Created: 2008-04-28 21:52
Updated: 2019-02-19 15:38
License: other


R YAML package

The R YAML package implements the libyaml YAML parser and emitter for R.

Table of Contents

What is YAML?

YAML is a human-readable markup language. With it, you can create easily readable documents that can be consumed by a variety of programming languages.


Hash of baseball teams per league:

- Boston Red Sox
- Detroit Tigers
- New York Yankees
- New York Mets
- Chicago Cubs
- Atlanta Braves

Data dictionary specification:

- field: ID
  description: primary identifier
  type: integer
  primary key: yes
- field: DOB
  description: date of birth
  type: date
  format: yyyy-mm-dd
- field: State
  description: state of residence
  type: string



You can install this package directly from CRAN by running (from within R): install.packages('yaml')


  1. Download the appropriate zip file or tar.gz file from the Github releases page.
  2. Run R CMD INSTALL <filename>

Git (via devtools)

  1. Install the devtools package from CRAN.
  2. In R, run the following:


The yaml package provides three functions: yaml.load, yaml.load_file and as.yaml.


yaml.load is the YAML parsing function. It accepts a YAML document as a string. Here's a simple example that parses a YAML sequence:

x <- "
- 1
- 2
- 3
yaml.load(x)  #=> [1] 1 2 3


A YAML scalar is the basic building block of YAML documents. Example of a YAML document with one element:


In this case, the scalar "1.2345" is typed as a float (or numeric) by the parser. yaml.load would return a numeric vector of length 1 for this document.

yaml.load("1.2345")  #=> [1] 1.2345


A YAML sequence is a list of elements. Here's an example of a simple YAML sequence:

- this
- is
- a
- simple
- sequence
- of
- scalars

If you pass a YAML sequence to yaml.load, a couple of things can happen. If all of the elements in the sequence are uniform, yaml.load will return a vector of that type (i.e. character, integer, real, or logical). If the elements are not uniform, yaml.load will return a list of the elements.


A YAML map is a list of paired keys and values, or hash, of elements. Here's an example of a simple YAML map:

one: 1
two: 2
three: 3
four: 4

Passing a map to yaml.load will produce a named list by default. That is, keys are coerced to strings. Since it is possible for the keys of a YAML map to be almost anything (not just strings), you might not want yaml.load to return a named list. If you want to preserve the data type of keys, you can pass as.named.list = FALSE to yaml.load. If as.named.list is FALSE, yaml.load will create a keys attribute for the list it returns instead of coercing the keys into strings.


yaml.load has the capability to accept custom handler functions. With handlers, you can customize yaml.load to do almost anything you want. Example of handler usage:

integer.handler <- function(x) { as.integer(x) + 123 }
yaml.load("123", handlers = list(int = integer.handler))  #=> [1] 246

Handlers are passed to yaml.load through the handlers argument. The handlers argument must be a named list of functions, where each name is the YAML type that you want to be handled by your function. The functions you provide must accept one argument and must return an R object.

Handler functions will be passed a string or list, depending on the original type of the object. In the example above, integer.handler was passed the string "123".

Sequence handlers

Custom sequence handlers will be passed a list of objects. You can then convert the list into whatever you want and return it. Example:

sequence.handler <- function(x) {
  tmp <- as.numeric(x)
  tmp / 5
string <- "
- foo
- bar
- 123
- 4.567
yaml.load(string, handlers = list(seq = sequence.handler))  #=> [1]      NA      NA 24.6000  0.9134
Map handlers

Custom map handlers work much in the same way as custom list handlers. A map handler function is passed a named list, or a list with a keys attribute (depending on the value of as.named.list). Example:

string <- "
- 1
- 2
- 3
- 4
yaml.load(string, handlers = list(map = function(x) { as.data.frame(x) }))


  a b
1 1 3
2 2 4


yaml.load_file does the same thing as yaml.load, except it reads a file from a connection. For example:

x <- yaml.load_file("Data/document.yml")

This function takes the same arguments as yaml.load, with the exception that the first argument is a filename or a connection.


