Re: json(b)_to_tsvector with numeric values

1) I don't like jsonb_all_to_tsvector too.. What if we will accept new variant 
to index? Let me suggest:

tsvector jsonb_to_tsvector([regclass,] jsonb, text[])

where text[] arg is actually a flags, array contains any combination of literals 
'numeric', 'string', 'boolean' (and even 'key' to index keys_ to point which 
types should be indexed. More than it, may be, it should a jsonb type for 
possible improvements in future. For now, it shouldbe a jsonb array type with 
string elements described above, example:

select jsonb_to_tsvector('{"a": "aaa in bbb ddd ccc", "b":123}',
                '["numeric", "boolean"]');


Form jsonb_to_tsvector('...', '["string"]) is effectively the same as current 
to_tsvector(jsonb)

2)
Now it fails, and I see something strange in resuling tsvector: 'true':9,13 and
'fals':9,13 - I don't see any bool keys in input json.

% more /home/teodor/pgsql/src/test/regress/regression.diffs
*** /home/teodor/pgsql/src/test/regress/expected/jsonb.out      2018-04-06 
16:34:59.424481000 +0300
--- /home/teodor/pgsql/src/test/regress/results/jsonb.out       2018-04-06 
16:36:48.095411000 +0300
***************
*** 4132,4138 ****
   select jsonb_all_to_tsvector('english', '{"a": "aaa in bbb ddd ccc", "b": 
123, "c": 456}'::jsonb);
                       jsonb_all_to_tsvector
   --------------------------------------------------------------
!  '123':7 '456':11 'aaa':1 'bbb':3 'ccc':5 'ddd':4 'true':9,13
   (1 row)

   -- ts_vector corner cases
--- 4132,4138 ----
   select jsonb_all_to_tsvector('english', '{"a": "aaa in bbb ddd ccc", "b": 
123, "c": 456}'::jsonb);
                       jsonb_all_to_tsvector
   --------------------------------------------------------------
!  '123':7 '456':11 'aaa':1 'bbb':3 'ccc':5 'ddd':4 'fals':9,13
   (1 row)

   -- ts_vector corner cases


Dmitry Dolgov wrote:
>> On 4 April 2018 at 16:09, Teodor Sigaev <teodor@sigaev.ru> wrote:
>>
>>>> Hm, seems, it's useful feature, but I suggest to make separate function
>>>> jsonb_any_to_tsvector and add support for boolean too (if you know better
>>>> name for function, do not hide it). Changing behavior of existing
>>>> function
>>>> is not obvious for users and, seems, should not backpatched.
>>>
>>>
>>> What do you think about having not a separate function, but a flag
>>> argument to
>>> the existing one (like `create` in `jsonb_set`), that will have false as
>>> default value? The result would be the same, but without an extra function
>>> with
>>> almost the same implementation.
>>
>>
>> tsvector jsonb_to_tsvector(jsonb[, bool]) ?
>> Agreed. Second arg should be optional.
> 
> Unfortunately, this idea with a flag argument can't be implemented easily
> (related discussion is here [1]). So I've modified the patch accordingly to
> your original suggestion about having separate functions
> `json(b)_all_to_tsvector`.
> 
> 1:
https://www.postgresql.org/message-id/flat/CA%2Bq6zcVJ%2BWx%2B-%3DkkN5UC0T-LtsJWnx0g9S0xSnn3jUWkriufDA%40mail.gmail.com
> 

-- 
Teodor Sigaev                                   E-mail: teodor@sigaev.ru
                                                    WWW: http://www.sigaev.ru/