def get_absolute_year(self, entity, commonParser=None):
matcher = self.absolute_year_pat.match(entity)
if matcher:
if matcher.group(1):
year_ = digitconv.getNumFromHan(matcher.group(1))
if commonParser:
if year_ < 100:
year_ple_tw = 2000 + year_
year_ple_ni = 1900 + year_
if abs(commonParser.date.year -
year_ple_tw) >= abs(commonParser.date.year -
year_ple_ni):
commonParser.date = commonParser.date.replace(
year=year_ple_ni)
else:
commonParser.date = commonParser.date.replace(
year=year_ple_tw)
else:
commonParser.date = commonParser.date.replace(
year=year_)
commonParser.timeUnit[1] = True
return year_
else:
matcher = self.mix_year_pat.match(entity)
if matcher:
year_ = digitconv.getNumFromHan(matcher.group("year"))
if commonParser:
if year_ < 100:
year_ple_tw = 2000 + year_
year_ple_ni = 1900 + year_
if abs(commonParser.date.year -
year_ple_tw) >= abs(commonParser.date.year -
year_ple_ni):
commonParser.date = commonParser.date.replace(
year=year_ple_ni)
else:
commonParser.date = commonParser.date.replace(
year=year_ple_tw)
else:
commonParser.date = commonParser.date.replace(
year=year_)
commonParser.timeUnit[1] = True
if matcher.group("month"):
month = digitconv.getNumFromHan(matcher.group("month"))
commonParser.date = commonParser.date.replace(
month=month)
commonParser.timeUnit[2] = True
if matcher.group("day"):
day = digitconv.getNumFromHan(matcher.group("day"))
commonParser.date = commonParser.date.replace(
day=day)
commonParser.timeUnit[3] = True
return year_
return None
def get_dev_numlist(self, numberlist):
"""
Function:
entity has the number list "一、二、三...."
Parameters:
1.numberlist: list, '0-9零一二三四五六七八九十'
Return: start_month, end_month
"""
season_start = digitconv.getNumFromHan(numberlist[0]) - 1
season_end = digitconv.getNumFromHan(numberlist[-1]) - 1
start_month = self.season_dict[season_start][0]
end_month = self.season_dict[season_end][1]
return start_month, end_month
def frequency_output(self, prefix_word, number, suffix_word):
"""
Function:
Process the time frequency, and get the output of the frequency.
Parameters:
1.prefix_word: str, "每|每隔"
2.number: str, "个|0-9零一二三四五六七八九|半"
3.suffix_word: str, "年|寒暑假|季度|月|周|星期|礼拜|日|晚上|时|分|秒"
Return:
1.time_fre_list: list, represent the "年、月、周、日、时、分、秒"
"""
time_fre_list = [0, 0, 0, 0, 0, 0, 0]
if (prefix_word == u'每逢'):
time_indx = self.fre_indx_whenever[suffix_word]
time_fre_list[time_indx[0]] = time_indx[1]
else:
shift = 0.0
if (prefix_word == u'每隔'):
shift = 1.0
fre_num = 1.0
try:
fre_num = int(digitconv.getNumFromHan(number))
except:
if (number == u'半' and prefix_word == u'每隔'):
fre_num = 0.5
shift = 0.5
elif (number == u'半'):
fre_num = 0.5
time_indx = self.fre_indx_dict[suffix_word]
time_fre_list[time_indx[0]] = shift + fre_num * time_indx[1]
return time_fre_list
def get_relative_year(self,
entity,
basetime='2018-11-01 12:00:00',
commonParser=None):
time = arrow.get(basetime)
base_year = time.year
util = util_tools.Util()
matcher = self.relative_year_pat.match(entity)
decorate_deviation = ''
if matcher:
if matcher.group(1):
decorate_deviation = matcher.group(1)
year_shift = util.rel_deviation(matcher.group(2),
decorate_deviation)
if commonParser:
commonParser.timeUnit[1] = True
commonParser.date = commonParser.date.shift(years=year_shift)
return year_shift + base_year
else:
matcher = self.relative_year_diviation_pat.match(entity)
