def test_last_row_simple(self):
s = self.scheduler()
t1 = Table(name=get_random_name("cst1"),
data={
'xmin': [1],
'xmax': [2]
})
t2 = Table(name=get_random_name("cst2"),
data={
'ymin': [3],
'ymax': [4]
})
cst1 = Constant(t1, scheduler=s)
cst2 = Constant(t2, scheduler=s)
join = Join(scheduler=s)
join.input.table = cst1.output.table
join.input.table = cst2.output.table
pr = Print(proc=self.terse, scheduler=s)
pr.input.df = join.output.table
s.start()
s.join()
#res = join.trace_stats(max_runs=1)
#pd.set_option('display.expand_frame_repr', False)
#print(res)
df = join.table()
last = df.last()
self.assertTrue(last['xmin']==1 and last['xmax']==2 and \
last['ymin']==3 and last['ymax']==4)
def __init__(self,
name,
index,
base=None,
storagegroup=None,
dshape=None,
fillvalue=None,
shape=None,
chunks=None,
data=None,
indices=None):
"""Create a new column.
if index is None and self.index return None, a new index and dataset are created.
"""
super(Column, self).__init__(name, index, base=base)
if storagegroup is None:
if index is not None:
storagegroup = index.storagegroup
else:
storagegroup = Group.default(name=get_random_name('column_'))
self._storagegroup = storagegroup
self.dataset = None
self._dshape = None
if self.index is None:
if data is not None: # check before creating everything
l = len(data)
if indices and l != len(indices):
raise ValueError('Bad index length (%d/%d)', len(indices),
l)
self._complete_column(dshape, fillvalue, shape, chunks, data)
if data is not None:
self.append(data, indices)
def __init__(
self,
name: Optional[str],
data: Any = None,
dshape: Optional[Union[str, DataShape]] = None,
fillvalues: Optional[Dict[str, Any]] = None,
storagegroup: Optional[Group] = None,
chunks: Optional[Chunks] = None,
create: Optional[bool] = None,
indices: Optional[Index] = None,
):
# pylint: disable=too-many-arguments, too-many-branches
super(Table, self).__init__()
if not (fillvalues is None or isinstance(fillvalues, Mapping)):
raise ValueError(
"Invalid fillvalues (%s) should be None or a dictionary" %
fillvalues)
if not (chunks is None or isinstance(chunks,
(integer_types, Mapping))):
raise ValueError(
"Invalid chunks (%s) should be None or a dictionary" % chunks)
if data is not None:
if create is not None and create is not True:
logger.warning("creating a Table with data and create=False")
create = True
self._chunks = chunks
# self._nrow = 0
self._name: str = get_random_name("table_") if name is None else name
# TODO: attach all randomly named tables to a dedicated, common parent node
if not (storagegroup is None or isinstance(storagegroup, Group)):
raise ValueError(
"Invalid storagegroup (%s) should be None or a Group" %
storagegroup)
if storagegroup is None:
assert Group.default
storagegroup = Group.default(self._name, create=create)
if storagegroup is None:
raise RuntimeError("Cannot get a valid default storage Group")
self._storagegroup = storagegroup
if dshape is None:
if data is None:
self._dshape = EMPTY_DSHAPE
else:
data = self.parse_data(data)
self._dshape = dshape_extract(data) or EMPTY_DSHAPE
else:
self._dshape = dshape_create(dshape)
assert dshape_table_check(self._dshape)
if create and self._dshape is EMPTY_DSHAPE:
raise ValueError("Cannot create a table without a dshape")
if self._dshape is EMPTY_DSHAPE or (not create and metadata.ATTR_TABLE
in self._storagegroup.attrs):
self._load_table()
else:
self._create_table(fillvalues or {})
if data is not None:
self.append(data, indices=indices)
def __init__(self,
name,
data=None,
dshape=None,
fillvalues=None,
