I'm trying to render a tree that is very broad... and it renders, as expected, in a long, skinny horizontal image.
Problem is that I need a graph suitable for a document. I would very much like to take and move the nodes that are rendered horizontally and "drag" them down so that the graph is more vertical... with the edges curving to accommodate this. Are there any clever ways to accomplish this? GraphViz settings? Third party tools that let me manipulate and fine tune the output? I work mostly in the Python ecosystem, but open to others. Also open to the use of tools like Visio and other pro drawing tools. Thanks!
Edit
After implementing the answer below by #sroush, and then tweaking a little further with Photoshop, got some nice results.
Tweaking the above in Photosop. Had to add the two curved edges after the secondary node by hand, but it's worth it. Much more presentable.
I assume you are using dot, and your graph "naturally" has only a few ranks (rows).
There are a few tweaks that will help a bit (reducing node horizontal footprints):
node [shape=rect] // snugger fit into rectangles
insert newlines into node labels e.g. xxx [label="Controller Board\n#19_8"])
Also try the unflatten program (https://www.graphviz.org/pdf/unflatten.1.pdf). It will increase the apparent number of ranks (rows).
See related question here with command line examples:
Distribute nodes on the same rank of a wide graph to different lines
You can use the minlen property to limit the minimum level span of some edges.This avoids the result becoming very long in the horizontal position.
For example:
digraph {
a->b
a->c
a->d
a->e
}
This will output the following image:
But when minlen is used, the picture will become longer vertically but shortened horizontally:
digraph {
a->b
a->c
a->d[minlen=2]
a->e[minlen=3]
}
Related
I read a couple of posts on position nodes in force layout but didn't find an answer to what I was looking for.
I have an object with nodes and links.
I' trying to create a graph which would show all the nodes top to bottom.
I was looking at the example code from here:
https://github.com/danielstern/force-graph-example
Here's a screenshot of the result:
I'm trying to find a way to position each node so the nodes without parents would be on the top and the ones connecting to them would be under them and so forth.
Here's an image to illustrate it:
Right now, all the nodes are scattered randomly.
I wanted to if I need to actually calculate the position of each node in a vertical view or is there a smarter/built-in way to achieve it.
I looked at this example which looked promising:
How to organise node positions in D3 Force layout
But in my case I don't have a way to differentiate between nodes levels so I don't think the yPostion would help.
I was also looking at thes post:
d3.js - How can I expand force directed graph horizontally?
According to #Lars Kotthoff:
"The point of the force layout is to automatically lay out a graph like this so that you don't have to specify the positions of the nodes yourself".
Since my graph is not really a tree, I don't think the tree view would match.
What would be my best approach to position the nodes?
Or perhaps there's a better library to achieve what I need?
I found this package:
d3-dag
It basically supports what I need:
"Often data sets are hierarchical, but are not in a tree structure..."
Here's an exmaple:
exmaple
I've spent hours searching for an answer to this, but in most cases either the
question is about plots/charts (rather than graphs as in "control flow graph"),
or the answer "just use graphviz" is a valid answer.
However I have some constraints and requirements that make "just use graphviz" a
non-answer.
The full graph is large enough that it's not possible to generate a graphviz
for all of it.
Nodes and edges will be dynamically added and removed.
Nodes have lots of information that will be hidden by default and will be
expanded on request (imagine every node as a table with expandable rows/cols)
I want to be able to show only a subset of the graph on request, e.g. for
features like "only show reachable part of the graph from this node" or "show
all simple paths from this node to this node".
Basically I want to be able to start drawing nodes and edges on a 2D plane, and
add new nodes and edges dynamically. It's fine if nodes/edges move around as new
stuff is added. While I don't yet have hard requirements for this, it'd be good
if it looked "nice" -- for example if a node has lots of incoming edges (this is
a directed graph) ideally it'd be in a central place on the plane with all other
nodes around it etc.
Anything that gets me going would be helpful. Thanks.
