View Source Estructura (estructura v1.6.0)
Estructura is a set of extensions for Elixir structures,
such as Access implementation, Enumerable and Collectable
implementations, validations and test data generation via StreamData.
Estructura simplifies the following
Accessimplementation for structsEnumerableimplementation for structs (as maps)Collectableimplementation for one of struct’s fields (asMapSetdoes)StreamDatageneration of structs for property-based testing
Use Options
use Estructura accepts four keyword arguments.
access: true | false | :lazywhether to generate theAccessimplementation, defaulttrue; whentrueor:lazy, it also producesput/3andget/3methods to be used withcoercionandvalidation, when:lazy, instances ofEstructura.Lazyare understood as valuescoercion: boolean() | [key()]whether to generate the bunch ofcoerce_×××/1functions to be overwritten by implementations, defaultfalsevalidation: boolean() | [key()]whether to generate the bunch ofvalidate_×××/1functions to be overwritten by implementations, defaultfalsecalculated: [{key(), formula}] when formula: binary() | Formulae.t() | (t() -> any())the calculated fieldsenumerable: boolean()whether to generate theEnumerableporotocol implementation, defaultfalsecollectable: false | key()whether to generate theCollectableprotocol implementation, defaultfalse; if non-falsey atom is given, it must point to a struct field whereCollectablewould collect. Should be one oflist(),map(),MapSet.t(),bitstribg()generator: %{optional(key()) => Estructura.Config.generator()}the instructions for the__generate__/{0,1}functions that would produce the target structure values suitable for usage inStreamDataproperty testing; the generated__generator__/1function is overwritable.
Please note, that setting coercion and/or validation to truthy values has effect
if and only if access has been also set to true.
Typical example of usage would be:
defmodule MyStruct do
use Estructura,
access: true,
coercion: [:foo], # requires `c:MyStruct.Coercible.coerce_foo/1` impl
validation: true, # requires `c:MyStruct.Validatable.validate_×××/1` impls
calculated: [foo: "length(bar)"], # requires `:formulae` dependency
enumerable: true,
collectable: :bar,
generator: [
foo: {StreamData, :integer},
bar: {StreamData, :list_of, [{StreamData, :string, [:alphanumeric]}]},
baz: {StreamData, :fixed_map,
[[key1: {StreamData, :integer}, key2: {StreamData, :integer}]]}
]
defstruct foo: 0, bar: [], baz: %{}
@impl MyStruct.Coercible
def coerce_foo(value) when is_integer(value), do: {:ok, value}
def coerce_foo(value) when is_float(value), do: {:ok, round(value)}
def coerce_foo(value) when is_binary(value) do
case Integer.parse(value) do
{value, ""} -> {:ok, value}
_ -> {:error, "#{value} is not a valid integer value"}
end
end
def coerce_foo(value),
do: {:error, "Cannot coerce value given for `foo` field (#{inspect(value)})"}
@impl MyStruct.Validatable
def validate_foo(value) when value >= 0, do: {:ok, value}
def validate_foo(_), do: {:error, ":foo must be positive"}
@impl MyStruct.Validatable
def validate_bar(value), do: {:ok, value}
@impl MyStruct.Validatable
def validate_baz(value), do: {:ok, value}
endThe above would allow the following to be done with the structure:
s = %MyStruct{}
put_in s, [:foo], "42"
#⇒ %MyStruct{foo: 42, bar: [], baz: %{}}
for i <- [1, 2, 3], into: s, do: i
#⇒ %MyStruct{foo: 0, bar: [1, 2, 3], baz: %{}}
Enum.map(s, &elem(&1, 1))
#⇒ [0, [], %{}]
MyStruct.__generator__() |> Enum.take(3)
#⇒ [
# %MyStruct{bar: [], baz: %{key1: 0, key2: 0}, foo: -1},
# %MyStruct{bar: ["g", "xO"], baz: %{key1: -1, key2: -2}, foo: 2},
# %MyStruct{bar: ["", "", ""], baz: %{key1: -3, key2: 1}, foo: -1}
# ]Calculated fields
When using Access, the calculated fields would be also updated upon the
update the fields then depend on.
Coercion
When coercion: true | [key()] is passed as an argument to use Estructura,
the nested behaviour Coercible is generated and the target module claims to implement it.
To make a coercion work with MyStruct.put/3 and put_in/3 provided
by Access implementation, the consumer module should implement MyStruct.Coercible
behaviour.
For the consumer convenience, the warnings for not implemented functions will be issued by compiler.
Validation
When validation: true | [key()] is passed as an argument to use Estructura,
the nested behaviour Validatable is generated and the target module claims to implement it.
To make a validation work with MyStruct.put/3 and put_in/3 provided
by Access implementation, the consumer module should implement MyStruct.Validatable
behaviour.
For the consumer convenience, the warnings for not implemented functions will be issued by compiler.
Generation
If generator keyword argument has been passed, MyStruct.__generate__/{0,1} can be
used to generate instances of this struct for StreamData property based tests.
property "generation" do
check all %MyStruct{foo: foo, bar: bar, baz: baz} <- MyStruct.__generator__() do
assert match?(%{key1: v1, key2: v2} when is_integer(v1) and is_integer(v2), baz)
assert is_integer(foo)
assert is_binary(bar)
end
endLazy
If access: :lazy is passed as an option, the struct content might be instantiated lazily,
upon first access through Kernel.×××_in/{2,3} family.
This might be explicitly helpful when the real content requires a significant time to load and/or store. Consider the full response from the web server, including the gzipped content, which might in turn be a huge text file. Or an attachment to an email.
Instead of unarchiving the content, one might use Lazy as
defmodule Response do
@moduledoc false
use Estructura, access: :lazy
def extract(file), do: {:ok, ZipHelper.unzip(file)}
defstruct __lazy_data__: nil,
file: Estructura.Lazy.new(&Response.extract/1)
end
response = %Response{__lazy_data__: zipped_content}
# immediate response
response |> get_in([:file])
# unzip and return
{unzipped, struct_with_cached_value} = response |> pop_in([:file])
# unzip and return the value, alter the struct with itSee Estructura.Lazy for details and options, see Estructura.LazyMap for
the implementation of lazy map.
Summary
Types
Diff return type
Types
Functions
Instantiates the struct by using Access from a map, passing all coercions and validations.
@spec diff(map() | struct(), map() | struct(), :diff) :: {map(), map()}
@spec diff(map() | struct(), map() | struct(), :overlap | :disjoint) :: map()
Calculates the difference between two estructures and returns a tuple with the first element containing same values and the second one with diffs.
This function accepts maps but this options should be used as a last resort because structs are 4–6 times faster.
Examples
defmodule M do
use Estructura, enumerable: true
defstruct a: true, b: false
end
Estructura.diff(struct(M, []), struct(M, b: true), :diff)
#⇒{%{a: true}, %{b: [false, true]}}
Estructura.diff(%{a: true, b: false}, %{a: true, b: true}, :overlap)
#⇒ %{a: true}
Estructura.diff(%{a: true, b: false}, %{a: true, b: true}, :disjoint)
#⇒ %{b: [false, true]}