projects_repository.py

from pymongo import MongoClient
from bson.objectid import ObjectId
from project import Project

class ProjectsRepository(object):
    """ Repository implementing CRUD operations on projects collection in MongoDB """

    def __init__(self):
        # initializing the MongoClient, this helps to 
        # access the MongoDB databases and collections 
        self.client = MongoClient(host='localhost', port=27017)
        self.database = self.client['projects']


    def create(self, project):
        if project is not None:
            self.database.projects.insert(project.get_as_json())            
        else:
            raise Exception("Nothing to save, because project parameter is None")


    def read(self, project_id=None):
        if project_id is None:
            return self.database.projects.find({})
        else:
            return self.database.projects.find({"_id":project_id})


    def update(self, project):
        if project is not None:
            # the save() method updates the document if this has an _id property 
            # which appears in the collection, otherwise it saves the data
            # as a new document in the collection
            self.database.projects.save(project.get_as_json())            
        else:
            raise Exception("Nothing to update, because project parameter is None")


    def delete(self, project):
        if project is not None:
            self.database.projects.remove(project.get_as_json())            
        else:
            raise Exception("Nothing to delete, because project parameter is None")

self.client = MongoClient(host='localhost', port=27017)

Create this client once for each process, and reuse it for all operations. It is a common mistake to create a new client for each request, which is very inefficient.

Shouldn't we use Connection Pooling by making client accessible at the module level ?