#!/usr/bin/env ruby

require 'rbbt/workflow'

require 'rbbt-util'
require 'fileutils'
require 'rbbt/util/simpleopt'
require 'rbbt/workflow/step'
require 'rbbt/util/misc'

require 'rbbt-util'
require 'rbbt/util/simpleopt'

$0 = "rbbt #{$previous_commands*""} #{ File.basename(__FILE__) }" if $previous_commands

options = SOPT.setup <<EOF
Examine the provenance of a job result

$ rbbt workflow prov <job-result>

-h--help Help
-p--plot* draw the dependency plot into <file.png>
EOF

SOPT.usage if options[:help]

file = ARGV.shift

$seen = []
def get_step(file)
  file = file.sub(/\.(info|files)/,'')
  $seen << file
  Step.new file
end

def status_msg(status)
  color = case status
          when :error, :aborted, :missing
            :red
          when :streaming
            :yellow
          when :done
            :green
          end
  Log.color(color, status)
end

def report_msg(status, name, path)

  parts = path.sub(/\{.*/,'').sub(/#{Regexp.quote(name)}$/,'').split "/"

  task = Log.color(:yellow, parts.pop)
  workflow = Log.color(:magenta, parts.pop)

  if not Open.remote?(path) and $main_mtime and ($main_mtime - File.mtime(path)) < 0
    status_msg(status) << " " << [workflow, task, path] * " " << " (#{Log.color(:red, "Mtime out of sync") })\n"
  else
    status_msg(status) << " " << [workflow, task, path] * " " << "\n"
  end
end

def report(step, offset = 0)
  info = step.info || {}
  path  = step.path
  status = info[:status] || :missing
  status = "remote" if Open.remote?(path)
  name = info[:name] || File.basename(path)
  status = :unsync if status == :done and not File.exist? path
  str = " " * offset
  str << report_msg(status, name, path)
  info[:dependencies].each do |task,name,path|
    new = ! $seen.include?(path)
    dep = get_step path
    if new
      str << report(dep, offset + 1)
    else
      str << Log.color(:blue, Log.uncolor(report(dep, offset+1)))
    end
  end if info[:dependencies]
  str
end

step = get_step file
$main_mtime = File.exist?(step.path) ? File.mtime(step.path) : nil

def adjacency(step)

  info = step.info || {}
  path  = step.path
  status = info[:status] || :missing
  status = "remote" if Open.remote?(path)
  if status == 'remote'
    workflow, task, name = path.sub(/\{.*/,'').split("/")[-3..-1]
  else
    workflow, task, name = path.split("/")[-3..-1]
  end
  name = name.split(":").last
  status = :unsync if status == :done and not File.exist? path
  edge_info = {:status =>  status, :workflow => workflow, :task => task, :name => name, :label => task, :shape => shapes[workflow], :color => status == 'remote' ? 'blue' : 'green'}
  id = Misc.digest(path)
  edges = []
  node_info = {}
  node_info[id] = edge_info
  if info[:dependencies]
    info[:dependencies].each do |task,name,path|
      dep = get_step path
      _id, _edges, _node_info = adjacency(dep)
      edges << [id, _id] 
      edges.concat _edges
      node_info.merge!(_node_info)
    end 
  end

  [id, edges, node_info]
end

if options[:plot]
  id, edges, node_info = adjacency(step)
  node_info[id][:color] = 'red'
  TmpFile.with_file do |edge_file|
    Open.write(edge_file) do |f|
      f.puts "from,to"
      edges.each do |from,to|
        f.puts [from,to]*","
      end
    end
    TmpFile.with_file do |node_info_file|
      Open.write(node_info_file) do |f|
        fields = node_info.first.last.keys
        f.puts "id," + fields * ","
        node_info.each do |id,info|
          f.puts ([id] + info.values_at(*fields)) * ","
        end
      end

      require 'rbbt/util/R'

      ppp Open.read edge_file
      ppp Open.read node_info_file

      R.run <<-EOF
      nodes <- read.csv("#{node_info_file}", header=T, as.is=T)
      links <- read.csv("#{edge_file}", header=T, as.is=T)

      library(igraph)

      net <- graph.data.frame(links, nodes, directed=T)
      net <- simplify(net, remove.multiple = F, remove.loops = T)

      png("#{options[:plot]}")
      plot(net, edge.arrow.size=0.4, vertex.label=net$label, vertex.color=net$color)
      dev.off()
      EOF
    end
  end

else
  puts report(step).strip
end