public cExport(ContainerControl sender, Delegate senderDelegate, cGlobalParas.PublishType pType,string FileName,System.Data.DataTable pData )
{
m_sender = sender;
m_senderDelegate = senderDelegate;
m_pType =pType ;
m_FileName = FileName;
m_pData = pData;
}
public void WriteLog(string TaskName,cGlobalParas.LogType LogType, string strLog)
{
string FileName = Program.getPrjPath() + "Log\\" + TaskName + DateTime.Now.Year;
if (DateTime.Now.Month.ToString().Length == 1)
FileName = FileName + "0" + DateTime.Now.Month.ToString();
else
FileName = FileName + DateTime.Now.Month.ToString();
if (DateTime.Now.Day.ToString().Length == 1)
FileName = FileName + "0" + DateTime.Now.Day.ToString();
else
FileName = FileName + DateTime.Now.Day.ToString();
FileName = FileName + ".txt";
string Log="";
if (!Directory.Exists(Path.GetDirectoryName(FileName)))
Directory.CreateDirectory(Path.GetDirectoryName(FileName));
FileStream myStream = File.Open(FileName, FileMode.Append, FileAccess.Write, FileShare.Write);
StreamWriter sw = new StreamWriter(myStream, System.Text.Encoding.GetEncoding("gb2312"));
if (LogType ==cGlobalParas.LogType.GatherError )
{
Log =TaskName + "\t" + DateTime.Now .ToString () + "\t" + "�ɼ�����" + "\t" + strLog ;
sw.WriteLine(Log);
}
else if (LogType ==cGlobalParas.LogType.PublishError )
{
Log =TaskName + "\t" + DateTime.Now .ToString () + "\t" + "��������" + "\t" + strLog ;
sw.WriteLine(Log);
}
sw.Close();
myStream.Close();
}
public cRunTaskEventArgs(cGlobalParas.MessageType MessType ,string RunName ,string RunPara)
{
base.MessType = MessType;
m_RunName = RunName;
m_RunPara = RunPara;
}
public cListenInitializedEventArgs(cGlobalParas.MessageType MessType)
{
m_MessType = MessType;
}
/// <summary>
///
/// </summary>
/// <param name="old_state">�ɵ�״̬</param>
/// <param name="new_statue">�µ�״̬</param>
public TaskStateChangedEventArgs(Int64 TaskID, cGlobalParas.TaskState oldState, cGlobalParas.TaskState newState)
{
//m_TaskID = TaskID;
base.TaskID = TaskID;
m_OldState = oldState;
m_NewState = newState;
}
public cGatherUrlCountArgs(Int64 TaskID, cGlobalParas.UpdateUrlCountType uType, int TrueUrlCount)
{
//m_TaskID = TaskID;
base.TaskID = TaskID;
m_TrueUrlCount = TrueUrlCount;
m_UType = uType;
}
//����һ��ͨѶ�Ľӿڷ����������ɼ�����Ĵ����������Ҫ�ɼ�����ҳ�����ô˷���
//�ɴ˷�������cGatherWeb.GetGatherData�����η�����Ŀ����Ϊ�˿��Դ����������
private DataTable GetGatherData(string Url, cGlobalParas.WebCode webCode, string cookie, string startPos, string endPos, string sPath, bool IsAjax)
{
cGatherWeb gWeb = new cGatherWeb();
gWeb.CutFlag = m_TaskSplitData.CutFlag;
DataTable tmpData ;
int AgainTime = 0;
GatherAgain:
try
{
tmpData = gWeb.GetGatherData(Url, m_WebCode, m_Cookie, m_gStartPos, m_gEndPos, m_SavePath, IsAjax);
}
catch (System.Exception ex)
{
AgainTime++;
if (AgainTime > m_AgainNumber)
{
if (m_IsErrorLog == true)
{
//���������־
}
throw ex;
}
else
{
if (m_Ignore404 == true && ex.Message.Contains ("404"))
{
if (m_IsErrorLog == true)
{
//���������־
}
throw ex;
}
else
{
e_Log(this, new cGatherTaskLogArgs(m_TaskID, ((int)cGlobalParas.LogType.Error).ToString() + "��ַ��" + Url + "���ʷ������������Ϣ��" + ex.Message + "���ȴ�3������\n", this.IsErrorLog));
