class Heartbeat:
"""heartbeat class - Will inform the user on regular intervals whether the bot is still alive"""
__log__ = logging.getLogger('flathunt')
def __init__(self, config, interval):
self.config = config
if not isinstance(self.config, Config):
raise Exception(
"Invalid config for hunter - should be a 'Config' object")
self.notifiers = self.config.get('notifiers', list())
if 'mattermost' in self.notifiers:
self.notifier = SenderMattermost(config)
elif 'telegram' in self.notifiers:
self.notifier = SenderTelegram(config)
else:
self.notifier = None
self.interval = interval2counter(interval)
def send_heartbeat(self, counter):
"""Send a new heartbeat message"""
# its time for a new heartbeat message and reset counter
if self.notifier is not None and self.interval is not None:
if counter % self.interval == 0:
self.notifier.send_msg(
'Beep Boop. This is a heartbeat message. Your bot is searching actively for flats.'
)
counter = 0
return counter
def hunt_flats(self, config, searchers, id_watch):
sender = SenderTelegram(config)
new_links = 0
processed = id_watch.get()
for url in config.get('urls', list()):
self.__log__.debug('Processing URL: ' + url)
try:
for searcher in searchers:
if re.search(searcher.URL_PATTERN, url):
results = searcher.get_results(url)
break
except requests.exceptions.ConnectionError:
self.__log__.warning("Connection to %s failed. Retrying. " %
url.split('/')[2])
continue
# on error, stop execution
if not results:
break
uniq_results = list({v['id']: v for v in results}.values())
for expose in uniq_results:
# check if already processed
if expose['id'] in processed:
continue
self.__log__.info('New offer: ' + expose['title'])
# to reduce traffic, some addresses need to be loaded on demand
address = expose['address']
if address.startswith('http'):
url = address
for searcher in searchers:
if re.search(searcher.URL_PATTERN, url):
address = searcher.load_address(url)
self.__log__.debug("Loaded address %s for url %s" %
(address, url))
break
# calculdate durations
message = config.get('message', "").format(
title=expose['title'],
rooms=expose['rooms'],
size=expose['size'],
price=expose['price'],
url=expose['url'],
durations=self.get_formatted_durations(config,
address)).strip()
# send message to all receivers
sender.send_msg(message)
new_links = new_links + 1
id_watch.add(expose['id'])
self.__log__.info(str(new_links) + ' new offer found')
def hunt_flats(self, config, searchers, id_watch):
sender = SenderTelegram(config)
new_links = 0
processed = id_watch.get()
for url in config.get("urls", list()):
self.__log__.debug("Processing URL: " + url)
try:
for searcher in searchers:
if re.search(searcher.URL_PATTERN, url):
results = searcher.get_results(url)
break
except requests.exceptions.ConnectionError:
self.__log__.warning("Connection to %s failed. Retrying. " %
url.split("/")[2])
continue
# on error, stop execution
if not results:
print("something broke, results empty")
break
for expose in results:
# check if already processed
if expose["id"] in processed:
continue
self.__log__.info("New offer: " + expose["title"])
# to reduce traffic, some addresses need to be loaded on demand
address = expose["address"]
if address.startswith("http"):
url = address
for searcher in searchers:
if re.search(searcher.URL_PATTERN, url):
address = searcher.load_address(url)
self.__log__.debug("Loaded address %s for url %s" %
(address, url))
break
# calculdate durations
message = (config.get("message", "").format(
title=expose["title"],
rooms=expose["rooms"],
size=expose["size"],
price=expose["price"],
url=expose["url"],
).strip())
# send message to all receivers
sender.send_msg(message)
new_links = new_links + 1
id_watch.add(expose["id"])
print(str(new_links) + " new offer found")
self.__log__.info(str(new_links) + " new offer found")
def test_send_no_message_if_no_receivers(self, m):
sender = SenderTelegram(
{"telegram": {
"bot_token": "dummy_token",
"receiver_ids": None
}})
self.assertEqual(None, sender.send_msg("result"),
"Expected no message to be sent")
class SenderTelegramTest(unittest.TestCase):
def setUp(self):
self.sender = SenderTelegram({ "telegram": { "bot_token": "dummy_token", "receiver_ids": [ 123 ] }})
@requests_mock.Mocker()
def test_send_message(self, m):
mock_response = '{"ok":true,"result":{"message_id":456,"from":{"id":1,"is_bot":true,"first_name":"Wohnbot","username":"******"},"chat":{"id":5,"first_name":"Arthur","last_name":"Taylor","type":"private"},"date":1589813130,"text":"hello arthur"}}'
m.get('https://api.telegram.org/botdummy_token/sendMessage?chat_id=123&text=result', text=mock_response)
self.assertEqual(None, self.sender.send_msg("result"), "Expected message to be sent")
def hunt_flats(self, config, searchers, id_watch):
sender = SenderTelegram(config)
new_links = 0
processed = id_watch.get()
for url in config.get('urls', list()):
self.__log__.debug('Processing URL: ' + url)
# TODO: improve control flow
try:
for searcher in searchers:
if re.search(searcher.URL_PATTERN, url):
results = searcher.get_results(url)
break
except requests.exceptions.ConnectionError:
self.__log__.warning("Connection to %s failed. Retrying. " % url.split('/')[2])
continue
except Exception as e:
self.__log__.warning("Unknown error: {}".format(e.with_traceback()))
continue
# on error, stop execution
if not results:
break
for expose in results:
# check if already processed
if expose['id'] in processed:
continue
self.__log__.info('New offer: ' + expose['title'])
# to reduce traffic, some addresses need to be loaded on demand
address = expose['address']
if address.startswith('http'): # ugh, TODO
url = address
for searcher in searchers:
if re.search(searcher.URL_PATTERN, url):
address = searcher.load_address(url)
self.__log__.debug("Loaded address %s for url %s" % (address, url))
break
# filter districts
blacklist = config.get('blacklist', list())
address = ' '.join(filter(lambda x: x not in blacklist, address.split()))
# add to visited list already now so that we can actually skip if entry does not match date filter
id_watch.add(expose['id'])
# get date if necessary
if not "date" in expose:
for searcher in searchers:
if re.search(searcher.URL_PATTERN, url):
expose["date"] = searcher.load_date(expose["url"])
self.__log__.debug("Loaded date {} for url {}".format(address, expose["url"]))
break
date_filter = config.get('date_filter', dict())
mismatched_date = False
for blacklisted_phrase in date_filter.get("blacklist_phrases", []):
if blacklisted_phrase in expose["date"]:
mismatched_date = True
break
if mismatched_date:
# go to next expose
self.__log__.info("Skipping entry, date {} matches blacklist".format(expose["date"]))
continue
# try to parse date string
try:
parsed_date = parser.parse(expose["date"], dayfirst=True).date()
date_min = date_filter.get('date_min')
date_max = date_filter.get('date_max')
if date_min and parsed_date < date_min:
self.__log__.info("Skipping entry, date {} too early".format(expose["date"]))
continue
if date_max and parsed_date > date_max:
self.__log__.info("Skipping entry, date {} too late".format(expose["date"]))
continue
except (ValueError, OverflowError):
self.__log__.debug("Could not parse date {} for url {} - ignoring filters".format(expose["date"], url))
# calculdate durations
message = config.get('message', "").format(
title=expose['title'],
date=expose['date'],
rooms=expose['rooms'],
size=expose['size'],
price=expose['price'],
url=expose['url'],
address=address,
durations=self.get_formatted_durations(config, address)).strip()
# send message to all receivers
sender.send_msg(message)
new_links = new_links + 1
self.__log__.info(str(new_links) + ' new offers found')
def hunt_flats(self, connection=None):
sender = SenderTelegram(self.config)
new_exposes = []
processed = self.id_watch.get(connection)
for url in self.config.get('urls', list()):
self.__log__.debug('Processing URL: ' + url)
try:
for searcher in self.searchers:
if re.search(searcher.URL_PATTERN, url):
results = searcher.get_results(url)
break
except requests.exceptions.ConnectionError:
self.__log__.warning("Connection to %s failed. Retrying. " %
url.split('/')[2])
continue
# on error, stop execution
if not results:
self.__log__.debug('No results for: ' + url)
continue
for expose in results:
# check if already processed
if expose['id'] in processed:
continue
self.__log__.info('New offer: ' + expose['title'])
# to reduce traffic, some addresses need to be loaded on demand
address = expose['address']
if address.startswith('http'):
url = address
for searcher in self.searchers:
if re.search(searcher.URL_PATTERN, url):
address = searcher.load_address(url)
self.__log__.debug("Loaded address %s for url %s" %
(address, url))
break
# calculate durations if enabled
durations_enabled = "google_maps_api" in self.config and self.config[
"google_maps_api"]["enable"]
if durations_enabled:
durations = self.get_formatted_durations(
self.config, address).strip()
message = self.config.get('message', "").format(
title=expose['title'],
rooms=expose['rooms'],
size=expose['size'],
price=expose['price'],
url=expose['url'],
address=address,
durations=""
if not durations_enabled else durations).strip()
# if no excludes, send messages
if len(self.excluded_titles) == 0:
# send message to all receivers
sender.send_msg(message)
new_exposes.append(expose)
self.id_watch.add(expose['id'], connection)
continue
# combine all the regex patterns into one
combined_excludes = "(" + ")|(".join(
self.excluded_titles) + ")"
found_objects = re.search(combined_excludes,
expose['title'].lower())
# send all non matching regex patterns
if not found_objects:
# send message to all receivers
sender.send_msg(message)
new_exposes.append(expose)
self.id_watch.add(expose['id'], connection)
self.__log__.info(str(len(new_exposes)) + ' new offers found')
self.id_watch.update_last_run_time(connection)
return new_exposes