Using mongoimport to read CSV into nested structure?

Question

I have a mongo document with a structure like: { "foo": { "bar1": "val1", "bar2": "val2"} } I'd like to import my data from a csv using mongoimport --type csv --headerline [...]

I am not sure how to format the field name in the csv to address the nested structure. For instance:

test.csv:

foo.bar1
example

returns { "_id" : ObjectId("4e9d9d25c5d8708e1f51cdbc"), "foo.bar1" : "example" } instead of the desired output:

{ "_id" : ObjectId("4e9d9d25c5d8708e1f51cdbc"), "foo: {"bar1" : "example"} }

The field name seems to be interpreted as a string regardless of its value. Things like foo[bar1] and foo: {bar1} are also used verbatim.

Lye

This may not currently be possible. From this thread: No way to do nesting with csv. Kind of a weird api. Suggestions/patches welcome – Lye

Answer 1

This isn't supported in the current (v2.0) version of mongoimport, but it should be coming soon. You can check out the JIRA ticket here, scheduled for v2.1:

• Dotted field names should produce nested objects in mongoimport

Until then, if you can translate your CSV data to JSON then you could use mongoimport --type json to import the nested data.

EDIT: This feature is released now, and available from 2.8.0-rc0

Synecology

Answer 2

You can add a column in CSV like parent_key.array_index.nested_key

CSV data example

JSON object image in mongo db after import

Gwenora