org.cdlib.xtf.textEngine
Class XtfDocNumMap

Object
  XtfDocNumMap

All Implemented Interfaces:

DocNumMap

public class XtfDocNumMap
extends Object
implements DocNumMap

Used to map chunk indexes to the corresponding document index, and vice-versa. Only performs the load when necessary (typically dynaXML uses the DocNumMap, while crossQuery doesn't.)

Author:

Martin Haye

Field Summary

private int	chunkOverlap Number of words one chunk overlaps with the next
private int	chunkSize Max number of words in a chunk
private int[]	docNums Array of indexes, one for each docInfo chunk
private int	high Used in binary searching
private int	low Used in binary searching
private int	nDocs Total number of docInfo chunks found
private int	prevNum Caches result of previous scan, used for speed
private IndexReader	reader Where to get the data from

Constructor Summary

XtfDocNumMap(IndexReader reader, int chunkSize, int chunkOverlap)
Make a map for the given reader.

Method Summary

int	getChunkOverlap() Get the number of words one chunk overlaps with the next
int	getChunkSize() Get the max number of words per chunk
int	getDocCount() Return a count of the number of documents (not chunks) in the index.
int	getDocNum(int chunkNumber) Given a chunk number, return the corresponding document number that it is part of.
int	getFirstChunk(int docNum) Given a document number, this method returns the number of its first chunk.
int	getLastChunk(int docNum) Given a document number, this method returns the number of its last chunk.
private void	load()
private void	scan(int num) Perform a binary search looking for the given number.

Methods inherited from class Object

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

Field Detail

reader

private IndexReader reader

Where to get the data from

chunkSize

private int chunkSize

Max number of words in a chunk

chunkOverlap

private int chunkOverlap

Number of words one chunk overlaps with the next

nDocs

private int nDocs

Total number of docInfo chunks found

docNums

private int[] docNums

Array of indexes, one for each docInfo chunk

prevNum

private int prevNum

Caches result of previous scan, used for speed

low

private int low

Used in binary searching

high

private int high

Used in binary searching

Constructor Detail

XtfDocNumMap

public XtfDocNumMap(IndexReader reader,
                    int chunkSize,
                    int chunkOverlap)
             throws IOException

Make a map for the given reader. This reads in all the docInfo chunks to determine the range of text chunks for each document.

Throws:

IOException

Method Detail

load

private void load()

getChunkSize

public int getChunkSize()

Get the max number of words per chunk

Specified by:

getChunkSize in interface DocNumMap

getChunkOverlap

public int getChunkOverlap()

Get the number of words one chunk overlaps with the next

Specified by:

getChunkOverlap in interface DocNumMap

getDocCount

public final int getDocCount()

Return a count of the number of documents (not chunks) in the index.

Specified by:

getDocCount in interface DocNumMap

getDocNum

public final int getDocNum(int chunkNumber)

Given a chunk number, return the corresponding document number that it is part of. Note that like all Lucene indexes, this is ephemeral and only applies to the given reader. If not found, returns -1; this can basically only happen if the chunk number is greater than all document numbers.

Specified by:

getDocNum in interface DocNumMap

Parameters:

chunkNumber - Chunk number to translate

Returns:

Document index, or -1 if no match.

getFirstChunk

public final int getFirstChunk(int docNum)

Given a document number, this method returns the number of its first chunk.

Specified by:

getFirstChunk in interface DocNumMap

getLastChunk

public final int getLastChunk(int docNum)

Given a document number, this method returns the number of its last chunk.

Specified by:

getLastChunk in interface DocNumMap

scan

private void scan(int num)

Perform a binary search looking for the given number. On exit, the 'low' and 'high' member variables will be indexes into the array that bracket the value.

Parameters:

num - The number to look for.