Customizing your Glue environment

Using a file as described in Configuring Glue via a startup file, you can customize many aspects of your Glue environment, which are described in the following sections.


Before we talk about the different components of the Glue environment that you can customize, we first need to look at registries. Glue is written so as to allow users to easily register new data viewers, tools, exporters, and more. Registering such components can be done via registries located in the glue.config sub-package. Registries include for example link_function, data_factory, colormaps, and so on. As demonstrated below, some registries can be used as decorators (see e.g. Custom Link Functions) and for others you can add items using the add method (see e.g. Custom Colormaps).

In the following sections, we show a few examples of registering new functionality, and a full list of available registries is given in Complete list of registries.

Custom Data Loaders

Glue lets you create custom data loader functions, to use from within the GUI.

Here’s a quick example: the default image loader in Glue reads each color in an RGB image into 3 two-dimensional components. Perhaps you want to be able to load these images into a single 3-dimensional component called cube. Here’s how you could do this:

from glue.config import data_factory
from glue.core import Data
from import imread

def is_jpeg(filename, **kwargs):
    return filename.endswith('.jpeg')

@data_factory('3D image loader', is_jpeg)
def read_jpeg(file_name):
    im = imread(file_name)
    return Data(cube=im)

Let’s look at this line-by-line:

  • The is_jpeg function takes a filename and keywords as input, and returns True if a data factory can handle this file
  • The @data_factory decorator is how Glue “finds” this function. Its two arguments are a label, and the is_jpeg identifier function
  • The first line in read_jpeg uses scikit-image to load an image file into a NumPy array.
  • The second line constructs a Data object from this array, and returns the result.

If you put this in your file, you will see a new file type when loading data:


If you open a file using this file type selection, Glue will pass the path of this file to your function, and use the resulting Data object.

If you are defining a data factory that may clash with an existing one, for example if you are defining a loader for a specific type of FITS file, then make sure that the identifier function (e.g. is_jpeg above) returns True only for that specific subset of FITS files. Then you can set the priority= keyword in the @data_factory decorator. The value should be an integer or floating-point number, with larger numbers indicating a higher priority.

For more examples of custom data loaders, see the example repository.

Custom importers

The Custom Data Loaders described above allow Glue to recognize more file formats than originally implemented, but it is also possible to write entire new ways of importing data, including new GUI dialogs. An example would be a dialog that allows the user to query and download online data.

Currently, an importer should be defined as a function that returns a list of Data objects. In future we may relax this latter requirement and allow existing tools in Glue to interpret the data.

An importer can be defined using the @importer decorator:

from glue.config import importer
from glue.core import Data

@importer("Import from custom source")
def my_importer():
    # Main code here
    return [Data(...), Data(...)]

The label in the @importer decorator is the text that will appear in the Import menu in Glue.

Custom Data/Subset Exporters


This section is about exporting the numerical values for datasets and subsets. To export the masks for subsets, see Custom subset mask importers and Custom subset mask exporters.

In addition to allowing you to create custom loaders and importers, glue lets you create custom exporters for datasets and subsets. These exporters can be accessed by control-clicking on specific datasets or subsets:


and selecting Export Data or Export Subsets.

A custom exporter looks like the following:

from glue.config import data_exporter

@data_exporter('My exporter')
def export_custom(filename, data):
   # write out the data here

The data argument to the function can be either a Data or a Subset object, and filename is a string which gives the file path. You can then write out the file in any way you like. Note that if you get a Subset object, you should make sure you export the data subset, not just the mask itself. For e.g. 2-dimensional datasets, we find that it is more intuitive to export arrays the same size as the original data but with the values not in the subset masked or set to NaN.

Custom subset mask importers

When right-clicking on datasets or subsets, it is possible to select to import subset masks from files (as well as export them). To define a new importer format, use the @subset_mask_importer decorator:

from glue.config import subset_mask_importer

@subset_mask_importer(label='My Format')
def my_subset_mask_importer(filename):
    # write code that reads in subset masks here

The function should return a dictionary where the labels are the names of the subsets, and the values are Numpy boolean arrays. The @subset_mask_importer decorator can also take an optional extension argument that takes a list of extensions (e.g. ['fits', 'fit']).

Custom subset mask exporters

When right-clicking on datasets or subsets, it is also possible to select to export subset masks to files. To define a new exporter format, use the @subset_mask_exporter decorator:

from glue.config import subset_mask_exporter

@subset_mask_exporter(label='My Format')
def my_subset_mask_exporter(filename, masks):
    # write code that writes out subset masks here

The masks argument will be given a dictionary where each key is the name of a subset, and each value is a Numpy boolean array. The @subset_mask_exporter decorator can also take an optional extension argument that takes a list of extensions (e.g. ['fits', 'fit']).