Thread.Sleep(3000);
e_Log(this, new cGatherTaskLogArgs(m_TaskID, ((int)cGlobalParas.LogType.Warning).ToString() + Url + "���ڽ��е�" + AgainTime + "������\n", this.IsErrorLog));
//��������
goto GatherAgain;
}
}
}
return tmpData;
}
//���ݵ�������ȡ��ҳ��ַ����һ������
public List<string> GetUrlsByRule(string Url, string UrlRule,cGlobalParas.WebCode webCode, string cookie)
{
string Url1;
List<string> Urls=new List<string> ();
if (UrlRule.Trim() == "")
{
Urls.Add(Url);
return Urls;
}
//�ж���ַ�Ƿ���ڲ�����������ڲ�����ȡ����һ��������ַ
if (Regex.IsMatch(Url, "{.*}"))
{
List<string> Urls1 = SplitWebUrl(Url ); //,IsUrlEncode ,UrlEncode
Url1 = Urls1[0].ToString();
}
else
{
Url1 = Url;
}
//������ַ��Դ�룬��������ȡ������ȡ��������ַ
//string UrlSource= cTool.GetHtmlSource(Url1,true );
cGatherWeb gW = new cGatherWeb();
string UrlSource = gW.GetHtml(Url1, webCode, cookie, "", "", true, false);
gW = null;
if (UrlSource == "")
{
return null ;
}
//string Rule=@"(?<=href=[\W])" + cTool.RegexReplaceTrans(UrlRule) + @"(\S[^'"">]*)(?=[\s'""])";
string Rule = "";
if (UrlRule.StartsWith("<Regex:"))
{
Rule = @"(?<=[href=|src=|open(][\W])";
//����ǰ
string strPre = UrlRule.Substring(UrlRule.IndexOf("<Regex:")+7, UrlRule.IndexOf(">")-7);
Rule += strPre;
//���������
string cma=@"(?<=<Common:)\S+?(?=>)";
Regex cmas = new Regex(cma, RegexOptions.IgnoreCase | RegexOptions.Multiline);
MatchCollection cs = cmas.Matches(UrlRule);
foreach (Match ma in cs)
{
Rule +=@"(\S*)" + ma.Value.ToString ();
}
//�����
if (Regex.IsMatch(UrlRule, "<End:"))
{
string s = UrlRule.Substring(UrlRule.IndexOf("<End:") + 5, UrlRule.Length - UrlRule.IndexOf("<End:") - 6);
Rule += @"(\S*)" + s;
}
else
{
Rule += @"(\S[^'"">]*)(?=[\s'""])";
}
}
else
{
Rule = @"(?<=[href=|src=|open(][\W])" + cTool.RegexReplaceTrans(UrlRule) + @"(\S[^'"">]*)(?=[\s'""])";
}
Regex re = new Regex(Rule, RegexOptions.IgnoreCase | RegexOptions.Multiline);
MatchCollection aa = re.Matches(UrlSource);
DataTable d1 = new DataTable();
d1.Columns.Add("Name");
foreach (Match ma in aa)
{
//Urls.Add(ma.Value.ToString());
d1.Rows.Add(ma.Value.ToString());
}
//����ʱ���ܻ��ȡ�ظ���ַ���б��������Ҫȥ��
//ȥ���ظ���
string[] strComuns = new string[d1.Columns.Count];
for (int m = 0; m < d1.Columns.Count; m++)
{
strComuns[m] = d1.Columns[m].ColumnName;
}
DataView dv = new DataView(d1);
DataTable d2 = dv.ToTable(true, strComuns);
for (int i = 0; i < d2.Rows.Count; i++)
{
if (string.Compare(d2.Rows[i][0].ToString ().Substring (0,4), "http", true) != 0)
{
string PreUrl = Url;
if (d2.Rows[i][0].ToString().Substring(0, 1) == "/")
{
PreUrl = PreUrl.Substring(7, PreUrl.Length - 7);
PreUrl = PreUrl.Substring(0, PreUrl.IndexOf("/"));
PreUrl = "http://" + PreUrl;
}
else
{
Match a = Regex.Match(PreUrl, ".*/");
PreUrl = a.Groups[0].Value.ToString();
}
Urls.Add(PreUrl + d2.Rows[i][0].ToString());
}
else
{
Urls.Add(d2.Rows[i][0].ToString());
}
}
return Urls;
}
public void InsertOnceLog(cGlobalParas.LogType lType, string PlanID, string PlanName, cGlobalParas.RunTaskType rType, string FileName, string Para)
{
if (!IsExist())
NewLogFile();
cXmlIO xmlconfig = new cXmlIO(Program.getPrjPath() + "tasks\\plan\\RunLog.xml");
