Skip to content

Washing your code: divide and conquer, or merge and relax

You’re reading an excerpt of my upcoming book on clean code for frontend developers, “Washing your code.”

Get the book now!

Washing your code book cover

Knowing how to organize code into modules or functions, and when the right time is to introduce an abstraction instead of duplicating code, is an important skill. Writing generic code that others can effectively use is yet another skill. There are just as many reasons to split the code as there are to keep it together. In this chapter, we’ll discuss some of these reasons.

#Let abstractions grow

We, developers, hate to do the same work twice. DRY is a mantra for many. However, when we have two or three pieces of code that kind of do the same thing, it may be still too early to introduce an abstraction, no matter how tempting it may feel.

Info The Don’t repeat yourself (DRY) principle demands that “every piece of knowledge must have a single, unambiguous, authoritative representation within a system”, which is often interpreted as any code duplication is strictly verboten.

Live with the pain of code duplication for a while; maybe it’s not so bad in the end, and the code is actually not exactly the same. Some level of code duplication is healthy and allows us to iterate and evolve code faster without the fear of breaking something.

It’s also hard to come up with a good API when we only consider a couple of use cases.

Managing shared code in large projects with many developers and teams is difficult. New requirements for one team may not work for another Team and break their code, or we end up with an unmaintainable spaghetti monster with dozens of conditions.

Imagine Team A is adding a comment form to their page: a name, a message, and a submit button. Then, Team B needs a feedback form, so they find Team A’s component and try to reuse it. Then, Team A also wants an email field, but they don’t know that Team B uses their component, so they add a required email field and break the feature for Team B users. Then, Team B needs a phone number field, but they know that Team A is using the component without it, so they add an option to show a phone number field. A year later, the two teams hate each other for breaking each other’s code, and the component is full of conditions and is impossible to maintain. Both teams would save a lot of time and have healthier relationships with each other if they maintained separate components composed of lower-level shared components, like an input field or a button.

Tip It might be a good idea to forbid other teams from using our code unless it’s designed and marked as shared. The Dependency cruiser is a tool that could help set up such rules.

Sometimes, we have to roll back an abstraction. When we start adding conditions and options, we should ask ourselves: is it still a variation of the same thing or a new thing that should be separated? Adding too many conditions and parameters to a module can make the API hard to use and the code hard to maintain and test.

Duplication is cheaper and healthier than the wrong abstraction.

Info See Sandi Metz’s article The Wrong Abstraction for a great explanation.

The higher the level of the code, the longer we should wait before we abstract it. Low-level utility abstractions are much more obvious and stable than business logic.

#Size doesn’t always matter

Code reuse isn’t the only, or even most important, reason to extract a piece of code into a separate function or module.

Code length is often used as a metric for when we should split a module or a function, but size alone doesn’t make code hard to read or maintain.

Splitting a linear algorithm, even a long one, into several functions and then calling them one after another rarely makes the code more readable. Jumping between functions (and even more so — files) is harder than scrolling, and if we have to look into each function’s implementation to understand the code, then the abstraction wasn’t the right one.

Info Egon Elbre wrote a nice article on psychology of code readability.

Here’s an example, adapted from the Google Testing Blog:

function createPizza(order) {
  const pizza = new Pizza({
    base: order.size,
    sauce: order.sauce,
    cheese: 'Mozzarella'
  });
 
  if (order.kind === 'Veg') {
    pizza.toppings = vegToppings;
  } else if (order.kind === 'Meat') {
    pizza.toppings = meatToppings;
  }
 
  const oven = new Oven();
 
  if (oven.temp !== cookingTemp) {
    while (oven.temp < cookingTemp) {
      time.sleep(checkOvenInterval);
      oven.temp = getOvenTemp(oven);
    }
  }
 
  if (!pizza.baked) {
    oven.insert(pizza);
    time.sleep(cookTime);
    oven.remove(pizza);
    pizza.baked = true;
  }
 
  const box = new Box();
  pizza.boxed = box.putIn(pizza);
  pizza.sliced = box.slicePizza(order.size);
  pizza.ready = box.close();
 
