Add support for rownames_to_stub + opt_interactive()

[olivroy]

fixes #1702

Do you think it is worth adding a new test?

gt(iris, rownames_to_stub = T) %>% 
    opt_interactive()

Maybe the width is too wide? I don't know how to tweak it.

This PR

Before

[rich-iannone]

This is great! I wouldn't worry about the width. There's a default width here that can be set for all widths in tab_options(), plus you can individually set them with cols_width(). The width rules are different in interactive tables because the underlying framework constructs a table without the <table> element, so there is no automatic sizing based on content.

[rich-iannone]

Also, I wouldn't worry about the adding of a test here.

[rich-iannone]

LGTM!

[rich-iannone]

I've reviewed and approved. If there is nothing more to change here, I'll merge once all tests pass.

[olivroy]

Thanks for the review. All good on my side, but let's wait for test-coverage to pass! I have tried to address #1706, but realized it was not as straight forward as I had imagined. I may take another look later on.

[rich-iannone]

Thanks for taking on these bugs! If you're interested in working on #1510, it's along the same lines (but not interactive) and also a pretty serious bug. I suspect the issue there is that we are not supplying enough column width values and columns get dropped because of that.

[olivroy]

Hi @rich-iannone.

I may try again later, but here is the code I used in case it is useful to you

tb <- mtcars %>% 
  dplyr::select(mpg, cyl, vs) %>% 
  head() %>% 
  gt(groupname_col = "vs", row_group_as_column = T)

one_width <- tb %>% 
  cols_width(
    mpg ~ px(100),
  )

all_widths <-  tb %>% 
  cols_width(
    mpg ~ px(100),
    cyl ~ px(100)
  )
#' Somehow the only thing inside the gt object that is not different  
waldo::compare(
  one_width,
  all_widths
)
waldo::compare(one_width %>% render_as_html(), all_widths %>% render_as_html(), x_arg = "working", y_arg =  "bug", max_diffs = Inf, )


one_width %>% render_as_html() %>% 
  .[1] %>% 
  stringr::str_split_1(pattern = "\n") %>% 
  .[1:8]

all_widths %>% render_as_html() %>% 
  .[1] %>% 
  stringr::str_split_1(pattern = "\n") %>% 
  .[1:8]

#' # As a comparison, with `row_group_as_column = FALSE`
#' 
tb2 <-  mtcars %>% 
  dplyr::select(mpg, cyl, vs) %>% 
  head() %>% 
  gt(groupname_col = "vs", row_group_as_column = F)


one_width2 <- tb2 %>% 
  cols_width(
    mpg ~ px(100),
  )

all_widths2 <-  tb2 %>% 
  cols_width(
    mpg ~ px(100),
    cyl ~ px(100)
  )
#' Somehow the only thing inside the gt object that is not different  

# same as difference between one_width and all_widths
waldo::compare(one_width2 %>% render_as_html(), all_widths2 %>% render_as_html(), x_arg = "one_width", y_arg = "all_widths")

# same as the comparison above, so the issue is not in how gt sets col widths, the issue lies in the render html part.
waldo::compare(
  one_width2,
  all_widths2
)
waldo::compare(one_width2 %>% render_as_html(), all_widths2 %>% render_as_html(), x_arg = "one_width", y_arg = "all_widths")


# not very useful
waldo::compare(
  all_widths2 %>% render_as_html(),
  all_widths %>% render_as_html(),
  x_arg = "working",
  y_arg = "bug",
  max_diffs = Inf
)

# not very useful.
waldo::compare(
  one_width2 %>% render_as_html(),
  one_width %>% render_as_html(),
  x_arg = "working",
  y_arg = "working row_group_as_col"
)

Next steps for checking,

• cols_width() is not the culprit here.
• There may be a <col/> missing?
• There are 2 consecutive ; which means that something wrong is written
• The issue may be further than anticipated when doing the gt table

The issue may be caught more by a general html validator as locally, the rendered code doesn't tell us much...

Do you have tip on which helper function's output I should look at to really find out what is going wrong?

[rich-iannone]

Thanks for looking into this one. For that issue, I think the problem is within get_table_defs(). We can continue the discussion there. As for this PR, I'll merge it now!