Soufun_News
using AnfleCrawler.Common;
using System;
using System.Collections.Generic;
using System.ComponentModel;
using System.IO;
using System.Linq;
using System.Net;
using System.Text;
using System.Threading.Tasks; namespace AnfleCrawler.DataAnalyzer
{
internal class Soufun_News : AnalyzerBase
{
private enum Kind
{
[Description("市场")]
Market = ,
[Description("政策")]
Policy = ,
[Description("公司")]
Company = ,
} private static readonly string[] FilterTags = new string[] { "script", "iframe" }; public override void Init(PageCrawler crawler)
{
string exp = string.Format("http://news.sh.soufun.com/more/[{0}]/[1-50].html", string.Join(",", Enum.GetValues(typeof(Kind)).Cast<int>()));
crawler.PushUrl(new StringPatternGenerator(exp), );
base.Init(crawler);
} protected override void AnalyzeInternal(PageLandEntity current)
{
var lander = Crawler.Lander;
dynamic repository = Repository;
var pHandler = CreateContentHandler(current);
switch (current.Depth)
{
case :
{
var dom = lander.GetDocument(pHandler);
foreach (var node in QueryNodes(dom.DocumentNode, ".contenttext"))
{
var linkNode = QueryNode(node, "a.link_01");
string url = GetHref(linkNode, current.Url).OriginalString;
int i = url.LastIndexOf(".");
Crawler.PushUrl(new Uri(url.Insert(i, "_all")), );
}
}
break;
case :
{
var dom = lander.GetDocument(pHandler);
var hackNode = QueryNode(dom.DocumentNode, "#newxq_B01_26");
string kind = QueryNodes(hackNode, "a").Last().InnerText;
string title = QueryNode(dom.DocumentNode, "h1").InnerText;
var contentNode = QueryNode(dom.DocumentNode, "#news_body");
foreach (string tag in FilterTags)
{
foreach (var node in QueryNodes(contentNode, tag, false).ToArray())
{
node.Remove();
}
}
var set = QueryNodes(dom.DocumentNode, "#newxq_B01_27 span").Take().ToArray();
string source = null;
DateTime publishDate;
DateTime.TryParse(set[].InnerText, out publishDate);
if (set.Length == )
{
source = set[].InnerText;
}
repository.SaveNews(current.Url, kind, source, title, contentNode.InnerHtml, publishDate);
Crawler.OutWrite("保存新闻 {0}", title);
}
break;
}
}
}
}
public void SaveNews(Uri pageUrl, string kind, string source, string title, string content, DateTime publishDate)
{
Guid rowID = CryptoManaged.MD5Hash(pageUrl.OriginalString);
using (var db = Create())
{
var q = from t in db.News
where t.RowID == rowID
select t;
var news = q.SingleOrDefault();
if (news == null)
{
db.News.Add(news = new News()
{
RowID = rowID,
SiteID = pageUrl.Authority,
});
}
news.Kind = kind;
news.Source = source;
news.Title = title;
news.Content = content;
news.PublishDate = publishDate;
db._SaveChanges();
}
}
Soufun_News的更多相关文章
随机推荐
- iPad和iPhone开发区别
原文:http://mobile.51cto.com/iphone-273895.htm iPad与iPhone 开发区别详解是本文要介绍的内容,先来看看他们的区别. 1.首先我们先从官方发布的SDK ...
- em
macro jumptocaller(){ JumpToLocation(GetSymbolLocation((GetCurSymbol ())))}
- 给NIOS II CPU增加看门狗定时器并使用
给NIOS II CPU增加看门狗定时器并使用 配置看门狗定时器: 设置计时溢出时间为1秒 计数器位宽为32位 勾选No Start/Stop control bits 勾选Fixed perio ...
- YTU 2345: 后序遍历二叉树
原文链接:https://www.dreamwings.cn/ytu2345/2611.html 2345: 后序遍历二叉树 时间限制: 1 Sec 内存限制: 128 MB 提交: 3 解决: ...
- 用jquery实现简单的表单验证
HTML代码: 1 <form action="" method="post" id="form-data"> 2 <di ...
- CocoaPods安装第三方出错:XCode7.3
错误[!] The dependency `Masonry (~> 0.6.1)` is not used in any concrete target. 在之前,我使用的版本是XCode7.0 ...
- js蛋疼的Class(获取class对象)
之前处理class对象被迫用jquery,以为JavaScript不支持class.确实百度某些大神也说js不待见class对象,就这样本菜放弃了原生js. 今天偶然发现一片文章吐槽jquery拥抱J ...
- 关于php语言的使用! ------php语言与JavaScript的使用 方法是相似
<script type="text/javascript"> </script>--js与PHP同是一种弱类型语言 弱类型语言只是不显示表现 定义变量时 ...
- 微软2017校招笔试题2 composition
题目 Alice writes an English composition with a length of N characters. However, her teacher requires ...
- PHP中的变量详解
php变量通过名只能我们就知道首先变量,是在程序执行期间,可以变化的量. 1.那变量是干嘛的呢,用变量就可以来保存我们值,这就是变量,那么我们接着来看,知道了变量是什么,以及它能干什么,我们再来看一下 ...