emdash 0.40.0 → 0.41.0

package/dist/fts-manager-CCaZTp8y.mjs

@@ -0,0 +1,443 @@
+import { u as __exportAll } from "./runner-B3eyB9Gb.mjs";
+import { t as validateIdentifier } from "./validate-DANX1-Qi.mjs";
+import { d as isSqlite, y as tableExists } from "./dialect-helpers-7_IqPdh9.mjs";
+import { t as SEARCH_TOKENIZERS } from "./types-CkE41ZAG.mjs";
+import { sql } from "kysely";
+
+//#region src/search/fts-manager.ts
+var fts_manager_exports = /* @__PURE__ */ __exportAll({ FTSManager: () => FTSManager });
+const DEFAULT_SEARCH_TOKENIZER = "porter unicode61";
+function isSearchTokenizer(value) {
+	return SEARCH_TOKENIZERS.some((tokenizer) => tokenizer === value);
+}
+function resolveSearchTokenizer(tokenize) {
+	if (tokenize === void 0) return DEFAULT_SEARCH_TOKENIZER;
+	if (!isSearchTokenizer(tokenize)) throw new Error(`Unsupported FTS5 tokenizer: "${String(tokenize)}"`);
+	return tokenize;
+}
+/**
+* FTS5 Manager
+*
+* Handles creation, deletion, and management of FTS5 virtual tables
+* for full-text search on content collections.
+*/
+var FTSManager = class {
+	constructor(db) {
+		this.db = db;
+	}
+	/**
+	* Validate a collection slug and its searchable field names.
+	* Must be called before any raw SQL interpolation.
+	*/
+	validateInputs(collectionSlug, searchableFields) {
+		validateIdentifier(collectionSlug, "collection slug");
+		if (searchableFields) for (const field of searchableFields) validateIdentifier(field, "searchable field name");
+	}
+	/**
+	* Get the FTS table name for a collection
+	* Uses _emdash_ prefix to clearly mark as internal/system table
+	*/
+	getFtsTableName(collectionSlug) {
+		validateIdentifier(collectionSlug, "collection slug");
+		return `_emdash_fts_${collectionSlug}`;
+	}
+	/**
+	* Get the content table name for a collection
+	*/
+	getContentTableName(collectionSlug) {
+		validateIdentifier(collectionSlug, "collection slug");
+		return `ec_${collectionSlug}`;
+	}
+	/**
+	* Check if an FTS table exists for a collection
+	*/
+	async ftsTableExists(collectionSlug) {
+		const ftsTable = this.getFtsTableName(collectionSlug);
+		return tableExists(this.db, ftsTable);
+	}
+	/**
+	* Create an FTS5 virtual table for a collection.
+	* FTS5 is SQLite-only; on other dialects this is a no-op.
+	*
+	* @param collectionSlug - The collection slug
+	* @param searchableFields - Array of field names to index
+	* @param weights - Optional field weights for ranking
+	* @param tokenize - Optional FTS5 tokenizer configuration
+	*/
+	async createFtsTable(collectionSlug, searchableFields, _weights, tokenize) {
+		const tokenizer = resolveSearchTokenizer(tokenize);
+		if (!isSqlite(this.db)) return;
+		this.validateInputs(collectionSlug, searchableFields);
+		const ftsTable = this.getFtsTableName(collectionSlug);
+		const columns = [
+			"id UNINDEXED",
+			"locale UNINDEXED",
+			...searchableFields
+		].join(", ");
+		await sql.raw(`
+			CREATE VIRTUAL TABLE IF NOT EXISTS "${ftsTable}" USING fts5(
+				${columns},
+				tokenize='${tokenizer}'
+			)
+		`).execute(this.db);
+		await this.createTriggers(collectionSlug, searchableFields);
+	}
+	/**
+	* SQL expression producing the indexed value for one searchable field.
+	*
+	* Portable Text fields are stored as JSON; indexing the raw JSON pollutes
+	* the index with structural tokens (`_type`, style values like `normal`,
+	* `_key` ULIDs) and makes snippets show JSON fragments. Extract the prose
+	* instead: every JSON string under a `text`, `alt`, `caption`, or `code`
+	* key (span text, image alt/caption, code blocks). This is a superset of
+	* `extractPlainText` in text-extraction.ts, which walks only known block
+	* shapes — the SQL variant takes those keys at any depth, so search and
+	* the extractPlainText consumers (vectorize/ai-search) can see different
+	* text for the same document. Only JSON documents
+	* (arrays/objects) are extracted; legacy rows holding a bare string or a
+	* JSON scalar (`Some title`, `2024`) are indexed as-is. Extraction must
+	* live in SQL because the sync triggers cannot call into JS.
+	*
+	* `ref` must be a validated column reference (`NEW.x`, `OLD.x`, `"x"`).
+	*/
+	searchValueExpr(ref, fieldType) {
+		if (fieldType !== "portableText") return ref;
+		return `CASE WHEN ${ref} IS NULL THEN NULL WHEN json_valid(${ref}) AND json_type(${ref}) IN ('array', 'object') THEN (SELECT group_concat(j.value, ' ') FROM json_tree(${ref}) AS j WHERE j.key IN ('text', 'alt', 'caption', 'code') AND j.type = 'text') ELSE ${ref} END`;
+	}
+	/**
+	* Field type per slug for a collection, for choosing the indexed-value
+	* expression. Fields missing from the schema fall back to raw indexing.
+	*/
+	async getFieldTypes(collectionSlug) {
+		const collection = await this.db.selectFrom("_emdash_collections").select("id").where("slug", "=", collectionSlug).executeTakeFirst();
+		if (!collection) return /* @__PURE__ */ new Map();
+		const rows = await this.db.selectFrom("_emdash_fields").select(["slug", "type"]).where("collection_id", "=", collection.id).execute();
+		return new Map(rows.map((r) => [r.slug, r.type]));
+	}
+	/**
+	* Create triggers to keep FTS table in sync with content table.
+	*
+	* The insert and update triggers only add rows to the FTS index when
+	* `deleted_at IS NULL`. This keeps soft-deleted content out of the
+	* search index and ensures the FTS row count matches the non-deleted
+	* content count (which `verifyAndRepairIndex` relies on).
+	*
+	* The FTS table stores its own values (no `content=` option), so removal
+	* is a plain `DELETE FROM fts WHERE rowid = OLD.rowid` — a harmless no-op
+	* for rows that were never indexed (soft-deleted content). The
+	* external-content `'delete'`-command choreography and its corruption
+	* modes (migration 039) do not apply to self-contained tables.
+	*
+	* `INSERT OR REPLACE` keeps the insert path idempotent: re-running a
+	* populate (search repair runs per isolate, so two isolates can race) converges
+	* on one index row per content row instead of failing on the rowid
+	* constraint.
+	*
+	* The trigger SQL emitted here MUST stay in lock-step with migration
+	* `064_fts_plain_text.ts`. If this changes again, add a new migration
+	* rather than editing shipped ones — migrations are forward-only.
+	*/
+	async createTriggers(collectionSlug, searchableFields) {
+		this.validateInputs(collectionSlug, searchableFields);
+		if (searchableFields.length === 0) throw new Error(`Cannot create FTS triggers for collection "${collectionSlug}": no searchable fields. Mark at least one field as searchable before enabling search.`);
+		const ftsTable = this.getFtsTableName(collectionSlug);
+		const contentTable = this.getContentTableName(collectionSlug);
+		const fieldTypes = await this.getFieldTypes(collectionSlug);
+		const fieldList = searchableFields.join(", ");
+		const newValueList = searchableFields.map((f) => this.searchValueExpr(`NEW.${f}`, fieldTypes.get(f))).join(", ");
+		await sql.raw(`
+			CREATE TRIGGER IF NOT EXISTS "${ftsTable}_insert"
+			AFTER INSERT ON "${contentTable}"
+			WHEN NEW.deleted_at IS NULL
+			BEGIN
+				INSERT OR REPLACE INTO "${ftsTable}"(rowid, id, locale, ${fieldList})
+				VALUES (NEW.rowid, NEW.id, NEW.locale, ${newValueList});
+			END
+		`).execute(this.db);
+		const changedCondition = [
+			"deleted_at",
+			"locale",
+			...searchableFields
+		].map((f) => `OLD.${f} IS NOT NEW.${f}`).join(" OR ");
+		await sql.raw(`
+			CREATE TRIGGER IF NOT EXISTS "${ftsTable}_update"
+			AFTER UPDATE ON "${contentTable}"
+			WHEN ${changedCondition}
+			BEGIN
+				DELETE FROM "${ftsTable}" WHERE rowid = OLD.rowid;
+				INSERT INTO "${ftsTable}"(rowid, id, locale, ${fieldList})
+				SELECT NEW.rowid, NEW.id, NEW.locale, ${newValueList}
+				WHERE NEW.deleted_at IS NULL;
+			END
+		`).execute(this.db);
+		await sql.raw(`
+			CREATE TRIGGER IF NOT EXISTS "${ftsTable}_delete"
+			AFTER DELETE ON "${contentTable}"
+			BEGIN
+				DELETE FROM "${ftsTable}" WHERE rowid = OLD.rowid;
+			END
+		`).execute(this.db);
+	}
+	/**
+	* Drop triggers for a collection
+	*/
+	async dropTriggers(collectionSlug) {
+		this.validateInputs(collectionSlug);
+		const ftsTable = this.getFtsTableName(collectionSlug);
+		await sql.raw(`DROP TRIGGER IF EXISTS "${ftsTable}_insert"`).execute(this.db);
+		await sql.raw(`DROP TRIGGER IF EXISTS "${ftsTable}_update"`).execute(this.db);
+		await sql.raw(`DROP TRIGGER IF EXISTS "${ftsTable}_delete"`).execute(this.db);
+	}
+	/**
+	* Drop the FTS table and triggers for a collection
+	*/
+	async dropFtsTable(collectionSlug) {
+		if (!isSqlite(this.db)) return;
+		this.validateInputs(collectionSlug);
+		const ftsTable = this.getFtsTableName(collectionSlug);
+		await this.dropTriggers(collectionSlug);
+		await sql.raw(`DROP TABLE IF EXISTS "${ftsTable}"`).execute(this.db);
+	}
+	/**
+	* Rebuild the FTS index for a collection
+	*
+	* This is useful after bulk imports or if the index gets out of sync.
+	*/
+	async rebuildIndex(collectionSlug, searchableFields, weights, tokenize) {
+		resolveSearchTokenizer(tokenize);
+		if (!isSqlite(this.db)) return;
+		await this.dropFtsTable(collectionSlug);
+		await this.createFtsTable(collectionSlug, searchableFields, weights, tokenize);
+		await this.populateFromContent(collectionSlug, searchableFields);
+	}
+	/**
+	* Populate the FTS table from existing content.
+	*
+	* `INSERT OR REPLACE` so a concurrent double-populate (search repair runs
+	* per isolate) converges instead of failing on the rowid constraint.
