In r-lib/brio: Basic R Input Output

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>",
  fig.path = "man/figures/README-",
  out.width = "100%"
)

brio - Basic R Input Output

Functions to handle basic input output, these functions always read and write UTF-8 files and provide more explicit control over line endings.

Reading files

library(brio)
write_lines(c("abc", "123"), "my-file")

# Write with windows newlines
write_lines(c("abc", "123"), "my-file-2", eol = "\r\n")

file_line_endings("my-file")

file_line_endings("my-file-2")

read_lines("my-file")

unlink(c("my-file", "my-file-2"))

Drop-ins

brio also has readLines() and writeLines() functions drop-in replacements for base::readLines() and base::writeLines(). These functions are thin wrappers around brio::read_lines() and brio::write_lines(), with deliberately fewer features than the base equivalents. If you want to convert a package to using brio you can add the following line and re-document.

#' @importFrom brio readLines writeLines

Benchmarks

Speed is not necessarily a goal of brio, but it does end up being a nice side effect.

gen_random <- function(characters, num_lines, min, max) {
  line_lengths <- sample.int(max - min, num_lines, replace = TRUE) + min
  vapply(line_lengths, function(len) paste(sample(characters, len, replace = TRUE), collapse = ""), character(1))
}

set.seed(42)

# generate 1000 random lines between 100-1000 characters long
data <- gen_random(letters, 1000, min = 100, max = 1000)

brio::write_lines(data, "benchmark")

Reading

Reading speeds are a decent amount faster with brio, mainly due to larger block sizes and avoidance of extra copies.

bench::mark(
  brio::read_lines("benchmark"),
  readr::read_lines("benchmark"),
  base::readLines("benchmark")
)

Writing

Write speeds are basically the same regardless of method, though brio does avoid some extra memory allocations.

bench::mark(
  brio::write_lines(data, "benchmark"),
  readr::write_lines(data, "benchmark"),
  base::writeLines(data, "benchmark"),
  check = FALSE
)

unlink("benchmark")