[SPARK-5752][SQL] Don't implicitly convert RDDs directly to DataFrames

- The old implicit would convert RDDs directly to DataFrames, and that added too many methods.
- toDataFrame -> toDF
- Dsl -> functions
- implicits moved into SQLContext.implicits
- addColumn -> withColumn
- renameColumn -> withColumnRenamed

Python changes:
- toDataFrame -> toDF
- Dsl -> functions package
- addColumn -> withColumn
- renameColumn -> withColumnRenamed
- add toDF functions to RDD on SQLContext init
- add flatMap to DataFrame

Author: Reynold Xin <rxin@databricks.com>
Author: Davies Liu <davies@databricks.com>

Closes #4556 from rxin/SPARK-5752 and squashes the following commits:

5ef9910 [Reynold Xin] More fix
61d3fca [Reynold Xin] Merge branch 'df5' of github.com:davies/spark into SPARK-5752
ff5832c [Reynold Xin] Fix python
749c675 [Reynold Xin] count(*) fixes.
5806df0 [Reynold Xin] Fix build break again.
d941f3d [Reynold Xin] Fixed explode compilation break.
fe1267a [Davies Liu] flatMap
c4afb8e [Reynold Xin] style
d9de47f [Davies Liu] add comment
b783994 [Davies Liu] add comment for toDF
e2154e5 [Davies Liu] schema() -> schema
3a1004f [Davies Liu] Dsl -> functions, toDF()
fb256af [Reynold Xin] - toDataFrame -> toDF - Dsl -> functions - implicits moved into SQLContext.implicits - addColumn -> withColumn - renameColumn -> withColumnRenamed
0dd74eb [Reynold Xin] [SPARK-5752][SQL] Don't implicitly convert RDDs directly to DataFrames
97dd47c [Davies Liu] fix mistake
6168f74 [Davies Liu] fix test
1fc0199 [Davies Liu] fix test
a075cd5 [Davies Liu] clean up, toPandas
663d314 [Davies Liu] add test for agg('*')
9e214d5 [Reynold Xin] count(*) fixes.
1ed7136 [Reynold Xin] Fix build break again.
921b2e3 [Reynold Xin] Fixed explode compilation break.
14698d4 [Davies Liu] flatMap
ba3e12d [Reynold Xin] style
d08c92d [Davies Liu] add comment
5c8b524 [Davies Liu] add comment for toDF
a4e5e66 [Davies Liu] schema() -> schema
d377fc9 [Davies Liu] Dsl -> functions, toDF()
6b3086c [Reynold Xin] - toDataFrame -> toDF - Dsl -> functions - implicits moved into SQLContext.implicits - addColumn -> withColumn - renameColumn -> withColumnRenamed
807e8b1 [Reynold Xin] [SPARK-5752][SQL] Don't implicitly convert RDDs directly to DataFrames

