Skip to content

Instantly share code, notes, and snippets.

@bigsnarfdude

bigsnarfdude/SparkSQLexperiments.ipynb

Created July 1, 2014 05:03
Show Gist options
  • Save bigsnarfdude/c252d3b13ec2507b5012 to your computer and use it in GitHub Desktop.
Save bigsnarfdude/c252d3b13ec2507b5012 to your computer and use it in GitHub Desktop.
Download ZIP
Load Spark SQL from File, JSON file, or arrays
Display the source blob
Display the rendered blob
Raw
SparkSQLexperiments.ipynb
{
"metadata": {
"name": "",
"signature": "sha256:bd3b1a4d62b88d99f662b693b788f433bbf49f9f1d7f66187514a7a3a04e21d8"
},
"nbformat": 3,
"nbformat_minor": 0,
"worksheets": [
{
"cells": [
{
"cell_type": "markdown",
"metadata": {},
"source": [
"# Experiments to get json files loaded into SQL table"
]
},
{
"cell_type": "code",
"collapsed": false,
"input": [
"cat sql.py"
],
"language": "python",
"metadata": {},
"outputs": [
{
"output_type": "stream",
"stream": "stdout",
"text": [
"#\r\n",
"# Licensed to the Apache Software Foundation (ASF) under one or more\r\n",
"# contributor license agreements. See the NOTICE file distributed with\r\n",
"# this work for additional information regarding copyright ownership.\r\n",
"# The ASF licenses this file to You under the Apache License, Version 2.0\r\n",
"# (the \"License\"); you may not use this file except in compliance with\r\n",
"# the License. You may obtain a copy of the License at\r\n",
"#\r\n",
"# http://www.apache.org/licenses/LICENSE-2.0\r\n",
"#\r\n",
"# Unless required by applicable law or agreed to in writing, software\r\n",
"# distributed under the License is distributed on an \"AS IS\" BASIS,\r\n",
"# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.\r\n",
"# See the License for the specific language governing permissions and\r\n",
"# limitations under the License.\r\n",
"#\r\n",
"\r\n",
"from pyspark.rdd import RDD, PipelinedRDD\r\n",
"from pyspark.serializers import BatchedSerializer, PickleSerializer\r\n",
"\r\n",
"from py4j.protocol import Py4JError\r\n",
"\r\n",
"__all__ = [\"SQLContext\", \"HiveContext\", \"LocalHiveContext\", \"TestHiveContext\", \"SchemaRDD\", \"Row\"]\r\n",
"\r\n",
"\r\n",
"class SQLContext:\r\n",
" \"\"\"Main entry point for SparkSQL functionality.\r\n",
"\r\n",
" A SQLContext can be used create L{SchemaRDD}s, register L{SchemaRDD}s as\r\n",
" tables, execute SQL over tables, cache tables, and read parquet files.\r\n",
" \"\"\"\r\n",
"\r\n",
" def __init__(self, sparkContext, sqlContext = None):\r\n",
" \"\"\"Create a new SQLContext.\r\n",
"\r\n",
" @param sparkContext: The SparkContext to wrap.\r\n",
"\r\n",
" >>> srdd = sqlCtx.inferSchema(rdd)\r\n",
" >>> sqlCtx.inferSchema(srdd) # doctest: +IGNORE_EXCEPTION_DETAIL\r\n",
" Traceback (most recent call last):\r\n",
" ...\r\n",
" ValueError:...\r\n",
"\r\n",
" >>> bad_rdd = sc.parallelize([1,2,3])\r\n",
" >>> sqlCtx.inferSchema(bad_rdd) # doctest: +IGNORE_EXCEPTION_DETAIL\r\n",
" Traceback (most recent call last):\r\n",
" ...\r\n",
" ValueError:...\r\n",
"\r\n",
" >>> allTypes = sc.parallelize([{\"int\" : 1, \"string\" : \"string\", \"double\" : 1.0, \"long\": 1L,\r\n",
" ... \"boolean\" : True}])\r\n",
" >>> srdd = sqlCtx.inferSchema(allTypes).map(lambda x: (x.int, x.string, x.double, x.long,\r\n",
" ... x.boolean))\r\n",
" >>> srdd.collect()[0]\r\n",
" (1, u'string', 1.0, 1, True)\r\n",
" \"\"\"\r\n",
" self._sc = sparkContext\r\n",
" self._jsc = self._sc._jsc\r\n",
" self._jvm = self._sc._jvm\r\n",
" self._pythonToJavaMap = self._jvm.PythonRDD.pythonToJavaMap\r\n",
"\r\n",
" if sqlContext:\r\n",
" self._scala_SQLContext = sqlContext\r\n",
