Sha256: d7e67fae60951d530afd2851922747432eb54e9bad0c8bfa9b35fbe8fb3e9af9
Contents?: true
Size: 727 Bytes
Versions: 6
Compression:
Stored size: 727 Bytes
Contents
#!/usr/bin/env ruby
require_relative './common'
# cat data/swk-100.tsv | ./histograms.rb --map | sort > data/swk-hist-map.tsv
# cat data/swk-hist-map.tsv | ./histograms.rb --reduce > data/swk-hist.tsv
class HistogramsMapper < Wukong::Streamer::ModelStreamer
self.model_klass = Logline
def process visit
yield [visit.path, visit.day_hr]
end
end
class HistogramsReducer < Wukong::Streamer::Reducer
def get_key(path, day_hr)
[path, day_hr]
end
def start!(*args)
@count = 0
super
end
def accumulate(path, day_hr)
@count += 1
end
def finalize
yield [key, @count]
end
end
# Wukong.run( HistogramsMapper )
Wukong.run( HistogramsMapper, HistogramsReducer, :sort_fields => 3 )
Version data entries
6 entries across 6 versions & 2 rubygems