There's only one global DownloadManager with its own NAM (for downloads not
associated with a page). We can't really decide whether that should be private
or not, so as a best-effort approximation we simply make it private if private
browsing was turned on when starting qutebrowser.
Before, we just returned the same data for both, but then we'll run into
same-origin restrictions as qute:history and qute:history/data are not the same
host.
Fixes#2304
In some cases, the finished handler fired before the error handler, e.g.
when downloading a 500 error page that is sent as attachment:
HTTP/1.1 500 Internal Server Error
Content-Type: application/octet-stream
Content-Disposition: inline; filename="attachment.jpg"
here we downloaded 0 bytes, fired the finished handler and after that
fired the error handler because of the 500 - but the finished handler
had already set our reply to None (and displayed the error message).
When starting a download due to unsupportedContent being emitted, we
need to use (and later adopt) the page's QNetworkAccessManager.
Since we need the whole adopting logic for that case anyways, let's keep
things as they were and always run downloads in per-tab QNAMs.
This reverts 53e360ec4b and fixes#2134.
When we use self._set_filename in self._set_fileobj, the file already
exists, so we need to force "overwriting" it.
Also, move temporary file handling to a dedicated _set_tempfile method,
so we can officially claim not supporting _set_fileobj with QtWebEngine
instead of supporting it with a hack.
This makes a lot of code eaiser, and we don't have per-tab settings yet
anyways. Also, with QtWebEngine, we can't honour any per-tab settings
for downloads...
