mns-scheduler 1.0.8.7__py3-none-any.whl → 1.4.3.2__py3-none-any.whl

mns_scheduler/company_info/{company_info_sync_api.py → base/sync_company_base_info_api.py}

@@ -9,9 +9,8 @@ sys.path.append(project_path)
 from datetime import datetime
 import pandas as pd
 from loguru import logger
-import mns_common.api.ths.ths_stock_api as ths_stock_api
-import mns_common.api.em.east_money_stock_v2_api as east_money_stock_v2_api
-import mns_scheduler.company_info.company_constant_data as company_constant_data_api
+import mns_common.api.ths.concept.web.ths_company_info_web as ths_company_info_web
+import mns_common.component.em.em_stock_info_api as em_stock_info_api
 import mns_common.component.common_service_fun_api as common_service_fun_api
 import mns_common.component.concept.ths_concept_common_service_api as ths_concept_common_service_api
 from mns_common.db.MongodbUtil import MongodbUtil
@@ -21,10 +20,12 @@ import mns_common.api.kpl.constant.kpl_constant as kpl_constant
 import mns_common.component.company.company_common_service_api as company_common_service_api
 import mns_common.component.k_line.common.k_line_common_service_api as k_line_common_service_api
 import mns_common.constant.db_name_constant as db_name_constant
+from functools import lru_cache
+import mns_scheduler.company_info.base.sync_company_hold_info_api as sync_company_hold_info_api
 
 mongodb_util = MongodbUtil('27017')
 # 分页大小
-MAX_PAGE_NUMBER = 6000
+MAX_PAGE_NUMBER = 2000
 import threading
 
