clue/graph

A mathematical graph/network library written in PHP

Note: While the APIs should be stable, this library is still considered beta software. Please see Contributing below for ways you can help out.

Quickstart examples

Once installed, let's initialize a sample graph:

<?php
require_once 'vendor/autoload.php';

use \Fhaculty\Graph\Graph as Graph;

$graph = new Graph();

// create some cities
$rome = $graph->createVertex('Rome');
$madrid = $graph->createVertex('Madrid');
$cologne = $graph->createVertex('Cologne');

// build some roads
$cologne->createEdgeTo($madrid);
$madrid->createEdgeTo($rome);
// create loop
$rome->createEdgeTo($rome);

Let's see which city (Vertex) has road (i.e. an edge pointing) to Rome

foreach ($rome->getVerticesEdgeFrom() as $vertex) {
    echo $vertex->getId().' leads to rome'.PHP_EOL;
    // result: Madrid and Rome itself
}

Features

This library is built around the concept of mathematical graph theory (i.e. it is not a charting library for drawing a graph of a function). In essence, a graph is a set of nodes with any number of connections inbetween. In graph theory, vertices (plural of vertex) are an abstract representation of these nodes, while connections are represented as edges. Edges may be either undirected ("two-way") or directed ("one-way", aka di-edges, arcs).

Depending on how the edges are constructed, the whole graph can either be undirected, can be a directed graph (aka digraph) or be a mixed graph. Edges are also allowed to form loops (i.e. an edge from vertex A pointing to vertex A again). Also, multiple edges from vertex A to vertex B are supported as well (aka parallel edges), effectively forming a multigraph (aka pseudograph). And of course, any combination thereof is supported as well. While many authors try to differentiate between these core concepts, this library tries hard to not impose any artificial limitations or assumptions on your graphs.

Graph drawing

The library supports visualizing graph images, including them into webpages, opening up images from within CLI applications and exporting them as PNG, JPEG or SVG file formats (among many others). Because graph drawing is a complex area on its own, the actual layouting of the graph is left up to the excelent GraphViz "Graph Visualization Software" and we merely provide some convenient APIs to interface with GraphViz.

Common algorithms

Besides graph drawing, one of the most common things to do with graphs is running algorithms to solve common graph problems. Therefor this library includes implementations for a number of commonly used graph algorithms:

• Search
  • Deep first (DFS)
  • Breadth first search (BFS)
• Shortest path
  • Dijkstra
  • Moore-Bellman-Ford (MBF)
  • Counting number of hops (simple BFS)
• Minimum spanning tree (MST)
  • Kruskal
  • Prim
• Traveling salesman problem (TSP)
  • Bruteforce algorithm
  • Minimum spanning tree heuristic (TSP MST heuristic)
  • Nearest neighbor heuristic (NN heuristic)
• Maximum flow
  • Edmonds-Karp
• Minimum cost flow (MCF)
  • Cycle canceling
  • Successive shortest path
• Maximum matching
  • Flow algorithm

Install

The recommended way to install this library is through composer. New to composer?

{
    "require": {
        "clue/graph": "0.7.*"
    }
}

Optional recommendation: In order to be able to use the graph drawing feature you'll have to install GraphViz (dot executable). Users of Debian/Ubuntu-based distributions may simply invoke sudo apt-get install graphviz, Windows users have to download GraphViZ for Windows and remaining users should install from GraphViz homepage.

Tests

This library uses phpunit for its extensive testsuite. You can either use a global installation or rely on the one composer installs when you first run $ composer install. This sets up the developer environment, so that you can now run it from the project root directory:

$ php vendor/bin/phpunit`

Contributing

As stated above, this library is still considered beta software and requires some contributions. While constructing graphs, plotting them and running all algorithms on them has been tested with sample graphs, there's a fair chance that we're missing some special cases for your graphs - partly because of some implied assumptions and mostly because of a lack of basic unit tests.

If you encounter any issues, please don't hesitate to drop us a line, file a bug report or even best provide us with a patch / pull request and/or unit test to reproduce your problem.

Besides directly working with the code, any additional documentation, additions to our readme or even fixing simple typos are appreciated just as well.

Any feedback and/or contribution is welcome!

Check out #graphp on irc.freenode.net.

License

Released under the terms of the permissive MIT license.