nokoxpath.rb

#!/usr/bin/env ruby -w

require "rubygems"
require "nokogiri"
require "open-uri"

PROGRAM_BASE = File.basename $PROGRAM_NAME

def usage
  puts <<-USAGE
Usage: #{PROGRAM_BASE} <query> [<file>|<URL>]...
Search <query> in each HTML or XML document read from <file>, <URL> or stdin.
USAGE
  exit 64
end

def search(io, query)
  Nokogiri.parse(io.read) do |doc|
    doc.remove_namespaces!
    return doc.search(query)
  end
end

usage if ARGV.empty?
usage if ARGV.first =~ /-h|--?help/
query = ARGV.shift
puts search($stdin, query) if ARGV.empty?
ARGV.each do |path|
  open(path) do |io|
    puts search(io, query)
  end
end