Build Twitter Scala API Library for Spark Streaming using sbt

If you need to import Twitter API library in Scala:

import org.apache.spark.streaming.twitter._

You will need to build the jar file and place the jar file in the classpath. In our case, default classpath is $SPARK_HOME/jars

To start with, know your Spark and Scala version, which can be found by running $SPARK_HOME/bin/spark-shell

$SPARK_HOME/bin/spark-shell
Setting default log level to "WARN".
To adjust logging level use sc.setLogLevel(newLevel). For SparkR, use setLogLevel(newLevel).
20/06/12 14:06:22 WARN NativeCodeLoader: Unable to load native-hadoop library for your platform... using builtin-java classes where applicable
Spark context Web UI available at http://10.0.2.15:4040
Spark context available as 'sc' (master = local[*], app id = local-1591995983793).
Spark session available as 'spark'.
Welcome to
      ____              __
     / __/__  ___ _____/ /__
    _\ \/ _ \/ _ `/ __/  '_/
   /___/ .__/\_,_/_/ /_/\_\   version 2.2.1
      /_/

Using Scala version 2.11.8 (Java HotSpot(TM) 64-Bit Server VM, Java 1.8.0_171)
Type in expressions to have them evaluated.
Type :help for more information.

In this example, Spark version is 2.2.1, Scala version is 2.11.8

To build the "twitter jar" file, you need to manually create directory structure, let's call twitter for example, that contains subfolder project and src. Subfolder names are significant. The directory structure is like below:

(base) [hadoop@master twitter]$ tree -a
.
├── build.sbt
├── project
│   ├── assembly.sbt
│   └── build.properties
└── src
    └── main
        └── scala

Install tree command on Linux, here is CentOS, do below

sudo yum install tree -y

test tree command to see if it works

tree -d

There are 1 files from the root directory called build.sbt, you need to create:

vi build.sbt

// this file was written for spark 3.0.0-preview and scala 2.12.10
// George Jen
// Jen Tek LLC

version := "1"
name := "JentekLLC-spark-streaming-from-Twitter"
scalaVersion := "2.11.8"

libraryDependencies += "org.apache.spark" %% "spark-core" % "2.2.1" % "provided"
libraryDependencies += "org.apache.spark" %% "spark-streaming" % "2.2.1" % "provided"
libraryDependencies += "org.apache.spark" %% "spark-sql" % "2.2.1" % "provided"
libraryDependencies += "org.twitter4j" % "twitter4j-core" % "4.0.4"
libraryDependencies += "org.twitter4j" % "twitter4j-stream" % "4.0.4"
libraryDependencies += "org.apache.bahir" %% "spark-streaming-twitter" % "2.2.1"

assemblyMergeStrategy in assembly := {
  case PathList("org","aopalliance", xs @ _*) => MergeStrategy.last
  case PathList("javax", "inject", xs @ _*) => MergeStrategy.last
  case PathList("javax", "servlet", xs @ _*) => MergeStrategy.last
  case PathList("javax", "activation", xs @ _*) => MergeStrategy.last
  case PathList("org", "apache", xs @ _*) => MergeStrategy.last
  case PathList("com", "google", xs @ _*) => MergeStrategy.last
  case PathList("com", "esotericsoftware", xs @ _*) => MergeStrategy.last
  case PathList("com", "codahale", xs @ _*) => MergeStrategy.last
  case PathList("com", "yammer", xs @ _*) => MergeStrategy.last
  case "about.html" => MergeStrategy.rename
  case "META-INF/ECLIPSEF.RSA" => MergeStrategy.last
  case "META-INF/mailcap" => MergeStrategy.last
  case "META-INF/mimetypes.default" => MergeStrategy.last
  case "plugin.properties" => MergeStrategy.last
  case "log4j.properties" => MergeStrategy.last
  case x =>
    val oldStrategy = (assemblyMergeStrategy in assembly).value
    oldStrategy(x)
}

You need to set your Scala version, in my example, 2.11.8. Your one might be different.

scalaVersion := "2.11.8"

Then fill in your Spark version in these lines, mine is 2.2.1

libraryDependencies += "org.apache.spark" %% "spark-core" % "2.2.1" % "provided"

libraryDependencies += "org.apache.spark" %% "spark-streaming" % "2.2.1" % "provided"

libraryDependencies += "org.apache.spark" %% "spark-sql" % "2.2.1" % "provided"

libraryDependencies += "org.apache.bahir" %% "spark-streaming-twitter" % "2.2.1"

Also change the line spark-streaming-twitter to your Spark version, mine is 2.2.1

libraryDependencies += "org.apache.bahir" %% "spark-streaming-twitter" % "2.2.1"

Save and exit

Inside sub-folder project, there are 2 files: assembly.sbt and build.properties, you need to create both:

vi project/assembly.sbt

addSbtPlugin("com.eed3si9n" % "sbt-assembly" % "0.14.10")

Save and exit

vi project/build.properties

sbt.version=1.3.6

Save and exit

You do not need place any source code file in src/main/scala folder in this excercise. We only need to download needed twitter jar files from Maven repository and assemble into one single combined jar file by sbt.

