def test_dump_to_sqlite_db_dumps(self):
if os.path.exists(
os.path.dirname(os.path.abspath(__file__)) +
'/data/output.db'):
os.remove(
os.path.dirname(os.path.abspath(__file__)) + '/data/output.db')
field_list = [
'id_str', 'coordinates.coordinates.0', 'coordinates.coordinates.1',
'user.id_str', 'user.lang', 'lang', 'text', 'user.screen_name',
'user.location', 'user.description', 'created_at',
'user.friends_count', 'user.followers_count', 'retweet_count',
'entities.urls.0.expanded_url', 'entities.urls.1.expanded_url',
'entities.urls.2.expanded_url', 'entities.urls.3.expanded_url',
'entities.urls.4.expanded_url', 'entities.hashtags.0.text',
'entities.hashtags.1.text'
]
output_path = os.path.dirname(
os.path.realpath(__file__)) + '/' + 'data/output.db'
collection = BsonCollection(
os.path.dirname(os.path.realpath(__file__)) + '/' +
config['bson']['valid'])
collection.dump_to_sqlite_db(output_path, field_list)
self.assertTrue(os.path.getsize(output_path) > 0)
if os.path.exists(
os.path.dirname(os.path.abspath(__file__)) +
'/data/output.db'):
os.remove(
os.path.dirname(os.path.abspath(__file__)) + '/data/output.db')
def test_dump_to_sqlite_db_dumps_the_right_stuff(self):
if os.path.exists(
os.path.dirname(os.path.abspath(__file__)) +
'/data/output.db'):
os.remove(
os.path.dirname(os.path.abspath(__file__)) + '/data/output.db')
field_list = [
'id_str', 'coordinates.coordinates.0', 'coordinates.coordinates.1',
'user.id_str', 'user.lang', 'lang', 'text', 'user.screen_name',
'user.location', 'user.description', 'created_at',
'user.friends_count', 'user.followers_count', 'retweet_count',
'entities.urls.0.expanded_url', 'entities.urls.1.expanded_url',
'entities.urls.2.expanded_url', 'entities.urls.3.expanded_url',
'entities.urls.4.expanded_url', 'entities.hashtags.0.text',
'entities.hashtags.1.text'
]
output_path = os.path.dirname(
os.path.realpath(__file__)) + '/' + 'data/output.db'
collection = BsonCollection(
os.path.dirname(os.path.realpath(__file__)) + '/' +
config['bson']['valid'])
collection.dump_to_sqlite_db(output_path, field_list)
con = sqlite3.connect(output_path)
cur = con.cursor()
row = [
elem for row in cur.execute("SELECT * FROM data LIMIT 1;")
for elem in row
]
con.close()
self.assertTrue(len(row) > 0)
self.assertEqual(
set(row),
set([
'661275583813431296', 'NULL', 'NULL', '379851447', 'en', 'de',
'Susan Lindauer, Rtd US Army LTC Potter: Jade Helm https://t.co/VA4bQRudLt #jadehelm #newworldorder #usa #tyranny #threat',
'Col_Connaughton', 'London UK',
'#gaza #palestine #israel #BDS MAD EVIL ISRAEL MURDERS BABIES CIVILIANS to STEAL PALESTINIAN LAND RESOURCES with USA UK HELP. To stop my tweets, BLOCK or MUTE me',
'Mon Nov 02 20:15:59 +0000 2015', 2019, 3159, 0,
'https://www.youtube.com/watch?v=0nJqymxVpwc', 'NULL', 'NULL',
'NULL', 'NULL', 'jadehelm', 'newworldorder'
]))
if os.path.exists(
os.path.dirname(os.path.abspath(__file__)) +
'/data/output.db'):
os.remove(
os.path.dirname(os.path.abspath(__file__)) + '/data/output.db')
