private void GetKeyWordData()
{
DataTable dt = new DataTable();
//dt = cmd.GetTabel("select uid,keyword as 关键字,case messagealarm when 0 then '不报警' else '报警' end as 短信报警 ,case musicalarm when 0 then '不报警' else '报警' end as 声音报警 ,case has when 0 then '不包含' else '包含' end as 是否包含 from keywords order by uid desc");
dt = cmd.GetTabel("select uid,keyword as 关键字,case messagealarm when 0 then '不报警' else '报警' end as 短信报警 ,case musicalarm when 0 then '不报警' else '报警' end as 声音报警 from keywords where has=1 order by uid desc");
dataGridView1.DataSource = dt;
dataGridView1.Columns[0].Visible = false;
}
private void GetWebHrefData()
{
DataTable dt = new DataTable();
dt = cmd.GetTabel("select uid,name as 别名 ,url as 网址 ,likeurl as 相似链接 ,case pid when 1 then '博客' when 2 then '论坛' else '其他' end as 网站类别 from webaddress order by uid desc");
dataGridView1.DataSource = dt;
dataGridView1.Columns[0].Visible = false;
//dataGridView1.Columns[2].AutoSizeMode = DataGridViewAutoSizeColumnMode.Fill;
dataGridView1.Columns[2].Width = 300;
//dataGridView1.Columns[3].Width = 400;
dataGridView1.Columns[3].AutoSizeMode = DataGridViewAutoSizeColumnMode.Fill;
dataGridView1.Columns[4].DefaultCellStyle.Alignment = DataGridViewContentAlignment.MiddleCenter;
}
private void GetPartWordData()
{
DataTable dt = new DataTable();
string sql = "select uid , word , part from partword ";
if (pWordView.Checked)
{
sql += "where part=1";
}
if (mWordView.Checked)
{
sql += "where part=0";
}
dt = cmd.GetTabel(sql + " order by uid desc");
dataGridView1.DataSource = dt;
dataGridView1.Columns[0].Visible = false;
}
/// <summary>
/// 通用网站抓取程序
/// </summary>
private void GeneralWebSpider()
{
DataBaseServer.MySqlCmd cmd = new MySqlCmd();
//得到关键字列表
DataTable dtkey = cmd.GetTabel("select * from Keywords");
//得到抓取网站的地址
DataTable dtXs = cmd.GetTabel("Select * from WebAddress");
#region 读取相似度表中的数据据,循环抓取
for (int xs = 0; xs < dtXs.Rows.Count; xs++)
{
string url = dtXs.Rows[xs]["url"].ToString();
string sheng = dtXs.Rows[xs]["sheng"].ToString();
string shi = dtXs.Rows[xs]["shi"].ToString();
string xian = dtXs.Rows[xs]["xian"].ToString();
string name = dtXs.Rows[xs]["Name"].ToString();
//读取相似链接
string Similar = dtXs.Rows[xs]["likeurl"].ToString();
bool isCatchSubWeb = false;
BasicWebSpider(url, Similar, dtkey, sheng, shi, xian, name, isCatchSubWeb);
//防止拉黑
Thread.Sleep(2000);
}
#endregion
}
private void BlogWebSpider()
{
#region 处理关键字
MySqlCmd cmd = new MySqlCmd();
//得到关键字列表
DataTable dtkey;
DataTable dtParts;
if (selectKID == -1)
{
//全部
dtkey = cmd.GetTabel("select * from Keywords");
}
else
{
dtkey = cmd.GetTabel("select * from Keywords where kid=" + selectKID);
}
dtParts = cmd.GetTabel("SELECT * FROM partword");
#endregion
HtmlParse.Parse parse = new HtmlParse.Parse();
//parse.ReportCatchProcess += new HtmlParse.ReportCatchProcessEventHandler(Blog_ReportCatchProcess);
List<ModelReleaseInfo> webDatas = new List<ModelReleaseInfo>();
//按关键字循环
for (int kw = 0; kw < dtkey.Rows.Count; kw++)
{
//处理关键字
if (selectKID != -1 && selectKwName != "全部")
{
if (dtkey.Rows[kw]["name"].ToString().Trim() != selectKwName) continue;
}
string keyword = dtkey.Rows[kw]["KeyWord"].ToString().Trim();
