def data_shuffle(data):
data["FILE_URL_"] = data["URL_"]
data["FILE_NAME_"] = data["TITLE_"][:-2]
try:
pdf_content = parse(pdf_url=data["URL_"])
except Exception:
publish_date = re.findall(r"cmbir/(\d{8})/", data["URL_"])
if publish_date:
data["NOTICE_TIME_"] = "-".join([
publish_date[0][:4], publish_date[0][4:6], publish_date[0][6:]
])
else:
publish_date = re.findall(
r"[〇一二三四五六七八九十]{4} ?年[〇一二三四五六七八九十]{1,2} ?月[〇一二三四五六七八九十]{1,3} ?日",
pdf_content)
if publish_date:
s_publish_date = list(set(publish_date))
each_date = s_publish_date[-1].replace(" ", "")
d_year = each_date[:4]
d_month = each_date[each_date.find("年") + 1:each_date.find("月")]
d_day = each_date[each_date.find("月") + 1:each_date.find("日")]
f_year = "".join([YEAR_DICT[i] for i in d_year])
data["NOTICE_TIME_"] = "-".join(
[f_year, WORD_DICT[d_month], WORD_DICT[d_day]])
else:
publish_date = re.findall(r"cmbir/(\d{8})/", data["URL_"])
if publish_date:
data["NOTICE_TIME_"] = "-".join([
publish_date[0][:4], publish_date[0][4:6],
publish_date[0][6:]
])
return data
def data_shuffle(data):
result = parse(data["PDF_"])
pro_name = re.findall(r"产品名称：([\w ]+)", result)
if pro_name:
data["PRO_NAME_"] = pro_name[0]
regist_code = re.findall(r"C\d{13}", result)
if regist_code:
data["REGIST_CODE_"] = regist_code[0]
# 类型
# pro_type = re.findall(r"产品类型：([\w ]+)", result)
# if pro_type:
# data[""]
real_days = re.findall(r"期限:(\d+) ?天?", result)
if real_days:
data["REAL_DAYS_"] = real_days[0]
yield_ = re.findall(r"([^:：]+)%", result)
if yield_:
if "-" in yield_[0]:
data["YIELD_HIGH_"] = yield_[0].split("-")[1]
data["YIELD_LOW_"] = yield_[0].split("-")[0]
return data
def data_shuffle(data):
year = re.findall(r"2\d{3}", data["TITLE_"])
if year:
data["YEAR_"] = year[0] # 年份
else:
try:
pdf_content = parse(pdf_url=data["PDF_URL_"])
except Exception:
pass
else:
year = re.findall(r"(2\d{3}) ?年?年度", pdf_content)
if year:
s_year = list(set(year))
if len(s_year) == 1:
data["YEAR_"] = s_year[0]
else:
s_year.sort()
data["YEAR_"] = s_year[-1]
else:
year = re.findall(r"(2 ?\d ?\d ?\d) 年", pdf_content)
if year:
year.sort()
data["YEAR_"] = year[-1].replace(" ", "")
else:
year = re.findall(r"(2 ?\d ?\d ?\d)", pdf_content)
if year:
year.sort()
data["YEAR_"] = year[-1]
if not data.get("YEAR_"):
year = re.findall(r"/(2\d{3})/", data["PDF_URL_"])
if year:
data["YEAR_"] = year[0]
data["FILE_URL_"] = data["PDF_URL_"]
data["FILE_NAME_"] = data["PDF_NAME_"]
data["TITLE_"] = data["TITLE_"].replace('•', '')
return data
def data_shuffle(data):
result = parse(pdf_url=data["PDF_"])
# 登记编码
regist_code = re.findall(r"C\d{13}", result)
if regist_code:
data["REGIST_CODE_"] = regist_code[0]
