def __init__(self, ff_file_name, debug=False):
from foyer.forcefield import preprocess_forcefield_files
try:
preprocessed_ff_file_name = preprocess_forcefield_files(
[ff_file_name])
ff_tree = etree.parse(preprocessed_ff_file_name[0])
self.validate_xsd(ff_tree)
self.atom_type_names = ff_tree.xpath(
"/ForceField/AtomTypes/Type/@name")
self.atom_types = ff_tree.xpath("/ForceField/AtomTypes/Type")
self.validate_class_type_exclusivity(ff_tree)
# Loading forcefield should succeed, because XML can be parsed and
# basics have been validated.
from foyer.forcefield import Forcefield
self.smarts_parser = Forcefield(preprocessed_ff_file_name,
validation=False).parser
finally:
for ff_file_name in preprocessed_ff_file_name:
os.remove(ff_file_name)
self.validate_smarts(debug=debug)
self.validate_overrides()
def test_structure_to_hoomdsimulation(self, ethane):
import hoomd
from foyer.forcefield import Forcefield
from mbuild.formats.hoomd_simulation import create_hoomd_simulation
ff = Forcefield(name="oplsaa")
structure = ff.apply(ethane)
sim = hoomd.context.SimulationContext()
with sim:
create_hoomd_simulation(structure, 2.5)
sim_forces = hoomd.context.current.forces
pair_force = import_("hoomd.md.pair")
charge_force = import_("hoomd.md.charge")
special_pair_force = import_("hoomd.md.special_pair")
bond_force = import_("hoomd.md.bond")
angle_force = import_("hoomd.md.angle")
dihedral_force = import_("hoomd.md.dihedral")
assert isinstance(sim_forces[0], pair_force.lj)
assert isinstance(sim_forces[1], charge_force.pppm)
assert isinstance(sim_forces[2], pair_force.ewald)
assert isinstance(sim_forces[3], special_pair_force.lj)
assert isinstance(sim_forces[4], special_pair_force.coulomb)
assert isinstance(sim_forces[5], bond_force.harmonic)
assert isinstance(sim_forces[6], angle_force.harmonic)
assert isinstance(sim_forces[7], dihedral_force.opls)
def test_hoomdsimulation_restart(self):
import gsd.hoomd
import hoomd
from foyer.forcefield import Forcefield
from mbuild.formats.hoomd_simulation import create_hoomd_simulation
box = mb.Compound()
box.add(mb.Compound(name="Ar", pos=[1, 1, 1]))
box.add(mb.Compound(name="Ar", pos=[1, 1, 1]))
ff = Forcefield(forcefield_files=get_fn("lj.xml"))
structure = ff.apply(box)
structure.box = [10, 10, 10, 90, 90, 90]
sim = hoomd.context.SimulationContext()
with sim:
hoomd_obj, ref_vals = create_hoomd_simulation(
structure, 2.5, restart=get_fn("restart.gsd"))
sim_forces = hoomd.context.current.forces
pair_force = import_("hoomd.md.pair")
assert isinstance(sim_forces[0], pair_force.lj)
snap = hoomd_obj[0]
with gsd.hoomd.open(get_fn("restart.gsd")) as f:
rsnap = f[0]
assert np.array_equal(snap.particles.position,
rsnap.particles.position)
def test_param_structure_to_snapshot(self, ethane):
from foyer.forcefield import Forcefield
from mbuild.formats.hoomd_snapshot import to_hoomdsnapshot
ff = Forcefield(name="oplsaa")
structure = ff.apply(ethane)
snap, _ = to_hoomdsnapshot(structure)
assert snap.particles.N == 8
assert snap.bonds.N == 7
assert snap.angles.N == 12
assert snap.dihedrals.N == 9
assert snap.pairs.N == 9
def __init__(self, ff_file_name):
from foyer.forcefield import preprocess_forcefield_files
preprocessed_ff_file_name = preprocess_forcefield_files([ff_file_name])
ff_tree = etree.parse(preprocessed_ff_file_name[0])
self.validate_xsd(ff_tree)
