def setUp(self):
engine = create_engine('sqlite:///:memory:', echo=False)
db.create_tables(engine)
Session = sessionmaker(bind=engine)
self.session = Session()
self.word_count_repository = WordCountRepository(self.session)
class TestWordCountRepository(unittest.TestCase):
def setUp(self):
engine = create_engine('sqlite:///:memory:', echo=False)
db.create_tables(engine)
Session = sessionmaker(bind=engine)
self.session = Session()
self.word_count_repository = WordCountRepository(self.session)
def tearDown(self):
self.session.close()
def test_create_word_count_with_one_word(self):
self.word_count_repository.create_word_count('test', 2, 1)
result_word_count = self.word_count_repository\
.get_word_counts_by_website_id(1)
assert result_word_count[0][1] == 'test'
assert result_word_count[0][2] == 2
def test_create_word_count_with_three_words(self):
test_words = [('test', 2, 100),
('pero', 10, 21),
('gusto', 2, 23000)]
for test_word in test_words:
self.word_count_repository.create_word_count(test_word[0],
test_word[1],
test_word[2])
for test_word in test_words:
result_word_count = self.word_count_repository\
.get_word_counts_by_website_id(test_word[2])
assert result_word_count[0][1] == test_word[0]
assert result_word_count[0][2] == test_word[1]
from dataAccessLayer.helper import db
from sqlalchemy import create_engine
from sqlalchemy.orm import sessionmaker
if __name__ == "__main__":
#Set database up
engine = create_engine('sqlite:///../../database/hegemone_dev.sqlite3',
echo=False)
db.setup_db(engine)
Session = sessionmaker(bind=engine)
session = Session()
#Set repositories up
website_repository = WebsiteRepository(session)
word_count_repository = WordCountRepository(session)
#Get websites
websites = website_repository.get_all_websites()
#Crawl websites
for website_tuple in websites:
raw_text = converter.get_text(website_tuple[1], 0)
word_counts = WordFrequency.count_words(raw_text)
#Store word counts in database
website_id = website_tuple[0]
for word in word_counts.keys():
word_count_repository.create_word_count(word,
word_counts[word],
website_id)