Add more signals when computing document language
Browsers don't use only the HTML lang attribute to compute document language. We rely on the layout engine to compute language if a layout object is available, , otherwise we use our own logic. We also use our own logic if the layout engine cannot figure out the language. Our own logic first looks at the lang attribute on the current object, or any of its ancestors, followed by the "content-language" meta tag, the accept language header and the browsers default UI language. The latter two steps provide a mere guess, but are better than simply returning en-US, which is our current implementation for IAccessible2. This certainly is an improvement over the current situation, though not perfect by all means. R=dmazzoni@chromium.org Bug: 831186 Tested: Manually using foreign language sites without lang attributes and Chrome's the "--lang" command line flag, automatically using layout tests Change-Id: Ia80c31cc6dabdcea2aeec177aeda6b29d695019e Reviewed-on: https://chromium-review.googlesource.com/1011332 Commit-Queue: Nektarios Paisios <nektar@chromium.org> Reviewed-by:Dominic Mazzoni <dmazzoni@chromium.org> Reviewed-by:
Nektarios Paisios <nektar@chromium.org> Cr-Commit-Position: refs/heads/master@{#551416}
Showing
Please register or sign in to comment