gensim: added Gensim 2.1.0 and related tools

lib/maintainers.nix

@@ -478,6 +478,7 @@
   schneefux = "schneefux <schneefux+nixos_pkg@schneefux.xyz>";
   schristo = "Scott Christopher <schristopher@konputa.com>";
   scolobb = "Sergiu Ivanov <sivanov@colimite.fr>";
+  sdll = "Sasha Illarionov <sasha.delly@gmail.com>";
   sepi = "Raffael Mancini <raffael@mancini.lu>";
   seppeljordan = "Sebastian Jordan <sebastian.jordan.mail@googlemail.com>";
   shanemikel = "Shane Pearlman <shanemikel1@gmail.com>";

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

@@ -0,0 +1,53 @@
+{ stdenv
+, buildPythonPackage
+, fetchPypi
+, python
+, numpy
+, scipy
+, six
+, smart_open
+, testfixtures
+, unittest2
+, pyro4
+, pyemd
+, scikitlearn
+, morfessor
+}:
+
+buildPythonPackage rec {
+  name = "${pname}-${version}";
+  pname = "gensim";
+  version = "2.1.0";
+
+  src = fetchPypi {
+    inherit pname version;
+    sha256 = "1wn7bji9b80wn1yggmh7a0dlwzdjr6cp24x4p33j2rf29lxnm2kc";
+  };
+
+  propagatedBuildInputs = [
+    numpy
+    scipy
+    six
+    smart_open
+  ];
+
+  buildInputs = [
+    testfixtures
+    unittest2
+    pyro4
+    pyemd
+    scikitlearn
+    morfessor
+  ];
+
+  checkPhase = ''
+    ${python.interpreter} setup.py test
+  '';
+
+  meta = with stdenv.lib; {
+    description = "a Python library for topic modelling, document indexing and similarity retrieval with large corpora";
+    homepage = https://github.com/RaRe-Technologies/gensim;
+    license = licenses.lgpl21;
+    maintainers = with maintainers; [ sdll ];
+    };
+}

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

@@ -0,0 +1,27 @@
+{ stdenv
+, buildPythonPackage
+, fetchPypi
+, python
+}:
+
+buildPythonPackage rec {
+  name = "${pname}-${version}";
+  pname = "Morfessor";
+  version = "2.0.2a4";
+
+  src = fetchPypi {
+    inherit pname version;
+    sha256 = "1c0qbr241vs7zamj6s3l14fgw3ssca4b58kx351jvgc1pypb70s3";
+  };
+
+  checkPhase = ''
+    ${python.interpreter} setup.py test
+  '';
+
+  meta = with stdenv.lib; {
+    description = "Python Implementation and Extensions for Morfessor Baseline, a tool for unsupervised and semi-supervised morphological segmentation";
+    homepage = https://morfessor.readthedocs.org/;
+    license = licenses.bsdOriginal;
+    maintainers = with maintainers; [ sdll ];
+    };
+}

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

@@ -0,0 +1,32 @@
+{ stdenv
+, buildPythonPackage
+, fetchPypi
+, python
+, numpy
+}:
+
+buildPythonPackage rec {
+  name = "${pname}-${version}";
+  pname = "pyemd";
+  version = "0.4.3";
+
+  src = fetchPypi {
+    inherit pname version;
+    sha256 = "1as95f9ppwrxj4hs8980b89vyjza3d8cxf0gf0x75n2gxlymgss8";
+  };
+
+  propagatedBuildInputs = [
+    numpy
+  ];
+
+  checkPhase = ''
+    ${python.interpreter} setup.py test
+  '';
+
+  meta = with stdenv.lib; {
+    description = "Fast EMD for Python";
+    homepage = https://github.com/wmayner/pyemd;
+    license = licenses.mit;
+    maintainers = with maintainers; [ sdll ];
+    };
+}

