Now the program can make only one query but it can accept multiple
optional filters based on the fields. Each extra argument is a filter
specification, and filters are all joined using a logical AND.
The individual filters have the format X:fitler, there X is a field
specification:
t = titulo
d = desc
a = autor
f = formato
c = comentarios
C = cds
F = fecha
D = downloads
And filter is a string that should be found on that field (case
insensitive). If the format specifier is not known (or there isn't one)
the filter string is looked in all the fields.
Those tags can have HTML comments inside of them, and they are not
parsed by HTMLParser, so we just ignore them to avoid issues, we don't
really need them for anything anyway.