As an example, suppose we have the following dictionary list in memory
dict_list = [
{
'id': 1,
'from_value_key': 'x'
},
{
'id': 2,
'from_value_key': 'y'
},
{
'id': 4,
'from_value_key': 'z'
}
]
db.collection
id: 1, other_key_and_values...
id: 2, other_key_and_values...
id: 3, other_key_and_values...
I want to apply a batch update process to the above db collections.
The contents of the db collection after updating are as follows
id: 1, other_key_and_values..., to_value_key: 'x'
id: 2, other_key_and_values..., to_value_key: 'y'
id: 3, other_key_and_values... #The id: 3 does not exist in the dictionary list side, so the setting of to_value_key does not occur.
#Ignore id: 4 because it exists only in the dictionary list.
I think it is possible to create multiple UpdateOne and use bulk_write, but is it possible to use, for example, $lookup, $unwind, $merge, etc. in the aggregation pipline to perform the update process in batches?
Assuming id
field is unique and there is a unique index built around it, you can put your documents in a $documents
stage and $merge
into the collection.
db.aggregate([
{
"$documents": [
// your python dict here
{
"id": 1,
"from_value_key": "x"
},
{
"id": 2,
"from_value_key": "y"
},
{
"id": 4,
"from_value_key": "z"
}
]
},
{
"$merge": {
"into": "collection",
"on": "id",
"whenMatched": "merge"
}
}
])
^ This won't work, as playground is yet to support $documents
. It is just to demonstrate the syntax.