def test_match_digi_clock():
fn = 'digi_clock'
tf = [('kedden 8:45', [[Hour(8, fn), Minute(45, fn)]]),
('kedden 08:45', [[Hour(8, fn), Minute(45, fn)]]),
('ma este 18:12-kor', [[Hour(18, fn), Minute(12, fn)]])]
for inp, exp in tf:
out = match_digi_clock(inp)
date_parts = []
for e in out:
date_parts.append(e['date_parts'])
assert date_parts == exp
def match_digi_clock(s: str) -> List[Dict[str, Any]]:
"""
Match digi clock format.
:param s: textual input
:return: tuple of date parts
"""
match = re.findall(R_DIGI, s)
res = []
for group in match:
group = [int(m.lstrip('0')) for m in group if m.lstrip('0')]
if len(group) == 2:
h, m = group
res.append({
'match':
group,
'date_parts': [Hour(h, 'digi_clock'),
Minute(m, 'digi_clock')]
})
elif len(group) == 1:
res.append({
'match': group,
'date_parts': [Hour(group[0], 'digi_clock')]
})
return res
def match_now(s: str, now: datetime) -> List[Dict[str, Any]]:
if match_weekday(s, now):
return []
match = re.match(r'.*\bmost\b.*', s.lower())
if match:
date_parts = [
Year(now.year, 'now'),
Month(now.month, 'now'),
Day(now.day, 'now'),
Hour(now.hour, 'now'),
Minute(now.minute, 'now')
]
return [{'match': 'most', 'date_parts': date_parts}]
return []
def test_match_now():
now = datetime(2020, 12, 30, 12, 1)
fn = 'now'
tf = [('kedden 8:45', []), ('most kedden', []),
('legyen most', [[
Year(now.year, fn),
Month(now.month, fn),
Day(now.day, fn),
Hour(now.hour, fn),
Minute(now.minute, fn)
]])]
for inp, exp in tf:
out = match_now(inp, now)
date_parts = []
for e in out:
date_parts.append(e['date_parts'])
assert date_parts == exp
def match_n_periods_compared_to_now(s: str, now: datetime) -> List[Dict[str, Any]]:
# TODO: implement n-periods-before-now functionality
fn = 'n_date_periods_compared_to_now'
regexes = [
(R_NWEEKS_FROM_NOW, 'w'),
(R_NDAYS_FROM_NOW, 'd'),
(R_NHOURS_FROM_NOW, 'h'),
(R_NMINS_FROM_NOW, 'm')
]
res = []
for regex, freq in regexes:
groups = re.findall(regex, s)
for group in groups:
date_parts = {'match': group, 'date_parts': []}
n = group[1]
if n:
n = word_to_num(n)
if freq == 'w':
res_dt = (now + timedelta(days=7*n))
y, m, d = res_dt.year, res_dt.month, res_dt.day
date_parts['date_parts'].extend([Year(y, fn), Month(m, fn), Day(d, fn)])
elif freq == 'd':
res_dt = (now + timedelta(days=n))
y, m, d = res_dt.year, res_dt.month, res_dt.day
date_parts['date_parts'].extend([Year(y, fn), Month(m, fn), Day(d, fn)])
elif freq == 'h':
res_dt = (now + timedelta(hours=n))
y, m, d, h = res_dt.year, res_dt.month, res_dt.day, res_dt.hour
date_parts['date_parts'].extend([Year(y, fn), Month(m, fn), Day(d, fn), Hour(h, fn)])
elif freq == 'm':
res_dt = (now + timedelta(minutes=n))
y, mo, d, h, mi = res_dt.year, res_dt.month, res_dt.day, res_dt.hour, res_dt.minute
date_parts['date_parts'].extend([Year(y, fn), Month(mo, fn),
Day(d, fn), Hour(h, fn), Minute(mi, fn)])
res.append(date_parts)
return res
tf = [('kedden 8:45', [[Hour(8, fn), Minute(45, fn)]]),
('kedden 08:45', [[Hour(8, fn), Minute(45, fn)]]),
('ma este 18:12-kor', [[Hour(18, fn), Minute(12, fn)]])]
for inp, exp in tf:
out = match_digi_clock(inp)
date_parts = []
for e in out:
date_parts.append(e['date_parts'])
assert date_parts == exp
time_word_fn = 'time_words'
time_word_scenarios = [
('reggel nyolc előtt hat perccel',
[[Hour(7, time_word_fn), Minute(54, time_word_fn)]]),
('reggel nyolc előtt nyolcvan perccel',
[[Hour(6, time_word_fn), Minute(40, time_word_fn)]]),
('este 8 előtt 12 perccel',
[[Hour(19, time_word_fn),
Minute(48, time_word_fn)]]),
('nyolc óra nyolc perckor',
[[Hour(8, time_word_fn), Minute(8, time_word_fn)]]),
('ma reggel hat óra', [[Hour(6, time_word_fn)]]),
('ma reggel', [[Daypart(1, time_word_fn)]]),
('ma délután háromkor', [[Hour(15, time_word_fn)]]),
('ma délután három után negyvenhat perckor',
[[Hour(15, time_word_fn),
Minute(46, time_word_fn)]]),
('ma délután haemdknc után negyvenhat perckor',
[[Daypart(3, time_word_fn)]]), ('ma', []), ('ötvenöt perckor', []),
def match_time_words(s: str) -> List[Dict[str, Any]]:
"""
:param s: textual input
:return: tuple of date parts
"""