Custom menubar tools

In some cases, it might be desirable to add tools to Glue that can operate on any aspects of the data or subsets, and can be accessed from the menubar. To do this, you can define a function that takes two arguments (the session object, and the data collection object), and decorate it with the @menubar_plugin decorator, giving it the label that will appear in the Tools menubar:

from glue.config import menubar_plugin

@menubar_plugin("Do something")
def my_plugin(session, data_collection):
    # do anything here

The function can do anything, such as launch a QWidget, or anything else (such as a web browser, etc.), and does not need to return anything (instead it can operate by directly modifying the data collection or subsets).

Custom Colormaps

You can add additional matplotlib colormaps to Glue’s image viewer by adding the following code into

from glue.config import colormaps
from import Paired
colormaps.add('Paired', Paired)

Custom Actions

You can add menu items to run custom functions when selecting datasets, subset groups or subsets in the data collection. To do this, you should define a function to be called when the menu item is selected, and use the @layer_action decorator:

from glue.config import layer_action

@layer_action('Do something')
def callback(selected_layers, data_collection):
    print("Called with %s, %s" % (selected_layers, data_collection))

The layer_action decorator takes an optional single keyword argument that can be set to True or False to indicate whether the action should only appear when a single dataset, subset group, or subset is selected. If single is True, the following keyword arguments can be used to further control when to show the action:

  • data: only show the action when selecting a dataset
  • subset_group: only show the action when selecting a subset group
  • subset: only show the action when selecting a subset

These default to False, so setting e.g.:

@layer_action('Do something', single=True, data=True, subset=True)

means that the action will appear when a single dataset or subset is selected but not when a subset group is selected.

The callback function is called with two arguments. If single is True, the first argument is the selected layer, otherwise it is the list of selected layers. The second argument is the DataCollection object.

Custom Preference Panes

You can also add custom panes in the Qt preferences dialog. To do this, you should create a Qt widget that encapsulates the preferences you want to include, and you should make sure that this widget has a finalize method that will get called when the preferences dialog is closed. This method should then set any settings appropriately in the application state. The following is an example of a custom preference pane:

from glue.config import settings, preference_panes
from qtpy import QtWidgets

class MyPreferences(QtWidgets.QWidget):

    def __init__(self, parent=None):

        super(MyPreferences, self).__init__(parent=parent)

        self.layout = QtWidgets.QFormLayout()

        self.option1 = QtWidgets.QLineEdit()
        self.option2 = QtWidgets.QCheckBox()

        self.layout.addRow("Option 1", self.option1)
        self.layout.addRow("Option 2", self.option2)



    def finalize(self):
        settings.OPTION1 = self.option1.text()
        settings.OPTION2 = self.option2.isChecked()

settings.add('OPTION1', '')
settings.add('OPTION2', False, bool)
preference_panes.add('My preferences', MyPreferences)

This example then looks this the following once glue is loaded:


Complete list of registries

A few registries have been demonstrated above, and a complete list of main registries are listed below. All can be imported from glue.config - each registry is an instance of a class, given in the second column, and which provides more information about what the registry is and how it can be used.

Registry name Registry class
qt_client glue.config.QtClientRegistry
viewer_tool glue.config.ViewerToolRegistry
data_factory glue.config.DataFactoryRegistry
data_exporter glue.config.DataExporterRegistry
subset_mask_importer glue.config.SubsetMaskImporterRegistry
subset_mask_exporter glue.config.SubsetMaskExporterRegistry
link_function glue.config.LinkFunctionRegistry
link_helper glue.config.LinkHelperRegistry
colormaps glue.config.ColormapRegistry
exporters glue.config.ExporterRegistry
settings glue.config.SettingRegistry
preference_panes glue.config.PreferencePanesRegistry
fit_plugin glue.config.ProfileFitterRegistry
layer_action glue.config.LayerActionRegistry

Deferring loading of plug-in functionality (advanced)

In some cases, you may want to defer the loading of your component/functionality until it is actually needed. To do this:

  • Place the code for your plugin in a file or package that could be imported from the (but don’t import it directly - it just has to be importable)
  • Include a function called setup alongside the plugin, and this function should contain code to actually add your custom tools to the appropriate registries.
  • In, you can then add the plugin file or package to a registry by using the lazy_add method and pass a string giving the name of the package or sub-package containing the plugin.

Imagine that you have created a data viewer MyQtViewer. You could directly register it using:

from glue.config import qt_client

but if you want to defer the loading of the MyQtViewer class, you can place the definition of MyQtViewer in a file called e.g. that is located in the same directory as your file. This file should look something like:

class MyQtViewer(...):

def setup():
    from glue.config import qt_client

then in, you can do:

from glue.config import qt_client

With this in place, the setup in your plugin will only get called if the Qt data viewers are needed, but you will avoid unecessarily importing Qt if you only want to access glue.core.