require "stringio"
require "fileutils"
require "pathname"

$dryrun = ARGV.include?("--dryrun") 
puts "running in dry run mode" if $dryrun


outputDirIndex = ARGV.index("-o")
$output_dir = outputDirIndex ? ARGV[outputDirIndex+1] || "target" : "target"



$dblist = {
  "hg17" => ["refGene"],
  "hg18" => ["refGene"],
  "mm9" => ["refGene"],
  "rn4" => ["refGene"],
  "dm3" => ["refGene"],
  "sacCer1" => ["sgdGene", "all_est"],
  "oryLat1" => ["all_est", "ensGene"],
}

revisionNameIndex = ARGV.index("-r")
if revisionNameIndex
  revisionName = ARGV[revisionNameIndex+1] || "hg17"
  $dblist.clear
  $dblist.store(revisionName, [])
#  $dblist.delete_if { |k, v| k != revisionName }
end

dbNameIndex = ARGV.index("-d")
if dbNameIndex
  dbName = ARGV[dbNameIndex+1] || "refGene"
  $dblist.each_key do |revision|
    $dblist[revision].clear
    $dblist[revision] << dbName
  end
end


def eval(command)
  result = StringIO.new("", "w")
  IO.popen(command) do |p|
    p.each_line { |l| result.puts(l) }
  end
  return result.string
end

def run(command)
  puts command
  if !$dryrun
    IO.popen(command) do |p|
      p.each_line { |l| puts(l) }
    end
  end
end


def createDir(dir)
   puts "create a directory: #{dir}" if !(File.directory?(dir))
   if !$dryrun
     FileUtils.mkdir_p(dir)  
   end
end


# download data
$dblist.each_key do |revision|
  $dblist[revision].each do |dbName|
    puts "retrieving #{revision} #{dbName}"
    out = "#{$output_dir}/#{revision}/#{dbName}"
    archive = "#{dbName}.txt.gz"
    text = "#{dbName}.txt"
    sql = "#{dbName}.sql"
    sqlitedb = "#{dbName}.db"
    
    createDir(out)
    run("curl -o#{out}/#{archive} http://hgdownload.cse.ucsc.edu/goldenPath/#{revision}/database/#{archive}") if ! File.exist?("#{out}/#{archive}") 
    
    run("curl -o#{out}/#{sql}.orig http://hgdownload.cse.ucsc.edu/goldenPath/#{revision}/database/#{sql}") if ! File.exist?("#{out}/#{sql}.orig")
    run("cat #{out}/#{sql}.orig | perl mysql2sqlite.pl > #{out}/#{sql}") if ! File.exist?("#{out}/#{sql}")

    run("gunzip -c #{out}/#{archive} > #{out}/#{text}") if ! File.exist?("#{out}/#{text}")
    if !File.exist?("#{out}/#{sqlitedb}")
      run("sqlite3 #{out}/#{sqlitedb} < #{out}/#{sql}") 
      run("sqlite3 -separator \"	\" #{out}/#{sqlitedb} \".import #{out}/#{text} entry\"")
    end
  end
end