def __init__(self, conf: Any = None):
p = ParamDict(FUGUE_DASK_DEFAULT_CONF)
p.update(ParamDict(conf))
super().__init__(p)
self._fs = FileSystem()
self._log = logging.getLogger()
self._native = NativeExecutionEngine(conf=conf)
def test_load_parquet_folder(self):
e = self.engine
native = NativeExecutionEngine()
a = ArrayDataFrame([[6, 1]], "c:int,a:long")
b = ArrayDataFrame([[2, 7], [4, 8]], "c:int,a:long")
path = os.path.join(self.tmpdir, "a", "b")
native.save_df(a, os.path.join(path, "a.parquet"))
native.save_df(b, os.path.join(path, "b.parquet"))
FileSystem().touch(os.path.join(path, "_SUCCESS"))
c = e.load_df(path, format_hint="parquet", columns=["a", "c"])
df_eq(c, [[1, 6], [7, 2], [8, 4]], "a:long,c:int", throw=True)
def test_load_avro_folder(self):
# TODO: switch to c:int,a:long when we can preserve schema to avro
e = self.engine
native = NativeExecutionEngine()
a = ArrayDataFrame([[6, 1]], "c:long,a:long")
b = ArrayDataFrame([[2, 7], [4, 8]], "c:long,a:long")
path = os.path.join(self.tmpdir, "a", "b")
native.save_df(a, os.path.join(path, "a.avro"))
native.save_df(b, os.path.join(path, "b.avro"))
FileSystem().touch(os.path.join(path, "_SUCCESS"))
c = e.load_df(path, format_hint="avro", columns=["a", "c"])
df_eq(c, [[1, 6], [7, 2], [8, 4]], "a:long,c:long", throw=True)
def test_use_df(tmpdir):
# df generated inside dag
with FugueSQLWorkflow() as dag:
a = dag.df([[0], [1]], "a:int")
dag("""
b=CREATE[[0], [1]] SCHEMA a: int
OUTPUT a, b USING assert_eq
""")
dag.sql_vars["b"].assert_eq(a)
# external non-workflowdataframe
arr = ArrayDataFrame([[0], [1]], "a:int")
with FugueSQLWorkflow() as dag:
dag(
"""
b=CREATE[[0], [1]] SCHEMA a: int
OUTPUT a, b USING assert_eq
""",
a=arr,
)
dag.sql_vars["b"].assert_eq(dag.df([[0], [1]], "a:int"))
# from yield file
engine = NativeExecutionEngine(
conf={"fugue.workflow.checkpoint.path": os.path.join(tmpdir, "ck")})
with FugueSQLWorkflow(engine) as dag:
dag("CREATE[[0], [1]] SCHEMA a: int YIELD FILE AS b")
res = dag.yields["b"]
with FugueSQLWorkflow(engine) as dag:
dag(
"""
b=CREATE[[0], [1]] SCHEMA a: int
OUTPUT a, b USING assert_eq
""",
a=res,
)
# from yield dataframe
engine = NativeExecutionEngine()
with FugueSQLWorkflow(engine) as dag:
dag("CREATE[[0], [1]] SCHEMA a: int YIELD DATAFRAME AS b")
res = dag.yields["b"]
with FugueSQLWorkflow(engine) as dag:
dag(
"""
b=CREATE[[0], [1]] SCHEMA a: int
OUTPUT a, b USING assert_eq
""",
a=res,
)
def test_auto_persist():
dag1 = FugueWorkflow(NativeExecutionEngine())
df1 = dag1.df([[0]], "a:int")
df1.show()
df1.show()
id1 = dag1.spec_uuid()
dag2 = FugueWorkflow(NativeExecutionEngine({"fugue.workflow.auto_persist": True}))
df1 = dag2.df([[0]], "a:int")
df1.show()
df1.show()
id2 = dag2.spec_uuid()
dag3 = FugueWorkflow(NativeExecutionEngine())
df1 = dag3.df([[0]], "a:int").weak_checkpoint(level=None)
df1.show()
df1.show()
id3 = dag3.spec_uuid()
assert id1 == id2
assert id2 == id3
dag2 = FugueWorkflow(
NativeExecutionEngine(
{
"fugue.workflow.auto_persist": True,
"fugue.workflow.auto_persist_value": "abc",
}
)
)
df1 = dag2.df([[0]], "a:int")
df1.show()
df1.show()
id2 = dag2.spec_uuid()
dag3 = FugueWorkflow(NativeExecutionEngine())
df1 = dag3.df([[0]], "a:int").weak_checkpoint(level="abc")
df1.show()
df1.show()
id3 = dag3.spec_uuid()
assert id2 == id3
dag1 = FugueWorkflow(NativeExecutionEngine())
df1 = dag1.df([[0]], "a:int")
df1.show()
id1 = dag1.spec_uuid()