string strXml = "<LogType>" + lType + "</LogType>" +
"<PlanID>" + PlanID + "</PlanID>" +
"<PlanName>" + PlanName + "</PlanName>" +
"<FileName>" + FileName + "</FileName>" +
"<FilePara>" + Para + "</FilePara>" +
"<TaskType>" + rType + "</TaskType>" +
"<RunTime>" + DateTime.Now.ToString() + "</RunTime>";
xmlconfig.InsertElement("Logs", "Log", strXml);
xmlconfig.Save();
xmlconfig = null;
}
///����ָ���ĵ����������ҳ�浼������1.6�汾�У������˶�㵼���Ĺ���
///��ַ����������һ�Զ�Ĺ�ϵ����ÿһ����ĵ�����������һ�ԶࣨҲ����һ��һ�Ĺ�ϵ��
///�ڴ������Ǽ������������صĶ������յ���Ҫ�ɼ����ݵ���ַ
///��Ϊ�Ƕ�㵼�������������ڵݹ��һ���㷨
///������ַ�صĶ��DZ���ַ��������������ַ�����
public List<string> ParseUrlRule(string Url, List<cNavigRule> nRules,cGlobalParas.WebCode webCode, string cookie)
{
List<string> pUrls = new List<string>();
List<string> Urls = new List<string>();
pUrls.Add(Url);
//��һ�㵼���ֽⶼ�Ǵ�һ����һ��ַ���У�֮����
//ѡ�ϣ���Ϊ��ͳһ���ýӿڲ���
try
{
Urls = PUrlRule(pUrls, 1, nRules, webCode, cookie);
}
catch (System.Exception ex)
{
//����ʧ�ܣ���������������
return null;
}
return Urls;
}
/// <summary>
/// �ɼ���ҳ����
/// </summary>
/// <param name="Url">��ҳ��ַ</param>
/// <param name="StartPos">��ʼ�ɼ�λ��</param>
/// <param name="EndPos">��ֹ�ɼ�λ��</param>
/// <returns></returns>
public DataTable GetGatherData(string Url,cGlobalParas.WebCode webCode, string cookie, string startPos,string endPos,string sPath,bool IsAjax)
{
tempData = new DataTable("tempData");
int i ;
int j;
string strCut="";
bool IsDownloadFile = false;
#region ������ṹ����������ȡ����
//����ҳ���ȡ�ı�־������ṹ
for (i = 0; i < this.CutFlag.Count; i++)
{
tempData.Columns.Add(new DataColumn(this.CutFlag[i].Title, typeof(string)));
if (this.CutFlag[i].DataType !=(int) cGlobalParas.GDataType.Txt && IsDownloadFile ==false)
{
IsDownloadFile = true;
}
}
//�����û�ָ����ҳ���ȡλ�ù���������ʽ
for (i = 0; i < this.CutFlag.Count; i++)
{
strCut += "(?<" + this.CutFlag[i].Title + ">" + cTool.RegexReplaceTrans(this.CutFlag[i].StartPos) + ")";
//strCut += "(?<=" + cTool.RegexReplaceTrans(this.CutFlag[i].StartPos) + ")";
switch (this.CutFlag[i].LimitSign )
{
case (int)cGlobalParas.LimitSign.NoLimit :
strCut += ".*?";
break;
case (int)cGlobalParas.LimitSign.NoWebSign:
strCut += "[^<>]*?";
break;
case (int)cGlobalParas.LimitSign.OnlyCN:
strCut += "[\\u4e00-\\u9fa5]*?";
break;
case (int)cGlobalParas.LimitSign.OnlyDoubleByte:
strCut += "[^\\x00-\\xff]*?";
break;
case (int)cGlobalParas.LimitSign.OnlyNumber:
strCut += "[\\d]*?";
break;
case (int)cGlobalParas.LimitSign.OnlyChar:
strCut += "[\\x00-\\xff]*?";
break;
case (int)cGlobalParas.LimitSign.Custom:
//strCut += cTool.RegexReplaceTrans(this.CutFlag[i].RegionExpression.ToString());
strCut += this.CutFlag[i].RegionExpression.ToString();
break;
default:
strCut += "[\\S\\s]*?";
break;
}
strCut += "(?=" + cTool.RegexReplaceTrans(this.CutFlag[i].EndPos) + ")|";
}
#endregion
#region ��ȡ��ҳԴ��
int rowCount = this.CutFlag.Count;
//ȥ�����һ����|��
strCut = strCut.Substring(0, strCut.Length - 1);
//��ȡ��ҳ��Ϣ
//�жϴ����Url�Ƿ���ȷ���������ȷ���ؿ�����
if (Regex.IsMatch(Url, "[\"\\s]"))
{
Match aa = Regex.Match(Url, "[\"\\s]");
tempData = null;
return tempData;
}
try
{