(I don't know what label to add to this, adding "graph-theory" because I don't know what else to add)
Let's say you're placing rectangular tooltips on a screen of elements you want to provide information for. You want all these tooltips to be visible all at once and not cover any of the nodes any of the other tooltips are for.
You want each tooltip to be as close to the item its related to as feasible. What algorithm(s) exist to help solve this problem?
I've checked out rtrees, which seem to only help you find collisions, but don't help on the front of actually searching for free locations. I've found rectangle packing algorithms that search for a position unconstrained by a maximization function (like "be closest to this other element as possible").
I can imagine an algorithm that has some physics simulation where nodes and their tooltips are each connected by some kind of rubber band and plays it out until equilibrium, but I'd think that things could be calculated faster and less complicated than that.
Any related algorithms or libraries would be helpful. Bonus points for a javascript library : )
You might investigate map labeling algorithms.
See, for example, these lecture notes by Robero Tamassia #Brown:
PDF download.
I need to display a Directed Acyclic Graph in a web page. I am not looking for an off-the-shelf library or solution. I am looking for suggestions, recommendations or a push in the right direction.
1. DAG Visualization
I am not sure of how the nodes and relations will be represented. Viable solutions may be Treemaps, the good old node & line or a combination of that two. I don't have a problem if one node appears more than once on the screen.
I don't need all the nodes to appear on the screen from the start. The user may expand a node by double clicking or zooming for example.
I am open to all suggestions and advices.
2. Technology
There are some functionalities that the implementation must have:
drag & drop
zoom
events on mouse interaction with nodes
From my point of view, I have 2 options (Flash is out of the question):
a. HTML5 Canvas
Disadvantages: no vectors, basically just an image; no implicit mouse events on nodes;
Advantages: speed; popularity; animations
b. SVG
Disadvantages: low speed when there are many nodes;
Advantages: vector graphics; elements are in the DOM so you can have events and so on;
c. A mix of HTML5 Canvas & SVG
Assuming you want to dynamically update your graph, you could probably use python on the server with the pydot GraphViz module.
I have not tried this, but it's something worth looking into.
I have a map that I converted from a raster graphic into an SVG file by converting the differently coloured areas into paths.
I know how to do a basic point-in-polygon check given an array of edges, but the svg:path elements represent multiple polygons as well as masks (to account for seas etc) and extracting that information by parsing the d attribute seems rather heavy-handed.
Is there a JS library that allows me to simplify that check? I basically want to create random points and then check whether they are on land (i.e. inside the polygons) or water (i.e. outside).
As SVG elements seem to allow for mouse event handling, I would think that this shouldn't be much of a problem (i.e. if you can tell whether the mouse pointer is on top of an element, you are already solving the point-in-polygon problem).
EDIT: Complicating the matter a bit, I should mention that the svg:path elements seem to be based on curves rather than lines, so just parsing the d attribute to create an array of edges doesn't seem to be an option.
As the elements can take a fill attribute, a ghetto approach of rendering the SVG on a canvas and then finding the colour value of the pixel at the given point could work, but that seems like a really, really awful way to do it.
The answers on Hit-testing SVG shapes? may help you in this quest. There are issues with missing browser support, but you could perhaps use svgroot.checkIntersection to hit test a small (perhaps even 0 width/height would work?) rectangle within your polygon shape.
The approach I suggested as a last resort seems to be the easiest solution for this problem.
I found a nice JS library that makes it easy to render SVG on a canvas. With the SVG rendered, all it takes is a call to the 2D context's getImageData method for a 1x1 region at the point you want to check. I guess it helps to create a copy of the SVG with colour coding to make the check easier if your SVG is more complex than the one I'm using (you'll have to check the RGBA value byte-by-byte).
This feels terribly hackish as you're actually inspecting the pixels of a raster image, but the performance seems to be decent enough and the colour checks can be written in a way that allows for impurities (e.g. near the edges).
I guess if you want relative coordinates you could try creating a 1-to-1 sized canvas and then divide the pixel coordinates by the canvas dimensions.
If somebody comes up with a better answer, I'll accept it instead. Until then, this one serves as a placeholder in case someone comes here with the same problem looking for an easy solution.