# Loading forcefield should succeed, because XML can be parsed and
# basics have been validated.
from foyer.forcefield import Forcefield
self.smarts_parser = Forcefield(preprocessed_ff_file_name,
validation=False).parser
self.validate_smarts(ff_tree)
def test_lj_to_hoomd_forcefield(self):
import hoomd
from foyer.forcefield import Forcefield
from mbuild.formats.hoomd_forcefield import create_hoomd_forcefield
box = mb.Compound()
box.add(mb.Compound(name="Ar", pos=[1, 1, 1]))
box.add(mb.Compound(name="Ar", pos=[1, 1, 1]))
ff = Forcefield(forcefield_files=get_fn("lj.xml"))
structure = ff.apply(box)
structure.box = [10, 10, 10, 90, 90, 90]
snapshot, forces, ref_values = create_hoomd_forcefield(structure, 2.5)
assert isinstance(forces[0], hoomd.md.pair.LJ)
def test_hexa_coordinated():
ff = Forcefield(forcefield_files=get_fn('pf6.xml'))
mol2 = pmd.load_file(get_fn('pf6.mol2'), structure=True)
pf6 = ff.apply(mol2)
types = [a.type for a in pf6.atoms]
assert types.count('P') == 1
assert types.count('F1') == 2
assert types.count('F2') == 2
assert types.count('F3') == 2
assert len(pf6.bonds) == 6
assert all(bond.type for bond in pf6.bonds)
assert len(pf6.angles) == 15
assert all(angle.type for angle in pf6.angles)
def test_hexa_coordinated(self):
ff = Forcefield(forcefield_files=get_fn("pf6.xml"))
mol2 = pmd.load_file(get_fn("pf6.mol2"), structure=True)
pf6 = ff.apply(mol2)
types = [a.type for a in pf6.atoms]
assert types.count("P") == 1
assert types.count("F1") == 2
assert types.count("F2") == 2
assert types.count("F3") == 2
assert len(pf6.bonds) == 6
assert all(bond.type for bond in pf6.bonds)
assert len(pf6.angles) == 15
assert all(angle.type for angle in pf6.angles)
def test_hexa_coordinated():
ff = Forcefield(forcefield_files=get_fn('pf6.xml'))
mol2 = pmd.load_file(get_fn('pf6.mol2'), structure=True)
pf6 = ff.apply(mol2)
types = [a.type for a in pf6.atoms]
assert types.count('P') == 1
assert types.count('F1') == 2
assert types.count('F2') == 2
assert types.count('F3') == 2
assert len(pf6.bonds) == 6
assert all(bond.type for bond in pf6.bonds)
assert len(pf6.angles) == 15
assert all(angle.type for angle in pf6.angles)
def test_lj_to_hoomdsimulation(self):
import hoomd
from foyer.forcefield import Forcefield
from mbuild.formats.hoomd_simulation import create_hoomd_simulation
box = mb.Compound()
box.add(mb.Compound(name="Ar", pos=[1, 1, 1]))
box.add(mb.Compound(name="Ar", pos=[1, 1, 1]))
ff = Forcefield(forcefield_files=get_fn("lj.xml"))
structure = ff.apply(box)
structure.box = [10, 10, 10, 90, 90, 90]
sim = hoomd.context.SimulationContext()
with sim:
create_hoomd_simulation(structure, 2.5)
sim_forces = hoomd.context.current.forces
pair_force = import_("hoomd.md.pair")
assert isinstance(sim_forces[0], pair_force.lj)
def test_structure_to_hoomd_forcefield(self, ethane):
import hoomd
from foyer.forcefield import Forcefield
from mbuild.formats.hoomd_forcefield import create_hoomd_forcefield
ff = Forcefield(name="oplsaa")
structure = ff.apply(ethane)
snapshot, forces, ref_values = create_hoomd_forcefield(structure, 2.5)
assert isinstance(forces[0], hoomd.md.pair.LJ)
assert isinstance(forces[1], hoomd.md.pair.Ewald)
assert isinstance(forces[2], hoomd.md.long_range.pppm.Coulomb)
assert isinstance(forces[3], hoomd.md.special_pair.LJ)
assert isinstance(forces[4], hoomd.md.special_pair.Coulomb)
assert isinstance(forces[5], hoomd.md.bond.Harmonic)
assert isinstance(forces[6], hoomd.md.angle.Harmonic)
assert isinstance(forces[7], hoomd.md.dihedral.OPLS)
def test_atom_typing(
self, openff_topology_graph, gmso_topology_graph, parmed_topology_graph
):
# ToDo: More robust testing for atomtyping
opls = Forcefield(name="oplsaa")
openff_typemap = find_atomtypes(openff_topology_graph, forcefield=opls)
gmso_typemap = find_atomtypes(gmso_topology_graph, forcefield=opls)
parmed_typemap = find_atomtypes(parmed_topology_graph, forcefield=opls)
assert openff_typemap
assert gmso_typemap
assert parmed_typemap
def __init__(self, ff_file_name):
from foyer.forcefield import preprocess_forcefield_files
preprocessed_ff_file_name = preprocess_forcefield_files([ff_file_name])
ff_tree = etree.parse(preprocessed_ff_file_name[0])
self.validate_xsd(ff_tree)