GetHtml(Url, webCode, cookie, startPos, endPos, true, IsAjax);
}
catch (System.Web.HttpException ex)
{
throw ex;
}
#endregion
//��ʼ��ȡ��ȡ����
Regex re = new Regex(@strCut, RegexOptions.IgnoreCase | RegexOptions.Multiline );
MatchCollection mc = re.Matches(this.WebpageSource);
if (mc.Count == 0)
{
tempData = null;
return tempData;
}
DataRow drNew=null ;
i = 0;
#region ��ʼ�����ȡ�ַ�����ƴ��һ����
//��ʼ���ݲɼ������ݹ������ݱ�������
//�ڴ���Ҫ����ɼ������п��ܴ��е�����
//���汻ע�͵Ĵ���������������ݱ�Ĵ��룬������ִ�������
//Match ma;
int rows = 0; //ͳ�ƹ��ɼ��˶�����
int m = 0; //����ʹ��
try
{
while (m < mc.Count)
{
//�½�����
drNew = tempData.NewRow();
rows++;
for (i = 0; i < this.CutFlag.Count; i++)
{
if (m < mc.Count)
{
if (i == 0)
{
while (!mc[m].Value.StartsWith(this.CutFlag[i].StartPos, StringComparison.CurrentCultureIgnoreCase))
{
m++;
if (m >= mc.Count)
{
//�˳�����ѭ��
goto ExitWhile;
}
}
drNew[i] = mc[m].Value.Substring(this.CutFlag[i].StartPos.Length, mc[m].Value.Length - this.CutFlag[i].StartPos.Length);
m++;
}
else
{
if (mc[m].Value.StartsWith(this.CutFlag[i].StartPos, StringComparison.CurrentCultureIgnoreCase))
{
drNew[i] = mc[m].Value.Substring(this.CutFlag[i].StartPos.Length, mc[m].Value.Length - this.CutFlag[i].StartPos.Length);
m++;
}
else
{
if (mc[m].Value.StartsWith(this.CutFlag[i - 1].StartPos, StringComparison.CurrentCultureIgnoreCase))
{
m++;
i--;
}
else
{
if (i < this.CutFlag.Count - 1)
{
if (mc[m].Value.StartsWith(this.CutFlag[i + 1].StartPos, StringComparison.CurrentCultureIgnoreCase))
{
}
else
{
m++;
i--;
}
}
else
{
m++;
i--;
}
//���ɼ�ʱ������ȱ�ٲɼ����ݣ����ô˷������вɼ����ݲ���
//drNew[i] = "";
//continue;
}
}
}
}
}
tempData.Rows.Add(drNew);
drNew = null;
}
}
catch (System.Exception ex)
{
throw ex;
}
#endregion
#region ��ʼ����������ƣ����л�ȡ���ݼӹ�
ExitWhile:
//�ڴ��ж��Ƿ���Ҫ�����ʱ�������ݵ�����,��������汾1.2�������������������
for (i = 0; i < this.CutFlag.Count; i++)
{
switch (this.CutFlag[i].ExportLimit)
{
case (int)cGlobalParas.ExportLimit.ExportNoLimit :
break;
case (int)cGlobalParas.ExportLimit.ExportNoWebSign:
for (int index = 0; index < tempData.Rows.Count; index++)
{
tempData.Rows[index][i] = getTxt(tempData.Rows[index][i].ToString());
}
break;
case (int)cGlobalParas.ExportLimit.ExportPrefix:
for (int index = 0; index < tempData.Rows.Count; index++)
{
tempData.Rows[index][i] =this.CutFlag[i].ExportExpression + tempData.Rows[index][i].ToString();
}
break;
case (int)cGlobalParas.ExportLimit.ExportReplace:
for (int index = 0; index < tempData.Rows.Count; index++)
{
string oStr=this.CutFlag[i].ExportExpression .Substring (1,this.CutFlag[i].ExportExpression.IndexOf (",")-2);
string nStr = this.CutFlag[i].ExportExpression.Substring(this.CutFlag[i].ExportExpression.IndexOf(",") + 2, this.CutFlag[i].ExportExpression.Length - this.CutFlag[i].ExportExpression.IndexOf(",") - 3);
tempData.Rows[index][i] = tempData.Rows[index][i].ToString().Replace(oStr,nStr );
}
break;
case (int)cGlobalParas.ExportLimit.ExportSuffix:
for (int index = 0; index < tempData.Rows.Count; index++)
{
tempData.Rows[index][i] = tempData.Rows[index][i].ToString() + this.CutFlag[i].ExportExpression;
}
break;
case (int)cGlobalParas.ExportLimit.ExportTrimLeft:
