private async void SelectHangzhou()
{
/**************************************************************
*
* 目前只能爬取到第一页的数据信息
*
* Only crawl the first page data
*
**************************************************************/
textBox1.Text = "http://hotels.ctrip.com/hotel/hangzhou17";
WebSpider Spider = new WebSpider();
Spider.OnStartEvent += (s, e) =>
{
Invoke(new Action(() =>
{
textBox3.AppendText(e.Uri.ToString() + " 开始" + Environment.NewLine);
}));
};
Spider.OnExceptionEvent += (s, e) =>
{
Invoke(new Action(() =>
{
textBox3.AppendText(e.Uri.ToString() + " 异常：" + e.Exception.Message + Environment.NewLine);
}));
};
Spider.OnCompletedEvent += (s, e) =>
{
//<span class="hotel_num">1</span>杭州马可波罗假日酒店</a>
string pattern = "<span class=\"hotel_num\">[0-9]+</span>[^<]+";
string addr = "<p class=\"searchresult_htladdress\">[\\S]+";
MatchCollection mc = Regex.Matches(e.PageSource, pattern, RegexOptions.IgnoreCase);
MatchCollection mcaddr = Regex.Matches(e.PageSource, addr, RegexOptions.IgnoreCase);
StringBuilder sb = new StringBuilder();
int index = 0;
foreach (Match m in mc)
{
sb.Append(m.Value.Substring(m.Value.LastIndexOf('>') + 1) +
" 地址[" + mcaddr[index].Value.Substring(mcaddr[index].Value.LastIndexOf('>') + 1) + "]" + Environment.NewLine);
//sb.Append(Regex.Match(m.Value, "/hotel/.+[^\\\"]").Value.Substring(1) + Environment.NewLine);
//sb.Append(m.Value + Environment.NewLine);
index++;
}
Invoke(new Action(() =>
{
label1.Text = "共计数据：" + mc.Count;
textBox2.Text = sb.ToString();
textBox3.AppendText(e.Uri.ToString() + " 耗时：" + e.MilliSeconds + Environment.NewLine);
}));
};
await Spider.Start(new Uri(textBox1.Text));
}
private async void SelectCities()
{
textBox1.Text = "http://hotels.ctrip.com/citylist";
WebSpider Spider = new WebSpider();
Spider.OnStartEvent += (s, e) =>
{
Invoke(new Action(() =>
{
textBox3.AppendText(e.Uri.ToString() + " 开始" + Environment.NewLine);
}));
};
Spider.OnExceptionEvent += (s, e) =>
{
Invoke(new Action(() =>
{
textBox3.AppendText(e.Uri.ToString() + " 异常：" + e.Exception.Message + Environment.NewLine);
}));
};
Spider.OnCompletedEvent += (s, e) =>
{
string pattern = "<a href=\"/hotel/[a-z0-9]+\" title=\"[^\"]+\">[^<]+</a>";
MatchCollection mc = Regex.Matches(e.PageSource, pattern, RegexOptions.IgnoreCase);
StringBuilder sb = new StringBuilder();
foreach (Match m in mc)
{
sb.Append(Regex.Match(m.Value, ">[^<]+").Value.Substring(1) + Environment.NewLine);
//sb.Append(Regex.Match(m.Value, "/hotel/.+[^\\\"]").Value.Substring(1) + Environment.NewLine);
//sb.Append(m.Value + Environment.NewLine);
}
Invoke(new Action(() =>
{
label1.Text = "共计数据：" + mc.Count;
textBox2.Text = sb.ToString();
textBox3.AppendText(e.Uri.ToString() + " 耗时：" + e.MilliSeconds + Environment.NewLine);
}));
};
await Spider.Start(new Uri(textBox1.Text));
}