basic.py 文件源码

def parse_downurl(self,response):
        try:
            antivirus1 =response.css("#static_antivirus").extract()[0]
            antivirus = Selector(response).css("#static_antivirus").extract()[0]
            # ?Static Analysis ------ Antivirus????
            antiresult  = re.findall("((Microsoft|Kaspersky|ESET\-NOD32)</td>\n\s*<td>\n\s*<span class=\"text\-error\")",antivirus.encode("utf-8"),re.S)
            # ?????????????????eset????????????????????
            if antiresult == []:
                return
            # ?????????????
            url = response.xpath("//a[contains(@class,'btn-primary')]/@href").extract()[0].encode('utf-8')
            url = urlparse.urljoin("https://malwr.com",url)

            item = MalwrItem()
            item['file_urls'] = [url]
            return item
        except Exception,e:
            pass
        return