org.terrier.utility
Class ApplicationSetup

java.lang.Object
  org.terrier.utility.ApplicationSetup

public class ApplicationSetup
extends Object

This class retrieves and provides access to all the constants and parameters for the system. When it is statically initialised, it loads the properties file specified by the system property terrier.setup. If this is not specified, then the default value is the value of the terrier.home system property, appended by etc/terrier.properties.
eg java -Dterrier.home=$TERRIER_HOME -Dterrier.setup=$TERRIER_HOME/etc/terrier.properties TrecTerrier

System Properties used:

terrier.setup	Specifies where the terrier.properties file can be found.
terrier.home	Specified where Terrier has been installed, if the terrier.properties file cannot be found, or the terrier.properties file does not specify the terrier.home in it. NB:In the future, this may further default to $TERRIER_HOME from the environment.
file.separator	What separates directory names in this platform. Set automatically by Java
line.separator	What separates lines in a file on this platform. Set automatically by Java

In essence, for Terrier to function properly, you need to specify one of the following on the command line:

• terrier.setup pointing to a terrier.properties file containing a terrier.home value.
OR
• terrier.home, and Terrier will use a properties file at etc/terrier.properties, if it finds one.

Any property defined in the properties file can be overridden as follows:

• If the system property terrier.usecontext is equal to true, then a Context object is used to override the properties defined in the file.
• If the system property terrier.usecontext is equal to false, then system properties are used to override the properties defined in the file.

Author:

Gianni Amati, Vassilis Plachouras, Ben He, Craig Macdonald

Nested Class Summary

static interface

ApplicationSetup.TerrierApplicationPlugin Interface for plugins.

Field Summary

protected static Properties	appProperties The properties object in which the properties from the file are read.
static boolean	BLOCK_INDEXING Specifies whether block information will be used for indexing.
static int	BLOCK_SIZE The size of a block of terms in a document.
static String	COLLECTION_SPEC The name of the file that contains the list of resources to be processed during indexing.
static String	DEFAULT_LOG4J_CONFIG Default log4j config Terrier loads if no TERRIER_ETC/terrier-log.xml file exists
static int	DOCS_CHECK_SINGLEPASS Number of documents between each memory check in the single pass inversion method.
static String	EOL The new line character used by the operating system.
static int	EXPANSION_DOCUMENTS The number of top ranked documents considered for expanding the query.
static String	EXPANSION_MODELS Deprecated. As of Terrier 3.5 The name of the file which contains the query expansion methods used. The corresponding property is expansion.models and the default value is qemodels. If a relative path is given, it is prefixed with TERRIER_ETC.
static int	EXPANSION_TERMS The number of terms added to the original query.
static boolean	FIELD_QUERYING Specifies whether fields will be used for querying.
static String	FILE_SEPARATOR The file separator used by the operating system.
static boolean	IGNORE_EMPTY_DOCUMENTS Ignore or not empty documents.
protected static List<ApplicationSetup.TerrierApplicationPlugin>	loadedPlugins list of loaded plugins
static String	LOG4J_CONFIG The configuration file used by log4j
static int	MAX_BLOCKS The maximum number of blocks in a document.
static int	MAX_TERM_LENGTH The maximum size of a term.
static int	MEMORY_THRESHOLD_SINGLEPASS Memory threshold in the single pass inversion method.
static String	TERRIER_ETC The directory under which the configuration files of Terrier are stored.
static String	TERRIER_HOME The directory under which the application is installed.
static String	TERRIER_INDEX_PATH The name of the directory where the inverted file and other data structures are stored.
static String	TERRIER_INDEX_PREFIX The prefix of the data structures' filenames.
static String	TERRIER_SHARE The name of the directory where installation independent read-only data is stored.
static String	TERRIER_VAR The name of the directory where the data structures and the output of Terrier are stored.
static String	TERRIER_VERSION Current Terrier version
static String	TREC_MODELS Deprecated. As of Terrier 3.5 The filename of the file that contains the weighting models to be used. The corresponding property is trec.models and the default value is trec.models. If a relative path is given, then it is prefixed with TERRIER_ETC.
static String	TREC_QRELS Deprecated. As of Terrier 3.5 The name of the file that contains a list of qrels files to be used for evaluation. The corresponding property is trec.qrels and its default value is trec.qrels. If a relative path is given, TERRIER_ETC will be prefixed.
static String	TREC_RESULTS The name of the directory where the results are stored.
static String	TREC_RESULTS_SUFFIX The suffix of the files, where the results are stored.
static String	TREC_TOPICS_LIST Deprecated. As of Terrier 3.5 The name of the file that contains a list of files where queries are stored. The corresponding property is trec.topics.list and the default value is trec.topics.list. If a relative path is given, TERRIER_ETC will be prefixed.
protected static Properties	UsedAppProperties

Constructor Summary

ApplicationSetup()

