/// <summary>
/// 以指定的UserID为起点开始爬行
/// </summary>
/// <param name="lUid"></param>
public void Start()
{
//获取上次中止处的用户ID并入队
long lLastUID = SysArg.GetCurrentID(SysArgFor.USER_INFO);
if (lLastUID > 0)
{
queueUserForUserInfoRobot.Enqueue(lLastUID);
}
while (queueUserForUserInfoRobot.Count == 0)
{
if (blnAsyncCancelled)
{
return;
}
Thread.Sleep(GlobalPool.SleepMsForThread); //若队列为空，则等待
}
AdjustFreq();
SetCrawlerFreq();
Log("The initial requesting interval is " + crawler.SleepTime.ToString() + "ms. " + api.ResetTimeInSeconds.ToString() + "s, " + api.RemainingIPHits.ToString() + " IP hits and " + api.RemainingUserHits.ToString() + " user hits left this hour.");
User user;
//对队列循环爬行
while (true)
{
if (blnAsyncCancelled)
{
return;
}
while (blnSuspending)
{
if (blnAsyncCancelled)
{
return;
}
Thread.Sleep(GlobalPool.SleepMsForThread);
}
//将队头取出
//lCurrentID = queueUserForUserInfoRobot.RollQueue();
lCurrentID = queueUserForUserInfoRobot.FirstValue;
//日志
Log("Recording current UserID: " + lCurrentID.ToString() + "...");
SysArg.SetCurrentID(lCurrentID, SysArgFor.USER_INFO);
#region 用户基本信息
if (blnAsyncCancelled)
{
return;
}
while (blnSuspending)
{
if (blnAsyncCancelled)
{
return;
}
Thread.Sleep(GlobalPool.SleepMsForThread);
}
Log("Crawling information of User " + lCurrentID.ToString() + "...");
user = crawler.GetUserInfo(lCurrentID);
//日志
AdjustFreq();
SetCrawlerFreq();
Log("Requesting interval is adjusted as " + crawler.SleepTime.ToString() + "ms. " + api.ResetTimeInSeconds.ToString() + "s, " + api.RemainingIPHits.ToString() + " IP hits and " + api.RemainingUserHits.ToString() + " user hits left this hour.");
if (user != null && user.user_id > 0)
{
//若数据库中不存在当前用户的基本信息，则爬取，加入数据库
if (!User.ExistInDB(lCurrentID))
{
//日志
Log("Saving User " + lCurrentID.ToString() + " into database...");
user.Add();
}
else
{
//日志
Log("Updating the information of User " + lCurrentID.ToString() + "...");
user.Update();
}
if (InvalidUser.ExistInDB(lCurrentID))
{
//日志
Log("Removing User " + lCurrentID.ToString() + " from invalid users...");
InvalidUser.RemoveFromDB(lCurrentID);
}
//日志
Log("The information of User " + lCurrentID.ToString() + " crawled.");
queueUserForUserInfoRobot.RollQueue();
}
else if (user == null) //用户不存在
{
Log("Recording invalid User " + lCurrentID.ToString() + "...");
InvalidUser iu = new InvalidUser();
iu.user_id = lCurrentID;
iu.Add();
//将该用户ID从各个队列中去掉
Log("Removing invalid User " + lCurrentID.ToString() + " from all queues...");
queueUserForUserRelationRobot.Remove(lCurrentID);
queueUserForUserInfoRobot.Remove(lCurrentID);
if (GlobalPool.TagRobotEnabled)
{
queueUserForUserTagRobot.Remove(lCurrentID);
}
if (GlobalPool.StatusRobotEnabled)
{
queueUserForStatusRobot.Remove(lCurrentID);
}
}
else if (user.user_id == -1) //forbidden
{
int iSleepSeconds = GlobalPool.GetAPI(SysArgFor.USER_INFO).ResetTimeInSeconds;
