More of a question

I have a query

SELECT "words"."word", count("files"."file") as "count" FROM "words", "phones", "files" WHERE "words"."word_id" = "phones"."word_id" AND "words"."word_id" = "files"."word_id" AND "phones"."phone1" = 'SH' GROUP BY "words"."word"

What is confusing me is if I run through ODBC and libreoffice base the above returns records and also fairly quickly.

If I open the database direct with sqlitebrowser on ubuntu 22.04 with a db created in python I get the following

Execution finished without errors.
Result: 0 rows returned in 24529ms
At line 1:
SELECT "words"."word", count("files"."file") as "count" FROM "words", "phones", "files" WHERE "words"."word_id" = "phones"."word_id" AND "words"."word_id" = "files"."word_id" AND "phones"."phone1" = 'SH' GROUP BY "words"."word"

1stly as seen its very slow but also how can it return zero rows when the exact same via ODBC does.
Maybe its the snap of sqlitebrowser on Ubuntu 22.04 as never had anything like this on 20.04

Start by reproducing the supplied SQL query in SQLiteBrowser on Ubuntu 22.04 using the database created in Python, then compare it with the ODBC and LibreOffice Base results. Done means identifying a reproducible cause for the differing rows or query time, or documenting the missing information needed to investigate it.