def simple_parser(self, res):
result_date = parse(res, now=self.get_now(), allowed_results=[date])
result_time = parse(res, now=self.get_now(), allowed_results=[time])
res_tot = parse(res, now=self.get_now(), allowed_results=[datetime])
res_text = get_clear_text(res)
if result_time != self.get_now_time():
return [result_time, result_date, res_text, True]
else:
return [result_time, result_date, res_text, False]
return
def clear_posts_text_to_parse(domain=''):
posts = get_wall_posts(domain=domain)
for post in posts:
concert = post['text']
if 'лет' in concert:
clear_concert_text = concert.replace(
'лет',
'') # here we reaplace word 'years' becouse parser get mistake
concert_date = parse(clear_concert_text)
return concert, concert_date
elif '|' in concert: # we split text to create list where [1] object will be date "day(1) mounth(january)" and here we miss posts without "|" char
clear_concert_text = concert.split('|')
concert_date = parse(clear_concert_text[1])
return concert, concert_date
def message(update, context):
eta, text = None, None
mo = RE_MESSAGE.match(update.message.text)
if mo:
eta = arrow.get(arrow.get(mo.group(1)).datetime, tz.gettz())
text = mo.group(2)
else:
try:
eta = arrow.get(parse(update.message.text, allowed_results=(datetime,), tz=os.environ['TZ']))
text = get_last_clear_text(update.message.text)
except:
logging.exception('error parsing date')
logging.debug('eta: %s'%eta)
logging.debug('text: %s'%text)
if text:
sqs.delay_message(
eta = eta.to('UTC').naive,
text = text,
chat_id = update.effective_chat.id,
reply_to_message_id = update.message.message_id,
quote = True
)
bot.send_message(
text = _('will notify you at: %s')%eta.format(),
chat_id = update.effective_chat.id,
reply_to_message_id = update.message.message_id,
quote = True
)
else:
start(update, context)
def proc_rus_date(self, s):
if re.search(
"[0-9]{1,2}-[0-9]{1,2}-[0-9]{4}",
s): # костыль, чтобы rutimeparser понимал даты вида dd-mm-yyyy
res = re.search("([0-9]{1,2})-([0-9]{1,2})-([0-9]{4})", s)
s = "{}.{}.{}".format(res.group(1), res.group(2), res.group(3))
if 'феврал' in s.lower() and re.search(
"\\b29\\b", s
): # без этого костыля rutimeparser не понимает, что у февраля 29 дней
if re.search("\\b[0-9]{4}\\b", s):
year = int(re.search("\\b([0-9]{4})\\b", s).group(1))
else:
year = 2020
d = datetime.date(year, 2, 29)
else:
d = parse(s, allowed_results=[datetime.date, None])
if not d:
return None
if re.search(
"\\b[0-9]{4}\\b", s
): # костыль, потому что rutimeparser внезапно (!) неправильно видит год
year = int(re.search("\\b([0-9]{4})\\b", s).group(1))
if year != d.year:
d = datetime.date(year, d.month, d.day)
else:
now = datetime.datetime.now()
if now.year != d.year:
d = datetime.date(now.year, d.month, d.day)
return d
def test_weekday_and_time(self):
good = datetime(now.year, 4, 1, 17, 43)
if good < now:
good = datetime(now.year + 1, 4, 1, 17, 43)
wd = datetime.weekday(good)
days = 7 - (wd - 3) % 7
good = good + timedelta(days=days)
src_text = 'в апреле в четверг в 17:43'
with self.subTest(text=src_text):
for text in make_mutations(src_text):
result = parse(text, now=now)
self.compare(result, good, False)
def test_weekday(self):
year = now.year
good = datetime(year, 4, 1).date()
if good < now.date():
good = datetime(year + 1, 4, 1).date()
wd = datetime.weekday(good)
days = 7 - (wd - 2) % 7
good = good + timedelta(days=days)
for text in make_mutations('в апреле в среду'):
with self.subTest(text=text):
result = parse(text, now=now)
self.compare(result, good, True)
def test_delta(self):
texts = (
'8 апреля у меня поезд. Напомни за день.',
'9 апреля у меня поезд. Напомни за 2 дня.',
'6 апреля через день',
)
year = now.year
good = datetime(year, 4, 7)
if good < now:
good = datetime(year + 1, 4, 7)
for text in make_mutations(*texts):
with self.subTest(text=text):
result = parse(text, now=now)
self.compare(result, good, True)
def import_dt(text,
timezone='Europe/Moscow'
): # Функция забирает дату и время из сообщения
result = parse(text,
tz=timezone,
now=None,
remove_junk=True,
allowed_results=(datetime, None),
default_time=time(12, 0),
default_datetime=None)
dt_text = result
localtz = pytz.timezone(servertz)
timezone = pytz.timezone(timezone)
if result is None:
return 'null', 'null'
if datetime.now(tz=timezone) > result:
return 'old', 'old'
result = result.astimezone(tz=localtz)
result = ('{:%H:%M %d.%m.%Y}'.format(result))
dt_text = ('{:%H:%M %d.%m.%Y}'.format(dt_text))
return result, dt_text
def test_common(self):
for src_text, good, is_date in common_tests:
with self.subTest(src_text=src_text):
for text in make_mutations(src_text):
result = parse(text, now=now)
self.compare(result, good, is_date)