determine-content-type.rb

require 'pathname'

def main(middlewares, files)
  files.each do |filename|
    pathname = Pathname.new(filename).expand_path
    raise "No such file: #{filename}" unless pathname.file?
    content_type = middlewares.inject(nil) {|ct, mw|
      mw.call(pathname) || ct
    }
    $stderr.puts [pathname.to_path, content_type].join(': ')
  end
end

require 'rack/mime'
class Extension
  def initialize(map={})
    @mime_types = Rack::Mime::MIME_TYPES.merge(map)
  end

  def call(pathname)
    @mime_types.fetch(pathname.extname, 'application/octet-stream')
  end
end

require 'epub/parser'
EPUB::OCF::PhysicalContainer.adapter = :File
class EPUBContent
  def initialize(dirs)
    @media_types = dirs.each_with_object({}) {|dir, map|
      dir = Pathname.new(dir).expand_path
      raise "No such directory: #{dir}" unless dir.directory?
      epub = EPUB::Parser.parse(dir)
      epub_file = Pathname.new(epub.epub_file).expand_path
      Pathname.glob("#{epub_file}/META-INF/*.xml").each do |path|
        map[path] = 'application/xml'
      end
      epub.ocf.container.rootfiles.each do |rootfile|
        map[epub_file.join(rootfile.full_path)] = rootfile.media_type
      end
      epub.resources.each do |resource|
        map[epub_file.join(resource.entry_name)] = resource.media_type
      end
    }
  end

  def call(pathname)
    @media_types[pathname]
  end
end

epub_base_dir = "#{ENV['HOME']}/ruby/projects/epub-parser/test"
epub_dirs = Pathname.glob("#{epub_base_dir}/**/META-INF/container.xml").map {|path| path.dirname.dirname}
determiners = [
  Extension.new({'.epub' => 'application/epub+zip'}),
  EPUBContent.new(epub_dirs)
]

main(determiners, ARGV)