---
--@author Sven Klemm <sven@c3d2.de>

module(... or "sedusa", package.seeall)

require 'stdnse'
require 'http'
require 'xml'

hints = {}
verify = {}

document_cache = {}

Document = {
  new = function( doc )
    doc.xml = nil
    if string.len( doc.body ) > 0 then
      doc.xml = xml.parse_html( doc.body )
    end
    return doc
  end
}

http_get = function( u, options )
  stdnse.print_debug( 2, "Fetching %s", u )
  local document = http.get_url( u, options )
  return Document.new( document )
end

get_document = function( url )
  local document

  if not document_cache[ url ] then
    document = http_get( url )
    document_cache[ url ] = document
  else
    document = document_cache[ url ]
  end
  if document.status == 301 or
     document.status == 302
   then
    -- take care of relative redirects
    if not document.header.location:match('https?://') then
      local base = url:match('^(.*/)[^/]*$')
      document.header.location = base .. document.header.location
    end
    document = get_document( document.header.location )
  end

  return document
end

detect_app = function( u, no_verify )
  local app,doc,exps,index,xpath
  local apps = {}

  doc = get_document( u )

  for app, exps in pairs( sedusa.hints ) do
    for index, xpath in pairs( exps ) do
      if doc.xml:find( xpath ) then
        if no_verify then
          table.insert( apps, app )
        else
          if sedusa.verify[app] then
            local version = sedusa.verify[app]( u )
            if version then
              table.insert( apps, app .. " " .. version )
            else
              table.insert( apps, app )
            end
          else
            stdnse.print_debug( "No verify function for %s found.", app )
            table.insert( apps, app )
          end
        end
        break
      end
    end
  end

  return apps
end

-- setup default detector for some web applications
for key, app in pairs( {'b2evolution', 'bBlog', 'C3D2-Web', 'DokuWiki', 'gitweb', 'Midgard', 'Nucleus CMS', 'Pentabarf', 'PhpWiki', 'Plone', 'PostNuke', 'TYPO3', 'vBulletin'} ) do
  hints[app] = { '//meta[@name="generator" and starts-with(@content,"' .. app .. '")]' }
  verify[app] = function( url )
    local document = get_document( url )
    -- look in generator meta tag
    local generator = document.xml:find( '//meta[@name="generator"]/@content' )
    if generator and string.match( generator, app .. " (.*)" ) then
      return string.match( generator, app .. " (.*)" )
    end
  end
end

hints['Bugzilla'] = {
  '//div[@id="banner"]/p[@id="banner-version"]/a[@href="http://www.bugzilla.org/"]/span[text()="Bugzilla"]',
  '//div[@id="header"]/table/tr/td[@id="title"]/p[starts-with(text(),"Bugzilla")]',
}

verify['Bugzilla'] = function( u )
  local doc = get_document( u )
  local version = doc.xml:find( '//div[@id="banner"]/p[@id="banner-version" and a[@href="http://www.bugzilla.org/"]/span[text()="Bugzilla"]]/span[starts-with(text(),"Version ")]/text()' )
  if version then
    return version:match('Version (.*)')
  end
  version = doc.xml:find( '//div[@id="header"]/table/tr/td[@id="information"]/p[@class="header_addl_info"]/text()' )
  if version then
    return version:match('version (.*)')
  end
end

hints['Burning Board'] = {
  '//span[@class="smallfont" and text()="Powered by "]/b[starts-with(text(),"WoltLab Burning Board")]',
  '//span[@class="smallfont"]/a[@href="http://www.woltlab.de"]/b[contains(text(),"Burning Board")]',
}

verify['Burning Board'] = function( u )
  local doc = get_document( u )
  local version = doc.xml:find( hints['Burning Board'][1] )
  if not version then version = doc.xml:find( hints['Burning Board'][2] ) end
  if version then
    return version:match('Burning Board (.*)')
  end
end

hints['CommunityServer'] = {
  '//meta[@name="GENERATOR" and starts-with(@content, "CommunityServer")]/@content',
  '//a[@href="http://communityserver.org/r.ashx?1" and @target="_blank"]/img[starts-with(@alt,"Powered by CommunityServer")]',
}

verify['CommunityServer'] = function( u )
  local doc = get_document( u )
  local version = doc.xml:find( hints['CommunityServer'][1] )
  if version then
    return version:match("^CommunityServer (.*)$")
  end
end

hints['Cisco VPN 3000 Concentrator'] = {
  '//title[starts-with(@text,"Cisco Systems, Inc. VPN 3000 Concentrator [")]',
  '//form/table/tr/td[@align="right" and @colspan="2" and text()="VPN 3000 Concentrator"]',
}

