org.apache.lucene.benchmark.byTask.feeds.demohtml

Class HTMLParser

java.lang.Object

org.apache.lucene.benchmark.byTask.feeds.demohtml.HTMLParser

All Implemented Interfaces:

HTMLParserConstants

public class HTMLParser
extends Object
implements HTMLParserConstants

Field Summary

Fields

Modifier and Type	Field and Description
Token	jj_nt Next token.
static int	SUMMARY_LENGTH
Token	token Current token.
HTMLParserTokenManager	token_source Generated Token Manager.

Fields inherited from interface org.apache.lucene.benchmark.byTask.feeds.demohtml.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(HTMLParserTokenManager tm) Constructor with generated Token Manager.
HTMLParser(InputStream stream) Constructor with InputStream.
HTMLParser(InputStream stream, String encoding) Constructor with InputStream and supplied encoding
HTMLParser(Reader stream) Constructor.

Method Summary

Methods

Modifier and Type	Method and Description
Token	ArgValue()
void	CommentTag()
Token	Decl()
void	disable_tracing() Disable tracing.
void	enable_tracing() Enable tracing.
ParseException	generateParseException() Generate ParseException.
Properties	getMetaTags()
Token	getNextToken() Get the next Token.
Reader	getReader()
String	getSummary()
String	getTitle()
Token	getToken(int index) Get the specific Token.
void	HTMLDocument()
void	ReInit(HTMLParserTokenManager tm) Reinitialise.
void	ReInit(InputStream stream) Reinitialise.
void	ReInit(InputStream stream, String encoding) Reinitialise.
void	ReInit(Reader stream) Reinitialise.
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

Generated Token Manager.

token

public Token token

Current token.

jj_nt

public Token jj_nt

Next token.

Constructor Detail

HTMLParser

public HTMLParser(InputStream stream)

Constructor with InputStream.

HTMLParser

public HTMLParser(InputStream stream,
          String encoding)

Constructor with InputStream and supplied encoding

HTMLParser

public HTMLParser(Reader stream)

Constructor.

HTMLParser

public HTMLParser(HTMLParserTokenManager tm)

Constructor with generated Token Manager.

Method Detail

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

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)

Reinitialise.

ReInit

public void ReInit(InputStream stream,
          String encoding)

Reinitialise.

ReInit

public void ReInit(Reader stream)

Reinitialise.

ReInit

public void ReInit(HTMLParserTokenManager tm)

Reinitialise.

getNextToken

public final Token getNextToken()

Get the next Token.

getToken

public final Token getToken(int index)

Get the specific Token.

generateParseException

public ParseException generateParseException()

Generate ParseException.

enable_tracing

public final void enable_tracing()

Enable tracing.

disable_tracing

public final void disable_tracing()

Disable tracing.