def test_cluster_same_feed(self):
article = ArticleController().read(category_id__ne=None).first()
cluster = article.cluster
# all is enabled, article in cluster
update_on_all_objs(articles=cluster.articles,
cluster_enabled=True,
cluster_same_feed=True)
article = self.create_article_from(cluster, cluster.main_article.feed)
self.assertInCluster(article, cluster)
# feed's disabled, won't cluster
FeedController().update(
{'id__in': [a.feed_id for a in cluster.articles]},
{'cluster_same_feed': False})
article = self.create_article_from(cluster, cluster.main_article.feed)
self.assertNotInCluster(article, cluster)
# category's disabled, won't cluster
FeedController().update(
{'id__in': [a.feed_id for a in cluster.articles]},
{'cluster_same_feed': None})
CategoryController().update({'id': cluster.main_article.category.id},
{'cluster_same_feed': False})
article = self.create_article_from(cluster, cluster.main_article.feed)
self.assertNotInCluster(article, cluster)
# user's disable, won't cluster
CategoryController().update({'id': cluster.main_article.category.id},
{'cluster_same_feed': None})
UserController().update({'id': cluster.user_id},
{'cluster_same_feed': False})
article = self.create_article_from(cluster, cluster.main_article.feed)
self.assertNotInCluster(article, cluster)
# reenabling user, will cluster
UserController().update({'id': cluster.user_id},
{'cluster_same_feed': True})
article = self.create_article_from(cluster, cluster.main_article.feed)
self.assertInCluster(article, cluster)
def test_delete_main_cluster_handling(self):
suffix = 'suffix'
clu = ClusterController().get(id=10)
acontr = ArticleController(clu.user_id)
fcontr = FeedController(clu.user_id)
old_title = clu.main_title
old_feed_title, old_art_id = clu.main_feed_title, clu.main_article_id
for art_to_del in acontr.read(link=clu.main_article.link,
id__ne=clu.main_article.id):
acontr.delete(art_to_del.id)
other_feed = fcontr.read(id__ne=clu.main_article.feed_id).first()
update_on_all_objs(articles=[clu.main_article],
feeds=[other_feed],
cluster_enabled=True)
acontr.create(
feed_id=other_feed.id,
entry_id=clu.main_article.entry_id + suffix,
link=clu.main_article.link,
title=clu.main_article.title + suffix,
content=clu.main_article.content + suffix,
date=clu.main_article.date + timedelta(1),
retrieved_date=clu.main_article.retrieved_date + timedelta(1),
)
ClusterController(clu.user_id).clusterize_pending_articles()
clu = ClusterController().get(id=10)
self.assertEqual(2, len(clu.articles))
fcontr.delete(clu.main_article.feed_id)
new_cluster = ClusterController(clu.user_id).get(id=clu.id)
self.assertEqual(1, len(new_cluster.articles))
self.assertNotEqual(old_title, new_cluster.main_title)
self.assertNotEqual(old_feed_title, new_cluster.main_feed_title)
self.assertNotEqual(old_art_id, new_cluster.main_article_id)
def test_adding_to_cluster_by_link(self):
ccontr = ClusterController()
cluster = ccontr.read().first()
ccontr.update({'id': cluster.id}, {
'read': True,
'read_reason': 'marked'
})
cluster = ccontr.get(id=cluster.id)
self.assertTrue(cluster.read)
article = cluster.articles[0]
articles_count = len(cluster.articles)
fcontr = FeedController(cluster.user_id)
acontr = ArticleController(cluster.user_id)
fcontr.update({'id': article.feed_id}, {'cluster_wake_up': True})
feed = fcontr.read(id__ne=article.feed_id).first()
update_on_all_objs(articles=[article],
feeds=[feed],
cluster_enabled=True)
self._clone_article(acontr, article, feed)
ccontr.clusterize_pending_articles()
cluster = ccontr.get(id=cluster.id)
self.assertEqual(articles_count + 1, len(cluster.articles))
self.assertFalse(cluster.read)
def test_cluster_enabled(self):
ccontr = ClusterController()
cluster = ccontr.read().first()
feed = FeedController(cluster.user_id).read(
category_id__ne=None,
id__nin=[art.feed_id for art in cluster.articles]).first()
category = feed.category
# clustering works when all is true
update_on_all_objs(articles=cluster.articles,
feeds=[feed],
cluster_enabled=True)
article = self.create_article_from(cluster, feed)
self.assertInCluster(article, cluster)
# disabling on user desactivate all clustering by default
update_on_all_objs(articles=cluster.articles,
feeds=[feed],
cluster_enabled=None)
UserController().update({'id': cluster.user_id},
{'cluster_enabled': False})
article = self.create_article_from(cluster, feed)
self.assertNotInCluster(article, cluster)
# disabling on article's feed prevents from clustering
update_on_all_objs(articles=cluster.articles,
feeds=[feed],
cluster_enabled=True)
FeedController().update({'id': feed.id}, {'cluster_enabled': False})
article = self.create_article_from(cluster, feed)
