From 96cf0fc1f9995a07f177f40ff303fbdc1edc759d Mon Sep 17 00:00:00 2001
From: SukkaW <isukkaw@gmail.com>
Date: Sat, 18 Jan 2025 02:59:32 +0800
Subject: [PATCH] Perf: preload phishing data source

---
 Build/lib/get-phishing-domains.ts     | 15 +++++++++------
 Build/lib/parse-filter/domainlists.ts | 18 ++++++++++++++++++
 Build/lib/parse-filter/hosts.ts       | 20 ++++++++++++++++++++
 3 files changed, 47 insertions(+), 6 deletions(-)

diff --git a/Build/lib/get-phishing-domains.ts b/Build/lib/get-phishing-domains.ts
index f25b3764..99a073a9 100644
--- a/Build/lib/get-phishing-domains.ts
+++ b/Build/lib/get-phishing-domains.ts
@@ -1,5 +1,5 @@
-import { processHosts } from './parse-filter/hosts';
-import { processDomainLists } from './parse-filter/domainlists';
+import { processHostsWithPreload } from './parse-filter/hosts';
+import { processDomainListsWithPreload } from './parse-filter/domainlists';
 
 import * as tldts from 'tldts-experimental';
 
@@ -207,15 +207,18 @@ const processPhihsingDomains = cache(function processPhihsingDomains(domainArr:
   temporaryBypass: !isCI || DEBUG_DOMAIN_TO_FIND !== null
 });
 
+const downloads = [
+  ...PHISHING_DOMAIN_LISTS_EXTRA.map(entry => processDomainListsWithPreload(...entry)),
+  ...PHISHING_HOSTS_EXTRA.map(entry => processHostsWithPreload(...entry))
+];
+
 export function getPhishingDomains(parentSpan: Span) {
   return parentSpan.traceChild('get phishing domains').traceAsyncFn(async (span) => {
     const domainArr = await span.traceChildAsync('download/parse/merge phishing domains', async (curSpan) => {
       const domainArr: string[] = [];
 
-      await Promise.all([
-        ...PHISHING_DOMAIN_LISTS_EXTRA.map(entry => processDomainLists(curSpan, ...entry)),
-        ...PHISHING_HOSTS_EXTRA.map(entry => processHosts(curSpan, ...entry))
-      ]).then(domainGroups => domainGroups.forEach(appendArrayInPlaceCurried(domainArr)));
+      const domainGroups = await Promise.all(downloads.map(task => task(curSpan)));
+      domainGroups.forEach(appendArrayInPlaceCurried(domainArr));
 
       return domainArr;
     });
diff --git a/Build/lib/parse-filter/domainlists.ts b/Build/lib/parse-filter/domainlists.ts
index a0284079..428c73a0 100644
--- a/Build/lib/parse-filter/domainlists.ts
+++ b/Build/lib/parse-filter/domainlists.ts
@@ -48,3 +48,21 @@ export function processDomainLists(
     return domainSets;
   });
 }
+
+export function processDomainListsWithPreload(domainListsUrl: string, mirrors: string[] | null, includeAllSubDomain = false) {
+  const downloadPromise = fetchAssets(domainListsUrl, mirrors);
+
+  return (span: Span) => span.traceChildAsync(`process domainlist: ${domainListsUrl}`, async (span) => {
+    const text = await span.traceChildPromise('download', downloadPromise);
+    const domainSets: string[] = [];
+    const filterRules = text.split('\n');
+
+    span.traceChildSync('parse domain list', () => {
+      for (let i = 0, len = filterRules.length; i < len; i++) {
+        domainListLineCb(filterRules[i], domainSets, includeAllSubDomain, domainListsUrl);
+      }
+    });
+
+    return domainSets;
+  });
+}
diff --git a/Build/lib/parse-filter/hosts.ts b/Build/lib/parse-filter/hosts.ts
index 6df00c40..db4dd2ae 100644
--- a/Build/lib/parse-filter/hosts.ts
+++ b/Build/lib/parse-filter/hosts.ts
@@ -44,3 +44,23 @@ export function processHosts(
     return domainSets;
   });
 }
+
+export function processHostsWithPreload(hostsUrl: string, mirrors: string[] | null, includeAllSubDomain = false) {
+  const downloadPromise = fetchAssets(hostsUrl, mirrors);
+
+  return (span: Span) => span.traceChildAsync(`process hosts: ${hostsUrl}`, async (span) => {
+    const text = await span.traceChild('download').tracePromise(downloadPromise);
+
+    const domainSets: string[] = [];
+
+    const filterRules = text.split('\n');
+
+    span.traceChild('parse hosts').traceSyncFn(() => {
+      for (let i = 0, len = filterRules.length; i < len; i++) {
+        hostsLineCb(filterRules[i], domainSets, includeAllSubDomain, hostsUrl);
+      }
+    });
+
+    return domainSets;
+  });
+}