From f502a8ce4f4b65d2c2d4dbcb50f786808267aed5 Mon Sep 17 00:00:00 2001
From: h7x4 <h7x4@nani.wtf>
Date: Sat, 15 Mar 2025 14:33:13 +0100
Subject: [PATCH] kommode/gitea: add robots.txt

---
 flake.nix                                |   7 ++
 hosts/kommode/services/gitea/default.nix | 103 +++++++++++++++++++++++
 2 files changed, 110 insertions(+)

diff --git a/flake.nix b/flake.nix
index 8d95e54..b28f90c 100644
--- a/flake.nix
+++ b/flake.nix
@@ -150,6 +150,12 @@
           self.nixosModules.bluemap
         ];
       };
+      bob = stableNixosConfig "bob" {
+        modules = [
+          disko.nixosModules.disko
+          { disko.devices.disk.disk1.device = "/dev/vda"; }
+        ];
+      };
       ildkule = stableNixosConfig "ildkule" { };
       #ildkule-unstable = unstableNixosConfig "ildkule" { };
       shark = stableNixosConfig "shark" { };
@@ -161,6 +167,7 @@
         ];
         modules = [
           inputs.nix-gitea-themes.nixosModules.default
+          self.nixosModules.robots-txt
         ];
       };
 
diff --git a/hosts/kommode/services/gitea/default.nix b/hosts/kommode/services/gitea/default.nix
index a5c1bf6..6c5482e 100644
--- a/hosts/kommode/services/gitea/default.nix
+++ b/hosts/kommode/services/gitea/default.nix
@@ -193,6 +193,109 @@ in {
     };
   };
 
+  environment.robots-txt."gitea" = {
+    virtualHost = domain;
+    rules = [
+      {
+        pre_comment = ''
+          Gitea internals
+
+          See these for more information:
+          - https://gitea.com/robots.txt
+          - https://codeberg.org/robots.txt
+        '';
+        User-agent = "*";
+        Disallow = [
+          "/api/*"
+          "/avatars"
+          "/*/*/src/commit/*"
+          "/*/*/commit/*"
+          "/*/*/*/refs/*"
+          "/*/*/*/star"
+          "/*/*/*/watch"
+          "/*/*/labels"
+          "/*/*/activity/*"
+          "/vendor/*"
+          "/swagger.*.json"
+          "/repo/create"
+          "/repo/migrate"
+          "/org/create"
+          "/*/*/fork"
+          "/*/*/watchers"
+          "/*/*/stargazers"
+          "/*/*/forks"
+          "*/.git/"
+          "/*.git"
+          "/*.atom"
+          "/*.rss"
+        ];
+      }
+      {
+        pre_comment = "Language Spam";
+        Disallow = "/*?lang=";
+      }
+      {
+        pre_comment = ''
+          AI bots
+
+          Sourced from:
+          - https://www.vg.no/robots.txt
+          - https://codeberg.org/robots.txt
+        '';
+        User-agent = [
+          "AI2Bot"
+          "Ai2Bot-Dolma"
+          "Amazonbot"
+          "Applebot-Extended"
+          "Bytespider"
+          "CCBot"
+          "ChatGPT-User"
+          "Claude-Web"
+          "ClaudeBot"
+          "Crawlspace"
+          "Diffbot"
+          "FacebookBot"
+          "FriendlyCrawler"
+          "GPTBot"
+          "Google-Extended"
+          "ICC-Crawler"
+          "ImagesiftBot"
+          "Kangaroo Bot"
+          "Meta-ExternalAgent"
+          "OAI-SearchBot"
+          "Omgili"
+          "Omgilibot"
+          "PanguBot"
+          "PerplexityBot"
+          "PetalBot"
+          "Scrapy"
+          "SemrushBot-OCOB"
+          "Sidetrade indexer bot"
+          "Timpibot"
+          "VelenPublicWebCrawler"
+          "Webzio-Extended"
+          "YouBot"
+          "anthropic-ai"
+          "cohere-ai"
+          "cohere-training-data-crawler"
+          "facebookexternalhit"
+          "iaskspider/2.0"
+          "img2dataset"
+          "meta-externalagent"
+          "omgili"
+          "omgilibot"
+        ];
+        Disallow = "/";
+      }
+      {
+        Crawl-delay = "2";
+      }
+      {
+        Sitemap = "https://${domain}/sitemap.xml";
+      }
+    ];
+  };
+
   networking.firewall.allowedTCPPorts = [ sshPort ];
 
   systemd.services.gitea-dump = {