You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@flink.apache.org by ch...@apache.org on 2016/01/20 17:42:48 UTC

flink git commit: [FLINK-2439] [py] Expand DataSet feature coverage

Repository: flink
Updated Branches:
  refs/heads/master ab8470714 -> ac42d1504


[FLINK-2439] [py] Expand DataSet feature coverage


Project: http://git-wip-us.apache.org/repos/asf/flink/repo
Commit: http://git-wip-us.apache.org/repos/asf/flink/commit/ac42d150
Tree: http://git-wip-us.apache.org/repos/asf/flink/tree/ac42d150
Diff: http://git-wip-us.apache.org/repos/asf/flink/diff/ac42d150

Branch: refs/heads/master
Commit: ac42d150441f968dfb2eaa578928f0e0d987e828
Parents: ab84707
Author: zentol <s....@web.de>
Authored: Wed Jan 20 15:30:10 2016 +0100
Committer: zentol <s....@web.de>
Committed: Wed Jan 20 15:30:10 2016 +0100

----------------------------------------------------------------------
 .../flink/python/api/PythonPlanBinder.java      |   2 +-
 .../python/api/functions/util/KeyDiscarder.java |   5 +-
 .../flink/python/api/flink/plan/Constants.py    |   4 +
 .../flink/python/api/flink/plan/DataSet.py      | 113 +++++++++++++++++++
 .../flink/python/api/flink/plan/Environment.py  |   8 ++
 .../python/api/flink/plan/OperationInfo.py      |   1 +
 .../org/apache/flink/python/api/test_main2.py   |  10 ++
 7 files changed, 139 insertions(+), 4 deletions(-)
----------------------------------------------------------------------


http://git-wip-us.apache.org/repos/asf/flink/blob/ac42d150/flink-libraries/flink-python/src/main/java/org/apache/flink/python/api/PythonPlanBinder.java
----------------------------------------------------------------------
diff --git a/flink-libraries/flink-python/src/main/java/org/apache/flink/python/api/PythonPlanBinder.java b/flink-libraries/flink-python/src/main/java/org/apache/flink/python/api/PythonPlanBinder.java
index f07a975..5b0d846 100644
--- a/flink-libraries/flink-python/src/main/java/org/apache/flink/python/api/PythonPlanBinder.java
+++ b/flink-libraries/flink-python/src/main/java/org/apache/flink/python/api/PythonPlanBinder.java
@@ -473,7 +473,7 @@ public class PythonPlanBinder {
 
 	private void createDistinctOperation(PythonOperationInfo info) throws IOException {
 		DataSet op = (DataSet) sets.get(info.parentID);
-		sets.put(info.setID, info.keys.length == 0 ? op.distinct() : op.distinct(info.keys).name("Distinct").map(new KeyDiscarder()).name("DistinctPostStep"));
+		sets.put(info.setID, op.distinct(info.keys).name("Distinct").map(new KeyDiscarder()).name("DistinctPostStep"));
 	}
 
 	private void createFirstOperation(PythonOperationInfo info) throws IOException {

http://git-wip-us.apache.org/repos/asf/flink/blob/ac42d150/flink-libraries/flink-python/src/main/java/org/apache/flink/python/api/functions/util/KeyDiscarder.java
----------------------------------------------------------------------
diff --git a/flink-libraries/flink-python/src/main/java/org/apache/flink/python/api/functions/util/KeyDiscarder.java b/flink-libraries/flink-python/src/main/java/org/apache/flink/python/api/functions/util/KeyDiscarder.java
index 118bc9f..b2af7be 100644
--- a/flink-libraries/flink-python/src/main/java/org/apache/flink/python/api/functions/util/KeyDiscarder.java
+++ b/flink-libraries/flink-python/src/main/java/org/apache/flink/python/api/functions/util/KeyDiscarder.java
@@ -14,16 +14,15 @@ package org.apache.flink.python.api.functions.util;
 
 import org.apache.flink.api.common.functions.MapFunction;
 import org.apache.flink.api.java.functions.FunctionAnnotation.ForwardedFields;
-import org.apache.flink.api.java.tuple.Tuple;
 import org.apache.flink.api.java.tuple.Tuple2;
 
 /*
 Utility function to extract the value from a Key-Value Tuple.
 */
 @ForwardedFields("f1->*")
-public class KeyDiscarder implements MapFunction<Tuple2<Tuple, byte[]>, byte[]> {
+public class KeyDiscarder <T> implements MapFunction<Tuple2<T, byte[]>, byte[]> {
 	@Override
-	public byte[] map(Tuple2<Tuple, byte[]> value) throws Exception {
+	public byte[] map(Tuple2<T, byte[]> value) throws Exception {
 		return value.f1;
 	}
 }

