pythonPackages.jieba: init at 0.39

2 files changed, 35 insertions, 0 deletions

diff --git a/pkgs/development/python-modules/jieba/default.nix b/pkgs/development/python-modules/jieba/default.nix
new file mode 100644
index 000000000000..df21c0f41861
--- /dev/null
+++ b/pkgs/development/python-modules/jieba/default.nix
@@ -0,0 +1,33 @@
+{ lib, buildPythonPackage, fetchFromGitHub, glibcLocales, python, isPy3k }:
+
+buildPythonPackage rec {
+  pname = "jieba";
+  version = "0.39";
+
+  # no tests in PyPI tarball
+  src = fetchFromGitHub {
+    owner = "fxsjy";
+    repo = pname;
+    rev = "v${version}";
+    sha256 = "0hbq0l1jbgcvm58qg4p37im4jl5a9igvq3wlhlk22pmbkbvqqgzs";
+  };
+
+  checkInputs = [ glibcLocales ];
+
+  # UnicodeEncodeError
+  doCheck = isPy3k;
+
+  # Citing https://github.com/fxsjy/jieba/issues/384: "testcases is in a mess"
+  # So just picking random ones that currently work
+  checkPhase = ''
+    export LC_ALL=en_US.UTF-8
+    ${python.interpreter} test/test.py
+    ${python.interpreter} test/test_tokenize.py
+  '';
+
+  meta = with lib; {
+    description = "Chinese Words Segementation Utilities";
+    homepage = https://github.com/fxsjy/jieba;
+    license = licenses.mit;
+  };
+}
diff --git a/pkgs/top-level/python-packages.nix b/pkgs/top-level/python-packages.nix
index 1872f1e3bf5a..35524287f724 100644
--- a/pkgs/top-level/python-packages.nix
+++ b/pkgs/top-level/python-packages.nix
@@ -4988,6 +4988,8 @@ in {
 
   jdcal = callPackage ../development/python-modules/jdcal { };
 
+  jieba = callPackage ../development/python-modules/jieba { };
+
   internetarchive = callPackage ../development/python-modules/internetarchive {};
 
   JPype1 = callPackage ../development/python-modules/JPype1 {};