JSON to Table¶
Convert a JSON file to a Table
There are two kinds of tables that can be created:
- Single row – where the JSON structure is simply flattened
- Multiple rows - where the JSON structure is recursively expanded to create several rows
If a single-row table is created, there is an option to minimize the column names to remove unnecessary path information from the JSON keys.
For example from the JSON:
{
"version":"1.0",
"software":"sfd",
"items" : {
"a":"1",
"b":"2",
"c":"3"
}
}
we can create the following single-row table
version software items.a items.b items.c
----------------------------------------------------
1.0 sfd 1 2 3
and the column names can be minimized to
version software a b c ------------------------------------- 1.0 sfd 1 2 3
If a multiple rows-table is created, the recursive algorithm might identify keys and therefore columns that are lacking some values. One can choose to fill in the missing values with a empty string, a nan string or mask the value.
For example from the JSON:
{
"version":"1.0",
"software":"sfd",
"items" : [
{
"a":"1",
"b":"2",
"c":"3"
},
{
"a":"67",
"b":"77",
"d":"97"
}
]
}
we can create the following multiple-rows table
version software a b c d
-------------------------------------------
1.0 sfd 1 2 3 ---
1.0 sfd 67 77 --- 97
where the c
column is masked in the second row and the d
column is masked in the first row.
If the algorithm that creates tnhe multi-row table fails to produce the desired table, it might be worth using other nodes to remove, select or split the JSON structure on some key.
- Input ports:
input: json
Input JSON object
- Output ports:
output: table
Output table
- Configuration:
- (no label) (table_kind)
- What kind of table to create
- Minimize colum names (minimize_col_names)
- Create column names that are minimal
- Use zero-like values instead of masks (nomask)
- When unchecked data cells that are missing will be masked. When checked such cells are instead assigned 0, 0.0, False, “”, etc. depending on the type of the value column.