When downloading files having non US-ASCII characters in their names, the downloaded documents have all their accentuated characters removed.
When a file is being downloaded, the HTTP header contains two versions of the filename: filename and filename* where filename* uses the encoding defined in RFC 5987 and allows it to have non US-ASCII characters.
However, the filename* is not parsed by some browsers, who are defaulting to the filename parameter.
For instance, dépôt.log becomes dpt.log
This issue occurs with Chrome-based browsers, such as Chrome, Chromium and Microsoft Edge.