Re: [PHP] best way to determine (MIME) content type of a stream of bytes?

On Mon, 8 Mar 2010, Ashley Sheridan wrote:

> On Mon, 2010-03-08 at 10:57 -0500, Robert P. J. Day wrote:
>
> > hi, i'm interested in the most comprehensive way to determine the
> > content type of a stream of bytes that's been uploaded to a PHP
> > script?  assuming that the bytes are uploaded simply via a POST
> > parameter, i can see that there are a couple ways to do it:
> >
> >   * getimagesize()
> >   * FileInfo
> >
> > i've been doing some testing this morning and a few video formats
> > handed to FileInfo come back as "application/octet-stream" which
> > isn't particularly informative.  and i want to support as many
> > different formats of image, audio and video as possible.
> >
> >   so ... what's the best way?  oh, by the way, when i used
> > fileinfo, i didn't bother handing over a magic file.  i'm starting
> > to think that would make a difference.  and is there a noticeable
> > advantage to upgrading to PHP 5.3 since the server (centos 5.4) is
> > currently running only PHP 5.1.6.  thanks.

> If you're wanting to grab details about a clip, what about using
> mplayer for dealing with video clips. It has more than a few command
> line options that can return various levels of detail about a media
> file. You could use the extension of the clip as a hint about what
> way you can determine a files exact type. So, if a file came in with
> a jpg, png or gif extension, you could use GD functions to determine
> if it's really an image. If it's a .avi, .mpg, .mp4, .mp3, .ogg, you
> could use mplayer to deal with it.

  in order to make life as difficult as possible, all i can assume is
an incoming stream of bytes.  i will have no idea where it came from,
or its original file name.  all of the mime/type identification has to
be done by the PHP script on the server end, based solely on the
content.  (i'm fairly sure that means a "magic" file will have to be
involved.)

> This does seem to ba a bit of an area where PHP is lacking. Even the
> manual pages are cryptic. It seems to suggest that the Mime
> functions which we should use in-place of deprecated ones themselves
> rely on those same deprecated functions!

  i have noticed that.  the "mime_content_type()" function looked like
a good candidate but it's marked as deprecated.  the best option
appears to be the Fileinfo stuff.

rday
--

========================================================================
Robert P. J. Day                               Waterloo, Ontario, CANADA

            Linux Consulting, Training and Kernel Pedantry.

Web page:                                          http://crashcourse.ca
Twitter:                                       http://twitter.com/rpjday
========================================================================