/// <summary>
/// Extracts links from the contents of a document.
/// </summary>
/// <param name="content">The contents of the document.</param>
/// <param name="contentUrl">The url of the document.</param>
/// <returns>
/// An <see cref="ArrayList"/> of <see cref="InternetUrlToIndex"/> objects, one for
/// each link found in the content.
/// </returns>
/// <remarks>This method <b>ALWAYS</b> returns an empty ArrayList.</remarks>
public override ArrayList ExtractLinks(byte[] content, ref InternetUrlToCrawl contentUrl)
{
ArrayList links = new ArrayList();
ParserEventArgs e = new ParserEventArgs(contentUrl.Url);
OnExtractLinksComplete(e);
return(links);
}
/// <summary>
/// Performs the extraction of links from a text document. It can extract simple
/// links that are separated from the rest of the text using spaces or line brakes
/// or any other delimiters. The results are returned as an <see cref="ArrayList"/>
/// of <see cref="InternetUrlToIndex"/> objects.
/// </summary>
/// <remarks>
/// Besides the parsing and extraction of Urls, ExtractLinks also performs other
/// tasks as well, such as:<br/>
/// <list type="bullet">
/// <item>
/// <description>Filtering of urls to resources of unsupported content-type, e.g. css, images, etc.</description>
/// </item>
/// <item>
/// <description>Filtering of multimple links to the same url and to the document itself.</description>
/// </item>
/// <item>
/// <description>Filtering of session id variables in dynamic Urls and limiting
/// of the number of GET variables in dynamic Urls.</description>
/// </item>
/// <item>
/// <description>Flagging of Urls according to their country domain.</description>
/// </item>
/// </list>
/// <b>Update History</b>
/// <list type="table">
/// <listheader>
/// <term>Date</term>
/// <description>Description</description>
/// </listheader>
/// <item>
/// <term>15/09/04</term>
/// <description>First release. A lot more needs to be done.</description>
/// </item>
/// </list>
/// </remarks>
/// <param name="content">The text that must be parsed for links. IIt is passed as
/// an array of bytes containing the text contents in UTF8 binary format, in order
/// to reduce memory consumption.</param>
/// <param name="contentUrl">The Url from which the content comes.</param>
/// <returns>
/// An <see cref="ArrayList"/> of <see cref="InternetUrlToIndex"/> objects, one for
/// each link found in the content.
/// </returns>
public override ArrayList ExtractLinks(byte[] content, ref InternetUrlToCrawl contentUrl)
{
ArrayList retVal = null;
try
{
mutex.WaitOne();
string html = Encoding.UTF8.GetString(content);
retVal = ExtractLinks(ref html, ref contentUrl);
}
catch
{}
finally
{
mutex.ReleaseMutex();
}
return(retVal);
}
/// <summary>
/// Checks if at least 30 seconds have passed since the last request to a given host
/// was made, in order not to slammer it with simultaneous or frequent requests.
/// </summary>
/// <param name="targetUrl">
/// A <see cref="InternetUrlToCrawl"/> that is served by a host we wish to check.
/// </param>
/// <returns>
/// An integer containing the number of milliseconds a crawler thread must wait
/// before visiting this host.
/// </returns>
public int FilterUrl(ref InternetUrlToCrawl targetUrl)
{
string hostName = InternetUtils.HostName(targetUrl);
return(FilterHost(ref hostName));
}
/// <summary>
/// Selects and returns a set of urls that are ready to be crawled.
/// </summary>
/// <param name="ci">The <see cref="ClientInfo"/> of the client requesting urls to crawl.</param>
/// <param name="data">An array of <see cref="InternetUrlToCrawl"/> objects containing the selected urls.</param>
/// <returns>Null if the operation succeeds, or <see cref="SerializedException"/>
/// encapsulating the error that occured if the operation fails.</returns>
public SerializedException SelectUrlsToCrawl(ClientInfo ci, ref InternetUrlToCrawl[] data)
{
SerializedException sx = null;
try
{
if (!ConnectToDatabase())
{
throw new CWDBConnectionFailedException();
}
//we must use a transaction to make sure that if something goes wrong the
//changes to the database will be rolled back.
