org.apache.lucene.demo.html

Class HTMLParser

java.lang.Object

org.apache.lucene.demo.html.HTMLParser

All Implemented Interfaces:

HTMLParserConstants

public class HTMLParser
extends java.lang.Object
implements HTMLParserConstants

Field Summary

Fields

Modifier and Type	Field and 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 and Description
HTMLParser(java.io.File file) Deprecated. Use HTMLParser(FileInputStream) instead
HTMLParser(HTMLParserTokenManager tm)
HTMLParser(java.io.InputStream stream)
HTMLParser(java.io.InputStream stream, java.lang.String encoding)
HTMLParser(java.io.Reader stream)

Method Summary

Methods

Modifier and Type	Method and Description
Token	ArgValue()
void	CommentTag()
Token	Decl()
void	disable_tracing()
void	enable_tracing()
ParseException	generateParseException()
java.util.Properties	getMetaTags()
Token	getNextToken()
java.io.Reader	getReader()
java.lang.String	getSummary()
java.lang.String	getTitle()
Token	getToken(int index)
void	HTMLDocument()
void	ReInit(HTMLParserTokenManager tm)
void	ReInit(java.io.InputStream stream)
void	ReInit(java.io.InputStream stream, java.lang.String encoding)
void	ReInit(java.io.Reader stream)
void	ScriptTag()
void	Tag()

Methods inherited from class java.lang.Object

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

Field Detail

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 Detail

HTMLParser

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

Deprecated. Use HTMLParser(FileInputStream) instead

Throws:

java.io.FileNotFoundException

HTMLParser

public HTMLParser(java.io.InputStream stream)

HTMLParser

public HTMLParser(java.io.InputStream stream,
          java.lang.String encoding)

HTMLParser

public HTMLParser(java.io.Reader stream)

HTMLParser

public HTMLParser(HTMLParserTokenManager tm)

Method Detail

getTitle

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

Throws:

java.io.IOException

java.lang.InterruptedException

getMetaTags

public java.util.Properties getMetaTags()
                                 throws java.io.IOException,
                                        java.lang.InterruptedException

Throws:

java.io.IOException

java.lang.InterruptedException

getSummary

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

Throws:

java.io.IOException

java.lang.InterruptedException

getReader

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

Throws:

java.io.IOException

HTMLDocument

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

Throws:

ParseException

java.io.IOException

Tag

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

Throws:

ParseException

java.io.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(java.io.InputStream stream)

ReInit

public void ReInit(java.io.InputStream stream,
          java.lang.String encoding)

ReInit

public void ReInit(java.io.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()