parser: tokens in tsearch2

I need to "glue" codes to "concatenate" them as a single lexeme to
other tsvectors in a way that a code won't be split but will remain
one lexeme.

eg. if input is "ab12 456/z7_92$44'a37" I don't want it get parsed
into
{ab12} {456} {z7} {92} {44} {a37}
but I want to keep the highest possible number of "symbols" in the
string so that it ends up in something like:
{ab12456/z792$44'a37}

So I need to know where the parser will cut a string so I can

_code:=regexp_replace(_code,'[/$ \t-]','');
replacing the minimum number of characters to avoid _code get split
in more than one lexeme.

thanks

--
Ivan Sergio Borgonovo
http://www.webthatworks.it