAP/sqlite.db form that already worked. In the osgeo/grass-gis:main-alpine image the first case fails on main with the db.tables error from the issue, and both pass with this change. If you meant something else in CONTRIBUTING let me know please