| Safe Haskell | None |
|---|---|
| Language | Haskell2010 |
Data.Generic.Labels
Description
This module provides functionality for converting between different collections of fields.
Projections
To project out certain fields of a record, use project.
For instance:
data IBXD x = IBXD { i :: Int, b :: Bool, x :: x, d :: Double }
deriving stock Generic
data XI x = XI { x :: c, i :: Int }
deriving stock Generic
We can project out the two fields of interest from the first record type:
ibxd_to_xi :: IBXD x -> XI x ibxd_to_xi = project
Injections
Going the other way, we can use inject to override fields of the larger record
with those from the smaller record:
xi_into_ibxd :: XI x -> IBXD x -> IBXD x xi_into_ibxd = inject
Adapters
project and inject are two instances of the more general adapt function,
which allows us to only specify the missing arguments in the above example.
xi_plus_bd_makes_ibxd :: XI x -> ( "b" := Bool, "d" := Double ) -> IBXD x xi_plus_bd_makes_ibxd = adapt
In this situation, we are building up a record of type IBXD x out of two parts.
More generally, adapt allows for fields in the first argument to override
fields in the second argument, which provides a convenient mechanism for
named optional arguments.
adapt :: _ => givenArgs -> optionalArgs -> allArgs
For instance, if we have a function f which takes in several named arguments
type AllArgsTuple = ( "arg1" := Ty1, "arg2" := Ty2, "arg3" := Ty3, "arg4" := Ty4 ) f :: AllArgsTuple -> r
and we have default values for some of those arguments, e.g.
type DefaultArgsTuple = ( "arg2" := Ty2, "arg3" := Ty3 ) f_defaults :: DefaultArgsTuple f_defaults = ( arg3 := val3 )
then we can create a corresponding function f_defaulting,
which allows user to only pass the remaining (required) arguments:
f_defaulting :: CheckedAdapt args DefaultArgsTuple AllArgsTuple => args -> r f_defaulting args = adapt args f_defaults
Synopsis
- class UncheckedAdapt args opt all => Adapt args opt all where
- adapt :: args -> opt -> all
- class UncheckedInject small big => Inject small big where
- inject :: small -> big -> big
- class UncheckedProject big small => Project big small where
- project :: big -> small
- newtype (lbl :: Symbol) := (a :: Type) where
- class UncheckedAdapt args opt all where
- uncheckedAdapt :: args -> opt -> all
- class UncheckedAdapt small big big => UncheckedInject small big where
- uncheckedInject :: small -> big -> big
- class UncheckedAdapt big big small => UncheckedProject big small where
- uncheckedProject :: big -> small
Converting between collections of fields
class UncheckedAdapt args opt all => Adapt args opt all where Source #
Methods
Arguments
| :: args | Provided arguments |
| -> opt | Default values of optional arguments |
| -> all | Combination of provided arguments and non-overridden defaults |
Create an adapter, to inject a smaller type into a larger one, providing defaults for optional values.
myAdapt :: ( "i" := Int, "f" := Float ) -> ( "f" := Float, "b" := Bool, "i" := Int ) myAdapt args = adapt args ( #b := False )
> myAdapt ( f := 17.1 ) > ( b = False, #i := 3 )
Here myAdapt re-arranges the arguments into the result,
passing in additional (default) values that are overriden
when they occur in the arguments.
Includes custom validation, e.g. to disallow duplicate arguments.
Use uncheckedAdapt to disable this validation
(you might get strange errors!).
Instances
| (UncheckedAdapt args opt all, CheckAdapt args opt all) => Adapt args opt all Source # | |
Defined in Data.Generic.Labels | |
class UncheckedInject small big => Inject small big where Source #
Methods
inject :: small -> big -> big Source #
Inject a smaller type into a larger one, overriding the fields in the larger type with those from the smaller type.
myInject
:: ( "i" := Int, "f" := Float )
-> ( "f" := Float, "b" := Bool, "i" := Int )
-> ( "f" := Float, "b" := Bool, "i" := Int )
myInject = inject
Here myInject overrides the fields of the second argument
with those provided in the first argument.
> myInject ( f := 17.1 ) ( b := False, #i := 22 ) > ( b := False, #i := 3 )
Includes custom validation, e.g. to disallow duplicate arguments.
Use uncheckedInject to disable this validation
(you might get strange errors!).
Instances
| (UncheckedInject small big, CheckInject small big) => Inject small big Source # | |
Defined in Data.Generic.Labels | |
class UncheckedProject big small => Project big small where Source #
Methods
project :: big -> small Source #
Project a smaller type out from a larger one, discarding the rest.
myProject :: ( "f" := Float, "b" := Bool, "i" := Int ) -> ( "i" := Int, "f" := Float ) myProject = project
Here myProject projects out a sub-component of the whole type,
in this case discarding the boolean while re-arranging the other fields.
