[WebDriver][socket] Titles containing multibyte characters cannot be retrieved correctly

haruhisa-shin · fujii · commit 0a3175f4f36a · 2024-06-06T00:08:54.000-07:00
https://bugs.webkit.org/show_bug.cgi?id=270063 Reviewed by Alexey Proskuryakov. If the document title contains multibyte characters such as Japanese or entity references, the "Get Title" result will be garbled. The title string is obtained by JavaScript's "document.title()", and this data is encoded in UTF8. However, the StringBuilder.append() function used to create HTTP messages uses fromLatin1() internally to generate strings from byte data. This seems to be causing the multibyte characters to be garbled. This patch changes to use String::fromUTF8() before concatenation to restore the correct WTF::String even if it contains multibyte characters. Also, the change of get.py is regression test for this issue. This is an export from web-platform-tests/wpt#46584. * Source/WebDriver/socket/HTTPServerSocket.cpp: (WebDriver::HTTPRequestHandler::packHTTPMessage const): * WebDriverTests/imported/w3c/webdriver/tests/classic/get_title/get.py: (test_strip_and_collapse): (test_title_included_entity_references): (test_title_included_multibyte_char): Canonical link: https://commits.webkit.org/279767@main
diff --git a/Source/WebDriver/socket/HTTPServerSocket.cpp b/Source/WebDriver/socket/HTTPServerSocket.cpp
@@ -127,7 +127,7 @@ String HTTPRequestHandler::packHTTPMessage(HTTPRequestHandler::Response&& respon
     builder.append(EOL);
 
     if (!response.data.isNull())
-        builder.append(response.data.span());
+        builder.append(String::fromUTF8(response.data.span()));
 
     return builder.toString();
 }
diff --git a/WebDriverTests/imported/w3c/webdriver/tests/classic/get_title/get.py b/WebDriverTests/imported/w3c/webdriver/tests/classic/get_title/get.py
@@ -1,4 +1,4 @@
-from tests.support.asserts import assert_error, assert_success
+﻿from tests.support.asserts import assert_error, assert_success
 
 
 def get_title(session):
@@ -54,3 +54,17 @@ def test_strip_and_collapse(session, inline):
 
     result = get_title(session)
     assert_success(result, "a b c d e")
+
+
+def test_title_included_entity_references(session, inline):
+    session.url = inline("<title>&reg; &copy; &cent; &pound; &yen;</title>")
+
+    result = get_title(session)
+    assert_success(result, u'® © ¢ £ ¥')
+
+
+def test_title_included_multibyte_char(session, inline):
+    session.url = inline(u"<title>日本語</title>")
+
+    result = get_title(session)
+    assert_success(result, u"日本語")

Original file line number	Diff line number	Diff line change
`@@ -127,7 +127,7 @@ String HTTPRequestHandler::packHTTPMessage(HTTPRequestHandler::Response&& respon`
`127`	`127`	`builder.append(EOL);`
`128`	`128`
`129`	`129`	`if (!response.data.isNull())`
`130`		`- builder.append(response.data.span());`
	`130`	`+ builder.append(String::fromUTF8(response.data.span()));`
`131`	`131`
`132`	`132`	`return builder.toString();`
`133`	`133`	`}`