"\r\n",
" @property\r\n",
" def _ssql_ctx(self):\r\n",
" \"\"\"Accessor for the JVM SparkSQL context.\r\n",
"\r\n",
" Subclasses can override this property to provide their own\r\n",
" JVM Contexts.\r\n",
" \"\"\"\r\n",
" if not hasattr(self, '_scala_SQLContext'):\r\n",
" self._scala_SQLContext = self._jvm.SQLContext(self._jsc.sc())\r\n",
" return self._scala_SQLContext\r\n",
"\r\n",
" def inferSchema(self, rdd):\r\n",
" \"\"\"Infer and apply a schema to an RDD of L{dict}s.\r\n",
"\r\n",
" We peek at the first row of the RDD to determine the fields names\r\n",
" and types, and then use that to extract all the dictionaries. Nested\r\n",
" collections are supported, which include array, dict, list, set, and\r\n",
" tuple.\r\n",
"\r\n",
" >>> srdd = sqlCtx.inferSchema(rdd)\r\n",
" >>> srdd.collect() == [{\"field1\" : 1, \"field2\" : \"row1\"}, {\"field1\" : 2, \"field2\": \"row2\"},\r\n",
" ... {\"field1\" : 3, \"field2\": \"row3\"}]\r\n",
" True\r\n",
"\r\n",
" >>> from array import array\r\n",
" >>> srdd = sqlCtx.inferSchema(nestedRdd1)\r\n",
" >>> srdd.collect() == [{\"f1\" : array('i', [1, 2]), \"f2\" : {\"row1\" : 1.0}},\r\n",
" ... {\"f1\" : array('i', [2, 3]), \"f2\" : {\"row2\" : 2.0}}]\r\n",
" True\r\n",
"\r\n",
" >>> srdd = sqlCtx.inferSchema(nestedRdd2)\r\n",
" >>> srdd.collect() == [{\"f1\" : [[1, 2], [2, 3]], \"f2\" : set([1, 2]), \"f3\" : (1, 2)},\r\n",
" ... {\"f1\" : [[2, 3], [3, 4]], \"f2\" : set([2, 3]), \"f3\" : (2, 3)}]\r\n",
" True\r\n",
" \"\"\"\r\n",
" if (rdd.__class__ is SchemaRDD):\r\n",
" raise ValueError(\"Cannot apply schema to %s\" % SchemaRDD.__name__)\r\n",
" elif not isinstance(rdd.first(), dict):\r\n",
" raise ValueError(\"Only RDDs with dictionaries can be converted to %s: %s\" %\r\n",
" (SchemaRDD.__name__, rdd.first()))\r\n",
"\r\n",
" jrdd = self._pythonToJavaMap(rdd._jrdd)\r\n",
" srdd = self._ssql_ctx.inferSchema(jrdd.rdd())\r\n",
" return SchemaRDD(srdd, self)\r\n",
"\r\n",
" def registerRDDAsTable(self, rdd, tableName):\r\n",
" \"\"\"Registers the given RDD as a temporary table in the catalog.\r\n",
"\r\n",
" Temporary tables exist only during the lifetime of this instance of\r\n",
" SQLContext.\r\n",
"\r\n",
" >>> srdd = sqlCtx.inferSchema(rdd)\r\n",
" >>> sqlCtx.registerRDDAsTable(srdd, \"table1\")\r\n",
" \"\"\"\r\n",
" if (rdd.__class__ is SchemaRDD):\r\n",
" jschema_rdd = rdd._jschema_rdd\r\n",
" self._ssql_ctx.registerRDDAsTable(jschema_rdd, tableName)\r\n",
" else:\r\n",
" raise ValueError(\"Can only register SchemaRDD as table\")\r\n",
"\r\n",
" def parquetFile(self, path):\r\n",
" \"\"\"Loads a Parquet file, returning the result as a L{SchemaRDD}.\r\n",
"\r\n",
" >>> import tempfile, shutil\r\n",
" >>> parquetFile = tempfile.mkdtemp()\r\n",
" >>> shutil.rmtree(parquetFile)\r\n",
" >>> srdd = sqlCtx.inferSchema(rdd)\r\n",
" >>> srdd.saveAsParquetFile(parquetFile)\r\n",
" >>> srdd2 = sqlCtx.parquetFile(parquetFile)\r\n",
" >>> sorted(srdd.collect()) == sorted(srdd2.collect())\r\n",
" True\r\n",
" \"\"\"\r\n",
" jschema_rdd = self._ssql_ctx.parquetFile(path)\r\n",
" return SchemaRDD(jschema_rdd, self)\r\n",
"\r\n",
"\r\n",
" def jsonFile(self, path):\r\n",
" \"\"\"Loads a text file storing one JSON object per line,\r\n",
" returning the result as a L{SchemaRDD}.\r\n",
" It goes through the entire dataset once to determine the schema.\r\n",
"\r\n",
" >>> import tempfile, shutil\r\n",
" >>> jsonFile = tempfile.mkdtemp()\r\n",
" >>> shutil.rmtree(jsonFile)\r\n",
" >>> ofn = open(jsonFile, 'w')\r\n",
" >>> for json in jsonStrings:\r\n",
" ... print>>ofn, json\r\n",
" >>> ofn.close()\r\n",
" >>> srdd = sqlCtx.jsonFile(jsonFile)\r\n",
" >>> sqlCtx.registerRDDAsTable(srdd, \"table1\")\r\n",
" >>> srdd2 = sqlCtx.sql(\"SELECT field1 AS f1, field2 as f2, field3 as f3 from table1\")\r\n",
" >>> srdd2.collect() == [{\"f1\": 1, \"f2\": \"row1\", \"f3\":{\"field4\":11}},\r\n",
" ... {\"f1\": 2, \"f2\": \"row2\", \"f3\":{\"field4\":22}},\r\n",
" ... {\"f1\": 3, \"f2\": \"row3\", \"f3\":{\"field4\":33}}]\r\n",
" True\r\n",
" \"\"\"\r\n",
" jschema_rdd = self._ssql_ctx.jsonFile(path)\r\n",
" return SchemaRDD(jschema_rdd, self)\r\n",
"\r\n",
" def jsonRDD(self, rdd):\r\n",
" \"\"\"Loads an RDD storing one JSON object per string, returning the result as a L{SchemaRDD}.\r\n",
" It goes through the entire dataset once to determine the schema.\r\n",
"\r\n",
" >>> srdd = sqlCtx.jsonRDD(json)\r\n",
" >>> sqlCtx.registerRDDAsTable(srdd, \"table1\")\r\n",
" >>> srdd2 = sqlCtx.sql(\"SELECT field1 AS f1, field2 as f2, field3 as f3 from table1\")\r\n",
" >>> srdd2.collect() == [{\"f1\": 1, \"f2\": \"row1\", \"f3\":{\"field4\":11}},\r\n",
" ... {\"f1\": 2, \"f2\": \"row2\", \"f3\":{\"field4\":22}},\r\n",
" ... {\"f1\": 3, \"f2\": \"row3\", \"f3\":{\"field4\":33}}]\r\n",
" True\r\n",
" \"\"\"\r\n",
" def func(split, iterator):\r\n",
" for x in iterator:\r\n",
" if not isinstance(x, basestring):\r\n",
" x = unicode(x)\r\n",
" yield x.encode(\"utf-8\")\r\n",
" keyed = PipelinedRDD(rdd, func)\r\n",
" keyed._bypass_serializer = True\r\n",
" jrdd = keyed._jrdd.map(self._jvm.BytesToString())\r\n",
" jschema_rdd = self._ssql_ctx.jsonRDD(jrdd.rdd())\r\n",
" return SchemaRDD(jschema_rdd, self)\r\n",
"\r\n",
" def sql(self, sqlQuery):\r\n",
" \"\"\"Return a L{SchemaRDD} representing the result of the given query.\r\n",
"\r\n",
" >>> srdd = sqlCtx.inferSchema(rdd)\r\n",
" >>> sqlCtx.registerRDDAsTable(srdd, \"table1\")\r\n",
" >>> srdd2 = sqlCtx.sql(\"SELECT field1 AS f1, field2 as f2 from table1\")\r\n",
" >>> srdd2.collect() == [{\"f1\" : 1, \"f2\" : \"row1\"}, {\"f1\" : 2, \"f2\": \"row2\"},\r\n",
" ... {\"f1\" : 3, \"f2\": \"row3\"}]\r\n",
" True\r\n",
" \"\"\"\r\n",
" return SchemaRDD(self._ssql_ctx.sql(sqlQuery), self)\r\n",
"\r\n",
" def table(self, tableName):\r\n",
" \"\"\"Returns the specified table as a L{SchemaRDD}.\r\n",
"\r\n",
" >>> srdd = sqlCtx.inferSchema(rdd)\r\n",
" >>> sqlCtx.registerRDDAsTable(srdd, \"table1\")\r\n",
" >>> srdd2 = sqlCtx.table(\"table1\")\r\n",
" >>> sorted(srdd.collect()) == sorted(srdd2.collect())\r\n",
" True\r\n",
" \"\"\"\r\n",
" return SchemaRDD(self._ssql_ctx.table(tableName), self)\r\n",
"\r\n",
" def cacheTable(self, tableName):\r\n",
" \"\"\"Caches the specified table in-memory.\"\"\"\r\n",
" self._ssql_ctx.cacheTable(tableName)\r\n",
"\r\n",
" def uncacheTable(self, tableName):\r\n",
" \"\"\"Removes the specified table from the in-memory cache.\"\"\"\r\n",
" self._ssql_ctx.uncacheTable(tableName)\r\n",
"\r\n",
"\r\n",
"class HiveContext(SQLContext):\r\n",
" \"\"\"A variant of Spark SQL that integrates with data stored in Hive.\r\n",
"\r\n",
" Configuration for Hive is read from hive-site.xml on the classpath.\r\n",
" It supports running both SQL and HiveQL commands.\r\n",
" \"\"\"\r\n",
"\r\n",
" @property\r\n",
" def _ssql_ctx(self):\r\n",
" try:\r\n",
" if not hasattr(self, '_scala_HiveContext'):\r\n",
" self._scala_HiveContext = self._get_hive_ctx()\r\n",
" return self._scala_HiveContext\r\n",
" except Py4JError as e:\r\n",
" raise Exception(\"You must build Spark with Hive. Export 'SPARK_HIVE=true' and run \" \\\r\n",
" \"sbt/sbt assembly\" , e)\r\n",
"\r\n",
" def _get_hive_ctx(self):\r\n",
