[ENH] Do not save the original name and time stamp of gzip files (#295)

* [ENH] Do not save the original name and time stamp of gzip files

Fixes nipreps/fmriprep#1480

* fix indentation

* add new test

* do not hardlink gzipped files

Author: oesteban

niworkflows/utils/misc.py

@@ -140,7 +140,7 @@ def _copy_any(src, dst):
 
     src_isgz = src.endswith('.gz')
     dst_isgz = dst.endswith('.gz')
-    if src_isgz == dst_isgz:
+    if not src_isgz and not dst_isgz:
         copyfile(src, dst, copy=True, use_hardlink=True)
         return False  # Make sure we do not reuse the hardlink later
 
@@ -149,10 +149,16 @@ def _copy_any(src, dst):
         os.unlink(dst)
 
     src_open = gzip.open if src_isgz else open
-    dst_open = gzip.open if dst_isgz else open
     with src_open(src, 'rb') as f_in:
-        with dst_open(dst, 'wb') as f_out:
-            copyfileobj(f_in, f_out)
+        with open(dst, 'wb') as f_out:
+            if dst_isgz:
+                # Remove FNAME header from gzip (poldracklab/fmriprep#1480)
+                gz_out = gzip.GzipFile('', 'wb', 9, f_out, 0.)
+                copyfileobj(f_in, gz_out)
+                gz_out.close()
+            else:
+                copyfileobj(f_in, f_out)
+
     return True
 
 

niworkflows/utils/tests/test_utils.py

@@ -0,0 +1,24 @@
+"""Test utils"""
+from pathlib import Path
+from tempfile import TemporaryDirectory
+from subprocess import check_call
+from niworkflows.utils.misc import _copy_any
+
+
+def test_copy_gzip():
+    with TemporaryDirectory() as tmpdir:
+        tmppath = Path(tmpdir)
+        filepath = tmppath / 'name1.txt'
+        filepath2 = tmppath / 'name2.txt'
+        assert not filepath2.exists()
+        open(str(filepath), 'w').close()
+        check_call(['gzip', '-N', str(filepath)])
+        assert not filepath.exists()
+
+        gzpath1 = '%s/%s' % (tmppath, 'name1.txt.gz')
+        gzpath2 = '%s/%s' % (tmppath, 'name2.txt.gz')
+        _copy_any(gzpath1, gzpath2)
+        assert Path(gzpath2).exists()
+        check_call(['gunzip', '-N', '-f', gzpath2])
+        assert not filepath.exists()
+        assert filepath2.exists()