decoding.py
    Bitarray's decode method is implemented in C.  Since the C code
    might be hard to read, we have implemented exactly the same
    algorithm in Python.  It is about 20 times slower than it's
    C counterpart, since (recursive) function calls are more expensive
    in Python than in C.


huffman.py
    Demonstrates building a Huffman tree.  Given an input file,
    calculates the number of occurrences for each character;
    from those frequencies, a Huffman tree is build; and by traversing
    the tree, the Huffman code is evaluated.
    Also allows encoding and decoding of a file, see -h option.


mandel.py
    Generates a .ppm image file of size 8000x6000 of the Mandelbrot set.
    Despite it's size, the output image file has only a size of slightly
    over 6 Million bytes (uncompressed) because each pixel is stored in
    one bit.
    Requires numpy and scipy (see http://scipy.org/).
    Not supported by Python 3.x.


ndarray.py
    Demonstrates how to efficiently convert boolean data from a bitarray
    to a numpy.ndarray of dtype bool.
    Requires numpy.


pbm.py
    Defines a simple class called PBM (Portable Bit Map) which allows:
    - addressing pixels by their coordinates
    - storing and loading .ppm (P4), which is the same as .pbm, files


sieve.py
    Sieve of Eratosthenes is a simple, ancient algorithm for finding all
    prime numbers up to a specified integer.  In this exmaple, the algorithm
    is implemented using the numpy ndarray as well as the bitarray object.
    Thanks Steve for emailing this example.


smallints.py
    A class is defined which allows efficiently storing an array of
    integers represented by a specified number of bits (1 through 8).
    For example, an array with 1000 5 bit integers can be created,
    allowing each element in the array to take values form 0 to 31,
    while the size of the object is 625 (5000/8) bytes.
    Thanks to David Kammeyer for the idea to apply a bitarray in this way.