‘rectpacker’ implements the skyline algorithm for packing rectangles
into a box using stb_rect_pack.h from
stb.
I wrote this package as I needed to assemble graphical assets into a single image - hopefully minimising the size of the final image.
This package can be installed from CRAN
install.packages('rectpacker')You can install the latest development version from GitHub with:
# install.package('remotes')
remotes::install_github('coolbutuseless/rectpacker')Pre-built source/binary versions can also be installed from R-universe
install.packages('rectpacker', repos = c('https://coolbutuseless.r-universe.dev', 'https://cloud.r-project.org'))Pack 100 randomly sized rectangles into a 50x25 box.
Note that not all rectangles will fit into this box, but the algorithm
does the best it can. If a rectangle cannot fit, then packed = FALSE
library(ggplot2)
library(rectpacker)
set.seed(1)
N <- 100
widths <- sample(7, N, replace = TRUE)
heights <- sample(6, N, replace = TRUE)
rects <- pack_rects(50, 25, widths, heights)
# How many of the 100 rectangles actually got packed?
table(rects$packed)#>
#> FALSE TRUE
#> 18 82
# Keep only the rects which were able to be packed into the box
rects <- subset(rects, packed == TRUE)
head(rects)#> # A tibble: 6 × 6
#> idx w h packed x y
#> <int> <int> <int> <lgl> <int> <int>
#> 1 1 4 3 TRUE 33 19
#> 2 2 7 6 TRUE 14 0
#> 3 3 1 2 TRUE 49 21
#> 4 5 5 3 TRUE 23 19
#> 5 6 7 6 TRUE 0 0
#> 6 8 6 4 TRUE 0 12
# Plot the rectangles
ggplot(rects) +
geom_rect(
aes(
xmin = x,
ymin = y,
xmax = x + w,
ymax = y + h,
fill = as.factor(idx)
),
col = 'black'
) +
annotate('rect', xmin = 0, ymin = 0, xmax = 50, ymax = 25, fill = NA, col = 'red') +
coord_equal() +
theme_minimal() +
theme(legend.position = 'none') +
scale_fill_viridis_d()
Often it is necessary to find the size of a box which will fit all rectangles without too much wasted space.
calc_small_box() performs a brute force search for a suitable box
within the set of boxes of a given aspect ratio.
Note: This search is not guaranteed to find the absolute minimal box size.
set.seed(1)
N <- 100
rect_widths <- sample(N, N, T)
rect_heights <- sample(N, N, T)
box <- calc_small_box(rect_widths, rect_heights)
box#> $width
#> [1] 664
#>
#> $height
#> [1] 420
rects <- pack_rects(box$width, box$height, rect_widths, rect_heights)
ggplot(rects) +
geom_rect(
aes(
xmin = x,
ymin = y,
xmax = x + w,
ymax = y + h,
fill = as.factor(idx)
),
col = 'black'
) +
annotate('rect', xmin = 0, ymin = 0, xmax = box$width, ymax = box$height, fill = NA, col = 'red') +
coord_equal() +
theme_minimal() +
theme(legend.position = 'none') +
scale_fill_viridis_d(option = 'D')