Input:
select to_tsvector('simple', '["one","two","three"]'::jsonb)
returns
'one':1 'three':5 'two':3
Expected:
'one':1 'three':3 'two':2
Which words have index 2 and 4? Is this a bug? How do I use it in a valid way?
With those gaps in the index numbering any search using a followed-by operator (ex: 'one' <-> 'zwo'
) is doomed to fail.
How I can repair that using immutable existing functions? The jsonb originates from a call to jsonb_path_query_array()
which could be changed to jsonb_path_query()
but I have no clue how to consolidate a set of jsonb values into a single tsvector.