完成songInfo的更新
This commit is contained in:
parent
7b866cdaf9
commit
ade6e4f9e9
@ -2,11 +2,27 @@ CREATE DATABASE `neteaseMusic` CHARACTER SET 'utf8mb4' COLLATE 'utf8mb4_general_
|
||||
USE `neteaseMusic`;
|
||||
CREATE TABLE `song` (
|
||||
`song_id` int(10) unsigned NOT NULL COMMENT '歌曲id',
|
||||
`title` varchar(200) NOT NULL COMMENT '歌曲名',
|
||||
`image` varchar(200) NOT NULL COMMENT '封面图 http://p1.music.126.net/ 后面的部分',
|
||||
`pub_date` varchar(100) NOT NULL COMMENT '发布日期',
|
||||
`title` varchar(200)NOT NULL COMMENT '歌曲名',
|
||||
`type` tinyint(4) NOT NULL COMMENT ' 0: 一般类型 1: 通过云盘上传的音乐，网易云不存在公开对应 2: 通过云盘上传的音乐，网易云存在公开对应',
|
||||
`alias` json DEFAULT NULL COMMENT '歌曲别名（JSON格式数组）',
|
||||
`pop` float DEFAULT NULL COMMENT '歌曲热度',
|
||||
`fee` tinyint(4) DEFAULT NULL COMMENT '版权 0: 免费或无版权 1: VIP 歌曲 4: 购买专辑 8: 非会员可免费播放低音质，会员可播放高音质及下载 fee 为 1 或 8 的歌曲均可单独购买 2 元单曲',
|
||||
`quality` varchar(500)NOT NULL COMMENT '高/中/低/无损质量文件信息',
|
||||
`cd` varchar(255) NOT NULL COMMENT 'None或如"04", "1/2", "3", "null"的字符串，表示歌曲属于专辑中第几张CD，对应音频文件的Tag',
|
||||
`no` int(10) unsigned DEFAULT NULL COMMENT '表示歌曲属于CD中第几曲，0表示没有这个字段，对应音频文件的Tag',
|
||||
`dj_id` int(10) unsigned DEFAULT NULL COMMENT '0: 不是DJ节目 其他：是DJ节目，表示DJ ID',
|
||||
`s_id` int(10) unsigned DEFAULT NULL COMMENT '对于t == 2的歌曲，表示匹配到的公开版本歌曲ID',
|
||||
`origin_cover_type` tinyint(4) NOT NULL DEFAULT '0' COMMENT '0: 未知 1: 原曲 2: 翻唱',
|
||||
`image` varchar(200)DEFAULT NULL COMMENT '封面图 http://p1.music.126.net/ 后面的部分',
|
||||
`pub_date` varchar(100)DEFAULT NULL COMMENT '发布日期(弃用)',
|
||||
`pub_time` bigint(20) unsigned DEFAULT NULL COMMENT '发布日期 毫秒为单位的Unix时间戳',
|
||||
`no_copyright_rcmd` varchar(255) DEFAULT NULL COMMENT 'None表示可以播，非空表示无版权',
|
||||
`mv` int(10) unsigned DEFAULT NULL COMMENT '非零表示有MV ID',
|
||||
`single` tinyint(4) DEFAULT NULL COMMENT '0: 有专辑信息或者是DJ节目 1: 未知专辑',
|
||||
`version` int(10) unsigned NOT NULL DEFAULT '1' COMMENT '歌曲版本信息',
|
||||
`create_time` datetime NOT NULL DEFAULT CURRENT_TIMESTAMP COMMENT '爬取时间',
|
||||
`modify_time` datetime NOT NULL DEFAULT CURRENT_TIMESTAMP ON UPDATE CURRENT_TIMESTAMP COMMENT '最后更新时间',
|
||||
`data_version` tinyint(4) NOT NULL DEFAULT '1' COMMENT '数据记录版本(如果有字段调整则整体+1)',
|
||||
PRIMARY KEY (`song_id`),
|
||||
KEY `song_id` (`song_id`)
|
||||
) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_general_ci;
|
||||
|
||||
@ -1,11 +1,12 @@
|
||||
const fs = require('fs');
|
||||
const path = require('path');
|
||||
|
||||
const requestUtils = require('../../../utils/requestUtils');
|
||||
const sleepUtils = require('../../../utils/sleepUtils');
|
||||
|
||||
const dbUtils = global.dbUtils;
|
||||
|
||||
const { song_detail } = require('NeteaseCloudMusicApi');
|
||||
|
||||
// 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 ...");
|
||||
@ -20,6 +21,10 @@ async function fetchAll({ args = {} }) {
|
||||
${args.order ? `ORDER BY song_id ${args.order}` : ''}
|
||||
${args.limit ? `LIMIT ${args.limit}` : ''}
|
||||
`;
|
||||
// 更新现有数据
|
||||
