Report damaged local state encoding through the installed CLI error contract

[rudycelekli]

A damaged UTF-8 state.json raises UnicodeDecodeError outside the installed client's DataSyncError boundary. Both offline search and sync can then raise a traceback instead of returning the documented structured local-state error.

Classify encoding failures alongside other unreadable/invalid state files as invalid_local_state. Preserve the original file and the existing error status/exit contract; no update or repair is performed automatically.

Validation: all three new regressions fail on unchanged f056cb9. Real invalid bytes exercise DataStore.state, installed sync --json, and installed search --json; after the fix, the CLI emits the structured error on stderr and leaves bytes unchanged. New regressions plus data-sync suite: 23 passed. No network fallback is used.

Full required CI sequence passed on an independent clean detached checkout, with recursive submodule initialization: Ruff check, Ruff format check, normalize-catalog, classify, build-data-release, and pytest (1,468 passed). This is the installed state-file boundary, separate from #722's catalog/index/shard decoding and #728's interrupted remote transfers. No source data or generated artifacts are changed.

Prepared with Codex (GPT) assistance. Hosted CI remains pending.

[rudycelekli]

330226 GPT — AI assistance was used to reproduce, implement, and test this change. Damaged UTF-8 local state now uses the existing JSON error contract; focused tests and the complete required CI sequence passed locally.

[ktwu01]

Hi @rudycelekli, thanks, adding UnicodeDecodeError to that except is the right fix. One trim request: please move tests/test_local_state_encoding.py into tests/test_data_sync.py, where the malformed-JSON invalid_local_state contract is already tested. The two cases read best side by side, and could even be one parametrized test (b"{not json" / b"\xff"). (Same pattern as noted on #749.)

---

Generated by Claude Code

[rudycelekli]

Moved the regression coverage into tests/test_data_sync.py, preserving every moved test/helper AST and the production fix. Rebased onto current main 3547a5c and reran all six required clean-checkout stages, including recursive submodules: all passed (1,635 tests). Updated head: 73b29da5764c32356ac8abd26bd88cf5254bfaae. Hosted checks are pending on that head.

330226 GPT-6 — OpenAI Codex assisted this repair; the test and generator results above were executed.

[ktwu01]

Closing as not planned. One author now has a cap of 10 open issues + PRs, and this account had 56 open PRs and 11 open issues, mostly paired with each other. We merged the three changes that fix silent data loss in the corpus (#723, #774, and #773 once rebased) and are closing the rest so review stays possible.

This is not a verdict that the change is wrong. If you think this one matters, reopen it later when you are under the cap, one at a time, with a reproduction from a real user or caller rather than a hypothetical edge case.

不是否定这个改动本身。为保证评审能跟上，同一作者最多同时保留 10 个未关闭的 issue/PR；该账号曾有 56 个未关闭 PR 和 11 个 issue。已合并修复语料静默丢失的三个改动，其余先关闭。若认为本条重要，请在低于上限后逐个重新打开，并附真实用户或调用方的复现。

---

Generated by Claude Code