if matcher:
year_ = 1
if matcher.group(1):
year_ = digitconv.getNumFromHan(matcher.group(1))
year_shift = util.rel_deviation_num(matcher.group(2), year_)
if commonParser:
commonParser.timeUnit[7] = True
commonParser.date = commonParser.date.shift(
years=year_shift)
return base_year + year_shift
return None
def get_dev_up2down(self, basemonth, deviation):
"""
Function:
process the sample like "上*", "下*" and "这"
Parameters:
1.basemonth: str, current month of basetime
2.deviation: str, get the value according to the pattern "[1-4一二三四|下|上|这]"
"""
#get the now season according to base month
shift_season = 0
basemonth = int(basemonth)
for season in self.season_dict.keys():
if (basemonth >= self.season_dict[season][0]
and basemonth <= self.season_dict[season][1]):
shift_season = season
break
if u'上' in deviation:
for i in range(len(deviation)):
shift_season -= 1
elif u'下' in deviation:
for i in range(len(deviation)):
shift_season += 1
elif deviation in u'1-4一二三四':
shift_season = digitconv.getNumFromHan(deviation) - 1
#get the shift of year and season
year = shift_season / 4
shift_season = shift_season % 4
return year, shift_season
def get_dev_aroundlist(self, year, season_number, around_text):
"""
Function:
entity has the around_text "去|今|明" "前|后"
Parameters:
1.year: int
2.season_number: str, '0-9零一二三四五六七八九十'
3.around_text: list, ["去|今", "前|后"]
Return: start_year, end_year, start_month, end_month
"""
start_month = 0
end_month = 0
start_year = 0
end_year = 0
if (around_text[0] == u'去'):
start_year = year - 1
end_year = year - 1
elif (around_text[0] == u'今'):
start_year = year
end_year = year
elif (around_text[0] == u'明'):
start_year = year + 1
end_year = year + 1
shift = digitconv.getNumFromHan(season_number) - 1
if (around_text[1] == u'前'):
start_month = 1
end_month = 3 + 3 * shift
elif (around_text[1] == u'后'):
start_month = 10 - 3 * shift
end_month = 12
return start_year, end_year, start_month, end_month
def get_relative_month(self, entity, basetime='2018-11-01 12:00:00', commonParser=None):
matcher = self.relative_month_pat.match(entity)
decorate_deviation,deviation = '',''
util = util_tools.Util()
time = arrow.get(basetime)
base_month = time.month
if matcher:
if matcher.group(1):
decorate_deviation = matcher.group(1)
shift_month = util.rel_deviation(matcher.group(2), decorate_deviation)
if commonParser:
commonParser.timeUnit[2] = True
if shift_month != 0:
commonParser.date = commonParser.date.shift(months=shift_month)
return base_month + shift_month
else:
matcher = self.relative_month_diviation_pat.match(entity)
if matcher:
month_ = digitconv.getNumFromHan(matcher.group(1))
shift_month = util.rel_deviation_num(matcher.group(2), month_)
if commonParser:
commonParser.timeUnit[7] = True
commonParser.date = commonParser.date.shift(months=shift_month)
return shift_month + base_month
return None
def get_absolute_second(self, entity, commonParser=None):
matcher = self.absolute_second_pat.match(entity)
if matcher:
chsecond = matcher.group(1)
second_ = digitconv.getNumFromHan(chsecond)
if commonParser:
commonParser.timeUnit[7] = True
commonParser.date = commonParser.date.replace(second=second_)
return second_
return None
def get_absolute_minute(self, entity, commonParser=None):
matcher = self.absolute_minute_pat.match(entity)
ke_matcher = self.ke_pat.match(entity)