Log("Service is forbidden now. I will wait for " + iSleepSeconds.ToString() + "s to continue...");
for (int i = 0; i < iSleepSeconds; i++)
{
if (blnAsyncCancelled)
{
return;
}
Thread.Sleep(1000);
}
}
else if (user.user_id == -2) //timeout
{
int iSleepSeconds = GlobalPool.GetAPI(SysArgFor.USER_INFO).ResetTimeInSeconds;
Log("Time out. I will crawl user " + lCurrentID.ToString() + " again...");
}
#endregion
}
}
/// <summary>
/// 开始爬行取微博评论
/// </summary>
public void Start()
{
//获取上次中止处的微博ID并入队
long lLastStatusID = SysArg.GetCurrentID(SysArgFor.COMMENT);
if (lLastStatusID > 0)
{
queueStatus.Enqueue(lLastStatusID);
}
while (queueStatus.Count == 0)
{
if (blnAsyncCancelled)
{
return;
}
Thread.Sleep(GlobalPool.SleepMsForThread); //若队列为空，则等待
}
AdjustRealFreq();
SetCrawlerFreq();
Log("The initial requesting interval is " + crawler.SleepTime.ToString() + "ms. " + api.ResetTimeInSeconds.ToString() + "s, " + api.RemainingIPHits.ToString() + " IP hits and " + api.RemainingUserHits.ToString() + " user hits left this hour.");
//对队列无限循环爬行，直至有操作暂停或停止
while (true)
{
bool blnForbidden = false;
if (blnAsyncCancelled)
{
return;
}
while (blnSuspending)
{
if (blnAsyncCancelled)
{
return;
}
Thread.Sleep(GlobalPool.SleepMsForThread);
}
//将队头取出
lCurrentID = queueStatus.FirstValue;
//lCurrentID = queueStatus.RollQueue();
//日志
Log("Recording current StatusID: " + lCurrentID.ToString() + "...");
SysArg.SetCurrentID(lCurrentID, SysArgFor.COMMENT);
#region 微博相应评论
if (blnAsyncCancelled)
{
return;
}
while (blnSuspending)
{
if (blnAsyncCancelled)
{
return;
}
Thread.Sleep(GlobalPool.SleepMsForThread);
}
//日志
Log("Crawling the comments of Status " + lCurrentID.ToString() + "...");
int iPage = 1;
//爬取当前微博的评论
LinkedList <Comment> lstComment = new LinkedList <Comment>();
LinkedList <Comment> lstTemp = new LinkedList <Comment>();
lstTemp = crawler.GetCommentsOf(lCurrentID, iPage);
//日志
AdjustFreq();
SetCrawlerFreq();
Log("Requesting interval is adjusted as " + crawler.SleepTime.ToString() + "ms. " + api.ResetTimeInSeconds.ToString() + "s, " + api.RemainingIPHits.ToString() + " IP hits and " + api.RemainingUserHits.ToString() + " user hits left this hour.");
while (lstTemp.Count > 0)
{
if (blnAsyncCancelled)
{
return;
}
while (blnSuspending)
{
if (blnAsyncCancelled)
{
return;
}
Thread.Sleep(GlobalPool.SleepMsForThread);
}
while (lstTemp.Count > 0)
{
if (lstTemp.First.Value.comment_id > 0)
{
lstComment.AddLast(lstTemp.First.Value);
lstTemp.RemoveFirst();
}
else
{
blnForbidden = true;
lstTemp.Clear();
int iSleepSeconds = GlobalPool.GetAPI(SysArgFor.USER_INFO).ResetTimeInSeconds;
Log("Service is forbidden now. I will wait for " + iSleepSeconds.ToString() + "s to continue...");
for (int i = 0; i < iSleepSeconds; i++)
{
if (blnAsyncCancelled)
{
return;
}
Thread.Sleep(1000);
}
continue;
}
}
iPage++;
lstTemp = crawler.GetCommentsOf(lCurrentID, iPage);
//日志
AdjustFreq();
