Converts the selected columns content to a JSON value row-wise or column-wise. It also has an option to "undo" - with some limitations - the JSON to Table transformation, and can create JSON values for each row based on the column names.
Example input table:

a.b	a.c	d
b0	c0	d0
b1	c1	d1

With the different parameters, the following JSON values are generated:
Row-oriented (datatable as one JSON value):

[ {
  "a.b" : "b0",
  "a.c" : "c0",
  "d" : "d0"
}, {
  "a.b" : "b1",
  "a.c" : "c1",
  "d" : "d1"
} ]

Keep rows (column names treated as paths, default separator: .):

{
  "a" : {
    "b" : "b0",
    "c" : "c0"
  },
  "d" : "d0"
}

{
  "a" : {
    "b" : "b1",
    "c" : "c1"
  },
  "d" : "d1"
}

Column-oriented (row keys as JSON value, with the key: "ROWID"):

{
  "ROWID" : [ "Row0", "Row1" ],
  "a.b" : [ "b0", "b1" ],
  "a.c" : [ "c0", "c1" ],
  "d" : [ "d0", "d1" ]
}

Options

Input columns

The selected columns will be transformed.

Row keys

Configure how row keys should be handled in the JSON output.

• Omit row key: The row keys will be omitted, not used in the generated JSON value (which are arrays when the rows are not kept).
• Row key as JSON value with key: The row keys will be included in the generated JSON (array) value with the specified key.
• Row key as JSON key: The row keys are added to the generated JSON value as a key, in this case not an array, but an object is created.

Row key name

The key name to use when row keys are included as JSON values.

Column names as paths

When enabled, column names are treated as hierarchical paths using the specified separator.

Path separator

When column names as paths is enabled, this value will be used to find the keys for JSON columns based on the column names.

Aggregation direction

Configure how to aggregate the selected columns into JSON.

• Row-oriented (n input rows → 1 output cell): The values from the selected columns are collected and concatenated row-wise to a JSON object/array, after that these are combined by the rows to a single JSON value.
• Keep rows (n input rows → n output cells): The selected columns are combined to a new JSON column containing the values from the columns and the name of the columns as keys. This option does not combine the rows of the input table.
• Column-oriented (n input rows → 1 output cell): The values from the selected columns are collected and concatenated column-wise to a JSON object/array, after that these are combined by the columns to a single JSON value.

Remove source columns

When checked, the source columns are removed when the rows are kept. It has no effect for the other aggregation direction options, when all columns are removed.

Missing values

Configure how missing values should be handled in the JSON output.

• are omitted: Missing values from the input table do not generate a key in the resulting JSON structure, they are omitted completely. Note that in a column-oriented transformation missing cells will still be inserted as null values in the column's array because otherwise the row arrays for different columns may have different numbers of entries. This would make it impossible to reconstruct the original table.
• are inserted as 'null': Missing values from the input table are inserted as null values.

Output column name

Name of the resulting JSON column.

Views

This node has no views

Workflows

• _01_TestWF_Fails_Then_SucceedsKNIME Hub
• 00 Politica de Carteira Agrupada2KNIME Hub
• 00_call_local_workflow_(JSON)KNIME Hub
• 00_check_prime_numberKNIME Hub
• 00_Topic_Modeling_with_Verified_ComponentsKNIME Hub
• Show all 529 workflows

Links

• Flow_variables to support Object as wellby izaychik63 on 2020-08-11

  forum.knime.com/p/99623
• String manipulation for requests bodyby danielesser on 2020-11-15

  forum.knime.com/p/108268
• Pulling Row Data into JSON Content Bodyby qqilihq on 2022-05-07

  forum.knime.com/p/156188

Developers