def from_line(
cls,
line: str,
column_names: Optional[List[str]] = None,
scheme: Optional['MafScheme'] = None,
line_number: Optional[int] = None,
validation_stringency: ValidationStringency = ValidationStringency.
Strict,
logger: logging.Logger = Logger.RootLogger,
) -> 'MafRecord':
"""
Parses a record from a single tab-delimited line.
:param column_names: the expected names of the columns, in order,
otherwise will use the scheme.
:param line: the line to parse.
:param scheme: an optional MafScheme
:param line_number: the optional line number.
:param validation_stringency: the optional validation stringency for
the record
:param logger the logger to which to write errors
:return:
"""
record = cls(line_number=line_number,
validation_stringency=validation_stringency)
if column_names is None:
if scheme is None:
raise ValueError("Either column_names or scheme must be given")
column_names = scheme.column_names()
def add_errors(error: MafValidationError) -> None:
record.validation_errors.append(error)
column_values = line.rstrip("\r\n").split(cls.ColumnSeparator)
if len(column_names) != len(column_values):
add_errors(
MafValidationError(
MafValidationErrorType.RECORD_MISMATCH_NUMBER_OF_COLUMNS,
f"Found '{len(column_values)}' columns but expected '{len(column_names)}'",
line_number=line_number,
))
record.validate(logger=logger, reset_errors=False)
return record
for column_index, (column_name, column_value) in enumerate(
zip(column_names, column_values)):
column = None
scheme_column_class = (scheme.column_class(
name=column_name) if scheme else None)
# A validation error will be found later if we don't find the
# column name
if scheme_column_class is None:
column = MafColumnRecord(key=column_name,
value=column_value,
column_index=column_index)
else:
try:
scheme_column_class = scheme.column_class(
name=column_name) # type: ignore
column = scheme_column_class.build( # type: ignore
name=column_name,
value=column_value,
column_index=column_index,
)
except Exception as error:
add_errors(
MafValidationError(
MafValidationErrorType.RECORD_INVALID_COLUMN_VALUE,
f"Could not build column '{column_index+1}' with name '{column_name}' scheme '{scheme.version()}': {error}", # type: ignore
line_number=line_number,
))
if column is not None:
column_validation_errors = column.validate(
scheme=scheme, line_number=line_number)
record.validation_errors.extend(
column_validation_errors) # type: ignore
if len(column_validation_errors) == 0:
record[column_name] = column
# process validation errors
record.validate(logger=logger, reset_errors=False)
return record
def from_line(cls,
line,
column_names=None,
scheme=None,
line_number=None,
validation_stringency=None,
logger=Logger.RootLogger):
"""
Parses a record from a single tab-delimited line.
:param column_names: the expected names of the columns, in order,
otherwise will use the scheme.
:param line: the line to parse.
:param scheme: an optional MafScheme
:param line_number: the optional line number.
:param validation_stringency: the optional validation stringency for
the record
:param logger the logger to which to write errors
:return:
"""
record = MafRecord(line_number=line_number,
validation_stringency=validation_stringency)
if column_names is None:
if scheme is None:
raise ValueError("Either column_names or scheme must be given")
column_names = scheme.column_names()
def add_errors(error):
record.validation_errors.append(error)
column_values = line.rstrip("\r\n").split(MafRecord.ColumnSeparator)
if len(column_names) != len(column_values):
add_errors(
MafValidationError(
MafValidationErrorType.RECORD_MISMATCH_NUMBER_OF_COLUMNS,
"Found '%d' columns but expected '%d'" %
(len(column_values), len(column_names)),
line_number=line_number))
else:
for column_index, column_name_and_value in \
enumerate(zip(column_names, column_values)):
column_name = column_name_and_value[0]
column_value = column_name_and_value[1]
column = None
scheme_column_class = \
scheme.column_class(name=column_name) if scheme else None
# A validation error will be found later if we don't find the
# column name
if scheme_column_class is None:
column = MafColumnRecord(key=column_name,
value=column_value,
column_index=column_index)
else:
try:
scheme_column_class = \
scheme.column_class(name=column_name)
column = scheme_column_class.build(
name=column_name,
value=column_value,
column_index=column_index)
except Exception as error:
add_errors(
MafValidationError(
MafValidationErrorType.
RECORD_INVALID_COLUMN_VALUE,
"Could not build column '%d' with name '%s' "
"with the scheme '%s': %s" %
(column_index + 1, column_name,
scheme.version(), str(error)),
line_number=line_number,
))
if column is not None:
column_validation_errors = \
column.validate(scheme=scheme, line_number=line_number)
record.validation_errors.extend(column_validation_errors)
if len(column_validation_errors) == 0:
record[column_name] = column
# process validation errors
record.validate(logger=logger, reset_errors=False)
return record