# Loading forcefield should succeed, because XML can be parsed and
# basics have been validated.
from foyer.forcefield import Forcefield
self.smarts_parser = Forcefield(preprocessed_ff_file_name, validation=False).parser
self.validate_smarts(ff_tree)
def __init__(self, ff_file_name, debug=False):
from foyer.forcefield import preprocess_forcefield_files
preprocessed_ff_file_name = preprocess_forcefield_files([ff_file_name])
ff_tree = etree.parse(preprocessed_ff_file_name[0])
self.validate_xsd(ff_tree)
self.atom_type_names = ff_tree.xpath('/ForceField/AtomTypes/Type/@name')
self.atom_types = ff_tree.xpath('/ForceField/AtomTypes/Type')
self.validate_class_type_exclusivity(ff_tree)
# Loading forcefield should succeed, because XML can be parsed and
# basics have been validated.
from foyer.forcefield import Forcefield
self.smarts_parser = Forcefield(preprocessed_ff_file_name, validation=False).parser
self.validate_smarts(debug=debug)
self.validate_overrides()
class Validator(object):
def __init__(self, ff_file_name):
from foyer.forcefield import preprocess_forcefield_files
preprocessed_ff_file_name = preprocess_forcefield_files([ff_file_name])
ff_tree = etree.parse(preprocessed_ff_file_name[0])
self.validate_xsd(ff_tree)
# Loading forcefield should succeed, because XML can be parsed and
# basics have been validated.
from foyer.forcefield import Forcefield
self.smarts_parser = Forcefield(preprocessed_ff_file_name, validation=False).parser
self.validate_smarts(ff_tree)
# TODO: figure out what exceptions are raised, and provide good error messages
def validate_xsd(self, ff_tree, xsd_file=None):
if xsd_file is None:
xsd_file = join(split(abspath(__file__))[0], 'forcefields', 'ff.xsd')
xmlschema_doc = etree.parse(xsd_file)
xmlschema = etree.XMLSchema(xmlschema_doc)
try:
xmlschema.assertValid(ff_tree)
except DocumentInvalid as ex:
message = ex.error_log.last_error.message
line = ex.error_log.last_error.line
# rewrite error message for constraint violation
if ex.error_log.last_error.type_name == "SCHEMAV_CVC_IDC":
if "missing_atom_type_in_nonbonded" in message:
atomtype = message[message.find("[") + 1:message.find("]")]
raise ValidationError(
"Atom type {} is found in NonbondedForce at line {} but"
" undefined in AtomTypes".format(atomtype, line), ex, line)
elif "nonunique_atomtype_name" in message:
atomtype = message[message.find("[") + 1:message.find("]")]
raise ValidationError(
"Atom type {} is defined a second time at line {}".format(
atomtype, line), ex, line)
raise
def validate_smarts(self, ff_tree):
results = ff_tree.xpath('/ForceField/AtomTypes/Type')
atom_types = ff_tree.xpath('/ForceField/AtomTypes/Type/@name')
missing_smarts = []
for r in results:
smarts_string = r.attrib.get('def')
name = r.attrib['name']
if smarts_string is None:
missing_smarts.append(name)
continue
# make sure smarts string can be parsed
try:
self.smarts_parser.parse(smarts_string)
except ParseError as ex:
if " col " in ex.args[0]:
column = ex.args[0][ex.args[0].find(" col ") + 5:].strip()
column = " at character {} of {}".format(column, smarts_string)
else:
column = ""
raise ValidationError("Malformed SMARTS string{} on line {}".format(
column, r.sourceline), ex, r.sourceline)
# make sure referenced labels exist
smarts_graph = SMARTSGraph(smarts_string, parser=self.smarts_parser,
name=name, overrides=r.attrib.get('overrides'))
for atom_expr in nx.get_node_attributes(smarts_graph, 'atom').values():
labels = atom_expr.select('has_label')
for label in labels:
atom_type = label.tail[0][1:]
if atom_type not in atom_types:
raise ValidationError(
"Reference to undefined atomtype {} in SMARTS string" " '{}' at line {}".format(
atom_type, r.attrib['def'], r.sourceline), None, r.sourceline)
warn("The following atom types do not have smarts definitions: {}".format(', '.join(missing_smarts)),
ValidationWarning)
def validate_overrides(self, ff_tree):
results = ff_tree.xpath('/ForceField/AtomTypes/Type[@name]')
for r in results:
smarts_string = r.attrib['def']
print(smarts_string)
print(r.sourceline)
# make sure smarts string can be parsed
self.smarts.parse(smarts_string)
class Validator(object):
def __init__(self, ff_file_name, debug=False):
from foyer.forcefield import preprocess_forcefield_files
preprocessed_ff_file_name = preprocess_forcefield_files([ff_file_name])
ff_tree = etree.parse(preprocessed_ff_file_name[0])
self.validate_xsd(ff_tree)
self.atom_type_names = ff_tree.xpath('/ForceField/AtomTypes/Type/@name')
self.atom_types = ff_tree.xpath('/ForceField/AtomTypes/Type')
self.validate_class_type_exclusivity(ff_tree)
