def _build_symbols(self, results: _RESULT) -> Sequence[str]:
if 'symbols' in results:
# Safeguard, in case a different parser already
# did this. Not currently available in a default parser
return results.pop('symbols')
# Build the symbols of the atoms
for required_key in ('ion_types', 'species'):
if required_key not in results:
raise ParseError(
'Did not find required key "{}" in parsed header results.'.
format(required_key))
ion_types = results.pop('ion_types')
species = results.pop('species')
if len(ion_types) != len(species):
raise ParseError(
('Expected length of ion_types to be same as species, '
'but got ion_types={} and species={}').format(
len(ion_types), len(species)))
# Expand the symbols list
symbols = []
for n, sym in zip(ion_types, species):
symbols.extend(n * [sym])
return symbols
def build(self, lines: _CHUNK) -> Atoms:
"""Apply outcar chunk parsers, and build an atoms object"""
self.update_parser_headers() # Ensure header is in sync
results = self.parse(lines)
symbols = self.header['symbols']
constraint = self.header.get('constraint', None)
atoms_kwargs = dict(symbols=symbols, constraint=constraint)
# Find some required properties in the parsed results.
# Raise ParseError if they are not present
for prop in ('positions', 'cell'):
try:
atoms_kwargs[prop] = results.pop(prop)
except KeyError:
raise ParseError(
'Did not find required property {} during parse.'.format(
prop))
atoms = Atoms(**atoms_kwargs)
kpts = results.pop('kpts', None)
calc = SinglePointDFTCalculator(atoms, **results)
if kpts is not None:
calc.kpts = kpts
calc.name = 'vasp'
atoms.calc = calc
return atoms
def parse(self, cursor: _CURSOR, lines: _CHUNK) -> _RESULT:
line = lines[cursor].strip()
parts = line.split()
nkpts = int(parts[3])
nbands = int(parts[-1])
results = {'nkpts': nkpts, 'nbands': nbands}
# We also now get the k-point weights etc.,
# because we need to know how many k-points we have
# for parsing that
# Move cursor down to next delimiter
delim2 = 'k-points in reciprocal lattice and weights'
for offset, line in enumerate(lines[cursor:], start=1):
line = line.strip()
if delim2 in line:
# build k-points
ibzkpts = np.zeros((nkpts, 3))
kpt_weights = np.zeros(nkpts)
for nk in range(nkpts):
line = lines[cursor + offset + nk + 3].strip()
parts = line.split()
ibzkpts[nk] = list(map(float, parts[:3]))
kpt_weights[nk] = float(parts[-1])
results['ibzkpts'] = ibzkpts
results['kpt_weights'] = kpt_weights
break
else:
raise ParseError('Did not find the K-points in the OUTCAR')
return results
def parse(self, cursor: _CURSOR, lines: _CHUNK) -> _RESULT:
line = lines[cursor].strip()
parts = line.split()
# Determine in what position we'd expect to find the symbol
if '1/r potential' in line:
# This denotes an AE potential
# Currently only H_AE
# " H 1/r potential "
idx = 1
else:
# Regular PAW potential, e.g.
# "PAW_PBE H1.25 07Sep2000" or
# "PAW_PBE Fe_pv 02Aug2007"
idx = 2
sym = parts[idx]
# remove "_h", "_GW", "_3" tags etc.
sym = sym.split('_')[0]
# in the case of the "H1.25" potentials etc.,
# remove any non-alphabetic characters
sym = ''.join([s for s in sym if s.isalpha()])
if sym not in atomic_numbers:
# Check that we have properly parsed the symbol, and we found
# an element
raise ParseError(
f'Found an unexpected symbol {sym} in line {line}')
self.species.append(sym)
return self._make_returnval()
def get_from_header(self, key: str) -> Any:
"""Get a key from the header, and raise a ParseError
if that key doesn't exist"""
try:
return self.header[key]
except KeyError:
raise ParseError(
'Parser requested unavailable key "{}" from header'.format(
key))
def _read_zmatrix(zmatrix_contents, zmatrix_vars=None):
''' Reads a z-matrix (zmatrix_contents) using its list of variables
(zmatrix_vars), and returns atom positions and symbols '''
try:
atoms = parse_zmatrix(zmatrix_contents, defs=zmatrix_vars)
except (ValueError, AssertionError) as e:
raise ParseError(
"Failed to read Z-matrix from "
"Gaussian input file: ", e)
except KeyError as e:
raise ParseError("Failed to read Z-matrix from "
"Gaussian input file, as symbol: {}"
"could not be recognised. Please make "
"sure you use element symbols, not "
"atomic numbers in the element labels.".format(e))
positions = atoms.positions
symbols = atoms.get_chemical_symbols()
return positions, symbols
def _get_zmatrix_line(line):
''' Converts line into the format needed for it to
be added to the z-matrix contents '''
line_list = line.split()
if len(line_list) == 8 and line_list[7] == '1':
raise ParseError("ERROR: Could not read the Gaussian input file"
", as the alternative Z-matrix format using "
"two bond angles instead of a bond angle and "
"a dihedral angle is not supported.")
