# Literals

Literals are used for values such as strings, numbers, and dates. They can be untyped, languaged-tagged or typed (but following the RDF 1.1 spec untyped literals are in fact just xsd:string typed literals)

# Untyped literals

In general literals are created with the RDF.Literal.new constructor function or its alias function RDF.literal:


The actual value can be accessed via the RDF.Literal.value/1 function:

iex> RDF.literal("foo") |> RDF.Literal.value()

An untyped literal can also be created with the ~L sigil or the ~l which supports string interpolation:

use RDF
# or an explicit: import RDF.Sigils


~L"foo #{fun(args)"

# Language-tagged literals

A language-tagged literal can be created by providing the language option with a BCP47 (opens new window)-conform language or by adding the language as a modifier to the ~L or ~l sigil:

import RDF.Sigils

RDF.literal("foo", language: "en")


Note: Only languages without subtags are supported as modifiers of the ~L sigil, i.e. if you want to use en-US as a language tag, you would have to use the constructor functions.

# Typed literals

A typed literal can be created by providing the datatype option with an IRI of a datatype. Most of the time this will be an XML schema datatype (opens new window):

alias RDF.NS

RDF.literal("42", datatype: NS.XSD.integer)

It is also possible to create a typed literal by using a native Elixir non-string value, for which the following datatype mapping will be applied:

Elixir datatype XSD datatype
string xsd:string
boolean xsd:boolean
integer xsd:integer
float xsd:double
Decimal (opens new window) xsd:decimal
Time xsd:time
Date xsd:date
DateTime xsd:dateTime
NaiveDateTime xsd:dateTime
URI xsd:AnyURI

So the former example literal can be created equivalently like this:


The value/1 function returns the literal value as the native Elixir value according to the above mapping. When a known XSD datatype is specified, the given value will be converted automatically if needed and possible.

iex> RDF.literal(42, datatype: NS.XSD.double) |> RDF.Literal.value()

iex> RDF.literal("0042", datatype: NS.XSD.byte) |> RDF.Literal.value()


For some datatypes where the value space of the XSD datatype is larger than what the corresponding Elixir datatype supports, you might also get a tuple with annotations back. For example, xsd:dates and xsd:times support timezones while Elixir's Date and Time structs don't support that. In case of an xsd:date with a timezone you'll get a tuple like this {~D[2014-09-01], "-08:00"}. For xsd:times with timezones you'll instead just get a tuple like {~T[23:00:00], true} with a boolean signifying that it has a timezone, since the timezone offset was already normalized in the value (the original timezone offset is kept in the lexical form).

For all of the supported RDF and XSD datatypes there are RDF.Literal.Datatype modules available that implement the semantics of the respective datatype. They also provide a new constructor function that allows the creation of RDF.Literals with the respective datatype. These constructor can also be called via the alias functions on the top-level RDF respective RDF.XSD namespace.

use RDF # this automatically defines an alias RDF.XSD
# we'll consider this alias to be defined throughout this guide implicitly

RDF.LangString.new("foo", language: "en")
RDF.langString("foo", language: "en")

Besides the RDF.LangString datatype the following XSD datatypes are provided as RDF.Literal.Datatypes:

XSD datatype RDF.Literal.Datatype
xsd:boolean RDF.XSD.Boolean
xsd:float RDF.XSD.Float
xsd:double RDF.XSD.Double
xsd:decimal RDF.XSD.Decimal
xsd:integer RDF.XSD.Integer
xsd:long RDF.XSD.Long
xsd:int RDF.XSD.Int
xsd:short RDF.XSD.Short
xsd:byte RDF.XSD.Byte
xsd:nonPositiveInteger RDF.XSD.NonPositiveInteger
xsd:negativeInteger RDF.XSD.NegativeInteger
xsd:nonNegativeInteger RDF.XSD.NonNegativeInteger
xsd:positiveInteger RDF.XSD.PositiveInteger
xsd:unsignedLong RDF.XSD.UnsignedLong
xsd:unsignedInt RDF.XSD.UnsignedInt
xsd:unsignedShort RDF.XSD.UnsignedShort
xsd:unsignedByte RDF.XSD.UnsignedByte
xsd:string RDF.XSD.String
xsd:dateTime RDF.XSD.DateTime
xsd:date RDF.XSD.Date
xsd:time RDF.XSD.Time
xsd:base64Binary RDF.XSD.Base64Binary

For literals with an unknown datatype, i.e. a datatype without a RDF.Literal.Datatype module the generic RDF.Literal.Generic implementation s used. For those generic literals the RDF.Literal.value/1 function simply returns the initially given value unvalidated and unconverted.

