Perf: improve reject set dedupe worker performance

Build/build-reject-domainset.js

@@ -198,6 +198,17 @@ const threads = require('os').cpus().length - 1;
     workerData: [...domainSets]
   });
 
+  console.log(`Launching ${threads} threads...`)
+
+  const tasksArray = Array.from(domainSets)
+    .reduce((result, element, index) => {
+      const chunk = index % threads;
+      result[chunk] ??= [];
+
+      result[chunk].push(element);
+      return result;
+    }, []);
+
   (await Promise.all(
     Array.from(domainSets)
       .reduce((result, element, index) => {
@@ -208,11 +219,15 @@ const threads = require('os').cpus().length - 1;
         return result;
       }, [])
       .map(chunk => piscina.run(
-        { chunk },
-        { name: 'dedupe' }
+        { chunk }
       ))
-  )).forEach(set => {
-    set.forEach(i => domainSets.delete(i));
+  )).forEach((result, taskIndex) => {
+    const chunk = tasksArray[taskIndex];
+    result.forEach((value, index) => {
+      if (value === 1) {
+        domainSets.delete(chunk[index])
+      }
+    })
   });
 
   console.log(`Deduped ${previousSize - domainSets.size} rules!`);

Build/worker/build-reject-domainset-worker.js

@@ -1,11 +1,12 @@
-const { workerData } = require('piscina');
+const { workerData, move } = require('piscina');
 
 const len = workerData.length;
 
-exports.dedupe = ({ chunk }) => {
-  const outputToBeRemoved = new Set();
+module.exports = ({ chunk }) => {
+  const chunkLength = chunk.length;
+  const outputToBeRemoved = new Int32Array(chunkLength);
 
-  for (let i = 0, l = chunk.length; i < l; i++) {
+  for (let i = 0; i < chunkLength; i++) {
     const domainFromInput = chunk[i];
 
     for (let j = 0; j < len; j++) {
@@ -26,7 +27,7 @@ exports.dedupe = ({ chunk }) => {
         }
 
         if (shouldBeRemoved) {
-          outputToBeRemoved.add(domainFromInput);
+          outputToBeRemoved[i] = 1;
           break;
         }
       }
@@ -34,12 +35,12 @@ exports.dedupe = ({ chunk }) => {
 
       if (domainFromInput.length >= domainFromFullSet.length) {
         if (domainFromInput.endsWith(domainFromFullSet)) {
-          outputToBeRemoved.add(domainFromInput);
+          outputToBeRemoved[i] = 1;
           break;
         }
       }
     }
   }
 
-  return outputToBeRemoved;
+  return move(outputToBeRemoved);
 };