 # 定义一个全局锁，用于保护 result 变量的访问
@@ -33,200 +34,15 @@ result_lock = threading.Lock()
 result = []
 
 
-# 计算实际流通比例
-def calculate_circu_ratio(symbol):
-    query = {"symbol": symbol}
-    stock_gdfx_free_top_10 = mongodb_util.descend_query(query, 'stock_gdfx_free_top_10', "period", 10)
-    if stock_gdfx_free_top_10.shape[0] == 0:
-        mv_circulation_ratio = 1
-    else:
-        # 排除香港结算公司 大于5%减持不用发公告  香港中央结算    HKSCC
-        stock_gdfx_free_top_10['is_hk'] = stock_gdfx_free_top_10['shareholder_name'].apply(
-            lambda shareholder_name: "HK" if shareholder_name.startswith('香港中央结算') or shareholder_name.startswith(
-                'HKSCC') else "A")
-
-        # 持股大于5%      # 排除香港结算公司 大于5%减持不用发公告  香港中央结算    HKSCC
-        stock_gdfx_free_top_10 = stock_gdfx_free_top_10.loc[
-            (stock_gdfx_free_top_10['circulation_ratio'] >= 5) & (stock_gdfx_free_top_10['is_hk'] == 'A')]
-
-        circulation_ratio = sum(stock_gdfx_free_top_10['circulation_ratio'])
-        mv_circulation_ratio = round((100 - circulation_ratio) / 100, 2)
-        # 防止错误数据
-        if mv_circulation_ratio < 0:
-            mv_circulation_ratio = 1
-    return mv_circulation_ratio
-
-
-def filed_sort(company_info):
-    return company_info[[
-        "_id",
-        "name",
-        "industry",
-        "first_sw_industry",
-        "first_industry_code",
-        "second_sw_industry",
-        "second_industry_code",
-        "third_sw_industry",
-        "third_industry_code",
-        "ths_concept_name",
-        "ths_concept_code",
-        "ths_concept_sync_day",
-        "em_industry",
-        "em_concept",
-        "business_nature",
-        "actual_controller_name",
-        "actual_controller_rate",
-        "final_controller_name",
-        "final_controller_rate",
-        "mv_circulation_ratio",
-        "flow_mv_sp",
-        "total_mv_sp",
-        "now_price",
-        "total_share",
-        "flow_share",
-        "total_mv",
-        "flow_mv",
-        "flow_mv_level",
-        "holder_controller_name",
-        "holder_controller_rate",
-        "area",
-        "list_date",
-        "deal_days",
-        "pe_ttm",
-        "pb",
-        "ROE",
-        "classification",
-        "base_business",
-        "address",
-        "market_id",
-        "symbol",
-        "amount",
-        "sync_date",
-        "ths_concept_list_info",
-        "kpl_plate_name",
-        "kpl_most_relative_name",
-        "kpl_plate_list_info",
-        'operate_profit',
-        'total_operate_income'
-    ]]
-
-
-def get_east_money_stock_info():
-    all_real_time_quotes = east_money_stock_v2_api.get_all_real_time_quotes()
-    all_real_time_quotes = all_real_time_quotes[['symbol',
-                                                 'name',
-                                                 "now_price",
-                                                 'total_mv',
-                                                 'flow_mv',
-                                                 'pe_ttm',
-                                                 'sz_sh',
-                                                 'area',
-                                                 'pb',
-                                                 'list_date',
-                                                 'ROE',
-                                                 'total_share',
-                                                 'flow_share',
-                                                 'industry',
-                                                 'amount',
-                                                 'concept']]
-
-    return all_real_time_quotes
-
-
-def create_index():
-    mongodb_util.create_index('company_info',
-                              [("classification", 1)])
-    mongodb_util.create_index('company_info',
-                              [("industry", 1)])
-    mongodb_util.create_index('company_info',
-                              [("flow_mv", 1)])
-    mongodb_util.create_index('company_info',
-                              [("list_date", 1)])
-    mongodb_util.create_index('company_info',
-                              [("symbol", 1)])
-
-
-# 修改行业信息
-def fix_company_industry(symbol):
-    if symbol is not None:
-        query = {"symbol": symbol}
-        company_info = mongodb_util.find_query_data('company_info_base', query)
-    else:
-        company_info = mongodb_util.find_all_data('company_info_base')
-
-    company_info = company_info.set_index(['second_sw_industry'], drop=False)
-    del company_info['industry']
-    # fix industry  name
-    industry_final_fix_df = company_constant_data_api.get_industry_final_fix_df()
-    industry_final_fix_df = industry_final_fix_df.set_index(['second_sw_industry'], drop=True)
-    company_info = pd.merge(company_info, industry_final_fix_df, how='outer',
-                            left_index=True, right_index=True)
-    # 将申万第三行业做为行业
-    company_info = company_constant_data_api.fix_industry_use_sw_third(company_info.copy())
-    company_info['industry'] = company_info['industry'].fillna('综合')
-    company_info = filed_sort(company_info)
-    company_info['company_type'] = company_info['business_nature']
-
-    # 将list_date列中的所有NaN值设置为99990909
-    company_info['list_date'].fillna(20990909.0, inplace=True)
-
-    # 将日期数值转换为日期时间格式
-    company_info['list_date_01'] = pd.to_datetime(company_info['list_date'], format='%Y%m%d')
-
-    company_info['list_date'] = company_info['list_date'].apply(
-        lambda x: pd.to_numeric(x, errors="coerce"))
-
-    now_date = datetime.now()
-
-    # 计算日期差值 距离现在上市时间
-    company_info['diff_days'] = company_info.apply(
-        lambda row: (now_date - row['list_date_01']).days, axis=1)
-
-    str_now_date = now_date.strftime('%Y-%m-%d %H:%M:%S')
-    company_info['sync_date'] = str_now_date
-
-    try:
-        # 次新股
-        sub_stock = ths_stock_api.ths_stock_concept('885598')
-        sub_stock_symbol_list = list(sub_stock['symbol'])
-    except BaseException as e:
-        logger.error("出现异常:{},{}", symbol, e)
-        query = {'concept_code': 885598}
-        ths_stock_concept_detail = mongodb_util.find_query_data('ths_stock_concept_detail', query)
-        sub_stock_symbol_list = list(ths_stock_concept_detail['symbol'])
-    company_info.loc[:, 'sub_stock'] = False
-    company_info.loc[company_info['symbol'].isin(sub_stock_symbol_list), 'sub_stock'] = True
-
-    try:
-        company_info.dropna(subset=['symbol'], axis=0, inplace=True)
-        company_info.dropna(subset=['_id'], axis=0, inplace=True)
-        mongodb_util.save_mongo(company_info, db_name_constant.COMPANY_INFO)
-        # 保存历史数据
-        save_company_info_his(company_info)
-    except BaseException as e:
-        logger.error("出现异常:{},{}", symbol, e)
-
-    return company_info
-
-
-def save_company_info_his(company_info_df):
-    now_date = datetime.now()
-    str_day = now_date.strftime('%Y-%m-%d')
-    company_info_df['symbol'] = company_info_df['_id']
-    company_info_df['str_day'] = str_day
-    company_info_df['_id'] = company_info_df['_id'] + "_" + str_day
-    remove_query = {'str_day': str_day}
-    tag = mongodb_util.remove_data(remove_query, db_name_constant.COMPANY_INFO_HIS)
-    success = tag.acknowledged
-    if success:
-        mongodb_util.insert_mongo(company_info_df, db_name_constant.COMPANY_INFO_HIS)
-
+# 同步公司基本信息
 