SqlTransaction transaction = dbcon.BeginTransaction(IsolationLevel.Serializable); //perhaps | repeatableread
try
{
//first select the urls to crawl
SqlCommand cmd = new SqlCommand("cw_select_urls_to_crawl", dbcon, transaction);
cmd.CommandType = CommandType.StoredProcedure;
cmd.CommandTimeout = 120;
SqlDataAdapter da = new SqlDataAdapter(cmd);
DataSet ds = new DataSet();
da.Fill(ds);
da.Dispose();
cmd.Dispose();
//now delete them from the table of urls to crawl
data = new InternetUrlToCrawl[ds.Tables[0].Rows.Count];
if (data.Length > 0)
{
int i = 0;
foreach (DataRow dr in ds.Tables[0].Rows)
{
try
{
InternetUrlToCrawl url = new InternetUrlToCrawl((int)dr[0], (string)dr[1]);
if (dr[2] != DBNull.Value)
{
url.CRC = (long)dr[2];
}
if (dr[3] != DBNull.Value)
{
url.FlagDomain = (DomainFlagValue)((byte)dr[3]);
}
if (dr[4] != DBNull.Value)
{
url.RobotsDisallowedPaths = (string)dr[4];
}
else
{
RobotsTxtEntry entry = settings.Robots.GetEntry(InternetUtils.HostName(url));
if (entry != null)
{
url.RobotsDisallowedPaths = ConcatenatePaths(entry.DisallowedPaths);
}
else
{
url.FlagFetchRobots = true;
}
}
data[i++] = url;
}
catch
{
continue;
}
}
SqlCommand statscmd = new SqlCommand("cw_update_client_statistics", dbcon, transaction);
statscmd.CommandType = CommandType.StoredProcedure;
statscmd.CommandTimeout = 120;
statscmd.Parameters.Add("@client_id", SqlDbType.UniqueIdentifier);
statscmd.Parameters.Add("@assigned", SqlDbType.BigInt);
statscmd.Parameters.Add("@returned", SqlDbType.BigInt);
statscmd.Parameters.Add("@type", SqlDbType.TinyInt);
statscmd.Parameters[0].Value = ci.ClientID;
statscmd.Parameters[1].Value = data.Length;
statscmd.Parameters[2].Value = DBNull.Value;
statscmd.Parameters[3].Value = 0;
statscmd.ExecuteNonQuery();
statscmd.Dispose();
transaction.Commit();
}
}
catch (Exception ex)
{
transaction.Rollback();
if (settings.LogLevel <= CWLogLevel.LogWarning)
{
settings.Log.LogWarning("SelectUrlsToCrawl failed, Transaction was rolled back: " + ex.ToString());
}
throw ex;
}
finally
{
UpdateClientLastActive(ci);
LogClientAction(ci, CWClientActions.LogSendUrlsToCrawl);
if (!DisconnectFromDatabase())
{
throw new CWDBConnectionFailedException("Disconnect from database failure.");
}
}
}
catch (Exception e)
{
sx = new SerializedException(e.GetType().ToString(), e.Message, e.ToString());
if (settings.LogLevel <= CWLogLevel.LogWarning)
{
settings.Log.LogWarning("SelectUrlsToCrawl failed: " + e.ToString());
}
}
return(sx);
}
/// <summary> /// Extracts the hypertext references (links) contained in a document. /// </summary> /// <param name="content"> /// An array of bytes holding the content of the document that will be parsed for links. /// </param> /// <param name="contentUrl"> /// An <see cref="InternetUrlToCrawl"/> object encapsulating the Uri address of the /// document to be parsed for links and its associated robots.txt file. /// </param> /// <returns> /// An <see cref="ArrayList"/> of <see cref="InternetUrlToIndex"/> objects encapsulating /// the links contained in the parsed document. /// </returns> public abstract ArrayList ExtractLinks(byte[] content, ref InternetUrlToCrawl contentUrl);
/// <summary> /// Extracts the hypertext references (links) contained in a document. /// </summary> /// <param name="content"> /// The content of the document that will be parsed for links. /// </param> /// <param name="contentUrl"> /// An <see cref="InternetUrlToCrawl"/> object encapsulating the Uri address of the /// document to be parsed for links and its associated robots.txt file. /// </param> /// <returns> /// An <see cref="ArrayList"/> of <see cref="InternetUrlToIndex"/> objects encapsulating /// the links contained in the parsed document. /// </returns> public abstract ArrayList ExtractLinks(ref string content, ref InternetUrlToCrawl contentUrl);
/// <summary>
/// Extracts links from the contents of a SWF document.
/// </summary>
/// <param name="content">The contents of the SWF document.</param>
/// <param name="contentUrl">The url of the SWF document.</param>
/// <returns>
/// An <see cref="ArrayList"/> of <see cref="InternetUrlToIndex"/> objects, one for
/// each link found in the content.