string keyTitle = dtkey.Rows[kw]["Name"].ToString().Trim();
int kid = 0;
int.TryParse(dtkey.Rows[kw]["kid"].ToString().Trim(), out kid);
#region 按关键字检索
//组成查询字串
string url = "http://www.sogou.com/web?interation=196647&query=" + keyword + "&ie=utf8";
string html = HtmlUtil.HttpGet(url, Encoding.UTF8);
#endregion
List<ModelReleaseInfo> mris = parse.ParseSogouBlog(html, keyword, kid);
if (mris != null && mris.Count() > 0)
{
DataPersistenceControl.GetInstance().Add(mris);
}
//防止拉黑
Thread.Sleep(Interval2m);
}
}
private void BaiduWebWebSpider()
{
#region 处理关键字
MySqlCmd cmd = new MySqlCmd();
//得到关键字列表
DataTable dtkey;
DataTable dtParts;
if (selectKID == -1)
{
//全部
dtkey = cmd.GetTabel("select * from Keywords");
}
else
{
dtkey = cmd.GetTabel("select * from Keywords where kid=" + selectKID);
}
dtParts = cmd.GetTabel("SELECT * FROM partword");
#endregion
HtmlParse.Parse parse = new HtmlParse.Parse();
//parse.ReportCatchProcess += new HtmlParse.ReportCatchProcessEventHandler(BaiduWeb_ReportCatchProcess);
List<ModelReleaseInfo> webDatas = new List<ModelReleaseInfo>();
//按关键字循环
for (int kw = 0; kw < dtkey.Rows.Count; kw++)
{
//处理关键字
if (selectKID != -1 && selectKwName != "全部")
{
if (dtkey.Rows[kw]["name"].ToString().Trim() != selectKwName) continue;
}
string keyword = dtkey.Rows[kw]["KeyWord"].ToString().Trim();
string keyTitle = dtkey.Rows[kw]["Name"].ToString().Trim();
int kid = 0;
int.TryParse(dtkey.Rows[kw]["kid"].ToString().Trim(), out kid);
#region 百度检索
for (int i = 0; i < 5; i++)
{
String encodeKey = CrawlHtml.UrlEncode(keyword);
string url = string.Format(@"http://www.baidu.com/s?wd={0}&pn={1}&ie=utf-8", encodeKey, i * 10);
string html = HtmlUtil.HttpGet(url, Encoding.UTF8);
List<ModelReleaseInfo> mris = parse.ParseBaiduWeb(html, keyword, kid);
if (mris != null && mris.Count() > 0)
{
DataPersistenceControl.GetInstance().Add(mris);
}
else
{
break;
}
//防止拉黑
Thread.Sleep(Interval50s);
}
#endregion
#region bing检索
for (int i = 0; i < 10; i++)
{
//组成查询字串
string url = string.Format("http://cn.bing.com/search?q={0}&first={1}&FORM=PERE", keyword, i * 10 + 1);
string html = HtmlUtil.HttpGet(url, Encoding.UTF8);
List<ModelReleaseInfo> mris = parse.ParseBingWeb(html, keyword, kid);
if (mris != null && mris.Count() > 0)
{
//写入数据库
DataPersistenceControl.GetInstance().Add(mris);
}
else
{
break;
}
//防止拉黑
Thread.Sleep(Interval30s);
}
#endregion
#region 搜狗检索
for (int i = 0; i < 10; i++)
{
//组成查询字串
string url = string.Format("http://www.sogou.com/web?query={0}&page={1}&ie=utf8", keyword, i + 1);
string html = HtmlUtil.HttpGet(url, Encoding.UTF8);
List<ModelReleaseInfo> mris = parse.ParseSogouWeb(html, keyword, kid);
if (mris != null && mris.Count() > 0)
{
//写入数据库
DataPersistenceControl.GetInstance().Add(mris);
}
else
{
break;
}
//防止拉黑
Thread.Sleep(Interval2m);
}
#endregion
#region 中搜检索
for (int i = 0; i < 10; i++)
{
//组成查询字串
string url = string.Format("http://www.zhongsou.com/third?w={0}&b={1}", keyword, i + 1);
string html = HtmlUtil.HttpGet(url, Encoding.Default);
List<ModelReleaseInfo> mris = parse.ParseZhongsouWeb(html, keyword, kid);
if (mris != null && mris.Count() > 0)
{
//写入数据库
