private void btnAdd_Click(object sender, EventArgs e)
{
int index = 0;
string key = "";
if (data_conf.areas == null)
{
data_conf.areas = new Hashtable();
}
data_area_t area = new data_area_t();
area.enable = true;
area.from = "";
area.to = "";
area.regex = "";
area.mode = AREA_MODE.REGEX;
area.info_path = "";
index = data_conf.areas.Count;
do
{
index++;
key = "区域" + index.ToString();
}while(data_conf.areas.ContainsKey(key));
area.id = key;
data_conf.areas.Add(area.id, area);
chklArea.Items.Add(area.id, area.enable);
}
public string getInfo(string content, data_area_t area)
{
int i = 0;
string[] buf=new string[16];
string regbuf;
string ret = "";
MatchCollection mc;
Regex regex;
string path = area.info_path;
string regex_str = area.regex;
// parse path
buf = path.Split(new char[1] { '/' }, StringSplitOptions.RemoveEmptyEntries);
for (i = 0; i < buf.Length; i++)
{
regbuf = string.Format("(?<=<({0}).*>).*(?=<\\/\\1>)", buf[i]);
regex = new Regex(regbuf, RegexOptions.IgnoreCase | RegexOptions.Multiline);
Match m = regex.Match(content);
content = m.Value;
}
// parse regex
regex = new Regex(regex_str, RegexOptions.IgnoreCase | RegexOptions.Multiline);
mc = regex.Matches(content);
foreach (System.Text.RegularExpressions.Match item in mc)
{
//Console.WriteLine(item.Value);
// parse text
if (AREA_MODE.TEXT == area.mode)
{
ret += getText(item.Value);
}
else
{
ret += item.Value;
}
ret += "\r\n";
}
return ret;
}
public string getIterInfo(string content, data_area_t area)
{
string ret = "";
string buf = "";
int index = 0;
int len = 0;
int i = 0;
do
{
buf = getSourceByTag(content, area.iter_tag,ref index, out len);
if (len > 0)
{
i++;
ret += i.ToString() + "\r\n";
ret += getText(buf);
ret += "\r\n";
}
}while(index < content.Length);
//Console.WriteLine(ret);
return ret;
}
private void Form2_Load(object sender, EventArgs e)
{
//CHtmlTag debug = new CHtmlTag(); int debug1 = 0; int debug2 = 0;
//string debugstr=debug.getSourceByTag("<li><li>test</li></li>", "li", ref debug1, out debug2);
if (DateTime.Now.Year > 2011) {
MessageBox.Show("�������Ѿ�����");
this.Close();
}
data_area_t area;
work1 = new CWdManager();
work1.load_game();
/*http://s.taobao.com/search?q=iphone&keyword=&commend=all&ssid=s5-e&search_type=item&atype=&tracelog=&sourceId=tb.index
http://s.taobao.com/search?q=iphone&commend=all&ssid=s5-e&filterFineness=2&s=40#J_FilterTabBar
http://s.taobao.com/search?q=iphone&commend=all&filterFineness=2&ssid=s5-e&atype=b&s=80#J_FilterTabBar
*
*
��һҳ http://s.taobao.com/search?q=iphone&commend=all&atype=b&filterFineness=2&ssid=s5-e&s=120#J_FilterTabBar
s=80��ҳ��*/
data_conf = new data_conf_t();
data_conf.id = "������Ϣ��ȫ��Ʒ��֤��֤����";
data_conf.cols = 8;
data_conf.url = "http://www.isccc.gov.cn/zsgg/08/398933.shtml";
//data_conf.url = "http://127.0.0.1/list.htm";
data_conf.encoding = "UTF-8";
data_conf.areas = new Hashtable();
// ����1:title
area = new data_area_t("����");
area.enable = true;
area.from = "<head>";
area.to = "</head>";
area.info_path = "";
area.regex = "(?<=<title>)[^<]*(?=</title>)";
area.mode = AREA_MODE.REGEX;
data_conf.areas.Add(area.id, area);
// ����2:����
area = new data_area_t("��������");
//area.from = "���ߣ��й���Ϣ��ȫ��֤����";
//area.to = "��֤��ҵչʾ";
area.from = "<TABLE style=\"BORDER-RIGHT:";
area.to = "</table>";
area.regex = "(?<=<(tr).*>).*(?=<\\/\\1>)";
area.mode = AREA_MODE.ITERATOR;
area.info_path = "";
area.iter_tag = "tr";
data_conf.areas.Add(area.id, area);
work1.data_confs.Add(data_conf.id, data_conf);
data_conf = new data_conf_t("weibo.com");
data_conf.cols = 0;
data_conf.url = "http://weibo.com";
data_conf.encoding = "UTF-8";
data_conf.areas = new Hashtable();
work1.data_confs.Add(data_conf.id, data_conf);
data_conf = new data_conf_t("weibo_��㿴��.com");
data_conf.cols = 0;
data_conf.url = "http://127.0.0.1/weibo_suibiankankan.htm";
data_conf.encoding = "UTF-8";
data_conf.areas = new Hashtable();
// ����1:title
area = new data_area_t("title");
area.enable = true;
area.from = "<head>";
area.to = "</head>";
area.info_path = "";
area.regex = "(?<=<title>)[^<]*(?=</title>)";
area.mode = AREA_MODE.REGEX;
data_conf.areas.Add(area.id, area);
// ����2
area = new data_area_t("��㿴���˺��б�");
area.enable = true;
area.from = "<ul class=\"MIB_feed\">";
area.to = "<div class=\"feed_bt\">";
area.info_path = "li";
area.regex = "(?<=<a href=\")\\S+(?=\" title=\")";
area.mode = AREA_MODE.REGEX;
data_conf.areas.Add(area.id, area);
work1.data_confs.Add(data_conf.id, data_conf);
data_conf = new data_conf_t("taobao.com(iphone)");
data_conf.cols = 0;
data_conf.url = "http://127.0.0.1/iphone.htm";
data_conf.url = "http://s.taobao.com/search?q=iphone&keyword=&commend=all&ssid=s5-e&search_type=item&atype=&tracelog=&sourceId=tb.index";
data_conf.encoding = "GB2312";
data_conf.areas = new Hashtable();
// ����1:title
area = new data_area_t("title");
area.enable = true;
area.from = "<head>";
area.to = "</head>";
area.info_path = "";
area.regex = "(?<=<title>)[^<]*(?=</title>)";
area.mode = AREA_MODE.REGEX;
data_conf.areas.Add(area.id, area);
// ����2
area = new data_area_t("��Ʒ�б�");
area.enable = true;
area.from = "<!-- mall -->";
area.to = "</form>";
area.info_path = "";
area.regex = ".*";
area.iter_tag = "li";
area.mode = AREA_MODE.ITERATOR;
data_conf.areas.Add(area.id, area);
work1.data_confs.Add(data_conf.id, data_conf);
comboBox1.Items.Clear();
foreach (DictionaryEntry item in work1.data_confs)
{
data_conf_t conf = (data_conf_t)item.Value;
comboBox1.Items.Add(conf.id);
}
comboBox1.SelectedIndex = 0;
}