public bool IsSameUser(UserRank otherUser)
{
if (this.status == (int)userStatus.notUsed || otherUser.status == (int)userStatus.notUsed)
{
return(false);
}
foreach (UserInfo user1 in userInfos)
{
foreach (UserInfo user2 in otherUser.userInfos)
{
if (user1.status == user2.status)
{
if (user1.status == (int)userStatus.fixedNick)
{
if (user1.IdorIp.Equals(user2.IdorIp))
{
Console.WriteLine(user1.IdorIp + " 같다 " + user2.IdorIp);
return(true);
}
}
else if (user1.status == (int)userStatus.fluidNick)
{
if (user1.Nick.Equals(user2.Nick) && user1.Nick != "ㅇㅇ")
{
return(true);
}
}
}
}
}
return(false);
}
public void Crawler()
{
int previousPageGallNum = 1000000000;
Console.WriteLine(initDate.ToString() + endDate.ToString());
string url = gallUrl + "&page=";
var client = new WebClient();
client.Encoding = System.Text.Encoding.UTF8;
//Dictionary value => count, replyNum, gallCount, gallRecommend
Dictionary <UserInfo, int[]> userDic = new Dictionary <UserInfo, int[]>();
int currentPage = this.initPage;
while (true)
{
string text;
try
{
text = client.DownloadString(url + currentPage.ToString());
if (string.IsNullOrEmpty(text))
{
continue;
}
}
catch
{
continue;
}
hap.HtmlDocument textHap = new hap.HtmlDocument();
textHap.LoadHtml(text);
hap.HtmlNodeCollection nicks = textHap.DocumentNode.SelectNodes("//tr[@class='ub-content us-post']");
//Console.WriteLine(nicks.Count);
//Console.WriteLine("==================" + currentPage.ToString() + "==================");
try
{
foreach (hap.HtmlNode nick in nicks)
{
int gallNum, replyNum, gallCount, gallRecommend;
DateTime gallDate; string subject;
gallNum = GetOnlyInt(nick.SelectSingleNode("./td[@class='gall_num']").InnerText);
gallDate = DateTime.ParseExact(nick.SelectSingleNode("./td[@class='gall_date']").Attributes["title"].Value,
"yyyy-MM-dd HH:mm:ss", null);
Console.WriteLine(gallNum.ToString() + " " + gallDate.ToString());
if (gallNum >= previousPageGallNum)
{
Console.WriteLine(previousPageGallNum.ToString() + " " + gallNum.ToString());
Console.WriteLine("번호 에러");
continue;
}
if (DateTime.Compare(gallDate, initDate) < 0 || DateTime.Compare(gallDate, endDate) > 0)
{
Console.WriteLine("날짜 에러");
continue;
}
hap.HtmlNode user = nick.SelectSingleNode("./td[@class='gall_writer ub-writer']");
UserInfo tempUserInfo = new UserInfo(user.Attributes["data-nick"].Value);
if (user.Attributes["data-uid"].Value == "")
{
tempUserInfo.setFluidNick(user.Attributes["data-ip"].Value);
}
else
{
tempUserInfo.setFixedNick(user.Attributes["data-uid"].Value);
}
//replyNum and subject are in <td class='gall_tit ub-word'></td>
hap.HtmlNode subjectNode = nick.SelectSingleNode("./td[2]");
try
{
if (subjectNode.Attributes["class"].Value == "gall_subject")
{
subjectNode = nick.SelectSingleNode("./td[3]");
}
subject = subjectNode.SelectSingleNode("./a[1]").InnerText;
if (subjectNode.SelectNodes("./a").Count == 2)
{
replyNum = GetOnlyInt(subjectNode.SelectSingleNode("./a[@class='reply_numbox']/span").InnerText);
}
else
{
replyNum = 0;
}
}
catch
{
subject = "NullSubjectException";
replyNum = 0;
}
// Console.WriteLine("댓글: " + replyNum.ToString());
gallCount = GetOnlyInt(nick.SelectSingleNode("./td[@class='gall_count']").InnerText);
gallRecommend = GetOnlyInt(nick.SelectSingleNode("./td[@class='gall_recommend']").InnerText);
//Dictionary value => count, replyNum, gallCount, gallRecommend
if (userDic.ContainsKey(tempUserInfo))
{
userDic[tempUserInfo][0] += 1;
userDic[tempUserInfo][1] += replyNum;
userDic[tempUserInfo][2] += gallCount;
userDic[tempUserInfo][3] += gallRecommend;
}
else
{
int[] tempInts = new int[] { 1, replyNum, gallCount, gallRecommend };
userDic.Add(tempUserInfo, tempInts);
}
UserData tempUserData = new UserData(tempUserInfo);
tempUserData.DataInput(gallNum, replyNum, gallCount, gallRecommend, gallDate, subject);
//gallDatas.Add(tempUserData);
}
}
catch
{
if (ErrorOccured != null)
{
ErrorOccured(text, null);
}
currentPage++;
continue;
}
previousPageGallNum = GetOnlyInt(nicks[nicks.Count - 1].SelectSingleNode("./td[@class='gall_num']").InnerText);
DateTime currentDate = DateTime.ParseExact(nicks[nicks.Count - 1].
SelectSingleNode("./td[@class='gall_date']").Attributes["title"].Value, "yyyy-MM-dd HH:mm:ss", null);
if (currentPage >= endPage || DateTime.Compare(currentDate, initDate) < 0)
{
break;
}
else
{
System.Collections.ArrayList arr = new System.Collections.ArrayList();
string str = currentPage.ToString() + " 페이지, 날짜: " + currentDate.ToString();
arr.Add(str); arr.Add(currentDate); arr.Add(currentPage - initPage);
if (newPageHappened != null)
{
newPageHappened(arr, null);
}
currentPage++;
}
}
//Dictionary value => count, replyNum, gallCount, gallRecommend
foreach (KeyValuePair <UserInfo, int[]> user in userDic)
{
UserInfo tempUser = user.Key;
tempUser.count = user.Value[0];
tempUser.replyNum = user.Value[1];
tempUser.gallCount = user.Value[2];
tempUser.gallRecommend = user.Value[3];
UserRank tempUserRank = new UserRank(tempUser, user.Value[0], user.Value[1], user.Value[2], user.Value[3]);
userList.Add(tempUserRank);
}
var sorted = from userRank in userList
orderby userRank.count descending
select userRank;
userList = sorted.ToList <UserRank>();
if (CrawlingEnded != null)
{
CrawlingEnded(userList, null);
}
string tempDataDir = Directory.GetCurrentDirectory() + "\\temp-data\\";
Directory.CreateDirectory(tempDataDir);
string filename = tempDataDir + gallId + DateTime.Now.ToString("_yyyy-MM-dd_HH-mm-ss");
SaveResult(filename);
}