Commit Graph

3 Commits

Author SHA1 Message Date
9f271b0d37 Fix the cap hook in CI, and the re-exports its --all-files run exposed
The hook entry was an absolute /home/kuhy/utils path, copied from a config
written for ~/.claude where it only ever ran locally. On an Actions runner
that is exit code 127 — "No such file or directory" — so gatelock-precommit
and crdt-sync-precommit failed on every commit of this refactor while the
hook silently checked nothing. Now repo-relative, matching the sibling hooks
(`entry: scripts/check_subproject_scope.sh`): pre-commit always runs from the
git root, so it resolves the same in both places.

Fixing that made the CI configs runnable locally, and `--all-files` then found
what per-file runs had missed — ruff pruning imports that exist *for* their
re-export. `import crdt_sync` was broken three separate ways (FileSyncStateStore,
encode_key, FileCredentialStore). Each affected module now states an explicit
__all__ so the autofixer cannot take them again, and the two files that pushed
past the cap once __all__ was added are split further:

  _firebase.py 260 -> 247 + _fberrors 25
  _sync.py     257 -> 188 + _syncargs 91

Test patch targets followed the code they patch: gatelock._outputs.shutil ->
_xrandr.shutil, gatelock._arbiter.Path -> _claims.Path.

gatelock and crdt-sync both pass `pre-commit run --all-files` in full;
389 and 259 tests at 100% branch coverage; the root gate exits 0.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01CFSGeWxHPY5rHUGrPqhcmG
2026-08-22 15:52:59 +02:00
3589d166cd crdt-sync: split five source files — package now fully under the cap
Some checks are pending
crdt-sync pre-commit / pre-commit (push) Waiting to run
crdt-sync tests / test (3.10) (push) Waiting to run
crdt-sync tests / test (3.11) (push) Waiting to run
crdt-sync tests / test (3.12) (push) Waiting to run
gatelock pre-commit / pre-commit (push) Waiting to run
gatelock tests / test (3.10) (push) Waiting to run
gatelock tests / test (3.11) (push) Waiting to run
gatelock tests / test (3.12) (push) Waiting to run
palette drift / palette (push) Waiting to run
_sync.py         456 -> 245 + _syncstate 131 + _pull 106 + _revisions 34
  _firebase_auth   412 -> 242 + _credentials 146 + _autherrors 69
  _firebase.py     336 -> 250 + _fbkeys 100
  google_id_token  316 -> 232 + tool/_oauth_callback 105
  _github.py       255 -> 238 + _gherrors 30

revision_of became its own leaf module because both the push and pull halves
need it; without that the two would have imported each other. _fbkeys holds
the RTDB key escaping — the escape tables moved with the functions that read
them, since the pair has to round-trip exactly or a device silently stops
seeing its own writes.

crdt-sync/tool/ is outside the coverage source, so the split there has no test
net: verified by running it the documented way instead
(`python3 -m tool.google_id_token --help` prints its usage, and
`from tool.google_id_token import TokenError, fetch_id_token` — what
seed_session.py does — still resolves).

An AST pass over each pair caught what the suite would not have: _is_revoked
needs the error type, so it stayed with it rather than following the
credentials; _TERMINAL_AUTH_REASONS and _CONSENT_TIMEOUT_SECONDS had to be
carried to their new callers.

259 passed at 100% branch coverage. crdt-sync/tool/seed_session.py and
firebase-debug.log are pre-existing local changes and are deliberately not in
this commit.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01CFSGeWxHPY5rHUGrPqhcmG
2026-08-22 15:30:52 +02:00
5f64a8959b Sign in with Google, and stop trusting a revoked session
Some checks failed
crdt-sync pre-commit / pre-commit (push) Has been cancelled
crdt-sync tests / test (3.10) (push) Has been cancelled
crdt-sync tests / test (3.11) (push) Has been cancelled
crdt-sync tests / test (3.12) (push) Has been cancelled
Two changes to the shared sync credential path, both driven by devices
that reported "Connected" while every sync failed.

Google sign-in (signInWithGoogle, both languages) replaces typing the
sync account's password on a fresh install. The account is linked once
from the PC with tool/link_google.py, which asserts the pinned uid --
without that assertion an unlinked identity is *signed up* rather than
refused, and the new uid is then denied every read and write by the
security rules.

The second change is the one with teeth. hasSession() is a presence
check, so a refresh token the server had revoked still read as a live
session: the desktop todo build showed "Connected" and then failed every
sync with `HTTP 400 (TOKEN_EXPIRED)`. A refresh that fails for a terminal
reason now clears the stored session, so hasSession() answers honestly.

Deliberately narrow: only TOKEN_EXPIRED, USER_DISABLED, USER_NOT_FOUND,
INVALID_REFRESH_TOKEN, INVALID_GRANT_TYPE and MISSING_REFRESH_TOKEN
clear it. A network error or a 5xx must not -- a device that signs
itself out whenever the wifi drops needs a manual sign-in on every
device to recover, which is worse than the stale banner this fixes.
Verified against the live endpoint, not just mocks: has_session flips
True -> False on a real rejection.

crdt_sync_dart 0.9.0 -> 0.10.0; the five apps pin the matching tag.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0192EkoAyFBEWTvAN5vVgmSy
2026-08-11 23:58:02 +02:00