# 风险等级
risk_level = re.findall(r"本产品为 ?([\w ]+（\w+）)理财产品", result)
if risk_level:
data["SOURCE_RISK_LEVEL_"] = risk_level[0]
if "R1" in risk_level[0]:
data["RISK_LEVEL_CODE_"] = "R1"
elif "R2" in risk_level[0]:
data["RISK_LEVEL_CODE_"] = "R2"
elif "R3" in risk_level[0]:
data["RISK_LEVEL_CODE_"] = "R3"
elif "R4" in risk_level[0]:
data["RISK_LEVEL_CODE_"] = "R4"
elif "R5" in risk_level[0]:
data["RISK_LEVEL_CODE_"] = "R5"
else:
pass
# print(result)
return data
def data_shuffle(data):
result = parse(data["PDF_"])
# 产品名称
pro_name = re.findall(r"(.*理财产品)说明书", result)
if pro_name:
data["PRO_NAME_"] = pro_name[0]
# 发行机构
data["PRO_ORG_"] = "浙商银行股份有限公司"
# 登记编码
regist_code = re.findall(r"C\d{13}", result)
if regist_code:
data["REGIST_CODE_"] = regist_code[0]
# 运作模式
# opt_mode = re.findall(r"非保本浮动收益型 ")
# data["OPT_MODE_"] =
# # 收益类型
# data["YIELD_TYPE_"] =
# 募集币种
currency_type = re.findall(r"([人美][民元]币?) \n", result)
if currency_type:
data["CURRENCY_TYPE_"] = currency_type[0]
# 起购金额 递增单位
start_funds = re.findall(r"[^\n]+[递增][增加]", result)
if start_funds:
start = re.findall(r"\d+ [千百十]?万元", start_funds[0])
increase = re.findall(r"(\d+ [千万百]?元)的整数倍", start_funds[0])
if start:
data["START_FUNDS_"] = start[0].replace(" ", "")
if increase:
data["INCREASE_"] = increase[0].replace(" ", "")
# # 原始风险等级
source_risk = re.findall(r"[低中高]{1,2}风险", result)
if source_risk:
# print(source_risk)
data["SOURCE_RISK_LEVEL_"] = source_risk[0]
# # 风险等级
if data["SOURCE_RISK_LEVEL_"]:
if data["SOURCE_RISK_LEVEL_"] == "低风险":
data["RISK_LEVEL_CODE_"] = "R1"
elif data["SOURCE_RISK_LEVEL_"] == "中低风险":
data["RISK_LEVEL_CODE_"] = "R2"
elif data["SOURCE_RISK_LEVEL_"] == "较低风险":
data["RISK_LEVEL_CODE_"] = "R2"
elif data["SOURCE_RISK_LEVEL_"] == "中等风险":
data["RISK_LEVEL_CODE_"] = "R3"
elif data["SOURCE_RISK_LEVEL_"] == "中高风险":
data["RISK_LEVEL_CODE_"] = "R4"
elif data["SOURCE_RISK_LEVEL_"] == "高风险":
data["RISK_LEVEL_CODE_"] = "R5"
raise_date = re.findall(r"\d{4} ?年 \d{1,2} 月 \d{1,2} 日.*?[-—].*?日", result)
if raise_date:
r_date = re.findall(r"\d{4} 年 \d{1,2} 月 \d{1,2} 日", raise_date[0])
# 募集起始日期
r_start = re.sub(r"[日 ]", "", r_date[0])
data["RAISE_START_"] = re.sub(r"[年月]", "-", r_start)
# 募集结束日期
r_end = re.sub(r"[日 ]", "", r_date[0])
data["RAISE_END_"] = re.sub(r"[年月]", "-", r_end)
all_date = re.findall(r"\d{4} 年 \d{1,2} 月 \d{1,2} 日", result)
re_date = list()
for each in all_date:
each_rp = re.findall(r" \d ", each)
for i in each_rp:
each = each.replace(i, "0" + i.strip())
each = re.sub(r"[年月日 ]", "", each)
if each not in re_date:
re_date.append(each)
re_date.sort()
if len(re_date) >= 3:
# 产品起始日期
data["PRO_START_"] = re_date[2]
# 产品结束日期
data["PRO_END_"] = re_date[-1]
