Popularity
1.5
Declining
Activity
0.0
Stable
10
3
3

Programming language: Scala
License: Apache License 2.0
Tags: Big Data    
Latest version: v1.0

Shadoop alternatives and similar packages

Based on the "Big Data" category.
Alternatively, view shadoop alternatives based on common mentions on social networks and blogs.

Do you think we are missing an alternative of Shadoop or a related project?

Add another 'Big Data' Package

README

Shadoop

A Hadoop DSL and lightweight wrapper for Scala

Build Status

This fork of ScalaHadop is mostly just cherry-picked commits from the forks by @hito-asa, @ivmaykov and @oscarrenalis, of the original work by @bsdfish. In addition there are a few extra features and a cleaned up Maven build.

This code provides some syntactic sugar on top of Hadoop in order to make it more usable from Scala. Take a look at src/main/scala/net/renalias/scoop/examples/WordCount.scala for more details.

License

Apache License, Version 2.0

Usage

Basic Usage

A basic mapper looks like:

val mapper = new Mapper[LongWritable, Text, Text, LongWritable] {
    mapWith {
        (k, v) =>
            (v split " |\t").map(x => (new Text(x), new LongWritable(1L))).toList
    }
}

a reducer looks like this:

val reducer = new Reducer[Text, LongWritable, Text, LongWritable] {
    reduceWith {
        (k, v) =>
            List((k, (0L /: v)((total, next) => total + next)))
    }
}

and, the pipeline to bind them together may look like this:

TextInput[LongWritable, Text]("/tmp/input.txt") -->
MapReduceTask(mapper, reducer, "Word Count")    -->
TextOutput[Text, LongWritable]("/tmp/output")   execute

The key difference here between standard mappers and reducers is that the map and reduce parts are written as side-effect free functions that accept a key and a value, and return an iterable; code behind the scenes will take care of updating Hadoop's Context object.

Some note still remains to be done to polish the current interface, to remove things like .toList from the mapper and the creation of Hadoop's specific Text and LongWritable objects.

Note that implicit conversion is used to convert between LongWritable and longs, as well as Text and Strings. The types of the input and output parameters only need to be stated as the generic specializers of the class it extends.

These mappers and reducers can be chained together with the --> operator:

object WordCount extends ScalaHadoop {
  def run(args: Array[String]) : Int = {
    TextInput[LongWritable, Text](args(0)) -->
    MapReduceTask(mapper, reducer, "Main task") -->
    TextOutput[Text, LongWritable](args(1)) execute

    0 //result code
  }
}

Multiple map/reduce

Multiple map/reduce runs may be chained together:

object WordsWithSameCount extends ScalaHadoop {
  def run(args: Array[String]) : Int = {
    TextInput[LongWritable, Text](args(0)) -->
    MapReduceTask(tokenizerMap1, sumReducer, "Sum") -->
    MapReduceTask(flipKeyValueMap, wordListReducer, "Reduce") -->
    TextOutput[LongWritable, Text](args(1)) execute

    0 //result code
  }
}

Contributors


*Note that all licence references and agreements mentioned in the Shadoop README section above are relevant to that project's source code only.