for (int index = 0; index < tempData.Rows.Count; index++)
{
int len = tempData.Rows[index][i].ToString().Length;
int lefti = int.Parse(this.CutFlag[i].ExportExpression.ToString());
if (tempData.Rows[index][i].ToString().Length > lefti)
{
tempData.Rows[index][i] = tempData.Rows[index][i].ToString().Substring(lefti, len - lefti);
}
}
break;
case (int)cGlobalParas.ExportLimit.ExportTrimRight:
for (int index = 0; index < tempData.Rows.Count; index++)
{
int len = tempData.Rows[index][i].ToString().Length;
int righti = int.Parse(this.CutFlag[i].ExportExpression.ToString());
if (tempData.Rows[index][i].ToString().Length > righti)
{
tempData.Rows[index][i] = tempData.Rows[index][i].ToString().Substring(0, len - righti);
}
}
break;
case (int)cGlobalParas.ExportLimit.ExportTrim :
for (int index = 0; index < tempData.Rows.Count; index++)
{
tempData.Rows[index][i] = tempData.Rows[index][i].ToString().Trim();
}
break;
case (int)cGlobalParas.ExportLimit.ExportRegexReplace :
for (int index = 0; index < tempData.Rows.Count; index++)
{
//string oStr=cTool.RegexReplaceTrans( this.CutFlag[i].ExportExpression .Substring (1,this.CutFlag[i].ExportExpression.IndexOf (",")-2));
//string nStr = this.CutFlag[i].ExportExpression.Substring(this.CutFlag[i].ExportExpression.IndexOf(",") + 2, this.CutFlag[i].ExportExpression.Length - this.CutFlag[i].ExportExpression.IndexOf(",") - 3);
//tempData.Rows[index][i] = Regex.Replace(tempData.Rows[index][i].ToString(), oStr, nStr, RegexOptions.IgnoreCase | RegexOptions.Multiline);
string oStr = this.CutFlag[i].ExportExpression.Substring(1, this.CutFlag[i].ExportExpression.IndexOf(",") - 2);
string nStr = this.CutFlag[i].ExportExpression.Substring(this.CutFlag[i].ExportExpression.IndexOf(",") + 2, this.CutFlag[i].ExportExpression.Length - this.CutFlag[i].ExportExpression.IndexOf(",") - 3);
tempData.Rows[index][i] = Regex.Replace(tempData.Rows[index][i].ToString(), oStr, nStr, RegexOptions.IgnoreCase | RegexOptions.Multiline);
}
break;
default :
break;
}
}
#endregion
#region ��Բɼ���Ҫ�����ļ����ֶν����ļ����ش���
//�ж��Ƿ�����������ļ�����������У���ʼ���أ���Ϊ�˹���������������ͼƬʹ��
//������ר�õ����ع��ߣ����Զ����ش����û�е��������̴߳���
try
{
if (IsDownloadFile == true)
{
if (sPath == "")
{
sPath = Program.getPrjPath() + "data\\tem_file";
}
if (!Directory.Exists(sPath))
{
Directory.CreateDirectory(sPath);
}
string FileUrl = "";
string DownloadFileName = "";
for (i = 0; i < rows; i++)
{
for (j = 0; j < this.CutFlag.Count; j++)
{
if (this.CutFlag[j].DataType != (int)cGlobalParas.GDataType.Txt)
{
FileUrl = tempData.Rows[i][j].ToString();
//��ʼ��ȡ�����ļ�����
Regex s = new Regex(@"(?<=/)[^/]*", RegexOptions.IgnoreCase | RegexOptions.Multiline);
MatchCollection urlstr = s.Matches(FileUrl);
if (urlstr.Count == 0)
DownloadFileName = FileUrl;
else
DownloadFileName = urlstr[urlstr.Count - 1].ToString();
DownloadFileName = sPath + "\\" + DownloadFileName;
if (string.Compare ( FileUrl.Substring(0, 4) , "http",true )==0)
{
DownloadFile(FileUrl, DownloadFileName);
}
else
{
if (FileUrl.Substring(0, 1) == "/")
{
Url = Url.Substring(7, Url.Length - 7);
Url = FileUrl.Substring(0, Url.IndexOf("/"));
Url = "http://" + Url;
