Library
Module
Module type
Parameter
Class
Class type
Conformist is a library for creating and validating schemas. It provides run-time types without using any ppx. It can be used to validate incoming data and to translate it to static types for safe usage.
Let's start with an example. We have a static type that represents a user.
type gender = Male | Female | Other
type user = {
gender : gender;
email : string;
birthday : int * int * int;
nr_of_siblings : int;
comment : string option;
wants_premium : bool;
}
In order to create a conformist schema, we need a constructor that takes all the record fields and create a user
.
let user gender email birthday nr_of_siblings comment wants_premium
=
{
gender;
email;
birthday;
nr_of_siblings;
comment;
wants_premium;
}
Now we can create a schema.
let gender_decoder = function
| "male" -> Ok Male
| "female" -> Ok Female
| "other" -> Ok Other
| _ -> Error "Unknown gender provided"
let gender_encoder = function
| Male -> "male"
| Female -> "female"
| Other -> "other"
module C = Conformist
let user_schema =
C.make
C.Field.
[
C.custom gender_decoder gender_encoder "gender" ~meta:();
C.string "email";
C.date "birthday";
C.int ~default:0 "nr_of_siblings";
C.optional (C.string "comment");
C.bool "wants_premium");
]
user
Try to delete/swap lines of that list, to change the constructor or the user
type. The code doesn't compile anymore!
user_schema
showcases the creation of a custom type and optional types.
This is how you can decode a user given some input:
let user =
let input =
[
("gender", [ "male" ]);
("email", [ "test@example.com" ]);
("birthday", [ "2020-12-01" ]);
("nr_of_siblings", [ "3" ]);
("comment", [ "hello" ]);
("wants_premium", [ "true" ]);
]
in
C.decode Schema.user_schema input
Decoding doesn't validate the data, it just makes sure that the types are correct and translates strings to the correct static types.
We can validate data based on our validator per field.
let validation_errors =
let input =
[
("gender", [ "male" ]);
("email", [ "test@example.com" ]);
("birthday", [ "2020-12-01" ]);
("nr_of_siblings", [ "3" ]);
("comment", [ "hello" ]);
("wants_premium", [ "true" ]);
]
in
C.validate Schema.user_schema input
Note that if decoding of a field fails, validation fails as well since before a field is validated it gets decoded.
Every member of the list in the example is a field. Use the provided fold_left
to traverse the list of fiels. Helper functions are provided that operate on fields.
module Field : sig ... end
type 'a decoder = string -> ('a, string) result
A 'a decoder
tries to turn a string into a value of type 'a
. It returns a descriptive errors message upon failure.
A 'a validator
takes something of type 'a
and returns an error string if validation fails, None
if everything is ok
val custom :
'a decoder ->
'a encoder ->
?default:'a ->
?type_:string ->
?meta:'b ->
?validator:'a validator ->
string ->
('b, 'a) Field.t
Use custom decoder encoder ?default ?type_ ?meta ?validator field_name
to create a field with a custom type that is not supported out-of-the box. Provide a custom decoder
with a descriptive error message so conformist knows how to turn a string into your custom value.
A string representation of the static type_
can also be provided, by default the field_name
is taken.
A default
value can be provided.
Use optional ?meta field
to turn any field into an optional value. Note that the field must still be contained in the final input (when decoding or validating), but it can be an empty list or an empty string. If the data is not provided in the input, no validation logic is executed.
val bool :
?default:bool ->
?meta:'a ->
?msg:string ->
string ->
('a, bool) Field.t
bool ?default ?meta ?msg field_name
creates a field with field_name
some meta
data and a custom decode error message msg
that decodes to a boolean.
A default
value can be provided.
val float :
?default:float ->
?meta:'a ->
?msg:string ->
?validator:float validator ->
string ->
('a, float) Field.t
float ?meta ?msg ?validator field_name
creates a field that decodes to a float with field_name
some meta
data, a custom decode error message msg
and a validator
.
A default
value can be provided.
val int :
?default:int ->
?meta:'a ->
?msg:string ->
?validator:int validator ->
string ->
('a, int) Field.t
int ?meta ?msg ?validator field_name
creates a field that decodes to a int with field_name
some meta
data, a custom decode error message msg
and a validator
.
A default
value can be provided.
val string :
?default:string ->
?meta:'a ->
?validator:string validator ->
string ->
('a, string) Field.t
string ?meta ?validator field_name
creates a field that decodes to a string with field_name
some meta
data and a validator
. Note that this field does not need to be decoded, but it can still be validated.
A default
value can be provided.
val date :
?default:date ->
?meta:'a ->
?msg:string ->
?validator:(int * int * int) validator ->
string ->
('a, date) Field.t
string ?meta ?validator field_name
creates a field that decodes to a date with field_name
some meta
data and a validator
.
A default
value can be provided.
A schema is a list of fields. Input data can be decoded and validated using a schema.
val empty : ('a, unit, unit) t
empty
creates an empty schema.
val make : ('a, 'b, 'c) Field.list -> 'b -> ('a, 'b, 'c) t
make fields constructor
create a schema.
val fold_left :
f:('res -> 'meta Field.any_field -> 'res) ->
init:'res ->
('meta, 'args, 'ty) t ->
'res
fold_left ~f ~init schema
can be used to traverse the list of fields of schema
. Use the functions meta
, name
, validate
and optional
in f.
An empty validation_error
means that the schema is valid.
The input
represents unsafe data that needs to be validated and decoded. This is typically some user input or other data that needs to be sanitization.
decode schema input
tries to create a value of the static type 'ty
. Note that a successfully decoded value means that the strings contain the expected types, but no validation logic was executed.
val validate : ('meta, 'ctor, 'ty) t -> input -> validation_error
validate schema input
runs the field validators on decoded data. Note that a field that fails to decode will also fail validation, but a decoded field might still fail validation.