#!/usr/bin/ruby -w

##############################################################
# GLOBAL CONSTANTS

DEFAULT_DATABASE_PATH = "/project/marcotte/marcotte/ms/database"
DEFAULT_MZXML_PATH = "."
DEFAULT_OUTDIR = "pepxml"
DEFAULT_PARAMS_GLOB = "*.params"
DEFAULT_PARAMS_FILE = Dir[DEFAULT_PARAMS_GLOB].first
DEFAULT_PEPXML_VERSION = 18
DEFAULT_MS_MODEL = 'LCQ'
DEFAULT_MASS_ANALYZER = 'Ion Trap'
##############################################################

require 'spec_id'
require 'optparse'
require 'ostruct'
require 'fileutils'

# establish the default database path after examining env vars
def_dbpath = nil
db_env_var = ENV["BIOWORKS_DBPATH"]
if db_env_var
  def_dbpath = db_env_var
else
  def_dbpath = DEFAULT_DATABASE_PATH
end

opt = OpenStruct.new

opt_obj = OptionParser.new do |op|
  op.banner = "\nusage: #{File.basename(__FILE__)} [options] <file>.srf ...
usage: #{File.basename(__FILE__)} [options] bioworks.xml"
  op.on_head "
  Takes .srf files or the xml exported output of Bioworks multi-consensus view 
  (no filtering) and outputs pepXML files (to feed the trans-proteomic pipeline).

Options:"
  op.on('-h', '--help', "display this and more notes and exit") {|v| opt.help = v }
  op.on('-o', '--outdir path', "output directory     d: '#{DEFAULT_OUTDIR}'") {|v| opt.outdir = v }

  op.separator ""
  op.separator "bioworks.xml files may require additional options:"
  op.separator ""
  op.on('-p', '--params file', "sequest params file  d: '#{DEFAULT_PARAMS_FILE}'") {|v| opt.params = v }
  op.on('-d', '--dbpath path', "path to databases    d: '#{DEFAULT_DATABASE_PATH}'") {|v| opt.dbpath = v }
  op.on('-m', '--mspath path', "path to MS files     d: '#{DEFAULT_MZXML_PATH}'") {|v| opt.mspath = v }
  op.on('--model <LCQ|Orbi|string>', "MS model             d: '#{DEFAULT_MS_MODEL}'") {|v| opt.model = v }
  op.on('--mass_analyzer <string>',  "Mass Analyzer        d: '#{DEFAULT_MASS_ANALYZER}'") {|v| opt.mass_analyzer = v }
  op.on('-v', '--version pepxml_version', "pepxml version       d: '#{DEFAULT_PEPXML_VERSION}'") {|v| opt.pepxml_version = v.to_i }

end

more_notes = "
Notes:

  mspath: Directory to RAW or mzXML (version 1) files.
          This option is not used with Bioworks 3.3 files.
  outdir: Path will be created if it does not already exist.
  model : LCQ -> 'LCQ Deca XP Plus'
        : Orbi -> 'LTQ Orbitrap'
        : other string -> That's the string that will be used.

  options with spaces should be quoted: e.g., \"Time of Flight\"

Database Path:

  If the database path in the sequest.params file is valid, that will be used.
  Otherwise, will try (in order):
      1. --dbpath or -d option
      1. environmental variable BIOWORKS_DBPATH (currently: '#{db_env_var}')
      2. constant at top of this script         (currently: '#{DEFAULT_DATABASE_PATH}')
  "



opt_obj.parse!

# intercept before argv count
if opt.help
  puts opt_obj
  puts more_notes
  exit
end

if ARGV.size < 1
  puts opt_obj 
  exit
end



opt.outdir ||= DEFAULT_OUTDIR

## Create dbpath if does not exist
if opt.outdir
  FileUtils.mkpath(opt.outdir) unless File.exist? opt.outdir
end

files = ARGV.to_a

if files[0] =~ /\.srf/i
  opt.dbpath ||= def_dbpath
  files.each do |file|
    hash = {
      :backup_db_path => opt.dbpath || def_dbpath,
      :out_path => opt.outdir,
    }
    xml_obj = SpecID::Sequest::PepXML.new_from_srf(file, hash)
    xml_obj.to_pepxml(xml_obj.base_name + ".xml")
  end
else
  ## Ensure params file exists (unless opt given)
  opt.params ||= DEFAULT_PARAMS_FILE
  params_obj = SpecID::Sequest::Params.new(opt.params)
  # Ensure the database exists!
  unless File.exist?( params_obj.database )
    if opt.dbpath
      params_obj.database_path = opt.dbpath 
    else
      params_obj.database_path = def_dbpath
    end
  end

  opt.mspath ||= DEFAULT_MZXML_PATH
  opt.pepxml_version ||= DEFAULT_PEPXML_VERSION
  opt.model ||= DEFAULT_MS_MODEL
  opt.mass_analyzer ||= DEFAULT_MASS_ANALYZER

  case opt.model
  when "LCQ"
    model = 'LCQ Deca XP Plus'
  when "Orbi"
    model = 'LTQ Orbitrap'
  else
    model = opt.model
  end


  bioworks = files[0]
  xml_objs = SpecID::Sequest::PepXML.set_from_bioworks(params_obj, bioworks, opt.mspath, opt.outdir, opt.pepxml_version, 'trypsin', 'ThermoFinnigan', model)

  xml_objs.each do |obj|
    obj.to_pepxml(obj.base_name + ".xml")
  end
end