return (line.strip() + '\n')
def build_header(fd: TextIO) -> _CHUNK:
"""Build a chunk containing the header data"""
lines = []
for line in fd:
lines.append(line)
if 'Iteration' in line:
# Start of SCF cycle
return lines
# We never found the SCF delimiter, so the OUTCAR must be incomplete
raise ParseError('Incomplete OUTCAR')
def _get_cartesian_atom_coords(symbol, pos):
'''Returns the coordinates: pos as a list of floats if they
are cartesian, and not in z-matrix format'''
if len(pos) < 3 or (pos[0] == '0' and symbol != 'TV'):
# In this case, we have a z-matrix definition, so
# no cartesian coords.
return
elif len(pos) > 3:
raise ParseError("ERROR: Gaussian input file could "
"not be read as freeze codes are not"
" supported. If using cartesian "
"coordinates, these must be "
"given as 3 numbers separated "
"by whitespace.")
else:
try:
return list(map(float, pos))
except ValueError:
raise (ParseError("ERROR: Molecule specification in"
"Gaussian input file could not be read"))
def _get_charge_mult(chgmult_section):
'''return a dict with the charge and multiplicity from
a list chgmult_section that contains the charge and multiplicity
line, read from a gaussian input file'''
chgmult_match = _re_chgmult.match(str(chgmult_section))
try:
chgmult = chgmult_match.group(0).split()
return {'charge': int(chgmult[0]), 'mult': int(chgmult[1])}
except (IndexError, AttributeError):
raise ParseError("ERROR: Could not read the charge and multiplicity "
"from the Gaussian input file. These must be 2 "
"integers separated with whitespace or a comma.")
def atomtypes_outpot(posfname, numsyms):
"""Try to retrieve chemical symbols from OUTCAR or POTCAR
If getting atomtypes from the first line in POSCAR/CONTCAR fails, it might
be possible to find the data in OUTCAR or POTCAR, if these files exist.
posfname -- The filename of the POSCAR/CONTCAR file we're trying to read
numsyms -- The number of symbols we must find
"""
import os.path as op
import glob
# First check files with exactly same name except POTCAR/OUTCAR instead
# of POSCAR/CONTCAR.
fnames = [
posfname.replace('POSCAR', 'POTCAR').replace('CONTCAR', 'POTCAR')
]
fnames.append(
posfname.replace('POSCAR', 'OUTCAR').replace('CONTCAR', 'OUTCAR'))
# Try the same but with compressed files
fsc = []
for fn in fnames:
fsc.append(fn + '.gz')
fsc.append(fn + '.bz2')
for f in fsc:
fnames.append(f)
# Finally try anything with POTCAR or OUTCAR in the name
vaspdir = op.dirname(posfname)
fs = glob.glob(vaspdir + '*POTCAR*')
for f in fs:
fnames.append(f)
fs = glob.glob(vaspdir + '*OUTCAR*')
for f in fs:
fnames.append(f)
tried = []
files_in_dir = os.listdir('.')
for fn in fnames:
if fn in files_in_dir:
tried.append(fn)
at = get_atomtypes(fn)
if len(at) == numsyms:
return at
raise ParseError('Could not determine chemical symbols. Tried files ' +
str(tried))
def get_atomtypes(fname):
"""Given a file name, get the atomic symbols.
The function can get this information from OUTCAR and POTCAR
format files. The files can also be compressed with gzip or
bzip2.
