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_change_publishedcontents_impacts_chapter(self):
if not self.manager.connected_to_es:
return
# 1. Create middle-size content and index it
text = "test"
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) # change title to be sure it will match
chapter1 = ContainerFactory(parent=tuto_draft, db_object=tuto)
chapter1.repo_update(text, text, text)
extract = ExtractFactory(container=chapter1, db_object=tuto)
extract.repo_update(text, text)
published = 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.save()
self.manager.es_bulk_indexing_of_model(PublishedContent)
self.manager.refresh_index()
self.assertEqual(
len(
self.manager.setup_search(Search().query(
MatchAll())).execute()), 2) # indexing ok
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, 2)
chapters = [r for r in response if r.meta.doc_type == "chapter"]
self.assertEqual(chapters[0].meta.doc_type,
FakeChapter.get_es_document_type())
self.assertEqual(chapters[0].meta.id,
published.content_public_slug + "__" + chapter1.slug)
# 2. Change tuto: delete chapter and insert new one !
tuto = PublishableContent.objects.get(pk=tuto.pk)
tuto_draft = tuto.load_version()
tuto_draft.children[0].repo_delete() # chapter 1 is gone !
another_text = "another thing"
self.assertTrue(
text not in another_text
) # to prevent a future modification from breaking this test
chapter2 = ContainerFactory(parent=tuto_draft, db_object=tuto)
chapter2.repo_update(another_text, another_text, another_text)
extract2 = ExtractFactory(container=chapter2, db_object=tuto)
extract2.repo_update(another_text, another_text)
published = publish_content(tuto, tuto_draft, is_major_update=False)
tuto.sha_public = tuto_draft.current_version
tuto.sha_draft = tuto_draft.current_version
tuto.public_version = published
tuto.save()
self.manager.es_bulk_indexing_of_model(PublishedContent)
self.manager.refresh_index()
self.assertEqual(
len(
self.manager.setup_search(Search().query(
MatchAll())).execute()), 2) # 2 objects, not 3 !
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()
contents = [r for r in response if r.meta.doc_type != "chapter"]
self.assertEqual(response.hits.total,
len(contents)) # no chapter found anymore
result = self.client.get(reverse("search:query") + "?q=" +
another_text + "&models=content",
follow=False)
self.assertEqual(result.status_code, 200)
response = result.context["object_list"].execute()
chapters = [r for r in response if r.meta.doc_type == "chapter"]
self.assertEqual(response.hits.total, 1)
self.assertEqual(chapters[0].meta.doc_type,
FakeChapter.get_es_document_type())
self.assertEqual(chapters[0].meta.id, published.content_public_slug +
"__" + chapter2.slug) # got new chapter
def test_category_and_subcategory_impact_search(self):
"""If two contents do not belong to the same (sub)category"""
if not self.manager.connected_to_es:
return
text = "Did you ever hear the tragedy of Darth Plagueis The Wise?"
# 1. Create two contents with different subcategories
category_1 = "category 1"
subcategory_1 = SubCategoryFactory(title=category_1)
category_2 = "category 2"
subcategory_2 = SubCategoryFactory(title=category_2)
tuto_1 = PublishableContentFactory(type="TUTORIAL")
tuto_1_draft = tuto_1.load_version()
tuto_1.title = text
tuto_1.authors.add(self.user)
tuto_1.subcategory.add(subcategory_1)
tuto_1.save()
tuto_1_draft.description = text
tuto_1_draft.repo_update_top_container(text, tuto_1.slug, text, text)
chapter_1 = ContainerFactory(parent=tuto_1_draft, db_object=tuto_1)
extract_1 = ExtractFactory(container=chapter_1, db_object=tuto_1)
extract_1.repo_update(text, text)
published_1 = publish_content(tuto_1,
tuto_1_draft,
is_major_update=True)
tuto_1.sha_public = tuto_1_draft.current_version
tuto_1.sha_draft = tuto_1_draft.current_version
tuto_1.public_version = published_1
tuto_1.save()