FileUrl = Url + FileUrl;
}
else if (FileUrl.IndexOf("/") <= 0)
{
Url = Url.Substring(0, Url.LastIndexOf("/") + 1);
FileUrl = Url + FileUrl;
}
else
{
Url = Url.Substring(0, Url.LastIndexOf("/") + 1);
FileUrl = Url + FileUrl;
}
DownloadFile(FileUrl, DownloadFileName);
}
}
}
}
}
}
catch (System.Exception ex)
{
throw ex;
}
#endregion
return tempData;
}
/// <summary>
/// ��ȡָ����ַԴ��
/// </summary>
/// <param name="url">��ַ</param>
/// <param name="webCode">��ҳ����</param>
/// <param name="cookie">��ҳcookie</param>
/// <param name="startPos">��ȡ��ҳԴ�����ʼλ��</param>
/// <param name="endPos">��ȡ��ҳԴ�����ֹλ��</param>
/// <param name="IsCutnr">�Ƿ��ȡ�س����з���Ĭ��Ϊtrue����ȡ</param>
/// <returns></returns>
public string GetHtml(string url, cGlobalParas.WebCode webCode, string cookie, string startPos, string endPos,bool IsCutnr,bool IsAjax)
{
//�ж���ҳ����
Encoding wCode;
string PostPara = "";
CookieContainer CookieCon = new CookieContainer();
HttpWebRequest wReq ;
if (Regex.IsMatch(url, @"<POST>.*</POST>", RegexOptions.IgnoreCase))
{
wReq = (HttpWebRequest)WebRequest.Create(@url.Substring (0,url.IndexOf ("<POST>")));
}
else
{
wReq = (HttpWebRequest)WebRequest.Create(@url );
}
wReq.UserAgent = "Mozilla/4.0 (compatible; MSIE 8.0; Windows NT 5.0; .NET CLR 1.1.4322; .NET CLR 2.0.50215;)";
//wReq.Headers.Add("Accept-Language", "zh-cn,en-us;q=0.5");
//wReq.Headers.Add("Accept-Encoding", "gzip, deflate");
//wReq.SendChunked = true;
//wReq.TransferEncoding = "utf-8";
Match a = Regex.Match(url, @"(http://).[^/]*[?=/]", RegexOptions.IgnoreCase);
string url1 = a.Groups[0].Value.ToString();
wReq.Referer = url1;
//�ж��Ƿ���cookie
if (cookie != "")
{
CookieCollection cl = new CookieCollection();
//foreach (string sc in cookie.Split(';'))
//{
// string ss = sc.Trim();
// cl.Add(new Cookie(ss.Split('=')[0].Trim(), ss.Split('=')[1].Trim(), "/"));
//}
foreach (string sc in cookie.Split(';'))
{
string ss = sc.Trim();
if (ss.IndexOf("&") > 0)
{
foreach (string s1 in ss.Split('&'))
{
string s2 = s1.Trim();
string s4 = s2.Substring(s2.IndexOf("=")+1, s2.Length - s2.IndexOf("=")-1);
cl.Add(new Cookie(s2.Split('=')[0].Trim(), s4, "/"));
}
}
else
{
string s3 = sc.Trim();
cl.Add(new Cookie(s3.Split('=')[0].Trim(), s3.Split('=')[1].Trim(), "/"));
}
}
CookieCon.Add(new Uri(url), cl);
wReq.CookieContainer = CookieCon;
}
//�ж��Ƿ���POST����
if (Regex.IsMatch(url, @"(?<=<POST>)[\S\s]*(?=</POST>)", RegexOptions.IgnoreCase))
{
Match s = Regex.Match(url, @"(?<=<POST>).*(?=</POST>)", RegexOptions.IgnoreCase);
PostPara = s.Groups[0].Value.ToString();
byte[] pPara = Encoding.ASCII.GetBytes(PostPara);
wReq.ContentType = "application/x-www-form-urlencoded";
wReq.ContentLength = pPara.Length;
wReq.Method = "POST";
System.IO.Stream reqStream = wReq.GetRequestStream();
reqStream.Write(pPara, 0, pPara.Length);
reqStream.Close();
}
else
{
wReq.Method = "GET";
}
//����ҳ�泬ʱʱ��Ϊ12��
wReq.Timeout = 12000;
HttpWebResponse wResp = (HttpWebResponse)wReq.GetResponse();
System.IO.Stream respStream = wResp.GetResponseStream();
string strWebData = "";
switch (webCode)
{
case cGlobalParas.WebCode.auto:
try
{
wCode = Encoding.Default;
string cType = wResp.ContentType.ToLower();
