Projet

Général

Profil

0001-settings-decrease-A2_DUPLICATES_THRESHOLD-to-0.2-504.patch

Benjamin Dauvergne, 25 janvier 2021 11:34

Télécharger (3,46 ko)

Voir les différences: 

Subject: [PATCH] settings: decrease A2_DUPLICATES_THRESHOLD to 0.2 (#50445)

The original limit of 0.7 is kept for the find-duplicates web-service API.

 src/authentic2/api_views.py            | 2 +-
 src/authentic2/app_settings.py         | 2 +-
 src/authentic2/custom_user/managers.py | 4 ++--
 tests/test_custom_user.py              | 4 ++--
 4 files changed, 6 insertions(+), 6 deletions(-)
src/authentic2/api_views.py
871 871 

  		




  872
  872
  
    
        attributes = data.pop('attributes', {})

  		




  873
  873
  
    
        birthdate = attributes.get('birthdate')

  		




  874
  
  
    
        qs = User.objects.find_duplicates(first_name, last_name, birthdate=birthdate)

  		




  
  874
  
    
        qs = User.objects.find_duplicates(first_name, last_name, birthdate=birthdate, threshold=0.7)

  		




  875
  875
  
    

  		




  876
  876
  
    
        return Response({

  		




  877
  877
  
    
            'data': DuplicateUserSerializer(qs, many=True).data,

  		












  

    
      src/authentic2/app_settings.py
    
  





  346
  346
  
    
        default=True,

  		




  347
  347
  
    
        definition='If an active token exists, warn user before generating a new one.'),

  		




  348
  348
  
    
    A2_DUPLICATES_THRESHOLD=Setting(

  		




  349
  
  
    
        default=0.7,

  		




  
  349
  
    
        default=0.2,

  		




  350
  350
  
    
        definition='Trigram similarity threshold for considering user as duplicate.'),

  		




  351
  351
  
    
    A2_DUPLICATES_BIRTHDATE_BONUS=Setting(

  		




  352
  352
  
    
        default=0.3,

  		












  

    
      src/authentic2/custom_user/managers.py
    
  





  92
  92
  
    
        qs = qs.order_by('dist', 'last_name', 'first_name')

  		




  93
  93
  
    
        return qs

  		




  94
  94
  
    

  		




  95
  
  
    
    def find_duplicates(self, first_name=None, last_name=None, fullname=None, birthdate=None, limit=5):

  		




  
  95
  
    
    def find_duplicates(self, first_name=None, last_name=None, fullname=None, birthdate=None, limit=5, threshold=None):

  		




  96
  96
  
    
        with connection.cursor() as cursor:

  		




  97
  97
  
    
            cursor.execute(

  		




  98
  
  
    
                "SET pg_trgm.similarity_threshold = %f" % app_settings.A2_DUPLICATES_THRESHOLD

  		




  
  98
  
    
                "SET pg_trgm.similarity_threshold = %f" % (threshold or app_settings.A2_DUPLICATES_THRESHOLD)

  		




  99
  99
  
    
            )

  		




  100
  100
  
    

  		




  101
  101
  
    
        if fullname is not None:

  		












  

    
      tests/test_custom_user.py
    
  





  127
  127
  
    
    # dist attribute signals queryset from find_duplicates()

  		




  128
  128
  
    
    assert hasattr(User.objects.free_text_search('darmettein')[0], 'dist')

  		




  129
  129
  
    

  		




  130
  
  
    
    assert User.objects.free_text_search('lea darmettein').count() == 1

  		




  
  130
  
    
    assert User.objects.free_text_search('lea darmettein').filter(dist__lte=0.3).count() == 1

  		




  131
  131
  
    
    assert hasattr(User.objects.free_text_search('darmettein')[0], 'dist')

  		




  132
  132
  
    

  		




  133
  133
  
    

  		




  134
  134
  
    
def test_fts_legacy(fts):

  		




  135
  
  
    
    assert User.objects.free_text_search('rue des peupliers').count() == 2

  		




  
  135
  
    
    assert User.objects.free_text_search('rue des peupliers').count() == 3

  		




  136
  136
  
    

  		




  137
  137
  
    

  		




  138
  138
  
    
def test_fts_legacy_and_trigram(fts):

  		




  139
  
  
    
-