  return pizza;
}

I have so many questions about the API of the Pizza class, but let’s see what improvements the authors suggest:

function prepare(order) {
  const pizza = new Pizza({
    base: order.size,
    sauce: order.sauce,
    cheese: 'Mozzarella'
  });
  addToppings(pizza, order.kind);
  return pizza;
}
 
function addToppings(pizza, kind) {
  if (kind === 'Veg') {
    pizza.toppings = vegToppings;
  } else if (kind === 'Meat') {
    pizza.toppings = meatToppings;
  }
}
 
function bake(pizza) {
  const oven = new Oven();
  heatOven(oven);
  bakePizza(pizza, oven);
}
 
function heatOven(oven) {
  if (oven.temp !== cookingTemp) {
    while (oven.temp < cookingTemp) {
      time.sleep(checkOvenInterval);
      oven.temp = getOvenTemp(oven);
    }
  }
}
 
function bakePizza(pizza, oven) {
  if (!pizza.baked) {
    oven.insert(pizza);
    time.sleep(cookTime);
    oven.remove(pizza);
    pizza.baked = true;
  }
}
 
function pack(pizza) {
  const box = new Box();
  pizza.boxed = box.putIn(pizza);
  pizza.sliced = box.slicePizza(pizza.size);
  pizza.ready = box.close();
}
 
function createPizza(order) {
  const pizza = prepare(order);
  bake(pizza);
  pack(pizza);
  return pizza;
}

What was already complex and convoluted is now even more complex and convoluted, and half of the code is just function calls. This doesn’t make the code any easier to understand, but it does make it almost impossible to work with. The article doesn’t show the complete code of the refactored version, perhaps to make the point more compelling.

Pierre “catwell” Chapuis suggests in his blog post to add comments instead of new functions:

function createPizza(order) {
  // Prepare pizza
  const pizza = new Pizza({
    base: order.size,
    sauce: order.sauce,
    cheese: 'Mozzarella'
  });
 
  // Add toppings
  if (order.kind == 'Veg') {
    pizza.toppings = vegToppings;
  } else if (order.kind == 'Meat') {
    pizza.toppings = meatToppings;
  }
 
  const oven = new Oven();
 
  if (oven.temp !== cookingTemp) {
    // Heat oven
    while (oven.temp < cookingTemp) {
      time.sleep(checkOvenInterval);
      oven.temp = getOvenTemp(oven);
    }
  }
 
  if (!pizza.baked) {
    // Bake pizza
    oven.insert(pizza);
    time.sleep(cookTime);
    oven.remove(pizza);
    pizza.baked = true;
  }
 
  // Box and slice
  const box = new Box();
  pizza.boxed = box.putIn(pizza);
  pizza.sliced = box.slicePizza(order.size);
  pizza.ready = box.close();
 
  return pizza;
}

This is already much better than the split version. An even better solution would be improving the APIs and making the code more clear. Pierre suggests that preheating the oven shouldn’t be part of the createPizza() function (and baking many pizzas myself, I totally agree!) because in real life the oven is already there and probably already hot from the previous pizza. Pierre also suggests that the function should return the box, not the pizza, because in the original code, the box kind of disappears after all the slicing and packaging magic, and we end up with the sliced pizza in our hands.

There are many ways to cook a pizza, just as there are many ways to code a problem. The result may look the same, but some solutions are easier to understand, modify, reuse, and delete than others.

Naming can also be a problem too when all the extracted functions are parts of the same algorithm. We need to invent names that are clearer than the code and shorter than comments — not an easy task.

Info We talk about commenting code in the Avoid comments chapter, and about naming in the Naming is hard chapter.

You probably won’t find many small functions in my code. In my experience, the most useful reasons to split code are change frequency and change reason.

#Separate code that changes often

Let’s start with change frequency. Business logic changes much more often than utility functions. It makes sense to separate often-changing code from code that is very stable.

The comment form we discussed earlier in this chapter is an example of the former; a function that converts camelCase strings to kebab-case is an example of the latter. The comment form is likely to change and diverge over time when new business requirements arise; the case conversion function is unlikely to change at all and it’s safe to reuse in many places.