else:
pass
yield_ = re.findall(r"[1-9]\.\d{1,2}%", result)
yield_.sort()
if yield_:
# 预期最低收益率
data["YIELD_LOW_"] = yield_[0]
# 预期最高收益率
data["YIELD_HIGH_"] = yield_[-1]
# 实际天数 本理财计划存续期为 7 年
real_days = re.findall(r"(\d+ ?[年月日天]) ?\n", result)
re_days = list()
for days in real_days:
if "年" in days:
days = re.sub(r"[年 ]", "", days)
days = int(days) * 365
re_days.append(days)
elif "月" in days:
days = re.sub(r"[年 ]", "", days)
days = int(days) * 365
re_days.append(days)
else:
days = re.sub(r"[日天 ]", "", days)
days = int(days)
re_days.append(days)
re_days.sort()
if re_days:
data["REAL_DAYS_"] = str(re_days[-1])
return data
def data_shuffle(data):
# 产品名称
data["PRO_NAME_"] = data["PDF_NAME_"]
# 产品编码
# pro_code = re.findall(r"(.*?)中银", data["PDF_NAME_"])
pro_code = re.findall(r"([\da-zA-Z\-]{6,})", data["PDF_NAME_"])
if pro_code:
data["PRO_CODE_"] = pro_code[0]
# 产品期限
real_days = re.findall(r"(\d+)天", data["PDF_NAME_"])
if real_days:
data["REAL_DAYS_"] = real_days[0]
# 募集起始日期
raise_start = re.findall(r"\d{4}年\d{1,2}月\d{1,2}", data["PDF_NAME_"])
if raise_start:
start_date = re.sub(r"[\u4e00-\u9fa5]", "-", raise_start[0])
data["RAISE_START_"] = start_date
pdf_result = parse(pdf_url=data["PDF_"])
# print(pdf_result)
# 清洗 PDF
# 风险等级
risk_level = re.findall(r"(\w+风险)产品", pdf_result)
if risk_level:
data["SOURCE_RISK_LEVEL_"] = risk_level[0]
if data["SOURCE_RISK_LEVEL_"] == "低风险":
data["RISK_LEVEL_CODE_"] = "R1"
elif data["SOURCE_RISK_LEVEL_"] == "中低风险":
data["RISK_LEVEL_CODE_"] = "R2"
elif data["SOURCE_RISK_LEVEL_"] == "较低风险":
data["RISK_LEVEL_CODE_"] = "R2"
elif data["SOURCE_RISK_LEVEL_"] == "中等风险":
data["RISK_LEVEL_CODE_"] = "R3"
elif data["SOURCE_RISK_LEVEL_"] == "中高风险":
data["RISK_LEVEL_CODE_"] = "R4"
elif data["SOURCE_RISK_LEVEL_"] == "高风险":
data["RISK_LEVEL_CODE_"] = "R5"
# 匹配表格中内容
# 一、理财计划基本信息 二、理财计划投资范围、投资种类及比例 [投理][资财][对计][象划]及?投资[限范][制围]
first_shuffle = re.findall(r"[一1]、 *理?财?[产计][品划]基本信息(.*)\n[二2]、",
pdf_result, re.S)
if first_shuffle:
# 保证收益型 非保本浮动收益型 无固定期限非保本浮动收益型，实际产品期限受制于提前终止条款。
# break_even = re.findall(r"产品类型 *\n([^\n]*)", first_shuffle[0])
# break_even = re.findall(r"\n(\w*收益型)[；，]?\w*。? *\n", first_shuffle[0])
# 售卖时间范围 理财计划认购期 【2018】年【11】月【21】日-【2018】年【11】月【22】日
# time_limit = re.findall(r"[认募][购集]期 *\d{4} ?年?\d{1,2} ?月?\d{1,2} ?日?[-至] ?\d{4} ?年?\d{1,2} ?月?\d{1,2} ?日?", first_shuffle[0])
time_limit = re.findall(
r"\d{4}】? ?年?【?\d{1,2}】? ?月?【?\d{1,2}】? ?日?[-至] ?【?\d{4}】? ?年?【?\d{1,2}】? ?月?【?\d{1,2}】? ?日?",
first_shuffle[0])
if not time_limit:
time_limit = re.findall(
r"\d{4}】? ?年?【?\d{1,2}】? ?月?【?\d{1,2}】? ?日?[-至] ?【?\d{4}】? ?年?【?\d{1,2}】? ?月?【?\d{1,2}】? ?日?",
first_shuffle[0])
if time_limit:
