Skip to content


Switch branches/tags

Latest commit

- Module structure following example from NetworkX and other projects.
  This adds an `import pydot` to ``, which is cyclic/circular,
  but necessary to make the exceptions available. Cyclic imports are
  not uncommon and Python can handle this. Using the exceptions in
  pydot source code now requires prefixing `pydot.`, for example:

      raise pydot.Error("Message")

- Exception hierarchy changes. See ChangeLog and class docstrings in
  this commit for details. Additional background notes:

  - Removal of the unused exception class `InvocationException`: It was
    introduced in 842173c of 2008 (v1.0.2), but got in disuse when
    commits 9b3c1a1 and bc639e7 of 2016 (v1.2.0) fell back to using
    `Exception` and `assert` (`AssertionError`) again. If we ever need
    a custom class like this again in the future, it would probably
    have a different signature for context data (e.g. a DOT string,
    input and output), different parent class (or classes, e.g.
    `PydotException, CalledProcessError`) and perhaps a different name
    (e.g. ending in `...Error`), so no need to keep the old class
    around for that.

Further additions to the exception hierarchy are likely before the
final release of pydot 2.0.

Discussed in #171, #230 and #271.

Git stats


Failed to load latest commit information.

Build Status PyPI Code style: black



  • is an interface to Graphviz
  • can parse and dump into the DOT language used by GraphViz,
  • is written in pure Python,

and networkx can convert its graphs to pydot.

Development occurs at GitHub, where you can report issues and contribute code.


The examples here will show you the most common input, editing and output methods.


No matter what you want to do with pydot, it will need some input to start with. Here are 3 common options:

  1. Import a graph from an existing DOT-file.

    Use this method if you already have a DOT-file describing a graph, for example as output of another program. Let's say you already have this (based on an example from Wikipedia):

    graph my_graph {
       a [label="Foo"];
       b [shape=circle];
       a -- b -- c [color=blue];

    Just read the graph from the DOT-file:

    import pydot
    graphs = pydot.graph_from_dot_file("")
    graph = graphs[0]
  2. or: Parse a graph from an existing DOT-string.

    Use this method if you already have a DOT-string describing a graph in a Python variable:

    import pydot
    dot_string = """graph my_graph {
        a [label="Foo"];
        b [shape=circle];
        a -- b -- c [color=blue];
    graphs = pydot.graph_from_dot_data(dot_string)
    graph = graphs[0]
  3. or: Create a graph from scratch using pydot objects.

    Now this is where the cool stuff starts. Use this method if you want to build new graphs from Python.

    import pydot
    graph = pydot.Dot("my_graph", graph_type="graph", bgcolor="yellow")
    # Add nodes
    my_node = pydot.Node("a", label="Foo")
    # Or, without using an intermediate variable:
    graph.add_node(pydot.Node("b", shape="circle"))
    # Add edges
    my_edge = pydot.Edge("a", "b", color="blue")
    # Or, without using an intermediate variable:
    graph.add_edge(pydot.Edge("b", "c", color="blue"))

    Imagine using these basic building blocks from your Python program to dynamically generate a graph. For example, start out with a basic pydot.Dot graph object, then loop through your data while adding nodes and edges. Use values from your data as labels, to determine shapes, edges and so forth. This way, you can easily build visualizations of thousands of interconnected items.

  4. or: Convert a NetworkX graph to a pydot graph.

    NetworkX has conversion methods for pydot graphs:

    import networkx
    import pydot
    # See NetworkX documentation on how to build a NetworkX graph.
    graph = networkx.drawing.nx_pydot.to_pydot(my_networkx_graph)


You can now further manipulate your graph using pydot methods:

  • Add further nodes and edges:

    graph.add_edge(pydot.Edge("b", "d", style="dotted"))
  • Edit attributes of graph, nodes and edges:



Here are 3 different output options:

  1. Generate an image.

    To generate an image of the graph, use one of the create_*() or write_*() methods.

    • If you need to further process the output in Python, the create_* methods will get you a Python bytes object:

      output_graphviz_svg = graph.create_svg()
    • If instead you just want to save the image to a file, use one of the write_* methods:

  2. Retrieve the DOT string.

    There are two different DOT strings you can retrieve:

    • The "raw" pydot DOT: This is generated the fastest and will usually still look quite similar to the DOT you put in. It is generated by pydot itself, without calling Graphviz.

      # As a string:
      output_raw_dot = graph.to_string()
      # Or, save it as a DOT-file:
    • The Graphviz DOT: You can use it to check how Graphviz lays out the graph before it produces an image. It is generated by Graphviz.

      # As a bytes literal:
      output_graphviz_dot = graph.create_dot()
      # Or, save it as a DOT-file:
  3. Convert to a NetworkX graph.

    Here as well, NetworkX has a conversion method for pydot graphs:

    my_networkx_graph = networkx.drawing.nx_pydot.from_pydot(graph)

More help

For more help, see the docstrings of the various pydot objects and methods. For example, help(pydot), help(pydot.Graph) and help(pydot.Dot.write).

More documentation contributions welcome.


From PyPI using pip:

pip install pydot

From source:

python install


  • pyparsing: used only for loading DOT files, installed automatically during pydot installation.

  • GraphViz: used to render graphs as PDF, PNG, SVG, etc. Should be installed separately, using your system's package manager, something similar (e.g., MacPorts), or from its source.


Distributed under an MIT license.



Original author: Ero Carrera