if matcher:
minute_ = digitconv.getNumFromHan(matcher.group(1))
if commonParser:
if matcher.group(2):
commonParser.timeUnit[7] = True
else:
commonParser.timeUnit[6] = True
if minute_ == 60:
commonParser.date = commonParser.date.replace(
hour=commonParser.date.hour + 1).replace(minute=0)
elif 0 <= minute_ < 60:
commonParser.date = commonParser.date.replace(
minute=minute_)
return None
else:
commonParser.timeFormatInfo = u'时间表述异常，分钟应该在[0,60]范围内！'
return None
return minute_
elif ke_matcher:
minuteToParse = digitconv.getNumFromHan(ke_matcher.group("minute"))
minute_ = minuteToParse * 15
if commonParser:
if matcher.group(2):
commonParser.timeUnit[7] = True
else:
commonParser.timeUnit[6] = True
if minute_ == 60:
commonParser.date = commonParser.date.replace(
hour=commonParser.date.hour + 1).replace(minute=0)
elif 0 <= minute_ < 60:
commonParser.date = commonParser.date.replace(
minute=minute_)
return None
else:
commonParser.timeFormatInfo = u'时间表述异常，分钟应该在[0,60]范围内！'
return None
return minute_
return None
def get_absolute_day(self, entity, commonParser=None):
matcher = self.absolute_day_pat.match(entity)
if matcher:
if matcher.group("day"):
day_ = matcher.group("day")[1:]
day = digitconv.getNumFromHan(day_)
if commonParser:
commonParser.date = commonParser.date.replace(day=day)
commonParser.timeUnit[3] = True
commonParser.lunar = True
return day_
return None
def get_absolute_day(self, entity, commonParser=None):
matcher = self.absolute_day_pat.match(entity)
if matcher:
if matcher.group(1):
day_ = digitconv.getNumFromHan(matcher.group(1))
if commonParser:
commonParser.timeUnit[3] = True
if 1 <= day_ <= 31:
commonParser.date = commonParser.date.replace(day=day_)
else:
commonParser.timeFormatInfo = u'时间表述异常，日应该在[1,31]范围内！'
return None
return day_
return None
def get_absolute_month(self, entity, commonParser=None):
matcher = self.absolute_month_pat.match(entity)
if matcher:
if matcher.group(1):
month_ = digitconv.getNumFromHan(matcher.group(1))
if commonParser:
commonParser.timeUnit[2] = True
if 1 <= month_ <= 12:
commonParser.date = commonParser.date.replace(month=month_)
else:
commonParser.timeFormatInfo = u'时间表述异常，月份应该在[1,12]范围内！'
return None
return month_
return None
def duration_output(self, number, suffix_word):
"""
Function:
Process the time duration, and get the output of the duration.
Parameters:
1.number: str, samples like the "2"; list, [number1, number2] and samples like "2.3";
list, [number1, '半'] and samples like the "一个半"
2.suffix_word: str, "世纪|年|周年|月|星期|礼拜|周末|周|天|工作日|夜|晚|小时|分钟|秒"
Return:
1.time_dur_list: list, represent the "年、月、周、日、时、分、秒"
"""
time_dur_list = [0, 0, 0, 0, 0, 0, 0]
shift = 0.0
if (type(number) == list):
if (number[1] == u'个半'):
#process the "一个半"
shift = digitconv.getNumFromHan(number[0]) + 0.5
elif (number[1] == u'整'):
shift = digitconv.getNumFromHan(number[0])
else:
shift = str(digitconv.getNumFromHan(number[0])) + "." + \
str(digitconv.getNumFromHan(number[1]))
shift = float(shift)
else:
if (number == u'半'):
shift = 0.5
elif (number == u'整'):
shift = 1.0
else:
shift = float(digitconv.getNumFromHan(number))
time_indx = self.duration_dict[suffix_word]
time_dur_list[time_indx[0]] = shift * time_indx[1]
return time_dur_list
def season_end(self, basetime, season_number, season_word):
"""
Function:
process the "季度末"
Parameter:
1.basetime: str
