def entry_into_force_time(url):
'''
获取生效时间
:param url:
:return: str 2016-01-10
'''
first_res = requests.get(url=baes_url + url, headers=headers)
first_res.encoding = 'utf-8'
soup = BeautifulSoup(first_res.text, 'lxml')
date_y = get_date_object(
str(soup.find('div',
class_='article_opt').find('i').text).strip()).year
p_list = soup.find('div', class_='allZoom').find_all('p')
take_effect_text = re.findall(r'(\w+月\w+日正式生效)', p_list[0].text)
if take_effect_text:
date_m_d = re.findall(r'([0-9]+)', take_effect_text[0])
if len(date_m_d) == 2:
com_date = '%s%s%s' % (date_y, date_m_d[0], date_m_d[1])
else:
com_date = '%s%s%s' % (date_m_d[0], date_m_d[1], date_m_d[2])
else:
take_effect_text = re.findall(r'(\w+月\w+日正式生效)', p_list[1].text)
if take_effect_text:
date_m_d = re.findall(r'([0-9]+)', take_effect_text[0])
if len(date_m_d) == 2:
com_date = '%s%s%s' % (date_y, date_m_d[0], date_m_d[1])
else:
com_date = '%s%s%s' % (date_m_d[0], date_m_d[1], date_m_d[2])
else:
take_effect_text = re.findall(r'(\w+月\w+日正式生效)', p_list[-1].text)
if not take_effect_text:
print('请查看具体页面:%s' % url)
date_m_d = re.findall(r'([0-9]+)', take_effect_text[0])
if len(date_m_d) == 2:
com_date = '%s%s%s' % (date_y, date_m_d[0], date_m_d[1])
else:
com_date = '%s%s%s' % (date_m_d[0], date_m_d[1], date_m_d[2])
com_date = get_date_object(com_date, "%Y%m%d").date()
return com_date
'first': [],
'second': [],
'third': [],
}
constituent_shares_list = LmInnerReportDBMgr.get_data_list('lm_index_data', {'code': '000016'},
'constituent_shares_list')
__list = sorted(constituent_shares_list, key=lambda item: item['date'], reverse=True)
max_date = __list[0]['date']
min_date = __list[-1]['date']
# 1、获取此指数从头到尾的所有日期
date_list = getEveryDay(min_date, max_date, 1)
date_list = sorted(date_list, reverse=True)
# 2、将日期按照规则分配到date_dict中
for date in date_list:
date_obj = get_date_object(date)
# [每一年的10-31到09-01]
if 9 <= date_obj.month <= 10:
date_dict['first'].append(date)
# [每一年08-31到05-01]
elif 5 <= date_obj.month <= 8:
date_dict['second'].append(date)
# [最新日期到此年限或上一年的11-01(要判断最新日期月份是不是12月，如果是12月那么就是当前年份的11-01，如果是1月或2月或3月或4月，就用上一年份的11-01)]
# [最新日期此年的4月30到一下年的11-01日]
else:
date_dict['third'].append(date)
print(date_dict['third'])
print(__list[0]['date'])
print(__list[0]['cfg'])
if __list[0]['date'] in date_dict['first']:
def lrb(response, stock_code):
'''
:param response: 利润表的数据 str
:param stock_code: 股票代码
:return:
'''
zd_list = [
"YEAREND_DATE", "GROSS_REVENUE", "OPERATIONG_RECEIPT",
"INTEREST_REVENUE", "EARNED_PREMIUM", "POUNDAGE_AND_COMMISSION_INCOME",
"SALES_INCOME_OF_REAL_ESTATE", "OTHER_BUSINESS_INCOME",
"TOTAL_OPERATING_COST", "COST_IN_BUSINESS", "INTEREST_EXPENSE",
"POUNDAGE_AND_COMMISSION_EXPENSES", "SALE_COST_OF_REAL_ESTATE",
"RESEARCH_AND_DEV_ELOPMENT_COST", "SURRENDER_VALUE",
"PAYMENT_NET_EXPENDITURE", "EXTRACT_INSURANCE_CONTRACT_RESERVE_NET",
"EXPENDITURE_DIVIDEND_POLICY", "REINSURANCE_EXPENSES",
"OTHER_BUSINESS_COSTS", "BUSINESS_TAXES_AND_SURCHARGES",
