Configuring Plot Tools

Bokeh comes with a number of interactive tools. There are three categories of tool interactions:

For each type of gesture, one tool can be active at any given time, and the active tool is indicated on the toolbar by a highlight next to to the tool icon. Actions are immediate or modal operations that are only activated when their button in the toolbar is pressed. Inspectors are passive tools that report information or annotate the plot in some way.

Positioning the Toolbar

By default, Bokeh plots come with a toolbar above the plot. In this section you will learn how to specify a different location for the toolbar, or to remove it entirely.

The toolbar location can be specified by passing the toolbar_location parameter to the figure() function. Valid values are:

  • "above"
  • "below"
  • "left"
  • "right"

If you would like to hide the toolbar entirely, pass None.

Below is some code that positions the toolbar below the plot. Try running the code and changing the toolbar_location value.

from bokeh.plotting import figure, output_file, show

output_file("toolbar.html")

# create a new plot with the toolbar below
p = figure(plot_width=400, plot_height=400,
           title=None, toolbar_location="below")

p.circle([1, 2, 3, 4, 5], [2, 5, 8, 2, 7], size=10)

show(p)

Note that the toolbar position clashes with the default axes, in this case setting the toolbar_sticky option to False will move the toolbar to outside of the region where the axis is drawn.

from bokeh.plotting import figure, output_file, show

output_file("toolbar.html")

# create a new plot with the toolbar below
p = figure(plot_width=400, plot_height=400,
           title=None, toolbar_location="below",
           toolbar_sticky=False)

p.circle([1, 2, 3, 4, 5], [2, 5, 8, 2, 7], size=10)

show(p)

Specifying Tools

At the lowest bokeh.models level, tools are added to a Plot by passing instances of Tool objects to the add_tools method:

plot = Plot()
plot.add_tools(LassoSelectTool())
plot.add_tools(WheelZoomTool())

This explicit way of adding tools works with any Bokeh Plot or Plot subclass, such as Figure.

Tools can be specified by passing the tools parameter to the figure() function. The tools parameter accepts a list of tool objects, for instance:

tools = [BoxZoomTool(), ResetTool()]

Tools can also be supplied conveniently with a comma-separate string containing tool shortcut names:

tools = "pan,wheel_zoom,box_zoom,reset"

However, this method does not allow setting properties of the tools.

Finally, it is also always possible to add new tools to a plot by passing a tool object to the add_tools method of a plot. This can also be done in conjunction with the tools keyword described above:

plot = figure(tools="pan,wheel_zoom,box_zoom,reset")
plot.add_tools(BoxSelectTool(dimensions=["width"]))

Setting the Active Tools

Bokeh toolbars can have (at most) one active tool from each kind of gesture (drag, scroll, tap). By default, Bokeh will use a default pre-defined order of preference to choose one of each kind from the set of configured tools, to be active.

However it is possible to exert control over which tool is active. At the lowest bokeh.models level, this is accomplished by using the active_drag, active_inspect, active_scroll, and active_tap properties of Toolbar. These properties can take the following values:

  • None — there is no active tool of this kind
  • "auto" — Bokeh chooses a tool of this kind to be active (possibly none)
  • a Tool instance — Bokeh sets the given tool to be the active tool

Additionally, the active_inspect tool may accept: * A sequence of Tool instances to be set as the active tools

As an example:

# configure so that no drag tools are active
plot.toolbar.active_drag = None

# configure so that Bokeh chooses what (if any) scroll tool is active
plot.toolbar.active_scroll = "auto"

# configure so that a specific PolySelect tap tool is active
plot.toolbar.active_tap = poly_select

# configure so that a sequence of specific inspect tools are active
# note: this only works for inspect tools
plot.toolbar.active_inspect = [hover_tool, crosshair_tool]

The default value for all of these properties is "auto".

Active tools can be specified by passing the these properties as keyword arguments to the figure() function. It is also possible to pass any one of the string names for, ease of configuration:

# configures the lasso tool to be active
plot = figure(tools="pan,lasso_select,box_select", active_drag="lasso_select")

Pan/Drag Tools

These tools are employed by panning (on touch devices) or left-dragging (on mouse devices). Only one pan/drag tool may be active at a time. Where applicable, Pan/Drag tools will respect any max and min values set on ranges.

BoxSelectTool

  • name: 'box_select'
  • icon: box_select_icon

The box selection tool allows the user to define a rectangular selection region by left-dragging a mouse, or dragging a finger across the plot area. The box select tool may be configured to select across only one dimension by setting the dimensions property to a list containing width or height.

