PyGraphviz

Warnings

• breaking PyGraphviz 0.32 introduced significant API changes, being a rewrite not backward compatible with earlier versions (e.g., 0.2x). Programs written for older versions will not work without modification.
  Fix: Review the PyGraphviz 0.32+ documentation and API for the new methods and object structures. The API became similar to NetworkX.
• gotcha The Graphviz C library (not the Python `graphviz` package) is a mandatory prerequisite and must be installed separately on your operating system *before* installing `pygraphviz`. Failure to do so typically results in `ImportError: libagraph.so.1: cannot open shared object file` or similar compilation errors during `pip install`.
  Fix: Install the Graphviz C library using your system's package manager (e.g., `sudo apt-get install graphviz graphviz-dev` on Debian/Ubuntu, `brew install graphviz` on macOS) or via official installers for Windows. Ensure Graphviz binaries and libraries are on your system's PATH.
• gotcha Installing PyGraphviz on Windows can be particularly challenging. It often requires specific Graphviz versions (e.g., 2.46+ for PyGraphviz 1.7+), a C/C++ compiler (like Visual C++ Build Tools), and potentially manually specifying Graphviz include and library paths during `pip install` using `--config-settings`.
  Fix: Follow the detailed Windows installation instructions in the official PyGraphviz documentation, including installing Visual C++ Build Tools and potentially using `pip install --config-settings` to point to Graphviz directories. Consider using conda or official Graphviz binaries.
• breaking PyGraphviz regularly drops support for older Python versions. For example, PyGraphviz 1.14 dropped Python 3.8 support. PyGraphviz 1.9 dropped Python 3.7, and 1.7 dropped 3.6.
  Fix: Always check the `requires_python` metadata on PyPI or the official documentation for compatibility with your Python version. Upgrade Python if necessary or use an older PyGraphviz version compatible with your environment.
• breaking Specific minimum Graphviz C library versions are required by newer PyGraphviz releases. For instance, PyGraphviz 1.7 requires Graphviz 2.46.0 or higher, and PyGraphviz 1.11 also requires Graphviz 2.46+.
  Fix: Ensure your system's Graphviz C library installation meets or exceeds the minimum version specified in the PyGraphviz documentation. Upgrade Graphviz if it's too old.

Install

• pip install pygraphviz Standard pip installation (requires Graphviz C library)
• sudo apt-get install graphviz graphviz-dev pip install pygraphviz Ubuntu/Debian with Graphviz
• brew install graphviz pip install pygraphviz macOS (Homebrew) with Graphviz

Imports

• AGraph
  wrong:

  import pygraphviz.AGraph

  correct:

  from pygraphviz import AGraph

  AGraph is the primary class for graph objects and is typically imported directly or via 'import pygraphviz as pgv; pgv.AGraph()'.

Quickstart

This quickstart demonstrates how to create a directed graph, add nodes and edges with attributes, set global graph attributes, lay out the graph using a specified engine (e.g., 'dot'), and render it to a PNG image file.

import pygraphviz as pgv

# Create a directed graph
G = pgv.AGraph(directed=True, strict=False)

# Add nodes
G.add_node('A', shape='box', color='red')
G.add_node('B', shape='circle')
G.add_node('C', label='Node C')

# Add edges
G.add_edge('A', 'B', label='connects', penwidth=2)
G.add_edge('B', 'C', color='blue', style='dashed')
G.add_edge('C', 'A')

# Set graph attributes
G.graph_attr['label'] = 'My Sample Graph'
G.graph_attr['overlap'] = 'false'
G.graph_attr['splines'] = 'true'

# Layout the graph (using dot engine by default)
# and draw to a file
G.layout(prog='dot')
output_filename = 'sample_graph.png'
G.draw(output_filename)

print(f"Graph saved to {output_filename}")
# print(G) # Uncomment to see the DOT language representation