To use an intelligent structure model in a mapping, add a Structure Parser transformation or a Hierarchy Builder transformation to the mapping.
An intelligent structure model is required for Structure Parser transformations and is optional for Hierarchy Builder transformations.
When you configure the transformation, select or create an intelligent structure model, select the type of input that the transformation expects to receive, and select the output that you pass to downstream transformations. The model can generate relational, XML, JSON, JSON Lines, or Hadoop format output.
Note: If the output contains data of type DATE, Intelligent Structure Discovery represents the data in the mapping as String, not as Date or DateTime.
Models that you use with a Structure Parser transformation in a mapping can contain up to 12,000 fields. The Structure Parser transformation might fail to load models that contain more than 12,000 fields.
Tip: When you create a model that isn't JSON, XML, or XSD-based, and the model contains nested repeating groups, you can reduce the number of fields in the model by normalizing the output data. Intelligent Structure Discovery normalizes the input data by default for JSON, XML, and XSD-based models that contain nested repeating groups.