From 8ae06f3a7a399bddd2eac2690e94040dbb600eb3 Mon Sep 17 00:00:00 2001
From: Florent Heyworth <florent@heyworth.tech>
Date: Thu, 31 Mar 2022 12:46:04 +0200
Subject: [PATCH] add html utf8 bom signature closes #267

---
 internal/magic/text.go     |  1 +
 mimetype_test.go           |  1 +
 testdata/html.utf8bom.html | 23 +++++++++++++++++++++++
 3 files changed, 25 insertions(+)
 create mode 100644 testdata/html.utf8bom.html
diff --git a/internal/magic/text.go b/internal/magic/text.go
index e2a03caf..f4d50465 100644
--- a/internal/magic/text.go
+++ b/internal/magic/text.go
@@ -13,6 +13,7 @@ import (
 var (
 	// HTML matches a Hypertext Markup Language file.
 	HTML = markup(
+		append([]byte{0xEF, 0xBB, 0xBF}, []byte("<!DOCTYPE HTML")...),
 		[]byte("<!DOCTYPE HTML"),
 		[]byte("<HTML"),
 		[]byte("<HEAD"),
diff --git a/mimetype_test.go b/mimetype_test.go
index 2b926932..442cac84 100644
--- a/mimetype_test.go
+++ b/mimetype_test.go
@@ -80,6 +80,7 @@ var files = map[string]string{
 	"html.svg.html":      "text/html; charset=utf-8",
 	"html.usascii.html":  "text/html; charset=us-ascii",
 	"html.utf8.html":     "text/html; charset=utf-8",
+	"html.utf8bom.html":  "text/html; charset=utf-8",
 	"html.withbr.html":   "text/html; charset=utf-8",
 	"ico.ico":            "image/x-icon",
 	"ics.dos.ics":        "text/calendar",
diff --git a/testdata/html.utf8bom.html b/testdata/html.utf8bom.html
new file mode 100644
index 00000000..3c47ba38
--- /dev/null
+++ b/testdata/html.utf8bom.html
@@ -0,0 +1,23 @@
+﻿<!DOCTYPE html>
+<!--
+Some comment
+
+-->
+<html dir="ltr" mozdisallowselectionprint>
+  <head>
+    <meta charset="utf-8">
+    <meta name="viewport" content="width=device-width, initial-scale=1, maximum-scale=1">
+    <meta name="some name" content="notranslate">
+    <title>test</title>
+
+
+    <link rel="stylesheet" href="html.utf8bom.css">
+
+
+
+  </head>
+
+  <body tabindex="1">
+    <div id="printContainer"></div>
+  </body>
+</html>