pkgs/development/python-modules/pyro4/default.nix

@@ -0,0 +1,63 @@
+{ stdenv
+, buildPythonPackage
+, fetchPypi
+, python
+, serpent
+, dill
+, msgpack
+, six
+, isPy27
+}:
+let 
+  selectors34 = buildPythonPackage rec {
+    name = "${pname}-${version}";
+    pname = "selectors34";
+    version = "1.1";
+
+    src = fetchPypi {
+      inherit pname version;      
+      sha256 = "0lmw1fp60qzcl1a0n0vfzfn7j7mrkbkiag6ipkmilij6j0xp9cw4";
+    };
+
+    propagatedBuildInputs = [ six ];
+
+    doCheck = isPy27;
+
+    meta = with stdenv.lib; {
+      description = "a backport of the selectors module from Python 3.4";
+      homepage = https://github.com/berkerpeksag/selectors34;
+      license = licenses.psfl;
+      maintainers = with maintainers; [ sdll ];
+    };
+
+  };
+in buildPythonPackage rec {
+  name = "${pname}-${version}";
+  pname = "Pyro4";
+  version = "4.59";
+
+  src = fetchPypi {
+    inherit pname version;
+    sha256 = "0f1anx4gg5w719hvh104bixldzdylkw04m3zmganyfx8sbdxlfba";
+  };
+
+  propagatedBuildInputs = [
+    selectors34
+    serpent
+  ];
+  buildInputs = [    
+    msgpack
+    dill
+  ];
+
+  checkPhase = ''
+    ${python.interpreter} setup.py test
+  '';
+
+  meta = with stdenv.lib; {
+    description = "distributed object middleware for Python (RPC)";
+    homepage = https://github.com/irmen/Pyro4;
+    license = licenses.mit;
+    maintainers = with maintainers; [ sdll ];
+    };
+}

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

@@ -0,0 +1,28 @@
+{ stdenv
+, buildPythonPackage
+, fetchPypi
+, python
+}:
+
+buildPythonPackage rec {
+
+  name = "${pname}-${version}";
+  pname = "serpent";
+  version = "1.19";
+
+  src = fetchPypi {
+    inherit pname version;
+    sha256 = "1kh76xjkn9ilcjgipg4yai4gx6zhv2nrvbvm6pgp36c8gy1jqswl";
+  };
+
+  checkPhase = ''
+    ${python.interpreter} setup.py test
+  '';
+
+  meta = with stdenv.lib; {
+    description = "serializer for literal Python expressions";
+    homepage = https://github.com/irmen/Serpent;
+    license = licenses.mit;
+    maintainers = with maintainers; [ sdll ];
+    };
+}

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

@@ -0,0 +1,65 @@
+{ stdenv
+, buildPythonPackage
+, fetchFromGitHub
+, fetchPypi
+, python
+, requests
+, boto
+, moto
+, responses
+, isPy3k
+, mock
+}:
+let
+  bz2file = buildPythonPackage rec {
+    name = "${pname}-${version}";
+    pname = "bz2file";
+    version = "0.98";
+
+    src = fetchPypi {
+      inherit pname version;      
+      sha256 = "126s53fkpx04f33a829yqqk8fj4png3qwg4m66cvlmhmwc8zihb4";
+    };
+
+    # fails to find a test.support module
+    doCheck = false;
+
+    meta = with stdenv.lib; {
+      description = "a Python library for reading and writing bzip2-compressed files";
+      homepage = https://github.com/nvawda/bz2file;
+      license = licenses.asl20;
+      maintainers = with maintainers; [ sdll ];
+    };    
+  };
+in buildPythonPackage rec {
+  name = "${pname}-${version}";
+  pname = "smart_open";
+  version = "1.5.2";
+
+  src = fetchPypi {
+    inherit pname version;
+    sha256 = "1nkw6i9ijmbmx01j8ssb3n7v2va9in2y5z3ksficcxyh62mdf7i3";
+  };
+
+  propagatedBuildInputs = [
+    boto
+    requests
+    bz2file
+  ];
+  buildInputs = [
+    mock
+    moto
+    responses    
+  ];
+
+  checkPhase = ''
+    ${python.interpreter} setup.py test
+  '';
+
+  meta = with stdenv.lib; {
+    description = "Utils for streaming large files (S3, HDFS, gzip, bz2...)";
+    homepage = https://github.com/RaRe-Technologies/smart_open;
+    license = licenses.mit;
+    maintainers = with maintainers; [ sdll ];
+    };
+}

pkgs/top-level/python-packages.nix

@@ -31346,6 +31346,18 @@ EOF
 
   distro = callPackage ../development/python-modules/distro { };
 
+  smart_open = callPackage ../development/python-modules/smart_open { };
+
+  serpent = callPackage ../development/python-modules/serpent { };
+
+  pyro4 = callPackage ../development/python-modules/pyro4 { };    
+
+  pyemd = callPackage ../development/python-modules/pyemd { };
+
+  morfessor = callPackage ../development/python-modules/morfessor { };
+
+  gensim = callPackage ../development/python-modules/gensim { };  
+
 });
 
 in fix' (extends overrides packages)