emphatic

emphatic is a tool for exploratory analysis of tabular data. It allows the user to visually colour elements of the data, yet still keep all values visible.

Conceptually, emphatic highlighting could be considered to lie between tabular output and graphical output - like a table it shows all values, but like graphs these values are used to control appearance (i.e. colour).

emphatic follows the conventions of dplyr for selecting rows and columns; advanced colouring is via colour scales provided by ggplot2.

Supported output

This highlighting works

  • in the console
  • when output to Excel (using write_xlsx())
  • rendered in Rmarkdown documents (to HTML, latex and typst outputs)
  • rendered in Quarto documents (to HTML, latex and typst outputs)
  • rendered to SVG
  • rendered as multiple frames to animated SVG
Click here to show/hide gif demo

What’s in the box

  • hl() for user-defined highlighting of data.frames
  • hl_diff() for highlighting differences between two objects
  • hl_grep() highlight regular expression matches in an object or string
  • Manual conversion to other formats using:
  • Conversion to the correct output format is performed automatically when knitting an Rmarkdown or Quarto document.

Installation

You can install from GitHub with:

# install.packages('remotes')
remotes::install_github('coolbutuseless/emphatic')

Highlighting of data.frames with hl()

hl() lets you specify a palette, and the rows/columns of the data.frame the palette should apply to.

hl() calls are cumulative - so the required highlighting can be built up one-step-at-a-time.

To add highlighing to a data.frame:

  • Specify a palette
    • a single colour or vector of colours
    • a ggplot2 Scale object e.g. scale_colour_continuous()
  • Specify rows and columns
    • default: highlight all rows and all columns.
    • numeric vector giving row/column indices e.g. c(1, 2, 8), 1:8
    • character vector giving row/column names e.g. c('mpg', 'wt')
    • vector of bare column names e.g. c(mpg, wt), mpg:wt
    • tidyselect-style selectors: starts_with(), ends_with(), everything(), all_of(), any_of(), matches(), contains(), row_number(), n()
    • row selection using filtering operations e.g. cyl == 6 & mpg > 20

Simple example

By default, colouring will be applied to all rows and columns, and the supplied vector of colours will be recycled to meet the required length.

mtcars |>
  head(15) |>
  hl(c('red', 'white', 'blue'))

Complex example

A more complex example showing how to highlight the mtcars dataset where:

  • Highlight the row for the car with the minimum horsepower
  • Determine an expressive colouring for mpg using scale_colour_viridis_c() - where low values are darker, and high values are a bright yellow
  • Apply the scale’s colouring to all columns from mpg to disp 
mtcars |>
  head(15) |>
  hl('hotpink', rows = hp == min(hp)) |>
  hl(
    palette     = ggplot2::scale_colour_viridis_c(option = 'A'),
    cols        = mpg,      # Where the colour scale is calculated
    scale_apply = mpg:disp, # Where the colour scale is applied
    show_legend = TRUE
  )

Highlight difference between two objects with hl_diff()

The Levenshtein edit distance is calculated between the string representation of two objects and these edits are then coloured 🟢 = insert, 🔴 = delete, 🔵 = substitute.

x <- "Paris in the the spring?"
y <- "Not Paris in the spring!"
hl_diff(x, y)

Levenshtein’s edit distance naturally applies to strings, but hl_diff() can visualise the difference between arbitrary objects by first converting them to a string representation. Coercion to a string is controlled by the coerce argument, and defaults to the output if the objects were print()ed.

In this example, the difference between the mean() and median() function definitions is highlighted.

hl_diff(mean, median, coerce = 'print', sep = " ")

Highlight regular expression matches in objects with hl_grep()

hl_grep() highlights the regular expression matches within a string or objects coerced into a string representation.

Highlight regular expression matches in a character string 

txt <- "Among the few possessions he left to his heirs was a set of 
Encyclopedia Britannica in storage at the Lindbergh Palace Hotel under
the  names Ari and Uzi Tenenbaum. No-one spoke at the funeral, and 
Father Petersen's leg had not yet mended, but it was agreed among them 
that Royal would have found the event to be most satisfactory.
[Chas, now wearing a black Adidas tracksuit, nods to his sons]"

hl_grep(txt, "event.*satisfactory", coerce = 'character')

Highlight regular expression matches within an object

Other R objects (functions, lists, data.frames, etc) can also be highlighted with regular expressions. How an object is coerced into string representation is controlled by the coerce argument.

In this example, the function body for mode() is searched for the word switch:

hl_grep(mode, 'switch')

Animated SVG

Multiple emphatic objects may be rendered to an svg animation using as_svg_anim()

objs <- list(
  hl_grep("hello", "there"),
  hl_grep("goodbye", "good boy")
) 

svg <- as_svg_anim(objs, width = 600, height = 300, duration = 2, 
                   playback = 'infinite', font_size = "2em")

Options

  • hl_opts() create a named list of default options accepted by the functions in this package
  • hl_adjust() to adjust options after creation.
  • These options are initialised at package start time using Sys.getenv(). Set these values as environment variables in your .Rprofile to save your preferred settings across different sessions. e.g.
    • Sys.setenv(HL_NA = "<none>") prior to loading package or in .Rprofile
    • options(HL_NA = ".") at any time
Option Description
HL_NA String to use for NA values. Default “NA”
HL_FULL_COLOUR Should full colour ANSI codes be used when outputting to the console? Default: FALSE on Rstudio, but TRUE on all other R consoles
HL_TEXT_MODE How to handle text if no text colour has been explicitly specified by the user
"contrast" (default) automatically select a colour which contrasts with the background
"asis" do not change the colour from the console’s default
"remove" remove all text without a user-defined colour
HL_TEXT_CONTRAST When text_mode = "contrast" this numeric value (in range [0, 1]) adjusts the visibility of the text. Default: 1 (high contrast)
HL_GREP_COL The fill colour to use with hl_grep() if no colour is specified. Default: “#0F19F0”
HL_SUB_COL, HL_INS_COL, HL_DEL_COL the default colours to use with hl_diff() for substitution, insertion and deletion (respectively).

Vignettes

See the online documentation for vignettes and more examples.