Customization

QR code & watermarks

You can customize the QR code (or disable it altogether) by specifying a custom watermark in the call to gofigr.jupyter.configure(). See gofigr.watermarks.DefaultWatermark for other parameters.

%reload_ext gofigr

from gofigr.jupyter import *
configure(analysis=FindByName("Documentation examples", create=True),
          auto_publish=True,
          watermark=DefaultWatermark(show_qr_code=False))

Annotators

GoFigr implements Annotators, an extensible framework for automatically including relevant data when publishing revisions. You can use this mechanism to automatically attach the output of pip freeze, for example.

Out of the box, GoFigr includes the following annotators:

gofigr.annotators.NotebookMetadataAnnotator: name & path of the current notebook
gofigr.annotators.CellCodeAnnotator: code of the Jupyter cell
gofigr.annotators.CellIdAnnotator: Jupyter Cell ID (only available in Jupyter Lab)
gofigr.annotators.PipFreezeAnnotator: output of pip freeze
gofigr.annotators.SystemAnnotator: output of uname -a
gofigr.annotators.EnvironmentAnnotator: Python version and kernel path
gofigr.annotators.BackendAnnotator: Figure backend (e.g. matplotlib, plotly)
gofigr.annotators.HistoryAnnotator: Jupyter execution history

Notebook name & path

GoFigr uses ipynbname to infer the name & path of the currently running notebook. While this works well most of the time, it may not work in certain configurations. For example, it doesn’t work if the notebook is executed programmatically with nbconvert. Other exceptions may exist, as well.

GoFigr will show a warning if the notebook name or path cannot be obtained. To fix, you can specify them in the call to configure:

%reload_ext gofigr

from gofigr.jupyter import *
configure(analysis=FindByName("Documentation examples", create=True),
          auto_publish=True,
          notebook_name="my notebook",
          notebook_path="/path/to/notebook.ipynb")

Implementing custom annotators

To implement a custom annotator, simply subclass gofigr.jupyter.Annotator. For example, here’s how pip freeze is implemented:

class PipFreezeAnnotator(Annotator):
    """Annotates revisions with the output of pip freeze"""
    def annotate(self, revision):
        try:
            output = subprocess.check_output(["pip", "freeze"]).decode('ascii')
        except subprocess.CalledProcessError as e:
            output = e.output

        revision.data.append(_GF_EXTENSION.gf.TextData(name="pip freeze", contents=output))
        return revision

You can annotate revisions with:

Please note that you need to instantiate those from the GoFigr object, e.g. gf.CodeData, gf.TableData, etc.

Specifying annotators

You can override the default annotators in the call to gofigr.jupyter.configure():

%reload_ext gofigr

from gofigr.jupyter import *
from gofigr.watermarks import DefaultWatermark

configure(..., annotators=DEFAULT_ANNOTATORS)

Widgets

When used with Jupyter, GoFigr will display a widget under each published figure. The widget is customizable – you can override it by passing widget_class to gofigr.jupyter.configure().

For a full list of supported widget classes, see gofigr.widget.

Detailed (default)

%load_ext gofigr
from gofigr.jupyter import *
from gofigr.widget import *

configure(..., widget_class=DetailedWidget)

Compact

%load_ext gofigr
from gofigr.jupyter import *
from gofigr.widget import *

configure(..., widget_class=CompactWidget)

Minimal

%load_ext gofigr
from gofigr.jupyter import *
from gofigr.widget import *

configure(..., widget_class=MinimalWidget)