Basic Transform Hook Script Breaks Flow?

[jaredfraley]

Hi Everyone,

Here is a basic transform function script to just split the tracking into an array.

Yes, it would be best to just combine both records into one but that caused even more issues down the line so I am just working through one issue at a time.

The primary issue is the inability for any script to function. As soon as I add a script, even a basic one like this, the export shows 0 even though there is valid data to be exported and the debug logs output "Qualified Record Ids ["29208443"]"

If I remove the script, the flow functions properly except for the fact that I need the tracking numbers in an array to handle separately.

Why is a simple script breaking the entire flow, preventing exported data to flow down the line?

[tylerlamparter]

You're running into this issue because the script you're using is written for a transform script, but you're actually placing it inside a preSavePage script. The two script types expect different data structures and return formats, which is why your flow breaks when you use the wrong one.

In a transform script, you return a single record. But in preSavePage, you must return an object containing data, errors, abort, and newErrorsAndRetryData. If you don’t, the export shows zero records because integrator.io doesn’t know what to pass forward.

Here’s how your logic should look inside a preSavePage script:

function preSavePage(options) {
  options.data.forEach(page => {
    page.forEach(record => {
      if (record.tracking) {
        record.tracking = record.tracking.split(',');
      }
    });
  });

  return {
    data: options.data,
    errors: options.errors,
    abort: false,
    newErrorsAndRetryData: []
  };
}

This matches your input structure (which is an array of pages), splits the tracking string into an array, and returns the properly formatted object.

For reference:

Stub for preSavePage:

/*
* preSavePageFunction stub:
*
* The name of the function can be changed to anything you like.
*
* The function will be passed one 'options' argument that has the following fields:
*   'data' - an array of records representing one page of data. A record can be an object {} or array [] depending on the data source.
*   'files' - file exports only. files[i] contains source file metadata for data[i]. i.e. files[i].fileMeta.fileName.
*   'errors' - an array of errors where each error has the structure {code: '', message: '', source: '', retryDataKey: ''}.
*   'retryData' - a dictionary object containing the retry data for all errors: {retryDataKey: { data: <record>, stage: '', traceKey: ''}}.
*   '_exportId' - the _exportId currently running.
*   '_connectionId' - the _connectionId currently running.
*   '_flowId' - the _flowId currently running.
*   '_integrationId' - the _integrationId currently running.
*   '_apiId' - the _apiId currently running.
*   '_parentIntegrationId' - the parent of the _integrationId currently running.
*   'pageIndex' - 0 based. context is the batch export currently running.
*   'lastExportDateTime' - delta exports only.
*   'currentExportDateTime' - delta exports only.
*   'settings' - all custom settings in scope for the export currently running.
*   'sandbox' - boolean value indicating whether the script is invoked for sandbox.
*   'testMode' - boolean flag indicating test mode and previews.
*   'job' - the job currently running.
*
* The function needs to return an object that has the following fields:
*   'data' - your modified data.
*   'errors' - your modified errors.
*   'abort' - instruct the batch export currently running to stop generating new pages of data.
*   'newErrorsAndRetryData' - return brand new errors linked to retry data: [{retryData: <record>, errors: [<error>]}].
* Throwing an exception will signal a fatal error and stop the flow.
*/
function preSavePage (options) {
  // sample code that simply passes on what has been exported
  return {
    data: options.data,
    errors: options.errors,
    abort: false,
    newErrorsAndRetryData: []
  }
} 

Stub for transform:

/*
* transformFunction stub:
*
* The name of the function can be changed to anything you like.
*
* The function will be passed one 'options' argument that has the following fields:
*   'record' - object {} or array [] depending on the data source.
*   'settings' - all custom settings in scope for the transform currently running.
*   'testMode' - boolean flag indicating test mode and previews.
*   'job' - the job currently running.
*
* The function needs to return the transformed record.
* Throwing an exception will return an error for the record.
*/
function transform (options) {
  return options.record
}

And just to tie it all together visually:

[jaredfraley]

Thanks! I did not catch that.