" return self._jvm.HiveContext(self._jsc.sc())\r\n",
"\r\n",
" def hiveql(self, hqlQuery):\r\n",
" \"\"\"\r\n",
" Runs a query expressed in HiveQL, returning the result as a L{SchemaRDD}.\r\n",
" \"\"\"\r\n",
" return SchemaRDD(self._ssql_ctx.hiveql(hqlQuery), self)\r\n",
"\r\n",
" def hql(self, hqlQuery):\r\n",
" \"\"\"\r\n",
" Runs a query expressed in HiveQL, returning the result as a L{SchemaRDD}.\r\n",
" \"\"\"\r\n",
" return self.hiveql(hqlQuery)\r\n",
"\r\n",
"\r\n",
"class LocalHiveContext(HiveContext):\r\n",
" \"\"\"Starts up an instance of hive where metadata is stored locally.\r\n",
"\r\n",
" An in-process metadata data is created with data stored in ./metadata.\r\n",
" Warehouse data is stored in in ./warehouse.\r\n",
"\r\n",
" >>> import os\r\n",
" >>> hiveCtx = LocalHiveContext(sc)\r\n",
" >>> try:\r\n",
" ... supress = hiveCtx.hql(\"DROP TABLE src\")\r\n",
" ... except Exception:\r\n",
" ... pass\r\n",
" >>> kv1 = os.path.join(os.environ[\"SPARK_HOME\"], 'examples/src/main/resources/kv1.txt')\r\n",
" >>> supress = hiveCtx.hql(\"CREATE TABLE IF NOT EXISTS src (key INT, value STRING)\")\r\n",
" >>> supress = hiveCtx.hql(\"LOAD DATA LOCAL INPATH '%s' INTO TABLE src\" % kv1)\r\n",
" >>> results = hiveCtx.hql(\"FROM src SELECT value\").map(lambda r: int(r.value.split('_')[1]))\r\n",
" >>> num = results.count()\r\n",
" >>> reduce_sum = results.reduce(lambda x, y: x + y)\r\n",
" >>> num\r\n",
" 500\r\n",
" >>> reduce_sum\r\n",
" 130091\r\n",
" \"\"\"\r\n",
"\r\n",
" def _get_hive_ctx(self):\r\n",
" return self._jvm.LocalHiveContext(self._jsc.sc())\r\n",
"\r\n",
"\r\n",
"class TestHiveContext(HiveContext):\r\n",
"\r\n",
" def _get_hive_ctx(self):\r\n",
" return self._jvm.TestHiveContext(self._jsc.sc())\r\n",
"\r\n",
"\r\n",
"# TODO: Investigate if it is more efficient to use a namedtuple. One problem is that named tuples\r\n",
"# are custom classes that must be generated per Schema.\r\n",
"class Row(dict):\r\n",
" \"\"\"A row in L{SchemaRDD}.\r\n",
"\r\n",
" An extended L{dict} that takes a L{dict} in its constructor, and\r\n",
" exposes those items as fields.\r\n",
"\r\n",
" >>> r = Row({\"hello\" : \"world\", \"foo\" : \"bar\"})\r\n",
" >>> r.hello\r\n",
" 'world'\r\n",
" >>> r.foo\r\n",
" 'bar'\r\n",
" \"\"\"\r\n",
"\r\n",
" def __init__(self, d):\r\n",
" d.update(self.__dict__)\r\n",
" self.__dict__ = d\r\n",
" dict.__init__(self, d)\r\n",
"\r\n",
"\r\n",
"class SchemaRDD(RDD):\r\n",
" \"\"\"An RDD of L{Row} objects that has an associated schema.\r\n",
"\r\n",
" The underlying JVM object is a SchemaRDD, not a PythonRDD, so we can\r\n",
" utilize the relational query api exposed by SparkSQL.\r\n",
"\r\n",
" For normal L{pyspark.rdd.RDD} operations (map, count, etc.) the\r\n",
" L{SchemaRDD} is not operated on directly, as it's underlying\r\n",
" implementation is an RDD composed of Java objects. Instead it is\r\n",
" converted to a PythonRDD in the JVM, on which Python operations can\r\n",
" be done.\r\n",
" \"\"\"\r\n",
"\r\n",
" def __init__(self, jschema_rdd, sql_ctx):\r\n",
" self.sql_ctx = sql_ctx\r\n",
" self._sc = sql_ctx._sc\r\n",
" self._jschema_rdd = jschema_rdd\r\n",
"\r\n",
" self.is_cached = False\r\n",
" self.is_checkpointed = False\r\n",
" self.ctx = self.sql_ctx._sc\r\n",
" self._jrdd_deserializer = self.ctx.serializer\r\n",
"\r\n",
" @property\r\n",
" def _jrdd(self):\r\n",
" \"\"\"Lazy evaluation of PythonRDD object.\r\n",
"\r\n",
" Only done when a user calls methods defined by the\r\n",
" L{pyspark.rdd.RDD} super class (map, filter, etc.).\r\n",
" \"\"\"\r\n",
" if not hasattr(self, '_lazy_jrdd'):\r\n",
