ORG.oclc.resources.html
Class HTMLResource

java.lang.Object
  |
  +--ORG.oclc.resources.html.HTMLResource

public class HTMLResource

extends Object

implements NetworkResource

Field Summary

protected boolean	allContentChecksAreDone
protected String	cleanContent
protected boolean	cleanContentIsSet
protected String	closeTag
protected boolean	containsCode
protected boolean	containsCodeIsSet
protected boolean	containsForm
protected boolean	containsFormIsSet
protected boolean	containsFrames
protected boolean	containsFramesIsSet
protected boolean	containsImageMap
protected boolean	containsImageMapIsSet
protected boolean	containsLayers
protected boolean	containsLayersIsSet
protected boolean	containsMetaTags
protected boolean	containsMetaTagsIsSet
protected boolean	containsNewsCues
protected boolean	containsNewsCuesIsSet
protected boolean	containsStyles
protected boolean	containsStylesIsSet
protected boolean	containsTechnicalCues
protected boolean	containsTechnicalCuesIsSet
protected boolean	containsTemporalCues
protected boolean	containsTemporalCuesIsSet
protected boolean	containsXML
protected boolean	containsXMLIsSet
protected String	content
protected String	copyrightFromCRText
protected boolean	copyRightIsSet
protected String	dateFromCRText
protected Vector	descriptions
protected boolean	descriptionsIsSet
protected String	endOfOpenTag
protected String	excerpt
protected boolean	excerptIsSet
protected Vector	languages
protected boolean	languagesIsSet
protected int	linkRatio
protected boolean	linkRatioIsSet
protected Vector	links
protected boolean	looksLike404
protected boolean	looksLike404IsSet
protected String	lowerCaseContent
protected boolean	lowerCaseContentIsSet
protected Vector	metaTags
protected int	noTagsLength
protected int	noTagsNoAnchorsLength
protected Vector	oneoffs
protected String	openTag
protected boolean	processedTagsIsSet
protected String	pubFromCRText
protected String	rdf
protected String	startOfOpenTag
protected String	title
protected boolean	titleIsSet
protected Vector	xmlTags

Constructor Summary

HTMLResource(String doc)

Method Summary

String	cleanChunk(String chunk, int len)
boolean	containsCode()
boolean	containsForm()
boolean	containsFrames()
boolean	containsImageMap()
boolean	containsLayers()
boolean	containsMetaTags()
boolean	containsNewsCues()
boolean	containsPattern(String uncompiledPattern, int limit)
boolean	containsStyles()
boolean	containsTechnicalCues()
boolean	containsTemporalCues()
boolean	containsXML()
void	doAllContentChecks()
String	getCleanContent()
String	getContent()
String	getCopyrightFromText()
Vector	getDescriptions() Returns the content attributes from meta tags that have a name or http-equiv attribute that contains "description".
String	getExcerpt()
String	getExcerpt(boolean refresh)
String	getExcerpt(int limit)
String	getExcerpt(int limit, boolean refresh)
Vector	getLanguages() Returns the content attributes from meta tags that have a name or http-equiv attribute that contains "language".
int	getLinkRatio()
Vector	getLinks()
String	getLowerCaseContent()
Vector	getMetaTags()
Vector	getOneoffs()
String	getPublisherFromText()
String	getRDF()
String	getTitle()
Vector	getXMLTags()
boolean	looksLike404()
static void	main(String[] args)
void	processTags()
protected void	setLowerCaseContent()
String	toString()