http://git-wip-us.apache.org/repos/asf/flink/blob/ac42d150/flink-libraries/flink-python/src/main/python/org/apache/flink/python/api/flink/plan/Constants.py
----------------------------------------------------------------------
diff --git a/flink-libraries/flink-python/src/main/python/org/apache/flink/python/api/flink/plan/Constants.py b/flink-libraries/flink-python/src/main/python/org/apache/flink/python/api/flink/plan/Constants.py
index 8881463..be9fc6d 100644
--- a/flink-libraries/flink-python/src/main/python/org/apache/flink/python/api/flink/plan/Constants.py
+++ b/flink-libraries/flink-python/src/main/python/org/apache/flink/python/api/flink/plan/Constants.py
@@ -45,6 +45,10 @@ class _Identifier(object):
     SINK_TEXT = "sink_text"
     SINK_PRINT = "sink_print"
     BROADCAST = "broadcast"
+    FIRST = "first"
+    DISTINCT = "distinct"
+    PARTITION_HASH = "partition_hash"
+    REBALANCE = "rebalance"
 
 
 class WriteMode(object):

http://git-wip-us.apache.org/repos/asf/flink/blob/ac42d150/flink-libraries/flink-python/src/main/python/org/apache/flink/python/api/flink/plan/DataSet.py
----------------------------------------------------------------------
diff --git a/flink-libraries/flink-python/src/main/python/org/apache/flink/python/api/flink/plan/DataSet.py b/flink-libraries/flink-python/src/main/python/org/apache/flink/python/api/flink/plan/DataSet.py
index eda8d02..3132651 100644
--- a/flink-libraries/flink-python/src/main/python/org/apache/flink/python/api/flink/plan/DataSet.py
+++ b/flink-libraries/flink-python/src/main/python/org/apache/flink/python/api/flink/plan/DataSet.py
@@ -277,6 +277,36 @@ class DataSet(object):
         self._env._sets.append(child)
         return child_set
 
+    def distinct(self, *fields):
+        """
+        Returns a distinct set of a tuple DataSet using field position keys.
+
+        :param fields: One or more field positions on which the distinction of the DataSet is decided.
+        :return: The distinct DataSet.
+        """
+        f = None
+        if len(fields) == 0:
+            f = lambda x: (x,)
+            fields = (0,)
+        if isinstance(fields[0], TYPES.FunctionType):
+            f = lambda x: (fields[0](x),)
+        if isinstance(fields[0], KeySelectorFunction):
+            f = lambda x: (fields[0].get_key(x),)
+        if f is None:
+            f = lambda x: tuple([x[key] for key in fields])
+        return self.map(lambda x: (f(x), x)).name("DistinctPreStep")._distinct(tuple([x for x in range(len(fields))]))
+
+    def _distinct(self, fields):
+        self._info.types = _createKeyValueTypeInfo(len(fields))
+        child = OperationInfo()
+        child_set = DataSet(self._env, child)
+        child.identifier = _Identifier.DISTINCT
+        child.parent = self._info
+        child.keys = fields
+        self._info.children.append(child)
+        self._env._sets.append(child)
+        return child_set
+
     def filter(self, operator):
         """
         Applies a Filter transformation on a DataSet.
@@ -302,6 +332,22 @@ class DataSet(object):
         self._env._sets.append(child)
         return child_set
 
+    def first(self, count):
+        """
+        Returns a new set containing the first n elements in this DataSet.
+
+        :param count: The desired number of elements.
+        :return: A DataSet containing the elements.
+        """
+        child = OperationInfo()
+        child_set = DataSet(self._env, child)
+        child.identifier = _Identifier.FIRST
+        child.parent = self._info
+        child.count = count
+        self._info.children.append(child)
+        self._env._sets.append(child)
+        return child_set
+
     def flat_map(self, operator):
         """
         Applies a FlatMap transformation on a DataSet.
@@ -426,6 +472,52 @@ class DataSet(object):
         self._env._sets.append(child)
         return child_set
 
+    def partition_by_hash(self, *fields):
+        f = None
+        if len(fields) == 0:
+            f = lambda x: (x,)
+        if isinstance(fields[0], TYPES.FunctionType):
+            f = lambda x: (fields[0](x),)
+        if isinstance(fields[0], KeySelectorFunction):
+            f = lambda x: (fields[0].get_key(x),)
+        if f is None:
+            f = lambda x: tuple([x[key] for key in fields])
+        return self.map(lambda x: (f(x), x)).name("HashPartitionPreStep")._partition_by_hash(tuple([x for x in range(len(fields))]))
+
+    def _partition_by_hash(self, fields):
+        """
+        Hash-partitions a DataSet on the specified key fields.
+        Important:This operation shuffles the whole DataSet over the network and can take significant amount of time.
+
+        :param fields: The field indexes on which the DataSet is hash-partitioned.
+        :return: The partitioned DataSet.
+        """
+        self._info.types = _createKeyValueTypeInfo(len(fields))
+        child = OperationInfo()
+        child_set = DataSet(self._env, child)
+        child.identifier = _Identifier.PARTITION_HASH
+        child.parent = self._info
+        child.keys = fields
+        self._info.children.append(child)
+        self._env._sets.append(child)
+        return child_set
+
+    def rebalance(self):
+        """
+        Enforces a re-balancing of the DataSet, i.e., the DataSet is evenly distributed over all parallel instances of the
+        following task. This can help to improve performance in case of heavy data skew and compute intensive operations.
+        Important:This operation shuffles the whole DataSet over the network and can take significant amount of time.
+
+        :return: The re-balanced DataSet.
+        """
+        child = OperationInfo()
+        child_set = DataSet(self._env, child)
+        child.identifier = _Identifier.REBALANCE
+        child.parent = self._info
+        self._info.children.append(child)
+        self._env._sets.append(child)
+        return child_set
+
     def union(self, other_set):
         """
         Creates a union of this DataSet with an other DataSet.
@@ -445,6 +537,10 @@ class DataSet(object):
         self._env._sets.append(child)
         return child_set
 
