COUCHBASE PYTHON LIBRARY

This library provides methods to connect to both the couchbase memcached interface and the couchbase rest api interface.

This version requires Python 2.6 or later.

You'll need to install the following Python library requirements via pip:

pip install requests

Open Issues: http://www.couchbase.org/issues/browse/PYCBC

USAGE

Two simple use cases to set and get a key in the default bucket and then create a new bucket using the memcached and rest clients::

#!/usr/bin/env python

from couchbase.couchbaseclient import CouchbaseClient
from couchbase.rest_client import RestConnection

client = CouchbaseClient("http://localhost:8091/pools/default",
                         "default", "", False)
client.set("key1", 0, 0, "value1")
client.get("key1")

server_info = {"ip": "localhost",
               "port": 8091,
               "username": "Administrator",
               "password": "password"}
rest = RestConnection(server_info)
rest.create_bucket(bucket='newbucket',
                   ramQuotaMB=100,
                   authType='none',
                   saslPassword='',
                   replicaNumber=1,
                   proxyPort=11215,
                   bucketType='membase')

Example code that creates buckets and then does sets, gets and views using the unified client::

from couchbase import Couchbase

# connect to a couchbase server
cb = Couchbase('localhost:8091',
               username='Administrator',
               password='password')

# fetch a Bucket with subscript
default_bucket = cb['default']
# set a value with subscript (nearly equivalent to .set)
default_bucket['key1'] = 'value1'

# fetch a bucket with a function
default_bucket2 = cb.bucket('default')
# set a json value with subscript (nearly equivalent to .set)
default_bucket2['key2'] = {'value': 'value2', 'expiration': 0}

# set a value with a function
default_bucket.set('key3', 0, 0, 'value3')

# fetch a key with a function
print 'key1 ' + str(default_bucket.get('key1'))
print 'key2 ' + str(default_bucket2.get('key2'))
# fetch a key with subscript
print 'key3 ' + str(default_bucket2['key3'])

# create a new bucket
try:
    newbucket = cb.create('newbucket', ram_quota_mb=100, replica=1)
except:
    newbucket = cb['newbucket']

# set a JSON document using the more "pythonic" interface
newbucket['json_test'] = {'type': 'item', 'value': 'json test'}
print 'json_test ' + str(newbucket[doc_id])
# use the more verbose API which allows for setting expiration & flags
newbucket.set('key4', 0, 0, {'type': 'item', 'value': 'json test'})
print 'key4 ' + str(newbucket[doc_id])

design_doc = {"views":
              {"all_by_types":
               {"map":
                '''function (doc, meta) {
                     emit([meta.type, doc.type, meta.id], doc.value);
                     // row output: ['json', 'item', 'key4'], 'json test'
                   }'''
                },
               },
              }
# save a design document
newbucket['_design/testing'] = design_doc

all_by_types_view = newbucket['_design/testing'].views()[0]
rows = all_by_types_view.results({'stale': False})
for row in rows:
    print row

# delete the 'newbucket' bucket
cb.delete('newbucket')

RUNNING TESTS

Requirements:

• easy_install nose
• pip install nose-testconfig

Thanks to nose's setup.py integration, test running is as simple as

python setup.py nosetests

If you want to customize the nose settings which are stored in setup.cfg. The default will generate coverage reports (placed in './cover'), and stop on the first error found.

Additionally, to run these tests on a version of Couchbase Server greater than 1.8, you'll need to enable the flush_all setting.

In 1.8.1 use cbflushctl:

cbflushctl localhost:11210 set flushall_enabled true

In 2.0.0 use cbepctl:

cbepctl localhost:11210 set flush_param flushall_enabled true

BASIC BENCHMARKING

We like things to go fast, and we can't know how fast they're going without measuring them. To check the various Python SDK pieces against python-memcached and pylibc, we've created a simple cProfile-based performance reporting tool.

To run this (on a testing cluster, not on dev or production), do:

python couchbase/benchmarks/benchmark.py

To read the profile output do:

python couchbase/benchmarks/profiles/{name_of_profile_output_file}

It's early stage stuff as yet, but it should be helpful for quick progress comparison, and to help track down places the SDK can improve.