"""
fpath = Path(fname)
atomtypes = []
atomtypes_alt = []
if fpath.suffix == '.gz':
import gzip
opener = gzip.open
elif fpath.suffix == '.bz2':
import bz2
opener = bz2.BZ2File
else:
opener = open
with opener(fpath) as f:
for line in f:
if 'TITEL' in line:
atomtypes.append(line.split()[3].split('_')[0].split('.')[0])
elif 'POTCAR:' in line:
atomtypes_alt.append(
line.split()[2].split('_')[0].split('.')[0])
if len(atomtypes) == 0 and len(atomtypes_alt) > 0:
# old VASP doesn't echo TITEL, but all versions print out species lines
# preceded by "POTCAR:", twice
if len(atomtypes_alt) % 2 != 0:
raise ParseError(
f'Tried to get atom types from {len(atomtypes_alt)} "POTCAR": '
'lines in OUTCAR, but expected an even number')
atomtypes = atomtypes_alt[0:len(atomtypes_alt) // 2]
return atomtypes
def atomtypes_outpot(posfname, numsyms):
"""Try to retrieve chemical symbols from OUTCAR or POTCAR
If getting atomtypes from the first line in POSCAR/CONTCAR fails, it might
be possible to find the data in OUTCAR or POTCAR, if these files exist.
posfname -- The filename of the POSCAR/CONTCAR file we're trying to read
numsyms -- The number of symbols we must find
"""
posfpath = Path(posfname)
# Check files with exactly same path except POTCAR/OUTCAR instead
# of POSCAR/CONTCAR.
fnames = [posfpath.with_name('POTCAR'),
posfpath.with_name('OUTCAR')]
# Try the same but with compressed files
fsc = []
for fnpath in fnames:
fsc.append(fnpath.parent / (fnpath.name + '.gz'))
fsc.append(fnpath.parent / (fnpath.name + '.bz2'))
for f in fsc:
fnames.append(f)
# Code used to try anything with POTCAR or OUTCAR in the name
# but this is no longer supported
tried = []
for fn in fnames:
if fn in posfpath.parent.iterdir():
tried.append(fn)
at = get_atomtypes(fn)
if len(at) == numsyms:
return at
raise ParseError('Could not determine chemical symbols. Tried files ' +
str(tried))
def _validate_params(parameters):
'''Checks whether all of the required parameters exist in the
parameters dict and whether it contains any unsupported settings
'''
# Check for unsupported settings
unsupported_settings = {
"z-matrix", "modredun", "modredundant", "addredundant", "addredun",
"readopt", "rdopt"
}
for s in unsupported_settings:
for v in parameters.values():
if v is not None and s in str(v):
raise ParseError(
"ERROR: Could not read the Gaussian input file"
", as the option: {} is currently unsupported.".format(s))
for k in list(parameters.keys()):
if "popt" in k:
parameters["opt"] = parameters.pop(k)
warnings.warn("The option {} is currently unsupported. "
"This has been replaced with {}.".format(
"POpt", "opt"))
return
def read_eigenvalues(line, f, debug=False):
"""
Read the Eigenvalues in the `.out` file and returns the eigenvalue
First, it assumes system have two spins and start reading until it reaches
the end('*****...').
eigenvalues[spin][kpoint][nbands]
For symmetry reason, `.out` file prints the eigenvalues at the half of the
K points. Thus, we have to fill up the rest of the half.
However, if the calculation was conducted only on the gamma point, it will
raise the 'gamma_flag' as true and it will returns the original samples.
