Info: A tornado-based asynchronous version of the pymongo driver for MongoDB.
Author: Dan Yamins [email protected]
APyMongo is an asynchronous version of the PyMongo driver for MongoDB. APyMongo uses the tornado iostream eventloop to drive asychronous requests. A primary use of APyMongo is to serve MongoDB-backed websites in an efficient asynchronous manner via the tornado web server, but it can be used wherever one wants to drive multiple efficient highthrouput read-write connections to a MongoDB instance.
APyMongo tries to stick as closely as possible to PyMongo, both in terms of codebase and API.
APyMongo was developed by the GovData Project, which is sponsored by the Institute for Quantitative Social Science at Harvard.
APyMongo is still the process of "beta testing" -- obviously this is an open beta test -- and has yet to have a formal release. Don't expect everything to work! (But please let us know via the mailing list if things break.)
For now, the project can only be obtained from the github repo (https://github.com/govdata/APyMongo).
The installation process is:
- Install mongodb (http://www.mongodb.org/) if you havent already.
- Install Tornado (www.tornadoweb.org) if you havent already. (NB: You MUST be using a recent pull from the tornado github repo. APyMongo depends on a recent bugfix in the tornado.iostream module.)
- Pull the APyMongo repo
- Run "python setup.py install" in the APyMongo directory.
Here's a basic example that can be used in a Tornado web server:
import json
import tornado.web
import apymongo
from apymongo import json_util
class TestHandler(tornado.web.RequestHandler):
@tornado.web.asynchronous
def get(self):
connection = apymongo.Connection()
collection = conn['testdb']['testcollection']
coll.find_one(callback=self.handle)
def handle(self,response):
self.write(json.dumps(response,default=json_util.default))
self.finish()
For more information, see the examples section of the docs. There, you can find code that will show you how to insert records, query to find records, and count records. You can also see how to use APyMongo to build a streaming data handler.
To use a given example:
-
Make sure you have installed MongoDB, Tornado (and of course APyMongo itself), and that a MongoDB instance is running on localhost:27017 (the default).
-
cd /path/to/apymongo/doc/examples
-
python [desired_example_file.py]
-
Open a web broweser and point it to localhost:8000
Currently, there is no separate documentation for this project. That is by design, as APyMongo's API is essentially identical to pymongo's except for the following:
-
Every pymongo method that actually hits the database for a response now has a callback argument, a single-argument executable to which tornado will pass the contents of the response when it is ready to be read. In other words, you can no longer do e.g.:
r = collection.find_one()
but must instead do e.g.:
def callback(r):
#handle response ...
collection.find_one(callback)
This goes for ALL methods that hit the database, including even such ``simple" things as connection.database_names.
-
Cursors have no next method. Instead, to obtain the equivalent of ``list(cursor.find())", use the apymongo.cursor.loop method.
-
The Connection method has a io_loop argument, to which you can pass an existing tornado.io_loop object for the streams to attach to.
Questions should be posted here:
You'll have to join the mailing list group to post a question. I'll try to respond to questions as quickly as I can.
You can also post a bug report via the github issue tracker. (Probably you should email the list first, and then once I confirm something is a bug, start an issue on the tracker.)
Mongo: APyMongo works for the same MongoDB distributions that PyMongo works on.
Python: APyMongo requires Python >=2.4.
Tornado: >= 1.1
Additional dependencies are:
To run the tests, install nose and run nosetests or python setup.py test in the root of the distribution. Tests are located in the test/ directory.
Currently, the tests are very scant (and -- something using the AsyncTestCase in the tornado.testing framework is not working quite right.) I wouldn't be surprised if some things are broken in stupid ways, but if you let me know about a problem on the mailing list ([email protected]), I'll try to handle it right away.
APymongo currently does not handle:
- master-slave connections.
- DBRef derefercing in son manipulators.
- the explain method
All of these should be no problem to support in principle -- hopefully I (or some needy user) will get to these soon.
Contributions are welcome! Simply fork the project using Github, do your business, and issue a pull request when ready.
APyMongo was developed for the GovData project (https://github.com/govdata/govdata-core), where a version of it is buried deep in the govdata core code. While APyMongo was being modularized for separate relase, we learned of asyncmongo, an existing asynchronous python-language MongoDB driver that also uses the tornado iostream.
Because asyncmongo has a somewhat different API, we decided to release APyMongo as a separate project.