all

pymysql.connect(hostlocalhost,port3306,userroot,password123456,databasetest)#

创建游标对象self.cursor

__del__(self):self.cursor.close()self.db.close()print(关闭数据库)2.3

生成header

timeout2)else:print(未输入send_type直接返回None)res

Nonereturn

fhttps://www.baidu.com/CTMDS/pub/PUB010100.do?methodhandle04compId{row[companyId]}#

发起请求#

htmlTextres.text)else:print(f{urlDetail}请求失败)2.6

find_all()

的所有子节点并判断这些节点是否符合过滤条件最后以列表形式将符合条件的内容返回语法格式如下

find_all(

的直接子节点。

text用来搜文档中的字符串内容该参数可以接受字符串

、正则表达式

class_[col-sm-8])divtextListO***r

[re.sub(r\s,

row[recordStatus]cancelRecordTime

)compLevel

medical_register(company_id,area_name,record_no,comp_name,address,link_man,link_tel,record_status,comp_level,record_time,cancel_record_time,o***r_org_add)

sql2

fvalues({companyId},{areaName},{recordNo},{compName},{address},{linkMan},{linkTel},{recordStatus},{compLevel},{recordTime},{cancelRecordTime},{o***rOrgAdd})sql3

sql1

执行sqlself.cursor.execute(sql3)#

提交self.db.commit()#

(company_id,professional_name,principal_investigator,job_title)

values(%s,%s,%s,%s)self.cursor.execute(sql4,

tdTextList)#

处理数据self.parseData(dataListdataList)else:print(f{self.url}请求失败)except

Exception

mySpider(https://www.baidu.com/CTMDS/pub/PUB010100.do?methodhandle05)spider.run()3

完整代码

pymysql.connect(hostlocalhost,port3306,userroot,passwordlogicfeng,databasetest2)#

创建游标对象self.cursor

__del__(self):self.cursor.close()self.db.close()print(关闭数据库)#

获取一个headerdef

timeout2)else:print(未输入send_type直接返回None)res

Nonereturn

fhttps://www.baidu.com/CTMDS/pub/PUB010100.do?methodhandle04compId{row[companyId]}#

发起请求#

htmlTextres.text)else:print(f{urlDetail}请求失败)#

解析页面def

class_[col-sm-8])divtextListO***r

[re.sub(r\s,

row[recordStatus]cancelRecordTime

)compLevel

medical_register(company_id,area_name,record_no,comp_name,address,link_man,link_tel,record_status,comp_level,record_time,cancel_record_time,o***r_org_add)

sql2

fvalues({companyId},{areaName},{recordNo},{compName},{address},{linkMan},{linkTel},{recordStatus},{compLevel},{recordTime},{cancelRecordTime},{o***rOrgAdd})sql3

sql1

执行sqlself.cursor.execute(sql3)#

提交self.db.commit()#

(company_id,professional_name,principal_investigator,job_title)

values(%s,%s,%s,%s)self.cursor.execute(sql4,

tdTextList)#

处理数据self.parseData(dataListdataList)else:print(f{self.url}请求失败)except

Exception

mySpider(https://www.百度.com/CTMDS/pub/PUB010100.do?methodhandle05)spider.run()