def setUp(self):
self.test_geolocation_finder = GeolocationFinder()
self.setup_mocks()
def __init__(self):
self.validator = ValidatorClass(pathToPickleFiles)
self.geo_finder = GeolocationFinder()
self.database_handler = DatabaseHandler(dbURL, dbPort, dbUser,
dbPasswd)
class GeolocationFinderTests(unittest.TestCase):
def setUp(self):
self.test_geolocation_finder = GeolocationFinder()
self.setup_mocks()
def setup_mocks(self):
self.attrs = {'address': 'test_address', 'latitude': '0000', 'longitude': '0000'}
self.test_geolocation = Mock(**self.attrs)
self.test_geolocation_finder.geolocation = Mock(**self.attrs)
def test_GetLocation_ReturnsNoneNoneNone_IfLocationIsNone(self):
# Checks
self.assertEqual((None, None, None), self.test_geolocation_finder.get_location(None))
def test_get_location_sets_geolocation_from_cache_When_geolocation_exists_in_cache(self):
# Arrange
self.test_geolocation_finder.location_cache['Dublin'] = self.test_geolocation
# Execute
with patch.object(GeolocationFinder, 'set_geolocation_from_cache') as mock_method:
self.test_geolocation_finder.get_location('Dublin')
# Checks
mock_method.assert_called_once_with('Dublin')
def test_get_location_sets_geolocation_from_geolocator_when_not_present_in_cache(self):
# Execute
with patch.object(GeolocationFinder, 'set_geolocation_from_geolocator',
return_value=self.test_geolocation) as mock_method:
self.test_geolocation_finder.get_location('Dublin')
# Checks
mock_method.assert_called_once_with('Dublin')
def test_get_location_sets_geolocation_from_geolocator_adds_geolocation_to_cache(self):
# Arrange
self.test_geolocation_finder.location_cache = {}
# Execute
with patch.object(Nominatim, 'geocode', return_value=self.test_geolocation) as mock_method:
self.test_geolocation_finder.get_location('Dublin')
# Checks
self.assertTrue('Dublin' in self.test_geolocation_finder.location_cache)
def test_set_geolocation_from_cache_sets_the_geolocation(self):
# Arrange
self.test_geolocation_finder.location_cache['Dublin'] = self.test_geolocation
# Execute
self.test_geolocation_finder.set_geolocation_from_cache('Dublin')
# Checks
self.assertEqual(self.test_geolocation_finder.geolocation, self.test_geolocation)
def test_set_geolocation_from_geolocator_sets_geolocation(self):
# Execute
with patch.object(Nominatim, 'geocode', return_value=self.test_geolocation):
self.test_geolocation_finder.set_geolocation_from_geolocator('Dublin')
# Checks
self.assertEqual(self.test_geolocation_finder.geolocation, self.test_geolocation)
def test_set_geolocation_from_geolocator_sets_geolocation_to_None_when_GeocoderTimedOut_exception_raised(self):
# Execute
with patch.object(Nominatim, 'geocode', side_effect=GeocoderTimedOut):
self.test_geolocation_finder.set_geolocation_from_geolocator('Dublin')
# Checks
self.assertEqual(self.test_geolocation_finder.geolocation, None)
def test_get_addr_lat_long_raises_AttributeError_and_returns_None_None_None_when_no_geolocation_passed(self):
# Arrange
self.test_geolocation_finder.geolocation = None
# Execute
address, lat, long = self.test_geolocation_finder.get_addr_lat_long()
# Checks
self.assertEqual(address, None)
self.assertEqual(lat, None)
self.assertEqual(long, None)
def test_get_addr_lat_long(self):
# Execute
address, lat, long = self.test_geolocation_finder.get_addr_lat_long()
# Checks
self.assertEqual(address, 'test_address')
self.assertEqual(lat, '0000')
self.assertEqual(long, '0000')
class DataCollector(StreamListener):
def __init__(self):
self.validator = ValidatorClass(pathToPickleFiles)
self.geo_finder = GeolocationFinder()
self.database_handler = DatabaseHandler(dbURL, dbPort, dbUser,
dbPasswd)
def on_data(self, raw_data):
"""
When Listener detects a tweet with the keywords this method is called to handle the tweet.
