def setUp(self):
TestBase.setUp(self)
self.config = getConfig()
self._tsv_files_path = self.config.get(
"TumblrImporter", "tsv_test_files_syntax_feature_generator")
self._db = DB()
self._db.setUp()
self._tumblr_parser = TumblrImporter(self._db)
self._author_guid = "150ff707-a6eb-3051-8f3c-f623293c714b"
self._tumblr_parser.setUp(self._tsv_files_path)
self._tumblr_parser.execute()
authors = self._db.get_authors_by_domain(Domains.MICROBLOG)
posts = self._db.get_posts_by_domain(Domains.MICROBLOG)
parameters = {"authors": authors, "posts": posts}
self._syntax_feature_generator = SyntaxFeatureGenerator(
self._db, **parameters)
self._syntax_feature_generator.execute()
self._author_features = self._db.get_author_features_by_author_guid(
author_guid=self._author_guid)
self._author_features_dict = self._create_author_features_dictionary(
self._author_features)
def testSyntaxFeatureGenerator(self):
parameters = {"authors": self.authors, "posts": self.posts}
syntax_feature = SyntaxFeatureGenerator(self._db, **parameters)
syntax_feature.execute()
allFeatures = self._db.get_author_features()
if allFeatures is not None:
for feature in allFeatures:
if feature.author_guid == 'TestUser1' and feature.attribute_name == "average_hashtags":
self.assertEqual(feature.attribute_value, '1.0')
elif feature.author_guid == 'TestUser4' and feature.attribute_name == "average_user_mentions":
self.assertEqual(feature.attribute_value, '1.0')
elif feature.author_guid == 'TestUser2' and feature.attribute_name == "average_links":
self.assertEqual(feature.attribute_value, '1.0')
elif feature.author_guid == 'TestUser3' and feature.attribute_name == "average_post_lenth":
self.assertEqual(feature.attribute_value, '5.0')
self._db.session.close()
def setUp(self):
self._db = DB()
self._db.setUp()
self._posts = []
self._author = None
self.syntax_feature_generator = SyntaxFeatureGenerator(self._db, **{})
class TestTumblrImporterSyntaxFeatureGenerator(TestBase):
def setUp(self):
TestBase.setUp(self)
self.config = getConfig()
self._tsv_files_path = self.config.get(
"TumblrImporter", "tsv_test_files_syntax_feature_generator")
self._db = DB()
self._db.setUp()
self._tumblr_parser = TumblrImporter(self._db)
self._author_guid = "150ff707-a6eb-3051-8f3c-f623293c714b"
self._tumblr_parser.setUp(self._tsv_files_path)
self._tumblr_parser.execute()
authors = self._db.get_authors_by_domain(Domains.MICROBLOG)
posts = self._db.get_posts_by_domain(Domains.MICROBLOG)
parameters = {"authors": authors, "posts": posts}
self._syntax_feature_generator = SyntaxFeatureGenerator(
self._db, **parameters)
self._syntax_feature_generator.execute()
self._author_features = self._db.get_author_features_by_author_guid(
author_guid=self._author_guid)
self._author_features_dict = self._create_author_features_dictionary(
self._author_features)
def test_average_hashtags(self):
attribute_value = self._author_features_dict["average_hashtags"]
attribute_value = float(attribute_value)
self.assertEquals(0.5, attribute_value)
def test_average_links(self):
attribute_value = self._author_features_dict["average_links"]
attribute_value = float(attribute_value)
self.assertEquals(0.5, attribute_value)
def test_average_user_mentions(self):
attribute_value = self._author_features_dict["average_user_mentions"]
attribute_value = float(attribute_value)
self.assertEquals(0.5, attribute_value)
def test_average_post_lenth(self):
attribute_value = self._author_features_dict["average_post_lenth"]
attribute_value = float(attribute_value)
self.assertEquals(6.0, attribute_value)
def tearDown(self):
self._db.deleteDB()
pass
def _create_author_features_dictionary(self, author_features):