" self._lazy_jrdd = self._toPython()._jrdd\r\n",
" return self._lazy_jrdd\r\n",
"\r\n",
" @property\r\n",
" def _id(self):\r\n",
" return self._jrdd.id()\r\n",
"\r\n",
" def saveAsParquetFile(self, path):\r\n",
" \"\"\"Save the contents as a Parquet file, preserving the schema.\r\n",
"\r\n",
" Files that are written out using this method can be read back in as\r\n",
" a SchemaRDD using the L{SQLContext.parquetFile} method.\r\n",
"\r\n",
" >>> import tempfile, shutil\r\n",
" >>> parquetFile = tempfile.mkdtemp()\r\n",
" >>> shutil.rmtree(parquetFile)\r\n",
" >>> srdd = sqlCtx.inferSchema(rdd)\r\n",
" >>> srdd.saveAsParquetFile(parquetFile)\r\n",
" >>> srdd2 = sqlCtx.parquetFile(parquetFile)\r\n",
" >>> sorted(srdd2.collect()) == sorted(srdd.collect())\r\n",
" True\r\n",
" \"\"\"\r\n",
" self._jschema_rdd.saveAsParquetFile(path)\r\n",
"\r\n",
" def registerAsTable(self, name):\r\n",
" \"\"\"Registers this RDD as a temporary table using the given name.\r\n",
"\r\n",
" The lifetime of this temporary table is tied to the L{SQLContext}\r\n",
" that was used to create this SchemaRDD.\r\n",
"\r\n",
" >>> srdd = sqlCtx.inferSchema(rdd)\r\n",
" >>> srdd.registerAsTable(\"test\")\r\n",
" >>> srdd2 = sqlCtx.sql(\"select * from test\")\r\n",
" >>> sorted(srdd.collect()) == sorted(srdd2.collect())\r\n",
" True\r\n",
" \"\"\"\r\n",
" self._jschema_rdd.registerAsTable(name)\r\n",
"\r\n",
" def insertInto(self, tableName, overwrite = False):\r\n",
" \"\"\"Inserts the contents of this SchemaRDD into the specified table.\r\n",
"\r\n",
" Optionally overwriting any existing data.\r\n",
" \"\"\"\r\n",
" self._jschema_rdd.insertInto(tableName, overwrite)\r\n",
"\r\n",
" def saveAsTable(self, tableName):\r\n",
" \"\"\"Creates a new table with the contents of this SchemaRDD.\"\"\"\r\n",
" self._jschema_rdd.saveAsTable(tableName)\r\n",
"\r\n",
" def schemaString(self):\r\n",
" \"\"\"Returns the output schema in the tree format.\"\"\"\r\n",
" return self._jschema_rdd.schemaString()\r\n",
"\r\n",
" def printSchema(self):\r\n",
" \"\"\"Prints out the schema in the tree format.\"\"\"\r\n",
" print self.schemaString()\r\n",
"\r\n",
" def count(self):\r\n",
" \"\"\"Return the number of elements in this RDD.\r\n",
"\r\n",
" Unlike the base RDD implementation of count, this implementation\r\n",
" leverages the query optimizer to compute the count on the SchemaRDD,\r\n",
" which supports features such as filter pushdown.\r\n",
"\r\n",
" >>> srdd = sqlCtx.inferSchema(rdd)\r\n",
" >>> srdd.count()\r\n",
" 3L\r\n",
" >>> srdd.count() == srdd.map(lambda x: x).count()\r\n",
" True\r\n",
" \"\"\"\r",
"\r\n",
" return self._jschema_rdd.count()\r\n",
"\r\n",
" def _toPython(self):\r\n",
" # We have to import the Row class explicitly, so that the reference Pickler has is\r\n",
" # pyspark.sql.Row instead of __main__.Row\r\n",
" from pyspark.sql import Row\r\n",
" jrdd = self._jschema_rdd.javaToPython()\r\n",
" # TODO: This is inefficient, we should construct the Python Row object\r\n",
" # in Java land in the javaToPython function. May require a custom\r\n",
" # pickle serializer in Pyrolite\r\n",
" return RDD(jrdd, self._sc, BatchedSerializer(\r\n",
" PickleSerializer())).map(lambda d: Row(d))\r\n",
"\r\n",
" # We override the default cache/persist/checkpoint behavior as we want to cache the underlying\r\n",
" # SchemaRDD object in the JVM, not the PythonRDD checkpointed by the super class\r\n",
" def cache(self):\r\n",
" self.is_cached = True\r\n",
" self._jschema_rdd.cache()\r\n",
" return self\r\n",
"\r\n",
" def persist(self, storageLevel):\r\n",
" self.is_cached = True\r\n",
