kvfile

A simple Key-Value store that's file based - so can accommodate large data sets with a small memory footprint.

Internally will use the faster leveldb as a storage backend or sqlite3 as fallback if leveldb is not available.

The Basics

The API should feel familiar to anyone working with Python. It exposes get, keys and items for reading from the DB, and set for setting a value in the DB.

Initializing

import datetime
import decimal

from kvfile import KVFile

kv = KVFile()

Setting values

kv.set('s', 'value')
kv.set('i', 123)
kv.set('d', datetime.datetime.fromtimestamp(12325))
kv.set('n', decimal.Decimal('1234.56'))
kv.set('ss', set(range(10)))
kv.set('o', dict(d=decimal.Decimal('1234.58'), 
                 n=datetime.datetime.fromtimestamp(12325)))

Getting values

assert kv.get('s') == 'value'
assert kv.get('i') == 123
assert kv.get('d') == datetime.datetime.fromtimestamp(12325)
assert kv.get('n') == decimal.Decimal('1234.56')
assert kv.get('ss') == set(range(10))
assert kv.get('o') == dict(d=decimal.Decimal('1234.58'), 
                           n=datetime.datetime.fromtimestamp(12325))

Listing values

keys() and items() methods return a generator yielding the values for efficient stream processing.

The returned data is sorted ascending (by default) based on the keys

assert list(kv.keys()) == ['d', 'i', 'n', 'o', 's', 'ss']
assert list(kv.items()) == [
  ('d', datetime.datetime.fromtimestamp(12325)), 
  ('i', 123), 
  ('n', decimal.Decimal('1234.56')), 
  ('o', {'d': decimal.Decimal('1234.58'), 
         'n': datetime.datetime.fromtimestamp(12325)}), 
  ('s', 'value'), 
  ('ss', {0, 1, 2, 3, 4, 5, 6, 7, 8, 9})
]

Set the reverse argument to True for the keys() and items() methods to sort in descending order.

Bulk inserting data

The SQLite DB backend can be very slow when bulk inserting data. You can use the insert method to insert efficiently in bulk.

kv.insert(((str(i), ':{}'.format(i)) for i in range(50000)))

The batch size is 1000 by default, you should modify it depending on the size of your data and available memory.

kv.insert(((str(i), ':{}'.format(i)) for i in range(50000)), batch_size=40000)

If you are inserting data from a generator and need to use the inserted data, use insert_generator method:

for key, value in kv.insert_generator(((str(i), ':{}'.format(i)) for i in range(50)), batch_size=10):
    print(key, value)

Installing leveldb

On Debian based Linux:

$ apt-get install libleveldb-dev libleveldb1

On Alpine based Linux:

$ apk --repository http://dl-3.alpinelinux.org/alpine/edge/testing/ --update add leveldb leveldb-dev

On OS X:

$ brew install leveldb