def build(cls, column, value=None, default=None, scheme=None):
# None must be '' in this case
default = default if default is not None else 'None'
if value is None:
return MafColumnRecord.build(column, default, scheme=scheme)
else:
return MafColumnRecord.build(column, value, scheme=scheme)
def __init__(self, chromosome, start, end, ref="A", alts=None):
super(DummyRecordWithAllele, self).__init__()
self.add(MafColumnRecord("Chromosome", chromosome))
self.add(MafColumnRecord("Start_Position", start))
self.add(MafColumnRecord("End_Position", end))
self.add(MafColumnRecord("Reference_Allele", ref))
self._alts = alts if alts else []
def build(cls, column, value=None, default=None, scheme=None):
if value is None:
assert default is not None, "BooleanColumn {0} can't be None".format(
column)
return MafColumnRecord.build(column, default, scheme=scheme)
else:
return MafColumnRecord.build(column, value, scheme=scheme)
def build(cls, key, value, scheme, default=None):
if value is None and default is not None:
return MafColumnRecord.build(key, default, scheme=scheme)
elif value is None and scheme.column_class(key).is_nullable():
return scheme.column_class(key).build_nullable(key, scheme=scheme)
elif isinstance(value, list):
return MafColumnRecord.build(key,
';'.join(sorted(value)),
scheme=scheme)
else:
return MafColumnRecord.build(key, value, scheme=scheme)
def build(cls, key, value, scheme, default=None, fn=None, **kwargs):
if fn is not None:
return MafColumnRecord.build(key,
fn(value, **kwargs),
scheme=scheme)
elif value is None and default is not None:
return MafColumnRecord.build(key, default, scheme=scheme)
elif value is None and scheme.column_class(key).is_nullable():
return scheme.column_class(key).build_nullable(key, scheme=scheme)
else:
return MafColumnRecord.build(key, value, scheme=scheme)
def __init__(self, chromosome, start, end,
tumor_barcode = None,
normal_barcode = None):
super(DummyRecord, self).__init__()
self.add(MafColumnRecord("Chromosome", chromosome))
self.add(MafColumnRecord("Start_Position", start))
self.add(MafColumnRecord("End_Position", end))
if tumor_barcode:
self.add(MafColumnRecord("Tumor_Sample_Barcode",
tumor_barcode))
if normal_barcode:
self.add(MafColumnRecord("Matched_Norm_Sample_Barcode",
normal_barcode))
def test_basic_column(self):
column = MafColumnRecord.build(name="key",
value="value",
column_index=0,
description="Foo Bar")
self.assertFalse(column.is_nullable())
self.assertEqual([], column.__nullable_values__())
self.assertIsNone(column.__nullable_dict__())
self.assertEqual([], column.__nullable_keys__())
self.assertFalse(column.is_null())
self.assertEqual(str(column), "value")
def test_build_with_scheme(self):
scheme = _TestScheme()
# column found in the scheme, column index inferred
column = MafColumnRecord.build(name="float",
value=2.1,
description="Foo Bar",
scheme=scheme)
self.assertEqual(str(column), "2.1")
self.assertEqual(column.column_index, 1)
# column found in the scheme, column index validated
column = MafColumnRecord.build(name="float",
value=2.1,
column_index=1,
description="Foo Bar",
scheme=scheme)
self.assertEqual(str(column), "2.1")
self.assertEqual(column.column_index, 1)
# error: name not found in scheme
with self.assertRaises(KeyError):
MafColumnRecord.build(name="not-a-key",
value=2.1,
column_index=0,
description="Foo Bar",
scheme=scheme)
# error: mismatching column index
with self.assertRaises(ValueError):
MafColumnRecord.build(name="float",
value=2.1,
column_index=0,
description="Foo Bar",
scheme=scheme)
def __get_key_from_int(key: Union[str, int],
column: MafColumnRecord) -> str:
"""Returns the name that should be used for the column and sets the
column index if not already set"""
column_index = int(key)
if column_index < 0:
raise KeyError
# set the column_index on column if not already set, otherwise
# ensure they are the same
if column.column_index is None:
column.column_index = column_index
elif column_index != column.column_index:
raise ValueError(
f"Column index mismatch: {column.column_index} is not {column_index}"
)
return column.key
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 __setitem__(self, key: Union[int, str],
column: MafColumnRecord) -> None:
"""If a record already exists with the column name `key`, we either
check that the `column_index`es are the same, or set the `column_index`
if it is not set on the record. Otherwise, if the `column_index` is not
set and no existing record is found with the column name `key`, then
append it, and set `column_index` in `column`.
If the `column_index` is greater than the length, then extend the number
of columns, inserting `None` for values for any columns between the old
and new length; these columns will have no name.
:param key: The key may be one of four things: 1. If an `int`,
it is assumed to be the `column_index` of the column. The
`column_index` of`column` will be set to `int(key)` if not already set.
If already set, then `column.column_index` should equal `int(key)`. 2.
If an `MafColumnRecord`, then the `MafColumnRecord` should have the same
`key` as the provided column. 3. If `None`, then `None` will always be
returned. 4. Otherwise, it should be the column name, and be the same as
the `key` in the provided column.
:param column: an instance of `MafColumnRecord`."""
if not isinstance(column, MafColumnRecord):
raise TypeError(f"{type(column)} is not 'MafColumnRecord'")
if isinstance(key, int):
key = self.__get_key_from_int(key, column)
elif isinstance(key, MafColumnRecord):
# make sure the keys are the same
key = self.__get_key_from_column(key, column)
elif not isinstance(key, str):
raise TypeError("Column name must be a string")
elif column.key != key:
raise ValueError(
f"Adding a column with name '{column.key}' but key was '{key}'"
)
assert key == column.key
# if there already is a record with the same key, make sure that it has
# the same column index, otherwise set the column index if the current
# record doesn't have one. If there is no record with the same and the
# column index is not set, set it as the next column in the list.
if key in self.__columns_dict:
if column.column_index is None:
column.column_index = self.__columns_dict[key].column_index
else:
if self.__columns_dict[key].column_index != column.column_index:
raise ValueError(
f"Existing column's index '{self.__columns_dict[key].column_index}' does not match replacement column's index '{column.column_index}'"
)
elif column.column_index is None:
# set the column index to the next column
column.column_index = len(self.__columns_list)
self.__columns_dict[key] = column
assert column.column_index is not None
# extend the list if the index is out of range
if len(self) <= column.column_index:
num_more = column.column_index - len(self) + 1
self.__columns_list.extend([None] * num_more)
# Developer Note: due to padding, the number of items in the dictionary
# may be less than the number of items in the list. Use validate to
# catch this later.
self.__columns_list[column.column_index] = column
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
