/// <summary>
/// The constructor is private so that only the class itself can create an instance.
/// </summary>
private WordsCache()
{
settings = PluginSettings.Instance();
dbProvider = DBConnectionStringProvider.Instance();
settings.DBConnectionString = dbProvider.ProvideDBConnectionString("CrawlWave.ServerPlugins.WordExtraction");
dbcon = new SqlConnection(settings.DBConnectionString);
words = new Hashtable();
stemming = Stemming.Instance();
culture = new CultureInfo("el-GR");
LoadCache();
}
/// <summary>
/// The constructor is private so that only the class itself can create an instance.
/// </summary>
private PluginSettings()
{
settings = new EbayPluginSettings();
settings.PauseBetweenOperations = false;
settings.PauseDelay = 0;
settings.UseTransactions = false;
settings.DBActionTimeout = 60;
dbProvider = DBConnectionStringProvider.Instance();
connectionString = dbProvider.ProvideDBConnectionString(Assembly.GetExecutingAssembly().GetName().Name);
LoadSettings();
}
/// <summary>
/// The constructor is private so that only the class itself can create an instance.
/// </summary>
private Globals()
{
dbProvider = DBConnectionStringProvider.Instance();
appName = "CrawlWave.ServerManager";
appPath = GetAppPath();
loadedForms = new Hashtable(8);
foreach (string formName in formNames)
{
loadedForms.Add(formName, null);
}
log = new FileEventLogger(appPath + appName + ".log", true, appName);
}
/// <summary>
/// The constructor is private so that only the class itself can create an instance.
/// </summary>
private PluginSettings()
{
settings = new USPluginSettings();
settings.Threshold = 1000;
settings.SelectionSize = 10000;
settings.SelectionMode = false;
settings.PauseBetweenOperations = false;
settings.PauseDelay = 0;
settings.UseTransactions = false;
settings.DBTimeout = 60;
dbProvider = DBConnectionStringProvider.Instance();
connectionString = dbProvider.ProvideDBConnectionString(Assembly.GetExecutingAssembly().GetName().Name);
LoadSettings();
}
/// <summary>
/// Constructs a new instance of the <see cref="PageRankPlugin"/> class.
/// </summary>
public PageRankPlugin()
{
description = "CrawlWave PageRank Plugin";
name = "CrawlWave.ServerPlugins.PageRank";
percent = 0;
version = Assembly.GetExecutingAssembly().GetName().Version.ToString();
string connectionString = DBConnectionStringProvider.Instance().ProvideDBConnectionString(name);
dbcon = new SqlConnection(connectionString);
pluginThread = null;
mustStop = false;
dataDependent = true;
enabled = true;
}
/// <summary>
/// The constructor is private so that only the class itself can create an instance.
/// </summary>
private PluginSettings()
{
settings = new INPluginSettings();
settings.Threads = 5;
settings.CleanUrls = true;
settings.CheckUrls = true;
settings.PauseBetweenOperations = false;
settings.PauseDelay = 0;
dbProvider = DBConnectionStringProvider.Instance();
rnd = new Random();
appName = Assembly.GetExecutingAssembly().GetName().Name;
inputFile = String.Empty;
outputFile = String.Empty;
LoadSettings();
}
/// <summary>
/// Create a new instance of the <see cref="EbayPlugin"/> class.
/// </summary>
public EbayPlugin()
{
mutex = new Mutex();
settings = PluginSettings.Instance();
name = "CrawlWave.ServerPlugins.Ebay";
description = "CrawlWave Ebay Plugin";
dataDependent = true;
state = PluginState.Stopped;
enabled = true;
version = Assembly.GetExecutingAssembly().GetName().Version.ToString();
percent = 0;
mustStop = false;
pluginThread = null;
dbProvider = DBConnectionStringProvider.Instance();
settings.DBConnectionString = dbProvider.ProvideDBConnectionString(name);
dbcon = new SqlConnection(settings.DBConnectionString);
regUser = new Regex("<h1>eBay My World:\\s*(?<user>[^<]*)<img", RegexOptions.Compiled | RegexOptions.IgnoreCase | RegexOptions.Multiline);
regFeedback = new Regex("Feedback score: <b>(?<feedback>\\d+)</b><span class=\"vSep\">\\|</span>Positive feedback: <b>(?<positive>\\d+\\.?\\d)%</b>", RegexOptions.Compiled | RegexOptions.IgnoreCase | RegexOptions.Multiline);
//eBay My World: c7gman<img
//Feedback score: <b>285</b><span class="vSep">|</span>Positive feedback: <b>100%</b>
}
/// <summary>
/// Create a new instance of the <see cref="WordExtractorPlugin"/> class.
