The curl() and curl_download() functions provide highly configurable drop-in replacements for base url() and download.file() with better performance, support for encryption (https, ftps), gzip compression, authentication, and other libcurl goodies. The core of the package implements a framework for performing fully customized requests where data can be processed either in memory, on disk, or streaming via the callback or connection interfaces. Some knowledge of libcurl is recommended; for a more-user-friendly web client see the 'httr' package which builds on this package with http specific tools and logic.

Note that you cannot use devtools::install_github() because it uses curl ;)

install.packages("https://github.com/jeroen/curl/archive/master.tar.gz", repos = NULL)

About the R package:

• Vignette: curl: A Modern and Flexible Web Client for R

Other resources:

• libcurl handle options overview (use with handle_setopt in R)

There are three download interfaces (memory, disk and streaming). Always start by setting up a request handle:

library(curl)
h <- new_handle(copypostfields = "moo=moomooo")
handle_setheaders(h,
  "Content-Type" = "text/moo",
  "Cache-Control" = "no-cache",
  "User-Agent" = "A cow"
)

Perform request and download response in memory:

# Perform the request
req <- curl_fetch_memory("http://httpbin.org/post", handle = h)

# Show some outputs
parse_headers(req$headers)
cat(rawToChar(req$content))
str(req)

Or alternatively, write response to disk:

tmp <- tempfile()
curl_download("https://httpbin.org/post", tmp, handle = h)
readLines(tmp)

Or stream response via Connection interface:

con <- curl("https://httpbin.org/post", handle = h)
open(con)

# Get 3 lines
readLines(con, n = 3)

# Get remaining lines and close connection
readLines(con)
close(con)

Binary packages for OS-X or Windows can be installed directly from CRAN:

install.packages("curl")

Installation from source on Linux requires libcurl. On Debian or Ubuntu use libcurl4-openssl-dev:

sudo apt-get install -y libcurl-dev

On Fedora, CentOS or RHEL use libcurl-devel:

sudo yum install libcurl-devel

On OS-X libcurl is included with the system so nothing extra is needed. However if you want to build against the most recent version of libcurl, install and force-link curl from homebrew:

brew install curl
brew link --force curl

Note that on OS-X you must recompile the R package from source after force-linking curl, otherwise you get a version conflict with the system version of libcurl.

Because devtools and httr depend on curl, installing with install_github does not work well. The easiest way to install the development version of curl is a clean R session:

install.packages("https://github.com/jeroen/curl/archive/master.tar.gz", repos = NULL)

Of course windows users need Rtools to compile from source.

To use HTTP/2 the libcurl library has to be built --with-nghttp2 which is usually not the default. Use curl_version() in R to check if it is enabled. To enable it on OSX use:

brew install curl --with-nghttp2
brew link curl --force

And then reinstall the curl package from source:

install.packages("https://github.com/jeroen/curl/archive/master.tar.gz", repos = NULL)

Check again curl_version() to see if HTTP/2 is enabled now.