多线程带智能采集策略的采集系统

    去年年底的时候曾经发过一个数据采集器网页数据采集器,那是专门针对某一个网站来进行采集的,如果需要采集新的网站内容,就需要修改代码并重新编译。

    昨晚完成了一个带智能策略的采集系统。其实,这个策略的方案三年前就想好了,那时候打算用VB做,做了一半就搁置了。现在用C#才终于把这个方案实现了。

    整个方案大概是这样的:

    需要建立一个AC数据库,MSSQL也行,有四个表:PageType用于记录页面的种类,比如列表页和详细页两类;Url表用于记录要采集的网址,另外还有一个字段TypeID标明该网址属于哪一种页面类型,比如是列表页还是详细页;Rule表记录着各种规则,主要有三个字段,FromTypeID源页类型,ToTypeID目的页类型,Pattern规则;CjPage用于存储采集到的网页内容,还包含网址和页面种类。

    采集策略的核心就在于规则库Rule。

    工作过程大概这样:
1,采集线程从Url表抽取一个网址,并马上在表中将其删除,为了防止冲突,这个过程需要用多线程同步解决;
2,用WebClient请求该网址的页面内容;
3,取得内容后,给线程池的线程来分析处理,本线程回到1,继续去Url表取下一个网址;
4,线程池在有空闲线程时,会调用分析函数ParsePage去处理上次获得的页面内容;
5,先到Rule中取所有FromTypeID为当前网址TypeID;
6,如果没有取到任何规则Rule,则将本页内容写入到CjPage中;
7,如果取到规则,那么遍历规则,为每条规则执行ParseUrl方法;
8,ParseUrl根据规则的Pattern匹配到页面内容中的所有网址,并记录到Url中,规则的ToTypeID就是Url的TypeID。

    至此,整个流程就完成了。下面举一个实际例子来说明一下:
    我要截取动网开发者网络的所有ASP文章http://www.cndw.com/tech/asp/
    首先,在页面类型库中加入列表页和详细页两行,再把http://www.cndw.com/tech/asp/写入到Url中,页面类型是列表页;
    其次,在Rule中加入两条规则:
        一,从列表页取得详细页的网址FromTypeID=1  ToTypeID=2,Pattern是· <a href="([^>]*)" target=_blank>,这条规则将会识别列表页上的所有详细页的链接,并记入到Url中,TypeID是详细页;
        二,从列表页取得列表页的网址FromTypeID=1  ToTypeID=1,Pattern是<a href='([^>]*)'>下一页<\/a>,这条规则将会取得当前列表页上的下一页的链接,并记入到Url中,TypeID还是列表页。
    采集器工作时,如果采集的是详细页的内容,将会直接写入到CjPage中,因为没有FromTypeID=2的规则;而采集的是列表页的内容时,就要做两件事了,因为有两条FromTypeID=1的规则,一件事是识别当前列表页中所有文章的链接并存入Url,另一件事是识别下一列表页链接并存入Url。
    由于规则具有递归性,使得采集器能递归采集到所有的文章。

    下面是一些核心源码(没有公开的都是一些数据层的添删改查的代码):

以下是代码片段:

using System;
using System.Collections.Generic;
using System.Text;
using System.Net;
using System.Threading;
using CJData;
using System.Text.RegularExpressions;
using NLog;

namespace CJ
{
    /// <summary>
    /// 写日志委托
    /// </summary>
    /// <param name="log"></param>
    public delegate void WriteLogCallBack(String log);
    /// <summary>
    /// 采集
    /// </summary>
    public class CaiJi
    {
        private WebClient _wc;

        public WebClient Wc
        {
            get
            {
                if (_wc == null) _wc = new WebClient();
                return _wc;
            }
        }
        private Thread thread;

        public String Name = "";
        public event WriteLogCallBack OnWriteLog;

        /// <summary>
        /// 开始工作
        /// </summary>
        public void Start()
        {
            if (thread != null) return;
            thread = new Thread(new ThreadStart(Work));
            thread.Start();
        }
        /// <summary>
        /// 停止工作
        /// </summary>
        public void Stop()
        {
            if (thread != null) thread.Abort();
            thread = null;
        }

