def load_topics(cli, size, fake, *_, **__):
"""
Load topics
"""
nb_topics = size * 10
cli.stdout.write(f"Nombres de Topics à créer : {nb_topics}")
tps1 = time.time()
nb_forums = Forum.objects.count()
if nb_forums == 0:
cli.stdout.write(
"Il n'y a aucun forum actuellement. " "Vous devez rajouter les forums dans vos fixtures (forum)"
)
return
forums = list(Forum.objects.all())
nb_users = User.objects.count()
if nb_users == 0:
cli.stdout.write(
"Il n'y a aucun membre actuellement. " "Vous devez rajouter les membres dans vos fixtures (member)"
)
return
profiles = list(Profile.objects.all())
nb_tags = Tag.objects.count()
if nb_tags == 0:
cli.stdout.write("Il n'y a aucun tag actuellement. " "Vous devez rajouter les tags dans vos fixtures (tag)")
return
for i in range(0, nb_topics):
with contextlib.suppress(IntegrityError):
topic = TopicFactory(forum=forums[i % nb_forums], author=profiles[i % nb_users].user)
topic.solved_by = profiles[i % nb_users].user if i % 5 else None
topic.is_locked = i % 10 == 0
topic.is_sticky = i % 15 == 0
nb_rand_tags = random.randint(0, 5)
add_generated_tags_to_topic(nb_rand_tags, nb_tags, topic)
topic.title = fake.text(max_nb_chars=80)
topic.subtitle = fake.text(max_nb_chars=200)
topic.save()
PostFactory(topic=topic, author=topic.author, position=1)
sys.stdout.write(f" Topic {i + 1}/{nb_topics} \r")
sys.stdout.flush()
tps2 = time.time()
cli.stdout.write(f"\nFait en {tps2 - tps1} sec")
def test_upercase_and_lowercase_search_give_same_results(self):
"""Pretty self-explanatory function name, isn't it ?"""
if not self.manager.connected_to_es:
return
# 1. Index lowercase stuffs
text_lc = "test"
topic_1_lc = TopicFactory(forum=self.forum,
author=self.user,
title=text_lc)
tag_lc = TagFactory(title=text_lc)
topic_1_lc.tags.add(tag_lc)
topic_1_lc.subtitle = text_lc
topic_1_lc.save()
post_1_lc = PostFactory(topic=topic_1_lc, author=self.user, position=1)
post_1_lc.text = post_1_lc.text_html = text_lc
post_1_lc.save()
tuto_lc = PublishableContentFactory(type="TUTORIAL")
tuto_draft_lc = tuto_lc.load_version()
tuto_lc.title = text_lc
tuto_lc.authors.add(self.user)
subcategory_lc = SubCategoryFactory(title=text_lc)
tuto_lc.subcategory.add(subcategory_lc)
tuto_lc.tags.add(tag_lc)
tuto_lc.save()
tuto_draft_lc.description = text_lc
tuto_draft_lc.repo_update_top_container(text_lc, tuto_lc.slug, text_lc,
text_lc)
chapter1_lc = ContainerFactory(parent=tuto_draft_lc, db_object=tuto_lc)
extract_lc = ExtractFactory(container=chapter1_lc, db_object=tuto_lc)
extract_lc.repo_update(text_lc, text_lc)
published_lc = publish_content(tuto_lc,
tuto_draft_lc,
is_major_update=True)
tuto_lc.sha_public = tuto_draft_lc.current_version
tuto_lc.sha_draft = tuto_draft_lc.current_version
tuto_lc.public_version = published_lc
tuto_lc.save()
# 2. Index uppercase stuffs
text_uc = "TEST"
topic_1_uc = TopicFactory(forum=self.forum,
author=self.user,
title=text_uc)
topic_1_uc.tags.add(
tag_lc) # Note: a constraint forces tags title to be unique
topic_1_uc.subtitle = text_uc
topic_1_uc.save()
post_1_uc = PostFactory(topic=topic_1_uc, author=self.user, position=1)
post_1_uc.text = post_1_uc.text_html = text_uc
post_1_uc.save()
tuto_uc = PublishableContentFactory(type="TUTORIAL")
tuto_draft_uc = tuto_uc.load_version()
tuto_uc.title = text_uc
tuto_uc.authors.add(self.user)
tuto_uc.subcategory.add(subcategory_lc)
tuto_uc.tags.add(tag_lc)
tuto_uc.save()
tuto_draft_uc.description = text_uc
tuto_draft_uc.repo_update_top_container(text_uc, tuto_uc.slug, text_uc,
text_uc)
chapter1_uc = ContainerFactory(parent=tuto_draft_uc, db_object=tuto_uc)
extract_uc = ExtractFactory(container=chapter1_uc, db_object=tuto_uc)
extract_uc.repo_update(text_uc, text_uc)
published_uc = publish_content(tuto_uc,
tuto_draft_uc,
is_major_update=True)
tuto_uc.sha_public = tuto_draft_uc.current_version
tuto_uc.sha_draft = tuto_draft_uc.current_version
tuto_uc.public_version = published_uc
tuto_uc.save()
# 3. Index and search:
self.assertEqual(
len(
self.manager.setup_search(Search().query(
MatchAll())).execute()), 0)
# index
for model in self.indexable:
if model is FakeChapter:
continue
self.manager.es_bulk_indexing_of_model(model)
self.manager.refresh_index()
result = self.client.get(reverse("search:query") + "?q=" + text_lc,
follow=False)
self.assertEqual(result.status_code, 200)
response_lc = result.context["object_list"].execute()
self.assertEqual(response_lc.hits.total, 8)
result = self.client.get(reverse("search:query") + "?q=" + text_uc,
follow=False)
self.assertEqual(result.status_code, 200)
response_uc = result.context["object_list"].execute()
self.assertEqual(response_uc.hits.total, 8)
for responses in zip(
response_lc,
response_uc): # we should get results in the same order!
