emdash 0.40.1 → 0.41.0

package/dist/search-Bt1PL1jf.mjs

@@ -1,448 +0,0 @@
-import { u as __exportAll } from "./runner-1y4J9fJB.mjs";
-import { t as validateIdentifier } from "./validate-DANX1-Qi.mjs";
-import { o as resolveConfiguredLocale } from "./config-CvmjJcVw.mjs";
-import { i as InvalidCursorError, o as decodeCursor, s as encodeCursor } from "./types-BkUg2yHx.mjs";
-import { t as FTSManager } from "./fts-manager-S-K0laoS.mjs";
-import { s as getDb } from "./loader-BzqeR1ZN.mjs";
-import { sql } from "kysely";
-import { toPlainText } from "@portabletext/toolkit";
-
-//#region src/search/query.ts
-/**
-* Marker stored in the `id` slot of a search pagination cursor.
-*
-* Search results are merged from per-collection FTS queries and re-sorted by
-* score, so there is no single stable keyset column to encode (the way
-* `getEmDashCollection` encodes `(orderValue, id)`). We page over the merged,
-* score-sorted set by offset instead, carried as the cursor's `orderValue`.
-* Reusing `encodeCursor`/`decodeCursor` keeps the opaque base64-JSON shape and
-* the `InvalidCursorError` handling consistent with the rest of the API.
-*/
-const SEARCH_CURSOR_MARKER = "search";
-/** Encode the next-page offset into an opaque search cursor. */
-function encodeSearchCursor(offset) {
-	return encodeCursor(String(offset), SEARCH_CURSOR_MARKER);
-}
-/**
-* Decode a search cursor back to its offset. Throws `InvalidCursorError` for a
-* malformed cursor or one that doesn't carry a non-negative integer offset, so
-* a client pagination bug surfaces immediately rather than silently restarting
-* from the first page.
-*/
-/** Upper bound on a decoded search offset, to cap the per-collection row fetch a forged cursor can trigger. */
-const MAX_SEARCH_OFFSET = 1e4;
-function decodeSearchOffset(cursor) {
-	const { orderValue, id } = decodeCursor(cursor);
-	if (id !== SEARCH_CURSOR_MARKER) throw new InvalidCursorError(cursor);
-	const offset = Number(orderValue);
-	if (!Number.isInteger(offset) || offset < 0 || offset > MAX_SEARCH_OFFSET) throw new InvalidCursorError(cursor);
-	return offset;
-}
-/** Pattern to split on whitespace for query term extraction */
-const WHITESPACE_SPLIT_PATTERN = /\s+/;
-const FTS_OPERATORS_PATTERN = /\b(AND|OR|NOT|NEAR)\b/i;
-const DOUBLE_QUOTE_PATTERN = /"/g;
-/**
-* Detect FTS5 query syntax errors. Match specifically on the SQLite FTS5
-* error fingerprints rather than a broad "fts5" / "syntax error" filter
-* (which would also swallow internal table-corruption errors). The two
-* fingerprints we care about are:
-*
-*  - "fts5: syntax error near …" — unbalanced quotes, stray operators,
-*    other malformed user input
-*  - "unknown special query: …" — bare special tokens like `^*` that
-*    parse but don't resolve to a real FTS5 directive
-*/
-function isFts5SyntaxError(error) {
-	if (!(error instanceof Error)) return false;
-	const message = error.message.toLowerCase();
-	return message.includes("fts5: syntax error") || message.includes("unknown special query");
-}
-/**
-* Search across multiple collections
-*
-* Public API that auto-injects the database.
-*
-* @param query - Search query (FTS5 syntax supported)
-* @param options - Search options
-* @returns Search results with pagination
-*
-* @example
-* ```typescript
-* import { search } from "emdash";
-*
-* const results = await search("hello world", {
-*   collections: ["posts", "pages"],
-*   limit: 20
-* });
-* ```
-*/
-async function search(query, options = {}) {
-	return searchWithDb(await getDb(), query, options);
-}
-/**
-* Search across multiple collections (with explicit db)
-*
-* @internal Use `search()` in templates. This variant is for admin routes
-* that already have a database handle.
-*
-* @param db - Kysely database instance
-* @param query - Search query (FTS5 syntax supported)
-* @param options - Search options
-* @returns Search results with pagination
-*/
-async function searchWithDb(db, query, options = {}) {
-	const ftsManager = new FTSManager(db);
-	const limit = options.limit ?? 20;
-	const status = options.status ?? "published";
-	const offset = options.cursor ? decodeSearchOffset(options.cursor) : 0;
-	let collections = options.collections;
-	if (!collections || collections.length === 0) collections = await getSearchableCollections(db);
-	if (collections.length === 0) return { items: [] };
-	const perCollectionLimit = offset + limit + 1;
-	const allResults = [];
-	const titleColumns = await ftsManager.getCollectionsWithTitleColumn(collections);
-	for (const collection of collections) {
-		const config = await ftsManager.getSearchConfig(collection);
-		if (!config?.enabled) continue;
-		const collectionResults = await searchSingleCollection(db, collection, query, {
-			status,
-			locale: options.locale,
-			limit: perCollectionLimit,
-			scope: options.scope
-		}, config.weights, titleColumns.has(collection), config.titleField);
-		allResults.push(...collectionResults);
-	}
-	allResults.sort((a, b) => b.score - a.score);
-	return {
-		items: allResults.slice(offset, offset + limit),
-		nextCursor: allResults.length > offset + limit ? encodeSearchCursor(offset + limit) : void 0
-	};
-}
-/**
-* Search within a single collection
-*
-* @param db - Kysely database instance
-* @param collection - Collection slug
-* @param query - Search query (FTS5 syntax supported)
-* @param options - Search options
-* @returns Search results with pagination
-*
-* @example
-* ```typescript
-* const results = await searchCollection(db, "posts", "hello world", {
-*   limit: 10
-* });
-* ```
-*/
-async function searchCollection(db, collection, query, options = {}) {
-	const config = await new FTSManager(db).getSearchConfig(collection);
-	if (!config?.enabled) return { items: [] };
-	const limit = options.limit ?? 20;
-	const offset = options.cursor ? decodeSearchOffset(options.cursor) : 0;
-	const fetched = await searchSingleCollection(db, collection, query, {
-		status: options.status,
-		locale: options.locale,
-		limit: offset + limit + 1,
-		scope: options.scope
-	}, config.weights, void 0, config.titleField);
-	return {
-		items: fetched.slice(offset, offset + limit),
-		nextCursor: fetched.length > offset + limit ? encodeSearchCursor(offset + limit) : void 0
-	};
-}
-/**
-* Internal function to search a single collection
-*/
-async function searchSingleCollection(db, collection, query, options, weights, hasTitle, titleField) {
-	validateIdentifier(collection, "collection slug");
-	const ftsManager = new FTSManager(db);
-	const ftsTable = ftsManager.getFtsTableName(collection);
-	const contentTable = ftsManager.getContentTableName(collection);
-	const limit = options.limit ?? 20;
-	const status = options.status ?? "published";
-	const locale = options.locale ? resolveConfiguredLocale(options.locale) : void 0;
-	if (!await ftsManager.ftsTableExists(collection)) return [];
-	const escapedQuery = escapeQuery(query);
-	if (!escapedQuery) return [];
-	const searchableFields = await ftsManager.getSearchableFields(collection);
-	let matchQuery = escapedQuery;
-	if (options.scope === "title") {
-		const titleColumn = titleField && searchableFields.includes(titleField) ? titleField : searchableFields.includes("title") ? "title" : void 0;
-		if (!titleColumn) return [];
-		validateIdentifier(titleColumn, "title field");
-		matchQuery = `${titleColumn} : (${escapeQuery(query, false)})`;
-	}
-	let titleExpr;
-	if (titleField) {
-		validateIdentifier(titleField, "title field");
-		titleExpr = sql`c.${sql.ref(titleField)}`;
-	} else titleExpr = hasTitle ?? await ftsManager.hasTitleColumn(collection) ? sql`c.title` : sql`NULL`;
-	let bm25Args = "";
-	if (weights && searchableFields.length > 0) {
-		const weightValues = ["0", "0"];
-		for (const field of searchableFields) weightValues.push(String(weights[field] ?? 1));
-		bm25Args = weightValues.join(", ");
-	}
-	const bm25Expr = bm25Args ? `bm25("${ftsTable}", ${bm25Args})` : `bm25("${ftsTable}")`;
-	let results;
-	try {
-		results = await sql`
-		SELECT
-			c.id,
-			c.slug,
-			c.locale,
-			${titleExpr} as title,
-			snippet("${sql.raw(ftsTable)}", 2, '<mark>', '</mark>', '...', 32) as snippet,
-			${sql.raw(bm25Expr)} as score
-		FROM "${sql.raw(ftsTable)}" f
-		JOIN "${sql.raw(contentTable)}" c ON f.id = c.id
-		WHERE "${sql.raw(ftsTable)}" MATCH ${matchQuery}
-		AND c.status = ${status}
-		AND c.deleted_at IS NULL
-		${locale ? sql`AND c.locale = ${locale}` : sql``}
-		ORDER BY score
-		LIMIT ${limit}
-	`.execute(db);
-	} catch (error) {
-		if (isFts5SyntaxError(error)) return [];
-		throw error;
-	}
-	return results.rows.map((row) => ({
-		collection,
-		id: row.id,
-		slug: row.slug,
-		locale: row.locale,
-		title: row.title ?? void 0,
-		snippet: row.snippet === null ? void 0 : sanitizeSnippet(row.snippet),
-		score: Math.abs(row.score)
-	}));
-}
-const SNIPPET_AMP_RE = /&/g;
-const SNIPPET_LT_RE = /</g;
-const SNIPPET_GT_RE = />/g;
-const SNIPPET_QUOT_RE = /"/g;
-const SNIPPET_APOS_RE = /'/g;
-/**
-* Make an FTS5 snippet safe to render with `set:html` / `innerHTML`.