# Loading forcefield should succeed, because XML can be parsed and
# basics have been validated.
from foyer.forcefield import Forcefield
self.smarts_parser = Forcefield(preprocessed_ff_file_name, validation=False).parser
self.validate_smarts(debug=debug)
self.validate_overrides()
@staticmethod
def validate_xsd(ff_tree, xsd_file=None):
if xsd_file is None:
xsd_file = join(split(abspath(__file__))[0], 'forcefields', 'ff.xsd')
xmlschema_doc = etree.parse(xsd_file)
xmlschema = etree.XMLSchema(xmlschema_doc)
error_texts = {'missing_atom_type_in_nonbonded':
("Atom type {} is found in NonbondedForce at line {}"
" but undefined in AtomTypes"),
'nonunique_atomtype_name':
"Atom type {} is defined a second time at line {}",
'atomtype_name_key':
"Atom type {} is defined a second time at line {}"
}
def create_error(keyword, message, line):
atomtype = message[message.find("[") + 1:message.find("]")]
error_text = error_texts[keyword].format(atomtype, line)
return ValidationError(error_text, ex, line)
try:
xmlschema.assertValid(ff_tree)
except DocumentInvalid as ex:
message = ex.error_log.last_error.message
line = ex.error_log.last_error.line
# rewrite error message for constraint violation
if ex.error_log.last_error.type_name == "SCHEMAV_CVC_IDC":
for keyword in error_texts:
if keyword in message:
raise create_error(keyword, message, line)
else:
raise ValidationError('Unhandled XML validation error. '
'Please consider submitting a bug report.', ex, line)
raise
def validate_class_type_exclusivity(self, ff_tree):
sections = {'HarmonicBondForce/Bond': 2,
'HarmonicAngleForce/Angle': 3,
'RBTorsionForce/Proper': 4}
errors = []
for element, num_atoms in sections.items():
valid_attribs = set()
for n in range(1, num_atoms + 1):
valid_attribs.add('class{}'.format(n))
valid_attribs.add('type{}'.format(n))
for entry in ff_tree.xpath('/ForceField/{}'.format(element)):
attribs = [valid for valid in valid_attribs
if entry.attrib.get(valid) is not None]
if num_atoms != len(attribs):
error = ValidationError(
'Invalid number of "class" and/or "type" attributes for'
' {} at line {}'.format(element, entry.sourceline),
None, entry.sourceline)
errors.append(error)
number_endings = Counter([a[-1] for a in attribs])
if not all(1 == x for x in number_endings.values()):
error = ValidationError(
'Only one "class" or "type" attribute may be defined'
' for each atom in a bonded force. See line {}'.format(entry.sourceline),
None, entry.sourceline)
errors.append(error)
referenced_types = []
for valid in valid_attribs:
if valid.startswith('type'):
atomtype = entry.attrib.get(valid)
if atomtype:
referenced_types.append(atomtype)
for atomtype in referenced_types:
if atomtype not in self.atom_type_names:
error = ValidationError(
'Atom type {} is found in {} at line {} but'
' undefined in AtomTypes'.format(atomtype, element.split('/')[0], entry.sourceline),
None, entry.sourceline)
errors.append(error)
raise_collected(errors)
def validate_smarts(self, debug):
missing_smarts = []
errors = []
for entry in self.atom_types:
smarts_string = entry.attrib.get('def')
name = entry.attrib['name']
if smarts_string is None:
missing_smarts.append(name)
continue
# make sure smarts string can be parsed
try:
self.smarts_parser.parse(smarts_string)
except lark.ParseError as ex:
if " col " in ex.args[0]:
column = ex.args[0][ex.args[0].find(" col ") + 5:].strip()
column = " at character {} of {}".format(column, smarts_string)
else:
column = ""
malformed = ValidationError(
"Malformed SMARTS string{} on line {}".format(column, entry.sourceline),
ex, entry.sourceline)
errors.append(malformed)
continue
# make sure referenced labels exist
smarts_graph = SMARTSGraph(smarts_string, parser=self.smarts_parser,
