org.apache.lucene.demo.html
Class HTMLParser

java.lang.Object
  extended by org.apache.lucene.demo.html.HTMLParser
All Implemented Interfaces:
HTMLParserConstants

public class HTMLParser
extends Object
implements HTMLParserConstants


Field Summary
 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
HTMLParser(File file)
          Deprecated. Use HTMLParser(FileInputStream) instead
HTMLParser(HTMLParserTokenManager tm)
           
HTMLParser(InputStream stream)
           
HTMLParser(InputStream stream, String encoding)
           
HTMLParser(Reader stream)
           
 
Method Summary
 Token ArgValue()
           
 void CommentTag()
           
 Token Decl()
           
 void disable_tracing()
           
 void enable_tracing()
           
 ParseException generateParseException()
           
 Properties getMetaTags()
           
 Token getNextToken()
           
 Reader getReader()
           
 String getSummary()
           
 String getTitle()
           
 Token getToken(int index)
           
 void HTMLDocument()
           
 void ReInit(HTMLParserTokenManager tm)
           
 void ReInit(InputStream stream)
           
 void ReInit(InputStream stream, String encoding)
           
 void ReInit(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(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 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)

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()


Copyright © 2000-2010 Apache Software Foundation. All Rights Reserved.