SetCrawlerFreq();
Log("Requesting interval is adjusted as " + crawler.SleepTime.ToString() + "ms. " + api.ResetTimeInSeconds.ToString() + "s, " + api.RemainingIPHits.ToString() + " IP hits and " + api.RemainingUserHits.ToString() + " user hits left this hour.");
}
if (blnForbidden)
{
continue;
}
//日志
Log(lstComment.Count.ToString() + " comments of Status " + lCurrentID.ToString() + " crawled.");
Comment comment;
while (lstComment.Count > 0)
{
if (blnAsyncCancelled)
{
return;
}
while (blnSuspending)
{
if (blnAsyncCancelled)
{
return;
}
Thread.Sleep(GlobalPool.SleepMsForThread);
}
comment = lstComment.First.Value;
if (!Comment.Exists(comment.comment_id))
{
//日志
Log("Saving Comment " + comment.comment_id.ToString() + " into database...");
comment.Add();
}
if (queueUserForUserRelationRobot.Enqueue(comment.user.user_id))
{
Log("Adding Commenter " + comment.user.user_id.ToString() + " to the user queue of User Relation Robot...");
}
if (GlobalPool.UserInfoRobotEnabled && queueUserForUserInfoRobot.Enqueue(comment.user.user_id))
{
Log("Adding Commenter " + comment.user.user_id.ToString() + " to the user queue of User Information Robot...");
}
if (GlobalPool.TagRobotEnabled && queueUserForUserTagRobot.Enqueue(comment.user.user_id))
{
Log("Adding Commenter " + comment.user.user_id.ToString() + " to the user queue of User Tag Robot...");
}
if (GlobalPool.StatusRobotEnabled && queueUserForStatusRobot.Enqueue(comment.user.user_id))
{
Log("Adding Commenter " + comment.user.user_id.ToString() + " to the user queue of Status Robot...");
}
if (!User.ExistInDB(comment.user.user_id))
{
Log("Saving Commenter " + comment.user.user_id.ToString() + " into database...");
comment.user.Add();
}
lstComment.RemoveFirst();
if (comment.reply_comment != null)
{
lstComment.AddLast(comment.reply_comment);
}
}//while for lstComment
queueStatus.RollQueue();
//日志
Log("Comments of Status " + lCurrentID.ToString() + " crawled.");
#endregion
}
}
/// <summary>
/// 以指定的UserID为起点开始爬行
/// </summary>
/// <param name="lUid"></param>
public void Start()
{
//获取上次中止处的用户ID并入队
long lLastUID = SysArg.GetCurrentID(SysArgFor.USER_TAG);
if (lLastUID > 0)
{
queueUserForUserTagRobot.Enqueue(lLastUID);
}
while (queueUserForUserTagRobot.Count == 0)
{
if (blnAsyncCancelled)
{
return;
}
Thread.Sleep(GlobalPool.SleepMsForThread); //若队列为空，则等待
}
AdjustRealFreq();
SetCrawlerFreq();
Log("The initial requesting interval is " + crawler.SleepTime.ToString() + "ms. " + api.ResetTimeInSeconds.ToString() + "s, " + api.RemainingIPHits.ToString() + " IP hits and " + api.RemainingUserHits.ToString() + " user hits left this hour.");
//对队列无限循环爬行，直至有操作暂停或停止
while (true)
{
if (blnAsyncCancelled)
{
return;
}
while (blnSuspending)
{
if (blnAsyncCancelled)
{
return;
}
Thread.Sleep(GlobalPool.SleepMsForThread);
}
//将队头取出
//lCurrentID = queueUserForUserTagRobot.RollQueue();
