def parse_cif_ase(fileobj) -> Iterator[CIFBlock]:
"""Parse a CIF file using ase CIF parser."""
if isinstance(fileobj, str):
with open(fileobj, 'rb') as fileobj:
data = fileobj.read()
else:
data = fileobj.read()
if isinstance(data, bytes):
data = data.decode('latin1')
data = format_unicode(data)
lines = [e for e in data.split('\n') if len(e) > 0]
if len(lines) > 0 and lines[0].rstrip() == '#\\#CIF_2.0':
warnings.warn('CIF v2.0 file format detected; `ase` CIF reader might '
'incorrectly interpret some syntax constructions, use '
'`pycodcif` reader instead')
lines = [''] + lines[::-1] # all lines (reversed)
while lines:
line = lines.pop().strip()
if not line or line.startswith('#'):
continue
yield parse_block(lines, line)
def parse_cif_ase(fileobj):
"""Parse a CIF file using ase CIF parser"""
blocks = []
if isinstance(fileobj, basestring):
fileobj = open(fileobj, 'rb')
data = fileobj.read()
if isinstance(data, bytes):
data = data.decode('latin1')
data = format_unicode(data)
data = [e for e in data.split('\n') if len(e) > 0]
if len(data) > 0 and data[0].rstrip() == '#\\#CIF_2.0':
warnings.warn('CIF v2.0 file format detected; `ase` CIF reader might '
'incorrectly interpret some syntax constructions, use '
'`pycodcif` reader instead')
lines = [''] + data[::-1] # all lines (reversed)
while True:
if not lines:
break
line = lines.pop()
line = line.strip()
if not line or line.startswith('#'):
continue
blocks.append(parse_block(lines, line))
return blocks
def convert_value(value):
"""Convert CIF value string to corresponding python type."""
value = value.strip()
if re.match('(".*")|(\'.*\')$', value):
return format_unicode(value[1:-1])
elif re.match(r'[+-]?\d+$', value):
return int(value)
elif re.match(r'[+-]?(?:\d+(?:\.\d*)?|\.\d+)(?:[eE][+-]?\d+)?$', value):
return float(value)
elif re.match(r'[+-]?(?:\d+(?:\.\d*)?|\.\d+)(?:[eE][+-]?\d+)?\(\d+\)$',
value):
return float(value[:value.index('(')]) # strip off uncertainties
elif re.match(r'[+-]?(?:\d+(?:\.\d*)?|\.\d+)(?:[eE][+-]?\d+)?\(\d+$',
value):
warnings.warn('Badly formed number: "{0}"'.format(value))
return float(value[:value.index('(')]) # strip off uncertainties
else:
return format_unicode(value)