gnu: Add python-cykhash.

* gnu/packages/python-xyz.scm (python-cykhash): New variable.

Change-Id: I7f46da1f53b10957c102ca21e74f144955676730

1 files changed, 27 insertions, 0 deletions

diff --git a/gnu/packages/python-xyz.scm b/gnu/packages/python-xyz.scm
index 433f6ec0fc..5d9d98e4c3 100644
--- a/gnu/packages/python-xyz.scm
+++ b/gnu/packages/python-xyz.scm
@@ -8482,6 +8482,33 @@ Mallard using the @command{ducktype} tool.  The yelp-tools package
 provides additional functionality on the produced Mallard documents.")
     (license license:expat)))
 
+(define-public python-cykhash
+  (package
+    (name "python-cykhash")
+    (version "2.0.1")
+    (source
+     (origin
+       (method url-fetch)
+       (uri (pypi-uri "cykhash" version))
+       (sha256
+        (base32 "1xwpxff4whfvkwvcyhzhdcj5zzq89vvdjmnqy664s4a9yp4lnydl"))))
+    (build-system pyproject-build-system)
+    (native-inputs (list python-cython
+                         python-setuptools
+                         python-wheel))
+    (home-page "https://github.com/realead/cykhash")
+    (synopsis "Khash-sets and maps")
+    (description
+     "This package is a Cython wrapper for khash-sets/maps.  It brings
+functionality of
+@url{https://github.com/attractivechaos/klib/blob/master/khash.h, khash} to
+Python and Cython and can be used seamlessly in numpy or pandas.  Numpy's
+world is lacking the concept of a (hash-)set.  This shortcoming is fixed and
+efficient (memory- and speedwise compared to pandas) @code{unique} and
+@code{isin} are implemented.  Python-set/dict have a big memory-footprint.
+For some datatypes the overhead can be reduced by using khash by factor 4-8.")
+    (license license:expat)))
+
 (define-public python-cython
   (package
     (name "python-cython")