/// <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);
}
public async void OutputResponse(WebResponse response, byte[] buffer, int bytesRead)
{
try
{
Writer.OutputResponse(response, buffer, bytesRead);
// bail if we dont get anything
if (bytesRead == 0)
{
return;
}
// if we need to push it to this workspace
if (queueItem is ImmediateWorkspaceQueueItem wqi)
{
try
{
var sb = new StringBuilder();
if (response is HttpWebResponse httpWebResponse)
{
sb.AppendFormat("status: {0} {1}\n", httpWebResponse.StatusCode, httpWebResponse.StatusDescription);
}
foreach (string key in response.Headers)
{
sb.AppendFormat("{0}: {1}\n", key, response.Headers[key]);
}
responseHeaders = sb.ToString();
using (var ms = new MemoryStream(buffer))
{
var builder = new WorkspaceResultBuilder();
var result = builder.Build(ms, requestString, responseHeaders, out string evalText);
result.Tags.AddRange(wqi.Tags);
var hash = MD5Hash.Compute(buffer);
var db = await DatabasePool.GetDatabaseAsync();
try
{
await db.AddWorkspaceResults(wqi.DestinationWorkspaceId, new[] { result });
// OPTIMISE: Remove ToArray and just pass IEnumerable<>
await db.AddWebResourceDataCache(hash, buffer.Take(bytesRead).ToArray());
}
catch (Exception ex)
{
Utility.LogException(ex);
}
finally
{
await DatabasePool.GiveBackToPool(db);
}
}
}
catch (Exception ex)
{
Utility.LogException(ex);
}
}
}
catch (Exception ex)
{
Utility.LogException(ex);
}
}