re_time = re.findall(r"\d{4} ?年 ?\d{1,2} ?月 ?\d{1,2} ?日",
time_limit[0])
if not re_time:
re_time = re.findall(
r"(【?\d{4}】? ?[年\\/-] ?【?\d{1,2}】? ?[月\\/-] ?【?\d{1,2}】? ?日?)",
time_limit[0])
# else:
# re_time = re.findall(r"(【?\d{4}】? ?[年\\/-] ?【?\d{1,2}】? ?[月\\/-] ?【?\d{1,2}】? ?日?)", first_shuffle[0])
if re_time:
if len(re_time) > 1:
# 结束售卖时间
raise_end = re.sub(r"[【】 ]", "", re_time[1])
data["RAISE_END_"] = re.sub(r"[年月]", "-", raise_end[:-1])
# 期限
inverst_period = re.findall(r"\n【?(\d*)】? ?天（?.*）? *\n",
first_shuffle[0])
if inverst_period:
data["REAL_DAYS_"] = inverst_period[0]
# 收益率
yield_rate = re.findall(r"(【?\d\.\d{2}】?%) *\n", first_shuffle[0])
if yield_rate:
if len(yield_rate) == 1:
data["YIELD_HIGH_"] = re.sub(r"[【】 %]", "", yield_rate[0])
data["YIELD_LOW_"] = data["YIELD_HIGH_"]
else:
data["YIELD_LOW_"] = re.sub(r"[【】 %]", "", yield_rate[0])
data["YIELD_HIGH_"] = re.sub(r"[【】 %]", "", yield_rate[1])
# 起购金额
funds = re.findall(r"起点([^\n]*)倍[递累][^\n]*", first_shuffle[0])
if funds:
re_funds = re.findall(r"\d+.*?元", funds[0])
if re_funds:
if len(re_funds) == 1:
funds = re.findall(r"认购起点金额.*?倍", first_shuffle[0], re.S)
if funds:
re_funds = re.findall(r"\d+.*?元", funds[0])
if re_funds:
if len(re_funds) == 2:
# data["START_FUNDS_"] = re_funds[0].replace(" ", "")
data["START_FUNDS_"] = re.sub(
r"[^个十百千万亿元\d.]", "", re_funds[0])
data["INCREASE_"] = re.sub(
r"[^个十百千万亿元\d.]", "", re_funds[0])
else:
pass
else:
data["START_FUNDS_"] = re_funds[0].replace(" ", "")
data["INCREASE_"] = re_funds[1].replace(" ", "")
else:
pass
# print(data["PDF_"])
# print(pdf_result)
return data
def __shuffle(self, data):
re_data = dict()
re_data["ENTITY_CODE_"] = data["ENTITY_CODE_"]
re_data["ENTITY_NAME_"] = data["ENTITY_NAME_"]
re_data["URL_"] = data["URL_"]
if "中国理财网" in data["ENTITY_NAME_"]:
serial_number = req_for_serial_number(code="JRCP_LCCP_INFO")
re_data["ID_"] = serial_number
re_data["PRO_NAME_"] = data["PRO_NAME_"]
re_data["PRO_ORG_"] = data["PRO_ORG_"]
re_data["REGIST_CODE_"] = data["REGIST_CODE_"]
re_data["PRO_STATUS_"] = data["PRO_STATUS_"]
re_data["OPT_MODE_"] = data["OPT_MODE_"]
re_data["YIELD_TYPE_"] = data["YIELD_TYPE_"]
# re_data["YIELD_TYPE_CODE_"] = data[""]
re_data["CURRENCY_TYPE_"] = data["CURRENCY_TYPE_"]
# re_data["CURRENCY_TYPE_CODE_"] = data[""]
re_data["START_FUNDS_"] = data["START_FUNDS_"]
try:
if float(data["START_FUNDS_"]) <= 10000:
re_data["START_FUNDS_CODE_"] = "S0_1"
elif 10000 < float(data["START_FUNDS_"]) <= 50000:
re_data["START_FUNDS_CODE_"] = "S1_5"
elif 50000 < float(data["START_FUNDS_"]) < 100000:
re_data["START_FUNDS_CODE_"] = "S5_10"
elif 100000 < float(data["START_FUNDS_"]):
