200000 regular expressions generalized from
200000 protein names must be matched in text in
order to annotate the text with links into
a protein database.

My first guess to try this with antlr would
be to rewrite the regexps as an antlr lexer
grammar. Each regexp would be a rule, and the
rule would rewrite the token to be a link
to the database.

I know antlr only from the overview and
quick examples of its homepage. Before I dig
any deeper, I would appreciate hints as to
whether there is
any chance of success. Concerns arising from
my limited knowledge of antlr include:

"warning:lexical nondeterminism between rules"

copying "unmatched" input unchanged to the output

size

Thanks,
Harald.