adding test_keep_largest_overlapped_keywords

meedan · Aug 21, 2024 · 255fe2d · 255fe2d
1 parent 6602091
commit 255fe2d
Showing 1 changed file with 5 additions and 2 deletions.
diff --git a/test/lib/model/test_yake_keywords.py b/test/lib/model/test_yake_keywords.py
@@ -42,7 +42,10 @@ def test_run_yake_real(self):
         })
         results = self.yake_model.run_yake(**self.yake_model.get_params(message))
         self.assertEqual(results, {"keywords": [('love Meedan', 0.0013670273525686505)]})
-
+    def test_keep_largest_overlapped_keywords(self):
+        keywords_test = [('Alegre',0), ('AlegreAlegre', 0), ('Alegre Alegre', 0), ("Presto", 0)]
+        expected = [('AlegreAlegre', 0), ('Alegre Alegre', 0), ('Presto', 0)]
+        self.assertEqual(self.yake_model.keep_largest_overlapped_keywords(keywords_test), expected)
     def test_get_params_with_defaults(self):
         message = schemas.parse_message({
             "body": {
@@ -51,7 +54,7 @@ def test_get_params_with_defaults(self):
             },
             "model_name": "yake_keywords__Model"
         })
-        expected = {'text': 'Some Text', 'language': "en", 'max_ngram_size': 3, 'deduplication_threshold': 0.25, 'deduplication_algo': 'seqm', 'window_size': 0, 'num_of_keywords': 10}
+        expected = {'text': 'Some Text', 'language': "auto", 'max_ngram_size': 3, 'deduplication_threshold': 0.25, 'deduplication_algo': 'seqm', 'window_size': 0, 'num_of_keywords': 10}
         self.assertEqual(self.yake_model.get_params(message), expected)
 
     def test_get_params_with_specifics(self):