re_data["START_FUNDS_CODE_"] = "S10_"
except Exception:
re_data["START_FUNDS_"] = 0
org = {
'01': '国有银行',
'02': '股份制银行',
'03': '城商行',
'04': '外资银行',
'05': '农村合作金融机构',
'06': '其他',
'07': '其他',
'08': '其他',
'09': '其他',
'00': '其他',
'10': '理财子公司'
}
re_data["SOURCE_RISK_LEVEL_"] = data["SOURCE_RISK_LEVEL_"]
re_data['ORG_TYPE_'] = org.get(data.get('ORG_TYPE_'))
re_data["RAISE_START_"] = data["RAISE_START_"]
re_data["RAISE_END_"] = data["RAISE_END_"]
re_data["PRO_START_"] = data["PRO_START_"]
re_data["PRO_END_"] = data["PRO_END_"]
re_data["YIELD_LOW_"] = data["YIELD_LOW_"]
re_data["YIELD_HIGH_"] = data["YIELD_HIGH_"]
re_data["REAL_DAYS_"] = data["REAL_DAYS_"]
re_data["INVEST_TYPE_"] = data["INVEST_TYPE_"]
re_data["DATE_TYPE_"] = data["DATE_TYPE_"]
re_data["YIELD_"] = data["YIELD_"]
re_data["RAISE_TYPE_"] = data["RAISE_TYPE_"]
re_data["INVEST_PROPERTIES_"] = data["INVEST_PROPERTIES_"]
re_data["BUS_START_"] = data["BUS_START_"]
re_data["BUS_END_"] = data["BUS_END_"]
re_data["START_VALUE_"] = data["START_VALUE_"]
re_data["PRO_VALUE_"] = data["PRO_VALUE_"]
re_data["TOTAL_VALUE_"] = data["TOTAL_VALUE_"]
re_data["RECENT_YIELD_"] = data["RECENT_YIELD_"]
re_data["PRO_TYPE_"] = data["PRO_TYPE_"]
re_data["SALE_AREA_"] = data["SALE_AREA_"]
if "PROVINCE_NAME_" in data:
re_data["PROVINCE_NAME_"] = data["PROVINCE_NAME_"]
if "PROVINCE_NAME_" in data:
re_data["PROVINCE_CODE_"] = data["PROVINCE_CODE_"]
if "CITY_NAME_" in data:
re_data["CITY_NAME_"] = data["CITY_NAME_"]
if "CITY_CODE_" in data:
re_data["CITY_CODE_"] = data["CITY_CODE_"]
# re_data["REDEEM_"] = data[""]
# re_data["INCREASE_"] = data[""]
# re_data["INVEST_RANGE_"] = data[""]
bank_list = list()
bank_code_list = list()
for each in self.bank_list:
if each["NAME_"] in data.get("ENTITY_NAME_", ""):
bank_list.append(each["NAME_"])
bank_code_list.append(each["CODE_"])
if bank_list:
re_data["BANK_NAME_"] = "|".join(bank_list)
if bank_code_list:
re_data["BANK_CODE_"] = "|".join(bank_code_list)
# del re_data["CREATE_TIME_"]
# del re_data["SPIDER_TIME_"]
# del re_data["M_STATUS_"]
# del re_data["DELETE_STATUS_"]
# del re_data["DATA_STATUS_"]
# del re_data["PUBLISH_STATUS_"]
re_data = super(BranchFinProduct,
self).generic_shuffle(data=data,
re_data=re_data,
field=None)
if not data["YIELD_LOW_"]:
re_data['YIELD_LOW_'] = '--'
if not data["YIELD_HIGH_"]:
re_data['YIELD_HIGH_'] = '--'
if not data["START_FUNDS_"]:
re_data['START_FUNDS_'] = '--'
return {"TABLE_NAME_": TABLE_NAME("CRMLCCP"), "DATA_": re_data}
else:
source = re.findall(r"(https?://.*?)/", data["URL_"])
re_data["SOURCE_"] = source[0]
re_data["SOURCE_NAME_"] = data["ENTITY_NAME_"]
serial_number = req_for_serial_number(code="JRCP_LCCP")
