How to pass additional style and color options to graphviz?

My application has many streams of different sources, how can I pass some additional style of color params to graphviz? FOr example I would like streams that start via xls files to be red, streams that start via xml files to be yellow and streams that take input from websockets to be blue. It would also be nice to pass color and style options to intermediate nodes like map, combine latest, etc. Is this possible?

[CJ-Wright]

Currently these are set as attributes on the classes, see https://github.com/python-streamz/streamz/blob/master/streamz/core.py#L156-L159

There are then plugged into graphviz here, https://github.com/python-streamz/streamz/blob/master/streamz/graph.py#L80-L84

In the future we might consider having a _graphviz dict and just unpack that into graphviz rather than having individual attributes.

Thanks! @CJ-Wright I'l look into injecting my styles there.

@CJ-Wright Would you be interested in a pull request that adds an optional dict keyword arg called graphviz_styles that if supplied overwrites the static styles?

[martindurant]

Sounds reasonable to me

[CJ-Wright]

I think that would be great!

[badiku]

there are many options or parameters for function graphviz.Digraph,
like graph_attr, node_attr, edge_attr.
but currently, in function to_graphviz, this line:

streamz/streamz/graph.py

Line 134 in f9b4ce3

gvz = graphviz.Digraph(graph_attr=graph_attr)

it pass only graph_attr .

to pass more options to function graphviz.Digraph,
I modified this line to:

gvz = graphviz.Digraph(**graph_attr)

so as to add node_attr like fontname to support Other language.

import graphviz
graphviz.Digraph?

Init signature:
graphviz.Digraph(
    name=None,
    comment=None,
    filename=None,
    directory=None,
    format=None,
    engine=None,
    encoding='utf-8',
    graph_attr=None,
    node_attr=None,
    edge_attr=None,
    body=None,
    strict=False,
)

[martindurant]

@badiku : that sounds reasonable (please submit a PR), but do note that the individual nodes and edges on the lines that follow have their own kwargs too, taken from node-specific attributes.