org.terrier.querying

Class Decorate

java.lang.Object

org.terrier.querying.Decorate

All Implemented Interfaces:

PostFilter, PostProcess, Process

public class Decorate
extends Object
implements PostProcess, PostFilter

This class decorates a result set with metadata. This metadata can be highlighted, can have a query biased summary created, and also be escaped for display in another format. Controls:

• summaries - comma or semicolon delimited list of the key names for which a query biased summary should be created. e.g. summaries:snippet
• emphasis - comma or semicolon delimited list of they key names that should have boldened for occurrences of the query terms. e.g. emphasis:title;snippet
• earlyDecorate - comma or semicolon delimited list of the key names that should be decorated early, e.g. to support another PostProcess using them.
• escape - comma or semicolon delimited list of the key names that should be escaped e.g. escape:title;snippet;url. Currently, per-key type escaping is not supported. The default escape type is defined using the property decorate.escape.

Properties:

• decorate.escape - default escape type for metadata. Default is HTML. Possible escape types include XML, JAVASCRIPT, and URL. See utility.StringTools.ESCAPE

Since:

3.0

Author:

Craig Macdonald, Vassilis Plachouras, Ben He

Field Summary

Fields

Modifier and Type	Field and Description
protected static Pattern	cleanQuery
protected static String[]	CONTROL_VALUE_DELIMS delimiters for breaking down the values of controls further
protected static Pattern	controlNonVisibleCharacters
protected Matcher	controlNonVisibleCharactersMatcher
protected static StringTools.ESCAPE	defaultEscape what is the default escape sequence
protected Set<String>	earlyKeys
protected Set<String>	emphasisKeys
protected Map<String,StringTools.ESCAPE>	escapeKeys
protected Pattern	highlight highlighting pattern for the current query
protected gnu.trove.TObjectIntHashMap<String>	keys
protected LRUMap<Integer,String[]>	metaCache The cache used for the meta data.
protected MetaIndex	metaIndex The meta index server.
protected String[]	metaKeys
protected String[]	qTerms query terms of the current query
protected Summariser	summariser
protected Set<String>	summaryKeys

Fields inherited from interface org.terrier.querying.PostFilter

FILTER_ADJUSTED, FILTER_OK, FILTER_REMOVE

Constructor Summary

Constructors

Constructor and Description
Decorate()

Method Summary

Methods

Modifier and Type	Method and Description
protected boolean	checkControl(String control_name, SearchRequest srq)
byte	filter(Manager m, SearchRequest q, ResultSet rs, int rank, int docid) Called for each result in the resultset, used to filter out unwanted results.
protected Pattern	generateEmphasisPattern(String[] _qTerms) Creates a regular expression pattern to highlight query terms metadata.
String	getInfo() Returns the name of the post processor.
protected String[]	getMetadata(String[] metaKeys, int docid)
protected String[][]	getMetadata(String[] metaKeys, int[] docids)
void	new_query(Manager m, SearchRequest q, ResultSet rs) Called before the processing of a resultset using this PostFilter is applied.
void	process(Manager manager, SearchRequest q) decoration at the postprocess stage.

Methods inherited from class java.lang.Object

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

Field Detail

CONTROL_VALUE_DELIMS

protected static final String[] CONTROL_VALUE_DELIMS

delimiters for breaking down the values of controls further

metaCache

protected LRUMap<Integer,String[]> metaCache

The cache used for the meta data. Implements a Least-Recently-Used policy for retaining the most recently accessed metadata.

metaIndex

protected MetaIndex metaIndex

The meta index server. It is provided by the manager.

controlNonVisibleCharacters

protected static final Pattern controlNonVisibleCharacters

defaultEscape

protected static final StringTools.ESCAPE defaultEscape

what is the default escape sequence

controlNonVisibleCharactersMatcher

protected Matcher controlNonVisibleCharactersMatcher

cleanQuery

protected static final Pattern cleanQuery

highlight

protected Pattern highlight

highlighting pattern for the current query

qTerms

protected String[] qTerms

query terms of the current query

keys

protected gnu.trove.TObjectIntHashMap<String> keys

summaryKeys

protected Set<String> summaryKeys

emphasisKeys

protected Set<String> emphasisKeys

escapeKeys

protected Map<String,StringTools.ESCAPE> escapeKeys

earlyKeys

protected Set<String> earlyKeys

summariser

protected Summariser summariser

metaKeys

protected String[] metaKeys

Constructor Detail

Decorate

public Decorate()

Method Detail

new_query

public void new_query(Manager m,
             SearchRequest q,
             ResultSet rs)

Called before the processing of a resultset using this PostFilter is applied. Can be used to save information for the duration of the query.

Specified by:

new_query in interface PostFilter

Parameters:

m - The manager controlling this query

q - The search request being processed

rs - the resultset that is being iterated through

filter

public byte filter(Manager m,
          SearchRequest q,
          ResultSet rs,
          int rank,
          int docid)

Called for each result in the resultset, used to filter out unwanted results.

Specified by:

filter in interface PostFilter

Parameters:

m - The manager controlling this query

q - The search request being processed

rank - which array index (rank) in the resultset have we reached

docid - The docid of the currently being procesed result.

process

public void process(Manager manager,
           SearchRequest q)

decoration at the postprocess stage. only decorate if required for future postfilter or postprocesses.

Specified by:

process in interface Process

Parameters:

manager - The manager instance handling this search session.

q - the current query being processed

getMetadata

protected String[] getMetadata(String[] metaKeys,
                   int docid)

getMetadata

protected String[][] getMetadata(String[] metaKeys,
                     int[] docids)

generateEmphasisPattern

protected Pattern generateEmphasisPattern(String[] _qTerms)

Creates a regular expression pattern to highlight query terms metadata.

Parameters:

_qTerms - query terms

Returns:

Pattern to apply

checkControl

protected boolean checkControl(String control_name,
                   SearchRequest srq)

getInfo

public String getInfo()

Returns the name of the post processor.

Specified by:

getInfo in interface Process

Returns:

String the name of the post processor.