def _enter_query(self, city_name):
Logger.debug(self, "Searching for city name: '{}'".format(city_name))
try:
search_input = self._find_element_by_id('kgw')
search_input.clear()
search_input.send_keys(city_name)
sleep(0.3)
search_button = self._find_element_by_id('search')
search_button.click()
except (NoSuchElementException, ElementClickInterceptedException) as e:
Logger.error(self, e, self._enter_query.__name__)
raise KrkgwPageException('Failed to enter query')
def _get_result(self):
Logger.debug(self, 'Getting result...')
sleep(2)
try:
distance_output_xpath = '//*[@id="section-directions-trip-0"]/div/div[1]/div[1]/div[2]/div'
self._wait_for_element_by_xpath(distance_output_xpath, timeout=10)
distance_output = self._find_element_by_xpath(
distance_output_xpath)
return distance_output.text.strip()
except (TimeoutException, NoSuchElementException,
ElementClickInterceptedException,
StaleElementReferenceException) as e:
Logger.error(self, e, self._get_result.__name__)
raise MapPageException('Failed to get result')
def _get_result_details(self):
try:
name_output_xpath = '//*[@id="zawartosc"]/table[1]/tbody/tr[3]/td[2]'
self._wait_for_visibility_by_xpath(name_output_xpath, timeout=10)
name_output = self._find_element_by_xpath(name_output_xpath)
name = name_output.text
address_output_xpath = '//*[@id="zawartosc"]/table[1]/tbody/tr[4]/td[2]'
self._wait_for_visibility_by_xpath(address_output_xpath,
timeout=10)
address_output = self._find_element_by_xpath(address_output_xpath)
address = address_output.text
except (TimeoutException, NoSuchElementException) as e:
Logger.error(self, e, self._get_result_details.__name__)
raise KrkgwPageException('Failed to get result details')
return name, address
def __init__(self,
data_dir_path='data',
report_dir_path='reports',
allow_indirect_matches=False,
allow_duplicates=False,
allow_blacklisted=False):
super(GoogleSearcher,
self).__init__(data_dir_path=data_dir_path,
allow_indirect_matches=allow_indirect_matches,
allow_duplicates=allow_duplicates)
self._allow_blacklisted = allow_blacklisted
Logger.info(self,
'Allowing blacklisted: {}'.format(self._allow_blacklisted))
self._black_list = self._source.get_black_list()
self._engine = EngineFactory.get_engine('google_page')
self._keys = self._source.get_keys()
self._reporter = Reporter(report_dir_path, self.engine_name)
def _enter_query(self, main_city, target_city):
Logger.debug(
self,
"Searching distance from main city: '{}' to target city: '{}'".
format(main_city, target_city))
sleep(2)
try:
from_input = self._find_element_by_xpath(
'//*[@id="sb_ifc50"]/input')
from_input.clear()
from_input.send_keys(main_city)
to_input = self._find_element_by_xpath('//*[@id="sb_ifc51"]/input')
to_input.clear()
to_input.send_keys(target_city)
to_input.send_keys(Keys.ENTER)
except NoSuchElementException as e:
Logger.error(self, e, self._enter_query.__name__)
raise MapPageException('Failed to enter query')
def search(self):
Logger.info(self, 'Searching...')
self._reporter.generate_new_report_file_path()
self._add_city_header(self._main_city)
self._engine.start()
for target_city in self._cities:
results = None
try:
results = self._engine.search(self._main_city, target_city)
except EngineException as e:
Logger.error(self, e, self.search.__name__)
finally:
self._add_results(results, target_city)
self._reporter.save_report(self._results)
self._reporter.save_report(self._results)
self._reporter.save_to_excel(self._results, self._main_city)
self._engine.stop()
def search(self):
Logger.info(self, 'Searching...')
self._reporter.generate_new_report_file_path()
self._engine.start()
for city in self._cities:
self._add_city_header(city)
results = []
try:
results = self._engine.search(self._get_city_name(city))
except EngineException as e:
Logger.error(self, e)
finally:
self._add_results(results, city)
self._reporter.save_report(self._results)
self._add_results_count()
self._reporter.save_report(self._results)
self._engine.stop()
def _get_one_of_multiple_results(self, result_link):
try:
result_link.click()
sleep(1)
name_output_xpath = '//div[@class="SPZz6b"]//span'
self._wait_for_element_by_xpath(name_output_xpath, timeout=2)
name_output = self._find_element_by_xpath(name_output_xpath)
name = name_output.text
address_output_class = 'LrzXr'
self._wait_for_element_by_class_name(address_output_class,
timeout=1)
address_output = self._find_element_by_class_name(
address_output_class)
address = address_output.text
except (TimeoutException, NoSuchElementException,
ElementClickInterceptedException,
StaleElementReferenceException) as e:
Logger.error(self, e, self._get_one_of_multiple_results.__name__)
raise GooglePageException('Failed to get one of multiple results')
return name, address
def __init__(self, data_dir_path):
self._data_dir_path = data_dir_path
self._cities_file_path = self._assemble_data_file_path('cities.txt')
self._keys_file_path = self._assemble_data_file_path('keys.txt')
self._black_list_filer_path = self._assemble_data_file_path('black_list.txt')
Logger.info(self, 'Cities file path: {}'.format(self._cities_file_path))
Logger.info(self, 'Keys file path: {}'.format(self._keys_file_path))
Logger.info(self, 'Black list file path: {}'.format(self._black_list_filer_path))
def _get_multiple_results(self):
Logger.debug(self, 'Getting multiple results...')
results = []
result_link_class = 'dbg0pd'
try:
self._open_map()
self._wait_for_element_by_class_name(result_link_class, 2)
result_links = self._find_elements_by_class_name(result_link_class)
except (TimeoutException, NoSuchElementException,
GooglePageException) as e:
Logger.error(self, e, self._get_multiple_results.__name__)
return results
for result_link in result_links:
try:
name, address = self._get_one_of_multiple_results(result_link)
results.append(name + '\n' + self._format_address(address) +
'\n')
except GooglePageException as e:
Logger.error(self, e, self._get_multiple_results.__name__)
return results
def get_black_list(self):
Logger.info(self, 'Getting black list...')
return self._get_items_from_file(self._black_list_filer_path)
def get_keys(self):
Logger.info(self, 'Getting keys...')
return self._get_items_from_file(self._keys_file_path)
def get_cities(self):
Logger.info(self, 'Getting cities...')
return self._get_items_from_file(self._cities_file_path)
def stop(self):
self._driver.close()
self._driver = None
Logger.info(self, 'Engine stopped')
def start(self):
self._driver = Chrome(executable_path=self._executable_path)
self._driver.maximize_window()
if self._url is not None:
self._driver.get(self._url)
Logger.info(self, 'Engine started')