⚓ Filter | Parse 🔰 #8238
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| name: ⚓ Filter | Parse 🔰 | |
| # This will likely increase as Repo Size gets bigger over time | |
| #MAX_RUNTIME: 30-45 Mins | |
| on: | |
| workflow_dispatch: # Whenever manually run from actions-tab | |
| schedule: | |
| #- cron: "45 23 * * *" #( 11:45 PM UTC --> 05:30 AM Morning ) | |
| - cron: "45 * * * *" #( Every 45 Mins ) | |
| #------------------------------------------------------------------------------------# | |
| env: | |
| RCLONE_CF_R2_PUB: "${{ secrets.RCLONE_CF_R2_PUB }}" | |
| #https://github.com/Azathothas/Wordlists/tree/main/Misc#user-agents | |
| SSH_META: "${{ secrets.SF_SSH_HOSTS }}" | |
| #------------------------------------------------------------------------------------# | |
| jobs: | |
| Parse: | |
| runs-on: ubuntu-latest | |
| timeout-minutes: 45 # Can't afford to finish the 2000 Minutes quotes | |
| permissions: | |
| contents: write | |
| steps: | |
| - name: Debloat Runner | |
| run: | | |
| #Presets | |
| set -x ; set +e | |
| #--------------# | |
| bash <(curl -qfsSL "https://raw.github.com/Azathothas/Arsenal/main/misc/Github/Runners/Ubuntu/debloat.sh") | |
| continue-on-error: true | |
| - name: Checkout repository | |
| uses: actions/checkout@v4 | |
| with: | |
| path: "main" | |
| filter: "blob:none" #https://github.blog/2020-12-21-get-up-to-speed-with-partial-clone-and-shallow-clone/ | |
| - name: Setup Env | |
| run: | | |
| ##presets | |
| set +x ; set +e | |
| #-------------# | |
| ##CoreUtils | |
| sudo apt update -y | |
| sudo apt install bc coreutils curl dos2unix fdupes jq moreutils wget -y | |
| sudo apt-get install apt-transport-https apt-utils ca-certificates coreutils dos2unix gnupg2 jq moreutils p7zip-full rename rsync software-properties-common texinfo tmux util-linux wget -y 2>/dev/null ; sudo apt-get update -y 2>/dev/null | |
| ##Setup rClone | |
| mkdir -p "$HOME/.config/rclone" | |
| echo "${{ secrets.RCLONE_CF_R2_PUB }}" > "$HOME/.config/rclone/rclone.conf" | |
| ##User-Agent | |
| USER_AGENT="$(curl -qfsSL 'https://pub.ajam.dev/repos/Azathothas/Wordlists/Misc/User-Agents/ua_chrome_macos_latest.txt')" && export USER_AGENT="$USER_AGENT" | |
| echo "USER_AGENT=$USER_AGENT" >> "$GITHUB_ENV" | |
| continue-on-error: true | |
| - name: Install eget | |
| run: | | |
| # Presets | |
| set -x ; set +e | |
| #--------------# | |
| # eget for bins | |
| sudo wget "https://bin.ajam.dev/x86_64_Linux/eget" -O "/usr/local/bin/eget" | |
| sudo chmod +xwr "/usr/local/bin/eget" | |
| continue-on-error: false | |
| - name: Install Deps | |
| run: | | |
| # Presets | |
| set -x ; set +e | |
| #--------------# | |
| #7z | |
| sudo eget "https://bin.ajam.dev/x86_64_Linux/7z" --to "/usr/local/bin/7z" | |
| sudo eget "https://bin.ajam.dev/x86_64_Linux/7z" --to "/usr/bin/7z" | |
| #anew | |
