Bugfix: don't compress non-numeric arrays (#26)

tdegeus · May 25, 2020 · fecfebf · fecfebf
1 parent 454bf2c
commit fecfebf
Show file tree

Hide file tree

Showing 4 changed files with 30 additions and 5 deletions.
diff --git a/GooseHDF5/__init__.py b/GooseHDF5/__init__.py
@@ -3,7 +3,7 @@
 warnings.filterwarnings("ignore")
 
 
-__version__ = '0.7.1'
+__version__ = '0.7.2'
 
 
 def abspath(path):
@@ -504,6 +504,23 @@ def copydatasets(source, dest, source_datasets, dest_datasets=None, root=None):
         group = posixpath.split(dest_path)[0]
         source.copy(source_path, dest[group], posixpath.split(dest_path)[1])
 
+
+def isnumeric(a):
+    r'''
+Returns ``True`` is an array contains numeric values.
+    '''
+
+    import numpy as np
+
+    if type(a) == str:
+        return False
+
+    if np.issubdtype(a.dtype, np.number):
+        return True
+
+    return False
+
+
 def _equal_value(a, b):
 
     import numpy as np

diff --git a/GooseHDF5/cli/G5repack.py b/GooseHDF5/cli/G5repack.py
@@ -16,6 +16,7 @@
 (c - MIT) T.W.J. de Geus | tom@geus.me | www.geus.me | github.com/tdegeus/GooseHDF5
 '''
 
+from .. import isnumeric
 from .. import getpaths
 from .. import __version__
 import docopt
@@ -33,7 +34,7 @@ def copy_dataset(old, new, path, compress):
 
     data = old[path][...]
 
-    if data.size == 1 or not compress:
+    if data.size == 1 or not compress or not isnumeric(data):
         new[path] = old[path][...]
     else:
         dset = new.create_dataset(path, data.shape, compression="gzip")

diff --git a/docs/tools.rst b/docs/tools.rst
@@ -63,7 +63,7 @@ G5print
         Print datasets in a HDF5-file.
 
     Usage:
-        G5print [options] <source> <dataset>...
+        G5print [options] <source> [<dataset>...]
 
     Arguments:
         <source>    HDF5-file.
@@ -139,5 +139,6 @@ G5repack
         <source>    HDF5-file.
 
     Options:
+        -c, --compress  Apply compression (using the loss-less GZip algorithm).
         -h, --help      Show help.
             --version   Show version.
diff --git a/test/cli/G5repack.py b/test/cli/G5repack.py
@@ -8,17 +8,23 @@ def run(cmd):
     return out
 
 a = np.random.random(1000)
+b = 'foo'
 
 with h5py.File('a.hdf5', 'w') as source:
     source['/a'] = a
+    source['/b'] = b
 
 output = run('G5repack -c a.hdf5')
 
 with h5py.File('a.hdf5', 'r') as source:
-    b = source['/a'][...]
+    a_r = source['/a'][...]
+    b_r = source['/b'][...]
 
 os.remove('a.hdf5')
 
-if not np.allclose(a, b):
+if not np.allclose(a, a_r):
+    raise IOError('Test failed')
+
+if b != b_r:
     raise IOError('Test failed')