nokogiri with pipes

nokoe

#!/usr/bin/env ruby

#
# Nokoe: Nokogiri with pipes
#
#   Requirements:
#
# gem install trollop
# gem install nokogiri
#

require 'rubygems'
require 'trollop'
require 'nokogiri'
require 'open-uri'

opts= Trollop::options do
  version "#{$0} - 1.0 (c) aurelian 2011-2012"
  banner <<-EOS
  #{$0} - Nokogiri (http://nokogiri.org/) with pipes.

Usage: #{$0} <file> [options]

  $ cat file.html | #{$0} --selector "//body" > foo.html
  $ #{$0} --selector "nav#siteLinks" < file.html > foo.html
  $ #{$0} --file "http://google.com" --selector "//body/a/*"
  $ #{$0} --file "http://twitter.com" --irb # like `nokogiri http://twitter.com`
  $ #{$0} http://twitter.com --irb # same as above, without using --file.

Options:
EOS
  opt :file, "filename/url to read in", :type => String
  opt :selector, "Nokogiri selector", :default => "/", :type => String
  opt :irb, "drops you to IRB, @doc holds the document", :default => false
  opt :encoding, "encoding", :default => "utf-8", :type => String
end

if $stdin.stat.size > 0
  text= $stdin.read
  $stdin.reopen(File.open("/dev/tty", "r")) if opts[:irb]
elsif file= opts[:file] || ARGV.shift
  begin
    text= open(file).read 
  rescue StandardError => error
    Trollop::die(:file, "#{file} must _exist_ dammit!\n\t#{error.message}")
  end
else
  Trollop::die :file, "cannot read incoming stream" # be creative.
end

@doc= Nokogiri.parse(text, nil, opts[:encoding]).search(opts[:selector])

if opts[:irb]
  require 'irb'
  puts "-- starting irb session. @doc is your document."
  IRB.start
else
  $stdout << @doc
end