Merge pull request #253417 from npatsakula/natasha

python311Packages.natasha: init at 1.6.0

pkgs/development/python-modules/ipymarkup/default.nix

@@ -0,0 +1,29 @@
+{ lib
+, buildPythonPackage
+, fetchPypi
+, intervaltree
+}:
+
+buildPythonPackage rec {
+  pname = "ipymarkup";
+  version = "0.9.0";
+  format = "setuptools";
+
+  src = fetchPypi {
+    inherit pname version;
+    hash = "sha256-P0v6EP1mKTIBr4SEp+a8tyO/LjPZpqoAiCZxr5yiaRE=";
+  };
+
+  propagatedBuildInputs = [ intervaltree ];
+  pythonImportCheck = [ "ipymarkup" ];
+
+  # Upstream has no tests:
+  doCheck = false;
+
+  meta = with lib; {
+    description = "Collection of NLP visualizations for NER and syntax tree markup";
+    homepage = "https://github.com/natasha/ipymarkup";
+    license = licenses.mit;
+    maintainers = with maintainers; [ npatsakula ];
+  };
+}

pkgs/development/python-modules/natasha/default.nix

@@ -0,0 +1,35 @@
+{ lib
+, buildPythonPackage
+, fetchPypi
+, pymorphy2
+, razdel
+, navec
+, slovnet
+, yagry
+, ipymarkup
+, pytestCheckHook
+}:
+
+buildPythonPackage rec {
+  pname = "natasha";
+  version = "1.6.0";
+  format = "setuptools";
+
+  src = fetchPypi {
+    inherit pname version;
+    hash = "sha256-Rgguazgq06a8B9jeRnfHD5VTR+Xrd+8OCsQUfaGLEq0=";
+  };
+
+  propagatedBuildInputs = [ pymorphy2 navec razdel slovnet yagry ipymarkup ];
+
+  nativeCheckInputs = [ pytestCheckHook ];
+  pytestFlagsArray = [ "tests/" ];
+  pythonImportCheck = [ "natasha" ];
+
+  meta = with lib; {
+    description = "NLP framework for Russian language";
+    homepage = "https://github.com/natasha/natasha";
+    license = licenses.mit;
+    maintainers = with maintainers; [ npatsakula ];
+  };
+}

pkgs/development/python-modules/navec/default.nix

@@ -0,0 +1,32 @@
+{ lib
+, buildPythonPackage
+, fetchPypi
+, numpy
+, razdel
+, gensim
+, pytestCheckHook
+}:
+
+buildPythonPackage rec {
+  pname = "navec";
+  version = "0.10.0";
+  format = "setuptools";
+
+  src = fetchPypi {
+    inherit pname version;
+    hash = "sha256-TyNHSxwnmvbGBfhOeHPofEfKWLDFOKP50w2QxgnJ/SE=";
+  };
+
+  propagatedBuildInputs = [ numpy razdel ];
+  nativeCheckInputs = [ pytestCheckHook gensim ];
+  # TODO: remove when gensim usage will be fixed in `navec`.
+  disabledTests = [ "test_gensim" ];
+  pythonImportCheck = [ "navec" ];
+
+  meta = with lib; {
+    description = "Compact high quality word embeddings for Russian language";
+    homepage = "https://github.com/natasha/navec";
+    license = licenses.mit;
+    maintainers = with maintainers; [ npatsakula ];
+  };
+}

pkgs/development/python-modules/razdel/default.nix

@@ -0,0 +1,27 @@
+{ lib
+, buildPythonPackage
+, fetchPypi
+, pytestCheckHook
+}:
+
+buildPythonPackage rec {
+  pname = "razdel";
+  version = "0.5.0";
+  format = "setuptools";
+
+  src = fetchPypi {
+    inherit pname version;
+    hash = "sha256-QzTA/f401OiIzw7YVJaMnfFPClR9+Qmnf0Y0+f/mJuY=";
+  };
+
+  nativeCheckInputs = [ pytestCheckHook ];
+  pytestFlagsArray = [ "razdel" ];
+  pythonImportCheck = [ "razdel" ];
+
+  meta = with lib; {
+    description = "Rule-based system for Russian sentence and word tokenization";
+    homepage = "https://github.com/natasha/razdel";
+    license = licenses.mit;
+    maintainers = with maintainers; [ npatsakula ];
+  };
+}

