[CLI] Adding compression & copy of attributes (#24)

GooseHDF5/__init__.py

@@ -3,7 +3,7 @@
 warnings.filterwarnings("ignore")
 
 
-__version__ = '0.6.0'
+__version__ = '0.7.0'
 
 
 def abspath(path):
@@ -504,16 +504,15 @@ 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 _equal(a, b):
+def _equal_value(a, b):
 
     import numpy as np
 
-    if isinstance(a, h5py.Group) and isinstance(b, h5py.Group):
-        return True
-
-    if not isinstance(a, h5py.Dataset) or not isinstance(b, h5py.Dataset):
-        raise IOError('Not a Dataset')
+    if type(a) == str:
+        if type(b) == str:
+            return a == b
+        else:
+            return False
 
     if np.issubdtype(a.dtype, np.number) and np.issubdtype(b.dtype, np.number):
         if np.allclose(a, b):
@@ -530,12 +529,28 @@ def _equal(a, b):
         else:
             return False
 
-    if list(a) == list(b):
+    return list(a) == list(b)
+
+
+def _equal(a, b):
+
+    if isinstance(a, h5py.Group) and isinstance(b, h5py.Group):
         return True
-    else:
-        return False
 
-    return True
+    if not isinstance(a, h5py.Dataset) or not isinstance(b, h5py.Dataset):
+        raise IOError('Not a Dataset')
+
+    for key in a.attrs:
+        if key not in b.attrs:
+            return False
+        if not _equal_value(a.attrs[key], b.attrs[key]):
+            return False
+
+    for key in b.attrs:
+        if key not in a.attrs:
+            return False
+
+    return _equal_value(a, b)
 
 
 def equal(source, dest, source_dataset, dest_dataset=None):

GooseHDF5/cli/G5repack.py

@@ -9,6 +9,7 @@
     <source>    HDF5-file.
 
 Options:
+    -c, --compress  Apply compression (using the loss-less GZip algorithm).
     -h, --help      Show help.
         --version   Show version.
 
@@ -28,6 +29,19 @@ def check_isfile(fname):
     if not os.path.isfile(fname):
         raise IOError('"{0:s}" does not exist'.format(fname))
 
+def copy_dataset(old, new, path, compress):
+
+    data = old[path][...]
+
+    if data.size == 1 or not compress:
+        new[path] = old[path][...]
+    else:
+        dset = new.create_dataset(path, data.shape, compression="gzip")
+        dset[:] = data
+
+    for key in old[path].attrs:
+        new[path].attrs[key] = old[path].attrs[key]
+
 def main():
 
     args = docopt.docopt(__doc__, version=__version__)
@@ -42,6 +56,6 @@ def main():
         with h5py.File(filename, 'r') as source:
             with h5py.File(tempname, 'w') as tmp:
                 for path in getpaths(source):
-                    tmp[path] = source[path][...]
+                    copy_dataset(source, tmp, path, args['--compress'])
 
         os.replace(tempname, filename)

test/cli/G5compare.py

@@ -49,10 +49,24 @@ def run(cmd):
         source['/d/equal'] = d
         other['/e/equal'] = d
 
+        # attribute
+
+        f = np.random.random(25)
+
+        source['/f/equal'] = f
+        source['/f/equal'].attrs['key'] = f
+        source['/f/not_equal'] = f
+        source['/f/not_equal'].attrs['key'] = f
+
+        other['/f/equal'] = f
+        other['/f/equal'].attrs['key'] = f
+        other['/f/not_equal'] = f
+        other['/f/not_equal'].attrs['key'] = np.random.random(25)
+
 
 output = sorted(run("G5compare a.hdf5 b.hdf5 -r /d/equal:/e/equal"))
 
-expected_output = sorted(['!= /a/not_equal', '!= /b/not_equal', '!= /c/not_equal'])
+expected_output = sorted(['!= /a/not_equal', '!= /b/not_equal', '!= /c/not_equal', '!= /f/not_equal'])
 
 os.remove('a.hdf5')
 os.remove('b.hdf5')

test/cli/G5repack.py

@@ -7,18 +7,18 @@ def run(cmd):
     out = list(filter(None, subprocess.check_output(cmd, shell=True).decode('utf-8').split('\n')))
     return out
 
-a = np.random.random(5)
+a = np.random.random(1000)
 
 with h5py.File('a.hdf5', 'w') as source:
     source['/a'] = a
 
-output = run('G5repack a.hdf5')
+output = run('G5repack -c a.hdf5')
 
 with h5py.File('a.hdf5', 'r') as source:
     b = source['/a'][...]
 
 os.remove('a.hdf5')
 
-if not np.all(a == b):
+if not np.allclose(a, b):
     raise IOError('Test failed')