Skip to main content

Mapping

For the mapping, a inputMapping section in the Endpoint schema has to be added and this defines how the incoming data will be used to build up a new object with the desired structure.

The basic structure of this schema is key:value based. This means that the key of the object defines where to look for a value from the incoming document, and the value of that key defines where to populate/build the new property on the outgoing document. The root inputMapping property is used to define the mapping of the incoming document and populate/build the outgoing object.

The example below lists the different mapping syntax available:

{
"inputMapping": {
"rootProperty": "rootProperty", // map incoming root property to an outgoing root property
"rootObject.object1.object2.property": "rootObject.property", // map incoming nested property to an outgoing nested property
"array[]": "array", // map incoming root array to an outgoing array.
"rootArray[1]": "rootArray[1]", // map incoming root array at index 1 to an outgoing array at index 1 (useful when using an output.json template to override a specific index value).
"rootArray[].property": "rootArray[].property", // map all incoming array nested property to an outgoing array nested property. Note: not specifying an index for the array will push in the new value, instead of overriding it at the specified index
"rootArray[1].property": "rootArray[1].property", // map incoming array nested property at index 1 to an outgoing array nested property at index 1 (useful when using an output.json template to override a specific index value).
"rootObject.array[].property": "rootObject.object1.array[].property", // map incoming property that is an array of objects to an outgoing object with a nested object containing an array of objects
}
}