Github user mateiz commented on a diff in the pull request:
https://github.com/apache/spark/pull/455#discussion_r13147863
--- Diff: python/pyspark/context.py ---
@@ -311,6 +311,98 @@ def wholeTextFiles(self, path, minPartitions=None):
return RDD(self._jsc.wholeTextFiles(path, minPartitions), self,
PairDeserializer(UTF8Deserializer(),
UTF8Deserializer()))
+ def sequenceFile(self, name, key_class="org.apache.hadoop.io.Text",
value_class="org.apache.hadoop.io.Text",
+ key_wrapper="", value_wrapper="", minSplits=None):
+ """
+ Read a Hadoop SequenceFile with arbitrary key and value Writable
class from HDFS,
+ a local file system (available on all nodes), or any
Hadoop-supported file system URI.
+ The mechanism is as follows:
+ 1. A Java RDD is created from the SequenceFile or other
InputFormat, and the key and value Writable classes
+ 2. Serialization is attempted via Pyrolite pickling
+ 3. If this fails, the fallback is to call 'toString' on each
key and value
+ 4. C{PickleSerializer} is used to deserialize pickled objects
on the Python side
+ >>> sorted(sc.sequenceFile(tempdir +
"/sftestdata/sfint/").collect())
+ [(1, u'aa'), (1, u'aa'), (2, u'aa'), (2, u'bb'), (2, u'bb'), (3,
u'cc')]
+ >>> sorted(sc.sequenceFile(tempdir +
"/sftestdata/sfdouble/").collect())
+ [(1.0, u'aa'), (1.0, u'aa'), (2.0, u'aa'), (2.0, u'bb'), (2.0,
u'bb'), (3.0, u'cc')]
+ >>> sorted(sc.sequenceFile(tempdir +
"/sftestdata/sftext/").collect())
+ [(u'1', u'aa'), (u'1', u'aa'), (u'2', u'aa'), (u'2', u'bb'),
(u'2', u'bb'), (u'3', u'cc')]
+ >>> sorted(sc.sequenceFile(tempdir +
"/sftestdata/sfbool/").collect())
+ [(1, False), (1, True), (2, False), (2, False), (2, True), (3,
True)]
+ >>> sorted(sc.sequenceFile(tempdir +
"/sftestdata/sfnull/").collect())
+ [(1, None), (1, None), (2, None), (2, None), (2, None), (3, None)]
+ >>> sorted(sc.sequenceFile(tempdir +
"/sftestdata/sfmap/").collect())
+ [(1, {2.0: u'aa'}), (1, {3.0: u'bb'}), (2, {1.0: u'aa'}), (2,
{1.0: u'cc'}), (2, {3.0: u'bb'}), (3, {2.0: u'dd'})]
+ >>> r = sorted(sc.sequenceFile(tempdir +
"/sftestdata/sfclass").collect())[0]
+ >>> r == (u'1', {u'__class__':
u'org.apache.spark.api.python.TestWritable', u'double': 54.0, u'int': 123,
u'str': u'test1'})
+ True
+ """
+ minSplits = minSplits or min(self.defaultParallelism, 2)
+ jrdd = self._jvm.PythonRDD.sequenceFile(self._jsc, name,
key_class, value_class, key_wrapper, value_wrapper,
+ minSplits)
+ return RDD(jrdd, self, PickleSerializer())
+
+ def newAPIHadoopFile(self, name, inputformat_class,
key_class="org.apache.hadoop.io.Text",
+ value_class="org.apache.hadoop.io.Text",
key_wrapper="toString",
+ value_wrapper="toString", conf={}):
+ """
+ Read a 'new API' Hadoop InputFormat with arbitrary key and value
class from HDFS,
+ a local file system (available on all nodes), or any
Hadoop-supported file system URI.
+ The mechanism is the same as for sc.sequenceFile.
+
+ A Hadoop configuration can be passed in as a Python dict. This
will be converted into a Configuration in Java
+ """
+ jconf = self._jvm.java.util.HashMap()
+ for k, v in conf.iteritems():
+ jconf[k] = v
+ jrdd = self._jvm.PythonRDD.newAPIHadoopFile(self._jsc, name,
inputformat_class, key_class, value_class,
+ key_wrapper,
value_wrapper, jconf)
+ return RDD(jrdd, self, PickleSerializer())
+
+ def newAPIHadoopRDD(self, inputformat_class,
key_class="org.apache.hadoop.io.Text",
+ value_class="org.apache.hadoop.io.Text",
key_wrapper="", value_wrapper="", conf={}):
--- End diff --
For all of these methods I'd suggest not having a default key_class and
value_class. Did you find it inconvenient to always pass them? I'm worried that
a user will miss them and then get hard to explain ClassCastExceptions. It
seems like a small ask to make these required parameters by default.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