# Validation

The RDF.Literal.valid?/1 function checks if a given literal is valid according to the semantics in its RDF.Literal.Datatype implementation.

iex> RDF.Literal.valid?(XSD.integer("42"))

iex> RDF.Literal.valid?(XSD.integer("foo"))

Since the semantics of RDF.Literal.Generic literals is unknown they are always considered to be valid.

If you want to prohibit the creation of invalid literals, you can use the new! constructor function of the RDF.Literal.Datatype or RDF.Literal, which will fail in case of invalid values.

# Lexical and canonical form

A RDF literal is bound to the lexical form of the initially given value. This lexical representation can be retrieved with the RDF.Literal.lexical/1 function:

iex> RDF.Literal.lexical(XSD.integer("0042"))

iex> RDF.Literal.lexical(XSD.integer(42))

The RDF.Literal.canonical/1 function normalizes the given literal to the canonical lexical form according to its datatype:

iex> RDF.integer("0042") 
...> |> RDF.Literal.canonical()
...> |> RDF.Literal.lexical()

iex> RDF.Literal.canonical(RDF.integer("0042")) == 
...> RDF.Literal.canonical(RDF.integer("42"))

For RDF.Literal.Generic literals the canonical function returns the given literal unchanged.

Since the canonical form is undefined for invalid literals, nil is returned in this case.

If you're just interested in the canonical lexical form as a string you can also use the RDF.Literal.canonical_lexical/1 function, which is also a bit faster, since the intermediary canonicalization is not needed.

iex> RDF.Literal.canonical_lexical(XSD.integer("0042"))

# Equivalence

Although two literals might have the same value, they are not equal if they don't have the same lexical form:

iex> RDF.Literal.value(XSD.integer("0042")) == 
...> RDF.Literal.value(XSD.integer("42"))

iex> XSD.integer("0042") == XSD.integer("42")

The RDF.Literal.equal_value?/2 function however, does a pure value-based equivalence comparison. It also takes into account compatibilities between different types, eg. derived datypes. Since it is the basis for the implementation of SPARQLs = operator in SPARQL.ex everything that is equivalent in terms of this operator will match. Literals which aren't comparable in general due to their type and would result in an error match in terms of the SPARQL = operator (meaning that also the negation wouldn't match) will return nil. Above this, it also coerces native Elixir values to RDF.Literals before doing the comparison.

iex> XSD.integer("0042") 
...> |> RDF.Literal.equal_value?(XSD.integer("42"))

iex> XSD.integer("0042") 
...> |> RDF.Literal.equal_value?(42)

iex> XSD.integer("0042") 
...> |> RDF.Literal.equal_value?(XSD.short(42))

iex> XSD.anyURI("http://example.com") 
...> |> RDF.Literal.equal_value?(RDF.iri("http://example.com"))

iex> XSD.integer("0042") |> RDF.Literal.equal_value?(XSD.string("42"))

# Defining custom datatypes

You can define your own custom datatype by implementing the RDF.Literal.Datatype behaviour. Defining a completely independent dataype however, will probably be the exception and goes beyond the scope of this introductary guide. Most of the time you want to introduce a custom datatype by constraining one of the existing XSD datatypes through datatype derivation.


It should be noted that a triple store won't know how to handle your custom datatype unless it's a well-known datatype he supports, so they should be introduced cautiously. But at least in the RDF.ex libraries they will behave like the predefined XSD datatypes. In particular you can apply the respective SPARQL functions within SPARQL.ex on them.