2.season_number: str, '[1-4一二三四]';
3.season_word: str, "季度末"
"""
year = arrow.get(basetime).year
if (season_word == u'季度末'):
season_number = digitconv.getNumFromHan(season_number)
start_month = self.season_dict[season_number - 1][1]
end_month = self.season_dict[season_number - 1][1]
return year, year, start_month, end_month
def get_dev_around(self, year, basemonth, season_number, around_text):
"""
Function:
entity has the around_text "前|后"
Parameters:
1.year: int
2.basemonth: str,
3.season_number: str, '0-9零一二三四五六七八九十' or "上下"
4.around_text: str, "前|后" or "去"
Return: start_year, end_year, start_month, end_month
"""
basemonth = int(basemonth)
start_month = 0
end_month = 0
for season in self.season_dict.keys():
if (basemonth >= self.season_dict[season][0]
and basemonth <= self.season_dict[season][1]):
start_month = self.season_dict[season][0]
end_month = self.season_dict[season][1]
break
shift = 0
#if the season_number have the "上下"
if (u'上' in season_number or u'下' in season_number):
if (u'上' in season_number):
shift = -(len(season_number) + 1)
end_month = start_month - 1
start_month = start_month + shift * 3
if (u'下' in season_number):
shift = len(season_number) + 1
start_month = (end_month + 1) % 12
end_month = end_month + shift * 3
else:
if (around_text == u'前' or around_text == u'上'):
shift = -digitconv.getNumFromHan(season_number)
end_month = start_month - 1
start_month = start_month + shift * 3
elif (around_text == u'后' or around_text == u'下'):
shift = digitconv.getNumFromHan(season_number)
start_month = (end_month + 1) % 12
end_month = end_month + shift * 3
elif (around_text == u'去'):
shift = digitconv.getNumFromHan(season_number) - 1
start_month = self.season_dict[shift][0]
end_month = self.season_dict[shift][1]
year = year - 1
return year, year, start_month, end_month
#carry and abdicate the year
start_year = year
end_year = year
if (start_month <= 0 and end_month >= 0):
start_month += -1
start_year = year - (abs(start_month) / 12 + 1)
end_year = year
start_month = 12 - abs(start_month + 1) % 12
elif (start_month >= 0 and end_month >= 12):
start_year = year + (end_month / 12)
end_year = year + (end_month / 12)
end_month = end_month % 12
#when the end_month=12, the value of it will be 0. so reset the value of end_month
if (end_month == 0):
end_month = 12
end_year -= 1
return start_year, end_year, start_month, end_month
def get_absolute_hour(self, entity, commonParser=None):
hour_xu_matcher = self.hour_xu_pat.match(entity)
qunima_matcher = self.qunima_pat.match(entity)
gundan_matcher = self.gundan_pat.match(entity)
hour_minute_guo_matcher = self.hour_minute_guo_pat.match(entity)
hour_guo_ke_matcher = self.hour_guo_ke_pat.match(entity)
hour_minute_matcher = self.hour_minute_pat.match(entity)
hour_ke_matcher = self.hour_ke_pat.match(entity)
hour_matcher = self.hour_pat.match(entity)
apm = Apm()
if qunima_matcher:
minuteToParse = digitconv.getNumFromHan(qunima_matcher.group("minute"))
chhour = qunima_matcher.group(4)
hour_ = digitconv.getNumFromHan(chhour)
minute_ = digitconv.getNumFromHan(minuteToParse)
if qunima_matcher.group("guo"):
guocha = qunima_matcher.group("guo")
if guocha == u'差':
hourcha = minute_ / 60;
hour_ = hour_ - hourcha - 1;
minute_ = 60 - minute_ % 60;
return None
if commonParser:
if commonParser.timeUnit[4]:
hour_ = apm.adjustHours(entity, hour_, commonParser)
