Ensure library.xml is setup

README.md

@@ -87,6 +87,7 @@ await server.stop(); // Kill server
 ```
 
 #### List Local ZIMs
+
 ```js
 const local = await server.list();
 [

package.json

@@ -1,6 +1,6 @@
 {
 	"name": "@ztimson/zim-utils",
-	"version": "0.3.0",
+	"version": "0.3.5",
 	"description": "Native, dependency-light ZIM archive reader/searcher and Kiwix catalog downloader for Node.js",
 	"author": "Zak Timson",
 	"license": "MIT",

src/manager.js

@@ -10,24 +10,26 @@ import {zimCatalog, zimCatalogInfo, CATALOG_URL} from './catalog.js';
 export class ZimManager {
 	#catalogUrl;
 	#dir;
-	server;
+	#server;
 	#ownsServer;
 
-	/** @param {{catalog?: string, server?: KiwixServer, port?: number, host?: string, binDir?: string}} [opts]
-	 * server: reuse an existing KiwixServer; if omitted, one is created (not yet started) and exposed via `.server` for reuse elsewhere. */
-	constructor(dir, {catalog = CATALOG_URL, server, port, host, binDir} = {}) {
+	/** @param {{catalog?: string, server?: KiwixServer, port?: number, host?: string, binDir?: string, url?: string}} [opts]
+	 * server: reuse an existing KiwixServer instance directly.
+	 * url: attach to an already-running kiwix-serve by address instead (e.g. started elsewhere in your codebase
+	 * with no shared reference) - mutually exclusive with `server`, ignored if `server` is given. */
+	constructor(dir, {catalog = CATALOG_URL, server, port, host, binDir, url} = {}) {
 		this.#catalogUrl = catalog;
 		this.#dir = dir;
 		this.#ownsServer = !server;
-		this.server = server ?? new KiwixServer(dir, {port, host, binDir});
+		this.#server = server ?? new KiwixServer(dir, {port, host, binDir, url});
 	}
 
 	/** The KiwixServer backing this manager - reuse it directly for content/search access, or pass into another ZimManager. */
-	get server() { return this.server; }
+	get server() { return this.#server; }
 
 	async #ensureServer() {
-		if (!this.server.running) await this.server.start();
-		return this.server;
+		if (!this.#server.running) await this.#server.start();
+		return this.#server;
 	}
 
 	async #download(url, destPath) {
@@ -72,14 +74,15 @@ export class ZimManager {
 
 	/** Stops the internally-owned KiwixServer, if this manager created its own (no-op if one was passed in). */
 	async close() {
-		if (this.#ownsServer) await this.server.stop();
+		if (this.#ownsServer) await this.#server.stop();
 	}
 
 	async delete(nameOrFile) {
 		const local = await this.list();
-		const match = local.find(l => l.name === nameOrFile || l.file === nameOrFile);
+		const match = local.find(l => l.name === nameOrFile || l.href === nameOrFile);
 		if (!match) throw new Error(`ZIM not found locally: ${nameOrFile}`);
-		await fs.promises.rm(path.join(this.#dir, match.file), {force: true});
+		const file = match.href + (match.href.endsWith('.zim') ? '' : '.zim');
+		await fs.promises.rm(path.join(this.#dir, file), {force: true});
 		const server = await this.#ensureServer();
 		await server.reload();
 		return {file: match.file, status: 'deleted'};
@@ -87,9 +90,9 @@ export class ZimManager {
 
 	/** Downloads/updates a single ZIM by direct href, matching against any existing local copy by name. */
 	async download(href, {force = false} = {}) {
-		await fs.promises.mkdir(this.#dir, {recursive: true});
 		const {url: finalUrl} = await this.#resolveUrl(href);
 		const filename = path.basename(new URL(finalUrl).pathname).replace(/\.meta4$/i, '');
+		await fs.promises.mkdir(this.#dir, {recursive: true});
 		const name = filename.replace(/\.zim$/i, '').replace(/_\d{4}-\d{2}(?:_\d+)?$/, '');
 
 		const local = await this.list();

src/server.js

@@ -6,8 +6,7 @@ import net from 'node:net';
 import fs from 'node:fs';
 import path from 'node:path';
 import {fileURLToPath} from 'node:url';
-import {decodeHtml, fromXml} from '@ztimson/utils';
-import {fuzzyMatch, weightedScore} from './utils.js';
+import {fromXml} from '@ztimson/utils';
 
 const execFileAsync = promisify(execFile);
 const __dirname = path.dirname(fileURLToPath(import.meta.url));
@@ -28,38 +27,64 @@ function findFreePort() {
 
