Chapter 8 Cell content

By default, the displayed content of each cell will be the result of a simple formatting. The content can also be composed as the result of a concatenation of several chunks.

The tables will all display cell contents representing the contents of the corresponding input data.frame cell. If it’s a string, it will remain as is, if it’s a number, it will be transformed into a string with a certain number of decimal places, if it’s a date, it will be transformed into a string representing a date, and so on.

The default display is conditioned by the type of data, i.e. columns of type double will be formatted with colformat_double(), columns of type integer will be formatted with colformat_int(), columns of type character will be formatted with colformat_char() and so on.

First, we explain the use of the colformat_* functions and then we will see how to fill cells with richer contents (i.e. being able to mix images, texts and hyperlinks) with the compose(as_paragraph()) functions. Function compose() also exists in the ‘purrr’ package which often causes issues. The identical function mk_par() has been created to avoid that issue package conflict issues, i.e. flextable::mk_par() and flextable::compose() are the same exact functions. For clarity, only mk_par() is used in this book.

Before showing tables, let’s specify their post-processing before printing with function autofit():

set_flextable_defaults(
  post_process_html = autofit,
  post_process_pdf = autofit,
  post_process_docx = autofit
  )

8.1 Soft returns and tabulations

A cell is made of one single paragraph of text. Paragraphs can contain several chunks of text with different formatting but also images. (See keypoints.)

When working with flextable, if a string contains \n or \t, it will be treated as a soft return (not a new paragraph!) or a tabulation.

We don’t recommand to change your data so that it contains \n or \t. Instead we recommand to use mk_par(), prepend_chunks() or append_chunks(). See [#multi-content] for more details about these functions.

dat <- data.frame(a = c("Grand total", letters[1:4]))
flextable(dat) |>
  prepend_chunks(i = ~ a != "Grand total", j = "a", as_chunk("\t"))

8.2 Simple formatting of cell content

Theses are high level functions that should satisfy most of the usual formatting needs. They can be used to define the formatting of one or more columns and eventually on a subset of rows.

Each accept a prefix and suffix argument that can be used to add a currency symbol for example. Also they all have na_str argument (defaut to ““), the string to use when data are not available.

colformat_num() let you format columns of type numeric as it is formatted in your R console. This is the default function applied to all numeric columns when a flextable is created. It does not support digits argument, it uses default R options through calls to format().
colformat_double() with arguments digits and big.mark: let you format columns of type double.
colformat_int() with arguments big.mark: let you format columns of type integer.
colformat_char(): let you format columns of type character.
colformat_lgl(): let you format columns of type logical.
colformat_image(): let you format image paths as images.
colformat_date() and colformat_datetime(): let you format columns of type date and datetime (POSIX).

The following illustration is using dataset “people” (see table 15.4).

flextable(people) %>% 
  colformat_num(
    big.mark = " ", decimal.mark = ",",
    na_str = "na") %>% 
  colformat_int(big.mark = " ") %>% 
  colformat_char(j = "eye_color", prefix = "color: ") %>% 
  colformat_date(fmt_date = "%d/%m/%Y")

Keep in mind that by setting the default display values, these calls can be avoided in most cases. They can be updated with function set_flextable_defaults().

old_settings <- set_flextable_defaults(
  digits = 1, decimal.mark = ",", big.mark = " ",
  na_str = "na", fmt_date = "%d/%m/%Y")

flextable(people)

do.call(set_flextable_defaults, old_settings)

8.3 Replace displayed labels

The function labelizor() replaces text values in a flextable with labels. The labels are defined with character named vector or with a function that transforms the existing text.

