@clearcms/index

DB-rebuilder for the clear content protocol. Walks a clear bucket and reconstructs the libSQL items_index + item_revisions rows from the JSON files inside.

Why

Files = source of truth. The DB is a derived index. If you rsync a bucket between machines, restore a backup, or edit files outside the admin, the admin auto-heals on next boot — but to do that, something has to walk the bucket and rebuild the rows.

That's this package.

Usage

import { rebuild } from '@clearcms/index';

const result = await rebuild({
  bucketRoot: '/home/ms/clear/markonian/bucket',
  databaseUrl: 'file:/home/ms/clear/markonian/clear.db',
  mode: 'full',         // 'full' wipes items + revisions + drafts and re-derives them
  // mode: 'incremental' upserts only — leaves rows the bucket doesn't know about alone
});

// result: { itemsScanned, itemsWritten, revisionsScanned, revisionsWritten, errors, durationMs }

What's in scope

• collections/<type>/<locale>/<slug>.json → items_index row (status='published')
• drafts/<type>/<locale>/<slug>.json → items_index row (status='draft')
• trash/<type>/<locale>/<slug>.json → items_index row with deletedAt set
• revisions/<type>/<locale>/<slug>/<ISO>-<author>.json → item_revisions row

What's NOT in scope

• Collection / field schemas — these live in the DB only today; full mode intentionally does not wipe collections, fields, settings, members, sessions, api_tokens. A bucket-resident manifest format is out of scope for v0.
• Globals (content/globals/*.json) — served directly off the bucket; no derived index.
• Media (media/*) — already indexed in the media table elsewhere.

CLI

clear-admin reindex [--mode=full|incremental] [--collection=<slug>] [--dry-run]

Inputs come from env vars (CLEAR_STORAGE_ROOT, CLEAR_DB_URL); see clear-admin --help.