malware-filter/src/build.js

'use strict'

const { stream: gotStream } = require('got')
const unzip = require('extract-zip')
const { basename, join, parse: pathParse } = require('path')
const { mkdir } = require('fs/promises')
const { createWriteStream } = require('fs')
const { pipeline } = require('stream/promises')
const envVar = process.env

const rootPath = join(__dirname, '..')
const tmpPath = join(rootPath, 'tmp')
const publicPath = join(rootPath, 'public')
const artifacts = {
  'https://gitlab.com/curben/urlhaus-filter/-/jobs/artifacts/main/download?job=pages': 'urlhaus-filter.zip',
  'https://gitlab.com/curben/phishing-filter/-/jobs/artifacts/main/download?job=pages': 'phishing-filter.zip',
  'https://gitlab.com/curben/pup-filter/-/jobs/artifacts/main/download?job=pages': 'pup-filter.zip',
  'https://gitlab.com/curben/tracking-filter/-/jobs/artifacts/main/download?job=pages': 'tracking-filter.zip'
}
const ghMirror = {
  'urlhaus-filter': 'https://nightly.link/curbengh/urlhaus-filter/workflows/pages/main/public.zip',
  'phishing-filter': 'https://nightly.link/curbengh/phishing-filter/workflows/pages/main/public.zip',
  'pup-filter': 'https://nightly.link/curbengh/pup-filter/workflows/pages/main/public.zip',
  'tracking-filter': 'https://nightly.link/curbengh/tracking-filter/workflows/pages/main/public.zip'
}
const oisdFilters = {
  'https://abp.oisd.nl/basic/': 'oisd_abp_light.txt',
  'https://abp.oisd.nl/': 'oisd_abp.txt',
  'https://dbl.oisd.nl/basic/': 'oisd_dbl_light.txt',
  'https://dbl.oisd.nl/': 'oisd_dbl.txt',
  'https://dblw.oisd.nl/': 'oisd_dblw_light.txt',
  'https://dblw.oisd.nl/basic/': 'oisd_dblw.txt',
  'https://hosts.oisd.nl/basic/': 'oisd_hosts_light.txt',
  'https://hosts.oisd.nl/': 'oisd_hosts.txt',
  'https://dnsmasq.oisd.nl/basic/': 'oisd_dnsmasq_light.txt',
  'https://dnsmasq.oisd.nl/': 'oisd_dnsmasq.txt',
  'https://rpz.oisd.nl/basic/': 'oisd_rpz_light.txt',
  'https://rpz.oisd.nl/': 'oisd_rpz.txt',
}

const dl = async (link, filename) => {
  const zipPath = join(tmpPath, filename)
  let isMirror = false

  console.log(`Downloading ${filename} from "${link}"`)
  try {
    await pipeline(
      gotStream(link),
      createWriteStream(zipPath)
    )
  } catch ({ message }) {
    console.error(JSON.stringify({
      error: message,
      link,
      filename
    }))

    const mirrorLink = ghMirror[pathParse(filename).name]
    console.log(`Downloading ${filename} from "${mirrorLink}"`)
    isMirror = true

    try {
      await pipeline(
        gotStream(mirrorLink),
        createWriteStream(zipPath)
      )
    } catch ({ message }) {
      throw new Error(JSON.stringify({
        error: message,
        link: mirrorLink,
        filename
      }))
    }
  }

  console.log(`Extracting ${basename(zipPath)}...`)
  if (isMirror === false) {
    await unzip(zipPath, { dir: rootPath })
  } else {
    await unzip(zipPath, { dir: publicPath })
  }
}

const oisdDl = async (link, filename) => {
  const txtPath = join(publicPath, filename)
  console.log(`Downloading ${filename} from "${link}"`)
  try {
    await pipeline(
      gotStream(link),
      createWriteStream(txtPath)
    )
  } catch ({ message }) {
    console.error(JSON.stringify({
      error: message,
      link,
      filename
    }))
  }
}

const f = async () => {
  await mkdir(tmpPath, { recursive: true })
  await mkdir(publicPath, { recursive: true })
  await Promise.all(Object.entries(artifacts).map(([link, filename]) => { return dl(link, filename) }))
}

const oisd = async () => {
  await mkdir(publicPath, { recursive: true })
  if (Object.prototype.hasOwnProperty.call(envVar, 'CI_PROJECT_PATH') && envVar.CI_PROJECT_PATH === 'curben/malware-filter') {
    await Promise.all(Object.entries(oisdFilters).map(([link, filename]) => { return oisdDl(link, filename) }))
  }
}

f()
oisd()