storagegroup=None,
chunks=None,
create=None,
indices=None):
# pylint: disable=too-many-arguments, too-many-branches
super(Table, self).__init__()
if not (fillvalues is None or isinstance(fillvalues, Mapping)):
raise ValueError(
'Invalid fillvalues (%s) should be None or a dictionary' %
fillvalues)
if not (chunks is None or isinstance(chunks,
(integer_types, Mapping))):
raise ValueError(
'Invalid chunks (%s) should be None or a dictionary' % chunks)
if data is not None:
if create is not None:
logger.warning('creating a Table with data and create=False')
create = True
self._chunks = chunks
#self._nrow = 0
self._name = get_random_name('table_') if name is None else name
# TODO: attach all randomly named tables to a dedicated, common parent node
if not (storagegroup is None or isinstance(storagegroup, Group)):
raise ValueError(
'Invalid storagegroup (%s) should be None or a Group' %
storagegroup)
if storagegroup is None:
storagegroup = Group.default(self._name, create=create)
if storagegroup is None:
raise RuntimeError('Cannot get a valid default storage Group')
self._storagegroup = storagegroup
if dshape is None:
if data is None:
self._dshape = None
else:
data = self.parse_data(data)
self._dshape = dshape_extract(data)
else:
self._dshape = dshape_create(dshape)
assert dshape_table_check(self._dshape)
if create and self._dshape is None:
raise ValueError('Cannot create a table without a dshape')
if self._dshape is None or (not create and metadata.ATTR_TABLE
in self._storagegroup.attrs):
self._load_table()
else:
self._create_table(fillvalues or {})
if data is not None:
self.append(data, indices=indices)
def __init__(self,
name: Optional[str] = None,
parent: Optional[GroupImpl] = None):
if name is None:
name = get_random_name("mmapstorage_")
super(MMapGroup, self).__init__(name, parent=parent)
if parent is not None:
if name in parent.dict:
raise ValueError(
"Cannot create group {}, already exists".format(name))
parent.dict[name] = self
self._is_init = False
def test_last_row_simple(self) -> None:
s = self.scheduler()
t1 = Table(name=get_random_name("cst1"), data={"xmin": [1], "xmax": [2]})
t2 = Table(name=get_random_name("cst2"), data={"ymin": [3], "ymax": [4]})
cst1 = Constant(t1, scheduler=s)
cst2 = Constant(t2, scheduler=s)
join = Join(scheduler=s)
join.input[0] = cst1.output.result
join.input[0] = cst2.output.result
pr = Print(proc=self.terse, scheduler=s)
pr.input[0] = join.output.result
aio.run(s.start())
# res = join.trace_stats(max_runs=1)
# pd.set_option('display.expand_frame_repr', False)
# print(res)
last = notNone(join.table.last())
self.assertTrue(
last["xmin"] == 1
and last["xmax"] == 2
and last["ymin"] == 3
and last["ymax"] == 4
)
def create_group(name=None, create=True):
root = StorageEngine.engines()['mmap']
if name in root.dict:
if create:
name = get_random_name(name[:16]+'_')
else:
return root.dict[name]
# TODO : specify this behaviour
#grp = root.dict[name]
#if not isinstance(grp, MMapGroup):
# raise ValueError("{} already exists and is not a group".format(name))
#return grp
return MMapGroup(name, parent=root)
def __init__(
self,
name: str,
index: Optional[IndexTable],
base: Optional[BaseColumn] = None,
storagegroup: Optional[Group] = None,
dshape: Optional[Union[None, DataShape, str]] = None,
fillvalue: Optional[Any] = None,
shape: Optional[Shape] = None,
chunks: Optional[Chunks] = None,
indices: Optional[Index] = None,
data: Optional[Any] = None,
) -> None:
"""Create a new column.
if index is None and self.index return None, a new index and
dataset are created.
