def test_reduce_agg():
schema = Schema(timestamp="timestamp*", category="str*", value="int")
values = {
"timestamp": [1589455901, 1589455901, 1589455902, 1589455902],
"category": list("abab"),
"value": [1, 2, 3, 4],
}
frm = Frame(schema, values)
for op in AST.aggregates:
if op == "quantile":
# quantile not avail with binning
continue
new_frm = frm.reduce(category="category", value=f"({op} self.value)")
if op == "min":
assert list(new_frm["value"]) == [1, 2]
elif op == "max":
assert list(new_frm["value"]) == [3, 4]
elif op == "sum":
assert list(new_frm["value"]) == [4, 6]
elif op in ("mean", "average"):
assert list(new_frm["value"]) == [2, 3]
elif op == "first":
assert list(new_frm["value"]) == [1, 2]
elif op == "last":
assert list(new_frm["value"]) == [3, 4]
elif op in ("count", "len"):
assert list(new_frm["value"]) == [2, 2]
else:
raise ValueError(f'op "{op}" not tested')
for op in AST.aggregates:
if op == "quantile":
# quantile not avail with binning
continue
new_frm = frm.reduce(timestamp='(floor self.timestamp "D")',
value=f"({op} self.value)")
if op == "min":
assert list(new_frm["value"]) == [1]
elif op == "max":
assert list(new_frm["value"]) == [4]
elif op == "sum":
assert list(new_frm["value"]) == [10]
elif op in ("mean", "average"):
assert list(new_frm["value"]) == [2.5]
elif op == "first":
assert list(new_frm["value"]) == [1]
elif op == "last":
assert list(new_frm["value"]) == [4]
elif op in ("count", "len"):
assert list(new_frm["value"]) == [4]
else:
raise ValueError(f'op "{op}" not tested')
def test_alias():
res = AST.parse("(as (asarray (list 1 2 3)) 'new_name')").eval()
arr = res.value
alias = res.name
assert all(arr == asarray([1, 2, 3]))
assert alias == "new_name"
frm = Frame(schema, values)
frm = frm.reduce("(as self.timestamp 'ts')")
assert all(frm["ts"] == asarray(values["timestamp"], "M"))
def test_reduce_without_agg():
schema = Schema(timestamp="timestamp*", category="str*", value="int")
values = {
"timestamp": [1589455901, 1589455901, 1589455902, 1589455902],
"category": list("abab"),
"value": [1, 2, 3, 4],
}
frm = Frame(schema, values)
# No changes to column
assert frm == frm.reduce(timestamp="timestamp",
category="category",
value="value")
# Mapping on one column
res = frm.reduce(value="(% self.value 2)")["value"]
assert list(res) == [1, 0, 1, 0]
# Mapping over two columns
expected = frm["timestamp"] + frm["value"]
new_frm = frm.reduce(new_col="(+ self.value self.timestamp)")
assert all(new_frm["new_col"] == expected)
