def test_element():
x = symbol('x', '5 * 3 * float32')
assert isscalar(x[1, 2].dshape)
assert x[1, 2].dshape == dshape('float32')
assert str(x[1, 2]) == 'x[1, 2]'
x = symbol('x', '5 * float32')
assert isscalar(x[3].dshape)
def test_element():
x = symbol('x', '5 * 3 * float32')
assert isscalar(x[1, 2].dshape)
assert x[1, 2].dshape == dshape('float32')
assert str(x[1, 2]) == 'x[1, 2]'
x = symbol('x', '5 * float32')
assert isscalar(x[3].dshape)
def compute_signature(expr):
"""Get the ``numba`` *function signature* corresponding to ``DataShape``
Examples
--------
>>> from blaze import symbol
>>> s = symbol('s', 'int64')
>>> t = symbol('t', 'float32')
>>> d = symbol('d', 'datetime')
>>> expr = s + t
>>> compute_signature(expr)
float64(int64, float32)
>>> expr = d.truncate(days=1)
>>> compute_signature(expr)
datetime64(D)(datetime64(us))
>>> expr = d.day + 1
>>> compute_signature(expr) # only looks at leaf nodes
int64(datetime64(us))
Notes
-----
* This could potentially be adapted/refactored to deal with
``datashape.Function`` types.
* Cannot handle ``datashape.Record`` types.
"""
assert datashape.isscalar(expr.schema)
restype = get_numba_type(expr.schema)
argtypes = [get_numba_type(e.schema) for e in expr._leaves()]
return restype(*argtypes)
def compute_signature(expr):
"""Get the ``numba`` *function signature* corresponding to ``DataShape``
Examples
--------
>>> from blaze import symbol
>>> s = symbol('s', 'int64')
>>> t = symbol('t', 'float32')
>>> d = symbol('d', 'datetime')
>>> expr = s + t
>>> compute_signature(expr)
float64(int64, float32)
>>> expr = d.truncate(days=1)
>>> compute_signature(expr)
datetime64(D)(datetime64(us))
>>> expr = d.day + 1
>>> compute_signature(expr) # only looks at leaf nodes
int64(datetime64(us))
Notes
-----
* This could potentially be adapted/refactored to deal with
``datashape.Function`` types.
* Cannot handle ``datashape.Record`` types.
"""
assert datashape.isscalar(expr.schema)
restype = get_numba_type(expr.schema)
argtypes = [get_numba_type(e.schema) for e in expr._leaves()]
return restype(*argtypes)
def fillvalue_for_expr(expr):
"""表达式默认值"""
fillmissing = _FILLVALUE_DEFAULTS.copy()
fillmissing.update({
int64_dtype: 0, # # 默认值 0
categorical_dtype: '未知', # 类似object_dtype，默认值为None
})
ret = {}
for name, type_ in expr.dshape.measure.fields:
if name in (AD_FIELD_NAME, SID_FIELD_NAME, TS_FIELD_NAME):
continue
if isscalar(type_):
n_type = datashape_type_to_numpy(type_)
ret[name] = fillmissing[n_type]
return ret
def fillvalue_for_expr(expr):
"""为表达式填充空白值
Arguments:
expr {Expr} -- 要使用的blaze表达式
"""
fillmissing = _FILLVALUE_DEFAULTS.copy()
fillmissing.update({
int64_dtype: -9999,
categorical_dtype: 'NA',
})
ret = {}
for name, type_ in expr.dshape.measure.fields:
if isscalar(type_):
n_type = datashape_type_to_numpy(type_)
ret[name] = fillmissing[n_type]
return ret
def fillvalue_for_expr(expr):
"""表达式默认值"""
fillmissing = _FILLVALUE_DEFAULTS.copy()
fillmissing.update({int64_dtype: -1})
ret = {}
# 传入类型 对象、浮点、整数、日期、逻辑
for name, type_ in expr.dshape.measure.fields:
if name in (AD_FIELD_NAME, SID_FIELD_NAME, TS_FIELD_NAME):
continue
if isscalar(type_):
n_type = datashape_type_to_numpy(type_)
if pd.core.dtypes.common.is_float_dtype(n_type):
ret[name] = fillmissing[float32_dtype]
elif pd.core.dtypes.common.is_datetime64_any_dtype(n_type):
ret[name] = fillmissing[datetime64ns_dtype]
elif pd.core.dtypes.common.is_integer_dtype(n_type):
ret[name] = fillmissing[int64_dtype]
elif pd.core.dtypes.common.is_object_dtype(n_type):
ret[name] = '' # fillmissing[object_dtype]
elif pd.core.dtypes.common.is_bool_dtype(n_type):
ret[name] = fillmissing[bool_dtype]
return ret
def compute_signature(expr):
"""Get the ``numba`` *function signature* corresponding to ``DataShape``
Examples
--------
>>> from blaze import symbol
>>> s = symbol('s', 'int64')
>>> t = symbol('t', 'float32')
>>> from numba import float64, int64, float32
>>> expr = s + t
>>> compute_signature(expr) == float64(int64, float32)
True
Notes
-----
* This could potentially be adapted/refactored to deal with
``datashape.Function`` types.
* Cannot handle ``datashape.Record`` types.
