Browse Source

Perf: improve reject set dedupe worker performance

SukkaW 3 years ago
parent
commit
7a7a97e4d0
1 changed files with 12 additions and 5 deletions
  1. 12 5
      Build/worker/build-reject-domainset-worker.js

+ 12 - 5
Build/worker/build-reject-domainset-worker.js

@@ -1,25 +1,32 @@
 const { workerData, move } = require('piscina');
 
 const len = workerData.length;
+// pre check if fullset domain is starts with a "."
+// This avoid calling chatCodeAt repeatedly
+const fullsetDomainStartsWithADot = workerData.map(domain => domain.charCodeAt(0) === 46);
 
 module.exports = ({ chunk }) => {
   const chunkLength = chunk.length;
-  const outputToBeRemoved = new Int32Array(chunkLength);
+  const outputToBeRemoved = new Int8Array(chunkLength);
 
   for (let i = 0; i < chunkLength; i++) {
     const domainFromInput = chunk[i];
 
     for (let j = 0; j < len; j++) {
+      // Check if domainFromFullset starts with a "."
+      if (!fullsetDomainStartsWithADot[j]) continue;
+      // domainFromFullSet is now startsWith a "."
+
       const domainFromFullSet = workerData[j];
 
       if (domainFromFullSet === domainFromInput) continue;
-      if (domainFromFullSet.charCodeAt(0) !== 46) continue;
-      // domainFromFullSet is now startsWith a "."
+
+      const domainFromInputLen = domainFromInput.length;
 
       if (domainFromInput.charCodeAt(0) !== 46) {
         let shouldBeRemoved = true;
 
-        for (let k = 0, l2 = domainFromInput.length; k < l2; k++) {
+        for (let k = 0; k < domainFromInputLen; k++) {
           if (domainFromFullSet.charCodeAt(k + 1) !== domainFromInput.charCodeAt(k)) {
             shouldBeRemoved = false;
             break;
@@ -33,7 +40,7 @@ module.exports = ({ chunk }) => {
       }
       // domainFromInput is now startsWith a "."
 
-      if (domainFromInput.length >= domainFromFullSet.length) {
+      if (domainFromInputLen >= domainFromFullSet.length) {
         if (domainFromInput.endsWith(domainFromFullSet)) {
           outputToBeRemoved[i] = 1;
           break;