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

Nested Class Summary

(package private) static class	HTMLParser.JJCalls
private static class	HTMLParser.LookaheadSuccess
private class	HTMLParser.MyPipedInputStream

Field Summary

(package private) boolean	afterSpace
(package private) boolean	afterTag
(package private) java.lang.String	currentMetaContent
(package private) java.lang.String	currentMetaTag
(package private) java.lang.String	eol
(package private) boolean	inMetaTag
(package private) boolean	inStyle
(package private) boolean	inTitle
private HTMLParser.JJCalls[]	jj_2_rtns
private int	jj_endpos
private java.util.Vector	jj_expentries
private int[]	jj_expentry
private int	jj_gc
private int	jj_gen
(package private) SimpleCharStream	jj_input_stream
private int	jj_kind
private int	jj_la
private int[]	jj_la1
private static int[]	jj_la1_0
private Token	jj_lastpos
private int[]	jj_lasttokens
private HTMLParser.LookaheadSuccess	jj_ls
Token	jj_nt
private int	jj_ntk
private boolean	jj_rescan
private Token	jj_scanpos
private boolean	jj_semLA
(package private) int	length
boolean	lookingAhead
(package private) java.util.Properties	metaTags
(package private) java.io.Reader	pipeIn
private HTMLParser.MyPipedInputStream	pipeInStream
(package private) java.io.Writer	pipeOut
private java.io.PipedOutputStream	pipeOutStream
(package private) java.lang.StringBuffer	summary
static int	SUMMARY_LENGTH
(package private) java.lang.StringBuffer	title
(package private) boolean	titleComplete
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, LET, NUM, Punct, Quote1Text, Quote2Text, ScriptEnd, ScriptStart, ScriptText, SP, Space, TagEnd, TagName, tokenImage, WithinComment1, WithinComment2, WithinQuote1, WithinQuote2, WithinScript, WithinTag, Word

Constructor Summary

HTMLParser(java.io.File file)

HTMLParser(HTMLParserTokenManager tm)

HTMLParser(java.io.InputStream stream)

HTMLParser(java.io.Reader stream)

Method Summary

(package private) void	addMetaTag()
(package private) void	addSpace()
(package private) void	addText(java.lang.String text)
(package private) void	addToSummary(java.lang.String text)
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()
private boolean	jj_2_1(int xla)
private boolean	jj_2_2(int xla)
private boolean	jj_3_1()
private boolean	jj_3_2()
private void	jj_add_error_token(int kind, int pos)
private Token	jj_consume_token(int kind)
private static void	jj_la1_0()
private int	jj_ntk()
private void	jj_rescan_token()
private void	jj_save(int index, int xla)
private boolean	jj_scan_token(int kind)
void	ReInit(HTMLParserTokenManager tm)
void	ReInit(java.io.InputStream stream)
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

title

java.lang.StringBuffer title

summary

java.lang.StringBuffer summary

metaTags

java.util.Properties metaTags

currentMetaTag

java.lang.String currentMetaTag

currentMetaContent

java.lang.String currentMetaContent

length

int length

titleComplete

boolean titleComplete

inTitle

boolean inTitle

inMetaTag

boolean inMetaTag

inStyle

boolean inStyle

afterTag

boolean afterTag

afterSpace

boolean afterSpace

eol

java.lang.String eol

pipeIn

java.io.Reader pipeIn

pipeOut

java.io.Writer pipeOut

pipeInStream

private HTMLParser.MyPipedInputStream pipeInStream

pipeOutStream

private java.io.PipedOutputStream pipeOutStream

token_source

public HTMLParserTokenManager token_source

jj_input_stream

SimpleCharStream jj_input_stream

token

public Token token

jj_nt

public Token jj_nt

jj_ntk

private int jj_ntk

jj_scanpos

private Token jj_scanpos

jj_lastpos

private Token jj_lastpos

jj_la

private int jj_la

lookingAhead

public boolean lookingAhead

jj_semLA

private boolean jj_semLA

jj_gen

private int jj_gen

jj_la1

private final int[] jj_la1

jj_la1_0

private static int[] jj_la1_0

jj_2_rtns

private final HTMLParser.JJCalls[] jj_2_rtns

jj_rescan

private boolean jj_rescan

jj_gc

private int jj_gc

jj_ls

private final HTMLParser.LookaheadSuccess jj_ls

jj_expentries

private java.util.Vector jj_expentries

jj_expentry

private int[] jj_expentry

jj_kind

private int jj_kind

jj_lasttokens

private int[] jj_lasttokens

jj_endpos

private int jj_endpos

Constructor Detail

HTMLParser

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

Throws:

java.io.FileNotFoundException

HTMLParser

public HTMLParser(java.io.InputStream stream)

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

addToSummary

void addToSummary(java.lang.String text)

addText

void addText(java.lang.String text)
       throws java.io.IOException

Throws:

java.io.IOException

addMetaTag

void addMetaTag()
          throws java.io.IOException

Throws:

java.io.IOException

addSpace

void addSpace()
        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

jj_2_1

private final boolean jj_2_1(int xla)

jj_2_2

private final boolean jj_2_2(int xla)

jj_3_1

private final boolean jj_3_1()

jj_3_2

private final boolean jj_3_2()

jj_la1_0

private static void jj_la1_0()

ReInit

public void ReInit(java.io.InputStream stream)

ReInit

public void ReInit(java.io.Reader stream)

ReInit

public void ReInit(HTMLParserTokenManager tm)

jj_consume_token

private final Token jj_consume_token(int kind)
                              throws ParseException

Throws:

ParseException

jj_scan_token

private final boolean jj_scan_token(int kind)

getNextToken

public final Token getNextToken()

getToken

public final Token getToken(int index)

jj_ntk

private final int jj_ntk()

jj_add_error_token

private void jj_add_error_token(int kind,
                                int pos)

generateParseException

public ParseException generateParseException()

enable_tracing

public final void enable_tracing()

disable_tracing

public final void disable_tracing()

jj_rescan_token

private final void jj_rescan_token()

jj_save

private final void jj_save(int index,
                           int xla)