Re: [tsvector] to_tsvector called multiple times

Sven R. Kunze wrote:
> However, are you sure, I am using snowball? Maybe, I am reading the
> documenation wrong:

test=> SELECT * FROM ts_debug('german', 'system');
alias | description | token | dictionaries | dictionary | lexemes
-----------+-----------------+--------+---------------+-------------+---------
asciiword | Word, all ASCII | system | {german_stem} | german_stem | {syst}
(1 row)

test=> \dFd german_stem
List of text search dictionaries
Schema | Name | Description
------------+-------------+--------------------------------------
pg_catalog | german_stem | snowball stemmer for german language
(1 row)

> http://www.postgresql.org/docs/9.3/static/textsearch-dictionaries.html
> but it seems as it depends on which packages (ispell, hunspell, myspell,
> snowball + corresponding languages) my system has installed.
>
> Is there an easy way to determine which of these packages PostgreSQL
> uses AND what for?

If you use a standard PostgreSQL distribution, you will have no ispell
dictionary (as the documentation you quote says).
You can always list all dictionaries with "\dFd" in psql.

> Sure. That might be the problem. It occurs to me that stems (if detected
> as such) should be left alone.
> In case a stem is real German word, it should be stemmed to itself anyway
> If not, it might help not to stem in order to avoid errors.

Yes, but that would mean that you have a way to determine from a string
whether it is a word or a stem or both, and the software does not do that.

Yours,
Laurenz Albe