fig.add_vline on date axis fails with date, timestamp and string as x value

[stoney95]

I build a line chart with dates on the x-axis. I want to add a vertical line at a certain x-position. Using a timestamp, date or date-formated string as value for x fails.

Example

import plotly.graph_objects as go

fig = go.Figure()

x = ["2024-08-01", "2024-09-01, "2024-0-01"]
y = [1,3,2]

fig.add_trace(go.Scatter(x=x, y=y))

fig.add_vline(x="2024-05-01", annotation_text="Example")

Error Message

File "...\src\plots.py", line 45, in plot_score
  fig.add_vline(
File "...\.venv\Lib\site-packages\plotly\graph_objs\_figure.py", line 1049, in add_vline
  return super(Figure, self).add_vline(
          ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "...\.venv\Lib\site-packages\plotly\basedatatypes.py", line 4095, in add_vline
  self._process_multiple_axis_spanning_shapes(
File "...\.venv\Lib\site-packages\plotly\basedatatypes.py", line 4036, in _process_multiple_axis_spanning_shapes
  augmented_annotation = shapeannotation.axis_spanning_shape_annotation(
                          ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "...\.venv\Lib\site-packages\plotly\shapeannotation.py", line 216, in axis_spanning_shape_annotation
  shape_dict = annotation_params_for_line(
                ^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "...\.venv\Lib\site-packages\plotly\shapeannotation.py", line 63, in annotation_params_for_line
  eX = _mean(X)
        ^^^^^^^^
File "...\.venv\Lib\site-packages\plotly\shapeannotation.py", line 7, in _mean
  return float(sum(x)) / len(x)
                 ^^^^^^
TypeError: unsupported operand type(s) for +: 'int' and 'str'

Expected behavior

The line is added at the provided date.

Additional context

Using fig.add_vrect with dates works

[gvwilson]

@stoney95 please let us know which version of plotly you are using - thanks.

[stoney95]

@gvwilson I am using plotly==5.24.1 and python==3.11

[FBruzzesi]

I will be stating the obvious here: in annotation_param_per_line the computation of mean X and mean Y could be postponed for only those cases that actually requires them. In this way at least all other operation could run, leaving out the following combinations:

• vline with annotation_position="left" or annotation_position="right"
• hline with annotation_position="top" or annotation_position="bottom"

For these cases a check on the X and/or Y type to be numeric could be performed (maybe even in _prepare_position) and a more informative error message provided. @gvwilson what do you think?

[gvwilson]

I will defer to @emilykl on this one - sadly (tragically) I'm not familiar enough with the code for any of this to be obvious to me any longer :-(

[austinspagnolo]

This bug was reported a few years ago; there's a workaround on the issue: #3065
Unfortunately, this bug is still very real. :(

I experience the same error with add_vrect or add_vline for datetime.datetime objects for the x axis. Tested today on python 3.10.6 using plotly 5.22.0 and python 3.9.15 with plotly 5.24.1.

[FBruzzesi]

@emilykl if no one is taking a look at this I am happy to give it a try.

I wonder what's the expected behavior for date, datetime, timestamp and string types for the cases mentioned above, namely:

• vline with annotation_position="left" or annotation_position="right"
• hline with annotation_position="top" or annotation_position="bottom"