def doTask(eid, url, ename, param):
'''
循环6次尝试获取指定url的内容
然后存储到远程数据库
返回True,否则False
'''
param['eid'] = str(eid)
param['ename'] = ename
# param['url'] = param['preUrl'] + url
param['url'] = param['preUrlip'] + url
proxyList = jTool.getProxy('proxy.txt')
return rp.crawlUrl(param['conn'], param, proxyList)
def operLog(logFileName):
file = open(logFileName)
line = file.readline()
param = makeParam()
param['taskTable'] = 'item_url_task'
param['num'] = '999'
proxyList = jTool.getProxy('proxy.txt')
pcount = len(proxyList)-1
proxy = proxyList[random.randint(0, pcount)]
while line:
if line[0]=='G':
tmp = line.split(',')
for t in tmp:
tt = t.split(':')
if len(tt)>1:
param[tt[0].strip()] = tt[1]
if len(tt)>2:
param[tt[0]] = tt[1]+':'+tt[2]+':'+tt[3].strip('\n')
param['id'] = param['eid']
rp.crawlUrl(param['conn'], param, proxy)
if line[0]=='P':
# print line
pass
line = file.readline()
def doTask(task, param, proxy):
param['id'] = str(task[0])
param['url'] = param['preUrlip'] + task[1]
param['ename'] = task[2]
# try:
return rp.crawlUrl(param['conn'], param, proxy)
def makeParam():
paramDic = {}
paramDic['station'] = '信用浙江'
paramDic['begin_url'] = 'http://www.zjcredit.gov.cn:8000/CreditQuery.aspx?sectionID=02'
paramDic['query_url'] = 'http://www.zjcredit.gov.cn:8000/ListQuery.aspx'
paramDic['post_data_dic'] = {'isIntermediary': 'False', 'isOpen': 'False', 'pageLength': '20', 'recordTotal': '1778190', 'sectionID': '02', 'sortDirection': '1', 'sortField': 'CreditID'}
paramDic['preUrl'] = 'http://www.zjcredit.gov.cn:8000/EnterpriseInfo.aspx?creditID='
paramDic['preUrlip'] = 'http://218.108.28.28:8000/EnterpriseInfo.aspx?creditID='
paramDic['basePostUrl'] = 'http://www.zjcredit.gov.cn:8000/GetInfoByDataSupplier.aspx'
paramDic['basePostUrlip'] = 'http://218.108.28.28:8000/GetInfoByDataSupplier.aspx'
paramDic['dbHost'] = 'localhost'
paramDic['dbUser'] = '******'
paramDic['dbPasswd'] = 'root'
paramDic['rdb'] = 'rawData'
conn = jTool.initCursor(paramDic['dbHost'], paramDic['dbUser'], paramDic['dbPasswd'], paramDic['rdb'])
paramDic['conn'] = conn
return paramDic
param = makeParam()
param['taskTable'] = 'item_url_task'
param['num'] = '999'
proxyList = jTool.getProxy('proxy.txt')
pcount = len(proxyList)-1
proxy = proxyList[random.randint(0, pcount)]
param['url'] = 'http://www.zjcredit.gov.cn:8000/EnterpriseInfo.aspx?creditID=F651B7F17FEEDA7A'
param['eid'] = '69999'
param['id'] = '69999'
param['ename'] = '湖州市邮政局千金邮电所'
rp.crawlUrl(param['conn'], param, proxy)
