def validate(self, line: str) -> None:
try:
record_id = record_id_from_line(line)
except ValueError:
raise InvalidS1Format("Does not contain S1 separator")
try:
spec = spec_from_record_id(record_id)
except KeyError:
raise UnknownRecordType(
f"Encountered unknown record id of '{record_id}'")
try:
# Tossing away this data, but catching field length errors
parse_line_with_spec(line, spec)
except InvalidFieldLength:
raise InvalidS1Line()
def handle_new_line(self, line: str) -> Tuple[RecordSpec, RecordType]:
record_id = record_id_from_line(line)
try:
spec = spec_from_record_id(record_id)
except KeyError:
raise UnknownRecordType()
record = parse_line_with_spec(line, spec)
self.total_lines += 1
return spec, record
def test_parse_repeat(correct_repeat_line, repeat_spec):
parsed = parse_line_with_spec(correct_repeat_line, repeat_spec)
assert parsed == {
"record_id": "2",
"a": "1",
"b": "2",
"c": "3",
"repeats": [{"d": "4", "e": "5"}, {"d": "6", "e": "7"}],
}
def test_parsing_invalid_length_no_repeat(bad_line, basic_spec):
with pytest.raises(InvalidFieldLength):
parse_line_with_spec(bad_line, basic_spec)
def test_incorrect_record_ids(basic_spec):
""" Should raise when a record_id doesn't match the spec.record_id """
bad_line = "abcd|a|b|c"
assert basic_spec.record_id == "1"
with pytest.raises(ValueError):
parse_line_with_spec(bad_line, basic_spec)
def test_basic_parse(correct_basic_line, basic_spec):
parsed = parse_line_with_spec(correct_basic_line, basic_spec)
assert parsed == {"record_id": "1", "a": "foo", "b": "bar", "c": "bat"}