paperless: init at 2.7.0

pkgs/applications/office/paperless/default.nix

@@ -0,0 +1,170 @@
+{ stdenv
+, lib
+, fetchFromGitHub
+, makeWrapper
+, callPackage
+
+, python3
+, file
+, imagemagick7
+, ghostscript
+, optipng
+, poppler
+, tesseract
+, unpaper
+}:
+
+## Usage
+
+# ${paperless}/bin/paperless wraps manage.py
+
+# ${paperless}/share/paperless/setup-env.sh can be sourced from a
+# shell script to setup a Paperless environment
+
+# paperless.withConfig is a convenience function to setup a
+# configured Paperless instance. (See ./withConfig.nix)
+
+# For WSGI with gunicorn, use a shell script like this:
+# let
+#   pythonEnv = paperless.python.withPackages (ps: paperless.runtimePackages ++ [ ps.gunicorn ]);
+# in
+#   writers.writeBash "run-gunicorn" ''
+#     source ${paperless}/share/paperless/setup-env.sh
+#     PYTHONPATH=$paperlessSrc ${pythonEnv}/bin/gunicorn paperless.wsgi
+#   ''
+
+let
+  paperless = stdenv.mkDerivation rec {
+    name = "paperless-${version}";
+    version = "2.7.0";
+
+    src = fetchFromGitHub {
+      owner = "the-paperless-project";
+      repo = "paperless";
+      rev = version;
+      sha256 = "0pkmyky1crjnsg7r0gfk0fadisfsgzlsq6afpz16wx4hp6yvkkf7";
+    };
+
+    nativeBuildInputs = [ makeWrapper ];
+
+    doCheck = true;
+    dontInstall = true;
+
+    pythonEnv      = python.withPackages (_: runtimePackages);
+    pythonCheckEnv = python.withPackages (_: (runtimePackages ++ checkPackages));
+
+    unpackPhase = ''
+      srcDir=$out/share/paperless
+      mkdir -p $srcDir
+      cp -r --no-preserve=mode $src/src/* $src/LICENSE $srcDir
+    '';
+
+    buildPhase = let
+      # Paperless has explicit runtime checks that expect these binaries to be in PATH
+      extraBin = lib.makeBinPath [ imagemagick7 ghostscript optipng tesseract unpaper ];
+    in ''
+      ${python.interpreter} -m compileall $srcDir
+
+      makeWrapper $pythonEnv/bin/python $out/bin/paperless \
+        --set PATH ${extraBin} --add-flags $out/share/paperless/manage.py
+
+      # A shell snippet that can be sourced to setup a paperless env
+      cat > $out/share/paperless/setup-env.sh <<EOF
+      export PATH="$pythonEnv/bin:${extraBin}''${PATH:+:}$PATH"
+      export paperlessSrc=$out/share/paperless
+      EOF
+    '';
+
+    checkPhase = ''
+      source $out/share/paperless/setup-env.sh
+      tmpDir=$(realpath testsTmp)
+      mkdir $tmpDir
+      export HOME=$tmpDir
+      export PAPERLESS_MEDIADIR=$tmpDir
+      cd $paperlessSrc
+      # Prevent tests from writing to the derivation output
+      chmod -R -w $out
+      # Disable cache to silence a pytest warning ("could not create cache")
+      $pythonCheckEnv/bin/pytest -p no:cacheprovider
+    '';
+
+    passthru = {
+      withConfig = callPackage ./withConfig.nix {};
+      inherit python runtimePackages checkPackages tesseract;
+    };
+
+    meta = with lib; {
+      description = "Scan, index, and archive all of your paper documents";
+      homepage = https://github.com/the-paperless-project/paperless;
+      license = licenses.gpl3;
+      maintainers = [ maintainers.earvstedt ];
+    };
+  };
+
+  python = python3.override {
+    packageOverrides = self: super: {
+      # Paperless only supports Django 2.0
+      django = django_2_0 super;
+      pyocr = pyocrWithUserTesseract super;
+      # These are pre-release versions, hence they are private to this pkg
+      django-filter = self.callPackage ./python-modules/django-filter.nix {};
+      django-crispy-forms = self.callPackage ./python-modules/django-crispy-forms.nix {};
+    };
+  };
+
+  django_2_0 = pyPkgs: pyPkgs.django_2_1.overrideDerivation (_: rec {
+    pname = "Django";
+    version = "2.0.12";
+    name = "${pname}-${version}";
+    src = pyPkgs.fetchPypi {
+      inherit pname version;
+      sha256 = "15s8z54k0gf9brnz06521bikm60ddw5pn6v3nbvnl47j1jjsvwz2";
+    };
+  });
+
+  runtimePackages = with python.pkgs; [
+    dateparser
+    dateutil
+    django
+    django-cors-headers
+    django-crispy-forms
+    django-filter
+    django_extensions
+    djangoql
+    djangorestframework
+    factory_boy
+    filemagic
+    fuzzywuzzy
+    langdetect
+    pdftotext
+    pillow
+    psycopg2
+    pyocr
+    python-dotenv
+    python-gnupg
+    pytz
+    termcolor
+  ] ++ (lib.optional stdenv.isLinux inotify-simple);
+
+  checkPackages = with python.pkgs; [
+    pytest
+    pytest-django
+    pytest-env
+    pytest_xdist
+  ];
+
+  pyocrWithUserTesseract = pyPkgs:
+    let
+      pyocr = pyPkgs.pyocr.override { inherit tesseract; };
+    in
+      if pyocr.outPath == pyPkgs.pyocr.outPath then
+        pyocr
+      else
+        # The user has provided a custom tesseract derivation that might be
+        # missing some languages that are required for PyOCR's tests. Disable them to
+        # avoid build errors.
+        pyocr.overridePythonAttrs (attrs: {
+          doCheck = false;
+        });
+in
+  paperless