Imagine that we’re making a nice-looking table to display some data. We may think we’ll never need this table design again, so we decide to keep all the code for the table in a single module.

Next sprint, we get a task to add another column to the table, so we copy the code of an existing column and change a few lines there. Next sprint, we need to add another table with the same design. Next sprint, we need to change the design of the tables…

Our table module has at least three reasons to change, or responsibilities:

  • new business requirements, like a new table column;
  • UI or behavior changes, like adding sorting or column resizing;
  • design changes, like replacing borders with striped row backgrounds.

This makes the module harder to understand and harder to change. Presentational code adds a lot of verbosity, making it harder to understand the business logic. To make a change in any of the responsibilities, we need to read and modify more code. This makes it harder and slower to iterate on either.

Having a generic table as a separate module solves this problem. Now, to add another column to a table, we only need to understand and modify one of the two modules. We don’t need to know anything about the generic table module except its public API. To change the design of all tables, we only need to change the generic table module’s code and likely don’t need to touch individual tables at all.

However, depending on the complexity of the problem, it’s okay, and often better, to start with a monolithic approach and extract an abstraction later.

Even code reuse can be a valid reason to separate code: if we use some component on one page, we’ll likely need it on another page soon.

#Keep together code that changes at the same time

It might be tempting to extract every function into its own module. However, it has downsides too:

  • Other developers may think that they can reuse the function somewhere else, but in reality, this function is likely not generic or tested enough to be reused.
  • Creating, importing, and switching between multiple files creates unnecessary overhead when the function is only used in one place.
  • Such functions often stay in the codebase long after the code that used them is gone.

I prefer to keep small functions that are used only in one module at the beginning of the module. This way, we don’t need to import them to use in the same module, but reusing them somewhere else would be awkward.

function FormattedAddress({ address, city, country, district, zip }) {
  return [address, zip, district, city, country]
    .filter(Boolean)
    .join(', ');
}
 
function getMapLink({ name, address, city, country, zip }) {
  return `https://www.google.com/maps/?q=${encodeURIComponent(
    [name, address, zip, city, country].filter(Boolean).join(', ')
  )}`;
}
 
function ShopsPage({ url, title, shops }) {
  return (
    <PageWithTitle url={url} title={title}>
      <Stack as="ul" gap="l">
        {shops.map(shop => (
          <Stack key={shop.name} as="li" gap="m">
            <Heading level={2}>
              <Link href={shop.url}>{shop.name}</Link>
            </Heading>
            {shop.address && (
              <Text variant="small">
                <Link href={getMapLink(shop)}>
                  <FormattedAddress {...shop} />
                </Link>
              </Text>
            )}
          </Stack>
        ))}
      </Stack>
    </PageWithTitle>
  );
}

In the code above, we have a component (FormattedAddress) and a function (getMapLink()) that are only used in this module, so they’re defined at the top of the file.

If we need to test these functions (and we should!), we can export them from the module and test them together with the main function of the module.

The same applies to functions that are intended to be used only together with a certain function or component. Keeping them in the same module makes it clearer that all functions belong together and makes these functions more discoverable.

Another benefit is that when we delete a module, we automatically delete its dependencies. Code in shared modules often stays in the codebase forever because it’s hard to know if it’s still used (though TypeScript makes this easier).

Info Such modules are sometimes called deep modules: a relatively large modules that encapsulate complex problems but has a simple APIs. The opposite of deep modules are shallow modules: many small modules that need to interact with each other.

If we often have to change several modules or functions at the same time, it might be better to merge them into a single module or function. This approach is sometimes called colocation.

Here are a couple of examples of colocation:

  • React components: keeping everything a component needs in the same file, including markup (JSX), styles (CSS in JS), and logic, rather than separating each into its own file, likely in a separate folder.
  • Tests: keeping tests next to the module file rather than in a separate folder.
  • Ducks convention for Redux: keep related actions, action creators, and reducers in the same file rather than having them in three files in separate folders.

