Writes compressed output of given format

package/MDAnalysis/coordinates/GRO.py

@@ -326,7 +326,7 @@ def __init__(self, filename, convert_units=None, n_atoms=None, **kwargs):
                w.write(u.atoms)
 
         """
-        self.filename = util.filename(filename, ext='gro')
+        self.filename = util.filename(filename, ext='gro', keep = True)
         self.n_atoms = n_atoms
         self.reindex = kwargs.pop('reindex', True)
 

package/MDAnalysis/core/groups.py

@@ -2975,7 +2975,7 @@ def improper(self):
                 "improper only makes sense for a group with exactly 4 atoms")
         return topologyobjects.ImproperDihedral(self.ix, self.universe)
 
-    def write(self, filename=None, file_format="PDB",
+    def write(self, filename=None, file_format=None,
               filenamefmt="{trjname}_{frame}", frames=None, **kwargs):
         """Write `AtomGroup` to a file.
 
@@ -3054,8 +3054,7 @@ def write(self, filename=None, file_format="PDB",
         if filename is None:
             trjname, ext = os.path.splitext(os.path.basename(trj.filename))
             filename = filenamefmt.format(trjname=trjname, frame=trj.frame)
-        filename = util.filename(filename, ext=file_format.lower(), keep=True)
-
+        filename = util.filename(filename, ext= file_format or 'PDB', keep=True)
         # Some writer behave differently when they are given a "multiframe"
         # argument. It is the case of the PDB writer tht writes models when
         # "multiframe" is True.
@@ -3080,14 +3079,7 @@ def write(self, filename=None, file_format="PDB",
         # Try and select a Class using get_ methods (becomes `writer`)
         # Once (and if!) class is selected, use it in with block
         try:
-            # format keyword works differently in get_writer and get_selection_writer
-            # here it overrides everything, in get_sel it is just a default
-            # apply sparingly here!
-            format = os.path.splitext(filename)[1][1:]  # strip initial dot!
-            format = format or file_format
-            format = format.strip().upper()
-
-            writer = get_writer_for(filename, format=format, multiframe=multiframe)
+            writer = get_writer_for(filename, format=file_format, multiframe=multiframe)
         except (ValueError, TypeError):
             pass
         else:
@@ -3106,7 +3098,7 @@ def write(self, filename=None, file_format="PDB",
         try:
             # here `file_format` is only used as default,
             # anything pulled off `filename` will be used preferentially
-            writer = get_selection_writer_for(filename, file_format)
+            writer = get_selection_writer_for(filename, file_format or 'PDB')
         except (TypeError, NotImplementedError):
             pass
         else:

package/MDAnalysis/lib/util.py

@@ -249,6 +249,7 @@ def filename(name, ext=None, keep=False):
        Also permits :class:`NamedStream` to pass through.
     """
     if ext is not None:
+        ext = ext.lower()
         if not ext.startswith(os.path.extsep):
             ext = os.path.extsep + ext
         root, origext = os.path.splitext(name)

testsuite/MDAnalysisTests/core/test_atomgroup.py

@@ -167,7 +167,27 @@ def test_write_frame_none(self, u, tmpdir, extension):
             u.atoms.positions[None, ...], new_positions, decimal=2
         )
 
+    @pytest.mark.parametrize('extension', ('xtc', 'dcd', 'pdb', 'xyz'))
+    def test_compressed_write_frame_none(self, u, tmpdir, extension):
+        for ext in ('.gz', '.bz2'):
+            destination = str(tmpdir / 'test.' + extension + ext)
+            u.atoms.write(destination, frames=None)
+            u_new = mda.Universe(destination)
+            new_positions = np.stack([ts.positions for ts in u_new.trajectory])
+            assert_array_almost_equal(
+                u.atoms.positions[None, ...], new_positions, decimal=2
+        )
+
     def test_write_frames_all(self, u, tmpdir):
+        for ext in ('.gz', '.bz2'):
+            destination = str(tmpdir / 'test.dcd') + ext
+            u.atoms.write(destination, frames='all')
+            u_new = mda.Universe(destination)
+            ref_positions = np.stack([ts.positions for ts in u.trajectory])
+            new_positions = np.stack([ts.positions for ts in u_new.trajectory])
+            assert_array_almost_equal(new_positions, ref_positions)
+
+    def test_compressed_rite_frames_all(self, u, tmpdir):
         destination = str(tmpdir / 'test.dcd')
         u.atoms.write(destination, frames='all')
         u_new = mda.Universe(destination)
@@ -238,6 +258,16 @@ def test_write_atoms(self, universe, outfile):
             err_msg=("atom coordinate mismatch between original and {0!s} file"
                      "".format(self.ext)))
 
+    def test_compressed_write_atoms(self, universe, outfile):
+        for compressed_ext in ('.gz', '.bz2'):
+            universe.atoms.write(outfile + compressed_ext)
+            u2 = self.universe_from_tmp(outfile + compressed_ext)
+            assert_almost_equal(
+                universe.atoms.positions, u2.atoms.positions,
+                self.precision,
+                err_msg=("atom coordinate mismatch between original and {0!s} file"
+                        "".format(self.ext)))
+
     def test_write_empty_atomgroup(self, universe, outfile):
         sel = universe.select_atoms('name doesntexist')
         with pytest.raises(IndexError):