Pymongo- Selecting Sub-documents From Collection By Regex

Lets take for example the following collections: { '_id': '0', 'docs': [ {'value': 'abcd', 'key': '1234'}, {'value': 'abef', 'key': '5678'} ] } { '_

Solution 1:

You need an aggregation pipeline that matches each subdocument separately, then re-joins the matching subdocuments into arrays:

from pprint import pprint
from bson import Regex

regex = Regex(r'ab')
pprint(list(col.aggregate([{
    '$unwind': '$docs'
}, {
    '$match': {'docs.value': regex}
}, {
    '$group': {
        '_id': '$_id',
        'docs': {'$push': '$docs'}
    }
}])))

I assume "col" is a variable pointing to your PyMongo Collection object. This outputs:

[{u'_id': u'1', 
  u'docs': [{u'key': u'5678', u'value': u'abgh'}]},
 {u'_id': u'0',
  u'docs': [{u'key': u'1234', u'value': u'abcd'},
            {u'key': u'5678', u'value': u'abef'}]}]

The "r" prefix to the string makes it a Python "raw" string to avoid any trouble with regex code. In this case the regex is just "ab" so the "r" prefix isn't necessary, but it's good practice now so you don't make a mistake in the future.