+	*/
+	async populateFromContent(collectionSlug, searchableFields) {
+		if (!isSqlite(this.db)) return;
+		this.validateInputs(collectionSlug, searchableFields);
+		const ftsTable = this.getFtsTableName(collectionSlug);
+		const contentTable = this.getContentTableName(collectionSlug);
+		const fieldTypes = await this.getFieldTypes(collectionSlug);
+		const fieldList = searchableFields.join(", ");
+		const valueList = searchableFields.map((f) => this.searchValueExpr(`"${contentTable}"."${f}"`, fieldTypes.get(f))).join(", ");
+		await sql.raw(`
+			INSERT OR REPLACE INTO "${ftsTable}"(rowid, id, locale, ${fieldList})
+			SELECT rowid, id, locale, ${valueList} FROM "${contentTable}"
+			WHERE deleted_at IS NULL
+		`).execute(this.db);
+	}
+	/**
+	* Index up to `limit` content rows with `rowid > afterRowid`, in rowid
+	* order, so a large collection can be populated across several requests.
+	* Returns the last rowid covered, or null when no rows remain. Returns
+	* null on dialects without FTS5.
+	*/
+	async populateRange(collectionSlug, searchableFields, afterRowid, limit) {
+		if (!isSqlite(this.db)) return null;
+		this.validateInputs(collectionSlug, searchableFields);
+		const ftsTable = this.getFtsTableName(collectionSlug);
+		const contentTable = this.getContentTableName(collectionSlug);
+		const last = (await sql`
+			SELECT MAX(rowid) AS last FROM (
+				SELECT rowid FROM ${sql.ref(contentTable)}
+				WHERE rowid > ${afterRowid}
+				ORDER BY rowid
+				LIMIT ${limit}
+			)
+		`.execute(this.db)).rows[0]?.last ?? null;
+		if (last === null) return null;
+		const fieldTypes = await this.getFieldTypes(collectionSlug);
+		const fieldList = searchableFields.join(", ");
+		const valueList = searchableFields.map((f) => this.searchValueExpr(`"${contentTable}"."${f}"`, fieldTypes.get(f))).join(", ");
+		await sql`
+			INSERT OR REPLACE INTO ${sql.ref(ftsTable)}(rowid, id, locale, ${sql.raw(fieldList)})
+			SELECT rowid, id, locale, ${sql.raw(valueList)} FROM ${sql.ref(contentTable)}
+			WHERE deleted_at IS NULL AND rowid > ${afterRowid} AND rowid <= ${last}
+		`.execute(this.db);
+		return Number(last);
+	}
+	/**
+	* Get the search configuration for a collection
+	*/
+	async getSearchConfig(collectionSlug) {
+		const result = await this.db.selectFrom("_emdash_collections").select(["search_config", "title_field"]).where("slug", "=", collectionSlug).executeTakeFirst();
+		if (!result?.search_config) return null;
+		try {
+			const parsed = JSON.parse(result.search_config);
+			if (typeof parsed !== "object" || parsed === null || !("enabled" in parsed) || typeof parsed.enabled !== "boolean") return null;
+			const config = { enabled: parsed.enabled };
+			if (result.title_field) config.titleField = result.title_field;
+			if ("weights" in parsed && typeof parsed.weights === "object" && parsed.weights !== null) {
+				const weights = {};
+				for (const [k, v] of Object.entries(parsed.weights)) if (typeof v === "number") weights[k] = v;
+				config.weights = weights;
+			}
+			if ("tokenize" in parsed) {
+				if (!isSearchTokenizer(parsed.tokenize)) return null;
+				config.tokenize = parsed.tokenize;
+			}
+			return config;
+		} catch {
+			return null;
+		}
+	}
+	/**
+	* Update the search configuration for a collection
+	*/
+	async setSearchConfig(collectionSlug, config) {
+		if (config.tokenize !== void 0) resolveSearchTokenizer(config.tokenize);
+		await this.db.updateTable("_emdash_collections").set({ search_config: JSON.stringify(config) }).where("slug", "=", collectionSlug).execute();
+	}
+	/**
+	* Get searchable fields for a collection
+	*/
+	async getSearchableFields(collectionSlug) {
+		const collection = await this.db.selectFrom("_emdash_collections").select("id").where("slug", "=", collectionSlug).executeTakeFirst();
+		if (!collection) return [];
+		return (await this.db.selectFrom("_emdash_fields").select("slug").where("collection_id", "=", collection.id).where("searchable", "=", 1).execute()).map((f) => f.slug);
+	}
+	/**
+	* Whether a collection has a user-defined `title` field.
+	*
+	* `title` is not a system column on `ec_*` tables -- it exists only when a
+	* collection defines a field with slug `title`. Search and suggestion SQL
+	* that selects `c.title` must check this first; otherwise collections
+	* without a title field raise "no such column: c.title".
+	*/
+	async hasTitleColumn(collectionSlug) {
+		return (await this.getCollectionsWithTitleColumn([collectionSlug])).has(collectionSlug);
+	}
+	/**
+	* Bulk variant of `hasTitleColumn()`: which of the given collections have
+	* a user-defined `title` field. One query instead of one `hasTitleColumn`
+	* round-trip pair per collection -- callers that check this once per
+	* collection in a loop (multi-collection search, suggestions) should use
+	* this instead (AGENTS.md: "one query beats two").
+	*/
+	async getCollectionsWithTitleColumn(collectionSlugs) {
+		if (collectionSlugs.length === 0) return /* @__PURE__ */ new Set();
+		const rows = await this.db.selectFrom("_emdash_fields as f").innerJoin("_emdash_collections as c", "c.id", "f.collection_id").select(["c.slug as collection_slug"]).where("f.slug", "=", "title").execute();
+		const withTitle = new Set(rows.map((r) => r.collection_slug));
+		return new Set(collectionSlugs.filter((slug) => withTitle.has(slug)));
+	}
+	/**
+	* Enable search for a collection.
+	*
+	* Uses rebuildIndex to ensure a clean state -- drop any existing FTS
+	* table/triggers, recreate them, and populate from content. This avoids
+	* duplicate rows when triggers have already populated the index (e.g.
+	* during seeding where content is inserted before search is enabled).
+	*/
+	async enableSearch(collectionSlug, options) {
+		if (options?.tokenize !== void 0) resolveSearchTokenizer(options.tokenize);
+		if (!isSqlite(this.db)) throw new Error("Full-text search is only available with SQLite databases");
+		const searchableFields = await this.getSearchableFields(collectionSlug);
+		if (searchableFields.length === 0) throw new Error(`No searchable fields defined for collection "${collectionSlug}". Mark at least one field as searchable before enabling search.`);
+		const existing = await this.getSearchConfig(collectionSlug);
+		const weights = options?.weights ?? existing?.weights;
+		const tokenize = options?.tokenize ?? existing?.tokenize;
+		await this.rebuildIndex(collectionSlug, searchableFields, weights, tokenize);
+		await this.setSearchConfig(collectionSlug, {
+			enabled: true,
+			weights,
+			tokenize
+		});
+	}
+	/**
+	* Disable search for a collection
+	*
+	* Drops the FTS table and triggers.
+	*/
+	async disableSearch(collectionSlug) {
+		if (!isSqlite(this.db)) return;
+		await this.dropFtsTable(collectionSlug);
+		const existing = await this.getSearchConfig(collectionSlug);
+		await this.setSearchConfig(collectionSlug, {
+			enabled: false,
+			weights: existing?.weights,
+			tokenize: existing?.tokenize
+		});
+	}
+	/**
+	* Get index statistics for a collection
+	*/
+	async getIndexStats(collectionSlug) {
+		if (!isSqlite(this.db)) return null;
+		this.validateInputs(collectionSlug);
+		const ftsDocsizeTable = `${this.getFtsTableName(collectionSlug)}_docsize`;
+		if (!await this.ftsTableExists(collectionSlug)) return null;
+		return { indexed: (await sql`
+			SELECT COUNT(*) as count FROM "${sql.raw(ftsDocsizeTable)}"
+		`.execute(this.db)).rows[0]?.count ?? 0 };
+	}
+	/**
+	* Verify FTS index integrity and rebuild if drift is detected.
+	*
+	* Cheap belt-and-braces check, run lazily on the first search request
+	* per isolate. The expensive cases (corrupted indexes from pre-fix
+	* EmDash versions, broken legacy triggers) are handled at boot time by
+	* migration `039_fix_fts5_triggers`, not here. This routine sticks to:
+	*
+	*   1. FTS table missing while config says search is enabled -> rebuild.
+	*   2. Row count mismatch between content table and FTS docsize -> rebuild.
+	*
+	* Returns true if the index was rebuilt, false if it was healthy.
+	*/
+	async verifyAndRepairIndex(collectionSlug) {
+		if (!isSqlite(this.db)) return false;
+		this.validateInputs(collectionSlug);
+		const ftsDocsizeTable = `${this.getFtsTableName(collectionSlug)}_docsize`;
+		const contentTable = this.getContentTableName(collectionSlug);
+		const fields = await this.getSearchableFields(collectionSlug);
+		const config = await this.getSearchConfig(collectionSlug);
+		if (!await this.ftsTableExists(collectionSlug)) {
+			if (!config?.enabled || fields.length === 0) return false;
+			console.warn(`FTS index for "${collectionSlug}" is missing. Rebuilding.`);
+			await this.rebuildIndex(collectionSlug, fields, config.weights, config.tokenize);
+			return true;
+		}
+		const contentCount = await sql`
+			SELECT COUNT(*) as count FROM ${sql.ref(contentTable)}
+			WHERE deleted_at IS NULL
+		`.execute(this.db);
+		const ftsCount = await sql`
+			SELECT COUNT(*) as count FROM "${sql.raw(ftsDocsizeTable)}"
+		`.execute(this.db);
+		const contentRows = contentCount.rows[0]?.count ?? 0;
+		const ftsRows = ftsCount.rows[0]?.count ?? 0;
+		if (contentRows !== ftsRows) {
+			console.warn(`FTS index for "${collectionSlug}" has ${ftsRows} rows but content table has ${contentRows}. Rebuilding.`);
+			if (fields.length > 0) await this.rebuildIndex(collectionSlug, fields, config?.weights, config?.tokenize);
+			return true;
+		}
+		return false;
+	}
+	/**
+	* Verify and repair FTS indexes for all search-enabled collections.
+	*
+	* Intended to run at startup to auto-heal any corruption from
+	* previous process crashes.
