mirror of
https://github.com/typeable/validationt.git
synced 2024-11-25 14:40:35 +03:00
76 lines
2.6 KiB
Markdown
76 lines
2.6 KiB
Markdown
![Travis CI Badge](https://img.shields.io/travis/typeable/validationt)
|
|
|
|
# ValidationT
|
|
|
|
A very small validation data library.
|
|
|
|
## Usage
|
|
|
|
Suppose you want to validate some data from the user. Say, that the password adheres to some rules.
|
|
|
|
You might do something like this:
|
|
|
|
```haskell
|
|
validatePassword :: Monad m => String -> ValidationT [String] m ()
|
|
validatePassword password = do
|
|
vLength password
|
|
vAlpha password
|
|
vNum password
|
|
where
|
|
vLength p = when (length p < 8)
|
|
$ vWarning ["The password should be at least 8 characters long."]
|
|
vAlpha p = unless (any isAlpha p)
|
|
$ vWarning ["The password should contain at least one alphabetic character."]
|
|
vNum p = unless (any isDigit p)
|
|
$ vWarning ["The password should contain at least one numeric character."]
|
|
```
|
|
|
|
`ValidationT e m a` essentially just gathers the errors thrown by `vWarning`.
|
|
|
|
```haskell
|
|
vWarning :: (Monad m, Monoid e) => e -> ValidationT e m ()
|
|
```
|
|
|
|
`vWarning` `mappend`s the given `e` to the already collected errors. This is why the warnings are contained inside a list.
|
|
|
|
There is also `vError`. The only difference between `vWarning` and `vError` is that `vError` stops further execution (and further collection of errors and warnings).
|
|
|
|
You would use the validation like this:
|
|
|
|
```haskell
|
|
main :: IO ()
|
|
main = do
|
|
password <- getLine
|
|
result <- runValidationTEither . validatePassword $ password
|
|
putStrLn $ case result of
|
|
Left errs -> unlines err
|
|
Right () -> "You are fine."
|
|
```
|
|
|
|
You could, of course, do more complicated things like use `vWarningL` and `vErrorL` to add an error to a `mempty` structure, which gets `mappend`ed with other errors.
|
|
|
|
The library comes with a `MonoidMap` `newtype` wrapper around `Map`, which `mappend`s the values themselves on conflict. This can be useful if you have a multiple points of failure and you want to distinguich between them -- validating a username and a password for example:
|
|
|
|
```haskell
|
|
data Piece
|
|
= Password
|
|
| UserName
|
|
deriving (Eq, Show, Ord)
|
|
|
|
validatePassword :: Monad m => String -> ValidationT (MonoidMap Piece [String]) m ()
|
|
validatePassword password = do
|
|
vLength password
|
|
vAlpha password
|
|
vNum password
|
|
where
|
|
warning = mmSingleton UserName
|
|
vLength p = when (length p < 8)
|
|
$ warning ["The password should be at least 8 characters long."]
|
|
vAlpha p = unless (any isAlpha p)
|
|
$ warning ["The password should contain at least one alphabetic character."]
|
|
vNum p = unless (any isDigit p)
|
|
$ warning ["The password should contain at least one numeric character."]
|
|
```
|
|
|
|
(`mmSingleton` is a covenience initializer for `MonoidMap`.)
|