name=name, overrides=entry.attrib.get('overrides'))
for atom_expr in nx.get_node_attributes(smarts_graph, name='atom').values():
labels = atom_expr.find_data('has_label')
for label in labels:
atom_type = label.children[0][1:]
if atom_type not in self.atom_type_names:
undefined = ValidationError(
"Reference to undefined atomtype '{}' in SMARTS "
"string '{}' at line {}".format(atom_type, entry.attrib['def'], entry.sourceline),
None, entry.sourceline)
errors.append(undefined)
raise_collected(errors)
if missing_smarts and debug:
warn("The following atom types do not have smarts definitions: {}".format(
', '.join(missing_smarts)), ValidationWarning)
if missing_smarts and not debug:
warn("There are {} atom types that are missing a smarts definition. "
"To view the missing atom types, re-run with debug=True when "
"applying the forcefield.".format(len(missing_smarts)), ValidationWarning)
def validate_overrides(self):
errors = []
for entry in self.atom_types:
overrides = entry.attrib.get('overrides')
if not overrides:
continue
overridden_types = [at.strip() for at in overrides.split(',') if at]
for atom_type in overridden_types:
if atom_type not in self.atom_type_names:
undefined = ValidationError(
"Reference to undefined atomtype '{}' in 'overrides' "
"'{}' at line {}".format(atom_type, entry.attrib['overrides'], entry.sourceline),
None, entry.sourceline)
errors.append(undefined)
raise_collected(errors)
class Validator(object):
"""Verifies formatting of force field XML."""
def __init__(self, ff_file_name, debug=False):
from foyer.forcefield import preprocess_forcefield_files
try:
preprocessed_ff_file_name = preprocess_forcefield_files(
[ff_file_name])
ff_tree = etree.parse(preprocessed_ff_file_name[0])
self.validate_xsd(ff_tree)
self.atom_type_names = ff_tree.xpath(
"/ForceField/AtomTypes/Type/@name")
self.atom_types = ff_tree.xpath("/ForceField/AtomTypes/Type")
self.validate_class_type_exclusivity(ff_tree)
# Loading forcefield should succeed, because XML can be parsed and
# basics have been validated.
from foyer.forcefield import Forcefield
self.smarts_parser = Forcefield(preprocessed_ff_file_name,
validation=False).parser
finally:
for ff_file_name in preprocessed_ff_file_name:
os.remove(ff_file_name)
self.validate_smarts(debug=debug)
self.validate_overrides()
@staticmethod
def validate_xsd(ff_tree, xsd_file=None):
"""Check consistency with forcefields/ff.xsd."""
if xsd_file is None:
xsd_file = join(
split(abspath(__file__))[0], "forcefields", "ff.xsd")
xmlschema_doc = etree.parse(xsd_file)
xmlschema = etree.XMLSchema(xmlschema_doc)
error_texts = {
"missing_atom_type_in_nonbonded":
("Atom type {} is found in NonbondedForce at line {}"
" but undefined in AtomTypes"),
"nonunique_atomtype_name":
"Atom type {} is defined a second time at line {}",
"atomtype_name_key":
"Atom type {} is defined a second time at line {}",
}
def create_error(keyword, message, line):
atomtype = message[message.find("[") + 1:message.find("]")]
error_text = error_texts[keyword].format(atomtype, line)
return ValidationError(error_text, ex, line)
try:
xmlschema.assertValid(ff_tree)
except DocumentInvalid as ex:
message = ex.error_log.last_error.message
line = ex.error_log.last_error.line
# rewrite error message for constraint violation
if ex.error_log.last_error.type_name == "SCHEMAV_CVC_IDC":
for keyword in error_texts:
if keyword in message:
raise create_error(keyword, message, line)
else:
raise ValidationError(
"Unhandled XML validation error. "
"Please consider submitting a bug report.",
ex,
line,
)
raise
def validate_class_type_exclusivity(self, ff_tree):
"""Assert unique bond/angle/dihedral definitions and class lengths."""
