niveum/configs/bots/tlg-wotd.nix

{
  pkgs,
  lib,
  config,
  niveumPackages,
  unstablePackages,
  ...
}: let
  mastodonEndpoint = "https://social.krebsco.de";
in {
  systemd.services.bot-tlg-wotd = {
    # TODO reenable
    # once https://github.com/NixOS/nixpkgs/pull/462893 is in stable NixOS
    enable = true;
    wants = ["network-online.target"];
    startAt = "9:30";
    path = [ pkgs.jq pkgs.curl pkgs.recode pkgs.deno pkgs.imagemagick pkgs.gawk pkgs.gnugrep pkgs.coreutils ];
    environment = {
      NPM_CONFIG_CACHE = "/tmp";
      CLTK_DATA = "/tmp";
    };
    script = ''
      set -efux

      chat_id=@tlgwotd

      export TELEGRAM_TOKEN="$(cat "$CREDENTIALS_DIRECTORY/telegram-token")"
      export MASTODON_TOKEN="$(cat "$CREDENTIALS_DIRECTORY/mastodon-token")"

      json_data=$(curl -sSL http://stephanus.tlg.uci.edu/Iris/Wotd | recode html..utf8)

      word=$(echo "$json_data" | jq -r '.word')
      compact_word=$(echo "$word" | sed 's/,.*$//')
      definition=$(echo "$json_data" | jq -r '.definition | sub("<.*>"; "") | rtrimstr(" ")')
      first_occurrence=$(echo "$json_data" | jq -r '.firstOccurrence')
      total_occurrences=$(echo "$json_data" | jq -r '.totalOccurrences')
      telegram_caption="*$word* ‘$definition’

    First occurrence (century): $first_occurrence
    Number of occurrences (in all Ancient Greek texts): $total_occurrences"
      mastodon_caption="$word ‘$definition’

    First occurrence (century): $first_occurrence
    Number of occurrences (in all Ancient Greek texts): $total_occurrences"

      #ancientgreek #classics #wotd #wordoftheday

      transliteration=$(${pkgs.writers.writePython3 "translit.py" {
        libraries = py: [ py.cltk ];
      } ''
        import sys
        from cltk.phonology.grc.transcription import Transcriber

        probert = Transcriber("Attic", "Probert")
        text = " ".join(sys.argv[1:])
        ipa = probert.transcribe(text)

        print(ipa)
      ''} "$compact_word")


      photo_path=/tmp/output.png

      hex_to_rgb() {
          hex="$1"
          r=$(printf "%d" "0x$(echo "$hex" | cut -c2-3)")
          g=$(printf "%d" "0x$(echo "$hex" | cut -c4-5)")
          b=$(printf "%d" "0x$(echo "$hex" | cut -c6-7)")
          echo "$r $g $b"
      }

      calculate_luminance() {
          r="$1"
          g="$2"
          b="$3"

          r_l=$(echo "$r" | awk '{print ($1 / 255 <= 0.03928) ? $1 / 255 / 12.92 : (($1 / 255 + 0.055) / 1.055)^2.4}')
          g_l=$(echo "$g" | awk '{print ($1 / 255 <= 0.03928) ? $1 / 255 / 12.92 : (($1 / 255 + 0.055) / 1.055)^2.4}')
          b_l=$(echo "$b" | awk '{print ($1 / 255 <= 0.03928) ? $1 / 255 / 12.92 : (($1 / 255 + 0.055) / 1.055)^2.4}')

          echo "$r_l $g_l $b_l" | awk '{print 0.2126*$1 + 0.7152*$2 + 0.0722*$3}'
      }


      hex_color="#$(echo "$compact_word" | md5sum | cut -c 1-6)"
      if echo "$hex_color" | grep -qE '^#[0-9A-Fa-f]{6}$'; then
          set -- $(hex_to_rgb "$hex_color")
          r="$1"
          g="$2"
          b="$3"
      fi

      luminance=$(calculate_luminance "$r" "$g" "$b")

      threshold="0.1"
      echo "$r $g $b"
      if [ "$(echo "$luminance" | awk -v threshold="$threshold" '{print ($1 > threshold)}')" -eq 1 ]; then
          color1="black"
          color2="#333"
      else
          color1="white"
          color2=lightgrey
      fi

      magick -size 1400x846 \
          xc:"$hex_color" \
          -font "${pkgs.gentium}/share/fonts/truetype/GentiumBookPlus-Bold.ttf" \
          -fill "$color1" \
          -pointsize 150 -gravity west \
          -annotate +100-160 "$compact_word" \
          -font "${pkgs.gentium}/share/fonts/truetype/GentiumBookPlus-Regular.ttf" \
          -fill "$color2" \
          -pointsize 60 -gravity west \
          -annotate +100+00 "$transliteration" \
          -fill "$color1" \
          -annotate +100+120 "‘$definition’" \
          -fill "$color2" \
          -pointsize 40 -gravity southwest \
          -annotate +100+60 "attested $total_occurrences times" \
          -pointsize 40 -gravity southeast \
          -annotate +100+60 "$(date -I)" \
          "$photo_path"

      curl -X POST "https://api.telegram.org/bot$TELEGRAM_TOKEN/sendPhoto" \
           -F "chat_id=\"$chat_id\"" \
           -F "photo=@$photo_path" \
           -F parse_mode=Markdown \
           -F caption="$telegram_caption"

      mastodon_upload_response=$(curl -X POST "${mastodonEndpoint}/api/v2/media" \
          -H "Authorization: Bearer $MASTODON_TOKEN" \
          -F "file=@$photo_path" \
          -F "description=$word ‘$definition’")
      mastodon_image_id=$(echo $mastodon_upload_response | jq -r .id)
      curl -X POST "${mastodonEndpoint}/api/v1/statuses" \
          -H "Authorization: Bearer $MASTODON_TOKEN" \
          -d "status=$mastodon_caption" \
          -d "visibility=public" \
          -d "media_ids[]=$mastodon_image_id"
    '';
    serviceConfig = {
      Type = "oneshot";
      DynamicUser = true;
      StateDirectory = "tlgwotd";
      PrivateTmp = true;
      LoadCredential = [
        "telegram-token:${config.age.secrets.telegram-token-kmein.path}"
        "mastodon-token:${config.age.secrets.mastodon-token-tlgwotd.path}"
      ];
    };
  };

  age.secrets = {
    mastodon-token-tlgwotd.file = ../../secrets/mastodon-token-tlgwotd.age;
  };

  niveum.passport.services = [
    {
      title = "Thesaurus Linguae Graecae Word of the Day";
      description = "sends <a href=\"https://stephanus.tlg.uci.edu/\">TLG</a>'s word of the day to Telegram.";
      link = "https://t.me/tlgwotd";
    }
  ];
}