SkipDict

A skip dict is a Python dictionary which is permanently sorted by value. This package provides a fast, idiomatic implementation written in C with an extensive test suite.

The data structure uses a skip list internally.

An example use is a leaderboard where the skip dict provides logarithmic access to the score and ranking of each user, as well as efficient iteration in either direction from any node.

Compatible with Python 2.7+ and Python 3.3+.

Usage

The skip dict works just like a normal dictionary except it maps only to floating point values:

from skipdict import SkipDict

skipdict = SkipDict(maxlevel=4)

skipdict['foo'] = 1.0
skipdict['bar'] = 2.0

The SkipDict optionally takes a dictionary or a sequence of (key, value) pairs:

skipdict = SkipDict({'foo': 1.0, 'bar': 2.0})
skipdict = SkipDict(('foo', 1.0), ('bar', 1.0), ('bar', 1.0))

Note that duplicates are automatically aggregated to a sum. To illustrate this, we can count the occurrences of letters in a text:

skipdict = SkipDict(
    (char, 1) for char in
    "Everything popular is wrong. - Oscar Wilde"
)

# The most frequent letter is a space.
skipdict.keys()[-1] == " "

The skipdict is sorted by value which means that iteration and standard mapping protocol methods such as keys(), values() and items() return items in sorted order.

Each of these methods have been extended with optional range arguments min and max which limit iteration based on value. In addition, the iterator objects support the item and slice protocols:

>>> skipdict.keys(min=2.0)[0]
'bar'
>>> skipdict.keys(max=2.0)[1:]
['bar']

Note that the methods always return an iterator. Use list to expand to a sequence:

>>> iterator = skipdict.keys()
>>> list(iterator)
['bar']

The index(value) method returns the first key that has exactly the required value. If the value is not found then a KeyError exception is raised.

>>> skipdict.index(2.0)
'bar'

Alternatives

Francesco Romani wrote pyskiplist which also provides an implementation of the skip list datastructure for CPython written in C.

Paul Colomiets wrote sortedsets which is an implementation in pure Python. The randomized test cases were adapted from this package.

License

Copyright (c) 204 Malthe Borch <[email protected]>

This software is provided "as-is" under the BSD License.