Detect XML UTF-16-BE & UTF-16-LE via pattern matching

sindresorhus · Sep 14, 2021 · b6e0319 · b6e0319
1 parent 74c852b
commit b6e0319
Show file tree

Hide file tree

Showing 4 changed files with 14 additions and 0 deletions.
diff --git a/core.js b/core.js
@@ -1203,6 +1203,15 @@ async function _fromTokenizer(tokenizer) {
 		};
 	}
 
+	if (check([0xFE, 0xFF, 0, 60, 0, 63, 0, 120, 0, 109, 0, 108]) // UTF-16-BOM-LE
+		|| check([0xFF, 0xFE, 60, 0, 63, 0, 120, 0, 109, 0, 108, 0])
+	) { // UTF-16-BOM-LE
+		return {
+			ext: 'xml',
+			mime: 'application/xml',
+		};
+	}
+
 	// -- Unsafe signatures --
 
 	if (

diff --git a/fixture/fixture-utf16-be-bom.xml b/fixture/fixture-utf16-be-bom.xml
diff --git a/fixture/fixture-utf16-le-bom.xml b/fixture/fixture-utf16-le-bom.xml
diff --git a/test.js b/test.js
@@ -206,6 +206,11 @@ const names = {
 	webm: [
 		'fixture-null', // EBML DocType with trailing null character
 	],
+	xml: [
+		'fixture',
+		'fixture-utf16-be-bom', // UTF-16 little endian encoded XML, with BOM
+		'fixture-utf16-le-bom', // UTF-16 big endian encoded XML, with BOM
+	],
 };
 
 // Define an entry here only if the file type has potential