IFrame renderer throws exception on plots in the first cell

[gokikamukow]

Symptom: when running the following conversion on a notebook that produces a plot in the first cell:

PLOTLY_RENDERER=iframe_connected jupyter nbconvert --execute --to html --no-input my_notebook.ipynb

the following error is received:

File ~/.virtualenvs/myenv/lib/python3.9/site-packages/plotly/io/_base_renderers.py:612, in IFrameRenderer.build_filename(self)
    610 def build_filename(self):
    611     ip = IPython.get_ipython() if IPython else None
--> 612     cell_number = list(ip.history_manager.get_tail(1))[0][1] + 1 if ip else 0
    613     filename = "{dirname}/figure_{cell_number}.html".format(
    614         dirname=self.html_directory, cell_number=cell_number
    615     )
    616     return filename

IndexError: list index out of range

The iframe renderer saves plots to a separate file and the filename is supposed to be based off the current cell number. The line of code that retrieves this cell number makes use of the IPython HistoryManager.get_tail method which, by default, returns the already executed cells, not the currently executing cell. When within the first cell it therefore returns an empty list and this is what causes the 'list index our of range'. The fix is very simple: add include_latest=True to the get_tail call so that the currently executing cell is returned instead of the previous one.

Meanwhile a simple workaround is to always include a cell that does not produce a plot at the start of the notebook.

[gokikamukow]

To be explicit, the fix that worked for me locally was to change this line:

cell_number = list(ip.history_manager.get_tail(1))[0][1] + 1 if ip else 0

to:

cell_number = list(ip.history_manager.get_tail(1, include_latest=True))[0][1] + 1 if ip else 0

[jeertmans]

Hi @gokikamukow, many thanks for finding the actual reason behind this bug!
It would be nice if this fix could be integrated into Plotly.py :)