WebMethods

WebMethods.io: Custom Transformers

Michael Mikhailidi

• 4 min read
ttt
Image Source: Karan Bhatia/Unsplash

In previous integration posts, I briefly touched on the powerful mechanism of the Webmethods.io Transformers. Let's explore the data manipulation tools using a real-life example.

Transform allows you to customize the workflow output or data you send to the subsequent actions. Webmetohds.io Integration will enable you to define one or more Transforms for a connector or a link. Workflow elements with Transform definitions display a small blue icon, as in the diagram below.

Transform Definitions and Availability

As you may see, the GPT Chat connector and the link between GPT Chat and Switch have defined connectors - red $t1 and $t2 labels on the diagram below.

The GPT Could use Transform $t1 output (blue label) for the OpenAI API interactions, but $t2 output is unavailable because the declaration is further down the flow. However, all components after the Swith could benefit from both Transform outputs.

❗
You can define multiple transformers for a single workflow item, however you cant "chain" Transforms on the same level.

The Webmethods.io Integration offers a long list of predefined transformations. The list below contains only categories of Transforms:

  • Array - List manipulation operation
  • Custom - Node.js custom code.
  • Date - Date/Time manipulations
  • JSON - JSON - String transformations
  • Math - Mathematical functions
  • Object - Object operations
  • String - String type functions
  • Util - Transforms with no categories.

The document provides a detailed outline of general Transform usage. I'll focus on using the Node.js transformation and be brief on creating and mapping a new Transform.

❔
For time to time, ChatGPT completion texts contain unnecessary surrounding spaces. You cold, those spaces, using the standard Trim transformation. Unfortunately, the resulting still unnecessary spaces, enclosed by double quotes.

I should rely on the existing toolset by adding a few steps to use standard operations to remove spaces and quotes. But I'm all for efficiency and custom transformation with a few "spaghetti" code lines that solve my formatting problem with a single operation.

The steps below use the integration workflow I designed for this blog. If you haven't seen it before, check my previous posts on this topic for more context on operations below.

  1. To open the configuration dialog, click on the link to the right of the GPT Chat connector and select the ⛭ icon.
Wire Configuration Dialog.
  1. Expand the Transform branch on the left side and click "Add New" to start the new Transform definition wizard.
  2. Set the name for the new transformation (in my case, "TrimGPTCompletion") and select the type as "Custom/Node.js."
Custom Node.js Transform Configuration
  1. Delete str1 and str2 declarations on lines 6- 7 and leave the cursor on the blank space in the code editor area.
  2. Expand your GPT Chat response. On my screenshots, it has the name "$a3 - Create completions," then drill down to the "choices" list.
  3. Click on the list name, and the corresponding variable reference will appear in your code.
Use Workflow Parameters and Data
  1. Now that we know how to refer to the workflow data, let's use the code below. The code loops through the list of choices and returns a list of trimmed strings without extra quotes. 

/* for completing this action and exporting output to 
next action use $export(error, data) */


 
var trimmed = [];

$a3.data.choices.forEach(reply=>{
  trimmed.push(reply.text.trim().replace(/^"/g,'') \
               .replace(/"$/g,'').trim());  
});

$export(null, { done : trimmed });
  1. Ensure your code editor does not indicate syntax errors or warnings, and click the "Done" button twice to close the wire configuration screen.
  2. Use the "Save" button in the top right corner to preserve workflow changes.

The following steps are not well-documented, but I show you how to use custom transform results.

  1. Select the X connector, then click the ⛭ icon to open the connector configuration screen.
  2. Expand the Transform section on the right to see the new Transform element. Note that you could use the transformation but can't delete or change the transformation.
Custom Transform for X Connector
  1. Drag TrimGPTCompletion to the "text" field of the X connector. The input template should be similar to {{$transform.t1}}.
  2. Since that variable refers to the whole object, we export from the custom operation. Change the template text to {{$transform.t1.output.done[0]}}. This way, we address the first list item of the output attribute "done."
  1. Click "Next" and complete the connector configuration, testing the transformation results.

A few takeaways:

  • Transformations offer you a rich toolset to manipulate workflow data
  • Use custom connectors and transformations only when you have no choice or the "no-code" approach is too "expensive."

In case you missed previous publications: