From d6fa31b7863168bdeb050e8074dc3734ee8ba35c Mon Sep 17 00:00:00 2001
From: Benjamin Dauvergne <bdauvergne@entrouvert.com>
Date: Mon, 13 Jul 2020 23:09:12 +0200
Subject: [PATCH 2/4] csv_import: ignore empty values when checking uniqueness
 (#44805)

---
 src/authentic2/csv_import.py |  5 ++++
 tests/test_csv_import.py     | 45 ++++++++++++++++++++++++++++++++++++
 2 files changed, 50 insertions(+)

diff --git a/src/authentic2/csv_import.py b/src/authentic2/csv_import.py
index 7e682c1c..96558a36 100644
--- a/src/authentic2/csv_import.py
+++ b/src/authentic2/csv_import.py
@@ -576,6 +576,9 @@ class UserCsvImporter(object):
             if header.name == SOURCE_ID:
                 unique_key = (SOURCE_ID, row[SOURCE_NAME].value, cell.value)
             elif header.key or header.globally_unique or header.unique:
+                if not cell.value:
+                    # empty values are not checked
+                    continue
                 unique_key = (header.name, cell.value)
             else:
                 continue
@@ -595,6 +598,8 @@ class UserCsvImporter(object):
         for cell in row:
             if (not cell.header.globally_unique and not cell.header.unique) or (user and not cell.header.update):
                 continue
+            if not cell.value:
+                continue
             qs = ou_users
             if cell.header.globally_unique:
                 qs = users
diff --git a/tests/test_csv_import.py b/tests/test_csv_import.py
index 70cbc602..97fdf2ec 100644
--- a/tests/test_csv_import.py
+++ b/tests/test_csv_import.py
@@ -518,3 +518,48 @@ tnoel@entrouvert.com,Thomas,Noël,'''
     assert importer.run()
     assert importer.has_errors
     assert importer.rows[0].cells[-1].errors[0].code == 'data-error'
+
+
+def test_check_empty_value_for_uniqueness(profile, user_csv_importer_factory):
+    ou = get_default_ou()
+    ou.username_is_unique = True
+    ou.email_is_unique = True
+    ou.save()
+
+    Attribute.objects.update(required=False)
+    User.objects.create(username='john.doe', email='john.doe@gmail.com')
+
+    content = '''_source_name,_source_id,username,email,first_name,last_name
+remote,1,,john.doe1@gmail.com,John,Doe1
+remote,2,john.doe2,,John,Doe2
+remote,3,,,John,Doe3
+remote,4,,,John,Doe4
+remote,5,,john.doe1@gmail.com,,
+remote,6,john.doe2,,,
+remote,7,,john.doe@gmail.com,,
+remote,8,john.doe,,,
+'''
+    for i, line in enumerate(content.splitlines()):
+        print(i, line.count(','))
+    importer = user_csv_importer_factory(content)
+    importer.run()
+    assert importer.has_errors
+    assert not importer.errors
+    assert importer.headers_by_name['username'].unique
+    assert importer.headers_by_name['username'].globally_unique
+    assert importer.headers_by_name['email'].unique
+    assert not importer.headers_by_name['email'].globally_unique
+    row_errors = {row.line: [error.description for error in row.errors] for row in importer.rows if row.errors}
+    assert row_errors == {
+        6: ['Unique constraint on column "email" failed: value already appear on line 2',
+            'Unique constraint on column "email" failed'],
+        7: ['Unique constraint on column "username" failed: value already appear on line 3'],
+        9: ['Unique constraint on column "username" failed']
+    }
+    cell_errors = {
+        row.line: {
+            cell.header.name: [error for error in cell.errors] for cell in row.cells if cell.errors
+        }
+        for row in importer.rows if any(cell.errors for cell in row.cells)
+    }
+    assert cell_errors == {}
-- 
2.27.0