def inspect_file(path):
"""Inspect SDFile structure
Returns:
tuple: (data label list, number of records)
"""
with open(path, 'rb') as f:
labels, count = inspect(tx.decode(line) for line in f)
return labels, count
def mols_from_text(text, no_halt=True, assign_descriptors=True):
"""Returns molecules generated from sdfile text
Throws:
StopIteration: if the text does not have molecule
ValueError: if Unsupported symbol is found
"""
if isinstance(text, bytes):
t = tx.decode(text)
else:
t = text
# Lazy line splitter. More efficient memory usage than str.split.
exp = re.compile(r"[^\n]*\n|.")
sp = (x.group(0) for x in re.finditer(exp, t))
for c in mol_supplier(sp, no_halt, assign_descriptors):
yield c
def mols_from_file(path, no_halt=True, assign_descriptors=True):
"""Compound supplier from CTAB text file (.mol, .sdf)"""
with open(path, 'rb') as f:
fd = (tx.decode(line) for line in f)
for c in mol_supplier(fd, no_halt, assign_descriptors):
yield c
