The steps to achieve this uses multiple graphviz tools which can be piped together.
The following line is a possible configuration, graph.dot being the file which contains your graph(s). You may have to fiddle with the options.
ccomps -x graph.dot | dot | gvpack -array3 | neato -Tpng -n2 -o graph.png
And here's the explanation:
1. Separate disconnected graphs
Tool: ccomps
decomposes graphs into their connected components
The -x
option (Only the connected components are printed, as separate graphs) is probably all that is needed.
2. Layout each graph
Tool: dot
Each directed graph is layed out, one by one. This step is needed to get the position of the nodes and edges.
3. Pack all layed out graphs into one
Tool: gvpack
reads in a stream of graphs, combines the graphs into a single layout,
and produces a single graph serving as the union of the input graphs.
You should read the documentation of the options for this tool and play with the options. For example, -array
is used to lay out the graphs in a grid like manner, and offers several flags to control the layout.
4. Create the output
Tool: neato
The option -n2
tells neato to not layout the input graphs but to use the existing position attributes.
Example graph:
digraph G {
subgraph G1 {
a->{b; c;};
}
subgraph G2 {
d -> {e; f;};
}
subgraph G3 {
g -> h;
}
subgraph G4 {
i -> j;
}
subgraph G5 {
{k; l;} -> m;
}
}
Edit: Sorting the digraphs in gvpack
In order to determine the order of appearance od the subgraphs in the combined layout created by gvpack
, each subgraph will need a sortv
attribute.
For example, the following graphs:
digraph G1 {
sortv=1;
a->{b; c;};
}
digraph G2 {
sortv=2;
d -> {e; f;};
}
digraph G3 {
sortv=3;
g -> h;
}
digraph G4 {
sortv=4;
i -> j;
}
digraph G5 {
sortv=5;
{k; l;} -> m;
}
can be transformed using
dot graph.dot | gvpack -array_u | neato -Tpng -n2 -o graph.png
resulting in