couchdb-pythonviews – A Python View Server for CouchDB¶

Installation¶

couchdb-pythonviews requires setuptools. If you do not have it installed already you will want to:

$ curl -O http://peak.telecommunity.com/dist/ez_setup.py
$ python ez_setup.py

The view server line protocol is 100% JSON so installing an optimized JSON parser significantly improves performance. couchdb-pythonviews checks for JSON libraries in this order; jsonlib2, simplejson, json. If you have one of the optimized libraries installed (jsonlib2 or simplejson with C speedups) it will use it.

Map and Reduce¶

This section assumes you have some knowledge of how CouchDB’s map and reduce system works.

This Python view server works a bit differently than the default JavaScript view server in CouchDB. Instead of defining a single function in the design document attribute for a given function the Python view server requires that you use a decorator to call out which function in the Python code is a given function type.:

@map_function
def my_map_function(doc):
    emit(doc['type'], 1)

This means that, if you like, you can stick multiple view functions in a single Python file which will sync the code blob to multiple attributes.:

@map_function
def my_map_function(doc):
    emit(doc['type'], 1)

@reduce_function
def my_reduce_function(values, rereduce):
    return sum(values)

This code can now be used as both the map and reduce attributes and inside the view server it will only get compiled once.

Reduce is a special function, it uses the positional argument names to decide what should be passed to it. This means that the argument names you use in your reduce function matter. Valid argument names are keys, values, ids, length, and rereduce.:

@reduce_function
def mega_reduce(keys, values, ids, length, rereduce):
    # I did *not* need all of these args
    if rereduce:
        return sum(values)
    return length

The Python view server also allows you to write explicit rereduce functions.:

@reduce_function
def full_length_reduce(length):
    return length

@rereduce_function
def full_length_rereduce(values):
    return sum(values)