From cb251da8b576b376afcba11eed7a69bf5571c87b Mon Sep 17 00:00:00 2001
From: Jonathan Helmus <jjhelmus@gmail.com>
Date: Sat, 16 Sep 2017 12:09:02 -0500
Subject: [PATCH 1/2] ENH: pyfive.Dataset.dims attribute

The dims attribute provide access to dimension labels and scales on
HDF5 datasets.
---
 pyfive/high_level.py | 40 ++++++++++++++++++++++++++++++++++++++--
 1 file changed, 38 insertions(+), 2 deletions(-)

diff --git a/pyfive/high_level.py b/pyfive/high_level.py
index fbc41344..bc52bf5d 100644
--- a/pyfive/high_level.py
+++ b/pyfive/high_level.py
@@ -1,6 +1,6 @@
 """ High-level classes for reading HDF5 files.  """
 
-from collections import Mapping, deque
+from collections import Mapping, deque, Sequence
 from io import open     # Python 2.7 requires for a Buffered Reader
 
 import numpy as np
@@ -299,7 +299,7 @@ def fillvalue(self):
     @property
     def dims(self):
         """ dims attribute. """
-        raise NotImplementedError
+        return DimensionManager(self)
 
     @property
     def attrs(self):
@@ -307,3 +307,39 @@ def attrs(self):
         if self._attrs is None:
             self._attrs = self._dataobjects.get_attributes()
         return self._attrs
+
+
+class DimensionManager(Sequence):
+    """ Represents a collection of dimensions associated with a dataset. """
+    def __init__(self, dset):
+        ndim = len(dset.shape)
+        dim_list = [[]]*ndim
+        if 'DIMENSION_LIST' in dset.attrs:
+            dim_list = dset.attrs['DIMENSION_LIST']
+        dim_labels = [b'']*ndim
+        if 'DIMENSION_LABELS' in dset.attrs:
+            dim_labels = dset.attrs['DIMENSION_LABELS']
+        self._dims = [
+            DimensionProxy(dset.file, label, refs) for
+            label, refs in zip(dim_labels, dim_list)]
+
+    def __len__(self):
+        return len(self._dims)
+
+    def __getitem__(self, x):
+        return self._dims[x]
+
+
+class DimensionProxy(Sequence):
+    """ Represents a HDF5 "dimension". """
+
+    def __init__(self, dset_file, label, refs):
+        self.label = label.decode('utf-8')
+        self._refs = refs
+        self._file = dset_file
+
+    def __len__(self):
+        return len(self._refs)
+
+    def __getitem__(self, x):
+        return self._file[self._refs[x]]

From fe71d7b919a8d13af51e5cca3c92e8569d06313c Mon Sep 17 00:00:00 2001
From: Jonathan Helmus <jjhelmus@gmail.com>
Date: Sat, 16 Sep 2017 12:12:30 -0500
Subject: [PATCH 2/2] TST: unit tests for dim scales and labels

---
 tests/dim_scales.hdf5    | Bin 0 -> 8524 bytes
 tests/make_dim_scales.py |  39 ++++++++++++++++++++++++
 tests/test_dim_scales.py |  64 +++++++++++++++++++++++++++++++++++++++
 3 files changed, 103 insertions(+)
 create mode 100644 tests/dim_scales.hdf5
 create mode 100644 tests/make_dim_scales.py
 create mode 100644 tests/test_dim_scales.py

diff --git a/tests/dim_scales.hdf5 b/tests/dim_scales.hdf5
new file mode 100644
index 0000000000000000000000000000000000000000..fc775f37fb7fb0343570ea8b26e8b7380e4693f1
GIT binary patch
literal 8524
zcmeHMy-piJ5T3Jv9K-~ppajT|E*(V*k}4&RIResQB21AElR$w4BoZ!cD$3+dQse<r
zy5s?P1WL;E$;{4thkeKne~x5j7wyi@?Ck9Bx3i0P?%VR>()ify7)c648ls}BqWEno
zd7%>UXs=l%F#6T#_ncM&l0TwplTVuctlq|Y920B8KB#r0`5~=E5P14uA+Q{m-phhN
z*s9+cJuyEsmbMRf>aUn-NWrOT^vG!AwaP_Ls_@%L#h1!}MpefY1JaR$yb8=MDQ`V7
zoo@0x`@i%ohQw!0Qsl~E3Y|Qc%Q?d#pBSm+gBoi}s<{$H(auq0SS2`5kpG%hc>^72
zA+6n}Qmw2l1~kI6)f8djU5w6~eD8S|1rw*Y&PZ>)Icq+f>8&>nyZ3tQMb><Z4!zay
zthGSL&Q?H^G{t+pxUw2osw-=i&7|};PO2DahMhpLk?b79kPiao06(ZmTedj@Wjimf
z($0(jUG{GU0`}q0px5``mLUs|;XyeSNIgS+DWq~zs>-4ta=utAwc%1Nmy($KsRslk
zCnlsyX_fzPXg?W2QIYe%d9by=!?wg4(fLzHsHKEUm8Yy#{)Pu|zM$7V&nJdQp=0QN
z<I&9E@$!mzbWHLFjO<50^Skm|;`ZFO6Z;Q(WOIi9Lf$kv_7QUE<Nk)cVfuF6rhW!H
zxVE4VJGg$}{kUhqgF-%P<3L_C1NsAvf1+7C^zV1khj&4ruWXmbGy1UeAS1`P6DChj
zi`mKEzsE3O`kr_MJOUm8kAO$OBj6G62zUfM0v-X6fJfji5s>$`yzS*3FK>2@ab;t+
zoiQ!E?Ln^}(>jfKC{F3C*m3=i;T&aRID1druvds_X>1umi@?6d++B?CEYg{z(FVsN
z_V(jh=R9wU6VZYix@B3QFU~?h6TE{RvDcOA29bo8WoI9AmK6af#EL-0_gc1OiL12p
z5)Ki+M7zXM3iOjp(VL(;kGmc(#p`jU9B(FW+(+C&#@*hkZ}H~|r`xN*{I+-ZcMcAB
zKOdNK=Pq2fkp1?6^&xfznvLxb>in^uL7l(&k)%PNzr^g_Uf72G`)_IIkpq%yO43sA
bgOv4^fk!~%<x`yHv*RbuA32NL^)vqgBrm==