Here’s how the file tree changes with colocation:

SeparatedColocated
React components
src/components/Button.tsxsrc/components/Button.tsx
styles/Button.css
Tests
src/util/formatDate.tssrc/util/formatDate.ts
tests/formatDate.tssrc/util/formatDate.test.ts
Ducks
src/actions/feature.jssrc/ducks/feature.js
src/actionCreators/feature.js
src/reducers/feature.js

Info To learn more about colocation, read Kent C. Dodds’s article.

A common complaint about colocation is that it makes components too large. In such cases, it’s better to extract some parts into their own components, along with the markup, styles, and logic.

The idea of colocation also conflicts with separation of concerns — an outdated idea that led web developers to keep HTML, CSS, and JavaScript in separate files (and often in separate parts of the file tree) for too long, forcing us edit three files at the same time to make even the most basic changes to web pages.

Info The change reason is also known as the single responsibility principle, which states that “every module, class, or function should have responsibility over a single part of the functionality provided by the software, and that responsibility should be entirely encapsulated by the class.”

#Sweep that ugly code under the rug

Sometimes, we have to work with an API that’s especially difficult to use or prone to errors. For example, it requires several steps in a specific order or calling a function with multiple parameters that are always the same. This is a good reason to create a utility function to make sure we always do it right. As a bonus, we can now write tests for this piece of code.

String manipulations — such as URLs, filenames, case conversion, or formatting — are good candidates for abstraction. Most likely, there’s already a library for what we’re trying to do.

Consider this example:

const file = 'pizza.jpg';
const prefix = file.slice(0, -4);
// → 'pizza'

It takes some time to realize that this code removes the file extension and returns the base name. Not only is it unnecessary and hard to read, but it also assumes the extension is always three characters, which might not be the case.

Let’s rewrite it using a library, the built-in Node.js’ path module:

const file = 'pizza.jpg';
const prefix = path.parse(file).name;
// → 'pizza'

Now, it’s clear what’s happening, there are no magic numbers, and it works with file extensions of any length.

Other candidates for abstraction include dates, device capabilities, forms, data validation, internationalization, and more. I recommend looking for an existing library before writing a new utility function. We often underestimate the complexity of seemingly simple functions.

Here are a few examples of such libraries:

  • Lodash: utility functions of all kinds.
  • Date-fns: functions to work with dates, such as parsing, manipulation, and formatting.
  • Zod: schema validation for TypeScript.

#Bless the inline refactoring!

Sometimes, we get carried away and create abstractions that neither simplify the code nor make it shorter:

// my_feature_util.js
const noop = () => {};
 
export const Utility = {
  noop
  // Many more functions…
};
 
// MyComponent.js
function MyComponent({ onClick }) {
  return <button onClick={onClick}>Hola!</button>;
}
 
MyComponent.defaultProps = {
  onClick: Utility.noop
};

Another example:

const findByReference = (wrapper, reference) =>
  wrapper.find(reference);
 
const favoriteTaco = findByReference(
  ['Al pastor', 'Cochinita pibil', 'Barbacoa'],
  x => x === 'Cochinita pibil'
);
 
// → 'Cochinita pibil'

The best thing we can do in such cases is to apply the almighty inline refactoring: replace each function call with its body. No abstraction, no problem.

The first example becomes:

function MyComponent({ onClick }) {
  return <button onClick={onClick}>Hola!</button>;
}
 
MyComponent.defaultProps = {
  onClick: () => {}
};

And the second example becomes:

const favoriteTaco = [
  'Al pastor',
  'Cochinita pibil',
  'Barbacoa'
].find(x => x === 'Cochinita pibil');
 
// → 'Cochinita pibil'

The result is not just shorter and more readable; now readers don’t need to guess what these functions do, as we now use JavaScript native functions and features without home-baked abstractions.

