public void SetText(NewsObject no)
{
label9.Text = no.type_name;
label2.Text = no.title;
label4.Text = no.pubdate;
label6.Text = no.source;
int i = int.Parse(count.Text);
i++;
count.Text = i.ToString();
i = int.Parse(all.Text);
i++;
all.Text = i.ToString();
string time = string.Format("{0: HH:mm:ss}", DateTime.Now);
string s = " ";
if (label13.Text.StartsWith(s))
{
s = "";
}
if (i%2 == 0)
{
label16.Text = "";
}
else
{
label16.Text = "√:";
}
if (no.status == 0)
label13.Text = s+"插入成功 --"+time;
else
label13.Text = s+"成功更新 --"+time;
}
private int InsertItem(NewsObject no)
{
string title = no.title.Replace("'","''");
string type_id = no.type_id;
string content = no.content.Replace("'", "''");
string source = no.source;
string pubdate = no.pubdate;
string misc = no.misc;
if (content.Contains("?"))
content = content.Replace('?', ' ');
string hrefs = MyClass.ExtractStr(content, "href", "<", ">", 1000, true);
foreach (string item in hrefs.Split(','))
{
if (item.StartsWith("a ") || item.StartsWith("A "))
content = content.Replace("<" + item + ">", "<a>");
}
string sql = "INSERT INTO data_xp_news(title,type_id,content,source,misc,pubdate) VALUES('{0}',{1},'{2}','{3}','{4}','{5}')";
sql = string.Format(sql, title.Replace("'", "''"), type_id, content.Replace("'", "''"), source.Replace("'", "''"), misc, pubdate);
if (!MyClass.SqlExists("select * from data_xp_news where title='" + title + "'"))
{
MyClass.ExecuteNonQuery(sql);
return 0;
}
else
{
sql = string.Format("update data_xp_news set content='{0}',source='{1}',misc='{2}',pubdate='{3}' where title='{4}'", content, source, misc, pubdate, title);
MyClass.ExecuteNonQuery(sql);
return 1;
}
}
NewsObject GetNewsObject(string title, string url,string type)
{
NewsObject no = new NewsObject();
string allc = ""; MyClass.GetUrltoHtml(url,"utf-8");
if (type == "本地新闻")//本地新闻提取方法
{
allc = MyClass.GetUrltoHtml(url, "utf-8");
string str1 = MyClass.ExtractStr(allc, "class=\"article\"", "<div", "div>", 1, true,",");
if (str1 != "")
str1 = "<div" + str1 + "div>";
string time = MyClass.GetNowTime();
string resource = "";
DateTime dt = DateTime.Now;
try
{
time = MyClass.ExtractStr(allc, "class=\"info\"", ">", " 来", 1, false, ",").Trim();
resource = MyClass.ExtractStr(allc, "class=\"info", "来源:", "</div>", 1, false, ",").Split('<')[2].Split('>')[1];
dt = DateTime.Parse(time);
}
catch (Exception)
{
}
time = MyClass.FormateTime(dt);
no.title = title;
no.pubdate = time;
no.source = resource;
no.content = str1;
}
else
{
allc = MyClass.GetUrltoHtml(url,"gb2312");
string str1 = MyClass.ExtractStr(allc, "id=\"contentText\"", "<div", "div>", 1, true,",");
if (str1 != "")
str1 = "<div" + str1 + "div>";
string time = MyClass.ExtractStr(allc, "class=\"l\"", ">", "<", 1, false, ",");
string resource = MyClass.ExtractStr(allc, "id=\"media_span\"", ">", "<", 1, false, ",");
if (resource == "")
{
resource = MyClass.ExtractStr(allc, "itemprop=\"name\"", ">", "<", 1, false, ",");
}
if (time == "")
{
time = MyClass.ExtractStr(allc, "class=\"time\"", ">", "<", 1, false, ",");
}
DateTime dt;
try
{
dt = DateTime.Parse(time);
time = MyClass.FormateTime(dt);
}
catch (Exception)
{
time = MyClass.FormateTime(DateTime.Now);
}
no.title = title;
no.pubdate = time;
no.source = resource;
no.content = str1;
}
return no;
}
NewsObject GetNewsObject(string title, string url,string type)
{
