example.py 文件源码

def items(self, task, response):
        items = []
        document = lxml.html.document_fromstring(html=response.text)
        products = document.xpath("//div[@class='product']")
        for product in products:
            iid = int(product.xpath(".//@product-id")[0])
            name = product.xpath(".//h2/text()")[0]
            desc = product.xpath(".//p/text()")[0]
            category = product.xpath(".//span/text()")[0]
            price = float(product.xpath(".//em/text()")[0])
            images = product.xpath(".//div//img/@src")
            item = Product(
                iid=iid,
                url=response.url,
                name=name,
                category=category,
                desc=desc,
                price=price,
                images=images,
            )
            items.append(item)
        return items