Fix crash on -a --self-check by adding exception handling to site check coroutines (#2466)

* Initial plan

* Fix crash on -a --self-check by adding exception handling in site_self_check and self_check

Wrap the body of site_self_check in try/except to catch unexpected errors
and always return a valid changes dict. Also add a safety-net try/except
in self_check around awaiting individual site check futures so that a
single site failure doesn't crash the entire self-check process.

Agent-Logs-Url: https://github.com/soxoj/maigret/sessions/5e27d620-5cbb-43d2-a9f9-ecb53a29904d

Co-authored-by: soxoj <31013580+soxoj@users.noreply.github.com>

* Restore @pytest.mark.slow on test_maigret_results

Agent-Logs-Url: https://github.com/soxoj/maigret/sessions/5e27d620-5cbb-43d2-a9f9-ecb53a29904d

Co-authored-by: soxoj <31013580+soxoj@users.noreply.github.com>

* Document --self-check error resilience, --auto-disable, and --diagnose in docs/

Update command-line-options.rst with expanded --self-check description
and new --auto-disable and --diagnose entries. Add a "Database self-check"
section to features.rst explaining error-resilient behaviour and usage
examples. Update usage-examples.rst to reference --auto-disable.

Agent-Logs-Url: https://github.com/soxoj/maigret/sessions/af1f0f09-9112-4902-8475-e81d235ff3ed

Co-authored-by: soxoj <31013580+soxoj@users.noreply.github.com>

---------

Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com>
Co-authored-by: soxoj <31013580+soxoj@users.noreply.github.com>
This commit is contained in:
Copilot
2026-04-07 19:44:09 +02:00
committed by GitHub
parent 6834483360
commit 23adc178ea
5 changed files with 222 additions and 119 deletions
+29
View File
@@ -170,6 +170,35 @@ Maigret will do retries of the requests with temporary errors got (connection fa
One attempt by default, can be changed with option ``--retries N``.
Database self-check
-------------------
Maigret includes a self-check mode (``--self-check``) that validates every site
in the database by looking up its known-claimed and known-unclaimed usernames
and verifying that the detection results match expectations.
The self-check is **error-resilient**: if an individual site check raises an
unexpected exception (e.g. a network error or a parsing failure), the error is
caught, logged, and recorded as an issue — the remaining sites continue to be
checked without interruption. This means the process always runs to completion,
even when checking hundreds of sites with ``-a --self-check``.
Use ``--auto-disable`` together with ``--self-check`` to automatically disable
sites that fail checks. Without it, issues are only reported. Use ``--diagnose``
to print detailed per-site diagnosis including the check type, specific issues,
and recommendations.
.. code-block:: console
# Report-only mode (no changes to the database)
maigret --self-check
# Automatically disable failing sites and save updates
maigret -a --self-check --auto-disable
# Show detailed diagnosis for each failing site
maigret -a --self-check --diagnose
Archives and mirrors checking
-----------------------------