Match charSetMatch = Regex.Match(cType, "(?<=charset=)([^<]*)*", RegexOptions.IgnoreCase | RegexOptions.Multiline);
string webCharSet = charSetMatch.ToString();
wCode = System.Text.Encoding.GetEncoding(webCharSet);
}
catch
{
wCode = Encoding.Default;
}
break;
case cGlobalParas.WebCode.gb2312:
wCode = Encoding.GetEncoding("gb2312");
break;
case cGlobalParas.WebCode.gbk:
wCode = Encoding.GetEncoding("gbk");
break;
case cGlobalParas.WebCode.utf8:
wCode = Encoding.UTF8;
break;
default:
wCode = Encoding.UTF8;
break;
}
if (wResp.ContentEncoding == "gzip")
{
GZipStream myGZip = new GZipStream(respStream, CompressionMode.Decompress);
System.IO.StreamReader reader;
reader = new System.IO.StreamReader(myGZip, wCode);
strWebData = reader.ReadToEnd();
reader.Close();
reader.Dispose();
}
else
{
System.IO.StreamReader reader;
reader = new System.IO.StreamReader(respStream, wCode);
strWebData = reader.ReadToEnd();
reader.Close();
reader.Dispose();
}
//ȥ���س����з���
if (IsCutnr == true)
{
strWebData = Regex.Replace(strWebData, "([\\r\\n])[\\s]+", "", RegexOptions.IgnoreCase | RegexOptions.Multiline);
strWebData = Regex.Replace(strWebData, "\\n", "", RegexOptions.IgnoreCase | RegexOptions.Multiline);
strWebData.Replace("\\r\\n", "");
}
//��ȡ��ҳ��ı����ʽ,����Դ�����һ���ж�,�����û��Ƿ�ָ������ҳ����
//Match charSetMatch = Regex.Match(strWebData, "<meta([^<]*)charset=([^<]*)\"", RegexOptions.IgnoreCase | RegexOptions.Multiline);
//string webCharSet = charSetMatch.Groups[2].Value;
//string charSet = webCharSet;
//if (charSet != null && charSet != "" && Encoding.GetEncoding(charSet) != wCode)
//{
// byte[] myDataBuffer;
// myDataBuffer = System.Text.Encoding.GetEncoding(charSet).GetBytes(strWebData);
// strWebData = Encoding.GetEncoding(charSet).GetString(myDataBuffer);
//}
//���ս�ȡ��ҳ����ʼ��־����ֹ��־���н�ȡ
//�����ʼ����ֹ��ȡ��ʶ��һ��Ϊ�գ����н�ȡ
if (startPos != "" && endPos != "")
{
string Splitstr = "(" + startPos + ").*?(" + endPos + ")";
Match aa = Regex.Match(strWebData, Splitstr);
strWebData = aa.Groups[0].ToString();
}
if (IsAjax == true)
{
strWebData = System.Web.HttpUtility.UrlDecode(strWebData, Encoding.UTF8);
}
this.m_WebpageSource = strWebData;
return strWebData;
}
public Int64 InsertTaskComplete(Int64 TaskID,cGlobalParas.GatherResult tSate)
{
///�����жϴ������ִ�е�Ŀ¼�Ƿ����
///��Ŀ¼�ǹ̶�Ŀ¼�������ϵͳ\\data
string cPath = Program.getPrjPath() + "data";
if (!System.IO.Directory.Exists(cPath))
{
System.IO.Directory.CreateDirectory(cPath);
}
///�Ƚ��������ժҪ��Ϣ���ص�index.xml�ļ���
Task.cTaskRun t = new Task.cTaskRun();
t.LoadSingleTask(TaskID);
//��ʼ����xml�ڵ�����
LoadTaskData();
//int MaxID=GetCount () + 1;
string txml="";
txml = "<TaskID>" + t.GetTaskID (0) + "</TaskID>";
txml += "<TaskName>" + t.GetTaskName(0) + "</TaskName>";
txml += "<TaskState>" + tSate + "</TaskState>";
txml += "<TaskType>" + (int)t.GetTaskType(0) + "</TaskType>";
txml += "<RunType>" + (int)t.GetTaskRunType(0) + "</RunType>";
txml += "<ExportFile>" + t.GetExportFile(0) + "</ExportFile>";
txml += "<tempFile>" + t.GetTempFile(0) + "</tempFile>";
txml += "<UrlCount>" + t.GetUrlCount(0) + "</UrlCount>";
txml += "<GatheredUrlCount>" + t.GetGatheredUrlCount(0) + "</GatheredUrlCount>";
