If the CSV is only one line, JSON data with a different layout will be created.

Anonymous — Sat, 16 Nov 2024 04:10:37 GMT

Hi.

I have a CSV file that looks like this:

id,name,age
1,john,27
2,paul,30

I want to create JSON array like this:

[
    {"id":1,"name":"john","age":27},
    {"id":2,"name":"paul","age":30}
]

There is no problem if there are more than 2 rows in the CSV file.

But, if there is 1 line, it will be output like this.

Input:

id,name,age
1,john,27

Output:

{
    "dummyloop":{"id":1,"name":"john","age":27}
}

Why does the layout change?
I want the following results:

[{"id":1,"name":"john","age":27}]

I present a job with the minimum configuration that can reproduce this phenomenon.

(In fact, the job continues after this.)

Job OverviewtFileInputDelimitedtMaptFileInputJSONField - ComponenttFileInputJSONField - SchematFileInputJSONField - JSON tree

Q:Why use a dummy loop?

A:The job continues after this.
This is the final layout:

{
  "app" : 9,
  "records" : [
    {"id":1,"name":"john","age":27},
    {"id":2,"name":"paul","age":30}
  ]
}

Thank you.

Re: If the CSV is only one line, JSON data with a different layout will be created.

Anonymous — Tue, 19 Nov 2019 08:07:28 GMT

Hi yanchi
If there is only one row, you need to add an attribute @class under dummyloop element, and set its value as "array", see image.

It will generate a string like:

{"dummyloop":[{"id":"1","name":"john","age":"27"}]}

After that, you can remove other characters with this expression on a tJavaRow:

output_row.newColumn = input_row.newColumn.substring(input_row.newColumn.indexOf("["), input_row.newColumn.lastIndexOf("]")+1);

The result will be:

[{"id":"1","name":"john","age":"27"}]

Regards

Shong