commonParser.timeUnit[5] = True
commonParser.timeUnit[6] = True
if minute_ == 60:
hour_ = hour_ + 1
elif 0 <= minute_ < 60:
commonParser.date = commonParser.date.replace(hour=hour_).replace(minute=minute_)
return None
else:
commonParser.timeFormatInfo = u'时间表述异常，分钟应该在[0,60]范围内！'
return None
if hour_ == 24:
commonParser.date = commonParser.date.replace(days=commonParser.date.day + 1).replace(
hour=0).replace(minute=0)
else:
if 0 <= hour_ < 24:
commonParser.date = commonParser.date.replace(
hour=hour_).replace(minute=0)
else:
commonParser.timeFormatInfo = u'时间表述异常，小时应该在[0,24]范围内！'
return None
return hour_
elif gundan_matcher:
minuteToParse = digitconv.getNumFromHan(qunima_matcher.group("minute"))
if minuteToParse != 0:
minute_ = minuteToParse * 15
chhour = qunima_matcher.group(4)
hour_ = digitconv.getNumFromHan(chhour)
if gundan_matcher.group("guo"):
guocha = qunima_matcher.group("guo")
if guocha == u'差':
hourcha = minute_ / 60
hour_ = hour_ - hourcha - 1
minute_ = 60 - minute_ % 60
return None
if commonParser:
if commonParser.timeUnit[4]:
hour_ = apm.adjustHours(entity, hour_, commonParser)
commonParser.timeUnit[5] = True
commonParser.timeUnit[6] = True
if minute_ == 60:
hour_ = hour_ + 1
elif 0 <= minute_ < 60:
commonParser.date = commonParser.date.replace(hour=hour_).replace(minute=minute_)
return None
else:
commonParser.timeFormatInfo = u'时间表述异常，分钟应该在[0,60]范围内！'
return None
if hour_ == 24:
commonParser.date = commonParser.date.replace(days=commonParser.date.day + 1).replace(
hour=0).replace(minute=0)
else:
if 0 <= hour_ < 24:
commonParser.date = commonParser.date.replace(
hour=hour_).replace(minute=0)
else:
commonParser.timeFormatInfo = u'时间表述异常，小时应该在[0,24]范围内！'
return None
return hour_
elif hour_minute_guo_matcher:
hourToParse = hour_minute_guo_matcher.group(1)
hour_ = digitconv.getNumFromHan(hourToParse)
minute_ = digitconv.getNumFromHan(hour_minute_guo_matcher.group("minute"))
if hour_minute_guo_matcher.group("guo"):
guocha = hour_minute_guo_matcher.group("guo")
if guocha == u"差" or guocha == u"少":
hourcha = minute_ / 60
hour_ = hour_ - hourcha-1
minute_ = 60-minute_ % 60
if commonParser:
if commonParser.timeUnit[4]:
hour_ = apm.adjustHours(entity, hour_, commonParser)
commonParser.timeUnit[5] = True
commonParser.timeUnit[6] = True
if minute_ == 60:
hour_ = hour_ + 1
elif 0 <= minute_ < 60:
commonParser.date = commonParser.date.replace(hour=hour_).replace(minute=minute_)
return None
else:
commonParser.timeFormatInfo = u'时间表述异常，分钟应该在[0,60]范围内！'
return None
if hour_ == 24:
commonParser.date = commonParser.date.replace(days=commonParser.date.day + 1).replace(
hour=0).replace(minute=0)
else:
if 0 <= hour_ < 24:
commonParser.date = commonParser.date.replace(
hour=hour_).replace(minute=0)
else:
commonParser.timeFormatInfo = u'时间表述异常，小时应该在[0,24]范围内！'
return None
return hour_
elif hour_guo_ke_matcher:
chhour = hour_guo_ke_matcher.group(1)
hour_ = digitconv.getNumFromHan(chhour)
minuteToParse = digitconv.getNumFromHan(hour_guo_ke_matcher.group("minute"))
minute_ = minuteToParse * 15
if hour_guo_ke_matcher.group("guo"):
guocha = hour_guo_ke_matcher.group("guo")
if guocha == u"差" or guocha == u"少":
hourcha = minute_ / 60
hour_ = hour_ - hourcha - 1
minute_ = 60 - minute_ % 60
if commonParser:
if commonParser.timeUnit[4]:
hour_ = apm.adjustHours(entity, hour_, commonParser)