sections = {
"HarmonicBondForce/Bond": 2,
"HarmonicAngleForce/Angle": 3,
"RBTorsionForce/Proper": 4,
}
errors = []
for element, num_atoms in sections.items():
valid_attribs = set()
for n in range(1, num_atoms + 1):
valid_attribs.add("class{}".format(n))
valid_attribs.add("type{}".format(n))
for entry in ff_tree.xpath("/ForceField/{}".format(element)):
attribs = [
valid for valid in valid_attribs
if entry.attrib.get(valid) is not None
]
if num_atoms != len(attribs):
error = ValidationError(
'Invalid number of "class" and/or "type" attributes for'
" {} at line {}".format(element, entry.sourceline),
None,
entry.sourceline,
)
errors.append(error)
number_endings = Counter([a[-1] for a in attribs])
if not all(1 == x for x in number_endings.values()):
error = ValidationError(
'Only one "class" or "type" attribute may be defined'
" for each atom in a bonded force. See line {}".format(
entry.sourceline),
None,
entry.sourceline,
)
errors.append(error)
referenced_types = []
for valid in valid_attribs:
if valid.startswith("type"):
atomtype = entry.attrib.get(valid)
if atomtype:
referenced_types.append(atomtype)
for atomtype in referenced_types:
if atomtype not in self.atom_type_names:
error = ValidationError(
"Atom type {} is found in {} at line {} but"
" undefined in AtomTypes".format(
atomtype,
element.split("/")[0],
entry.sourceline,
),
None,
entry.sourceline,
)
errors.append(error)
raise_collected(errors)
def validate_smarts(self, debug):
"""Check SMARTS definitions for missing or non-parseable."""
missing_smarts = []
errors = []
for entry in self.atom_types:
smarts_string = entry.attrib.get("def")
if not smarts_string:
warn("You have empty smart definition(s)", ValidationWarning)
continue
name = entry.attrib["name"]
if smarts_string is None:
missing_smarts.append(name)
continue
# make sure smarts string can be parsed
try:
self.smarts_parser.parse(smarts_string)
except lark.ParseError as ex:
if " col " in ex.args[0]:
column = ex.args[0][ex.args[0].find(" col ") + 5:].strip()
column = " at character {} of {}".format(
column, smarts_string)
else:
column = ""
malformed = ValidationError(
"Malformed SMARTS string{} on line {}".format(
column, entry.sourceline),
ex,
entry.sourceline,
)
errors.append(malformed)
continue
# make sure referenced labels exist
smarts_graph = SMARTSGraph(
smarts_string,
parser=self.smarts_parser,
name=name,
overrides=entry.attrib.get("overrides"),
)
for atom_expr in nx.get_node_attributes(smarts_graph,
name="atom").values():
labels = atom_expr.find_data("has_label")
for label in labels:
atom_type = label.children[0][1:]
if atom_type not in self.atom_type_names:
undefined = ValidationError(
"Reference to undefined atomtype '{}' in SMARTS "
"string '{}' at line {}".format(
atom_type, entry.attrib["def"],
entry.sourceline),
None,
entry.sourceline,
)
errors.append(undefined)
raise_collected(errors)
if missing_smarts and debug:
warn(
"The following atom types do not have smarts definitions: {}".
format(", ".join(missing_smarts)),
ValidationWarning,
)
if missing_smarts and not debug:
warn(
"There are {} atom types that are missing a smarts definition. "
"To view the missing atom types, re-run with debug=True when "
"applying the forcefield.".format(len(missing_smarts)),
ValidationWarning,
)
def validate_overrides(self):
"""Assert all overrides are defined elsewhere in force field."""
errors = []
for entry in self.atom_types:
overrides = entry.attrib.get("overrides")
if not overrides:
continue
overridden_types = [
at.strip() for at in overrides.split(",") if at
]
for atom_type in overridden_types:
if atom_type not in self.atom_type_names:
undefined = ValidationError(
"Reference to undefined atomtype '{}' in 'overrides' "
"'{}' at line {}".format(
atom_type,
entry.attrib["overrides"],
entry.sourceline,
),
None,
entry.sourceline,
)
errors.append(undefined)
raise_collected(errors)
class Validator(object):
def __init__(self, ff_file_name):
from foyer.forcefield import preprocess_forcefield_files
preprocessed_ff_file_name = preprocess_forcefield_files([ff_file_name])
ff_tree = etree.parse(preprocessed_ff_file_name[0])
self.validate_xsd(ff_tree)
self.atom_type_names = ff_tree.xpath('/ForceField/AtomTypes/Type/@name')
self.atom_types = ff_tree.xpath('/ForceField/AtomTypes/Type')
self.validate_class_type_exclusivity(ff_tree)