+	*/
+	async verifyAndRepairAll() {
+		if (!isSqlite(this.db)) return 0;
+		const collections = await this.db.selectFrom("_emdash_collections").select("slug").where("search_config", "is not", null).execute();
+		let repaired = 0;
+		for (const { slug } of collections) {
+			if (!(await this.getSearchConfig(slug))?.enabled) continue;
+			try {
+				if (await this.verifyAndRepairIndex(slug)) repaired++;
+			} catch (error) {
+				console.error(`Failed to verify/repair FTS index for "${slug}":`, error);
+			}
+		}
+		return repaired;
+	}
+};
+
+//#endregion
+export { fts_manager_exports as n, FTSManager as t };
+//# sourceMappingURL=fts-manager-CCaZTp8y.mjs.map

package/dist/fts-manager-CCaZTp8y.mjs.map

@@ -0,0 +1 @@
+{"version":3,"file":"fts-manager-CCaZTp8y.mjs","names":["dialectTableExists"],"sources":["../src/search/fts-manager.ts"],"sourcesContent":["/**\n * FTS5 Manager\n *\n * Manages FTS5 virtual tables and triggers for search indexing.\n */\n\nimport type { Kysely } from \"kysely\";\nimport { sql } from \"kysely\";\n\nimport { isSqlite, tableExists as dialectTableExists } from \"../database/dialect-helpers.js\";\nimport type { Database } from \"../database/types.js\";\nimport { validateIdentifier } from \"../database/validate.js\";\nimport { SEARCH_TOKENIZERS } from \"./types.js\";\nimport type { SearchConfig, SearchTokenizer } from \"./types.js\";\n\nconst DEFAULT_SEARCH_TOKENIZER: SearchTokenizer = \"porter unicode61\";\n\nfunction isSearchTokenizer(value: unknown): value is SearchTokenizer {\n\treturn SEARCH_TOKENIZERS.some((tokenizer) => tokenizer === value);\n}\n\nfunction resolveSearchTokenizer(tokenize?: SearchTokenizer): SearchTokenizer {\n\tif (tokenize === undefined) return DEFAULT_SEARCH_TOKENIZER;\n\tif (!isSearchTokenizer(tokenize)) {\n\t\tthrow new Error(`Unsupported FTS5 tokenizer: \"${String(tokenize)}\"`);\n\t}\n\treturn tokenize;\n}\n\n/**\n * FTS5 Manager\n *\n * Handles creation, deletion, and management of FTS5 virtual tables\n * for full-text search on content collections.\n */\nexport class FTSManager {\n\tconstructor(private db: Kysely<Database>) {}\n\n\t/**\n\t * Validate a collection slug and its searchable field names.\n\t * Must be called before any raw SQL interpolation.\n\t */\n\tprivate validateInputs(collectionSlug: string, searchableFields?: string[]): void {\n\t\tvalidateIdentifier(collectionSlug, \"collection slug\");\n\t\tif (searchableFields) {\n\t\t\tfor (const field of searchableFields) {\n\t\t\t\tvalidateIdentifier(field, \"searchable field name\");\n\t\t\t}\n\t\t}\n\t}\n\n\t/**\n\t * Get the FTS table name for a collection\n\t * Uses _emdash_ prefix to clearly mark as internal/system table\n\t */\n\tgetFtsTableName(collectionSlug: string): string {\n\t\tvalidateIdentifier(collectionSlug, \"collection slug\");\n\t\treturn `_emdash_fts_${collectionSlug}`;\n\t}\n\n\t/**\n\t * Get the content table name for a collection\n\t */\n\tgetContentTableName(collectionSlug: string): string {\n\t\tvalidateIdentifier(collectionSlug, \"collection slug\");\n\t\treturn `ec_${collectionSlug}`;\n\t}\n\n\t/**\n\t * Check if an FTS table exists for a collection\n\t */\n\tasync ftsTableExists(collectionSlug: string): Promise<boolean> {\n\t\tconst ftsTable = this.getFtsTableName(collectionSlug);\n\t\treturn dialectTableExists(this.db, ftsTable);\n\t}\n\n\t/**\n\t * Create an FTS5 virtual table for a collection.\n\t * FTS5 is SQLite-only; on other dialects this is a no-op.\n\t *\n\t * @param collectionSlug - The collection slug\n\t * @param searchableFields - Array of field names to index\n\t * @param weights - Optional field weights for ranking\n\t * @param tokenize - Optional FTS5 tokenizer configuration\n\t */\n\tasync createFtsTable(\n\t\tcollectionSlug: string,\n\t\tsearchableFields: string[],\n\t\t_weights?: Record<string, number>,\n\t\ttokenize?: SearchTokenizer,\n\t): Promise<void> {\n\t\tconst tokenizer = resolveSearchTokenizer(tokenize);\n\t\tif (!isSqlite(this.db)) return;\n\t\tthis.validateInputs(collectionSlug, searchableFields);\n\t\tconst ftsTable = this.getFtsTableName(collectionSlug);\n\n\t\t// Build the column list for FTS5\n\t\t// id and locale are UNINDEXED (used for joining/filtering, not searched)\n\t\tconst columns = [\"id UNINDEXED\", \"locale UNINDEXED\", ...searchableFields].join(\", \");\n\n\t\t// Create the FTS5 virtual table. The table stores its own copy of the\n\t\t// indexed values (no `content=` option): Portable Text fields are\n\t\t// indexed as extracted plain text — see searchValueExpr — which cannot\n\t\t// mirror the raw JSON in the ec_* column, and external-content FTS5\n\t\t// requires the index to exactly mirror the backing table's values\n\t\t// (snippet() reads them, and the 'delete' command must be fed the\n\t\t// inserted values or the index corrupts — see migration 039's history).\n\t\t// Storing the extracted text also makes snippet() return prose.\n\t\tawait sql\n\t\t\t.raw(`\n\t\t\tCREATE VIRTUAL TABLE IF NOT EXISTS \"${ftsTable}\" USING fts5(\n\t\t\t\t${columns},\n\t\t\t\ttokenize='${tokenizer}'\n\t\t\t)\n\t\t`)\n\t\t\t.execute(this.db);\n\n\t\t// Create triggers for automatic sync\n\t\tawait this.createTriggers(collectionSlug, searchableFields);\n\t}\n\n\t/**\n\t * SQL expression producing the indexed value for one searchable field.\n\t *\n\t * Portable Text fields are stored as JSON; indexing the raw JSON pollutes\n\t * the index with structural tokens (`_type`, style values like `normal`,\n\t * `_key` ULIDs) and makes snippets show JSON fragments. Extract the prose\n\t * instead: every JSON string under a `text`, `alt`, `caption`, or `code`\n\t * key (span text, image alt/caption, code blocks). This is a superset of\n\t * `extractPlainText` in text-extraction.ts, which walks only known block\n\t * shapes — the SQL variant takes those keys at any depth, so search and\n\t * the extractPlainText consumers (vectorize/ai-search) can see different\n\t * text for the same document. Only JSON documents\n\t * (arrays/objects) are extracted; legacy rows holding a bare string or a\n\t * JSON scalar (`Some title`, `2024`) are indexed as-is. Extraction must\n\t * live in SQL because the sync triggers cannot call into JS.\n\t *\n\t * `ref` must be a validated column reference (`NEW.x`, `OLD.x`, `\"x\"`).\n\t */\n\tprivate searchValueExpr(ref: string, fieldType: string | undefined): string {\n\t\tif (fieldType !== \"portableText\") return ref;\n\t\treturn (\n\t\t\t`CASE WHEN ${ref} IS NULL THEN NULL ` +\n\t\t\t`WHEN json_valid(${ref}) AND json_type(${ref}) IN ('array', 'object') THEN (` +\n\t\t\t`SELECT group_concat(j.value, ' ') FROM json_tree(${ref}) AS j ` +\n\t\t\t`WHERE j.key IN ('text', 'alt', 'caption', 'code') AND j.type = 'text') ` +\n\t\t\t`ELSE ${ref} END`\n\t\t);\n\t}\n\n\t/**\n\t * Field type per slug for a collection, for choosing the indexed-value\n\t * expression. Fields missing from the schema fall back to raw indexing.\n\t */\n\tprivate async getFieldTypes(collectionSlug: string): Promise<Map<string, string>> {\n\t\tconst collection = await this.db\n\t\t\t.selectFrom(\"_emdash_collections\")\n\t\t\t.select(\"id\")\n\t\t\t.where(\"slug\", \"=\", collectionSlug)\n\t\t\t.executeTakeFirst();\n\t\tif (!collection) return new Map();\n\n\t\tconst rows = await this.db\n\t\t\t.selectFrom(\"_emdash_fields\")\n\t\t\t.select([\"slug\", \"type\"])\n\t\t\t.where(\"collection_id\", \"=\", collection.id)\n\t\t\t.execute();\n\t\treturn new Map(rows.map((r) => [r.slug, r.type]));\n\t}\n\n\t/**\n\t * Create triggers to keep FTS table in sync with content table.