public LabelingForm()
{
InitializeComponent();
labelingSerializer = new Serializer(false);
statusSerializer = new Serializer(false);
doneLabeling = new List<Labeling>();
// Set menu item
MenuItem miLoad = new MenuItem("Load");
miLoad.Click += new EventHandler(menuLoad_click);
MenuItem miSave = new MenuItem("Save");
miSave.Click += new EventHandler(menuSave_click);
MenuItem miClear = new MenuItem("Clear");
miClear.Click += new EventHandler(menuClear_click);
MenuItem miFiltering = new MenuItem("Filtering");
miFiltering.Click += new EventHandler(menuFiltering_click);
this.Menu = new MainMenu();
this.Menu.MenuItems.Add(miLoad);
this.Menu.MenuItems.Add(miSave);
this.Menu.MenuItems.Add(miClear);
this.Menu.MenuItems.Add(miFiltering);
}
// 用依然可以使用的Home_Timeline获取最新的一大串微博
public static void fetchUserStatusInFile(string listPath, string toPath, int topMax, bool followFirst = false)
{
// 关注所有的list
Crawler crawler = new Crawler();
if (followFirst) crawler.followUserByUserList(listPath);
// 找到用户最新的微博ID
long max = long.MinValue;
Serializer rr = new Serializer(false);
if (File.Exists(toPath))
{
rr.ReadStream(toPath);
Status ts = (Status)rr.ReadNextObject();
while (ts != null)
{
if (max < long.Parse(ts.ID)) max = long.Parse(ts.ID);
ts = (Status)rr.ReadNextObject();
}
rr.CloseStream();
}
rr.WriteStream(toPath);
int count = 0, page = 1, numOfRecord = 0;
while(count < topMax){
int need = Math.Min(topMax - count, 100);
Console.WriteLine(need);
List<Status> ls = crawler.getUserStatusByHome(need, page);
Console.WriteLine(need + " " + page);
foreach (Status s in ls)
{
// 如果是最新的就写入到记录文件中
if (max < long.Parse(s.ID))
{
rr.WriteObject(s);
Console.WriteLine(s.ID);
numOfRecord++;
}
}
count += need;
page++;
}
rr.CloseStream();
Console.WriteLine(numOfRecord + " recorded");
}
public static void fetchUserStatus(string userName, int topMax)
{
// 找到用户名对应的文件
string filePath = pathUserList;
// 找到用户最新的微博ID
long max = long.MinValue;
Serializer rr = new Serializer(false);
if(File.Exists(filePath))
{
rr.ReadStream(filePath);
Status ts = (Status)rr.ReadNextObject();
while (ts != null)
{
if (max < long.Parse(ts.ID)) max = long.Parse(ts.ID);
ts = (Status)rr.ReadNextObject();
}
rr.CloseStream();
}
// 抓用户最新的微博
Crawler cr = new Crawler();
List<Status> ls = cr.getUserStatusByUserName(userName, topMax, 1);
rr.WriteStream(filePath);
foreach (Status s in ls)
{
// 如果是最新的就写入到记录文件中
if (max < long.Parse(s.ID))
{
rr.WriteObject(s);
Console.WriteLine(s.ID);
}
}
rr.CloseStream();
}