hints['Drupal'] = {
  '//div[@id="block-user-0" and h2[text()="User login"]]/div[@class="content"]/form[@id="user-login-form"]/div[div[@class="form-item"]/input[@type="text" and @class="form-text required"]]',
  '//p[@class="foot2"]/a[@href="http://drupal.org"]/img[contains(@title,"Powered by Drupal")]',
}

hints['Flyspray'] = {
  '//p[@id="footer"]/a[@href="http://flyspray.org/" and @class="offsite" and starts-with(text(),"Powered by Flyspray")]',
  '//p[@id="footer"]/a[@href="http://flyspray.rocks.cc/" and @class="offsite" and starts-with(text(),"Powered by Flyspray")]',
}

verify['Flyspray'] = function( u )
  local doc = get_document( u )
  local foot = doc.xml:find( hints['Flyspray'][1] )
  if foot then return foot:match('Powered by Flyspray (.*)') end
end

hints['Joomla!'] = {
  '//meta[@name="Generator" and starts-with(@content,"Joomla!")]'
}

hints['GForge'] = {
  '//a[@href="http://gforge.org"]/img[@alt="Powered by GForge"]',
}

hints['Instiki'] = {
  '//head/link[@rel="stylesheet" and contains(@href,"/stylesheets/instiki.css")]',
  '//div[@id="footer"]/div[text()="This site is running on "]/a[@href="http://instiki.org" and text()="Instiki"]'
}

hints['MediaWiki'] = {
  '//body[contains(@class, "mediawiki")]',
  '//div[@id="footer"]/div[@id="f-poweredbyico"]/a[@href="http://www.mediawiki.org/"]/img',
  '//head/script[@type="text/javascript" and contains(@text,"var wgVersion")]'
}

verify['MediaWiki'] = function( u )
  local document = get_document( u )
  -- find out location of index.php
  local link = document.xml:find( '//a[contains(@href,"index.php?title=") and starts-with(@href, "/")]/@href' )
  if link then
    link = link:match( "^(.*/index.php[?]title=).*")
  else
    link = document.xml:find( '//script[@type="text/javascript" and contains(text(), "var wgScriptPath = ")]' )
    link = link:gsub( "[\r\n]", "" )
    if link then
      link = link:match( 'var wgScriptPath = "([^"]+)"') .. '/index.php?title='
    end
  end

  -- Special:Version has the most exact version including svn revision
  local version = get_document( url.absolute( u, link .. "Special:Version" ) )
  if version.xml:find('//div/ul/li[a[@href="http://www.mediawiki.org/" and text()="MediaWiki"]]') then
    return string.match( version.xml:find('//div/ul/li[a[@href="http://www.mediawiki.org/" and text()="MediaWiki"]]'), "MediaWiki: ([^\r\n]+)" )
  end
  if version.xml:find('//table[@id="sv-software"]/tr[td[a[@href="http://www.mediawiki.org/" and text()="MediaWiki"]]]') then
    return version.xml:find('//table[@id="sv-software"]/tr[td[a[@href="http://www.mediawiki.org/" and text()="MediaWiki"]]]/td[2]')
  end

  -- the version without revision is included on every page as javascript variable on recent versions
  local js = document.xml:find('//script[@type="text/javascript" and contains(text(), "var wgVersion = ")]')
  if js then
    js = js:gsub( "[\r\n]", "" )
    return js:match( 'var wgVersion = "([^"]+)"')
  end

  -- get version from atom feed
  local atom = get_document( url.absolute( u, link .. "Special:Recentchanges&feed=atom" ) )
  if atom.xml:find( '//generator[starts-with(text(), "MediaWiki")]' ) then
    return string.match( atom.xml:find( '//generator/text()' ), "^MediaWiki (.*)$" )
  end
end

hints['OpenWRT'] = {
  '//head/title[text()="OpenWrt Administrative Console"]'
}

verify['OpenWRT'] = function( u )
  local document = get_document( u )
  local webif = document.xml:find( '//head/meta[@http-equiv="refresh"]/@content' )
  if webif and webif:match('0; URL=') then
    document = get_document( url.absolute( u, webif:match( '0; URL=(.*)' )) )
    local version = document.xml:find( '//div[@id="header"]/div[@id="header-title"]/div[@id="short-status"]/ul/li[strong[text()="Version:"]]/text()' )
    if version then
      return version:match( ' +(.*)' )
    end
  end
end

hints['PHP-Nuke'] = {
  '//meta[@name="GENERATOR" and starts-with(@content,"PHP-Nuke")]'
}