\n\t *\n\t * The insert and update triggers only add rows to the FTS index when\n\t * `deleted_at IS NULL`. This keeps soft-deleted content out of the\n\t * search index and ensures the FTS row count matches the non-deleted\n\t * content count (which `verifyAndRepairIndex` relies on).\n\t *\n\t * The FTS table stores its own values (no `content=` option), so removal\n\t * is a plain `DELETE FROM fts WHERE rowid = OLD.rowid` — a harmless no-op\n\t * for rows that were never indexed (soft-deleted content). The\n\t * external-content `'delete'`-command choreography and its corruption\n\t * modes (migration 039) do not apply to self-contained tables.\n\t *\n\t * `INSERT OR REPLACE` keeps the insert path idempotent: re-running a\n\t * populate (search repair runs per isolate, so two isolates can race) converges\n\t * on one index row per content row instead of failing on the rowid\n\t * constraint.\n\t *\n\t * The trigger SQL emitted here MUST stay in lock-step with migration\n\t * `064_fts_plain_text.ts`. If this changes again, add a new migration\n\t * rather than editing shipped ones — migrations are forward-only.\n\t */\n\tprivate async createTriggers(collectionSlug: string, searchableFields: string[]): Promise<void> {\n\t\tthis.validateInputs(collectionSlug, searchableFields);\n\t\tif (searchableFields.length === 0) {\n\t\t\tthrow new Error(\n\t\t\t\t`Cannot create FTS triggers for collection \"${collectionSlug}\": no searchable fields. ` +\n\t\t\t\t\t`Mark at least one field as searchable before enabling search.`,\n\t\t\t);\n\t\t}\n\t\tconst ftsTable = this.getFtsTableName(collectionSlug);\n\t\tconst contentTable = this.getContentTableName(collectionSlug);\n\t\tconst fieldTypes = await this.getFieldTypes(collectionSlug);\n\t\tconst fieldList = searchableFields.join(\", \");\n\t\tconst newValueList = searchableFields\n\t\t\t.map((f) => this.searchValueExpr(`NEW.${f}`, fieldTypes.get(f)))\n\t\t\t.join(\", \");\n\n\t\t// Insert trigger - only index non-deleted content\n\t\tawait sql\n\t\t\t.raw(`\n\t\t\tCREATE TRIGGER IF NOT EXISTS \"${ftsTable}_insert\"\n\t\t\tAFTER INSERT ON \"${contentTable}\"\n\t\t\tWHEN NEW.deleted_at IS NULL\n\t\t\tBEGIN\n\t\t\t\tINSERT OR REPLACE INTO \"${ftsTable}\"(rowid, id, locale, ${fieldList})\n\t\t\t\tVALUES (NEW.rowid, NEW.id, NEW.locale, ${newValueList});\n\t\t\tEND\n\t\t`)\n\t\t\t.execute(this.db);\n\n\t\t// Update trigger - drop the old index row, re-insert when the row is\n\t\t// still visible. Trash (deleted_at set) ends at DELETE only; restore\n\t\t// ends at DELETE (no-op) + re-insert.\n\t\t//\n\t\t// The WHEN guard compares raw column values (null-safe IS NOT) so the\n\t\t// trigger fires only when an indexed value, the row's locale, or its\n\t\t// trash state actually changed. Without it every UPDATE re-tokenizes\n\t\t// the whole document — metadata-only saves (status flips, scheduling,\n\t\t// version bumps) and the publish path's rewrite-identical-values\n\t\t// UPDATEs dominate save CPU and WAL volume. deleted_at must stay in\n\t\t// the guard or trash/restore stop syncing the index.\n\t\tconst changedCondition = [\"deleted_at\", \"locale\", ...searchableFields]\n\t\t\t.map((f) => `OLD.${f} IS NOT NEW.${f}`)\n\t\t\t.join(\" OR \");\n\t\tawait sql\n\t\t\t.raw(`\n\t\t\tCREATE TRIGGER IF NOT EXISTS \"${ftsTable}_update\"\n\t\t\tAFTER UPDATE ON \"${contentTable}\"\n\t\t\tWHEN ${changedCondition}\n\t\t\tBEGIN\n\t\t\t\tDELETE FROM \"${ftsTable}\" WHERE rowid = OLD.rowid;\n\t\t\t\tINSERT INTO \"${ftsTable}\"(rowid, id, locale, ${fieldList})\n\t\t\t\tSELECT NEW.rowid, NEW.id, NEW.locale, ${newValueList}\n\t\t\t\tWHERE NEW.deleted_at IS NULL;\n\t\t\tEND\n\t\t`)\n\t\t\t.execute(this.db);\n\n\t\t// Delete trigger\n\t\tawait sql\n\t\t\t.raw(`\n\t\t\tCREATE TRIGGER IF NOT EXISTS \"${ftsTable}_delete\"\n\t\t\tAFTER DELETE ON \"${contentTable}\"\n\t\t\tBEGIN\n\t\t\t\tDELETE FROM \"${ftsTable}\" WHERE rowid = OLD.rowid;\n\t\t\tEND\n\t\t`)\n\t\t\t.execute(this.db);\n\t}\n\n\t/**\n\t * Drop triggers for a collection\n\t */\n\tprivate async dropTriggers(collectionSlug: string): Promise<void> {\n\t\tthis.validateInputs(collectionSlug);\n\t\tconst ftsTable = this.getFtsTableName(collectionSlug);\n\n\t\tawait sql.raw(`DROP TRIGGER IF EXISTS \"${ftsTable}_insert\"`).execute(this.db);\n\t\tawait sql.raw(`DROP TRIGGER IF EXISTS \"${ftsTable}_update\"`).execute(this.db);\n\t\tawait sql.raw(`DROP TRIGGER IF EXISTS \"${ftsTable}_delete\"`).execute(this.db);\n\t}\n\n\t/**\n\t * Drop the FTS table and triggers for a collection\n\t */\n\tasync dropFtsTable(collectionSlug: string): Promise<void> {\n\t\tif (!isSqlite(this.db)) return;\n\t\tthis.validateInputs(collectionSlug);\n\t\tconst ftsTable = this.getFtsTableName(collectionSlug);\n\n\t\t// Drop triggers first\n\t\tawait this.dropTriggers(collectionSlug);\n\n\t\t// Drop the FTS table\n\t\tawait sql.raw(`DROP TABLE IF EXISTS \"${ftsTable}\"`).execute(this.db);\n\t}\n\n\t/**\n\t * Rebuild the FTS index for a collection\n\t *\n\t * This is useful after bulk imports or if the index gets out of sync.\n\t */\n\tasync rebuildIndex(\n\t\tcollectionSlug: string,\n\t\tsearchableFields: string[],\n\t\tweights?: Record<string, number>,\n\t\ttokenize?: SearchTokenizer,\n\t): Promise<void> {\n\t\tresolveSearchTokenizer(tokenize);\n\t\tif (!isSqlite(this.db)) return;\n\t\t// Drop existing table and triggers\n\t\tawait this.dropFtsTable(collectionSlug);\n\n\t\t// Recreate table and triggers\n\t\tawait this.createFtsTable(collectionSlug, searchableFields, weights, tokenize);\n\n\t\t// Populate from existing content\n\t\tawait this.populateFromContent(collectionSlug, searchableFields);\n\t}\n\n\t/**\n\t * Populate the FTS table from existing content.\n\t *\n\t * `INSERT OR REPLACE` so a concurrent double-populate (search repair runs\n\t * per isolate) converges instead of failing on the rowid constraint.\n\t */\n\tasync populateFromContent(collectionSlug: string, searchableFields: string[]): Promise<void> {\n\t\tif (!isSqlite(this.db)) return;\n\t\tthis.validateInputs(collectionSlug, searchableFields);\n\t\tconst ftsTable = this.getFtsTableName(collectionSlug);\n\t\tconst contentTable = this.getContentTableName(collectionSlug);\n\t\tconst fieldTypes = await this.getFieldTypes(collectionSlug);\n\t\tconst fieldList = searchableFields.join(\", \");\n\t\t// Table-qualified references: json_tree exposes columns named\n\t\t// key/value/type/path/..., and inside the extraction subquery a bare\n\t\t// column reference binds to those instead of the ec_* column.\n\t\tconst valueList = searchableFields\n\t\t\t.map((f) => this.searchValueExpr(`\"${contentTable}\".\"${f}\"`, fieldTypes.get(f)))\n\t\t\t.join(\", \");\n\n\t\t// Insert all existing content into FTS table\n\t\tawait sql\n\t\t\t.raw(`\n\t\t\tINSERT OR REPLACE INTO \"${ftsTable}\"(rowid, id, locale, ${fieldList})\n\t\t\tSELECT rowid, id, locale, ${valueList} FROM \"${contentTable}\"\n\t\t\tWHERE deleted_at IS NULL\n\t\t`)\n\t\t\t.execute(this.db);\n\t}\n\n\t/**\n\t * Index up to `limit` content rows with `rowid > afterRowid`, in rowid\n\t * order, so a large collection can be populated across several requests.\n\t * Returns the last rowid covered, or null when no rows remain. Returns\n\t * null on dialects without FTS5.\n\t */\n\tasync populateRange(\n\t\tcollectionSlug: string,\n\t\tsearchableFields: string[],\n\t\tafterRowid: number,\n\t\tlimit: number,\n\t): Promise<number | null> {\n\t\tif (!isSqlite(this.db)) return null;\n\t\tthis.validateInputs(collectionSlug, searchableFields);\n\t\tconst ftsTable = this.getFtsTableName(collectionSlug);\n\t\tconst contentTable = this.getContentTableName(collectionSlug);\n\t\tconst range = await sql<{ last: number | null }>`\n\t\t\tSELECT MAX(rowid) AS last FROM (\n\t\t\t\tSELECT rowid FROM ${sql.ref(contentTable)}\n\t\t\t\tWHERE rowid > ${afterRowid}\n\t\t\t\tORDER BY rowid\n\t\t\t\tLIMIT ${limit}\n\t\t\t)\n\t\t`.execute(this.db);\n\t\tconst last = range.rows[0]?.last ?? null;\n\t\tif (last === null) return null;\n\n\t\tconst fieldTypes = await this.getFieldTypes(collectionSlug);\n\t\tconst fieldList = searchableFields.join(\", \");\n\t\tconst valueList = searchableFields\n\t\t\t.map((f) => this.searchValueExpr(`\"${contentTable}\".