-*
-* SQLite's `snippet()` function splices literal `<mark>` and `</mark>`
-* markers around matched terms but does not escape the surrounding
-* source text. Posts that legitimately contain `<`, `>`, `&`, `"` or
-* `'` would render as broken markup, and a `<script>` literal in a
-* title (or any other indexed field) would execute when displayed.
-*
-* The fix: HTML-escape the whole string, which turns the markers into
-* `&lt;mark&gt;` / `&lt;/mark&gt;`. Then restore those two patterns to
-* their original tag form. The result is "the indexed text with all
-* HTML metacharacters escaped, plus a small set of literal `<mark>`
-* highlight tags around matched terms" — which matches the API's
-* documented contract.
-*/
-function sanitizeSnippet(snippet) {
-	return snippet.replace(SNIPPET_AMP_RE, "&amp;").replace(SNIPPET_LT_RE, "&lt;").replace(SNIPPET_GT_RE, "&gt;").replace(SNIPPET_QUOT_RE, "&quot;").replace(SNIPPET_APOS_RE, "&#39;").replaceAll("&lt;mark&gt;", "<mark>").replaceAll("&lt;/mark&gt;", "</mark>");
-}
-/**
-* Get search suggestions for autocomplete
-*
-* @param db - Kysely database instance
-* @param query - Partial search query
-* @param options - Suggestion options
-* @returns Array of suggestions
-*/
-async function getSuggestions(db, query, options = {}) {
-	const limit = options.limit ?? 5;
-	const locale = options.locale ? resolveConfiguredLocale(options.locale) : void 0;
-	let collections = options.collections;
-	if (!collections || collections.length === 0) collections = await getSearchableCollections(db);
-	if (collections.length === 0) return [];
-	const suggestions = [];
-	const ftsManager = new FTSManager(db);
-	const titleColumns = await ftsManager.getCollectionsWithTitleColumn(collections);
-	for (const collection of collections) {
-		const config = await ftsManager.getSearchConfig(collection);
-		if (!config?.enabled) continue;
-		let titleExpr;
-		if (config.titleField) {
-			validateIdentifier(config.titleField, "title field");
-			titleExpr = sql`c.${sql.ref(config.titleField)}`;
-		} else if (titleColumns.has(collection)) titleExpr = sql`c.title`;
-		else continue;
-		validateIdentifier(collection, "collection slug");
-		const ftsTable = ftsManager.getFtsTableName(collection);
-		const contentTable = ftsManager.getContentTableName(collection);
-		const prefixQuery = escapeQuery(query);
-		if (!prefixQuery) continue;
-		let results;
-		try {
-			results = await sql`
-				SELECT
-					c.id,
-					c.slug,
-					${titleExpr} AS title
-				FROM "${sql.raw(ftsTable)}" f
-				JOIN "${sql.raw(contentTable)}" c ON f.id = c.id
-				WHERE "${sql.raw(ftsTable)}" MATCH ${prefixQuery}
-				AND c.status = 'published'
-				AND c.deleted_at IS NULL
-				AND ${titleExpr} IS NOT NULL
-				${locale ? sql`AND c.locale = ${locale}` : sql``}
-				ORDER BY bm25("${sql.raw(ftsTable)}")
-				LIMIT ${limit}
-			`.execute(db);
-		} catch (error) {
-			if (isFts5SyntaxError(error)) continue;
-			throw error;
-		}
-		for (const row of results.rows) suggestions.push({
-			collection,
-			id: row.id,
-			slug: row.slug,
-			title: row.title
-		});
-	}
-	return suggestions.slice(0, limit);
-}
-/**
-* Get search statistics for all collections
-*/
-async function getSearchStats(db) {
-	const ftsManager = new FTSManager(db);
-	const collections = await getSearchableCollections(db);
-	const stats = { collections: {} };
-	for (const collection of collections) {
-		const collectionStats = await ftsManager.getIndexStats(collection);
-		if (collectionStats) stats.collections[collection] = collectionStats;
-	}
-	return stats;
-}
-/**
-* Get list of collections with search enabled
-*/
-async function getSearchableCollections(db) {
-	return (await db.selectFrom("_emdash_collections").select(["slug", "search_config"]).execute()).filter((r) => {
-		if (!r.search_config) return false;
-		try {
-			return JSON.parse(r.search_config).enabled === true;
-		} catch {
-			return false;
-		}
-	}).map((r) => r.slug);
-}
-/**
-* Escape a query string for FTS5
-*
-* Handles special characters and prevents injection.
-*/
-function escapeQuery(query, allowOperators = true) {
-	if (!query || typeof query !== "string") return "";
-	query = query.trim();
-	if (query.length === 0) return "";
-	if (query.startsWith("\"") && query.endsWith("\"") && query.length >= 2) return `"${query.slice(1, -1).replace(DOUBLE_QUOTE_PATTERN, "\"\"")}"`;
-	const escaped = query.replace(DOUBLE_QUOTE_PATTERN, "\"\"");
-	if (allowOperators && FTS_OPERATORS_PATTERN.test(query)) return escaped;
-	const terms = escaped.split(WHITESPACE_SPLIT_PATTERN).filter((t) => t.length > 0);
-	if (terms.length === 0) return "";
-	return terms.map((t) => `"${t}"*`).join(" ");
-}
-
-//#endregion
-//#region src/search/text-extraction.ts
-/**
-* Text Extraction
-*
-* Extracts plain text from Portable Text blocks for FTS indexing.
-* Uses @portabletext/toolkit as base with extensions for custom block types.
-*/
-/**
-* Validate that a value looks like a Portable Text block array.
-* Each element must have at least a `_type` string property.
-*/
-function isPortableTextArray(value) {
-	return value.every((item) => typeof item === "object" && item !== null && "_type" in item && typeof item._type === "string");
-}
-/**
-* Extract additional text from custom block types that toPlainText doesn't handle
-*/
-function extractCustomBlockText(block) {
-	if (block._type === "code" && "code" in block && typeof block.code === "string") return block.code;
-	if (block._type === "image") {
-		const parts = [];
-		if ("alt" in block && typeof block.alt === "string" && block.alt) parts.push(block.alt);
-		if ("caption" in block && typeof block.caption === "string" && block.caption) parts.push(block.caption);
-		return parts.join(" ");
-	}
-	return "";
-}
-/**
-* Extract plain text from Portable Text blocks
-*
-* Uses @portabletext/toolkit's toPlainText for standard blocks,
-* plus extracts text from custom block types (code, images with alt/caption).
-*
-* @param blocks - Array of Portable Text blocks (or a JSON string)
-* @returns Plain text content
-*
-* @example
-* ```typescript
-* const text = extractPlainText([
-*   {
-*     _type: "block",
-*     _key: "abc",
-*     children: [{ _type: "span", _key: "s1", text: "Hello World" }]
-*   }
-* ]);
-* // Returns: "Hello World"
-* ```
-*/
-function extractPlainText(blocks) {
-	if (!blocks) return "";
-	let parsedBlocks;
-	if (typeof blocks === "string") try {
-		parsedBlocks = JSON.parse(blocks);
-	} catch {
-		return blocks;
-	}
-	else parsedBlocks = blocks;
-	if (!Array.isArray(parsedBlocks)) return "";
-	return [toPlainText(parsedBlocks.map((b) => {
-		const obj = { _type: b._type };
-		for (const [key, val] of Object.entries(b)) obj[key] = val;
-		return obj;
-	})), ...parsedBlocks.map(extractCustomBlockText).filter((text) => text.length > 0)].filter((t) => t.length > 0).join("\n");
-}
-/**
-* Extract searchable text from a content entry
-*
-* Extracts text from specified fields, handling both plain text and Portable Text.