self.assertEqual(responses[0].meta.id, responses[1].meta.id)
def test_boosts(self):
"""Check if boosts are doing their job"""
if not self.manager.connected_to_es:
return
# 1. Create topics (with identical titles), posts (with identical texts), an article and a tuto
text = "test"
topic_1_solved_sticky = TopicFactory(forum=self.forum,
author=self.user)
topic_1_solved_sticky.title = text
topic_1_solved_sticky.subtitle = ""
topic_1_solved_sticky.solved_by = self.user
topic_1_solved_sticky.is_sticky = True
topic_1_solved_sticky.save()
post_1 = PostFactory(topic=topic_1_solved_sticky,
author=self.user,
position=1)
post_1.text = post_1.text_html = text
post_1.save()
post_2_useful = PostFactory(topic=topic_1_solved_sticky,
author=self.user,
position=2)
post_2_useful.text = post_2_useful.text_html = text
post_2_useful.is_useful = True
post_2_useful.like = 5
post_2_useful.dislike = 2 # l/d ratio above 1
post_2_useful.save()
topic_2_locked = TopicFactory(forum=self.forum,
author=self.user,
title=text)
topic_2_locked.title = text
topic_2_locked.subtitle = ""
topic_2_locked.is_locked = True
topic_2_locked.save()
post_3_ld_below_1 = PostFactory(topic=topic_2_locked,
author=self.user,
position=1)
post_3_ld_below_1.text = post_3_ld_below_1.text_html = text
post_3_ld_below_1.like = 2
post_3_ld_below_1.dislike = 5 # l/d ratio below 1
post_3_ld_below_1.save()
tuto = PublishableContentFactory(type="TUTORIAL")
tuto_draft = tuto.load_version()
tuto.title = text
tuto.authors.add(self.user)
tuto.save()
tuto_draft.repo_update_top_container(text, tuto.slug, text, text)
chapter1 = ContainerFactory(parent=tuto_draft, db_object=tuto)
chapter1.repo_update(text, "Who cares ?", "Same here")
ExtractFactory(container=chapter1, db_object=tuto)
published_tuto = publish_content(tuto,
tuto_draft,
is_major_update=True)
tuto.sha_public = tuto_draft.current_version
tuto.sha_draft = tuto_draft.current_version
tuto.public_version = published_tuto
tuto.save()
article = PublishedContentFactory(type="ARTICLE", title=text)
published_article = PublishedContent.objects.get(content_pk=article.pk)
opinion_not_picked = PublishedContentFactory(type="OPINION",
title=text)
published_opinion_not_picked = PublishedContent.objects.get(
content_pk=opinion_not_picked.pk)
opinion_picked = PublishedContentFactory(type="OPINION", title=text)
opinion_picked.sha_picked = opinion_picked.sha_draft
opinion_picked.date_picked = datetime.datetime.now()
opinion_picked.save()
published_opinion_picked = PublishedContent.objects.get(
content_pk=opinion_picked.pk)
for model in self.indexable:
if model is FakeChapter:
continue
self.manager.es_bulk_indexing_of_model(model)
self.manager.refresh_index()
self.assertEqual(
len(
self.manager.setup_search(Search().query(
MatchAll())).execute()), 10)
# 2. Reset all boosts to 1
for doc_type in settings.ZDS_APP["search"]["boosts"]:
for key in settings.ZDS_APP["search"]["boosts"][doc_type]:
settings.ZDS_APP["search"]["boosts"][doc_type][key] = 1.0
# 3. Test posts
result = self.client.get(reverse("search:query") + "?q=" + text +
"&models=" + Post.get_es_document_type(),
follow=False)
self.assertEqual(result.status_code, 200)
response = result.context["object_list"].execute()
self.assertEqual(response.hits.total, 3)
# score are equals without boost:
self.assertTrue(response[0].meta.score == response[1].meta.score ==
response[2].meta.score)
settings.ZDS_APP["search"]["boosts"]["post"]["if_first"] = 2.0
result = self.client.get(reverse("search:query") + "?q=" + text +
"&models=" + Post.get_es_document_type(),
follow=False)
self.assertEqual(result.status_code, 200)
response = result.context["object_list"].execute()
self.assertEqual(response.hits.total, 3)
self.assertTrue(response[0].meta.score == response[1].meta.score >
response[2].meta.score)
self.assertEqual(response[2].meta.id, str(
post_2_useful.pk)) # post 2 is the only one not first
settings.ZDS_APP["search"]["boosts"]["post"]["if_first"] = 1.0
settings.ZDS_APP["search"]["boosts"]["post"]["if_useful"] = 2.0
result = self.client.get(reverse("search:query") + "?q=" + text +
