Files

T

Cobra 45a8d6163e coe-mirror: mirror Nightmare_Eclipse Forgejo repos to pCloud

Standalone Python tool. Hits git.churchofmalware.org public API every
12h (00:00/12:00), downloads any repo HEAD whose SHA hasn't already
been snapshotted to pCloud, with local-staging fallback when the FUSE
mount is unavailable. Matrix notification stub for when the home
server is back.

Devil's-advocate findings folded into the design:
  - shutil.copy2 + size verify + unlink, not os.rename, for FUSE safety
  - no pCloud daemon auto-start (no user unit; DISPLAY=:0 unreliable);
    relies on Persistent timer + pending flush for recovery
  - two-call API path retained because the repo list response does not
    include commit.sha

Audit-driven security fixes:
  - repo names whitelist-validated (^[a-zA-Z0-9._-]+$)
  - urllib.parse.quote on all interpolated URL segments
  - same-file no-op guard in place_into_pcloud
  - existence check moved to main() so cached files don't generate
    spurious notify entries

13 pytest tests passing.

2026-06-11 09:34:55 -04:00

2.4 KiB

Raw Blame History

coe-mirror

Mirror Nightmare_Eclipse repos from Church of Malware to pCloud.

What it does

Monitors Nightmare_Eclipse on git.churchofmalware.org for new or updated repos. Downloads per-commit zip snapshots to ~/pCloudDrive/Hacking/Nightmare_Eclipse/. Falls back to local staging when pCloud is down. Runs at 00:00 and 12:00 daily with boot catch-up.

Filename convention

{repo}_{short_sha}.zip where short_sha is the first 8 chars of the HEAD commit SHA on the default branch. New commits produce new files; old snapshots stay as history.

Install

cd ~/tools/coe-mirror
bash install.sh

Manual run

systemctl --user start coe-mirror.service
# or directly:
python3 ~/tools/coe-mirror/coe_mirror.py

Logs

File: ~/.local/share/coe-mirror/coe-mirror.log (rotated at 5 MB, 3 backups)
Journal: journalctl --user -u coe-mirror.service -n 100

Storage estimate

8 repos × 2 runs/day × ~1 commit/repo/day ≈ 16 zips/day ≈ ~5 MB/month at current zip sizes (~265 KB each). Old snapshots accumulate — no automatic pruning. Manually prune ~/pCloudDrive/Hacking/Nightmare_Eclipse/ if needed.

Behavior when pCloud is down

Downloads land in ~/.local/share/coe-mirror/pending/. Next run (or any subsequent invocation) flushes them to pCloud once the mount is healthy. No auto-start of the pCloud daemon — by design, because the daemon is a GUI AppImage with no systemd user unit. If pCloud is offline, ensure the user session is active and pCloud is running before the next scheduled run, or trigger manually.

Catch-up after machine was off

The timer is Persistent=true and OnBootSec=2min. Missed windows run automatically on next boot. The tool also diffs against the pCloud directory, so even if a run is skipped entirely, the next run catches up everything that's missing.

Matrix notification

Currently a stub. Real implementation deferred until the home Matrix server is back. Reference: ~/tools/devtrack/scripts/matrix_alerts.py shows the auth + send pattern (Infisical creds MATRIX_USER_ID and MATRIX_ACCESS_TOKEN from the devtrack folder, POST to /_matrix/client/r0/rooms/{room_id}/send/m.room.message).

Testing

cd ~/tools/coe-mirror
python3 -m pytest tests/ -v

Uninstall

systemctl --user disable --now coe-mirror.timer
rm ~/.config/systemd/user/coe-mirror.{service,timer}
systemctl --user daemon-reload

2.4 KiB Raw Blame History Unescape Escape