anubis: weigh the souls of the AI scrapers

configs/bots/tlg-wotd.nix

@@ -44,7 +44,7 @@ in {
 
       transliteration=$(${pkgs.writers.makePythonWriter pkgs.python311 pkgs.python311Packages pkgs.python3Packages "translit.py" {
         # revert to pkgs.writers.writePython3 once https://github.com/NixOS/nixpkgs/pull/353367 is merged
-        libraries = [ pkgs.python311Packages.cltk ];
+        libraries = [ pkgs.python3Packages.cltk ];
       } ''
         import sys
         from cltk.phonology.grc.transcription import Transcriber

flake.nix

@@ -413,6 +413,7 @@
         vg = pkgs.callPackage packages/vg.nix {};
         vim = pkgs.callPackage packages/vim.nix {niveumPackages = self.packages.${system};};
         obsidian-vim = pkgs.callPackage packages/obsidian-vim.nix {};
+        radio-news = pkgs.callPackage packages/radio-news.nix {};
         vimPlugins-cheat-sh-vim = pkgs.callPackage packages/vimPlugins/cheat-sh.nix {};
         vimPlugins-icalendar-vim = pkgs.callPackage packages/vimPlugins/icalendar-vim.nix {};
         vimPlugins-jq-vim = pkgs.callPackage packages/vimPlugins/jq-vim.nix {};

packages/radio-news.nix

@@ -0,0 +1,47 @@
+{ writers, lib, gnused, curl, jq, yq }:
+writers.writeBashBin "radio-news" ''
+  set -efu
+  PATH=$PATH:${lib.makeBinPath [gnused curl jq yq]}
+
+  EVENTS=$(
+    curl https://www.goodnewsnetwork.org/feed/ \
+      | xq '
+        .rss.channel.item
+        | map(select((.pubDate|strptime("%a, %d %b %Y %H:%M:%S %z")) as $date | ($date | mktime) > (now - (60 * 60 * 24))) | {title, description})
+      '
+  )
+
+  SYSTEM_PROMPT=$(cat <<EOF
+  You are a news anchor writing a short news digest for a radio broadcast.
+  Summarize the following news headlines into a cohesive, engaging script under 400 words.
+  Keep it professional, concise as possible, and easy to follow.
+  Please no unnecessary explanations why the news is good.
+
+  Begin the digest with: "Here's your good news update."
+  EOF
+  )
+
+  REQUEST=$(cat <<EOF
+  {
+      "system_instruction": {
+        "parts": [
+          {
+            "text": $(jq -Rs <<< "$SYSTEM_PROMPT")
+          }
+        ]
+      },
+      "contents": [
+        {
+          "parts": [
+            {
+              "text": $(jq -Rs <<< "$EVENTS")
+            }
+          ]
+        }
+      ]
+    }
+  EOF
+  )
+
+  echo "$REQUEST" | curl "https://generativelanguage.googleapis.com/v1beta/models/gemini-1.5-flash-8b:generateContent?key=$GEMINI_API_KEY" -s -H "Content-Type: application/json" -d @-
+''

systems/makanek/gitea.nix

@@ -1,20 +1,45 @@
+{ config, ... }:
 let
   inherit (import ../../lib) sshPort;
-  domain = "https://code.kmein.de";
+  domain = "code.kmein.de";
 in {
+  services.anubis = {
+    defaultOptions.settings = {
+      USER_DEFINED_DEFAULT = true;
+    };
+    instances = {
+      "gitea".settings = {
+        TARGET = "http://localhost:${toString config.services.gitea.settings.server.HTTP_PORT}";
+        USER_DEFINED_INSTANCE = true;
+        OG_PASSTHROUGH = true;
+        SERVE_ROBOTS_TXT = true;
+      };
+    };
+  };
+
+  users.users.nginx.extraGroups = [ config.services.anubis.instances."gitea".group ];
+
   services.gitea = {
     enable = true;
-    appName = "code.kmein.de";
+    appName = domain;
     settings = {
-      server.ROOT_URL = domain;
+      server.ROOT_URL = "https://${domain}";
+      server.DOMAIN = domain;
       server.SSH_PORT = sshPort;
       service.DISABLE_REGISTRATION = true;
     };
   };
-  services.nginx.virtualHosts."code.kmein.de" = {
+  services.nginx.virtualHosts.${domain} = {
     forceSSL = true;
     enableACME = true;
-    locations."/".extraConfig = "proxy_pass http://localhost:3000;";
+    # locations."/".extraConfig = "proxy_pass http://localhost:3000;";
+    locations = {
+      "/" = {
+        proxyPass = "http://unix:${config.services.anubis.instances."gitea".settings.BIND}";
+        proxyWebsockets = true;
+      };
+      "/metrics".proxyPass = "http://unix:${config.services.anubis.instances."gitea".settings.METRICS_BIND}";
+    };
   };
 