"&models=" + Post.get_es_document_type(),
follow=False)
self.assertEqual(result.status_code, 200)
response = result.context["object_list"].execute()
self.assertEqual(response.hits.total, 3)
self.assertTrue(response[0].meta.score > response[1].meta.score ==
response[2].meta.score)
self.assertEqual(response[0].meta.id,
str(post_2_useful.pk)) # post 2 is useful
settings.ZDS_APP["search"]["boosts"]["post"]["if_useful"] = 1.0
settings.ZDS_APP["search"]["boosts"]["post"]["ld_ratio_above_1"] = 2.0
result = self.client.get(reverse("search:query") + "?q=" + text +
"&models=" + Post.get_es_document_type(),
follow=False)
self.assertEqual(result.status_code, 200)
response = result.context["object_list"].execute()
self.assertEqual(response.hits.total, 3)
self.assertTrue(response[0].meta.score > response[1].meta.score ==
response[2].meta.score)
self.assertEqual(response[0].meta.id, str(
post_2_useful.pk)) # post 2 have a l/d ratio of 5/2
settings.ZDS_APP["search"]["boosts"]["post"]["ld_ratio_above_1"] = 1.0
settings.ZDS_APP["search"]["boosts"]["post"][
"ld_ratio_below_1"] = 2.0 # no one would do that in real life
result = self.client.get(reverse("search:query") + "?q=" + text +
"&models=" + Post.get_es_document_type(),
follow=False)
self.assertEqual(result.status_code, 200)
response = result.context["object_list"].execute()
self.assertEqual(response.hits.total, 3)
self.assertTrue(response[0].meta.score > response[1].meta.score ==
response[2].meta.score)
self.assertEqual(response[0].meta.id, str(
post_3_ld_below_1.pk)) # post 3 have a l/d ratio of 2/5
settings.ZDS_APP["search"]["boosts"]["post"]["ld_ratio_below_1"] = 1.0
# 4. Test topics
result = self.client.get(reverse("search:query") + "?q=" + text +
"&models=" + Topic.get_es_document_type(),
follow=False)
self.assertEqual(result.status_code, 200)
response = result.context["object_list"].execute()
self.assertEqual(response.hits.total, 2)
# score are equals without boost:
self.assertTrue(response[0].meta.score == response[1].meta.score)
settings.ZDS_APP["search"]["boosts"]["topic"]["if_sticky"] = 2.0
result = self.client.get(reverse("search:query") + "?q=" + text +
"&models=" + Topic.get_es_document_type(),
follow=False)
self.assertEqual(result.status_code, 200)
response = result.context["object_list"].execute()
self.assertEqual(response.hits.total, 2)
self.assertTrue(response[0].meta.score > response[1].meta.score)
self.assertEqual(response[0].meta.id,
str(topic_1_solved_sticky.pk)) # topic 1 is sticky
settings.ZDS_APP["search"]["boosts"]["topic"]["if_sticky"] = 1.0
settings.ZDS_APP["search"]["boosts"]["topic"]["if_solved"] = 2.0
result = self.client.get(reverse("search:query") + "?q=" + text +
"&models=" + Topic.get_es_document_type(),
follow=False)
self.assertEqual(result.status_code, 200)
response = result.context["object_list"].execute()
self.assertEqual(response.hits.total, 2)
self.assertTrue(response[0].meta.score > response[1].meta.score)
self.assertEqual(response[0].meta.id,
str(topic_1_solved_sticky.pk)) # topic 1 is solved
settings.ZDS_APP["search"]["boosts"]["topic"]["if_solved"] = 1.0
settings.ZDS_APP["search"]["boosts"]["topic"][
"if_locked"] = 2.0 # no one would do that in real life
result = self.client.get(reverse("search:query") + "?q=" + text +
"&models=" + Topic.get_es_document_type(),
follow=False)
self.assertEqual(result.status_code, 200)
response = result.context["object_list"].execute()
self.assertEqual(response.hits.total, 2)
self.assertTrue(response[0].meta.score > response[1].meta.score)
self.assertEqual(response[0].meta.id,
str(topic_2_locked.pk)) # topic 2 is locked
settings.ZDS_APP["search"]["boosts"]["topic"][
"if_locked"] = 1.0 # no one would do that in real life
# 5. Test published contents
result = self.client.get(reverse("search:query") + "?q=" + text +
"&models=content",
follow=False)
self.assertEqual(result.status_code, 200)
response = result.context["object_list"].execute()
self.assertEqual(response.hits.total, 5)
# score are equals without boost:
self.assertTrue(
response[0].meta.score == response[1].meta.score == response[2].
