Package org.deeplearning4j.text.tokenization.tokenizer

Class BertWordPieceTokenizer

java.lang.Object

org.deeplearning4j.text.tokenization.tokenizer.BertWordPieceTokenizer

All Implemented Interfaces:

Tokenizer

Direct Known Subclasses:

BertWordPieceStreamTokenizer

public class BertWordPieceTokenizer
extends Object
implements Tokenizer

Field Summary

Fields

Modifier and Type	Field	Description
static Pattern	splitPattern

Constructor Summary

Constructors

Constructor	Description
BertWordPieceTokenizer(String tokens, NavigableMap<String,Integer> vocab, TokenPreProcess preTokenizePreProcessor, TokenPreProcess tokenPreProcess)

Method Summary

Modifier and Type	Method	Description
protected void	checkIfEmpty(Map<String,Integer> m, String candidate)
int	countTokens()	The number of tokens in the tokenizer
protected String	findLongestSubstring(NavigableMap<String,Integer> vocab, String candidate)
List<String>	getTokens()	Returns a list of all the tokens
boolean	hasMoreTokens()	An iterator for tracking whether more tokens are left in the iterator not
String	nextToken()	The next token (word usually) in the string
void	setTokenPreProcessor(TokenPreProcess tokenPreProcessor)	Set the token pre process

Methods inherited from class java.lang.Object

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

Field Detail

splitPattern

public static final Pattern splitPattern

Constructor Detail

BertWordPieceTokenizer

public BertWordPieceTokenizer(String tokens,
                              NavigableMap<String,Integer> vocab,
                              TokenPreProcess preTokenizePreProcessor,
                              TokenPreProcess tokenPreProcess)

Method Detail

hasMoreTokens

public boolean hasMoreTokens()

Description copied from interface: Tokenizer

An iterator for tracking whether more tokens are left in the iterator not

Specified by:

hasMoreTokens in interface Tokenizer

Returns:

whether there is anymore tokens to iterate over

countTokens

public int countTokens()

Description copied from interface: Tokenizer

The number of tokens in the tokenizer

Specified by:

countTokens in interface Tokenizer

Returns:

the number of tokens

nextToken

public String nextToken()

Description copied from interface: Tokenizer

The next token (word usually) in the string

Specified by:

nextToken in interface Tokenizer

Returns:

the next token in the string if any

getTokens

public List<String> getTokens()

Description copied from interface: Tokenizer

Returns a list of all the tokens

Specified by:

getTokens in interface Tokenizer

Returns:

a list of all the tokens

setTokenPreProcessor

public void setTokenPreProcessor(TokenPreProcess tokenPreProcessor)

Description copied from interface: Tokenizer

Set the token pre process

Specified by:

setTokenPreProcessor in interface Tokenizer

Parameters:

tokenPreProcessor - the token pre processor to set

findLongestSubstring

protected String findLongestSubstring(NavigableMap<String,Integer> vocab,
                                      String candidate)

checkIfEmpty

protected void checkIfEmpty(Map<String,Integer> m,
                            String candidate)