package Hadoop::Runner;
use Moose;
 
has 'mapper' => (does => 'Hadoop::Mapper', required => 1);
has 'reducer' => (does => 'Hadoop::Reducer');
has 'combiner' => (does => 'Hadoop::Reducer');
has 'partitioner' => (does => 'Hadoop::Partitioner');
 
has 'input_format' => (isa => 'InputFormat', default => 'TextInputFormat');
has 'output_format' => (isa => 'OutputFormat', default => 'TextOutputFormat');
has 'output_compression' => (isa => 'Bool', default => 0);
 
has 'hadoop_prefix' => (isa => 'Str', default => '/SGE/HADOOP/active');
has 'keep_env' => (isa => 'ArrayRef[Str]', default => sub { ["PERLLIB", "PERL5LIB"] });
 
sub run();

• mapper – a Hadoop::Mapper to use
• reducer – an optional Hadoop::Reducer to use
• combiner – an optional Hadoop::Reducer to use as combiner
• partitioner – an optional Hadoop::Partitioner to use
• input_format – one of TextInputFormat, KeyValueTextInputFormat, SequenceFileInputFormat
• output_format – one of TextOutputFormat, SequenceFileOutputFormat
• output_compression – Bool flag controlling the compression of output
• hadoop_prefix – the prefix of Hadoop instalation. Default value is fine in UFAL cluster.
• keep_env – which environment variables are preserved when running perl mappers, reducers, combiners and partitioners

script.pl run [-jt/--jobtracker jobtracker | -c/--cluster machines [-w/--wait secs]] [-r/--reducers reducers] [generic Hadoop options] input_path output_path
script.pl map number_of_reducers
script.pl reduce
script.pl combine

package Hadoop::Mapper;
use Moose::Role;
 
requires 'map';
 
sub setup() {}
sub cleanup {}

• sub map($self, $key, $value, $context) – executed for every (key, value) input pair. The variable '$content' has following methods:
  • $content->write($key, $value) – output the ($key, $value) pair
  • $content->counter($group, $name, $increment) – increases the counter $name in the group $group by $increment
• sub setup($self, $context) – executed once before any input (key, value) pairs are processed. The $context can be used to both produce (key, value) pairs and increment counters.
• sub cleanup($self, $context) – executed once after all input (key, value) pairs are processed. The $context can be used to both produce (key, value) pairs and increment counters.

package Hadoop::Reduce;
use Moose::Role;
 
requires 'reduce';
 
sub setup() {}
sub cleanup {}

• sub reduce($self, $key, $values, $context) – executed for every $key. The $values is an iterator with the following methods:
  • $values->value() – returns the current value, undef if there is any.
  • $values->next() – advance to next value. Returns true if there is any, false otherwise.
  • At the beginning there is no current value, the first value should be obtained by calling 'next'.
• sub reduce($self, $key, $values, $context) – the variable $content has following methods:
  • $content->write($key, $value) – output the ($key, $value) pair
  • $content->counter($group, $name, $increment) – increases the counter $name in the group $group by $increment
• sub setup($self, $context) – executed once before any input keys are processed. The $context can be used to both produce (key, value) pairs and increment counters.
• sub cleanup($self, $context) – executed once after all input keys are processed. The $context can be used to both produce (key, value) pairs and increment counters.

package Hadoop::Partitioner;
use Moose::Role;
 
requires 'getPartition';
 
sub setup {}
sub cleanup {}

• sub getPartition($self, $key, $value, $partitions) – executed for every output (key, value) pair. It must return a number of partition in range 0..$partitions-1, where the output (key, value) pair should be placed.
• sub setup($self) – executed once before any input (key, value) pairs are processed.
• sub cleanup($self) – executed once after all input (key, value) pairs are processed.

• HADOOP_TASK_ID – available in every mapper and reducer. The serial number of the mapper and reducer task (in range 0..number_of_tasks-1).
• HADOOP_WORK_OUTPUT_PATH – available in every reducer, and also in every mapper of a reduce-less job. It contains an existing directory where the reducer can output files. If the reducer finishes successfully, all files and subdirectories will be moved to output directory of the job.