def parse(self, response):
sel = Selector(response)
urls = sel.xpath("//ul[@class='vT-srch-result-list-bid']/li").extract()
for i in range(len(urls)):
item = zhaotou_zhaobItem()
item['crl_30101001'] = "".join(
sel.xpath(
"(//ul[@class='vT-srch-result-list-bid']/li)[%s]/a/@href" %
(i + 1)).extract())
item['crl_10416003'] = \
"".join(
sel.xpath("(//ul[@class='vT-srch-result-list-bid']/li)[%s]/span/text()" % (i + 1)).extract()).split(
"|")[0]
item['crl_10416018'] = \
"".join(
sel.xpath("(//ul[@class='vT-srch-result-list-bid']/li)[%s]/span/text()" % (i + 1)).extract()).split(
"|")[1]
item['crl_10416022'] = \
"".join(
sel.xpath("(//ul[@class='vT-srch-result-list-bid']/li)[%s]/span/text()" % (i + 1)).extract()).split(
"|")[2]
item['crl_10416002'] = "".join(
sel.xpath(
"(//ul[@class='vT-srch-result-list-bid']/li)[%s]/span/a/text()"
% (i + 1)).extract())
item['crl_10416008'] = "".join(
sel.xpath(
"(//ul[@class='vT-srch-result-list-bid']/li)[%s]/span/strong[last()]//text()"
% (i + 1)).extract())
yield Request(item['crl_30101001'],
callback=self.parse_item,
meta={'item': item})
def parse_item(self, response):
l = ItemLoader(zhaotou_zhaobItem(), response)
l.add_value("crl_30101001", response.url)
l.add_xpath("crl_30101002", "//div[@class='entry']//text()", MapCompose(unicode.strip))
l.add_value("crl_30101003", u"陕西采购与招标网陕西省招投标协会")
l.add_value("crl_30101004", u"公开招标")
l.add_xpath("crl_10416001", "//h1/text()", MapCompose(unicode.strip))
l.add_xpath("crl_10416003", "//span[@class='time']/text()", MapCompose(unicode.strip))
return l.load_item()
def parse_item(self, response):
l = ItemLoader(zhaotou_zhaobItem(), response)
# URL
l.add_value("crl_30101001", response.url)
# 网页内容
l.add_xpath("crl_30101002", "//div[@class='wz']//text()",
MapCompose(unicode.strip))
l.add_value("crl_30101003", u"电梯采购网")
l.add_value("crl_30101004", u"公开招标")
# 项目标题
l.add_xpath("crl_10416001", "///h2//text()", MapCompose(unicode.strip))
# 发布时间
l.add_xpath("crl_10416003", "//div[@class='zz']//text()",
MapCompose(unicode.strip))
return l.load_item()
def parse_item(self, response):
l = ItemLoader(zhaotou_zhaobItem(), response)
# URL
l.add_value("crl_30101001", response.url)
# 网页内容
l.add_xpath("crl_30101002", "//div[@id='vsb_content']//text()",
MapCompose(unicode.strip))
l.add_value("crl_30101003", u"西安市财政局")
l.add_value("crl_30101004", u"公开招标")
# 项目标题
l.add_xpath("crl_10416001", "//td[@class='titlestyle3181']//text()",
MapCompose(unicode.strip))
# 发布时间
l.add_xpath("crl_10416003", "//span[@class='timestyle3181']//text()",
MapCompose(unicode.strip))
return l.load_item()
def parse_item(self, response):
l = ItemLoader(zhaotou_zhaobItem(), response)
l.add_value("crl_30101001", response.url)
l.add_xpath(
"crl_30101002",
"//div[@id='vsb_content_2']//text() | //div[@id='vsb_content']//text()",
MapCompose(unicode.strip))
l.add_value("crl_30101003", u"西安高新技术产业开发区")
l.add_value("crl_30101004", u"公开招标")
l.add_xpath("crl_10416001",
"//tbody/tr[1]/td/table/tbody/tr[1]/td//text()",
MapCompose(unicode.strip))
l.add_xpath("crl_10416003", "(//td[@nowrap='nowrap'][11])[1]//text()",
MapCompose(unicode.strip))
return l.load_item()