The read_yaml function is a convenience function that works similarly to functions in the readr package. You can use it instead of yaml.load_file if you prefer.


as.yaml is used to convert R objects into YAML strings. Example as.yaml usage:

x <- as.yaml(1:5)
cat(x, "\n")

Output from above example:

- 1
- 2
- 3
- 4
- 5

Notable arguments


You can control the number of spaces used to indent by setting the indent option. By default, indent is 2.

For example:

cat(as.yaml(list(foo = list(bar = 'baz')), indent = 3))


   bar: baz

By default, sequences that are within a mapping context are not indented.

For example:

cat(as.yaml(list(foo = 1:10)))


- 1
- 2
- 3
- 4
- 5
- 6
- 7
- 8
- 9
- 10

If you want sequences to be indented in this context, set the indent.mapping.sequence option to TRUE.

For example:

cat(as.yaml(list(foo = 1:10), indent.mapping.sequence=TRUE))


  - 1
  - 2
  - 3
  - 4
  - 5
  - 6
  - 7
  - 8
  - 9
  - 10

The column.major option determines how a data frame is converted into YAML. By default, column.major is TRUE.

Example of as.yaml when column.major is TRUE:

x <- data.frame(a=1:5, b=6:10)
y <- as.yaml(x, column.major = TRUE)
cat(y, "\n")


- 1
- 2
- 3
- 4
- 5
- 6
- 7
- 8
- 9
- 10


x <- data.frame(a=1:5, b=6:10)
y <- as.yaml(x, column.major = FALSE)
cat(y, "\n")


- a: 1
  b: 6
- a: 2
  b: 7
- a: 3
  b: 8
- a: 4
  b: 9
- a: 5
  b: 10

You can specify custom handler functions via the handlers argument. This argument must be a named list of functions, where the names are R object class names (i.e., 'numeric', 'data.frame', 'list', etc). The function(s) you provide will be passed one argument (the R object) and can return any R object. The returned object will be emitted normally.

Special features

Verbatim(-ish) text

Character vectors that have a class of 'verbatim' will not be quoted in the output YAML document except when the YAML specification requires it. This means that you cannot do anything that would result in an invalid YAML document, but you can emit strings that would otherwise be quoted. This is useful for changing how logical vectors are emitted. For example:

as.yaml(c(TRUE, FALSE), handlers = list(
  logical = function(x) {
    result <- ifelse(x, "true", "false")
    class(result) <- "verbatim"
Custom tags

You can specify YAML tags for R objects by setting the 'tag' attribute to a character vector of length 1. If you set a tag for a vector, the tag will be applied to the YAML sequence as a whole, unless the vector has only 1 element. If you wish to tag individual elements, you must use a list of 1-length vectors, each with a tag attribute. Likewise, if you set a tag for an object that would be emitted as a YAML mapping (like a data frame or a named list), it will be applied to the mapping as a whole. Tags can be used in conjunction with YAML deserialization functions like yaml.load via custom handlers, however, if you set an internal tag on an incompatible data type (like !seq 1.0), errors will occur when you try to deserialize the document.


The write_yaml function is a convenience function that works similarly to functions in the readr package. It calls as.yaml and writes the result to a file or a connection.

Additional documentation

For more information, run help(package='yaml') or example('yaml-package') for some examples.


There is a Makefile for use with GNU Make to help with development. There are several make targets for building, debugging, and testing. You can run these by executing make <target-name> if you have the make program installed.

Target name Description
compile Compile the source files
check Run CRAN checks
gct-check Run CRAN checks with gctorture
test Run unit tests
gdb-test Run unit tests with gdb
valgrind-test Run unit tests with valgrind
tarball Create tarball suitable for CRAN submission
all Default target, runs compile and test

Implicit tag discovery

The algorithm used whenever there is no YAML tag explicitly provided is located in the implicit.re file. This file is used to create the implicit.c file via the re2c program. If you want to change this algorithm, make your changes in implicit.re, not implicit.c. The make targets will automatically update the C file as needed, but you'll need to have the re2c program installed for it to work.


The VERSION file is used to track the current version of the package. Warnings are displayed if the DESCRIPTION and CHANGELOG files are not properly updated when creating a tarball. This is to help prevent problems during the CRAN submission process.

Cookies help us deliver our services. By using our services, you agree to our use of cookies Learn more