dag2 = FugueWorkflow(NativeExecutionEngine({"fugue.workflow.auto_persist": True}))
df1 = dag2.df([[0]], "a:int")
df1.show() # auto persist will not trigger
id2 = dag2.spec_uuid()
dag3 = FugueWorkflow(NativeExecutionEngine())
df1 = dag3.df([[0]], "a:int").weak_checkpoint(level=None)
df1.show()
id3 = dag3.spec_uuid()
assert id1 == id2
assert id2 == id3 # checkpoint, including auto_persist doesn't change determinism
def test_auto_persist():
dag1 = FugueWorkflow(NativeExecutionEngine())
df1 = dag1.df([[0]], "a:int")
df1.show()
df1.show()
id1 = dag1.spec_uuid()
dag2 = FugueWorkflow(
NativeExecutionEngine({"fugue.workflow.auto_persist": True}))
df1 = dag2.df([[0]], "a:int")
df1.show()
df1.show()
id2 = dag2.spec_uuid()
dag3 = FugueWorkflow(NativeExecutionEngine())
df1 = dag3.df([[0]], "a:int").persist()
df1.show()
df1.show()
id3 = dag3.spec_uuid()
assert id1 != id2
assert id2 == id3
dag2 = FugueWorkflow(
NativeExecutionEngine({
"fugue.workflow.auto_persist": True,
"fugue.workflow.auto_persist_value": "abc"
}))
df1 = dag2.df([[0]], "a:int")
df1.show()
df1.show()
id2 = dag2.spec_uuid()
dag3 = FugueWorkflow(NativeExecutionEngine())
df1 = dag3.df([[0]], "a:int").persist("abc")
df1.show()
df1.show()
id3 = dag3.spec_uuid()
assert id2 == id3
dag1 = FugueWorkflow(NativeExecutionEngine())
df1 = dag1.df([[0]], "a:int")
df1.show()
id1 = dag1.spec_uuid()
dag2 = FugueWorkflow(
NativeExecutionEngine({"fugue.workflow.auto_persist": True}))
df1 = dag2.df([[0]], "a:int")
df1.show() # auto persist will not trigger
id2 = dag2.spec_uuid()
dag3 = FugueWorkflow(NativeExecutionEngine())
df1 = dag3.df([[0]], "a:int").persist()
df1.show()
id3 = dag3.spec_uuid()
assert id1 == id2
assert id2 != id3
def test_use_soecial_df(tmpdir):
# external non-workflowdataframe
arr = ArrayDataFrame([[0], [1]], "a:int")
fsql(
"""
b=CREATE[[0], [1]] SCHEMA a: int
a = SELECT * FROM a.x
OUTPUT a, b USING assert_eq
a = SELECT x.* FROM a.x AS x
OUTPUT a, b USING assert_eq
c=CREATE [[0,0],[1,1]] SCHEMA a:int,b:int
d = SELECT x.*,y.a AS b FROM a.x x INNER JOIN a.x y ON x.a=y.a
OUTPUT c, d USING assert_eq
""",
{
"a.x": arr
},
).run()
# from yield file
engine = NativeExecutionEngine(
conf={"fugue.workflow.checkpoint.path": os.path.join(tmpdir, "ck")})
with FugueSQLWorkflow(engine) as dag:
dag("CREATE[[0], [1]] SCHEMA a: int YIELD FILE AS b")
res = dag.yields["b"]
with FugueSQLWorkflow(engine) as dag:
dag(
"""
b=CREATE[[0], [1]] SCHEMA a: int
a = SELECT * FROM a.x
OUTPUT a, b USING assert_eq
""",
{"a.x": res},
)
def __init__(
self,
execution_engine: Any = None,
cache: Any = NoOpCache,
workflow_engine: Any = None,
hooks: Any = WorkflowHooks,
):
if execution_engine is None:
ee: ExecutionEngine = NativeExecutionEngine()
else:
ee = to_instance(execution_engine, ExecutionEngine)
self._fugue_engine = ee
self._lock = RLock()
self._results: Dict[Any, DataFrame] = {}
if workflow_engine is None:
workflow_engine = ParallelExecutionEngine(
self.execution_engine.conf.get("fugue.workflow.concurrency",
1), self)
super().__init__(
cache=cache,
engine=workflow_engine,
hooks=hooks,
logger=ee.log,
config=ee.conf,
)
def __init__(
self,
execution_engine: Any = None,
cache: Any = NoOpCache,
workflow_engine: Any = None,
hooks: Any = WorkflowHooks,
):
if execution_engine is None:
ee: ExecutionEngine = NativeExecutionEngine()
else:
ee = to_instance(execution_engine, ExecutionEngine)
self._fugue_engine = ee
self._lock = RLock()
