/// <summary>
/// ��ȡһ��ҳ������ݲ�����
/// </summary>
/// <param name="pageUri">ҳ���ַ</param>
/// <param name="number">ά��һ������,�ж������Ƿ����</param>
/// <returns></returns>
private DataPack GetPageData(string pageUri, ref int number, DataPackFunc func)
{
DataPack dp;
int bufferLength = 10;
byte[] buffer = new byte[bufferLength]; //���ص����ݻ�����
StringBuilder sb = new StringBuilder(); //���췵�صĽ��
Match match; //����ƥ��
//ҳ���ַ��ҳ���ַ����ƥ�䣡
if (!Regex.IsMatch(pageUri, this.FormatedPageUriRule))
{
++number;
state.FailCount++;
return(null);
//throw new ArgumentException("ҳ���ַ��ҳ���ַ����ƥ�䣡", pageUri);
}
//���ҳ���ַΪ���·�������������
if (absoluteUriRegex.IsMatch(pageUri))
{
pageUri = GetBasePath(pageUri) + pageUri;
}
//����ҳ���HTML
string html = String.Empty;
try
{
HttpWebRequest req = (HttpWebRequest)WebRequest.Create(pageUri);
req.Timeout = this.RequestTimeOut;
Stream stream = req.GetResponse().GetResponseStream();
html = sb.ToString();
using (StreamReader sr = new StreamReader(stream, this.Encode))
{
html = sr.ReadToEnd();
}
}
catch (Exception exc)
{
state.FailCount++;
return(null);
}
//������ص�����
#if DEBUG
Console.WriteLine("\r\n------------------------------\r\n�õ�ƥ����б�����Ϊ:{0}", html);
#endif
dp = new DataPack(Rules, pageUri);
foreach (string propertyName in this.Rules)
{
match = Regex.Match(html, this.Rules[propertyName]);
if (match != null)
{
dp[propertyName] = match.Groups[1].Value;
}
}
#if DEBUG
Console.WriteLine("\r\n-------------------------\r\n");
foreach (KeyValuePair <string, string> pair in dp)
{
Console.WriteLine("{0}->{1}\r\n", pair.Key, pair.Value);
}
#endif
//���¼���
++number;
#if DEBUG
Console.WriteLine("flish");
#endif
//ִ�л�ִ����
if (func != null)
{
func(dp);
}
//���һ���ɹ��ļ���
state.SuccessCount++;
return(dp);
}