Merge pull request #59 from alimanfoo/vlen-array-none

VLen... encode() element can be None

Author: alimanfoo

docs/release.rst

@@ -1,6 +1,13 @@
 Release notes
 =============
 
+.. _release_0.5.2:
+
+0.5.2
+-----
+
+* Add support for encoding None values in VLen... codecs (:issue:`59`).
+
 .. _release_0.5.1:
 
 0.5.1

numcodecs/tests/test_vlen_array.py

@@ -11,8 +11,10 @@
     from numcodecs.vlen import VLenArray
 except ImportError:  # pragma: no cover
     raise nose.SkipTest("vlen-array not available")
-from numcodecs.tests.common import (check_config, check_repr, check_encode_decode_array,
-                                    check_backwards_compatibility)
+from numcodecs.tests.common import (check_config, check_repr,
+                                    check_encode_decode_array,
+                                    check_backwards_compatibility,
+                                    assert_array_items_equal)
 
 
 arrays = [
@@ -88,3 +90,14 @@ def test_decode_errors():
         codec.decode(enc, out=123)
     with assert_raises(ValueError):
         codec.decode(enc, out=np.zeros(10, dtype='i4'))
+
+
+def test_encode_none():
+    a = np.array([[1, 3], None, [4, 7]], dtype=object)
+    codec = VLenArray(int)
+    enc = codec.encode(a)
+    dec = codec.decode(enc)
+    expect = np.array([np.array([1, 3]),
+                       np.array([]),
+                       np.array([4, 7])], dtype=object)
+    assert_array_items_equal(expect, dec)

numcodecs/tests/test_vlen_bytes.py

@@ -12,7 +12,8 @@
 except ImportError:  # pragma: no cover
     raise nose.SkipTest("vlen-bytes not available")
 from numcodecs.tests.common import (check_config, check_repr, check_encode_decode_array,
-                                    check_backwards_compatibility, greetings)
+                                    check_backwards_compatibility, greetings,
+                                    assert_array_items_equal)
 
 
 greetings_bytes = [g.encode('utf-8') for g in greetings]
@@ -53,7 +54,7 @@ def test_encode_errors():
     with assert_raises(TypeError):
         codec.encode([1234, 5678])
     with assert_raises(TypeError):
-        codec.encode(np.zeros(10, dtype='i4'))
+        codec.encode(np.ones(10, dtype='i4'))
 
 
 def test_decode_errors():
@@ -80,3 +81,12 @@ def test_decode_errors():
         codec.decode(enc, out=123)
     with assert_raises(ValueError):
         codec.decode(enc, out=np.zeros(10, dtype='i4'))
+
+
+def test_encode_none():
+    a = np.array([b'foo', None, b'bar'], dtype=object)
+    codec = VLenBytes()
+    enc = codec.encode(a)
+    dec = codec.decode(enc)
+    expect = np.array([b'foo', b'', b'bar'], dtype=object)
+    assert_array_items_equal(expect, dec)

numcodecs/tests/test_vlen_utf8.py

@@ -12,7 +12,8 @@
 except ImportError:  # pragma: no cover
     raise nose.SkipTest("vlen-utf8 not available")
 from numcodecs.tests.common import (check_config, check_repr, check_encode_decode_array,
-                                    check_backwards_compatibility, greetings)
+                                    check_backwards_compatibility, greetings,
+                                    assert_array_items_equal)
 
 
 arrays = [
@@ -49,7 +50,7 @@ def test_encode_errors():
     with assert_raises(TypeError):
         codec.encode([1234, 5678])
     with assert_raises(TypeError):
-        codec.encode(np.zeros(10, dtype='i4'))
+        codec.encode(np.ones(10, dtype='i4'))
 
 
 def test_decode_errors():
@@ -76,3 +77,12 @@ def test_decode_errors():
         codec.decode(enc, out=123)
     with assert_raises(ValueError):
         codec.decode(enc, out=np.zeros(10, dtype='i4'))
+
+
+def test_encode_utf8():
+    a = np.array([u'foo', None, u'bar'], dtype=object)
+    codec = VLenUTF8()
+    enc = codec.encode(a)
+    dec = codec.decode(enc)
+    expect = np.array([u'foo', u'', u'bar'], dtype=object)
+    assert_array_items_equal(expect, dec)