lCurrentID = queueUserForUserTagRobot.FirstValue;
//日志
Log("Recording current UserID: " + lCurrentID.ToString() + "...");
SysArg.SetCurrentID(lCurrentID, SysArgFor.USER_TAG);
#region 用户标签信息
if (blnAsyncCancelled)
{
return;
}
while (blnSuspending)
{
if (blnAsyncCancelled)
{
return;
}
Thread.Sleep(GlobalPool.SleepMsForThread);
}
//日志
Log("Crawling tags of User " + lCurrentID.ToString() + "...");
LinkedList <Tag> lstTag = crawler.GetTagsOf(lCurrentID);
if (lstTag.Count > 0 && lstTag.First.Value.tag_id > 0)
{
//日志
Log(lstTag.Count.ToString() + " tags crawled.");
//日志
AdjustFreq();
SetCrawlerFreq();
Log("Requesting interval is adjusted as " + crawler.SleepTime.ToString() + "ms. " + api.ResetTimeInSeconds.ToString() + "s, " + api.RemainingIPHits.ToString() + " IP hits and " + api.RemainingUserHits.ToString() + " user hits left this hour.");
while (lstTag.Count > 0)
{
if (blnAsyncCancelled)
{
return;
}
while (blnSuspending)
{
if (blnAsyncCancelled)
{
return;
}
Thread.Sleep(GlobalPool.SleepMsForThread);
}
Tag tag = lstTag.First.Value;
if (!Tag.Exists(tag.tag_id))
{
//日志
Log("Saving Tag " + tag.tag_id.ToString() + " into database...");
tag.Add();
}
else
{
//日志
//Log( "Tag " + tag.tag_id.ToString() + " exists." );
Log("Updating Tag " + tag.tag_id.ToString() + " into database...");
tag.Update();
}
if (!UserTag.Exists(lCurrentID, tag.tag_id))
{
//日志
Log("Recording User " + lCurrentID.ToString() + " has Tag " + tag.tag_id.ToString() + "...");
UserTag user_tag = new UserTag();
user_tag.user_id = lCurrentID;
user_tag.tag_id = tag.tag_id;
user_tag.Add();
}
else
{
//日志
Log("Tag " + tag.tag_id.ToString() + " of User " + lCurrentID.ToString() + " exists.");
}
lstTag.RemoveFirst();
}
queueUserForUserTagRobot.RollQueue();
//日志
Log("Tags of User " + lCurrentID.ToString() + " crawled.");
}
else if (lstTag.Count > 0 && lstTag.First.Value.tag_id == -1)
{
lstTag.Clear();
int iSleepSeconds = GlobalPool.GetAPI(SysArgFor.USER_INFO).ResetTimeInSeconds;
Log("Service is forbidden now. I will wait for " + iSleepSeconds.ToString() + "s to continue...");
for (int i = 0; i < iSleepSeconds; i++)
{
if (blnAsyncCancelled)
{
return;
}
Thread.Sleep(1000);
}
continue;
}
else if (lstTag.Count > 0 && lstTag.First.Value.tag_id == -2)
{
int iSleepSeconds = GlobalPool.GetAPI(SysArgFor.USER_INFO).ResetTimeInSeconds;
Log("Error! The error message is \"" + lstTag.First.Value.tag + "\". I will wait for " + iSleepSeconds.ToString() + "s to continue...");
lstTag.Clear();
for (int i = 0; i < iSleepSeconds; i++)
{
if (blnAsyncCancelled)
{
return;
}
Thread.Sleep(1000);
}
continue;
}
else
{
queueUserForUserTagRobot.RollQueue();
//日志
Log("Tags of User " + lCurrentID.ToString() + " crawled.");
}
#endregion
}
}
/// <summary>
/// 开始爬取微博数据
/// </summary>
public void Start()
{
//获取上次中止处的用户ID并入队
long lLastUID = SysArg.GetCurrentID(SysArgFor.STATUS);
if (lLastUID > 0)
{