DataPersistenceControl.GetInstance().Add(mris);
}
else
{
break;
}
//防止拉黑
Thread.Sleep(Interval2m);
}
#endregion
#region 好搜检索
for (int i = 0; i < 10; i++)
{
//组成查询字串
string url = string.Format("http://www.haosou.com/s?q={0}&pn={1}", keyword, i + 1);
string html = HtmlUtil.HttpGet(url, Encoding.UTF8);
List<ModelReleaseInfo> mris = parse.ParseHaosouWeb(html, keyword, kid);
if (mris != null && mris.Count() > 0)
{
//写入数据库
DataPersistenceControl.GetInstance().Add(mris);
}
else
{
break;
}
//防止拉黑
Thread.Sleep(Interval50s);
}
#endregion
}
}
public DataTable GetLatestData(int kid, string eventName)
{
DataPage firstPage = Finder.util.Comm.GetPageInfo();
//string sql = "Select * From ReleaseInfo where pid={0} and deleted=0 order by collectdate desc limit 0,100";
string sql = @"select b.Name eventname, a.* from releaseinfo a left join keywords b on a.keywords=b.KeyWord
where b.Name is not null and a.deleted=0 and a.uid between " + firstPage.CurrenPageStartUid + " and " + firstPage.CurrenPageEndUid;
if (kid != -1)
{
sql += " and a.kid=" + kid;
if (!string.IsNullOrEmpty(eventName) && eventName != "全部")
{
sql += " and b.Name='" + eventName + "'";
}
}
sql += " order by a.collectdate desc limit 0,50";
try
{
DataBaseServer.MySqlCmd dbobj = new MySqlCmd();
return dbobj.GetTabel(sql);
}
catch (Exception ex)
{
return null;
}
}
//刷新微博数据,前台感觉是在搜索微博
private void reFreshWeibo()
{
if (!Program.ProClose)
{
lbweibo.Visible = true;
lbweibo.Text = "正在搜索微博数据...";
lbweibo.ForeColor = Color.DarkBlue;
string wblog = "select uid,releasename AS title,contexts,releasedate,infosource,keywords,releasename,collectdate,snapshot,webname,pid,part,reposts,comments from ReleaseInfowb order by uid desc limit 0,300";
DataBaseServer.MySqlCmd cmd = new MySqlCmd();
DataTable dtwBlog = new DataTable();
dtwBlog = cmd.GetTabel(wblog);
dvWBlog.DataSource = dtwBlog;
dvWBlog.Refresh();
lbweibo.Text = "一轮搜索完毕！";
lbweibo.ForeColor = Color.Red;
}
}
private void MediaWebSpider()
{
#region 处理关键字
MySqlCmd cmd = new MySqlCmd();
//得到关键字列表
DataTable dtkey;
DataTable dtParts;
if (selectKID == -1)
{
//全部
dtkey = cmd.GetTabel("select * from Keywords");
}
else
{
dtkey = cmd.GetTabel("select * from Keywords where kid=" + selectKID);
}
dtParts = cmd.GetTabel("SELECT * FROM partword");
#endregion
HtmlParse.Parse parse = new HtmlParse.Parse();
//parse.ReportCatchProcess += new HtmlParse.ReportCatchProcessEventHandler(parse_ReportCatchProcess);
List<ModelReleaseInfo> webDatas = new List<ModelReleaseInfo>();
//按关键字循环
for (int kw = 0; kw < dtkey.Rows.Count; kw++)
{
//处理关键字
if (selectKID != -1 && selectKwName != "全部")
{
if (dtkey.Rows[kw]["name"].ToString().Trim() != selectKwName) continue;
}
string keyword = dtkey.Rows[kw]["KeyWord"].ToString().Trim();
//string keyTitle = dtkey.Rows[kw]["Name"].ToString().Trim();
int kid = 0;
int.TryParse(dtkey.Rows[kw]["kid"].ToString().Trim(), out kid);
#region 百度检索
//组成查询字串
string url = "http://news.baidu.com/ns?rn=100&word=" + keyword;
string html = HtmlUtil.HttpGet(url, Encoding.UTF8);
List<ModelReleaseInfo> mris = parse.ParseBaiduNews(html, keyword, kid);
if (mris != null && mris.Count() > 0)
{
//写入数据库
