fix ensemble merge conflicts

cctk/gaussian_file.py

@@ -85,7 +85,7 @@ def __init__(
             if not all(isinstance(job, JobType) for job in job_types):
                 raise TypeError(f"invalid job type {job}")
 
-        self.molecules = ConformationalEnsemble()
+        self.ensemble = ConformationalEnsemble()
         self.route_card = route_card
         self.link0 = link0
         self.footer = footer
@@ -94,7 +94,7 @@ def __init__(
         self.success = success
 
     def __str__(self):
-        return f"GaussianFile (title=\"{str(self.title)}\", {len(self.molecules)} entries in Ensemble)"
+        return f"GaussianFile (title=\"{str(self.title)}\", {len(self.ensemble)} entries in Ensemble)"
 
     @classmethod
     def write_molecule_to_file(cls, filename, molecule, route_card, link0={"mem": "32GB", "nprocshared": 16}, footer=None, title="title", append=False, print_symbol=False):
@@ -156,7 +156,7 @@ def write_file(self, filename, molecule=None, route_card=None, link0=None, foote
         Args:
             filename (str): path to the new file
             molecule (int): which molecule to use -- passed to ``self.get_molecule()``.
-                Default is -1 (e.g. the last molecule), but positive integers will select from self.molecules (1-indexed).
+                Default is -1 (e.g. the last molecule), but positive integers will select from self.ensemble(1-indexed).
                 A ``Molecule`` object can also be passed, in which case that molecule will be written to the file.
             route_card (str): route card for new file
             link0 (dict): dictionary of Link 0 commands (e.g. {"mem": "32GB", "nprocshared": 16}
@@ -318,7 +318,7 @@ def read_file(cls, filename, return_lines=False):
                 pass
 
             for mol, prop in zip(molecules, properties):
-                f.molecules.add_molecule(mol, properties=prop)
+                f.ensemble.add_molecule(mol, properties=prop)
 
             f.check_has_properties()
             files.append(f)
@@ -414,7 +414,7 @@ def _read_gjf_file(cls, filename, return_lines=False):
         job_types = cls._assign_job_types(header)
 
         f = GaussianFile(job_types=job_types, route_card=header, link0=link0, footer=footer, title=title)
-        f.molecules.add_molecule(Molecule(atomic_numbers, geometry, charge=charge, multiplicity=multip))
+        f.ensemble.add_molecule(Molecule(atomic_numbers, geometry, charge=charge, multiplicity=multip))
         if return_lines:
             return f, lines
         else:
@@ -424,7 +424,7 @@ def get_molecule(self, num=None):
         """
         Returns the last molecule (from an optimization job) or the only molecule (from other jobs).
 
-        If ``num`` is specified, returns ``self.molecules[num]``
+        If ``num`` is specified, returns ``self.ensemble.molecule_list()[num]``
         """
         # some methods pass num=None, which overrides setting the default above
         if num is None:
@@ -433,7 +433,7 @@ def get_molecule(self, num=None):
         if not isinstance(num, int):
             raise TypeError("num must be int")
 
-        return self.molecules[num]
+        return self.ensemble.molecule_list()[num]
 
     @classmethod
     def _assign_job_types(cls, header):
@@ -460,11 +460,11 @@ def check_has_properties(self):
         """
         Checks that the file has all the appropriate properties for its job types, and raises ValueError if not.
 
-        This only checks the last molecule in ``self.molecules``, for now.
+        This only checks the last molecule in ``self.ensemble``, for now.
         """
         for job_type in self.job_types:
             for prop in EXPECTED_PROPERTIES[job_type.value]:
-                if not self.molecules.has_property(-1, prop):
+                if not self.ensemble.has_property(-1, prop):
                     raise ValueError(f"expected property {prop} for job type {job_type}, but it's not there!")
 
