Surge_by_SukkaW/Build/worker/build-reject-domainset-worker.js
2022-10-08 17:50:18 +08:00

63 lines
1.8 KiB
JavaScript

const Piscina = require('piscina');
// pre check if fullset domain is starts with a "."
// This avoid calling chatCodeAt repeatedly
const { canExcludeFromDedupe } = require('../lib/parse-filter')
// workerData is an array of string, sorted by length, short first
const fullsetDomainStartsWithADot = Piscina.workerData
const totalLen = fullsetDomainStartsWithADot.length;
module.exports.dedupe = ({ chunk }) => {
const chunkLength = chunk.length;
const outputToBeRemoved = new Int8Array(chunkLength);
for (let i = 0; i < chunkLength; i++) {
const domainFromInput = chunk[i];
if (canExcludeFromDedupe(domainFromInput)) {
continue;
}
for (let j = 0; j < totalLen; j++) {
const domainFromFullSet = fullsetDomainStartsWithADot[j];
// domainFromFullSet is now startsWith a "."
if (domainFromFullSet === domainFromInput) continue;
const domainFromInputLen = domainFromInput.length;
const domainFromFullSetLen = domainFromFullSet.length;
// !domainFromInput.starsWith('.') && `.${domainFromInput}` === domainFromFullSet
if (domainFromInput.charCodeAt(0) !== 46) {
if (domainFromInputLen + 1 === domainFromFullSetLen) {
let shouldBeRemoved = true;
for (let k = 0; k < domainFromInputLen; k++) {
if (domainFromFullSet.charCodeAt(k + 1) !== domainFromInput.charCodeAt(k)) {
shouldBeRemoved = false;
break;
}
}
if (shouldBeRemoved) {
outputToBeRemoved[i] = 1;
break;
}
}
}
// domainFromInput is now startsWith a "."
if (domainFromInputLen >= domainFromFullSetLen) {
if (domainFromInput.endsWith(domainFromFullSet)) {
outputToBeRemoved[i] = 1;
break;
}
}
}
}
return Piscina.move(outputToBeRemoved);
};