self.assertNotInCluster(article, cluster)
# disabling on feed from cluster's articles prevents from clustering
update_on_all_objs(articles=cluster.articles,
feeds=[feed],
cluster_enabled=True)
FeedController().update(
{'id__in': [a.feed_id for a in cluster.articles]},
{'cluster_enabled': False})
article = self.create_article_from(cluster, feed)
self.assertNotInCluster(article, cluster)
# disabling on article's category prevents from clustering
CategoryController(cluster.user_id).update({'id': category.id},
{'cluster_enabled': False})
article = self.create_article_from(cluster, feed)
self.assertNotInCluster(article, cluster)
update_on_all_objs(articles=cluster.articles,
feeds=[feed],
cluster_enabled=True)
article = self.create_article_from(cluster, feed)
self.assertInCluster(article, cluster)
def test_MarkClustersAsRead_put_only_singles(self):
feed = FeedController(self.user.id).read()[0]
update_on_all_objs(feeds=[feed],
cluster_same_feed=True,
cluster_enabled=True)
# creating a new article that will cluster
ArticleController(self.user.id).create(entry_id='new entry_id',
title='new title',
content='new content',
feed_id=feed.id,
link=feed.articles[0].link)
ClusterController(self.user.id).clusterize_pending_articles()
self.assertClusterCount(18, {'filter': 'unread'})
# one per feed
self._mark_as_read(2, {'only_singles': True, 'filter': 'unread'})
self.assertClusterCount(1, {'filter': 'unread'})
def test_cluster_tfidf_control(self):
article = ArticleController().read(category_id__ne=None).first()
cluster = article.cluster
# leaving one cluster with one article
clu_ids = [c.id for c in ClusterController().read(id__ne=cluster.id)]
art_ids = [
a.id
for a in ArticleController().read(id__ne=cluster.main_article_id)
]
ArticleController().update({'id__in': art_ids}, {'cluster_id': None})
for clu_id in clu_ids:
ClusterController().delete(clu_id)
for art_id in art_ids:
ArticleController().delete(art_id)
self.assertEqual(1, ClusterController().read().count())
self.assertEqual(1, ArticleController().read().count())
feed1 = FeedController(cluster.user_id).create(
title='new feed',
cluster_conf={
'tfidf_min_score': -1,
'tfidf_min_sample_size': 1
})
update_on_all_objs(articles=cluster.articles,
feeds=[feed1],
cluster_tfidf_enabled=True,
cluster_enabled=True)
feed2 = FeedController(cluster.user_id).create(
cluster_enabled=True,
cluster_tfidf_enabled=False,
title='new feed',
cluster_conf={
'tfidf_min_score': -1,
'tfidf_min_sample_size': 1
})
article = self.create_article_from(cluster,
feed1,
link=cluster.main_article.link +
'do not match link')
self.assertInCluster(article, cluster, ClusterReason.tf_idf)
article = self.create_article_from(cluster,
feed2,
link=cluster.main_article.link +
'do not match link either')
self.assertNotInCluster(article, cluster)
def test_no_cluster_same_category_on_original_category(self):
article = ArticleController().read(category_id__ne=None).first()
art_cat_id = article.category_id
cat_ctrl = CategoryController(article.user_id)
cluster = article.cluster
feed = FeedController(cluster.user_id).create(title='new feed',
category_id=art_cat_id)
update_on_all_objs(articles=cluster.articles,
feeds=[feed],
cluster_same_category=None,
cluster_enabled=True)
cat_ctrl.update({'id': art_cat_id}, {'cluster_same_category': False})
article = self.create_article_from(cluster, feed)
self.assertNotInCluster(article, cluster)
cat_ctrl.update({'id': art_cat_id}, {'cluster_same_category': True})
article = self.create_article_from(cluster, feed)
self.assertInCluster(article, cluster)
def test_cluster_disabled_on_original_category(self):
article = ArticleController().read(category_id__ne=None).first()
art_cat_id = article.category_id
cat_ctrl = CategoryController(article.user_id)
cluster = article.cluster
fctrl = FeedController(cluster.user_id)
feed = fctrl.create(title='new feed', category_id=art_cat_id)
fno_cat = fctrl.create(title='category-less')
update_on_all_objs(users=[cluster.user], cluster_enabled=None)
cat_ctrl.update({}, {'cluster_enabled': False})
article = self.create_article_from(cluster, feed)
self.assertEqual(1, len(article.cluster.articles))
self.assertNotInCluster(article, cluster)
article = self.create_article_from(cluster, fno_cat)
self.assertEqual(1, len(article.cluster.articles))
self.assertNotInCluster(article, cluster)
cat_ctrl.update({'id': art_cat_id}, {'cluster_enabled': True})
article = self.create_article_from(cluster, fno_cat)
self.assertEqual(2, len(article.cluster.articles))
self.assertInCluster(article, cluster)
article = self.create_article_from(cluster, feed)
self.assertEqual(3, len(article.cluster.articles))
self.assertInCluster(article, cluster)