self._results: Dict[Any, DataFrame] = {}
if workflow_engine is None:
workflow_engine = ParallelExecutionEngine(
self.execution_engine.conf.get(
FUGUE_CONF_WORKFLOW_CONCURRENCY,
FUGUE_DEFAULT_CONF[FUGUE_CONF_WORKFLOW_CONCURRENCY],
),
self,
)
super().__init__(
cache=cache,
engine=workflow_engine,
hooks=hooks,
logger=ee.log,
config=ee.conf,
)
def __init__(self):
self._funcs: Dict[str, Callable] = {}
self._sql_funcs: Dict[str, Callable] = {}
self.register_default(
lambda conf, **kwargs: NativeExecutionEngine(conf=conf))
self.register_default_sql_engine(
lambda engine, **kwargs: engine.sql_engine)
def test_workflow_conf():
dag = FugueSQLWorkflow(NativeExecutionEngine({"x": 10}))
assert 10 == dag.conf.get_or_throw("x", int)
assert not dag.conf.get_or_throw("fugue.sql.compile.ignore_case", bool)
dag = FugueSQLWorkflow(
NativeExecutionEngine({
"x": 10,
"fugue.sql.compile.ignore_case": True
}))
assert 10 == dag.conf.get_or_throw("x", int)
assert dag.conf.get_or_throw("fugue.sql.compile.ignore_case", bool)
dag = FugueSQLWorkflow(NativeExecutionEngine({"x": 10}),
{"fugue.sql.compile.ignore_case": "true"})
assert 10 == dag.conf.get_or_throw("x", int)
assert dag.conf.get_or_throw("fugue.sql.compile.ignore_case", bool)
def test_workflow_conf():
dag = FugueSQLWorkflow(
NativeExecutionEngine({
"x": 10,
"fugue.sql.compile.simple_assign": "false"
}))
assert 10 == dag.conf.get_or_throw("x", int)
assert not dag.conf.get_or_throw("fugue.sql.compile.simple_assign", bool)
assert not dag.conf.get_or_throw("fugue.sql.compile.ignore_case", bool)
def test_conf_override():
with raises(FugueSQLSyntaxError):
FugueSQLWorkflow()("create [[0]] schema a:int")
with FugueSQLWorkflow(
NativeExecutionEngine({"fugue.sql.compile.ignore_case":
"true"})) as dag:
a = dag.df([[0], [1]], "a:int")
dag("""
b = create [[0],[1]] schema a:int
output a,b using assert_eq""")
def test_load_csv_folder(self):
e = self.engine
native = NativeExecutionEngine()
a = ArrayDataFrame([[6.1, 1.1]], "c:double,a:double")
b = ArrayDataFrame([[2.1, 7.1], [4.1, 8.1]], "c:double,a:double")
path = os.path.join(self.tmpdir, "a", "b")
native.save_df(a,
os.path.join(path, "a.csv"),
format_hint="csv",
header=True)
native.save_df(b,
os.path.join(path, "b.csv"),
format_hint="csv",
header=True)
FileSystem().touch(os.path.join(path, "_SUCCESS"))
c = e.load_df(
path,
format_hint="csv",
header=True,
infer_schema=True,
columns=["a", "c"],
)
df_eq(c, [[1.1, 6.1], [7.1, 2.1], [8.1, 4.1]],
"a:double,c:double",
throw=True)
from fugue.execution.execution_engine import ExecutionEngine, SQLEngine
from fugue.execution.factory import (
make_execution_engine,
make_sql_engine,
register_default_execution_engine,
register_default_sql_engine,
register_execution_engine,
register_sql_engine,
)
from fugue.execution.native_execution_engine import (
NativeExecutionEngine,
SqliteEngine,
QPDPandasEngine,
)
register_execution_engine("native",
lambda conf: NativeExecutionEngine(conf),
on_dup="ignore")
register_execution_engine("pandas",
lambda conf: NativeExecutionEngine(conf),
on_dup="ignore")
register_sql_engine("sqlite",
lambda engine: SqliteEngine(engine),
on_dup="ignore")
register_sql_engine("qpdpandas",
lambda engine: QPDPandasEngine(engine),
on_dup="ignore")
register_sql_engine("qpd_pandas",
lambda engine: QPDPandasEngine(engine),
on_dup="ignore")
class DaskExecutionEngine(ExecutionEngine):
"""The execution engine based on `Dask <https://docs.dask.org/>`_.
