On 24/01/2011 18:49, Dirk Meyer wrote:
> Hi,
>
> On 24.01.2011 01:13, Jason Tackaberry wrote:
>> On Sun, 2011-01-23 at 14:08 +0100, Dirk Meyer wrote:
>>> Add file hash based on calculations from
>>> http://trac.opensubtitles.org/projects/opensubtitles/wiki/HashSourceCodes
>> Based on the description in the URL, this code is wrong:
>>
>>> +            filehash = 0
>>> +            if filesize>= 65536 * 2:
>>> +                filehash = filesize
>>> +                [...]
>> That site says:
>>
>>          In natural language it calculates: size + 64bit chksum of the
>>          first and last 64k (even if they overlap because the file is
>>          smaller than 128k)
>>
>> Your implementation computes a 0 hash if the file size is less than
>> 128k.  But then the Python implementation at that page (which makes me
>> bleed from the eyeballs) returns "SizeError" (!!!) in this case too.
> My implementation is the one from the side without the things that made
> both our eyeballs bleed. That is why I returned 0 which is at least a
> number. I did not look much at the code, I only tried the sample file
> and was happy. The reason I added the code is that themoviedb has
> support to detect the movie based on the hash -- but both files I tried
> gave no result. But hey, it is a new feature on the side, let's wait
> some weeks and see what will happen.
>
>> So I find myself more than a little irritated having read this page,
>> trying to understand how to implement this.  I mean, I realize they're
>> not writing software to launch shuttles into space here, but bloody
>> hell, a _little_ more rigor would be nice for people like us who are
>> trying to make a compatible implementation.
> Well, I guess they do not have corner case files. And unlike us they
> only use it for videos. But since we already have the code, I added it
> to all files. Maybe it is needed later, maybe not.
>
>> Ok, I'm done ranting.  I feel better.
> :)
>
>> I've modified kaa.metadata's
>> implementation to handle files less than 128k so the first and second
>> hashes will overlap, and for files less than 64k, it will read as many
>> full 64-bit values until it reaches EOF.  This makes it work like the
>> first C implementation if the unsigned fsize problem were fixed.
> Thanks.
>
> But BTW, right now I'm playing with kaa.webmetadata which was the reason
> I started coding again. Expect more kaa and Freevo code soon.
:) Nice to hear.
>
> Dischi
>
John.

------------------------------------------------------------------------------
Special Offer-- Download ArcSight Logger for FREE (a $49 USD value)!
Finally, a world-class log management solution at an even better price-free!
Download using promo code Free_Logger_4_Dev2Dev. Offer expires 
February 28th, so secure your free ArcSight Logger TODAY! 
http://p.sf.net/sfu/arcsight-sfd2d
_______________________________________________
Freevo-devel mailing list
Freevo-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/freevo-devel

Reply via email to