re_data["ID_"] = serial_number
re_data["SOURCE_TYPE_"] = ""
# if "PRO_NAME_" not in data:
# return
re_data["PRO_NAME_"] = data["PRO_NAME_"]
f_index = data["ENTITY_NAME_"].find("-")
re_data["PRO_ORG_"] = data["ENTITY_NAME_"][:f_index]
if "PRO_CODE_" in data:
re_data["PRO_CODE_"] = data["PRO_CODE_"]
# 登记编码
if "REGIST_CODE_" in data:
re_data["REGIST_CODE_"] = data["REGIST_CODE_"]
else:
if "PDF_" in data:
try:
text = parse(data["PDF_"])
registration_code = re.findall(r"C\d{13}", text)
if registration_code:
re_data["REGIST_CODE_"] = registration_code[0]
except Exception as e:
self.logger.exception(
f"2.1--err: PDF."
f" 原始数据 collection = {self.m_client.mongo_collection};"
f" ENTITY_CODE_ = {data.get('ENTITY_CODE_', 'None')};"
f" 原始数据 _id = {data['_id']};"
f" error: {e}.")
# 预售(PRE)、在售(ON)、停售(STOP)
# 全部为 在售
re_data["PRO_STATUS_"] = "ON"
if "OPT_MODE_" in data:
re_data["OPT_MODE_"] = data["OPT_MODE_"]
if "YIELD_TYPE_" in data:
re_data["YIELD_TYPE_"] = data["YIELD_TYPE_"]
# re_data["YIELD_TYPE_CODE_"] = data[""]
if "CURRENCY_TYPE_" in data:
re_data["CURRENCY_TYPE_"] = data["CURRENCY_TYPE_"]
# re_data["CURRENCY_TYPE_CODE_"] = data[""]
# 起购金额
if "START_FUNDS_" in data:
start_funds = data["START_FUNDS_"].replace(" ", "")
start_funds = start_funds.replace("亿", "00000000")
start_funds = start_funds.replace("千万", "0000000")
start_funds = start_funds.replace("百万", "000000")
start_funds = start_funds.replace("十万", "00000")
start_funds = start_funds.replace("万", "0000")
start_funds = start_funds.replace("千", "000")
start_funds = start_funds.replace("百", "00")
start_funds = start_funds.replace("元", "")
re_data["START_FUNDS_"] = start_funds
try:
if float(re_data["START_FUNDS_"]) <= 10000:
re_data["START_FUNDS_CODE_"] = "S0_1"
elif 10000 < float(re_data["START_FUNDS_"]) <= 50000:
re_data["START_FUNDS_CODE_"] = "S1_5"
elif 50000 < float(re_data["START_FUNDS_"]) <= 100000:
re_data["START_FUNDS_CODE_"] = "S5_10"
elif 100000 < float(re_data["START_FUNDS_"]):
re_data["START_FUNDS_CODE_"] = "S10_"
except Exception as e:
re_data["START_FUNDS_"] = 0
if "RISK_LEVEL_CODE_" in data:
re_data["RISK_LEVEL_"] = self.risk_dict[
data["RISK_LEVEL_CODE_"]]
re_data["RISK_LEVEL_CODE_"] = data["RISK_LEVEL_CODE_"]
if "RISK_LEVEL_" in data:
re_data["SOURCE_RISK_LEVEL_"] = data["RISK_LEVEL_"]
elif "SOURCE_RISK_LEVEL_" in data:
re_data["SOURCE_RISK_LEVEL_"] = data["SOURCE_RISK_LEVEL_"]
# # 募集起始日期
if "RAISE_START_" in data:
re_data["RAISE_START_"] = data["RAISE_START_"]
# # 募集结束日期
if "RAISE_END_" in data:
re_data["RAISE_END_"] = data["RAISE_END_"]
# # 产品起始日期
if "PRO_START_" in data:
re_data["PRO_START_"] = data["PRO_START_"]
# # 产品结束日期
if "PRO_END_" in data:
re_data["PRO_END_"] = data["PRO_END_"]
# 预期最低收益率
if "YIELD_LOW_" in data:
re_data["YIELD_LOW_"] = data["YIELD_LOW_"].replace("%", "")
# 预期最高收益率
if "YIELD_HIGH_" in data:
re_data["YIELD_HIGH_"] = data["YIELD_HIGH_"].replace("%", "")
# 实际天数
if "REAL_DAYS_" in data:
data["REAL_DAYS_"] = data["REAL_DAYS_"].replace(" ", "")
if "年" in data["REAL_DAYS_"]:
re_data["REAL_DAYS_"] = data["REAL_DAYS_"].replace("年", "")
try:
re_data["REAL_DAYS_"] = int(
re_data["REAL_DAYS_"]) * 365
except Exception:
re_data["REAL_DAYS_"] = 0
elif "月" in data:
re_data["REAL_DAYS_"] = data["REAL_DAYS_"].replace("月", "")
try:
re_data["REAL_DAYS_"] = int(re_data["REAL_DAYS_"]) * 30
except Exception:
re_data["REAL_DAYS_"] = 0
else:
re_data["REAL_DAYS_"] = data["REAL_DAYS_"].replace("天", "")
else:
if "PRO_START_" in data and "PRO_END_" in data:
t_start = arrow.get(data["PRO_START_"], "YYY-MM-DD")
t_end = arrow.get(data["PRO_END_"], "YYYY-MM-DD")
real_days = t_end - t_start
data["REAL_DAYS_"] = real_days.days
if "INVEST_TYPE_" in data:
re_data["INVEST_TYPE_"] = data["INVEST_TYPE_"]
# # 投资者类型
if "PRO_TYPE_" in data:
re_data["PRO_TYPE_"] = data["PRO_TYPE_"]
if "SALE_AREA_" in data:
re_data["SALE_AREA_"] = data["SALE_AREA_"]
# # 可否赎回
if "REDEEM_" in data:
if "不" in data["REDEEM_"]:
re_data["REDEEM_"] = "N"
else:
re_data['REDEEM_'] = "Y"
if "INCREASE_" in data:
increase = data["INCREASE_"].replace(" ", "")
increase = increase.replace("亿", "00000000")
increase = increase.replace("千万", "0000000")
increase = increase.replace("百万", "000000")
increase = increase.replace("十万", "00000")
increase = increase.replace("万", "0000")
increase = increase.replace("千", "000")
increase = increase.replace("百", "00")
increase = increase.replace("元", "")
re_data["INCREASE_"] = increase
# re_data["INVEST_RANGE_"] = data["INVEST_RANGE_"]
re_data["RECOMMEND_"] = "N"
re_data["GOOD_SALE_"] = "N"
re_data["NEW_SALE_"] = "N"
re_data["SALE_SOURCE_"] = "NET"
bank_list = list()
bank_code_list = list()
for each in self.bank_list:
if each["NAME_"] in data.get("ENTITY_NAME_", ""):
bank_list.append(each["NAME_"])
bank_code_list.append(each["CODE_"])
if bank_list:
re_data["BANK_NAME_"] = "|".join(bank_list)
if bank_code_list:
re_data["BANK_CODE_"] = "|".join(bank_code_list)
if not data["YIELD_LOW_"]:
re_data['YIELD_LOW_'] = '--'
if not data["YIELD_HIGH_"]:
re_data['YIELD_HIGH_'] = '--'
if not data["START_FUNDS_"]:
re_data['START_FUNDS_'] = '--'
re_data = super(BranchFinProduct,
self).generic_shuffle(data=data,
re_data=re_data,
field=None)
re_data["PUBLISH_TIME_"] = re_data["SPIDER_TIME_"]
return {
"TABLE_NAME_": TABLE_NAME("CHA_BRANCH_FINANCIAL_PRODUCT"),
"DATA_": re_data
}