Thread: BUG #14278: Problem searching spanish words with accent mark outside the stem
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paco@hernandezgomez.com wrote:

> Search without accent mark is not working correctly when the accent mark is
> outside the stem of the word.

I think it'd be better to apply unaccent() to both the stored text
before ts_vectorization and to the query terms.   That would reliably
remove all diacritics (eñes too, though I suppose nobody would search
for their ñandúes by writing nandú, so it's not as severe).


--
Álvaro Herrera                http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services

On 7 August 2016 at 23:58, Alvaro Herrera <alvherre@2ndquadrant.com> wrote:
> paco@hernandezgomez.com wrote:
>
>> Search without accent mark is not working correctly when the accent mark=
 is
>> outside the stem of the word.
>
> I think it'd be better to apply unaccent() to both the stored text
> before ts_vectorization and to the query terms.   That would reliably
> remove all diacritics (e=C3=B1es too, though I suppose nobody would searc=
h
> for their =C3=B1and=C3=BAes by writing nand=C3=BA, so it's not as severe)=
.
>
>

problem is that unaccent() is stable so can't be in the index
expression, so OP would need to create a ts_vector field to store a
preprocessed version of the string (one in which ts_vector('spanish',
unaccent()) has been already executed. and query over that field.

<cough> or create an immutable version of unaccent() </cough>


--=20
Jaime Casanova                      www.2ndQuadrant.com
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services

Thank you very much =C3=81lvaro.

It is perfect for us, once you apply unaccent.

We have finally altered spanish config with:

alter text search configuration spanish
  alter mapping for hword, hword_part, word with unaccent, spanish_stem;

And it is working perfectly.

Thanks again. ;-)




--
View this message in context: http://postgresql.nabble.com/BUG-14278-Proble=
m-searching-spanish-words-with-accent-mark-outside-the-stem-tp5914833p59153=
53.html
Sent from the PostgreSQL - bugs mailing list archive at Nabble.com.