pylibbvg 2.0.5

A python interface for decompressing Boldi-Vigna graphs

This library implements a decoder for a Boldi-Vigna graph structure, a highly compressed means of storing web-graphs. The library is written in pure C. It includes a Python interface, and a means to implement multi-threaded graph algorithms.

Features

• In-memory and on-disk graph storage

• Sequential iteration over graph edges

• Parallel iteration over graph edges

• random access to graph edges

Installation

1. Compile the C library:

   make

2. Compile the Cython source code(Cython is assumed to have installed):

   python setup.py build_ext --inplace

3. To clean the library:

   python setup.py clean
   make clean

Synopsis

To use the library in Python code:

import bvg
G = bvg.BVGraph('wb-cs.stanford', 0) # sequential scan
print 'nodes = ' + str(G.nverts)
print 'edges = ' + str(G.nedges)
edges_and_degrees = G.edges_and_degrees()
for (src, dst, degree) in edges_and_degrees:
    print 'node ' + str(src) + ' has degree ' + str(degree)
    print 'edge: ' + str(src) + ' -> ' + str(dst)

License

GPL version 2.

Copyright by David F. Gleich, 2007-2014