G5compare: adding option to check for data-type (#28)

GooseHDF5/__init__.py

@@ -3,7 +3,7 @@
 warnings.filterwarnings("ignore")
 
 
-__version__ = '0.7.3'
+__version__ = '0.7.4'
 
 
 def abspath(path):

GooseHDF5/cli/G5compare.py

@@ -1,6 +1,7 @@
 '''G5compare
   Compare two HDF5 files. If the function does not output anything all datasets are present in both
-  files, and all the content of the datasets is equals
+  files, and all the content of the datasets is equal.
+  Each output line corresponds to a mismatch between the files.
 
 Usage:
   G5compare [options] [--renamed ARG]... <source> <other>
@@ -10,6 +11,7 @@
   <other>     HDF5-file.
 
 Options:
+  -t, --dtype           Verify that the type of the datasets match.
   -r, --renamed=ARG     Renamed paths, separated by a separator (see below).
   -s, --ifs=ARG         Separator used to separate renamed fields. [default: :]
   -h, --help            Show help.
@@ -35,20 +37,25 @@ def check_isfile(fname):
         raise IOError('"{0:s}" does not exist'.format(fname))
 
 
-def check_dataset(source, dest, source_dataset, dest_dataset=None):
+def check_dataset(source, dest, source_dataset, dest_dataset, check_dtype):
     r'''
 Check if the datasets (read outside) "a" and "b" are equal. If not print a message with the "path"
 to the screen and return "False".
     '''
 
     if not equal(source, dest, source_dataset, dest_dataset):
-        print('!= {0:s}'.format(source_dataset))
+        print(' !=  {0:s}'.format(source_dataset))
         return False
 
+    if check_dtype:
+        if source[source_dataset].dtype != dest[dest_dataset].dtype:
+            print('type {0:s}'.format(source_dataset))
+            return False
+
     return True
 
 
-def _check_plain(source, other):
+def _check_plain(source, other, check_dtype):
     r'''
 Support function for "check_plain."
     '''
@@ -63,19 +70,19 @@ def _check_plain(source, other):
 
     for path in getpaths(source):
         if path in other:
-            check_dataset(source, other, path)
+            check_dataset(source, other, path, path, check_dtype)
 
 
-def check_plain(source_name, other_name):
+def check_plain(source_name, other_name, check_dtype):
     r'''
 Check all datasets (without allowing for renamed datasets).
     '''
     with h5py.File(source_name, 'r') as source:
         with h5py.File(other_name, 'r') as other:
-            _check_plain(source, other)
+            _check_plain(source, other, check_dtype)
 
 
-def _check_renamed(source, other, renamed):
+def _check_renamed(source, other, renamed, check_dtype):
     r'''
 Support function for "check_renamed."
     '''
@@ -89,26 +96,26 @@ def _check_renamed(source, other, renamed):
 
     for _, path in s2o.items():
         if path not in o2s:
-            print('-> {0:s}'.format(path))
+            print(' ->  {0:s}'.format(path))
 
     for _, path in o2s.items():
         if path not in s2o:
-            print('<- {0:s}'.format(path))
+            print(' <-  {0:s}'.format(path))
 
     for new_path, path in s2o.items():
         if new_path in o2s:
-            check_dataset(source, other, path, new_path)
+            check_dataset(source, other, path, new_path, check_dtype)
 
 
-def check_renamed(source_name, other_name, renamed):
+def check_renamed(source_name, other_name, renamed, check_dtype):
     r'''
 Check all datasets while allowing for renamed datasets.
 renamed = [['source_name1', 'other_name1'], ['source_name2', 'other_name2'], ...]
     '''
 
     with h5py.File(source_name, 'r') as source:
         with h5py.File(other_name, 'r') as other:
-            _check_renamed(source, other, renamed)
+            _check_renamed(source, other, renamed, check_dtype)
 
 
 def main():
@@ -122,9 +129,9 @@ def main():
     check_isfile(args['<other>'])
 
     if len(args['--renamed']) == 0:
-        check_plain(args['<source>'], args['<other>'])
+        check_plain(args['<source>'], args['<other>'], args['--dtype'])
         sys.exit(0)
 
     renamed = [i.split(args['--ifs']) for i in args['--renamed']]
 
-    check_renamed(args['<source>'], args['<other>'], renamed)
+    check_renamed(args['<source>'], args['<other>'], renamed, args['--dtype'])

test/cli/G5compare.py

@@ -66,7 +66,7 @@ def run(cmd):
 
 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', '!= /f/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')