Kazaam was created with the goal of supporting easy and fast transformations of JSON data with Golang. This functionality provides us with an easy mechanism for taking intermediate JSON message representations and transforming them to formats required by arbitrary third-party APIs.

Inspired by Jolt, Kazaam supports JSON to JSON transformation via a transform "specification" also defined in JSON. A specification is comprised of one or more "operations". See Specification Support, below, for more details.

API Documentation is available at http://godoc.org/gopkg.in/qntfy/kazaam.v3.

Kazaam is primarily designed to be used as a library for transforming arbitrary JSON. It ships with six built-in transform types, described below, which provide significant flexibility in reshaping JSON data.

Also included when you go get Kazaam, is a binary implementation, kazaam that can be used for development and testing of new transform specifications.

Finally, Kazaam supports the implementation of custom transform types. We encourage and appreciate pull requests for new transform types so that they can be incorporated into the Kazaam distribution, but understand sometimes time-constraints or licensing issues prevent this. See the API documentation for details on how to write and register custom transforms.

Due to performance considerations, Kazaam does not fully validate that input data is valid JSON. The IsJson() function is provided for convenience if this functionality is needed, it may significantly slow down use of Kazaam.

Kazaam currently supports the following transforms:

• shift
• concat
• coalesce
• extract
• timestamp
• uuid
• default
• pass

The shift transform is the current Kazaam workhorse used for remapping of fields. The specification supports jsonpath-esque JSON accesses and sets. Concretely

{
  "operation": "shift",
  "spec": {
    "object.id": "doc.uid",
    "gid2": "doc.guid[1]",
    "allGuids": "doc.guidObjects[*].id"
  }
}

executed on a JSON message with format

{
  "doc": {
    "uid": 12345,
    "guid": ["guid0", "guid2", "guid4"],
    "guidObjects": [{"id": "guid0"}, {"id": "guid2"}, {"id": "guid4"}]
  },
  "top-level-key": null
}

would result in

{
  "object": {
    "id": 12345
  },
  "gid2": "guid2",
  "allGuids": ["guid0", "guid2", "guid4"]
}

The jsonpath implementation supports a few special cases:

• Array accesses: Retrieve nth element from array
• Array wildcarding: indexing an array with [*] will return every matching element in an array
• Top-level object capture: Mapping $ into a field will nest the entire original object under the requested key

The shift transform also supports a "require" field. When set to true, Kazaam will throw an error if any of the paths in the source JSON are not present.

The concat transform allows the combination of fields and literal strings into a single string value.

{
    "operation": "concat",
    "spec": {
        "sources": [{
            "value": "TEST"
        }, {
            "path": "a.timestamp"
        }],
        "targetPath": "a.timestamp",
        "delim": ","
    }
}

executed on a JSON message with format

{
    "a": {
        "timestamp": 1481305274
    }
}

would result in

{
    "a": {
        "timestamp": "TEST,1481305274"
    }
}

Notes:

• sources: list of items to combine (in the order listed)
  • literal values are specified via value
  • field values are specified via path (supports the same addressing as shift)
• targetPath: where to place the resulting string
  • if this an existing path, the result will replace current value.
• delim: Optional delimiter

The concat transform also supports a "require" field. When set to true, Kazaam will throw an error if any of the paths in the source JSON are not present.

A coalesce transform provides the ability to check multiple possible keys to find a desired value. The first matching key found of those provided is returned.

{
  "operation": "coalesce",
  "spec": {
    "firstObjectId": ["doc.guidObjects[0].uid", "doc.guidObjects[0].id"]
  }
}

executed on a json message with format

{
  "doc": {
    "uid": 12345,
    "guid": ["guid0", "guid2", "guid4"],
    "guidObjects": [{"id": "guid0"}, {"id": "guid2"}, {"id": "guid4"}]
  }
}

would result in

{
  "doc": {
    "uid": 12345,
    "guid": ["guid0", "guid2", "guid4"],
    "guidObjects": [{"id": "guid0"}, {"id": "guid2"}, {"id": "guid4"}]
  },
  "firstObjectId": "guid0"
}

