Configuring space symbols in text search configuration

Question

I'm using the following text search configuration:

CREATE TEXT SEARCH DICTIONARY tsd
  (TEMPLATE = pg_catalog.SIMPLE);
  
CREATE TEXT SEARCH CONFIGURATION tsc_test
  (COPY = german);
ALTER TEXT SEARCH CONFIGURATION tsc_test
  ALTER MAPPING FOR asciihword, asciiword, hword, hword_asciipart, hword_part, word
  WITH tsd;

I'm trying to perform full text search on a table, but I can't figure out how to define the text search configuration correctly. My problem is, that

SELECT to_tsvector('tsc_test', 'foo bar');

and

SELECT to_tsvector('tsc_test', 'foo_bar');

results in the same tsvector object, because the _ (and others, like #(){}, etc.) are interpreted as space symbols (see SELECT ts_debug ('tsc_test', 'foo_bar');). I need them to be distinct, however.

Is there a way to adjust my text search configuration so it treats only spaces as word separators and not anything else?

Answer 1

There is no really good way to do that.

A workaround may be to replace everything that is not a space or a character with – say – a digit:

CREATE FUNCTION transmogrify(text) RETURNS text
   LANGUAGE sql IMMUTABLE STRICT AS
$$SELECT regexp_replace($1, '[^[:alpha:] ]', '0', 'g')$$;

SELECT to_tsvector('german', transmogrify('foo_bar'))
    @@ to_tsquery('german', transmogrify('foo'));

 ?column? 
══════════
 f
(1 row)

SELECT to_tsvector('german', transmogrify('foo_bar'))
    @@ to_tsquery('german', transmogrify('foo_bar'));

 ?column? 
══════════
 t
(1 row)

Of course that will cause trouble if there are zeros in your text, but as I said, there is no perfectly satisfying way to do this.