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_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_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)
#TODO: helper func to raise 1 or more erros
if len(errors) > 1:
raise MultipleValidationError(errors)
elif len(errors) == 1:
raise errors[0]
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)
if len(errors) > 1:
raise MultipleValidationError(errors)
elif len(errors) == 1:
raise errors[0]
if missing_smarts:
warn(
"The following atom types do not have smarts definitions: {}".
format(', '.join(missing_smarts)), ValidationWarning)
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_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):
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)
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)
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)
