coder-xiaomo/tools
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity GitHub Gitee

playlist修bug

Browse Source
This commit is contained in:
程序员小墨 2022-10-20 00:30:09 +08:00
parent 896e66de53
commit 0862db5db9
2 changed files with 38 additions and 21 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

55
netease_music/src/getInfo/playlistUtils.js
View File

@ -12,8 +12,12 @@ const dbUtils = global.dbUtils;
const { playlist_catlist, playlist_hot, playlist_detail } = require('NeteaseCloudMusicApi');
async function fetchAll({ args }) {
// 睡眠时间设置长一些，不然容易触发500错误
global.sleepTime = Math.max(500, global.sleepTime);
console.log("global.sleepTime", global.sleepTime);
console.log("playlist 需要一口气爬完，中途不能停止，否则下次又要重头爬（歌单不会重复爬取，但是分页列表会）");
console.log("start fetching playlists ...");
console.log("playlist 需要一口气爬完，中途不能停止，否则下次又要重头爬（歌单不会重复爬取，但是分页列表会）")
// 从数据库中查出所有的网易云分类
let result = await dbUtils.query(`SELECT title FROM category WHERE netease_group_chinese IS NOT NULL`);
@ -23,10 +27,12 @@ async function fetchAll({ args }) {
for (let i = 0; i < cate.length; i++) {
const categoryName = cate[i];
try {
console.log("开始爬取分类：", categoryName);
await fetchCategory({ categoryName: categoryName, progress: `${i + 1}/${cate.length}` });
} catch (err) {
console.error(err);
}
await sleepUtils.sleep(global.sleepTime);
}
console.log("爬取完毕");
await sleepUtils.sleep(2000);
@ -58,35 +64,42 @@ async function fetchCategory({ categoryName, progress }) {
return;
}
// 从数据库查出已爬取的歌单ids，并从 playlistIds 中排除这部分歌单
var exceptPlaylistIds = await dbUtils.query(`
SELECT playlist_id FROM playlist WHERE playlist_id IN ?
`, [[playlistIds]]);
exceptPlaylistIds = exceptPlaylistIds.map(playlist => playlist.playlist_id);
if (playlistIds.length > 0) {
// 从数据库查出已爬取的歌单ids，并从 playlistIds 中排除这部分歌单
var exceptPlaylistIds = await dbUtils.query(`
SELECT playlist_id FROM playlist WHERE playlist_id IN ?
`, [[playlistIds]]);
exceptPlaylistIds = exceptPlaylistIds.map(playlist => playlist.playlist_id);
var finalPlaylistIds = playlistIds.filter(playlistId => exceptPlaylistIds.indexOf(playlistId) == -1);
var finalPlaylistIds = playlistIds.filter(playlistId => exceptPlaylistIds.indexOf(playlistId) == -1);
// console.log("playlistIds", playlistIds);
// console.log("exceptPlaylistIds", exceptPlaylistIds);
// console.log("finalPlaylistIds", finalPlaylistIds);
console.log("finalPlaylistIds.length", finalPlaylistIds.length);
// console.log("playlistIds", playlistIds);
// console.log("exceptPlaylistIds", exceptPlaylistIds);
// console.log("finalPlaylistIds", finalPlaylistIds);
console.log(categoryName, "offset", offset, "playlistIds.length", playlistIds.length, "finalPlaylistIds.length", finalPlaylistIds.length);
// console.log(url);
for (let i = 0; i < finalPlaylistIds.length; i++) {
await global.checkIsExit();
const playlistId = finalPlaylistIds[i];
// console.log(offset, i, finalPlaylistIds.length);
console.log(`分类: ${progress} | 歌单: ${offset + i + 1}/${offset + finalPlaylistIds.length} | playlist: ${playlistId}`);
try {
await fetch({ playlistId: playlistId });
} catch (err) {
console.error(err);
for (let i = 0; i < finalPlaylistIds.length; i++) {
await global.checkIsExit();
const playlistId = finalPlaylistIds[i];
// console.log(offset, i, finalPlaylistIds.length);
console.log(`分类: ${progress} | 歌单: ${offset + i + 1}/${offset + finalPlaylistIds.length} | playlist: ${playlistId}`);
try {
await fetch({ playlistId: playlistId });
} catch (err) {
console.error(err);
}
await sleepUtils.sleep(global.sleepTime);
}
await sleepUtils.sleep(global.sleepTime);
} else {
console.log("失败url", url);
await dbUtils.query('INSERT INTO log (`id`, `name`, `msg`) VALUES (?, ?, ?)', [0, 'playlist_fetch', `失败 可能是爬太快了\n${url}`]);
}
// 最有一页判断标识
if (html.indexOf(`class="zbtn znxt js-disabled">下一页</a>`) > -1) haveNext = false;
offset += perPage;
await sleepUtils.sleep(global.sleepTime);
}
}

4
netease_music/todo.txt
View File

@ -24,6 +24,10 @@ node index --utils playlist
后期：
批量查库修改为一条SQL搞定
歌单定时更新（rel表中添加一个del字段，先将歌单下面的全部置为删除状态，再插入的时候把已有歌曲的标记重新修改为正常状态）
评论的更新
爬取歌单playlist