Add a paged lyrics quality queue

server/admin-tools.js

@@ -0,0 +1,54 @@
+import { audit } from './audit-log.js';
+
+const clientOf = db => typeof db?.execute === 'function' ? db : db?.db;
+const parse = value => { try { return JSON.parse(value || '{}'); } catch { return {}; } };
+const number = value => Number(value) || 0;
+
+export function registerAdminToolsRoutes(app, { db, adminAuth, media = null, backupDir = null }) {
+  const sql = clientOf(db);
+
+  app.get('/api/admin/lyrics/quality', adminAuth, async c => {
+    const minLines = Math.max(1, Math.min(200, Number(c.req.query('minLines')) || 8));
+    const offset = Math.max(0, Math.min(1_000_000, Number(c.req.query('offset')) || 0));
+    const limit = Math.max(1, Math.min(100, Number(c.req.query('limit')) || 50));
+    const category = ['all', 'missing', 'short', 'automatic', 'untimed', 'reports'].includes(c.req.query('category')) ? c.req.query('category') : 'all';
+    const conditions = {
+      missing: 'n.video_id IS NULL',
+      short: 'n.video_id IS NOT NULL AND COALESCE(json_array_length(json_extract(n.data,\'$.lines\')),0) < ?',
+      automatic: "n.source='auto'",
+      untimed: "EXISTS (SELECT 1 FROM json_each(n.data,'$.lines') line WHERE json_extract(line.value,'$.t') IS NULL)",
+      reports: "EXISTS (SELECT 1 FROM lyric_flags f WHERE f.video_id=m.id AND f.status='open')",
+    };
+    const parts = category === 'all'
+      ? [conditions.missing, conditions.short, conditions.automatic, conditions.untimed, conditions.reports]
+      : [conditions[category]];
+    const where = `WHERE (${parts.join(' OR ')})`;
+    const args = category === 'short' ? [minLines] : category === 'all' ? [minLines] : [];
+    const query = `SELECT m.id,m.card,m.updated_at,n.data AS lyrics,n.source AS lyricSource,n.updated_at AS lyricUpdated,
+      COALESCE(json_array_length(json_extract(n.data,'$.lines')),0) AS lineCount,
+      EXISTS(SELECT 1 FROM json_each(n.data,'$.lines') line WHERE json_extract(line.value,'$.t') IS NULL) AS untimed,
+      (SELECT COUNT(*) FROM lyric_flags f WHERE f.video_id=m.id AND f.status='open') AS openReports
+      FROM video_meta m LEFT JOIN video_notes n ON n.video_id=m.id AND n.kind='lyrics'
+      ${where} ORDER BY openReports DESC,lyricUpdated ASC,m.updated_at DESC LIMIT ? OFFSET ?`;
+    const rows = await sql.execute({ sql: query, args: [...args, limit, offset] });
+    const count = await sql.execute({ sql: `SELECT COUNT(*) AS total FROM video_meta m LEFT JOIN video_notes n ON n.video_id=m.id AND n.kind='lyrics' ${where}`, args });
+    const songs = rows.rows.map(row => {
+      const card = parse(row.card);
+      let untimed = !!row.untimed;
+      let lines = number(row.lineCount);
+      if (row.lyrics) {
+        const data = parse(row.lyrics); lines ||= Array.isArray(data.lines) ? data.lines.length : 0;
+        if (Array.isArray(data.lines)) untimed ||= data.lines.some(line => line.t == null);
+      }
+      const reasons = [];
+      if (!row.lyrics) reasons.push('missing');
+      else if (lines < minLines) reasons.push('short');
+      if (row.lyricSource === 'auto') reasons.push('automatic');
+      if (untimed) reasons.push('untimed');
+      if (number(row.openReports)) reasons.push('reports');
+      return { id: row.id, title: card.title || row.id, channel: card.channel || '', lines, source: row.lyricSource || null, untimed, openReports: number(row.openReports), reasons, updatedAt: number(row.lyricUpdated) };
+    });
+    return c.json({ ok: true, minLines, category, offset, limit, total: number(count.rows[0]?.total), songs }, 200, { 'Cache-Control': 'no-store' });
+  });
+
+}