" javaStorageLevel = self.ctx._getJavaStorageLevel(storageLevel)\r\n",
" self._jschema_rdd.persist(javaStorageLevel)\r\n",
" return self\r\n",
"\r\n",
" def unpersist(self):\r\n",
" self.is_cached = False\r\n",
" self._jschema_rdd.unpersist()\r\n",
" return self\r\n",
"\r\n",
" def checkpoint(self):\r\n",
" self.is_checkpointed = True\r\n",
" self._jschema_rdd.checkpoint()\r\n",
"\r\n",
" def isCheckpointed(self):\r\n",
" return self._jschema_rdd.isCheckpointed()\r\n",
"\r\n",
" def getCheckpointFile(self):\r\n",
" checkpointFile = self._jschema_rdd.getCheckpointFile()\r\n",
" if checkpointFile.isDefined():\r\n",
" return checkpointFile.get()\r\n",
" else:\r\n",
" return None\r\n",
"\r\n",
" def coalesce(self, numPartitions, shuffle=False):\r\n",
" rdd = self._jschema_rdd.coalesce(numPartitions, shuffle)\r\n",
" return SchemaRDD(rdd, self.sql_ctx)\r\n",
"\r\n",
" def distinct(self):\r\n",
" rdd = self._jschema_rdd.distinct()\r\n",
" return SchemaRDD(rdd, self.sql_ctx)\r\n",
"\r\n",
" def intersection(self, other):\r\n",
" if (other.__class__ is SchemaRDD):\r\n",
" rdd = self._jschema_rdd.intersection(other._jschema_rdd)\r\n",
" return SchemaRDD(rdd, self.sql_ctx)\r\n",
" else:\r\n",
" raise ValueError(\"Can only intersect with another SchemaRDD\")\r\n",
"\r\n",
" def repartition(self, numPartitions):\r\n",
" rdd = self._jschema_rdd.repartition(numPartitions)\r\n",
" return SchemaRDD(rdd, self.sql_ctx)\r\n",
"\r\n",
" def subtract(self, other, numPartitions=None):\r\n",
" if (other.__class__ is SchemaRDD):\r\n",
" if numPartitions is None:\r\n",
" rdd = self._jschema_rdd.subtract(other._jschema_rdd)\r\n",
" else:\r\n",
" rdd = self._jschema_rdd.subtract(other._jschema_rdd, numPartitions)\r\n",
" return SchemaRDD(rdd, self.sql_ctx)\r\n",
" else:\r\n",
" raise ValueError(\"Can only subtract another SchemaRDD\")\r\n",
"\r\n",
"def _test():\r\n",
" import doctest\r\n",
" from array import array\r\n",
" from pyspark.context import SparkContext\r\n",
" globs = globals().copy()\r\n",
" # The small batch size here ensures that we see multiple batches,\r\n",
" # even in these small test examples:\r\n",
" sc = SparkContext('local[4]', 'PythonTest', batchSize=2)\r\n",
" globs['sc'] = sc\r\n",
" globs['sqlCtx'] = SQLContext(sc)\r\n",
" globs['rdd'] = sc.parallelize([{\"field1\" : 1, \"field2\" : \"row1\"},\r\n",
" {\"field1\" : 2, \"field2\": \"row2\"}, {\"field1\" : 3, \"field2\": \"row3\"}])\r\n",
" jsonStrings = ['{\"field1\": 1, \"field2\": \"row1\", \"field3\":{\"field4\":11}}',\r\n",
" '{\"field1\" : 2, \"field2\": \"row2\", \"field3\":{\"field4\":22}}',\r\n",
" '{\"field1\" : 3, \"field2\": \"row3\", \"field3\":{\"field4\":33}}']\r\n",
" globs['jsonStrings'] = jsonStrings\r\n",
" globs['json'] = sc.parallelize(jsonStrings)\r\n",
" globs['nestedRdd1'] = sc.parallelize([\r\n",
" {\"f1\" : array('i', [1, 2]), \"f2\" : {\"row1\" : 1.0}},\r\n",
" {\"f1\" : array('i', [2, 3]), \"f2\" : {\"row2\" : 2.0}}])\r\n",
" globs['nestedRdd2'] = sc.parallelize([\r\n",
" {\"f1\" : [[1, 2], [2, 3]], \"f2\" : set([1, 2]), \"f3\" : (1, 2)},\r\n",
" {\"f1\" : [[2, 3], [3, 4]], \"f2\" : set([2, 3]), \"f3\" : (2, 3)}])\r\n",
" (failure_count, test_count) = doctest.testmod(globs=globs,optionflags=doctest.ELLIPSIS)\r\n",
" globs['sc'].stop()\r\n",
" if failure_count:\r\n",
" exit(-1)\r\n",
"\r\n",
"\r\n",
"if __name__ == \"__main__\":\r\n",
" _test()\r\n",
"\r\n"
]
}
],
"prompt_number": 1
},
{
"cell_type": "code",
"collapsed": false,
"input": [
"from sql import *"
],
"language": "python",
"metadata": {},
"outputs": [],
"prompt_number": 2
},
{
"cell_type": "code",