\"${f}\"`, fieldTypes.get(f)))\n\t\t\t.join(\", \");\n\t\tawait sql`\n\t\t\tINSERT OR REPLACE INTO ${sql.ref(ftsTable)}(rowid, id, locale, ${sql.raw(fieldList)})\n\t\t\tSELECT rowid, id, locale, ${sql.raw(valueList)} FROM ${sql.ref(contentTable)}\n\t\t\tWHERE deleted_at IS NULL AND rowid > ${afterRowid} AND rowid <= ${last}\n\t\t`.execute(this.db);\n\t\treturn Number(last);\n\t}\n\n\t/**\n\t * Get the search configuration for a collection\n\t */\n\tasync getSearchConfig(collectionSlug: string): Promise<SearchConfig | null> {\n\t\tconst result = await this.db\n\t\t\t.selectFrom(\"_emdash_collections\")\n\t\t\t.select([\"search_config\", \"title_field\"])\n\t\t\t.where(\"slug\", \"=\", collectionSlug)\n\t\t\t.executeTakeFirst();\n\n\t\tif (!result?.search_config) {\n\t\t\treturn null;\n\t\t}\n\n\t\ttry {\n\t\t\tconst parsed: unknown = JSON.parse(result.search_config);\n\t\t\tif (\n\t\t\t\ttypeof parsed !== \"object\" ||\n\t\t\t\tparsed === null ||\n\t\t\t\t!(\"enabled\" in parsed) ||\n\t\t\t\ttypeof parsed.enabled !== \"boolean\"\n\t\t\t) {\n\t\t\t\treturn null;\n\t\t\t}\n\t\t\tconst config: SearchConfig = { enabled: parsed.enabled };\n\t\t\tif (result.title_field) config.titleField = result.title_field;\n\t\t\tif (\"weights\" in parsed && typeof parsed.weights === \"object\" && parsed.weights !== null) {\n\t\t\t\t// weights is a JSON-parsed object — safe to treat as Record<string, number>\n\t\t\t\tconst weights: Record<string, number> = {};\n\t\t\t\tfor (const [k, v] of Object.entries(parsed.weights)) {\n\t\t\t\t\tif (typeof v === \"number\") {\n\t\t\t\t\t\tweights[k] = v;\n\t\t\t\t\t}\n\t\t\t\t}\n\t\t\t\tconfig.weights = weights;\n\t\t\t}\n\t\t\tif (\"tokenize\" in parsed) {\n\t\t\t\tif (!isSearchTokenizer(parsed.tokenize)) {\n\t\t\t\t\treturn null;\n\t\t\t\t}\n\t\t\t\tconfig.tokenize = parsed.tokenize;\n\t\t\t}\n\t\t\treturn config;\n\t\t} catch {\n\t\t\treturn null;\n\t\t}\n\t}\n\n\t/**\n\t * Update the search configuration for a collection\n\t */\n\tasync setSearchConfig(collectionSlug: string, config: SearchConfig): Promise<void> {\n\t\tif (config.tokenize !== undefined) {\n\t\t\tresolveSearchTokenizer(config.tokenize);\n\t\t}\n\t\tawait this.db\n\t\t\t.updateTable(\"_emdash_collections\")\n\t\t\t.set({ search_config: JSON.stringify(config) })\n\t\t\t.where(\"slug\", \"=\", collectionSlug)\n\t\t\t.execute();\n\t}\n\n\t/**\n\t * Get searchable fields for a collection\n\t */\n\tasync getSearchableFields(collectionSlug: string): Promise<string[]> {\n\t\tconst collection = await this.db\n\t\t\t.selectFrom(\"_emdash_collections\")\n\t\t\t.select(\"id\")\n\t\t\t.where(\"slug\", \"=\", collectionSlug)\n\t\t\t.executeTakeFirst();\n\n\t\tif (!collection) {\n\t\t\treturn [];\n\t\t}\n\n\t\tconst fields = await this.db\n\t\t\t.selectFrom(\"_emdash_fields\")\n\t\t\t.select(\"slug\")\n\t\t\t.where(\"collection_id\", \"=\", collection.id)\n\t\t\t.where(\"searchable\", \"=\", 1)\n\t\t\t.execute();\n\n\t\treturn fields.map((f) => f.slug);\n\t}\n\n\t/**\n\t * Whether a collection has a user-defined `title` field.\n\t *\n\t * `title` is not a system column on `ec_*` tables -- it exists only when a\n\t * collection defines a field with slug `title`. Search and suggestion SQL\n\t * that selects `c.title` must check this first; otherwise collections\n\t * without a title field raise \"no such column: c.title\".\n\t */\n\tasync hasTitleColumn(collectionSlug: string): Promise<boolean> {\n\t\tconst withTitle = await this.getCollectionsWithTitleColumn([collectionSlug]);\n\t\treturn withTitle.has(collectionSlug);\n\t}\n\n\t/**\n\t * Bulk variant of `hasTitleColumn()`: which of the given collections have\n\t * a user-defined `title` field. One query instead of one `hasTitleColumn`\n\t * round-trip pair per collection -- callers that check this once per\n\t * collection in a loop (multi-collection search, suggestions) should use\n\t * this instead (AGENTS.md: \"one query beats two\").\n\t */\n\tasync getCollectionsWithTitleColumn(collectionSlugs: string[]): Promise<Set<string>> {\n\t\tif (collectionSlugs.length === 0) return new Set();\n\n\t\tconst rows = await this.db\n\t\t\t.selectFrom(\"_emdash_fields as f\")\n\t\t\t.innerJoin(\"_emdash_collections as c\", \"c.id\", \"f.collection_id\")\n\t\t\t.select([\"c.slug as collection_slug\"])\n\t\t\t.where(\"f.slug\", \"=\", \"title\")\n\t\t\t.execute();\n\n\t\tconst withTitle = new Set(rows.map((r) => r.collection_slug));\n\t\treturn new Set(collectionSlugs.filter((slug) => withTitle.has(slug)));\n\t}\n\n\t/**\n\t * Enable search for a collection.\n\t *\n\t * Uses rebuildIndex to ensure a clean state -- drop any existing FTS\n\t * table/triggers, recreate them, and populate from content. This avoids\n\t * duplicate rows when triggers have already populated the index (e.g.\n\t * during seeding where content is inserted before search is enabled).\n\t */\n\tasync enableSearch(\n\t\tcollectionSlug: string,\n\t\toptions?: { weights?: Record<string, number>; tokenize?: SearchTokenizer },\n\t): Promise<void> {\n\t\tif (options?.tokenize !== undefined) {\n\t\t\tresolveSearchTokenizer(options.tokenize);\n\t\t}\n\t\tif (!isSqlite(this.db)) {\n\t\t\tthrow new Error(\"Full-text search is only available with SQLite databases\");\n\t\t}\n\t\t// Get searchable fields\n\t\tconst searchableFields = await this.getSearchableFields(collectionSlug);\n\n\t\tif (searchableFields.length === 0) {\n\t\t\tthrow new Error(\n\t\t\t\t`No searchable fields defined for collection \"${collectionSlug}\". ` +\n\t\t\t\t\t`Mark at least one field as searchable before enabling search.`,\n\t\t\t);\n\t\t}\n\n\t\tconst existing = await this.getSearchConfig(collectionSlug);\n\t\tconst weights = options?.weights ?? existing?.weights;\n\t\tconst tokenize = options?.tokenize ?? existing?.tokenize;\n\n\t\t// Rebuild from scratch to ensure clean state (no duplicate rows)\n\t\tawait this.rebuildIndex(collectionSlug, searchableFields, weights, tokenize);\n\n\t\t// Update search config\n\t\tawait this.setSearchConfig(collectionSlug, {\n\t\t\tenabled: true,\n\t\t\tweights,\n\t\t\ttokenize,\n\t\t});\n\t}\n\n\t/**\n\t * Disable search for a collection\n\t *\n\t * Drops the FTS table and triggers.\n\t */\n\tasync disableSearch(collectionSlug: string): Promise<void> {\n\t\tif (!isSqlite(this.db)) return;\n\t\tawait this.dropFtsTable(collectionSlug);\n\t\tconst existing = await this.getSearchConfig(collectionSlug);\n\t\tawait this.setSearchConfig(collectionSlug, {\n\t\t\tenabled: false,\n\t\t\tweights: existing?.weights,\n\t\t\ttokenize: existing?.tokenize,\n\t\t});\n\t}\n\n\t/**\n\t * Get index statistics for a collection\n\t */\n\tasync getIndexStats(\n\t\tcollectionSlug: string,\n\t): Promise<{ indexed: number; lastRebuilt?: string } | null> {\n\t\tif (!isSqlite(this.db)) return null;\n\t\tthis.validateInputs(collectionSlug);\n\t\tconst ftsTable = this.getFtsTableName(collectionSlug);\n\t\tconst ftsDocsizeTable = `${ftsTable}_docsize`;\n\n\t\t// Check if table exists\n\t\tif (!(await this.ftsTableExists(collectionSlug))) {\n\t\t\treturn null;\n\t\t}\n\n\t\t// Count indexed rows\n\t\tconst result = await sql<{ count: number }>`\n\t\t\tSELECT COUNT(*) as count FROM \"${sql.raw(ftsDocsizeTable)}\"\n\t\t`.execute(this.db);\n\n\t\treturn {\n\t\t\tindexed: result.rows[0]?.count ?? 