/// </summary>
public WordExtractorPlugin()
{
mutex = new Mutex();
settings = PluginSettings.Instance();
name = "CrawlWave.ServerPlugins.WordExtraction";
description = "CrawlWave Word Extraction Plugin";
dataDependent = false;
state = PluginState.Stopped;
enabled = true;
version = Assembly.GetExecutingAssembly().GetName().Version.ToString();
percent = 0;
mustStop = false;
pluginThread = null;
if (settings.UseDatabase)
{
dbProvider = DBConnectionStringProvider.Instance();
settings.DBConnectionString = dbProvider.ProvideDBConnectionString(name);
dbcon = new SqlConnection(settings.DBConnectionString);
}
wordExtractor = WordExtractor.Instance();
cache = WordsCache.Instance();
backoff = new Backoff(BackoffSpeed.Slow, 30000);
}
private void PerformExtraction()
{
try
{
events.Enqueue(new EventLoggerEntry(CWLoggerEntryType.Info, DateTime.Now, "CrawlWave Word Extraction Plugin thread has started with ID 0x" + Thread.CurrentThread.GetHashCode().ToString("x4")));
//the user may have enabled database just before starting the plugin
if (settings.UseDatabase)
{
if (dbcon == null)
{
dbProvider = DBConnectionStringProvider.Instance();
settings.DBConnectionString = dbProvider.ProvideDBConnectionString(name);
dbcon = new SqlConnection(settings.DBConnectionString);
}
}
while (!mustStop)
{
//Select a page from the database, perform word extraction and store the
//results back in the database
int UrlID = 0;
string data = String.Empty;
int waitSeconds = 0;
SelectUrlForWordExtraction(out UrlID, out data);
if (UrlID != 0)
{
try
{
backoff.Reset();
SortedList words = wordExtractor.ExtractWords(data);
if (words.Count != 0)
{
//add all the words to the database if they don't exist already
string word = String.Empty;
short word_count = 0;
int word_id = -1;
foreach (DictionaryEntry de in words)
{
word = (string)de.Key;
cache.AddStemmedWord(word);
}
//remove all the old words related to this url from the database
RemoveUrlWords(UrlID);
//now add relationships between the url and its words
foreach (DictionaryEntry d in words)
{
word = (string)d.Key;
word_count = (short)((int)d.Value);
word_id = cache[word];
AddUrlWord(UrlID, word_id, word_count);
}
}
UpdateUrlDataLastProcess(UrlID);
}
catch (Exception e)
{
events.Enqueue(new EventLoggerEntry(CWLoggerEntryType.Warning, DateTime.Now, "WordExtractionPlugin failed to extract words from Url with ID " + UrlID.ToString() + ": " + e.ToString()));
continue;
}
}
else
{
waitSeconds = backoff.Next() / 1000;
for (int i = 0; i < waitSeconds; i++)
{
Thread.Sleep(1000);
if (mustStop)
{
break;
}
}
}
Report();
if (settings.PauseBetweenOperations)
{
waitSeconds = PauseInSeconds();
for (int i = 0; i < waitSeconds; i++)
{
Thread.Sleep(1000);
if (mustStop)
{
break;
}
}
}
}
}
catch (ThreadAbortException)
{
//The thread was asked to abort, which means it must return at once
return;
}
catch (ThreadInterruptedException)
{
//The thread has been asked to Join. We have nothing to do but return.
return;
}
finally
{
events.Enqueue(new EventLoggerEntry(CWLoggerEntryType.Info, DateTime.Now, Thread.CurrentThread.Name + " has stopped."));
}
}