/// </returns>
/// <exception cref="ArgumentNullException">If the input buffer is null or empty.</exception>
public override ArrayList ExtractLinks(byte[] content, ref InternetUrlToCrawl contentUrl)
{
ArrayList links = null;
if ((content == null) || (content.Length == 0))
{
throw new ArgumentNullException("content", "The input buffer cannot be empty or null.");
}
try
{
mutex.WaitOne();
string FileName = globals.AppWorkPath + Guid.NewGuid().ToString();
string swfFileName = FileName + ".swf";
string htmFileName = FileName + ".htm";
FileStream swf = null;
StreamReader htm = null;
try
{
//store the swf file
swf = new FileStream(swfFileName, FileMode.Create);
swf.Write(content, 0, content.Length);
swf.Close();
swf = null;
//convert it to html
bool success = converter.ConvertSwfFile(swfFileName, htmFileName);
if (success)
{
htm = new StreamReader(htmFileName, encoding);
string html = htm.ReadToEnd();
htm.Close();
htm = null;
links = parser.ExtractLinks(ref html, ref contentUrl);
}
}
catch (Exception ex)
{
if (swf != null)
{
try
{
swf.Close();
}
catch
{}
}
if (htm != null)
{
try
{
htm.Close();
}
catch
{}
}
if (globals.Settings.LogLevel <= CWLogLevel.LogInfo)
{
globals.FileLog.LogInfo("SwfParser failed to extract links from " + contentUrl.Url + ": " + ex.ToString());
}
}
finally
{
File.Delete(swfFileName);
File.Delete(htmFileName);
}
}
catch (Exception ex)
{
if (globals.Settings.LogLevel <= CWLogLevel.LogWarning)
{
globals.FileLog.LogWarning("SwfParser failed to extract links from " + contentUrl.Url + ": " + ex.Message);
}
}
finally
{
GC.Collect();
mutex.ReleaseMutex();
}
ParserEventArgs e = new ParserEventArgs(contentUrl.Url);
OnExtractLinksComplete(e);
return(links);
}
/// <summary>
/// Extracts links from the contents of a SWF document.
/// </summary>
/// <param name="content">The contents of the SWF document.</param>
/// <param name="contentUrl">The url of the PDF document.</param>
/// <returns>
/// An <see cref="ArrayList"/> of <see cref="InternetUrlToIndex"/> objects, one for
/// each link found in the content.
/// </returns>
/// <exception cref="NotSupportedException">Whenever this method is called.</exception>
/// <remarks>
/// Since a SWF document can not be converted to a string this method <b>ALWAYS</b>
/// throws a <see cref="NotSupportedException"/>.
/// </remarks>
public override ArrayList ExtractLinks(ref string content, ref InternetUrlToCrawl contentUrl)
{
throw new NotSupportedException();
}
/// <summary>
/// Checks if the Robots Exclusion Standard allows the crawler to visit a url.
/// </summary>
/// <param name="targetUrl">The url that is to be validated.</param>
/// <param name="sourceUrl">The <see cref="InternetUrlToCrawl"/> containing the targetUrl.</param>
/// <param name="robotsMeta">A <see cref="RobotsMetaTagValue"/> flag indicating the
/// restrictions posed by the robots meta tag contained in the sourceUrl.</param>
/// <returns> A <see cref="Boolean"/> value indicating whether the crawler is
/// allowed (false) or disallowed (true) to visit the target Url.</returns>
/// <remarks>This method is safe for multi-threaded operations. However only one
/// thread will be able to perform a check at any given time.
/// </remarks>
public bool FilterUrl(string targetUrl, InternetUrlToCrawl sourceUrl, RobotsMetaTagValue robotsMeta)
{
bool retVal = false; //assume that it's allowed to crawl the targetUrl
try
{
mutex.WaitOne();
//perhaphs we should use the hash code of the hostnames as keys.
string targetHost = InternetUtils.HostName(targetUrl);
string sourceHost = InternetUtils.HostName(sourceUrl);
RobotsTxtEntry robots = null;
//Do we need to fetch the robots.txt for the source Url?
if (sourceUrl.FlagFetchRobots)
{
//we must fetch the robots.txt from the source url host and update sourceUrl.
