root/lang/ruby/misc/hpricot-shell.rb @ 1081

#!ruby -Ku
=begin
hpricot 用コマンドライン要素ファインダ

hpricot の XPath/CSS3 実装は完全ではなく
また、独自実装している部分があるので注意が必要
=end


require "rubygems"
# gem install hpricot
require "hpricot"
# gem install term-ansicolor
require "term/ansicolor"
require "open-uri"
require "readline"

class XPathFinderCommand
        include Term::ANSIColor

        def self.run(argv)
                new.run(argv.dup)
        end

        def run(argv)
                Readline.completion_append_character = ""
                Readline.completion_case_fold = true
                Readline.completion_proc = method(:completion)
                @prompt = "> "
                @doc = nil
                @uri = nil

                if argv.first
                        cmd_get(argv.first)
                end


                begin
                        while l = Readline.readline(@prompt, true)
                                next if l.strip.empty?

                                cmd, arg = l.chomp.split(/\s+/, 2)
                                cmd = "cmd_#{cmd}"
                                if respond_to? cmd
                                        send(cmd, arg)
                                else
                                        # puts "#{cmd} command is not found."
                                        cmd_xpath(l.chomp)
                                end
                        end
                rescue Interrupt
                        puts
                        retry
                end
        end

        def completion(word)
                # hpricot の XPath 実装だとめんどい
#               case word
#               when /^(.*)\/\/(\w*)$/
#                       context = Regexp.last_match[1]
#                       match   = Regexp.last_match[2]
#                       candi   = @doc.search("#{context}//*").select {|e| e.elem? }.map {|e|
#                               e.name
#                       }.grep(/^#{match}/).map {|i|
#                               "#{context}//#{i}"
#                       }
#               when /^(.*)\/([^\/]*)$/
#                       context = Regexp.last_match[1]
#                       match   = Regexp.last_match[2]
#                       p context
#                       p @doc.search("#{context}/child::*")
#                       candi   = @doc.search("#{context}/child::*").select {|e| e.elem? }.map {|e|
#                               case
#                               when e[:id]
#                                       "#{e.name}[@id='#{e[:id]}']"
#                               when e[:class]
#                                       "#{e.name}[@class='#{e[:class]}']"
#                               else
#                                       e.name
#                               end
#                       }.grep(/^#{match}/).map {|i|
#                               "#{context}/#{i}"
#                       }
#               end
        end

        def cmd_get(arg)
                @uri = URI(arg)
                @doc = Hpricot(open(@uri).read)
                @prompt = "#{@uri.host + @uri.path}> "
        rescue Exception => e
                puts e
        end

        def cmd_xpath(arg)
                puts green { @doc.search(arg).map {|e| e.xpath }.join("\n") }
        end
        alias cmd_x cmd_xpath

        def cmd_exit(arg)
                exit
        end
end


if ARGV.empty?
        ARGV << "http://subtech.g.hatena.ne.jp/"
end

XPathFinderCommand.run(ARGV)