 /** Owns a kiwix-serve process's full lifecycle: library.xml, start/stop/reload, content + search access. */
 export class KiwixServer {
+	static #empty = '<?xml version="1.0" encoding="UTF-8" ?>\n<library version="20110515"></library>\n';
+
 	#dir;
 	#host;
 	#port;
 	#binDir;
 	#libraryPath;
 	#child = null;
+	#remote; // baseUrl string if attached to an externally-managed kiwix-serve, else null
 
 	get port() { return this.#port; }
-	get running() { return !!this.#child; }
-	get baseUrl() { return this.#child ? `http://${this.#host}:${this.#port}` : null; }
+	get running() { return !!this.#remote || !!this.#child; }
+	get baseUrl() { return this.#remote || (this.#child ? `http://${this.#host}:${this.#port}` : null); }
 
-	/** @param {{port?: number, host?: string, binDir?: string}} [opts] port defaults to an auto-picked free port; host defaults to localhost-only; binDir defaults to the bundled ./bin next to this package. */
-	constructor(dir, {port, host = '127.0.0.1', binDir = DEFAULT_BIN_DIR} = {}) {
+	/** @param {{port?: number, host?: string, binDir?: string, url?: string}} [opts]
+	 * url: attach to an already-running kiwix-serve (e.g. one started elsewhere in your codebase) instead of
+	 * spawning/owning one - start/stop/reload become no-ops, and library.xml is read over HTTP instead of disk. */
+	constructor(dir, {port, host = '127.0.0.1', binDir = DEFAULT_BIN_DIR, url} = {}) {
 		this.#dir = dir;
 		this.#host = host;
 		this.#port = port;
 		this.#binDir = binDir;
 		this.#libraryPath = path.join(dir, 'library.xml');
+		this.#remote = url ? url.replace(/\/$/, '') : null;
+		if (!this.#remote) this.#ensureLocalStore();
+	}
+
+	#ensureLocalStore() {
+		fs.mkdirSync(this.#dir, {recursive: true});
+		if (!fs.existsSync(this.#libraryPath)) fs.writeFileSync(this.#libraryPath, KiwixServer.#empty);
 	}
 
 	#assertRunning() {
-		if (!this.#child) throw new Error('KiwixServer is not running - call start() first');
+		if (!this.running) throw new Error('KiwixServer is not running - call start() first');
 	}
 
 	#bin(name) {
 		return path.join(this.#binDir, process.platform === 'win32' ? `${name}.exe` : name);
 	}
 
-	/** Rebuilds library.xml from scratch by scanning `dir` for .zim files - a full rebuild is simpler and less bug-prone than tracking incremental add/remove. */
+	/** Reads library.xml from disk if we own the server, or over HTTP if attached to a remote one. */
+	async #fetchLibraryXml() {
+		if (!this.#remote) return fs.promises.readFile(this.#libraryPath, 'utf8').catch(() => '');
+		try {
+			const res = await fetch(`${this.#remote}/library.xml`);
+			return res.ok ? await res.text() : '';
+		} catch {
+			return '';
+		}
+	}
+
+	/** Rebuilds library.xml from scratch by scanning `dir` for .zim files - no-op if attached to a remote server. */
 	async #rebuildLibrary() {
+		if (this.#remote) return;
 		await fs.promises.rm(this.#libraryPath, {force: true});
-		for (const f of await this.#zimFiles()) await execFileAsync(this.#bin('kiwix-manage'), [this.#libraryPath, 'add', path.join(this.#dir, f)]);
+		const files = await this.#zimFiles();
+		if (!files.length) return fs.promises.writeFile(this.#libraryPath, KiwixServer.#empty);
+		for (const f of files) await execFileAsync(this.#bin('kiwix-manage'), [this.#libraryPath, 'add', path.join(this.#dir, f)]);
 	}
 
 	async #waitUntilReady() {
@@ -78,15 +103,9 @@ export class KiwixServer {
 		return (await fs.promises.readdir(this.#dir).catch(() => [])).filter(f => f.endsWith('.zim'));
 	}
 
-	/** URL for a single asset, e.g. for pulling one html/image file when you already know the path (from search). */
-	contentUrl(book, pathInZim = '') {
-		this.#assertRunning();
-		return `${this.baseUrl}/content/${book}/${pathInZim}`;
-	}
-
 	/** Rebuilds library.xml and starts kiwix-serve. Resolves once the server is responding. */
 	async start() {
-		if (this.#child) return;
+		if (this.#remote || this.#child) return;
 		await fs.promises.mkdir(this.#dir, {recursive: true});
 		await this.#rebuildLibrary();
 		this.#port ??= await findFreePort();
@@ -102,9 +121,9 @@ export class KiwixServer {
 		}
 	}
 
