Package org.apache.lucene.benchmark.byTask.tasks

Class AnalyzerFactoryTask

java.lang.Object

org.apache.lucene.benchmark.byTask.tasks.PerfTask

org.apache.lucene.benchmark.byTask.tasks.AnalyzerFactoryTask

All Implemented Interfaces:

Cloneable

public class AnalyzerFactoryTask
extends PerfTask

Analyzer factory construction task. The name given to the constructed factory may be given to NewAnalyzerTask, which will call AnalyzerFactory.create().

Params are in the form argname:argvalue or argname:"argvalue" or argname:'argvalue'; use backslashes to escape '"' or "'" inside a quoted value when it's used as the enclosing quotation mark,

Specify params in a comma separated list of the following, in order:

1. Analyzer args:
   • Required: name:analyzer-factory-name
   • Optional: positionIncrementGap:int value (default: 0)
   • Optional: offsetGap:int value (default: 1)
2. zero or more CharFilterFactory's, followed by
3. exactly one TokenizerFactory, followed by
4. zero or more TokenFilterFactory's

Each component analysis factory may specify luceneMatchVersion (defaults to Version.LATEST) and any of the args understood by the specified *Factory class, in the above-describe param format.

Example:

     -AnalyzerFactory(name:'strip html, fold to ascii, whitespace tokenize, max 10k tokens',
                      positionIncrementGap:100,
                      HTMLStripCharFilter,
                      MappingCharFilter(mapping:'mapping-FoldToASCII.txt'),
                      WhitespaceTokenizer(luceneMatchVersion:LUCENE_5_0_0),
                      TokenLimitFilter(maxTokenCount:10000, consumeAllTokens:false))
     [...]
     -NewAnalyzer('strip html, fold to ascii, whitespace tokenize, max 10k tokens')
 

AnalyzerFactory will direct analysis component factories to look for resources under the directory specified in the "work.dir" property.

Field Summary

Fields inherited from class org.apache.lucene.benchmark.byTask.tasks.PerfTask

logStep, NEW_LINE, params, stopNow

Constructor Summary

Constructors

Constructor	Description
AnalyzerFactoryTask(PerfRunData runData)

Method Summary

Modifier and Type	Method	Description
int	doLogic()	Perform the task once (ignoring repetitions specification) Return number of work items done by this task.
int	lineno(StreamTokenizer stok)	Returns the current line in the algorithm file
<T> Class<? extends T>	lookupAnalysisClass(String className, Class<T> expectedType)	This method looks up a class with its fully qualified name (FQN), or a short-name class-simplename, or with a package suffix, assuming "org.apache.lucene.analysis." as the package prefix (e.g.
void	setParams(String params)	Sets the params.
boolean	supportsParams()	Sub classes that support parameters must override this method to return true.

Methods inherited from class org.apache.lucene.benchmark.byTask.tasks.PerfTask

clone, close, getAlgLineNum, getBackgroundDeltaPriority, getDepth, getLogMessage, getName, getParams, getRunData, getRunInBackground, isDisableCounting, runAndMaybeStats, setAlgLineNum, setDepth, setDisableCounting, setName, setRunInBackground, setup, shouldNeverLogAtStart, shouldNotRecordStats, stopNow, tearDown, toString

Methods inherited from class java.lang.Object

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

Constructor Detail

AnalyzerFactoryTask

public AnalyzerFactoryTask(PerfRunData runData)

Method Detail

doLogic

public int doLogic()

Description copied from class: PerfTask

Perform the task once (ignoring repetitions specification) Return number of work items done by this task. For indexing that can be number of docs added. For warming that can be number of scanned items, etc.

Specified by:

doLogic in class PerfTask

Returns:

number of work items done by this task.

setParams

public void setParams(String params)

Sets the params. Analysis component factory names may optionally include the "Factory" suffix.

Overrides:

setParams in class PerfTask

Parameters:

params - analysis pipeline specification: name, (optional) positionIncrementGap, (optional) offsetGap, 0+ CharFilterFactory's, 1 TokenizerFactory, and 0+ TokenFilterFactory's

lookupAnalysisClass

public <T> Class<? extends T> lookupAnalysisClass(String className,
                                                  Class<T> expectedType)
                                           throws ClassNotFoundException

This method looks up a class with its fully qualified name (FQN), or a short-name class-simplename, or with a package suffix, assuming "org.apache.lucene.analysis." as the package prefix (e.g. "standard.ClassicTokenizerFactory" -> "org.apache.lucene.analysis.standard.ClassicTokenizerFactory").

If className contains a period, the class is first looked up as-is, assuming that it is an FQN. If this fails, lookup is retried after prepending the Lucene analysis package prefix to the class name.

If className does not contain a period, the analysis SPI *Factory.lookupClass() methods are used to find the class.

Parameters:

className - The name or the short name of the class.

expectedType - The superclass className is expected to extend

Returns:

the loaded class.

Throws:

ClassNotFoundException - if lookup fails

supportsParams

public boolean supportsParams()

Description copied from class: PerfTask

Sub classes that support parameters must override this method to return true.

Overrides:

supportsParams in class PerfTask

Returns:

true iff this task supports command line params.

lineno

public int lineno(StreamTokenizer stok)

Returns the current line in the algorithm file