feat!: update methods to apply to a single provider

[florian-lefebvre]

unifont was extracted from @nuxt/fonts and inherited some of its APIs. I personally think it is too opinionated, from my experience using it in Astro.

I am aware this a breaking change but unifont is still in v0 so if downstream projects consume this package with ~ range, all should be fine.

This PR suggests a simpler API that still allows downstream consumers to replicate the previous behavior using the new unifont.providers API:

const unifont = await createUnifont([providers.google(), providers.fontsource()])

-const { fonts } = await unifont.resolveFont('Poppins')
+const fonts = []
+for (const provider of unifont.providers) {
+	const result = await unifont.resolveFont({ fontFamily: 'Poppins', provider })
+	if (result.fonts.length >= 0) {
+		fonts.push(...result.fonts)
+		break
+	}
+}

Changes are documented in the README

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 98.90%. Comparing base (207a1a2) to head (af01c6f).

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #287      +/-   ##
==========================================
- Coverage   98.91%   98.90%   -0.01%     
==========================================
  Files          11       11              
  Lines         460      457       -3     
  Branches      113      111       -2     
==========================================
- Hits          455      452       -3     
  Misses          5        5              

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[danielroe]

could you help me understand what you think is too opinionated here?

are you thinking of a non serial strategy?

[florian-lefebvre]

For me, unifont's purpose is to initialize all providers then be able to retrieve fonts data from them through a unified API. The opinionated bit I'd like to get rid of is the fact that you can a given method for all providers at once, that's something I don't do in Astro.

I think it's simpler and gives more control to users if they can call eg. resolveFont for one provider, and decide themsleves if they want to loop over all providers, or a subset etc.

Happy to discuss the API, maybe it can be updated but really the point of this PR is to remove the loop in each unifont method

[danielroe]

understood. what about an optional second argument which is a provider name?

without it, it searches all providers?

[florian-lefebvre]

This is the current API with providers as the 3rd argument isn't it? To be clear I know what I'm proposing in this PR is not required per say, it's possible to scope to one provider already. I guess the main argument against it is backward compat but since we're in v0 and with very few dependent packages, I thought that could be fine