mahara-contributors team mailing list archive

[Richard Mansfield <591373@xxxxxxxxxxxxxxxxxx>]

Damien:

I think mime_content_type would still have had a go at it even after the
browser says it's application/octet-stream.

But I've added your suggestion now here too:
http://gitorious.org/mahara/mahara/commit/851d6d3a44dac9445a96867c3120d91977d95fa7

So after this we'll be trying three times: browser/proxy, followed by
mime_content_type, followed by filename extension.

-- 
Mime-type recognition fails for file uploads when Mahara behind reverse proxy
https://bugs.launchpad.net/bugs/591373
You received this bug notification because you are a member of Mahara
Contributors, which is subscribed to Mahara.

Status in Mahara ePortfolio: Fix Committed
Status in Mahara 1.2 series: Fix Committed

Bug description:
When Mahara is hosted behind a reverse proxy, files uploaded are not recognized as their proper types. For example: if I go to "My Portfolio" and "My Files" to upload an gzip file, the file will upload properly but will not be properly recognized as being an archive. The result is that the generic file icon is used and the "Unzip" link is not displayed.

After looking at the code, I can see that Mahara is depending on the browser to send the proper mime-type when the file is uploaded. In our case, this value is being set to a generic value (application/octet-stream) when passed through our reverse proxy. The code that is responsible for handling new archive files (from archive/file/lib.php) is here:

    public static function new_archive($path, $data) {
        if (!isset($data->filetype)) {
            return self::archive_from_file($path, $data);
        }
        $descriptions = self::archive_file_descriptions();
        $validtypes = self::archive_mime_types();
        if (isset($validtypes[$data->filetype])) {
            return self::archive_from_file($path, $data, $descriptions[$validtypes[$data->filetype]->description]);
        }
        return false;
    }

This code checks for a mime-type first (sent by the browser) and if it finds one, assumes that it is correct. Perhaps it would be better if the code tried to determine the mime-type using the file extension in the case where the browser sends "application/octet-stream". Like this:

    public static function new_archive($path, $data) {
        if (!isset($data->filetype)) {
            return self::archive_from_file($path, $data);
        }
        if ($data->filetype == "application/octet-stream" &&
            $result = self::archive_from_file($path, $data)) {
            return $result;
        }
        $descriptions = self::archive_file_descriptions();
        $validtypes = self::archive_mime_types();
        if (isset($validtypes[$data->filetype])) {
            return self::archive_from_file($path, $data, $descriptions[$validtypes[$data->filetype]->description]);
        }
        return false;
    }

When I make this change, everything works properly for me.