How to generate a "dialect" with gettext

[Marcel J.]

Hi!

I'm pretty new to gettext.

I want to develop a kind of "special" translation and don't know what's 
the best way to do it.

I want to use .po files, that have already been translated and only 
change some words in those files. Up to this point, there's no problem.

However, I don't know how to deal with updates of the "official" 
translation.
Let me give an example:
There is an "official" translated .po file for <program> v1.0 (let's 
call it <o-.po1.0>)
I take this file and change a few words to my "dialect" (let's call this 
file <d-.po1.0>)

After a while a new version <programm> v2.0 comes out.
Again an "official" translated .po file is developed (called <o-.po2.0>)
<o-.po2.0> is pretty much the same as <o-.po1.0>. Still there are a few 
new definitions, some definitions were deleted and a couple of them have 
been changed.
How can I create a <d-.po2.0> now? Of course I can go through the whole 
<o-.po2.0> and make all the same changes again and again, but I hope 
there's a way to keep all the changes, once they are done.

The routine I would need would:
compare <o-.po2.0> with <o-.po1.0>
    if a phrase is the same in <o-.po2.0> and <o-.po1.0>
        this phrase should be the same in <d-.po2.0> and <d-.po1.0>
    if a phrase is different in <o-.po2.0> and <o-.po1.0>
        this phrase should be the same in <d-.po2.0> and <o-.po2.0>
        and get a "fuzzy"-flag
    if a phrase is new in <o-.po2.0>
        this phrase should be the same in <d-.po2.0> and <o-.po2.0>
        and get a "fuzzy"-flag
    if a phrase was deleted in <o-.po2.0>
        this phrase should not be in <d-.po2.0>

How can this be achieved?

Thanks in advance!

Marcel