-	/** Gracefully stops kiwix-serve, if running. */
+	/** Gracefully stops kiwix-serve, if we own it. No-op if attached to a remote instance. */
 	async stop() {
-		if (!this.#child) return;
+		if (this.#remote || !this.#child) return;
 		const child = this.#child;
 		await new Promise(resolve => {
 			child.once('exit', resolve);
@@ -118,16 +137,18 @@ export class KiwixServer {
 		await this.start();
 	}
 
-	/** Rebuilds library.xml from disk and hot-reloads kiwix-serve via SIGHUP - no downtime/restart needed. */
+	/** Rebuilds library.xml from disk and restarts kiwix-serve. No-op if attached to a remote instance -
+	 * whoever owns that process is responsible for reloading it. */
 	async reload() {
-		if(!this.#child) return;
+		if (this.#remote || !this.#child) return;
 		await this.restart();
 	}
 
-	/** Local catalog listing - same flat shape as the online catalog (catalog.js), plus a `file` field since these already live on disk. */
+	/** Local catalog listing - same flat shape as the online catalog (catalog.js), plus a `file` field. */
 	async list() {
 		this.#assertRunning();
-		const xml = await fs.promises.readFile(this.#libraryPath, 'utf8').catch(() => '');
+		const xml = await this.#fetchLibraryXml();
+		if (!xml) return [];
 		const entries = fromXml(xml);
 		return (entries?.library?.book || []).map(e => {
 			const tags = e.tags.split(';');
@@ -147,7 +168,7 @@ export class KiwixServer {
 				articleCount: +e.articleCount || 0,
 				sizeMb: +(Number(e.size) / 1024).toFixed(1) || 0,
 				href: name,
-				icon: e.favicon,
+				icon: `data:${e.faviconMimetype || 'image/png'};base64,${e.favicon}`,
 				viewer: `${this.baseUrl}/content/${name}`,
 			};
 		});
@@ -169,35 +190,28 @@ export class KiwixServer {
 
 	/** Fetches a single asset's raw bytes straight from kiwix-serve. */
 	async raw(href) {
-		const res = await fetch(this.fetch(href));
-		if (!res.ok) return null;
+		const res = await fetch(this.link(href));
+		if(!res.ok) return null;
 		return {mimetype: res.headers.get('content-type'), data: Buffer.from(await res.arrayBuffer())};
 	}
 
-	/**
-	 * Two-pass fulltext search across every local ZIM: xapian prefilter (kiwix-serve's
-	 * own index), then fuzzy-reranked by title so the strongest matches surface first.
-	 * Returns a flat array matching the catalog/list shape: {id, title, name, category, ..., href, score}
-	 */
+	/** Fulltext search across every local ZIM via kiwix-serve's own xapian index */
 	async search(terms, limit = 20) {
 		this.#assertRunning();
-		const termList = String(terms).split(',').map(t => t.trim()).filter(Boolean);
+		const termList = String(terms).split(/[,\s]+/).map(t => t.trim()).filter(Boolean);
 		if (!termList.length) return [];
 
 		const params = new URLSearchParams({pattern: termList.join(' '), format: 'xml', pageLength: String(limit)});
 		const res = await fetch(`${this.baseUrl}/search?${params}`);
 		if (!res.ok) return [];
 
-		const xml = await res.text();
-		let found = fromXml(xml);
-		found = found?.rss?.channel?.item || [];
-
+		const found = fromXml(await res.text())?.rss?.channel?.item || [];
 		const books = await this.list();
 		const bookMap = new Map(books.map(b => [b.title, b]));
 
-		const enriched = found.map(hit => {
+		return found.map(hit => {
 			const book = bookMap.get(hit.book.title);
-			if(!book) return null;
+			if (!book) return null;
 			const prefix = `/content/${book.href}/`;
 			const page = hit.link.startsWith(prefix) ? hit.link.slice(prefix.length) : hit.link.replace(/^\/+/, '');
 			return {
@@ -210,10 +224,8 @@ export class KiwixServer {
 				icon: book.icon,
 				viewer: this.baseUrl + hit.link,
 				summary: hit.description,
-				score: weightedScore(hit.title, termList) + weightedScore(hit.description, termList),
+				score: +hit.score || 0,
 			};
-		}).filter(hit => !!hit && hit.score > 0);
-
-		return enriched.toSorted((a, b) => b.score - a.score).slice(0, limit);
+		}).filter(Boolean);
 	}
 }

src/utils.js

@@ -40,10 +40,3 @@ export function fuzzyMatch(target, ...terms) {
 		similarities,
 	};
 }
-
-export function weightedScore(text, termList) {
-	if (!text) return 0;
-	const covered = termList.reduce((sum, t) => sum + t.length, 0);
-	const coverage = Math.min(1, covered / text.length);
-	return fuzzyMatch(text, ...termList).max * coverage;
-}