From 8bca6b4ec8a1d6147c0d877c9cf5d8441505aa20 Mon Sep 17 00:00:00 2001
From: Borewit <Borewit@users.noreply.github.com>
Date: Fri, 17 Sep 2021 11:17:47 +0200
Subject: [PATCH] Support XML encoding with UTF-8 including BOM field (#491)

---
 core.js                      | 7 +++++++
 fixture/fixture-utf8-bom.xml | 4 ++++
 test.js                      | 4 ++++
 3 files changed, 15 insertions(+)
 create mode 100644 fixture/fixture-utf8-bom.xml

diff --git a/core.js b/core.js
index d3d6ab53..407b0e8d 100644
--- a/core.js
+++ b/core.js
@@ -1066,6 +1066,13 @@ async function _fromTokenizer(tokenizer) {
 		};
 	}
 
+	if (check([0xEF, 0xBB, 0xBF]) && checkString('<?xml', {offset: 3})) { // UTF-8-BOM
+		return {
+			ext: 'xml',
+			mime: 'application/xml',
+		};
+	}
+
 	// -- 9-byte signatures --
 
 	if (check([0x49, 0x49, 0x52, 0x4F, 0x08, 0x00, 0x00, 0x00, 0x18])) {
diff --git a/fixture/fixture-utf8-bom.xml b/fixture/fixture-utf8-bom.xml
new file mode 100644
index 00000000..2f016080
--- /dev/null
+++ b/fixture/fixture-utf8-bom.xml
@@ -0,0 +1,4 @@
+<?xml version="1.0" encoding="UTF-16"?>
+<FileType>
+	<encoding>UTF-8 with BOM</encoding>
+</FileType>
diff --git a/test.js b/test.js
index b27bddd3..610a6f24 100644
--- a/test.js
+++ b/test.js
@@ -206,6 +206,10 @@ const names = {
 	webm: [
 		'fixture-null', // EBML DocType with trailing null character
 	],
+	xml: [
+		'fixture',
+		'fixture-utf8-bom', // UTF-8 with BOM
+	],
 };
 
 // Define an entry here only if the file type has potential