#!/usr/bin/env ruby
require 'optparse'
require 'rubygems'
require 'sip'

starttime = Time.now

options = { :db => nil, :table => nil, :config => 'sip.yml', :debug => false, :tfile => nil, :overwrite => false }
parser = OptionParser.new { |opts|
  opts.banner = "Usage: sip [options]"

  opts.on('-h', '--help', 'displays usage information') { 
    puts opts
    exit
  }

  opts.on('-o', '--overwrite', 'force full overwrite') {
    options[:overwrite] = true
  }

  opts.on('-d', '--debug', 'Print debugging information') { |d|
    options[:debug] = true
  }

  opts.on('-D <database>', '--database <database>', 'only import from given database') { |db|
    options[:db] = db
  }

  opts.on('-c <config file>', '--config <config file>', 'use specified config') { |c|
    options[:config] = c
  }

  opts.on('-t <table>', '--table <table>', 'only import from given table (must specify -D as well)') { |t|
    options[:table] = t
  }

  opts.on('-f <file>', '--transpart-file <file>', 'translation/partition file location (default: looks for file in path)') { |t|
    options[:tfile] = t
  }
}.parse!


if ENV['HADOOP_HOME'].nil? or ENV['HIVE_HOME'].nil?
   puts "You must first set both the HADOOP_HOME and HIVE_HOME environment variables."
   exit 1
end

debug = Proc.new { |t| puts "#{Sip::Utils::hostname} #{Time.now.strftime("%Y-%m-%d %H:%M:%S")}: #{t}" if options[:debug] }

debug.call "Using config file #{options[:config]}"
config = Sip::Config.load_file options[:config]
config.set_temp options, [:debug, :overwrite]

config['tfile'] = Sip::Utils::tfile_path(options[:tfile] || config['tfile'])
config['hdfs_tmpdir'] ||= "/tmp/sip"

if config['tfile'].nil?
  puts "Could not find transforation / partition (transpart) script.  Please use -f option to specify a location."
  exit 1
end
debug.call "Using transformation / partitioning script '#{config['tfile']}'"

sipper = Sip::Sipper.new config
config['databases'].each { |dbname, dbconfig|
  next if not options[:db].nil? and options[:db] != dbname
  dbconfig['tables'].each { |tablename, tableconfig|
    next if not options[:db].nil? and not options[:table].nil? and options[:table] != tablename
    debug.call "creating script for table #{tablename} in database #{dbname}"
    script_count = sipper.create_scripts dbconfig, tableconfig
    if script_count > 0
      debug.call "Executing #{script_count} scripts..."
      sipper.run_scripts    
      debug.call "Running hive import..."
      sipper.run_hive_import tableconfig
    end
  }
}

# Save new version of config
debug.call "Writing new version of config file to #{options[:config]}"
sipper.config.save_file options[:config]

debug.call "Finished running import in #{Time.now - starttime} seconds"