Leading/trailing spaces ignored in annotations, only in SVG output

[andom]

I position row total annotations at the end of horizontal bar plots, using the row total as x position value and left-aligning the string. I’d like to keep a gap, so the numbers stand a bit apart from the bars for legibility. This is possible by prepending 1-2 spaces to the annotation. When using fig.show() or pio.show(fig, "svg"), it displays as expected.

However, when using pio.write_image(fig, filename), leading spaces in the annotation string are missing.

And to add, I surround the annotation string with <b>...</b> and spaces both before and after the initial <b> are ignored.

Any tricks to circumvent this, or a place to report this? It seems like undesired behavior.

Cheers

[AIMPED]

Hey @andom, does the ticksuffix help?

Workaround of "tickvall_standoff" fails!? | styling a figure | corporate design

Many of us have to write reports or make presentations with visualizations. The challenge sometimes isn’t the underlying data or the visualization itself, but the corporate design. In my case, the corporate design of the organization is based on an excel template for diagrams, with all its disadvantages. My goal is to automate most of it with plotly and convince my academic organization to move to plotly to a certain extent. I simply love plotly and its capabilities! Therefore I would like to s…

[andom]

Unfortunately not, as I am using fig.add_annotation(), not tick labelling for these totals. Let me show you…

fig.show() & pio.show(fig, "svg"):

pio.write_image(fig, "filename.svg"):

Note also the slight spacing differences in the within-bar annotations (which aren’t my problem here though).

[AIMPED]

@andom

using the row total as x position value and left-aligning the string.

Why don’t you use x-position + your distance (float) as position?

[andom]

I would prefer a solution that works without the x-position, because it is currently determined by the x-position of the bar end, and bar lengths in x-coordinates can vary by orders of magnitude. So to attain a somewhat constant clearance, I’d have to work with the x range and compute a fraction of it. --Possible but cumbersome. I wanted to either find a way without this step or at least find out in what forum or site to flag this behavior.

[AIMPED]

I think I do not understand the issue. I was referring to something like this.

import plotly.graph_objects as go
import numpy as np

YOUR_DISTANCE = 0.5

# dummy data
y_saving = [1.3586, 2.2623000000000002, 4.9821999999999997, 6.5096999999999996, 7.4812000000000003, 7.5133000000000001, 15.2148, 17.520499999999998]
y_net_worth = [93453.919999999998, 81666.570000000007, 69889.619999999995, 78381.529999999999, 141395.29999999999, 92969.020000000004, 66090.179999999993, 122379.3]
x = ['Japan', 'United Kingdom', 'Canada', 'Netherlands','United States', 'Belgium', 'Sweden', 'Switzerland']

# Creating two subplots
fig = go.Figure(
    go.Bar(
        x=y_saving,
        y=x,
        marker=dict(
            color='rgba(50, 171, 96, 0.6)',
            line=dict(
                color='rgba(50, 171, 96, 1.0)',
                width=1),
        ),
        name='Household savings, percentage of household disposable income',
        orientation='h',
    )
)


annotations = []

y_s = np.round(y_saving, decimals=2)
y_nw = np.rint(y_net_worth)

# Adding labels
for ydn, yd, xd in zip(y_nw, y_s, x):
    # labeling the bar net worth
    annotations.append(dict(
        xref='x1', 
        yref='y1',
        y=xd, 
        x=yd + YOUR_DISTANCE,
        text=str(yd) + '%',
        showarrow=False))

fig.update_layout(annotations=annotations)

fig.show()

[andom]

Thanks for this reply! My issue rests in the fact that I don’t know beforehand the order of magnitude of the numbers that will be displayed. It can be 0…100 percent, or multiples of a billion. Therefore, expressing the clearance within the x coordinate system (YOUR_DISTANCE) is difficult and requires, I think, dealing with the overall x range and expressing the clearance as smth like “.0001 * <max. x>”. I get the impression that I will have to go this route.

*I would have liked the number positioning problem to be separate from the question of the data value range. By using the data-driven x range for a typographical matter seems to mix data and presentation, which feels more like drawing than plotting. [/opinion]