add: insert labelled songs into songs table

lib/db/allData.ts

@@ -8,7 +8,9 @@ export async function videoExistsInAllData(client: Client, aid: number) {
 }
 
 export async function userExistsInBiliUsers(client: Client, uid: number) {
-	return await client.queryObject<{ exists: boolean }>(`SELECT EXISTS(SELECT 1 FROM bili_user WHERE uid = $1)`, [uid])
+	return await client.queryObject<{ exists: boolean }>(`SELECT EXISTS(SELECT 1 FROM bili_user WHERE uid = $1)`, [
+		uid,
+	]);
 }
 
 export async function getUnlabelledVideos(client: Client) {
@@ -36,16 +38,17 @@ export async function getVideoInfoFromAllData(client: Client, aid: number) {
 		const q = await client.queryObject<BiliUserType>(
 			`SELECT * FROM bili_user WHERE uid = $1`,
 			[row.uid],
-		)
+		);
 		const userRow = q.rows[0];
-		if (userRow)
+		if (userRow) {
 			authorInfo = userRow.desc;
 		}
+	}
 	return {
 		title: row.title,
 		description: row.description,
 		tags: row.tags,
-		author_info: authorInfo
+		author_info: authorInfo,
 	};
 }
 
@@ -57,7 +60,7 @@ export async function getUnArchivedBiliUsers(client: Client) {
 		LEFT JOIN bili_user bu ON ad.uid = bu.uid
 		WHERE bu.uid IS NULL;
 		`,
-		[]
+		[],
 	);
 	const rows = queryResult.rows;
 	return rows.map((row) => row.uid);

lib/db/songs.ts

@@ -0,0 +1,29 @@
+import { Client } from "https://deno.land/x/postgres@v0.19.3/mod.ts";
+
+export async function getNotCollectedSongs(client: Client) {
+	const queryResult = await client.queryObject<{ aid: number }>(`
+		SELECT lr.aid
+		FROM labelling_result lr
+		WHERE lr.label != 0
+		AND NOT EXISTS (
+			SELECT 1
+			FROM songs s
+			WHERE s.aid = lr.aid
+		);
+	`);
+	return queryResult.rows.map((row) => row.aid);
+}
+
+export async function aidExistsInSongs(client: Client, aid: number) {
+	const queryResult = await client.queryObject<{ exists: boolean }>(
+		`
+	    SELECT EXISTS (
+	        SELECT 1
+	        FROM songs
+	        WHERE aid = $1
+	    );
+	`,
+		[aid],
+	);
+	return queryResult.rows[0].exists;
+}

lib/ml/benchmark.ts

@@ -70,13 +70,13 @@ function calculateMetrics(labels: number[], predictions: number[], elapsedTime:
 	speed: string;
 } {
 	// 输出label和prediction不一样的index列表
-    const arr = []
+	const arr = [];
 	for (let i = 0; i < labels.length; i++) {
 		if (labels[i] !== predictions[i] && predictions[i] == 0) {
-            arr.push([i + 1, labels[i], predictions[i]])
+			arr.push([i + 1, labels[i], predictions[i]]);
 		}
 	}
-    console.log(arr)
+	console.log(arr);
 	// 初始化混淆矩阵
 	const classCount = Math.max(...labels, ...predictions) + 1;
 	const matrix = Array.from({ length: classCount }, () => Array.from({ length: classCount }, () => 0));
@@ -146,7 +146,7 @@ async function evaluateModel(session: ort.InferenceSession): Promise<{
 			const embeddings = await getONNXEmbeddings([
 				sample.title,
 				sample.description,
-                sample.tags.join(",")
+				sample.tags.join(","),
 			], session);
 
 			const probabilities = await runClassification(embeddings);

lib/mq/exec/classifyVideo.ts

@@ -5,6 +5,8 @@ import { classifyVideo } from "lib/ml/filter_inference.ts";
 import { ClassifyVideoQueue } from "lib/mq/index.ts";
 import logger from "lib/log/logger.ts";
 import { lockManager } from "lib/mq/lockManager.ts";
+import { aidExistsInSongs } from "lib/db/songs.ts";
+import { insertIntoSongs } from "lib/mq/task/collectSongs.ts";
 
 export const classifyVideoWorker = async (job: Job) => {
 	const client = await db.connect();
@@ -23,6 +25,11 @@ export const classifyVideoWorker = async (job: Job) => {
 	}
 	await insertVideoLabel(client, aid, label);
 