"collapsed": false,
"input": [
"import tempfile, shutil"
],
"language": "python",
"metadata": {},
"outputs": [],
"prompt_number": 3
},
{
"cell_type": "code",
"collapsed": false,
"input": [
"from array import array\n",
"from pyspark.sql import SQLContext"
],
"language": "python",
"metadata": {},
"outputs": [],
"prompt_number": 4
},
{
"cell_type": "code",
"collapsed": false,
"input": [
"sqlCtx = SQLContext(sc)"
],
"language": "python",
"metadata": {},
"outputs": [],
"prompt_number": 5
},
{
"cell_type": "code",
"collapsed": false,
"input": [
"srdd = sqlCtx.jsonFile('jsonFile')"
],
"language": "python",
"metadata": {},
"outputs": [],
"prompt_number": 23
},
{
"cell_type": "code",
"collapsed": false,
"input": [
"srdd.collect() == [{\"f1\": 1, \"f2\": \"row1\", \"f3\":{\"field4\":11}},\n",
" {\"f1\": 2, \"f2\": \"row2\", \"f3\":{\"field4\":22}},\n",
" {\"f1\": 3, \"f2\": \"row3\", \"f3\":{\"field4\":33}}]"
],
"language": "python",
"metadata": {},
"outputs": [
{
"metadata": {},
"output_type": "pyout",
"prompt_number": 26,
"text": [
"True"
]
}
],
"prompt_number": 26
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"#https://www.mail-archive.com/commits@spark.apache.org/msg01502.html"
]
},
{
"cell_type": "code",
"collapsed": false,
"input": [
"from array import array\n",
"from pyspark.sql import SQLContext"
],
"language": "python",
"metadata": {},
"outputs": [],
"prompt_number": 25
},
{
"cell_type": "code",
"collapsed": false,
"input": [
"sqlCtx = SQLContext(sc)\n",
"rdd = sc.parallelize([\n",
" {\"f1\" : array('i', [1, 2]), \"f2\" : {\"row1\" : 1.0}},\n",
" {\"f1\" : array('i', [2, 3]), \"f2\" : {\"row2\" : 2.0}}])"
],
"language": "python",
"metadata": {},
"outputs": [],
"prompt_number": 21
},
{
"cell_type": "code",
"collapsed": false,
"input": [
"srdd = sqlCtx.inferSchema(rdd)\n",
"srdd.collect() == [{\"f1\" : array('i', [1, 2]), \"f2\" : {\"row1\" : 1.0}},\n",
" {\"f1\" : array('i', [2, 3]), \"f2\" : {\"row2\" : 2.0}}]"
],
"language": "python",
"metadata": {},
"outputs": [
{
"metadata": {},
"output_type": "pyout",
"prompt_number": 22,
"text": [
"True"
]
}
],
"prompt_number": 22
},
{
"cell_type": "code",
"collapsed": false,
"input": [
"rdd = sc.parallelize([\n",
" {\"f1\" : [[1, 2], [2, 3]], \"f2\" : set([1, 2]), \"f3\" : (1, 2)},\n",
" {\"f1\" : [[2, 3], [3, 4]], \"f2\" : set([2, 3]), \"f3\" : (2, 3)}])"
],
"language": "python",
"metadata": {},
"outputs": [],
"prompt_number": 23
},
{
"cell_type": "code",
"collapsed": false,
"input": [
"srdd = sqlCtx.inferSchema(rdd)\n",
"srdd.collect() == [{\"f1\" : [[1, 2], [2, 3]], \"f2\" : set([1, 2]), \"f3\" : (1, 2)},\n",
" {\"f1\" : [[2, 3], [3, 4]], \"f2\" : set([2, 3]), \"f3\" : (2, 3)}]"
],
"language": "python",
"metadata": {},
"outputs": [
{
"metadata": {},
"output_type": "pyout",
"prompt_number": 26,
"text": [
"True"
]
}
],
"prompt_number": 26
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"# Load Spark SQL from JSON files"
]
},
{
"cell_type": "code",
"collapsed": false,
"input": [
"jsonFile\n",
"!cat 'jsonFile'"
],
"language": "python",
"metadata": {},
"outputs": [
{
"output_type": "stream",
"stream": "stdout",
"text": [
"{\"f1\": 1, \"f2\": \"row1\", \"f3\":{\"field4\":11}}\r\n",
"{\"f1\": 2, \"f2\": \"row2\", \"f3\":{\"field4\":22}}\r\n",
"{\"f1\": 3, \"f2\": \"row3\", \"f3\":{\"field4\":33}}\r\n"
]
}
],
"prompt_number": 29
},
{
"cell_type": "code",
"collapsed": false,
"input": [
"jsonStrings = !cat jsonFile"
],
"language": "python",
"metadata": {},
"outputs": [],
"prompt_number": 15
},
{
"cell_type": "code",
"collapsed": false,
"input": [
"ofn = open(jsonFile, 'w')"
],
"language": "python",
