org.apache.lenya.lucene.html
Class HTMLParser

java.lang.Object
  org.apache.lenya.lucene.html.HTMLParser

All Implemented Interfaces:

HTMLParserConstants

public class HTMLParser
extends java.lang.Object
implements HTMLParserConstants

HTML Parser

Field Summary

Token	jj_nt
boolean	lookingAhead
static int	SUMMARY_LENGTH
Token	token
HTMLParserTokenManager	token_source

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

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

Constructor Summary

HTMLParser(java.io.File file)
Creates a new HTMLParser object.

HTMLParser(HTMLParserTokenManager tm)
Creates a new HTMLParser object.

HTMLParser(java.io.InputStream stream)
Creates a new HTMLParser object.

HTMLParser(java.io.Reader stream)
Creates a new HTMLParser object.

Method Summary

Token	ArgValue() DOCUMENT ME!
void	CommentTag() DOCUMENT ME!
Token	Decl() DOCUMENT ME!
void	disable_tracing() DOCUMENT ME!
void	enable_tracing() DOCUMENT ME!
ParseException	generateParseException() DOCUMENT ME!
java.lang.String	getKeywords() Get keywords
Token	getNextToken() DOCUMENT ME!
java.io.Reader	getReader() DOCUMENT ME!
java.lang.String	getSummary() DOCUMENT ME!
java.lang.String	getTitle() DOCUMENT ME!
Token	getToken(int index) DOCUMENT ME!
void	HTMLDocument() DOCUMENT ME!
void	ReInit(HTMLParserTokenManager tm) DOCUMENT ME!
void	ReInit(java.io.InputStream stream) DOCUMENT ME!
void	ReInit(java.io.Reader stream) DOCUMENT ME!
void	Tag() DOCUMENT ME!

Methods inherited from class java.lang.Object

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

Field Detail

SUMMARY_LENGTH

public static final int SUMMARY_LENGTH

See Also:

Constant Field Values

token_source

public HTMLParserTokenManager token_source

token

public Token token

jj_nt

public Token jj_nt

lookingAhead

public boolean lookingAhead

Constructor Detail

HTMLParser

public HTMLParser(java.io.File file)
           throws java.io.FileNotFoundException

Creates a new HTMLParser object.

Parameters:

file - DOCUMENT ME!

Throws:

java.io.FileNotFoundException - DOCUMENT ME!

HTMLParser

public HTMLParser(java.io.InputStream stream)

Creates a new HTMLParser object.

Parameters:

stream - DOCUMENT ME!

HTMLParser

public HTMLParser(java.io.Reader stream)

Creates a new HTMLParser object.

Parameters:

stream - DOCUMENT ME!

HTMLParser

public HTMLParser(HTMLParserTokenManager tm)

Creates a new HTMLParser object.

Parameters:

tm - DOCUMENT ME!

Method Detail

getTitle

public java.lang.String getTitle()
                          throws java.io.IOException,
                                 java.lang.InterruptedException

DOCUMENT ME!

Returns:

DOCUMENT ME!

Throws:

java.io.IOException - DOCUMENT ME!

java.lang.InterruptedException - DOCUMENT ME!

getKeywords

public java.lang.String getKeywords()
                             throws java.io.IOException,
                                    java.lang.InterruptedException

Get keywords

Returns:

keywords

Throws:

java.io.IOException - DOCUMENT ME!

java.lang.InterruptedException - DOCUMENT ME!

getSummary

public java.lang.String getSummary()
                            throws java.io.IOException,
                                   java.lang.InterruptedException

DOCUMENT ME!

Returns:

DOCUMENT ME!

Throws:

java.io.IOException - DOCUMENT ME!

java.lang.InterruptedException - DOCUMENT ME!

getReader

public java.io.Reader getReader()
                         throws java.io.IOException

DOCUMENT ME!

Returns:

DOCUMENT ME!

Throws:

java.io.IOException - DOCUMENT ME!

HTMLDocument

public final void HTMLDocument()
                        throws ParseException,
                               java.io.IOException

DOCUMENT ME!

Throws:

ParseException - DOCUMENT ME!

java.io.IOException - DOCUMENT ME!

Tag

public final void Tag()
               throws ParseException,
                      java.io.IOException

DOCUMENT ME!

Throws:

ParseException - DOCUMENT ME!

java.io.IOException - DOCUMENT ME!

ArgValue

public final Token ArgValue()
                     throws ParseException

DOCUMENT ME!

Returns:

DOCUMENT ME!

Throws:

ParseException - DOCUMENT ME!

java.lang.Error - DOCUMENT ME!

Decl

public final Token Decl()
                 throws ParseException

DOCUMENT ME!

Returns:

DOCUMENT ME!

Throws:

ParseException - DOCUMENT ME!

java.lang.Error - DOCUMENT ME!

CommentTag

public final void CommentTag()
                      throws ParseException

DOCUMENT ME!

Throws:

ParseException - DOCUMENT ME!

ReInit

public void ReInit(java.io.InputStream stream)

DOCUMENT ME!

Parameters:

stream - DOCUMENT ME!

ReInit

public void ReInit(java.io.Reader stream)

DOCUMENT ME!

Parameters:

stream - DOCUMENT ME!

ReInit

public void ReInit(HTMLParserTokenManager tm)

DOCUMENT ME!

Parameters:

tm - DOCUMENT ME!

getNextToken

public final Token getNextToken()

DOCUMENT ME!

Returns:

DOCUMENT ME!

getToken

public final Token getToken(int index)

DOCUMENT ME!

Parameters:

index - DOCUMENT ME!

Returns:

DOCUMENT ME!

generateParseException

public final ParseException generateParseException()

DOCUMENT ME!

Returns:

DOCUMENT ME!

enable_tracing

public final void enable_tracing()

DOCUMENT ME!

disable_tracing

public final void disable_tracing()

DOCUMENT ME!