def from_file(filename: str) -> 'Result':
"""Create a Result from a transaction file."""
#Why is this here??? This is really confusing in practice
#if filename is None or not Path(filename).exists(): return Result()
json_encode = Cartesian(JsonEncode())
json_decode = Cartesian(JsonDecode())
Pipe.join(DiskSource(filename), [json_decode, ResultPromote(), json_encode], DiskSink(filename, 'w')).run()
return Result.from_transactions(Pipe.join(DiskSource(filename), [json_decode]).read())
class ReaderSimulation(Simulation):
def __init__(self,
reader: Filter[Iterable[str], Any],
source: Union[str, Source[Iterable[str]]],
label_column: Union[str, int],
with_header: bool = True) -> None:
self._reader = reader
if isinstance(source, str) and source.startswith('http'):
self._source = Pipe.join(HttpSource(source), [ResponseToLines()])
elif isinstance(source, str):
self._source = DiskSource(source)
else:
self._source = source
self._label_column = label_column
self._with_header = with_header
self._interactions = cast(Optional[Sequence[Interaction]], None)
def read(self) -> Iterable[Interaction]:
"""Read the interactions in this simulation."""
return self._load_interactions()
def _load_interactions(self) -> Sequence[Interaction]:
parsed_rows_iter = iter(self._reader.filter(self._source.read()))
if self._with_header:
header = next(parsed_rows_iter)
else:
header = []
if isinstance(self._label_column, str):
label_col_index = header.index(self._label_column)
else:
label_col_index = self._label_column
parsed_cols = list(Transpose().filter(parsed_rows_iter))
label_col = parsed_cols.pop(label_col_index)
feature_rows = list(Transpose().filter(Flatten().filter(parsed_cols)))
is_sparse_labels = len(label_col) == 2 and isinstance(
label_col[0], tuple) and isinstance(label_col[1], tuple)
if is_sparse_labels:
dense_labels: List[Any] = ['0'] * len(feature_rows)
for label_row, label_val in zip(*label_col): #type:ignore
dense_labels[label_row] = label_val
else:
dense_labels = list(label_col)
return ClassificationSimulation(feature_rows, dense_labels).read()
def __repr__(self) -> str:
return str(self._source)
def __init__(self,
reader: Filter[Iterable[str], Any],
source: Union[str, Source[Iterable[str]]],
label_column: Union[str, int],
with_header: bool = True) -> None:
self._reader = reader
if isinstance(source, str) and source.startswith('http'):
self._source = Pipe.join(HttpSource(source), [ResponseToLines()])
elif isinstance(source, str):
self._source = DiskSource(source)
else:
self._source = source
self._label_column = label_column
self._with_header = with_header
self._interactions = cast(Optional[Sequence[Interaction]], None)
def from_file(arg) -> 'Benchmark': #type: ignore
"""Instantiate a Benchmark from a config file."""
if isinstance(arg, str) and arg.startswith('http'):
content = '\n'.join(ResponseToLines().filter(
HttpSource(arg).read()))
elif isinstance(arg, str) and not arg.startswith('http'):
content = '\n'.join(DiskSource(arg).read())
else:
content = arg.read() #type: ignore
return CobaRegistry.construct(CobaConfig.Benchmark['file_fmt']).filter(
JsonDecode().filter(content))
def test_from_definition_source(self):
if Path("coba/tests/.temp/from_file.env").exists():
Path("coba/tests/.temp/from_file.env").unlink()
try:
Path("coba/tests/.temp/from_file.env").write_text(
'{ "environments" : { "OpenmlSimulation": 150 } }')
env = Environments.from_file(
DiskSource("coba/tests/.temp/from_file.env"))
self.assertEqual(1, len(env))
self.assertEqual(150, env[0].params['openml'])
self.assertEqual(False, env[0].params['cat_as_str'])
finally:
if Path("coba/tests/.temp/from_file.env").exists():
Path("coba/tests/.temp/from_file.env").unlink()
def __init__(self, transaction_log: Optional[str] = None) -> None:
if not transaction_log or not Path(transaction_log).exists():
version = None
else:
version = JsonDecode().filter(next(DiskSource(transaction_log).read()))[1]
if version == 3:
self._transactionIO = TransactionIO_V3(transaction_log)
elif version == 4:
self._transactionIO = TransactionIO_V4(transaction_log)
elif version is None:
self._transactionIO = TransactionIO_V4(transaction_log)
else:
raise CobaException("We were unable to determine the appropriate Transaction reader for the file.")
def test_is_picklable(self):
pickle.dumps(DiskSource("coba/tests/.temp/test.gz"))
def test_simple_with_gz(self):
Path("coba/tests/.temp/test.gz").write_bytes(gzip.compress(b'a\nb\nc'))
self.assertEqual(["a", "b", "c"],
list(DiskSource("coba/tests/.temp/test.gz").read()))
def test_simple_sans_gz(self):
Path("coba/tests/.temp/test.log").write_text("a\nb\nc")
self.assertEqual(["a", "b", "c"],
list(DiskSource("coba/tests/.temp/test.log").read()))
def __init__(self, log_file: Optional[str] = None, minify:bool=True) -> None:
self._log_file = log_file
self._minify = minify
self._source = DiskSource(log_file) if log_file else ListSource()
self._sink = DiskSink(log_file) if log_file else ListSink(self._source.items)