NewsObject no = new NewsObject();
string allc = MyClass.GetUrltoHtml(url);
allc = MyClass.GetUrltoHtml(url,"gb2312");
string str1 = MyClass.ExtractStr(allc, "", "<div id=\"tupian_div\"", "<div class=\"left_name\"", 1);
if (str1 != "")
{
string script = MyClass.ExtractStr(str1, "", "<script", "</script>", 100, false,"，");
if(script!="")
foreach (string sc in script.Split('，'))
{
if (sc != "")
str1 = str1.Replace("<script" + sc + "</script>", "");
}
if (str1.Contains("<img"))
{
string str11 = MyClass.ExtractStr(str1, "<img ", "<img", ">", 100, true);
if (str11 != "")
{
string[] imgs = str11.Split(',');
foreach (string ig in imgs)
{
string igsrc=MyClass.ExtractStr(ig, "src", "\"", "\"");
if (igsrc != "" && !igsrc.StartsWith("http"))
{
if (igsrc.StartsWith("/"))
{
string snurl = "http://" + new Uri(url).Host + igsrc;
str1 = str1.Replace(igsrc, snurl);
}
else
{
int ids = url.LastIndexOf("/");
if (ids != -1)
{
string snurl = url.Remove(ids+1) + igsrc;
str1 = str1.Replace(igsrc, snurl);
}
}
}
}
}
}
// str1 = "<div" + str1 + "div>";
str1 = "<div id=\"tupian_div\"" + str1 ;
//string str2 = MyClass.ExtractStr(allc, "class=\"left-time\"", "<div", "div>", 1, true);
//str2 = "<div" + str2 + "div>";
string str3 = MyClass.ExtractStr(allc, "class=\"left_name\"", "<div", "div>", 1, true);
str3 = "<div" + str3 + "div>";
str1 =str1 + str3;
//if (str2 != "")
//{
// str1 = str1.Replace(str2, "");
//}
}
string time = MyClass.GetNowTime();
string[] timeandresource = MyClass.ExtractStr(allc, "\"left-t\"", ">", "<").Trim().Split(' ');
string resource = "";
if (timeandresource.Length == 2)
{
resource = timeandresource[1].Replace("来源：", "");
}
if (timeandresource.Length >0)
{
time = timeandresource[0].Replace("年", "-").Replace("月", "-").Replace("日", "");
}
DateTime dt;
try
{
dt = DateTime.Parse(time);
time = MyClass.FormateTime(dt);
}
catch (Exception)
{
time = MyClass.FormateTime(DateTime.Now);
}
no.title = title;
no.pubdate = time;
no.source = resource;
no.content = str1;
return no;
}
void GetNews()
{
Dictionary<string, string> bg = GetBigTypes();
progressBar1.Maximum = (bg.Count)*10+1;
progressBar1.PerformStep();
int index = 0;
foreach (KeyValuePair<string,string> item in bg)
{
index++;
label15.Text = index.ToString()+"/"+bg.Count.ToString();
object type=MyClass.ExecuteScalar(string.Format("SELECT id from data_xp_newstype where typename='{0}'", item.Key));
if(type==null)return;
string typeid = type.ToString();
Dictionary<string,string> sm=GetTitles(item.Value);
if (sm == null)
return;
totle.Text = sm.Count.ToString();
count.Text = "0";
int j = 0;
foreach (KeyValuePair<string, string> item1 in sm)
{
if (!newisContinue)
{
return;
}
j++;
NewsObject no=new NewsObject();
bool isget = false;
string sql="select * from data_xp_news where title='" + item1.Key + "'";
object[] neos= MyClass.GetObjects(sql, no.GetType());
if(neos!=null && neos.Length!=0)
{
no=(NewsObject)neos[0];
isget = true;
}
if (!checkBox1.Checked || !isget)
{
no = GetNewsObject(item1.Key, item1.Value, item.Key);
no.type_id = typeid;
no.type_name = item.Key;
if (no.content != "")
{
no.status = InsertItem(no);
}
}
no.type_id = typeid;
no.type_name = item.Key;
label1.Invoke(new SetTextDelegate(SetText), no);
progressBar1.Value= 10*(index-1) + j*10/sm.Count+1;
label14.Text = (((index-1) * 100 / bg.Count +j * 10 / sm.Count)).ToString() + "%";
}
}
}
