private bool SaveSiteContents(string domain, SiteContents siteContents, bool overwrite)
{
if (!SiteInfo.IsValidDomain(domain))
{
return(false);
}
var container = GetContentsBlobContainer();
var blob = container.GetBlockBlobReference(domain + _siteFileExt);
if (!overwrite && blob.Exists())
{
return(false);
}
try
{
blob.UploadText(siteContents.SerializeToJson());
}
catch
{
return(false);
}
return(true);
}
public SiteCrawler(string domain, ISiteRepository siteRepository, CancellationToken cancelToken)
{
Trace.TraceInformation("Crawler: Starting processing domain " + domain);
if (!SiteInfo.IsValidDomain(domain))
{
throw new ArgumentException("Invalid domain");
}
if (siteRepository == null)
{
throw new ArgumentNullException("siteRepository is null");
}
if (cancelToken == null)
{
throw new ArgumentNullException("cancelToken is null");
}
_domain = domain;
_rawDomain = _domain.StartsWith("www.") ? _domain.Substring(4) : _domain;
_siteRepository = siteRepository;
_cancelToken = cancelToken;
DesiredNumberOfPages = 220;
CrawlDelay = 100;
MaxSimultaneousRequests = 20;
_binaryExtensions = " " + _binaryExtensions + " ";
}
private SiteInfo LoadSiteInfo(string domain)
{
if (!SiteInfo.IsValidDomain(domain))
{
return(null);
}
var table = GetInfoTable();
TableOperation operation = TableOperation.Retrieve <SiteInfoEntity>(SiteInfoEntity.StaticPartitionKey, domain);
try
{
var result = table.Execute(operation);
var entity = result.Result as SiteInfoEntity;
if (entity == null)
{
return(null);
}
return(entity.ToSiteInfo());
}
catch
{
return(null);
}
}
private SiteContents LoadSiteContents(string domain)
{
if (!SiteInfo.IsValidDomain(domain))
{
return(null);
}
var container = GetContentsBlobContainer();
var blob = container.GetBlockBlobReference(domain + _siteFileExt);
if (!blob.Exists())
{
return(null);
}
try
{
string st = blob.DownloadText();
return(SiteContents.DeserializeFromJson(st));
}
catch
{
return(null);
}
}
public bool SiteExists(string domain)
{
if (!SiteInfo.IsValidDomain(domain))
{
return(false);
}
return(LoadSiteInfo(domain) != null);
}
public bool SiteExists(string domain)
{
if (!SiteInfo.IsValidDomain(domain))
{
return(false);
}
return(File.Exists(GetFullSiteInfoFileName(domain)));
}
public bool ProcessSite(string domain)
{
domain = SiteInfo.NormalizeDomain(domain);
if (!SiteInfo.IsValidDomain(domain))
{
throw new ArgumentException("Invalid domain");
}
return(_crawler.ProcessSite(domain));
}
private bool SaveSiteInfoToFile(SiteInfo siteInfo, bool overwrite)
{
if (siteInfo == null)
{
return(false);
}
if (!SiteInfo.IsValidDomain(siteInfo.Domain))
{
return(false);
}
string fileName = GetFullSiteInfoFileName(siteInfo.Domain);
if (!overwrite && File.Exists(fileName))
{
return(false);
}
int retryCount = 3;
do
{
try
{
using (var stream = new FileStream(fileName, FileMode.Create, FileAccess.Write, FileShare.None))
{
using (var writer = new StreamWriter(stream))
{
writer.Write(siteInfo.SerializeToJson());
}
}
return(true);
}
catch (IOException)
{
retryCount--;
if (retryCount == 0)
{
return(false);
}
Thread.Sleep(2);
}
catch
{
return(false);
}
}while (retryCount > 0);
return(true);
}
public ActionResult Index(SiteAddressModel siteAddress)
{
if (ModelState.IsValid)
{
if ((siteAddress.Address == null) || !SiteInfo.IsValidDomain(SiteInfo.NormalizeDomain(siteAddress.Address)))
{
ModelState.AddModelError("", "Invalid domain");
return(View(siteAddress));
}
return(RedirectToAction("ShowMap", "Map", new { domain = SiteInfo.NormalizeDomain(siteAddress.Address) }));
}
return(View(siteAddress));
}
public Site GetSite(string domain, bool includeContents, long?contentsTimeStamp = null, bool createIfNecessary = false)
{
