def m_datetime_only_hour_minute(seq):
"""
时间字符串只保留时分
:param seq: 日期字符串 或 日期字符串组成的列表
:return: 日期字符串只保留时分
example：
:seq ['2017-01-01 12:20:00', '2017-02-28 09:10:30']
:return ['12:20', '09:10']
"""
if not isinstance(seq, list):
try:
seq = datetime.strptime(seq, "%Y-%m-%d %H:%M:%S")
except:
raise FeatureProcessError('%s time format error ' % seq)
else:
try:
seq = map(lambda x: datetime.strptime(x, "%Y-%m-%d %H:%M:%S"), seq)
except:
raise FeatureProcessError('%s time format error ' % seq)
seq = float(str(seq.hour) + '.' + str(seq.minute))
return seq
def f_assert_must_digit(value_list, args=False):
"""
检测列表中的元素是否为数字
:param value_list: 待检测列表
:param args: 负数是否通过 false 不通过报异常 True 负数通过
:return: 异常或原值
example：
:value_list [-2,'-2', 3]
:args false
：return 异常
:value_list [-2,'-2', 3]
:args True
：return [-2,'-2', 3]
"""
for value in value_list:
if args:
if not str(value).lstrip('-').isdigit():
raise FeatureProcessError(
'%s negative number=%s f_assert_must_digit Error' %
(value_list, args))
else:
if not str(value).isdigit():
raise FeatureProcessError(
'%s negative number=%s f_assert_must_digit Error' %
(value_list, args))
return value_list
def m_city_name_to_level2_0(city_name):
"""
获取城市名所对应的level
:param city_name: 城市
:return: code
example：
:address '北京'
:return 1
"""
ccf = CityCodeField.objects.filter(city_name_cn=city_name, is_delete=False)
if not ccf:
raise FeatureProcessError(
'not find %s level config in database table' % city_name)
company_addr_city_level = ccf[0].city_level
if not str(company_addr_city_level).isdigit():
raise FeatureProcessError(
'%s city_level config error in database table' % city_name)
seq = int(company_addr_city_level)
if seq == 1:
seq = "一线"
elif seq == 2:
seq = "二线"
elif seq == 3:
seq = "三线"
elif seq == 4:
seq = "四线"
elif seq == 5:
seq = "其他"
return seq
def f_assert_must_digit_or_float(value_list, args=False):
"""
检测列表中的元素是否为数字或float, args=false 负数报异常 True 负数通过
:param value_list: 待检测列表
:param args: 负数是否通过 false 不通过报异常 True 负数通过
:return: 异常或原值
example：
:value_list [-2.0,'-2', 3]
:args false
：return 异常
:value_list [-2.0,'-2', 3]
:args True
：return [-2.0,'-2', 3]
"""
for value in value_list:
if args:
if not (str(value).count('.') <= 1
and str(value).replace('.', '').lstrip('-').isdigit()):
raise FeatureProcessError(
'%s negative number=%s f_assert_must_digit_or_float Error'
% (value_list, args))
else:
if not (str(value).count('.') <= 1
and str(value).replace('.', '').isdigit()):
raise FeatureProcessError(
'%s negative number=%s f_assert_must_digit_or_float Error'
% (value_list, args))
return value_list
def f_assert_not_null(seq):
"""检测值是否非空或值得列表是否存在非空元素"""
if seq in (None, '', [], {}, ()):
raise FeatureProcessError("value: %s f_assert_not_null Error" % seq)
if isinstance(seq, list):
for value in seq:
if value in (None, '', {}, [], ()):
raise FeatureProcessError("value: %s f_assert_not_null Error" %
seq)
return seq
def func_exec_operator_chain(value, chains):
chain_list = chains.split('->') if chains else []
for operator in chain_list:
if '#' in operator:
judge = eval(operator[1:])
if judge:
continue
else:
raise FeatureProcessError('exec_operator_chain Error value: %s assert: %s' % (value, operator[1:]))
try:
value = eval(operator)
except:
raise FeatureProcessError('exec_operator_chain Error value: %s operator: %s' % (value, operator))