Sequence:
- Load the json data
- Validate tweet
- Store if valid
:param raw_data:
:return: nothing
"""
try:
# Load the raw data
json_data = json.loads(raw_data)
# Get some required details from json data
user_id, text, language, location, timestamp = self.get_data_from_json_data(
json_data)
# Check if text in tweet is valid before processing
if text != 'invalid' and self.validator.validate_text_from_tweet(
text):
record = {'created': timestamp, 'user_language': language}
# Check if tweet contains a valid location
if self.validator.validate_location(
location) and location != 'None':
# get location details of user
address, latitude, longitude = self.geo_finder.get_location(
location)
# If location has not returned None for lat and long, construct and record the map point in database
if (latitude is not None) and (longitude is not None) \
and (latitude != 'None') and (longitude != 'None'):
self.add_to_record(address, latitude, longitude,
record)
self.record_map_point(latitude, longitude, timestamp,
text)
# Check if language is english
if self.language_is_english(language):
self.database_handler.write_english_tweet_to_database(
record)
except TypeError:
logger.logging.exception('Error during on_data method')
except ValueError:
logger.logging.exception('Error during on_data method')
def language_is_english(self, language):
"""
Checks is language provided is english
:param language:
:return boolean value True/False
"""
return (language == 'en') or (language == 'en-gb')
def add_to_record(self, address, latitude, longitude, record):
"""
Add location values to record which is a dictionary
:param address: string value for address
:param latitude: float value for latitude
:param longitude: float value for longitude
:param record: dictionary
:return: nothing
"""
record['address'] = address
record['latitude'] = latitude
record['longitude'] = longitude
def record_map_point(self, latitude, longitude, timestamp, text):
"""
creats a record(dictionary) for map point and calls the database handler to store it
:param latitude: float value for latitude
:param longitude: float value for longitude
:param timestamp: string value for timestamp
:param text: string value for text
:return: nothing
"""
map_point_record = {
'date': int(timestamp),
'lat': latitude,
'long': longitude,
'text': text
}
self.database_handler.write_map_point_to_database(map_point_record)
def get_data_from_json_data(self, json_data):
"""
extracts appropriate data from json data, if KeyError occurs sets attribute to unknown or none
:param json_data:
:return: user_id(string), text(string), user_language(string), location(string), timestamp(string)
"""
try:
user_id = json_data['user']['id_str']
except KeyError:
logger.logging.exception('KeyError while accessing user ID')
user_id = 'unknown'
try:
user_language = json_data['user']['lang']
except KeyError:
logger.logging.exception('KeyError while accessing user language')
user_language = 'unknown'
try:
location = json_data['user']['location']
except KeyError:
logger.logging.exception('KeyError while accessing user location')
location = None
try:
text = json_data['text'].lower()
except KeyError:
# if keyError is raised set the text to a banned word so it will not be accepted
text = 'invalid text'
logger.logging.exception('KeyError while accessing tweet text')
# Get time tweet picked up
timestamp = self.get_timestamp()
return user_id, text, user_language, location, timestamp
def get_timestamp(self):
"""
creates a timestamp in string format
:return: timestamp(string)
"""
now = datetime.datetime.now()
day = str(now.day)
month = str(now.month)
year = str(now.year)
if len(day) == 1:
day = '0' + day
if len(month) == 1:
month = '0' + month
timestamp = year + month + day
return timestamp
def on_error(self, status_code):
logging.error('Twitter Stream returned status code:' +
str(status_code))
def __init__(self):
self.validator = ValidatorClass(pathToPickleFiles)
self.geo_finder = GeolocationFinder()
self.database_handler = DatabaseHandler(dbURL, dbPort,dbUser, dbPasswd)
class DataCollector(StreamListener):
def __init__(self):
self.validator = ValidatorClass(pathToPickleFiles)
self.geo_finder = GeolocationFinder()
self.database_handler = DatabaseHandler(dbURL, dbPort,dbUser, dbPasswd)
def on_data(self, raw_data):
"""
When Listener detects a tweet with the keywords this method is called to handle the tweet.
Sequence:
- Load the json data
- Validate tweet
- Store if valid
:param raw_data:
:return: nothing
"""
try:
# Load the raw data
json_data = json.loads(raw_data)
# Get some required details from json data
user_id, text, language, location, timestamp = self.get_data_from_json_data(json_data)
# Check if text in tweet is valid before processing
if text != 'invalid' and self.validator.validate_text_from_tweet(text):
record = {'created': timestamp, 'user_language': language}
# Check if tweet contains a valid location
if self.validator.validate_location(location) and location != 'None':
# get location details of user
address, latitude, longitude = self.geo_finder.get_location(location)
# If location has not returned None for lat and long, construct and record the map point in database
if (latitude is not None) and (longitude is not None) \
and (latitude != 'None') and (longitude != 'None'):
self.add_to_record(address, latitude, longitude, record)
self.record_map_point(latitude, longitude, timestamp, text)
# Check if language is english
if self.language_is_english(language):
self.database_handler.write_english_tweet_to_database(record)
except TypeError:
logger.logging.exception('Error during on_data method')
except ValueError:
logger.logging.exception('Error during on_data method')
def language_is_english(self, language):
"""