0,\n\t\t};\n\t}\n\n\t/**\n\t * Verify FTS index integrity and rebuild if drift is detected.\n\t *\n\t * Cheap belt-and-braces check, run lazily on the first search request\n\t * per isolate. The expensive cases (corrupted indexes from pre-fix\n\t * EmDash versions, broken legacy triggers) are handled at boot time by\n\t * migration `039_fix_fts5_triggers`, not here. This routine sticks to:\n\t *\n\t *   1. FTS table missing while config says search is enabled -> rebuild.\n\t *   2. Row count mismatch between content table and FTS docsize -> rebuild.\n\t *\n\t * Returns true if the index was rebuilt, false if it was healthy.\n\t */\n\tasync verifyAndRepairIndex(collectionSlug: string): Promise<boolean> {\n\t\tif (!isSqlite(this.db)) return false;\n\t\tthis.validateInputs(collectionSlug);\n\t\tconst ftsTable = this.getFtsTableName(collectionSlug);\n\t\tconst ftsDocsizeTable = `${ftsTable}_docsize`;\n\t\tconst contentTable = this.getContentTableName(collectionSlug);\n\t\tconst fields = await this.getSearchableFields(collectionSlug);\n\t\tconst config = await this.getSearchConfig(collectionSlug);\n\n\t\tif (!(await this.ftsTableExists(collectionSlug))) {\n\t\t\tif (!config?.enabled || fields.length === 0) {\n\t\t\t\treturn false;\n\t\t\t}\n\n\t\t\tconsole.warn(`FTS index for \"${collectionSlug}\" is missing. Rebuilding.`);\n\t\t\tawait this.rebuildIndex(collectionSlug, fields, config.weights, config.tokenize);\n\t\t\treturn true;\n\t\t}\n\n\t\t// Row count parity check against the docsize shadow table, which\n\t\t// tracks rows actually present in the full-text index.\n\t\tconst contentCount = await sql<{ count: number }>`\n\t\t\tSELECT COUNT(*) as count FROM ${sql.ref(contentTable)}\n\t\t\tWHERE deleted_at IS NULL\n\t\t`.execute(this.db);\n\n\t\tconst ftsCount = await sql<{ count: number }>`\n\t\t\tSELECT COUNT(*) as count FROM \"${sql.raw(ftsDocsizeTable)}\"\n\t\t`.execute(this.db);\n\n\t\tconst contentRows = contentCount.rows[0]?.count ?? 0;\n\t\tconst ftsRows = ftsCount.rows[0]?.count ?? 0;\n\n\t\tif (contentRows !== ftsRows) {\n\t\t\tconsole.warn(\n\t\t\t\t`FTS index for \"${collectionSlug}\" has ${ftsRows} rows but content table has ${contentRows}. Rebuilding.`,\n\t\t\t);\n\t\t\tif (fields.length > 0) {\n\t\t\t\tawait this.rebuildIndex(collectionSlug, fields, config?.weights, config?.tokenize);\n\t\t\t}\n\t\t\treturn true;\n\t\t}\n\n\t\treturn false;\n\t}\n\n\t/**\n\t * Verify and repair FTS indexes for all search-enabled collections.\n\t *\n\t * Intended to run at startup to auto-heal any corruption from\n\t * previous process crashes.\n\t */\n\tasync verifyAndRepairAll(): Promise<number> {\n\t\tif (!isSqlite(this.db)) return 0;\n\n\t\tconst collections = await this.db\n\t\t\t.selectFrom(\"_emdash_collections\")\n\t\t\t.select(\"slug\")\n\t\t\t.where(\"search_config\", \"is not\", null)\n\t\t\t.execute();\n\n\t\tlet repaired = 0;\n\t\tfor (const { slug } of collections) {\n\t\t\tconst config = await this.getSearchConfig(slug);\n\t\t\tif (!config?.enabled) continue;\n\n\t\t\ttry {\n\t\t\t\tconst wasRepaired = await this.verifyAndRepairIndex(slug);\n\t\t\t\tif (wasRepaired) repaired++;\n\t\t\t} catch (error) {\n\t\t\t\tconsole.error(`Failed to verify/repair FTS index for \"${slug}\":`, error);\n\t\t\t}\n\t\t}\n\n\t\treturn repaired;\n\t}\n}\n"],"mappings":";;;;;;;;AAeA,MAAM,2BAA4C;AAElD,SAAS,kBAAkB,OAA0C;AACpE,QAAO,kBAAkB,MAAM,cAAc,cAAc,MAAM;;AAGlE,SAAS,uBAAuB,UAA6C;AAC5E,KAAI,aAAa,OAAW,QAAO;AACnC,KAAI,CAAC,kBAAkB,SAAS,CAC/B,OAAM,IAAI,MAAM,gCAAgC,OAAO,SAAS,CAAC,GAAG;AAErE,QAAO;;;;;;;;AASR,IAAa,aAAb,MAAwB;CACvB,YAAY,AAAQ,IAAsB;EAAtB;;;;;;CAMpB,AAAQ,eAAe,gBAAwB,kBAAmC;AACjF,qBAAmB,gBAAgB,kBAAkB;AACrD,MAAI,iBACH,MAAK,MAAM,SAAS,iBACnB,oBAAmB,OAAO,wBAAwB;;;;;;CASrD,gBAAgB,gBAAgC;AAC/C,qBAAmB,gBAAgB,kBAAkB;AACrD,SAAO,eAAe;;;;;CAMvB,oBAAoB,gBAAgC;AACnD,qBAAmB,gBAAgB,kBAAkB;AACrD,SAAO,MAAM;;;;;CAMd,MAAM,eAAe,gBAA0C;EAC9D,MAAM,WAAW,KAAK,gBAAgB,eAAe;AACrD,SAAOA,YAAmB,KAAK,IAAI,SAAS;;;;;;;;;;;CAY7C,MAAM,eACL,gBACA,kBACA,UACA,UACgB;EAChB,MAAM,YAAY,uBAAuB,SAAS;AAClD,MAAI,CAAC,SAAS,KAAK,GAAG,CAAE;AACxB,OAAK,eAAe,gBAAgB,iBAAiB;EACrD,MAAM,WAAW,KAAK,gBAAgB,eAAe;EAIrD,MAAM,UAAU;GAAC;GAAgB;GAAoB,GAAG;GAAiB,CAAC,KAAK,KAAK;AAUpF,QAAM,IACJ,IAAI;yCACiC,SAAS;MAC5C,QAAQ;gBACE,UAAU;;IAEtB,CACA,QAAQ,KAAK,GAAG;AAGlB,QAAM,KAAK,eAAe,gBAAgB,iBAAiB;;;;;;;;;;;;;;;;;;;;CAqB5D,AAAQ,gBAAgB,KAAa,WAAuC;AAC3E,MAAI,cAAc,eAAgB,QAAO;AACzC,SACC,aAAa,IAAI,qCACE,IAAI,kBAAkB,IAAI,kFACO,IAAI,qFAEhD,IAAI;;;;;;CAQd,MAAc,cAAc,gBAAsD;EACjF,MAAM,aAAa,MAAM,KAAK,GAC5B,WAAW,sBAAsB,CACjC,OAAO,KAAK,CACZ,MAAM,QAAQ,KAAK,eAAe,CAClC,kBAAkB;AACpB,MAAI,CAAC,WAAY,wBAAO,IAAI,KAAK;EAEjC,MAAM,OAAO,MAAM,KAAK,GACtB,WAAW,iBAAiB,CAC5B,OAAO,CAAC,QAAQ,OAAO,CAAC,CACxB,MAAM,iBAAiB,KAAK,WAAW,GAAG,CAC1C,SAAS;AACX,SAAO,IAAI,IAAI,KAAK,KAAK,MAAM,CAAC,EAAE,MAAM,EAAE,KAAK,CAAC,CAAC;;;;;;;;;;;;;;;;;;;;;;;;;CA0BlD,MAAc,eAAe,gBAAwB,kBAA2C;AAC/F,OAAK,eAAe,gBAAgB,iBAAiB;AACrD,MAAI,iBAAiB,WAAW,EAC/B,OAAM,IAAI,MACT,8CAA8C,eAAe,wFAE7D;EAEF,MAAM,WAAW,KAAK,gBAAgB,eAAe;EACrD,MAAM,eAAe,KAAK,oBAAoB,eAAe;EAC7D,MAAM,aAAa,MAAM,KAAK,cAAc,eAAe;EAC3D,MAAM,YAAY,iBAAiB,KAAK,KAAK;EAC7C,MAAM,eAAe,iBACnB,KAAK,MAAM,KAAK,gBAAgB,OAAO,KAAK,WAAW,IAAI,EAAE,CAAC,CAAC,CAC/D,KAAK,KAAK;AAGZ,QAAM,IACJ,IAAI;mCAC2B,SAAS;sBACtB,aAAa;;;8BAGL,SAAS,uBAAuB,UAAU;6CAC3B,aAAa;;IAEtD,CACA,QAAQ,KAAK,GAAG;EAalB,MAAM,mBAAmB;GAAC;GAAc;GAAU,GAAG;GAAiB,CACpE,KAAK,MAAM,OAAO,EAAE,cAAc,IAAI,CACtC,KAAK,OAAO;AACd,QAAM,IACJ,IAAI;mCAC2B,SAAS;sBACtB,aAAa;UACzB,iBAAiB;;mBAER,SAAS;mBACT,SAAS,uBAAuB,UAAU;4CACjB,aAAa;;;IAGrD,CACA,QAAQ,KAAK,GAAG;AAGlB,QAAM,IACJ,IAAI;mCAC2B,SAAS;sBACtB,aAAa;;mBAEhB,SAAS;;IAExB,CACA,QAAQ,KAAK,GAAG;;;;;CAMnB,MAAc,aAAa,gBAAuC;AACjE,OAAK,eAAe,eAAe;EACnC,MAAM,WAAW,KAAK,gBAAgB,eAAe;AAErD,QAAM,IAAI,IAAI,2BAA2B,SAAS,UAAU,CAAC,QAAQ,KAAK,GAAG;AAC7E,QAAM,IAAI,IAAI,2BAA2B,SAAS,UAAU,CAAC,QAAQ,KAAK,GAAG;AAC7E,QAAM,IAAI,IAAI,2BAA2B,SAAS,UAAU,CAAC,QAAQ,KAAK,GAAG;;;;;CAM9E,MAAM,aAAa,gBAAuC;AACzD,MAAI,CAAC,SAAS,KAAK,GAAG,CAAE;AACxB,OAAK,eAAe,eAAe;EACnC,MAAM,WAAW,KAAK,gBAAgB,eAAe;AAGrD,QAAM,KAAK,aAAa,eAAe;AAGvC,QAAM,IAAI,IAAI,yBAAyB,SAAS,GAAG,CAAC,QAAQ,KAAK,GAAG;;;;;;;CAQrE,MAAM,aACL,gBACA,kBACA,SACA,UACgB;AAChB,yBAAuB,SAAS;AAChC,MAAI,CAAC,SAAS,KAAK,GAAG,CAAE;AAExB,QAAM,KAAK,aAAa,eAAe;AAGvC,QAAM,KAAK,eAAe,gBAAgB,kBAAkB,SAAS,SAAS;AAG9E,QAAM,KAAK,oBAAoB,gBAAgB,iBAAiB;;;;;;;;CASjE,MAAM,oBAAoB,gBAAwB,kBAA2C;AAC5F,MAAI,CAAC,SAAS,KAAK,GAAG,CAAE;AACxB,OAAK,eAAe,gBAAgB,iBAAiB;EACrD,MAAM,WAAW,KAAK,gBAAgB,eAAe;EACrD,MAAM,eAAe,KAAK,oBAAoB,eAAe;EAC7D,MAAM,aAAa,MAAM,KAAK,cAAc,eAAe;EAC3D,MAAM,YAAY,iBAAiB,KAAK,KAAK;EAI7C,MAAM,YAAY,iBAChB,KAAK,MAAM,KAAK,gBAAgB,IAAI,aAAa,KAAK,EAAE,IAAI,WAAW,IAAI,EAAE,CAAC,CAAC,CAC/E,KAAK,KAAK;AAGZ,QAAM,IACJ,IAAI;6BACqB,SAAS,uBAAuB,UAAU;+BACxC,UAAU,SAAS,aAAa;;IAE3D,CACA,QAAQ,KAAK,GAAG;;;;;;;;CASnB,MAAM,cACL,gBACA,kBACA,YACA,OACyB;AACzB,MAAI,CAAC,SAAS,KAAK,GAAG,CAAE,QAAO;AAC/B,OAAK,eAAe,gBAAgB,iBAAiB;EACrD,MAAM,WAAW,KAAK,gBAAgB,eAAe;EACrD,MAAM,eAAe,KAAK,oBAAoB,eAAe;EAS7D,MAAM,QARQ,MAAM,GAA4B;;wBAE1B,IAAI,IAAI,aAAa,CAAC;oBAC1B,WAAW;;YAEnB,MAAM;;IAEd,QAAQ,KAAK,GAAG,EACC,KAAK,IAAI,QAAQ;AACpC,MAAI,SAAS,KAAM,QAAO;EAE1B,MAAM,aAAa,MAAM,KAAK,cAAc,eAAe;EAC3D,MAAM,YAAY,iBAAiB,KAAK,KAAK;EAC7C,MAAM,YAAY,iBAChB,KAAK,MAAM,KAAK,gBAAgB,IAAI,aAAa,KAAK,EAAE,IAAI,WAAW,IAAI,EAAE,CAAC,CAAC,CAC/E,KAAK,KAAK;AACZ,QAAM,GAAG;4BACiB,IAAI,IAAI,SAAS,CAAC,sBAAsB,IAAI,IAAI,UAAU,CAAC;+BACxD,IAAI,IAAI,UAAU,CAAC,QAAQ,IAAI,IAAI,aAAa,CAAC;0CACtC,WAAW,gBAAgB,KAAK;IACtE,QAAQ,KAAK,GAAG;AAClB,SAAO,OAAO,KAAK;;;;;CAMpB,MAAM,gBAAgB,gBAAsD;EAC3E,MAAM,SAAS,MAAM,KAAK,GACxB,WAAW,sBAAsB,CACjC,OAAO,CAAC,iBAAiB,cAAc,CAAC,CACxC,MAAM,QAAQ,KAAK,eAAe,CAClC,kBAAkB;AAEpB,MAAI,CAAC,QAAQ,cACZ,QAAO;AAGR,MAAI;