hints['phpMyAdmin'] = {
  '//title[starts-with(text(), "phpMyAdmin")]',
  '//link[@rel="stylesheet" and  @type="text/css" and contains(@href,"css/phpmyadmin.css.php")]',
  '//a[@href="http://www.phpmyadmin.net" and @class="logo" and @target="_blank"]/img[@alt="phpMyAdmin"]'
}

verify['phpMyAdmin'] = function( url )
  local document = get_document( url )
  local title = document.xml:find( '//title[starts-with(text(), "phpMyAdmin")]/text()' )
  local version = title:match('^phpMyAdmin (.*)$')
  if version then
    return version
  end
end

hints['phpSysInfo'] = {
  '//a[@href="http://phpsysinfo.sourceforge.net" and @target="_blank" and starts-with(text(),"phpSysInfo")]'
}

verify['phpSysInfo'] = function( url )
  local document = get_document( url )
  local version = document.xml:find( '//a[@href="http://phpsysinfo.sourceforge.net" and @target="_blank" and starts-with(text(),"phpSysInfo")]/text()' )
  version = version:match('^phpSysInfo[-](.*)$')
  if version then
    return version
  end
end

verify['PhpWiki'] = function( url )
  local document = get_document( url )
  local generator = document.xml:find( '//meta[@name="PHPWIKI_VERSION"]/@content' )
  if generator then return generator end
end

verify['Plone'] = function( u )
  local document = get_document( u )
  -- look in header
  if document.header.server then
    local version = document.header.server:match( 'Plone/(%d.%d.%d)' )
    if version then return version end
  end
end

hints['Serendipity'] = {
  '//meta[@name="Powered-By" and starts-with(@content, "Serendipity")]',
  '//div[@id="serendipity_banner"]',
  '//div[@class="serendipity_entry_body"]',
  '//div[@class="serendipity_entryFooter"]',
}

verify['Serendipity'] = function( u )
  local document = get_document( u )
  local generator = document.xml:find( '//meta[@name="Powered-By"]/@content' )
  if generator and string.match( generator, 'Serendipity v[.](.*)' ) then
    return string.match( generator, 'Serendipity v[.](.*)' )
  end
  local rss_url = document.xml:find( '//link[@rel="alternate" and @type="application/rss+xml"]/@href' )
  if rss_url then
    document = get_document( url.absolute( u, rss_url ) )
    generator = document.xml:find( '//generator')
    if generator and string.match( generator, 'Serendipity [0-9.]+' ) then
      return string.match( generator, 'Serendipity ([0-9.]+)' )
    end
  end
end

hints['SMF'] = {
  '//span[@class="smalltext"]/a[@href="http://www.simplemachines.org/" and contains(text(),"Powered by SMF")]',
}

verify['SMF'] = function(u)
  local doc = get_document( u )
  local foot = doc.xml:find( hints['SMF'][1] )
  if foot then
    return foot:match('Powered by SMF (.*)')
  end
end

hints['Trac'] = {
  '//div[@id="footer"]/a[@id="tracpowered" and @href="http://trac.edgewall.org/"]/img[@alt="Trac Powered"]',
}

verify['Trac'] = function( u )
  local document = get_document( u )
  local version = document.xml:find( '//div[@id="footer" and a[@id="tracpowered" and @href="http://trac.edgewall.org/"]]/p[@class="left"]/a/strong/text()' )
  if version and version:match( 'Trac (.*)' ) then
    return version:match( 'Trac (.*)' )
  end
end

hints['WordPress'] = {
  '//meta[@name="generator" and starts-with(@content,"WordPress")]',
  '//head/link[@rel="stylesheet" and @type="text/css" and contains( @href, "/wp-content/")]',
  '//head/style[contains( text(), "/wp-content/")]',
  '//div[@id="content"]/div[@class="post" and starts-with(@id, "post-") and div[@class="posttitle"] and div[@class="postmeta"] and div[@class="postbody"] and div[@class="postfooter"]]',
  '//ul/li/a[@href="http://wordpress.org/" and starts-with(@title,"Powered by WordPress")]',
}

verify['WordPress'] = function( u )
  local document = get_document( u )
  -- look in generator meta tag
  local generator = document.xml:find( '//meta[@name="generator"]/@content' )
  if generator and string.match( generator, "WordPress (.*)" ) then
    return string.match( generator, "WordPress (.*)" )
  end
  -- look in atom feed
  local atom = document.xml:find( '//link[@rel="alternate" and @type="application/atom+xml"]/@href' )
  local feed = get_document( atom )
  if feed.xml:find( '//generator[text()="WordPress"]/@version' ) then
    return feed.xml:find( '//generator[text()="WordPress"]/@version' )
  end
end