JSON parser and domain

JSON (JavaScript Object Notation) is a simple data-interchange format based on a subset of the JavaScript programming language.

IBM® Integration Bus provides support for a JSON domain. Messages in the JSON domain are processed by the JSON parser and serializer. The JSON parser interprets a bit stream using the JSON grammar, and generates a corresponding JSON domain logical message tree. When processing data for output, the JSON serializer generates a JSON formatted bit stream from a JSON domain logical message tree.

The JSON parser and serializer do not support message validation because JSON message modeling is not supported.

JSON is a language-independent text format, based on two structures:
  • Objects (name-value pairs) with the following types:
    • String
    • Number
    • Boolean
    • Null
  • Ordered collections of values (arrays)
Objects and arrays can be nested.

The JSON parser only accepts a JSON object or a JSON array as the top level value type in both an input bit stream and an output bit stream. It is not possible to use the JSON parser to output a JSON string literal, number, boolean, or null value without first embedding it in an object or array.

For more information about JSON message structure, see JSON message details.

JSON data streams can be parsed from any coded character set ID (CCSID) that is supported by the integration node. The data stream is parsed according to the CCSID that is defined by the transport when the JSON parser is invoked from an input or request node, or defined by the CCSID parameter in a PARSE clause on a CREATE function call. If no CCSID is specified, or if a value of 0 is set, the parser attempts to detect (by examining the first few bytes of the data stream) whether one of the following Unicode encodings is being used:
  • UTF-8
  • UTF-16BE
  • UTF-16LE
  • UTF-32BE
  • UTF-32LE

If a UTF-* CCSID is explicitly specified, the JSON parser tolerates the corresponding Byte Order Mark (BOM) at the beginning of the data stream.

JSON data streams are parsed into a logical message tree under the Data element below the JSON parser root. The logical tree structure is shown in the Example JSON message.

The Data element can be accessed and manipulated from ESQL as JSON.Data, from Java™ as JSON/Data, from the Graphical Data Mapping editor by using the Add User Defined function to define JSON->Data, or from XPath as $Body/Data. The JSON parser issues an error if a bit stream is not formatted according to the JSON grammar.

The JSON serializer serializes message trees into a JSON format data stream. The CCSID can be defined by either the integration node properties tree, or by transport headers in the message assembly. If no CCSID is defined, the serializer defaults to the queue manager default CCSID for all nodes except HTTP nodes, which default to UTF-8 encoding.

When the JSON serializer is invoked through an ASBITSTREAM function call, the CCSID is defined by the CCSID parameter. If no CCSID parameter is provided, or if the value is set to 0, the JSON serializer defaults to using UTF-8 encoding.

To process messages with the JSON parser, select JSON as the Message Domain on the relevant node in the message flow. The XSLTransform node does not support the JSON domain.

The integration node sets the HTTP Content-Type header to application/json when serializing a JSON message tree, unless an explicit value is set by the message flow.

JSON objects are modeled in the integration node message tree as a sequence of NameValue elements. The parser builds the message tree in the order in which it encounters the members in the bit stream. The serializer writes the object members into the bit stream in the tree order.

JSON arrays are modeled in the integration node message tree as a Name element with a JSON parser-specific type flag of JSON.Array and ordered children. The children can be any of the following types of array:
  • An array containing simple values; for example:
    "array1" : [ "thing1", 1 ]
    The following message tree is produced:
    (0x01001000:Array): array1 = (  
        (0x03000000:NameValue):Item = 'thing1' (CHARACTER)
        (0x03000000:NameValue):Item = 1 (INTEGER)
     )

    The JSON parser assigns the name Item to the NameValue elements.

  • An array containing objects; for example:
    "array2" : [ {"a" : 1}, {"b" : 2} ]
    The following message tree is produced:
    (0x01001000:Array):array2 = (
      (0x01000000:Object):Item = (
        (0x03000000:NameValue):a = 1 (INTEGER)
      )
      (0x01000000:Object):Item = (
        (0x03000000:NameValue):b = 2 (INTEGER)
      )
    )
  • A multidimensional array; for example:
    "array3" : [ [1.1], [2.1] ]
    The following message tree is produced:
    (0x01001000:Array):array3 = (
      (0x01001000:Array):Item = (
        (0x03000000:NameValue):Item = 1.1E+0 (FLOAT)
      )
      (0x01001000:Array):Item = (
        (0x03000000:NameValue):Item = 2.1E+0 (FLOAT)
      )
    )

Example JSON message

The following example shows a simple JSON message:

{
    "name" : "John Doe",
    "age" : -1.0,
    "known" : false,
    "address" : { "street" : null,
                  "city" : "unknown" },
    "belongings" : ["item1", "item2", "item3"]
}

This JSON input produces the following integration node logical message tree:

(0x01000000:Object):JSON            = ( ['json' : 0xd55fc8]
  (0x01000000:Object):Data = (
    (0x03000000:NameValue):name       = 'John Doe' (CHARACTER)
    (0x03000000:NameValue):age        = -1E+0 (FLOAT)
    (0x03000000:NameValue):known      = FALSE (BOOLEAN)
    (0x01000000:Object   ):address    = (
      (0x03000000:NameValue):street = NULL
      (0x03000000:NameValue):city   = 'unknown' (CHARACTER)
    )
    (0x01001000:Array    ):belongings = (
      (0x03000000:NameValue):Item = 'item1' (CHARACTER)
      (0x03000000:NameValue):Item = 'item2' (CHARACTER)
      (0x03000000:NameValue):Item = 'item3' (CHARACTER)
    )
  )
)