An extract transform provides the ability to select a sub-object and have kazaam return that sub-object as the top-level object. For example

{
  "operation": "extract",
  "spec": {
    "path": "doc.guidObjects[0].path.to.subobject"
  }
}

executed on a json message with format

{
  "doc": {
    "uid": 12345,
    "guid": ["guid0", "guid2", "guid4"],
    "guidObjects": [{"path": {"to": {"subobject": {"name": "the.subobject", "field", "field.in.subobject"}}}}, {"id": "guid2"}, {"id": "guid4"}]
  }
}

would result in

{
  "name": "the.subobject",
  "field": "field.in.subobject"
}

A timestamp transform parses and formats time strings using the golang syntax. Note: this operation is done in-place. If you want to preserve the original string(s), pair the transform with shift.

{
  "operation": "timestamp",
  "timestamp[0]": {
    "inputFormat": "Mon Jan _2 15:04:05 -0700 2006",
    "outputFormat": "2006-01-02T15:04:05-0700"
  }
}

executed on a json message with format

{
  "timestamp": [
    "Sat Jul 22 08:15:27 +0000 2017",
    "Sun Jul 23 08:15:27 +0000 2017",
    "Mon Jul 24 08:15:27 +0000 2017"
  ]
}

would result in

{
  "timestamp": [
    "2017-07-22T08:15:27+0000",
    "Sun Jul 23 08:15:27 +0000 2017",
    "Mon Jul 24 08:15:27 +0000 2017"
  ]
}

A uuid transform generates a UUID based on the spec. Currently supports UUIDv3, UUIDv4, UUIDv5.

For version 4 is a very simple spec

{
    "operation": "uuid",
    "spec": {
        "doc.uuid": {
            "version": 4, //required
        }
    }
}

executed on a json message with format

{
  "doc": {
    "author_id": 11122112,
    "document_id": 223323,
    "meta": {
      "id": 23
    }
  }
}

would result in

{
  "doc": {
    "author_id": 11122112,
    "document_id": 223323,
    "meta": {
      "id": 23
    }
    "uuid": "f03bacc1-f4e0-4371-a5c5-e8160d3d6c0c"
  }
}

For UUIDv3 & UUIDV5 are a bit more complex. These require a Name Space which is a valid UUID already, and a set of paths, which generate UUID's based on the value of that path. If that path doesn't exist in the incoming document, a default field will be used instead. Note both of these fields must be strings. Additionally you can use the 4 predefined namespaces such as DNS, URL, OID, & X500 in the name space field otherwise pass your own UUID.

{
   "operation":"uuid",
   "spec":{
      "doc.uuid":{
         "version":5,
         "nameSpace":"DNS",
         "names":[
            {"path":"doc.author_name", "default":"some string"},
            {"path":"doc.type", "default":"another string"},
         ]
      }
   }
}

executed on a json message with format

{
  "doc": {
    "author_name": "jason",
    "type": "secret-document"
    "document_id": 223323,
    "meta": {
      "id": 23
    }
  }
}

would result in

{
  "doc": {
    "author_name": "jason",
    "type": "secret-document",
    "document_id": 223323,
    "meta": {
      "id": 23
    },
    "uuid": "f03bacc1-f4e0-4371-a7c5-e8160d3d6c0c"
  }
}

A default transform provides the ability to set a key's value explicitly. For example

{
  "operation": "default",
  "spec": {
    "type": "message"
  }
}

would ensure that the output JSON message includes {"type": "message"}.

A pass transform, as the name implies, passes the input data unchanged to the output. This is used internally when a null transform spec is specified, but may also be useful for testing.

To start, go get the versioned repository:

go get gopkg.in/qntfy/kazaam.v3

If you want to create an executable binary from this project, follow these steps (you'll need go installed and $GOPATH set):

go get gopkg.in/qntfy/kazaam.v3
cd $GOPATH/src/gopkg.in/qntfy.kazaam.v3/kazaam
go install

This will create an executable in $GOPATH/bin like you would expect from the normal go build behavior.

See godoc examples.