commonParser.timeUnit[5] = True
commonParser.timeUnit[6] = True
if minute_ == 60:
hour_ = hour_ + 1
elif 0 <= minute_ < 60:
commonParser.date = commonParser.date.replace(hour=hour_).replace(minute=minute_)
return None
else:
commonParser.timeFormatInfo = u'时间表述异常，分钟应该在[0,60]范围内！'
return None
if hour_ == 24:
commonParser.date = commonParser.date.replace(days=commonParser.date.day + 1).replace(
hour=0).replace(minute=0)
else:
if 0 <= hour_ < 24:
commonParser.date = commonParser.date.replace(
hour=hour_).replace(minute=0)
else:
commonParser.timeFormatInfo = u'时间表述异常，小时应该在[0,24]范围内！'
return None
return hour_
elif hour_minute_matcher:
chhour = hour_minute_matcher.group(1)
hour_ = digitconv.getNumFromHan(chhour)
chminute = hour_minute_matcher.group("minute")
minute_ = digitconv.getNumFromHan(chminute)
if commonParser:
if commonParser.timeUnit[4]:
hour_ = apm.adjustHours(entity, hour_, commonParser)
commonParser.timeUnit[5] = True
commonParser.timeUnit[6] = True
if minute_ == 60:
hour_ = hour_+1
elif 0 <= minute_ < 60:
commonParser.date = commonParser.date.replace(hour=hour_).replace(minute=minute_)
return None
else:
commonParser.timeFormatInfo = u'时间表述异常，分钟应该在[0,60]范围内！'
return None
if hour_ == 24:
commonParser.date = commonParser.date.replace(days=commonParser.date.day+1).replace(hour=0).replace(minute=0)
else:
if 0 <= hour_ < 24:
commonParser.date = commonParser.date.replace(
hour=hour_).replace(minute=0)
else:
commonParser.timeFormatInfo = u'时间表述异常，小时应该在[0,24]范围内！'
return None
return hour_
elif hour_ke_matcher:
hour_ = digitconv.getNumFromHan(hour_ke_matcher.group(1))
chminute = hour_ke_matcher.group("minute")
minute_ = digitconv.getNumFromHan(chminute) * 15
if commonParser:
if commonParser.timeUnit[4]:
hour_ = apm.adjustHours(entity, hour_, commonParser)
commonParser.timeUnit[5] = True
commonParser.timeUnit[6] = True
if minute_ == 60:
hour_ = hour_+1
elif 0 <= minute_ < 60:
commonParser.date = commonParser.date.replace(hour=hour_).replace(minute=minute_)
return None
else:
commonParser.timeFormatInfo = u'时间表述异常，分钟应该在[0,60]范围内！'
return None
if hour_ == 24:
commonParser.date = commonParser.date.replace(days=commonParser.date.day+1).replace(hour=0).replace(minute=0)
else:
if 0 <= hour_ < 24:
commonParser.date = commonParser.date.replace(
hour=hour_).replace(minute=0)
else:
commonParser.timeFormatInfo = u'时间表述异常，小时应该在[0,24]范围内！'
return None
return hour_
elif hour_matcher:
chhour = hour_matcher.group(1)
hour_ = digitconv.getNumFromHan(chhour)
minuteToParse = hour_matcher.group(2)
hour_ = apm.adjustHours(entity, hour_, commonParser)
if minuteToParse == u"半":
minute_ = 30
else:
minute_ = digitconv.getNumFromHan(minuteToParse)
if commonParser:
commonParser.timeUnit[5] = True
commonParser.timeUnit[6] = True
commonParser.date = commonParser.date.replace(hour=hour_).replace(minute=minute_)
return hour_
elif hour_xu_matcher:
chhour = hour_xu_matcher.group(1)
hour_ = digitconv.getNumFromHan(chhour)
if commonParser:
if commonParser.timeUnit[4]:
hour_ = apm.adjustHours(entity, hour_, commonParser)
commonParser.timeUnit[5] = True
if u'许' in entity or u'整' in entity:
commonParser.timeUnit[6] = True
commonParser.timeUnit[7] = True
if hour_ == 24:
commonParser.date = commonParser.date.shift(days=1).replace(hour=0)
else:
commonParser.date = commonParser.date.replace(hour=hour_)
return hour_
return None