Chore: add fetch UA

Build/build-speedtest-domainset.ts

@@ -6,9 +6,11 @@ import domainSorter from './lib/stable-sort-domain';
 import { Sema } from 'async-sema';
 import * as tldts from 'tldts';
 import { task } from './lib/trace-runner';
+import { fetchWithRetry } from './lib/fetch-retry';
+
 const s = new Sema(3);
 
-const latestTopUserAgentsPromise = fetch('https://unpkg.com/top-user-agents@latest/index.json')
+const latestTopUserAgentsPromise = fetchWithRetry('https://unpkg.com/top-user-agents@latest/index.json')
   .then(res => res.json() as Promise<string[]>);
 
 const querySpeedtestApi = async (keyword: string): Promise<(string | null)[]> => {
@@ -23,7 +25,7 @@ const querySpeedtestApi = async (keyword: string): Promise<(string | null)[]> =>
     const key = `fetch speedtest endpoints: ${keyword}`;
     console.time(key);
 
-    const res = await fetch(`https://www.speedtest.net/api/js/servers?engine=js&search=${keyword}&limit=100`, {
+    const res = await fetchWithRetry(`https://www.speedtest.net/api/js/servers?engine=js&search=${keyword}&limit=100`, {
       headers: {
         dnt: '1',
         Referer: 'https://www.speedtest.net/',

Build/build-telegram-cidr.ts

@@ -1,5 +1,5 @@
 // @ts-check
-import { fetchWithRetry } from './lib/fetch-retry';
+import { defaultRequestInit, fetchWithRetry } from './lib/fetch-retry';
 import { createReadlineInterfaceFromResponse } from './lib/fetch-remote-text-by-line';
 import path from 'path';
 import { isIPv4, isIPv6 } from 'net';
@@ -9,7 +9,7 @@ import { task } from './lib/trace-runner';
 
 export const buildTelegramCIDR = task(__filename, async () => {
   /** @type {Response} */
-  const resp = await fetchWithRetry('https://core.telegram.org/resources/cidr.txt');
+  const resp = await fetchWithRetry('https://core.telegram.org/resources/cidr.txt', defaultRequestInit);
   const lastModified = resp.headers.get('last-modified');
   const date = lastModified ? new Date(lastModified) : new Date();
 

Build/download-previous-build.ts

@@ -8,6 +8,7 @@ import { pipeline } from 'stream/promises';
 import { readFileByLine } from './lib/fetch-remote-text-by-line';
 import { isCI } from 'ci-info';
 import { task, traceAsync } from './lib/trace-runner';
+import { defaultRequestInit, fetchWithRetry } from './lib/fetch-retry';
 
 const IS_READING_BUILD_OUTPUT = 1 << 2;
 const ALL_FILES_EXISTS = 1 << 3;
@@ -51,7 +52,7 @@ export const downloadPreviousBuild = task(__filename, async () => {
   await traceAsync(
     'Download and extract previous build',
     () => Promise.all([
-      fetch('https://codeload.github.com/sukkalab/ruleset.skk.moe/tar.gz/master'),
+      fetchWithRetry('https://codeload.github.com/sukkalab/ruleset.skk.moe/tar.gz/master', defaultRequestInit),
       fsp.mkdir(extractedPath, { recursive: true })
     ]).then(([resp]) => pipeline(
       Readable.fromWeb(resp.body!),
@@ -82,7 +83,7 @@ export const downloadPublicSuffixList = task(__filename, async () => {
   const publicSuffixPath = path.join(publicSuffixDir, 'public_suffix_list_dat.txt');
 
   const [resp] = await Promise.all([
-    fetch('https://publicsuffix.org/list/public_suffix_list.dat'),
+    fetchWithRetry('https://publicsuffix.org/list/public_suffix_list.dat', defaultRequestInit),
     fsp.mkdir(publicSuffixDir, { recursive: true })
   ]);
 

Build/lib/fetch-remote-text-by-line.ts

@@ -1,5 +1,5 @@
 import type { BunFile } from 'bun';
-import { fetchWithRetry } from './fetch-retry';
+import { fetchWithRetry, defaultRequestInit } from './fetch-retry';
 
 const decoder = new TextDecoder('utf-8');
 