Note

To make a multiple selection, press the SHIFT key. To clear the selection, press the ESC key.

BoxZoomTool

  • name: 'box_zoom'
  • icon: box_zoom_icon

The box zoom tool allows the user to define a rectangular region to zoom the plot bounds too, by left-dragging a mouse, or dragging a finger across the plot area.

LassoSelectTool

  • name: 'lasso_select'
  • icon: lasso_select_icon

The lasso selection tool allows the user to define an arbitrary region for selection by left-dragging a mouse, or dragging a finger across the plot area.

Note

To make a multiple selection, press the SHIFT key. To clear the selection, press the ESC key.

PanTool

  • name: 'pan', 'xpan', 'ypan',
  • icon: pan_icon

The pan tool allows the user to pan the plot by left-dragging a mouse or dragging a finger across the plot region.

It is also possible to constrain the pan tool to only act on either just the x-axis or just the y-axis by setting the dimensions property to a list containing width or height. Additionally, there are tool aliases 'xpan' and 'ypan', respectively.

Click/Tap Tools

These tools are employed by tapping (on touch devices) or left-clicking (on mouse devices). Only one click/tap tool may be active at a time.

PolySelectTool

  • name: 'poly_select'
  • icon: poly_select_icon

The polygon selection tool allows the user to define an arbitrary polygonal region for selection by left-clicking a mouse, or tapping a finger at different locations.

Note

Complete the selection by making a double left-click or tapping. To make a multiple selection, press the SHIFT key. To clear the selection, press the ESC key.

TapTool

  • name: 'tap'
  • icon: tap_icon

The tap selection tool allows the user to select at single points by clicking a left mouse button, or tapping with a finger.

Note

To make a multiple selection, press the SHIFT key. To clear the selection, press the ESC key.

Scroll/Pinch Tools

These tools are employed by pinching (on touch devices) or scrolling (on mouse devices). Only one scroll/pinch tool may be active at a time.

WheelZoomTool

  • name: 'wheel_zoom', 'xwheel_zoom', 'ywheel_zoom'
  • icon: wheel_zoom_icon

The wheel zoom tool will zoom the plot in and out, centered on the current mouse location. It will respect any min and max values and ranges preventing zooming in and out beyond these.

It is also possible to constraint the wheel zoom tool to only act on either just the x-axis or just the y-axis by setting the dimensions property to a list containing width or height. Additionally, there are tool aliases 'xwheel_zoom' and 'ywheel_zoom', respectively.

WheelPanTool

  • name: 'xwheel_pan', 'ywheel_pan'
  • icon: wheel_pan_icon

The wheel pan tool will translate the plot window along the specified dimension without changing the window’s aspect ratio. The tool will respect any min and max values and ranges preventing panning beyond these values.

Actions

Actions are operations that are activated only when their button in the toolbar is tapped or clicked. They are typically modal or immediate-acting.

UndoTool

  • name: 'undo'
  • icon: undo_icon

The undo tool allows to restore previous state of the plot.

RedoTool

  • name: 'redo'
  • icon: redo_icon

The redo tool reverses the last action performed by undo tool.

ResetTool

  • name: 'reset'
  • icon: reset_icon

The reset tool will restore the plot ranges to their original values.

SaveTool

  • name: 'save'
  • icon: save_icon

The save tool pops up a modal dialog that allows the user to save a PNG image of the plot.

ZoomInTool

  • name: 'zoom_in', 'xzoom_in', 'yzoom_in'
  • icon: zoom_in_icon

The zoom-in tool will increase the zoom of the plot. It will respect any min and max values and ranges preventing zooming in and out beyond these.

It is also possible to constraint the wheel zoom tool to only act on either just the x-axis or just the y-axis by setting the dimensions property to a list containing width or height. Additionally, there are tool aliases 'xzoom_in' and 'yzoom_in', respectively.

ZoomOutTool

  • name: 'zoom_out', 'xzoom_out', 'yzoom_out'
  • icon: zoom_out_icon

The zoom-out tool will decrease the zoom level of the plot. It will respect any min and max values and ranges preventing zooming in and out beyond these.

It is also possible to constraint the wheel zoom tool to only act on either just the x-axis or just the y-axis by setting the dimensions property to a list containing width or height. Additionally, there are tool aliases 'xzoom_in' and 'yzoom_in', respectively.