     @classmethod

cctk/mae_file.py

@@ -14,7 +14,7 @@ class MAEFile(File):
 
     Attributes:
         name (str): name of file
-        molecules (Ensemble): ``Ensemble`` or ``ConformationalEnsemble`` object
+        ensemble (Ensemble): ``Ensemble`` or ``ConformationalEnsemble`` object
     """
 
     def __init__(self, name=None):
@@ -42,12 +42,12 @@ def read_file(cls, filename, name=None, **kwargs):
         atomic_numbers = np.array([get_number(z) for z in symbols], dtype=np.int8)
 
         if conformers == True:
-            file.molecules = ConformationalEnsemble()
+            file.ensemble = ConformationalEnsemble()
         else:
-            file.molecules = Ensemble()
+            file.ensemble = Ensemble()
 
         for geom in geometries:
-            file.molecules.add_molecule(Molecule(atomic_numbers, geom, bonds=bonds.edges))
+            file.ensemble.add_molecule(Molecule(atomic_numbers, geom, bonds=bonds.edges))
 
         return file, p_names, p_vals
 
@@ -268,7 +268,7 @@ def get_molecule(self, num=None):
         """
         Returns the last molecule from the ensemble.
 
-        If ``num`` is specified, returns ``self.molecules[num]``
+        If ``num`` is specified, returns ``self.ensemble.molecules[num]``
         """
         # some methods pass num=None, which overrides setting the default above
         if num is None:
@@ -277,4 +277,4 @@ def get_molecule(self, num=None):
         if not isinstance(num, int):
             raise TypeError("num must be int")
 
-        return self.molecules[num]
+        return self.ensemble.molecules[num]

cctk/mol2_file.py

@@ -14,7 +14,7 @@ class MOL2File(File):
 
     Attributes:
         name (str): name of file
-        molecules (Ensemble): ``Ensemble`` or ``ConformationalEnsemble`` object
+        ensemble (Ensemble): ``Ensemble`` or ``ConformationalEnsemble`` object
     """
 
     def __init__(self, name=None):
@@ -40,12 +40,12 @@ def read_file(cls, filename, name=None, **kwargs):
         atomic_numbers = np.array([get_number(z) for z in symbols], dtype=np.int8)
 
         if conformers == True:
-            file.molecules = ConformationalEnsemble()
+            file.ensemble = ConformationalEnsemble()
         else:
-            file.molecules = Ensemble()
+            file.ensemble = Ensemble()
 
         for geom in geometries:
-            file.molecules.add_molecule(Molecule(atomic_numbers, geom, bonds=bonds.edges))
+            file.ensemble.add_molecule(Molecule(atomic_numbers, geom, bonds=bonds.edges))
 
         return file
 
@@ -261,7 +261,7 @@ def get_molecule(self, num=None):
         """
         Returns the last molecule from the ensemble.
 
-        If ``num`` is specified, returns ``self.molecules[num]``
+        If ``num`` is specified, returns ``self.ensemble.molecules[num]``
         """
         # some methods pass num=None, which overrides setting the default above
         if num is None:
@@ -270,7 +270,7 @@ def get_molecule(self, num=None):
         if not isinstance(num, int):
             raise TypeError("num must be int")
 
-        return self.molecules[num]
+        return self.ensemble.molecules[num]
 
     @classmethod
     def write_molecule_to_file(cls, filename, molecule, title=None):
@@ -298,17 +298,16 @@ def write_molecule_to_file(cls, filename, molecule, title=None):
 
         super().write_file(filename, text)
 
-    def write_file(self, filename, molecule=None, **kwargs):
+    def write_file(self, filename, molecule=-1, **kwargs):
         """
         Write a ``.mol2`` file, using object attributes.
 
