def update_feed(self, url):
try:
feed = FeedHandler.parse_feed(url[0])
except:
feed = False
traceback.print_exc() # ???
if feed:
print(f'{url[0]}:')
print(f'Longitud de feed: {len(feed)}')
url_items = self.db.get_url_items(url=url[0])
for item in url_items:
url_items[item]['active'] = False
new_items = []
for item in feed:
hash = str(
CityHash64(item['summary'] + item['title'] + item['link']))
if not (hash in url_items):
new_items.append(item)
url_items[hash] = {
'active': True,
'last_date': DateHandler.get_datetime_now()
}
for item, value in url_items.copy().items():
if not value['active']:
print(f'Desactivando {item}')
if not value['active'] and DateHandler.is_older_than_days(
value['last_date'], 5):
print(f'Borrando {item}')
url_items.pop(item)
self.db.update_url_items(url=url[0], items=url_items)
telegram_users = self.db.get_users_for_url(url=url[0])
for user in telegram_users:
if user[6]: # is_active
if not feed:
message = "Something went wrong when I tried to parse the URL: \n\n " + \
url[0] + "\n\nCould you please check that for me? Remove the url from your subscriptions using the /remove command, it seems like it does not work anymore!"
self.bot.send_message(chat_id=user[0],
text=message,
parse_mode=ParseMode.HTML)
return
for post in new_items:
self.send_message(post=post, user=user)
def get(self, bot, update, args):
"""
Manually parses an rss feed
"""
telegram_user = update.message.from_user
if len(args) > 2:
message = "To get the last news of your subscription please use /get <entryname> [optional: <count 1-10>]. Make sure you first add a feed using the /add command."
update.message.reply_text(message)
return
if len(args) == 2:
args_entry = args[0]
args_count = int(args[1])
else:
args_entry = args[0]
args_count = 4
if not(1 <= args_count <= 10):
message = "Count parameter (if used) must be between 1 and 10.\n" + \
"Usage: /get <entryname> [optional: <count 1-10>]"
update.message.reply_text(message)
return
url = self.db.get_user_bookmark(telegram_id=telegram_user.id, alias=args_entry)
if url is None:
message = "I can not find an entry with label " + \
args_entry + " in your subscriptions! Please check your subscriptions using /list and use the delete command again!"
update.message.reply_text(message)
return
entries = FeedHandler.parse_feed(url[0], args_count)
for entry in entries:
message = f"[{url[1]}] <a href='{entry.link}'>{entry.title}</a>"
print(message)
try:
update.message.reply_text(message, parse_mode=ParseMode.HTML)
except Unauthorized:
self.db.update_user(telegram_id=telegram_user.id, is_active=0)
except TelegramError:
# handle all other telegram related errors
pass
def update_feed(self, url):
telegram_users = self.db.get_users_for_url(url=url[0])
for user in telegram_users:
if user[6]: # is_active
try:
for post in FeedHandler.parse_feed(url[0]):
self.send_newest_messages(
url=url, post=post, user=user)
except:
traceback.print_exc()
message = "Something went wrong when I tried to parse the URL: \n\n " + \
url[0] + "\n\nCould you please check that for me? Remove the url from your subscriptions using the /remove command, it seems like it does not work anymore!"
self.bot.send_message(
chat_id=user[0], text=message, parse_mode=ParseMode.HTML)
self.db.update_url(url=url[0], last_updated=str(
DateHandler.get_datetime_now()))
def update_feed(self, url):
if not self._finished.isSet():
try:
get_url_info = self.db.get_update_url(url)
last_url = get_url_info['last_url']
date_last_url = DateHandler.parse_datetime(
get_url_info['last_update'])
feed = FeedHandler.parse_feed(
url, 4, date_last_url + timedelta(days=-1))
for post in feed:
if not hasattr(post, "published") and not hasattr(
post, "daily_liturgy"):
logger.warning('not published' + url)
continue
# for index, post in enumerate(feed):
date_published = DateHandler.parse_datetime(post.published)
if hasattr(post, "daily_liturgy"):
if date_published > date_last_url and post.link != last_url \
and post.daily_liturgy != '':
message = post.title + '\n' + post.daily_liturgy
result = self.send_newest_messages(message, url)
if post == feed[-1] and result:
self.update_url(url=url,
last_update=date_published,
last_url=post.link)
elif date_published > date_last_url and post.link != last_url:
message = post.title + '\n' + post.link
result = self.send_newest_messages(message, url)
if result:
self.update_url(url=url,
last_update=date_published,
last_url=post.link)
else:
pass
return True, url
except TypeError as e:
logger.error(f"TypeError {url} {str(e)}")
return False, url, 'update_feed'
except TelegramError as e:
logger.error(
f"except update_feed TelegramError {url} {str(e)}")
return False, url, 'update_feed'
def update_feed(self, url):
telegram_users = self.db.get_users_for_url(url=url[0])
telegram_channels = self.db.get_channels_for_url(url=url[0])
print("Processing url: {0}".format(url[0]))
try:
posts = FeedHandler.parse_feed(url[0])
except ValueError:
traceback.print_exc()
return
for post in posts:
print("Processing post: {0}".format(post.id))
for user in telegram_users:
if user[6]: # is_active
self.send_newest_messages(url=url, post=post, user=user)
for channel in telegram_channels:
self.send_newest_messages(url=url, post=post, user=channel)
self.db.update_url(url=url[0], last_updated=str(
DateHandler.get_datetime_now()))
def update_feed(url):
try:
get_url_info = db.get_update_url(url)
last_url = get_url_info['last_url']
date_last_url = DateHandler.parse_datetime(get_url_info['last_update'])
feed = FeedHandler.parse_feed(url, 4,
date_last_url + timedelta(days=-1))
for post in feed:
if not hasattr(post, "published") and not hasattr(
post, "daily_liturgy"):
logger.warning('not published' + url)
continue
date_published = DateHandler.parse_datetime(post.published)
if hasattr(post, "daily_liturgy"):
if date_published > date_last_url and post.link != last_url \
and post.daily_liturgy != '':
message = post.title + '\n' + post.daily_liturgy
result = send_newest_messages(message=message,
url=url,
disable_page_preview=True)
if post == feed[-1] and result:
update_url(url=url,
last_update=date_published,
last_url=post.link)
elif date_published > date_last_url and post.link != last_url:
message = post.title + '\n' + post.link
result = send_newest_messages(message=message, url=url)
if result:
update_url(url=url,
last_update=date_published,
last_url=post.link)
else:
pass
except TypeError as _:
logger.error(f"TypeError {url} {str(_)}")
def test_parse_feed(self):
url = "https://lorem-rss.herokuapp.com/feed"
feed = FeedHandler.parse_feed(url)
self.assertIsNotNone(url)
url = "https://lorem-rss.herokuapp.com/feed"
def test_parse_feed_amount(self):
url = "https://lorem-rss.herokuapp.com/feed"
feed = FeedHandler.parse_feed(url, 5)
self.assertIsNotNone(url)
self.assertEqual(len(feed), 5)