Please read |ExecutionEngineTutorial| to understand this important Fugue concept
:param conf: |ParamsLikeObject| defaults to None, read |FugueConfig| to
learn Fugue specific options
:Notice:
You should setup Dask single machine or distributed environment in the
:doc:`common <dask:setup>` way. Before initializing :class:`~.DaskExecutionEngine`
"""
def __init__(self, conf: Any = None):
p = ParamDict(FUGUE_DASK_DEFAULT_CONF)
p.update(ParamDict(conf))
super().__init__(p)
self._fs = FileSystem()
self._log = logging.getLogger()
self._native = NativeExecutionEngine(conf=conf)
def __repr__(self) -> str:
return "DaskExecutionEngine"
@property
def log(self) -> logging.Logger:
return self._log
@property
def fs(self) -> FileSystem:
return self._fs
@property
def default_sql_engine(self) -> SQLEngine:
return QPDDaskEngine(self)
@property
def pl_utils(self) -> DaskUtils:
"""Pandas-like dataframe utils"""
return DaskUtils()
def to_df(self,
df: Any,
schema: Any = None,
metadata: Any = None) -> DaskDataFrame:
"""Convert a data structure to :class:`~fugue_dask.dataframe.DaskDataFrame`
:param data: :class:`~fugue.dataframe.dataframe.DataFrame`,
:class:`dask:dask.dataframe.DataFrame`,
pandas DataFrame or list or iterable of arrays
:param schema: |SchemaLikeObject|, defaults to None.
:param metadata: |ParamsLikeObject|, defaults to None
:return: engine compatible dataframe
:Notice:
* if the input is already :class:`~fugue_dask.dataframe.DaskDataFrame`,
it should return itself
* For list or iterable of arrays, ``schema`` must be specified
* When ``schema`` is not None, a potential type cast may happen to ensure
the dataframe's schema.