-*
-* @param entry - Content entry data
-* @param fields - Field names to extract text from
-* @returns Object mapping field names to extracted text
-*/
-function extractSearchableFields(entry, fields) {
-	const result = {};
-	for (const field of fields) {
-		const value = entry[field];
-		if (value === null || value === void 0) {
-			result[field] = "";
-			continue;
-		}
-		if (typeof value === "string") if (value.startsWith("[")) result[field] = extractPlainText(value);
-		else result[field] = value;
-		else if (Array.isArray(value)) if (isPortableTextArray(value)) result[field] = extractPlainText(value);
-		else result[field] = JSON.stringify(value);
-		else if (typeof value === "object") result[field] = JSON.stringify(value);
-		else if (typeof value === "number" || typeof value === "boolean") result[field] = `${value}`;
-		else result[field] = "";
-	}
-	return result;
-}
-
-//#endregion
-//#region src/search/index.ts
-var search_exports = /* @__PURE__ */ __exportAll({ searchWithDb: () => searchWithDb });
-
-//#endregion
-export { getSuggestions as a, searchWithDb as c, getSearchStats as i, extractPlainText as n, search as o, extractSearchableFields as r, searchCollection as s, search_exports as t };
-//# sourceMappingURL=search-Bt1PL1jf.mjs.map

package/dist/search-Bt1PL1jf.mjs.map

@@ -1 +0,0 @@
-{"version":3,"file":"search-Bt1PL1jf.mjs","names":[],"sources":["../src/search/query.ts","../src/search/text-extraction.ts","../src/search/index.ts"],"sourcesContent":["/**\n * Search Query Functions\n *\n * Programmatic API for searching content using FTS5.\n */\n\nimport type { Kysely } from \"kysely\";\nimport { sql } from \"kysely\";\n\nimport { encodeCursor, decodeCursor, InvalidCursorError } from \"../database/repositories/types.js\";\nimport type { Database } from \"../database/types.js\";\nimport { validateIdentifier } from \"../database/validate.js\";\nimport { resolveConfiguredLocale } from \"../i18n/config.js\";\nimport { getDb } from \"../loader.js\";\nimport { FTSManager } from \"./fts-manager.js\";\nimport type {\n\tSearchOptions,\n\tCollectionSearchOptions,\n\tSearchResult,\n\tSearchResponse,\n\tSuggestOptions,\n\tSuggestion,\n\tSearchStats,\n} from \"./types.js\";\n\n/**\n * Marker stored in the `id` slot of a search pagination cursor.\n *\n * Search results are merged from per-collection FTS queries and re-sorted by\n * score, so there is no single stable keyset column to encode (the way\n * `getEmDashCollection` encodes `(orderValue, id)`). We page over the merged,\n * score-sorted set by offset instead, carried as the cursor's `orderValue`.\n * Reusing `encodeCursor`/`decodeCursor` keeps the opaque base64-JSON shape and\n * the `InvalidCursorError` handling consistent with the rest of the API.\n */\nconst SEARCH_CURSOR_MARKER = \"search\";\n\n/** Encode the next-page offset into an opaque search cursor. */\nfunction encodeSearchCursor(offset: number): string {\n\treturn encodeCursor(String(offset), SEARCH_CURSOR_MARKER);\n}\n\n/**\n * Decode a search cursor back to its offset. Throws `InvalidCursorError` for a\n * malformed cursor or one that doesn't carry a non-negative integer offset, so\n * a client pagination bug surfaces immediately rather than silently restarting\n * from the first page.\n */\n/** Upper bound on a decoded search offset, to cap the per-collection row fetch a forged cursor can trigger. */\nconst MAX_SEARCH_OFFSET = 10_000;\n\nfunction decodeSearchOffset(cursor: string): number {\n\tconst { orderValue, id } = decodeCursor(cursor);\n\tif (id !== SEARCH_CURSOR_MARKER) {\n\t\tthrow new InvalidCursorError(cursor);\n\t}\n\tconst offset = Number(orderValue);\n\tif (!Number.isInteger(offset) || offset < 0 || offset > MAX_SEARCH_OFFSET) {\n\t\tthrow new InvalidCursorError(cursor);\n\t}\n\treturn offset;\n}\n\n/** Pattern to split on whitespace for query term extraction */\nconst WHITESPACE_SPLIT_PATTERN = /\\s+/;\nconst FTS_OPERATORS_PATTERN = /\\b(AND|OR|NOT|NEAR)\\b/i;\nconst DOUBLE_QUOTE_PATTERN = /\"/g;\n\n/**\n * Detect FTS5 query syntax errors. Match specifically on the SQLite FTS5\n * error fingerprints rather than a broad \"fts5\" / \"syntax error\" filter\n * (which would also swallow internal table-corruption errors). The two\n * fingerprints we care about are:\n *\n *  - \"fts5: syntax error near …\" — unbalanced quotes, stray operators,\n *    other malformed user input\n *  - \"unknown special query: …\" — bare special tokens like `^*` that\n *    parse but don't resolve to a real FTS5 directive\n */\nfunction isFts5SyntaxError(error: unknown): boolean {\n\tif (!(error instanceof Error)) return false;\n\tconst message = error.message.toLowerCase();\n\treturn message.includes(\"fts5: syntax error\") || message.includes(\"unknown special query\");\n}\n\n/**\n * Search across multiple collections\n *\n * Public API that auto-injects the database.\n *\n * @param query - Search query (FTS5 syntax supported)\n * @param options - Search options\n * @returns Search results with pagination\n *\n * @example\n * ```typescript\n * import { search } from \"emdash\";\n *\n * const results = await search(\"hello world\", {\n *   collections: [\"posts\", \"pages\"],\n *   limit: 20\n * });\n * ```\n */\nexport async function search(query: string, options: SearchOptions = {}): Promise<SearchResponse> {\n\tconst db = await getDb();\n\treturn searchWithDb(db, query, options);\n}\n\n/**\n * Search across multiple collections (with explicit db)\n *\n * @internal Use `search()` in templates. This variant is for admin routes\n * that already have a database handle.\n *\n * @param db - Kysely database instance\n * @param query - Search query (FTS5 syntax supported)\n * @param options - Search options\n * @returns Search results with pagination\n */\nexport async function searchWithDb(\n\tdb: Kysely<Database>,\n\tquery: string,\n\toptions: SearchOptions = {},\n): Promise<SearchResponse> {\n\tconst ftsManager = new FTSManager(db);\n\tconst limit = options.limit ?? 