"""
def prind(*line, end='\n'):
if debug:
print(*line, end=end)
prind("Read eigenvalues output")
current_line = f.tell()
f.seek(0) # Seek for the kgrid information
while line != '':
line = f.readline().lower()
if 'scf.kgrid' in line:
break
f.seek(current_line) # Retrun to the original position
kgrid = read_tuple_integer(line)
if kgrid != ():
prind('Non-Gamma point calculation')
prind('scf.Kgrid is %d, %d, %d' % kgrid)
gamma_flag = False
# f.seek(f.tell()+57)
else:
prind('Gamma point calculation')
gamma_flag = True
line = f.readline()
line = f.readline()
eigenvalues = []
eigenvalues.append([])
eigenvalues.append([]) # Assume two spins
i = 0
while True:
# Go to eigenvalues line
while line != '':
line = f.readline()
prind(line)
ll = line.split()
if line.isspace():
continue
elif len(ll) > 1:
if ll[0] == '1':
break
elif "*****" in line:
break
# Break if it reaches the end or next parameters
if "*****" in line or line == '':
break
# Check Number and format is valid
try:
# Suppose to be looks like
# 1 -2.33424746491277 -2.33424746917880
ll = line.split()
# Check if it reaches the end of the file
assert line != ''
assert len(ll) == 3
float(ll[1]); float(ll[2])
except (AssertionError, ValueError):
raise ParseError("Cannot read eigenvalues")
# Read eigenvalues
eigenvalues[0].append([])
eigenvalues[1].append([])
while not (line == '' or line.isspace()):
eigenvalues[0][i].append(float(rn(line, 2)))
eigenvalues[1][i].append(float(rn(line, 1)))
line = f.readline()
prind(line, end='')
i += 1
prind(line)
if gamma_flag:
return np.asarray(eigenvalues)
eigen_half = np.asarray(eigenvalues)
prind(eigen_half)
# Fill up the half
spin, half_kpts, bands = eigen_half.shape
even_odd = np.array(kgrid).prod() % 2
eigen_values = np.zeros((spin, half_kpts*2-even_odd, bands))
for i in range(half_kpts):
eigen_values[0, i] = eigen_half[0, i, :]
eigen_values[1, i] = eigen_half[1, i, :]
eigen_values[0, 2*half_kpts-1-i-even_odd] = eigen_half[0, i, :]
eigen_values[1, 2*half_kpts-1-i-even_odd] = eigen_half[1, i, :]
return eigen_values
def _get_atoms_from_molspec(molspec_section):
''' Takes a string: molspec_section which contains the molecule
specification section of a gaussian input file, and returns an atoms
object that represents this.'''
# These will contain info that will be attached to the Atoms object:
symbols = []
positions = []
pbc = np.zeros(3, dtype=bool)
cell = np.zeros((3, 3))
npbc = 0
# Will contain a dictionary of nuclear properties for each atom,
# that will later be saved to the parameters dict:
nuclear_props = []
# Info relating to the z-matrix definition (if set)
zmatrix_type = False
zmatrix_contents = ""
zmatrix_var_section = False
zmatrix_vars = ""
for line in molspec_section:
# Remove any comments and replace '/' and ',' with whitespace,
# as these are equivalent:
line = line.split('!')[0].replace('/', ' ').replace(',', ' ')
if (line.split()):
if zmatrix_type:
# Save any variables set when defining the z-matrix:
if zmatrix_var_section:
zmatrix_vars += line.strip() + '\n'
continue
elif 'variables' in line.lower():
zmatrix_var_section = True
continue
elif 'constants' in line.lower():
zmatrix_var_section = True
warnings.warn("Constants in the optimisation are "
"not currently supported. Instead "
"setting constants as variables.")
continue
symbol, pos = _get_atoms_info(line)
current_nuclear_props = _get_nuclear_props(line)
if not zmatrix_type:
pos = _get_cartesian_atom_coords(symbol, pos)
if pos is None:
zmatrix_type = True
if symbol.upper() == 'TV' and pos is not None:
pbc[npbc] = True
cell[npbc] = pos
npbc += 1
else:
nuclear_props.append(current_nuclear_props)
if not zmatrix_type:
symbols.append(symbol)
positions.append(pos)
if zmatrix_type:
zmatrix_contents += _get_zmatrix_line(line)
# Now that we are past the molecule spec. section, we can read
# the entire z-matrix (if set):
if len(positions) == 0:
if zmatrix_type:
if zmatrix_vars == '':
zmatrix_vars = None
positions, symbols = _read_zmatrix(zmatrix_contents, zmatrix_vars)
try:
atoms = Atoms(symbols, positions, pbc=pbc, cell=cell)
except (IndexError, ValueError, KeyError) as e:
raise ParseError("ERROR: Could not read the Gaussian input file, "
"due to a problem with the molecule "
"specification: {}".format(e))
nuclear_props = _get_nuclear_props_for_all_atoms(nuclear_props)
return atoms, nuclear_props
def _validate_symbol_string(string):
if "-" in string:
raise ParseError("ERROR: Could not read the Gaussian input file, as"
" molecule specifications for molecular mechanics "
"calculations are not supported.")
return string