Tabular.vim : how to align on the first occurrence of 2 different delimiters placed at the beginning of Words?

I have installed the Tabular plugin, which works very well for me, as long as there are no complicated regexes involved…

But I have this list :

one @abc @rstuvw &foo  
three @defg &bar 
four @mn @opq &kludge &hack  
twelve @hijkl &baz &quux

I wish to align it that way (on @… first, then on &…) :

one    @abc @rstuvw &foo  
three  @defg        &bar 
four   @mn @opq     &kludge &hack  
twelve @hijkl       &baz &quux

which means I have 3 problems at the same time :

align on the first occurrence
of 2 different delimiters (@ and &)
which are not really delimiters but "special characters" at the beginning of Words

This is far beyond my understanding of both regexes and Tabular.vim

How should I proceed ?

Solution

The help file explains this problem, you can use this command:

:Tabularize /^[^@]*\zs@/l1l0

A little explaination:

^ means the begin of the line
[^@]* match everything that isn't a @. The * means 0 or more times, as much as you can
\zs put the start of the regex here (everything from this point is matched)
@ the 'this point' in the previous sentence means the @ symbol
/l1l0 means align the 1st block to the left and add 1 space (l1) and align the 2nd block to the left and add 0 spaces (l0)

You need to do this in 2 commands. To make your life easier you can name the pattern and use that name:

:AddTabularPattern f_at /^[^@]*\zs@/l1l0
:AddTabularPattern f_and /^[^&]*\zs&/l1l0

Now you can run

:Tabularize f_at
:Tabularize f_and

You can even map these methods to generate easy shortcuts. Read more about this here