| sudo eget "https://bin.ajam.dev/x86_64_Linux/anew" --to "/usr/local/bin/anew" | |
| #anew-rs | |
| sudo eget "https://bin.ajam.dev/x86_64_Linux/anew-rs" --to "/usr/local/bin/anew-rs" | |
| #asn | |
| sudo eget "https://bin.ajam.dev/x86_64_Linux/asn" --to "/usr/local/bin/asn" | |
| #asnmap | |
| sudo eget "https://bin.ajam.dev/x86_64_Linux/asnmap" --to "/usr/local/bin/asnmap" | |
| #csvtk | |
| sudo eget "https://bin.ajam.dev/x86_64_Linux/csvtk" --to "/usr/local/bin/csvtk" | |
| #cut-cdn | |
| sudo eget "https://bin.ajam.dev/x86_64_Linux/cut-cdn" --to "/usr/local/bin/cut-cdn" | |
| #dasel | |
| sudo eget "https://bin.ajam.dev/x86_64_Linux/dasel" --to "/usr/local/bin/dasel" | |
| #dirstat-rs for tree | |
| sudo eget "https://bin.ajam.dev/x86_64_Linux/ds" --to "/usr/local/bin/ds" | |
| #dnsx | |
| sudo eget "https://bin.ajam.dev/x86_64_Linux/dnsx" --to "/usr/local/bin/dnsx" | |
| #HTTPx | |
| sudo eget "https://bin.ajam.dev/x86_64_Linux/httpx" --to "/usr/local/bin/httpx" | |
| #inscope | |
| sudo eget "https://bin.ajam.dev/x86_64_Linux/inscope" --to "/usr/local/bin/inscope" | |
| #jq | |
| sudo eget "https://bin.ajam.dev/x86_64_Linux/jq" --to "/usr/local/bin/jq" | |
| sudo eget "https://bin.ajam.dev/x86_64_Linux/jq" --to "/usr/bin/jq" | |
| #mapcidr | |
| sudo eget "https://bin.ajam.dev/x86_64_Linux/mapcidr" --to "/usr/local/bin/mapcidr" | |
| #scopegen | |
| sudo eget "https://bin.ajam.dev/x86_64_Linux/scopegen" --to "/usr/local/bin/scopegen" | |
| #scopeview | |
| sudo eget "https://bin.ajam.dev/x86_64_Linux/scopeview" --to "/usr/local/bin/scopeview" | |
| #spk | |
| sudo eget "https://bin.ajam.dev/x86_64_Linux/spk" --to "/usr/local/bin/spk" | |
| #rclone | |
| sudo eget "https://bin.ajam.dev/x86_64_Linux/rclone" --to "/usr/local/bin/rclone" | |
| #yq | |
| sudo eget "https://bin.ajam.dev/x86_64_Linux/yq" --to "/usr/local/bin/yq" | |
| continue-on-error: true | |
| - name: Archive Stales | |
| run: | | |
| # Presets | |
| set -x ; set +e | |
| #--------------# | |
| #Create Archive Dir | |
| mkdir -p "$GITHUB_WORKSPACE/main/Raw/Archive" ; cd "$GITHUB_WORKSPACE/main/Raw/Archive" | |
| #Remove +7 days | |
| #find "$GITHUB_WORKSPACE/main/Raw/Archive" -name "*.7z" -type f -mtime +8 -exec rm {} \; | |
| #Generate a range for the last 8 days | |
| for i in {0..7}; do TZ='Asia/Kathmandu' date -d "$i days ago" +'%Y_%m_%d'; done > "/tmp/7day_range.txt" | |
| #Purge | |
| find "$GITHUB_WORKSPACE/main/Raw/Archive" -type f -exec basename {} \; | grep -v -f "/tmp/7day_range.txt" | xargs -I {} echo ./{} | sort -u | xargs -I {} rm {} -rf | |
| # Find .7z files older than 24(+1) hours & Archive | |
| find "$GITHUB_WORKSPACE/main/Raw/Latest" -maxdepth 1 -type f ! -name "*$(date +'%Y_%m_%d')*" -exec mv {} "$GITHUB_WORKSPACE/main/Raw/Archive" \; | |
| #find "$GITHUB_WORKSPACE/main/Raw/Latest" -name "*.7z" -type f -mmin +1500 -exec mv {} "$GITHUB_WORKSPACE/main/Raw/Archive" \; | |
