Add minSize option which moves small languages into other result

src/cli.ts

@@ -22,6 +22,7 @@ program
 	.option('-j|--json [bool]', 'Display the output as JSON', false)
 	.option('-t|--tree <traversal>', 'Which part of the output JSON to display (dot-delimited)')
 	.option('-F|--listFiles [bool]', 'Whether to list every matching file under the language results', false)
+	.option('-m|--minSize <size>', 'Minimum file size to show language results for (must have a unit: b, kb, mb, %)')
 	.option('-q|--quick [bool]', 'Skip complex language analysis (alias for -{A|I|H|S}=false)', false)
 	.option('-o|--offline [bool]', 'Use packaged data files instead of fetching latest from GitHub', false)
 	.option('-V|--keepVendored [bool]', 'Prevent skipping over vendored/generated files', false)

src/index.ts

@@ -431,6 +431,30 @@ async function analyse(rawPaths?: string | string[], opts: T.Options = {}): Prom
 		results.languages.bytes += fileSize;
 	}
 
+	// Ignore languages with a bytes/% size less than the declared min size
+	if (opts.minSize) {
+		const totalSize = results.languages.bytes;
+		const minSizeAmt = parseFloat(opts.minSize.replace(/[a-z]+$/i, '')); // '2KB' -> 2
+		const minSizeUnit = opts.minSize.replace(/^\d+/, '').toLowerCase(); // '2KB' -> 'kb'
+		const conversionFactors: Record<string, (n: number) => number> = {
+			'b': n => n,
+			'kb': n => n * 1e3,
+			'mb': n => n * 1e6,
+			'%': n => n * totalSize / 100,
+		};
+		const minBytesSize = conversionFactors[minSizeUnit](+minSizeAmt);
+		// Apply specified minimums: delete language results that do not reach the threshold
+		for (const [lang, data] of Object.entries(results.languages.results)) {
+			if (data.bytes < minBytesSize) {
+				// Add data to 'Other' result
+				results.languages.results.other ??= { type: 'data' /*arbitrary*/, bytes: 0 };
+				results.languages.results.other.bytes += data.bytes;
+				// Remove language result
+				delete results.languages.results[lang];
+			}
+		}
+	}
+
 	// Set counts
 	results.files.count = Object.keys(results.files.results).length;
 	results.languages.count = Object.keys(results.languages.results).length;

src/types.ts

@@ -18,6 +18,7 @@ export interface Options {
 	keepVendored?: boolean
 	keepBinary?: boolean
 	relativePaths?: boolean
+	minSize?: `${number}${Lowercase<'B' | 'KB' | 'MB' | '%'>}`
 	childLanguages?: boolean
 	quick?: boolean
 	offline?: boolean
@@ -39,7 +40,7 @@ export interface Results {
 	languages: {
 		count: Integer
 		bytes: Bytes
-		results: Record<Language, {
+		results: Record<Language | 'other', {
 			bytes: Bytes
 			type: Category
 			parent?: Language