You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@spark.apache.org by "Darcy Shen (Jira)" <ji...@apache.org> on 2021/03/19 02:54:00 UTC
[jira] [Updated] (SPARK-34799) Return User-defined types from
Pandas UDF
[ https://issues.apache.org/jira/browse/SPARK-34799?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Darcy Shen updated SPARK-34799:
-------------------------------
Description:
Because Pandas UDF uses pyarrow to passing data, it does not currently support UserDefinedTypes, as what normal python udf does.
For example:
{code:python}
class BoxType(UserDefinedType):
@classmethod
def sqlType(cls) -> StructType:
return StructType(
fields=[
StructField("xmin", DoubleType(), False),
StructField("ymin", DoubleType(), False),
StructField("xmax", DoubleType(), False),
StructField("ymax", DoubleType(), False),
]
)
@pandas_udf(
returnType=StructType([StructField("boxes", ArrayType(Box()))]
)
def pandas_pf(s: pd.DataFrame) -> pd.DataFrame:
yield s
{code}
The logs show
{code}
try:
to_arrow_type(self._returnType_placeholder)
except TypeError:
> raise NotImplementedError(
"Invalid return type with scalar Pandas UDFs: %s is "
E NotImplementedError: Invalid return type with scalar Pandas UDFs: StructType(List(StructField(boxes,ArrayType(Box,true),true))) is not supported
{code}
> Return User-defined types from Pandas UDF
> -----------------------------------------
>
> Key: SPARK-34799
> URL: https://issues.apache.org/jira/browse/SPARK-34799
> Project: Spark
> Issue Type: Sub-task
> Components: PySpark, SQL
> Affects Versions: 3.0.2, 3.1.1
> Reporter: Darcy Shen
> Priority: Major
>
> Because Pandas UDF uses pyarrow to passing data, it does not currently support UserDefinedTypes, as what normal python udf does.
> For example:
> {code:python}
> class BoxType(UserDefinedType):
> @classmethod
> def sqlType(cls) -> StructType:
> return StructType(
> fields=[
> StructField("xmin", DoubleType(), False),
> StructField("ymin", DoubleType(), False),
> StructField("xmax", DoubleType(), False),
> StructField("ymax", DoubleType(), False),
> ]
> )
> @pandas_udf(
> returnType=StructType([StructField("boxes", ArrayType(Box()))]
> )
> def pandas_pf(s: pd.DataFrame) -> pd.DataFrame:
> yield s
> {code}
> The logs show
> {code}
> try:
> to_arrow_type(self._returnType_placeholder)
> except TypeError:
> > raise NotImplementedError(
> "Invalid return type with scalar Pandas UDFs: %s is "
> E NotImplementedError: Invalid return type with scalar Pandas UDFs: StructType(List(StructField(boxes,ArrayType(Box,true),true))) is not supported
> {code}
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@spark.apache.org
For additional commands, e-mail: issues-help@spark.apache.org