fix: ignore legacy manual notes during sync#60
Conversation
Legacy or hand-edited Markdown files that lack a NexaNote frontmatter
id were being re-imported on every sync, producing duplicate notes and
weird "__Md…" titles derived from synthesised slugs.
Add a per-data-dir SyncState registry that tracks adopted remote paths
(remote_path → local_id) and ignored remote paths (with reasons). The
pull flow now resolves a remote note in this order:
1. by note id (frontmatter id matches a local note),
2. by remote_path (we adopted this path under another id before),
3. otherwise — if the id looks legacy/synthetic — record an ignore
marker and skip silently on every later sync.
The SyncReport gains a `notes_ignored_legacy` counter so the diagnostic
summary surfaces how many manual files were skipped.
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: f18bc6b274
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| if local_note is None and legacy_id: | ||
| reason = ( | ||
| "no NexaNote frontmatter id; legacy/manual Markdown file" |
There was a problem hiding this comment.
md. note ids
The new legacy guard treats every id starting with md. as non-importable, so a client can never pull those notes on first sync even when they were produced by NexaNote itself (plain-markdown notes are synthesized with PLAIN_MD_ID_PREFIX = "md." in FileNoteStore and are pushed with that id). In a multi-device setup, device A can upload such a note, but device B will always skip it at this branch and never materialize the note locally, which is a functional sync regression rather than just duplicate prevention.
Useful? React with 👍 / 👎.
| if self.sync_state.is_ignored(remote_path): | ||
| self.sync_state.touch_ignored(remote_path) | ||
| report.notes_ignored_legacy += 1 | ||
| return |
There was a problem hiding this comment.
This early return makes ignore decisions sticky forever for a given remote_path: once a path is marked ignored, pull never fetches note.json again, so it cannot recover if that remote folder is later fixed to include a valid NexaNote id. In practice, a user who migrates or repairs a previously legacy/manual note at the same path will still be skipped indefinitely unless they manually delete .nexanote_sync_state.json.
Useful? React with 👍 / 👎.
2 participants
Summary
Stops the WebDAV sync engine from re-importing legacy / hand-edited
Markdown files (no NexaNote frontmatter id) on every pull, which was
producing duplicate notes and
__Md…title artifacts derived fromsynthesised slugs.
nexanote/sync/sync_state.py— per-data_dirJSON registrytracking
adoptedremote paths (with the local note id they map to)and
ignoredremote paths (with a reason and timestamps). Loaded atengine init, saved at the end of every
sync()even on errors._pull_notenow resolves a remote note in three stages:md.…synthesised prefix ormissing) — record an ignore marker and skip silently on every
later sync. A previously-adopted path whose local note has been
purged is also ignored, never re-imported.
SyncReport.notes_ignored_legacysurfaces the count in thediagnostic summary.
Test plan
tests/test_sync_legacy_ignore.py— 18 new tests covering:.md(no NexaNote id) is ignored, never imported,__Md…title artifacts leak in,remote_pathfallback avoids duplicates when ids change,SyncStatehandles missing/corrupt JSON gracefully.pytest tests/ -q→ 279 passed).Generated by Claude Code