marmot.tokenize.preprocess

Class WikiReader

java.lang.Object

marmot.tokenize.preprocess.WikiReader

All Implemented Interfaces:

java.util.Iterator<Pair>

public class WikiReader
extends java.lang.Object
implements java.util.Iterator<Pair>

Constructor Summary

Constructors

Constructor and Description
WikiReader(InternalReader untokenized, InternalReader tokenized, boolean expand)
WikiReader(java.lang.String untokenized_file, java.lang.String tokenized_file, boolean expand)

Method Summary

Modifier and Type	Method and Description
protected void	expandPair()
protected java.lang.String	fixLine(java.lang.String line)
boolean	hasNext()
Pair	next()
static InternalReader	openFile(java.lang.String file)
java.util.List<Pair>	readAll()
void	readNext()
protected java.lang.String	readNonEmptyLine(InternalReader reader)
void	remove()

Methods inherited from class java.lang.Object

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

Methods inherited from interface java.util.Iterator

forEachRemaining

Constructor Detail

WikiReader

public WikiReader(InternalReader untokenized,
                  InternalReader tokenized,
                  boolean expand)

WikiReader

public WikiReader(java.lang.String untokenized_file,
                  java.lang.String tokenized_file,
                  boolean expand)

Method Detail

openFile

public static InternalReader openFile(java.lang.String file)

hasNext

public boolean hasNext()

Specified by:

hasNext in interface java.util.Iterator<Pair>

fixLine

protected java.lang.String fixLine(java.lang.String line)

readNonEmptyLine

protected java.lang.String readNonEmptyLine(InternalReader reader)

readNext

public void readNext()

expandPair

protected void expandPair()

next

public Pair next()

Specified by:

next in interface java.util.Iterator<Pair>

remove

public void remove()

Specified by:

remove in interface java.util.Iterator<Pair>

readAll

public java.util.List<Pair> readAll()