Bug 270063 – [WebDriver][socket] Titles containing multibyte characters cannot be retrieved correctly

• Status: RESOLVED
• Resolution: FIXED
• Component: WebDriver

[haruhisa.shin]

This is a problem with the WebDriver of socket implementation.
I have confirmed this problem with wincairo and playstation.

If the document title contains multibyte characters such as Japanese or entity references, the "Get Title" result will be garbled.

For example:
<title>foobar&reg;</title>
<title>日本語</title>

The title is obtained by JavaScript's "document.title()" and is UTF-8 encoded in WebDriverService::sendResponse.
https://github.com/WebKit/WebKit/blob/main/Source/WebDriver/WebDriverService.cpp#L332

However, when the result is concatenated in HttpServer, StringBuilder.append decodes the characters with fromLatin1().
This seems to be causing the multibyte characters to be garbled.
https://github.com/WebKit/WebKit/blob/main/Source/WebDriver/socket/HTTPServerSocket.cpp#L131
https://github.com/WebKit/WebKit/blob/main/Source/WTF/wtf/text/StringBuilder.h#L233

I think that String::fromUTF8() should be used before concatenating strings.

[haruhisa.shin]

Pull request: https://github.com/WebKit/WebKit/pull/25085

[Radar WebKit Bug Importer]

<rdar://problem/123987149>

[EWS]

Committed 279767@main (0a3175f4f36a): <https://commits.webkit.org/279767@main>

Reviewed commits have been landed. Closing PR #25085 and removing active labels.