Encode/Decode base64 content

These functions encode and decode strings using base64 representations. base64 can be used as a single entry point with an argument to encode or decode. The other two functions perform the specific action.

programming, IO
base64(txt, encode = !inherits(txt, "base64"), mode = "character")

character string to encode or decode


logical value indicating whether the desired action is to encode or decode the object. If txt has (S3) class base64, the default is to decode this.


a character string which is either "raw" or "character". This controls the type of vector that is returned. If this is "raw", a raw vector is created. Otherwise, a character vector of length 1 is returned and its element is the text version of the original data given in txt.


This calls the routines in libcurl. These are not declared in the curl header files. So the support may need to be handled carefully on some platforms, e.g. Microsoft Windows.


If encode is TRUE, a character vector with a class named base64. If decode is TRUE, a simple string.


This is currently not vectorized.

We might extend this to work with raw objects.


libcurl - Wikipedia's explanation of base 64 encoding -

  • base64
  • base64Encode
  • base64Decode
    # encode and then decode a simple string.
  txt = "Some simple text for base 64 to handle"
  x = base64(txt)

    # encode to a raw vector
  x = base64("Simple text", TRUE, "raw")

    # decode to a character string.
  ans = base64Decode(x)
  ans == txt
    # decoded to a raw format.
  ans = base64Decode(x, "raw")

   # Binary data
#  f = paste(R.home(), "doc", "html", "logo.jpg", sep = .Platform$file.sep)
  f = system.file("examples", "logo.jpg", package = "RCurl")
  img = readBin(f, "raw",[1, "size"])
  b64 = base64Encode(img, "raw")
  back = base64Decode(b64, "raw")
  identical(img, back)

   # alternatively, we can encode to a string and then decode back again
   # to raw and see that we preserve the date.
  enc = base64Encode(img, "character")
  dec = base64Decode(enc, "raw")
  identical(img, dec)

   # The following would be the sort of computation we could do if we
   # could have in-memory raw connections.
   # We would save() some objects to such an in-memory binary/raw connection
   # and then encode the resulting raw vector into a character vector.
   # Then we can insert that into a message, e.g. an email message or
   # an XML document and when we receive it in a different R session
   # we would get the string and reverse the encoding from the string to
   # a raw vector
   # In the absence of that in-memory connection  facility in save(),
   # we can use a file.
  x = 1:10

   # save two objects - a function and a vector
  f = paste(tempfile(), "rda", sep = ".")
  save(base64, x, file = f)

   # now read the results back from that file as a raw vector
  data = readBin(f, "raw",[1,"size"])

   # base64 encode it
  txt = base64Encode(data, "character")

  if(require(XML)) {
    tt = xmlTree("r:data", namespaces = c(r = ""))
    out = saveXML(tt)

    doc = xmlRoot(xmlTreeParse(out, asText = TRUE))
    rda = base64Decode(xmlValue(doc), "raw")
    f = tempfile()
    writeBin(rda, f)
    e = new.env()
    load(f, e)

   # we'd like to be able to do
   #  con = rawConnection(raw(), 'r+')
   #  save(base64, x, file = con)
   #  txt = base64Encode(rawConnectionValue(con), "character")
   # ... write and read xml stuff
   #  val = xmlValue(doc)
   #  rda = base64Decode(val, "raw")
   #  e = new.env()
   #  input = rawConnection(o, "r")
   #  load(input, e)
# }
Documentation reproduced from package RCurl, version 1.98-1.1, License: BSD_3_clause + file LICENSE

Community examples

Looks like there are no examples yet.