Method Summary

static void	bootstrapInitialisation() forces ApplicatinSetup initilisation
static void	clearAllProperties() Clears ApplicationSetup of all properties
static void	configure(InputStream propertiesStream) Loads the common Terrier properties from the specified InputStream
ApplicationSetup.TerrierApplicationPlugin	getPlugin(String name) Return a loaded plugin by name.
static Properties	getProperties() Returns the underlying properties object for ApplicationSetup
static String	getProperty(String propertyKey, String defaultValue) Returns the value for the specified property, given a default value, in case the property was not defined during the initialization of the system.
static Properties	getUsedProperties() Returns a properties object detailing all the properties fetched during the lifetime of this class.
static void	loadCommonProperties() Loads the ApplicationSetup variables, e.g.
static String	makeAbsolute(String filename, String DefaultPath) Checks whether the given filename is absolute and if not, it adds on the default path to make it absolute.
static void	setDefaultProperty(String propertyKey, String defaultValue) set a property value only if it has not already been set
static void	setProperty(String propertyKey, String value) Sets a value for the specified property.
protected static void	setupPlugins() Calls the initialise method of any plugins named in terrier.plugins

Methods inherited from class java.lang.Object

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

Field Detail

TERRIER_VERSION

public static final String TERRIER_VERSION

Current Terrier version

See Also:

Constant Field Values

DEFAULT_LOG4J_CONFIG

public static final String DEFAULT_LOG4J_CONFIG

Default log4j config Terrier loads if no TERRIER_ETC/terrier-log.xml file exists

Since:

1.1.0

See Also:

Constant Field Values

appProperties

protected static final Properties appProperties

The properties object in which the properties from the file are read.

UsedAppProperties

protected static final Properties UsedAppProperties

FILE_SEPARATOR

public static String FILE_SEPARATOR

The file separator used by the operating system. Defaults to the system property file.separator.

EOL

public static String EOL

The new line character used by the operating system. Defaults to the system property line.separator.

TERRIER_HOME

public static String TERRIER_HOME

The directory under which the application is installed. It corresponds to the property terrier.home and it should be set in the properties file, or as a property on the command line.

TERRIER_ETC

public static String TERRIER_ETC

The directory under which the configuration files of Terrier are stored. The corresponding property is terrier.etc and it should be set in the properties file. If a relative path is given, TERRIER_HOME will be prefixed.

TERRIER_SHARE

public static String TERRIER_SHARE

The name of the directory where installation independent read-only data is stored. Files like stopword lists, and example and testing data are examples. The corresponding property is terrier.share and its default value is share. If a relative path is given, then TERRIER_HOME will be prefixed.

TERRIER_VAR

public static String TERRIER_VAR

The name of the directory where the data structures and the output of Terrier are stored. The corresponding property is terrier.var and its default value is var. If a relative path is given, TERRIER_HOME will be prefixed.

TERRIER_INDEX_PATH

public static String TERRIER_INDEX_PATH

The name of the directory where the inverted file and other data structures are stored. The default value is InvFileCollection but it can be overridden with the property terrier.index.path. If a relative path is given, TERRIER_VAR will be prefixed.

COLLECTION_SPEC

public static String COLLECTION_SPEC

The name of the file that contains the list of resources to be processed during indexing. The contents of this file are collection implementation dependent. For example, for a TREC collection, this file must contain the list of files to index. The corresponding property is collection.spec and by default its value is collection.spec. If a relative path is given, TERRIER_ETC will be prefixed.

TREC_RESULTS

public static String TREC_RESULTS

The name of the directory where the results are stored. The corresponding property is trec.results and the default value is results. If a relative path is given, TERRIER_VAR will be prefixed.

TREC_TOPICS_LIST

public static String TREC_TOPICS_LIST

Deprecated. As of Terrier 3.5 The name of the file that contains a list of files where queries are stored. The corresponding property is trec.topics.list and the default value is trec.topics.list. If a relative path is given, TERRIER_ETC will be prefixed.

TREC_QRELS

public static String TREC_QRELS

Deprecated. As of Terrier 3.5 The name of the file that contains a list of qrels files to be used for evaluation. The corresponding property is trec.qrels and its default value is trec.qrels. If a relative path is given, TERRIER_ETC will be prefixed.

TREC_RESULTS_SUFFIX

public static String TREC_RESULTS_SUFFIX

The suffix of the files, where the results are stored. It corresponds to the property trec.results.suffix and the default value is .res.

TREC_MODELS

public static String TREC_MODELS

Deprecated. As of Terrier 3.5 The filename of the file that contains the weighting models to be used. The corresponding property is trec.models and the default value is trec.models. If a relative path is given, then it is prefixed with TERRIER_ETC.

MAX_TERM_LENGTH

public static int MAX_TERM_LENGTH

The maximum size of a term. It corresponds to the the property max.term.length, and the default value is 20.

