Surge_by_SukkaW/Build/tools-dedupe-src.ts
2025-11-10 23:01:29 +08:00

97 lines
3.1 KiB
TypeScript

import { fdir as Fdir } from 'fdir';
import path from 'node:path';
import fsp from 'node:fs/promises';
import { SOURCE_DIR } from './constants/dir';
import { readFileByLine } from './lib/fetch-text-by-line';
import { processLine } from './lib/process-line';
import { HostnameSmolTrie, HostnameTrie } from './lib/trie';
import { task } from './trace';
const ENFORCED_WHITELIST = [
'hola.sk',
'hola.org',
'hola-shopping.com',
'mynextphone.io',
'iadmatapk.nosdn.127.net',
'httpdns.bilivideo.com',
'httpdns-v6.gslb.yy.com',
'twemoji.maxcdn.com',
'samsungcloudsolution.com',
'samsungcloudsolution.net',
'samsungqbe.com',
'ntp.api.bz',
'cdn.tuk.dev',
'vocadb-analytics.fly.dev'
];
const WHITELIST: string[] = ['httpdns.bilivideo.com', 'ntp.api.bz', 'httpdns-v6.gslb.yy.com', 'img.vim-cn.com', 'img.jjbb.me', 'thingproxy.freeboard.io', 'assets.chess24.com', 'cdn.chess24.com', 'static-assets.freeanimehentai.net', 'static.javcdn.info', 'cdn.vidible.tv', 'it.apache.contactlab.it', 'mirror.netinch.com', 'de.freedif.org', 'league1.maoyuncloud.cn', 'spl.ztvx8.com', 'zls.xz6d.com', 'iadmatapk.nosdn.127.net', 'show.buzzcity.net', 'click.buzzcity.net', 'apps.buzzcity.net', 'content-cdn.y2mate.com', 'images.voguehk.com', 'cdn.amh.moe', 'statics.mnnews.tw'];
task(require.main === module, __filename)(async (span) => {
const files = await span.traceChildAsync('crawl thru all files', () => new Fdir()
.withFullPaths()
.filter((filepath, isDirectory) => {
if (isDirectory) return true;
const extname = path.extname(filepath);
return extname !== '.js' && extname !== '.ts';
})
.crawl(SOURCE_DIR)
.withPromise());
const whiteTrie = span.traceChildSync('build whitelist trie', () => {
const trie = new HostnameSmolTrie(WHITELIST);
ENFORCED_WHITELIST.forEach((item) => trie.whitelist(item));
return trie;
});
await Promise.all(files.map(file => span.traceChildAsync('dedupe ' + file, () => dedupeFile(file, whiteTrie))));
});
async function dedupeFile(file: string, whitelist: HostnameSmolTrie) {
const result: string[] = [];
const trie = new HostnameTrie();
let line: string | null = '';
for await (const l of readFileByLine(file)) {
line = processLine(l);
if (!line) {
if (l.startsWith('# $ skip_dedupe_src')) {
return;
}
result.push(l); // keep all comments and blank lines
continue;
}
if (trie.has(line)) {
continue; // drop duplicate
}
if (whitelist.has(line)) {
continue; // drop whitelisted items
}
trie.add(line);
result.push(line);
}
return fsp.writeFile(file, result.join('\n') + '\n');
}
// function isDomainSuffix(whiteItem: string, incomingItem: string) {
// const whiteIncludeDomain = whiteItem[0] === '.';
// whiteItem = whiteItem[0] === '.' ? whiteItem.slice(1) : whiteItem;
// if (whiteItem === incomingItem) {
// return true; // as long as exact match, we don't care if subdomain is included or not
// }
// if (whiteIncludeDomain) {
// return incomingItem.endsWith('.' + whiteItem);
// }
// return false;
// }