author_features_dict = {}
for author_feature in author_features:
attribute_name = author_feature.attribute_name
attribute_value = author_feature.attribute_value
author_features_dict[attribute_name] = attribute_value
return author_features_dict
class TestSyntaxFeatureGenerator(TestCase):
def setUp(self):
self._db = DB()
self._db.setUp()
self._posts = []
self._author = None
self.syntax_feature_generator = SyntaxFeatureGenerator(self._db, **{})
def tearDown(self):
self._db.session.close()
pass
def test_average_hashtags(self):
self._add_author(u"author_guid")
self._add_post(u"post1", u"#content 1", "2017-06-12 05:00:00")
self._add_post(u"post2", u"content 1", "2017-06-12 05:00:00")
self._add_post(u"post3", u"content 1 #tag #yes", "2017-06-12 05:00:00")
self._add_post(u"post4", u"content #1 #test #dont #fail #please",
"2017-06-12 05:00:00")
self.syntax_feature_generator.execute()
author_feature = self._db.get_author_feature(
u"author_guid", u"SyntaxFeatureGenerator_average_hashtags")
self.assertAlmostEqual(float(author_feature.attribute_value),
8.0 / 4,
places=4)
def test_average_links(self):
self._add_author(u"author_guid")
self._add_post(u"post1", u"#content 1", "2017-06-12 05:00:00")
self._add_post(u"post2", u"https://www.google.co.il 1",
"2017-06-12 05:00:00")
self._add_post(u"post3", u"content 1 #tag http://www.google.co.il",
"2017-06-12 05:00:00")
self._add_post(
u"post4",
u"http://www.bank.co.il #1 #test #dont http://www.ynet.co.il https://www.msn.co.il",
"2017-06-12 05:00:00")
self.syntax_feature_generator.execute()
author_feature = self._db.get_author_feature(
u"author_guid", u"SyntaxFeatureGenerator_average_links")
self.assertAlmostEqual(float(author_feature.attribute_value),
5.0 / 4,
places=4)
def test_average_user_mentions(self):
self._add_author(u"author_guid")
self._add_post(u"post1", u"@content 1", "2017-06-12 05:00:00")
self._add_post(u"post2", u"content 1", "2017-06-12 05:00:00")
self._add_post(u"post3", u"content 1 @tag #@es", "2017-06-12 05:00:00")
self._add_post(u"post4", u"content #1 @test @dont @fail #please",
"2017-06-12 05:00:00")
self.syntax_feature_generator.execute()
author_feature = self._db.get_author_feature(
u"author_guid", u"SyntaxFeatureGenerator_average_user_mentions")
self.assertAlmostEqual(float(author_feature.attribute_value),
6.0 / 4,
places=4)
def test_average_post_lenth(self):
self._add_author(u"author_guid")
self._add_post(u"post1", u"content 1", "2017-06-12 05:00:00")
self._add_post(u"post2", u"content 1", "2017-06-12 05:00:00")
self._add_post(u"post3", u"content 1 @tag #@es", "2017-06-12 05:00:00")
self._add_post(u"post4", u"content #1 @test @dont @fail #please",
"2017-06-12 05:00:00")
self.syntax_feature_generator.execute()
author_feature = self._db.get_author_feature(
u"author_guid", u"SyntaxFeatureGenerator_average_post_lenth")
self.assertAlmostEqual(float(author_feature.attribute_value),
14.0 / 4,
places=4)
def _add_author(self, author_guid):
author = Author()
author.author_guid = author_guid
author.author_full_name = u'test author'
author.name = u'test'
author.domain = u'tests'
author.statuses_count = 0
self._db.add_author(author)
self._author = author
def _add_post(self, title, content, date_str, domain=u'Microblog'):
post = Post()
post.author = self._author.author_guid
post.author_guid = self._author.author_guid
post.content = content
post.title = title
post.domain = domain
post.post_id = title
post.guid = post.post_id
post.date = convert_str_to_unicode_datetime(date_str)
post.created_at = post.date
self._db.addPost(post)
self._posts.append(post)
self._author.statuses_count += 1