def createSpider(self, kwags):
site = kwags.get("site", "")
site = site
str_ptn = kwags.get("ptns", "")
ptn = re.compile(str_ptn)
ptns = [ptn]
onScapyDic = {"onScapy": OnScapy(), "print": print_OnScapy(), "logfile": logfile_OnScapy()}
max_size = kwags.get("max_size", "100")
max_size = int(max_size)
onScapy = kwags.get("output_type", "onScapy")
onscapy = onScapyDic[onScapy]
logfileName = kwags.get("output_name", None)
multi = kwags.get("multi_thread", "False")
threadnumber = kwags.get("threadnumber", None)
if not site:
raise Exception("not site input")
if logfileName and onScapy == "logfile":
onscapy = logfile_OnScapy(logfileName)
spiderCls = Spider
if multi == "True":
spiderCls = MultiSpider
spider = spiderCls(site, ptns, max_size, onScapy=onscapy)
if multi == "True" and threadnumber:
threadnumber = int(threadnumber)
spider.setThreadNumber(threadnumber)
return spider
def test_cls_MultiScapy_scapy__logfile_OnScapy():
site = "http://www.hao123.com/"
max_size = 100
ptns = producePtns()
name = 'test_result/test_cls_MultiScapy_scapy__logfile_OnScapy'
onscapy = logfile_OnScapy(name=name)
multispider = MultiSpider(site,ptns,max_size=max_size,onScapy=onscapy)
multispider.scapy()
def test_cls_Spider_scapy__logfile_OnScapy():
site = "http://www.hao123.com/"
max_size = 100
logfile = "test_result/logfile"
ptns = producePtns()
onscapy = logfile_OnScapy(name=logfile)
spider = Spider(site,ptns,max_size=max_size,onScapy=onscapy)
spider.scapy()
def test_cls_logfile_OnScapy():
name = "test_result/logfile"
url,html = 'url','html'
on_scapy = logfile_OnScapy(name=name)
on_scapy.beforeScapy()
on_scapy.filterHtml(url, html)
on_scapy.finishScapy()
url = url+'\n'
f = open(name)
logfile = f.read()
f.close()
assert logfile==url,('logfile_OnScapy class error',logfile,url)