queueUserForStatusRobot.Enqueue(lLastUID);
}
while (queueUserForStatusRobot.Count == 0)
{
if (blnAsyncCancelled)
{
return;
}
Thread.Sleep(GlobalPool.SleepMsForThread); //若队列为空，则等待
}
AdjustRealFreq();
SetCrawlerFreq();
Log("The initial requesting interval is " + crawler.SleepTime.ToString() + "ms. " + api.ResetTimeInSeconds.ToString() + "s, " + api.RemainingIPHits.ToString() + " IP hits and " + api.RemainingUserHits.ToString() + " user hits left this hour.");
//对队列无限循环爬行，直至有操作暂停或停止
while (true)
{
if (blnAsyncCancelled)
{
return;
}
while (blnSuspending)
{
if (blnAsyncCancelled)
{
return;
}
Thread.Sleep(GlobalPool.SleepMsForThread);
}
//将队头取出
//lCurrentID = queueUserForStatusRobot.RollQueue();
lCurrentID = queueUserForStatusRobot.FirstValue;
//日志
Log("Recording current UserID: " + lCurrentID.ToString() + "...");
SysArg.SetCurrentID(lCurrentID, SysArgFor.STATUS);
#region 用户微博信息
if (blnAsyncCancelled)
{
return;
}
while (blnSuspending)
{
if (blnAsyncCancelled)
{
return;
}
Thread.Sleep(GlobalPool.SleepMsForThread);
}
//日志
Log("Getting the latest Status ID of User " + lCurrentID.ToString() + "...");
//获取数据库中当前用户最新一条微博的ID
long lCurrentSID = Status.GetLastStatusIDOf(lCurrentID);
if (blnAsyncCancelled)
{
return;
}
while (blnSuspending)
{
if (blnAsyncCancelled)
{
return;
}
Thread.Sleep(GlobalPool.SleepMsForThread);
}
Status status;
#region 后续微博
//日志
Log("Crawling statuses after Status " + lCurrentSID.ToString() + " of User " + lCurrentID.ToString() + "...");
//爬取数据库中当前用户最新一条微博的ID之后的微博，存入数据库
LinkedList <Status> lstStatus = crawler.GetStatusesOfSince(lCurrentID, lCurrentSID);
if (lstStatus.Count > 0 && lstStatus.First.Value.status_id > 0)
{
//日志
Log(lstStatus.Count.ToString() + " statuses crawled.");
//日志
AdjustFreq();
SetCrawlerFreq();
Log("Requesting interval is adjusted as " + crawler.SleepTime.ToString() + "ms. " + api.ResetTimeInSeconds.ToString() + "s, " + api.RemainingIPHits.ToString() + " IP hits and " + api.RemainingUserHits.ToString() + " user hits left this hour.");
while (lstStatus.Count > 0)
{
if (blnAsyncCancelled)
{
return;
}
while (blnSuspending)
{
if (blnAsyncCancelled)
{
return;
}
Thread.Sleep(GlobalPool.SleepMsForThread);
}
status = lstStatus.First.Value;
SaveStatus(status);
lstStatus.RemoveFirst();
}
queueUserForStatusRobot.RollQueue();
//日志
Log("Statuses of User " + lCurrentID.ToString() + " crawled.");
}
else if (lstStatus.Count > 0 && lstStatus.First.Value.status_id == -1)
{
lstStatus.Clear();
int iSleepSeconds = GlobalPool.GetAPI(SysArgFor.STATUS).ResetTimeInSeconds;
Log("Service is forbidden now. I will wait for " + iSleepSeconds.ToString() + "s to continue...");
for (int i = 0; i < iSleepSeconds; i++)
{
if (blnAsyncCancelled)
{
return;
}
Thread.Sleep(1000);
}
continue;
}
else
{
queueUserForStatusRobot.RollQueue();
//日志
Log("Statuses of User " + lCurrentID.ToString() + " crawled.");
}
#endregion
#endregion
}
}
