keyvalue2csv

#!/usr/bin/env ruby
#
# Reads input given in key=value log format and outputs it as CSV.
# Specify the keys as arguments in the order that they should be
# output. Without arguments, it outputs all values.
#
# E.g.
# cat /tmp/production.log | keyvalue2csv time controller action
# 2014-01-14 13:02:04 +0100,products,index,195.87
# 2014-01-14 13:02:04 +0100,products,index,25.82
# 2014-01-14 13:02:04 +0100,products,index,23.38
#
# Or pipe it into psql:
# cat /tmp/production.log | keyvalue2csv time controller action |\
#   psql some_database -c "\copy logs (time, controller, action) FROM STDIN CSV"

require 'csv'

slicer = if ARGV.any?
           -> (h) { ARGV.map { |key| h.fetch(key) } }
         else
           -> (h) { h.values }
         end

csv = CSV.new(STDOUT)

KEY_RE = '[a-z_]+?='

STDIN.each_line do |line|
  pairs = line.scan(/
                     (?<=^|\s)         # must be preceeded by start or space
                     #{KEY_RE}.*?      # the key=value pair
                     (?=\s#{KEY_RE}|$) # must be followed by another pair or end
                    /x)
  next if pairs.empty?
  csv << slicer.call(Hash[*pairs.map { |pair| pair.split('=', 2) }.flatten])
end