| # Update new time metadata | |
| find "$GITHUB_WORKSPACE/main/Raw/Archive" -name "*.7z" -exec touch {} \; | |
| continue-on-error: true | |
| - name: Parse New .7z | |
| run: | | |
| # Presets | |
| set -x ; set +e | |
| #--------------# | |
| #chdir to /tmp | |
| mkdir -p "/tmp/Unzipped" ; cd "/tmp/Unzipped" | |
| #Copy .7z | |
| cp $GITHUB_WORKSPACE/main/Raw/Latest/*.7z /tmp/Unzipped | |
| #Extract | |
| find "/tmp/Unzipped" -iname "certstream*.7z" -exec sh -c '7z x "{}" -o"$(dirname "{}")/$(basename "{}" .7z)"' \; | |
| #Del .7z | |
| find "/tmp/Unzipped" -type f -iname "certstream*.7z" -exec rm {} \; 2>/dev/null | |
| #Cat {} + | |
| #find "/tmp/Unzipped" -type f -iname "certstream_domains.txt" -exec cat {} \; 2>/dev/null | sort -u -o "/tmp/certstream_domains.txt" ; wc -l < "/tmp/certstream_domains.txt" | |
| find "/tmp/Unzipped" -type f -iname "certstream_domains.txt" -exec bash -c 'file -b --mime-type "$1" | grep -q "text/plain" && cat "$1"' bash {} \; 2>/dev/null | sort -u -o "/tmp/certstream_domains.txt" ; wc -l < "/tmp/certstream_domains.txt" | |
| #Stats | |
| echo -e "\n[+] Total Size (certstream_domains_latest.txt): $(du -h /tmp/certstream_domains.txt | awk '{print $1}')\n" | |
| continue-on-error: true | |
| - name: Parse Old .7z | |
| run: | | |
| # Presets | |
| set -x ; set +e | |
| #--------------# | |
| #chdir to /tmp | |
| mkdir -p "/tmp/Unzipped_7days" ; cd "/tmp/Unzipped_7days" | |
| #Copy .7z | |
| cp $GITHUB_WORKSPACE/main/Raw/Archive/*.7z /tmp/Unzipped_7days | |
| #Extract | |
| find "/tmp/Unzipped_7days" -iname "certstream*.7z" -exec sh -c '7z x "{}" -o"$(dirname "{}")/$(basename "{}" .7z)"' \; | |
| #Del .7z | |
| find "/tmp/Unzipped_7days" -type f -iname "certstream*.7z" -exec rm {} \; 2>/dev/null | |
| #Cat {} + | |
| #find "/tmp/Unzipped_7days" -type f -iname "certstream_domains.txt" -exec cat {} \; 2>/dev/null | sort -u -o "/tmp/certstream_domains_7days.txt" ; wc -l < "/tmp/certstream_domains_7days.txt" | |
| find "/tmp/Unzipped_7days" -type f -iname "certstream_domains.txt" -exec bash -c 'file -b --mime-type "$1" | grep -q "text/plain" && cat "$1"' bash {} \; 2>/dev/null | sort -u -o "/tmp/certstream_domains_7days.txt" ; wc -l < "/tmp/certstream_domains_7days.txt" | |
| #Stats | |
| export SSL_7DAYS_TOTAL="$(wc -l < /tmp/certstream_domains_7days.txt)" | |
| echo "SSL_7DAYS_TOTAL=$SSL_7DAYS_TOTAL" >> $GITHUB_ENV | |
| echo -e "\n[+] Total Subs (certstream_domains_7days.txt): $SSL_7DAYS_TOTAL" | |
| echo -e "[+] Total Size (certstream_domains_7days.txt): $(du -h /tmp/certstream_domains_7days.txt | awk '{print $1}')\n" | |
| continue-on-error: true | |
| - name: Parse External Sources (NRD) | |
| run: | | |
| # Presets | |
| set -x ; set +e | |
| #--------------# | |
| ##cubdomain | |
| #Get Base URL | |