tuto_2 = PublishableContentFactory(type="TUTORIAL")
tuto_2_draft = tuto_2.load_version()
tuto_2.title = text
tuto_2.authors.add(self.user)
tuto_2.subcategory.add(subcategory_2)
tuto_2.save()
tuto_2_draft.description = text
tuto_2_draft.repo_update_top_container(text, tuto_2.slug, text, text)
chapter_2 = ContainerFactory(parent=tuto_2_draft, db_object=tuto_2)
extract_2 = ExtractFactory(container=chapter_2, db_object=tuto_2)
extract_2.repo_update(text, text)
published_2 = publish_content(tuto_2,
tuto_2_draft,
is_major_update=True)
tuto_2.sha_public = tuto_2_draft.current_version
tuto_2.sha_draft = tuto_2_draft.current_version
tuto_2.public_version = published_2
tuto_2.save()
# 2. Index:
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,
follow=False)
self.assertEqual(result.status_code, 200)
response = result.context["object_list"].execute()
self.assertEqual(response.hits.total, 4) # Ok
# 3. Test
result = self.client.get(reverse("search:query") + "?q=" + text +
"&model=content&subcategory=" +
subcategory_1.slug,
follow=False)
self.assertEqual(result.status_code, 200)
response = result.context["object_list"].execute()
self.assertEqual(response.hits.total, 2)
self.assertEqual([
int(r.meta.id) for r in response
if r.meta.doc_type == "publishedcontent"
][0], published_1.pk)
self.assertEqual([
r.meta.id for r in response if r.meta.doc_type == "chapter"
][0], tuto_1.slug + "__" + chapter_1.slug)
result = self.client.get(reverse("search:query") + "?q=" + text +
"&model=content&subcategory=" +
subcategory_2.slug,
follow=False)
self.assertEqual(result.status_code, 200)
response = result.context["object_list"].execute()
self.assertEqual(response.hits.total, 2)
self.assertEqual([
int(r.meta.id) for r in response
if r.meta.doc_type == "publishedcontent"
][0], published_2.pk)
self.assertEqual([
r.meta.id for r in response if r.meta.doc_type == "chapter"
][0], tuto_2.slug + "__" + chapter_2.slug)
def test_basic_search(self):
"""Basic search and filtering"""
if not self.manager.connected_to_es:
return
# 1. Index and test search:
text = "test"
topic_1 = TopicFactory(forum=self.forum, author=self.user, title=text)
post_1 = PostFactory(topic=topic_1, author=self.user, position=1)
post_1.text = post_1.text_html = text
post_1.save()
# create a middle-size content and publish it
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) # change title to be sure it will match
chapter1 = ContainerFactory(parent=tuto_draft, db_object=tuto)
extract = ExtractFactory(container=chapter1, db_object=tuto)
extract.repo_update(text, text)
published = 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.save()
# nothing has been indexed yet:
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,
follow=False)
self.assertEqual(result.status_code, 200)
response = result.context["object_list"].execute()
self.assertEqual(response.hits.total, 4) # get 4 results
# 2. Test filtering:
topic_1 = Topic.objects.get(pk=topic_1.pk)
post_1 = Post.objects.get(pk=post_1.pk)
published = PublishedContent.objects.get(pk=published.pk)
ids = {
"topic": [topic_1.es_id],
"post": [post_1.es_id],
"content": [
published.es_id,
published.content_public_slug + "__" + chapter1.slug
],
}
search_groups = [
k for k, v in settings.ZDS_APP["search"]["search_groups"].items()
]
group_to_model = {
k: v[1]
for k, v in settings.ZDS_APP["search"]["search_groups"].items()
}
for doc_type in search_groups:
result = self.client.get(reverse("search:query") + "?q=" + text +
"&models=" + doc_type,
follow=False)
self.assertEqual(result.status_code, 200)
response = result.context["object_list"].execute()
self.assertEqual(response.hits.total,
len(ids[doc_type])) # get 1 result of each …
for i, r in enumerate(response):
self.assertIn(
r.meta.doc_type,
group_to_model[doc_type]) # … and only of the right type …
self.assertEqual(r.meta.id,
ids[doc_type][i]) # … with the right id !