In many cases, a bit of repetition is good. Consider this example:

const baseSpacing = 8;
const spacing = {
  tiny: baseSpacing / 2,
  small: baseSpacing,
  medium: baseSpacing * 2,
  large: baseSpacing * 3,
  xlarge: baseSpacing * 4,
  xxlarge: baseSpacing * 5
};

It looks perfectly fine and won’t raise any questions during code review. However, when we try to use these values, autocompletion only shows number instead of the actual values (see an illustration). This makes it harder to choose the right value.

Autocompletion with calculated values

We could inline the baseSpacing constant:

const spacing = {
  tiny: 4,
  small: 8,
  medium: 16,
  large: 24,
  xlarge: 32,
  xxlarge: 40
};

Now, we have less code, it’s just as easy to understand, and autocompletion shows the actual values (see the illustration). And I don’t think this code will change often — probably never.

Autocompletion with literal values

#Separate “what” and “how”

Consider this excerpt from a form validation function:

function validate(values) {
  const errors = {};
 
  if (!values.name || (values.name && values.name.trim() === '')) {
    errors.name = 'Name is required';
  }
 
  if (!values.login || (values.login && values.login.trim() === '')) {
    errors.login = 'Login is required';
  }
 
  if (values.login && values.login.indexOf(' ') > 0) {
    errors.login = 'No spaces are allowed in login';
  }
 
  // This goes on and on for a dozen of other fields…
 
  return errors;
}

It’s quite difficult to grasp what’s going on here: validation logic is mixed with error messages, many checks are repeated…

We can split this function into several pieces, each responsible for one thing only:

  • a list of validations for a particular form;
  • a collection of validation functions, such as isEmail();
  • a function that validates all form values using a list of validations.

We can describe the validations declaratively as an array:

const validations = [
  {
    field: 'name',
    validation: hasStringValue,
    message: 'Name is required'
  },
  {
    field: 'login',
    validation: hasStringValue,
    message: 'Login is required'
  },
  {
    field: 'login',
    validation: hasNoSpaces,
    message: 'No spaces are allowed in login'
  }
  // All other validations…
];

Each validation function and the function that runs validations are pretty generic, so we can either abstract them or use a third-party library.

Now, we can add validation for any form by describing which fields need which validations and which error to show when a certain check fails.

Info See the Avoid conditions chapter for the complete code and a more detailed explanation of this example.

I call this process separation of “what” and “how”:

  • the “what” is the data — the list of validations for a particular form;
  • the “how” is the algorithms — the validation functions and the validation runner function.

The benefits are:

  • Readability often, we can define the “what” declaratively, using basic data structures such as arrays and objects.
  • Maintainability we change the “what” more often than the “how”, and now they are separated. We can import the “what” from a file, such as JSON, or load it from a database, making updates possible without code changes, or allowing non-developers to do them.
  • Reusability often, the “how” is generic, and we can reuse it, or even import it from a third-party library.
  • Testability each validation and the validation runner function are isolated, and we can test them separately.

#Avoid monster utility files

Many projects have a file called utils.js, helpers.js, or misc.js where developers throw in utility functions when they can’t find a better place for them. Often, these functions are never reused anywhere else and stay in the utility file forever, so it keeps growing. That’s how monster utility files are born.

Monster utility files have several issues:

  • Poor discoverability since all functions are in the same file, we can’t use the fuzzy file opener in our code editor to find them.
  • They may outlive their callers often such functions are never reused again and stay in the codebase, even after the code that was using them is removed.
  • Not generic enough such functions are often made for a single use case and won’t cover other use cases.

These are my rules of thumb:

  • If the function is small and used only once, keep it in the same module where it’s used.
  • If the function is long or used more than once, put it in a separate file inside util, shared, or helpers folder.
  • If we want more organization, instead of creating files like utils/validators.js, we can group related functions (each in its own file) into a folder: utils/validators/isEmail.js.

#Avoid default exports

JavaScript modules have two types of exports. The first is named exports:

// button.js
export function Button() {
  /* … */
}

Which can be imported like this:

import { Button } from './button';

And the second is default exports:

// button.js
export default function Button() {
  /* … */
}

Which can be imported like this:

import Button from './button';

I don’t really see any advantages to default exports, but they have several issues:

  • Poor refactoring renaming a module with a default export often leaves existing imports unchanged. This doesn’t happen with named exports, where all imports are updated after renaming a function.
  • Inconsistency default-exported modules can be imported using any name, which reduces the consistency and greppability of the codebase. Named exports can also be imported using a different name using the as keyword to avoid naming conflicts, but it’s more explicit and is rarely done by accident.