20;\n\tconst status = options.status ?? \"published\";\n\tconst offset = options.cursor ? decodeSearchOffset(options.cursor) : 0;\n\n\t// Get searchable collections\n\tlet collections = options.collections;\n\tif (!collections || collections.length === 0) {\n\t\tcollections = await getSearchableCollections(db);\n\t}\n\n\tif (collections.length === 0) {\n\t\treturn { items: [] };\n\t}\n\n\t// To rank the merged window [offset, offset + limit) correctly, each\n\t// collection must contribute its own top (offset + limit) rows — in the\n\t// worst case the entire window comes from one collection. The extra +1\n\t// row detects whether a further page exists.\n\tconst perCollectionLimit = offset + limit + 1;\n\n\t// Search each collection and merge results\n\tconst allResults: SearchResult[] = [];\n\tconst titleColumns = await ftsManager.getCollectionsWithTitleColumn(collections);\n\n\tfor (const collection of collections) {\n\t\tconst config = await ftsManager.getSearchConfig(collection);\n\t\tif (!config?.enabled) {\n\t\t\tcontinue;\n\t\t}\n\n\t\tconst collectionResults = await searchSingleCollection(\n\t\t\tdb,\n\t\t\tcollection,\n\t\t\tquery,\n\t\t\t{\n\t\t\t\tstatus,\n\t\t\t\tlocale: options.locale,\n\t\t\t\tlimit: perCollectionLimit,\n\t\t\t\tscope: options.scope,\n\t\t\t},\n\t\t\tconfig.weights,\n\t\t\ttitleColumns.has(collection),\n\t\t\tconfig.titleField,\n\t\t);\n\n\t\tallResults.push(...collectionResults);\n\t}\n\n\t// Sort by score descending\n\tallResults.sort((a, b) => b.score - a.score);\n\n\t// Page the merged set by offset. A nextCursor is issued only when at least\n\t// one result exists past this page.\n\tconst items = allResults.slice(offset, offset + limit);\n\tconst hasMore = allResults.length > offset + limit;\n\tconst nextCursor = hasMore ? encodeSearchCursor(offset + limit) : undefined;\n\n\treturn { items, nextCursor };\n}\n\n/**\n * Search within a single collection\n *\n * @param db - Kysely database instance\n * @param collection - Collection slug\n * @param query - Search query (FTS5 syntax supported)\n * @param options - Search options\n * @returns Search results with pagination\n *\n * @example\n * ```typescript\n * const results = await searchCollection(db, \"posts\", \"hello world\", {\n *   limit: 10\n * });\n * ```\n */\nexport async function searchCollection(\n\tdb: Kysely<Database>,\n\tcollection: string,\n\tquery: string,\n\toptions: CollectionSearchOptions = {},\n): Promise<SearchResponse> {\n\tconst ftsManager = new FTSManager(db);\n\tconst config = await ftsManager.getSearchConfig(collection);\n\n\tif (!config?.enabled) {\n\t\treturn { items: [] };\n\t}\n\n\tconst limit = options.limit ?? 20;\n\tconst offset = options.cursor ? decodeSearchOffset(options.cursor) : 0;\n\n\t// Over-fetch the [offset, offset + limit) window plus one row to detect a\n\t// further page, then slice. Keeps the FTS SQL (LIMIT-only) unchanged and\n\t// the cursor contract identical to the cross-collection path.\n\tconst fetched = await searchSingleCollection(\n\t\tdb,\n\t\tcollection,\n\t\tquery,\n\t\t{\n\t\t\tstatus: options.status,\n\t\t\tlocale: options.locale,\n\t\t\tlimit: offset + limit + 1,\n\t\t\tscope: options.scope,\n\t\t},\n\t\tconfig.weights,\n\t\tundefined,\n\t\tconfig.titleField,\n\t);\n\n\tconst items = fetched.slice(offset, offset + limit);\n\tconst hasMore = fetched.length > offset + limit;\n\tconst nextCursor = hasMore ? encodeSearchCursor(offset + limit) : undefined;\n\n\treturn { items, nextCursor };\n}\n\n/**\n * Internal function to search a single collection\n */\nasync function searchSingleCollection(\n\tdb: Kysely<Database>,\n\tcollection: string,\n\tquery: string,\n\toptions: CollectionSearchOptions,\n\tweights?: Record<string, number>,\n\thasTitle?: boolean,\n\ttitleField?: string,\n): Promise<SearchResult[]> {\n\t// Validate before any raw SQL interpolation\n\tvalidateIdentifier(collection, \"collection slug\");\n\n\tconst ftsManager = new FTSManager(db);\n\tconst ftsTable = ftsManager.getFtsTableName(collection);\n\tconst contentTable = ftsManager.getContentTableName(collection);\n\tconst limit = options.limit ?? 20;\n\tconst status = options.status ?? \"published\";\n\tconst locale = options.locale ? resolveConfiguredLocale(options.locale) : undefined;\n\n\t// Check if FTS table exists\n\tif (!(await ftsManager.ftsTableExists(collection))) {\n\t\treturn [];\n\t}\n\n\t// Escape the query for FTS5\n\tconst escapedQuery = escapeQuery(query);\n\tif (!escapedQuery) {\n\t\treturn [];\n\t}\n\n\t// Get searchable fields for snippet generation\n\tconst searchableFields = await ftsManager.getSearchableFields(collection);\n\n\t// Title scope restricts the FTS5 match to the collection's title column.\n\t// The scoped column must be one of the indexed searchable fields; a\n\t// collection whose title is not indexed cannot match by title. Operators\n\t// are disabled so every term stays quoted inside the filter's group.\n\tlet matchQuery = escapedQuery;\n\tif (options.scope === \"title\") {\n\t\tconst titleColumn =\n\t\t\ttitleField && searchableFields.includes(titleField)\n\t\t\t\t? titleField\n\t\t\t\t: searchableFields.includes(\"title\")\n\t\t\t\t\t? \"title\"\n\t\t\t\t\t: undefined;\n\t\tif (!titleColumn) {\n\t\t\treturn [];\n\t\t}\n\t\tvalidateIdentifier(titleColumn, \"title field\");\n\t\tmatchQuery = `${titleColumn} : (${escapeQuery(query, false)})`;\n\t}\n\n\t// `title` is an optional user-defined field, not a system column. Only\n\t// select it when the collection actually has one; otherwise the query\n\t// errors with \"no such column: c.title\". Multi-collection callers\n\t// precompute this in bulk and pass it in; single-collection callers fall\n\t// back to the per-collection check.\n\t// A configured titleField drives the result title so search shows\n\t// the same value as the content list. It's a validated existing field, so\n\t// the column exists; otherwise fall back to the optional `title` column.