return value
def f_assert_must_basestring(value_list):
"""检测列表中的元素是否为字符串"""
for value in value_list:
if not isinstance(value, basestring):
raise FeatureProcessError('%s f_assert_must_basestring Error' %
value_list)
return value_list
def f_assert_must_dict(value_list):
"""检测列表中的元素是否为dict类型"""
for value in value_list:
if not isinstance(value, dict):
raise FeatureProcessError('%s f_assert_must_dict Error' %
value_list)
return value_list
def m_single_check_code(seq, feature_name):
"""
获取单值匹配(不是区间)所对应的Code
:param feature_name: 特征名称
:param seq: 特征对应的返回值
:return: code
example：
:feature_name education_degree_code
:seq: 20
:return 2
"""
if not seq:
return []
res = ''
feature_code = FeatureCodeMapping.objects.filter(
feature_name=feature_name, )
num_map = {
int(conf.mapped_value): conf.unitary_value
for conf in feature_code
}
for key, value in num_map.iteritems():
if seq[0] == value:
res = key
break
if res in ('', None):
raise FeatureProcessError("don't find %s=%s code value" %
(feature_name, seq))
return res
def func_exec_chain(data, chains):
chain_list = chains.split('->') if chains else []
for func in chain_list:
args = ''
try:
f = re.search("(.*)\\((.*)\\)", func)
if f:
func = f.group(1)
args = f.group(2)
func = eval(func)
except:
raise FeatureProcessError("exec_chain Error: don't find function %s" % func)
if args:
if args.startswith("[") or args.startswith("{"):
args = eval(args)
else:
args = args.split(',')
args_tmp = []
for i in args:
if i.startswith("\'"):
arg = i.strip("\'")
elif i.startswith("""\""""):
arg = i.strip("\"")
else:
arg = eval(i)
args_tmp.append(arg)
args = args_tmp
if len(args) == 1:
args = args[0]
# print args
data = func(data, args)
else:
data = func(data)
return data
def get(self, feature_name):
if self._feature_code_context is None:
self._load_data_from_db()
if not isinstance(self._feature_code_context, dict):
raise FeatureProcessError(
"Feature code context is not initialize.")
return self._feature_code_context.get(feature_name, None)
def m_city_name_to_code(city_name):
"""
获取城市名所对应的Code
:param city_name: 城市
:return: code
example：
:address '北京'
:return 1
"""
ccf = CityCodeField.objects.filter(city_name_cn=city_name, is_delete=False)
if not ccf:
raise FeatureProcessError('not find %s code config in database table' %
city_name)
company_addr_city_code = ccf[0].city_code
if not str(company_addr_city_code).isdigit():
raise FeatureProcessError(
'%s city_code config error in database table' % city_name)
seq = company_addr_city_code
return seq
def f_assert_must_percent(value_list):
"""
检测是否是百分数
"""
for value in value_list:
if not (str(value)[-1] == '%' and
(str(value[:-1]).count('.') <= 1
and str(value[:-1]).replace('.', '').isdigit())):
raise FeatureProcessError('%s f_assert_must_percent Error' %
value_list)
return value_list
def m_sex_to_code(seq):
"""
返回性别的code值
:param seq: 包含性别的字符串
:return: code
example：
:seq '男生'
:return '男'
"""
if '男' in seq:
seq = '男'
elif '女' in seq:
seq = '女'
else:
raise FeatureProcessError("'don't know sex")
return seq
def m_check_code(seq, args=None):
"""
将数值转化成对应的code
:param seq: 上一步得到的数据
:param args: [feature_name,操作符]
:return: 对应code
example：
:data: 20
:args ['education_degree_code','gte_lt']
:return 2
"""
if not seq:
return []
feature_name = args[0]
