public static Robot ParseRobotTxtToRobotObject(string robotTxt, string urlTargetTest)
{
Robot robot = new Robot();
robot.Domain = urlTargetTest.ToLower().Replace("/robots.txt", string.Empty);
try
{
var matchUserAgent = GetRegexMatches(robotTxt, "User-agent: (.+)", RegexOptions.None);
var matchDisallow = GetRegexMatches(robotTxt, "Disallow: (.+)", RegexOptions.None);
var matchAllow = GetRegexMatches(robotTxt, "Allow: (.+)", RegexOptions.None);
var matchComments = GetRegexMatches(robotTxt, "# (.+)", RegexOptions.None);
var matchSiteMap = GetRegexMatches(robotTxt, "Sitemap: (.+)", RegexOptions.None);
ParseUserAgentData(matchUserAgent, robot);
ParseDisallowData(matchDisallow, robot);
ParseAllowData(matchAllow, robot);
ParseCommentsData(matchComments, robot);
ParseSiteMapData(matchSiteMap, robot);
}
catch (Exception ex)
{
robot = null;
throw new Exception("An error occurred when I try to parse Robot object", ex);
}
return robot;
}
private static void ParseUserAgentData(MatchCollection matchUserAgent, Robot robot)
{
if (matchUserAgent != null && matchUserAgent.Count > 0)
{
robot.UserAgent = new List<string>();
foreach (var userAgent in matchUserAgent)
robot.UserAgent.Add(userAgent.ToString().TrimEnd('\r', '\n').ToLower().Replace("user-agent: ", string.Empty));
}
}
private static void ParseDisallowData(MatchCollection matchDisallow, Robot robot)
{
if (matchDisallow != null && matchDisallow.Count > 0)
{
robot.Disallows = new List<string>();
foreach (var disallow in matchDisallow)
robot.Disallows.Add(disallow.ToString().TrimEnd('\r', '\n').ToLower().Replace("disallow: ", string.Empty));
}
}
private static void ParseCommentsData(MatchCollection matchComments, Robot robot)
{
if (matchComments != null && matchComments.Count > 0)
{
robot.Comments = new List<string>();
foreach (var comments in matchComments)
robot.Comments.Add(comments.ToString().ToLower().TrimEnd('\r', '\n').Replace("# ", string.Empty));
}
}
private static void ParseSiteMapData(MatchCollection matchSiteMap, Robot robot)
{
if (matchSiteMap != null && matchSiteMap.Count > 0)
{
robot.SiteMap = new List<string>();
foreach (var siteMap in matchSiteMap)
robot.Comments.Add(siteMap.ToString().TrimEnd('\r', '\n').ToLower().Replace("sitemap: ", string.Empty));
}
}
private static void ProcessResponse(HttpWebResponse response)
{
Util.Util.ChangeConsoleColorToGreen();
Console.WriteLine("TARGET FOUND :)");
Console.WriteLine();
Util.Util.ChangeConsoleColorToDefault();
Console.WriteLine("Getting Robots.txt...");
Console.WriteLine();
string robotsTxt = Util.Util.ParseResponseStreamToText(response);
Util.Util.ChangeConsoleColorToGreen();
Console.WriteLine("VOILÀ!");
Console.WriteLine();
Util.Util.ChangeConsoleColorToDefault();
Console.WriteLine("Begin of file");
Console.WriteLine("-------------------------------------------");
Console.WriteLine();
Console.WriteLine(robotsTxt);
Console.WriteLine();
Console.WriteLine("-------------------------------------------");
Console.WriteLine("End of file");
Console.WriteLine();
robot = Util.Util.ParseRobotTxtToRobotObject(robotsTxt, _urlTarget);
if (robot.Disallows != null && robot.Disallows.Count > 0)
{
Console.WriteLine("Listing 'disallow' directories...");
Console.WriteLine();
foreach (var disallowDirectory in robot.Disallows)
Console.WriteLine(disallowDirectory);
Console.WriteLine();
AskForAttack();
}
else
{
Console.WriteLine("There is no 'disallow' directories on the target.");
Console.WriteLine();
FinishExecution();
}
}
