def parse_file(lines, require_order=None):
"""
>>> lines = ['# q:str ans:str s:int c:int rel:float\\n', 'url\\tph\\t1000\\t15\\t1.2\\n']
>>> list(parse_file(lines))
[Rec(q='url', ans='ph', s=1000, c=15, rel=1.2)]
"""
lines_iter = iter(lines)
header = list(islice(lines_iter, 1))
if len(header) == 1:
data_desc = parse_header(header[0])
if require_order:
if not data_desc.order.is_ordered_by(parse_header_order(require_order)):
raise Exception("require_order check of {0} failure on {1}.".format(require_order, data_desc.order))
field_spec = []
for field in data_desc.fields:
field_spec.append((field.name, TYPE_MAP[field.type]))
parser = FileParser(field_separator="\t", rec_parser=RecParser(field_spec))
return ParsedFile(data_desc, parser(lines_iter))
else:
raise Exception("No header")
def desc(self):
return parse_header(self.header)