\n\tlet titleExpr;\n\tif (titleField) {\n\t\tvalidateIdentifier(titleField, \"title field\");\n\t\ttitleExpr = sql`c.${sql.ref(titleField)}`;\n\t} else {\n\t\tconst collectionHasTitle = hasTitle ?? (await ftsManager.hasTitleColumn(collection));\n\t\ttitleExpr = collectionHasTitle ? sql`c.title` : sql`NULL`;\n\t}\n\n\t// Build weight string for bm25 if weights provided\n\t// Format: bm25(table, weight1, weight2, ...)\n\t// First two weights are for 'id' and 'locale' columns (UNINDEXED, so 0)\n\tlet bm25Args = \"\";\n\tif (weights && searchableFields.length > 0) {\n\t\tconst weightValues = [\"0\", \"0\"]; // id column, locale column\n\t\tfor (const field of searchableFields) {\n\t\t\tweightValues.push(String(weights[field] ?? 1));\n\t\t}\n\t\tbm25Args = weightValues.join(\", \");\n\t}\n\n\t// Build and execute the search query\n\t// Using raw SQL because Kysely doesn't have FTS5 support\n\tconst bm25Expr = bm25Args ? `bm25(\"${ftsTable}\", ${bm25Args})` : `bm25(\"${ftsTable}\")`;\n\n\t// Snippet column index is 2 (after id=0, locale=1, first searchable field=2)\n\tlet results;\n\ttry {\n\t\tresults = await sql<{\n\t\t\tid: string;\n\t\t\tslug: string | null;\n\t\t\tlocale: string;\n\t\t\ttitle: string | null;\n\t\t\tsnippet: string | null;\n\t\t\tscore: number;\n\t\t}>`\n\t\tSELECT\n\t\t\tc.id,\n\t\t\tc.slug,\n\t\t\tc.locale,\n\t\t\t${titleExpr} as title,\n\t\t\tsnippet(\"${sql.raw(ftsTable)}\", 2, '<mark>', '</mark>', '...', 32) as snippet,\n\t\t\t${sql.raw(bm25Expr)} as score\n\t\tFROM \"${sql.raw(ftsTable)}\" f\n\t\tJOIN \"${sql.raw(contentTable)}\" c ON f.id = c.id\n\t\tWHERE \"${sql.raw(ftsTable)}\" MATCH ${matchQuery}\n\t\tAND c.status = ${status}\n\t\tAND c.deleted_at IS NULL\n\t\t${locale ? sql`AND c.locale = ${locale}` : sql``}\n\t\tORDER BY score\n\t\tLIMIT ${limit}\n\t`.execute(db);\n\t} catch (error) {\n\t\t// FTS5 returns syntax errors for queries with unbalanced quotes,\n\t\t// stray operators, or other malformed input. Treat these as\n\t\t// \"no matches\" so the user gets an empty result rather than an\n\t\t// internals-leaking error. Other errors (table missing, IO) still\n\t\t// propagate. Intentionally not logged: any anonymous client can\n\t\t// trigger this path, and the underlying error message embeds the\n\t\t// raw query, so logging would be both noisy and a log-injection\n\t\t// vector.\n\t\tif (isFts5SyntaxError(error)) {\n\t\t\treturn [];\n\t\t}\n\t\tthrow error;\n\t}\n\n\treturn results.rows.map((row) => ({\n\t\tcollection,\n\t\tid: row.id,\n\t\tslug: row.slug,\n\t\tlocale: row.locale,\n\t\ttitle: row.title ?? undefined,\n\t\t// SQLite's snippet() returns NULL when the targeted column is\n\t\t// NULL for that row — even if the row matched via a different\n\t\t// searchable column. Skip sanitization in that case so we don't\n\t\t// throw on `null.replace`. The SearchResult.snippet field is\n\t\t// already optional, so omitting it is the documented contract.\n\t\tsnippet: row.snippet === null ? undefined : sanitizeSnippet(row.snippet),\n\t\tscore: Math.abs(row.score), // bm25 returns negative scores\n\t}));\n}\n\n// Module-scope regexes so the engine doesn't recompile per call —\n// snippet sanitization runs on every search result.\nconst SNIPPET_AMP_RE = /&/g;\nconst SNIPPET_LT_RE = /</g;\nconst SNIPPET_GT_RE = />/g;\nconst SNIPPET_QUOT_RE = /\"/g;\nconst SNIPPET_APOS_RE = /'/g;\n\n/**\n * Make an FTS5 snippet safe to render with `set:html` / `innerHTML`.\n *\n * SQLite's `snippet()` function splices literal `<mark>` and `</mark>`\n * markers around matched terms but does not escape the surrounding\n * source text. Posts that legitimately contain `<`, `>`, `&`, `\"` or\n * `'` would render as broken markup, and a `<script>` literal in a\n * title (or any other indexed field) would execute when displayed.\n *\n * The fix: HTML-escape the whole string, which turns the markers into\n * `&lt;mark&gt;` / `&lt;/mark&gt;`. Then restore those two patterns to\n * their original tag form. The result is \"the indexed text with all\n * HTML metacharacters escaped, plus a small set of literal `<mark>`\n * highlight tags around matched terms\" — which matches the API's\n * documented contract.\n */\nfunction sanitizeSnippet(snippet: string): string {\n\treturn snippet\n\t\t.replace(SNIPPET_AMP_RE, \"&amp;\")\n\t\t.replace(SNIPPET_LT_RE, \"&lt;\")\n\t\t.replace(SNIPPET_GT_RE, \"&gt;\")\n\t\t.replace(SNIPPET_QUOT_RE, \"&quot;\")\n\t\t.replace(SNIPPET_APOS_RE, \"&#39;\")\n\t\t.replaceAll(\"&lt;mark&gt;\", \"<mark>\")\n\t\t.replaceAll(\"&lt;/mark&gt;\", \"</mark>\");\n}\n\n/**\n * Get search suggestions for autocomplete\n *\n * @param db - Kysely database instance\n * @param query - Partial search query\n * @param options - Suggestion options\n * @returns Array of suggestions\n */\nexport async function getSuggestions(\n\tdb: Kysely<Database>,\n\tquery: string,\n\toptions: SuggestOptions = {},\n): Promise<Suggestion[]> {\n\tconst limit = options.limit ?? 5;\n\tconst locale = options.locale ? resolveConfiguredLocale(options.locale) : undefined;\n\n\t// Get searchable collections\n\tlet collections = options.collections;\n\tif (!collections || collections.length === 0) {\n\t\tcollections = await getSearchableCollections(db);\n\t}\n\n\tif (collections.length === 0) {\n\t\treturn [];\n\t}\n\n\tconst suggestions: Suggestion[] = [];\n\tconst ftsManager = new FTSManager(db);\n\tconst titleColumns = await ftsManager.getCollectionsWithTitleColumn(collections);\n\n\tfor (const collection of collections) {\n\t\tconst config = await ftsManager.getSearchConfig(collection);\n\t\tif (!config?.enabled) {\n\t\t\tcontinue;\n\t\t}\n\n\t\t// Suggestions are title-based (Suggestion.title is required). A configured\n\t\t// titleField supplies it -- a validated existing column -- so it\n\t\t// takes precedence and keeps autocomplete consistent with search results.