org.apache.manifoldcf.crawler.connectors.webcrawler
Class WebcrawlerConnector.CanonicalizationPolicy

java.lang.Object
  org.apache.manifoldcf.crawler.connectors.webcrawler.WebcrawlerConnector.CanonicalizationPolicy

Enclosing class:

WebcrawlerConnector

protected static class WebcrawlerConnector.CanonicalizationPolicy
extends java.lang.Object

Class representing a URL regular expression match, for the purposes of determining canonicalization policy

Field Summary

protected java.util.regex.Pattern	matchPattern
protected boolean	removeAspSession
protected boolean	removeBVSession
protected boolean	removeJavaSession
protected boolean	removePhpSession
protected boolean	reorder

Constructor Summary

WebcrawlerConnector.CanonicalizationPolicy(java.util.regex.Pattern matchPattern, boolean reorder, boolean removeJavaSession, boolean removeAspSession, boolean removePhpSession, boolean removeBVSession)

Method Summary

boolean	canRemoveAspSession()
boolean	canRemoveBvSession()
boolean	canRemoveJavaSession()
boolean	canRemovePhpSession()
boolean	canReorder()
boolean	checkMatch(java.lang.String url)

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Field Detail

matchPattern

protected java.util.regex.Pattern matchPattern

reorder

protected boolean reorder

removeJavaSession

protected boolean removeJavaSession

removeAspSession

protected boolean removeAspSession

removePhpSession

protected boolean removePhpSession

removeBVSession

protected boolean removeBVSession

Constructor Detail

WebcrawlerConnector.CanonicalizationPolicy

public WebcrawlerConnector.CanonicalizationPolicy(java.util.regex.Pattern matchPattern,
                                                  boolean reorder,
                                                  boolean removeJavaSession,
                                                  boolean removeAspSession,
                                                  boolean removePhpSession,
                                                  boolean removeBVSession)

Method Detail

checkMatch

public boolean checkMatch(java.lang.String url)

canReorder

public boolean canReorder()

canRemoveJavaSession

public boolean canRemoveJavaSession()

canRemoveAspSession

public boolean canRemoveAspSession()

canRemovePhpSession

public boolean canRemovePhpSession()

canRemoveBvSession

public boolean canRemoveBvSession()