sql = `SELECT song_id FROM song WHERE data_version = 1`;
|
||||
// 测试用
|
||||
// sql = `SELECT song_id FROM song_artist_relation group by song_id limit 10`;
|
||||
console.log(sql);
|
||||
|
||||
var songIds = await dbUtils.query(sql, []);
|
||||
@ -27,7 +32,7 @@ async function fetchAll({ args = {} }) {
|
||||
|
||||
// 0 - 100, 200 - 399, 400 - ..., ... - songIds.length-1
|
||||
// 0 1 2 count-1
|
||||
var step = 270;
|
||||
var step = 1000;
|
||||
var total = songIds.length;
|
||||
var count = Math.ceil(total / step);
|
||||
for (let i = 0; i < count; i++) {
|
||||
@ -45,56 +50,67 @@ async function fetchAll({ args = {} }) {
|
||||
|
||||
// 获取音乐详情
|
||||
async function fetch({ songIdArray, 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: ${songId}`);
|
||||
return;
|
||||
}
|
||||
|
||||
// https://neteasecloudmusicapi-docs.4everland.app/#/?id=%e8%8e%b7%e5%8f%96%e6%ad%8c%e6%9b%b2%e8%af%a6%e6%83%85
|
||||
try {
|
||||
// 每一次大概可以取到270条以上
|
||||
var songResult = await song_detail({
|
||||
ids: ["64956", "64956"].join(','),
|
||||
});
|
||||
fs.writeFileSync(path.join(__dirname, "../../temp", `song-${playlistId}.json`), JSON.stringify(playlistResult));
|
||||
// 每一次大概可以取到1000条以上
|
||||
var songResult = await song_detail({ ids: songIdArray.join(',') });
|
||||
// fs.writeFileSync(path.join(__dirname, "../../temp", `song-${songIdArray[0]}-${songIdArray[songIdArray.length - 1]}.json`), JSON.stringify(songResult));
|
||||
} catch (errors) {
|
||||
console.error(errors);
|
||||
return;
|
||||
}
|
||||
// console.log(songResult.body.songs.map(item => JSON.stringify(item)));
|
||||
|
||||
// console.log(playlistResult);
|
||||
let songAlbumRel = [], songArtistRel = [];
|
||||
let songInfoList = songResult.body.songs.map(song => {
|
||||
song.ar.forEach(item => songArtistRel.push([song.id, item.id]));
|
||||
songAlbumRel.push([song.id, song.al.id || 0])
|
||||
return {
|
||||
title: song.name, // 歌曲标题
|
||||
id: song.id, // 歌曲ID
|
||||
type: song.t, // 0: 一般类型 1: 通过云盘上传的音乐，网易云不存在公开对应 2: 通过云盘上传的音乐，网易云存在公开对应
|
||||
alias: JSON.stringify(song.alia), // 别名列表，第一个别名会被显示作副标题
|
||||
pop: song.pop, // 小数，常取[0.0, 100.0]中离散的几个数值, 表示歌曲热度
|
||||
fee: song.fee, // 版权 0: 免费或无版权 1: VIP 歌曲 4: 购买专辑 8: 非会员可免费播放低音质，会员可播放高音质及下载 fee 为 1 或 8 的歌曲均可单独购买 2 元单曲
|
||||
duration: song.dt, // 歌曲时长
|
||||
quality: JSON.stringify({ h: song.h, m: song.m, l: song.l, sq: song.sq }), // 高/中/低/无损质量文件信息
|
||||
version: song.version, // 歌曲版本信息
|
||||
cd: song.cd, // None或如"04", "1/2", "3", "null"的字符串，表示歌曲属于专辑中第几张CD，对应音频文件的Tag
|
||||
no: song.no, // 表示歌曲属于CD中第几曲，0表示没有这个字段，对应音频文件的Tag
|
||||
djId: song.djId, // 0: 不是DJ节目 其他：是DJ节目，表示DJ ID
|
||||
sId: song.s_id, // 对于t == 2的歌曲，表示匹配到的公开版本歌曲ID
|
||||
originCoverType: song.originCoverType, // 0: 未知 1: 原曲 2: 翻唱
|
||||
image: "",
|
||||
pubTime: song.publishTime, // 毫秒为单位的Unix时间戳
|
||||
mv: song.mv, // 非零表示有MV ID
|
||||
single: song.single, // 0: 有专辑信息或者是DJ节目 1: 未知专辑
|
||||
noCopyrightRcmd: song.noCopyrightRcmd ? JSON.stringify(song.noCopyrightRcmd) : "", // None表示可以播，非空表示无版权
