Repository: spark Updated Branches: refs/heads/branch-2.3 2fd7acabf -> 328dea6f8
[SPARK-23645][MINOR][DOCS][PYTHON] Add docs RE `pandas_udf` with keyword args ## What changes were proposed in this pull request? Add documentation about the limitations of `pandas_udf` with keyword arguments and related concepts, like `functools.partial` fn objects. NOTE: intermediate commits on this PR show some of the steps that can be taken to fix some (but not all) of these pain points. ### Survey of problems we face today: (Initialize) Note: python 3.6 and spark 2.4snapshot. ``` from pyspark.sql import SparkSession import inspect, functools from pyspark.sql.functions import pandas_udf, PandasUDFType, col, lit, udf spark = SparkSession.builder.getOrCreate() print(spark.version) df = spark.range(1,6).withColumn('b', col('id') * 2) def ok(a,b): return a+b ``` Using a keyword argument at the call site `b=...` (and yes, *full* stack trace below, haha): ``` ---> 14 df.withColumn('ok', pandas_udf(f=ok, returnType='bigint')('id', b='id')).show() # no kwargs TypeError: wrapper() got an unexpected keyword argument 'b' ``` Using partial with a keyword argument where the kw-arg is the first argument of the fn: *(Aside: kind of interesting that lines 15,16 work great and then 17 explodes)* ``` --------------------------------------------------------------------------- ValueError Traceback (most recent call last) <ipython-input-9-e9f31b8799c1> in <module>() 15 df.withColumn('ok', pandas_udf(f=functools.partial(ok, 7), returnType='bigint')('id')).show() 16 df.withColumn('ok', pandas_udf(f=functools.partial(ok, b=7), returnType='bigint')('id')).show() ---> 17 df.withColumn('ok', pandas_udf(f=functools.partial(ok, a=7), returnType='bigint')('id')).show() /Users/stu/ZZ/spark/python/pyspark/sql/functions.py in pandas_udf(f, returnType, functionType) 2378 return functools.partial(_create_udf, returnType=return_type, evalType=eval_type) 2379 else: -> 2380 return _create_udf(f=f, returnType=return_type, evalType=eval_type) 2381 2382 /Users/stu/ZZ/spark/python/pyspark/sql/udf.py in _create_udf(f, returnType, evalType) 54 argspec.varargs is None: 55 raise ValueError( ---> 56 "Invalid function: 0-arg pandas_udfs are not supported. " 57 "Instead, create a 1-arg pandas_udf and ignore the arg in your function." 58 ) ValueError: Invalid function: 0-arg pandas_udfs are not supported. Instead, create a 1-arg pandas_udf and ignore the arg in your function. ``` Author: Michael (Stu) Stewart <mstewart...@gmail.com> Closes #20900 from mstewart141/udfkw2. (cherry picked from commit 087fb3142028d679524e22596b0ad4f74ff47e8d) Signed-off-by: hyukjinkwon <gurwls...@apache.org> Signed-off-by: hyukjinkwon <gurwls...@apache.org> Project: http://git-wip-us.apache.org/repos/asf/spark/repo Commit: http://git-wip-us.apache.org/repos/asf/spark/commit/328dea6f Tree: http://git-wip-us.apache.org/repos/asf/spark/tree/328dea6f Diff: http://git-wip-us.apache.org/repos/asf/spark/diff/328dea6f Branch: refs/heads/branch-2.3 Commit: 328dea6f8ffcd515face7d64c29f7af71abd88a2 Parents: 2fd7aca Author: Michael (Stu) Stewart <mstewart...@gmail.com> Authored: Mon Mar 26 12:45:45 2018 +0900 Committer: hyukjinkwon <gurwls...@apache.org> Committed: Mon Mar 26 12:45:58 2018 +0900 ---------------------------------------------------------------------- python/pyspark/sql/functions.py | 4 ++++ 1 file changed, 4 insertions(+) ---------------------------------------------------------------------- http://git-wip-us.apache.org/repos/asf/spark/blob/328dea6f/python/pyspark/sql/functions.py ---------------------------------------------------------------------- diff --git a/python/pyspark/sql/functions.py b/python/pyspark/sql/functions.py index 316aa85..365be7b 100644 --- a/python/pyspark/sql/functions.py +++ b/python/pyspark/sql/functions.py @@ -2123,6 +2123,8 @@ def udf(f=None, returnType=StringType()): in boolean expressions and it ends up with being executed all internally. If the functions can fail on special rows, the workaround is to incorporate the condition into the functions. + .. note:: The user-defined functions do not take keyword arguments on the calling side. + :param f: python function if used as a standalone function :param returnType: the return type of the user-defined function. The value can be either a :class:`pyspark.sql.types.DataType` object or a DDL-formatted type string. @@ -2252,6 +2254,8 @@ def pandas_udf(f=None, returnType=None, functionType=None): .. note:: The user-defined functions do not support conditional expressions or short circuiting in boolean expressions and it ends up with being executed all internally. If the functions can fail on special rows, the workaround is to incorporate the condition into the functions. + + .. note:: The user-defined functions do not take keyword arguments on the calling side. """ # decorator @pandas_udf(returnType, functionType) is_decorator = f is None or isinstance(f, (str, DataType)) --------------------------------------------------------------------- To unsubscribe, e-mail: commits-unsubscr...@spark.apache.org For additional commands, e-mail: commits-h...@spark.apache.org