 def sync_company_base_info(symbol_list):
     global result
     result = []
     create_index()
     east_money_stock_info = get_east_money_stock_info()
+    east_money_stock_info = east_money_stock_info.sort_values(by=['list_date'], ascending=False)
+
     de_listed_stock_list = company_common_service_api.get_de_list_company()
     east_money_stock_info = east_money_stock_info.loc[~(
         east_money_stock_info['symbol'].isin(de_listed_stock_list))]
@@ -240,7 +56,7 @@ def sync_company_base_info(symbol_list):
     # 将日期数值转换为日期时间格式
     east_money_stock_info['list_date_01'] = pd.to_datetime(east_money_stock_info['list_date'], format='%Y%m%d')
 
-    # 开票啦实时数据
+    # 开盘啦实时数据
     kpl_real_time_quotes = kpl_real_time_quotes_api.get_kpl_real_time_quotes()
 
     if symbol_list is not None:
@@ -249,12 +65,16 @@ def sync_company_base_info(symbol_list):
     page_number = round(count / MAX_PAGE_NUMBER, 0) + 1
     page_number = int(page_number)
     threads = []
+
+    exist_company_df = mongodb_util.find_all_data(db_name_constant.COMPANY_INFO)
+
     # 创建多个线程来获取数据
     for page in range(page_number):  # 0到100页
         end_count = (page + 1) * MAX_PAGE_NUMBER
         begin_count = page * MAX_PAGE_NUMBER
-        page_df = east_money_stock_info.loc[begin_count:end_count]
-        thread = threading.Thread(target=single_thread_sync_company_info, args=(page_df, kpl_real_time_quotes))
+        page_df = east_money_stock_info.iloc[begin_count:end_count]
+        thread = threading.Thread(target=single_thread_sync_company_info,
+                                  args=(page_df, kpl_real_time_quotes, exist_company_df))
         threads.append(thread)
         thread.start()
 
@@ -263,23 +83,55 @@ def sync_company_base_info(symbol_list):
         thread.join()
 
     fail_df = east_money_stock_info.loc[east_money_stock_info['symbol'].isin(result)]
-    single_thread_sync_company_info(fail_df, kpl_real_time_quotes)
+    single_thread_sync_company_info(fail_df, kpl_real_time_quotes, exist_company_df)
+
+
+def get_east_money_stock_info():
+    all_real_time_quotes = em_stock_info_api.get_a_stock_info()
+    all_real_time_quotes = all_real_time_quotes[['symbol',
+                                                 'name',
+                                                 "now_price",
+                                                 'total_mv',
+                                                 'flow_mv',
+                                                 'pe_ttm',
+                                                 'sz_sh',
+                                                 'area',
+                                                 'pb',
+                                                 'list_date',
+                                                 'ROE',
+                                                 'total_share',
+                                                 'flow_share',
+                                                 'industry',
+                                                 'amount',
+                                                 "hk_stock_code",
+                                                 "hk_stock_name",
+                                                 'concept']]
+
+    return all_real_time_quotes
 