GACH,MAAM,SAAkB,KAAK,MAAM,OAAO,cAAc;AACxD,OACC,OAAO,WAAW,YAClB,WAAW,QACX,EAAE,aAAa,WACf,OAAO,OAAO,YAAY,UAE1B,QAAO;GAER,MAAM,SAAuB,EAAE,SAAS,OAAO,SAAS;AACxD,OAAI,OAAO,YAAa,QAAO,aAAa,OAAO;AACnD,OAAI,aAAa,UAAU,OAAO,OAAO,YAAY,YAAY,OAAO,YAAY,MAAM;IAEzF,MAAM,UAAkC,EAAE;AAC1C,SAAK,MAAM,CAAC,GAAG,MAAM,OAAO,QAAQ,OAAO,QAAQ,CAClD,KAAI,OAAO,MAAM,SAChB,SAAQ,KAAK;AAGf,WAAO,UAAU;;AAElB,OAAI,cAAc,QAAQ;AACzB,QAAI,CAAC,kBAAkB,OAAO,SAAS,CACtC,QAAO;AAER,WAAO,WAAW,OAAO;;AAE1B,UAAO;UACA;AACP,UAAO;;;;;;CAOT,MAAM,gBAAgB,gBAAwB,QAAqC;AAClF,MAAI,OAAO,aAAa,OACvB,wBAAuB,OAAO,SAAS;AAExC,QAAM,KAAK,GACT,YAAY,sBAAsB,CAClC,IAAI,EAAE,eAAe,KAAK,UAAU,OAAO,EAAE,CAAC,CAC9C,MAAM,QAAQ,KAAK,eAAe,CAClC,SAAS;;;;;CAMZ,MAAM,oBAAoB,gBAA2C;EACpE,MAAM,aAAa,MAAM,KAAK,GAC5B,WAAW,sBAAsB,CACjC,OAAO,KAAK,CACZ,MAAM,QAAQ,KAAK,eAAe,CAClC,kBAAkB;AAEpB,MAAI,CAAC,WACJ,QAAO,EAAE;AAUV,UAPe,MAAM,KAAK,GACxB,WAAW,iBAAiB,CAC5B,OAAO,OAAO,CACd,MAAM,iBAAiB,KAAK,WAAW,GAAG,CAC1C,MAAM,cAAc,KAAK,EAAE,CAC3B,SAAS,EAEG,KAAK,MAAM,EAAE,KAAK;;;;;;;;;;CAWjC,MAAM,eAAe,gBAA0C;AAE9D,UADkB,MAAM,KAAK,8BAA8B,CAAC,eAAe,CAAC,EAC3D,IAAI,eAAe;;;;;;;;;CAUrC,MAAM,8BAA8B,iBAAiD;AACpF,MAAI,gBAAgB,WAAW,EAAG,wBAAO,IAAI,KAAK;EAElD,MAAM,OAAO,MAAM,KAAK,GACtB,WAAW,sBAAsB,CACjC,UAAU,4BAA4B,QAAQ,kBAAkB,CAChE,OAAO,CAAC,4BAA4B,CAAC,CACrC,MAAM,UAAU,KAAK,QAAQ,CAC7B,SAAS;EAEX,MAAM,YAAY,IAAI,IAAI,KAAK,KAAK,MAAM,EAAE,gBAAgB,CAAC;AAC7D,SAAO,IAAI,IAAI,gBAAgB,QAAQ,SAAS,UAAU,IAAI,KAAK,CAAC,CAAC;;;;;;;;;;CAWtE,MAAM,aACL,gBACA,SACgB;AAChB,MAAI,SAAS,aAAa,OACzB,wBAAuB,QAAQ,SAAS;AAEzC,MAAI,CAAC,SAAS,KAAK,GAAG,CACrB,OAAM,IAAI,MAAM,2DAA2D;EAG5E,MAAM,mBAAmB,MAAM,KAAK,oBAAoB,eAAe;AAEvE,MAAI,iBAAiB,WAAW,EAC/B,OAAM,IAAI,MACT,gDAAgD,eAAe,kEAE/D;EAGF,MAAM,WAAW,MAAM,KAAK,gBAAgB,eAAe;EAC3D,MAAM,UAAU,SAAS,WAAW,UAAU;EAC9C,MAAM,WAAW,SAAS,YAAY,UAAU;AAGhD,QAAM,KAAK,aAAa,gBAAgB,kBAAkB,SAAS,SAAS;AAG5E,QAAM,KAAK,gBAAgB,gBAAgB;GAC1C,SAAS;GACT;GACA;GACA,CAAC;;;;;;;CAQH,MAAM,cAAc,gBAAuC;AAC1D,MAAI,CAAC,SAAS,KAAK,GAAG,CAAE;AACxB,QAAM,KAAK,aAAa,eAAe;EACvC,MAAM,WAAW,MAAM,KAAK,gBAAgB,eAAe;AAC3D,QAAM,KAAK,gBAAgB,gBAAgB;GAC1C,SAAS;GACT,SAAS,UAAU;GACnB,UAAU,UAAU;GACpB,CAAC;;;;;CAMH,MAAM,cACL,gBAC4D;AAC5D,MAAI,CAAC,SAAS,KAAK,GAAG,CAAE,QAAO;AAC/B,OAAK,eAAe,eAAe;EAEnC,MAAM,kBAAkB,GADP,KAAK,gBAAgB,eAAe,CACjB;AAGpC,MAAI,CAAE,MAAM,KAAK,eAAe,eAAe,CAC9C,QAAO;AAQR,SAAO,EACN,UALc,MAAM,GAAsB;oCACT,IAAI,IAAI,gBAAgB,CAAC;IACzD,QAAQ,KAAK,GAAG,EAGD,KAAK,IAAI,SAAS,GAClC;;;;;;;;;;;;;;;CAgBF,MAAM,qBAAqB,gBAA0C;AACpE,MAAI,CAAC,SAAS,KAAK,GAAG,CAAE,QAAO;AAC/B,OAAK,eAAe,eAAe;EAEnC,MAAM,kBAAkB,GADP,KAAK,gBAAgB,eAAe,CACjB;EACpC,MAAM,eAAe,KAAK,oBAAoB,eAAe;EAC7D,MAAM,SAAS,MAAM,KAAK,oBAAoB,eAAe;EAC7D,MAAM,SAAS,MAAM,KAAK,gBAAgB,eAAe;AAEzD,MAAI,CAAE,MAAM,KAAK,eAAe,eAAe,EAAG;AACjD,OAAI,CAAC,QAAQ,WAAW,OAAO,WAAW,EACzC,QAAO;AAGR,WAAQ,KAAK,kBAAkB,eAAe,2BAA2B;AACzE,SAAM,KAAK,aAAa,gBAAgB,QAAQ,OAAO,SAAS,OAAO,SAAS;AAChF,UAAO;;EAKR,MAAM,eAAe,MAAM,GAAsB;mCAChB,IAAI,IAAI,aAAa,CAAC;;IAErD,QAAQ,KAAK,GAAG;EAElB,MAAM,WAAW,MAAM,GAAsB;oCACX,IAAI,IAAI,gBAAgB,CAAC;IACzD,QAAQ,KAAK,GAAG;EAElB,MAAM,cAAc,aAAa,KAAK,IAAI,SAAS;EACnD,MAAM,UAAU,SAAS,KAAK,IAAI,SAAS;AAE3C,MAAI,gBAAgB,SAAS;AAC5B,WAAQ,KACP,kBAAkB,eAAe,QAAQ,QAAQ,8BAA8B,YAAY,eAC3F;AACD,OAAI,OAAO,SAAS,EACnB,OAAM,KAAK,aAAa,gBAAgB,QAAQ,QAAQ,SAAS,QAAQ,SAAS;AAEnF,UAAO;;AAGR,SAAO;;;;;;;;CASR,MAAM,qBAAsC;AAC3C,MAAI,CAAC,SAAS,KAAK,GAAG,CAAE,QAAO;EAE/B,MAAM,cAAc,MAAM,KAAK,GAC7B,WAAW,sBAAsB,CACjC,OAAO,OAAO,CACd,MAAM,iBAAiB,UAAU,KAAK,CACtC,SAAS;EAEX,IAAI,WAAW;AACf,OAAK,MAAM,EAAE,UAAU,aAAa;AAEnC,OAAI,EADW,MAAM,KAAK,gBAAgB,KAAK,GAClC,QAAS;AAEtB,OAAI;AAEH,QADoB,MAAM,KAAK,qBAAqB,KAAK,CACxC;YACT,OAAO;AACf,YAAQ,MAAM,0CAA0C,KAAK,KAAK,MAAM;;;AAI1E,SAAO"}

package/dist/gc-xowRjczI.mjs

@@ -0,0 +1,75 @@
+import "./dialect-helpers-7_IqPdh9.mjs";
+import "./base64-DvnNgP7e.mjs";
+import "./types-aOOxRSx6.mjs";
+import "./errors-Bt3goF8I.mjs";
+import "./canonical-CcSuuTlH.mjs";
+import "./digest-DVZ--cE9.mjs";
+import "./settings-DSfE7wOa.mjs";
+import "./errors-BN1QAnEY.mjs";
+import "./error-BfsNNrIE.mjs";
+import "./fence-Bb_IbCxI.mjs";
+import "./kinds-CYmRESWj.mjs";
+import "./paths-d51RFm8M.mjs";
+import "./version-DgYjhTud.mjs";
+import "./receipt-B-aU4hpJ.mjs";
+import { n as TransferApprovalRepository } from "./approvals-CAX9Yw33.mjs";
+import { i as TransferOperationRepository, t as TransferStage } from "./stage-CUjhED3P.mjs";
+import { i as stagingPrefix } from "./keys-Bck6LBfF.mjs";
+
+//#region src/transfer/gc.ts
+/** How long a finished import's staging is kept after its last state change. */
+const IMPORT_STAGING_RETENTION_SECONDS = 1440 * 60;
+const OBJECT_BATCH = 100;
+const ROW_BATCH = 500;
+async function collectTransferStaging(db, storage, options = {}) {
+	const operationLimit = options.operations ?? 5;
+	let objectBudget = options.objects ?? 1e3;
+	let rowBatchBudget = options.rowBatches ?? 20;
+	const result = {
+		expired: 0,
+		collected: 0,
+		objectsDeleted: 0,
+		rowsDeleted: 0
+	};
+	const operations = new TransferOperationRepository(db);
+	result.expired = (await operations.expireDue(operationLimit)).length;
+	await new TransferApprovalRepository(db).expireDue();
+	const pending = await operations.listUncollected(operationLimit, { importRetentionSeconds: IMPORT_STAGING_RETENTION_SECONDS });
+	for (const operation of pending) {
+		if (objectBudget <= 0 || rowBatchBudget <= 0) break;
+		try {
+			const stage = new TransferStage(storage, stagingPrefix(operation.kind, operation.id, operation.stagingSecret));
+			let objectsDone = false;
+			while (objectBudget > 0) {
+				const deleted = await stage.deleteSome(Math.min(OBJECT_BATCH, objectBudget));
+				objectBudget -= deleted;
+				result.objectsDeleted += deleted;
+				if (deleted === 0) {
+					objectsDone = true;
+					break;
+				}
+			}
+			if (!objectsDone) break;
+			let rowsDone = false;
+			while (rowBatchBudget > 0) {
+				rowBatchBudget--;
+				const deleted = await operations.deleteChildRows(operation.id, ROW_BATCH);
+				result.rowsDeleted += deleted;
+				if (deleted === 0) {
+					rowsDone = true;
+					break;
+				}
+			}
+			if (!rowsDone) break;
+			await operations.markCollected(operation.id);
+			result.collected++;
+		} catch (error) {
+			console.error(`[transfer] Failed to collect staging for operation ${operation.id}:`, error);
+		}
+	}
+	return result;
+}
+
+//#endregion
+export { collectTransferStaging };
+//# sourceMappingURL=gc-xowRjczI.mjs.map

package/dist/gc-xowRjczI.mjs.map

@@ -0,0 +1 @@
+{"version":3,"file":"gc-xowRjczI.mjs","names":[],"sources":["../src/transfer/gc.ts"],"sourcesContent":["/**\n * Bounded collection of transfer staging areas, run from system cleanup.\n *\n * Each tick expires overdue operations and approval grants, then collects\n * the staging area of a few finished operations: staged objects (package\n * files, the plan, and analysis state) first, then child rows, then the\n * operation is marked collected. Work stops when the tick's budgets run out\n * and resumes on the next tick. Operation rows, and the receipts they carry,\n * are kept.\n */\n\nimport type { Kysely } from \"kysely\";\n\nimport type { Database } from \"../database/types.js\";\nimport type { Storage } from \"../storage/types.js\";\nimport { TransferApprovalRepository } from \"./ops/approvals.js\";\nimport { TransferOperationRepository } from \"./ops/operations.js\";\nimport { stagingPrefix } from \"./staging/keys.js\";\nimport { TransferStage } from \"./staging/stage.js\";\n\n/** How long a finished import's staging is kept after its last state change. */\nexport const IMPORT_STAGING_RETENTION_SECONDS = 24 * 60 * 60;\n\nexport interface TransferStagingCollectionOptions {\n\t/** Operations to expire, and to collect, per tick. */\n\toperations?: number;\n\t/** Staged objects deleted per tick, across all operations. */\n\tobjects?: number;\n\t/** `deleteChildRows` calls per tick, across all operations. */\n\trowBatches?: number;\n}\n\nexport interface TransferStagingCollectionResult {\n\texpired: number;\n\tcollected: number;\n\tobjectsDeleted: number;\n\trowsDeleted: number;\n}\n\nconst OBJECT_BATCH = 100;\nconst ROW_BATCH = 500;\n\nexport async function collectTransferStaging(\n\tdb: Kysely<Database>,\n\tstorage: Storage,\n\toptions: TransferStagingCollectionOptions = {},\n): Promise<TransferStagingCollectionResult> {\n\tconst operationLimit = options.operations ?? 