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

Fields

Modifier and Type	Field and Description
protected boolean	lowercasing
protected java.util.regex.Pattern	matchPattern
protected boolean	removeAspSession
protected boolean	removeBVSession
protected boolean	removeJavaSession
protected boolean	removePhpSession
protected boolean	reorder

Constructor Summary

Constructors

Constructor and Description
CanonicalizationPolicy(java.util.regex.Pattern matchPattern, boolean reorder, boolean removeJavaSession, boolean removeAspSession, boolean removePhpSession, boolean removeBVSession, boolean lowercasing)

Method Summary

Modifier and Type	Method and Description
boolean	canLowercase()
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 final java.util.regex.Pattern matchPattern

reorder

protected final boolean reorder

removeJavaSession

protected final boolean removeJavaSession

removeAspSession

protected final boolean removeAspSession

removePhpSession

protected final boolean removePhpSession

removeBVSession

protected final boolean removeBVSession

lowercasing

protected final boolean lowercasing

Constructor Detail

CanonicalizationPolicy

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

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()

canLowercase

public boolean canLowercase()