Better searching

src/reader.js

@@ -3,7 +3,7 @@
 import fs from 'node:fs';
 import {decompress as lzmaDecompress} from 'lzma1';
 import {ZstdCodec} from 'zstd-codec';
-import {fuzzyMatch} from './utils.js';
+import {fuzzyMatch, titleFromUrl} from './utils.js';
 
 const HEADER_SIZE = 80;
 const NS_CONTENT = 'C';
@@ -219,7 +219,7 @@ export class ZimReader {
 		const termList = String(terms).split(',').map(t => t.trim()).filter(Boolean);
 		if (!termList.length) return [];
 
-		const candidates = this.#hasTitleListing
+		const candidates = (this.#hasTitleListing && this.#header.articleCount > 5000)
 			? await this.#titleIndexCandidates(termList[0])
 			: await this.#allDirents();
 
@@ -227,9 +227,16 @@ export class ZimReader {
 		for (const dirent of candidates) {
 			if (dirent.namespace !== NS_CONTENT) continue;
 			if (htmlOnly && !(this.#mimeTypes[dirent.mimetype] || '').startsWith('text/html')) continue;
-			const {max} = fuzzyMatch(dirent.title, ...termList);
-			scored.push({url: dirent.url, title: dirent.title, namespace: dirent.namespace, score: max});
+			const urlTitle = titleFromUrl(dirent.url);
+			const titleScore = fuzzyMatch(dirent.title, ...termList).max;
+			const urlScore = fuzzyMatch(urlTitle, ...termList).max;
+			scored.push({
+				url: dirent.url,
+				title: dirent.title.length > urlTitle.length ? dirent.title : urlTitle,
+				namespace: dirent.namespace,
+				score: Math.max(titleScore, urlScore)
+			});
 		}
-		return scored.toSorted((a, b) => b.score - a.score).slice(0, limit);
+		return scored.filter(a => a.score > 0).toSorted((a, b) => b.score - a.score).slice(0, limit);
 	}
 }