获取CIK 9/8

base/BaseCore.py

@@ -678,9 +678,19 @@ class BaseCore:
                 id = selects[0]
                 return id
 
-
-
-
+    # 更新企业的CIK
+    def updateCIK(self,social_code,cik):
+        try:
+            sql = f"UPDATE EnterpriseInfo SET CIK = '{cik}' WHERE SocialCode = '{social_code}'"
+            cnn = self.pool_caiji.connection()
+            cursor = cnn.cursor()
+            cursor.execute(sql)
+            cnn.commit()
+            cursor.close()
+            cnn.close()
+        except:
+            log = self.getLogger()
+            log.info('======保存企业CIK失败=====')
 
 
 

comData/annualReport_XQW/annualreportUS.py

@@ -19,6 +19,8 @@ urllib3.disable_warnings(urllib3.exceptions.InsecureRequestWarning)
 from bs4 import BeautifulSoup
 from kafka import KafkaProducer
 # from selenium import webdriver
+log = baseCore.getLogger()
+
 
 def paserUrl(html,listurl):
     # soup = BeautifulSoup(html, 'html.parser')
@@ -237,6 +239,39 @@ def getrequest(social_code,url,headers,data):
         result = ''
     return result
 
+def getCIK(social_code,code):
+    cik = ''
+    #"MNSO" post请求 获取企业CIK
+    payload = {"keysTyped":f"{code}","narrow":True}
+    data = json.dumps(payload)
+    result = getrequest(social_code,url,headers,data)
+    #判断接口返回的数据哪一条是该企业 根据股票代码
+    tickers = result['hits']['hits']
+    if len(tickers) == 0:
+        log.error(f'{code}....{social_code}....无hits')
+        return cik
+    for ticker in tickers:
+        try:
+            i_t_ = ticker['_source']['tickers']
+            i_ts = i_t_.split(', ')
+        except:
+            continue
+        for i_t in i_ts:
+            if i_t == code:
+                cik = ticker['_id']
+                if len(cik) < 10:
+                    cik = format(int(cik),'0>10d')
+                    baseCore.updateCIK(social_code,cik)
+                    break
+        if cik != '':
+            break
+    if cik == '':
+        log.error(f'{code}....{social_code}....无CIK')
+    else:
+        log.info(f'{code}....{social_code}....cik为{cik}')
+    return cik
+
+
 if __name__ == '__main__':
     headers = {
         'authority': 'efts.sec.gov',
@@ -308,7 +343,7 @@ if __name__ == '__main__':
         #         print(cik)
         #         break
         # break
-        spider(com_name,cik)
+        # spider(com_name,cik)
         # break