Inspectors

Inspectors are passive tools that annotate or otherwise report information about the plot, based on the current cursor position. Any number of inspectors may be active at any given time. The inspectors menu in the toolbar allows users to toggle the active state of any inspector.

CrosshairTool

  • name: 'crosshair'
  • menu icon: crosshair_icon

Th crosshair tool draws a crosshair annotation over the plot, centered on the current mouse position. The crosshair tool draw dimensions may be configured by setting the dimensions property to one of the enumerated values width, height, or both.

HoverTool

  • name: 'hover'
  • menu icon: hover_icon

The hover tool is a passive inspector tool. It is generally on at all times, but can be configured in the inspector’s menu associated with the toolbar.

Basic Tooltips

By default, the hover tool will generate a “tabular” tooltip where each row contains a label, and its associated value. The labels and values are supplied as a list of (label, value) tuples. For instance, the tooltip below on the left was created with the accompanying tooltips definition on the right.

hover_basic
hover.tooltips = [
    ("index", "$index"),
    ("(x,y)", "($x, $y)"),
    ("radius", "@radius"),
    ("fill color", "$color[hex, swatch]:fill_color"),
    ("foo", "@foo"),
    ("bar", "@bar"),
]

Field names that begin with $ are “special fields”. These often correspond to values that are intrinsic to the plot, such as the coordinates of the mouse in data or screen space. These special fields are listed here:

$index:index of selected point in the data source
$x:x-coordinate under the cursor in data space
$y:y-coordinate under the cursor in data space
$sx:x-coordinate under the cursor in screen (canvas) space
$sy:y-coordinate under the cursor in screen (canvas) space
$color:colors from a data source, with the syntax: $color[options]:field_name. The available options are: hex (to display the color as a hex value), and swatch to also display a small color swatch.

Field names that begin with @ are associated with columns in a ColumnDataSource. For instance the field name "@price" will display values from the "price" column whenever a hover is triggered. If the hover is for the 17th glyph, then the hover tooltip will correspondingly display the 17th price value.

Note that if a column name contains spaces, the it must be supplied by surrounding it in curly braces, e.g. @{adjusted close} will display values from a column named "adjusted close".

Here is a complete example of how to configure and use the hover tool with a default tooltip:

from bokeh.plotting import figure, output_file, show, ColumnDataSource
from bokeh.models import HoverTool

output_file("toolbar.html")

source = ColumnDataSource(data=dict(
    x=[1, 2, 3, 4, 5],
    y=[2, 5, 8, 2, 7],
    desc=['A', 'b', 'C', 'd', 'E'],
))

hover = HoverTool(tooltips=[
    ("index", "$index"),
    ("(x,y)", "($x, $y)"),
    ("desc", "@desc"),
])

p = figure(plot_width=400, plot_height=400, tools=[hover],
           title="Mouse over the dots")

p.circle('x', 'y', size=20, source=source)

show(p)

Hit-testing Behavior

The hover tool displays informational tooltips associated with individual glyphs. These tooltips can be configured to activate in in different ways with a mode property:

"mouse":only when the mouse is directly over a glyph
"vline":whenever the a vertical line from the mouse position intersects a glyph
"hline":whenever the a horizontal line from the mouse position intersects a glyph

The default configuration is mode = "mouse". This can be observed in the Basic Tooltips example above. The example below in Formatting Tooltip Fields demonstrates an example that sets mode = "vline".

Formatting Tooltip Fields

By default, values for fields (e.g. @foo) are displayed in a basic numeric format. However it is possible to control the formatting of values more precisely. Fields can be modified by appending a format specified to the end in curly braces. Some examples are below.

"@foo{0,0.000}"    # formats 10000.1234 as: 10,000.123

"@foo{(.00)}"      # formats -10000.1234 as: (10000.123)

"@foo{($ 0.00 a)}" # formats 1230974 as: $ 1.23 m

The examples above all use the default formatting scheme. But there are other formatting schemes that can be specified for interpreting format strings:

"numeral":Provides a wide variety of formats for numbers, currency, bytes, times, and percentages. The full set of formats can be found in the NumeralTickFormatter reference documentation.
"datetime":Provides formats for date and time values. The full set of formats is listed in the DatetimeTickFormatter reference documentation.
"printf":Provides formats similar to C-style “printf” type specifiers. See the PrintfTickFormatter reference documentation for complete details.