So, a custom datatype can be derived from a XSD datatype (with an existing RDF.Literal.Datatype implementation) by defining a new module with use RDF.XSD.Datatype.Restriction and constraining its value space. RDF.ex implements most of the XSD facets (opens new window) as RDF.XSD.Facet modules for this:

XSD facet RDF.XSD.Facet
length RDF.XSD.Facets.Length
minLength RDF.XSD.Facets.MinLength
maxLength RDF.XSD.Facets.MaxLength
maxInclusive RDF.XSD.Facets.MaxInclusive
maxExclusive RDF.XSD.Facets.MaxExclusive
minInclusive RDF.XSD.Facets.MinInclusive
minExclusive RDF.XSD.Facets.MinExclusive
totalDigits RDF.XSD.Facets.TotalDigits
fractionDigits RDF.XSD.Facets.FractionDigits
explicitTimezone RDF.XSD.Facets.ExplicitTimezone
pattern RDF.XSD.Facets.Pattern

Within the body of a module using RDF.XSD.Datatype.Restriction you can apply one or multiple of these facets with the def_facet_constraint macro and specifying a value for the facets. This table shows which facets can be applied on the primitive datatypes (and their derived datatypes):

Primitive datatype Applicable facets
string length, maxLength, minLength, pattern
boolean pattern
float maxExclusive, maxInclusive, minExclusive, minInclusive, pattern
double maxExclusive, maxInclusive, minExclusive, minInclusive, pattern
decimal maxExclusive, maxInclusive, minExclusive, minInclusive, pattern, totalDigits, fractionDigits
integer maxExclusive, maxInclusive, minExclusive, minInclusive, pattern, totalDigits
duration maxExclusive, maxInclusive, minExclusive, minInclusive, pattern
dateTime explicitTimezone, maxExclusive, maxInclusive, minExclusive, minInclusive, pattern
time explicitTimezone, maxExclusive, maxInclusive, minExclusive, minInclusive, pattern
date explicitTimezone, maxExclusive, maxInclusive, minExclusive, minInclusive, pattern
anyURI length, maxLength, minLength, pattern

Let's see how a custom datatype for the age of a person could be defined in an application:

defmodule MyApp.PersonAge do
  use RDF.XSD.Datatype.Restriction,
      name: "person_age",
      id: "http://example.com/person_age",
      base: RDF.XSD.NonNegativeInteger

  def_facet_constraint RDF.XSD.Facets.MaxInclusive, 150

This datatype can now constructed by either its new constructor or via the generic typed RDF.Literal constuctor and the specified datatype URI.

iex> MyApp.PersonAge.new(42) == 
...> RDF.literal(42, datatype: "http://example.com/person_age")

Within RDF.ex and the libraries on top of it (SPARQL.ex, ShEx.ex) this datatype can be used wherever a xsd:nonNegativeInteger or xsd:integer is expected.

iex> XSD.integer(42) 
...> |> RDF.Literal.equal_value?(MyApp.PersonAge.new(42))

# Type checking and reflection

The datatype IRI of any RDF.Literal can be retrieved with the RDF.Literal.datatype_id/1 function.

iex> XSD.integer(42) |> RDF.Literal.datatype_id()

iex> ~L"foo"en |> RDF.Literal.datatype_id()

iex> RDF.literal("foo", datatype: "http://example.com/dt") |> RDF.Literal.datatype_id()

Although you won't need this most of the time, since you can use all types of literals via the polymorphic RDF.Literal functions, the inverse operation is also possible. When a RDF.Literal.Datatype is defined for a datatype IRI, you can get the module dynamically by its IRI with the RDF.Literal.Datatype.get/1 function.

iex> RDF.Literal.Datatype.get("http://www.w3.org/2001/XMLSchema#integer")

iex> RDF.Literal.Datatype.get("http://example.com/custom/datatype")

# assuming there's no custom RDF.Literal.Datatype for http://example.com/dt defined
iex> RDF.Literal.Datatype.get("http://example.com/dt")

An RDF.Literal with a datatype for which a RDF.Literal.Datatype is defined can be pattern matched via its literal field and the module implementing the RDF.Literal.Datatype.

def fun(%RDF.Literal{literal: %XSD.Integer{}} = integer_literal), do: ...

def fun(%RDF.Literal{literal: %My.Custom.Datatype{}} = my_literal), do: ...

