Handle UTF-8 byte order marks in Greasemonkey scripts

See e.g. https://github.com/jerone/UserScripts/issues/135

doc/changelog.asciidoc

@@ -59,6 +59,7 @@ Fixed
 ~~~~~
 
 - Error when passing a substring with spaces to `:tab-take`.
+- Greasemonkey scripts which start with an UTF-8 BOM are now handled correctly.
 
 Removed
 ~~~~~~~

qutebrowser/browser/greasemonkey.py

@@ -234,7 +234,7 @@ class GreasemonkeyManager(QObject):
             if not os.path.isfile(script_filename):
                 continue
             script_path = os.path.join(scripts_dir, script_filename)
-            with open(script_path, encoding='utf-8') as script_file:
+            with open(script_path, encoding='utf-8-sig') as script_file:
                 script = GreasemonkeyScript.parse(script_file.read())
                 if not script.name:
                     script.name = script_filename

tests/unit/javascript/test_greasemonkey.py

@@ -130,6 +130,26 @@ def test_load_emits_signal(qtbot):
         gm_manager.load_scripts()
 
 
+def test_utf8_bom():
+    """Make sure UTF-8 BOMs are stripped from scripts.
+
+    If we don't strip them, we'll have a BOM in the middle of the file, causing
+    QtWebEngine to not catch the "// ==UserScript==" line.
+    """
+    script = textwrap.dedent("""
+        \N{BYTE ORDER MARK}// ==UserScript==
+        // @name qutebrowser test userscript
+        // ==/UserScript==
+    """.lstrip('\n'))
+    _save_script(script, 'bom.user.js')
+    gm_manager = greasemonkey.GreasemonkeyManager()
+
+    scripts = gm_manager.all_scripts()
+    assert len(scripts) == 1
+    script = scripts[0]
+    assert '// ==UserScript==' in script.code().splitlines()
+
+
 def test_required_scripts_are_included(download_stub, tmpdir):
     test_require_script = textwrap.dedent("""
         // ==UserScript==