literal 0
HcmV?d00001

diff --git a/tests/make_dim_scales.py b/tests/make_dim_scales.py
new file mode 100644
index 00000000..25cccec2
--- /dev/null
+++ b/tests/make_dim_scales.py
@@ -0,0 +1,39 @@
+#! /usr/bin/env python
+""" Create a HDF5 file with dimension scales and labels . """
+import h5py
+import numpy as np
+
+f = h5py.File('dim_scales.hdf5', 'w')
+
+common_args = {
+    'track_times': False,
+    'dtype': 'i4',
+}
+
+# dataset with dimension labels and scales
+f.create_dataset('dset1', data=np.ones((4, 3, 2)), **common_args)
+
+# dimension labels
+f['dset1'].dims[0].label = 'z'
+f['dset1'].dims[1].label = 'y'
+f['dset1'].dims[2].label = 'x'
+
+# dimension scales
+f.create_dataset('x1', data=[1, 2], **common_args)
+f.create_dataset('y1', data=[3, 4, 5], **common_args)
+f.create_dataset('z1', data=[0, 10, 20, 30], **common_args)
+f.create_dataset('x2', data=[99, 98], **common_args)
+
+f['dset1'].dims.create_scale(f['x1'], 'x1_name')
+f['dset1'].dims.create_scale(f['y1'], 'y1_name')
+f['dset1'].dims.create_scale(f['z1'], 'z1_name')
+
+f['dset1'].dims[0].attach_scale(f['z1'])
+f['dset1'].dims[1].attach_scale(f['y1'])
+f['dset1'].dims[2].attach_scale(f['x1'])
+f['dset1'].dims[2].attach_scale(f['x2'])
+
+# dataset with no dimension entries
+f.create_dataset('dset2', data=np.ones((4, 3, 2)), **common_args)
+
+f.close()
diff --git a/tests/test_dim_scales.py b/tests/test_dim_scales.py
new file mode 100644
index 00000000..ce97b4b5
--- /dev/null
+++ b/tests/test_dim_scales.py
@@ -0,0 +1,64 @@
+""" Unit tests for pyfive dimension scales. """
+
+import os
+
+from numpy.testing import assert_array_equal
+
+import pyfive
+
+DIRNAME = os.path.dirname(__file__)
+DIM_SCALES_HDF5_FILE = os.path.join(DIRNAME, './dim_scales.hdf5')
+
+
+def test_dim_labels():
+
+    hfile = pyfive.File(DIM_SCALES_HDF5_FILE)
+
+    # dataset with dimension labels
+    dims = hfile['dset1'].dims
+    assert dims[0].label == 'z'
+    assert dims[1].label == 'y'
+    assert dims[2].label == 'x'
+
+    # dataset with no dimension labels
+    dims = hfile['dset2'].dims
+    assert dims[0].label == ''
+    assert dims[1].label == ''
+    assert dims[2].label == ''
+
+    hfile.close()
+
+
+def test_dim_scales():
+
+    hfile = pyfive.File(DIM_SCALES_HDF5_FILE)
+
+    # dataset with dimension scales
+    dims = hfile['dset1'].dims
+
+    assert len(dims) == 3
+
+    assert len(dims[0]) == 1
+    assert len(dims[1]) == 1
+    assert len(dims[2]) == 2
+
+    assert dims[0][0].name == '/z1'
+    assert dims[1][0].name == '/y1'
+    assert dims[2][0].name == '/x1'
+    assert dims[2][1].name == '/x2'
+
+    assert_array_equal(dims[0][0][:], [0, 10, 20, 30])
+    assert_array_equal(dims[1][0][:], [3, 4, 5])
+    assert_array_equal(dims[2][0][:], [1, 2])
+    assert_array_equal(dims[2][1][:], [99, 98])
+
+    # dataset with no dimension scales
+    dims = hfile['dset2'].dims
+
+    assert len(dims) == 3
+
+    assert len(dims[0]) == 0
+    assert len(dims[1]) == 0
+    assert len(dims[2]) == 0
+
+    hfile.close()