 
 def single_thread_sync_company_info(east_money_stock_info,
-                                    kpl_real_time_quotes):
+                                    kpl_real_time_quotes, exist_company_df):
     global result
     fail_list = []
     for company_one in east_money_stock_info.itertuples():
         try:
-            company_info_type = ths_stock_api.get_company_info_detail(company_one.symbol)
+            # 同步公司控股子公司信息 异步执行
+            sync_company_hold_info_api.sync_company_hold_info(company_one.symbol)
+
+            company_info_type = ths_company_info_web.get_company_info_detail(company_one.symbol)
+            company_info_type = set_kzz_debt(company_info_type, company_one.symbol)
             company_info_type['first_industry_code'] = company_info_type['hycode'].apply(
                 lambda x: x[1:3] + '0000')
             company_info_type['second_industry_code'] = company_info_type['hy2code'].apply(
                 lambda x: x[1:5] + '00')
             company_info_type['third_industry_code'] = company_info_type['hy3code'].apply(
                 lambda x: x[1:7])
-
+            # company_info_type['main_business_list'] = company_info_type['main_business_list']
+            # company_info_type['most_profitable_business'] = company_info_type['most_profitable_business']
+            # company_info_type['most_profitable_business_rate'] = company_info_type['most_profitable_business_rate']
+            # company_info_type['most_profitable_business_profit'] = company_info_type['most_profitable_business_profit']
+            #
             company_info_type['first_sw_industry'] = company_info_type['hy']
             company_info_type['second_sw_industry'] = company_info_type['hy2']
             company_info_type['third_sw_industry'] = company_info_type['hy3']
@@ -293,6 +145,9 @@ def single_thread_sync_company_info(east_money_stock_info,
             company_info_type['em_industry'] = company_one.industry
             company_info_type['em_concept'] = company_one.concept
 
+            company_info_type['hk_stock_code'] = company_one.hk_stock_code
+            company_info_type['hk_stock_name'] = company_one.hk_stock_name
+
             company_info_type['now_price'] = company_one.now_price
             company_info_type['total_share'] = company_one.total_share
             company_info_type['flow_share'] = company_one.flow_share
@@ -311,16 +166,27 @@ def single_thread_sync_company_info(east_money_stock_info,
             company_info_type['total_mv_sp'] = company_one.total_mv_sp
             company_info_type['flow_mv_level'] = company_one.flow_mv_level
             company_info_type['classification'] = company_one.classification
-            company_info_type['mv_circulation_ratio'] = calculate_circu_ratio(company_one.symbol)
-            # 获取同花顺最新概念
-            company_info_type = ths_concept_common_service_api.set_ths_concept(company_one.symbol, company_info_type)
+
             now_date = datetime.now()
             str_now_date = now_date.strftime('%Y-%m-%d %H:%M:%S')
+
+            now_str_day = now_date.strftime('%Y-%m-%d')
             company_info_type['sync_date'] = str_now_date
-            fix_symbol_industry_df = company_constant_data_api.get_fix_symbol_industry()
+
+            result_dict = calculate_circulation_ratio(company_one.symbol, now_str_day)
+
+            company_info_type['mv_circulation_ratio'] = result_dict['mv_circulation_ratio']
+            company_info_type['qfii_type'] = result_dict['qfii_type']
+            company_info_type['qfii_number'] = result_dict['qfii_number']
+            company_info_type['share_holder_sync_day'] = result_dict['share_holder_sync_day']
+
+            # 获取同花顺最新概念
+            company_info_type = ths_concept_common_service_api.set_ths_concept(company_one.symbol, company_info_type)
+
+            fix_symbol_industry_df = company_constant_data.get_fix_symbol_industry()
             if company_one.symbol in list(fix_symbol_industry_df['symbol']):
                 # fix sw_industry
-                company_info_type = company_constant_data_api.fix_symbol_industry(company_info_type, company_one.symbol)
+                company_info_type = company_constant_data.fix_symbol_industry(company_info_type, company_one.symbol)
 
             # todo fix industry
             company_info_type['industry'] = company_info_type['second_sw_industry']
@@ -333,8 +199,9 @@ def single_thread_sync_company_info(east_money_stock_info,
             str_day = now_date.strftime('%Y-%m-%d')
             deal_days = k_line_common_service_api.get_deal_days(str_day, company_one.symbol)
             company_info_type['deal_days'] = deal_days
+
             # 设置年报信息
-            company_info_type = get_recent_year_income(company_one.symbol, company_info_type)
+            company_info_type = get_recent_year_income(company_one.symbol, company_info_type, exist_company_df)
 
             try:
                 if data_frame_util.is_not_empty(kpl_real_time_quotes):
@@ -355,7 +222,7 @@ def single_thread_sync_company_info(east_money_stock_info,
             except BaseException as e:
                 logger.warning("设置开盘啦数据异常:{},{}", company_one.symbol, e)
 
