full_text_mroonga_catalog: fix a PY3 regression

from 031408c6 (erp5_full_text_mroonga_catalog: defer fulltext index
and unindex using dedicated catalog keys., 2025-01-13)

bt5/erp5_full_text_mroonga_catalog/CatalogMethodTemplateItem/portal_catalog/erp5_mysql_innodb/SQLCatalog_deferUncatalogObjectActivity.py

+contains_more_than_uid = set(['uid']).isdisjoint
+
 uid_list = []
 for grouped_message in grouped_message_list:
   assert not grouped_message.args
-  assert grouped_message.kw.keys() == ['uid']
+  assert not contains_more_than_uid(grouped_message.kw)
   uid_list.append(grouped_message.kw['uid'])
 for method_id in context.getSqlDeferredUncatalogObjectList():
   method = getattr(context, method_id)

[Jérome Perrin]

@kazuhiko is this a good way ? I first did using ensure_list ( see 8631cc5d ) but I don't like ensure_list so much because it is "slower" on py3.

[Kazuhiko Shiozaki]

How about the following ?

@@ -1,8 +1,8 @@
 uid_list = []
 for grouped_message in grouped_message_list:
   assert not grouped_message.args
-  assert grouped_message.kw.keys() == ['uid']
-  uid_list.append(grouped_message.kw['uid'])
+  uid_list.append(grouped_message.kw.pop('uid'))
+  assert not grouped_message.kw
 for method_id in context.getSqlDeferredUncatalogObjectList():
   method = getattr(context, method_id)
   method(uid=uid_list)

[Jérome Perrin]

Thanks, but is it "safe" to mutate this dict like this ?

[Kazuhiko Shiozaki]

I believe it's safe, unless we use the dict in the code below of the same script.

[Kazuhiko Shiozaki]

Or... I can propose the following, if you prefer.

  assert len(grouped_message.kw) == 1
  uid_list.append(grouped_message.kw['uid'])

[Jérome Perrin]

ah sorry, I forgot to reply. I was a bit worried about not only using the dict in the same script, but also from the functions calling the script. I really like what you just suggested.