         Args:
             filename (str): path to the new file
             molecule (int): which molecule to use -- passed to ``self.get_molecule()``.
-                Default is -1 (e.g. the last molecule), but positive integers will select from self.molecules (1-indexed).
+                Default is -1 (e.g. the last molecule), but positive integers will select from self.ensemble.molecules (0-indexed).
                 A ``Molecule`` object can also be passed, in which case that molecule will be written to the file.
         """
-        if not isinstance(molecule, Molecule):
-            molecule = self.get_molecule(molecule)
-
+        if molecule is None or isinstance(molecule, (np.integer, int)):
+            molecule = self.ensemble.molecules[molecule]
         self.write_molecule_to_file(filename, molecule, **kwargs)

cctk/orca_file.py

@@ -13,13 +13,13 @@ class OrcaFile(File):
 
     Attributes:
         title (str): the title from the file
-        molecules (ConformationalEnsemble): `ConformationalEnsemble` instance
+        ensemble (ConformationalEnsemble): `ConformationalEnsemble` instance
         header (str): file header
     """
 
-    def __init__(self, molecules, title=None, header=None):
-        if molecules and isinstance(molecules, ConformationalEnsemble):
-            self.molecules = molecules
+    def __init__(self, ensemble, title=None, header=None):
+        if ensemble and isinstance(ensemble, ConformationalEnsemble):
+            self.ensemble = ensemble
         if title and (isinstance(title, str)):
             self.title = title
         if header and (isinstance(header, str)):
@@ -36,12 +36,14 @@ def write_file(self, filename, molecule=None, header=None):
         Args:
             filename (str): path to the new file
             molecule (int): which molecule to use -- passed to ``self.get_molecule()``.
-                Default is -1 (e.g. the last molecule), but positive integers will select from self.molecules (1-indexed).
+                Default is -1 (e.g. the last molecule), but positive integers will select from self.ensemble.molecules (0-indexed).
                 A ``Molecule`` object can also be passed, in which case that molecule will be written to the file.
             header (str): header for new file
         """
+        if molecule is None:
+            molecule = -1
         if not isinstance(molecule, Molecule):
-            molecule = self.get_molecule(molecule)
+            molecule = self.ensemble.molecules[molecule]
 
         if header is None:
             header = self.header
@@ -79,6 +81,8 @@ def write_molecule_to_file(cls, filename, molecule, header):
 
     def get_molecule(self, num=None):
         """
+        TODO: check indexing here
+
         Returns the last molecule (from an optimization job or other multi-molecule jobs) or the only molecule (from other jobs).
 
         If ``num`` is specified, returns that job (1-indexed for positive numbers). So ``job.get_molecule(3)`` will return the 3rd element of ``job.molecules``, not the 4th.
@@ -94,4 +98,4 @@ def get_molecule(self, num=None):
         if num > 0:
             num += -1
 
-        return self.molecules[num]
+        return self.ensemble.molecules[num]

cctk/pdb_file.py

@@ -54,7 +54,7 @@ def write_ensemble_to_trajectory(cls, filename, ensemble):
             filename (str): where to write the file
             ensemble (Ensemble): ``Ensemble`` object to write
         """
-        for idx, molecule in enumerate(ensemble.molecules()):
+        for idx, molecule in enumerate(ensemble.molecules):
             if idx == 0:
                 cls.write_molecule_to_file(filename, molecule, num=idx+1, append=False)
             else:

test/test_align.py

@@ -8,9 +8,9 @@ class TestAlign(unittest.TestCase):
     def test_RMSD(self):
         path = "test/static/gaussian_file.out"
         gaussian_file = cctk.GaussianFile.read_file(path)
-        conformational_ensemble = gaussian_file.molecules
-        m1 = conformational_ensemble[0]
-        m2 = conformational_ensemble[-1]
+        ensemble = gaussian_file.ensemble
+        m1 = ensemble.molecules[0]
+        m2 = ensemble.molecules[-1]
         RMSD = cctk.helper_functions.compute_RMSD(m1,m2)
         delta = abs(0.0006419131435567976 - RMSD)
         self.assertLess(delta, 0.0001)
@@ -20,10 +20,10 @@ def test_align(self):
         conformational_ensemble = cctk.ConformationalEnsemble()
         for filename in sorted(glob.glob(path)):
             gaussian_file = cctk.GaussianFile.read_file(filename)
-            e = gaussian_file.molecules
-            m = e[-1]
-            p = e[m]
-            conformational_ensemble.add_molecule(m,p)
+            ensemble = gaussian_file.ensemble
+            molecule = ensemble.molecules[-1]
+            property_dict = ensemble.get_property_dict(molecule)
+            conformational_ensemble.add_molecule(molecule,property_dict)
 
