MongoDB supports a rich query language; in fact, its support of dynamic queries is one of its more distinguishing features compared to other datastores in the NoSQL world. Mongo’s query language has a variety of query selectors that enable you to fashion some powerful document searches. One particular query selector that comes in handy from time to time is $exists.

Because Mongo is document oriented and thus lacks a rigid schema, documents (for better or for worse) can have varying structures within a collection. In practice, you probably don’t see vastly differing documents within a collection; however, from time to time, various document fields might differ (as in, they might not be present). Take for example, the classic example of a business card: some cards (i.e. documents) might list a fax number while others might omit it. As another example, as an application and its corresponding data evolves, new fields might be added (or removed).

The $exists query selector facilitates finding documents that have (or do not have) specific fields. On more than one occasion, I’ve employed this query selector to find documents in need of a surgical update. For example, in a collection dubbed words with word documents that each contain an embedded definitions document collection, I can find those particular definitions that do not have a part_of_speech element like so:

1

db.words.find({"definitions.part_of_speech":{"$exists":false}}).sort({"spelling":1}).limit(100)

Note that the $exists query selector takes a boolean – true or false.

As you can probably ascertain, Mongo’s $exists is slightly different than SQL’s exists – in fact, in SQL there is no way to fashion a query to find a row not containing a column! Can you dig it?