meta.score == response[3].meta.score == response[4].meta.score)
settings.ZDS_APP["search"]["boosts"]["publishedcontent"][
"if_article"] = 2.0
result = self.client.get(reverse("search:query") + "?q=" + text +
"&models=content",
follow=False)
self.assertEqual(result.status_code, 200)
response = result.context["object_list"].execute()
self.assertEqual(response.hits.total, 5)
self.assertTrue(response[0].meta.score > response[1].meta.score)
self.assertEqual(response[0].meta.id,
str(published_article.pk)) # obvious
settings.ZDS_APP["search"]["boosts"]["publishedcontent"][
"if_article"] = 1.0
settings.ZDS_APP["search"]["boosts"]["publishedcontent"][
"if_tutorial"] = 2.0
result = self.client.get(reverse("search:query") + "?q=" + text +
"&models=content",
follow=False)
self.assertEqual(result.status_code, 200)
response = result.context["object_list"].execute()
self.assertEqual(response.hits.total, 5)
self.assertTrue(response[0].meta.score > response[1].meta.score)
self.assertEqual(response[0].meta.id,
str(published_tuto.pk)) # obvious
settings.ZDS_APP["search"]["boosts"]["publishedcontent"][
"if_tutorial"] = 1.0
settings.ZDS_APP["search"]["boosts"]["publishedcontent"][
"if_opinion"] = 2.0
settings.ZDS_APP["search"]["boosts"]["publishedcontent"][
"if_opinion_not_picked"] = 4.0
# Note: in "real life", unpicked opinion would get a boost < 1.
result = self.client.get(reverse("search:query") + "?q=" + text +
"&models=content",
follow=False)
self.assertEqual(result.status_code, 200)
response = result.context["object_list"].execute()
self.assertEqual(response.hits.total, 5)
self.assertTrue(response[0].meta.score > response[1].meta.score >
response[2].meta.score)
self.assertEqual(
response[0].meta.id,
str(published_opinion_not_picked.pk)) # unpicked opinion got first
self.assertEqual(response[1].meta.id, str(published_opinion_picked.pk))
settings.ZDS_APP["search"]["boosts"]["publishedcontent"][
"if_opinion"] = 1.0
settings.ZDS_APP["search"]["boosts"]["publishedcontent"][
"if_opinion_not_picked"] = 1.0
settings.ZDS_APP["search"]["boosts"]["publishedcontent"][
"if_medium_or_big_tutorial"] = 2.0
result = self.client.get(reverse("search:query") + "?q=" + text +
"&models=content",
follow=False)
self.assertEqual(result.status_code, 200)
response = result.context["object_list"].execute()
self.assertEqual(response.hits.total, 5)
self.assertTrue(response[0].meta.score > response[1].meta.score)
self.assertEqual(response[0].meta.id,
str(published_tuto.pk)) # obvious
settings.ZDS_APP["search"]["boosts"]["publishedcontent"][
"if_medium_or_big_tutorial"] = 1.0
# 6. Test global boosts
# NOTE: score are NOT the same for all documents, no matter how hard it tries to, small differences exists
for model in self.indexable:
# set a huge number to overcome the small differences:
settings.ZDS_APP["search"]["boosts"][
model.get_es_document_type()]["global"] = 10.0
result = self.client.get(reverse("search:query") + "?q=" + text,
follow=False)
self.assertEqual(result.status_code, 200)
response = result.context["object_list"].execute()
self.assertEqual(response.hits.total, 10)
self.assertEqual(response[0].meta.doc_type,
model.get_es_document_type()) # obvious
settings.ZDS_APP["search"]["boosts"][
model.get_es_document_type()]["global"] = 1.0