Question

Coming from SQL I have this search condition

WHERE (col1 LIKE "%foo%" OR col2 LIKE "%foo%") AND
      (col1 LIKE "%bar%" OR col2 LIKE "%bar%")

which I want to convert to MongoDB.

I came up wit this, hopefully semantically identical query:

{
  $and: [
    {
      $or: [
        { col1: /.*foo.*/ },
        { col2: /.*foo.*/ }
      ]
    },
    {
      $or: [
        { col1: /.*bar.*/ },
        { col2: /.*bar.*/ }
      ]
    }
  ]
}

Is this the correct way or can it be improved?
Any suggestions about indexes (if they can be used at all)?

Answer 1

Yes, that’s the correct way to implement that query for MongoDB.

If you want an index to fully assist the query, it needs to be a compound index that includes both fields, because MongoDB queries can only use one index per query. So an index like this:

db.coll.ensureIndex({col1: 1, col2: 1})

You can confirm your query is using the index you expect by using explain().