These are supplied by configuring the formatters property of a hover tool. This property maps column names to format schemes. For example, to use the "datetime" scheme for formatting a column "close date", set the value:

hover_tool.formatters = { "close date": "datetime"}

If no formatter is specified for a column name, the default "numeral" formatter is assumed.

Note that format specifications are also compatible with column names that have spaces. For example `@{adjusted close}{($ 0.00 a)} applies a format to a column named “adjusted close”.

The example code below shows configuring a HoverTool with different formatters for different fields:

HoverTool(
    tooltips=[
        ( 'date',   '@date{%F}'            ),
        ( 'close',  '$@{adj close}{%0.2f}' ), # use @{ } for field names with spaces
        ( 'volume', '@volume{0.00 a}'      ),
    ],

    formatters={
        'date'      : 'datetime', # use 'datetime' formatter for 'date' field
        'adj close' : 'printf',   # use 'printf' formatter for 'adj close' field
                                  # use default 'numeral' formatter for other fields
    },

    # display a tooltip whenever the cursor is vertically in line with a glyph
    mode='vline'
)

You can see the output generated by this configuration by hovering the mouse over the plot below:

Custom Tooltip

It is also possible to supply a custom HTML template for a tooltip. To do this, pass an HTML string, with the Bokeh tooltip field name symbols wherever substitutions are desired. All of the information above regarding formats, etc. still applies. Note that you can also use the {safe} format after the column name to disable the escaping of HTML in the data source. An example is shown below:

from bokeh.plotting import figure, output_file, show, ColumnDataSource
from bokeh.models import HoverTool

output_file("toolbar.html")

source = ColumnDataSource(data=dict(
    x=[1, 2, 3, 4, 5],
    y=[2, 5, 8, 2, 7],
    desc=['A', 'b', 'C', 'd', 'E'],
    imgs=[
        'https://bokeh.pydata.org/static/snake.jpg',
        'https://bokeh.pydata.org/static/snake2.png',
        'https://bokeh.pydata.org/static/snake3D.png',
        'https://bokeh.pydata.org/static/snake4_TheRevenge.png',
        'https://bokeh.pydata.org/static/snakebite.jpg'
    ],
    fonts=[
        '<i>italics</i>',
        '<pre>pre</pre>',
        '<b>bold</b>',
        '<small>small</small>',
        '<del>del</del>'
    ]
))

hover = HoverTool( tooltips="""
    <div>
        <div>
            <img
                src="@imgs" height="42" alt="@imgs" width="42"
                style="float: left; margin: 0px 15px 15px 0px;"
                border="2"
            ></img>
        </div>
        <div>
            <span style="font-size: 17px; font-weight: bold;">@desc</span>
            <span style="font-size: 15px; color: #966;">[$index]</span>
        </div>
        <div>
            <span>@fonts{safe}</span>
        </div>
        <div>
            <span style="font-size: 15px;">Location</span>
            <span style="font-size: 10px; color: #696;">($x, $y)</span>
        </div>
    </div>
    """
)

p = figure(plot_width=400, plot_height=400, tools=[hover],
           title="Mouse over the dots")

p.circle('x', 'y', size=20, source=source)

show(p)

Controlling Level of Detail

Although the HTML canvas can comfortably display tens or even hundreds of thousands of glyphs, doing so can have adverse affects on interactive performance. In order to accommodate large-ish (but not enormous) data sizes, Bokeh plots offer “Level of Detail” (LOD) capability in the client.

Note

Another option, when dealing with very large data volumes, is to use the Bokeh Server to perform downsampling on data before it is sent to the browser. Such an approach is unavoidable past a certain data size. See Running a Bokeh Server for more information.

The basic idea is that during interactive operations (e.g., panning or zooming), the plot only draws some small fraction of data points. This hopefully allows the general sense of the interaction to be preserved mid-flight, while maintaining interactive performance. There are four properties on Plot objects that control LOD behavior:

lod_factor

property type: Int

Decimation factor to use when applying level-of-detail decimation.

lod_interval

property type: Int

Interval (in ms) during which an interactive tool event will enable level-of-detail downsampling.

lod_threshold

property type: Int

A number of data points, above which level-of-detail downsampling may be performed by glyph renderers. Set to None to disable any level-of-detail downsampling.

lod_timeout

property type: Int

Timeout (in ms) for checking whether interactive tool events are still occurring. Once level-of-detail mode is enabled, a check is made every lod_timeout ms. If no interactive tool events have happened, level-of-detail mode is disabled.