networkx-gdf 1.3

Python package to read and write NetworkX graphs as GDF (Graph Data Format).

networkx-gdf

Python package to read and write NetworkX graphs as GDF (Graph Data Format).

GDF is a compact file format originally implemented by GUESS. Although the software itself is not anymore maintained, the format is still supported by active open-source projects such as Gephi (see details here).

Requirements

• Python>=3.7
• networkx>=2.1
• pandas>=1.1.0

Install

Package is available to install on PyPI:

pip install networkx-gdf

Usage

Just two functions are implemented, for reading from and writing data to file.

from networkx_gdf import read_gdf, write_gdf

# Builds NetworkX graph object from file.
G = read_gdf("input_file.gdf")

# Writes NetworkX graph object to file.
write_gdf(G, "output_file.gdf")

Detailed usage

A few additional arguments are available both for reading and writing files:

Building NetworkX graph object from file

G = read_gdf(
    "input_file.gdf",
    # Required; file path to read data from.

    directed=None,
    # Optional; consider edges as directed or undirected.
    # True: returns a <nx.DiGraph> or <nx.MultiDiGraph> object.
    # False: returns a <nx.Graph> or <nx.MultiGraph> object.
    # None (default): decides based on "directed" edge attribute in file.

    multigraph=None,
    # Optional; consider multiple or single edges among nodes.
    # True: always returns a <nx.MultiGraph> or <nx.MultiDiGraph> object.
    # False: sums edge weights and returns a <nx.Graph> or <nx.DiGraph> object.
    # None (default): decides based on number of edges among the same pair of nodes.

    node_attr=None,
    # Optional; node attributes to import data from.
    # Accepts either a list containing attribute names or a boolean (True, False).
    # Defaults to True, which considers all node attributes.

    edge_attr=None,
    # Optional; edge attributes to import data from.
    # Accepts either a list containing attribute names or a boolean (True, False).
    # Defaults to True, which considers all edge attributes.
)

Writing NetworkX graph object to file

write_gdf(
    G,
    # Required; graph object from NetworkX.

    "output_file.gdf",
    # Required; file path to write data to.

    node_attr=None,
    # Optional; node attributes to export data from.
    # Accepts either a list containing attribute names or a boolean (True, False).
    # Defaults to True, which considers all node attributes.

    edge_attr=None,
    # Optional; edge attributes to export data from.
    # Accepts either a list containing attribute names or a boolean (True, False).
    # Defaults to True, which considers all edge attributes.
)

Note that any object types beyond integers/longs, floats/doubles, and booleans will be considered as strings.

Importing as a class

Alternatively, the module can also be imported as a class and inherited:

from networkx_gdf import GDF

class MyClass(GDF):
    ...

G = MyClass.read_gdf("input_file.gdf")
MyClass.write_gdf(G, "output_file.gdf")

Both functions are implemented as static methods, so initializing the class as an object is not required.

---

References

• NetworkX
• Pandas