A datatype pattern match on a literal can also be performed with the RDF.Guard.is_rdf_literal/2 guard, which checks if datatype module given as the second argument matches the literal given as the first argument.

use RDF 
# or an explicit: import RDF.Guards

def fun(value) when is_rdf_literal(value, XSD.Integer), do: ...

def fun(value) when is_rdf_literal(value, My.Custom.Datatype), do: ...

Literals with a datatype for which no RDF.Literal.Datatype is defined can be pattern matched via the datatype of the RDF.Literal.Generic datatype.

# assuming there's no custom RDF.Literal.Datatype for http://example.com/dt defined
def fun(%RDF.Literal{literal: %RDF.Literal.Generic{datatype: "http://example.com/dt"}} = literal), do: ...

Although pattern matching is the most elegant way for type checks, this only allows for exact datatype matches. The datatype?/1 functions on the individual RDF.Literal.Datatype modules are aware of derivations and check whether the datatype of a given literal is either the datatype for which the RDF.Literal.Datatype is defined or derived of this datatype.

iex> XSD.integer(42) |> XSD.Integer.datatype?()

iex> XSD.byte(42) |> XSD.Integer.datatype?()

iex> XSD.byte(42) |> XSD.UnsignedInteger.datatype?()

iex> XSD.byte(42) |> XSD.NegativeInteger.datatype?()

# assuming My.Custom.Datatype is derived from xsd:integer or one of its derived datatypes
iex> My.Custom.Datatype.new(42) |> XSD.Integer.datatype?()

# assuming there's a custom RDF.Literal.Datatype for http://example.com/dt defined and it is derived from a xsd:integer
iex> RDF.literal("foo", datatype: "http://example.com/dt") 
...> |> XSD.Integer.datatype?()

The RDF.XSD.Numeric.datatype?/1 function can also be handy. It checks if the datatype of a literal is one of the numeric XSD datatypes or derived from one of them.

iex> XSD.integer(42) |> XSD.Numeric.datatype?()

iex> XSD.string("foo") |> XSD.Numeric.datatype?()

# assuming My.Custom.Datatype is derived from a numeric XSD datatype
iex> My.Custom.Datatype.new(42) |> XSD.Numeric.datatype?()

# assuming there's no custom RDF.Literal.Datatype for http://example.com/dt defined or it is not derived from a numeric XSD datatype
iex> RDF.literal("foo", datatype: "http://example.com/dt") |> XSD.Numeric.datatype?()

The general purpose type checking function RDF.Literal.is_a?/2 supports all of these datatype?/1 functions

iex> XSD.byte(42) |> RDF.Literal.is_a?(XSD.Byte)

iex> XSD.byte(42) |> RDF.Literal.is_a?(XSD.Integer)

iex> XSD.byte(42) |> RDF.Literal.is_a?(XSD.Numeric)

iex> XSD.byte(42) |> RDF.Literal.is_a?(XSD.Datatype)

iex> RDF.langString("foo", language: "en") |> RDF.Literal.is_a?(XSD.Datatype)

# assuming there's no custom RDF.Literal.Datatype for http://example.com/dt
iex> RDF.literal("foo", datatype: "http://example.com/dt") 
...> |> RDF.Literal.is_a?(RDF.Literal.Generic)

Most of the functions on the RDF.Literal.Datatype modules are only applicable on literals of exact this datatype, but there are two notable exceptions which can be handy.

The valid?/1 function on RDF.Literal.Datatype modules is able to deal with derived datatypes and returns true if the given literal is valid AND of the proper type.

iex> XSD.byte(42) |> XSD.Integer.valid?()

iex> XSD.float(3.14) |> XSD.Integer.valid?()

The value/1 function on the RDF.Literal.Datatype modules also returns the value of literals when they are of a derived datatype (or nil if the datatype is not derived from this datatype.

iex> XSD.byte(42) |> XSD.Integer.value()

iex> XSD.float(3.14) |> XSD.Integer.value()