op = args[1]
fcm = FeatureCodeMapping.objects.filter(feature_name=feature_name, )
res = ''
num_map = {
int(conf.mapped_value): [conf.unitary_value, conf.dual_value]
for conf in fcm
}
for key, value in num_map.iteritems():
if op == 'gte_lt':
if float(value[0]) <= float(seq) < float(value[1]):
res = key
break
elif op == 'gt_lte':
if float(value[0]) < float(seq) <= float(value[1]):
res = key
break
elif op == 'in':
if seq in value[0]:
res = key
break
elif op == 'eq':
if seq == value[0]:
res = key
break
if res in ('', None):
raise FeatureProcessError("don't find %s=%s code value" %
(feature_name, seq))
return res
def m_max_flight_area(seq):
"""
一年内飞机出行中最多出行区域
:param seq: 飞行次数、国内次数、国外次数
:return: 飞行次数==0 返回0 国内次数>国外次数 返回inland 国外次数>国内次数 返回international
example：
:seq [2, 2, 3]
:return international
"""
""""""
flight_times = seq[0]
inland_count = seq[1]
international_count = seq[2]
if int(flight_times) == 0:
seq = 0
elif int(inland_count) >= int(international_count):
seq = 'inland'
elif int(inland_count) < int(international_count):
seq = 'international'
else:
raise FeatureProcessError("don't know max_flight_area code")
return seq
def f_assert_must_between(value_list, args):
"""
检测列表中的元素是否为数字或浮点数且在args的范围内
:param value_list: 待检测列表
:param args: 范围列表
:return: 异常或原值
example：
:value_list [2, 2, 3]
:args [1,3]
:value_list ['-2', '-3', 3]
:args ['-5',3]
"""
assert len(args) == 2
for value in value_list:
if not (str(value).count('.') <= 1
and str(value).replace('.', '').lstrip('-').isdigit()
and float(args[0]) <= float(value) <= float(args[1])):
raise FeatureProcessError('%s f_assert_must_between %s Error' %
(value_list, args))
return value_list
def f_assert_seq0_gte_seq1(value_list):
"""检测列表中的第一个元素是否大于等于第二个元素"""
if not value_list[0] >= value_list[1]:
raise FeatureProcessError('%s f_assert_seq0_gte_seq1 Error' %
value_list)
return value_list
def f_assert_jsonpath_true(seq):
"""假设jsonpath查询到的为true seq为[]空列表时代表没查到字段"""
if seq in ([], ):
raise FeatureProcessError("jsonpath not find field")
return seq
def m_to_code(seq, args=None):
"""
将数值转化成对应的code
:param seq: 上一步得到的数据
:param args: feature_name
:return: 对应code
example：
:data: 20
:args 'education_degree_code'
:return 2
"""
if not seq and seq != 0:
return []
if isinstance(seq, list):
seq = seq[0]
feature_name = args
fcm = FeatureCodeMapping.objects.filter(feature_name=feature_name,
is_delete=False)
res = ''
num_map = {
int(conf.mapped_value): [
conf.unitary_value, conf.dual_value, conf.arithmetic_type,
conf.feature_desc
]
for conf in fcm
}
for key, value in num_map.iteritems():
arithmetic_type = value[2]
if arithmetic_type == '[)':
if float(value[0]) <= float(seq) < float(value[1]):
res = key
break
elif arithmetic_type == '(]':
if float(value[0]) < float(seq) <= float(value[1]):
res = key
break
elif arithmetic_type == '()':
if float(value[0]) < float(seq) < float(value[1]):
res = key
break
elif arithmetic_type == '>=':
if float(seq) >= float(value[0]):
res = key
break
elif arithmetic_type == '<=':
if float(seq) <= float(value[0]):
res = key
break
elif arithmetic_type == 'in':
if seq in value[0]:
res = key
break
elif arithmetic_type == '==':
if str(seq) == value[0]:
res = value[3]
break
if res in ('', None):
raise FeatureProcessError("don't find %s=%s code value" %
(feature_name, seq))
return res