private void button2_Click(object sender, EventArgs e)
{
var oldIds = from t in db.QuestionSet select t.Id;
List<decimal> newIds = new List<decimal>();
if (txtPageCount.Text.Length>0)
{
var pageCount = int.Parse(txtPageCount.Text);
for (int i = pageCount; i >= 0; i--)
{
var url = string.Format("http://www.zhihu.com/topic/19776749/questions?page={0}",i);
HtmlAgilityPack.HtmlDocument doc = htmlWeb.Load(url);
var nodes = Utility.FindNodesByClass(doc, "question-item");
foreach (var node in nodes)
{
Question q = new Question();
q.Id = Utility.GetId(node);
//q.AnswerCount = GetAnswerCount(node); 拿不到，html里面没有
q.CreateTime = Utility.GetTime(node);
//q.FollowerCount = GetFollower(node);
q.Title = Utility.GetTitle(node);
q.Topic = Utility.GetTopic(node);
//q.ViewCount = GetViewCount(node);
bool contain = newIds.Contains(q.Id) || oldIds.Contains(q.Id);
if (!contain)
{
db.QuestionSet.AddObject(q);
newIds.Add(q.Id);
Application.DoEvents();
}
txtTitle.Text = q.Title;
}
db.SaveChanges();
}
}
}
/// <summary>
/// Create a new Question object.
/// </summary>
/// <param name="id">Initial value of the Id property.</param>
/// <param name="title">Initial value of the Title property.</param>
/// <param name="answerCount">Initial value of the AnswerCount property.</param>
/// <param name="topic">Initial value of the Topic property.</param>
/// <param name="followerCount">Initial value of the FollowerCount property.</param>
/// <param name="viewCount">Initial value of the ViewCount property.</param>
/// <param name="questionId">Initial value of the QuestionId property.</param>
public static Question CreateQuestion(global::System.Decimal id, global::System.String title, global::System.Decimal answerCount, global::System.String topic, global::System.Decimal followerCount, global::System.Decimal viewCount, global::System.Decimal questionId)
{
Question question = new Question();
question.Id = id;
question.Title = title;
question.AnswerCount = answerCount;
question.Topic = topic;
question.FollowerCount = followerCount;
question.ViewCount = viewCount;
question.QuestionId = questionId;
return question;
}
private void button2_Click(object sender, EventArgs e)
{
var oldIds = from t in db.QuestionSet select t.Id;
var oldAnswerIds = from t in db.AnswerSet select t.Id;
List<decimal> newIds = new List<decimal>();
List<decimal> newAnswerIds = new List<decimal>();
//WebBrowser browser = new WebBrowser();
browser.DocumentCompleted += new WebBrowserDocumentCompletedEventHandler(browser_DocumentCompleted);
browserDetail.DocumentCompleted += new WebBrowserDocumentCompletedEventHandler(browserDetail_DocumentCompleted);
var pageCount = (int)numericUpDown1.Value;
for (int i = pageCount; i > 0; i--)
{
var url = string.Format("http://www.zhihu.com/topic/19776749/top-answers?page={0}", i);
//NetworkCredential c = new NetworkCredential("*****@*****.**","password");
//HtmlAgilityPack.HtmlDocument doc = htmlWeb.Load(url,"get",null,c);
browser.Navigate(url);
while (!ready)
{
Application.DoEvents();
}
//System.Threading.Thread.Sleep(1000);
ready = false;
HtmlAgilityPack.HtmlDocument doc = new HtmlAgilityPack.HtmlDocument();
doc.LoadHtml(browser.Document.Body.OuterHtml);
var nodes = Utility.FindNodesByClass(doc, "feed-main");
foreach (var node in nodes)
{
//先找问题
Question q = new Question();
q.Id = Utility.GetTopId(node);
//q.AnswerCount = GetAnswerCount(node); 拿不到，html里面没有
//q.CreateTime = Utility.GetTopTime(node);
//q.FollowerCount = GetFollower(node);
q.Title = Utility.GetTopTitle(node);
q.Topic = Utility.GetTopic(node);
//q.ViewCount = GetViewCount(node);
bool contain = newIds.Contains(q.Id) || oldIds.Contains(q.Id);
if (!contain)
{
db.QuestionSet.AddObject(q);
newIds.Add(q.Id);
// Application.DoEvents();
}
Application.DoEvents();
txtTitle.Text = string.Format("P:{0},T:{1}", i, q.Title);
//再找答案
var answerId = Utility.GetAnswerId(node);
Answer a;
bool containAnswer = newAnswerIds.Contains(answerId) || oldAnswerIds.Contains(answerId);
if (!containAnswer)
{
a = new Answer();
a.Id = answerId;
db.AnswerSet.AddObject(a);
newIds.Add(a.Id);
//Application.DoEvents();
}
else
{
a = db.AnswerSet.First(t => t.Id == answerId);
}
a.Voteup = Utility.GetVoteCount(node);
a.Author = Utility.GetAnswerAuthor(node);
a.AuthorId = Utility.GetAnswerAuthorId(node);
var answerUrl = string.Format("http://www.zhihu.com/question/{0}/answer/{1}", q.Id, a.Id);
//var answerDoc = htmlWeb.Load(answerUrl);
browserDetail.Navigate(answerUrl);
while (!readyDetail)
{
Application.DoEvents();
}
readyDetail = false;
HtmlAgilityPack.HtmlDocument answerDoc = new HtmlAgilityPack.HtmlDocument();
answerDoc.LoadHtml(browserDetail.Document.Body.OuterHtml);
var rootNode = answerDoc.DocumentNode;
//var author = Utility.GetAnswerAuthor2(rootNode);
//if ( string.IsNullOrEmpty(a.Author) && !string.IsNullOrEmpty(author))
//{
// a.Author = author;
//}
a.CollectCount = Utility.GetCollectCount(rootNode);
a.CreateTime = Utility.GetAnswerTime(rootNode);
a.CommentCount = Utility.GetCommentCount(rootNode);
a.QuestionId = q.Id;
}
db.SaveChanges();
}
MessageBox.Show("Over!");
}
/// <summary>
/// Deprecated Method for adding a new object to the QuestionSet EntitySet. Consider using the .Add method of the associated ObjectSet<T> property instead.
/// </summary>
public void AddToQuestionSet(Question question)
{
base.AddObject("QuestionSet", question);
}