Methods inherited from class java.lang.Object

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

Field Detail

title

protected String title

titleIsSet

protected boolean titleIsSet

copyRightIsSet

protected boolean copyRightIsSet

pubFromCRText

protected String pubFromCRText

dateFromCRText

protected String dateFromCRText

copyrightFromCRText

protected String copyrightFromCRText

excerpt

protected String excerpt

excerptIsSet

protected boolean excerptIsSet

content

protected String content

cleanContent

protected String cleanContent

cleanContentIsSet

protected boolean cleanContentIsSet

lowerCaseContent

protected String lowerCaseContent

lowerCaseContentIsSet

protected boolean lowerCaseContentIsSet

noTagsLength

protected int noTagsLength

noTagsNoAnchorsLength

protected int noTagsNoAnchorsLength

linkRatio

protected int linkRatio

linkRatioIsSet

protected boolean linkRatioIsSet

allContentChecksAreDone

protected boolean allContentChecksAreDone

containsMetaTagsIsSet

protected boolean containsMetaTagsIsSet

containsMetaTags

protected boolean containsMetaTags

containsXMLIsSet

protected boolean containsXMLIsSet

containsXML

protected boolean containsXML

containsFormIsSet

protected boolean containsFormIsSet

containsForm

protected boolean containsForm

containsFramesIsSet

protected boolean containsFramesIsSet

containsFrames

protected boolean containsFrames

containsImageMapIsSet

protected boolean containsImageMapIsSet

containsImageMap

protected boolean containsImageMap

containsLayersIsSet

protected boolean containsLayersIsSet

containsLayers

protected boolean containsLayers

containsStylesIsSet

protected boolean containsStylesIsSet

containsStyles

protected boolean containsStyles

containsTemporalCuesIsSet

protected boolean containsTemporalCuesIsSet

containsTemporalCues

protected boolean containsTemporalCues

containsNewsCuesIsSet

protected boolean containsNewsCuesIsSet

containsNewsCues

protected boolean containsNewsCues

containsTechnicalCuesIsSet

protected boolean containsTechnicalCuesIsSet

containsTechnicalCues

protected boolean containsTechnicalCues

containsCodeIsSet

protected boolean containsCodeIsSet

containsCode

protected boolean containsCode

looksLike404IsSet

protected boolean looksLike404IsSet

looksLike404

protected boolean looksLike404

processedTagsIsSet

protected boolean processedTagsIsSet

rdf

protected String rdf

oneoffs

protected Vector oneoffs

links

protected Vector links

metaTags

protected Vector metaTags

xmlTags

protected Vector xmlTags

descriptions

protected Vector descriptions

descriptionsIsSet

protected boolean descriptionsIsSet

languages

protected Vector languages

languagesIsSet

protected boolean languagesIsSet

startOfOpenTag

protected String startOfOpenTag

endOfOpenTag

protected String endOfOpenTag

openTag

protected String openTag

closeTag

protected String closeTag

Constructor Detail

HTMLResource

public HTMLResource(String doc)

Method Detail

getTitle

public String getTitle()

Specified by:

getTitle in interface NetworkResource

getPublisherFromText

public String getPublisherFromText()

Specified by:

getPublisherFromText in interface NetworkResource

getCopyrightFromText

public String getCopyrightFromText()

Specified by:

getCopyrightFromText in interface NetworkResource

getExcerpt

public String getExcerpt(boolean refresh)

getExcerpt

public String getExcerpt()

Specified by:

getExcerpt in interface NetworkResource

getExcerpt

public String getExcerpt(int limit,
                         boolean refresh)

getExcerpt

public String getExcerpt(int limit)

Specified by:

getExcerpt in interface NetworkResource

getContent

public String getContent()

Specified by:

getContent in interface NetworkResource

setLowerCaseContent

protected void setLowerCaseContent()

getLowerCaseContent

public String getLowerCaseContent()

Specified by:

getLowerCaseContent in interface NetworkResource

cleanChunk

public String cleanChunk(String chunk,
                         int len)

getCleanContent

public String getCleanContent()

Specified by:

getCleanContent in interface NetworkResource

getLinkRatio

public int getLinkRatio()

Specified by:

getLinkRatio in interface NetworkResource

toString

public String toString()

Specified by:

toString in interface NetworkResource

Overrides:

toString in class Object

containsCode

public boolean containsCode()

containsTechnicalCues

public boolean containsTechnicalCues()

containsTemporalCues

public boolean containsTemporalCues()

containsNewsCues

public boolean containsNewsCues()

containsMetaTags

public boolean containsMetaTags()

containsXML

public boolean containsXML()

containsForm

public boolean containsForm()

containsFrames

public boolean containsFrames()

containsImageMap

public boolean containsImageMap()

containsLayers

public boolean containsLayers()

containsStyles

public boolean containsStyles()

looksLike404

public boolean looksLike404()

containsPattern

public boolean containsPattern(String uncompiledPattern,
                               int limit)

doAllContentChecks

public void doAllContentChecks()

processTags

public void processTags()

getMetaTags

public Vector getMetaTags()

getXMLTags

public Vector getXMLTags()

getOneoffs

public Vector getOneoffs()

getLinks

public Vector getLinks()

getRDF

public String getRDF()

main

public static void main(String[] args)

getDescriptions

public Vector getDescriptions()

Returns the content attributes from meta tags that have a name or http-equiv attribute that contains "description".

getLanguages

public Vector getLanguages()

Returns the content attributes from meta tags that have a name or http-equiv attribute that contains "language".