scalar-in-domain (ODW9505)
Preview (since 0.17.0.36) · Related issues · View source
Derived from the odoo linter.
Fix is sometimes available.
This rule is unstable and in preview. The --preview flag is required for use.
What it does
Checks for a domain condition whose operator is in or not in and whose value is a
single value rather than a list, such as ("country_ids", "in", record.country_id.id).
It looks at every (field, "in" | "not in", value) triple, so it covers the domains passed
to search, search_count, search_read, filtered_domain, _search or domain=, and
the ones kept in a variable, in a Python file and in the domain of an XML file: the
domain and filter_domain attributes, <attribute name="domain">, the domain and
domain_force fields of a record, and attrs. XML files are linted when they are passed
to Ruff or matched by extend-include.
The value is reported when it is certainly a scalar:
- a literal:
"draft",5,True,False,None; uid, or an attribute nameduidor_uid(self.env.uid);- the
idof a record:partner.id,user.id,self.env.company.id; - a local name whose only assignment in the same scope is one of the above.
In an XML file, where the names are the fields of the form, a bare name that is id,
active_id or ends in _id (donation_type_id) is reported too: it holds the id of a
many2one, or False. The same goes for a domain written as a string in the domain=
argument of a Python field (fields.Many2one(domain="[('company_ids', 'in', company_id)]")),
which Odoo evaluates with the fields of the record. A string that is an f-string, is
concatenated or holds an escape is not read.
Everything else is left alone, because it may well be a collection: a list, tuple, set or
dict display, a comprehension, a recordset, record.ids, records.mapped(...), any call,
subscript or conditional expression, and any other name. A Python name is never guessed
from its spelling: company_id = self.env.company holds a recordset, which a domain
accepts.
Why is this bad?
in and not in compare a field with a collection. Since 19.0 a domain that gets a scalar
logs The domain condition [...] should have a list value. and wraps it in a list, which
hides the mistake. A falsy scalar (False, None, 0, "", or the id of an empty
recordset) never reaches that step: the condition is replaced by a constant, always false
for in and always true for not in, whatever the records hold.
The step is _optimize_in_set.
Example
Use instead:
Fix safety
The fix is offered for Python only, and only when the replacement means the same thing:
a truthy literal is wrapped in a list (safe), and .id becomes .ids or uid is wrapped in
a list (unsafe, because the object may not be a record, and for a record that is new .id
is a NewId while .ids is not). A falsy literal and a name are reported without a fix:
[False] is not the same condition as False.