-            company_info_type = filed_sort(company_info_type)
+            company_info_type = company_constant_data.filed_sort(company_info_type)
             mongodb_util.save_mongo(company_info_type.copy(), 'company_info_base')
             logger.info("同步公司信息完成:{}", company_one.symbol + '-' + company_one.name)
         except BaseException as e:
@@ -366,6 +233,113 @@ def single_thread_sync_company_info(east_money_stock_info,
         result = fail_list
 
 
+# 计算实际流通比例
+def calculate_circulation_ratio(symbol, now_str_day):
+    query = {"symbol": symbol}
+    stock_gdfx_free_top_1 = mongodb_util.descend_query(query, 'stock_gdfx_free_top_10', "period", 1)
+    if stock_gdfx_free_top_1.shape[0] == 0:
+        mv_circulation_ratio = 1
+        qfii_number = 0
+        qfii_type = 'A股'
+        share_holder_sync_day = now_str_day
+    else:
+        period_time = list(stock_gdfx_free_top_1['period'])[0]
+
+        query_free = {'symbol': symbol, 'period': period_time}
+        stock_gdfx_free_top_10 = mongodb_util.find_query_data('stock_gdfx_free_top_10', query_free)
+
+        stock_gdfx_free_top_10['shares_number_str'] = stock_gdfx_free_top_10['shares_number'].astype(str)
+
+        stock_gdfx_free_top_10['id_key'] = stock_gdfx_free_top_10['symbol'] + '_' + stock_gdfx_free_top_10[
+            'period'] + '_' + stock_gdfx_free_top_10.shares_number_str
+
+        stock_gdfx_free_top_10.drop_duplicates('id_key', keep='last', inplace=True)
+
+        # 排除香港结算公司 大于5%减持不用发公告  香港中央结算    HKSCC
+        stock_gdfx_free_top_10['is_hk'] = stock_gdfx_free_top_10['shareholder_name'].apply(
+            lambda shareholder_name: "HK" if shareholder_name.startswith('香港中央结算') or shareholder_name.startswith(
+                'HKSCC') else "A")
+
+        # 持股大于5% 减持需要发公告
+        # 排除香港结算公司不发公共 小于5%减持不用发公告
+        # 香港中央结算    HKSCC
+        stock_free_top_greater_than_5 = stock_gdfx_free_top_10.loc[
+            (stock_gdfx_free_top_10['circulation_ratio'] >= 5) & (stock_gdfx_free_top_10['is_hk'] == 'A')]
+
+        stock_free_qfii = stock_gdfx_free_top_10.loc[stock_gdfx_free_top_10['shareholder_nature'] == 'QFII']
+
+        share_holder_sync_day = list(stock_gdfx_free_top_10['create_day'])[0]
+
+        # qfii 数量
+        qfii_number = stock_free_qfii.shape[0]
+        # qfii 类型
+        qfii_type = set_qfii_type(qfii_number, stock_free_qfii.copy())
+
+        circulation_ratio = sum(stock_free_top_greater_than_5['circulation_ratio'])
+        mv_circulation_ratio = round((100 - circulation_ratio) / 100, 2)
+        # 防止错误数据
+        if mv_circulation_ratio < 0:
+            mv_circulation_ratio = 1
+
+    result_dict = {
+        'mv_circulation_ratio': mv_circulation_ratio,
+        'qfii_type': qfii_type,
+        'qfii_number': qfii_number,
+        'share_holder_sync_day': share_holder_sync_day
+
+    }
+    return result_dict
+
+
+def set_qfii_type(qfii_number, stock_free_qfii):
+    if qfii_number > 0:
+        stock_free_qfii['new_change'] = stock_free_qfii['change']
+        stock_free_qfii.loc[stock_free_qfii['change_ratio'] == 0, 'new_change'] = 0
+        stock_free_qfii.loc[stock_free_qfii['change'] == '新进', 'new_change'] = \
+            stock_free_qfii['shares_number']
+        stock_free_qfii['new_change'] = stock_free_qfii['new_change'].astype(float)
+
+        stock_free_qfii_new_in = stock_free_qfii.loc[stock_free_qfii['change'] == '新进']
+        if data_frame_util.is_not_empty(stock_free_qfii_new_in):
+            qfii_type = 1
+            return qfii_type
+
+        stock_free_qfii_add = stock_free_qfii.loc[
+            (~stock_free_qfii['change'].isin(['不变', '新进'])) & (stock_free_qfii['new_change'] > 0)]
+
+        if data_frame_util.is_not_empty(stock_free_qfii_add):
+            qfii_type = 2
+            return qfii_type
+
+        stock_free_qfii_not_change = stock_free_qfii.loc[stock_free_qfii['change'] == '不变']
+
+        if data_frame_util.is_not_empty(stock_free_qfii_not_change):
+            qfii_type = 3
+            return qfii_type
+
+        stock_free_qfii_reduce = stock_free_qfii.loc[
+            (~stock_free_qfii['change'].isin(['不变', '新进'])) & (stock_free_qfii['new_change'] < 0)]
+
+        if data_frame_util.is_not_empty(stock_free_qfii_reduce):
+            qfii_type = 4
+            return qfii_type
+    else:
+        return 0
+
+
+def create_index():
+    mongodb_util.create_index('company_info',
+                              [("classification", 1)])
+    mongodb_util.create_index('company_info',
+                              [("industry", 1)])
+    mongodb_util.create_index('company_info',
+                              [("flow_mv", 1)])
+    mongodb_util.create_index('company_info',
+                              [("list_date", 1)])
+    mongodb_util.create_index('company_info',
+                              [("symbol", 1)])
+
+
 def set_kpl_data(kpl_real_time_quotes_one, company_info_type, company_one):
     if data_frame_util.is_not_empty(kpl_real_time_quotes_one):
         company_info_type['kpl_plate_name'] = list(kpl_real_time_quotes_one['plate_name_list'])[0]
@@ -394,22 +368,38 @@ def set_kpl_data(kpl_real_time_quotes_one, company_info_type, company_one):
     return company_info_type
 