* all other methods in the engine can take arbitrary dataframes and
call this method to convert before doing anything
"""
default_partitions = self.conf.get_or_throw(
FUGUE_DASK_CONF_DATAFRAME_DEFAULT_PARTITIONS, int)
if isinstance(df, DataFrame):
assert_or_throw(
schema is None and metadata is None,
ValueError(
"schema and metadata must be None when df is a DataFrame"),
)
if isinstance(df, DaskDataFrame):
return df
if isinstance(df, PandasDataFrame):
return DaskDataFrame(df.native,
df.schema,
df.metadata,
num_partitions=default_partitions)
return DaskDataFrame(
df.as_array(type_safe=True),
df.schema,
df.metadata,
num_partitions=default_partitions,
)
return DaskDataFrame(df,
schema,
metadata,
num_partitions=default_partitions)
def repartition(self, df: DataFrame,
partition_spec: PartitionSpec) -> DaskDataFrame:
df = self.to_df(df)
if partition_spec.empty:
return df
if len(partition_spec.partition_by) > 0:
return df
p = partition_spec.get_num_partitions(
**{
KEYWORD_ROWCOUNT: lambda: df.persist().count(), # type: ignore
KEYWORD_CORECOUNT: lambda: 2, # TODO: remove this hard code
})
if p > 0:
return DaskDataFrame(
df.native.repartition(npartitions=p),
schema=df.schema,
metadata=df.metadata,
type_safe=False,
)
return df
def map(
self,
df: DataFrame,
map_func: Callable[[PartitionCursor, LocalDataFrame], LocalDataFrame],
output_schema: Any,
partition_spec: PartitionSpec,
metadata: Any = None,
on_init: Optional[Callable[[int, DataFrame], Any]] = None,
) -> DataFrame:
presort = partition_spec.presort
presort_keys = list(presort.keys())
presort_asc = list(presort.values())
output_schema = Schema(output_schema)
input_schema = df.schema
on_init_once: Any = (None if on_init is None else RunOnce(
on_init,
lambda *args, **kwargs: to_uuid(id(on_init), id(args[0]))))
def _map(pdf: Any) -> dd.DataFrame:
if pdf.shape[0] == 0:
return PandasDataFrame([], output_schema).as_pandas()
if len(presort_keys) > 0:
pdf = pdf.sort_values(presort_keys, ascending=presort_asc)
input_df = PandasDataFrame(pdf.reset_index(drop=True),
input_schema,
pandas_df_wrapper=True)
if on_init_once is not None:
on_init_once(0, input_df)
cursor = partition_spec.get_cursor(input_schema, 0)
cursor.set(input_df.peek_array(), 0, 0)
output_df = map_func(cursor, input_df)
return output_df.as_pandas()
df = self.to_df(df)
if len(partition_spec.partition_by) == 0:
pdf = self.repartition(df, partition_spec)
result = pdf.native.map_partitions(_map,
meta=output_schema.pandas_dtype)
else:
df = self.repartition(
df, PartitionSpec(num=partition_spec.num_partitions))
result = self.pl_utils.safe_groupby_apply(
df.native,
partition_spec.partition_by,
_map,
meta=output_schema.pandas_dtype,
)
return DaskDataFrame(result, output_schema, metadata)
def broadcast(self, df: DataFrame) -> DataFrame:
return self.to_df(df)
def persist(
self,
df: DataFrame,
lazy: bool = False,
**kwargs: Any,
) -> DataFrame:
return self.to_df(df).persist()
def join(
self,
df1: DataFrame,
df2: DataFrame,
how: str,
on: List[str] = _DEFAULT_JOIN_KEYS,
metadata: Any = None,
) -> DataFrame:
key_schema, output_schema = get_join_schemas(df1, df2, how=how, on=on)
d = self.pl_utils.join(
self.to_df(df1).native,
self.to_df(df2).native,
join_type=how,
on=key_schema.names,
)
return DaskDataFrame(d, output_schema, metadata)
def union(
self,
df1: DataFrame,
df2: DataFrame,
distinct: bool = True,
metadata: Any = None,
) -> DataFrame:
assert_or_throw(
df1.schema == df2.schema,
lambda: ValueError(f"{df1.schema} != {df2.schema}"),
)
d = self.pl_utils.union(self.to_df(df1).native,
self.to_df(df2).native,
unique=distinct)
return DaskDataFrame(d, df1.schema, metadata)
def subtract(
self,
df1: DataFrame,
df2: DataFrame,
distinct: bool = True,
metadata: Any = None,
) -> DataFrame:
assert_or_throw(
distinct,
NotImplementedError("EXCEPT ALL for DaskExecutionEngine"))
assert_or_throw(
df1.schema == df2.schema,
lambda: ValueError(f"{df1.schema} != {df2.schema}"),
)
d = self.pl_utils.except_df(self.to_df(df1).native,
self.to_df(df2).native,
unique=distinct)
return DaskDataFrame(d, df1.schema, metadata)
def intersect(
self,
df1: DataFrame,
df2: DataFrame,