Let’s illustrate these two options with a table representing an aggregation.

library(palmerpenguins)

ft_pen <- penguins |> 
  select(species, island, ends_with("mm")) |> 
  group_by(species, island) |> 
  summarise(
    across(
      where(is.numeric), 
      .fns = list(
        avg = ~ mean(.x, na.rm = TRUE),
        sd = ~ sd(.x, na.rm = TRUE)
      )
    ),
    .groups = "drop") |> 
  rename_with(~ tolower(gsub("_mm_", "_", .x, fixed = TRUE))) |> 
  flextable() |> 
  colformat_double() |>
  separate_header() |> 
  theme_vanilla() |> 
  align(align = "center", part = "all") |>
  valign(valign = "center", part = "header") |> 
  autofit()
ft_pen

Now, let’s replace the names of calculated columns with prettier labels:

ft_pen <- labelizor(
  x = ft_pen, 
  part = "header", 
  labels = c("avg" = "Mean", "sd" = "Standard Deviation"))
ft_pen

And now, let’s format headers with title case:

ft_pen <- labelizor(
  x = ft_pen, 
  part = "header", 
  labels = stringr::str_to_title)
ft_pen

8.4 Multi content

The user can have more control over displayed content by using function compose() or mk_par() which are identical. The function enables the user to define the elements composing the paragraph and their respective formats. It can also be used to mix text chunks and images.

8.4.1 Function mk_par

flextable content can be defined with function mk_par().

It lets user control the formated content at the cell level of the table. It is possible to define content for a row subset and a column as well as for the whole column. One can mix images and text (but not with PowerPoint because PowerPoint can not do it).

The function requires a call to as_paragraph() which will concatenate text or images chunks as a paragraph.

The following illustration is using dataset “fishes” (see table 15.3). It shows how to control the format of displayed values and how to associate them with specific text formatting properties:

ft <- flextable(fishes) %>% 
  mk_par(j = "latin_name", 
          value = as_paragraph(
            as_chunk(latin_name, 
                     props = fp_text_default(color = "#C32900", bold = TRUE)))) %>% 
  mk_par(j = "french_name", 
          value = as_paragraph(
            as_chunk(french_name, 
                     props = fp_text_default(color = "#006699", bold = TRUE))))
ft

With this system, it’s easy to concatenate multiple values:

ft <- flextable(fishes, col_keys = c("dummy", "X00", "X01")) %>%
  mk_par(j = "dummy", value = as_paragraph(as_i(latin_name), as_b(french_name))) %>% 
  color(j = "dummy", color = "#006699") %>% 
  set_header_labels(dummy = "Species")
ft

Or to define specific title headers:

mk_par( 
  ft, j = "dummy", part = "header",
  value = as_paragraph(
    "Species ", 
    as_chunk("* latin/french name", 
             props = fp_text_default(color = "#006699", vertical.align = "superscript"))
    )
  )

Note that mk_par is not appending but is replacing the content.

8.4.2 Sugar functions for complex formatting

Functions as_b, as_i, as_sub, as_sup are special functions that can be used together. They set a value as bold, italic, subscripted or superscripted. This is particularly useful when the headers need complex formatting.

data <- structure(list(Species = structure(1:3, .Label = c("setosa", 
"versicolor", "virginica"), class = "factor"), col1 = c(5.006, 
5.936, 6.588)), class = "data.frame", row.names = c(NA, -3L))

ft <- flextable(data) %>%
  mk_par(
    part = "header", j = "Species",
    value = as_paragraph(as_i(as_b("Species")))
  ) %>%
  mk_par(
    part = "header", j = "col1",
    value = as_paragraph(as_b("µ"), as_sup("blah"))
  )
ft

8.4.3 Append or prepend content

prepend_chunks() or append_chunks() functions are handy functions that let users add in an existing paragraph one or more chunks. It can be any function dedicated to chunk creation, i.e. as_image(), as_chunk(), as_equation(), gg_chunk(), plot_chunk(). Function prepend_chunks() can be used to add tabulations for example.

ft <- flextable(data) %>%
  append_chunks(
    part = "body", i = 1, j = "Species",
    as_b(" some bold text")
  ) %>%
  prepend_chunks(
    part = "body", j = "col1", i = 2,
    as_chunk("hello ", props = fp_text_default(color = "red"))
  )
ft

8.5 Images

Function mk_par supports image insertion. Use function as_image in as_paragraph call.