 
-# 更新新上市公司信息
-def new_company_info_update():
-    east_money_stock_info = get_east_money_stock_info()
-    new_stock = common_service_fun_api.get_new_stock(east_money_stock_info)
-    for company_one in new_stock.itertuples():
-        try:
-            sync_company_base_info([company_one.symbol])
-            fix_company_industry(company_one.symbol)
+# 获取可转债信息
+@lru_cache(maxsize=None)
+def get_kzz_debt_info():
+    query = {}
+    kzz_debt_info_df = mongodb_util.find_query_data(db_name_constant.KZZ_DEBT_INFO, query)
+    kzz_debt_info_df = kzz_debt_info_df[[
+        'symbol',
+        'name',
+        'stock_code',
+        'apply_date',
+        'list_date',
+        'due_date'
+    ]]
+    return kzz_debt_info_df
 
-        except BaseException as e:
-            logger.error("出现异常:{}", e)
-    company_common_service_api.company_info_industry_cache_clear
+
+def set_kzz_debt(df, symbol):
+    kzz_debt_info_df_all = get_kzz_debt_info()
+    kzz_debt_info_df = kzz_debt_info_df_all.loc[kzz_debt_info_df_all['stock_code'] == symbol]
+    df.loc[:, 'kzz_debt_list'] = ''
+    if data_frame_util.is_not_empty(kzz_debt_info_df):
+        kzz_debt_info_df_list = kzz_debt_info_df.to_dict(orient='records')
+        df.at[0, 'kzz_debt_list'] = kzz_debt_info_df_list
+    return df
 
 
 # 保存申万行业分类
 def save_sw_data(company_info_type):
+    now_date = datetime.now()
+    hour = now_date.hour
+    if hour <= 15:
+        return company_info_type
     first_sw_info = company_info_type[[
         'first_sw_industry',
         'first_industry_code'
@@ -485,15 +475,38 @@ def save_sw_data(company_info_type):
 
