tools/netease_music/src/getInfo/songInfoUtils.old.js

const fs = require('fs');
const path = require('path');

const requestUtils = require('../../../utils/requestUtils');
const sleepUtils = require('../../../utils/sleepUtils');

const dbUtils = global.dbUtils;

// 从数据库中查询
async function getFromDatabase({ songId }) {
    // 查询出专辑
    let infoResultSet = await dbUtils.query('SELECT * FROM song WHERE song_id = ?', [songId]);
    if (infoResultSet.length == 0) return {};

    // 查出专辑与歌曲对应关系
    let albumRelationResultSet = await dbUtils.query('SELECT * FROM song_album_relation WHERE song_id = ?', [songId]);
    let artistRelationResultSet = await dbUtils.query('SELECT * FROM song_artist_relation WHERE song_id = ?', [songId]);

    // 拼装
    let songInfo = JSON.parse(JSON.stringify(infoResultSet[0]));
    songInfo.albumIds = albumRelationResultSet.map(album => album.album_id);
    songInfo.artistIds = artistRelationResultSet.map(artist => artist.artist_id);
    return songInfo;
}

// 从数据库中查出还缺少的歌曲，并进行爬取
async function fetchAll({ args = {} }) {
    console.log("start fetching songs ...");
    let whereClause = [
        args.min ? `song_id >  ${args.min}` : '1=1',
        args.max ? `song_id <= ${args.max}` : '1=1',
    ].join(' AND ');
    var sql = `
        SELECT DISTINCT song_id FROM song_artist_relation WHERE ${whereClause} AND song_id NOT IN ( SELECT song_id FROM song )
        UNION
        SELECT DISTINCT song_id FROM song_album_relation  WHERE ${whereClause} AND song_id NOT IN ( SELECT song_id FROM song )
        ${args.order ? `ORDER BY song_id ${args.order}` : ''}
        ${args.limit ? `LIMIT ${args.limit}` : ''}
    `;
    console.log(sql);

    var songIds = await dbUtils.query(sql, []);
    songIds = songIds.map(item => item.song_id);
    for (let i = 0; i < songIds.length; i++) {
        await global.checkIsExit();
        const songId = songIds[i];
        console.log(`${i + 1}/${songIds.length} | song: ${songId} | ${args.min || "?"}-${args.max || "?"}`);
        try {
            await fetch({ songId: songId });
        } catch (err) {
            console.error(err);
        }
        await sleepUtils.sleep(global.sleepTime);
    }
}

// 获取音乐详情
async function fetch({ songId, debug = false }) {
    let result = await dbUtils.query('SELECT count(*) as count FROM song WHERE song_id = ?', [songId]);
    if (result[0].count > 0 && !debug) {
        console.log(`数据库中已有数据，跳过 songId: ${songId}`);
        return;
    }

    let url = `https://music.163.com/song?id=${songId}`;
    try {
        // var html = fs.readFileSync(path.join(__dirname, "../../temp", `song-${songId}.html`), 'utf8');
        var html = await requestUtils.getApiResult(url);
        // fs.writeFileSync(path.join(__dirname, "../../temp", `song-${songId}.html`), html);
    } catch (errors) {
        console.error(errors);
        return;
    }

    if (html.includes(`<p class="note s-fc3">很抱歉，你要查找的网页找不到</p>`)) {
        let deleteResult1 = await dbUtils.query('DELETE FROM song_album_relation WHERE song_id = ?', [songId]);
        let deleteResult2 = await dbUtils.query('DELETE FROM song_artist_relation WHERE song_id = ?', [songId]);
        console.log(`song: ${songId} 不存在，从song_album_relation, song_artist_relation表中删除. affectedRows: ${deleteResult1.affectedRows}, ${deleteResult2.affectedRows}`);
        return;
    }

    // 正则匹配
    let regExResult = /\<script type\=\"application\/ld\+json\"\>([\S\s]*?)\<\/script\>/.exec(html);
    let songInfoJSONString = regExResult[1];
    let songInfoDict = JSON.parse(songInfoJSONString);
    // console.log(songInfoDict);

    // TODO 考虑歌曲别名 例如：https://music.163.com/#/song?id=26830207

    let title = /<meta property="og:title" content="(.*?)" \/>/.exec(html)[1];
    let image = /<meta property="og:image" content="http:\/\/p.\.music\.126\.net\/(.*?)" \/>/.exec(html)[1];
    let artist = /<meta property="og:music:artist" content="(.*?)" \/>/.exec(html)[1];
    let duration = /<meta property="music:duration" content="(.*?)"\/>/.exec(html)[1];
    try {
        var album = /<meta property="og:music:album" content="(.*?)"\/>/.exec(html)[1];
        var albumId = /<meta property="music:album" content="https:\/\/music\.163\.com\/album\?id=(.*?)"\/>/.exec(html)[1];
    } catch (err) {
        // 歌曲不在专辑中
    }

    const reg = /<meta property="music:musician" content="https:\/\/music\.163\.com\/artist\?id=(.*?)"\/>/g;
    let artistIds = [];
    let matched = null;
    while ((matched = reg.exec(html)) !== null) {
        artistIds.push(matched[1]);
    }

    let songInfo = {
        songId: songId,
        title: title,
        image: image,
        pubDate: songInfoDict.pubDate,
        artist: artist,
        artistIds: artistIds,
        album: album || null,
        albumId: albumId || null,
        duration: duration,
    };
    // console.log("songInfo", songInfo);
    if (albumId != null)
        dbUtils.query('INSERT IGNORE INTO song_album_relation SET ?', {
            song_id: songInfo.songId,
            album_id: songInfo.albumId,
        });
    artistIds.forEach(function (artistId) {
        dbUtils.query('INSERT IGNORE INTO song_artist_relation SET ?', {
            song_id: songInfo.songId,
            artist_id: artistId,
        });
    });
    dbUtils.query('INSERT IGNORE INTO song SET ?', {
        song_id: songInfo.songId,
        title: songInfo.title,
        image: songInfo.image,
        pub_date: songInfo.pubDate,
    });
    return songInfo;
}

module.exports = {
    getFromDatabase: getFromDatabase,
    fetch: fetch,
    fetchAll: fetchAll,
}