Package org.apache.lucene.demo.html

Class HTMLParser

java.lang.Object

org.apache.lucene.demo.html.HTMLParser

All Implemented Interfaces:

HTMLParserConstants

public class HTMLParser
extends Object
implements HTMLParserConstants

Field Summary

Fields

Modifier and Type	Field	Description
Token	jj_nt
boolean	lookingAhead
static int	SUMMARY_LENGTH
Token	token
HTMLParserTokenManager	token_source

Fields inherited from interface org.apache.lucene.demo.html.HTMLParserConstants

AfterEquals, ArgEquals, ArgName, ArgQuote1, ArgQuote2, ArgValue, CloseQuote1, CloseQuote2, Comment1, Comment2, CommentEnd1, CommentEnd2, CommentText1, CommentText2, DeclName, DEFAULT, Entity, EOF, HEX, LET, NUM, Punct, Quote1Text, Quote2Text, ScriptEnd, ScriptStart, ScriptText, SP, Space, TagEnd, TagName, tokenImage, WithinComment1, WithinComment2, WithinQuote1, WithinQuote2, WithinScript, WithinTag, Word

Constructor Summary

Constructors

Constructor	Description
HTMLParser(File file)	Deprecated. Use HTMLParser(FileInputStream) instead
HTMLParser(InputStream stream)
HTMLParser(InputStream stream, String encoding)
HTMLParser(Reader stream)
HTMLParser(HTMLParserTokenManager tm)

Method Summary

Modifier and Type	Method	Description
final Token	ArgValue()
final void	CommentTag()
final Token	Decl()
final void	disable_tracing()
final void	enable_tracing()
ParseException	generateParseException()
Exception	getException()
Properties	getMetaTags()
final Token	getNextToken()
Reader	getReader()
String	getSummary()
String	getTitle()
final Token	getToken(int index)
final void	HTMLDocument()
void	parse()
void	ReInit(InputStream stream)
void	ReInit(InputStream stream, String encoding)
void	ReInit(Reader stream)
void	ReInit(HTMLParserTokenManager tm)
final void	ScriptTag()
void	setException(Exception exception)
final void	Tag()

Methods inherited from class java.lang.Object

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

Field Details

SUMMARY_LENGTH

public static int SUMMARY_LENGTH

token_source

public HTMLParserTokenManager token_source

token

public Token token

jj_nt

public Token jj_nt

lookingAhead

public boolean lookingAhead

Constructor Details

HTMLParser

@Deprecated
public HTMLParser(File file)
           throws FileNotFoundException

Deprecated.

Use HTMLParser(FileInputStream) instead

Throws:

FileNotFoundException

HTMLParser

public HTMLParser(InputStream stream)

HTMLParser

public HTMLParser(InputStream stream,
 String encoding)

HTMLParser

public HTMLParser(Reader stream)

HTMLParser

public HTMLParser(HTMLParserTokenManager tm)

Method Details

getTitle

public String getTitle()
                throws IOException,
InterruptedException

Throws:

IOException

InterruptedException

getMetaTags

public Properties getMetaTags()
                       throws IOException,
InterruptedException

Throws:

IOException

InterruptedException

getSummary

public String getSummary()
                  throws IOException,
InterruptedException

Throws:

IOException

InterruptedException

getReader

public Reader getReader()
                 throws IOException

Throws:

IOException

parse

public void parse()
           throws IOException

Throws:

IOException

HTMLDocument

public final void HTMLDocument()
                        throws ParseException,
IOException

Throws:

ParseException

IOException

Tag

public final void Tag()
               throws ParseException,
IOException

Throws:

ParseException

IOException

ArgValue

public final Token ArgValue()
                     throws ParseException

Throws:

ParseException

Decl

public final Token Decl()
                 throws ParseException

Throws:

ParseException

CommentTag

public final void CommentTag()
                      throws ParseException

Throws:

ParseException

ScriptTag

public final void ScriptTag()
                     throws ParseException

Throws:

ParseException

ReInit

public void ReInit(InputStream stream)

ReInit

public void ReInit(InputStream stream,
 String encoding)

ReInit

public void ReInit(Reader stream)

ReInit

public void ReInit(HTMLParserTokenManager tm)

getNextToken

public final Token getNextToken()

getToken

public final Token getToken(int index)

generateParseException

public ParseException generateParseException()

enable_tracing

public final void enable_tracing()

disable_tracing

public final void disable_tracing()

setException

public void setException(Exception exception)

getException

public Exception getException()