"SELLING_EXPENSES", "MANAGEMENT_EXPENSES", "FINANCIAL_EXPENSES",
"ASSETS_IMPAIRMENT_LOSS", "FAIR_VALUE_CHANGE_INCOME",
"INVESTMENT_PROFIT",
"INVESTMENT_INCOME_FOR_VENTURES_AND_JOINT_VENTURES",
"EXCHANGE_EARNINGS", "FUTURES_PROFIT_AND_LOSS", "MANAGED_INCOME",
"SUBSIDY_INCOME", "OTHER_BUSINESS_PROFITS", "OPERATING_PROFIT",
"NON_BUSINESS_INCOME", "NON_BUSINESS_ESPENDITURE",
"DISPOSAL_LOSS_OF_NON_CURRENT_ASSETS", "TOTAL_PROFIT",
"INCOME_TAX_EXPENSE", "UNCINFIRMED_INVESTMENT_LOSS", "NET_MARGIN",
"VEST_IN_PARENT_COMPANY_PROPRIETOR_NET_MARGIN",
"NET_PROFIT_OF_THE_MERGED_PARTY_BEFORE_MERGER",
"MINORITY_SHAREHOLDER_PROFIT_AND_LOSS", "BASIC_PER_SHARE_PROFIT",
"BILUTED_EARNINGS_PER_SHARE"
]
lrb_data = [{j[0]: j[1]
for j in i} for i in (zip(zd_list, it) for it in zip(*[
item.strip().split(",")[1:-1]
for item in response.text.strip().split("\n")
]))]
for i in range(len(lrb_data)):
'''
处理单季度归属母公司净利润数据
'''
try:
first_data = lrb_data[i]
second_data = lrb_data[i + 1]
first_date_obj = get_date_object(first_data['YEAREND_DATE'])
second_date_obj = get_date_object(second_data['YEAREND_DATE'])
if first_date_obj.month == 3 or first_date_obj.year != second_date_obj.year:
# 3月的库里数据就是单季度的，则直接设置新字段存入mongo
# print('三月的数据,可直接存入库中...first: {},second: {}'.format(first_data['YEAREND_DATE'],
# second_data['YEAREND_DATE']))
SIGNLE_VEST_IN_PARENT_COMPANY_PROPRIETOR_NET_MARGIN = float(
first_data['VEST_IN_PARENT_COMPANY_PROPRIETOR_NET_MARGIN'])
elif first_date_obj.year == second_date_obj.year and (
first_date_obj.month - second_date_obj.month) == 3:
# 6月- 3月 9月 - 6月，或12月 - 9月，直接将数据做减法
# print('month相差3个月，则做减法输出 first: {},second: {}'.format(first_data['YEAREND_DATE'],
# second_data['YEAREND_DATE']))
try:
SIGNLE_VEST_IN_PARENT_COMPANY_PROPRIETOR_NET_MARGIN = float(
first_data[
'VEST_IN_PARENT_COMPANY_PROPRIETOR_NET_MARGIN']
) - float(second_data[
'VEST_IN_PARENT_COMPANY_PROPRIETOR_NET_MARGIN'])
except ValueError:
print('净利润数据为:--,无法计算,则忽略')
SIGNLE_VEST_IN_PARENT_COMPANY_PROPRIETOR_NET_MARGIN = ''
else:
print('未上市的财报不需要处理')
# print('有问题的数据 first:{},second:{}'.format(first_data['YEAREND_DATE'], second_data['YEAREND_DATE']))
SIGNLE_VEST_IN_PARENT_COMPANY_PROPRIETOR_NET_MARGIN = ''
except IndexError:
first_data = lrb_data[i]
# print('最后一条数据，不需要进行任何操作first: {}'.format(first_data['YEAREND_DATE']))
SIGNLE_VEST_IN_PARENT_COMPANY_PROPRIETOR_NET_MARGIN = float(
first_data['VEST_IN_PARENT_COMPANY_PROPRIETOR_NET_MARGIN'])
lrb_data[i][
'SIGNLE_VEST_IN_PARENT_COMPANY_PROPRIETOR_NET_MARGIN'] = SIGNLE_VEST_IN_PARENT_COMPANY_PROPRIETOR_NET_MARGIN
db_report_date = LmInnerReportDBMgr.get_code('lm_stock_data', stock_code,
'financial_report_date')
if db_report_date:
if date_turn_timestamp(db_report_date, lrb_data[0]['YEAREND_DATE']):
LmInnerReportDBMgr.save_set_result('lm_stock_data',
{'code': stock_code},
'report.lrb', lrb_data)
else:
LmInnerReportDBMgr.save_insert_result(
'lm_stock_data', {
'code': stock_code,
'financial_report_date': lrb_data[0]['YEAREND_DATE'],
'report': {
'lrb': lrb_data
}
})