static WebPage GetWebPage(string url)
{
string html = GetWebHTML(url);
if (html == null) return null;
int title_start_pos = html.IndexOf("<title>") + "<title>".Length;
int title_end_pos = html.IndexOf("</title>");
int pub_time_start_pos = html.IndexOf(">发布时间:") + ">发布时间:".Length;
//int content_start_pos = html.IndexOf("<font size='3'>") + "<td valign='top'><div><font size='3'>".Length;
//int content_end_pos = html.IndexOf("</font></div></td>");
WebPage page = new WebPage();
page.url = url;
page.html = html;
page.title = html.Substring(title_start_pos, title_end_pos - title_start_pos);
try
{
page.published_time = DateTime.Parse(html.Substring(pub_time_start_pos, 10));
}
catch
{
//Console.WriteLine("{0} {1}", url, html.Substring(pub_time_start_pos, 10));
page.published_time = DateTime.Now;
}
page.content = "";
return page;
}
public void InsertWebPage(WebPage page)
{
MongoCollection<WebPage> collection = mongo_database.GetCollection<WebPage>("webpages");
SafeModeResult smr = collection.Insert<WebPage>(page);
}