root/platform/pragger/Feed/nicovideo_find.rb @ 28583

## search nicovideo by tag. returning array of HTML page.
##
## If is not specified mail and password, your e-mail address and password must be set into ~/.netrc as machine nicovideo.jp.
##
## - module: Feed::nicovideo_find
##   config:
##     tag: IDOLM@STER (must)
##     maii: your mail address (optional)
##     password: your password (optional)
##
## or give some tags by a list:
##
## - module: Feed::nicovideo_find
##   config:
##     tag:
##       - IDOLM@STER
##       - IM@S
##
## Copyright (C) 2007, TADA Tadashi <sho@spc.gr.jp>, 
## 2007-2008 INOUE Yasuyuki <inoue.yasuyuki0@gmail.com>
## Original version is written by TADA Tadashi <sho@spc.gr.jp>
## You can redistribute it and/or modify it under GPL3 or any later version.

begin
        require 'rubygems'
rescue LoadError
end
require 'mechanize'
require 'cgi'

def nicovideo_find( config, data )
        plugin_name = "Feed::nicovideo_find"
        
        data = [] if config['reset']

        mail = config['mail']
        password = config['password']
        unless mail then
                require 'net/netrc'
                if netrc = Net::Netrc.locate( 'nicovideo.jp' ) then
                        mail, password = netrc.login, netrc.password
                else
                        $stderr.puts "#{plugin_name}: no nicovideo.jp entry in .netrc."
                        return []
                end
        end

        begin
                nico = "http://www.nicovideo.jp/"
                #sort = '?page=#{count}&sort=f'
                agent = WWW::Mechanize::new

                # login sequence
                page = agent.get( nico )
                login_form = page.forms.with.name( 'login_bar' ).first
                login_form.mail = mail
                login_form.password = password
                agent.submit( login_form )
        rescue NameError # no login form
                $stderr.puts "#{plugin_name}::nicovideo_find: login failure."
                return nil
        end

        # getting each tag pages
        config['tag'].each do |tag|
                (1..5).each do |count|
                        retry_count = 0
                        uri = "#{nico}tag/#{CGI::escape( tag )}?page=#{count}&sort=f"
                        begin
                                page = agent.get( uri )
                                tmp_data = page.body
                                data << tmp_data
                                break unless tmp_data.include? '次のページ&nbsp;&gt;&gt;</a>'
                        rescue EOFError # unknown error
                                if retry_count < 10 then
                                        retry_count += 1
                                        $stderr.puts "#{plugin_name}: retry #{tag} (#{retry_count})."
                                        sleep 10
                                        retry
                                end
                                $stderr.puts "#{plugin_name}::nicovideo_find: unknown error. retry aborted."
                        end
                        sleep 30
                end
        end
        data
end