   niveum.passport.services = [

systems/makanek/radio-news.nix

@@ -2,6 +2,7 @@
   config,
   pkgs,
   lib,
+  niveumPackages,
   ...
 }: let
   inherit (import ../../lib) serveHtml;
@@ -22,58 +23,14 @@ in {
     ];
     startAt = "*:50";
     script = ''
-      set -efu
-      PATH=$PATH:${lib.makeBinPath [pkgs.w3m pkgs.gnused pkgs.curl pkgs.jq pkgs.yq]}
+      PATH=$PATH:${lib.makeBinPath [pkgs.gnused pkgs.curl pkgs.jq]}
 
-      export GEMINI_API_KEY="$(cat "$CREDENTIALS_DIRECTORY/gemini-api-key")"
-
-      EVENTS=$(
-        curl https://www.goodnewsnetwork.org/feed/ \
-          | xq '
-            .rss.channel.item
-            | map(select((.pubDate|strptime("%a, %d %b %Y %H:%M:%S %z")) as $date | ($date | mktime) > (now - (60 * 60 * 24))) | {title, description})
-          '
-      )
-
-      SYSTEM_PROMPT=$(cat <<EOF
-      You are a news anchor writing a short news digest for a radio broadcast.
-      Summarize the following news headlines into a cohesive, engaging script under 400 words.
-      Keep it professional, concise, and easy to follow.
-
-      Begin the digest with: "Here's your good news update for $(date -u +"%B %d, %Y")."
-      EOF
-      )
-
-      REQUEST=$(cat <<EOF
+      GEMINI_API_KEY="$(cat "$CREDENTIALS_DIRECTORY/gemini-api-key")" ${niveumPackages.radio-news}/bin/radio-news | jq --arg from "$(date -u -Is | sed 's/+00:00/Z/')" --arg to "$(date -u -Is -d 'now + 30 minutes' | sed 's/+00:00/Z/')" '
       {
-          "system_instruction": {
-            "parts": [
-              {
-                "text": $(jq -Rs <<< "$SYSTEM_PROMPT")
-              }
-            ]
-          },
-          "contents": [
-            {
-              "parts": [
-                {
-                  "text": $(jq -Rs <<< "$EVENTS")
-                }
-              ]
-            }
-          ]
-        }
-      EOF
-      )
-
-      RESPONSE=$(echo "$REQUEST" | curl "https://generativelanguage.googleapis.com/v1beta/models/gemini-1.5-flash-8b:generateContent?key=$GEMINI_API_KEY" -s -H "Content-Type: application/json" -d @-)
-
-      echo "$RESPONSE" | jq --arg from "$(date -u -Is | sed 's/+00:00/Z/')" --arg to "$(date -u -Is -d 'now + 30 minutes' | sed 's/+00:00/Z/')" '
-        {
-          from: $from,
-          to: $to,
-          text: .candidates[].content.parts[].text
-        }' | curl -s -X POST http://radio-news.r -H "Content-Type: application/json" -d @-
+        from: $from,
+        to: $to,
+        text: .candidates[].content.parts[].text
+      }' | curl -s -X POST http://radio-news.r -H "Content-Type: application/json" -d @-
     '';
   };