# Loading forcefield should succeed, because XML can be parsed and
# basics have been validated.
from foyer.forcefield import Forcefield
self.smarts_parser = Forcefield(preprocessed_ff_file_name, validation=False).parser
self.validate_smarts()
self.validate_overrides()
@staticmethod
def validate_xsd(ff_tree, xsd_file=None):
if xsd_file is None:
xsd_file = join(split(abspath(__file__))[0], 'forcefields', 'ff.xsd')
xmlschema_doc = etree.parse(xsd_file)
xmlschema = etree.XMLSchema(xmlschema_doc)
error_texts = {'missing_atom_type_in_nonbonded':
("Atom type {} is found in NonbondedForce at line {}"
" but undefined in AtomTypes"),
'nonunique_atomtype_name':
"Atom type {} is defined a second time at line {}",
'atomtype_name_key':
"Atom type {} is defined a second time at line {}"
}
def create_error(keyword, message, line):
atomtype = message[message.find("[") + 1:message.find("]")]
error_text = error_texts[keyword].format(atomtype, line)
return ValidationError(error_text, ex, line)
try:
xmlschema.assertValid(ff_tree)
except DocumentInvalid as ex:
message = ex.error_log.last_error.message
line = ex.error_log.last_error.line
# rewrite error message for constraint violation
if ex.error_log.last_error.type_name == "SCHEMAV_CVC_IDC":
for keyword in error_texts:
if keyword in message:
raise create_error(keyword, message, line)
else:
raise ValidationError('Unhandled XML validation error. '
'Please consider submitting a bug report.', ex, line)
raise
def validate_class_type_exclusivity(self, ff_tree):
sections = {'HarmonicBondForce/Bond': 2,
'HarmonicAngleForce/Angle': 3,
'RBTorsionForce/Proper': 4}
errors = []
for element, num_atoms in sections.items():
valid_attribs = set()
for n in range(1, num_atoms + 1):
valid_attribs.add('class{}'.format(n))
valid_attribs.add('type{}'.format(n))
for entry in ff_tree.xpath('/ForceField/{}'.format(element)):
attribs = [valid for valid in valid_attribs
if entry.attrib.get(valid) is not None]
if num_atoms != len(attribs):
error = ValidationError(
'Invalid number of "class" and/or "type" attributes for'
' {} at line {}'.format(element, entry.sourceline),
None, entry.sourceline)
errors.append(error)
number_endings = Counter([a[-1] for a in attribs])
if not all(1 == x for x in number_endings.values()):
error = ValidationError(
'Only one "class" or "type" attribute may be defined'
' for each atom in a bonded force. See line {}'.format(entry.sourceline),
None, entry.sourceline)
errors.append(error)
referenced_types = []
for valid in valid_attribs:
if valid.startswith('type'):
atomtype = entry.attrib.get(valid)
if atomtype:
referenced_types.append(atomtype)
for atomtype in referenced_types:
if atomtype not in self.atom_type_names:
error = ValidationError(
'Atom type {} is found in {} at line {} but'
' undefined in AtomTypes'.format(atomtype, element.split('/')[0], entry.sourceline),
None, entry.sourceline)
errors.append(error)
raise_collected(errors)
def validate_smarts(self):
missing_smarts = []
errors = []
for entry in self.atom_types:
smarts_string = entry.attrib.get('def')
name = entry.attrib['name']
if smarts_string is None:
missing_smarts.append(name)
continue
# make sure smarts string can be parsed
try:
self.smarts_parser.parse(smarts_string)
except ParseError as ex:
if " col " in ex.args[0]:
column = ex.args[0][ex.args[0].find(" col ") + 5:].strip()
column = " at character {} of {}".format(column, smarts_string)
else:
column = ""
malformed = ValidationError(
"Malformed SMARTS string{} on line {}".format(column, entry.sourceline),
ex, entry.sourceline)
errors.append(malformed)
continue
# make sure referenced labels exist
smarts_graph = SMARTSGraph(smarts_string, parser=self.smarts_parser,
name=name, overrides=entry.attrib.get('overrides'))
for atom_expr in nx.get_node_attributes(smarts_graph, 'atom').values():
labels = atom_expr.select('has_label')
for label in labels:
atom_type = label.tail[0][1:]
if atom_type not in self.atom_type_names:
undefined = ValidationError(
