Package org.anchoranalysis.plugin.io.bean.file.copy.naming.cluster

Class ClusterByTimestamp

Object

AnchorBean<CopyFilesNaming<ClusterMembership>>

CopyFilesNaming<ClusterMembership>

ClusterByTimestamp

public class ClusterByTimestamp
extends CopyFilesNaming<ClusterMembership>

Associates particular timestamp with each file, and clusters.

The timestamp is chosen, in this order of priority:

• A date / time string extracted from the filename, if exists in particular patterns, falling back to creation-time, if none exists.
• Original photo-taken time from EXIF metadata if available, and the file has a jpg or jpeg extension.
• File creation time.

Timezones are assumed to be the current time-zone, if not otherwise indicated.

File modification time is not considered.

The clustered are named 01, 02, 03 etc. depending on the number of clusters.

The DBSCAN algorithm is used for clustering.

A special cluster OUTLIER_CLUSTER_IDENTIFIER may also be created, for points that were not density-reachable by others, and aren't part of any cluster in particular.

The relative-path of files are preserved, being added relative to the cluster subdirectory.

The default-patterns for matching filenames are:

• yyyy-mm-dd hh:mm:ss
• yyyymmdd_hhmmss
• yyyymmdd hhmmss

Author:

Owen Feehan

Constructor Summary

Constructors

Constructor	Description
ClusterByTimestamp()

Method Summary

Modifier and Type	Method	Description
ClusterMembership	beforeCopying(Path destinationDirectory, List<FileWithDirectoryInput> inputs)	To be called once before any calls to CopyFilesNaming.destinationPath(File, DirectoryWithPrefix, int, CopyContext).
Optional<Path>	destinationPathRelative(File file, DirectoryWithPrefix outputTarget, int index, CopyContext<ClusterMembership> context)	Calculates the relative-output path (to be appended to destDir)
int	getMinimumPerCluster()	The minimum number of files that must exist for a cluster.
double	getThresholdHours()	Files whose creation-time differs <= this parameter are joined into the same cluster.
List<TimestampPattern>	getTimestampPatterns()	The patterns which can be used to extract a date-time from a filename.
int	getTimeZoneOffset()	If >= 0, sets a specific time-offset in hours.
boolean	isPreserveSubdirectories()	If true, the entire relative-path is used when copying files into the cluster directory.
void	setMinimumPerCluster(int minimumPerCluster)	The minimum number of files that must exist for a cluster.
void	setPreserveSubdirectories(boolean preserveSubdirectories)	If true, the entire relative-path is used when copying files into the cluster directory.
void	setThresholdHours(double thresholdHours)	Files whose creation-time differs <= this parameter are joined into the same cluster.
void	setTimestampPatterns(List<TimestampPattern> timestampPatterns)	The patterns which can be used to extract a date-time from a filename.
void	setTimeZoneOffset(int timeZoneOffset)	If >= 0, sets a specific time-offset in hours.

Methods inherited from class org.anchoranalysis.plugin.io.bean.file.copy.naming.CopyFilesNaming

destinationPath

Methods inherited from class org.anchoranalysis.bean.AnchorBean

checkMisconfigured, describeBean, describeChildren, duplicateBean, fields, findFieldsOfClass, getBeanName, getLocalPath, localise, toString

Methods inherited from class java.lang.Object

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

Constructor Details

ClusterByTimestamp

public ClusterByTimestamp()

Method Details

beforeCopying

public ClusterMembership beforeCopying(Path destinationDirectory,
 List<FileWithDirectoryInput> inputs)
                                throws OperationFailedException

Description copied from class: CopyFilesNaming

To be called once before any calls to CopyFilesNaming.destinationPath(File, DirectoryWithPrefix, int, CopyContext).

Specified by:

beforeCopying in class CopyFilesNaming<ClusterMembership>

Parameters:

destinationDirectory - the directory to which files are copied.

inputs - the total number of files to copy.

Throws:

OperationFailedException

destinationPathRelative

public Optional<Path> destinationPathRelative(File file,
 DirectoryWithPrefix outputTarget,
 int index,
 CopyContext<ClusterMembership> context)
                                       throws OutputWriteFailedException

Description copied from class: CopyFilesNaming

Calculates the relative-output path (to be appended to destDir)

Specified by:

destinationPathRelative in class CopyFilesNaming<ClusterMembership>

Parameters:

file - file to be copied

outputTarget - the directory and prefix associated with the file for outputting

index - an increasing sequence of numbers for each file beginning at 0

context - the context for the copying

Returns:

the relative-path. if empty, the file should be skipped.

Throws:

OutputWriteFailedException

getThresholdHours

public double getThresholdHours()

Files whose creation-time differs <= this parameter are joined into the same cluster.

This is the principle parameter for affecting the sensitivity of the clustering. It is specified in hours between the date-time of two files.

A larger value encourages a smaller total number of clusters (or larger cluster-size). A smaller values encourages the opposite.

setThresholdHours

public void setThresholdHours(double thresholdHours)

Files whose creation-time differs <= this parameter are joined into the same cluster.

This is the principle parameter for affecting the sensitivity of the clustering. It is specified in hours between the date-time of two files.

A larger value encourages a smaller total number of clusters (or larger cluster-size). A smaller values encourages the opposite.

getMinimumPerCluster

public int getMinimumPerCluster()

The minimum number of files that must exist for a cluster.

setMinimumPerCluster

public void setMinimumPerCluster(int minimumPerCluster)

The minimum number of files that must exist for a cluster.

isPreserveSubdirectories

public boolean isPreserveSubdirectories()

If true, the entire relative-path is used when copying files into the cluster directory. If false, only the file-name is used.

setPreserveSubdirectories

public void setPreserveSubdirectories(boolean preserveSubdirectories)

If true, the entire relative-path is used when copying files into the cluster directory. If false, only the file-name is used.

getTimestampPatterns

public List<TimestampPattern> getTimestampPatterns()

The patterns which can be used to extract a date-time from a filename.

setTimestampPatterns

public void setTimestampPatterns(List<TimestampPattern> timestampPatterns)

The patterns which can be used to extract a date-time from a filename.

getTimeZoneOffset

public int getTimeZoneOffset()

If >= 0, sets a specific time-offset in hours. If == -1, then the offset is taken from the current system time-zone settings.

setTimeZoneOffset

public void setTimeZoneOffset(int timeZoneOffset)

If >= 0, sets a specific time-offset in hours. If == -1, then the offset is taken from the current system time-zone settings.