Filter Rows transform Icon Filter Rows

Description

The Filter Rows transform allows you to filter rows based on conditions and comparisons. It affects workflow if connecting the True or False outcome/hop to another transform. There are 3 path available as outputs (True/False/Main output). The main output follows the True path only.

Supported Engines

Hop Engine

Supported

Spark

Supported

Flink

Supported

Dataflow

Supported

Usage

Once this transform is connected to a previous transform (one or more and receiving input), you can click on the "<field>", "=" and "<value>" areas to construct a condition. Downstream transforms can be connected to the True or False hop.

Right-click a condition to edit, delete, move, or add a sub-condition.

You can use a REGEX expression in the "filter row" transform. See “Filter Rows” under the Transforms section for more.

To enter an IN LIST operator, use a string value separated by semicolons.
Lists also works on numeric values like integers. In this case, the list of values must be entered with a string data type, e.g.: 2;3;7;8.

Remember that all transforms are executed in parallel, so both true and false paths will run in some cases.

The filter rows transform detects only fields in the input stream. If you want to filter rows based on a variable value, you can modify the previous transform. For example, a table input transform can include the variable in the SQL as another field such as “select field1, field2, ${myvar} as field3 from table1”, then in the filter row condition you can say field1 = field3. Alternatively, you can use a ‘Get Variables’ transform to set parameters in fields. There are various SQL query statements that will always return results using IF EXISTS or IS NULL for example if you always require field(s) and results to be returned.

An example is found in the samples project called “filter-rows-basic.hpl”.

If you cannot select the field you would like because it does not exist yet because of metadata injection, see this doc: https://hop.apache.org//manual/latest/pipeline/transforms/filterrows.html#_mdi_example and the example in the samples project: metadata-injection/filter-rows-mdi-parent.hpl.

Essentially, to use an MDI field that is not selectable yet (say in a Filter rows transform), you can setup a transform with a similar field and with a similar condition. Then open the pipeline in a text editor and update the field name to the MDI field name that does not exist yet.

Options

Option Description

Transform name

Optionally, you can change the name of this transform to fit your needs.

Send 'true' data to transform

The rows for which the condition specified is true are sent to this transform

Send 'false' data to transform

The rows for which the condition specified are false are sent to this transform

The Condition

Add Condition

Click to add conditions. Add condition converts the original condition into a sub-level condition.

* Click a sub-condition to edit it by going down one level in the condition tree.

* Right-click a condition to edit, delete, move, or add a sub-condition.

Filtering

Filtering rows based on values from variables

The filter rows transform detects only fields in the input stream. If you want to filter rows based on a variable value, you can modify the previous transform like a Table input transform. For example, a table input transform can include the variable in the SQL as another field such as

SELECT field1, field2, ${myvar} AS field3 FROM table1
sql

, then in the filter row condition you can say field1 = field3. Alternatively, you can use a ‘Get Variables’ transform to set parameters in fields.

You can use a REGEX expression in the "filter row" transform.

A query:

SELECT field1,
field2,
${myvar} AS field3
FROM table
WHERE field1=xxxx
sql

Then in the filter row condition, you can have:

field1 = field3
bash

FYI - instead of adding a field to a Table Input as shown above, you can also use a "Get Variables" transform to add fields to the stream.

You can use a REGEX expression in the "filter row" transform.

Remember that all transforms are executed in parallel, so both true and false paths will run in some cases. In some cases it will be better to use a where filter in a SQL query than use a Filter rows transform. If you need to return input rows for either case (true/false), you can use various SQL query methods that will always return results using IF EXISTS or IS NULL for example.

Filtering special characters

To filter special characters like explicit EOF (e.g. from old cobol files) Use a REGEX expression in the "filter row" transform with the syntax: "\x{1A}" where \x mean HEX representation and 1A into parenthesis is the EOF char to match in HEX.

Metadata injection support

All fields of this transform support metadata injection. You can use this transform with ETL Metadata Injection to pass metadata to your pipeline at runtime.

If you cannot select the field you would like because it does not exist yet because of metadata injection, see this doc: https://hop.apache.org//manual/latest/pipeline/transforms/filterrows.html#_mdi_example and this example in the samples project: metadata-injection/filter-rows-mdi-parent.hpl.

Essentially, to use an MDI field that is not selectable yet (say in a Filter rows transform), you can setup a transform with a similar field and with a similar condition. Then open the pipeline in a text editor and update the field name to the MDI field name that does not exist yet.

Special considerations for the condition field

The Filter Rows transform is a special MDI scenario, since it has a nested structure of filter conditions. The condition is given in XML format. The condition XML has the same format as we store the pipeline metadata in a .HPL file in XML format. We do not have a DTD (Document Type Definition) for the .HPL XML format, nor the condition.

It is easy to get to an XML condition:

  1. Create a sample Filter transform with the different conditions you need. This sample transform gives you all the information, such as the values for the functions you use.

  2. Select the transform, copy it to the clipboard, and then paste it into a text editor. Alternatively, you can store the .HPL, and then open the .HPL in a text editor.

  3. Find the <condition> element and its nested elements and modify it accordingly to use it in your MDI scenario.

Sample

The samples project demonstrates some concepts in file "filter-rows-basic.hpl".

MDI Example

The example filter condition below injects the following filter conditions into a Filter Rows transform.

The full example pipelines are available in the samples project as metadata-injection/filter-rows-mdi-parent.hpl and metadata-injection/filter-rows-mdi-child.hpl

stateCode = FL
AND
housenr > 100
shell
<condition>
            <negated>N</negated>
            <conditions>
                <condition>
                    <negated>N</negated>
                    <leftvalue>stateCode</leftvalue>
                    <function>=</function>
                    <rightvalue/>
                    <value>
                        <name>constant</name>
                        <type>String</type>
                        <text>FL</text>
                        <length>-1</length>
                        <precision>-1</precision>
                        <isnull>N</isnull>
                        <mask/>
                    </value>
                </condition>
                <condition>
                    <negated>N</negated>
                    <operator>AND</operator>
                    <leftvalue>housenr</leftvalue>
                    <function>&gt;</function>
                    <rightvalue/>
                    <value>
                        <name>constant</name>
                        <type>Integer</type>
                        <text>100</text>
                        <length>-1</length>
                        <precision>0</precision>
                        <isnull>N</isnull>
                        <mask>####0;-####0</mask>
                    </value>
                </condition>
            </conditions>
        </condition>
xml