This document describes how to create and configure data sources in the application.

Data sources support two main field types:

Variables represent single values extracted from the data source.

snippet.json

{
  "type": "variable",
  "selector": "$.val1",
  "name": "v1",
  "dataType": "string"
}

• type: Must be „variable”
• selector: JSONPath expression to extract the value
• name: Field name in the output
• dataType: Data type (string, number, boolean)

Datasets represent tabular data with multiple rows and columns.

snippet.json

{
  "type": "dataset",
  "name": "table_data",
  "selector": "$.nested.nested_2",
  "columns": [
    {
      "name": "id",
      "selector": "$[*]",
      "dataType": "number"
    },
    {
      "name": "value",
      "selector": "$root:$.additionalData[*].val",
      "dataType": "number"
    }
  ]
}

• type: Must be „dataset”
• name: Dataset name in the output
• selector: JSONPath expression to extract the dataset
• columns: Array of column definitions
  • name: Column name
  • selector: JSONPath expression to extract column values
  • dataType: Data type (string, number, boolean)

The $root: prefix allows accessing data from the root context even when the current context is within a nested structure.

snippet.json

{
  "selector": "$root:$.additionalData[*].val"
}

This is particularly useful in dataset column selectors when you need to reference data outside the current dataset context.

1. Access root-level property from a nested dataset: json { "selector": "$root:$.metadata.version" }
2. Access array elements from root while in a nested context: json { "selector": "$root:$.additionalData[*].val" }
3. Use with filters: json { "selector": "$root:$.additionalData[?(@.name == 'N1')].values[*].val" }
4. Reference field values in selectors: json { "selector": "$root:$.someData[?(@.test_name == '{field:test_name}')].capacity[*].value" }

   This allows you to dynamically reference the value of another field (test_name in this example) within your selector.

Data splitting allows creating multiple outputs from a single data source based on a specific value.