meta-rules-dat/.github/workflows/build.yml

name: Build V2Ray rules dat files
on:
  schedule:
    - cron: "0 22 * * *"
  push:
    branches:
      - master
      - hidden
    paths-ignore:
      - "**/README.md"
jobs:
  build:
    name: Build
    runs-on: ubuntu-latest
    steps:
      - name: Setup Go 1.x
        uses: actions/setup-go@v2
        with:
          go-version: "^1.14"

      - name: Set $GOPATH and more variables
        run: |
          echo "::set-env name=RELEASE_NAME::Released on $(date +%Y%m%d%H%M)"
          echo "::set-env name=TAG_NAME::$(date +%Y%m%d%H%M)"
          echo "::set-env name=GEOIP_REPO::github.com/Loyalsoldier/geoip"
          echo "::set-env name=GEOSITE_REPO::github.com/v2fly/domain-list-community"
          echo "::set-env name=CHINA_DOMAINS_URL::https://raw.githubusercontent.com/felixonmars/dnsmasq-china-list/master/accelerated-domains.china.conf"
          echo "::set-env name=GOOGLE_DOMAINS_URL::https://raw.githubusercontent.com/felixonmars/dnsmasq-china-list/master/google.china.conf"
          echo "::set-env name=APPLE_DOMAINS_URL::https://raw.githubusercontent.com/felixonmars/dnsmasq-china-list/master/apple.china.conf"
          echo "::set-env name=GFWLIST_DOMAINS_URL::https://cokebar.github.io/gfwlist2dnsmasq/gfwlist_domain.txt"
          echo "::set-env name=GFWLIST_EXTRAS_DOMAINS_URL::https://raw.githubusercontent.com/pexcn/gfwlist-extras/master/gfwlist-extras.txt"
          echo "::set-env name=GREATFIRE_DOMAINS_URL::https://raw.githubusercontent.com/Loyalsoldier/cn-blocked-domain/release/domains.txt"
          echo "::set-env name=GEQ1AN_RULES_APPLE_URL::https://raw.githubusercontent.com/GeQ1an/Rules/master/QuantumultX/Filter/Apple.list"
          echo "::set-env name=GEQ1AN_RULES_MICROSOFT_URL::https://raw.githubusercontent.com/GeQ1an/Rules/master/QuantumultX/Filter/Optional/Microsoft.list"
          echo "::set-env name=GEQ1AN_RULES_GLOBAL_MEDIA_URL::https://raw.githubusercontent.com/GeQ1an/Rules/master/QuantumultX/Filter/GMedia.list"
          echo "::set-env name=GEQ1AN_RULES_OUTSIDE_URL::https://raw.githubusercontent.com/GeQ1an/Rules/master/QuantumultX/Filter/Outside.list"
          echo "::set-env name=GEQ1AN_RULES_REJECT_URL::https://raw.githubusercontent.com/GeQ1an/Rules/master/QuantumultX/Filter/AdBlock.list"
          echo "::set-env name=LHIE1_RULES_PROXY_URL::https://raw.githubusercontent.com/lhie1/Rules/master/Surge/Surge%203/Provider/Proxy.list"
          echo "::set-env name=LHIE1_RULES_REJECT_URL::https://raw.githubusercontent.com/lhie1/Rules/master/Surge/Surge%203/Provider/Reject.list"
          echo "::set-env name=ADAWAY_REJECT_URL::https://raw.githubusercontent.com/AdAway/adaway.github.io/master/hosts.txt"
          echo "::set-env name=EASYLISTCHINA_EASYLIST_REJECT_URL::https://easylist-downloads.adblockplus.org/easylistchina+easylist.txt"
          echo "::set-env name=V2FLY_DIRECT::https://raw.githubusercontent.com/v2fly/domain-list-community/release/cn.txt"
          echo "::set-env name=V2FLY_PROXY::https://raw.githubusercontent.com/v2fly/domain-list-community/release/geolocation-!cn.txt"
          echo "::set-env name=V2FLY_REJECT::https://raw.githubusercontent.com/v2fly/domain-list-community/release/category-ads-all.txt"
          echo "::set-env name=GOPATH::$(dirname $GITHUB_WORKSPACE)"
          echo "::add-path::$(dirname $GITHUB_WORKSPACE)/bin"
        shell: bash

      - name: Checkout the "hidden" branch
        uses: actions/checkout@v2.3.1
        with:
          ref: hidden