\n\t\t// Otherwise fall back to the optional `title` field; collections with\n\t\t// neither can't produce a suggestion and are skipped (selecting a missing\n\t\t// column would error).\n\t\tlet titleExpr;\n\t\tif (config.titleField) {\n\t\t\tvalidateIdentifier(config.titleField, \"title field\");\n\t\t\ttitleExpr = sql`c.${sql.ref(config.titleField)}`;\n\t\t} else if (titleColumns.has(collection)) {\n\t\t\ttitleExpr = sql`c.title`;\n\t\t} else {\n\t\t\tcontinue;\n\t\t}\n\n\t\t// Validate before raw SQL interpolation\n\t\tvalidateIdentifier(collection, \"collection slug\");\n\n\t\tconst ftsTable = ftsManager.getFtsTableName(collection);\n\t\tconst contentTable = ftsManager.getContentTableName(collection);\n\n\t\t// Use prefix search for autocomplete. `escapeQuery` already appends `*`\n\t\t// to each term for prefix matching, so we must not append another one.\n\t\tconst prefixQuery = escapeQuery(query);\n\t\tif (!prefixQuery) {\n\t\t\tcontinue;\n\t\t}\n\n\t\tlet results;\n\t\ttry {\n\t\t\tresults = await sql<{\n\t\t\t\tid: string;\n\t\t\t\tslug: string | null;\n\t\t\t\ttitle: string;\n\t\t\t}>`\n\t\t\t\tSELECT\n\t\t\t\t\tc.id,\n\t\t\t\t\tc.slug,\n\t\t\t\t\t${titleExpr} AS title\n\t\t\t\tFROM \"${sql.raw(ftsTable)}\" f\n\t\t\t\tJOIN \"${sql.raw(contentTable)}\" c ON f.id = c.id\n\t\t\t\tWHERE \"${sql.raw(ftsTable)}\" MATCH ${prefixQuery}\n\t\t\t\tAND c.status = 'published'\n\t\t\t\tAND c.deleted_at IS NULL\n\t\t\t\tAND ${titleExpr} IS NOT NULL\n\t\t\t\t${locale ? sql`AND c.locale = ${locale}` : sql``}\n\t\t\t\tORDER BY bm25(\"${sql.raw(ftsTable)}\")\n\t\t\t\tLIMIT ${limit}\n\t\t\t`.execute(db);\n\t\t} catch (error) {\n\t\t\t// Same swallow as searchSingleCollection: malformed prefix\n\t\t\t// queries should yield no suggestions, not surface DB errors.\n\t\t\t// Intentionally not logged (anonymous-triggerable, echoes\n\t\t\t// user input -- see searchSingleCollection for rationale).\n\t\t\tif (isFts5SyntaxError(error)) {\n\t\t\t\tcontinue;\n\t\t\t}\n\t\t\tthrow error;\n\t\t}\n\n\t\tfor (const row of results.rows) {\n\t\t\tsuggestions.push({\n\t\t\t\tcollection,\n\t\t\t\tid: row.id,\n\t\t\t\tslug: row.slug,\n\t\t\t\ttitle: row.title,\n\t\t\t});\n\t\t}\n\t}\n\n\treturn suggestions.slice(0, limit);\n}\n\n/**\n * Get search statistics for all collections\n */\nexport async function getSearchStats(db: Kysely<Database>): Promise<SearchStats> {\n\tconst ftsManager = new FTSManager(db);\n\tconst collections = await getSearchableCollections(db);\n\tconst stats: SearchStats = { collections: {} };\n\n\tfor (const collection of collections) {\n\t\tconst collectionStats = await ftsManager.getIndexStats(collection);\n\t\tif (collectionStats) {\n\t\t\tstats.collections[collection] = collectionStats;\n\t\t}\n\t}\n\n\treturn stats;\n}\n\n/**\n * Get list of collections with search enabled\n */\nasync function getSearchableCollections(db: Kysely<Database>): Promise<string[]> {\n\tconst results = await db\n\t\t.selectFrom(\"_emdash_collections\")\n\t\t.select([\"slug\", \"search_config\"])\n\t\t.execute();\n\n\treturn results\n\t\t.filter((r) => {\n\t\t\tif (!r.search_config) return false;\n\t\t\ttry {\n\t\t\t\tconst config = JSON.parse(r.search_config);\n\t\t\t\treturn config.enabled === true;\n\t\t\t} catch {\n\t\t\t\treturn false;\n\t\t\t}\n\t\t})\n\t\t.map((r) => r.slug);\n}\n\n/**\n * Escape a query string for FTS5\n *\n * Handles special characters and prevents injection.\n */\nfunction escapeQuery(query: string, allowOperators = true): string {\n\tif (!query || typeof query !== \"string\") {\n\t\treturn \"\";\n\t}\n\n\t// Trim whitespace\n\tquery = query.trim();\n\n\tif (query.length === 0) {\n\t\treturn \"\";\n\t}\n\n\t// If already a quoted phrase, escape only interior quotes and preserve phrase syntax\n\tif (query.startsWith('\"') && query.endsWith('\"') && query.length >= 2) {\n\t\tconst inner = query.slice(1, -1);\n\t\treturn `\"${inner.replace(DOUBLE_QUOTE_PATTERN, '\"\"')}\"`;\n\t}\n\n\t// Escape any existing quotes\n\tconst escaped = query.replace(DOUBLE_QUOTE_PATTERN, '\"\"');\n\n\t// If the query contains FTS5 operators (AND, OR, NOT, NEAR),\n\t// pass through with quotes escaped but operators preserved. Not allowed\n\t// when the match string is embedded in a column filter: unquoted parens\n\t// would let a clause escape the filter's group.\n\tif (allowOperators && FTS_OPERATORS_PATTERN.test(query)) {\n\t\treturn escaped;\n\t}\n\n\t// For simple queries, wrap each word to handle special chars\n\tconst terms = escaped.split(WHITESPACE_SPLIT_PATTERN).filter((t) => t.length > 0);\n\tif (terms.length === 0) {\n\t\treturn \"\";\n\t}\n\n\t// Join with implicit AND, add prefix matching (*) to all terms\n\t// This allows \"hel wor\" to match \"hello world\"\n\treturn terms.map((t) => `\"${t}\"*`).join(\" \");\n}\n","/**\n * Text Extraction\n *\n * Extracts plain text from Portable Text blocks for FTS indexing.\n * Uses @portabletext/toolkit as base with extensions for custom block types.\n */\n\nimport { toPlainText } from \"@portabletext/toolkit\";\n\nimport type { PortableTextBlock } from \"../content/converters/types.js\";\n\n/**\n * Validate that a value looks like a Portable Text block array.\n * Each element must have at least a `_type` string property.\n */\nfunction isPortableTextArray(value: unknown[]): value is PortableTextBlock[] {\n\treturn value.every(\n\t\t(item) =>\n\t\t\ttypeof item === \"object\" &&\n\t\t\titem !