txml += "<IsLogin>" + t.GetIsLogin(0) + "</IsLogin>";
txml += "<PublishType>" + (int)t.GetPublishType(0) + "</PublishType>";
xmlConfig.InsertElement("Tasks", "Task", txml);
xmlConfig.Save ();
return TaskID;
}
private string getCreateTablesql(cGlobalParas.DatabaseType dType ,string Encoding)
{
string strsql = "";
strsql = "create table " + this.m_pTaskData.DataTableName + "(";
for (int i=0;i<m_pTaskData.PublishData.Columns.Count ;i++)
{
switch (dType)
{
case cGlobalParas.DatabaseType.Access:
strsql += m_pTaskData.PublishData.Columns[i].ColumnName + " " + "text" + ",";
break;
case cGlobalParas.DatabaseType.MSSqlServer:
strsql += m_pTaskData.PublishData.Columns[i].ColumnName + " " + "text" + ",";
break;
case cGlobalParas.DatabaseType.MySql:
strsql += m_pTaskData.PublishData.Columns[i].ColumnName + " " + "text" + ",";
break;
default:
strsql += m_pTaskData.PublishData.Columns[i].ColumnName + " " + "text" + ",";
break;
}
}
strsql = strsql.Substring(0, strsql.Length - 1);
strsql += ")";
//�����mysql���ݿ⣬��Ҫ�������Ӵ����ַ����������ݱ�Ľ���
if (dType == cGlobalParas.DatabaseType.MySql)
{
if (Encoding == "" || Encoding == null)
Encoding = "utf8";
strsql += " CHARACTER SET " + Encoding + " ";
}
return strsql;
}
public cCommandEventArgs(cGlobalParas.MessageType MessType)
{
base.MessType = MessType;
}
///����������ҳ
///�ж��Ƿ�Ϊ���һ��������������Ҫע��һ�����⣬��Ϊ�п���
///�洢�ļ����ǰ���˳����еģ����ԣ�Ҫ���ݴ���ļ���Level����
///�жϣ��������ִ�������ҳ�Ľ��������ǰ���˳��ģ������
///������
public List<string> PUrlRule(List<string> pUrl,int Level, List<cNavigRule> nRules,cGlobalParas.WebCode webCode, string cookie)
{
List<string> tmpUrls;
List<string> Urls =new List<string> ();
if (nRules.Count == 0)
{
Urls.Add(pUrl[0].ToString());
return Urls;
}
string UrlRule="";
int i;
//����Level�õ���Ҫ��������ĵ�������
for (i = 0; i < nRules.Count; i++)
{
if (Level ==nRules[i].Level )
{
UrlRule =nRules [i].NavigRule ;
break;
}
}
for (i = 0; i < pUrl.Count; i++)
{
tmpUrls = new List<string>();
tmpUrls = GetUrlsByRule(pUrl[i].ToString(), UrlRule,webCode,cookie);
if (tmpUrls != null)
{
Urls.AddRange(tmpUrls);
}
}
//�ж��Ƿ�Ϊ����ĵ�����������أ�����������������
if (Level == nRules.Count)
{
return Urls;
}
else
{
List<string> rUrls= PUrlRule(Urls, Level + 1, nRules,webCode,cookie);
return rUrls;
}
}
private void ModifyPlanState(Int64 PlanID, cGlobalParas.PlanState pState)
{
cXmlIO xmlConfig = new cXmlIO(Program.getPrjPath() + "tasks\\plan\\plan.xml");
xmlConfig.EditNodeValue("Plans", "ID", PlanID.ToString () ,"PlanState",((int)pState).ToString ());
xmlConfig.Save ();
xmlConfig =null;
}
public void InsertLog(cGlobalParas.LogType lType,string PlanID,string PlanName, cGlobalParas.RunTaskType rType, string FileName, string Para)
{
string strXml = "<LogType>" + ((int)lType).ToString () + "</LogType>" +
"<PlanID>" + PlanID + "</PlanID>" +
"<PlanName>" + PlanName + "</PlanName>" +
"<FileName>" + FileName + "</FileName>" +
"<FilePara>" + Para + "</FilePara>" +
"<TaskType>" + ((int)rType).ToString () + "</TaskType>" +
"<RunTime>" + DateTime.Now.ToString() + "</RunTime>";
m_PlanFile.InsertElement("Logs", "Log", strXml);
m_PlanFile.Save();
}
