How to set background colors for the cells of multiple columns based on string contents

[ahbon123]

Hello, say I have a small dataset as follows:

library(gt)
library(tidyverse)

id <- c(1,2,3,4,5)
res1 <- c("true", "true", "false", "true", "false")
res2 <- c("false", NA, NA, "true", "true")
df <- data.frame(id, res1, res2)

df %>% 
  gt()

For columns res1 and res2, let's say if value content are trues, I'll need to highlight that cell with red background color, if falses, highlight with green color, for other cases, keep their colors as original.

Is it possible to achieve that using gt package?

[rich-iannone]

We need an easier way to set basic style attributes based on values. I think a new function for this should be the way to do it. The sub_values() function provides a nice (and forgiving) way to target values so I believe the new function ought to be modeled on that.

[ahbon123]

Thank you very much, I believe this function will have lots of application scenarios.

[jthomasmock]

Tagging on to this, there are a few requests I've seen on Twitter and gtExtras re: color the background of column 1 based on column 2. Right now, data_color() only operates on the current column, text_transform() only works on the current column (although this can be "abused"), and to properly color the background of a column you end up needing to use tab_style() multiple times.

I have a prototype that works like below:

set.seed(37)
base_tab <- head(mtcars) |>
  dplyr::slice_sample(n = 6) |>
  dplyr::arrange(desc(mpg)) |>
  gt::gt()

base_tab |>
  gt_color_by_col(mpg, cyl, domain = range(gtExtras::gt_index(base_tab, mpg)))

Essentially it loops through the column, values, and rows to apply a tab_style() with the minimal snippets of code below:

fn_data_clr_by_col <- function(gt_object, col2, value, row_id, color_add){

  gt_object |>
    tab_style(
      style = list(
        cell_fill(color = color_add),
        cell_text(color = gt:::ideal_fgnd_color(color_add))
      ),
      locations = cells_body({{ col2 }}, rows = row_id)
    )

}

for (i in 1:length(full_vals)){
    gt_object <- fn_data_clr_by_col(gt_object, {{ col2 }}, full_vals[i], i, color_add[i])
  }
  return(gt_object)

Full code:

Full Code

fn_data_clr_by_col <- function(gt_object, col2, value, row_id, color_add){

  gt_object |>
    tab_style(
      style = list(
        cell_fill(color = color_add),
        cell_text(color = gt:::ideal_fgnd_color(color_add))
      ),
      locations = cells_body({{ col2 }}, rows = row_id)
    )

}

gt_color_by_col <- function(
    gt_object, col1, col2, ..., domain = NULL, direction = 1, type = "continuous",
    palette = c("#af8dc3", "#e7d4e8", "#f7f7f7", "#d9f0d3", "#7fbf7b")){

  stopifnot("Table must be of class 'gt_tbl'" = "gt_tbl" %in% class(gt_object))

  full_vals <- gtExtras::gt_index(gt_object = gt_object, {{ col1 }})

  rng_vals <- range(full_vals, na.rm = TRUE)

  if(is.null(domain)){

    domain <- rng_vals
    warning(
      "Domain not specified, defaulting to observed range within each specified column.",
      call. = FALSE
    )
  }

  color_add <- scales::col_numeric(
    palette = if(grepl(x = palette[1], pattern = "::")){
      paletteer::paletteer_d(
        palette = palette,
        direction = direction,
        type = pal_type
      ) %>% as.character()
    } else {
      if(direction == -1){
        rev(palette)
      } else {
        palette
      }
    },
    ...,
    domain = domain
  )(full_vals)

  for (i in 1:length(full_vals)){
    gt_object <- fn_data_clr_by_col(gt_object, {{ col2 }}, full_vals[i], i, color_add[i])
  }
  return(gt_object)
}

[rich-iannone]

@jthomasmock if it doesn't already exist could you transfer your comments/ideas to a new issue? There's some great stuff there and it's such a common problem that I don't want to lose track of this.

[jthomasmock]

If it doesn't already exist could you transfer your comments/ideas to a new issue? There's some great stuff there and it's such a common problem that I don't want to lose track of this.

Done! I've moved it to: #1119 with some of the public requests I've seen