        private void Work()
        {
            int times = 0;
            while (times < 100)
            {
                Url url = Url.SelectOne();
                try
                {
                    if (url != null)
                    {
                        String page = Wc.DownloadString(url.UrlAddress);
                        if (!String.IsNullOrEmpty(page))
                        {
                            OnWriteLog(Name + " 成功抓取:" + url.UrlAddress);
                            times = 0;
                            ThreadPool.QueueUserWorkItem(new WaitCallback(ParsePage), new Object[] { url, page });
                        }
                    }
                    else
                    {
                        //OnWriteLog(Name + " 没有工作,休息半秒");
                        times++;
                        //没有工作,休息半秒
                        Thread.Sleep(500);
                    }
                }
                catch (ThreadAbortException e)
                {
                    OnWriteLog(Name + " 外部终止");
                    break;
                }
                catch (Exception e)
                {
                    times++;
                    OnWriteLog(Name + " 赚取" + url.UrlAddress + "出错,休息半秒。" + e.Message);
                    Trace.WriteLine(url.UrlAddress);
                    //出错,休息半秒
                    Thread.Sleep(500);
                }
            }
            OnWriteLog(Name + " 完成!");
        }

        private void ParsePage(Object state)
        {
            Object[] objs = (Object[])state;
            Url url = objs[0] as Url;
            String page = (String)objs[1];
            IList<Rule> rs = Rule.SelectAll(Rule._.FromTypeID, url.TypeID);
            //if (url.PageType.TypeName == "详细页")
            if (rs == null || rs.Count < 1)
            {
                CjPage cp = new CjPage();
                cp.CjTime = DateTime.Now;
                cp.Content = page;
                cp.Url = url.UrlAddress;
                cp.TypeID = url.TypeID;
                cp.Insert();
            }
            else
            {
                foreach (Rule r in rs)
                {
                    ParseUrl(url, r, page);
                }
            }
        }
        private void ParseUrl(Url u, Rule r, String page)
        {
            Regex reg = new Regex(r.Pattern, RegexOptions.Compiled | RegexOptions.IgnoreCase);
            MatchCollection ms = reg.Matches(page);
            foreach (Match m in ms)
            {
                Url url = new Url();
                url.TypeID = r.ToTypeID;
                url.UrlAddress = m.Groups[1].Value;
                if (!url.UrlAddress.StartsWith("http://"))
                {
                    if (url.UrlAddress.Substring(0, 1) == "/")
                    {
                        url.UrlAddress = u.UrlAddress.Substring(0, u.UrlAddress.IndexOf("/", 8)) + url.UrlAddress;
                    }
                    else
                    {
                        if (u.UrlAddress.Substring(u.UrlAddress.Length - 1) == "/")
                            url.UrlAddress = u.UrlAddress + url.UrlAddress;
                        else
                            if (u.UrlAddress.LastIndexOf("/") < u.UrlAddress.LastIndexOf("."))
                                url.UrlAddress = u.UrlAddress.Substring(0, u.UrlAddress.LastIndexOf("/") + 1) + url.UrlAddress;
                            else
                                url.UrlAddress = u.UrlAddress + "/" + url.UrlAddress;
                    }
                }
                url.Insert();
            }
        }
    }
}

以下是代码片段:

using System;
using System.Collections.Generic;
using System.ComponentModel;
using System.Data;
using System.Drawing;
using System.Text;
using System.Windows.Forms;
using System.Net;

namespace CJ
{
    public partial class Form1 : Form
    {
        public Form1()
        {
            InitializeComponent();
        }

        CaiJi[] cjs;
        private void button1_Click(object sender, EventArgs e)
        {
            Button btn = sender as Button;
            if (btn.Text == "停止")
            {
                foreach (CaiJi cj in cjs)
                {
                    if (cj != null) cj.Stop();
                }
                cjs = null;
                btn.Text = "开始";
                return;
            }

            richTextBox1.Text = "";
            btn.Text = "停止";

            int k = 100;
            if (!int.TryParse(textBox1.Text, out k)) k = 100;
            cjs = new CaiJi[k];
            for (int i = 0; i < cjs.Length; i++)
            {
                cjs[i] = new CaiJi();
                cjs[i].Name = "线程" + i.ToString("00");
                cjs[i].OnWriteLog += new WriteLogCallBack(cj_OnWriteLog);
            }
            foreach (CaiJi cj in cjs)
            {
                cj.Start();
            }
        }

        void cj_OnWriteLog(string log)
        {
            if (richTextBox1.InvokeRequired)
            {
                richTextBox1.Invoke(new WriteLogCallBack(cj_OnWriteLog), new object[] { log });
            }
            else
            {
                if (richTextBox1.Lines.Length > 3000) richTextBox1.Text = "";
                richTextBox1.Text = log + Environment.NewLine + richTextBox1.Text;
            }
        }
    }
}


我不相信神话,我只相信汗水!我不相信命运,我只相信双手!
上一篇:如何部署编译NDIS驱动的环境(内部资料)


下一篇:在递归函数中因不正确使用公共变量而形成死循环