== null &&\n\t\t\t\"_type\" in item &&\n\t\t\ttypeof item._type === \"string\",\n\t);\n}\n\n/**\n * Extract additional text from custom block types that toPlainText doesn't handle\n */\nfunction extractCustomBlockText(block: PortableTextBlock): string {\n\t// Code blocks - include the code content\n\tif (block._type === \"code\" && \"code\" in block && typeof block.code === \"string\") {\n\t\treturn block.code;\n\t}\n\n\t// Image blocks - include alt text and caption\n\tif (block._type === \"image\") {\n\t\tconst parts: string[] = [];\n\t\tif (\"alt\" in block && typeof block.alt === \"string\" && block.alt) {\n\t\t\tparts.push(block.alt);\n\t\t}\n\t\tif (\"caption\" in block && typeof block.caption === \"string\" && block.caption) {\n\t\t\tparts.push(block.caption);\n\t\t}\n\t\treturn parts.join(\" \");\n\t}\n\n\treturn \"\";\n}\n\n/**\n * Extract plain text from Portable Text blocks\n *\n * Uses @portabletext/toolkit's toPlainText for standard blocks,\n * plus extracts text from custom block types (code, images with alt/caption).\n *\n * @param blocks - Array of Portable Text blocks (or a JSON string)\n * @returns Plain text content\n *\n * @example\n * ```typescript\n * const text = extractPlainText([\n *   {\n *     _type: \"block\",\n *     _key: \"abc\",\n *     children: [{ _type: \"span\", _key: \"s1\", text: \"Hello World\" }]\n *   }\n * ]);\n * // Returns: \"Hello World\"\n * ```\n */\nexport function extractPlainText(blocks: PortableTextBlock[] | string | null | undefined): string {\n\tif (!blocks) {\n\t\treturn \"\";\n\t}\n\n\t// Handle JSON string input\n\tlet parsedBlocks: PortableTextBlock[];\n\tif (typeof blocks === \"string\") {\n\t\ttry {\n\t\t\tparsedBlocks = JSON.parse(blocks);\n\t\t} catch {\n\t\t\t// If it's not valid JSON, treat as plain text\n\t\t\treturn blocks;\n\t\t}\n\t} else {\n\t\tparsedBlocks = blocks;\n\t}\n\n\tif (!Array.isArray(parsedBlocks)) {\n\t\treturn \"\";\n\t}\n\n\t// Use official toPlainText for standard blocks.\n\t// toPlainText expects `{ _type: string; [key: string]: any }[]` but our blocks use\n\t// `unknown` index sigs. They're structurally compatible at runtime — spread each block\n\t// to satisfy the wider index signature without an unsafe cast.\n\tconst toolkitBlocks = parsedBlocks.map((b) => {\n\t\tconst obj: Record<string, unknown> & { _type: string } = { _type: b._type };\n\t\tfor (const [key, val] of Object.entries(b)) {\n\t\t\tobj[key] = val;\n\t\t}\n\t\treturn obj;\n\t});\n\tconst standardText = toPlainText(toolkitBlocks);\n\n\t// Extract text from custom block types that toPlainText doesn't handle\n\tconst customTexts = parsedBlocks.map(extractCustomBlockText).filter((text) => text.length > 0);\n\n\t// Combine both\n\tconst allTexts = [standardText, ...customTexts].filter((t) => t.length > 0);\n\treturn allTexts.join(\"\\n\");\n}\n\n/**\n * Extract searchable text from a content entry\n *\n * Extracts text from specified fields, handling both plain text and Portable Text.\n *\n * @param entry - Content entry data\n * @param fields - Field names to extract text from\n * @returns Object mapping field names to extracted text\n */\nexport function extractSearchableFields(\n\tentry: Record<string, unknown>,\n\tfields: string[],\n): Record<string, string> {\n\tconst result: Record<string, string> = {};\n\n\tfor (const field of fields) {\n\t\tconst value = entry[field];\n\n\t\tif (value === null || value === undefined) {\n\t\t\tresult[field] = \"\";\n\t\t\tcontinue;\n\t\t}\n\n\t\tif (typeof value === \"string\") {\n\t\t\t// Could be plain text or JSON Portable Text\n\t\t\tif (value.startsWith(\"[\")) {\n\t\t\t\tresult[field] = extractPlainText(value);\n\t\t\t} else {\n\t\t\t\tresult[field] = value;\n\t\t\t}\n\t\t} else if (Array.isArray(value)) {\n\t\t\t// Validate the array looks like Portable Text before treating it as such\n\t\t\tif (isPortableTextArray(value)) {\n\t\t\t\tresult[field] = extractPlainText(value);\n\t\t\t} else {\n\t\t\t\tresult[field] = JSON.stringify(value);\n\t\t\t}\n\t\t} else if (typeof value === \"object\") {\n\t\t\t// Object — serialize to JSON for searchable text\n\t\t\tresult[field] = JSON.stringify(value);\n\t\t} else if (typeof value === \"number\" || typeof value === \"boolean\") {\n\t\t\tresult[field] = `${value}`;\n\t\t} else {\n\t\t\tresult[field] = \"\";\n\t\t}\n\t}\n\n\treturn result;\n}\n","/**\n * Search Module\n *\n * Full-text search for EmDash using SQLite FTS5.\n */\n\n// Types\nexport type {\n\tSearchConfig,\n\tSearchTokenizer,\n\tSearchOptions,\n\tCollectionSearchOptions,\n\tSearchResult,\n\tSearchResponse,\n\tSuggestOptions,\n\tSuggestion,\n\tSearchStats,\n} from \"./types.js\";\nexport { SEARCH_TOKENIZERS } from \"./types.js\";\n\n// FTS Manager\nexport { FTSManager } from \"./fts-manager.js\";\n\n// Query functions (public API uses getDb() internally)\nexport { search, searchWithDb, searchCollection, getSuggestions, getSearchStats } from \"./query.js\";\n\n// Text extraction\nexport { extractPlainText, extractSearchableFields } from \"./text-extraction.js\";\n"],"mappings":";;;;;;;;;;;;;;;;;;;;AAmCA,MAAM,uBAAuB;;AAG7B,SAAS,mBAAmB,QAAwB;AACnD,QAAO,aAAa,OAAO,OAAO,EAAE,qBAAqB;;;;;;;;;AAU1D,MAAM,oBAAoB;AAE1B,SAAS,mBAAmB,QAAwB;CACnD,MAAM,EAAE,YAAY,OAAO,aAAa,OAAO;AAC/C,KAAI,OAAO,qBACV,OAAM,IAAI,mBAAmB,OAAO;CAErC,MAAM,SAAS,OAAO,WAAW;AACjC,KAAI,CAAC,OAAO,UAAU,OAAO,IAAI,SAAS,KAAK,SAAS,kBACvD,OAAM,IAAI,mBAAmB,OAAO;AAErC,QAAO;;;AAIR,MAAM,2BAA2B;AACjC,MAAM,wBAAwB;AAC9B,MAAM,uBAAuB;;;;;;;;;;;;AAa7B,SAAS,kBAAkB,OAAyB;AACnD,KAAI,EAAE,iBAAiB,OAAQ,QAAO;CACtC,MAAM,UAAU,MAAM,QAAQ,aAAa;AAC3C,QAAO,QAAQ,SAAS,qBAAqB,IAAI,QAAQ,SAAS,wBAAwB;;;;;;;;;;;;;;;;;;;;;AAsB3F,eAAsB,OAAO,OAAe,UAAyB,EAAE,EAA2B;AAEjG,QAAO,aADI,MAAM,OAAO,EACA,OAAO,QAAQ;;;;;;;;;;;;;AAcxC,eAAsB,aACrB,IACA,OACA,UAAyB,EAAE,EACD;CAC1B,MAAM,aAAa,IAAI,WAAW,GAAG;CACrC,MAAM,QAAQ,QAAQ,SAAS;CAC/B,MAAM,SAAS,QAAQ,UAAU;CACjC,MAAM,SAAS,QAAQ,SAAS,mBAAmB,QAAQ,OAAO,GAAG;CAGrE,IAAI,cAAc,QAAQ;AAC1B,KAAI,CAAC,eAAe,YAAY,WAAW,EAC1C,eAAc,MAAM,yBAAyB,GAAG;AAGjD,KAAI,YAAY,WAAW,EAC1B,QAAO,EAAE,OAAO,EAAE,EAAE;CAOrB,MAAM,qBAAqB,SAAS,QAAQ;CAG5C,MAAM,aAA6B,EAAE;CACrC,MAAM,eAAe,MAAM,WAAW,8BAA8B,YAAY;AAEhF,MAAK,MAAM,cAAc,aAAa;EACrC,MAAM,SAAS,MAAM,WAAW,gBAAgB,WAAW;AAC3D,MAAI,CAAC,QAAQ,QACZ;EAGD,MAAM,oBAAoB,MAAM,uBAC/B,IACA,YACA,OACA;GACC;GACA,QAAQ,QAAQ;GAChB,OAAO;GACP,OAAO,QAAQ;GACf,EACD,OAAO,SACP,aAAa,IAAI,WAAW,EAC5B,OAAO,WACP;AAED,aAAW,KAAK,GAAG,kBAAkB;;AAItC,YAAW,MAAM,GAAG,MAAM,EAAE,QAAQ,EAAE,MAAM;AAQ5C,