"""
assert datashape.isscalar(expr.schema)
restype = get_numba_type(expr.schema)
argtypes = [get_numba_type(e.schema) for e in expr._leaves()]
return restype(*argtypes)
def compute_signature(expr):
"""Get the ``numba`` *function signature* corresponding to ``DataShape``
Examples
--------
>>> from blaze import symbol
>>> s = symbol('s', 'int64')
>>> t = symbol('t', 'float32')
>>> from numba import float64, int64, float32
>>> expr = s + t
>>> compute_signature(expr) == float64(int64, float32)
True
Notes
-----
* This could potentially be adapted/refactored to deal with
``datashape.Function`` types.
* Cannot handle ``datashape.Record`` types.
"""
assert datashape.isscalar(expr.schema)
restype = get_numba_type(expr.schema)
argtypes = [get_numba_type(e.schema) for e in expr._leaves()]
return restype(*argtypes)
def from_blaze(expr,
deltas='auto',
loader=None,
resources=None,
no_deltas_rule=_valid_no_deltas_rules[0]):
"""Create a Pipeline API object from a blaze expression.
Parameters
----------
expr : Expr
The blaze expression to use.
deltas : Expr or 'auto', optional
The expression to use for the point in time adjustments.
If the string 'auto' is passed, a deltas expr will be looked up
by stepping up the expression tree and looking for another field
with the name of ``expr`` + '_deltas'. If None is passed, no deltas
will be used.
loader : BlazeLoader, optional
The blaze loader to attach this pipeline dataset to. If None is passed,
the global blaze loader is used.
resources : dict or any, optional
The data to execute the blaze expressions against. This is used as the
scope for ``bz.compute``.
no_deltas_rule : {'warn', 'raise', 'ignore'}
What should happen if ``deltas='auto'`` but no deltas can be found.
'warn' says to raise a warning but continue.
'raise' says to raise an exception if no deltas can be found.
'ignore' says take no action and proceed with no deltas.
Returns
-------
pipeline_api_obj : DataSet or BoundColumn
Either a new dataset or bound column based on the shape of the expr
passed in. If a table shaped expression is passed, this will return
a ``DataSet`` that represents the whole table. If an array-like shape
is passed, a ``BoundColumn`` on the dataset that would be constructed
from passing the parent is returned.
"""
deltas = _get_deltas(expr, deltas, no_deltas_rule)
if deltas is not None:
invalid_nodes = tuple(filter(is_invalid_deltas_node, expr._subterms()))
if invalid_nodes:
raise TypeError(
'expression with deltas may only contain (%s) nodes,'
" found: %s" % (
', '.join(map(getname, valid_deltas_node_types)),
', '.join(set(map(compose(getname, type), invalid_nodes))),
),
)
# Check if this is a single column out of a dataset.
if bz.ndim(expr) != 1:
raise TypeError(
'expression was not tabular or array-like,'
' %s dimensions: %d' % (
'too many' if bz.ndim(expr) > 1 else 'not enough',
bz.ndim(expr),
),
)
single_column = None
if isscalar(expr.dshape.measure):
# This is a single column. Record which column we are to return
# but create the entire dataset.
single_column = rename = expr._name
field_hit = False
if not isinstance(expr, traversable_nodes):
raise TypeError(
"expression '%s' was array-like but not a simple field of"
" some larger table" % str(expr),
)
while isinstance(expr, traversable_nodes):
if isinstance(expr, bz.expr.Field):
if not field_hit:
field_hit = True
else:
break
rename = expr._name
expr = expr._child
dataset_expr = expr.relabel({rename: single_column})
else:
dataset_expr = expr
measure = dataset_expr.dshape.measure
if not isrecord(measure) or AD_FIELD_NAME not in measure.names:
raise TypeError(
"The dataset must be a collection of records with at least an"
" '{ad}' field. Fields provided: '{fields}'\nhint: maybe you need"
" to use `relabel` to change your field names".format(
ad=AD_FIELD_NAME,
fields=measure,
),
)
_check_datetime_field(AD_FIELD_NAME, measure)
dataset_expr, deltas = _ensure_timestamp_field(dataset_expr, deltas)
if deltas is not None and (sorted(deltas.dshape.measure.fields) !=
sorted(measure.fields)):
raise TypeError(
'baseline measure != deltas measure:\n%s != %s' % (
measure,
deltas.dshape.measure,
),
)
# Ensure that we have a data resource to execute the query against.
_check_resources('dataset_expr', dataset_expr, resources)
_check_resources('deltas', deltas, resources)
# Create or retrieve the Pipeline API dataset.
ds = new_dataset(dataset_expr, deltas)
# Register our new dataset with the loader.
(loader if loader is not None else global_loader)[ds] = ExprData(
dataset_expr,
deltas,
resources,
)
if single_column is not None:
# We were passed a single column, extract and return it.
return getattr(ds, single_column)
return ds
def post_compute(expr, query, scope=None):
result = query.context.sql(sql_string(query.query))
if iscollection(expr.dshape) and isscalar(expr.dshape.measure):
result = result.map(lambda x: x[0])
return result.collect()
def post_compute(expr, query, scope=None):
result = query.context.sql(sql_string(query.query))
if iscollection(expr.dshape) and isscalar(expr.dshape.measure):
result = result.map(lambda x: x[0])
return result.collect()
