[darkrei08]

tools/profile.py keeps one household profile for TVs that should look the same: capture what a master unit has installed and disabled, check another unit against it, and sync the missing apps and the disabled set over ADB. Nothing is uninstalled; the exit codes are the ones check.py already uses.

• capture: writes profiles/<name>/ with the package lists and a JSON summary.
• check: compares a unit against the profile and prints what drifts.
• sync --source: installs the missing apps (split APKs in one session, abandoned when the TV refuses the commit) and mirrors the disabled set.

profiles/master/ is the profile measured on a Chromecast with Google TV 4K: 20 apps and 16 disabled packages. The guide is docs/profiles.md, and docs/TROUBLESHOOTING.md gains the note that ADB never stays on by itself, which the sync depends on.

Tested on: Chromecast with Google TV 4K (sabrina, Android 14), a second unit of the same model synced from it. tools/venv/bin/python -m unittest discover -s tests passes (2869 tests, 37 skipped).

Second of the three PRs the bot asked for on #3214; settings_sync.py and the launcher change come separately.

If this adds or changes a TV list

• not a TV list

If this touches the launcher

• not the launcher

If this adds or changes a theme

• not a theme

If this adds a theme key (a switch)

• not a switch

If this adds an add-on

• not an add-on

[github-actions]

maintainer-bot review · this one needs a maintainer decision

Thanks for the careful split from #3214 and the fake-TV harness, which lets sync be tested without a set. Copying apps between TVs is a new kind of feature, so this needs a scope call from the maintainer. Before that, it would help to send the disabling through disable.sh and to fold the drift check into check.py.

Why it goes to a maintainer: the bot asks for a maintainer decision.

Checks

• Note: possibly collides with Launcher: reach a TV's settings when they are a service, not an activity #3198 (both PRs modify docs/TROUBLESHOOTING.md with changes to different sections that may overlap during merge)
• Size: size/L (+945 -0, 11 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}

[darkrei08]

A reader's map for this one:

• tools/profile.py: capture is the only writer under profiles/; check and plan are read-only and produce the drift; sync is plan plus the installs and the disable calls. Nothing in it calls uninstall.
• install_split uses one install-create session and abandons it when the TV refuses the commit, so a partial split install cannot leave the unit half-installed.
• tests/profile_fake.py gives each simulated unit its own state file, keyed by TV_HOST, which is why a test can hold a master and a target at once.
• profiles/master/ is the measurement from the description; it is data, and a different model needs its own folder rather than an edit here.

Not tested: a unit whose Android version differs from the master's.

[darkrei08]

Picked up both notes.

The disabling step now refuses the core packages. sync filters its disable list through recovery.core_packages, which is the same pattern tests/test_core_packages.py keeps identical to disable.sh and disable.ps1, so the guard the boot-safety issue in #884 relies on applies here too, and it cannot drift from the two scripts. A refused name is printed with the reason and the rest of the sync still runs. tests/test_profile_sync.py pins it: with com.android.systemui in the profile, no pm disable-user for it reaches the TV, the refusal is on stderr, and a normal package in the same run still lands. Suite: 2870 tests, 0 failures.

What is still genuinely the maintainer's call: the whole destructive step could instead be delegated to tools/disable.sh with a temporary list, which would also bring the Home guard and the ~/.cleanyourtv recovery record for free. I did not do that here because it changes the tool's contract (disable.sh owns a debloat run and its record, sync owns a convergence plan), so I would rather implement it on your word than guess. Say which you prefer and I will.

On the #3198 collision: the two edits are not in the same region. That PR's docs/TROUBLESHOOTING.md hunk is around line 1135 (a Sony settings key that is a service rather than an activity); the note this PR adds sits at line 58, in the wireless-debugging block. A merge of both is two disjoint insertions. If you would still rather keep this PR to the tool alone, I can drop the note: it belongs to the ADB story, not to profile.py.

[Gathin23]

Thanks @darkrei08, the fake-TV harness and the core-package guard in sync are careful work. Cloning a household's apps from one TV to others is outside the debloat core, so it fits best in your own repo with one SHOWCASE.md row pointing to it. Please drop the always-on ADB note and adbautoenable there too. We never point people at keeping ADB on at boot.