> myProject ( b := False, #i := 3 ) > ( f := 17.1 )
Includes custom validation, e.g. to disallow duplicate arguments.
Use uncheckedProject to disable this validation
(you might get strange errors!).
Instances
| (UncheckedProject big small, CheckProject big small) => Project big small Source # | |
Defined in Data.Generic.Labels | |
Re-export of labelling functionality from Data.Label
newtype (lbl :: Symbol) := (a :: Type) infix 1 Source #
Bundled Patterns
| pattern (:=) :: Label lbl -> a -> lbl := a infix 1 | Add a With ( #bar := Just |
Instances
| GLens' (HasTotalLabelPSym lbl) (args :*: opts) all => GAdapt args opts (M1 m meta (Rec0 (lbl := all))) Source # | This instance is INCOHERENT because we assume that no type variable (say The end result is that, when we have both a built-in Haskell record field name as well as an explicit label, we prioritise the built-in record field name over the label. |
| (Generic args, Generic all, optFld ~ S1 ('MetaSel ('Just lbl) 'NoSourceUnpackedness 'NoSourceStrictness 'DecidedLazy) (Rec0 o), GAdapt (Rep args) optFld (Rep all)) => UncheckedAdapt args (lbl := o) all Source # | |
Defined in Data.Generic.Labels Methods uncheckedAdapt :: args -> (lbl := o) -> all Source # | |
| (KnownSymbol lbl, Show a) => Show (lbl := a) Source # | |
| (Generic opt, Generic all, argFld ~ S1 ('MetaSel ('Just lbl) 'NoSourceUnpackedness 'NoSourceStrictness 'DecidedLazy) (Rec0 a), GAdapt argFld (Rep opt) (Rep all)) => UncheckedAdapt (lbl := a) opt all Source # | |
Defined in Data.Generic.Labels Methods uncheckedAdapt :: (lbl := a) -> opt -> all Source # | |
| a ~ b => UncheckedAdapt (lbl := a) opt (lbl := b) Source # | |
Defined in Data.Generic.Labels Methods uncheckedAdapt :: (lbl := a) -> opt -> lbl := b Source # | |
| (Generic all, argFld ~ S1 ('MetaSel ('Just lbl1) 'NoSourceUnpackedness 'NoSourceStrictness 'DecidedLazy) (Rec0 a), optFld ~ S1 ('MetaSel ('Just lbl2) 'NoSourceUnpackedness 'NoSourceStrictness 'DecidedLazy) (Rec0 o), GAdapt argFld optFld (Rep all)) => UncheckedAdapt (lbl1 := a) (lbl2 := o) all Source # | |
Defined in Data.Generic.Labels Methods uncheckedAdapt :: (lbl1 := a) -> (lbl2 := o) -> all Source # | |
| (a ~ b, o ~ b) => UncheckedAdapt (lbl := a) (lbl := o) (lbl := b) Source # | |
Defined in Data.Generic.Labels | |
Unchecked functions (can behave unpredictably).
class UncheckedAdapt args opt all where Source #
Methods
uncheckedAdapt :: args -> opt -> all Source #
Create an adapter, without extra type-level validation.
Unchecked uses (e.g. presence of duplicate fields) can throw abstruse compile-time error messages or produce unexpected results at runtime.
Prefer using adapt whenever possible.
Instances
class UncheckedAdapt small big big => UncheckedInject small big where Source #
Methods
uncheckedInject :: small -> big -> big Source #
Inject a smaller type into a larger one, without extra type-level validation.
Unchecked uses (e.g. presence of duplicate fields) can throw abstruse compile-time error messages or produce unexpected results at runtime.
Prefer using inject whenever possible.
Instances
| UncheckedAdapt small big big => UncheckedInject small big Source # | |
Defined in Data.Generic.Labels Methods uncheckedInject :: small -> big -> big Source # | |
class UncheckedAdapt big big small => UncheckedProject big small where Source #
Methods
uncheckedProject :: big -> small Source #
Project a smaller type out from a larger one, without extra type-level validation.
Unchecked uses (e.g. presence of duplicate fields) can throw abstruse compile-time error messages or produce unexpected results at runtime.
Prefer using project whenever possible.
Instances
| UncheckedAdapt big big small => UncheckedProject big small Source # | |
Defined in Data.Generic.Labels Methods uncheckedProject :: big -> small Source # | |