Info We talk more about greppability in the Write greppable code section of the Other techniques chapter.

Unfortunately, some third-party APIs, such as React.lazy() require default exports, but for all other cases, I stick to named exports.

#Avoid barrel files

A barrel file is a module (usually named index.js or index.ts) that reexports a bunch of other modules:

// components/index.js
export { Box } from './Box';
export { Button } from './Button';
export { Link } from './Link';

The main advantage is cleaner imports. Instead of importing each module individually:

import { Box } from '../components/Box';
import { Button } from '../components/Button';
import { Link } from '../components/Link';

We can import all components from a barrel file:

import { Box, Button, Link } from '../components';

However, barrel files have several issues:

  • Maintenance cost we need to add an export of each new component in a barrel file, along with additional items such as types of utility functions.
  • Performance cost setting up tree shaking is complex, and barrel files often lead to increased bundle size or runtime costs. This can also slow down hot reload, unit tests, and linters.
  • Circular imports importing from a barrel file can cause a circular import when both modules are imported from the same barrel files (for example, the Button component imports the Box component).
  • Developer experience navigation to function definition navigates to the barrel file instead of the function’s source code; and autoimport can be confused whether to import from a barrel file instead of a source file.

Info TkDodo explains the drawbacks of barrel files in great detail.

The benefits of barrel files are too minor to justify their use, so I recommend avoiding them.

One type of barrel files I especially dislike is those that export a single component just to allow importing it as ./components/button instead of ./components/button/button.

#Stay hydrated

To troll the DRYers (developers who never repeat their code), someone coined another term: WET, write everything twice, or we enjoy typing, suggesting we should duplicate code at least twice until we replace it with an abstraction. It is a joke, and I don’t fully agree with the idea (sometimes it’s okay to duplicate some code more than twice), but it’s a good reminder that all good things are best in moderation.

Consider this example:

const stories = {
  YOUR_RECIPES: 'page--your-recipes',
  ALL_RECIPES: 'page--all-recipes',
  CUISINES: 'page--cuisines',
  RECIPE: 'page--recipe'
};
 
const testIds = {
  ADD_TO_FAVORITES: 'add-to-favs-button',
  QR_CLOSE: 'qr-close-button',
  QR_CODE: 'qr-code',
  MOBILE_CTA: 'transfer-button'
};
 
const copyTesters = {
  titleRecipe: /Cochinita Pibil Tacos/,
  titleYourRecipes: /Your favorite recipes on a single page/,
  addedToFavorites: /In favorites/
  // Many more lines…
};
 
it('your recipes', () => {
  visitStory(stories.RECIPE);
 
  cy.findByText(copyTesters.titleRecipe).should('exist');
 
  cy.findByTestId(testIds.ADD_TO_FAVORITES).click();
  cy.findByText(copyTesters.addedToFavorites).should('exist');
 
  // Lots of lines in similar style…
});

This is an extreme example of code DRYing, which doesn’t make the code more readable or maintainable, especially when most of these constants are used only once. Seeing variable names instead of actual strings here is unhelpful.

Let’s inline all these extra variables. (Unfortunately, inline refactoring in Visual Studio Code doesn’t support inlining object properties, so we have to do it manually.)

it('your recipes', () => {
  visitStory('page--recipe');
 
  cy.findByText(/Cochinita Pibil Tacos/).should('exist');
 
  cy.findByTestId('add-to-favs-button').click();
  cy.findByText(/In favorites/).should('exist');
 
  // Lots of lines in similar style…
});

Now, we have significantly less code, and it’s easier to understand what’s going on and easier to update or delete tests.

