Handle unicode in entity search

Fixes #1934

endpoints/api/search.py

@@ -13,6 +13,7 @@ from operator import itemgetter
 from stringscore import liquidmetal
 from util.names import parse_robot_username
 
+import anunidecode # Don't listen to pylint's lies. This import is required.
 import math
 
 @resource('/v1/entities/<prefix>')
@@ -27,6 +28,11 @@ class EntitySearch(ApiResource):
   @nickname('getMatchingEntities')
   def get(self, prefix, parsed_args):
     """ Get a list of entities that match the specified prefix. """
+
+    # Ensure we don't have any unicode characters in the search, as it breaks the search. Nothing
+    # being searched can have unicode in it anyway, so this is a safe operation.
+    prefix = prefix.encode('unidecode', 'ignore').replace(' ', '').lower()
+
     teams = []
     org_data = []
 

test/test_api_usage.py

@@ -738,6 +738,14 @@ class TestConductSearch(ApiTestCase):
 
 
 class TestGetMatchingEntities(ApiTestCase):
+  def test_unicode_search(self):
+    self.login(ADMIN_ACCESS_USER)
+
+    json = self.getJsonResponse(EntitySearch,
+                                params=dict(prefix='北京市', namespace=ORGANIZATION,
+                                            includeTeams='true'))
+    self.assertEquals(0, len(json['results']))
+
   def test_notinorg(self):
     self.login(NO_ACCESS_USER)