Introducing the ggreverse package for turning ggplots back in to code

2019-04-26

ggreverse

ggreverse takes a ggplot object and returns the code to create that plot.

This package was written as a learning exercise to help me figure out some of the internal structure of a ggplot object.

ToDo

  • Reverse engineering of facetting and scales from a plot object.
  • aes_string() is currently unsupported.
  • Using tidyeval in aes() calls is currently unsupported.
  • Lots of other stuff :)

Installation

You can install from GitHub with:

# install.packages("remotes")
remotes::install_github("coolbutuseless/ggreverse")

Example ggreverse::convert_to_code()

  1. Create a ggplot
  2. Convert the ggplot back into code using ggreverse
  3. Convert the code back into a plot
library(ggreverse)

plot_df <- mtcars

# Create a ggplot2 plot object
p <- ggplot(plot_df) +
  geom_point(aes(mpg, wt), size = 3) +
  labs(title = "hello") +
  theme_bw() + 
  coord_equal()

# Convert the plot object back into code
plot_code <- ggreverse::convert_to_code(p)
plot_code
Warning: Could not use colored = TRUE, as the package prettycode is not
installed. Please install it if you want to see colored output or see `?
print.vertical` for more information.
ggplot(data = plot_df) +
  geom_point(mapping = aes(x = mpg, y = wt), size = 3, position = position_identity(), stat = "identity") +
  labs(title = "hello", x = "mpg", y = "wt") +
  theme_bw(11) +
  coord_fixed()
# Parse the plot code back into a plot - which should match the original plot
eval(parse(text = plot_code))

SessionInfo

Developed against:

  • R 3.5.3
  • ggplot2 v3.1.1