pymongo在聚合查询中包含javascript

Question

I'm currently tasked with researching databases and am trying various queries using the pymongo library to investigate suitability for given projects.

My timestamps are saved in millisecond integer format and I'd like to do a simple sales by day aggregated query. I understand from here (answer by Alexandre Russel) that as the timestamps weren't uploaded in BSON format I can't use date and time functions to create bins, but can manipulate timestamps using embedded javascript.

As such I've written the following query:

 [{
    "$project": {
        "year": {
            "$year": {
                "$add": ["new Date(0)", "$data.horaContacto"]
            }
        },
        "month": {
            "$month": {
                "$add": ["new Date(0)", "$data.horaContacto"]
            }
        }
    }
}, {
    "$group": {
        "_id": {
            "year": "$year",
            "month": "$month"
        },
        "sales": {
            "$sum": {
                "$cond": ["$data.estadoVenta", 1, 0]
            }
        }
    }
}]

But get this error:

pymongo.errors.OperationFailure: exception: $add only supports numeric or date types, not String

I think whats happening is that the js "new Date(0)" is being interpreted by the mongo driver as a string, not applied as js. If I remove the encapsulating inverted double quotes then Python tries to interpret this code and errors accordingly. This is just one example and I'd like to include more js in queries in future tests but can't see a way to get it to play nicely with Python (having said this I'm fairly new to Python too).

Does anybody know if:

• I'm correct in assuming the error occurs because mongo interprets the JS as a string and tries to sum it directly?
• If I can indicate to mongo this is JS from Python without Python trying to intepret the code?

So far I've tried searching via Google and various combinations of single and double inverted commas.

Pasted below is a few rows of randomly generated test data if required:

Thanks,

James

{'_id': 0,'data': {'edad': '74','estadoVenta': True,'visits': [{'visitLength': 1819.349246663518,'visitNo': 1,'visitTime': 1480244647948.0}],'apellido2': 'Aguilar','apellido1': 'Garcia','horaContacto': 1464869545373.0,'preNombre': 'Agustin','_id': 0,'telefono': 630331272,'location': {'province': 'Aragón','city': 'Zaragoza','type': 'Point','coordinates': [-0.900203, 41.747726],'country': 'Spain'}}}, 
{'_id': 1,'data': {'edad': '87','estadoVenta': False,'visits': [{'visitLength': 2413.9938072105024,'visitNo': 1,'visitTime': 1465417353597.0}],'apellido2': 'Torres','apellido1': 'Acosta','horaContacto': 1473404147769.0,'preNombre': 'Sara','_id': 1,'telefono': 665968746,'location': {'province': 'Galicia','city': 'Cualedro','type': 'Point','coordinates': [-7.659321, 41.925328],'country': 'Spain'}}}, 
{'_id': 2,'data': {'edad': '48','estadoVenta': True,'visits': [{'visitLength': 2413.9938072105024,'visitNo': 1,'visitTime': 1465415138597.0}],'apellido2': 'Perez','apellido1': 'Sanchez','horaContacto': 1473404923569.0,'preNombre': 'Sara','_id': 2,'telefono': 665967346,'location': {'province': 'Galicia','city': 'Barcelona','type': 'Point','coordinates': [-7.659321, 41.925328],'country': 'Spain'}}}

Answer 1

The MongoDB aggregation framework cannot use any Javascript. You must specify all the data in your aggregation pipeline using BSON. PyMongo can translate a standard Python datetime to BSON, and you can send it as part of the aggregation pipeline, like so:

import datetime

epoch = datetime.datetime.fromtimestamp(0)
pipeline = [{
    "$project": {
        "year": {
            "$year": {
                "$add": [epoch, "$data.horaContacto"]
            }
        },
        # the rest of your pipeline here ....
    }
}]

cursor = db.collection.aggregate(pipeline)