linksys.py 文件源码

def parse_kb(self, response):
        mib = None

        # need to perform some nasty segmentation because different firmware versions are not clearly separated
        # reverse order to get MIB before firmware items
        for entry in reversed(response.xpath(
                "//div[@id='support-article-downloads']/div/p")):
            for segment in reversed(entry.extract().split("<br><br>")):
                resp = HtmlResponse(
                    url=response.url, body=segment, encoding=response.encoding)
                for href in resp.xpath("//a/@href").extract():
                    text = resp.xpath("//text()").extract()

                    if "MIBs" in href:
                        mib = href

                    elif "firmware" in href:
                        text = resp.xpath("//text()").extract()

                        item = FirmwareLoader(
                            item=FirmwareImage(), response=resp, date_fmt=["%m/%d/%Y"])
                        item.add_value("date", item.find_date(text))
                        item.add_xpath("url", "//a/@href")
                        item.add_value("mib", mib)
                        item.add_value("product", response.meta["product"])
                        item.add_value("vendor", self.name)
                        item.add_value(
                            "version", FirmwareLoader.find_version_period(text))
                        yield item.load_item()