public static Queue <HunterProxy> GetProxy(String r, List <String> pfilter)
{
Queue <HunterProxy> ResultList = new Queue <HunterProxy>();
Regex regex = new Regex(StrRegex);
Match m = regex.Match(r);
List <String> CommentRemovedFilter = new List <String>();
foreach (String s in pfilter)
{
String key = s.Substring(0, s.IndexOf('\'') >= 0 ? s.IndexOf('\'') : s.Length);
if (key.Trim() == "")
{
continue;
}
CommentRemovedFilter.Add(key);
}
while (m.Success)
{
bool Legal = true;
HunterProxy tempProxy = new HunterProxy();
tempProxy.IPAndPort = m.Result("${ipandport}");
tempProxy.Type = m.Result("${type}");
tempProxy.Speed = m.Result("${speed}");
tempProxy.Description = m.Result("${description}").Replace("\n", "").Replace("\r", "").Trim();
if (CommentRemovedFilter.Count > 0)
{
//过滤器有内容，则进行筛选。同时要去除以'开头的注释
foreach (String f in CommentRemovedFilter)
{
if (tempProxy.Description.Contains(f.Trim()))
{
Legal = true;
break;
}
Legal = false;
}
}
if (Legal)
{
ResultList.Enqueue(tempProxy);
}
m = m.NextMatch();
}
return(ResultList);
}
/// <summary>
/// 捕获带有filetype结尾的链接。proxy为代理，为null表示不使用代理。
/// </summary>
/// <returns>返回一个本次页面中捕获的链接序列</returns>
public List <string> HuntUris(HunterProxy proxy, HunterForm main)
{
Regex linkReg = null;
String htmlCode = null;
List <string> thisURL = new List <string>(); //记录本次匹配的所有URL项
try
{
linkReg = new Regex(strategy.StrategyData.configuration.Regex); //超链接+超链接文本
WebProxy webproxy;
if (proxy != null)
{
webproxy = new WebProxy(proxy.IPAndPort);
}
else
{
webproxy = null;
}
if (proxy != null)
{
mHunterConsole.WriteDetails("正在使用代理：" + proxy.IPAndPort + "(" + proxy.Description + ")");
}
mHunterConsole.WriteDetails("准备分析页面：" + urlAddress);
htmlCode = GetPageHtml(webproxy, main);
mHunterConsole.WriteHTML(htmlCode);
}
catch (WebException ex) //如果是返回超时，返回一个Count>0的随机结果
{
thisURL.Add("{/WebException/}" + new Random().Next().ToString());
mHunterConsole.WriteDetails("页面" + urlAddress + "请求失败。原因：" + ex.Message);
mHunterConsole.ReportAbandonURI(new UriResource(urlAddress, strategy.CurrentKeywordProgress,
strategy.CurrentSearchProgress, null), ex.Message);
return(thisURL);
}
catch (Exception ex)
{
mHunterConsole.WriteException(ex);
}
try
{
Match m = linkReg.Match(htmlCode);
while (m.Success)
{
allCount++;
mHunterConsole.outputAnalysedUris(DateTime.Now, allCount);
//得到一个网址后，保存起来
string linkText = m.Result("${text}");
thisURL.Add(linkText); //记录本次获取到的linkText
if (strategy.HasForbiddenWord(linkText))
{
m = m.NextMatch();
continue; //如果含有违禁词语 则放弃下载 继续下一个
}
//对linkText中的内容进行处理，去掉里面的尖括号
Regex r = new Regex("<(.*?)>");
linkText = r.Replace(linkText, "");
string uri = null;
try
{
uri = (strategy.StrategyData.configuration.Redirect.ToLower() == "true") ? GetTheRedirectUrl(m.Result("${url}")) : (m.Result("${url}"));
}
catch (WebException)
{
mHunterConsole.WriteDetails("链接" + (m.Result("${url}") + "重定向超时。"));
mHunterConsole.ReportAbandonURI(new UriResource((m.Result("${url}")), strategy.CurrentKeywordProgress,
strategy.CurrentSearchProgress, null), "重定向超时");
m = m.NextMatch();
continue;
}
if (uri.EndsWith("." + strategy.Filetype))
{
availableCount++;
mHunterConsole.outputAvailableUris(DateTime.Now, availableCount);
UriResource u = new UriResource(uri, strategy.CurrentKeywordProgress
, strategy.CurrentSearchProgress, linkText); //封装成一个Uri资源
if (!uriQueue.Contains(u)) //考虑在多线程中，可能会出现重复项目
{
uriQueue.Enqueue(u); //将一个资源放入队列
}
mHunterConsole.outputDownloadingUriInfo(DateTime.Now,
"找到的资源的URL:" + u.Url + Environment.NewLine +
"标题：" + u.Text + Environment.NewLine +
"关键字：" + strategy.GetKeyword(u.Keyword) + Environment.NewLine +
"搜索页码：" + u.index + Environment.NewLine +
"已列入下载队列。");
mHunterConsole.WriteDetails("正在获得有效URI：" + uri);
}
m = m.NextMatch();
}
}
catch (Exception ex)
{
mHunterConsole.WriteException(ex);
}
return(thisURL);
}
public string GetPageHtml(Encoding encoding, WebProxy proxy, HunterForm main)
{
