From 2b023c95c2d0ccd16614914405e925363a77e93f Mon Sep 17 00:00:00 2001 From: "yihua.huang" Date: Fri, 11 Apr 2014 11:43:04 +0800 Subject: [PATCH] qqmeishi demo --- .../webmagic/model/samples/QQMeishi.java | 27 +++++++++++++++++++ 1 file changed, 27 insertions(+) create mode 100644 webmagic-samples/src/main/java/us/codecraft/webmagic/model/samples/QQMeishi.java diff --git a/webmagic-samples/src/main/java/us/codecraft/webmagic/model/samples/QQMeishi.java b/webmagic-samples/src/main/java/us/codecraft/webmagic/model/samples/QQMeishi.java new file mode 100644 index 0000000..f4f8591 --- /dev/null +++ b/webmagic-samples/src/main/java/us/codecraft/webmagic/model/samples/QQMeishi.java @@ -0,0 +1,27 @@ +package us.codecraft.webmagic.model.samples; + +import us.codecraft.webmagic.Site; +import us.codecraft.webmagic.model.ConsolePageModelPipeline; +import us.codecraft.webmagic.model.OOSpider; +import us.codecraft.webmagic.model.annotation.ExtractBy; +import us.codecraft.webmagic.model.annotation.TargetUrl; + +/** + * @author code4crafter@gmail.com + * @date 14-4-11 + */ +@TargetUrl("http://meishi.qq.com/beijing/c/all[\\-p2]*") +@ExtractBy(value = "//ul[@id=\"promos_list2\"]/li",multi = true) +public class QQMeishi { + + @ExtractBy("//div[@class=info]/a[@class=title]/h4/text()") + private String shopName; + + @ExtractBy("//div[@class=info]/a[@class=title]/text()") + private String promo; + + public static void main(String[] args) { + OOSpider.create(Site.me(), new ConsolePageModelPipeline(), QQMeishi.class).addUrl("http://meishi.qq.com/beijing/c/all").thread(4).run(); + } + +}