[FFmpeg-devel] [PATCH] Add -fpre option.

[Michael Niedermayer]

On Mon, Nov 02, 2009 at 12:00:56PM -0200, Ramiro Polla wrote:
> On Mon, Nov 2, 2009 at 9:39 AM, Michael Niedermayer <michaelni at gmx.at> wrote:
> > On Mon, Nov 02, 2009 at 12:30:53PM +0100, Michael Niedermayer wrote:
> >> On Sun, Nov 01, 2009 at 10:53:37PM -0200, Ramiro Polla wrote:
> >> > msg7668 in issue659 spots another case where the preset files won't
> >> > open a valid path. Also "../" doesn't work for example. Attached patch
> >> > makes it the user's job to pass -fpre to open a preset file and FFmpeg
> >> > will just pass it on to open().
> >>
> >> your hack makes it impossible to specify different preset files
> >> for audio and video, that worked before ...
> >> also it complicates the interface
> 
> 'a', 'v', 's' are only used to write the codec name to the filename,
> and when specifying 'f' that seemed unnecessary.
> 
> >> is there a reason why open() is not always tried if all else fails?
> 
> patch attached opt_preset_1.diff
> 

> Although I prefer the other way around (try open() and if that fails
> snprintf() the filename)

that would be possibly exploitable
1. user downloads archive
2. user extracts files to new directory
3. user enters directors
4. user encodes a file from there with some common preset
5. sadly there is a file matching that common preset in that archive
   this gives an attacker full control over the command line parameters and
   the input file, definitly easier for her than just the input file


> 
> > or alternatively check for some prefix char and strip that from the
> > name to detect filenames or -fpre / fapre (audio) / fvpre (video) ...
> 
> patch attached opt_preset_2.diff but that makes no difference at the
> moment from my first patch. This way I can do:
> ffmpeg -i input -vcodec libx264 -fapre some_libx264_preset.ffpreset output
> 
> Is there any reason we separate [avs]pre besides to get the codec name
> to snprintf() into the filename? opt_preset() also allows specifying
> [avs]codec so a preset file could really change any codec and
> parameter...

hmmmmmm, i thought there was some other use, but it seems not
well, if its really useless then fpre alone should be enough of course
and iam of course ok with that patch

[...]
-- 
Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

Why not whip the teacher when the pupil misbehaves? -- Diogenes of Sinope
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
URL: <http://lists.mplayerhq.hu/pipermail/ffmpeg-devel/attachments/20091102/3166e447/attachment.pgp>