From 8cefcf740cda4f8566f9e2ec0e21abbf6595f21f Mon Sep 17 00:00:00 2001
From: SukkaW <isukkaw@gmail.com>
Date: Mon, 31 Mar 2025 00:53:07 +0800
Subject: [PATCH] Perf: compare and write file in worker

---
 Build/build-cdn-download-conf.ts   |  2 +-
 Build/lib/create-file.ts           |  4 +--
 Build/lib/create-file.worker.ts    | 55 ++++++++++++++++++++++++++++++
 Build/lib/writing-strategy/base.ts | 14 ++++++++
 4 files changed, 72 insertions(+), 3 deletions(-)
 create mode 100644 Build/lib/create-file.worker.ts

diff --git a/Build/build-cdn-download-conf.ts b/Build/build-cdn-download-conf.ts
index 6b4e881b..80c2e707 100644
--- a/Build/build-cdn-download-conf.ts
+++ b/Build/build-cdn-download-conf.ts
@@ -59,7 +59,7 @@ export const buildCdnDownloadConf = task(require.main === module, __filename)(as
     downloadDomainSet,
     steamDomainSet
   ] = await Promise.all([
-    getS3OSSDomainsPromise,
+    span.traceChildPromise('download public suffix list for s3', getS3OSSDomainsPromise),
     cdnDomainsListPromise,
     downloadDomainSetPromise,
     steamDomainSetPromise
diff --git a/Build/lib/create-file.ts b/Build/lib/create-file.ts
index 28904b0c..ef89a8b8 100644
--- a/Build/lib/create-file.ts
+++ b/Build/lib/create-file.ts
@@ -48,8 +48,6 @@ export async function fileEqual(linesA: string[], source: AsyncIterable<string>
 }
 
 export async function compareAndWriteFile(span: Span, linesA: string[], filePath: string) {
-  const linesALen = linesA.length;
-
   const isEqual = await span.traceChildAsync(`compare ${filePath}`, async () => {
     if (fs.existsSync(filePath)) {
       return fileEqual(linesA, readFileByLine(filePath));
@@ -65,6 +63,8 @@ export async function compareAndWriteFile(span: Span, linesA: string[], filePath
   }
 
   await span.traceChildAsync(`writing ${filePath}`, async () => {
+    const linesALen = linesA.length;
+
     // The default highwater mark is normally 16384,
     // So we make sure direct write to file if the content is
     // most likely less than 500 lines
diff --git a/Build/lib/create-file.worker.ts b/Build/lib/create-file.worker.ts
new file mode 100644
index 00000000..0cd1c69e
--- /dev/null
+++ b/Build/lib/create-file.worker.ts
@@ -0,0 +1,55 @@
+import Worktank from 'worktank';
+import os from 'node:os';
+import process from 'node:process';
+import type { Span } from '../trace';
+
+const pool = new Worktank({
+  name: 'process-phishing-domains',
+  size: Math.max(2, Math.max(1, ('availableParallelism' in os ? os.availableParallelism() : (os as typeof import('node:os')).cpus().length) - 1)),
+  timeout: 10000, // The maximum number of milliseconds to wait for the result from the worker, if exceeded the worker is terminated and the execution promise rejects
+  warmup: true,
+  autoterminate: 30000, // The interval of milliseconds at which to check if the pool can be automatically terminated, to free up resources, workers will be spawned up again if needed
+  env: {},
+  methods: {
+    // eslint-disable-next-line object-shorthand -- workertank
+    compareAndWriteFile: async function (
+      linesA: string[], filePath: string,
+      importMetaUrl: string
+    ): Promise<void> {
+      const { default: module } = await import('node:module');
+      const __require = module.createRequire(importMetaUrl);
+
+      const fs = __require('fs') as typeof import('fs');
+      const { readFileByLine } = __require('./fetch-text-by-line') as typeof import('./fetch-text-by-line');
+      const { fileEqual } = __require('./create-file') as typeof import('./create-file');
+      const path = __require('node:path') as typeof import('node:path');
+      const { fastStringArrayJoin } = __require('foxts/fast-string-array-join') as typeof import('foxts/fast-string-array-join');
+      const picocolors = __require('picocolors') as typeof import('picocolors');
+
+      let isEqual = false;
+      if (fs.existsSync(filePath)) {
+        isEqual = await fileEqual(linesA, readFileByLine(filePath));
+      } else {
+        console.log(`${filePath} does not exists, writing...`);
+        isEqual = false;
+      }
+
+      if (isEqual) {
+        console.log(picocolors.gray(picocolors.dim(`same content, bail out writing: ${filePath}`)));
+        return;
+      }
+
+      const dir = path.dirname(filePath);
+      if (!fs.existsSync(dir)) {
+        fs.mkdirSync(dir, { recursive: true });
+      }
+      fs.writeFileSync(filePath, fastStringArrayJoin(linesA, '\n') + '\n', { encoding: 'utf-8' });
+    }
+  }
+});
+
+export function compareAndWriteFileInWorker(span: Span, linesA: string[], filePath: string) {
+  return span.traceChildAsync(`compare and write ${filePath}`, () => pool.exec('compareAndWriteFile', [linesA, filePath, import.meta.url]));
+}
+
+process.on('beforeExit', () => pool.terminate());
diff --git a/Build/lib/writing-strategy/base.ts b/Build/lib/writing-strategy/base.ts
index 0e487ee4..58ae2d07 100644
--- a/Build/lib/writing-strategy/base.ts
+++ b/Build/lib/writing-strategy/base.ts
@@ -1,5 +1,6 @@
 import type { Span } from '../../trace';
 import { compareAndWriteFile } from '../create-file';
+import { compareAndWriteFileInWorker } from '../create-file.worker';
 
 /**
  * The class is not about holding rule data, instead it determines how the
@@ -76,6 +77,19 @@ export abstract class BaseWriteStrategy {
     if (!this.result) {
       return;
     }
+
+    if (this.result.length > 1000) {
+      return compareAndWriteFileInWorker(
+        span,
+        this.withPadding(
+          title,
+          description,
+          date,
+          this.result
+        ),
+        filePath
+      );
+    }
     return compareAndWriteFile(
       span,
       this.withPadding(