pkgs/applications/office/paperless/python-modules/django-crispy-forms.nix

@@ -0,0 +1,36 @@
+{ lib, buildPythonPackage, fetchFromGitHub
+, pytest, pytest-django, django }:
+
+buildPythonPackage rec {
+  pname = "django-crispy-forms";
+  version = "2019.04.21";
+
+  src = fetchFromGitHub {
+    owner = "django-crispy-forms";
+    repo = "django-crispy-forms";
+    rev = "e25a5326697e5b545689b3a914e516404a6911bb";
+    sha256 = "12zqa76q6i7j47aqvhilivpbdplgp9zw2q8zfcjzlgclrqafaj39";
+  };
+
+  # For reasons unknown, the source dir must contain a dash
+  # for the tests to run successfully
+  postUnpack = ''
+    mv $sourceRoot source-
+    export sourceRoot=source-
+  '';
+
+  checkInputs = [ pytest pytest-django django ];
+
+  checkPhase = ''
+    PYTHONPATH="$(pwd):$PYTHONPATH" \
+    DJANGO_SETTINGS_MODULE=crispy_forms.tests.test_settings \
+      pytest crispy_forms/tests
+  '';
+
+  meta = with lib; {
+    description = "The best way to have DRY Django forms";
+    homepage = https://github.com/maraujop/django-crispy-forms;
+    license = licenses.mit;
+    maintainers = with maintainers; [ earvstedt ];
+  };
+}

pkgs/applications/office/paperless/python-modules/django-filter.nix

@@ -0,0 +1,26 @@
+{ lib, buildPythonPackage, python, pythonOlder, fetchFromGitHub
+, django, django-crispy-forms, djangorestframework, mock, pytz }:
+
+buildPythonPackage rec {
+  pname = "django-filter";
+  version = "2.1.0-pre";
+  disabled = pythonOlder "3.4";
+
+  src = fetchFromGitHub {
+    owner = "carltongibson";
+    repo = pname;
+    rev = "24adad8c48bc9e7c7539b6510ffde4ce4effdc29";
+    sha256 = "0hv4w95jnlzp9vdximl6bb27fyi75001jhvsbs0ikkd8amq8iaj7";
+  };
+
+  checkInputs = [ django django-crispy-forms djangorestframework mock pytz ];
+
+  checkPhase = "${python.interpreter} runtests.py";
+
+  meta = with lib; {
+    description = "A reusable Django application for allowing users to filter querysets dynamically.";
+    homepage = https://github.com/carltongibson/django-filter;
+    license = licenses.bsd3;
+    maintainers = with maintainers; [ earvstedt ];
+  };
+}

pkgs/applications/office/paperless/withConfig.nix

@@ -0,0 +1,68 @@
+{ paperless, lib, writers }:
+
+## Usage
+#
+# nix-build --out-link ./paperless -E '
+# (import <nixpkgs> {}).paperless.withConfig {
+#   dataDir = /tmp/paperless-data;
+#   config = {
+#     PAPERLESS_DISABLE_LOGIN = "true";
+#   };
+# }'
+#
+# Setup DB
+# ./paperless migrate
+#
+# Consume documents in ${dataDir}/consume
+# ./paperless document_consumer --oneshot
+#
+# Start web interface
+# ./paperless runserver --noreload localhost:8000
+
+{ config ? {}, dataDir ? null, ocrLanguages ? null
+, paperlessPkg ? paperless, extraCmds ? "" }:
+with lib;
+let
+  paperless = if ocrLanguages == null then
+    paperlessPkg
+  else
+    (paperlessPkg.override {
+      tesseract = paperlessPkg.tesseract.override {
+        enableLanguages = ocrLanguages;
+      };
+    }).overrideDerivation (_: {
+      # `ocrLanguages` might be missing some languages required by the tests.
+      doCheck = false;
+    });
+
+  envVars = (optionalAttrs (dataDir != null) {
+    PAPERLESS_CONSUMPTION_DIR = "${dataDir}/consume";
+    PAPERLESS_MEDIADIR = "${dataDir}/media";
+    PAPERLESS_STATICDIR = "${dataDir}/static";
+    PAPERLESS_DBDIR = "${dataDir}";
+  }) // config;
+
+  envVarDefs = mapAttrsToList (n: v: ''export ${n}="${toString v}"'') envVars;
+  setupEnvVars = builtins.concatStringsSep "\n" envVarDefs;
+
+  setupEnv = ''
+    source ${paperless}/share/paperless/setup-env.sh
+    ${setupEnvVars}
+    ${optionalString (dataDir != null) ''
+      mkdir -p "$PAPERLESS_CONSUMPTION_DIR" \
+               "$PAPERLESS_MEDIADIR" \
+               "$PAPERLESS_STATICDIR" \
+               "$PAPERLESS_DBDIR"
+    ''}
+  '';
+
+  runPaperless = writers.writeBash "paperless" ''
+    set -e
+    ${setupEnv}
+    ${extraCmds}
+    exec python $paperlessSrc/manage.py "$@"
+  '';
+in
+  runPaperless // {
+    inherit paperless setupEnv;
+  }

pkgs/top-level/all-packages.nix

@@ -4952,6 +4952,8 @@ in
 
   paper-gtk-theme = callPackage ../misc/themes/paper { };
 
+  paperless = callPackage ../applications/office/paperless { };
+
   paperwork = callPackage ../applications/office/paperwork { };
 
   papertrail = callPackage ../tools/text/papertrail { };