Check binary file content when classifying

Author: Nixinova

changelog.md

@@ -1,6 +1,8 @@
 # Changelog
 
-## Next
+## 1.6.1
+*2021-08-17*
+- Changed binary file checking to check file content as well as extension.
 - Fixed filename matching not comparing the full base name to the list of filename matches.
 
 ## 1.6.0

package-lock.json

@@ -1,6 +1,6 @@
 {
 	"name": "linguist-js",
-	"version": "1.6.0",
+	"version": "1.6.1",
 	"description": "Analyse languages used in a folder. Powered by GitHub Linguist, although it doesn't need to be installed.",
 	"main": "dist/index.js",
 	"bin": {
@@ -36,6 +36,7 @@
 		"commander": "^8.1.0",
 		"cross-fetch": "^3.1.4",
 		"glob-to-regexp": "^0.4.1",
+		"isbinaryfile": "^4.0.8",
 		"js-yaml": "^4.1.0",
 		"node-cache": "^5.1.2",
 		"tiny-glob": "^0.2.9"

package.json

@@ -5,6 +5,7 @@ import yaml from 'js-yaml';
 import glob from 'tiny-glob';
 import glob2regex from 'glob-to-regexp';
 import binaryData from 'binary-extensions';
+import { isBinaryFile } from 'isbinaryfile';
 import Cache from 'node-cache';
 
 import * as T from './types';
@@ -185,7 +186,7 @@ export = async function analyse(root = '.', opts: T.Options = {}): Promise<T.Res
 	}
 	for (const file in results) {
 		// Skip binary files
-		if (!opts.keepBinary && binaryData.some(ext => file.endsWith(ext))) {
+		if (!opts.keepBinary && (binaryData.some(ext => file.endsWith(ext)) || await isBinaryFile(file))) {
 			continue;
 		}
 

src/index.ts

@@ -30,7 +30,10 @@ async function test() {
 	await linguist(samplesFolder).then(actual => {
 		console.log('Results:', actual);
 		console.log('Language data:', actual.languages.all, '\n');
-		if (!deepEqual(expected, actual)) throw new Error('Results differ from expected!');
+		if (!deepEqual(expected, actual)) {
+			console.warn('Results:', '\nexpected:', expected, '\nactual:', actual);
+			throw new Error('Results differ from expected!');
+		}
 	});
 }
 test();