Since:

1.1.0

IGNORE_EMPTY_DOCUMENTS

public static boolean IGNORE_EMPTY_DOCUMENTS

Ignore or not empty documents. That is, if it is true, then a document that does not contain any terms will have a corresponding entry in the .docid file and the total number of documents in the statistics will be the total number of documents in the collection, even if some of them are empty. It corresponds to the property ignore.empty.documents and the default value is false.

TERRIER_INDEX_PREFIX

public static String TERRIER_INDEX_PREFIX

The prefix of the data structures' filenames. It corresponds to the property terrier.index.prefix and the default value is data.

EXPANSION_TERMS

public static int EXPANSION_TERMS

The number of terms added to the original query. The corresponding property is expansion.terms and the default value is 10.

EXPANSION_DOCUMENTS

public static int EXPANSION_DOCUMENTS

The number of top ranked documents considered for expanding the query. The corresponding property is expansion.documents and the default value is 3.

EXPANSION_MODELS

public static String EXPANSION_MODELS

Deprecated. As of Terrier 3.5 The name of the file which contains the query expansion methods used. The corresponding property is expansion.models and the default value is qemodels. If a relative path is given, it is prefixed with TERRIER_ETC.

BLOCK_SIZE

public static int BLOCK_SIZE

The size of a block of terms in a document. The corresponding property is blocks.size and the default value is 1.

MAX_BLOCKS

public static int MAX_BLOCKS

The maximum number of blocks in a document. The corresponding property is blocks.max and the default value is 100000.

BLOCK_INDEXING

public static boolean BLOCK_INDEXING

Specifies whether block information will be used for indexing. The corresponding property is block.indexing and the default value is false. The value of this property cannot be modified after the index of a collection has been built.

FIELD_QUERYING

public static boolean FIELD_QUERYING

Specifies whether fields will be used for querying. The corresponding property is field.querying and the default value is false.

MEMORY_THRESHOLD_SINGLEPASS

public static int MEMORY_THRESHOLD_SINGLEPASS

Memory threshold in the single pass inversion method. If a memory check is below this value, the postings in memory are written to disk. Default is 50 000 000 (50MB) and 100 000 000 (100MB) for 32bit and 64bit JVMs respectively, and can be configured using the property memory.reserved.

DOCS_CHECK_SINGLEPASS

public static int DOCS_CHECK_SINGLEPASS

Number of documents between each memory check in the single pass inversion method. The default value is 20, and this can be configured using the property docs.check.

LOG4J_CONFIG

public static String LOG4J_CONFIG

The configuration file used by log4j

loadedPlugins

protected static List<ApplicationSetup.TerrierApplicationPlugin> loadedPlugins

list of loaded plugins

Constructor Detail

ApplicationSetup

public ApplicationSetup()

Method Detail

bootstrapInitialisation

public static void bootstrapInitialisation()

forces ApplicatinSetup initilisation

loadCommonProperties

public static void loadCommonProperties()

Loads the ApplicationSetup variables, e.g. ApplicationSetup.TERRIER_HOME

configure

public static void configure(InputStream propertiesStream)
                      throws IOException

Loads the common Terrier properties from the specified InputStream

Throws:

IOException

getProperty

public static String getProperty(String propertyKey,
                                 String defaultValue)

Returns the value for the specified property, given a default value, in case the property was not defined during the initialization of the system. The property values are read from the properties file. If the value of the property terrier.usecontext is true, then the properties file is overridden by the context. If the value of the property terrier.usecontext is false, then the properties file is overridden

Parameters:

propertyKey - The property to be returned

defaultValue - The default value used, in case it is not defined

Returns:

the value for the given property.

getUsedProperties

public static Properties getUsedProperties()

Returns a properties object detailing all the properties fetched during the lifetime of this class. It is of note that this is NOT the underlying appProperties table, as to update that would mean that properties fetched using their defaults, could not have different defaults in different places.

getProperties

public static Properties getProperties()

Returns the underlying properties object for ApplicationSetup

setProperty

public static void setProperty(String propertyKey,
                               String value)

Sets a value for the specified property. The properties set with this method are not saved in the properties file.

Parameters:

propertyKey - the name of the property to set.

value - the value of the property to set.

setDefaultProperty

public static void setDefaultProperty(String propertyKey,
                                      String defaultValue)

set a property value only if it has not already been set

Parameters:

propertyKey - the name of the property to set.

defaultValue - the value of the property to set.

setupPlugins

protected static void setupPlugins()

Calls the initialise method of any plugins named in terrier.plugins

getPlugin

public ApplicationSetup.TerrierApplicationPlugin getPlugin(String name)

Return a loaded plugin by name. Returns null if a plugin of that name has not been loaded

makeAbsolute

public static String makeAbsolute(String filename,
                                  String DefaultPath)

Checks whether the given filename is absolute and if not, it adds on the default path to make it absolute. If a URI scheme is present, the filename is assumed to be absolute

Parameters:

filename - String the filename to make absolute

DefaultPath - String the prefix to add

Returns:

the absolute filename

clearAllProperties

public static void clearAllProperties()

Clears ApplicationSetup of all properties