sqlite-utils query --functions mechanism for registering extra functions

It would be really cool if you could register additional custom SQL functions for use with the sqlite-utils query command - something like this:

sqlite-utils data.db 'update images set domain = extract_domain(url)' --functions '
from urllib.parse import urlparse

def extract_domain(url):
    return urlparse(url).netloc
'

Every function defined in that code block would be registered with the connection, unless the name began with an underscore.

[simonw]

Testing bulk --functions:

% sqlite-utils create-table chickens.db chickens id integer name text name_upper text 
% echo 'id,name
1,Blue
2,Snowy
3,Azi
4,Lila
5,Suna
6,Cardi' | sqlite-utils bulk chickens.db '
insert into chickens (id, name, name_upper) values (:id, :name, myupper(:name))
' - --functions '
def myupper(s):
    return s.upper()
' --csv
% sqlite-utils rows chickens.db chickens
[{"id": 1, "name": "Blue", "name_upper": "BLUE"},
 {"id": 2, "name": "Snowy", "name_upper": "SNOWY"},
 {"id": 3, "name": "Azi", "name_upper": "AZI"},
 {"id": 4, "name": "Lila", "name_upper": "LILA"},
 {"id": 5, "name": "Suna", "name_upper": "SUNA"},
 {"id": 6, "name": "Cardi", "name_upper": "CARDI"}]

[simonw]

Is it still nfortunately slow and tricky when playing with floats ?

Not sure what you mean here?