Nodes

Dot[source]

Dot(defaults=None, rankdir='LR', directed=True, compound=True, **kwargs)

Create a pydot.Dot graph with fastai/fastdot style defaults

uniq_name[source]

uniq_name(o)

quote[source]

quote(x, q='"')

Surround x with "

add_mapping[source]

add_mapping(graph_item, obj)

Node[source]

Node(obj, **kwargs)

Create a pydot.Node with a unique name

pydot uses the same name-based approach to identifying graph items as graphviz. However we would rather use python objects. Therefore, we patch pydot to use unique names.

g = Dot()
a = Node('a')
g.add_node(a)
g

If a 2-tuple is passed to add_node, then the 2nd element becomes the tooltip. You can also pass any kwargs that are accepted by graphviz.

g = Dot()
g.add_node(Node(['a', "My tooltip"], fillcolor='pink'))
g

Keyword args can also be arbitrary functions, which will called with the node's label.

g = Dot()
o = 'a'
g.add_node(Node(o, fillcolor=lambda o:'pink'))
g

object2graph[source]

object2graph(o)

Get graph item representing o

object2graph(o).get_fillcolor()

'"pink"'

Colors

The callable kwargs functionality can be used to map labels to colors in a consistent way..

obj2node_color[source]

obj2node_color(cm, minalpha, rangealpha, o)

Create a consistent mapping from objects to colors, using colormap cm

graph_colors1 = partial(obj2node_color, plt.get_cmap('rainbow'), 30, 160)
graph_colors2 = partial(obj2node_color, plt.get_cmap('tab20'), 30, 160)

These predefined color mapping functions provide a good range of colors and readable text.

g = Dot()
g.add_node(Node('a', fillcolor=graph_colors1))
g.add_node(Node('b', fillcolor=graph_colors1))
g

g = Dot()
g.add_node(Node('a', fillcolor=graph_colors2))
g.add_node(Node('b', fillcolor=graph_colors2))
g

We'll use the former color function as our default. You can change it by simply modifying node_defaults.

Clusters and Items

Cluster[source]

Cluster(obj='', **kwargs)

Create a pydot.Cluster with a unique name

g = Dot()
sg = Cluster('clus', tooltip='Cluster tooltip')
sg.add_node(Node(['a', "My tooltip"]))
sg.add_node(Node('b'))
g.add_subgraph(sg)
g

Graph.nodes[source]

Graph.nodes()

ith node in Graph

Graph.__getitem__[source]

Graph.__getitem__(i)

ith node in Graph

You can subscript into a Graph's Nodes by index:

print(sg[0].get_label())

"a"

Graph.add_item[source]

Graph.add_item(item, **kwargs)

Add a Cluster, Node, or Edge to the Graph

There's no good reason to have different methods for adding clusters vs nodes (as pydot requires), so we provide a single method.

g = Dot()
sg = Cluster('clus')
g.add_item(sg)
sg.add_item('a')
g

Graph.add_items[source]

Graph.add_items(*items, **kwargs)

Add items the Graph

graph_items[source]

graph_items(*items, **kwargs)

Add items to a new pydot.Dot

sg1 = Cluster('clus')
sg1.add_items('n1', 'n2')
sg2 = Cluster()
sg2.add_item('n')
graph_items(sg1,sg2)

Edges

Graph.first[source]

Graph.first()

First node in Graph, searching subgraphs recursively as needed

Graph.last[source]

Graph.last()

Lastt node in Graph, searching subgraphs recursively as needed

Node.with_compass[source]

Node.with_compass(compass=None)

Graph.with_compass[source]

Graph.with_compass(compass=None)

Node.connect[source]

Node.connect(item, compass1=None, compass2=None, **kwargs)

Connect two nodes or clusters

Graph.connect[source]

Graph.connect(item, compass1=None, compass2=None, **kwargs)

Connect two nodes or clusters

sg2 = Cluster('clus2')
n1 = sg2.add_item('n1', fillcolor='pink')
n2 = sg2.add_item('n2', fillcolor='lightblue')
sg2.add_item(n1.connect(n2))

sg1 = Cluster('clus1')
sg1.add_item(sg2)

a,b = Node('a'),Node('b')
edges = a.connect(b),a.connect(a),sg1.connect(b),sg2[0].connect(a)
g = Dot()
g.add_items(sg1, a, b, *edges)
g

object_connections[source]

object_connections(conns)

Create connections between all pairs in conns

This is a shortcut for creating connections between objects that are already in a graph.

a,b = 'a','b'
g = graph_items(a, b)
g.add_items(*object_connections([(a,b)]))
g

Sequential

Since it's common to want to connect a series sequentially, we provide some simple shortcuts for this functionality.

graph_edges_seq[source]

graph_edges_seq(items)

Add edges between each pair of nodes in items

Graph.add_edges_seq[source]

Graph.add_edges_seq(items)

Add edges between each pair of nodes in items

g = Dot()
its = g.add_items('a','b','c')
g.add_edges_seq(its)
g

seq_cluster[source]

seq_cluster(items, cluster_label='', **kwargs)

g = Dot()
g.add_item(seq_cluster(['a','b','c'], 'clust'))
g.add_item(seq_cluster(['1','2','c'], 'clust2'))
g

g = Dot()
g.add_item(seq_cluster(['a','b','c'], 'clust'))
g

sg1 = seq_cluster(['a','b','c'], 'clust1')
sg2 = seq_cluster(['a1','a2',sg1], 'clust2')
g = Dot()
g.add_item(sg2)
g

sg1 = seq_cluster(['inp'], 'clust1')
sg2 = seq_cluster(['a','b','c'], 'clust2')
sg2.add_items(sg1.connect(sg2[-1]), sg1.connect(sg2))
g = Dot()
g.add_items(sg1,sg2)
g

Point[source]

Point(label='pnt', **kwargs)

Create a Node with a 'point' shape

sg = Cluster('clus')
a,b,c = sg.add_items('a','b','c')
p = sg.add_item(Point())
sg.add_item(p.connect(c))
sg.add_items(p.connect(a), a.connect(b), b.connect(c))

g = Dot()
g.add_items(sg)
g