✨ Add occupancy_tolerance input

src/aiida_codtools/calculations/functions/primitive_structure_from_cif.py

@@ -9,7 +9,8 @@
 
 
 @calcfunction
-def primitive_structure_from_cif(cif, parse_engine, symprec, site_tolerance):
+def primitive_structure_from_cif(cif, parse_engine, symprec, site_tolerance, occupancy_tolerance):
+    # pylint: disable=too-many-return-statements
     """Attempt to parse the given `CifData` and create a `StructureData` from it.
 
     First the raw CIF file is parsed with the given `parse_engine`. The resulting `StructureData` is then passed through
@@ -21,12 +22,19 @@ def primitive_structure_from_cif(cif, parse_engine, symprec, site_tolerance):
     :param symprec: a `Float` node with symmetry precision for determining primitive cell in SeeKpath
     :param site_tolerance: a `Float` node with the fractional coordinate distance tolerance for finding overlapping
         sites. This will only be used if the parse_engine is pymatgen
+    :param occupancy_tolerance: a `Float` node with the occupancy tolerance below which occupancies will be scaled down
+        to 1. This will only be used if the parse_engine is pymatgen
     :return: the primitive `StructureData` as determined by SeeKpath
     """
     CifCleanWorkChain = WorkflowFactory('codtools.cif_clean')  # pylint: disable=invalid-name
 
     try:
-        structure = cif.get_structure(converter=parse_engine.value, site_tolerance=site_tolerance.value, store=False)
+        structure = cif.get_structure(
+            converter=parse_engine.value,
+            site_tolerance=site_tolerance.value,
+            occupancy_tolerance=occupancy_tolerance.value,
+            store=False
+        )
     except exceptions.UnsupportedSpeciesError:
         return CifCleanWorkChain.exit_codes.ERROR_CIF_HAS_UNKNOWN_SPECIES
     except InvalidOccupationsError:

src/aiida_codtools/cli/workflows/cif_clean.py

@@ -39,12 +39,15 @@
 @click.option(
     '-p', '--parse-engine', type=click.Choice(['ase', 'pymatgen']), default='pymatgen', show_default=True,
     help='Select the parse engine for parsing the structure from the cleaned cif if requested.')
+@click.option(
+    '-O', '--occupancy-tolerance', type=click.FLOAT, default=1.0, show_default=True,
+    help='If total occupancy of a site is between 1 and occupancy_tolerance, the occupancies will be scaled down to 1.')
 @click.option(
     '-d', '--daemon', is_flag=True, default=False, show_default=True,
     help='Submit the process to the daemon instead of running it locally.')
 @decorators.with_dbenv()
 def launch_cif_clean(cif_filter, cif_select, group_cif_raw, group_cif_clean, group_structure, group_workchain, node,
-    max_entries, skip_check, parse_engine, daemon):
+    max_entries, skip_check, parse_engine, occupancy_tolerance, daemon):
     """Run the `CifCleanWorkChain` on the entries in a group with raw imported CifData nodes.
 
     It will use the `cif_filter` and `cif_select` scripts of `cod-tools` to clean the input cif file. Additionally, if
@@ -132,8 +135,7 @@ def launch_cif_clean(cif_filter, cif_select, group_cif_raw, group_cif_clean, gro
     })
 
     node_parse_engine = get_input_node(orm.Str, parse_engine)
-    node_site_tolerance = get_input_node(orm.Float, 5E-4)
-    node_symprec = get_input_node(orm.Float, 5E-3)
+    node_occupancy_tolerance = get_input_node(orm.Float, occupancy_tolerance)
 
     for cif in nodes:
 
@@ -155,6 +157,7 @@ def launch_cif_clean(cif_filter, cif_select, group_cif_raw, group_cif_clean, gro
             },
             'parse_engine': node_parse_engine,
             'site_tolerance': node_site_tolerance,
+            'occupancy_tolerance': node_occupancy_tolerance,
             'symprec': node_symprec,
         }
 

src/aiida_codtools/workflows/cif_clean.py

@@ -34,6 +34,9 @@ def define(cls, spec):
             help='The symmetry precision used by SeeKpath for crystal symmetry refinement.')
         spec.input('site_tolerance', valid_type=orm.Float, default=lambda: orm.Float(5E-4),
             help='The fractional coordinate distance tolerance for finding overlapping sites (pymatgen only).')
+        spec.input('occupancy_tolerance', valid_type=orm.Float, default=orm.Float(1.0),
+            help='If total occupancy of a site is between 1 and occupancy_tolerance, the occupancies will be scaled'\
+                    + 'down to 1 (pymatgen only).')
         spec.input('group_cif', valid_type=orm.Group, required=False, non_db=True,
             help='An optional Group to which the final cleaned CifData node will be added.')
         spec.input('group_structure', valid_type=orm.Group, required=False, non_db=True,
@@ -141,6 +144,7 @@ def parse_cif_structure(self):
             'cif': self.ctx.cif,
             'parse_engine': self.inputs.parse_engine,
             'site_tolerance': self.inputs.site_tolerance,
+            'occupancy_tolerance': self.inputs.occupancy_tolerance,
             'symprec': self.inputs.symprec,
             'metadata': {
                 'call_link_label': 'primitive_structure_from_cif'