domain = SiteInfo.NormalizeDomain(domain);
if (!SiteInfo.IsValidDomain(domain))
{
throw new ArgumentException("Invalid domain");
}
var site = _siteRepository.GetSite(domain, includeContents, contentsTimeStamp);
bool needToCreate = (site == null) && createIfNecessary;
needToCreate = needToCreate || ((site != null) && (DateTime.UtcNow - site.Info.StatusTime > TimeSpan.FromDays(RefreshPeriodInDays)));
needToCreate = needToCreate || ((site != null) &&
((site.Info.Status == SiteStatus.ConnectionProblem) || (site.Info.Status == SiteStatus.RobotsTxtProblem)) &&
(DateTime.UtcNow - site.Info.StatusTime > TimeSpan.FromMinutes(10)));
var needToProcess = ((site != null) && ((site.Info.Status == SiteStatus.Added) || (site.Info.Status == SiteStatus.Processing)));
if ((site != null) && !site.Info.RefreshEnabled)
{
needToProcess = false;
}
if (needToCreate || needToProcess)
{
if (needToCreate)
{
site = new Site();
site.Info.Domain = domain;
if (_siteRepository.SaveSite(site, true))
{
_crawler.ProcessSite(domain);
}
else
{
site = null;
}
}
else
{
_crawler.ProcessSite(domain);
}
}
return(site);
}
private SiteContents LoadSiteContentsFromFile(string domain)
{
if (!SiteInfo.IsValidDomain(domain))
{
return(null);
}
string fileName = GetFullSiteContentsFileName(domain);
if (!File.Exists(fileName))
{
return(null);
}
int retryCount = 3;
do
{
try
{
using (var stream = new FileStream(fileName, FileMode.Open, FileAccess.Read, FileShare.None))
{
using (var reader = new StreamReader(stream))
{
string st = reader.ReadToEnd();
return(SiteContents.DeserializeFromJson(st));
}
}
}
catch (IOException)
{
retryCount--;
if (retryCount == 0)
{
return(null);
}
Thread.Sleep(2);
}
catch
{
return(null);
}
}while (retryCount > 0);
return(null);
}
public bool RemoveSite(string domain)
{
if (!SiteInfo.IsValidDomain(domain))
{
return(false);
}
if (!DeleteSiteInfo(domain))
{
return(false);
}
DeleteSiteContents(domain);
return(true);
}
public bool RemoveSite(string domain)
{
if (!SiteInfo.IsValidDomain(domain))
{
return(false);
}
try
{
File.Delete(GetFullSiteInfoFileName(domain));
File.Delete(GetFullSiteContentsFileName(domain));
}
catch
{
return(false);
}
return(true);
}
private bool DeleteSiteContents(string domain)
{
if (!SiteInfo.IsValidDomain(domain))
{
return(false);
}
var container = GetContentsBlobContainer();
var blob = container.GetBlockBlobReference(domain + _siteFileExt);
try
{
blob.Delete();
}
catch
{
return(false);
}
return(true);
}
private bool DeleteSiteInfo(string domain)
{
if (!SiteInfo.IsValidDomain(domain))
{
return(false);
}
var table = GetInfoTable();
try
{
// Create a retrieve operation
TableOperation retrieveOperation = TableOperation.Retrieve <SiteInfoEntity>(SiteInfoEntity.StaticPartitionKey, domain);
// Execute the operation.
TableResult retrievedResult = table.Execute(retrieveOperation);
// Assign the result to a CustomerEntity.
var deleteEntity = (SiteInfoEntity)retrievedResult.Result;
// Create the Delete TableOperation.
if (deleteEntity != null)
{
TableOperation deleteOperation = TableOperation.Delete(deleteEntity);
// Execute the operation.
table.Execute(deleteOperation);
return(true);
}
else
{
return(false);
}
}
catch
{
return(false);
}
}
private bool SaveSiteInfo(SiteInfo siteInfo, bool overwrite)
{
if (siteInfo == null)
{
return(false);
}
if (!SiteInfo.IsValidDomain(siteInfo.Domain))
{
return(false);
}
var table = GetInfoTable();
TableOperation operation;
var entity = new SiteInfoEntity(siteInfo);
if (overwrite)
{
operation = TableOperation.InsertOrReplace(entity);
}
else
{
operation = TableOperation.Insert(entity);
}
try
{
var res = table.Execute(operation);
}
catch
{
return(false);
}
return(true);
}