fix on cluster handling

src/tests/controllers/cluster_test.py

@@ -103,6 +103,7 @@ def gen_articles(factor):
                     Mock(valuable_tokens=['Sarkozy', 'garage', 'chanson'])] \
                             * factor
         ccontr = ClusterController()
+        ccontr.tfidf_min_score = 0.6
 
         acontr_cls.return_value.read.return_value = gen_articles(2)
 

src/web/controllers/article.py

@@ -97,17 +97,32 @@ def remove_from_cluster(self, article):
         Return True if the article is deleted at the end or not
         """
         from web.controllers import ClusterController
+        if not article.cluster_id:
+            return
         clu_ctrl = ClusterController(self.user_id)
-        cluster = clu_ctrl.read(id=article.cluster_id,
-                                main_article_id=article.id).first()
+        cluster = clu_ctrl.read(id=article.cluster_id).first()
         if not cluster:
-            return True
-        if len(cluster.articles) == 1:  # only on article in cluster, deleting
-            clu_ctrl.delete(cluster.id)
-            return False
-        clu_ctrl._enrich_cluster(cluster, cluster.articles[1],
-                                 cluster.read, cluster.liked, True)
-        return True
+            return
+
+        try:
+            new_art = next(new_art for new_art in cluster.articles
+                           if new_art.id != article.id)
+        except StopIteration:
+            # only on article in cluster, deleting cluster
+            clu_ctrl.delete(cluster.id, delete_articles=False)
+        else:
+            if cluster.main_article_id == article.id:
+                cluster.main_article_id = None
+                clu_ctrl._enrich_cluster(cluster, new_art,
+                                         cluster.read, cluster.liked,
+                                         force_article_as_main=True)
+        self.update({'id': article.id},
+                    {'cluster_id': None,
+                     'cluster_reason': None,
+                     'cluster_score': None,
+                     'cluster_tfidf_with': None,
+                     'cluster_tfidf_neighbor_size': None})
+        return
 
     def _delete(self, article, commit):
         Tag.query.filter(Tag.article_id == article.id).delete()
@@ -119,7 +134,5 @@ def _delete(self, article, commit):
 
     def delete(self, obj_id, commit=True):
         article = self.get(id=obj_id)
-        still_delete_article = self.remove_from_cluster(article)
-        if still_delete_article:
-            return self._delete(article, commit=commit)
-        return article
+        self.remove_from_cluster(article)
+        return self._delete(article, commit=commit)

src/web/controllers/cluster.py

@@ -29,6 +29,10 @@ class ClusterController(AbstractController):
     _db_cls = Cluster
     max_day_dist = timedelta(days=7)
 
+    def __init__(self, *args, **kwargs):
+        super().__init__(*args, **kwargs)
+        self.tfidf_min_score = MIN_SIMILARITY_SCORE
+
     def _get_cluster_by_link(self, article):
         cluster = self.read(user_id=article.user_id,
                          main_date__lt=article.date + self.max_day_dist,
@@ -73,7 +77,7 @@ def _get_cluster_by_similarity(self, article, min_sample_size=10):
             return
 
         best_match, score = get_best_match_and_score(article, neighbors)
-        if score > MIN_SIMILARITY_SCORE:
+        if score > self.tfidf_min_score:
             article.cluster_reason = ClusterReason.tf_idf
             article.cluster_score = int(score * 1000)
             article.cluster_tfidf_neighbor_size = len(neighbors)
@@ -263,12 +267,15 @@ def join_read(self, feed_id=None, **filters):
                      .order_by(Cluster.main_date.desc()).limit(JR_LENGTH),
                 filter_on_cat)
 
-    def delete(self, obj_id):
+    def delete(self, obj_id, delete_articles=True):
         from web.controllers import ArticleController
         self.update({'id': obj_id}, {'main_article_id': None}, commit=False)
         actrl = ArticleController(self.user_id)
-        for art in actrl.read(cluster_id=obj_id):
-            actrl._delete(art, commit=False)
+        if delete_articles:
+            for art in actrl.read(cluster_id=obj_id):
+                actrl._delete(art, commit=False)
+        else:
+            actrl.update({'cluster_id': obj_id}, {'cluster_id': None})
         return super().delete(obj_id)
 
     #