Search code examples
ubuntumatplotlibnetworkxfont-family

Available "font-family" entry for draw_networkx_labels


I have this code:

    labels_params = {"labels":{n: "" if n[0] == "." else n for n in G.nodes () },
                     "font_family":"sans-serif",
                     "alpha":.5,
                     "font_size":16 }

    draw_networkx_labels (G, pos, **labels_params)

My problem is that I wish to be able to display in my output graph more than 2 fonts .

I wish to know how can I detect all possible entries for "font_family".

I browsed the code of FontManager from Networkx, and I see only "sans-serif".

I am working in X11 under Ubuntu.


Solution

  • Poking around a bit in the draw_networkx_labels source, it does indeed come down to a call to ax.text (where ax is a matplotlib axis). So that means you should have as much configurability as you get with any normal MPL text (docs).

    Font names vs font families

    As far as I can tell, a font name is an instance of a font family; though because the generic families (eg 'serif') are often given as settings to font family variables. http://www.w3schools.com/css/css_font.asp

    This has confused me for a while, so please correct me if I'm wrong here.

    Finding the full list of fonts

    So if you use the technique from here:

    avail_font_names = [f.name for f in matplotlib.font_manager.fontManager.ttflist]
    

    you get all the (specific) options. Don't know where you find a fuller list than the font demo for generics.

    this post shows a method for searching by name:

    [i for i in matplotlib.font_manager.findSystemFonts(fontpaths=None, fontext='ttf') if 'times' in i.lower()]
    

    Using more than one font for labels in your graph

    From the avail_font_names list above, I picked out three for this example; you might have to substitute some of them depending on what you have installed.

    import matplotlib.pyplot as plt
    import networkx as nx
    
    font_names = ['Sawasdee', 'Gentium Book Basic', 'FreeMono', ]
    family_names = ['sans-serif', 'serif', 'fantasy', 'monospace']
    
    
    # Make a graph
    G  = nx.generators.florentine_families_graph()
    
    # need some positions for the nodes, so lay it out
    pos = nx.spring_layout(G)
    
    # create some maps for some subgraphs (not elegant way)
    subgraph_members  = [G.nodes()[i:i+3] for i in xrange(0, len(G.nodes()), 3)]
    
    plt.figure(1)
    nx.draw_networkx_nodes(G, pos)
    
    
    for i, nodes in enumerate(subgraph_members):
        f = font_names[(i % 3)]
        #f = family_names[(i % 4)]
        # extract the subgraph
        g = G.subgraph(subgraph_members[i])
        # draw on the labels with different fonts
        nx.draw_networkx_labels(g, pos, font_family=f, font_size=40)
    
    # show the edges too
    nx.draw_networkx_edges(G, pos)
    
    
    plt.show()    
    

    example figure showing different font labelling

    Resolving the 'Falling back to ...' warnings

    Note: if you get errors when of the form "UserWarning: findfont: Font family ['sans-serif'] not found. Falling back to ...", when trying to use fonts even if they do exist, this nabble dialog suggests clearing the font cache: (yup, this will irretrievably remove a file but it is auto-generated.)

    rm ~/.matplotlib/fontList.cache