| BASE_URL="$(curl -qfsSL "https://www.cubdomain.com/domains-registered-dates/1" -H "User-Agent: $USER_AGENT" | grep -oE 'https://www\.cubdomain\.com/domains-registered-by-date/[0-9]{4}-[0-9]{2}-[0-9]{2}/[0-9]+' | sort -u | sort -V | tail -n 1 | sed 's/\/[^/]*$//')" && export BASE_URL="$BASE_URL" && echo -e "\n[+] Base URL: $BASE_URL" | |
| #Get latest HTML | |
| curl -qfsSL "https://www.cubdomain.com/domains-registered-by-date/$(date +'%Y-%m-%d')/1" -H "User-Agent: $USER_AGENT" || curl -qfsSL "https://www.cubdomain.com/domains-registered-by-date/$(date -d 'yesterday' +'%Y-%m-%d')/1" -H "User-Agent: $USER_AGENT" > "/tmp/cubdomain_daily.html" | |
| curl -qfsSL "$BASE_URL/1" -H "User-Agent: $USER_AGENT" >> "/tmp/cubdomain_daily.html" | |
| #Get Number of Pages | |
| PAGES_NUM="$(grep -oE 'https://www\.cubdomain\.com/domains-registered-by-date/[0-9]{4}-[0-9]{2}-[0-9]{2}/[0-9]+' "/tmp/cubdomain_daily.html" | awk -F '/' '{print $NF}' | sort -u | sort -V | tail -n 1)" && export PAGES_NUM="$PAGES_NUM" && echo -e "\n[+] Total Pages: $PAGES_NUM" | |
| #Fetch | |
| for ((i=1; i<=$PAGES_NUM; i++)); do | |
| #echo -e "\n[+] Fetching Domains from : $BASE_URL/$i" | |
| curl -qfsSL "$BASE_URL/$i" -H "User-Agent: $USER_AGENT" | grep -o '<a href=https://www\.cubdomain\.com/site/[^>]*>[^<]*' | sed 's/.*>//' | |
| done > "/tmp/cubdomain_daily.txt" | |
| #Filter | |
| echo -e "\n[+] Parsing Domains" | |
| sed '/^\s*$/d' "/tmp/cubdomain_daily.txt" | sed 's/[[:space:]]//g' | sort -u -o "/tmp/cubdomain_daily.txt" | |
| echo -e "[+] Total Domains: $(wc -l < /tmp/cubdomain_daily.txt)" | |
| echo -e "[+] Total Size: $(du -sh /tmp/cubdomain_daily.txt | awk '{print $1}')" | |
| #anew | |
| cat "/tmp/cubdomain_daily.txt" | anew-rs -q "/tmp/certstream_domains.txt" | |
| cat "/tmp/cubdomain_daily.txt" | anew-rs -q "/tmp/nrd_latest.txt" | |
| #--------------# | |
| ##Shrestha: https://shreshtait.com/blog/2024/02/recently-registered-domains-download/ | |
| curl -qfsSL "https://raw.github.com/shreshta-labs/newly-registered-domains/main/nrd-1w.csv" | sed '/^\s*$/d' | sed 's/[[:space:]]//g' | sort -u -o "/tmp/shrestha_weekly.txt" | |
| #anew | |
| cat "/tmp/shrestha_weekly.txt" | anew-rs -q "/tmp/certstream_domains_7days.txt" | |
| cat "/tmp/shrestha_weekly.txt" | anew-rs -q "/tmp/nrd_weekly.txt" | |
| #--------------# | |
| # NOT RELIABLE | |
| # ##WhoisDS.com (https://github.com/PeterDaveHello/nrd-list-downloader) | |
| # pushd "$(mktemp -d)" > /dev/null 2>&1 | |
| # #Fetch | |
| # curl -qfsSL "https://pub.ajam.dev/repos/Azathothas/CertStream-Domains/.github/scripts/nrd-list-downloader.sh" | DAY_RANGE="2" bash | |
| # #cat | |
| # find . -type f -name '*.txt' -exec cat {} \; | sed '/^\s*$/d' | sed 's/[[:space:]]//g' | sort -u -o "/tmp/whoisds_daily.txt" | |
| # popd > /dev/null 2>&1 | |
| # #anew | |
| # cat "/tmp/whoisds_daily.txt" | anew-rs -q "/tmp/certstream_domains.txt" | |
| # cat "/tmp/whoisds_daily.txt" | anew-rs -q "/tmp/nrd_latest.txt" | |