QAAO;EAAE,OAJK,WAAW,MAAM,QAAQ,SAAS,MAAM;EAItC,YAHA,WAAW,SAAS,SAAS,QAChB,mBAAmB,SAAS,MAAM,GAAG;EAEtC;;;;;;;;;;;;;;;;;;AAmB7B,eAAsB,iBACrB,IACA,YACA,OACA,UAAmC,EAAE,EACX;CAE1B,MAAM,SAAS,MADI,IAAI,WAAW,GAAG,CACL,gBAAgB,WAAW;AAE3D,KAAI,CAAC,QAAQ,QACZ,QAAO,EAAE,OAAO,EAAE,EAAE;CAGrB,MAAM,QAAQ,QAAQ,SAAS;CAC/B,MAAM,SAAS,QAAQ,SAAS,mBAAmB,QAAQ,OAAO,GAAG;CAKrE,MAAM,UAAU,MAAM,uBACrB,IACA,YACA,OACA;EACC,QAAQ,QAAQ;EAChB,QAAQ,QAAQ;EAChB,OAAO,SAAS,QAAQ;EACxB,OAAO,QAAQ;EACf,EACD,OAAO,SACP,QACA,OAAO,WACP;AAMD,QAAO;EAAE,OAJK,QAAQ,MAAM,QAAQ,SAAS,MAAM;EAInC,YAHA,QAAQ,SAAS,SAAS,QACb,mBAAmB,SAAS,MAAM,GAAG;EAEtC;;;;;AAM7B,eAAe,uBACd,IACA,YACA,OACA,SACA,SACA,UACA,YAC0B;AAE1B,oBAAmB,YAAY,kBAAkB;CAEjD,MAAM,aAAa,IAAI,WAAW,GAAG;CACrC,MAAM,WAAW,WAAW,gBAAgB,WAAW;CACvD,MAAM,eAAe,WAAW,oBAAoB,WAAW;CAC/D,MAAM,QAAQ,QAAQ,SAAS;CAC/B,MAAM,SAAS,QAAQ,UAAU;CACjC,MAAM,SAAS,QAAQ,SAAS,wBAAwB,QAAQ,OAAO,GAAG;AAG1E,KAAI,CAAE,MAAM,WAAW,eAAe,WAAW,CAChD,QAAO,EAAE;CAIV,MAAM,eAAe,YAAY,MAAM;AACvC,KAAI,CAAC,aACJ,QAAO,EAAE;CAIV,MAAM,mBAAmB,MAAM,WAAW,oBAAoB,WAAW;CAMzE,IAAI,aAAa;AACjB,KAAI,QAAQ,UAAU,SAAS;EAC9B,MAAM,cACL,cAAc,iBAAiB,SAAS,WAAW,GAChD,aACA,iBAAiB,SAAS,QAAQ,GACjC,UACA;AACL,MAAI,CAAC,YACJ,QAAO,EAAE;AAEV,qBAAmB,aAAa,cAAc;AAC9C,eAAa,GAAG,YAAY,MAAM,YAAY,OAAO,MAAM,CAAC;;CAW7D,IAAI;AACJ,KAAI,YAAY;AACf,qBAAmB,YAAY,cAAc;AAC7C,cAAY,GAAG,KAAK,IAAI,IAAI,WAAW;OAGvC,aAD2B,YAAa,MAAM,WAAW,eAAe,WAAW,GAClD,GAAG,YAAY,GAAG;CAMpD,IAAI,WAAW;AACf,KAAI,WAAW,iBAAiB,SAAS,GAAG;EAC3C,MAAM,eAAe,CAAC,KAAK,IAAI;AAC/B,OAAK,MAAM,SAAS,iBACnB,cAAa,KAAK,OAAO,QAAQ,UAAU,EAAE,CAAC;AAE/C,aAAW,aAAa,KAAK,KAAK;;CAKnC,MAAM,WAAW,WAAW,SAAS,SAAS,KAAK,SAAS,KAAK,SAAS,SAAS;CAGnF,IAAI;AACJ,KAAI;AACH,YAAU,MAAM,GAOd;;;;;KAKC,UAAU;cACD,IAAI,IAAI,SAAS,CAAC;KAC3B,IAAI,IAAI,SAAS,CAAC;UACb,IAAI,IAAI,SAAS,CAAC;UAClB,IAAI,IAAI,aAAa,CAAC;WACrB,IAAI,IAAI,SAAS,CAAC,UAAU,WAAW;mBAC/B,OAAO;;IAEtB,SAAS,GAAG,kBAAkB,WAAW,GAAG,GAAG;;UAEzC,MAAM;GACb,QAAQ,GAAG;UACJ,OAAO;AASf,MAAI,kBAAkB,MAAM,CAC3B,QAAO,EAAE;AAEV,QAAM;;AAGP,QAAO,QAAQ,KAAK,KAAK,SAAS;EACjC;EACA,IAAI,IAAI;EACR,MAAM,IAAI;EACV,QAAQ,IAAI;EACZ,OAAO,IAAI,SAAS;EAMpB,SAAS,IAAI,YAAY,OAAO,SAAY,gBAAgB,IAAI,QAAQ;EACxE,OAAO,KAAK,IAAI,IAAI,MAAM;EAC1B,EAAE;;AAKJ,MAAM,iBAAiB;AACvB,MAAM,gBAAgB;AACtB,MAAM,gBAAgB;AACtB,MAAM,kBAAkB;AACxB,MAAM,kBAAkB;;;;;;;;;;;;;;;;;AAkBxB,SAAS,gBAAgB,SAAyB;AACjD,QAAO,QACL,QAAQ,gBAAgB,QAAQ,CAChC,QAAQ,eAAe,OAAO,CAC9B,QAAQ,eAAe,OAAO,CAC9B,QAAQ,iBAAiB,SAAS,CAClC,QAAQ,iBAAiB,QAAQ,CACjC,WAAW,gBAAgB,SAAS,CACpC,WAAW,iBAAiB,UAAU;;;;;;;;;;AAWzC,eAAsB,eACrB,IACA,OACA,UAA0B,EAAE,EACJ;CACxB,MAAM,QAAQ,QAAQ,SAAS;CAC/B,MAAM,SAAS,QAAQ,SAAS,wBAAwB,QAAQ,OAAO,GAAG;CAG1E,IAAI,cAAc,QAAQ;AAC1B,KAAI,CAAC,eAAe,YAAY,WAAW,EAC1C,eAAc,MAAM,yBAAyB,GAAG;AAGjD,KAAI,YAAY,WAAW,EAC1B,QAAO,EAAE;CAGV,MAAM,cAA4B,EAAE;CACpC,MAAM,aAAa,IAAI,WAAW,GAAG;CACrC,MAAM,eAAe,MAAM,WAAW,8BAA8B,YAAY;AAEhF,MAAK,MAAM,cAAc,aAAa;EACrC,MAAM,SAAS,MAAM,WAAW,gBAAgB,WAAW;AAC3D,MAAI,CAAC,QAAQ,QACZ;EASD,IAAI;AACJ,MAAI,OAAO,YAAY;AACtB,sBAAmB,OAAO,YAAY,cAAc;AACpD,eAAY,GAAG,KAAK,IAAI,IAAI,OAAO,WAAW;aACpC,aAAa,IAAI,WAAW,CACtC,aAAY,GAAG;MAEf;AAID,qBAAmB,YAAY,kBAAkB;EAEjD,MAAM,WAAW,WAAW,gBAAgB,WAAW;EACvD,MAAM,eAAe,WAAW,oBAAoB,WAAW;EAI/D,MAAM,cAAc,YAAY,MAAM;AACtC,MAAI,CAAC,YACJ;EAGD,IAAI;AACJ,MAAI;AACH,aAAU,MAAM,GAId;;;;OAIE,UAAU;YACL,IAAI,IAAI,SAAS,CAAC;YAClB,IAAI,IAAI,aAAa,CAAC;aACrB,IAAI,IAAI,SAAS,CAAC,UAAU,YAAY;;;UAG3C,UAAU;MACd,SAAS,GAAG,kBAAkB,WAAW,GAAG,GAAG;qBAChC,IAAI,IAAI,SAAS,CAAC;YAC3B,MAAM;KACb,QAAQ,GAAG;WACL,OAAO;AAKf,OAAI,kBAAkB,MAAM,CAC3B;AAED,SAAM;;AAGP,OAAK,MAAM,OAAO,QAAQ,KACzB,aAAY,KAAK;GAChB;GACA,IAAI,IAAI;GACR,MAAM,IAAI;GACV,OAAO,IAAI;GACX,CAAC;;AAIJ,QAAO,YAAY,MAAM,GAAG,MAAM;;;;;AAMnC,eAAsB,eAAe,IAA4C;CAChF,MAAM,aAAa,IAAI,WAAW,GAAG;CACrC,MAAM,cAAc,MAAM,yBAAyB,GAAG;CACtD,MAAM,QAAqB,EAAE,aAAa,EAAE,EAAE;AAE9C,MAAK,MAAM,cAAc,aAAa;EACrC,MAAM,kBAAkB,MAAM,WAAW,cAAc,WAAW;AAClE,MAAI,gBACH,OAAM,YAAY,cAAc;;AAIlC,QAAO;;;;;AAMR,eAAe,yBAAyB,IAAyC;AAMhF,SALgB,MAAM,GACpB,WAAW,sBAAsB,CACjC,OAAO,CAAC,QAAQ,gBAAgB,CAAC,CACjC,SAAS,EAGT,QAAQ,MAAM;AACd,MAAI,CAAC,EAAE,cAAe,QAAO;AAC7B,MAAI;AAEH,UADe,KAAK,MAAM,EAAE,cAAc,CAC5B,YAAY;UACnB;AACP,UAAO;;GAEP,CACD,KAAK,MAAM,EAAE,KAAK;;;;;;;AAQrB,SAAS,YAAY,OAAe,iBAAiB,MAAc;AAClE,KAAI,CAAC,SAAS,OAAO,UAAU,SAC9B,QAAO;AAIR,SAAQ,MAAM,MAAM;AAEpB,KAAI,MAAM,WAAW,EACpB,QAAO;AAIR,KAAI,MAAM,WAAW,KAAI,IAAI,MAAM,SAAS,KAAI,IAAI,MAAM,UAAU,EAEnE,QAAO,IADO,MAAM,MAAM,GAAG,GAAG,CACf,QAAQ,sBAAsB,OAAK,CAAC;CAItD,MAAM,UAAU,MAAM,QAAQ,sBAAsB,OAAK;AAMzD,KAAI,kBAAkB,sBAAsB,KAAK,MAAM,CACtD,QAAO;CAIR,MAAM,QAAQ,QAAQ,MAAM,yBAAyB,CAAC,QAAQ,MAAM,EAAE,SAAS,EAAE;AACjF,KAAI,MAAM,WAAW,EACpB,QAAO;AAKR,QAAO,MAAM,KAAK,MAAM,IAAI,EAAE,IAAI,CAAC,KAAK,IAAI;;;;;;;;;;;;;;;ACzlB7C,SAAS,oBAAoB,OAAgD;AAC5E,QAAO,MAAM,OACX,SACA,OAAO,SAAS,YAChB,SAAS,QACT,WAAW,QACX,OAAO,KAAK,UAAU,SACvB;;;;;AAMF,SAAS,uBAAuB,OAAkC;AAEjE,KAAI,MAAM,UAAU,UAAU,UAAU,SAAS,OAAO,MAAM,SAAS,SACtE,QAAO,MAAM;AAId,KAAI,MAAM,UAAU,SAAS;EAC5B,MAAM,QAAkB,EAAE;AAC1B,MAAI,SAAS,SAAS,OAAO,MAAM,QAAQ,YAAY,MAAM,IAC5D,OAAM,KAAK,MAAM,IAAI;AAEtB,MAAI,aAAa,SAAS,OAAO,MAAM,YAAY,YAAY,MAAM,QACpE,OAAM,KAAK,MAAM,QAAQ;AAE1B,SAAO,MAAM,KAAK,IAAI;;AAGvB,QAAO;;;;;;;;;;;;;;;;;;;;;;;AAwBR,SAAgB,iBAAiB,QAAiE;AACjG,KAAI,CAAC,OACJ,QAAO;CAIR,IAAI;AACJ,KAAI,OAAO,WAAW,SACrB,KAAI;AACH,iBAAe,KAAK,MAAM,OAAO;SAC1B;AAEP,SAAO;;KAGR,gBAAe;AAGhB,KAAI,CAAC,MAAM,QAAQ,aAAa,CAC/B,QAAO;AAqBR,QADiB,CANI,YAPC,aAAa,KAAK,MAAM;EAC7C,MAAM,MAAmD,EAAE,OAAO,EAAE,OAAO;AAC3E,OAAK,MAAM,CAAC,KAAK,QAAQ,OAAO,QAAQ,EAAE,CACzC,KAAI,OAAO;AAEZ,SAAO;GACN,CAC6C,EAMf,GAHZ,aAAa,IAAI,uBAAuB,CAAC,QAAQ,SAAS,KAAK,SAAS,EAAE,CAG/C,CAAC,QAAQ,MAAM,EAAE,SAAS,EAAE,CAC3D,KAAK,KAAK;;;;;;;;;;;AAY3B,SAAgB,wBACf,OACA,QACyB;CACzB,MAAM,SAAiC,EAAE;AAEzC,MAAK,MAAM,SAAS,QAAQ;EAC3B,MAAM,QAAQ,MAAM;AAEpB,MAAI,UAAU,QAAQ,UAAU,QAAW;AAC1C,UAAO,SAAS;AAChB;;AAGD,MAAI,OAAO,UAAU,SAEpB,KAAI,MAAM,WAAW,IAAI,CACxB,QAAO,SAAS,iBAAiB,MAAM;MAEvC,QAAO,SAAS;WAEP,MAAM,QAAQ,MAAM,CAE9B,KAAI,oBAAoB,MAAM,CAC7B,QAAO,SAAS,iBAAiB,MAAM;MAEvC,QAAO,SAAS,KAAK,UAAU,MAAM;WAE5B,OAAO,UAAU,SAE3B,QAAO,SAAS,KAAK,UAAU,MAAM;WAC3B,OAAO,UAAU,YAAY,OAAO,UAAU,UACxD,QAAO,SAAS,GAAG;MAEnB,QAAO,SAAS;;AAIlB,QAAO"}