I’ve encountered so many hopeless abstractions in tests. For example, this pattern is very common:

let wrapper;
beforeEach(() => {
  wrapper = mount(<Pony color="pink" />);
});
 
test('pony has pink tail', () => {
  expect(wrapper.find('.tail').prop('value')).toBe('pink');
});
 
// More tests that use `wrapper`…

This pattern tries to avoid repeating mount(...) calls in each test case, but it makes tests more confusing than they need to be. Let’s inline the mount() calls:

test('pony has pink tail', () => {
  const wrapper = mount(<Pony color="pink" />);
  expect(wrapper.find('.tail').prop('value')).toBe('pink');
});
 
// More tests that use `wrapper`…

Additionally, the beforeEach pattern works only when we want to initialize each test case with the same values, which is rarely the case:

test('pony has pink tail', () => {
  const wrapper = mount(<Pony color="pink" />);
  expect(wrapper.find('.tail').prop('value')).toBe('pink');
});
 
test('pony can breath fire', () => {
  const wrapper = mount(<Pony color="pink" breathFire />);
  expect(wrapper.find('.fire')).toBeVisible();
});

To avoid some duplication when testing React components, I often add a defaultProps object and spread it inside each test case:

const defaultProps = { color: 'pink' };
 
test('pony has pink tail', () => {
  const wrapper = mount(<Pony {...defaultProps} />);
  expect(wrapper.find('.tail').prop('value')).toBe('pink');
});
 
test('pony can breath fire', () => {
  const wrapper = mount(<Pony {...defaultProps} breathFire />);
  expect(wrapper.find('.fire')).toBeVisible();
});

This way, we don’t have too much duplication, but at the same time, each test case is isolated and readable. The difference between test cases is now clearer because it’s easier to see the unique props of each test case.

Here’s a more extreme variation of the same problem:

let css;
let res;
 
beforeEach(() => {
  css = '';
  res = '';
});
 
test('works with basic selectors', () => {
  css = 'div\n{}';
  res = 'div\n';
  expect(getSelector(css)).toBe(res);
});
 
test('works with lobotomized owl selector', () => {
  css = '.stack > * + *\n{}';
  res = '.stack > * + *\n';
  expect(getSelector(css)).toBe(res);
});
 
// More tests that use `css` and `res`…

We can inline the beforeEach() function the same way we did in the previous example:

test('works with basic selectors', () => {
  const css = 'div\n{}';
  const expected = 'div\n';
  expect(getSelector(css)).toBe(expected);
});
 
test('works with lobotomized owl selector', () => {
  const css = '.stack > * + *\n{}';
  const expected = '.stack > * + *\n';
  expect(getSelector(css)).toBe(expected);
});
 
// More tests that use `css` and `res`…

I’d go even further and use test.each() method because we run the same test with a bunch of different inputs:

test.each([
  ['div\n{}', 'div\n'],
  ['.stack > * + *\n{}', '.stack > * + *\n']
  // More inputs…
])('selector: %s', (css, expected) => {
  expect(getSelector(css)).toBe(expected);
});

Now, we’ve gathered all the test inputs with their expected results in one place, making it easier to add new test cases.

Info Check out my Jest and Vitest cheat sheets.


The biggest challenge with abstractions is finding a balance between being too rigid and too flexible, and knowing when to start abstracting things and when to stop. It’s often worth waiting to see if we really need to abstract something — many times, it’s better not to.

It’s nice to have a global button component, but if it’s too flexible and has a dozen boolean props to switch between different variations, it will be difficult to use. However, if it’s too rigid, developers will create their own button components instead of using a shared one.

We should be vigilant about letting others reuse our code. Too often, this creates tight coupling between parts of the codebase that should be independent, slowing down development and leading to bugs.

Start thinking about:

  • Colocating related code in the same file or folder to make it easier to change, move, or delete.
  • Before adding another option to an abstraction, think whether this new use case truly belongs there.
  • Before merging several pieces of code that look similar, think whether they actually solve the same problems or just happened to look the same.
  • Before making tests DRY, think whether it would make them more readable and maintainable, or a bit of code duplication isn’t an issue.

Read other sample chapters of the book: