// import "../../lab_modules/033";
The Practical
I find many of the tutorials on fp-ts to be a bit too low level. They are great for learning the fundamentals, but I wanted to see some more practical examples. I've been using fp-ts in production for a while now, and I wanted to explore some commone use cases that I've encountered.
Before jumping into the examples, I want to mention another library that I found to be very useful. io-ts is a library for runtime type checking. It is a great companion to fp-ts, and I will be using it in some of the examples below.
DeepEqual of Nested Objects
Sometimes you need to compare two objects to see if they are equal. This is
pretty easy to do with primitive values, but it gets a bit more complicated when
you have nested objects. You can use JSON.stringify to compare the objects,
but this is not very efficient. It is also not very flexible, since it will
compare all of the properties on the object. What if you only want to compare
some of the properties?
import * as t from "io-ts";
import * as EQ from "io-ts/Eq";
const Filter = t.type({
types: t.array(t.string),
range: t.tuple([t.number, t.number]),
boolean: t.boolean,
user: t.type({
name: t.string,
id: t.number,
}),
});
type Filter = t.TypeOf<typeof Filter>;
const FilterEq = EQ.struct<Filter>({
types: EQ.array(EQ.string),
range: EQ.tuple(EQ.number, EQ.number),
boolean: EQ.boolean,
user: EQ.struct({
name: EQ.string,
id: EQ.number,
}),
});
const isEqual = (a: Filter, b: Filter) => FilterEq.equals(a, b);
const settings: Filter = {
types: ["a", "b"],
range: [1, 2],
boolean: true,
user: {
name: "John",
id: 1,
},
};
console.log(isEqual(settings, settings)); // Output: true
console.log(isEqual(settings, { ...settings, user: { name: "Fred", id: 2 } })); // Output: falseElegant IO Layer (Error Handling)
This is borrowed heavily from fp-ts and Even More Beautiful API Calls (w/ sum types!) but also expands on it a bit. This pattern is great for handeling APIs that are less trustworthy. It allows you to handle errors in a very elegant way, without having to crash the application.
import * as A from "fp-ts/lib/Array";
import * as O from "fp-ts/lib/Option";
import * as T from "fp-ts/Task";
import * as TE from "fp-ts/TaskEither";
import * as t from "io-ts";
import { pipe } from "fp-ts/lib/function";
import { failure } from "io-ts/PathReporter";
const ApiData = t.type({
status: t.string,
value: t.type({
id: t.number,
name: t.string,
}),
});
type ApiData = t.TypeOf<typeof ApiData>;
interface NetworkError {
type: "NetworkError";
message: string;
}
interface ParseError {
type: "ParseError";
errors: t.Errors;
}
type AppError = NetworkError | ParseError;
const handleErrors = (appError: AppError): T.Task<string> => {
switch (appError.type) {
case "NetworkError":
return T.of(`Network error: ${appError.message}`);
case "ParseError":
return pipe(
appError.errors,
failure,
A.head,
O.getOrElse(() => "Unknown error"),
T.of,
);
}
};
const getFromUrl = (url: string) =>
pipe(
TE.tryCatch(
() => fetch(url),
(e) => ({ type: "NetworkError", message: String(e) }),
),
TE.chain((x) => TE.of(x.json())),
TE.mapLeft(({ message }): AppError => ({ type: "NetworkError", message })),
TE.chain((json) =>
pipe(
ApiData.decode(json),
E.mapLeft((errors): AppError => ({ type: "ParseError", errors })),
TE.fromEither,
),
),
TE.fold(handleErrors, (data) => T.of(data.value.name)),
);
getFromUrl("https://api.chucknorris.io/jokes/random")().then(console.log); // Output: "Invalid value undefined supplied to : { status: string, value: { id: number, name: string } }/status: string"Branching Render Logic
This pattern is very useful in React applications. It allows you to branch your render logic based on the type of data that you have. This is very useful when you have a loading or error state that you want to handle differently.
import * as E from "fp-ts/lib/Either";
import * as IO from "fp-ts/lib/IO";
import * as O from "fp-ts/lib/Option";
import { pipe } from "fp-ts/lib/function";
export const ConditionalRender: FC<{ value?: number }> = ({ value }) => {
useEffect(() => {
return pipe(
value,
O.fromPredicate((value) => value == 10),
O.fold(
() => IO.of(undefined), // no side effect
() => IO.of(console.log("The value is very important!")), // conditional side effect
),
);
}, [value]);
return pipe(
value,
O.fromNullable,
O.fold(
() => <LoadingCard />, // Output: Loading UI Card
(value) =>
pipe(
value,
E.fromPredicate(
(value) => value !== 0, // Addition validation
() => "Error: Value is not 0",
),
E.map((value) =>
value > 0 ? `Positive (${value})` : `Negative (${value})`, // Transform value
),
E.fold(
(error) => <ErrorCard error={error} />, // Output: Error UI Card
(success) => <SuccessCard data={success} />, // Success UI Card
),
),
),
);
};Responsive Values
When working with repsonsiveness it can sometimes be nessisary to conditionally return values based on the screen size. This pattern allows you to do that in a very elegant way. While its not nessisary to use fp-ts for this, it does make it use of functional programming concepts such as currying and composition.
export const foldScreenWidth =
<T>(
xs: (width: number) => T,
sm?: (width: number) => T,
md?: (width: number) => T,
lg?: (width: number) => T,
) =>
(width: number) => {
switch (true) {
case width > 1200 && lg !== undefined:
return lg!(width);
case width > 992 && md !== undefined:
return md!(width);
case width > 768 && sm !== undefined:
return sm!(width);
default:
return xs(width);
}
};
const responsiveFontSize = foldScreenWidth(
() => `1rem`, // default for xs screen
() => `1.25rem`, // sm screen
undefined, // md screen
() => `2rem`, // lg screen
);
console.log(responsiveFontSize(1000)); // Output: "1.25rem"This can also be done using ADTs which is a bit more elegant, but also a bit more complicated. I've only used an ADT abstraction from morphic-ts but dont want to add that dependency to this project (right now).