def test_es_manager(self):
"""Test the behavior of the ``es_manager`` command"""
if not self.index_manager.connected_to_es:
return
# in the beginning: the void
self.assertTrue(self.index_manager.index not in
self.index_manager.es.cat.indices())
text = "Ceci est un texte de test"
# create a topic with a post
topic = TopicFactory(forum=self.forum, author=self.user, title=text)
post = PostFactory(topic=topic, author=self.user, position=1)
post.text = post.text_html = text
post.save()
topic = Topic.objects.get(pk=topic.pk)
post = Post.objects.get(pk=post.pk)
self.assertFalse(topic.es_already_indexed)
self.assertTrue(topic.es_flagged)
self.assertFalse(post.es_already_indexed)
self.assertTrue(post.es_flagged)
# create a middle-tutorial and publish it
tuto = PublishableContentFactory(type="TUTORIAL")
tuto.authors.add(self.user)
tuto.save()
tuto_draft = tuto.load_version()
chapter1 = ContainerFactory(parent=tuto_draft, db_object=tuto)
chapter1.repo_update(text, text, text)
extract1 = ExtractFactory(container=chapter1, db_object=tuto)
version = extract1.repo_update(text, text)
published = publish_content(tuto, tuto_draft, is_major_update=True)
tuto.sha_public = version
tuto.sha_draft = version
tuto.public_version = published
tuto.save()
published = PublishedContent.objects.get(content_pk=tuto.pk)
self.assertFalse(published.es_already_indexed)
self.assertTrue(published.es_flagged)
# 1. test "index-all"
call_command("es_manager", "index_all")
self.assertTrue(
self.index_manager.es.indices.exists(self.index_manager.index))
self.index_manager.index_exists = True
topic = Topic.objects.get(pk=topic.pk)
post = Post.objects.get(pk=post.pk)
self.assertTrue(topic.es_already_indexed)
self.assertFalse(topic.es_flagged)
self.assertTrue(post.es_already_indexed)
self.assertFalse(post.es_flagged)
published = PublishedContent.objects.get(content_pk=tuto.pk)
self.assertTrue(published.es_already_indexed)
self.assertFalse(published.es_flagged)
s = Search()
s.query(MatchAll())
results = self.index_manager.setup_search(s).execute()
self.assertEqual(len(results), 4) # get 4 results, one of each type
must_contain = {
"post": False,
"topic": False,
"publishedcontent": False,
"chapter": False
}
id_must_be = {
"post": str(post.pk),
"topic": str(topic.pk),
"publishedcontent": str(published.pk),
"chapter": tuto.slug + "__" + chapter1.slug,
}
for hit in results:
doc_type = hit.meta.doc_type
must_contain[doc_type] = True
self.assertEqual(hit.meta.id, id_must_be[doc_type])
self.assertTrue(all(must_contain))
# 2. test "clear"
self.assertTrue(self.index_manager.index
in self.index_manager.es.cat.indices()) # index in
call_command("es_manager", "clear")
self.assertFalse(
self.index_manager.es.indices.exists(self.index_manager.index))
self.index_manager.index_exists = False
# must reset every object
topic = Topic.objects.get(pk=topic.pk)
post = Post.objects.get(pk=post.pk)
self.assertFalse(topic.es_already_indexed)
self.assertTrue(topic.es_flagged)
self.assertFalse(post.es_already_indexed)
self.assertTrue(post.es_flagged)
published = PublishedContent.objects.get(content_pk=tuto.pk)
self.assertFalse(published.es_already_indexed)
self.assertTrue(published.es_flagged)
self.assertTrue(
self.index_manager.index
not in self.index_manager.es.cat.indices()) # index wiped out !
# 3. test "setup"
call_command("es_manager", "setup")
self.assertTrue(
self.index_manager.es.indices.exists(self.index_manager.index))
self.index_manager.index_exists = True
self.assertTrue(
self.index_manager.index
in self.index_manager.es.cat.indices()) # index back in ...
s = Search()
s.query(MatchAll())
results = self.index_manager.setup_search(s).execute()
self.assertEqual(len(results), 0) # ... but with nothing in it
result = self.index_manager.es.indices.get_settings(
index=self.index_manager.index)
settings_index = result[self.index_manager.index]["settings"]["index"]
self.assertTrue("analysis"
in settings_index) # custom analyzer was setup
# 4. test "index-flagged" once ...
call_command("es_manager", "index_flagged")
topic = Topic.objects.get(pk=topic.pk)
post = Post.objects.get(pk=post.pk)
self.assertTrue(topic.es_already_indexed)
self.assertFalse(topic.es_flagged)
self.assertTrue(post.es_already_indexed)
self.assertFalse(post.es_flagged)
published = PublishedContent.objects.get(content_pk=tuto.pk)
self.assertTrue(published.es_already_indexed)
self.assertFalse(published.es_flagged)
s = Search()
s.query(MatchAll())
results = self.index_manager.setup_search(s).execute()
self.assertEqual(len(results), 4) # get the 4 results back