distinct: bool = True,
metadata: Any = None,
) -> DataFrame:
assert_or_throw(
distinct,
NotImplementedError("INTERSECT ALL for DaskExecutionEngine"))
assert_or_throw(
df1.schema == df2.schema,
lambda: ValueError(f"{df1.schema} != {df2.schema}"),
)
d = self.pl_utils.intersect(self.to_df(df1).native,
self.to_df(df2).native,
unique=distinct)
return DaskDataFrame(d, df1.schema, metadata)
def distinct(
self,
df: DataFrame,
metadata: Any = None,
) -> DataFrame:
d = self.pl_utils.drop_duplicates(self.to_df(df).native)
return DaskDataFrame(d, df.schema, metadata)
def dropna(
self,
df: DataFrame,
how: str = "any",
thresh: int = None,
subset: List[str] = None,
metadata: Any = None,
) -> DataFrame:
d = self.to_df(df).native.dropna(how=how, thresh=thresh, subset=subset)
return DaskDataFrame(d, df.schema, metadata)
def fillna(
self,
df: DataFrame,
value: Any,
subset: List[str] = None,
metadata: Any = None,
) -> DataFrame:
assert_or_throw(
(not isinstance(value, list)) and (value is not None),
ValueError("fillna value can not be a list or None"),
)
if isinstance(value, dict):
assert_or_throw(
(None not in value.values()) and (any(value.values())),
ValueError(
"fillna dict can not contain None and needs at least one value"
),
)
mapping = value
else:
# If subset is none, apply to all columns
subset = subset or df.schema.names
mapping = {col: value for col in subset}
d = self.to_df(df).native.fillna(mapping)
return DaskDataFrame(d, df.schema, metadata)
def sample(
self,
df: DataFrame,
n: Optional[int] = None,
frac: Optional[float] = None,
replace: bool = False,
seed: Optional[int] = None,
metadata: Any = None,
) -> DataFrame:
assert_or_throw(
(n is None and frac is not None)
or (n is not None and frac is None),
ValueError("one and only one of n and frac should be set"),
)
# TODO: dask does not support sample by number of rows
d = self.to_df(df).native.sample(n=n,
frac=frac,
replace=replace,
random_state=seed)
return DaskDataFrame(d, df.schema, metadata)
def take(
self,
df: DataFrame,
n: int,
presort: str,
na_position: str = "last",
partition_spec: PartitionSpec = EMPTY_PARTITION_SPEC,
metadata: Any = None,
) -> DataFrame:
assert_or_throw(
isinstance(n, int),
ValueError("n needs to be an integer"),
)
d = self.to_df(df).native
meta = [(d[x].name, d[x].dtype) for x in d.columns]
if presort:
presort = parse_presort_exp(presort)
# Use presort over partition_spec.presort if possible
_presort: IndexedOrderedDict = presort or partition_spec.presort
def _partition_take(partition, n, presort):
if len(presort.keys()) > 0:
partition = partition.sort_values(
list(presort.keys()),
ascending=list(presort.values()),
na_position=na_position,
)
return partition.head(n)
if len(partition_spec.partition_by) == 0:
if len(_presort.keys()) == 0:
d = d.head(n)
else:
# Use the default partition
d = (d.map_partitions(
_partition_take, n, _presort,
meta=meta).reset_index(drop=True).compute())
# compute() brings this to Pandas so we can use pandas
d = d.sort_values(
list(_presort.keys()),
ascending=list(_presort.values()),
na_position=na_position,
).head(n)
else:
d = (d.groupby(partition_spec.partition_by,
dropna=False).apply(
_partition_take,
n=n,
presort=_presort,
meta=meta).reset_index(drop=True))
return DaskDataFrame(d, df.schema, metadata)
def load_df(
self,
path: Union[str, List[str]],
format_hint: Any = None,
columns: Any = None,
**kwargs: Any,
) -> DaskDataFrame:
return self.to_df(
load_df(path,
format_hint=format_hint,
columns=columns,
fs=self.fs,
**kwargs))
def save_df(
self,
df: DataFrame,
path: str,
format_hint: Any = None,
mode: str = "overwrite",
partition_spec: PartitionSpec = EMPTY_PARTITION_SPEC,
force_single: bool = False,
**kwargs: Any,
) -> None:
if force_single:
self._native.save_df(
df,
path=path,
format_hint=format_hint,
mode=mode,
partition_spec=partition_spec,
**kwargs,
)
else:
if not partition_spec.empty:
self.log.warning( # pragma: no cover
"partition_spec is not respected in %s.save_df", self)
self.fs.makedirs(os.path.dirname(path), recreate=True)
df = self.to_df(df)
save_df(df,
path,
format_hint=format_hint,
mode=mode,
fs=self.fs,
**kwargs)