5;\n\tlet objectBudget = options.objects ?? 1000;\n\tlet rowBatchBudget = options.rowBatches ?? 20;\n\tconst result: TransferStagingCollectionResult = {\n\t\texpired: 0,\n\t\tcollected: 0,\n\t\tobjectsDeleted: 0,\n\t\trowsDeleted: 0,\n\t};\n\n\tconst operations = new TransferOperationRepository(db);\n\tresult.expired = (await operations.expireDue(operationLimit)).length;\n\tawait new TransferApprovalRepository(db).expireDue();\n\n\tconst pending = await operations.listUncollected(operationLimit, {\n\t\timportRetentionSeconds: IMPORT_STAGING_RETENTION_SECONDS,\n\t});\n\tfor (const operation of pending) {\n\t\tif (objectBudget <= 0 || rowBatchBudget <= 0) break;\n\t\ttry {\n\t\t\tconst stage = new TransferStage(\n\t\t\t\tstorage,\n\t\t\t\tstagingPrefix(operation.kind, operation.id, operation.stagingSecret),\n\t\t\t);\n\t\t\tlet objectsDone = false;\n\t\t\twhile (objectBudget > 0) {\n\t\t\t\tconst deleted = await stage.deleteSome(Math.min(OBJECT_BATCH, objectBudget));\n\t\t\t\tobjectBudget -= deleted;\n\t\t\t\tresult.objectsDeleted += deleted;\n\t\t\t\tif (deleted === 0) {\n\t\t\t\t\tobjectsDone = true;\n\t\t\t\t\tbreak;\n\t\t\t\t}\n\t\t\t}\n\t\t\tif (!objectsDone) break;\n\n\t\t\tlet rowsDone = false;\n\t\t\twhile (rowBatchBudget > 0) {\n\t\t\t\trowBatchBudget--;\n\t\t\t\tconst deleted = await operations.deleteChildRows(operation.id, ROW_BATCH);\n\t\t\t\tresult.rowsDeleted += deleted;\n\t\t\t\tif (deleted === 0) {\n\t\t\t\t\trowsDone = true;\n\t\t\t\t\tbreak;\n\t\t\t\t}\n\t\t\t}\n\t\t\tif (!rowsDone) break;\n\n\t\t\tawait operations.markCollected(operation.id);\n\t\t\tresult.collected++;\n\t\t} catch (error) {\n\t\t\tconsole.error(`[transfer] Failed to collect staging for operation ${operation.id}:`, error);\n\t\t}\n\t}\n\treturn result;\n}\n"],"mappings":";;;;;;;;;;;;;;;;;;;;AAqBA,MAAa,mCAAmC,OAAU;AAkB1D,MAAM,eAAe;AACrB,MAAM,YAAY;AAElB,eAAsB,uBACrB,IACA,SACA,UAA4C,EAAE,EACH;CAC3C,MAAM,iBAAiB,QAAQ,cAAc;CAC7C,IAAI,eAAe,QAAQ,WAAW;CACtC,IAAI,iBAAiB,QAAQ,cAAc;CAC3C,MAAM,SAA0C;EAC/C,SAAS;EACT,WAAW;EACX,gBAAgB;EAChB,aAAa;EACb;CAED,MAAM,aAAa,IAAI,4BAA4B,GAAG;AACtD,QAAO,WAAW,MAAM,WAAW,UAAU,eAAe,EAAE;AAC9D,OAAM,IAAI,2BAA2B,GAAG,CAAC,WAAW;CAEpD,MAAM,UAAU,MAAM,WAAW,gBAAgB,gBAAgB,EAChE,wBAAwB,kCACxB,CAAC;AACF,MAAK,MAAM,aAAa,SAAS;AAChC,MAAI,gBAAgB,KAAK,kBAAkB,EAAG;AAC9C,MAAI;GACH,MAAM,QAAQ,IAAI,cACjB,SACA,cAAc,UAAU,MAAM,UAAU,IAAI,UAAU,cAAc,CACpE;GACD,IAAI,cAAc;AAClB,UAAO,eAAe,GAAG;IACxB,MAAM,UAAU,MAAM,MAAM,WAAW,KAAK,IAAI,cAAc,aAAa,CAAC;AAC5E,oBAAgB;AAChB,WAAO,kBAAkB;AACzB,QAAI,YAAY,GAAG;AAClB,mBAAc;AACd;;;AAGF,OAAI,CAAC,YAAa;GAElB,IAAI,WAAW;AACf,UAAO,iBAAiB,GAAG;AAC1B;IACA,MAAM,UAAU,MAAM,WAAW,gBAAgB,UAAU,IAAI,UAAU;AACzE,WAAO,eAAe;AACtB,QAAI,YAAY,GAAG;AAClB,gBAAW;AACX;;;AAGF,OAAI,CAAC,SAAU;AAEf,SAAM,WAAW,cAAc,UAAU,GAAG;AAC5C,UAAO;WACC,OAAO;AACf,WAAQ,MAAM,sDAAsD,UAAU,GAAG,IAAI,MAAM;;;AAG7F,QAAO"}

package/dist/hash-COVK27Pc.mjs

@@ -0,0 +1,37 @@
+//#region src/utils/hash.ts
+const MAX_CONTENT_HASH_BYTES = 8 * 1024 * 1024;
+function formatContentHash(hash) {
+	return `sha1:${Array.from(hash, (byte) => byte.toString(16).padStart(2, "0")).join("")}`;
+}
+/**
+* SHA-256 hash of a string, truncated to 16 hex chars (64 bits).
+* For cache invalidation / ETags — not for security.
+*/
+async function hashString(content) {
+	const buf = await crypto.subtle.digest("SHA-256", new TextEncoder().encode(content));
+	return Array.from(new Uint8Array(buf).slice(0, 8), (b) => b.toString(16).padStart(2, "0")).join("");
+}
+/**
+* Compute content hash using Web Crypto API
+*
+* Uses SHA-1 which is the fastest option in SubtleCrypto.
+* SHA-1 is cryptographically weak but fine for content deduplication
+* where we only need to detect identical files, not resist attacks.
+*
+* Returns hex string prefixed with "sha1:" for future-proofing
+*/
+async function computeContentHash(content) {
+	let buf;
+	if (content instanceof ArrayBuffer) buf = content;
+	else if (content.buffer instanceof ArrayBuffer && content.byteOffset === 0 && content.byteLength === content.buffer.byteLength) buf = content.buffer;
+	else {
+		buf = new ArrayBuffer(content.byteLength);
+		new Uint8Array(buf).set(content);
+	}
+	const hashBuffer = await crypto.subtle.digest("SHA-1", buf);
+	return formatContentHash(new Uint8Array(hashBuffer));
+}
+
+//#endregion
+export { computeContentHash as n, hashString as r, MAX_CONTENT_HASH_BYTES as t };
+//# sourceMappingURL=hash-COVK27Pc.mjs.map

package/dist/hash-COVK27Pc.mjs.map

@@ -0,0 +1 @@
+{"version":3,"file":"hash-COVK27Pc.mjs","names":[],"sources":["../src/utils/hash.ts"],"sourcesContent":["export const MAX_CONTENT_HASH_BYTES = 8 * 1024 * 1024;\n\nfunction formatContentHash(hash: Uint8Array): string {\n\tconst hashHex = Array.from(hash, (byte) => byte.toString(16).padStart(2, \"0\")).join(\"\");\n\treturn `sha1:${hashHex}`;\n}\n\n/**\n * SHA-256 hash of a string, truncated to 16 hex chars (64 bits).\n * For cache invalidation / ETags — not for security.\n */\nexport async function hashString(content: string): Promise<string> {\n\tconst buf = await crypto.subtle.digest(\"SHA-256\", new TextEncoder().encode(content));\n\treturn Array.from(new Uint8Array(buf).slice(0, 8), (b) => b.toString(16).padStart(2, \"0\")).join(\n\t\t\"\",\n\t);\n}\n\n/**\n * Compute content hash using Web Crypto API\n *\n * Uses SHA-1 which is the fastest option in SubtleCrypto.\n * SHA-1 is cryptographically weak but fine for content deduplication\n * where we only need to detect identical files, not resist attacks.\n *\n * Returns hex string prefixed with \"sha1:\" for future-proofing\n */\nexport async function computeContentHash(content: Uint8Array | ArrayBuffer): Promise<string> {\n\t// SubtleCrypto.digest() requires BufferSource (ArrayBuffer | ArrayBufferView<ArrayBuffer>).\n\t// Uint8Array.buffer is ArrayBufferLike which may include SharedArrayBuffer in the type system,\n\t// so we ensure we have a plain ArrayBuffer.\n\tlet buf: ArrayBuffer;\n\tif (content instanceof ArrayBuffer) {\n\t\tbuf = content;\n\t} else if (\n\t\tcontent.buffer instanceof ArrayBuffer &&\n\t\tcontent.byteOffset === 0 &&\n\t\tcontent.byteLength === content.buffer.byteLength\n\t) {\n\t\tbuf = content.buffer;\n\t} else {\n\t\tbuf = new ArrayBuffer(content.byteLength);\n\t\tnew Uint8Array(buf).set(content);\n\t}\n\tconst hashBuffer = await crypto.subtle.digest(\"SHA-1\", buf);\n\treturn formatContentHash(new Uint8Array(hashBuffer));\n}\n"],"mappings":";AAAA,MAAa,yBAAyB,IAAI,OAAO;AAEjD,SAAS,kBAAkB,MAA0B;AAEpD,QAAO,QADS,MAAM,KAAK,OAAO,SAAS,KAAK,SAAS,GAAG,CAAC,SAAS,GAAG,IAAI,CAAC,CAAC,KAAK,GAAG;;;;;;AAQxF,eAAsB,WAAW,SAAkC;CAClE,MAAM,MAAM,MAAM,OAAO,OAAO,OAAO,WAAW,IAAI,aAAa,CAAC,OAAO,QAAQ,CAAC;AACpF,QAAO,MAAM,KAAK,IAAI,WAAW,IAAI,CAAC,MAAM,GAAG,EAAE,GAAG,MAAM,EAAE,SAAS,GAAG,CAAC,SAAS,GAAG,IAAI,CAAC,CAAC,KAC1F,GACA;;;;;;;;;;;AAYF,eAAsB,mBAAmB,SAAoD;CAI5F,IAAI;AACJ,KAAI,mBAAmB,YACtB,OAAM;UAEN,QAAQ,kBAAkB,eAC1B,QAAQ,eAAe,KACvB,QAAQ,eAAe,QAAQ,OAAO,WAEtC,OAAM,QAAQ;MACR;AACN,QAAM,IAAI,YAAY,QAAQ,WAAW;AACzC,MAAI,WAAW,IAAI,CAAC,IAAI,QAAQ;;CAEjC,MAAM,aAAa,MAAM,OAAO,OAAO,OAAO,SAAS,IAAI;AAC3D,QAAO,kBAAkB,IAAI,WAAW,WAAW,CAAC"}