         comparison_atoms = [1,2,3,4,5,6]
         # added np.int64 here to check that the to_geometry parameter will take any int
@@ -42,12 +42,5 @@ def test_align(self):
         self.assertEqual(len(ensemble3), 1)
         cctk.GaussianFile.write_ensemble_to_file("test/static/phenylpropane_aligned3.gjf", ensemble3, "#p")
 
-        #for i,molecule in enumerate(aligned_ensemble):
-        #    filename = f"test/static/phenylpropane_{i+1}.gjf"
-        #    route_card = "#"
-        #    cctk.GaussianFile.write_molecule_to_file(filename, molecule, route_card)
-        #    rmsd = cctk.helper_functions.compute_RMSD(m0,molecule,comparison_atoms)
-        #    #print("after:",rmsd)
-
 if __name__ == '__main__':
     unittest.main()

test/test_core.py

@@ -35,11 +35,11 @@ def test_read(self):
         file = cctk.MOL2File.read_file(path)
         self.assertTrue(isinstance(file, cctk.MOL2File))
 
-        ensemble = file.molecules
+        ensemble = file.ensemble
         self.assertTrue(isinstance(ensemble, cctk.ConformationalEnsemble))
         self.assertEqual(len(ensemble), 1)
 
-        mol = ensemble[0]
+        mol = ensemble.molecules[0]
         self.assertTrue(isinstance(mol, cctk.Molecule))
         self.assertEqual(len(mol.atomic_numbers), 38)
         self.assertEqual(len(mol.geometry), 38)
@@ -50,10 +50,10 @@ def test_bulk_read(self):
         file = cctk.MOL2File.read_file(path)
         self.assertTrue(isinstance(file, cctk.MOL2File))
 
-        ensemble = file.molecules
+        ensemble = file.ensemble
         self.assertTrue(isinstance(ensemble, cctk.ConformationalEnsemble))
         self.assertEqual(len(ensemble), 597)
-        for mol in ensemble.molecules():
+        for mol in ensemble.molecules:
             self.assertEqual(len(mol.atomic_numbers), 38)
             self.assertEqual(len(mol.geometry), 38)
             self.assertEqual(mol.get_bond_order(1,2), 1)
@@ -82,10 +82,10 @@ def test_read(self):
         self.assertEqual(len(pnames), 597)
         self.assertEqual(len(pvals), 597)
 
-        ensemble = file.molecules
+        ensemble = file.ensemble
         self.assertTrue(isinstance(ensemble, cctk.ConformationalEnsemble))
         self.assertEqual(len(ensemble), 597)
-        for mol in ensemble.molecules():
+        for mol in ensemble.molecules:
             self.assertEqual(len(mol.atomic_numbers), 38)
             self.assertEqual(len(mol.geometry), 38)
             self.assertEqual(mol.get_bond_order(1,2), 1)
@@ -112,7 +112,7 @@ def test_writefile(self):
         ensemble = cctk.ConformationalEnsemble()
         ensemble.add_molecule(file.molecule)
 
-        orca_file = cctk.OrcaFile(molecules=ensemble, header=header)
+        orca_file = cctk.OrcaFile(ensemble=ensemble, header=header)
         orca_file.write_file(new_path)
 
         with open(path) as old:
@@ -128,7 +128,7 @@ class TestPDB(unittest.TestCase):
     def test_write_traj(self):
         path = "test/static/gaussian_file.out"
         file = cctk.GaussianFile.read_file(path)
-        mols = file.molecules
+        mols = file.ensemble
         self.assertTrue(isinstance(mols, cctk.ConformationalEnsemble))
 
         old_path = "test/static/traj.pdb"