Working with AutoShapes

Auto shapes are regular shape shapes. Squares, circles, triangles, stars, that sort of thing. There are 182 different auto shapes to choose from. 120 of these have adjustment “handles” you can use to change the shape, sometimes dramatically.

Many shape types share a common set of properties. We’ll introduce many of them here because several of those shapes are just a specialized form of AutoShape.

Adding an auto shape

The following code adds a rounded rectangle shape, one inch square, and positioned one inch from the top-left corner of the slide:

from pptx.enum.shapes import MSO_SHAPE

shapes = slide.shapes
left = top = width = height = Inches(1.0)
shape = shapes.add_shape(
    MSO_SHAPE.ROUNDED_RECTANGLE, left, top, width, height

See the MSO_AUTO_SHAPE_TYPE enumeration page for a list of all 182 auto shape types.

Understanding English Metric Units

In the prior example we set the position and dimension values to the expression Inches(1.0). What’s that about?

Internally, PowerPoint stores length values in English Metric Units (EMU). This term might be worth a quick Googling, but the short story is EMU is an integer unit of length, 914400 to the inch. Most lengths in Office documents are stored in EMU. 914400 has the great virtue that it is evenly divisible by a great many common factors, allowing exact conversion between inches and centimeters, for example. Being an integer, it can be represented exactly across serializations and across platforms.

As you might imagine, working directly in EMU is inconvenient. To make it easier, python-pptx provides a collection of value types to allow easy specification and conversion into convenient units:

>>> from pptx.util import Inches, Pt
>>> length = Inches(1)
>>> length
>>> length.inches
>>> length = Pt(72)
>>> length

More details are available in the API documentation for pptx.util

Shape position and dimensions

All shapes have a position on their slide and have a size. In general, position and size are specified when the shape is created. Position and size can also be read from existing shapes and changed:

>>> from pptx.enum.shapes import MSO_SHAPE
>>> left = top = width = height = Inches(1.0)
>>> shape = shapes.add_shape(
>>>     MSO_SHAPE.ROUNDED_RECTANGLE, left, top, width, height
>>> )
>>> shape.left,, shape.width, shape.height
(914400, 914400, 914400, 914400)
>>> shape.left.inches
>>> shape.left = Inches(2.0)
>>> shape.left.inches


AutoShapes have an outline around their outside edge. What appears within that outline is called the shape’s fill.

The most common type of fill is a solid color. A shape may also be filled with a gradient, a picture, a pattern (like cross-hatching for example), or may have no fill (transparent).

When a color is used, it may be specified as a specific RGB value or a color from the theme palette.

Because there are so many options, the API for fill is a bit complex. This code sets the fill of a shape to red:

>>> fill = shape.fill
>>> fill.solid()
>>> fill.fore_color.rgb = RGBColor(255, 0, 0)

This sets it to the theme color that appears as ‘Accent 1 - 25% Darker’ in the toolbar palette:

>>> from pptx.enum.dml import MSO_THEME_COLOR
>>> fill = shape.fill
>>> fill.solid()
>>> fill.fore_color.theme_color = MSO_THEME_COLOR.ACCENT_1
>>> fill.fore_color.brightness = -0.25

This sets the shape fill to transparent, or ‘No Fill’ as it’s called in the PowerPoint UI:

>>> shape.fill.background()

As you can see, the first step is to specify the desired fill type by calling the corresponding method on fill. Doing so actually changes the properties available on the fill object. For example, referencing .fore_color on a fill object after calling its .background() method will raise an exception:

>>> fill = shape.fill
>>> fill.solid()
>>> fill.fore_color
<pptx.dml.color.ColorFormat object at 0x10ce20910>
>>> fill.background()
>>> fill.fore_color
Traceback (most recent call last):
TypeError: a transparent (background) fill has no foreground color


The outline of an AutoShape can also be formatted, including setting its color, width, dash (solid, dashed, dotted, etc.), line style (single, double, thick-thin, etc.), end cap, join type, and others. At the time of writing, color and width can be set using python-pptx:

>>> line = shape.line
>>> line.color.rgb = RGBColor(255, 0, 0)
>>> line.color.brightness = 0.5  # 50% lighter
>>> line.width = Pt(2.5)

Theme colors can be used on lines too:

>>> line.color.theme_color = MSO_THEME_COLOR.ACCENT_6

Shape.line has the attribute .color. This is essentially a shortcut for:

>>> line.fill.solid()
>>> line.fill.fore_color

This makes sense for line formatting because a shape outline is most frequently set to a solid color. Accessing the fill directly is required, for example, to set the line to transparent:

>>> line.fill.background()

Line width

The shape outline also has a read/write width property:

>>> line.width
>>> line.width = Pt(2.0)

Adjusting an autoshape

Many auto shapes have adjustments. In PowerPoint, these show up as little yellow diamonds you can drag to change the look of the shape. They’re a little fiddly to work with via a program, but if you have the patience to get them right, you can achieve some remarkable effects with great precision.

Shape Adjustment Concepts

There are a few concepts it’s worthwhile to grasp before trying to do serious work with adjustments.

First, adjustments are particular to a specific auto shape type. Each auto shape has between zero and eight adjustments. What each of them does is arbitrary and depends on the shape design.

Conceptually, adjustments are guides, in many ways like the light blue ones you can align to in the PowerPoint UI and other drawing apps. These don’t show, but they operate in a similar way, each defining an x or y value that part of the shape will align to, changing the proportions of the shape.

Adjustment values are large integers, each based on a nominal value of 100,000. The effective value of an adjustment is proportional to the width or height of the shape. So a value of 50,000 for an x-coordinate adjustment corresponds to half the width of the shape; a value of 75,000 for a y-coordinate adjustment corresponds to 3/4 of the shape height.

Adjustment values can be negative, generally indicating the coordinate is to the left or above the top left corner (origin) of the shape. Values can also be subject to limits, meaning their effective value cannot be outside a prescribed range. In practice this corresponds to a point not being able to extend beyond the left side of the shape, for example.

Spending some time fooling around with shape adjustments in PowerPoint is time well spent to build an intuitive sense of how they behave. You also might want to have opc-diag installed so you can look at the XML values that are generated by different adjustments as a head start on developing your adjustment code.

The following code formats a callout shape using its adjustments:

callout_sp = shapes.add_shape(
    MSO_SHAPE.LINE_CALLOUT_2_ACCENT_BAR, left, top, width, height

# get the callout line coming out of the right place
adjs = callout_sp.adjustments
adjs[0] = 0.5   # vert pos of junction in margin line, 0 is top
adjs[1] = 0.0   # horz pos of margin ln wrt shape width, 0 is left side
adjs[2] = 0.5   # vert pos of elbow wrt margin line, 0 is top
adjs[3] = -0.1  # horz pos of elbow wrt shape width, 0 is margin line
adjs[4] = 3.0   # vert pos of line end wrt shape height, 0 is top
a5 = adjs[3] - (adjs[4] - adjs[0]) * height/width
adjs[5] = a5    # horz pos of elbow wrt shape width, 0 is margin line

# rotate 45 degrees counter-clockwise
callout_sp.rotation = -45.0