Re: libdvdread has trouble with some DVD images

Wed Jul 24 2002, Jeff Bowden wrote:
> The company I work for has a library of non-encrypted DVD images on a 
> large file server.  These images were created as backup copies of VHS 
> video tapes using some DVD authoring software.   I've been using tccat 
> from the transcode project (which uses libdvdread) to extract the mpeg-2 
> streams from these but a small percentage of them cause trouble.  The 
> exact error is:
> 
> tccat: ifo_read.c:324: ifoRead_VMG: Assertion `vmgi_mat->vmgm_vobs == 0 
> || (vmgi_mat->vmgm_vobs > vmgi_mat->vmgi_last_sector && 
> vmgi_mat->vmgm_vobs < vmgi_mat->vmg_last_sector)' failed.
> 
> For reference: The Mediamatics DVD player in Windows can play from these 
> images.  Also, if I make a physical DVD-R from one of these images, I 
> can play it in a Sony DVD player.
> 
> Any ideas for how to proceed in enhancing dvdread to be able to handle 
> these?
> 
> I modified ifo_read.c to dump the contents of the vmgi_mat struct.  Here 
> is example of its output:
> 
> vmgi_mat_t:
>    vmg_identifier[12] = "DVDVIDEO-VMG"
>    vmg_last_sector = 277
>    vmgi_last_sector = 47
>    zero_2 = 0
>    specification_version = 16
>    vmg_category = 0
>    vmg_nr_of_volumes = 1
>    vmg_this_volume_nr = 1
>    disc_side = 1
>    vmg_nr_of_title_sets = 2
>    provider_identifier[32] = "FLASH-DVD"
>    vmg_pos_code = 0
>    vmgi_last_byte = 1275
>    first_play_pgc = 1024
>    vmgm_vobs = 16
>    tt_srpt = 1
>    vmgm_pgci_ut = 2
>    ptl_mait = 0
>    vts_atrt = 45
>    txtdt_mgi = 0
>    vmgm_c_adt = 46
>    vmgm_vobu_admap = 47
>    nr_of_vmgm_audio_streams = 0
>    nr_of_vmgm_subp_streams = 1
> tccat: ifo_read.c:341: ifoRead_VMG: Assertion `vmgi_mat->vmgm_vobs == 0 
> || (vmgi_mat->vmgm_vobs > vmgi_mat->vmgi_last_sector && 
> vmgi_mat->vmgm_vobs < vmgi_mat->vmg_last_sector)' failed.
>
So it seems that it's the part that checks 
vmgi_mat->vmgm_vobs > vmgi_mat->vmgi_last_sector
that fails.  Indeed the ifo information can't end a sector 47
and the VIDEO_TS.VOB file begin at sector 16.
(Or well I guess it could if the files where 'nested' within each other..
VIDEO_TS.VOB occupying 16-44)
Can you 'mount' one of these and look at what it says is the file
lenght of VIDEO_TS.IFO and VIDEO_TS.VOB ?

Given that vts_atrt, vmgm_c_adt and vmgm_vobu_admap seem to start
at sector offsets 45,46 and 47  the ifo file would have to end at
47 or later.  vmg_last_sector beeing 277 would also indicate that
the VIDEO_TS.VOB file should be 182 sectors long, not 29 (44-16+1).
 
Since libdvdread doesn't actually use these values for anything this
assert could just be a waring instead.

-- 
Håkan Hjort