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>
/// Create a new Answer object.
/// </summary>
/// <param name="id">Initial value of the Id property.</param>
/// <param name="voteup">Initial value of the Voteup property.</param>
/// <param name="collectCount">Initial value of the CollectCount property.</param>
/// <param name="commentCount">Initial value of the CommentCount property.</param>
/// <param name="questionId">Initial value of the QuestionId property.</param>
public static Answer CreateAnswer(global::System.Decimal id, global::System.Decimal voteup, global::System.Decimal collectCount, global::System.Decimal commentCount, global::System.Decimal questionId)
{
Answer answer = new Answer();
answer.Id = id;
answer.Voteup = voteup;
answer.CollectCount = collectCount;
answer.CommentCount = commentCount;
answer.QuestionId = questionId;
return answer;
}
/// <summary>
/// Deprecated Method for adding a new object to the AnswerSet EntitySet. Consider using the .Add method of the associated ObjectSet<T> property instead.
/// </summary>
public void AddToAnswerSet(Answer answer)
{
base.AddObject("AnswerSet", answer);
}
/// <summary>
/// Deprecated Method for adding a new object to the AnswerSet EntitySet. Consider using the .Add method of the associated ObjectSet<T> property instead.
/// </summary>
public void AddToAnswerSet(Answer answer)
{
base.AddObject("AnswerSet", answer);
}
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!");
}