      - name: Get GeoLite2 zip file
        run: |
          curl -L -o GeoLite2-Country-CSV.zip "https://download.maxmind.com/app/geoip_download?edition_id=GeoLite2-Country-CSV&license_key=JvbzLLx7qBZT&suffix=zip"
          unzip GeoLite2-Country-CSV.zip
          rm -f GeoLite2-Country-CSV.zip
          mv GeoLite2* geoip

      - name: Generate geoip.dat file
        run: |
          go get -v -insecure $GEOIP_REPO
          geoip --country=./geoip/GeoLite2-Country-Locations-en.csv --ipv4=./geoip/GeoLite2-Country-Blocks-IPv4.csv --ipv6=./geoip/GeoLite2-Country-Blocks-IPv6.csv --ipv4CN=https://raw.githubusercontent.com/pexcn/daily/gh-pages/chnroute/chnroute.txt

      - name: Download geosite project
        run: |
          go get -v -insecure $GEOSITE_REPO

      - name: Get and add direct domains into temp-direct.txt file
        run: |
          curl -sSL $CHINA_DOMAINS_URL | perl -ne '/^server=\/([^\/]+)\// && print "$1\n"' > temp-direct.txt
          curl -sSL ${V2FLY_DIRECT} | perl -ne '/^(domain|full):([^:]+)(\n$|:@.+)/ && print "$2\n"' >> temp-direct.txt

      - name: Get and add proxy domains into temp-proxy.txt file
        run: |
          curl -sSL $GFWLIST_DOMAINS_URL | perl -ne '/^((?=^.{3,255})[a-zA-Z0-9][-_a-zA-Z0-9]{0,62}(\.[a-zA-Z0-9][-_a-zA-Z0-9]{0,62})+)/ && print "$1\n"' > temp-proxy.txt
          curl -sSL $GFWLIST_EXTRAS_DOMAINS_URL | perl -ne 'print if not /^(#|\s)/' >> temp-proxy.txt
          curl -sSL $GREATFIRE_DOMAINS_URL | perl -ne '/^((?=^.{3,255})[a-zA-Z0-9][-_a-zA-Z0-9]{0,62}(\.[a-zA-Z0-9][-_a-zA-Z0-9]{0,62})+)/ && print "$1\n"' >> temp-proxy.txt
          curl -sSL $GOOGLE_DOMAINS_URL | perl -ne '/^server=\/([^\/]+)\// && print "$1\n"' >> temp-proxy.txt
          curl -sSL $APPLE_DOMAINS_URL | perl -ne '/^server=\/([^\/]+)\// && print "$1\n"' >> temp-proxy.txt
          curl -sSL $GEQ1AN_RULES_APPLE_URL | awk -F ',' '/^(HOST|DOMAIN)(,|\-SUFFIX,).+Apple/ {print $2}' >> temp-proxy.txt
          curl -sSL $GEQ1AN_RULES_MICROSOFT_URL | awk -F ',' '/^(HOST|DOMAIN)(,|\-SUFFIX,).+Microsoft/ {print $2}' >> temp-proxy.txt
          curl -sSL $GEQ1AN_RULES_GLOBAL_MEDIA_URL | awk -F ',' '/^(HOST|DOMAIN)(,|\-SUFFIX,).+GMedia/ {print $2}' >> temp-proxy.txt
          curl -sSL $GEQ1AN_RULES_OUTSIDE_URL | awk -F ',' '/^(HOST|DOMAIN)(,|\-SUFFIX,).+Outside/ {print $2}' >> temp-proxy.txt
          curl -sSL $LHIE1_RULES_PROXY_URL | awk -F ',' '/^(HOST|DOMAIN)(,|\-SUFFIX,).+/ {print $2}' >> temp-proxy.txt
          curl -sSL ${V2FLY_PROXY} | perl -ne '/^(domain|full):([^:]+)(\n$|:@.+)/ && print "$2\n"' >> temp-proxy.txt

      - name: Get and add reject domains into temp-reject.txt file
        run: |
          curl -sSL $GEQ1AN_RULES_REJECT_URL | awk -F ',' '/^(HOST|DOMAIN)(,|\-SUFFIX,).+AdBlock/ {print $2}' > temp-reject.txt
          curl -sSL $LHIE1_RULES_REJECT_URL | awk -F ',' '/^DOMAIN(,|\-SUFFIX,).+/ {print $2}' >> temp-reject.txt
          curl -sSL $ADAWAY_REJECT_URL | grep "127.0.0.1" | sed '1d' | awk '{print $2}' >> temp-reject.txt
          curl -sSL $EASYLISTCHINA_EASYLIST_REJECT_URL | perl -ne '/^\|\|([-_0-9a-zA-Z]+(\.[-_0-9a-zA-Z]+){1,64})\^($|\$third-party$)/ && print "$1\n"' | perl -ne 'print if not /^[0-9]{1,3}(\.[0-9]{1,3}){3}$/' >> temp-reject.txt
          curl -sSL ${V2FLY_REJECT} | perl -ne '/^(domain|full):([^:]+)(\n$|:@.+)/ && print "$2\n"' >> temp-reject.txt