1 files changed, 170 insertions, 0 deletions

diff --git a/python/pyspark/sql/functions.py b/python/pyspark/sql/functions.py
new file mode 100644
index 0000000000..39aa550eeb
--- /dev/null
+++ b/python/pyspark/sql/functions.py
@@ -0,0 +1,170 @@
+#
+# Licensed to the Apache Software Foundation (ASF) under one or more
+# contributor license agreements.  See the NOTICE file distributed with
+# this work for additional information regarding copyright ownership.
+# The ASF licenses this file to You under the Apache License, Version 2.0
+# (the "License"); you may not use this file except in compliance with
+# the License.  You may obtain a copy of the License at
+#
+#    http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+#
+
+"""
+A collections of builtin functions
+"""
+
+from itertools import imap
+
+from py4j.java_collections import ListConverter
+
+from pyspark import SparkContext
+from pyspark.rdd import _prepare_for_python_RDD
+from pyspark.serializers import PickleSerializer, AutoBatchedSerializer
+from pyspark.sql.types import StringType
+from pyspark.sql.dataframe import Column, _to_java_column
+
+
+__all__ = ['countDistinct', 'approxCountDistinct', 'udf']
+
+
+def _create_function(name, doc=""):
+    """ Create a function for aggregator by name"""
+    def _(col):
+        sc = SparkContext._active_spark_context
+        jc = getattr(sc._jvm.functions, name)(_to_java_column(col))
+        return Column(jc)
+    _.__name__ = name
+    _.__doc__ = doc
+    return _
+
+
+_functions = {
+    'lit': 'Creates a :class:`Column` of literal value.',
+    'col': 'Returns a :class:`Column` based on the given column name.',
+    'column': 'Returns a :class:`Column` based on the given column name.',
+    'upper': 'Converts a string expression to upper case.',
+    'lower': 'Converts a string expression to upper case.',
+    'sqrt': 'Computes the square root of the specified float value.',
+    'abs': 'Computes the absolutle value.',
+
+    'max': 'Aggregate function: returns the maximum value of the expression in a group.',
+    'min': 'Aggregate function: returns the minimum value of the expression in a group.',
+    'first': 'Aggregate function: returns the first value in a group.',
+    'last': 'Aggregate function: returns the last value in a group.',
+    'count': 'Aggregate function: returns the number of items in a group.',
+    'sum': 'Aggregate function: returns the sum of all values in the expression.',
+    'avg': 'Aggregate function: returns the average of the values in a group.',
+    'mean': 'Aggregate function: returns the average of the values in a group.',
+    'sumDistinct': 'Aggregate function: returns the sum of distinct values in the expression.',
+}
+
+
+for _name, _doc in _functions.items():
+    globals()[_name] = _create_function(_name, _doc)
+del _name, _doc
+__all__ += _functions.keys()
+
+
+def countDistinct(col, *cols):
+    """ Return a new Column for distinct count of `col` or `cols`
+
+    >>> df.agg(countDistinct(df.age, df.name).alias('c')).collect()
+    [Row(c=2)]
+
+    >>> df.agg(countDistinct("age", "name").alias('c')).collect()
+    [Row(c=2)]
+    """
+    sc = SparkContext._active_spark_context
+    jcols = ListConverter().convert([_to_java_column(c) for c in cols], sc._gateway._gateway_client)
+    jc = sc._jvm.functions.countDistinct(_to_java_column(col), sc._jvm.PythonUtils.toSeq(jcols))
+    return Column(jc)
+
+
+def approxCountDistinct(col, rsd=None):
+    """ Return a new Column for approximate distinct count of `col`
+
+    >>> df.agg(approxCountDistinct(df.age).alias('c')).collect()
+    [Row(c=2)]
+    """
+    sc = SparkContext._active_spark_context
+    if rsd is None:
+        jc = sc._jvm.functions.approxCountDistinct(_to_java_column(col))
+    else:
+        jc = sc._jvm.functions.approxCountDistinct(_to_java_column(col), rsd)
+    return Column(jc)
+
+
+class UserDefinedFunction(object):
+    """
+    User defined function in Python
+    """
+    def __init__(self, func, returnType):
+        self.func = func
+        self.returnType = returnType
+        self._broadcast = None
+        self._judf = self._create_judf()
+
+    def _create_judf(self):
+        f = self.func  # put it in closure `func`
+        func = lambda _, it: imap(lambda x: f(*x), it)
+        ser = AutoBatchedSerializer(PickleSerializer())
+        command = (func, None, ser, ser)
+        sc = SparkContext._active_spark_context
+        pickled_command, broadcast_vars, env, includes = _prepare_for_python_RDD(sc, command, self)
+        ssql_ctx = sc._jvm.SQLContext(sc._jsc.sc())
+        jdt = ssql_ctx.parseDataType(self.returnType.json())
+        judf = sc._jvm.UserDefinedPythonFunction(f.__name__, bytearray(pickled_command), env,
+                                                 includes, sc.pythonExec, broadcast_vars,
+                                                 sc._javaAccumulator, jdt)
+        return judf
+
+    def __del__(self):
+        if self._broadcast is not None:
+            self._broadcast.unpersist()
+            self._broadcast = None
+
+    def __call__(self, *cols):
+        sc = SparkContext._active_spark_context
+        jcols = ListConverter().convert([_to_java_column(c) for c in cols],
+                                        sc._gateway._gateway_client)
+        jc = self._judf.apply(sc._jvm.PythonUtils.toSeq(jcols))
+        return Column(jc)
+
+
+def udf(f, returnType=StringType()):
+    """Create a user defined function (UDF)
+
+    >>> slen = udf(lambda s: len(s), IntegerType())
+    >>> df.select(slen(df.name).alias('slen')).collect()
+    [Row(slen=5), Row(slen=3)]
+    """
+    return UserDefinedFunction(f, returnType)
+
+
+def _test():
+    import doctest
+    from pyspark.context import SparkContext
+    from pyspark.sql import Row, SQLContext
+    import pyspark.sql.dataframe
+    globs = pyspark.sql.dataframe.__dict__.copy()
+    sc = SparkContext('local[4]', 'PythonTest')
+    globs['sc'] = sc
+    globs['sqlCtx'] = SQLContext(sc)
+    globs['df'] = sc.parallelize([Row(name='Alice', age=2), Row(name='Bob', age=5)]).toDF()
+    globs['df2'] = sc.parallelize([Row(name='Tom', height=80), Row(name='Bob', height=85)]).toDF()
+    (failure_count, test_count) = doctest.testmod(
+        pyspark.sql.dataframe, globs=globs,
+        optionflags=doctest.ELLIPSIS | doctest.NORMALIZE_WHITESPACE)
+    globs['sc'].stop()
+    if failure_count:
+        exit(-1)
+
+
+if __name__ == "__main__":
+    _test()