robots = FetchRobots(sourceHost);
sourceUrl.RobotsDisallowedPaths = ConcatenatePaths(robots.DisallowedPaths);
sourceUrl.FlagFetchRobots = false; //fetch it only once
//check if it exists in the Hashtable, if so update it, otherwise add it
if (robotsTable.ContainsKey(sourceHost))
{
robotsTable[sourceHost] = robots;
}
else
{
robotsTable.Add(sourceHost, robots);
}
}
else
{
//check if it exists in the Hashtable. If so check if it has expired, else just get it from InternetUrlToCrawl
if (!robotsTable.TryGetValue(sourceHost, out robots))
{
robots = new RobotsTxtEntry();
robots.DisallowedPaths = SplitPaths(sourceUrl.RobotsDisallowedPaths);
robotsTable.Add(sourceHost, robots);
}
else
{
if (robots.ExpirationDate < DateTime.Today)
{
robots = FetchRobots(sourceHost);
robotsTable[sourceHost] = robots;
}
}
}
if (targetHost != sourceHost)
{
//the target url is on a different host, we must get its robots.txt
if (!robotsTable.TryGetValue(targetHost, out robots))
{
robots = FetchRobots(targetHost);
robotsTable.Add(targetHost, robots);
}
else
{
if (robots.ExpirationDate < DateTime.Today)
{
robots = FetchRobots(targetHost);
robotsTable[targetHost] = robots;
}
}
}
if ((robotsMeta & RobotsMetaTagValue.NoFollow) > 0)
{
//if the meta tag has the NoFollow option set then we cannot crawl targetUrl
retVal = true;
}
else
{
robots = robotsTable[targetHost];
//if the DisallowedPaths is null then we can crawl targetUrl, otherwise we must check the disallowed paths
if (robots.DisallowedPaths != null)
{
for (int i = 0; i < robots.DisallowedPaths.Length; i++)
{
if (targetUrl.IndexOf(robots.DisallowedPaths[i]) != -1)
{
//we found a match. It is therefore not allowed to crawl targetUrl
retVal = true;
break; //stop searching as soon as we have a match
}
}
}
}
}
catch (Exception e)
{
if (globals.Settings.LogLevel <= CWLogLevel.LogWarning)
{
globals.FileLog.LogWarning("RobotsFilter failed to filter " + targetUrl + ": " + e.ToString());
}
}
finally
{
mutex.ReleaseMutex();
}
return(retVal);
}
/// <summary>
/// Extracts links from the contents of a PDF document.
/// </summary>
/// <param name="content">The contents of the PDF document.</param>
/// <param name="contentUrl">The url of the PDF document.</param>
/// <returns>
/// An <see cref="ArrayList"/> of <see cref="InternetUrlToIndex"/> objects, one for
/// each link found in the content.
/// </returns>
/// <exception cref="ArgumentNullException">If the input buffer is null or empty.</exception>
public override ArrayList ExtractLinks(byte[] content, ref InternetUrlToCrawl contentUrl)
{
ArrayList links = null;
if ((content == null) || (content.Length == 0))
{
throw new ArgumentNullException("content", "The input buffer cannot be empty or null.");
}
try
{
mutex.WaitOne();
string FileName = globals.AppWorkPath + Guid.NewGuid().ToString();
string pdfFileName = FileName + ".pdf";
string txtFileName = FileName + ".txt";
FileStream pdf = null;
StreamReader txt = null;
try
{
//store the pdf file
pdf = new FileStream(pdfFileName, FileMode.Create);
pdf.Write(content, 0, content.Length);
pdf.Close();
pdf = null;
//convert it to text
int success = converter.ConvertPdf2Text(pdfFileName, txtFileName);
if (success == 0)
{
txt = new StreamReader(txtFileName, encoding);
string text = txt.ReadToEnd();
txt.Close();
txt = null;
links = parser.ExtractLinks(ref text, ref contentUrl);
}
else
{
txt.Close();
txt = null;
}
}
catch (Exception ex)
{
if (pdf != null)
{
try
{
pdf.Close();
}
catch
{}
}
if (txt != null)
{
try
{
txt.Close();
}
catch
{}
}
if (globals.Settings.LogLevel <= CWLogLevel.LogInfo)
{
globals.FileLog.LogWarning("PdfParser failed to extract links from " + contentUrl.Url + ": " + ex.ToString());
}
}
finally
{
File.Delete(pdfFileName);
File.Delete(txtFileName);
}
}
catch
{
if (globals.Settings.LogLevel <= CWLogLevel.LogWarning)
{
globals.FileLog.LogWarning("PdfParser failed to extract links from " + contentUrl.Url);
}
}
finally
{
GC.Collect();
mutex.ReleaseMutex();
}
ParserEventArgs e = new ParserEventArgs(contentUrl.Url);
OnExtractLinksComplete(e);
return(links);
}
/// <summary>
/// Extracts links from the contents of a PDF document.
/// </summary>
/// <param name="content">The contents of the PDF document.</param>
/// <param name="contentUrl">The url of the PDF document.</param>
/// <returns>
/// An <see cref="ArrayList"/> of <see cref="InternetUrlToIndex"/> objects, one for
/// each link found in the content.