"Reference to undefined atomtype '{}' in SMARTS "
"string '{}' at line {}".format(atom_type, entry.attrib['def'], entry.sourceline),
None, entry.sourceline)
errors.append(undefined)
raise_collected(errors)
if missing_smarts:
warn("The following atom types do not have smarts definitions: {}".format(
', '.join(missing_smarts)), ValidationWarning)
def validate_overrides(self):
errors = []
for entry in self.atom_types:
overrides = entry.attrib.get('overrides')
if not overrides:
continue
overridden_types = [at.strip() for at in overrides.split(',') if at]
for atom_type in overridden_types:
if atom_type not in self.atom_type_names:
undefined = ValidationError(
"Reference to undefined atomtype '{}' in 'overrides' "
"'{}' at line {}".format(atom_type, entry.attrib['overrides'], entry.sourceline),
None, entry.sourceline)
errors.append(undefined)
raise_collected(errors)
class Validator(object):
def __init__(self, ff_file_name):
from foyer.forcefield import preprocess_forcefield_files
preprocessed_ff_file_name = preprocess_forcefield_files([ff_file_name])
ff_tree = etree.parse(preprocessed_ff_file_name[0])
self.validate_xsd(ff_tree)
# Loading forcefield should succeed, because XML can be parsed and
# basics have been validated.
from foyer.forcefield import Forcefield
self.smarts_parser = Forcefield(preprocessed_ff_file_name,
validation=False).parser
self.validate_smarts(ff_tree)
# TODO: figure out what exceptions are raised, and provide good error messages
def validate_xsd(self, ff_tree, xsd_file=None):
if xsd_file is None:
xsd_file = join(
split(abspath(__file__))[0], 'forcefields', 'ff.xsd')
xmlschema_doc = etree.parse(xsd_file)
xmlschema = etree.XMLSchema(xmlschema_doc)
try:
xmlschema.assertValid(ff_tree)
except DocumentInvalid as ex:
message = ex.error_log.last_error.message
line = ex.error_log.last_error.line
# rewrite error message for constraint violation
if ex.error_log.last_error.type_name == "SCHEMAV_CVC_IDC":
if "missing_atom_type_in_nonbonded" in message:
atomtype = message[message.find("[") + 1:message.find("]")]
raise ValidationError(
"Atom type {} is found in NonbondedForce at line {} but"
" undefined in AtomTypes".format(atomtype, line), ex,
line)
elif "nonunique_atomtype_name" in message:
atomtype = message[message.find("[") + 1:message.find("]")]
raise ValidationError(
"Atom type {} is defined a second time at line {}".
format(atomtype, line), ex, line)
raise
def validate_smarts(self, ff_tree):
results = ff_tree.xpath('/ForceField/AtomTypes/Type')
atom_types = ff_tree.xpath('/ForceField/AtomTypes/Type/@name')
missing_smarts = []
for r in results:
smarts_string = r.attrib.get('def')
name = r.attrib['name']
if smarts_string is None:
missing_smarts.append(name)
continue
# make sure smarts string can be parsed
try:
self.smarts_parser.parse(smarts_string)
except ParseError as ex:
if " col " in ex.args[0]:
column = ex.args[0][ex.args[0].find(" col ") + 5:].strip()
column = " at character {} of {}".format(
column, smarts_string)
else:
column = ""
raise ValidationError(
"Malformed SMARTS string{} on line {}".format(
column, r.sourceline), ex, r.sourceline)
# make sure referenced labels exist
smarts_graph = SMARTSGraph(smarts_string,
parser=self.smarts_parser,
name=name,
overrides=r.attrib.get('overrides'))
for atom_expr in nx.get_node_attributes(smarts_graph,
'atom').values():
labels = atom_expr.select('has_label')
for label in labels:
atom_type = label.tail[0][1:]
if atom_type not in atom_types:
raise ValidationError(
"Reference to undefined atomtype {} in SMARTS string"
" '{}' at line {}".format(atom_type,
r.attrib['def'],
r.sourceline), None,
r.sourceline)
warn(
"The following atom types do not have smarts definitions: {}".
format(', '.join(missing_smarts)), ValidationWarning)
def validate_overrides(self, ff_tree):
results = ff_tree.xpath('/ForceField/AtomTypes/Type[@name]')
for r in results:
smarts_string = r.attrib['def']
print(smarts_string)
print(r.sourceline)
# make sure smarts string can be parsed
self.smarts.parse(smarts_string)