[Python] Google App Engine Search api proxy after Python 2.7 to Python 3 Conversion

When using the Python 2.7 of google app engine we were able to use this import

from google.appengine.api import search



One such area

def search_update_index(self):
"""
Updates the search index for this instance.

This happens automatically on put.
"""
doc_id = self.search_get_document_id(self.key)

fields = [search.AtomField('class_name', name) for name in self.search_get_class_names()]

index = self.search_get_index()

if self.searchable_fields is None:
searchable_fields = []

for field, prop in self._properties.items():
if field == 'class':
continue
for class_, field_type in SEARCHABLE_PROPERTY_TYPES.items():
if isinstance(prop, class_):
searchable_fields.append(field)
else:
searchable_fields = self.searchable_fields

for f in set(searchable_fields):
prop = self._properties[f]
value = getattr(self, f)
field = None
field_found = False
for class_, field_type in SEARCHABLE_PROPERTY_TYPES.items():
if isinstance(prop, class_):
field_found = True
if value is not None:
if isinstance(value, list) or isinstance(value, tuple) or isinstance(value, set):
for v in value:
field = field_type(name=f, value=v)
elif isinstance(value, ndb.Key):
field = field_type(name=f, value=value.urlsafe())
elif isinstance(value, bytes):
field = field_type(name=f, value=int(value))
else:
field = field_type(name=f, value=value)
if not field_found:
raise ValueError('Cannot find field type for %r on %r' % (prop, self.__class__))

if field is not None:
fields.append(field)

document = search.Document(doc_id, fields=fields)
index.put(document)


When using this we seem to get errors like this now inside the Python 3 environment.

2025-05-16 19:40:50 sync[20250516t153825] assert stub, 'No api proxy found for service "%s"' % service
2025-05-16 19:40:50 sync[20250516t153825] ^^^^
2025-05-16 19:40:51 sync[20250516t153825] AssertionError: No api proxy found for service "search"


Should I expect to be able to use that api in the python 3 environment?

It looks like some form of middleware was developed for areas like ndb in python 2? I do not know if something similar was done for search.

Is this considered a layered reverse proxy?

Continue reading...