From 2963d76ca34c4ccddf81348194ca5335f30787a3 Mon Sep 17 00:00:00 2001
From: Shea Levy <shea@shealevy.com>
Date: Thu, 8 Mar 2018 10:22:37 -0500
Subject: [PATCH] pythonPackages.pyspark: Init at 2.3.0

---
 .../python-modules/pyspark/default.nix        | 28 +++++++++++++++++++
 pkgs/top-level/python-packages.nix            |  2 ++
 2 files changed, 30 insertions(+)
 create mode 100644 pkgs/development/python-modules/pyspark/default.nix

diff --git a/pkgs/development/python-modules/pyspark/default.nix b/pkgs/development/python-modules/pyspark/default.nix
new file mode 100644
index 00000000000..3447c64cf6b
--- /dev/null
+++ b/pkgs/development/python-modules/pyspark/default.nix
@@ -0,0 +1,28 @@
+{ buildPythonPackage, fetchPypi, stdenv, py4j }:
+
+buildPythonPackage rec {
+  pname = "pyspark";
+  version = "2.3.0";
+
+  src = fetchPypi {
+    inherit pname version;
+    sha256 = "0vlq07yqy6c7ayg401i0qynnliqz405bmw1r8alkck0m1s8kcd8b";
+  };
+
+  # pypandoc is broken with pandoc2, so we just lose docs.
+  postPatch = ''
+    sed -i "s/'pypandoc'//" setup.py
+  '';
+
+  propagatedBuildInputs = [ py4j ];
+
+  # Tests assume running spark...
+  doCheck = false;
+
+  meta = with stdenv.lib; {
+    description = "Apache Spark";
+    homepage = https://github.com/apache/spark/tree/master/python;
+    license = licenses.asl20;
+    maintainers = [ maintainers.shlevy ];
+  };
+}
diff --git a/pkgs/top-level/python-packages.nix b/pkgs/top-level/python-packages.nix
index 5149f1c4aa9..c5ab8def731 100644
--- a/pkgs/top-level/python-packages.nix
+++ b/pkgs/top-level/python-packages.nix
@@ -20654,6 +20654,8 @@ EOF
 
   spacy_models = callPackage ../development/python-modules/spacy/models.nix { };
 
+  pyspark = callPackage ../development/python-modules/pyspark { };
+
   sseclient = callPackage ../development/python-modules/sseclient { };
 
   textacy = callPackage ../development/python-modules/textacy { };