def grab_product(self, flag, url):
product = Product()
# 来源
product.flag = flag
# 源链接
product.url = url
# 状态正常
product.status = "1"
# 英文
product.language_id = "2"
# 品牌
product.brand = "Ted Baker"
pg = self.do_visit(url)
# 解析商品信息
if self.ana_product_info(product, pg):
# 保存商品信息
self.save_product(product)
# 保存描述信息
self.save_product_desc(product)
pimg = ProductImages()
pimg.spid = product.spider_product_id
# 解析图片信息
self.ana_product_images(pimg, pg)
# 保存图片信息
self.save_product_images(pimg)
psku = ProductSku()
# 解析并保存sku信息
self.ana_and_save_product_sku(psku, product)
else:
self.log_info("".join([url, " product not saved!"]))
def grab_product(self, flag, url):
product = Product()
# 来源
product.flag = flag
# 源链接
product.url = url
# 状态正常
product.status = "1"
# 英文
product.language_id = "2"
pg = self.do_visit(url)
# 解析商品信息
if self.ana_product_info(product, pg):
# 保存商品信息
self.save_product(product)
# 保存描述信息
self.save_product_desc(product)
# 保存图片信息
pimg = ProductImages()
pimg.spid = product.spider_product_id
if len(product.images) > 0:
pimg.images = ",".join(product.images)
self.save_product_images(pimg)
else:
# 记录没有图片的商品
self.log_info("".join([str(pimg.spid), " no images!"]))
psku = ProductSku()
# 保存sku信息
self.ana_and_save_product_sku(psku, product)
else:
self.log_info("".join([url, " product not saved!"]))
def save_product_images(product, pidao):
pimg = ProductImages()
pimg.spid = product.spider_product_id
images = product.images
img_arr = []
for img in images:
img_url = "".join([img['url'], "?wid=750"])
img_arr.append(img_url)
pimg.images = ",".join(img_arr)
if pidao.is_exists_product_images(pimg.spid):
# pidao.update_product_images(pimg)
print "".join([str(pimg.spid), " images is exists!"])
else:
pidao.save(pimg)
def grab_product(self, flag, url):
# 先爬取意大利站的信息
iturl = url.replace("/cn/", "/it/")
it_page = self.do_visit(iturl)
product = Product()
product.flag = flag
# 源链接
product.url = iturl
# 状态设为在架
product.status = "1"
# 先爬取的意大利站信息，记录描述的语言为英语
product.language_id = "2"
# 解析商品信息
if self.ana_product_info(product, it_page):
# 保存商品信息
self.save_product(product)
# 保存描述信息
self.save_product_desc(product)
pimg = ProductImages()
pimg.spid = product.spider_product_id
# 解析保存图片信息
self.ana_product_images(pimg, it_page)
self.save_product_images(pimg)
# 换成中国站，爬取中文描述
cnurl = iturl.replace("/it/", "/cn/")
cn_page = self.do_visit(cnurl)
# 设置描述语言为中文
product.language_id = "1"
# 解析并保存商品描述
self.ana_product_info(product, cn_page)
self.save_product_desc(product)
# 调用商品sku信息接口获取sku信息
product_url = "".join([
"https://www.farfetch.cn/it/product/GetDetailState?productId=",
product.resource_code, "&designerId=0"
])
skus = self.get_json(product_url)
# 解析并保存sku信息
self.ana_and_save_skus(product.spider_product_id, skus)
else:
# 未解析成功的记录爬取失败
self.log_info("".join([url, " product not find!"]))
def grab_product(self, flag, url):
product = Product()
# 来源
product.flag = flag
# 源链接
product.url = url
# 状态正常
product.status = "1"
# 英语
product.language_id = "2"
pg = self.do_visit(url)
# 解析商品信息
if self.ana_product_info(product, pg):
# 保存商品信息
self.save_product(product)
# 保存描述信息
self.save_product_desc(product)
pimg = ProductImages()
pimg.spid = product.spider_product_id
img_arr = []
if len(product.images) > 0:
for img in product.images:
# -20后缀结束的图片不是需要的图片
if "-20." in img:
continue
img_arr.append(img)
pimg.images = ",".join(img_arr)
# 保存图片信息
self.save_product_images(pimg)
else:
# 记录没有图片的商品
self.log_info("".join([str(pimg.spid), " no images!"]))
psku = ProductSku()
# 解析并保存sku信息
self.ana_and_save_product_sku(psku, product, pg)
else:
self.log_info("".join([url, " product not saved!"]))
def grab_product(self, flag, url):
# 将连接替换成意大利站点（意大利站是欧元价）
surl = url.replace("/cn/", "/it/")
pg = self.do_visit(surl)
product = Product()
# 来源
product.flag = flag
# 源链接
product.url = surl
# 此站只有女士
product.gender = "1"
# 状态正常
product.status = "1"
# 中文
product.language_id = "1"
# 解析商品信息
if self.ana_product_info(product, pg):
# 保存商品信息
self.save_product(product)
# 保存商品描述
self.save_product_desc(product)
pimg = ProductImages()
pimg.spid = product.spider_product_id
self.ana_product_images(pimg, pg)
self.save_product_images(pimg)
psku = ProductSku()
psku.spid = product.spider_product_id
self.ana_and_save_product_sku(psku, pg)
# 语言切换成英文
en_pg = self.do_visit(surl.replace("/zh/", "/en/"))
product.language_id = "2"
# 保存英文版商品信息
if self.ana_product_info(product, en_pg):
self.save_product_desc(product)
else:
self.log_info("".join([surl, " product not saved!"]))