"""
indexwasnone: bool = index is None
if index is None:
if data is not None: # check before creating everything
length = len(data)
if indices and length != len(indices):
raise ValueError("Bad index length (%d/%d)", len(indices),
length)
index = IndexTable()
super(Column, self).__init__(name, index, base=base)
if storagegroup is None:
if index is not None and hasattr(index, "storagegroup"):
# i.e. isinstance(index, Table)
storagegroup = getattr(index, "storagegroup")
assert isinstance(storagegroup, Group)
else:
assert Group.default
storagegroup = Group.default(name=get_random_name("column_"))
self._storagegroup = storagegroup
self.dataset: Optional[Dataset] = None
self._dshape: DataShape = EMPTY_DSHAPE
if isinstance(dshape, DataShape):
self._dshape = dshape
elif isinstance(dshape, str):
self._dshape = dshape_create(dshape)
if indexwasnone:
self._complete_column(dshape, fillvalue, shape, chunks, data)
if data is not None:
self.append(data, indices)
def __init__(self, column, min_column=None, max_column=None, reset_index=False, **kwds):
self._add_slots(kwds,'input_descriptors',
[SlotDescriptor('table', type=Table, required=True)])
super(Stats, self).__init__(table_slot='stats', **kwds)
self._column = column
self.default_step_size = 10000
if min_column is None:
min_column = '_' + str(column) + '_min'
if max_column is None:
max_column = '_' + str(column) + '_max'
self._min_column = min_column
self._max_column = max_column
self._reset_index = reset_index
# self.schema = [(self._min_column, np.dtype(float), np.nan),
# (self._max_column, np.dtype(float), np.nan),]
self.schema = '{'+self._min_column+': float64, '+self._max_column+': float64}'
self._table = Table(get_random_name('stats_'), dshape=self.schema)
def create_group(name: Optional[str] = None, create: bool = True) -> Group:
root = StorageEngine.engines()["mmap"]
assert isinstance(root, GroupImpl)
if name in root.dict:
if create:
name = get_random_name(name[:16] + "_")
else:
group = root.dict[name]
if isinstance(group, GroupImpl):
return group
raise ValueError(
f"Cannot create group {name}, already exists as {type(group)}"
)
# TODO : specify this behaviour
# grp = root.dict[name]
# if not isinstance(grp, MMapGroup):
# raise ValueError("{} already exists and is not a group".format(name))
# return grp
if create is False:
raise ValueError(f"group {name} does not exist")
return MMapGroup(name, parent=root)
def __init__(self, name=None, parent=None):
if name is None:
name = get_random_name("mmapstorage_")
super(MMapGroup, self).__init__(name, parent=parent)
self._directory = self.path()
metadata = os.path.join(self._directory, METADATA_FILE)
self._metadata = metadata
if os.path.exists(self._directory):
if not os.path.isdir(self._directory):
raise OSError('Cannot create group %s'%self._directory)
if not os.path.isfile(metadata):
raise ValueError('Cannot create group %s, "unsuitable directory'%
self._directory)
_read_attributes(self._attrs.attrs, metadata)
else:
os.makedirs(self._directory) # can raise exceptions
_write_attributes(self._attrs.attrs, metadata)
if parent is not None:
if name in parent.dict:
raise ValueError('Cannot create group {}, already exists'.format(name))
parent.dict[name] = self
def __init__(self,
column: Union[str, int],
min_column: Optional[str] = None,
max_column: Optional[str] = None,
reset_index: bool = False,
**kwds: Any) -> None:
super(Stats, self).__init__(**kwds)
self._column = column
self.default_step_size = 10000
if min_column is None:
min_column = "_" + str(column) + "_min"
if max_column is None:
max_column = "_" + str(column) + "_max"
self._min_column: str = min_column
self._max_column: str = max_column
self._reset_index = reset_index
# self.schema = [(self._min_column, np.dtype(float), np.nan),
# (self._max_column, np.dtype(float), np.nan),]
self.schema = ("{" + self._min_column + ": float64, " +
self._max_column + ": float64}")
self.result = Table(get_random_name("stats_"), dshape=self.schema)
