classify_wfids.rb

#!/usr/bin/env ruby

# Classifies a list of wfids into probably what workflow they come from,
# just based on the name.

require 'pp'
require 'set'

counters = Hash.new(0)

flows_per_kase = Hash.new(0)
wackies = Set.new
old_case_dates = Set.new
old_jp_dates = Set.new

STDIN.each_line do |line|
  case line
  when /^gr-.*-([[:xdigit:]]*$)/ then
    counters[:pcp_workflows] += 1
    counters[:kase_count] += flows_per_kase.has_key?($1) ? 0 : 1
    flows_per_kase[$1] += 1
  when /^[[:xdigit:]]+(-[[:xdigit:]]+){4,4}/ then
    counters[:ingestion_workflows] += 1
  when /^[[:xdigit:]]*$/ then
    counters[:multipart_workflows] += 1
  when /^([[:digit:]]{8,8})-[[:digit:]]{4,4}+-[a-z]+-[[:xdigit:]]+$/ then
    counters[:old_cases] += 1
    old_case_dates << $1.to_i
  when /^([[:digit:]]{8,8})-[[:digit:]]{4,4}+-[a-z]+-[a-z]+$/ then
    counters[:old_ruote_picked] += 1
    old_jp_dates << $1.to_i
  else
    wackies << line
  end
end

PP.pp counters
puts
puts "Wackys:"
PP.pp wackies