| #--------------# | |
| ##maaaaz/dnsdumps | |
| #Fetch | |
| pushd "$(mktemp -d)" > /dev/null 2>&1 | |
| curl -qfsSLJO "https://raw.github.com/maaaaz/dnsdumps/main/daily/today_new.gz" | |
| curl -qfsSLJO "https://raw.github.com/maaaaz/dnsdumps/main/daily/yesterday_new.gz" | |
| #Parse | |
| find . -type f -name "*.gz" -exec gzip -d {} \; && find . -type f -name "*.gz" -exec rm {} \; | |
| find . -type f -exec cat {} \; | sed '/^\s*$/d' | sed 's/[[:space:]]//g' | sort -u -o "/tmp/maaaaz_daily.txt" | |
| #anew | |
| cat "/tmp/maaaaz_daily.txt" | anew-rs -q "/tmp/certstream_domains.txt" | |
| cat "/tmp/maaaaz_daily.txt" | anew-rs -q "/tmp/nrd_latest.txt" | |
| popd > /dev/null 2>&1 | |
| #--------------# | |
| ##SystemJargon | |
| #fetch | |
| curl -qfsSL "https://raw.github.com/SystemJargon/nrd-lists/main/nrd-list-7-days-agh.txt" | grep -oE '\|\|(.*?)\^' | sed 's/||\(.*\)\^/\1/' | sed '/^\s*$/d' | sed 's/[[:space:]]//g' | sort -u -o "/tmp/systemjargon_weekly.txt" | |
| #anew | |
| cat "/tmp/systemjargon_weekly.txt" | anew-rs -q "/tmp/certstream_domains_7days.txt" | |
| cat "/tmp/systemjargon_weekly.txt" | anew-rs -q "/tmp/nrd_weekly.txt" | |
| continue-on-error: true | |
| - name: 🖳🇨🇭 Get Runners Metadata 🔒🔑 | |
| run: | | |
| # Presets | |
| set +x ; set +e | |
| #--------------# | |
| echo '```console' > "$GITHUB_WORKSPACE/main/SERVERS.md" | |
| find "$GITHUB_WORKSPACE/main/.github/SERVERS" -type f -name '*meta.txt' -exec cat {} \; | tee -a "$GITHUB_WORKSPACE/main/SERVERS.md" | |
| echo '```' >> "$GITHUB_WORKSPACE/main/SERVERS.md" | |
| #Ping Runners | |
| echo '${{ secrets.SF_SSH_HOSTS }}' > "/tmp/sf.sh" | |
| dos2unix --quiet "/tmp/sf.sh" ; chmod +xwr "/tmp/sf.sh" | |
| tmux new-session -s "segfault" -d "bash /tmp/sf.sh" | |
| sleep 10 ; while tmux has-session -t "segfault" ; do sleep 1 ; done | |
| echo -e "\n\n" ; cat "/tmp/sf.log" ; echo -e "\n\n" | |
| continue-on-error: true | |
| - name: Update README.md | |
| run: | | |
| # Presets | |
| set -x ; set +e | |
| #--------------# | |
| cd "$GITHUB_WORKSPACE/main" | |
| #cat INFO | |
| #cat "$GITHUB_WORKSPACE/main/INFO.md" > "$GITHUB_WORKSPACE/main/README.md" | |
| echo -e "\n" > "$GITHUB_WORKSPACE/main/README.md" | |
| #Last 24 Hr | |
| echo '---' >> "$GITHUB_WORKSPACE/main/README.md" | |
| echo -e "- #### [🖨️ **Stats** \`24Hr\`⏲️ ➼ $(date +'%Y_%m_%d')](https://pub.ajam.dev/datasets/certstream/all_latest.txt)" >> "$GITHUB_WORKSPACE/main/README.md" | |
| echo '```console' >> "$GITHUB_WORKSPACE/main/README.md" | |
| echo -e "" >> "$GITHUB_WORKSPACE/main/README.md" | |
| #Total | |
| echo -e "\n--> 🌐 Total" >> "$GITHUB_WORKSPACE/main/README.md" | |
| echo -e "[+] New/ReNewed SSL Certs (ALL): +$(wc -l < /tmp/certstream_domains.txt)" >> "$GITHUB_WORKSPACE/main/README.md" | |
| echo -e "[+] View/Download: https://pub.ajam.dev/datasets/certstream/all_latest.txt\n" >> "$GITHUB_WORKSPACE/main/README.md" | |