+	const exists = await aidExistsInSongs(client, aid);
+	if (!exists) {
+		await insertIntoSongs(client, aid);
+	}
+
 	client.release();
 
 	await job.updateData({

lib/mq/exec/getLatestVideos.ts

@@ -1,6 +1,8 @@
 import { Job } from "bullmq";
 import { queueLatestVideos } from "lib/mq/task/queueLatestVideo.ts";
 import { db } from "lib/db/init.ts";
+import { insertVideoInfo } from "lib/mq/task/getVideoInfo.ts";
+import { collectSongs } from "lib/mq/task/collectSongs.ts";
 
 export const getLatestVideosWorker = async (_job: Job): Promise<void> => {
 	const client = await db.connect();
@@ -10,3 +12,26 @@ export const getLatestVideosWorker = async (_job: Job): Promise<void> => {
 		client.release();
 	}
 };
+
+export const collectSongsWorker = async (_job: Job): Promise<void> => {
+	const client = await db.connect();
+	try {
+		await collectSongs(client);
+	} finally {
+		client.release();
+	}
+};
+
+export const getVideoInfoWorker = async (job: Job): Promise<number> => {
+	const client = await db.connect();
+	try {
+		const aid = job.data.aid;
+		if (!aid) {
+			return 3;
+		}
+		await insertVideoInfo(client, aid);
+		return 0;
+	} finally {
+		client.release();
+	}
+};

lib/mq/exec/getVideoInfo.ts

@@ -1,17 +0,0 @@
-import { Job } from "bullmq";
-import { db } from "lib/db/init.ts";
-import { insertVideoInfo } from "lib/mq/task/getVideoInfo.ts";
-
-export const getVideoInfoWorker = async (job: Job): Promise<number> => {
-	const client = await db.connect();
-	try {
-		const aid = job.data.aid;
-		if (!aid) {
-			return 3;
-		}
-		await insertVideoInfo(client, aid);
-		return 0;
-	} finally {
-		client.release();
-	}
-};

lib/mq/init.ts

@@ -11,6 +11,10 @@ export async function initMQ() {
 		every: 5 * MINUTE,
 		immediately: true,
 	});
+	await LatestVideosQueue.upsertJobScheduler("collectSongs", {
+		every: 3 * MINUTE,
+		immediately: true,
+	});
 
 	logger.log("Message queue initialized.");
 }

lib/mq/task/collectSongs.ts

@@ -0,0 +1,29 @@
+import { Client } from "https://deno.land/x/postgres@v0.19.3/mod.ts";
+import { aidExistsInSongs, getNotCollectedSongs } from "lib/db/songs.ts";
+import logger from "lib/log/logger.ts";
+
+export async function collectSongs(client: Client) {
+	const aids = await getNotCollectedSongs(client);
+	for (const aid of aids) {
+		const exists = await aidExistsInSongs(client, aid);
+		if (exists) continue;
+		await insertIntoSongs(client, aid);
+		logger.log(`Video ${aid} was added into the songs table.`, "mq", "fn:collectSongs");
+	}
+}
+
+export async function insertIntoSongs(client: Client, aid: number) {
+	await client.queryObject(
+		`
+			INSERT INTO songs (aid, bvid, published_at, duration)
+			VALUES (
+				$1,
+				(SELECT bvid FROM all_data WHERE aid = $1),
+				(SELECT published_at FROM all_data WHERE aid = $1),
+				(SELECT duration FROM all_data WHERE aid = $1)
+			)
+			ON CONFLICT DO NOTHING
+		`,
+		[aid],
+	);
+}

lib/mq/task/getVideoInfo.ts

@@ -34,8 +34,7 @@ export async function insertVideoInfo(client: Client, aid: number) {
 			`INSERT INTO bili_user (uid, username, "desc", fans) VALUES ($1, $2, $3, $4)`,
 			[uid, data.View.owner.name, data.Card.card.sign, data.Card.follower],
 		);
-	}
+	} else {
-	else {
 		await client.queryObject(
 			`UPDATE bili_user SET fans = $1 WHERE uid = $2`,
 			[data.Card.follower, uid],

lib/mq/task/queueLatestVideo.ts

@@ -26,12 +26,13 @@ export async function queueLatestVideos(
 			if (videoExists) {
 				continue;
 			}
-			await LatestVideosQueue.add("getVideoInfo", { aid }, { delay,
+			await LatestVideosQueue.add("getVideoInfo", { aid }, {
+				delay,
 				attempts: 100,
 				backoff: {
 					type: "fixed",
-					delay: SECOND * 5
+					delay: SECOND * 5,
-				}
+				},
 			});
 			videosFound.add(aid);
 			allExists = false;

src/worker.ts

@@ -1,10 +1,10 @@
 import { Job, Worker } from "bullmq";
-import { getLatestVideosWorker } from "lib/mq/executors.ts";
+import { collectSongsWorker, getLatestVideosWorker } from "lib/mq/executors.ts";
 import { redis } from "lib/db/redis.ts";
 import logger from "lib/log/logger.ts";
 import { lockManager } from "lib/mq/lockManager.ts";
 import { WorkerError } from "lib/mq/schema.ts";
-import { getVideoInfoWorker } from "lib/mq/exec/getVideoInfo.ts";
+import { getVideoInfoWorker } from "lib/mq/exec/getLatestVideos.ts";
 
 Deno.addSignalListener("SIGINT", async () => {
 	logger.log("SIGINT Received: Shutting down workers...", "mq");
@@ -28,6 +28,9 @@ const latestVideoWorker = new Worker(
 			case "getVideoInfo":
 				await getVideoInfoWorker(job);
 				break;
+			case "collectSongs":
+				await collectSongsWorker(job);
+				break;
 			default:
 				break;
 		}