      - name: Reserve `regex` and `keyword` type of rules from v2fly lists to "reserve" files
        run: |
          curl -sSL ${V2FLY_DIRECT} | perl -ne '/^((regex|keyword):[^:]+)(\n$|:@.+)/ && print "$1\n"' > direct-reserve.txt
          curl -sSL ${V2FLY_PROXY} | perl -ne '/^((regex|keyword):[^:]+)(\n$|:@.+)/ && print "$1\n"' > proxy-reserve.txt
          curl -sSL ${V2FLY_REJECT} | perl -ne '/^((regex|keyword):[^:]+)(\n$|:@.+)/ && print "$1\n"' > reject-reserve.txt

      - name: Add proxy, direct and reject domains from "hidden" branch to appropriate temp files
        run: |
          cat proxy.txt >> temp-proxy.txt
          cat direct.txt >> temp-direct.txt
          cat reject.txt >> temp-reject.txt

      - name: Sort and generate redundant lists
        run: |
          cat temp-proxy.txt | sort --ignore-case -u > proxy-list-with-redundant
          cat temp-direct.txt | sort --ignore-case -u > direct-list-with-redundant
          cat temp-reject.txt | sort --ignore-case -u > reject-list-with-redundant

      - name: Remove redundant domains
        run: |
          chmod +x findRedundantDomain.py
          ./findRedundantDomain.py ./direct-list-with-redundant ./direct-list-deleted-unsort
          ./findRedundantDomain.py ./proxy-list-with-redundant ./proxy-list-deleted-unsort
          ./findRedundantDomain.py ./reject-list-with-redundant ./reject-list-deleted-unsort
          [ ! -f "direct-list-deleted-unsort" ] && touch direct-list-deleted-unsort
          [ ! -f "proxy-list-deleted-unsort" ] && touch proxy-list-deleted-unsort
          [ ! -f "reject-list-deleted-unsort" ] && touch reject-list-deleted-unsort
          sort ./direct-list-deleted-unsort > ./direct-list-deleted-sort
          sort ./proxy-list-deleted-unsort > ./proxy-list-deleted-sort
          sort ./reject-list-deleted-unsort > ./reject-list-deleted-sort
          diff ./direct-list-deleted-sort ./direct-list-with-redundant | awk '/^>/{print $2}' > ./direct-list-without-redundant
          diff ./proxy-list-deleted-sort ./proxy-list-with-redundant | awk '/^>/{print $2}' > ./proxy-list-without-redundant
          diff ./reject-list-deleted-sort ./reject-list-with-redundant | awk '/^>/{print $2}' > ./reject-list-without-redundant

      - name: Remove domains from "need-to-remove" lists in "hidden" branch
        run: |
          diff ./direct-need-to-remove.txt ./direct-list-without-redundant | awk '/^>/{print $2}' > temp-cn.txt
          diff ./proxy-need-to-remove.txt ./proxy-list-without-redundant | awk '/^>/{print $2}' > temp-geolocation-\!cn.txt
          diff ./reject-need-to-remove.txt ./reject-list-without-redundant | awk '/^>/{print $2}' > temp-category-ads-all.txt

