from pymongo import MongoClient

uri = "mongodb://localhost:27017/"

client = MongoClient(uri)

db = client['test_db']
# or
# db = client.test_db

# collection = db['test_collection']
# or
collection = db.test_collection


collection.save({"hello":"world"})

print collection.find_one()

Once you got a collection object, queries use the same syntax as in the mongo shell. Some slight differences are:

• every key must be enclosed in brackets. For example:

  db.find({frequencies: {$exists: true}})
  

  becomes in pymongo (note the True in uppercase):

  db.find({"frequencies": { "$exists": True }})
  

• objects such as object ids or ISODate are manipulated using python classes. PyMongo uses its own ObjectId class to deal with object ids, while dates use the standard datetime package. For example, if you want to query all events between 2010 and 2011, you can do:

   from datetime import datetime
  
   date_from = datetime(2010, 1, 1)
   date_to = datetime(2011, 1, 1)
   db.find({ "date": { "$gte": date_from, "$lt": date_to } }):
  

Let's say you need to add a field to every document in a collection.

import pymongo

client = pymongo.MongoClient('localhost', 27017)
db = client.mydb.mycollection

for doc in db.find():
   db.update(
       {'_id': doc['_id']}, 
       {'$set': {'newField': 10} }, upsert=False, multi=False)

The find method returns a Cursor, on which you can easily iterate over using the for in syntax. Then, we call the update method, specifying the _id and that we add a field ($set). The parameters upsert and multi come from mongodb (see here for more info).