By default, the Recoll indexer only uses most non-alphanumeric characters as
separators, treating them as white space, so that inputs like all
words, and all,words
produce the same terms.
It may sometimes be useful to index some of these characters so that they can be
used as discriminants for searches. This can be done by setting
the indexedpunctuation configuration parameter. The value is an UTF-8
string, for example, setting:
indexedpunctuation = %€
would allow searching separately 100%
or 100€.
The affected characters are indexed as terms with their own term positions, and they
are their own separators, so that 100%
and 100 % would be equivalent inputs.
