Merge pull request #215750 from puppe/fix-paperless

nixos/paperless: download NLTK data

nixos/modules/services/misc/paperless.nix

···
   pkg = cfg.package;
 
   defaultUser = "paperless";
+  nltkDir = "/var/cache/paperless/nltk";
 
   # Don't start a redis instance if the user sets a custom redis connection
   enableRedis = !hasAttr "PAPERLESS_REDIS" cfg.extraConfig;
···
     PAPERLESS_DATA_DIR = cfg.dataDir;
     PAPERLESS_MEDIA_ROOT = cfg.mediaDir;
     PAPERLESS_CONSUMPTION_DIR = cfg.consumptionDir;
+    PAPERLESS_NLTK_DIR = nltkDir;
     GUNICORN_CMD_ARGS = "--bind=${cfg.address}:${toString cfg.port}";
   } // optionalAttrs (config.time.timeZone != null) {
     PAPERLESS_TIME_ZONE = config.time.timeZone;
···
     lib.mapAttrs (_: toString) cfg.extraConfig
   );
 
-  manage = let
-    setupEnv = lib.concatStringsSep "\n" (mapAttrsToList (name: val: "export ${name}=\"${val}\"") env);
-  in pkgs.writeShellScript "manage" ''
-    ${setupEnv}
-    exec ${pkg}/bin/paperless-ngx "$@"
-  '';
+  manage =
+    let
+      setupEnv = lib.concatStringsSep "\n" (mapAttrsToList (name: val: "export ${name}=\"${val}\"") env);
+    in
+    pkgs.writeShellScript "manage" ''
+      ${setupEnv}
+      exec ${pkg}/bin/paperless-ngx "$@"
+    '';
 
   # Secure the services
   defaultServiceConfig = {
···
       cfg.dataDir
       cfg.mediaDir
     ];
+    CacheDirectory = "paperless";
     CapabilityBoundingSet = "";
     # ProtectClock adds DeviceAllow=char-rtc r
     DeviceAllow = "";
···
 
     extraConfig = mkOption {
       type = types.attrs;
-      default = {};
+      default = { };
       description = lib.mdDoc ''
         Extra paperless config options.
 
···
             '${cfg.passwordFile}' '${cfg.dataDir}/superuser-password'
         '';
         Type = "oneshot";
+      };
+    };
+
+    # Download NLTK corpus data
+    systemd.services.paperless-download-nltk-data = {
+      wantedBy = [ "paperless-scheduler.service" ];
+      before = [ "paperless-scheduler.service" ];
+      after = [ "network-online.target" ];
+      serviceConfig = defaultServiceConfig // {
+        User = cfg.user;
+        Type = "oneshot";
+        # Enable internet access
+        PrivateNetwork = false;
+        # Restrict write access
+        BindPaths = [];
+        BindReadOnlyPaths = [
+          "/nix/store"
+          "-/etc/resolv.conf"
+          "-/etc/nsswitch.conf"
+          "-/etc/ssl/certs"
+          "-/etc/static/ssl/certs"
+          "-/etc/hosts"
+          "-/etc/localtime"
+        ];
+        ExecStart = let pythonWithNltk = pkg.python.withPackages (ps: [ ps.nltk ]); in ''
+          ${pythonWithNltk}/bin/python -m nltk.downloader -d '${nltkDir}' punkt snowball_data stopwords
+        '';
       };
     };
 

pkgs/applications/office/paperless-ngx/default.nix

···
 
   meta = with lib; {
     description = "Tool to scan, index, and archive all of your physical documents";
-    homepage = "https://paperless-ngx.readthedocs.io/";
+    homepage = "https://docs.paperless-ngx.com/";
     changelog = "https://github.com/paperless-ngx/paperless-ngx/releases/tag/v${version}";
     license = licenses.gpl3Only;
     maintainers = with maintainers; [ lukegb gador erikarvstedt ];