I have a PostgreSQL used to index text content. The SearchVector column is created successfully using the following code
UPDATE public."DocumentFiles" SET "SearchVector" = setweight(to_tsvector('pg_catalog.italian', coalesce("DocumentFileName", '')), 'A') || setweight(to_tsvector('pg_catalog.italian', coalesce("DocumentFileDescription", '')), 'B') || setweight(to_tsvector('pg_catalog.italian', coalesce("DocumentFileContentString", '')), 'B') WHERE "DocumentFileID" = 123;
The content looks like the following:
'011989':1A '5':7A 'cdp':2A 'contonu':10A 'elettr':6A 'grupp':8A 'impiant':5A 'manual':3A 'uso':4A
But if I try to run a query to get plurals or singular of manual (in Italian: manuale is one, manuali are 2 or more) it fails:
SELECT "DocumentFileID" FROM public."DocumentFiles" where "SearchVector"::tsvector @@ 'manuali'::tsquery;
return nothing
SELECT "DocumentFileID" FROM public."DocumentFiles" where "SearchVector"::tsvector @@ 'manuale'::tsquery;
return nothing
It only returns the record if I write exactly what is written in the searchvector field:
SELECT "DocumentFileID" FROM public."DocumentFiles" where "SearchVector"::tsvector @@ 'manual'::tsquery;
What’s wrong with it?
Advertisement
Answer
The problem is probably that the parameter default_text_search_configuration
is not set to italian
, so that a different stemming algorithm is used.
Be explicit and use to_tsquery('italian', 'manuali')
rather than 'manuali'::tsquery
.