parts = _raw_match_time_words(s)
if not parts:
return []
else:
group, daypart, hour_modifier, hour, minute = parts
# Only numbers can match dates as well, this is an attempt to remove false matches
hour_index = s.index(f'{hour}')
before_hour = s[:hour_index].split()
if before_hour:
months = [
'jan', 'feb', 'mar', 'apr', 'maj', 'jun', 'jul', 'aug', 'szep',
'okt', 'nov', 'dec'
]
for month in months:
if month in remove_accent(before_hour[-1]):
return []
# Fix false time match for input 'jövő hét'
if remove_accent(hour) == 'het':
hour_indeces = [
m.start() for m in re.finditer('het(?!fo)', remove_accent(s))
]
if hour_indeces:
before_hour = s[:hour_indeces[-1]].split()
if before_hour:
if 'jovo' in remove_accent(before_hour[-1]):
return []
res = []
am = True
date_parts = []
if daypart and hour:
if 'reggel' in daypart or 'delelott' in remove_accent(
daypart) or 'hajnal' in daypart:
am = True
elif 'delutan' in remove_accent(
daypart) or 'este' in daypart or 'ejjel' in remove_accent(
daypart):
am = False
if hour:
# SKIP the whole matching rule when any of these apply
# TODO: come up with a more elegant solution for this
# TODO: i.e: by implementing the possibility of one rule exclude another
# this is made redundant by the change in the patterns
non_hours = ['ev', 'perc']
for nh in non_hours:
if f' {nh}' in remove_accent(hour) or remove_accent(
hour).startswith(nh):
return []
if 'mulva' in remove_accent(s):
return []
hour_num = word_to_num(hour)
minute_num = word_to_num(minute)
if not daypart:
# default to business hour if daypart is not specified
if hour_num < 8:
hour_num += 12
if hour_modifier:
if 'haromnegyed' in remove_accent(hour_modifier):
hour_num = hour_num - 1 if hour_num - 1 >= 0 else 23
minute_num = 45
elif 'fel' in remove_accent(hour_modifier):
hour_num = hour_num - 1 if hour_num - 1 >= 0 else 23
minute_num = 30
elif 'negyed' in remove_accent(hour_modifier):
hour_num = hour_num - 1 if hour_num - 1 >= 0 else 23
minute_num = 15
if hour_num == NAN:
return []
else:
if hour_num < 12 and not am:
hour_num += 12
if minute or hour_modifier:
# this is made redundant by the change in the patterns
non_minutes = ['ev', 'ora']
for nm in non_minutes:
if f' {nm}' in remove_accent(minute) or remove_accent(
minute).startswith(nm):
return []
if 'elott' in remove_accent(minute) and not hour_modifier:
hour_num -= (minute_num // 60) + 1
hour_num = hour_num if hour_num >= 0 else 23
date_parts.extend([
Hour(hour_num, 'time_words'),
Minute(60 - (minute_num % 60), 'time_words')
])
elif 'elott' in remove_accent(minute) and hour_modifier:
n_minutes_before = word_to_num(minute)
if n_minutes_before != NAN:
minute_num -= n_minutes_before
if minute_num < 0:
hour_num += (minute_num // 60)
hour_num = hour_num if hour_num >= 0 else 23
minute_num = minute_num % 60
date_parts.extend([
Hour(hour_num, 'time_words'),
Minute(minute_num, 'time_words')
])
elif hour_modifier:
n_minutes_after = word_to_num(minute)
if n_minutes_after != NAN:
minute_num += n_minutes_after
if minute_num > 59:
hour_num += (minute_num // 60)
hour_num = hour_num if hour_num <= 23 else 0
minute_num = minute_num % 60
date_parts.extend([
Hour(hour_num, 'time_words'),
Minute(minute_num, 'time_words')
])
else:
date_parts.extend([
Hour(hour_num, 'time_words'),
Minute(minute_num, 'time_words')
])
else:
date_parts.append(Hour(hour_num, 'time_words'))
res.append({'match': group, 'date_parts': date_parts})
elif daypart:
if 'hajnal' in daypart:
res.append({
'match': group,
'date_parts': [Daypart(0, 'time_words')]
})
elif 'reggel' in daypart:
res.append({
'match': group,
'date_parts': [Daypart(1, 'time_words')]
})
elif 'delelott' in remove_accent(daypart):
res.append({
'match': group,
'date_parts': [Daypart(2, 'time_words')]
})
elif 'delutan' in remove_accent(daypart):
res.append({
'match': group,
'date_parts': [Daypart(3, 'time_words')]
})
elif 'este' in daypart:
res.append({
'match': group,
'date_parts': [Daypart(4, 'time_words')]
})
elif 'ejjel' in remove_accent(daypart):
res.append({
'match': group,
'date_parts': [Daypart(5, 'time_words')]
})
return res