Support wildcards in the title search

media-scrape/src/store.ts

@@ -187,6 +187,20 @@ export function normalizeTitleKey(title: string): string {
         .replace(/\s+/g, " ");
 }
 
+/** Lowercase a title query and collapse punctuation, but keep `*` as a wildcard marker. */
+export function normalizeTitleQuery(query: string): string {
+    return query
+        .toLowerCase()
+        .replace(/[^a-z0-9*]+/g, " ")
+        .trim()
+        .replace(/\s+/g, " ");
+}
+
+/** Escape LIKE metacharacters, then turn `*` into `%`. */
+function titleLikePattern(value: string): string {
+    return value.replace(/[\\%_]/g, match => `\\${match}`).replace(/\*/g, "%");
+}
+
 const SCHEMA = `
 CREATE TABLE IF NOT EXISTS messages (
     id TEXT PRIMARY KEY,
@@ -803,14 +817,22 @@ export class ScrapeStore {
             values.push(options.lengthGreaterThan);
         }
         if (options.titleQuery?.trim()) {
-            const query = normalizeTitleKey(options.titleQuery);
+            const query = normalizeTitleQuery(options.titleQuery);
             if (options.titleExact) {
-                conditions.push("title_key LIKE ?");
-                values.push(`%${query}%`);
+                // One phrase, substring of the title; `*` still acts as a wildcard.
+                conditions.push("title_key LIKE ? ESCAPE '\\'");
+                values.push(`%${titleLikePattern(query)}%`);
             } else {
                 for (const token of query.split(" ").filter(Boolean)) {
-                    conditions.push("(' ' || COALESCE(title_key, '')) LIKE ?");
-                    values.push(`% ${token}%`);
+                    const pattern = titleLikePattern(token);
+                    if (token.startsWith("*")) {
+                        // Leading `*`: anywhere in a word, not just the start.
+                        conditions.push("COALESCE(title_key, '') LIKE ? ESCAPE '\\'");
+                        values.push(`%${pattern}%`);
+                    } else {
+                        conditions.push("(' ' || COALESCE(title_key, '')) LIKE ? ESCAPE '\\'");
+                        values.push(`% ${pattern}%`);
+                    }
                 }
             }
         }

media-scrape/test/run.ts

@@ -31,6 +31,7 @@ import {
     dirSizeBytes,
     listScrapes,
     normalizeTitleKey,
+    normalizeTitleQuery,
     readScrapeMeta,
     ScrapeStore,
     writeScrapeMeta
@@ -1619,6 +1620,56 @@ void (async () => {
         store.close();
     });
 
+    await test("title wildcards match anywhere and stay literal for % and _", () => {
+        const dir = join(root, "title-wildcards");
+        mkdirSync(dir, { recursive: true });
+        const store = new ScrapeStore(dir);
+        const embed = (id: string, title: string) => ({
+            type: "video",
+            url: `https://www.youtube.com/watch?v=${id}`,
+            title,
+            provider: "YouTube",
+            videoUrl: `https://www.youtube.com/watch?v=${id}`,
+            imageUrl: null,
+            thumbnailUrl: null
+        });
+
+        store.insertPage([
+            message("1300", {
+                embeds: [
+                    embed("w1", "Scar Tissue"),
+                    embed("w2", "Discard the rest"),
+                    embed("w3", "Car wash"),
+                    embed("w4", "Cartoon physics"),
+                    embed("w5", "Carly and the r")
+                ]
+            })
+        ]);
+
+        const ids = (options: Record<string, unknown>) =>
+            store
+                .queryMedia({ limit: 50, ...options })
+                .map(row => row.url.split("v=")[1])
+                .sort();
+
+        // Word-prefix default: car starts a word.
+        assert.deepEqual(ids({ titleQuery: "car" }), ["w3", "w4", "w5"]);
+        // Leading *: anywhere, so scar and discard join in.
+        assert.deepEqual(ids({ titleQuery: "*car" }), ["w1", "w2", "w3", "w4", "w5"]);
+        // Trailing * behaves like the prefix form.
+        assert.deepEqual(ids({ titleQuery: "car*" }), ["w3", "w4", "w5"]);
+        // In-word wildcard after a word start.
+        assert.deepEqual(ids({ titleQuery: "c*r" }), ["w3", "w4", "w5"]);
+        assert.deepEqual(ids({ titleQuery: "c*r*" }), ["w3", "w4", "w5"]);
+        assert.deepEqual(ids({ titleQuery: "*c*r*" }), ["w1", "w2", "w3", "w4", "w5"]);
+        store.close();
+    });
+
+    await test("normalizeTitleQuery keeps stars and drops punctuation", () => {
+        assert.equal(normalizeTitleQuery("  *CaR, C*r*  "), "*car c*r*");
+        assert.equal(normalizeTitleQuery("Boat"), "boat");
+    });
+
     await test("normalizeTitleKey collapses punctuation and case", () => {
         assert.equal(normalizeTitleKey("Woman's Day: 3 Things!"), "woman s day 3 things");
         assert.equal(normalizeTitleKey("  A---B  "), "a b");