python爬取图片链接标签的src属性值_python 正则提取img标签和src

#coding=utf-8

importpymssqlimportredefconnectDB():

conn= pymssql.connect(server='****', user='User', password='****', database='*****',

charset='cp936')

cur=conn.cursor()

sql= 'select ProductID,Content from Products WHERE (not Content IS NULL )'cur.execute(sql)

row=cur.fetchone()

resultList=[]whilerow:#print("ProductID=%s,Content=%s" % (row[0], row[1]))

result = parseContent(row[1])ifresult:

tmp=[]#print("解析出的img为:")

#print(result)

#tmp.append(int(row[0])) # 将productID转成int类型，方便下面的比较

tmp.append(row[0])

tmp.append(result)

resultList.append(tmp)try:

row=cur.fetchone()exceptUnicodeDecodeError:continueconn.close()returnresultListdefparseContent(content):

pattern= ']*/>'result=re.findall(pattern, content)returnresultdefsaveImg(resultList):

productIdList=getExtraBookProductIDList()

conn= pymssql.connect(server='****', user='User', password='****', database='*****',

charset='cp936')

cur=conn.cursor()for result in resultList: #遍历解析出来的imgList

if result[0] inproductIdList:#提取src

imgSrc = getImgSrc(result[1])for img inimgSrc:

sql_1= """update ExtraBookInfo set YImage='{img}' WHERE ProductID='{pID}'""".format(

img=img, pID=result[0])print(sql_1)

cur.execute(sql_1)

conn.commit()else:#sql_2 = """insert into ExtraBookInfo (ProductID,YImage) values( '{pID}','{img}')""".format(

#pID=result[0], img=tmp)

for img inimgSrc:

cur.execute('insert into ExtraBookInfo ProductID,YImage values(%s,%s)', (result[0], img))

conn.commit()

conn.close()defgetExtraBookProductIDList():

conn= pymssql.connect(server='****', user='User', password='****', database='*****',

charset='cp936')

cur=conn.cursor()

sql= 'select ProductID from ExtraBookInfo'cur.execute(sql)

productIdList=[]

row=cur.fetchone()whilerow:

productIdList.append(row[0])try:

row=cur.fetchone()exceptUnicodeDecodeError:continueconn.close()returnproductIdListdefgetImgSrc(result):for r inresult:

pattern_2= 'http.*?\.jpg'p2=re.findall(pattern_2, r)print(p2)returnp2

resultList=connectDB()

saveImg(resultList)