edu.stanford.nlp.trees
Class DiskTreebank

java.lang.Object
  java.util.AbstractCollection<Tree>
      edu.stanford.nlp.trees.Treebank
          edu.stanford.nlp.trees.DiskTreebank

All Implemented Interfaces:

Iterable<Tree>, Collection<Tree>

public final class DiskTreebank
extends Treebank

A DiskTreebank is a Collection of s. A DiskTreebank object stores merely the information to get at a corpus of trees that is stored on disk. Access is usually via apply()'ing a TreeVisitor to each Tree in the Treebank or by using an iterator() to get an iteration over the Trees.

Author:

Christopher Manning

Constructor Summary

DiskTreebank()
Create a new DiskTreebank.

DiskTreebank(int initialCapacity)
Create a new Treebank.

DiskTreebank(int initialCapacity, TreeReaderFactory trf)
Create a new Treebank.

DiskTreebank(String encoding)
Create a new tree bank, set the encoding for file access.

DiskTreebank(TreeReaderFactory trf)
Create a new DiskTreebank.

DiskTreebank(TreeReaderFactory trf, String encoding)
Create a new DiskTreebank.

Method Summary

void	apply(TreeVisitor tp) Applies the TreeVisitor to to all trees in the Treebank.
void	clear() Empty a Treebank.
File	getCurrentFile() Return the File from which trees are currently being read by apply(), and pased to a TreePprocessor.
Iterator<Tree>	iterator() Return an Iterator over Trees in the Treebank.
void	loadPath(File path, FileFilter filt) Load trees from given directory.
static void	main(String[] args) Loads treebank and prints it.
void	transformOnRead(TreeTransformer tt) Transform all trees with this TreeTransformer as they are read.

Methods inherited from class edu.stanford.nlp.trees.Treebank

encoding, loadPath, loadPath, loadPath, loadPath, loadPath, remove, size, textualSummary, toString, transform, treeReaderFactory

Methods inherited from class java.util.AbstractCollection

add, addAll, contains, containsAll, isEmpty, removeAll, retainAll, toArray, toArray

Methods inherited from class java.lang.Object

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

Methods inherited from interface java.util.Collection

equals, hashCode

Constructor Detail

DiskTreebank

public DiskTreebank()

Create a new DiskTreebank. The trees are made with a LabeledScoredTreeReaderFactory.

Compatibility note: Until Sep 2004, this used to create a Treebank with a SimpleTreeReaderFactory, but this was changed as the old default wasn't very useful, especially to naive users.

DiskTreebank

public DiskTreebank(String encoding)

Create a new tree bank, set the encoding for file access.

Parameters:

encoding - The charset encoding to use for treebank file decoding

DiskTreebank

public DiskTreebank(TreeReaderFactory trf)

Create a new DiskTreebank.

Parameters:

trf - the factory class to be called to create a new TreeReader

DiskTreebank

public DiskTreebank(TreeReaderFactory trf,
                    String encoding)

Create a new DiskTreebank.

Parameters:

trf - the factory class to be called to create a new TreeReader

encoding - The charset encoding to use for treebank file decoding

DiskTreebank

public DiskTreebank(int initialCapacity)

Create a new Treebank. The trees are made with a LabeledScoredTreeReaderFactory.

Compatibility note: Until Sep 2004, this used to create a Treebank with a SimpleTreeReaderFactory, but this was changed as the old default wasn't very useful, especially to naive users.

Parameters:

initialCapacity - The initial size of the underlying Collection. For a DiskTreebank, this parameter is ignored.

DiskTreebank

public DiskTreebank(int initialCapacity,
                    TreeReaderFactory trf)

Create a new Treebank.

Parameters:

initialCapacity - The initial size of the underlying Collection, For a DiskTreebank, this parameter is ignored.

trf - the factory class to be called to create a new TreeReader

Method Detail

transformOnRead

public void transformOnRead(TreeTransformer tt)

Transform all trees with this TreeTransformer as they are read.

Parameters:

tt -

clear

public void clear()

Empty a Treebank.

Specified by:

clear in interface Collection<Tree>

Specified by:

clear in class Treebank

loadPath

public void loadPath(File path,
                     FileFilter filt)

Load trees from given directory. This version just records the paths to be processed, and actually processes them at apply time.

Specified by:

loadPath in class Treebank

Parameters:

path - file or directory to load from

filt - a FilenameFilter of files to load

apply

public void apply(TreeVisitor tp)

Applies the TreeVisitor to to all trees in the Treebank.

Specified by:

apply in class Treebank

Parameters:

tp - A class that can process trees.

getCurrentFile

public File getCurrentFile()

Return the File from which trees are currently being read by apply(), and pased to a TreePprocessor. This is useful if one wants to map the original file and directory structure over to a set of modified trees.

Returns:

the file that trees are currently being read from, or null if no file is currently open

iterator

public Iterator<Tree> iterator()

Return an Iterator over Trees in the Treebank. This is implemented by building per-file MemoryTreebanks for the files in the DiskTreebank. As such, it isn't as efficient as using apply().

Specified by:

iterator in interface Iterable<Tree>

Specified by:

iterator in interface Collection<Tree>

Specified by:

iterator in class AbstractCollection<Tree>

main

public static void main(String[] args)

Loads treebank and prints it. All files below the designated filePath within the given number range if any are loaded. You can normalize the trees or not (English-specific) and print trees one per line up to a certain length (for EVALB).

Usage: java edu.stanford.nlp.trees.DiskTreebank [-maxLength n|-normalize] filePath numberRanges

Parameters:

args - Array of command-line arguments