No description, website, or topics provided.
Scala Python Shell Java
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Failed to load latest commit information.


KijiExpress provides a simple data analysis language using KijiSchema and Scalding, allowing you to analyze data stored in Kiji and other data stores.

Getting Started

There are a couple different ways to get started with KijiExpress. The easiest way is running a script using KijiExpress shell. You can also run KijiExpress on compiled jobs, or on arbitrary jar files.

You can run KijiExpress in either local mode or HDFS mode. Local mode uses Scalding's local mode and runs the job locally on your machine. In HDFS mode, the job runs on a cluster.

Running Compiled Jobs

If you want, you can run compiled jobs as well. You must have python3 and the python-base package installed. To install python3 on your machine we recommend using the pyenv project found at: In order to install the python base package use pip install python-base, if you're on a system whose 'python' is still python2, you may have to: $ sudo apt-get install python3-pip.

Once python is installed running a job requires that you write a Scalding Job and compile it yourself into a .jar file. Then you can run your job with the command: job path/to/your/jarfile.jar name.of.your.jobclass <any arguments to your job>

or job path/to/your/jarfile.jar name.of.your.jobclass <any arguments to your job> --hdfs

You can see some examples of Jobs in the kiji-express-music tutorial.

Running Arbitrary Jars

The express tool can also run arbitrary jars, with KijiExpress and its dependencies on the classpath, with the command: jar /path/to/your/jarfile.jar name.of.main.class <any arguments to your main class>

This requires you to have a Java or Scala main class.

Jars on the classpath are automatically added to the classpath of tasks run in MapReduce jobs. If you want to add jars to the classpath, you can set the EXPRESS_CLASSPATH variable to a colon-separated list of paths to your jars and they will be appended to the KijiExpress classpath.

To see the classpath that KijiExpress is running with, you can run: classpath

Interactive Shell

You can use the express tool to run a Scala REPL. You can use the shell to run and prototype queries. The shell can be run in local mode (such that all jobs run in Cascading's local runner) or in HDFS mode (such that all jobs run on a Hadoop cluster). Use shell shell --local

to use local mode, and shell --hdfs

to use HDFS mode.

Scalding flows created in the REPL must be explicitly run. For example, this REPL query gets the latest value from the column info:track_plays of a Kiji table and writes the results to a TSV file.

express> KijiInput("kiji://.env/default/users", "info:track_plays" -> 'playSlice)
res0: =
express> res0.mapTo('playSlice -> 'play) { slice: KijiSlice[String] => slice.getFirstValue() }
res1: cascading.pipe.Pipe = Each([MapFunction[decl:'play']]
express> res1.write(Tsv("songPlays"))
res2: cascading.pipe.Pipe = Each([MapFunction[decl:'play']]

More Examples

See the kiji-express-examples project and kiji-express-music tutorial for more detailed examples of KijiExpress computations.