Version: 54
OS: All

What steps will reproduce the problem?
(1) Make an empty file site/ا/test.html.
(2) cd site
(3) Run python -m SimpleHTTPServer.
(4) Visit: http://127.0.0.1:8000/%D8%A7/test.html

The key here is that you used an IP address, not a domain, and that the path contains an Arabic character U+FE70 ARABIC FATHATAN ISOLATED FORM). This is essentially the same bug as  Issue 495933  but in the title, not the Omnibox (so it isn't security).

What is the expected output?
The tab title shows:
127.0.0.1:8000/‎ا‬/test.html

What do you see instead?
The tab title shows:
test.html/‪ا‬‬/127.0.0.1:8000

This is due to the Arabic character causing the entire URL to be rendered in an RTL context (so the domain is shown on the right). We fixed this in the Omnibox in  Issue 495933 ; we should fix it elsewhere.

RFC 3987 § 4.1 states: "Bidirectional IRIs MUST be rendered in the same way as they would be if they were in a left-to-right embedding; i.e., as if they were preceded by U+202A, LEFT-TO-RIGHT EMBEDDING (LRE), and followed by U+202C, POP DIRECTIONAL FORMATTING (PDF)."; we are not compliant here. Because it is non-editable, we can literally just insert U+202A and U+202C here.

There is a function url_formatter::FormatUrl which is responsible for making URLs presentable throughout Chrome. We could add it here, but it might have unintended consequences. A more targeted approach would be to add it in NavigationEntryImpl::GetTitleForDisplay.
 

Deleted: title-rtl.png 6.6 KB

	title-rtl.png 6.6 KB View Download