JSON to SQL Mapping

This is a map-type SmartModule that converts arbitrary JSON records into SQL model, which is a self-describing representation of SQL INSERT statements. This SmartModule is intended to be used in SQL Sink Connector, to execute a command in a SQL database.

Mapping

The mapping between incoming JSON records and the resulting SQL record is defined in the configuration of the SmartModule in the mapping init parameter. Let's look at the example:

- uses: infinyon/json-sql@x.y.z
  with:
    mapping:
      table: "target_table"
      map-columns:
        "device_id":
          json-key: "device.device_id"
          value:
            type: "int"
            required: true
        "device_type":
          json-key: "device.type"
          value:
            type: "text"
            default: "mobile"
        "record":
          json-key: "$"
          value:
            type: "jsonb"
            required: true

Here, we create insert statements to target_table database table. Each statement has three columns:

• device_id with int type. The value for this column will looked-up at the following hierarchy $.device.device_id of the input record. If it is not present, the error will be thrown, as the mapping states, it is a required field.
• device_type - text field with mobile marked as default value. Not required. Will be taken from $.device.type hierarchy.
• record - jsonb column that contains the whole input record.

With the given mapping, the Json-Sql SmartModule will convert the input:

{
  "device": {
    "device_id": 1
  }
}

into the following output:

{
  "Insert": {
    "table": "target_table",
    "values": [
      {
        "column": "record",
        "raw_value": "{\"device\":{\"device_id\":1}}",
        "type": "Json"
      },
      {
        "column": "device_type",
        "raw_value": "mobile",
        "type": "Text"
      },
      {
        "column": "device_id",
        "raw_value": "1",
        "type": "Int"
      }
    ]
  }
}

which is equivalent to the following SQL statement:

INSERT INTO 'target_table' (record, device_type, device_id) VALUES ('{"device":{"device_id":1}}', 'mobile', 1)

Data types

The list of supported types and corresponding types from SQL model:

Mapping	Model
int8, bigint	BigInt
int4, int, integer	Int
int2, smallint	SmallInt
bool, boolean	Bool
bytes, bytea	Bytes
text	Text
float4, float, real	Float
float8, "double precision", doubleprecision	DoublePrecision
decimal, numeric	Numeric
date	Date
time	Time
timestamp	Timestamp
json, jsonb	Json
uuid	Uuid

Usage example

To see Json-Sql SmartModule in action, we will use SMDK tool.

info

All smartmodule versions are marked with x.y.z. To find the latest version, run:

• fluvio hub smartmodule list

First, download the Smartmodule:

$ fluvio hub smartmodule download infinyon/json-sql@x.y.z

Second, we create a file transform.yaml with the mapping used before:

# transform.yaml
transforms:
  - uses: infinyon/json-sql@x.y.z
    with:
      mapping:
        table: "target_table"
        map-columns:
          "device_id":
            json-key: "device.device_id"
            value:
              type: "int"
              required: true
          "device_type":
            json-key: "device.type"
            value:
              type: "text"
              default: "mobile"
          "record":
            json-key: "$"
            value:
              type: "jsonb"
              required: true

Test the SmartModule to see the result:

$ smdk test --text '{"device":{"device_id":1}}' --transforms-file ./transform.yaml
{"Insert":{"table":"target_table","values":[{"column":"record","raw_value":"{\"device\":{\"device_id\":1}}","type":"Json"},{"column":"type","raw_value":"mobile","type":"Text"},{"column":"device_id","raw_value":"1","type":"Int"}]}}

As mentioned at the beginning of this page, the outputed records can be consumed by SQL Sink Connector to be executed on SQL database. By convention, SQL Sink connector expects JSON inputs.

For additional examples checkout the tutorials:

• [Build HTTP to SQL Pipeline]
• [Build MQTT to SQL Pipeline]