|
||||
|
||||
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,
|
||||
});
|
||||
artist: song.ar.map(item => item.id), // 歌手列表
|
||||
album: song.al.id || 0, // 专辑，如果是DJ节目(dj_type != 0)或者无专辑信息(single == 1)，则专辑id为0
|
||||
};
|
||||
});
|
||||
dbUtils.query('INSERT IGNORE INTO song SET ?', {
|
||||
song_id: songInfo.songId,
|
||||
title: songInfo.title,
|
||||
image: songInfo.image,
|
||||
pub_date: songInfo.pubDate,
|
||||
});
|
||||
return songInfo;
|
||||
|
||||
// console.log("songAlbumRel, songArtistRel", songAlbumRel, songArtistRel);
|
||||
// console.log("songInfoList", songInfoList);
|
||||
|
||||
await dbUtils.query('INSERT IGNORE INTO song_album_relation (song_id, album_id) VALUES ?', [songAlbumRel]);
|
||||
await dbUtils.query('INSERT IGNORE INTO song_artist_relation (song_id, artist_id) VALUES ?', [songArtistRel]);
|
||||
await dbUtils.query(`
|
||||
INSERT INTO song (
|
||||
song_id, title, type, alias, pop, fee, quality, cd,
|
||||
no, dj_id, s_id, origin_cover_type, pub_time,
|
||||
no_copyright_rcmd, mv, single, version, data_version
|
||||
) VALUES ? ON DUPLICATE KEY UPDATE
|
||||
title = VALUES(title), type = VALUES(type), alias = VALUES(alias), pop = VALUES(pop), fee = VALUES(fee), quality = VALUES(quality), cd = VALUES(cd),
|
||||
no = VALUES(no), dj_id = VALUES(dj_id), s_id = VALUES(s_id), origin_cover_type = VALUES(origin_cover_type), pub_time = VALUES(pub_time),
|
||||
no_copyright_rcmd = VALUES(no_copyright_rcmd), mv = VALUES(mv), single = VALUES(single), version = VALUES(version), data_version = VALUES(data_version)
|
||||
`, [songInfoList.map(songInfo => [
|
||||
songInfo.id, songInfo.title, songInfo.type, songInfo.alias, songInfo.pop, songInfo.fee, songInfo.quality, songInfo.cd,
|
||||
songInfo.no, songInfo.djId, songInfo.sId, songInfo.originCoverType, songInfo.pubTime,
|
||||
songInfo.noCopyrightRcmd, songInfo.mv, songInfo.single, songInfo.version, 2
|
||||
])]);
|
||||
// image 因为接口没有返回，所以不更新
|
||||
}
|
||||
|
||||
module.exports = {
|
||||
|
||||
@ -31,6 +31,8 @@ async function test() {
|
||||
|
||||
// 不是所有歌手都有个人主页 例如 https://music.163.com/#/artist?id=1079075
|
||||
|
||||
// let res = await songInfoUtils.fetchAll({ args: {} });
|
||||
|
||||
// let res = await albumInfoUtils.fetch({ albumId: "9156", debug: true });
|
||||
// let res = await artistInfoUtils.fetch({ artistId: "12023508" });
|
||||
// let res = await songInfoUtils.fetch({ songId: "437608327" });
|
||||
@ -40,8 +42,6 @@ async function test() {
|
||||
// let res = await artistInfoUtils.getFromDatabase({ artistId: "12023508" });
|
||||
// let res = await songInfoUtils.getFromDatabase({ songId: "437608327" });
|
||||
|
||||
let res = await dbUtils.query('INSERT IGNORE INTO song (`song_id`, `title`, `image`, `pub_date`) VALUES ?',
|
||||
[[[100, '4', '3', '4'], [200, '23', '4', '5']]]);
|
||||
console.log(res);
|
||||
}
|
||||
|
||||
|
||||
Loading…
x
Reference in New Issue
Block a user