Add binary file checking

changelog.md

@@ -1,5 +1,10 @@
 # Changelog
 
+## 1.6.0
+*2021-08-15*
+- Added checking of binary files to avoid including them in the language results.
+- Added `keepBinary` option to control whether binary files are checked (defaults to `false`).
+
 ## 1.5.5
 *2021-08-15*
 - Fixed a crash occurring when using the CLI.
@@ -27,8 +32,7 @@
 ## 1.5.0
 *2021-08-12*
 - Added checking of shebang (`#!`) lines for explicit language classification.
-- Added `checkShebang` API option to implement shebang checking (defaults to `true` unless `quick` is set).
-- Added `--checkShebang` CLI option (ditto).
+- Added `checkShebang` option to implement shebang checking (defaults to `true` unless `quick` is set).
 
 ## 1.4.5
 *2021-08-10*

package.json

@@ -1,6 +1,6 @@
 {
 	"name": "linguist-js",
-	"version": "1.5.5",
+	"version": "1.6.0",
 	"description": "Analyse languages used in a folder. Powered by GitHub Linguist, although it doesn't need to be installed.",
 	"main": "dist/index.js",
 	"bin": {
@@ -32,6 +32,7 @@
 	},
 	"homepage": "https://github.com/Nixinova/Linguist#readme",
 	"dependencies": {
+		"binary-extensions": "^2.2.0",
 		"commander": "^8.1.0",
 		"cross-fetch": "^3.1.4",
 		"glob-to-regexp": "^0.4.1",

src/index.ts

@@ -3,6 +3,7 @@ import fetch from 'cross-fetch';
 import yaml from 'js-yaml';
 import glob from 'tiny-glob';
 import glob2regex from 'glob-to-regexp';
+import binaryData from 'binary-extensions';
 import Cache from 'node-cache';
 
 import * as T from './types';
@@ -180,9 +181,11 @@ export = async function analyse(root = '.', opts: T.Options = {}): Promise<T.Res
 		// Fallback to null if no language matches
 		if (!results[file]) addResult(file, null);
 	}
-
-	// Parse heuristics if applicable
 	for (const file in results) {
+		// Skip binary files
+		if (!opts.keepBinary && binaryData.some(ext => file.endsWith(ext))) continue;
+
+		// Parse heuristics if applicable
 		heuristics:
 		for (const heuristics of heuristicsData.disambiguations) {
 			// Check heuristic extensions
@@ -215,9 +218,7 @@ export = async function analyse(root = '.', opts: T.Options = {}): Promise<T.Res
 				finalResults[file] ??= Array.isArray(lastLanguage) ? lastLanguage[0] : lastLanguage;
 			}
 		}
-	}
-	// If no heuristics, load the only language
-	for (const file in results) {
+		// If no heuristics, load the only language
 		finalResults[file] ??= results[file][0];
 	}
 

src/types.ts

@@ -8,6 +8,7 @@ export type Integer = number
 
 export interface Options {
 	keepVendored?: boolean
+	keepBinary?: boolean
 	quick?: boolean
 	checkIgnored?: boolean
 	checkAttributes?: boolean

test/test.js

@@ -22,7 +22,7 @@ async function test() {
 
 	await linguist(samplesFolder).then(actual => {
 		const assert = (a, b, msg) => console.assert(a === b, msg, [a, b]);
-		const getResults = obj => Object.entries(obj.results).flat().join(',');
+		const getResults = obj => Object.entries(obj.results).sort(([a], [b]) => a < b ? +1 : -1).flat().join(',');
 		console.log('Results:', actual);
 		console.log('TOML data:', actual.languages.all['TOML'], '\n');
 		assert(getResults(expected), getResults(actual), 'Results');