@@ -56,6 +56,6 @@ export async function* createReadlineInterfaceFromResponse(resp: Response): Asyn
   }
 }
 
-export function fetchRemoteTextAndCreateReadlineInterface(url: string | URL, opt?: RequestInit): Promise<AsyncGenerator<string>> {
-  return fetchWithRetry(url, opt).then(res => createReadlineInterfaceFromResponse(res));
+export function fetchRemoteTextAndCreateReadlineInterface(url: string | URL): Promise<AsyncGenerator<string>> {
+  return fetchWithRetry(url, defaultRequestInit).then(res => createReadlineInterfaceFromResponse(res));
 }

Build/lib/fetch-retry.ts

@@ -1,4 +1,10 @@
 // @ts-expect-error -- missing types
 import createFetchRetry from '@vercel/fetch-retry';
 
+export const defaultRequestInit: RequestInit = {
+  headers: {
+    'User-Agent': 'curl/8.1.2 (https://github.com/SukkaW/Surge)'
+  }
+}
+
 export const fetchWithRetry: typeof fetch = createFetchRetry(fetch);

Build/lib/get-gorhill-publicsuffix.ts

@@ -1,6 +1,7 @@
 import { toASCII } from 'punycode';
 import path from 'path';
 import { traceAsync } from './trace-runner';
+import { defaultRequestInit, fetchWithRetry } from './fetch-retry';
 import type { PublicSuffixList } from 'gorhill-publicsuffixlist';
 
 const publicSuffixPath = path.resolve(__dirname, '../../node_modules/.cache/public_suffix_list_dat.txt');
@@ -13,7 +14,7 @@ const getGorhillPublicSuffix = () => traceAsync('create gorhill public suffix in
   const [publicSuffixListDat, { default: gorhill }] = await Promise.all([
     await publicSuffixFile.exists()
       ? publicSuffixFile.text()
-      : fetch('https://publicsuffix.org/list/public_suffix_list.dat').then(r => {
+      : fetchWithRetry('https://publicsuffix.org/list/public_suffix_list.dat', defaultRequestInit).then(r => {
         console.log('public_suffix_list.dat not found, fetch directly from remote.');
         return r.text();
       }),

Build/lib/parse-filter.ts

@@ -1,5 +1,5 @@
 // @ts-check
-import { fetchWithRetry } from './fetch-retry';
+import { defaultRequestInit, fetchWithRetry } from './fetch-retry';
 import * as tldts from './cached-tld-parse';
 import { fetchRemoteTextAndCreateReadlineInterface } from './fetch-remote-text-by-line';
 import { NetworkFilter } from '@cliqz/adblocker';
@@ -198,7 +198,7 @@ export async function processFilterRules(
       filterRules = (
         await Promise.any(
           [filterRulesUrl, ...(fallbackUrls || [])].map(async url => {
-            const r = await fetchWithRetry(url, { signal: controller.signal });
+            const r = await fetchWithRetry(url, { signal: controller.signal, ...defaultRequestInit });
             const text = await r.text();
 
             controller.abort();

Build/lib/reject-data-source.ts

@@ -106,9 +106,10 @@ export const ADGUARD_FILTERS = [
   [
     'https://curbengh.github.io/urlhaus-filter/urlhaus-filter-agh-online.txt',
     [
-      'https://urlhaus-filter.pages.dev/urlhaus-filter-agh-online.txt'
+      'https://urlhaus-filter.pages.dev/urlhaus-filter-agh-online.txt',
       // Prefer mirror, since malware-filter.gitlab.io has not been updated for a while
-      // 'https://malware-filter.gitlab.io/urlhaus-filter/urlhaus-filter-agh-online.txt'
+      // 'https://malware-filter.gitlab.io/urlhaus-filter/urlhaus-filter-agh-online.txt',
+      'https://ublockorigin.github.io/uAssets/thirdparties/urlhaus-filter/urlhaus-filter-online.txt',
     ]
   ],
   // Curben's Phishing URL Blocklist