/// <summary>
/// Fetch a robot file for a uri
/// </summary>
/// <param name="anyUri">Any URI for which you want the robots file for</param>
/// <returns></returns>
public static async Task <RobotsFile> GetFile(Uri anyUri)
{
//log.Debug("Downloading robots: " + uri);
Site site = null;
RobotsFile robotsFile = null;
var robotsUri = MakeRobotsUri(anyUri);
bool needsVisiting = true;
try
{
var db = await DatabasePool.GetDatabaseAsync();
site = await db.GetSite(robotsUri);
await DatabasePool.GiveBackToPool(db);
if (site != null)
{
needsVisiting = site.RobotsNeedsVisiting;
}
else
{
site = MakeNewSite(anyUri);
}
if (needsVisiting)
{
if (site != null && site.IsBlocked)
{
Utility.LogInfo("Can't get robots file as site is blocked by policy: " + robotsUri);
return(null);
}
robotsFile = await DownloadRobots(robotsUri, site.LastRobotsFetched);
site.LastRobotsFetched = DateTime.UtcNow;
site.RobotsFile = robotsFile;
db = await DatabasePool.GetDatabaseAsync();
await db.SaveSite(site);
await DatabasePool.GiveBackToPool(db);
}
else
{
robotsFile = site.RobotsFile;
}
}
catch (Exception ex)
{
Utility.LogException(ex);
}
return(robotsFile);
}
/// <summary>
/// Download a robots file
/// </summary>
/// <param name="robotsUri"></param>
/// <param name="lastFetched"></param>
/// <returns></returns>
public static async Task <RobotsFile> DownloadRobots(Uri anyUri, DateTime?lastFetched)
{
RobotsFile robots = null;
var robotsUri = MakeRobotsUri(anyUri);
try
{
var ip = await Utility.GetHostIPAddress(robotsUri);
/*while (!await FetchoConfiguration.Current.HostCache.WaitToFetch(ip, 60000))
* Utility.LogInfo("IP Congestion {0}", ip);*/
var bb = new BufferBlock <IWebResourceWriter>();
using (var ms = new MemoryStream())
{
using (var packet = new WebDataPacketWriter(ms))
{
// this is annoying, I shouldn't have to create a buffer block to get a robots file
// or we should put robots into the standard flow of things
await bb.SendAsync(packet);
await(new HttpResourceFetcher()).Fetch(null, robotsUri, null, lastFetched, bb);
}
ms.Seek(0, SeekOrigin.Begin);
using (var packet = new WebDataPacketReader(CreateXmlReader(ms)))
{
using (var stream = packet.GetResponseStream())
{
if (stream == null)
{
robots = new RobotsFile();
}
else
{
robots = new RobotsFile(robotsUri, stream);
}
}
}
}
}
catch (Exception ex)
{
Utility.LogInfo("Fetching {0}:", robotsUri);
Utility.LogException(ex);
}
return(robots);
}
