[M-khalifa]

b64push.py counted the final move as done when "moved" appeared anywhere in the TV's answer, so a refused mv whose path contains the word (a target such as /data/local/tmp/removed.jpg) skipped the copy fallback and exited 0 with the target unchanged; manage.py read pm's Success the same way, so a failed add-on install whose message names a package containing the word counted as installed. Both now match the marker only as a whole line. In manage.py this covers install for an add-on without its own install script, and remove. Add-ons that ship their own install.sh or install.ps1 run pm inside that script, which still accepts Success anywhere in pm's answer; those scripts are a follow-up and are not touched here. Found by reading the code; the new test reproduces both with a fake TV.

Tested on: no TV needed (fake TV client); Windows 11, Python 3.11.

• python -m unittest discover -s tests: 3172 tests, no failure from this change; 2 errors in test_bench_other_app and test_bench_replay, which fail to import here (from tests import ... resolves to another tests package in site-packages; the same on every branch, files not touched); 494 skipped (the POSIX-shell and pwsh suites, which CI runs).
• New tests/test_success_lines.py (6 tests, no shell needed): the refused move that names moved exits 1 through the copy fallback; a failed install and a failed removal whose Failure line names a Success package are failures; Success with surrounding whitespace still counts; Success still counts for install and remove. With the whole-line match reverted, the two Failure-line cases fail; with the whitespace trimming reverted, the trimming case fails.

Related: #1588 (the copy fallback this keeps reachable)

Reviewed with cross-review 1.2.1 (SkillArray, MIT license): https://github.com/M-khalifa/SkillArray

[github-actions]

maintainer-bot review · suggests merge (p2)

Thanks for the careful fix and for the fake-TV test that reproduces both false successes. This looks ready to merge as it is, and the add-on install scripts make a good follow-up.

Checks

• Size: size/M (+142 -3, 3 files)

This is a suggestion; a maintainer makes the final call. A maintainer can comment /bot review to run it again.

_{Reviewed by maintainer-bot · runs on 0G Private Computer}

[Gathin23]

Thanks @M-khalifa, b64push.py and manage.py now accept 'moved' and 'Success' only as whole lines, so a refusal that happens to name such a path or package no longer counts as done.