Netscape and mysql

[seizmic]

I'm trying to use a database to control the contents of documents (~CVS).

I use the followign header when trying to download contents from the db.

mysql queries....
.
..
...

if ( strstr( getenv("HTTP_USER_AGENT"), "MSIE") != false) {
//IE
header('Content-Type: application/octetstream');
header('Content-Disposition: inline; filename= '. $row['name']);
header('Expires: 0');
header('Cache-Control: must-revalidate, post-check=0, pre-check=0');
header('Pragma: public');
} else {
//not IE-explorer
header('Content-Type: application/octet-stream');
header("Content-Disposition: inline; filename=\"".$row['name']."\"");
header('Expires: 0');
header('Pragma: no-cache');
}

echo "". base64_decode($row[content]) ."";

=> view.php

creating a link to view.php only gives me a download dialog box when using netscape, but when using IE it gives me an app. for that particular content ...say pdf (then the pdf plugin launches).

I've setup netscape to launch acrobat when ever a pdf is "clicked" it works fine on files.

/S

[Sxooter]

change header('Content-Type: application/octetstream');

to

header('Content-Type: application/pdf');

[seizmic]

Well that's only valid for ONE app.

Say I have a lot of them doc,pdf,xls and so on I want those to be auto detected!

/S

[bobthedog]

grin

Welcome to the donwerful world of cross browser compatability.

I have had this crop up again and again, and unfortunately I cannot find one single solution which will work for all platforms.

The answer goes along these lines....

IE works differently to Netscape on a lot of levels. One of these levels is how it determines the content being fed down to it from a web server. IE does not always rely on the content-type header that you set (or Apache sets), but actually starts the download and "sniffs" the file headers to see if it can determine what is actually being downloaded. If appropriate, it will then display content, or prompt you for where to save it etc.

This allows IE to be a bit cleverer than all the other browsers, but can also work to it's disadvantage, when trying to force it to make the actual download save to a users machine.

Netscape doesn't do this and seems to rely solely on the content-type header that you send to it.

So - the short answer is - yes you do need to store all those lovely MIME content-types so that you can dump the correct header to each browser as required. This means a bit more code, but at least it works 🙂

[Sxooter]

Let's be clear here. The http 1.1 standard VERY clearly states that one thing and one thing only shall tell a user agent what type of content is being sent, and that is the content-type header message. Anything else I.E. does to figure out what's coming is non-standard and therefore a bug.

The problem is that IE knows what is coming through but Windows doesn't listen, and goes by the extension or the content-disposition header, which is the wrong way to do it.

In fact, this methodology has been responsible for at least one major security hole in windows, maybe more (see bugtraq).

Mozilla is doing the right thing, Windows / IE is doing the wrong thing. Program for the right thing and let MS fix their own bugs. Nice thing is that if you do things the "right way" Windows will currently work fine.

[Sxooter]

A bit of help, if you're running on a linux/apache box or something similar, you should have a file in your apache/conf directory called mime.types that will let you look up the type with a simple script like so:

<?php
$file = "/usr/local/apache/conf/mime.types";
$fp = fopen($file,"r");
$body = fread ($fp,filesize($file));
$lines = split("\n",$body);
foreach ($lines as $line){
        if (substr($line,0,1)=="#") continue(1);
        unset($tmp);
        $tmp = split(" +",$line);
        if (count($tmp)>1){
                $mime["$tmp[1]"] = $tmp[0];
        }
}

$ext="jpg";
print $mime["$ext"]."\n";
?>