"metadata": {},
"outputs": [],
"prompt_number": 16
},
{
"cell_type": "code",
"collapsed": false,
"input": [
"for json in jsonStrings:\n",
" print>>ofn, json"
],
"language": "python",
"metadata": {},
"outputs": [],
"prompt_number": 17
},
{
"cell_type": "code",
"collapsed": false,
"input": [
"json"
],
"language": "python",
"metadata": {},
"outputs": [
{
"metadata": {},
"output_type": "pyout",
"prompt_number": 18,
"text": [
"'{\"f1\": 3, \"f2\": \"row3\", \"f3\":{\"field4\":33}}'"
]
}
],
"prompt_number": 18
},
{
"cell_type": "code",
"collapsed": false,
"input": [
"ofn.close()\n",
"!cat '/var/folders/dj/92mp96m54d90mdpqlmwbz1m40000gn/T/tmpMFXhlP'"
],
"language": "python",
"metadata": {},
"outputs": [
{
"output_type": "stream",
"stream": "stdout",
"text": [
"{'f1': 1, 'f2': 'row1', 'f3': {'field4': 11}}\r\n",
"{'f1': 2, 'f2': 'row2', 'f3': {'field4': 22}}\r\n",
"{'f1': 3, 'f2': 'row3', 'f3': {'field4': 33}}\r\n"
]
}
],
"prompt_number": 19
},
{
"cell_type": "code",
"collapsed": false,
"input": [
"srdd = sqlCtx.jsonFile(jsonFile)"
],
"language": "python",
"metadata": {},
"outputs": [],
"prompt_number": 20
},
{
"cell_type": "code",
"collapsed": false,
"input": [
"srdd.collect() == [{\"f1\": 1, \"f2\": \"row1\", \"f3\":{\"field4\":11}},\n",
" {\"f1\": 2, \"f2\": \"row2\", \"f3\":{\"field4\":22}},\n",
" {\"f1\": 3, \"f2\": \"row3\", \"f3\":{\"field4\":33}}]"
],
"language": "python",
"metadata": {},
"outputs": [
{
"metadata": {},
"output_type": "pyout",
"prompt_number": 22,
"text": [
"True"
]
}
],
"prompt_number": 22
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"#http://people.apache.org/~pwendell/spark-1.0.1-rc1-docs/sql-programming-guide.html"
]
},
{
"cell_type": "code",
"collapsed": false,
"input": [
"from pyspark.sql import SQLContext\n",
"sqlContext = SQLContext(sc)"
],
"language": "python",
"metadata": {},
"outputs": [],
"prompt_number": 30
},
{
"cell_type": "code",
"collapsed": false,
"input": [
"lines = sc.textFile(\"../../examples/src/main/resources/people.txt\")\n",
"parts = lines.map(lambda l: l.split(\",\"))\n",
"people = parts.map(lambda p: {\"name\": p[0], \"age\": int(p[1])})\n"
],
"language": "python",
"metadata": {},
"outputs": [],
"prompt_number": 33
},
{
"cell_type": "code",
"collapsed": false,
"input": [
"schemaPeople = sqlContext.inferSchema(people)\n",
"schemaPeople.registerAsTable(\"people\")"
],
"language": "python",
"metadata": {},
"outputs": [],
"prompt_number": 34
},
{
"cell_type": "code",
"collapsed": false,
"input": [
"teenagers = sqlContext.sql(\"SELECT name FROM people WHERE age >= 13 AND age <= 19\")\n"
],
"language": "python",
"metadata": {},
"outputs": [],
"prompt_number": 35
},
{
"cell_type": "code",
"collapsed": false,
"input": [
"teenNames = teenagers.map(lambda p: \"Name: \" + p.name)\n",
"for teenName in teenNames.collect():\n",
" print teenName"
],
"language": "python",
"metadata": {},
"outputs": [
{
"output_type": "stream",
"stream": "stdout",
"text": [
"Name: Justin\n"
]
}
],
"prompt_number": 36
},
{
"cell_type": "code",
"collapsed": false,
"input": [
"!cat ../../examples/src/main/resources/people.txt "
],
"language": "python",
"metadata": {},
"outputs": [
{
"output_type": "stream",
"stream": "stdout",
"text": [
"Michael, 29\r\n",
"Andy, 30\r\n",
"Justin, 19\r\n"
]
}
],
"prompt_number": 37
},
{
"cell_type": "code",
"collapsed": false,
"input": [],
"language": "python",
"metadata": {},
"outputs": []
},
{
"cell_type": "code",
"collapsed": false,
"input": [],
"language": "python",
"metadata": {},
"outputs": []
}
],
"metadata": {}
}
]
}
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment