Matplotlib table: use bold weight when calculating header width

[padgy]

When using Matplotlib as converter I was getting overlapping text for long column headers. This appears to be because the headers are bold but we do not set the weight as bold when calculating headers.

These changes fixed the issue for me.

[PaleNeutron]

Can you add long bold header to case test_styled in tests/test_df_image.py to make sure this feature works fine afterwards?

[padgy]

Can you add long bold header to case test_styled in tests/test_df_image.py to make sure this feature works fine afterwards?

@PaleNeutron I don't think adding a bold header style is necessary to test this. The matplotlib converter produces bold headers already regardless of the style used:

dataframe_image/dataframe_image/_matplotlib_table.py

Line 96 in eff1f4d

bold = el.name == "th"

I can add a long column header to the covid19.csv test data though if that is what you mean.

[PaleNeutron]

Can you add long bold header to case test_styled in tests/test_df_image.py to make sure this feature works fine afterwards?

@PaleNeutron I don't think adding a bold header style is necessary to test this. The matplotlib converter produces bold headers already regardless of the style used:

dataframe_image/dataframe_image/_matplotlib_table.py

Line 96 in eff1f4d

bold = el.name == "th"

I can add a long column header to the covid19.csv test data though if that is what you mean.

I mean some corner cases, for example, dataframe's index is th and what happens to it if it is very long?

And if user make one cell text bold by styler and the text is quite long, will it cause overlap?

But if you do not want to implement this is OK too. I can merge this after you add long header test case and a TODO at the place.

[padgy]

Can you add long bold header to case test_styled in tests/test_df_image.py to make sure this feature works fine afterwards?

@PaleNeutron I don't think adding a bold header style is necessary to test this. The matplotlib converter produces bold headers already regardless of the style used:

dataframe_image/dataframe_image/_matplotlib_table.py

Line 96 in eff1f4d

bold = el.name == "th"

I can add a long column header to the covid19.csv test data though if that is what you mean.

I mean some corner cases, for example, dataframe's index is th and what happens to it if it is very long?

And if user make one cell text bold by styler and the text is quite long, will it cause overlap?

But if you do not want to implement this is OK too. I can merge this after you add long header test case and a TODO at the place.

No problem! Just added a test for a dataframe with long column headers. If you run the test and check the images in the test output folder you should see the headers do not overlap.

As for edge cases, like making one cell bold etc, correct me if I am wrong, but it seems that bold styles applied to the body do not work in the current implementation? Likely due to the following:

https://github.com/dexplo/dataframe_image/blob/eff1f4dcc2256c704adefff8f04e01afe5d4e2de/dataframe_image/_matplotlib_table.py#L102C5-L112