 
 # 获取最近年报收入
-def get_recent_year_income(symbol, company_info_type):
+def get_recent_year_income(symbol, company_info_type, exist_company_df):
+    now_date = datetime.now()
+    hour = now_date.hour
+    # if hour <= 15:
+    #     exist_company_one_df = exist_company_df.loc[exist_company_df['_id'] == symbol]
+    #     if data_frame_util.is_not_empty(exist_company_one_df):
+    #         company_info_type['operate_profit'] = list(exist_company_one_df['operate_profit'])[0]
+    #         company_info_type['total_operate_income'] = list(exist_company_one_df['total_operate_income'])[0]
+    #         if 'operate_date_name' in exist_company_one_df:
+    #             company_info_type['operate_date_name'] = list(exist_company_one_df['total_operate_income'])[0]
+    #         else:
+    #             company_info_type['operate_date_name'] = '暂无年报'
+    #     else:
+    #         company_info_type['operate_profit'] = 0
+    #         company_info_type['total_operate_income'] = 0
+    #         company_info_type['operate_date_name'] = '暂无年报'
+    #     return company_info_type
     query = {'symbol': symbol, "REPORT_TYPE": "年报"}
     em_stock_profit = mongodb_util.descend_query(query, db_name_constant.EM_STOCK_PROFIT, 'REPORT_DATE', 1)
     if data_frame_util.is_not_empty(em_stock_profit):
         company_info_type['operate_profit'] = list(em_stock_profit['OPERATE_PROFIT'])[0]
-        company_info_type['total_operate_income'] = list(em_stock_profit['TOTAL_OPERATE_INCOME'])[0]
+        company_info_type['operate_date_name'] = list(em_stock_profit['REPORT_DATE_NAME'])[0]
+        total_operate_income = list(em_stock_profit['TOTAL_OPERATE_INCOME'])[0]
+        # 金融机构大多收入计入在这个字段中
+        if total_operate_income == 0:
+            total_operate_income = list(em_stock_profit['OPERATE_INCOME'])[0]
+
+        company_info_type['total_operate_income'] = total_operate_income
     else:
         company_info_type['operate_profit'] = 0
         company_info_type['total_operate_income'] = 0
+        company_info_type['operate_date_name'] = '暂无年报'
     company_info_type['operate_profit'] = round(
         company_info_type['operate_profit'] / common_service_fun_api.HUNDRED_MILLION, 2)
     company_info_type['total_operate_income'] = round(
@@ -501,12 +514,9 @@ def get_recent_year_income(symbol, company_info_type):
     return company_info_type
 
 
+import mns_scheduler.company_info.constant.company_constant_data as company_constant_data
+
 if __name__ == '__main__':
-    # while True:
-    #     df = ths_stock_api.get_company_info_detail('000001')
-    #     print(df)
-    # company_info_update()
-    # company_info_type = ths_stock_api.get_company_info_detail('836699')
     # sync_company_base_info()
     # fix_company_industry()
     # calculate_circu_ratio("601069")
@@ -514,6 +524,8 @@ if __name__ == '__main__':
     # 300293
     # sync_company_base_info(None)
     # new_company_info_update()
-    sync_company_base_info(['920002'])
-    fix_company_industry(None)
-    # group_by_industry()
+    # query = {"total_operate_income": 0}
+    # un_report_company_info = mongodb_util.find_query_data(db_name_constant.COMPANY_INFO, query)
+    # symbol_list = list(un_report_company_info['symbol'])
+    sync_company_base_info(['920009'])
+    sync_company_base_info(None)

mns_scheduler/company_info/base/sync_company_hold_info_api.py

@@ -0,0 +1,37 @@
+import sys
+import os
+
+file_path = os.path.abspath(__file__)
+end = file_path.index('mns') + 17
+project_path = file_path[0:end]
+sys.path.append(project_path)
+import mns_common.api.ths.company.ths_company_info_api as ths_company_info_api
+import mns_common.component.cookie.cookie_info_service as cookie_info_service
+import mns_common.utils.data_frame_util as data_frame_util
+from mns_common.db.MongodbUtil import MongodbUtil
+import mns_common.constant.db_name_constant as db_name_constant
+from datetime import datetime
+from mns_common.utils.async_fun import async_fun
+from loguru import logger
+
+mongodb_util = MongodbUtil('27017')
+
+
+# 同步公司控股子公司信息
+@async_fun
+def sync_company_hold_info(symbol):
+    try:
+        ths_cookie = cookie_info_service.get_ths_cookie()
+        company_hold_info_df = ths_company_info_api.get_company_hold_info(symbol, ths_cookie)
+
+        now_date = datetime.now()
+        sync_str_date = now_date.strftime('%Y-%m-%d %H:%M:%S')
+        if data_frame_util.is_not_empty(company_hold_info_df):
+            company_hold_info_df['sync_str_date'] = sync_str_date
+            mongodb_util.insert_mongo(company_hold_info_df, db_name_constant.COMPANY_HOLDING_INFO)
+    except BaseException as e:
+        logger.error("同步公司控股子公司信息:{},{}", symbol, e)
+
+
+if __name__ == '__main__':
+    sync_company_hold_info('300085')