      - name: Remove domains end with ".cn" in "temp-geolocation-!cn.txt" and write lists to data directory
        run: |
          cat temp-cn.txt | sort --ignore-case -u | perl -ne '/^((?=^.{3,255})[a-zA-Z0-9][-_a-zA-Z0-9]{0,62}(\.[a-zA-Z0-9][-_a-zA-Z0-9]{0,62})*)/ && print "$1\n"' > $GOPATH/src/$GEOSITE_REPO/data/cn
          cat temp-cn.txt | sort --ignore-case -u | perl -ne 'print if not /^((?=^.{3,255})[a-zA-Z0-9][-_a-zA-Z0-9]{0,62}(\.[a-zA-Z0-9][-_a-zA-Z0-9]{0,62})+)/' > direct-tld-list.txt
          cat temp-geolocation-\!cn.txt | sort --ignore-case -u | perl -ne '/^((?=^.{3,255})[a-zA-Z0-9][-_a-zA-Z0-9]{0,62}(\.[a-zA-Z0-9][-_a-zA-Z0-9]{0,62})*)/ && print "$1\n"' | perl -ne 'print if not /\.cn$/' > $GOPATH/src/$GEOSITE_REPO/data/geolocation-\!cn
          cat temp-geolocation-\!cn.txt | sort --ignore-case -u | perl -ne 'print if not /^((?=^.{3,255})[a-zA-Z0-9][-_a-zA-Z0-9]{0,62}(\.[a-zA-Z0-9][-_a-zA-Z0-9]{0,62})+)/' > proxy-tld-list.txt
          cat temp-category-ads-all.txt | sort --ignore-case -u | perl -ne '/^((?=^.{3,255})[a-zA-Z0-9][-_a-zA-Z0-9]{0,62}(\.[a-zA-Z0-9][-_a-zA-Z0-9]{0,62})*)/ && print "$1\n"' > $GOPATH/src/$GEOSITE_REPO/data/category-ads-all
          cat temp-category-ads-all.txt | sort --ignore-case -u | perl -ne 'print if not /^((?=^.{3,255})[a-zA-Z0-9][-_a-zA-Z0-9]{0,62}(\.[a-zA-Z0-9][-_a-zA-Z0-9]{0,62})+)/' > reject-tld-list.txt

      - name: Add `regex` and `keyword` type of rules back into "cn", "geolocation-!cn" and "category-ads-all" list
        run: |
          [ -f "direct-reserve.txt" ] && cat direct-reserve.txt >> $GOPATH/src/$GEOSITE_REPO/data/cn
          [ -f "proxy-reserve.txt" ] && cat proxy-reserve.txt >> $GOPATH/src/$GEOSITE_REPO/data/geolocation-\!cn
          [ -f "reject-reserve.txt" ] && cat reject-reserve.txt >> $GOPATH/src/$GEOSITE_REPO/data/category-ads-all
          cp $GOPATH/src/$GEOSITE_REPO/data/cn direct-list.txt
          cp $GOPATH/src/$GEOSITE_REPO/data/geolocation-\!cn proxy-list.txt
          cp $GOPATH/src/$GEOSITE_REPO/data/category-ads-all reject-list.txt

      - name: Add `google-cn` and `apple-cn` sub-lists for custom routing settings for third-party users due to accessibility in China mainland
        run: |
          curl -sSL $GOOGLE_DOMAINS_URL | perl -ne '/^server=\/([^\/]+)\// && print "$1\n"' > $GOPATH/src/$GEOSITE_REPO/data/google-cn
          curl -sSL $APPLE_DOMAINS_URL | perl -ne '/^server=\/([^\/]+)\// && print "$1\n"' > $GOPATH/src/$GEOSITE_REPO/data/apple-cn

      - name: Build geosite.dat file
        run: |
          domain-list-community --datapath=${{ env.GOPATH }}/src/${{ env.GEOSITE_REPO }}/data

      - name: Move and zip files and generate sha256 hash
        run: |
          install -Dp geoip.dat ./publish/geoip.dat
          install -Dp dlc.dat ./publish/geosite.dat
          install -p {proxy,direct,reject}-tld-list.txt ./publish/
          install -p {proxy,direct,reject}-list.txt ./publish/
          cd ./publish
          zip rules.zip {proxy,direct,reject}-list.txt geoip.dat geosite.dat
          sha256sum geoip.dat > geoip.dat.sha256sum
          sha256sum geosite.dat > geosite.dat.sha256sum
          sha256sum rules.zip > rules.zip.sha256sum

      - name: Release and upload assets
        uses: softprops/action-gh-release@v1
        with:
          name: ${{ env.RELEASE_NAME }}
          tag_name: ${{ env.TAG_NAME }}
          draft: false
          prerelease: false
          files: |
            ./publish/*
        env:
          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}

      - name: Git push assets to "release" branch
        run: |
          cd publish
          git init
          git config --local user.name "actions"
          git config --local user.email "action@github.com"
          git checkout -b release
          git add .
          git commit -m "${{ env.RELEASE_NAME }}"
          git remote add origin "https://${{ github.actor }}:${{ secrets.GITHUB_TOKEN }}@github.com/${{ github.repository }}"
          git push -f -u origin release