if ((strategy.StrategyData.configuration.UseIE && ((HunterMain)main).mHunterConfig.HunterCore == HunterConfig.Core.Default) ||
((HunterMain)main).mHunterConfig.HunterCore == HunterConfig.Core.IE)
{
lock (this)
{
try
{
if (proxy != null)
{
HunterProxy.InternetSetOption(proxy.Address.ToString().Replace("http://", "").Replace("/", ""));
}
else
{
HunterProxy.InternetSetOption(String.Empty);
}
string html = string.Empty;
AutoResetEvent are = new AutoResetEvent(false);
((HunterMain)main).IEBrowser.NewWindow += (object sender, CancelEventArgs e) =>
{
e.Cancel = true;
};
((HunterMain)main).IEBrowser.DocumentCompleted += (object sender, WebBrowserDocumentCompletedEventArgs e) =>
{
((HunterMain)main).IEBrowser.ScriptErrorsSuppressed = true;
html = ((HunterMain)main).IEBrowser.DocumentText;
((HunterMain)main).IEBrowser.Dispose();
((HunterMain)main).IEBrowser = new WebBrowser();
are.Set();
};
String[] parameters = { urlAddress };
main.Invoke(new StringHandler(((HunterMain)main).IEBrowser.Navigate), parameters);
are.WaitOne(15000);
if (html != String.Empty)
{
return(html);
}
else
{
throw new WebException();
}
}
catch (Exception) { throw; }
}
}
else
{
HttpWebRequest request = null;
HttpWebResponse response = null;
StreamReader reader = null;
try
{
request = (HttpWebRequest)WebRequest.Create(urlAddress);
if (proxy != null)
{
request.Proxy = proxy;
}
strategy.Disguise(request);
response = (HttpWebResponse)request.GetResponse();
if (response.StatusCode == HttpStatusCode.OK && response.ContentLength < 1024 * 1024)
{
reader = new StreamReader(response.GetResponseStream(), encoding);
string html = reader.ReadToEnd();
return(html);
}
}
catch (Exception) { throw; }
finally
{
if (response != null)
{
response.Close();
response = null;
}
if (reader != null)
{
reader.Close();
}
if (request != null)
{
request = null;
}
}
return(string.Empty);
}
}
public Hunter(HunterConsole oh, HunterConfig config, ProjectInfo _pj, HunterForm main)
{
try
{
MainForm = main;
Error = false;
mHunterConsole = oh;
projectInfo = _pj;
mHunterConfig = config;
ProxyFetcher = new HunterProxyFetcher(AvailableProxies);
projectInfo = ProjectInfo.LoadProject(_pj.mHunterConsole, _pj.projectPath, _pj.strategyPath, true);
downloadThreadNum = int.Parse(projectInfo.threadnum);
hunterThreads = new HunterDownloadThread[downloadThreadNum];
//获取代理的线程
ProxyGetThreads = new Thread[downloadThreadNum];
if (mHunterConfig.UseProxy == true)
{
FileStream fs = new FileStream("proxy.hip", FileMode.Open, FileAccess.Read);
StreamReader sr = new StreamReader(fs);
ProxyText = sr.ReadToEnd();
sr.Close();
fs.Close();
AllProxies = HunterProxy.GetProxy(ProxyText, mHunterConfig.ProxyFilterKeywords);
}
mHunterConsole.WriteMessage(projectInfo.ConfigInformation());
mHunterConsole.WriteMessage("");
mHunterConsole.WriteMessage(projectInfo.strategy.GetStrategyInformation());
xmlDatabase = new XMLDatabase(projectInfo.database, mHunterConsole);
xmlDatabase.openDatabase();
try
{
if (downloadThreadNum <= 0)
{
mHunterConsole.WriteMessage("配置错误：下载线程数不能小于0。");
return;
}
}
catch (Exception ex)
{
mHunterConsole.WriteException(ex);
return;
}
for (int i = 0; i < hunterThreads.Length; i++)
{
hunterThreads[i] = new HunterDownloadThread();
hunterThreads[i].downloadThread = new Thread(threadDownloadUris);
}
if (mHunterConfig.UseProxy)
{
for (int i = 0; i < ProxyGetThreads.Length; i++)
{
ProxyGetThreads[i] = new Thread(GetAvaliableProxies);
}
}
if (projectInfo.strategy.Keywords.Count <= 0)
{
projectInfo.mHunterConsole.WriteMessage("没有找到关键字，任务取消。");
Error = true;
return;
}
thHuntUris = new Thread(threadHuntUris);
thHuntUris.SetApartmentState(ApartmentState.STA);
hUri = new HunterUri(this);
projectInfo.strategy.RecordFirstWord();
mHunterConsole.WriteMessage("下载线程总数：" + hunterThreads.Length);
mHunterConsole.WriteMessage("读取配置完毕。");
mHunterConsole.WriteMessage("正在运行任务...");
}
catch (Exception e)
{
mHunterConsole.WriteException(e);
}
}
