label-pr.ts

/**
 * Applies size and semantic labels to a pull request, from the files it touches and its title.
 *
 * Loaded by the "Apply size and semantic labels" step of `pull-request.yml`, which hands it the
 * authenticated client and the event context that `actions/github-script` puts in scope.
 */

/** A file in the pull request, as `pulls.listFiles` returns it. */
interface PullRequestFile {
	readonly filename: string;
	readonly additions: number;
	readonly deletions: number;
}

interface PullRequestRef {
	readonly owner: string;
	readonly repo: string;
	readonly pull_number: number;
	readonly per_page: number;
}

interface IssueRef {
	readonly owner: string;
	readonly repo: string;
	readonly issue_number: number;
}

type ListFiles = (params: PullRequestRef) => Promise<{ data: PullRequestFile[] }>;

/** The slice of the Octokit client this script uses. */
interface GitHubApi {
	readonly paginate: (endpoint: ListFiles, params: PullRequestRef) => Promise<PullRequestFile[]>;
	readonly rest: {
		readonly pulls: { readonly listFiles: ListFiles };
		readonly issues: {
			readonly listLabelsOnIssue: (
				params: IssueRef,
			) => Promise<{ data: { name: string }[] }>;
			readonly addLabels: (params: IssueRef & { labels: string[] }) => Promise<unknown>;
			readonly removeLabel: (params: IssueRef & { name: string }) => Promise<unknown>;
		};
	};
}

/** The slice of `actions/github-script`'s `context` this script uses. */
interface GitHubContext {
	readonly repo: { readonly owner: string; readonly repo: string };
	readonly issue: { readonly number: number };
	readonly payload: { readonly pull_request?: { readonly title?: string } };
}

/** Generated or vendored files, which say nothing about how much work a pull request is to review. */
const GENERATED_PATTERNS = [
	// Generated code patterns
	/\.gen\.(ts|js|tsx|jsx)$/, // *.gen.ts files
	/\.generated\.(ts|js|tsx|jsx)$/, // *.generated.ts files

	// Webapp generated (entire api directory is generated by @hey-api/openapi-ts)
	/^webapp\/src\/api\//, // All files in webapp/src/api/
	/^webapp\/src\/routeTree\.gen\.ts$/, // TanStack Router generated

	// Server generated
	/^server\/openapi\.yaml$/, // Generated OpenAPI spec

	// Docs generated
	/^docs\/contributor\/erd\/.*\.mmd$/, // Generated Mermaid ERD diagrams

	/pnpm-lock\.yaml$/,
	/\.lock$/,
];

/** Conventional Commit type, and the label it earns. Several types share one label on purpose. */
const TYPE_LABELS: readonly (readonly [RegExp, string])[] = [
	[/^feat(\(.*\))?!?:/, "feature"],
	[/^fix(\(.*\))?!?:/, "bug"],
	[/^docs(\(.*\))?!?:/, "documentation"],
	[/^chore(\(.*\))?!?:/, "maintenance"],
	[/^refactor(\(.*\))?!?:/, "refactor"],
	[/^perf(\(.*\))?!?:/, "performance"],
	[/^test(\(.*\))?!?:/, "test"],
	[/^build(\(.*\))?!?:/, "maintenance"],
	[/^style(\(.*\))?!?:/, "maintenance"],
	[/^revert(\(.*\))?!?:/, "revert"],
	[/^ci(\(.*\))?!?:/, "ci"],
];

/** `!` immediately before the `:`; `[^)]*` keeps it from matching inside the scope parentheses. */
const BREAKING = /^\w+(\([^)]*\))?!:/;

const sizeLabelFor = (changedLines: number): string => {
	if (changedLines > 1000) return "size:XXL";
	if (changedLines > 499) return "size:XL";
	if (changedLines > 99) return "size:L";
	if (changedLines > 29) return "size:M";
	if (changedLines > 9) return "size:S";
	return "size:XS";
};

const changedLinesIn = (files: readonly PullRequestFile[]): number =>
	files.reduce((sum, file) => sum + file.additions + file.deletions, 0);

export default async function labelPullRequest({
	github,
	context,
}: {
	github: GitHubApi;
	context: GitHubContext;
}): Promise<void> {
	const issue = {
		owner: context.repo.owner,
		repo: context.repo.repo,
		issue_number: context.issue.number,
	};

	console.log("Calculating PR size via listFiles pagination...");
	const files = await github.paginate(github.rest.pulls.listFiles, {
		owner: context.repo.owner,
		repo: context.repo.repo,
		pull_number: context.issue.number,
		per_page: 100,
	});

	const isGenerated = (filename: string): boolean =>
		GENERATED_PATTERNS.some((pattern) => pattern.test(filename));
	const nonGeneratedFiles = files.filter((file) => !isGenerated(file.filename));
	const generatedFiles = files.filter((file) => isGenerated(file.filename));

	const changedLines = changedLinesIn(nonGeneratedFiles);
	console.log(`Changed lines (excluding generated): ${changedLines}`);
	console.log(
		`Generated files excluded: ${generatedFiles.length} files, ${changedLinesIn(generatedFiles)} lines`,
	);
	if (generatedFiles.length > 0) {
		console.log(`Excluded files: ${generatedFiles.map((file) => file.filename).join(", ")}`);
	}

	const sizeLabel = sizeLabelFor(changedLines);

	const title = context.payload.pull_request?.title ?? "";
	const typeLabels = TYPE_LABELS.filter(([pattern]) => pattern.test(title)).map(([, label]) => label);
	if (BREAKING.test(title) || title.includes("BREAKING CHANGE")) typeLabels.push("breaking");
	console.log(`Detected type labels from title "${title}":`, typeLabels);

	console.log("Fetching existing labels...");
	const labelsResponse = await github.rest.issues.listLabelsOnIssue(issue);
	const existingLabels = labelsResponse.data.map((label) => label.name);

	// 1. Handle Size Labels
	const sizeToRemove = existingLabels.filter(
		(name) => name.startsWith("size:") && name !== sizeLabel,
	);
	if (sizeToRemove.length > 0) {
		console.log(`Removing old size labels: ${sizeToRemove.join(", ")}`);
		for (const name of sizeToRemove) {
			await github.rest.issues.removeLabel({ ...issue, name });
		}
	}
	if (!existingLabels.includes(sizeLabel)) {
		console.log(`Adding size label: ${sizeLabel}`);
		await github.rest.issues.addLabels({ ...issue, labels: [sizeLabel] });
	}

	// 2. Handle Semantic Labels
	const labelsToAdd = typeLabels.filter((label) => !existingLabels.includes(label));
	if (labelsToAdd.length > 0) {
		console.log(`Adding semantic labels: ${labelsToAdd.join(", ")}`);
		await github.rest.issues.addLabels({ ...issue, labels: labelsToAdd });
	}
}