If you do not have sbt installed, you can get it at

https://www.scala-sbt.org/download.html

Download the zip file, sbt-1.3.10.zip (note, the file name may be changing over the time). On linux, do below to download and install sbt:

mkdir ~/sbt
cd ~/sbt
wget https://piccolo.link/sbt-1.3.10.zip
unzip sbt-1.3.10.zip
sudo cp sbt/bin/* /usr/local/bin

Now, you should be able to build Twitter jar file using sbt

cd ~/twitter
#Test below to see you have the right 
#directory structure with right files

(base) [hadoop@master twitter]$ tree -a
.
├── build.sbt
├── project
│   ├── assembly.sbt
│   └── build.properties
└── src
    └── main
        └── scala

#Now you run sbt at the root of the directory
#that has the build.sbt file

You must run the sbt command below at the root level of the project folder, that has the file build.sbt

sbt assembly

The first time, it will takes a while and download lots of files, eventually, it will show below:

(base) [hadoop@master twitter]$ sbt assembly
[info] Loading settings for project twitter-build from assembly.sbt ...
[info] Loading project definition from /opt/hadoop/sbt/twitter/project
[info] Loading settings for project twitter from build.sbt ...
[info] Set current project to JentekLLC-spark-streaming-from-Twitter (in build file:/opt/hadoop/sbt/twitter/)
[info] Strategy 'discard' was applied to 12 files (Run the task at debug level to see details)
[info] Strategy 'last' was applied to a file (Run the task at debug level to see details)
[info] Strategy 'rename' was applied to 3 files (Run the task at debug level to see details)
[success] Total time: 8 s, completed Jun 12, 2020 3:47:03 PM

It produces a new directory called target, with subdirectory called scala-<version>, simply cd into it

(base) [hadoop@master twitter]$ ls
build.sbt  project  src  target
(base) [hadoop@master twitter]$ cd target
(base) [hadoop@master target]$ ls
scala-2.11  streams
(base) [hadoop@master target]$ cd scala*
(base) [hadoop@master scala-2.11]$ pwd
/opt/hadoop/sbt/twitter/target/scala-2.11

Now you should see twitter jar file produced

(base) [hadoop@master scala-2.11]$ ls
JentekLLC-spark-streaming-from-Twitter-assembly-1.jar  update

Then copy the jar file into $SPARK_HOME/jars/, which is the default classpath of Java library

cp JentekLLC-spark-streaming-from-Twitter-assembly-1.jar $SPARK_HOME/jars/

Then verify the file JentekLLC-spark-streaming-from-Twitter-assembly-1.jar in $SPARK_HOME/jars/

ls $SPARK_HOME/jars/Jentek*
/opt/spark/jars/JentekLLC-spark-streaming-from-Twitter-assembly-1.jar

Since we are going to load Twitter Tweets into Hive table, set the permission of HDFS /tmp/hive by:

hdfs dfs -chmod -R 777 /tmp/hive
hdfs dfs -ls /tmp/
drwx------   - hadoop supergroup          0 2020-05-09 14:16 /tmp/hadoop-yarn
drwxrwxrwx   - hadoop supergroup          0 2018-05-05 22:00 /tmp/hive

Assuming you have already started up Hadoop, Hive, Spark, and Jupyter-Notebook server in the virtualenv spark by conda activate spark. If not, visit the relevant session of this eBook if needing direction. Then launch your web browser, connect to Jupyter notebook server, start a new notebook in Scala Spylon kernel and run below code

https://github.com/geyungjen/jentekllc/blob/master/Spark/Scala/STREAMING/twitter_streaming_project_code.ipynb

You need to replace with your own Twitter consumer key, consumer secret, access token, access token secret. Additionally, it is likely you might need to replace the IP address of the machine that runs HIVE.

Note:

The notebook runs fines with Jupyter-notebook (server) runs on Linux, but however, on Mac, when creating a table on HIVE, following error will result, despite hive support has been enabled.

val spark = SparkSession.builder().config("spark.master", "local[2]").appName("interfacing spark sql to hive metastore through thrift url below").config("hive.metastore.uris", "thrift://localhost:9083").enableHiveSupport().getOrCreate()

But when running below create table statement

spark.sql("CREATE TABLE IF NOT EXISTS tweets (datetime STRING, text STRING) USING hive")

org.apache.spark.sql.AnalysisException: Hive support is required to CREATE Hive TABLE (AS SELECT);;
'CreateTable `tweets`, Ignore
  at org.apache.spark.sql.execution.datasources.HiveOnlyCheck$.$anonfun$apply$4(rules.scala:446)
  at org.apache.spark.sql.execution.datasources.HiveOnlyCheck$.$anonfun$apply$4$adapted(rules.scala:444)
  at org.apache.spark.sql.catalyst.trees.TreeNode.foreach(TreeNode.scala:144)
  at org.apache.spark.sql.execution.datasources.HiveOnlyCheck$.apply(rules.scala:444)
  at org.apache.spark.sql.execution.datasources.HiveOnlyCheck$.apply(rules.scala:442)
  at org.apache.spark.sql.catalyst.analysis.CheckAnalysis.$anonfun$checkAnalysis$38(CheckAnalysis.scala:574)
  at org.apache.spark.sql.catalyst.analysis.CheckAnalysis.$anonfun$checkAnalysis$38$adapted(CheckAnalysis.scala:574)

This is likely related to Jupyter notebook Spylon Kernel on MacOS only, because the same code runs fines on spark-shell on the same Mac without problem

scala> val spark = SparkSession.builder().config("spark.master", "local[2]").appName("interfacing spark sql to hive metastore through thrift url below").config("hive.metastore.uris", "thrift://localhost:9083").enableHiveSupport().getOrCreate()
spark: org.apache.spark.sql.SparkSession = org.apache.spark.sql.SparkSession@7273bb91

scala> import spark.implicits._
import spark.implicits._

scala> spark.sql("CREATE TABLE IF NOT EXISTS tweets (datetime STRING, text STRING) USING hive")
res19: org.apache.spark.sql.DataFrame = []

If you need direction, view my video presentation below in its entirety:

https://www.youtube.com/watch?v=XBfL6Jx4U2I&feature=youtu.be