Make build faster

This commit is contained in:
SukkaW 2023-05-26 23:43:57 +08:00
parent 72f8d06bd2
commit c12a75b00a
3 changed files with 119 additions and 126 deletions

View File

@ -1,3 +1,4 @@
// @ts-check
const { promises: fsPromises } = require('fs');
const fse = require('fs-extra');
const { resolve: pathResolve } = require('path');
@ -12,7 +13,12 @@ const { HOSTS, ADGUARD_FILTERS, PREDEFINED_WHITELIST, PREDEFINED_ENFORCED_BACKLI
const { withBannerArray } = require('./lib/with-banner');
const { compareAndWriteFile } = require('./lib/string-array-compare');
/** Whitelists */
const filterRuleWhitelistDomainSets = new Set(PREDEFINED_WHITELIST);
/** @type {Set<string>} Dedupe domains inclued by DOMAIN-KEYWORD */
const domainKeywordsSet = new Set();
/** @type {Set<string>} Dedupe domains included by DOMAIN-SUFFIX */
const domainSuffixSet = new Set();
(async () => {
console.time('Total Time - build-reject-domain-set');
@ -46,7 +52,7 @@ const filterRuleWhitelistDomainSets = new Set(PREDEFINED_WHITELIST);
await Promise.all(ADGUARD_FILTERS.map(input => {
const promise = typeof input === 'string'
? processFilterRules(input, undefined, false)
: processFilterRules(input[0], input[1] ?? undefined, input[2] ?? false)
: processFilterRules(input[0], input[1] || undefined, input[2] ?? false)
return promise.then((i) => {
if (i) {
@ -118,17 +124,13 @@ const filterRuleWhitelistDomainSets = new Set(PREDEFINED_WHITELIST);
});
});
// Copy reject_sukka.conf for backward compatibility
await fse.copy(pathResolve(__dirname, '../Source/domainset/reject_sukka.conf'), pathResolve(__dirname, '../List/domainset/reject_sukka.conf'))
previousSize = domainSets.size - previousSize;
console.log(`Import ${previousSize} rules from reject_sukka.conf!`);
// Read DOMAIN Keyword
const domainKeywordsSet = new Set();
const domainSuffixSet = new Set();
await fsPromises.readFile(pathResolve(__dirname, '../List/non_ip/reject.conf'), { encoding: 'utf-8' }).then(data => {
await Promise.all([
// Copy reject_sukka.conf for backward compatibility
fse.copy(pathResolve(__dirname, '../Source/domainset/reject_sukka.conf'), pathResolve(__dirname, '../List/domainset/reject_sukka.conf')),
fsPromises.readFile(pathResolve(__dirname, '../List/non_ip/reject.conf'), { encoding: 'utf-8' }).then(data => {
data.split('\n').forEach(line => {
if (line.startsWith('DOMAIN-KEYWORD')) {
const [, ...keywords] = line.split(',');
@ -138,10 +140,9 @@ const filterRuleWhitelistDomainSets = new Set(PREDEFINED_WHITELIST);
domainSuffixSet.add(keywords.join(',').trim());
}
});
});
}),
// Read Special Phishing Suffix list
await fsPromises.readFile(pathResolve(__dirname, '../List/domainset/reject_phishing.conf'), { encoding: 'utf-8' }).then(data => {
fsPromises.readFile(pathResolve(__dirname, '../List/domainset/reject_phishing.conf'), { encoding: 'utf-8' }).then(data => {
data.split('\n').forEach(line => {
const trimmed = line.trim();
if (
@ -156,7 +157,8 @@ const filterRuleWhitelistDomainSets = new Set(PREDEFINED_WHITELIST);
domainSuffixSet.add(trimmed);
});
});
})
]);
console.log(`Import ${domainKeywordsSet.size} black keywords and ${domainSuffixSet.size} black suffixes!`);
@ -166,31 +168,7 @@ const filterRuleWhitelistDomainSets = new Set(PREDEFINED_WHITELIST);
console.time(`* Dedupe from black keywords/suffixes`);
for (const domain of domainSets) {
let isTobeRemoved = false;
for (const suffix of domainSuffixSet) {
if (domain.endsWith(suffix)) {
isTobeRemoved = true;
break;
}
}
if (!isTobeRemoved) {
for (const keyword of domainKeywordsSet) {
if (domain.includes(keyword)) {
isTobeRemoved = true;
break;
}
}
}
if (!isTobeRemoved) {
if (isInWhiteList(domain)) {
isTobeRemoved = true;
}
}
if (isTobeRemoved) {
if (isMatchKeyword(domain) || isMatchSuffix(domain) || isInWhiteList(domain)) {
domainSets.delete(domain);
}
}
@ -204,37 +182,30 @@ const filterRuleWhitelistDomainSets = new Set(PREDEFINED_WHITELIST);
const START_TIME = Date.now();
const domainSetsArray = Array.from(domainSets);
const piscina = new Piscina({
filename: pathResolve(__dirname, 'worker/build-reject-domainset-worker.js'),
workerData: preprocessFullDomainSetBeforeUsedAsWorkerData([...domainSets]),
workerData: preprocessFullDomainSetBeforeUsedAsWorkerData(Array.from(domainSetsArray)),
idleTimeout: 50,
minThreads: threads,
maxThreads: threads
});
console.log(`Launching ${threads} threads...`)
console.log(preprocessFullDomainSetBeforeUsedAsWorkerData(Array.from(domainSetsArray)).length);
const tasksArray = Array.from(domainSets)
.reduce((result, element, index) => {
console.log(`Launching ${threads} threads...`);
const tasksArray = domainSetsArray.reduce((result, element, index) => {
const chunk = index % threads;
result[chunk] ??= [];
result[chunk].push(element);
return result;
}, []);
}, /** @type {string[][]} */([]));
(
await Promise.all(
Array.from(domainSets)
.reduce((result, element, index) => {
const chunk = index % threads;
result[chunk] ??= [];
result[chunk].push(element);
return result;
}, [])
.map(chunk => piscina.run({ chunk }, { name: 'dedupe' }))
)
).forEach((result, taskIndex) => {
(await Promise.all(
tasksArray.map(chunk => piscina.run({ chunk }))
)).forEach((result, taskIndex) => {
const chunk = tasksArray[taskIndex];
for (let i = 0, len = result.length; i < len; i++) {
if (result[i]) {
@ -259,7 +230,7 @@ const filterRuleWhitelistDomainSets = new Set(PREDEFINED_WHITELIST);
}
return 0;
};
const sortedDomainSets = [...domainSets]
const sortedDomainSets = Array.from(domainSets)
.map((v) => {
return { v, domain: getDomain(v.charCodeAt(0) === 46 ? v.slice(1) : v)?.toLowerCase() || v };
})
@ -296,6 +267,35 @@ const filterRuleWhitelistDomainSets = new Set(PREDEFINED_WHITELIST);
}
})();
/**
* @param {string} domain
*/
function isMatchKeyword(domain) {
for (const keyword of domainKeywordsSet) {
if (domain.includes(keyword)) {
return true;
}
}
return false;
}
/**
* @param {string} domain
*/
function isMatchSuffix(domain) {
for (const suffix of domainSuffixSet) {
if (domain.endsWith(suffix)) {
return true;
}
}
return false;
}
/**
* @param {string} domain
*/
function isInWhiteList(domain) {
for (const white of filterRuleWhitelistDomainSets) {
if (domain === white || domain.endsWith(white)) {

