Hadoop Streaming with Node.js (CoffeeScript edition)

map.coffee

#!/usr/bin/env coffee

stdin  = process.openStdin()
stdout = process.stdout
input  = ''

stdin.setEncoding 'utf8'

stdin.on 'data', (data) ->
  return unless data?
  input += data;
  while input.match /\r?\n/
    input = RegExp.rightContext
    proc RegExp.leftContext

stdin.on 'end', ->
  proc input if input?

proc = (line) ->
  words = line.split ' '
  stdout.write "#{words[8]},1\n"

reduce.coffee

#!/usr/bin/env coffee

stdin   = process.openStdin()
stdout  = process.stdout
counter = {}
input   = ''

stdin.setEncoding 'utf8'
stdin.setEncoding 'utf8'

stdin.on 'data', (data) ->
  return unless data?
  input += data
  while input.match /\r?\n/
    input = RegExp.rightContext
    proc RegExp.leftContext

stdin.on 'end', ->
  proc input if input?
  stdout.write "#{k}:#{v}\n" for k, v of counter

proc = (line) ->
  words = line.split ','
  word  = words[0]
  # I'm gonna assume base10 radix here?
  count = parseInt words[1], 10
  if not counter[word] then counter[word] = 1
  else counter[word] += count

My first pass at translating an existing JavaScript gist to CoffeeScript. This could also help me learn about Hadoop. :)

Also, using the leftContext and rightContext properties of the RegExp object was a snazzy (though, I thought obsoleted?) feature of the original code. Hmm...