| echo '```' >> "$GITHUB_WORKSPACE/main/README.md" | |
| echo -e "" >> "$GITHUB_WORKSPACE/main/README.md" | |
| echo '---' >> "$GITHUB_WORKSPACE/main/README.md" | |
| #7Days | |
| echo -e "- #### [🖨️ **Stats** \`7Days\`⏲️ ➼ $(date +'%Y_%m_%d') <--> $(date -d "7 days ago" +'%Y_%m_%d')](https://pub.ajam.dev/datasets/certstream/all_weekly.txt)" >> "$GITHUB_WORKSPACE/main/README.md" | |
| echo '```console' >> "$GITHUB_WORKSPACE/main/README.md" | |
| echo -e "" >> "$GITHUB_WORKSPACE/main/README.md" | |
| #Total | |
| echo -e "\n--> 🌐 Total" >> "$GITHUB_WORKSPACE/main/README.md" | |
| echo -e "[+] New/ReNewed SSL Certs (ALL): +${{ env.SSL_7DAYS_TOTAL }}" >> "$GITHUB_WORKSPACE/main/README.md" | |
| echo -e "[+] View/Download: https://pub.ajam.dev/datasets/certstream/all_weekly.txt\n" >> "$GITHUB_WORKSPACE/main/README.md" | |
| echo '```' >> "$GITHUB_WORKSPACE/main/README.md" | |
| echo -e "" >> "$GITHUB_WORKSPACE/main/README.md" | |
| echo '---' >> "$GITHUB_WORKSPACE/main/README.md" | |
| echo -e "" >> "$GITHUB_WORKSPACE/main/README.md" | |
| #cat INFO | |
| cat "$GITHUB_WORKSPACE/main/INFO.md" >> "$GITHUB_WORKSPACE/main/README.md" | |
| continue-on-error: true | |
| - name: Get DateTime | |
| run: | | |
| # Date Time | |
| NEPALI_TIME=$(TZ='Asia/Kathmandu' date +'%Y-%m-%d (%I:%M:%S %p)') | |
| echo "NEPALI_TIME=$NEPALI_TIME" >> $GITHUB_ENV | |
| - name: Git Pull | |
| run: | | |
| cd "$GITHUB_WORKSPACE/main" && git pull origin main | |
| continue-on-error: true | |
| - uses: stefanzweifel/git-auto-commit-action@v5 | |
| with: | |
| repository: ./main | |
| commit_user_name: Azathothas # defaults to "github-actions[bot]" | |
| commit_user_email: AjamX101@gmail.com # defaults to "41898282+github-actions[bot]@users.noreply.github.com" | |
| commit_message: "✅ ⚓ Filter | Parsed 🔰 CertStream Data 🌊 <-- ${{ env.NEPALI_TIME }} ⌚" | |
| #push_options: '--force' | |
| - name: rClone Upload to R2 ("https://pub.ajam.dev/datasets/certstream") | |
| run: | | |
| # Presets | |
| set +x ; set +e | |
| #--------------# | |
| #touch "$HOME/.rclone.conf" | |
| echo "${{ secrets.RCLONE_CF_R2_PUB }}" > "$HOME/.rclone.conf" | |
| R2_ARCHIVE="$(date --utc +'%Y_%m_%d')" && export R2_ARCHIVE="${R2_ARCHIVE}" ; echo "R2_ARCHIVE: ${R2_ARCHIVE}" && echo "${R2_ARCHIVE}" > "/tmp/R2_ARCHIVE.txt" | |
| R2_ARCHIVE_P="$(date --utc -d "$(date --utc +'%Y-%m-%d') - 1 day" +'%Y_%m_%d')" && export R2_ARCHIVE_P="${R2_ARCHIVE_P}" ; echo "R2_ARCHIVE_P: ${R2_ARCHIVE_P}" && echo "${R2_ARCHIVE_P}" >> "/tmp/R2_ARCHIVE.txt" | |
| ##Fetch Today's Archive | |
| rm -rf "/tmp/R2ARCHIVE" ; mkdir -p "/tmp/R2ARCHIVE" ; cd "/tmp/R2ARCHIVE" | |
| readarray -t dates < "/tmp/R2_ARCHIVE.txt" | |
| for date in "${dates[@]}"; do | |
| echo -e "\n[+] Processing $date\n" | |
| find "$GITHUB_WORKSPACE/main/Raw/" -iname "*$date*.7z" -exec cp {} "/tmp/R2ARCHIVE" \; 2>/dev/null | |
