Response Syntax

{
  "header_footer": [
    {
      "value":"string",
      "key":"string"
    }
  ],
  "tables":[
    {
      "table":{
        "tab_index":0,
        "rows": [
          {
            "row_index":1,
            "columns":[
              {
                "col_index":1,
                "value":"string"
              }
            ]
          }
        ]
      }
    }
  ]
}

The AI response structure is page based. The API is designed to process a single page at a time, thus the result should only ever contain data for a single page. A single object will be returned that contains two JSON properties.

• header_footer (array) -
  A collection of key value pairs identified by the AI extraction as header or footer field data. Header and footer data can cross table boundaries, meaning data may be present in a header and also be part of a table. This is entirely dependent on document layout and can not be controlled or impacted by configuration.

  • key (string) -
    The key, identified by the AI engine, as the field descriptor for a given value. It is possible for the key to be an empty string.

  • value string -
    The extracted value for an associated key.

• tables (array) -
  A collection of tables, rows, and columns identified by the AI engine as relating to one or more tables of data.

  • table (object) -
    A single table object found on a page.

    • tab_index (integer) -
      The numerical index of a table. The index may be used in expressions to identify a specific table of interest. While the index for a table should be consistent, variability between documents may result in visually similar documents presenting a specific table in one sample and not in another. When this happens, the table index can become out of sync with what is expected if expressions are directly referencing table indices. Be sure to test thoroughly a range of document samples when directly referencing table a table index in an expression.

    • rows (array) -
      An array of rows for a parent table. Note, the keyword $row in any expression is replaced automatically with the current row being analyzed by GlobalCapture work collecting table data for injection into a capture process.

      • row_index (integer) -
        The numerical index of a row in a table. The topmost row or rows of any table may represent header data. Depending on a table’s structure, a tables line item data may not be present until the second or third row of a table, with preceding rows containing the table’s header column names.

      • columns (array) -
        An array of column object representing the individual cells of a parent row.

        • col_index (integer) -
          The numerical index of a column in a row (a cell).

        • value (string) -
          The extracted value for a specific cell.