View File

@ -99,7 +99,7 @@ async function processHosts (hostsUrl, includeAllSubDomain = false) {
/**
* @param {string | URL} filterRulesUrl
* @param {(string | URL)[] | undefined} fallbackUrls
* @param {readonly (string | URL)[] | undefined} [fallbackUrls]
* @returns {Promise<{ white: Set<string>, black: Set<string>, foundDebugDomain: boolean, parseFailed: boolean }>}
*/
async function processFilterRules(filterRulesUrl, fallbackUrls, includeThirdParties = false) {
@ -390,10 +390,13 @@ async function processFilterRules (filterRulesUrl, fallbackUrls, includeThirdPar
};
}
/**
* @param {string[]} data
*/
function preprocessFullDomainSetBeforeUsedAsWorkerData(data) {
return data.filter(domain => (
domain.charCodeAt(0) === 46
));
return data
.filter(domain => domain.charCodeAt(0) === 46)
.sort((a, b) => a.length - b.length);
}

View File

@ -1,54 +1,44 @@
const Piscina = require('piscina');
const { isCI } = require('ci-info');
// const { isCI } = require('ci-info');
const fullsetDomainStartsWithADot = Piscina.workerData
const totalLen = fullsetDomainStartsWithADot.length;
const log = isCI ? () => { } : console.log.bind(console);
// const log = isCI ? () => { } : console.log.bind(console);
module.exports.dedupe = ({ chunk }) => {
module.exports = ({ chunk }) => {
const chunkLength = chunk.length;
const outputToBeRemoved = new Int8Array(chunkLength);
for (let i = 0; i < chunkLength; i++) {
const domainFromInput = chunk[i];
const domainFromInputChunk = chunk[i];
for (let j = 0; j < totalLen; j++) {
const domainFromFullSet = fullsetDomainStartsWithADot[j];
const domainStartsWithADotAndFromFullSet = fullsetDomainStartsWithADot[j];
// domainFromFullSet is always startsWith "."
if (domainStartsWithADotAndFromFullSet === domainFromInputChunk) continue;
if (domainFromFullSet === domainFromInput) continue;
const domainFromInputLen = domainFromInput.length;
const domainFromFullSetLen = domainFromFullSet.length;
const domainFromInputLen = domainFromInputChunk.length;
const domainFromFullSetLen = domainStartsWithADotAndFromFullSet.length;
if (domainFromInputLen < domainFromFullSetLen) {
if (domainFromInputLen + 1 === domainFromFullSetLen) {
// !domainFromInput.starsWith('.') && `.${domainFromInput}` === domainFromFullSet
if (domainFromInput[0] !== '.' && domainFromInputLen + 1 === domainFromFullSetLen) {
let shouldBeRemoved = true;
for (let k = 0; k < domainFromInputLen; k++) {
if (domainFromFullSet[k + 1] !== domainFromInput[k]) {
shouldBeRemoved = false;
break;
}
}
if (shouldBeRemoved) {
if (domainFromInputChunk.charCodeAt(0) !== 46 && domainFromInputChunk.endsWith(domainStartsWithADotAndFromFullSet)) {
outputToBeRemoved[i] = 1;
log(domainFromInput, domainFromFullSet)
// log(domainFromInputChunk, domainStartsWithADotAndFromFullSet)
break;
}
} else {
break;
}
if (domainFromInputLen > domainFromFullSetLen) {
// domainFromInput is now startsWith a "."
if (domainFromInput.endsWith(domainFromFullSet)) {
} else if (domainFromInputChunk.endsWith(domainStartsWithADotAndFromFullSet)) {
outputToBeRemoved[i] = 1;
log(domainFromInput, domainFromFullSet)
// log(domainFromInputChunk, domainStartsWithADotAndFromFullSet)
break;
}
}
}
}
return Piscina.move(outputToBeRemoved);
};