From 70a2dfa89feebe1f08246f58c87692d11d888884 Mon Sep 17 00:00:00 2001
From: Viet Huynh <xa.lach@yahoo.com>
Date: Sun, 19 Nov 2023 17:46:39 +0700
Subject: [PATCH] refactored domain processing code to improve performance and
 memory usage

---
 cf_list_create.js | 60 ++++++++++++++++++++---------------------------
 lib/utils.js      | 18 +++++++-------
 2 files changed, 34 insertions(+), 44 deletions(-)

diff --git a/cf_list_create.js b/cf_list_create.js
index 68670cb..cfecb22 100644
--- a/cf_list_create.js
+++ b/cf_list_create.js
@@ -73,40 +73,31 @@ await readFile(resolve(`./${blocklistFilename}`), (line, rl) => {
 
   processedDomainCount++;
 
-  // Get all the levels of the domain and check from the highest
-  // because we are blocking all subdomains
-  // Example: fourth.third.example.com => ["example.com", "third.example.com", "fourth.third.example.com"]
-  const anyDomainExists = extractDomain(domain)
-    .reverse()
-    .some((item) => {
-      if (blocklist.has(item)) {
-        if (item === domain) {
-          // The exact domain is already blocked
-          console.log(`Found ${item} in blocklist already - Skipping`);
-          duplicateDomainCount++;
-        } else {
-          // The higher-level domain is already blocked
-          // so it's not necessary to block this domain
-          console.log(
-            `Found ${item} in blocklist already - Skipping ${domain}`
-          );
-          unnecessaryDomainCount++;
-        }
-
-        return true;
-      }
-
-      return false;
-    });
-
-  if (anyDomainExists) return;
-
   if (allowlist.has(domain)) {
     console.log(`Found ${domain} in allowlist - Skipping`);
     allowedDomainCount++;
     return;
   }
 
+  if (blocklist.has(domain)) {
+    console.log(`Found ${domain} in blocklist already - Skipping`);
+    duplicateDomainCount++;
+    return;
+  }
+
+  // Get all the levels of the domain and check from the highest
+  // because we are blocking all subdomains
+  // Example: fourth.third.example.com => ["example.com", "third.example.com", "fourth.third.example.com"]
+  for (const item of extractDomain(domain).slice(1)) {
+    if (!blocklist.has(item)) continue;
+
+    // The higher-level domain is already blocked
+    // so it's not necessary to block this domain
+    console.log(`Found ${item} in blocklist already - Skipping ${domain}`);
+    unnecessaryDomainCount++;
+    return;
+  }
+
   blocklist.set(domain, 1);
   domains.push(domain);
 
@@ -124,8 +115,8 @@ console.log("\n\n");
 console.log(`Number of processed domains: ${processedDomainCount}`);
 console.log(`Number of duplicate domains: ${duplicateDomainCount}`);
 console.log(`Number of unnecessary domains: ${unnecessaryDomainCount}`);
-console.log(`Number of blocked domains: ${domains.length}`);
 console.log(`Number of allowed domains: ${allowedDomainCount}`);
+console.log(`Number of blocked domains: ${domains.length}`);
 console.log(`Number of lists to be created: ${numberOfLists}`);
 console.log("\n\n");
 
@@ -143,12 +134,11 @@ console.log("\n\n");
 
   if (FAST_MODE) {
     await createZeroTrustListsAtOnce(domains);
-    // TODO: make this less repetitive
-    await notifyWebhook(`CF List Create script finished running (${domains.length} domains, ${numberOfLists} lists)`);
-    return;
+  } else {
+    await createZeroTrustListsOneByOne(domains);
   }
 
-  await createZeroTrustListsOneByOne(domains);
-
-  await notifyWebhook(`CF List Create script finished running (${domains.length} domains, ${numberOfLists} lists)`);
+  await notifyWebhook(
+    `CF List Create script finished running (${domains.length} domains, ${numberOfLists} lists)`
+  );
 })();
diff --git a/lib/utils.js b/lib/utils.js
index 1c60b57..a263a2f 100644
--- a/lib/utils.js
+++ b/lib/utils.js
@@ -19,18 +19,18 @@ export const isValidDomain = (value) =>
  * @param {string} domain The domain to be extracted.
  * @returns {string[]}
  */
-export const extractDomain = (domain) =>
-  domain.split(".").reduce((previous, current, index, array) => {
-    const nextIndex = index + 1;
+export const extractDomain = (domain) => {
+  const parts = domain.split(".");
+  const extractedDomains = [];
 
-    if (nextIndex > array.length - 1) return previous;
+  for (let i = 0; i < parts.length; i++) {
+    const subdomains = parts.slice(i).join(".");
 
-    const domain = [current, ...array.slice(nextIndex)].join(".");
+    extractedDomains.unshift(subdomains);
+  }
 
-    previous.push(domain);
-
-    return previous;
-  }, []);
+  return extractedDomains;
+};
 
 /**
  * Checks if the value is a comment.