private HtmlParser parser; //This will be used to extract links after the conversion
#endregion Fields
#region Constructors
/// <summary>
/// The constructor is private so that only the class itself can create an instance.
/// </summary>
private SwfParser()
{
//Initialize the synchronization mechanism
mutex=new Mutex();
//Initialize the Encoding
encoding = Encoding.UTF8;//GetEncoding("ISO-8859-7");
//Initialize the converters and parsers
converter = new CSwf2HtmlConverterClass();
parser = HtmlParser.Instance();
//Get a reference to the global variables and application settings
globals = Globals.Instance();
}
/// <summary>
/// Constructs a new istance of the <see cref="Crawler"/> class and initializes its
/// properties with the default values. There should be only one instance of Crawler
/// </summary>
public Crawler()
{
//first of all get a reference to the global variables because they are needed
//in order to initialize some variables.
globals = Globals.Instance();
mustStop = false;
stopping = false;
state = CrawlerState.Stopped;
stats = new long[10] {0,0,0,0,0,0,0,0,0,0};
numThreads = (int)globals.Settings.ConnectionSpeed;
runningThreads = 0;
sendResultsThread = null;
synchronizeThread = null;
crawlingThreads = null;
syncBackOff = new ExponentialBackoff(BackoffSpeed.Declining);
downloadBackOff = new ExponentialBackoff(BackoffSpeed.Fast);
urlsToCrawl = new Queue();
resultFileNames = new Queue();
crawledUrls = new ArrayList();
queueSize = 0;
dataFileName = String.Empty;
defaultEncoding = Encoding.GetEncoding("ISO-8859-7");
htmlParser = HtmlParser.Instance();
textParser = TextParser.Instance();
pdfParser = PdfParser.Instance();
swfParser = SwfParser.Instance();
robotsFilter = RobotsFilter.Instance();
domainFilter = DomainFilter.Instance();
hostRequestFilter =HostRequestFilter.Instance();
hostBanFilter = HostBanFilter.Instance();
proxy = WebServiceProxy.Instance();
}
/// <summary>
/// Constructs a new istance of the <see cref="Crawler"/> class and initializes its
/// properties with the default values. The constructor is private so that only the
/// class itself can create an instance.
/// </summary>
private Crawler()
{
//first of all get a reference to the global variables because they are needed
//in order to initialize some variables.
globals = Globals.Instance();
mustStop = false;
stopping = false;
state = CrawlerState.Stopped;
stats = new long[10] {0,0,0,0,0,0,0,0,0,0};
numThreads = (int)globals.Settings.ConnectionSpeed;
runningThreads = 0;
//sendResultsThread = null;
synchronizeThread = null;
crawlingThreads = null;
syncBackOff = new Backoff(BackoffSpeed.Declining, 30000);
downloadBackOff = new Backoff(BackoffSpeed.Fast);
urlsToCrawl = new Queue();
resultFileNames = new Queue();
crawledUrls = new ArrayList();
queueSize = 0;
dataFileName = String.Empty;
defaultEncoding = Encoding.GetEncoding("ISO-8859-7");
defaultGreekEncoding = Encoding.GetEncoding(1253);
contentRegex = new Regex("<meta\\s*http-equiv=([^>])*charset\\s*=\\s*([^>])*(utf-7|utf-8|utf-16|windows-1253)([^>])*>",RegexOptions.CultureInvariant|RegexOptions.Multiline|RegexOptions.IgnoreCase|RegexOptions.Compiled);
htmlParser = HtmlParser.Instance();
textParser = TextParser.Instance();
pdfParser = PdfParser.Instance();
swfParser = SwfParser.Instance();
nullParser = NullParser.Instance();
robotsFilter = RobotsFilter.Instance();
robotsFilter.LoadEntries();
domainFilter = DomainFilter.Instance();
hostRequestFilter = HostRequestFilter.Instance();
hostBanFilter = HostBanFilter.Instance();
//proxy = WebServiceProxy.Instance();
proxy = CrawlWaveServerProxy.Instance(globals);
}
/// <summary>
/// Provides a global access point for the single instance of the <see cref="HtmlParser"/>
/// class.
/// </summary>
/// <returns>A reference to the single instance of <see cref="HtmlParser"/>.</returns>
public static HtmlParser Instance()
{
if (instance==null)
{
//Make sure the call is thread-safe. We cannot use the private mutex since
//it hasn't yet been initialized - it gets initialized in the constructor.
Mutex imutex=new Mutex();
imutex.WaitOne();
if( instance == null )
{
instance = new HtmlParser();
}
imutex.Close();
}
return instance;
}