The following illustration is using dataset “Tennis players” (see table 15.1).

To display only one image per cell, you can use the colformat_image function. The dimensions of the image(s) must always be specified.

flextable(tennis_players) %>%
  colformat_image(j = "head", width = .5, height = 0.5) %>% 
  colformat_image(j = "flag", width = .5, height = 0.33)

For a more complex display, such as mixing text and image, use the mk_par and as_image functions.

flextable(tennis_players,
  col_keys = c(
    "Rank", "Player", "Percentage",
    "Games.Won", "Total.Games", "Matches"
  )
) %>%
  mk_par(
    j = "Player",
    value = as_paragraph(
      as_image(src = head, width = .5, height = 0.5),
      " ",
      as_image(src = flag, width = .5, height = 0.33),
      " ", as_chunk(x = Player, fp_text_default(color = "#337ab7"))
    )
  )

8.6 Images and limitations of PowerPoint

Using images in flextable is not supported when output format is PowerPoint. This is not a choice nor a unimplemented feature. This is because PowerPoint is not able to embed images in a table cell. That’s a PowerPoint limitation. The same limitation occurs with ggplot charts and mini charts.

If beeing able to display images in PowerPoint table is important to you, you can use the plot function or the save_as_image and embed the result in the PowerPoint. You will of course loose ability to edit the table in PowerPoint.

8.7 Mini charts

set_flextable_defaults(
  post_process_html = function(x){
    theme_alafoli(x) %>% 
      align(align = "center", part = "all") %>% 
      align(align = "left", part = "footer") %>% 
      autofit()
  }
)

8.7.1 base plots and ggplot objects

There are four types of mini plots available that can be inserted into flextables: ‘box’, ‘line’, ‘points’ and ‘density’. This requires to store a list column in your data.frame as these are functions that need more than a single point.

z <- as.data.table(iris)
z <- z[, list(
  data = list(.SD$Sepal.Length)
), by = "Species"]

flextable(z) %>% 
  mk_par(j = "data", value = as_paragraph(
    plot_chunk(value = data, type = "dens", col = "red")
    ))

ggplot objects are supported and can be inserted with gg_chunk(). It lets users implement any graphics with ggplot2.

It usually requires a structure containing data grouped by one or more factors and a function to call on each content that produces a ggplot graph (in which adding the theme_void() is common).

Below, the function used to draw bars:

gg_bars <- function(z) {
  z <- scale(z)
  z <- na.omit(z)
  z <- data.frame(x = seq_along(z), z = z, w = z < 0)
  ggplot(z, aes(x = x, y = z, fill = w)) +
    geom_col(show.legend = FALSE) +
    theme_void()
}

Now the dataset:

dat <- as.data.table(mtcars)
z <- dat[,
  lapply(.SD, function(x) list(gg_bars(x))),
  by = c("vs", "am"), .SDcols = c("mpg", "disp", "drat")
]

And now the flextable with the ggplots:

ft <- flextable(z)
ft <- mk_par(ft, 
    j = c("mpg", "disp", "drat"),
    value = as_paragraph(gg_chunk(value = ., height = .15, width = 1)),
    use_dot = TRUE
  )
ft

8.7.2 minibar

Function mk_par supports mini barplots insertion. Use function minibar in as_paragraph call:

flextable( head(iris, n = 10 )) %>% 
  mk_par(j = 1,
          value = as_paragraph(
            minibar(value = Sepal.Length, max = max(Sepal.Length), height = .15)
            ),
  part = "body")

8.7.3 linerange

Function mk_par supports mini linerange insertion. Use function linerange in as_paragraph call:

flextable( head(iris, n = 10 )) %>% 
  mk_par(j = 1,
          value = as_paragraph(
            linerange(value = Sepal.Length, max = max(Sepal.Length), height = .15)
            ),
  part = "body")