pkgs/development/python-modules/slovnet/default.nix

@@ -0,0 +1,35 @@
+{ lib
+, buildPythonPackage
+, fetchPypi
+, numpy
+, razdel
+, navec
+, pytestCheckHook
+}:
+
+buildPythonPackage rec {
+  pname = "slovnet";
+  version = "0.6.0";
+  format = "setuptools";
+
+  src = fetchPypi {
+    inherit pname version;
+    hash = "sha256-AtIle9ybnMHSQr007iyGHGSPcIPveJj+FGirzDge95k=";
+  };
+
+  propagatedBuildInputs = [ numpy navec razdel ];
+  nativeCheckInputs = [ pytestCheckHook ];
+  pytestFlagsArray = [ "tests/" ];
+  disabledTestPaths = [
+    # Tries to download model binary artifacts:
+    "tests/test_api.py"
+  ];
+  pythonImportCheck = [ "slovnet" ];
+
+  meta = with lib; {
+    description = "Deep-learning based NLP modeling for Russian language";
+    homepage = "https://github.com/natasha/slovnet";
+    license = licenses.mit;
+    maintainers = with maintainers; [ npatsakula ];
+  };
+}

pkgs/development/python-modules/yargy/default.nix

@@ -0,0 +1,29 @@
+{ lib
+, buildPythonPackage
+, fetchPypi
+, pymorphy2
+, pytestCheckHook
+}:
+
+buildPythonPackage rec {
+  pname = "yargy";
+  version = "0.16.0";
+  format = "setuptools";
+
+  src = fetchPypi {
+    inherit pname version;
+    hash = "sha256-yRfu+zKkDCPEa2yojWiScHLdAKuU6Q/V3GqwpitZtZM=";
+  };
+
+  propagatedBuildInputs = [ pymorphy2 ];
+  pythonImportCheck = [ "yargy" ];
+  nativeCheckInputs = [ pytestCheckHook ];
+  pytestFlagsArray = [ "tests" ];
+
+  meta = with lib; {
+    description = "Rule-based facts extraction for Russian language";
+    homepage = "https://github.com/natasha/yargu";
+    license = licenses.mit;
+    maintainers = with maintainers; [ npatsakula ];
+  };
+}

pkgs/top-level/python-packages.nix

@@ -5322,6 +5322,8 @@ self: super: with self; {
 
   ipykernel = callPackage ../development/python-modules/ipykernel { };
 
+  ipymarkup = callPackage ../development/python-modules/ipymarkup { };
+
   ipympl = callPackage ../development/python-modules/ipympl { };
 
   ipyparallel = callPackage ../development/python-modules/ipyparallel { };
@@ -7021,6 +7023,10 @@ self: super: with self; {
 
   nanoleaf = callPackage ../development/python-modules/nanoleaf { };
 
+  navec = callPackage ../development/python-modules/navec { };
+
+  natasha = callPackage ../development/python-modules/natasha { };
+
   nomadnet = callPackage ../development/python-modules/nomadnet { };
 
   nox = callPackage ../development/python-modules/nox { };
@@ -11065,6 +11071,8 @@ self: super: with self; {
 
   ray = callPackage ../development/python-modules/ray { redis = pkgs.redis; };
 
+  razdel = callPackage ../development/python-modules/razdel { };
+
   rbtools = callPackage ../development/python-modules/rbtools { };
 
   rchitect = callPackage ../development/python-modules/rchitect { };
@@ -11952,6 +11960,8 @@ self: super: with self; {
 
   slob = callPackage ../development/python-modules/slob { };
 
+  slovnet = callPackage ../development/python-modules/slovnet { };
+
   slowapi = callPackage ../development/python-modules/slowapi { };
 
   slugid = callPackage ../development/python-modules/slugid { };
@@ -14070,6 +14080,8 @@ self: super: with self; {
 
   yarg = callPackage ../development/python-modules/yarg { };
 
+  yagry = callPackage ../development/python-modules/yargy { };
+
   yark = callPackage ../development/python-modules/yark { };
 
   yarl = callPackage ../development/python-modules/yarl { };