Merge pull request #303388 from SuperSandro2000/paperless-only-enabled-languages

nixos/paperless: override enabled tesseract languages with the in paperless configured ones

nixos/modules/services/misc/paperless.nix

···
 with lib;
 let
   cfg = config.services.paperless;
-  pkg = cfg.package;
 
   defaultUser = "paperless";
   defaultFont = "${pkgs.liberation_ttf}/share/fonts/truetype/LiberationSerif-Regular.ttf";
···
   } // optionalAttrs (cfg.settings.PAPERLESS_ENABLE_NLTK or true) {
     PAPERLESS_NLTK_DIR = pkgs.symlinkJoin {
       name = "paperless_ngx_nltk_data";
-      paths = pkg.nltkData;
+      paths = cfg.package.nltkData;
     };
   } // optionalAttrs (cfg.openMPThreadingWorkaround) {
     OMP_NUM_THREADS = "1";
···
   manage = pkgs.writeShellScript "manage" ''
     set -o allexport # Export the following env vars
     ${lib.toShellVars env}
-    exec ${pkg}/bin/paperless-ngx "$@"
+    exec ${cfg.package}/bin/paperless-ngx "$@"
   '';
 
   # Secure the services
···
       description = "User under which Paperless runs.";
     };
 
-    package = mkPackageOption pkgs "paperless-ngx" { };
+    package = mkPackageOption pkgs "paperless-ngx" { } // {
+      apply = pkg: pkg.override {
+        tesseract5 = pkg.tesseract5.override {
+          # always enable detection modules
+          enableLanguages = if cfg.settings ? PAPERLESS_OCR_LANGUAGE then
+            [ "equ" "osd" ]
+              ++ lib.splitString "+" cfg.settings.PAPERLESS_OCR_LANGUAGE
+          else null;
+        };
+      };
+    };
 
     openMPThreadingWorkaround = mkEnableOption ''
       a workaround for document classifier timeouts.
···
       wants = [ "paperless-consumer.service" "paperless-web.service" "paperless-task-queue.service" ];
       serviceConfig = defaultServiceConfig // {
         User = cfg.user;
-        ExecStart = "${pkg}/bin/celery --app paperless beat --loglevel INFO";
+        ExecStart = "${cfg.package}/bin/celery --app paperless beat --loglevel INFO";
         Restart = "on-failure";
         LoadCredential = lib.optionalString (cfg.passwordFile != null) "PAPERLESS_ADMIN_PASSWORD:${cfg.passwordFile}";
       };
···
         versionFile="${cfg.dataDir}/src-version"
         version=$(cat "$versionFile" 2>/dev/null || echo 0)
 
-        if [[ $version != ${pkg.version} ]]; then
-          ${pkg}/bin/paperless-ngx migrate
+        if [[ $version != ${cfg.package.version} ]]; then
+          ${cfg.package}/bin/paperless-ngx migrate
 
           # Parse old version string format for backwards compatibility
           version=$(echo "$version" | grep -ohP '[^-]+$')
···
           if versionLessThan 1.12.0; then
             # Reindex documents as mentioned in https://github.com/paperless-ngx/paperless-ngx/releases/tag/v1.12.1
             echo "Reindexing documents, to allow searching old comments. Required after the 1.12.x upgrade."
-            ${pkg}/bin/paperless-ngx document_index reindex
+            ${cfg.package}/bin/paperless-ngx document_index reindex
           fi
 
-          echo ${pkg.version} > "$versionFile"
+          echo ${cfg.package.version} > "$versionFile"
         fi
       ''
       + optionalString (cfg.passwordFile != null) ''
···
         superuserStateFile="${cfg.dataDir}/superuser-state"
 
         if [[ $(cat "$superuserStateFile" 2>/dev/null) != $superuserState ]]; then
-          ${pkg}/bin/paperless-ngx manage_superuser
+          ${cfg.package}/bin/paperless-ngx manage_superuser
           echo "$superuserState" > "$superuserStateFile"
         fi
       '';
···
       after = [ "paperless-scheduler.service" ];
       serviceConfig = defaultServiceConfig // {
         User = cfg.user;
-        ExecStart = "${pkg}/bin/celery --app paperless worker --loglevel INFO";
+        ExecStart = "${cfg.package}/bin/celery --app paperless worker --loglevel INFO";
         Restart = "on-failure";
         # The `mbind` syscall is needed for running the classifier.
         SystemCallFilter = defaultServiceConfig.SystemCallFilter ++ [ "mbind" ];
···
       after = [ "paperless-scheduler.service" ];
       serviceConfig = defaultServiceConfig // {
         User = cfg.user;
-        ExecStart = "${pkg}/bin/paperless-ngx document_consumer";
+        ExecStart = "${cfg.package}/bin/paperless-ngx document_consumer";
         Restart = "on-failure";
       };
       environment = env;
···
           echo "PAPERLESS_SECRET_KEY is empty, refusing to start."
           exit 1
         fi
-        exec ${pkg.python.pkgs.gunicorn}/bin/gunicorn \
-          -c ${pkg}/lib/paperless-ngx/gunicorn.conf.py paperless.asgi:application
+        exec ${cfg.package.python.pkgs.gunicorn}/bin/gunicorn \
+          -c ${cfg.package}/lib/paperless-ngx/gunicorn.conf.py paperless.asgi:application
       '';
       serviceConfig = defaultServiceConfig // {
         User = cfg.user;
···
         CapabilityBoundingSet = [ "CAP_NET_BIND_SERVICE" ];
       };
       environment = env // {
-        PYTHONPATH = "${pkg.python.pkgs.makePythonPath pkg.propagatedBuildInputs}:${pkg}/lib/paperless-ngx/src";
+        PYTHONPATH = "${cfg.package.python.pkgs.makePythonPath cfg.package.propagatedBuildInputs}:${cfg.package}/lib/paperless-ngx/src";
       };
       # Allow the web interface to access the private /tmp directory of the server.
       # This is required to support uploading files via the web interface.

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

···
   # https://github.com/NixOS/nixpkgs/issues/298719
   # https://github.com/paperless-ngx/paperless-ngx/issues/5494
   python = python3.override {
-    packageOverrides = self: super: {
-      uvicorn = super.uvicorn.overridePythonAttrs (oldAttrs: {
+    packageOverrides = final: prev: {
+      # tesseract5 may be overwritten in the paperless module and we need to propagate that to make the closure reduction effective
+      ocrmypdf = prev.ocrmypdf.override { tesseract = tesseract5; };
+
+      uvicorn = prev.uvicorn.overridePythonAttrs (_: {
         version = "0.25.0";
         src = fetchFromGitHub {
           owner = "encode";
···
   doCheck = !stdenv.isDarwin;
 
   passthru = {
-    inherit python path frontend;
+    inherit python path frontend tesseract5;
     nltkData = with nltk-data; [ punkt snowball_data stopwords ];
     tests = { inherit (nixosTests) paperless; };
   };