Checks is language provided is english
:param language:
:return boolean value True/False
"""
return (language == 'en') or (language == 'en-gb')
def add_to_record(self, address, latitude, longitude, record):
"""
Add location values to record which is a dictionary
:param address: string value for address
:param latitude: float value for latitude
:param longitude: float value for longitude
:param record: dictionary
:return: nothing
"""
record['address'] = address
record['latitude'] = latitude
record['longitude'] = longitude
def record_map_point(self, latitude, longitude, timestamp, text):
"""
creats a record(dictionary) for map point and calls the database handler to store it
:param latitude: float value for latitude
:param longitude: float value for longitude
:param timestamp: string value for timestamp
:param text: string value for text
:return: nothing
"""
map_point_record = {'date': int(timestamp), 'lat': latitude, 'long': longitude, 'text': text}
self.database_handler.write_map_point_to_database(map_point_record)
def get_data_from_json_data(self, json_data):
"""
extracts appropriate data from json data, if KeyError occurs sets attribute to unknown or none
:param json_data:
:return: user_id(string), text(string), user_language(string), location(string), timestamp(string)
"""
try:
user_id = json_data['user']['id_str']
except KeyError:
logger.logging.exception('KeyError while accessing user ID')
user_id = 'unknown'
try:
user_language = json_data['user']['lang']
except KeyError:
logger.logging.exception('KeyError while accessing user language')
user_language = 'unknown'
try:
location = json_data['user']['location']
except KeyError:
logger.logging.exception('KeyError while accessing user location')
location = None
try:
text = json_data['text'].lower()
except KeyError:
# if keyError is raised set the text to a banned word so it will not be accepted
text = 'invalid text'
logger.logging.exception('KeyError while accessing tweet text')
# Get time tweet picked up
timestamp = self.get_timestamp()
return user_id, text, user_language, location, timestamp
def get_timestamp(self):
"""
creates a timestamp in string format
:return: timestamp(string)
"""
now = datetime.datetime.now()
day = str(now.day)
month = str(now.month)
year = str(now.year)
if len(day) == 1:
day = '0' + day
if len(month) == 1:
month = '0' + month
timestamp = year + month + day
return timestamp
def on_error(self, status_code):
logging.error('Twitter Stream returned status code:' + str(status_code))
def setUp(self):
self.test_geolocation_finder = GeolocationFinder()
self.setup_mocks()
class GeolocationFinderTests(unittest.TestCase):
def setUp(self):
self.test_geolocation_finder = GeolocationFinder()
self.setup_mocks()
def setup_mocks(self):
self.attrs = {
'address': 'test_address',
'latitude': '0000',
'longitude': '0000'
}
self.test_geolocation = Mock(**self.attrs)
self.test_geolocation_finder.geolocation = Mock(**self.attrs)
def test_GetLocation_ReturnsNoneNoneNone_IfLocationIsNone(self):
# Checks
self.assertEqual((None, None, None),
self.test_geolocation_finder.get_location(None))
def test_get_location_sets_geolocation_from_cache_When_geolocation_exists_in_cache(
self):
# Arrange
self.test_geolocation_finder.location_cache[
'Dublin'] = self.test_geolocation
# Execute
with patch.object(GeolocationFinder,
'set_geolocation_from_cache') as mock_method:
self.test_geolocation_finder.get_location('Dublin')
# Checks
mock_method.assert_called_once_with('Dublin')
def test_get_location_sets_geolocation_from_geolocator_when_not_present_in_cache(
self):
# Execute
with patch.object(GeolocationFinder,
'set_geolocation_from_geolocator',
return_value=self.test_geolocation) as mock_method:
self.test_geolocation_finder.get_location('Dublin')
# Checks
mock_method.assert_called_once_with('Dublin')
def test_get_location_sets_geolocation_from_geolocator_adds_geolocation_to_cache(
self):
# Arrange
self.test_geolocation_finder.location_cache = {}
# Execute
with patch.object(Nominatim,
'geocode',
return_value=self.test_geolocation) as mock_method:
self.test_geolocation_finder.get_location('Dublin')
# Checks
self.assertTrue(
'Dublin' in self.test_geolocation_finder.location_cache)
def test_set_geolocation_from_cache_sets_the_geolocation(self):
# Arrange
self.test_geolocation_finder.location_cache[
'Dublin'] = self.test_geolocation
# Execute
self.test_geolocation_finder.set_geolocation_from_cache('Dublin')
# Checks
self.assertEqual(self.test_geolocation_finder.geolocation,
self.test_geolocation)
def test_set_geolocation_from_geolocator_sets_geolocation(self):
# Execute
with patch.object(Nominatim,
'geocode',
return_value=self.test_geolocation):
self.test_geolocation_finder.set_geolocation_from_geolocator(
'Dublin')
# Checks
self.assertEqual(self.test_geolocation_finder.geolocation,
self.test_geolocation)
def test_set_geolocation_from_geolocator_sets_geolocation_to_None_when_GeocoderTimedOut_exception_raised(
self):
# Execute
with patch.object(Nominatim, 'geocode', side_effect=GeocoderTimedOut):
self.test_geolocation_finder.set_geolocation_from_geolocator(
'Dublin')
# Checks
self.assertEqual(self.test_geolocation_finder.geolocation, None)
def test_get_addr_lat_long_raises_AttributeError_and_returns_None_None_None_when_no_geolocation_passed(
self):
# Arrange
self.test_geolocation_finder.geolocation = None
# Execute
address, lat, long = self.test_geolocation_finder.get_addr_lat_long()
# Checks
self.assertEqual(address, None)
self.assertEqual(lat, None)
self.assertEqual(long, None)
def test_get_addr_lat_long(self):
# Execute
address, lat, long = self.test_geolocation_finder.get_addr_lat_long()
# Checks
self.assertEqual(address, 'test_address')
self.assertEqual(lat, '0000')
self.assertEqual(long, '0000')