+    def name(self, name):
+        self._info.name = name
+        return self
+
 
 class OperatorSet(DataSet):
     def __init__(self, env, info):
@@ -472,6 +568,23 @@ class Grouping(object):
     def _finalize(self):
         pass
 
+    def first(self, count):
+        """
+        Returns a new set containing the first n elements in this DataSet.
+
+        :param count: The desired number of elements.
+        :return: A DataSet containing the elements.
+        """
+        self._finalize()
+        child = OperationInfo()
+        child_set = DataSet(self._env, child)
+        child.identifier = _Identifier.FIRST
+        child.parent = self._info
+        child.count = count
+        self._info.children.append(child)
+        self._env._sets.append(child)
+        return child_set
+
     def reduce_group(self, operator, combinable=False):
         """
         Applies a GroupReduce transformation.

http://git-wip-us.apache.org/repos/asf/flink/blob/ac42d150/flink-libraries/flink-python/src/main/python/org/apache/flink/python/api/flink/plan/Environment.py
----------------------------------------------------------------------
diff --git a/flink-libraries/flink-python/src/main/python/org/apache/flink/python/api/flink/plan/Environment.py b/flink-libraries/flink-python/src/main/python/org/apache/flink/python/api/flink/plan/Environment.py
index 4f2c5e3..777f30b 100644
--- a/flink-libraries/flink-python/src/main/python/org/apache/flink/python/api/flink/plan/Environment.py
+++ b/flink-libraries/flink-python/src/main/python/org/apache/flink/python/api/flink/plan/Environment.py
@@ -281,6 +281,14 @@ class Environment(object):
             collect(set.id)
             collect(set.parent.id)
             for case in Switch(identifier):
+                if case(_Identifier.REBALANCE):
+                    break
+                if case(_Identifier.DISTINCT, _Identifier.PARTITION_HASH):
+                    collect(set.keys)
+                    break
+                if case(_Identifier.FIRST):
+                    collect(set.count)
+                    break
                 if case(_Identifier.SORT):
                     collect(set.field)
                     collect(set.order)

http://git-wip-us.apache.org/repos/asf/flink/blob/ac42d150/flink-libraries/flink-python/src/main/python/org/apache/flink/python/api/flink/plan/OperationInfo.py
----------------------------------------------------------------------
diff --git a/flink-libraries/flink-python/src/main/python/org/apache/flink/python/api/flink/plan/OperationInfo.py b/flink-libraries/flink-python/src/main/python/org/apache/flink/python/api/flink/plan/OperationInfo.py
index 3605d7f..6eb228c 100644
--- a/flink-libraries/flink-python/src/main/python/org/apache/flink/python/api/flink/plan/OperationInfo.py
+++ b/flink-libraries/flink-python/src/main/python/org/apache/flink/python/api/flink/plan/OperationInfo.py
@@ -41,6 +41,7 @@ class OperationInfo():
             self.sinks = []
             self.children = []
             self.path = None
+            self.count = 0
             self.values = []
             self.projections = []
             self.bcvars = []

http://git-wip-us.apache.org/repos/asf/flink/blob/ac42d150/flink-libraries/flink-python/src/test/python/org/apache/flink/python/api/test_main2.py
----------------------------------------------------------------------
diff --git a/flink-libraries/flink-python/src/test/python/org/apache/flink/python/api/test_main2.py b/flink-libraries/flink-python/src/test/python/org/apache/flink/python/api/test_main2.py
index 6bf1fab..787928c 100644
--- a/flink-libraries/flink-python/src/test/python/org/apache/flink/python/api/test_main2.py
+++ b/flink-libraries/flink-python/src/test/python/org/apache/flink/python/api/test_main2.py
@@ -142,6 +142,16 @@ if __name__ == "__main__":
     d2 \
         .union(d4) \
         .map_partition(Verify2([(1, 0.5, "hello", True), (2, 0.4, "world", False), (1, 0.5, "hello", True), (1, 0.4, "hello", False), (1, 0.5, "hello", True), (2, 0.4, "world", False)], "Union")).output()
+    d1 \
+        .first(1) \
+        .map_partition(Verify([1], "First")).output()
+    d1 \
+        .rebalance()
+    d6 \
+        .distinct() \
+        .map_partition(Verify([1, 12], "Distinct")).output()
+    d2 \
+        .partition_by_hash(3)
 
 
     #Execution