Skip to main content

​
DocumentProcessorRun Object

​
Properties

  • object (string)
    Type of the object, always ‘document_processor_run’.
  • id (string)
    The unique identifier for the document processor run.
  • edited (boolean)
    Whether the document processor run has been edited.
  • edits (object)
    A hash of the edits (original + edited values + operator notes, if any) made to the document processor run. Values conform to the value types of the fields.
  • type (string)
    The type of the document processor run. Either EXTRACT, CLASSIFY, SPLITTER, or INSTRUCT_QA.
  • config (object)
    The config of the document processor run. The shape of output depends on the document processor type. Details for each can be found in the Output and Config types page.
  • initialOutput (object)
    The initial raw output from the document processor. This is always set. The shape of output depends on the document processor type. Details for each can be found in the Output and Config types page.
  • reviewedOutput (object)
    An optional field that shows the output after review, if applicable. The shape of output depends on the document processor type. Details for each can be found in the Output and Config types page.
  • output (object)
    The final output, either reviewed or initial. You can rely on this to be always set as the final output (initial or post review when reviewed). The shape of output depends on the document processor type. Details for each can be found in the Output and Config types page.

​
Example

{
  "object": "document_processor_run",
  "id": "dpr_1234",
  "edited": true,
  "type": "EXTRACT",
  "initialOutput": {
    "is_invoice": {
      "id": "field6",
      "type": "boolean",
      "value": true,
      "confidence": 1,
      "references": []
    },
    "line_items": {
      "id": "field3",
      "type": "array",
      "value": [
        {
          "quantity": 1,
          "unit_price": {
            "amount": 4.316,
            "iso_4217_currency_code": "USD"
          },
          "description": "CHEESE CHEDDAR SHARP PRIN SYS2822312"
        }
      ],
      "confidence": 0.51,
      "references": []
    }
  },
  "reviewedOutput": {
    // Same structure as initialOutput, but with reviewed values
  },
  "output": {
    // Final output, either reviewed or initial
  }
}
The exact shape of initialOutput, reviewedOutput, and output will depend on the processor type (EXTRACT, CLASSIFY, etc.) and the specific configuration of that processor. The example above shows a partial extract processor output with invoice fields. Each field in these outputs typically includes:
  • An ID
  • The field type (string, boolean, array, etc.)
  • The extracted value
  • A confidence score
  • Any references to locations in the source document where the information was found