Filter

Short Description

Filter component filters input records according to a specified condition.

COMPONENT	SAME INPUT METADATA	SORTED INPUTS	INPUTS	OUTPUTS	JAVA	CTL	AUTO-PROPAGATED METADATA
Filter	-	x	1	1-2	-	-	✓

Ports

PORT TYPE	NUMBER	REQUIRED	DESCRIPTION	METADATA
Input	0	✓	For input data records	Any
Output	0	✓	For allowed data records	Input 0
	1	x	For rejected data records	Input 0

This component has one input port and one or two output ports. The first input port and the first output port are mandatory. The optional second output port can be used for rejected data if it is connected to another component.

Metadata

Filter has no metadata template.
Metadata can be propagated through this component.
All output metadata must be the same.

Filter Attributes

ATTRIBUTE	REQ	DESCRIPTION	POSSIBLE VALUES
BASIC
Filter expression	[1]	An expression according to which records are filtered. It is a CTL expression returning boolean.	e.g. `$in.0.field1==2`
ADVANCED
Filter class	[1]	Name of an external class defining which records pass the filter.

[1] One of these attributes must be specified. In case both Filter expression and Filter class is specified, the former will be used. The Java class referenced by the Filter class attribute is expected to implement the RecordFilter interface.

Details

Filter receives records through input port and filters them according to a logical expression. It sends all records corresponding to the filter expression to the first output port and all rejected records to the second output port.

Unmatched records are sent to the second output port only if it is connected. If no edge is connected to the second output port, unmatched records are discarded.

Filter Expression

To configure this component, specify the Filter expression. The filter expression can be very simple - just a comparison or a single CTL function returning boolean.

//#CTL2
$in.0.count == 1

//#CTL2
isInteger($in.0.field2)

The filter expression can consist of several subexpressions connected with logical operators.

//#CTL2
$in.0.isInteger($in.0.count) && $in.0.weightKgs <= 1500

To express precedence of particular subexpressions, use parentheses.

//#CTL2
( $in.0.weightKgs > 0 && $in.0.weightKgs < 1500 ) || $in.0.product == "C"

For these subexpressions, there is a set of functions that can be used and set of comparison operators (greater than, greater than or equal to, less than, less than or equal to, equal to, not equal to). The latter can be selected in the Filter editor dialog as the mathematical comparison signs (>, >=, <, <=, ==, !=) or as textual abbreviations (.gt., .ge., .lt., .le., .eq., .ne.).

All of the record field values should be expressed by their port numbers preceded by a dollar sign, dot and their names. For example, $in.0.employeeid.

Components with Similar Functionality

You can also use the Partition component as a filter instead of Filter. With the Partition component you can define much more sophisticated filter expressions and distribute data records among more output ports.
Or you can use the Map component as a filter.

Java Interface for Filter

Beside filter expression, it is possible to define filtering by Java class implementing org.jetel.component.RecordFilter interface. The class requires a default (no arguments) constructor. The interface consists of the following methods:

void init()
Called before isValid() is used.
void setTransformationGraph(TransformationGraph)
Associates a transformation graph with the filter class instance.
boolean isValid(DataRecord)
Is called for each incoming data record. The implementor shall answer true if the record passes the filter, false otherwise.

Examples

Filtering according to a single field value

This example shows the basic use case of the Filter.

The input contains data on the products sold in the last year. We are interested in figures related to one particular product, e.g. pencils. Filter out other products.

The metadata contains Product, Count and Location fields.

Pen       | 324 | Edinburgh
Pencil    | 543 | Edinburgh
Sharpener | 54  | Edinburgh
Pen       | 150 | Glasgow
Pencil    | 834 | Glasgow
Pen       | 92  | Stirling
Pencil    | 257 | Stirling

Solution
Use Filter. In the component, specify the Filter Expression.

ATTRIBUTE	VALUE
Filter Expression	$in.0.Product == "Pencil"

The following records have been sent to the first output port. Other records have been filtered out.

Pencil    | 543 | Edinburgh
Pencil    | 834 | Glasgow
Pencil    | 257 | Stirling

Note that comparison with == operator is case sensitive.
To compare the strings case-insensitively, convert both operands to the same case first.

//#CTL2
lowerCase( $in.0.Product ) == "pencil"

See lowerCase or upperCase.