CSV and TSV

Your CSV or TSV file must start with a header row. Atlas Data Federation utilizes the header row as field names. The dot-delimited field names in the header row become nested fields or objects in JSON format. For each dot in the field name, Data Federation creates another level of nesting.

Example

Suppose your federated database instance is reading a CSV file with content similar to the following:

company,location.state,location.city.name,location.city.street
"MongoDB", "California", "Palo Alto", "Forest Ave"

For the data fields in the above example CSV file, Data Federation creates the following JSON document:

{
   "company": "MongoDB",
   "location": {
      "state": "California",
      "city": {
         "name": "Palo Alto",
         "street": "Forest Ave",
   }
}

Data Federation requires all field names at the same level of nesting to be unique. The following are examples of invalid field names in the header row:

One field duplicates another field at the same level of nesting.
Example
Consider the following:
company,location,company
In the header, company is included twice at the same level of nesting.
One dot-delimited field duplicates another field at the same level of nesting.
Example
Consider the following:
company,location,location.city
In the header, location is both a stand-alone field and dot-delimited field at the same level of nesting.

Back

Parquet

Limitations