ccomps

decompose graphs into connected components

TLDR

Print components, each as a separate graph

$ ccomps -x [input.dot]

Write each component to a numbered file derived from outfile

$ ccomps -x -o [out.dot] [input.dot]

Print only the largest component

$ ccomps -z -x [input.dot]

Count the components, print only counts to stderr

$ ccomps -sv [input.dot]

Extract the component containing a specific node

$ ccomps -X [node_name] [input.dot]

Output components by size range (at least 5 nodes)

$ ccomps -x -X%5- [input.dot]

SYNOPSIS

ccomps [-svnxCe?] [-z] [-X[#%]v] [-o outfile] [files]

ccomps decomposes graphs into their connected components, printing the components to standard output. By default, every node ends up in some component; with -C the cluster subgraph structure is also used to determine connectivity.For undirected graphs it computes connected components; for directed graphs it treats edges as undirected (weakly connected). The tool is useful for filtering disconnected subgraphs out of a graph, isolating the component containing a node of interest, or pre-processing graphs before layout with dot, neato, etc.If no input file is given, the graph is read from standard input.

PARAMETERS

-s

Silent: print no output graph; useful with -v to just check connectivity.

-x

Print each connected component as a separate graph (instead of embedding them as subgraphs).

-X nodename_

Print only the component containing the named node.

-X# index[-[last]]

Print only the component(s) at the given index, or index range (last optional).

-X% min[-[max]]

Print only the component(s) whose node count falls in the given size range.

-z

Sort components by size, largest first. Only meaningful with -x or -X#.

-C

Use cluster subgraphs in addition to normal edge connectivity when computing components.

-e

Do not induce edges in the connected components.

-n

Do not project subgraph structure: components contain only relevant nodes and edges.

-v

Verbose: print counts of nodes, edges and components to stderr.

-o outfile

Write output to outfile. With multiple components, files are numbered (outfile, outfile\_1, ...).

Print usage information.

CAVEATS

Input must be in DOT format. For directed graphs only weakly connected components are computed. Memory usage scales with graph size. Node names must be unique for the -X nodename_ form to work as expected. The -X flag is overloaded: a plain name, a #-prefixed index, or a %-prefixed size selector.

HISTORY

ccomps is part of the Graphviz suite, originally developed at AT&T Labs Research starting in the early 1990s by Stephen North, Eleftherios Koutsofios, Emden Gansner and others. Graphviz was open-sourced in 2004 and continues to be actively maintained.