Orbit/lib/ntu_archive/archive_ntu_dd.rb

# encoding : utf-8

# Usage ===============
#
# ARGV[0] = folder path of xml files
# ARGV[1] = folder path of files
#
#======================


require 'mongo'
require 'nokogiri'
include Mongo

CSV_PATH = ARGV[0]
FILE_PATH = ARGV[1]
FIELDS = %w[title file createdate modifydate createuser modifyuser]
DB_BASE_NAME = "production"

TABLE = {
  "0" => ["gaTaco", "ntuga"],
  "1" => ["fdLai", "fdhome"],
  "5" => ["prLin", "property01"],
  "3" => ["gsTien", "ckh", "general01", "lynn"],
  "10" => ["cmChen", "construction01", "allen", "enAmeliaxu", "energy"],
#  "2" => ["caLin", "cashier01"],
  "6" => ["pcAmyok", "purchasing01", "purchasing02"],
  "7" => ["fmHsyu", "ntufss"],
  "4" => ["soChang", "social01"],
  "8" => ["meJune", "medicine01"],
  "9" => ["plKoa", "police"]
}

which_site = {}

data = []

Dir.foreach(ARGV[0]) do |file|
  if file =~ /\w\.\w/
    doc = Nokogiri::XML(File.open(File.join(ARGV[0], file)))
    doc.xpath("//row").each { |row| 
      xml_fields = row.children.map{ |row| [row.name, row.content] }
      fields = FIELDS.inject([]) do |fields, field|
        xml_fields.each do |xf|
          if xf[0].include?(field)
            fields << xf[1] 
            break
          end
        end
        fields
      end
      data << fields + [File.basename(file, '.xml')]
    }
    data.each do |d|
      title, file, c_time, m_time, creator, modifier, category = d
      if file =~ /\w\.\w/
        file = File.join(File.join(ARGV[1], file))
        site = ""
        TABLE.each do |key, value|
          if value.include? modifier
            site = key
            break
          elsif value.include? creator
            site = key
            break
          end
        end
        site = "0" if site.empty?
        if which_site[site]
          if which_site[site][category]
            which_site[site][category] += [[title, file, c_time, m_time]]
          else
            which_site[site][category] = [[title, file, c_time, m_time]]
          end
        else
          which_site[site] = {category => [[title, file, c_time, m_time]]}
        end
      end
    end
  end
end

def save_category archive_file_category, in_category
  category = {
    title: {"zh_tw" => in_category, "en" => in_category},
    key: "import_#{category}"
  }
  archive_file_category.save(category)
end

def save_file_multiples archive_file_multiples, grid ,archive_id,title, file, c_time, m_time
  a_file = File.open(file) rescue nil
  if a_file
    filename = File.basename(file)
    archive_file = {
      file_title: {"zh_tw" => title, "en" => title },
      choose_lang: ["zh_tw", "en"],
      file: filename,
      archive_file_id: archive_id,
      created_at: c_time,
      updated_at: m_time
    }
    a_file_id = archive_file_multiples.save(archive_file)
    grid.put(a_file, filename: "assets/archive_file_multiple/file/#{a_file_id}/#{filename}")
  end
end

def save_archive archive_files, category_id
  archive = {
    title: {"zh_tw" => " ", "en" => " "},
    archive_file_category_id: category_id,
    is_top: false,
    is_hot: false,
    is_hidden: false,
    created_at: c_time,
    updated_at: m_time
  }
  archive_files.save(archive)
end

which_site.each do |site, categories|

  db = MongoClient.new("localhost", 27017).db("#{DB_BASE_NAME}_#{site}")

  categories.each do |category, data|
    archive_file_category = db.collection('archive_file_categories')
    category_id = save_category archive_file_category, category

    archive_files = db.collection('archive_files')
    archive_id = save_archive archive_files, category_id

    grid = Grid.new(db)
    archive_file_multiples = db.collection('archive_file_multiples')
    file_ids = []

    data.each do |file|
      save_file_multiples(archive_file_multiples, grid ,archive_id , *file)
    end
  end

end