Merge branch 'cif-handle-unknown-spacegroup'

Resolve #21 on master.

Author: pavoljuhas

src/diffpy/Structure/Parsers/P_cif.py

@@ -479,6 +479,9 @@ def _parse_space_group_symop_operation_xyz(self, block):
                     short_name=new_short_name,
                     crystal_system=new_crystal_system,
                     symop_list=symop_list)
+        if self.spacegroup is None:
+            emsg = "CIF file has unknown space group identifier {!r}."
+            raise StructureFormatError(emsg.format(sgid))
         self._expandAsymmetricUnit()
         return
 

src/diffpy/Structure/SpaceGroups.py

@@ -84,6 +84,30 @@ def _buildSGLookupTable():
         _sg_lookup_table.setdefault(sg.short_name, sg)
         _sg_lookup_table.setdefault(sg.pdb_name, sg)
         _sg_lookup_table.setdefault(sg.alt_name, sg)
+    # extra aliases obtained from matching code in
+    # cctbx::sgtbx::symbols::find_main_symbol_dict_entry
+    alias_hmname = [
+        ('Pm3', 'P m -3'),
+        ('Pn3', 'P n -3'),
+        ('Fm3', 'F m -3'),
+        ('Fd3', 'F d -3'),
+        ('Im3', 'I m -3'),
+        ('Pa3', 'P a -3'),
+        ('Ia3', 'I a -3'),
+        ('Pm3m', 'P m -3 m'),
+        ('Pn3n', 'P n -3 n'),
+        ('Pm3n', 'P m -3 n'),
+        ('Pn3m', 'P n -3 m'),
+        ('Fm3m', 'F m -3 m'),
+        ('Fm3c', 'F m -3 c'),
+        ('Fd3m', 'F d -3 m'),
+        ('Fd3c', 'F d -3 c'),
+        ('Im3m', 'I m -3 m'),
+        ('Ia3d', 'I a -3 d'),
+    ]
+    for a, hm in alias_hmname:
+        hmbare = hm.replace(' ', '')
+        _sg_lookup_table.setdefault(a, _sg_lookup_table[hmbare])
     # make sure None does not sneak into the dictionary
     if None in _sg_lookup_table:
         del _sg_lookup_table[None]

src/diffpy/Structure/tests/TestP_cif.py

@@ -249,6 +249,14 @@ def test_nosites_cif(self):
         self.assertEqual(10.413, stru.lattice.c)
         return
 
+    def test_badspacegroup_cif(self):
+        """Test reading of CIF file with unrecognized space group.
+        """
+        ptest = self.ptest
+        filename = datafile('badspacegroup.cif')
+        self.assertRaises(StructureFormatError, ptest.parseFile, filename)
+        return
+
     def test_getParser(self):
         """Test passing of eps keyword argument by getParser function.
         """

src/diffpy/Structure/tests/TestSymmetryUtilities.py

@@ -61,6 +61,13 @@ def test_isSpaceGroupLatPar(self):
         self.assertTrue(isSpaceGroupLatPar(cubic, 3, 3, 3, 90, 90, 90))
         return
 
+    def test_sgtbx_spacegroup_aliases(self):
+        """check GetSpaceGroup for non-standard aliases from sgtbx.
+        """
+        self.assertIs(GetSpaceGroup('Fm3m'), GetSpaceGroup(225))
+        self.assertIs(GetSpaceGroup('Ia3d'), GetSpaceGroup('I a -3 d'))
+        return
+
     def test_expandPosition(self):
         """check expandPosition()
         """

src/diffpy/Structure/tests/testdata/badspacegroup.cif

@@ -0,0 +1,49 @@
+#------------------------------------------------------------------------------
+#$Date: 2015-01-27 21:58:39 +0200 (Tue, 27 Jan 2015) $
+#$Revision: 130149 $
+#$URL: svn://www.crystallography.net/cod/cif/1/10/01/1100106.cif $
+#------------------------------------------------------------------------------
+#
+# This file is available in the Crystallography Open Database (COD),
+# http://www.crystallography.net/
+#
+# All data on this site have been placed in the public domain by the
+# contributors.
+#
+data_1100106
+_publ_section_title              'Clay Montmorillonite'
+_chemical_name_common            Montmorillonite
+_symmetry_cell_setting           monoclinic
+_symmetry_space_group_name_H-M   C2:b1
+_audit_update_record             'Last update Thu Dec 18 14:12:16 CET 2003'
+_cell_angle_alpha                90
+_cell_angle_beta                 100.3(1)
+_cell_angle_gamma                90
+_cell_formula_units_Z            1
+_cell_length_a                   5.411(4)
+_cell_length_b                   9.0(5)
+_cell_length_c                   10.25(2)
+_refine_ls_goodness_of_fit_all   1.1109964
+_refine_ls_R_factor_all          0.04240644
+_refine_ls_wR_factor_all         0.06303699
+_cod_database_code               1100106
+loop_
+_atom_site_label
+_atom_site_type_symbol
+_atom_site_occupancy
+_atom_site_fract_x
+_atom_site_fract_y
+_atom_site_fract_z
+_atom_site_B_iso_or_equiv
+Al1 Al3+ 1.00000 0 0 0 0.5
+Al2 Al3+ 1.00000 0 0.321 0 0.5
+Mg1 Mg2+ 1.00000 0 0.654 0 0.5
+Si1 Si4+ 1.00000 0.432 0.333 0.27 0.5
+Si2 Si4+ 1.00000 0.432 0.662 0.27 0.5
+K1 K1+ 1.00000 0.5 0 0.5 0.5
+O1 O2- 1.00000 0.489 0.496 0.335 0.8
+O2 O2- 1.00000 0.173 0.725 0.335 0.8
+O3 O2- 1.00000 0.17 0.268 0.32 0.8
+O4 O2- 1.00000 0.334 -0.024 0.105 0.8
+O5 O2- 1.00000 0.417 0.656 0.109 0.8
+O6 O2- 1.00000 0.343 0.347 0.109 0.8