/// </returns>
/// <exception cref="NotSupportedException">Whenever this method is called.</exception>
/// <remarks>
/// Since a PDF document can not be converted to a string this method <b>ALWAYS</b>
/// throws a <see cref="NotSupportedException"/>.
/// </remarks>
public override ArrayList ExtractLinks(ref string content, ref InternetUrlToCrawl contentUrl)
{
throw new NotSupportedException("The ExtractLinks method of the PdfParser cannot accept a string as input.");
}
/// <summary>
/// Performs the extraction of links from a text document. It can extract simple
/// links that are separated from the rest of the text using spaces or line brakes
/// or any other delimiters. The results are returned as an <see cref="ArrayList"/>
/// of <see cref="InternetUrlToIndex"/> objects.
/// </summary>
/// <remarks>
/// Besides the parsing and extraction of Urls, ExtractLinks also performs other
/// tasks as well, such as:<br/>
/// <list type="bullet">
/// <item>
/// <description>Filtering of urls to resources of unsupported content-type, e.g. css, images, etc.</description>
/// </item>
/// <item>
/// <description>Filtering of multimple links to the same url and to the document itself.</description>
/// </item>
/// <item>
/// <description>Filtering of session id variables in dynamic Urls and limiting
/// of the number of GET variables in dynamic Urls.</description>
/// </item>
/// <item>
/// <description>Flagging of Urls according to their country domain.</description>
/// </item>
/// </list>
/// <b>Update History</b>
/// <list type="table">
/// <listheader>
/// <term>Date</term>
/// <description>Description</description>
/// </listheader>
/// <item>
/// <term>15/09/04</term>
/// <description>First release. A lot more needs to be done.</description>
/// </item>
/// </list>
/// </remarks>
/// <param name="content">The text that must be parsed for links. It is passed by
/// reference in order to reduce memory consumption.</param>
/// <param name="contentUrl">The Url from which the content comes.</param>
/// <returns>
/// An <see cref="ArrayList"/> of <see cref="InternetUrlToIndex"/> objects, one for
/// each link found in the content.
/// </returns>
public override ArrayList ExtractLinks(ref string content, ref InternetUrlToCrawl contentUrl)
{
ArrayList links = new ArrayList();
// It is important to notice that if the FlagFetchRobots of the contentUrl is
// true then the TextParser must remember this value because during the Robots
// Filtering it will become false so as not to download the robots.txt file
// every time a Url must be filtered.
//bool FlagFetchRobots = contentUrl.FlagFetchRobots;
try
{
//make sure only one thread will parse contents at a time.
//mutex.WaitOne();
if (contentUrl.FlagDomain != DomainFlagValue.MustVisit)
{
contentUrl.FlagDomain = ExtractDomainFlag(ref content);
if (contentUrl.FlagDomain != DomainFlagValue.MustVisit)
{
if (InternetUtils.HostName(contentUrl).Contains("ebay.com"))
{
contentUrl.FlagDomain = DomainFlagValue.MustVisit;
}
}
}
//perform the hyperlink matching
MatchCollection matches = hrefRegex.Matches(content);
if (matches.Count > 0)
{
string documentUrl = contentUrl.Url;
string baseUrl = BaseUrl(ref documentUrl);
byte priority = 0;
foreach (Match m in matches)
{
try
{
string url = m.Value.Trim();
url = NormalizeUrl(ref url, ref baseUrl);
priority = CleanUrlParams(ref url);
if (FilterUrl(ref url, ref documentUrl))
{
InternetUrlToIndex iurl = new InternetUrlToIndex(url);
iurl.Priority = priority;
iurl.FlagDomain = domainFilter.FilterUrl(ref url);
//[mod 24/2/05] No robots.txt checking is performed for non-greek urls
if (iurl.FlagDomain == DomainFlagValue.MustVisit)
{
iurl.FlagRobots = robotsFilter.FilterUrl(url, contentUrl, RobotsMetaTagValue.NoMeta);
}
else
{
iurl.FlagRobots = false;
}
if (!links.Contains(iurl))
{
links.Add(iurl);
}
}
}
catch
{
if (globals.Settings.LogLevel == CWLogLevel.LogInfo)
{
globals.FileLog.LogInfo("TextParser failed to parse " + m.Value);
}
continue;
}
}
}
}
catch (Exception ex)
{
if (globals.Settings.LogLevel <= CWLogLevel.LogWarning)
{
globals.FileLog.LogWarning(ex.Message);
}
}
finally
{
//mutex.ReleaseMutex();
}
//contentUrl.FlagFetchRobots = FlagFetchRobots;
ParserEventArgs e = new ParserEventArgs(contentUrl.Url);
OnExtractLinksComplete(e);
links.TrimToSize();
return(links);
}
