Skip to main content
Version: 2.1.3

Json

Description​

Json analysis of the specified fields of the original data set

tip

This transform ONLY supported by Spark.

Options​

nametyperequireddefault value
source_fieldstringnoraw_message
target_fieldstringnoroot
schema_dirstringno-
schema_filestringno-
common-optionsstringno-

source_field [string]​

Source field, if not configured, the default is raw_message

target_field [string]​

The target field, if it is not configured, the default is __root__ , and the result of Json parsing will be uniformly placed at the top of the Dataframe

schema_dir [string]​

Style directory, if not configured, the default is $seatunnelRoot/plugins/json/files/schemas/

schema_file [string]​

The style file name, if it is not configured, the default is empty, that is, the structure is not specified, and the system derives it by itself according to the input of the data source.

common options [string]​

Transform plugin common parameters, please refer to Transform Plugin for details

Schema Use cases​

  • json schema usage scenarios

The multiple data sources of a single task may contain different styles of json data. For example, the topicA style from Kafka is

{
"A": "a_val",
"B": "b_val"
}

The style from topicB is

{
"C": "c_val",
"D": "d_val"
}

When running Transform , you need to fuse the data of topicA and topicB into a wide table for calculation. You can specify a schema whose content style is:

{
"A": "a_val",
"B": "b_val",
"C": "c_val",
"D": "d_val"
}

Then the fusion output result of topicA and topicB is:

+-----+-----+-----+-----+
|A |B |C |D |
+-----+-----+-----+-----+
|a_val|b_val|null |null |
|null |null |c_val|d_val|
+-----+-----+-----+-----+

Examples​

Do not use target_field​

json {
source_field = "message"
}
  • Source
+----------------------------+
|message |
+----------------------------+
|{"name": "ricky", "age": 24}|
|{"name": "gary", "age": 28} |
+----------------------------+
  • Sink
+----------------------------+---+-----+
|message |age|name |
+----------------------------+---+-----+
|{"name": "gary", "age": 28} |28 |gary |
|{"name": "ricky", "age": 23}|23 |ricky|
+----------------------------+---+-----+

Use target_field​

json {
source_field = "message"
target_field = "info"
}
  • Souce
+----------------------------+
|message |
+----------------------------+
|{"name": "ricky", "age": 24}|
|{"name": "gary", "age": 28} |
+----------------------------+
  • Sink
+----------------------------+----------+
|message |info |
+----------------------------+----------+
|{"name": "gary", "age": 28} |[28,gary] |
|{"name": "ricky", "age": 23}|[23,ricky]|
+----------------------------+----------+

The results of json processing support select * from where info.age = 23 such SQL statements

Use schema_file​

json {
source_field = "message"
schema_file = "demo.json"
}
  • Schema

Place the following content in ~/seatunnel/plugins/json/files/schemas/demo.json of Driver Node:

{
"name": "demo",
"age": 24,
"city": "LA"
}
  • Source
+----------------------------+
|message |
+----------------------------+
|{"name": "ricky", "age": 24}|
|{"name": "gary", "age": 28} |
+----------------------------+
  • Sink
+----------------------------+---+-----+-----+
|message |age|name |city |
+----------------------------+---+-----+-----+
|{"name": "gary", "age": 28} |28 |gary |null |
|{"name": "ricky", "age": 23}|23 |ricky|null |
+----------------------------+---+-----+-----+

If you use cluster mode for deployment, make sure that the json schemas directory is packaged in plugins.tar.gz