| cd "/tmp/R2ARCHIVE" ; find "/tmp/R2ARCHIVE" -iname "*$date*.7z" -exec sh -c '7z x "{}" -o"$(dirname "{}")/$(basename "{}" .7z)"' \; 2>/dev/null | |
| find "/tmp/R2ARCHIVE" -iname "*$date*.7z" -exec rm -rf {} \; 2>/dev/null | |
| find "/tmp/R2ARCHIVE" -type f -iname "certstream_domains.txt" -exec bash -c 'file -b --mime-type "$1" | grep -q "text/plain" && cat "$1"' bash {} \; 2>/dev/null | sort -u -o "/tmp/$date.txt" ; wc -l "/tmp/$date.txt" | |
| find "/tmp/R2ARCHIVE" -type d -iname "*$date*" -exec rm -rf {} \; 2>/dev/null | |
| 7z a -t7z -mx="9" -mmt="$(($(nproc)+1))" "/tmp/$date.7z" "/tmp/$date.txt" | |
| cp "/tmp/$date.7z" "/tmp/R2ARCHIVE/$date.7z" ; du -sh "/tmp/R2ARCHIVE/$date.7z" | |
| echo -e "\n[+] Generated (/tmp/R2ARCHIVE/$date.7z) [$date]\n" | |
| done | |
| cd "/tmp/R2ARCHIVE" && rclone sync "./" "r2:/pub/datasets/certstream/Temp/" --user-agent="$USER_AGENT" --buffer-size="100M" --s3-upload-concurrency="500" --s3-chunk-size="100M" --multi-thread-streams="500" --checkers="2000" --transfers="1000" --retries="10" --check-first --checksum --copy-links --fast-list --progress | |
| ##Upload | |
| echo -e "\n[+] Uploading Results to R2 (rclone)\n" | |
| ##CertStream | |
| sort -u "/tmp/certstream_domains.txt" -o "/tmp/certstream_domains.txt" | |
| rclone copyto "/tmp/certstream_domains.txt" "r2:/pub/datasets/certstream/all_latest.txt" --user-agent="$USER_AGENT" --buffer-size="100M" --s3-upload-concurrency="500" --s3-chunk-size="100M" --multi-thread-streams="500" --checkers="2000" --transfers="1000" --retries="10" --check-first --checksum --copy-links --fast-list --progress | |
| sort -u "/tmp/certstream_domains_7days.txt" -o "/tmp/certstream_domains_7days.txt" | |
| rclone copyto "/tmp/certstream_domains_7days.txt" "r2:/pub/datasets/certstream/all_weekly.txt" --user-agent="$USER_AGENT" --buffer-size="100M" --s3-upload-concurrency="500" --s3-chunk-size="100M" --multi-thread-streams="500" --checkers="2000" --transfers="1000" --retries="10" --check-first --checksum --copy-links --fast-list --progress | |
| ##NRD | |
| sort -u "/tmp/nrd_latest.txt" -o "/tmp/nrd_latest.txt" | |
| rclone copyto "/tmp/nrd_latest.txt" "r2:/pub/datasets/certstream/nrd_latest.txt" --user-agent="$USER_AGENT" --buffer-size="100M" --s3-upload-concurrency="500" --s3-chunk-size="100M" --multi-thread-streams="500" --checkers="2000" --transfers="1000" --retries="10" --check-first --checksum --copy-links --fast-list --progress | |
| sort -u "/tmp/nrd_weekly.txt" -o "/tmp/nrd_weekly.txt" | |
| rclone copyto "/tmp/nrd_weekly.txt" "r2:/pub/datasets/certstream/nrd_weekly.txt" --user-agent="$USER_AGENT" --buffer-size="100M" --s3-upload-concurrency="500" --s3-chunk-size="100M" --multi-thread-streams="500" --checkers="2000" --transfers="1000" --retries="10" --check-first --checksum --copy-links --fast-list --progress | |
| continue-on-error: true |