Searching the Scene Graph

It is often useful to get a handle to a particular node deep within the scene graph, especially to get a sub-part of a model that was loaded from a single file. There are a number of methods dedicated to finding entrenched nodes and returning the NodePaths.

First, and most useful, is the ls() command:

This simply lists all of the children of the indicated NodePath, along with all of their children, and so on until the entire subgraph is printed out. It also lists the transforms and Render Attributes that are on each node. This is an especially useful command for when you’re running interactively with Python; it’s a good way to check that the scene graph is what you think it should be.

The two methods find() and findAllMatches() will return a NodePath and a NodePathCollection respectively. These methods require a path string as an argument. Searches can be based on name or type. In its simplest form this path consists of a series of node names separated by slashes, like a directory pathname. When creating the string each component may optionally consist of one of the following special names, instead of a node name.


Matches exactly one node of any name


Matches any sequence of zero or more nodes


Matches any node that is or derives from the given type


Matches any node that is the given type exactly


Matches any node that has the indicated tag


Matches any node whose tag matches the indicated value

Furthermore, a node name may itself contain standard filename globbing characters, like *, ?, and [a-z], that will be accepted as a partial match. (In fact, the ‘*’ special name may be seen as just a special case of this.) The globbing characters may not be used with the typename matches or with tag matches, but they may be used to match a tag’s value in the =tag=value syntax.

The special characters “@@”, appearing at the beginning of a node name, indicate a stashed node. Normally, stashed nodes are not returned by a find (but see the special flags, below), but a stashed node may be found if it is explicitly named with its leading @@ characters. By extension, “@@*” may be used to identify any stashed node.


"room//graph" will look for a node named “graph”, which is a child of an unnamed node, which is a child of a node named “room”, which is a child of the starting path.

"**/red*" will look for any node anywhere in the tree (below the starting path) with a name that begins with “red”.

"**/+PartBundleNode/**/head" will look for a node named “head”, somewhere below a PartBundleNode anywhere in the tree.

The argument may also be followed by one or more optional control flags. To use a control flag, add a semicolon after the argument, followed by at least one of the special flags with no extra spaces or punctuation.


Do not return hidden nodes


Return hidden nodes


Do not return stashed nodes unless explicitly referenced with @@


Return stashed nodes even without any explicit @@ characters


Node name comparisons are not case insensitive: case must match exactly


Node name comparisons are case insensitive: case is not important. This affects matches against the node name only; node type and tag strings are always case sensitive

The default flags are +h-s-i.

The find() method searches for a single node that matches the path string given. If there are multiple matches, the method returns the shortest match. If it finds no match, it will return an empty NodePath. On the other hand, findAllMatches() will return all NodePaths found, shortest first.


Some examples:


This will look for a node named “door”, which is a child of a node named “house”, which is a child of the starting path.


This will look for any node anywhere in the tree (below the starting path) with a name that begins with “red”.


This will search shipNP recursively using tag/value. Tag name is “type” and tag value is “weaponMount”. All matches found will be returned.

In addition there are also the methods getParent() and getChildren(). getParent() returns the NodePath of the parent node. getChildren() returns the children of the current node as a NodePathCollection.

The NodePathCollection can be treated like any Python sequence:

for child in myNodePath.getChildren():

For more information and a complete list of NodePath functions please see the NodePath page in the API Reference.