Tree Layout

API Reference ▸ Layouts

The tree layout produces tidy node-link diagrams of trees using the Rheingold–Tilford “tidy” algorithm. For example, a tree layout can be used to organize software classes in a package hierarchy:

Like most other layouts, the object returned by d3.layout.tree is both an object and a function. That is: you can call the layout like any other function, and the layout has additional methods that change its behavior. Like other classes in D3, layouts follow the method chaining pattern where setter methods return the layout itself, allowing multiple setters to be invoked in a concise statement.

# d3.layout.tree()

Creates a new tree layout with the default settings: the default sort order is null; the default children accessor assumes each input data is an object with a children array; the default separation function uses one node width for siblings, and two node widths for non-siblings; the default size is 1×1.

The tree layout is part of D3's family of hierarchical layouts. These layouts follow the same basic structure: the input argument to the layout is the root node of the hierarchy, and the output return value is an array representing the computed positions of all nodes. Note that these position objects are not the same as the input data passed to the layout function; the computed layout nodes wrap the data objects, and provide several attributes:

• parent - the parent node, or null for the root.
• children - the array of child nodes, or null for leaf nodes.
• depth - the depth of the node, starting at 0 for the root.
• x - the computed x-coordinate of the node position.
• y - the computed y-coordinate of the node position.
• data - the underlying data represented by this node.

Although the layout has a size in x and y, this represents an arbitrary coordinate system; for example, you can treat x as a radius and y as an angle to produce a radial rather than Cartesian layout.

# tree.sort([comparator])

If comparator is specified, sets the sort order of sibling nodes for the layout using the specified comparator function. If comparator is not specified, returns the current group sort order, which defaults to null for no sorting. The comparator function is invoked for pairs of nodes, being passed the input data for each node. The default comparator is null, which disables sorting and uses tree traversal order. For example, to sort sibling nodes in descending order by the associated input data's numeric value attribute, say:

function comparator(a, b) {
  return b.value - a.value;
}

Sorting by the node's name or key is also common. This can be done easily using d3.ascending or d3.descending.

# tree.children([children])

If children is specified, sets the specified children accessor function. If children is not specified, returns the current children accessor function, which by default assumes that the input data is an object with a children array:

function children(d) {
  return d.children;
}

The children accessor is first invoked for root node in the hierarchy. If the accessor returns null, then the node is assumed to be a leaf node at the layout traversal terminates. Otherwise, the accessor should return an array of data elements representing the child nodes.

Often, it is convenient to load the node hierarchy using d3.json, and represent the input hierarchy efficiently as a nested JSON object, rather than an explicit array of children. In this case, the leaf nodes of the hierarchy may be simple numbers rather than objects. Using the children accessor function, you can easily tell the layout how to traverse the input hierarchy. For example, if the input is:

{
  "analytics": {
    "cluster": {
      "AgglomerativeCluster": 3938,
      "CommunityStructure": 3812,
      "MergeEdge": 743
    },
    "graph": {
      "BetweennessCentrality": 3534,
      "LinkDistance": 5731
    }
  }
}

Then a suitable children accessor will iterate over the values in each object, if the current node is an object, or return null for non-objects to indicate that the specified input is a leaf node:

function children(d) {
  return isNaN(d) ? d3.values(d) : null;
}

However, note that with this accessor, the data associated with each node will only have access to the child nodes, and will not know its own key! To store the name of the current node, in addition to the child nodes, we can use the d3.entries helper:

function children(d) {
  return isNaN(d.value) ? d3.values(d.value) : null;
}

With this children accessor, the input to the layout must itself be an object with key and value attributes. This can be achieved by saying d3.entries(object)[0], where object is the root JSON object.

# tree.links(nodes)

Given the specified array of nodes, such as the computed nodes returned by the tree layout, returns an array of objects representing the links from parent to child for each node. Leaf nodes will not have any links. Each link is an object with two attributes:

• source - the parent node (as described above).
• target - the child node.

This method is useful for retrieving a set of link descriptions suitable for display, often in conjunction with the diagonal shape generator. For example:

svg.selectAll("path")
    .data(tree.links(nodes))
  .enter().append("svg:path")
    .attr("d", d3.svg.diagonal());

# tree.separation([separation])

If separation is specified, uses the specified function to compute separation between neighboring nodes. If separation is not specified, returns the current separation function, which defaults to:

function separation(a, b) {
  return a.parent == b.parent ? 1 : 2;
}

A variation that is more appropriate for radial layouts reduces the separation gap proportionally to the radius:

function separation(a, b) {
  return (a.parent == b.parent ? 1 : 2) / a.depth;
}

The separation function is passed two neighboring nodes a and b, and must return the desired separation between nodes. The nodes are typically siblings, though the nodes may also be cousins (or even more distant relations) if the layout decides to place such nodes adjacent.

# tree.size([size])

If size is specified, sets the available layout size to the specified two-element array of numbers representing x and y. If size is not specified, returns the current size, which defaults to 1×1. Although the layout has a size in x and y, this represents an arbitrary coordinate system. For example, to produce a radial layout where the tree breadth (x) in measured in degrees, and the tree depth (y) is a radius r in pixels, say [360, r].