#!/usr/bin/env ruby

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

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

options = SOPT.setup <<EOF

Optimize the files in a directory. 

$ rbbt system optimize [options] <directory>

Compresses the chromosome files with BGZ and optimizes TokyoCabinet databases. Some files
do not compress correctly with BGZ and are left un-compressed.

-h--help Print this help
-k--keep Keep original files

EOF
if options[:help]
  if defined? rbbt_usage
    rbbt_usage 
  else
    puts SOPT.usage
  end
  exit 0
end

directory = ARGV.shift
raise ParameterException, "No directory given" if directory.nil?

raise ParameterException, "Not a valid directory" unless File.directory? directory

directory = Path.setup directory.dup
keep = options[:keep]

require 'rbbt/util/misc/bgzf'

Log.info Log.color(:magenta, "Processing chromosomes")
chromosome_files = directory.glob('**/chromosome_*')
TSV.traverse chromosome_files, :type => :array, :bar => "Processing chromosomes" do |chr_file|
  next if chr_file =~ /.bgz$/
  compressed_file = chr_file + '.bgz'
  Log.info "Compressing #{ chr_file } into #{ compressed_file }"
  CMD.cmd("#{Bgzf.bgzip_cmd} '#{ chr_file }' -c > #{ compressed_file }")
  begin
    Open.read(compressed_file)
    if keep
      Log.info "File #{ chr_file } was correctly compressed. Keeping original"
    else
      Log.info "File #{ chr_file } was correctly compressed. Removing original"
      FileUtils.rm chr_file
    end
  rescue
    Log.warn "File #{ chr_file } was not correctly compressed. Removing compressed version and leaving original"
    FileUtils.rm compressed_file
  end
end

Log.info Log.color(:magenta, "Processing Tokyo Cabinet files")

all_files = directory.glob('**/*') + directory.glob('**/.*')
all_files.uniq!
TSV.traverse all_files, :type => :array, :bar => "Processing Tokyo Cabinet files" do |file|
  cmd = 'file "' << file << '"'
  file_type = CMD.cmd(cmd).read.strip.partition(/:\s+/).last
  next unless file_type =~ /Tokyo/
  type = file_type.split(", ")[1]
  case type
  when "Hash"
    cmd = 'tchmgr optimize "' << file << '"'
  when "B+ tree"
    cmd = 'tcbmgr optimize "' << file << '"'
  else
    next
  end
  size = File.size(file)
  cmd << ' -td -tl' if size > 2_000_000_000 or file_type =~ /deflate/
  Log.info "Optimizing #{type} database #{file}"
  io = CMD.cmd(cmd)
  while line = io.gets
    Log.debug line
  end
end