DataPersistenceControl.GetInstance().Add(mris);
}
#endregion
#region bing检索
for (int i = 0; i < 10; i++)
{
//组成查询字串
url = "";
html = "";
mris = null;
url = string.Format("http://cn.bing.com/news/search?q={0}&first={1}&FORM=PENR", keyword, i * 10 + 1);
html = HtmlUtil.HttpGet(url, Encoding.UTF8);
mris = parse.ParseBingNews(html, keyword, kid);
if (mris != null && mris.Count() > 0)
{
//写入数据库
DataPersistenceControl.GetInstance().Add(mris);
}
else
{
break;
}
//防止拉黑
Thread.Sleep(Interval50s);
}
#endregion
#region 搜狗新闻
for (int i = 0; i < 10; i++)
{
//组成查询字串
url = "";
html = "";
mris = null;
url = string.Format("http://news.sogou.com/news?query={0}&page={1}", keyword, i + 1);
html = HtmlUtil.HttpGet(url, Encoding.Default);
mris = parse.ParseSogouNews(html, keyword, kid);
if (mris != null && mris.Count() > 0)
{
//写入数据库
DataPersistenceControl.GetInstance().Add(mris);
}
else
{
break;
}
//防止拉黑（2分钟）
Thread.Sleep(Interval2m);
}
#endregion
#region 中搜新闻
for (int i = 0; i < 10; i++)
{
//组成查询字串
url = "";
html = "";
mris = null;
url = string.Format("http://zixun.zhongsou.com/n?w={0}&b={1}", keyword, i + 1);
html = HtmlUtil.HttpGet(url, Encoding.Default);
mris = parse.ParseZhongsouNews(html, keyword, kid);
if (mris != null && mris.Count() > 0)
{
//写入数据库
DataPersistenceControl.GetInstance().Add(mris);
}
else
{
break;
}
//防止拉黑（2分钟）
Thread.Sleep(Interval2m);
}
#endregion
#region 好搜新闻
for (int i = 0; i < 10; i++)
{
//组成查询字串
url = "";
html = "";
mris = null;
url = string.Format("http://news.haosou.com/ns?q={0}&pn={1}&tn=news&rank=rank&j=0", keyword, i + 1);
html = HtmlUtil.HttpGet(url, Encoding.UTF8);
mris = parse.ParseHaosouNews(html, keyword, kid);
if (mris != null && mris.Count() > 0)
{
//写入数据库
DataPersistenceControl.GetInstance().Add(mris);
}
else
{
break;
}
//防止拉黑
Thread.Sleep(Interval50s);
}
#endregion
}
}
/// <summary>
/// 查询采集信息
/// </summary>
/// <param name="t1">时间起</param>
/// <param name="t2">时间至</param>
/// <param name="infoType">数据类型</param>
/// <returns></returns>
public DataTable SelReleaseInfo(string t1, string t2, string pid)
{
DataPage firstPage = Finder.util.Comm.GetPageInfo();
string sql = "SELECT * FROM ReleaseInfo WHERE CollectDate BETWEEN '{1}' AND '{0}' AND pid={2} and uid between " + firstPage.CurrenPageStartUid + " and " + firstPage.CurrenPageEndUid + " ORDER BY CollectDate";
sql = string.Format(sql, t1, t2, pid);
try
{
DataBaseServer.MySqlCmd dbobj = new MySqlCmd();
return dbobj.GetTabel(sql);
}
catch (Exception ex)
{
throw new Exception("查询失败,位置:SelReleaseInfo.原因:" + ex.Message);
}
}
/// <summary>
/// 查询采集信息(按网址)
/// </summary>
/// <returns></returns>
public DataTable SelReleaseInfo(string sUrl)
{
DataPage firstPage = Finder.util.Comm.GetPageInfo();
string sql = "SELECT * FROM ReleaseInfo WHERE InfoSource ='{0}' and uid between " + firstPage.CurrenPageStartUid + " and " + firstPage.CurrenPageEndUid;
sql = string.Format(sql, sUrl);
try
{
DataBaseServer.MySqlCmd dbobj = new MySqlCmd();
return dbobj.GetTabel(sql);
}
catch (Exception ex)
{
throw new Exception("查询失败,位置:SelReleaseInfo.原因:" + ex.Message);
}
}
/// <summary>
/// 查询信息(按ID)
/// </summary>
/// <returns></returns>
public DataTable SelReleaseInfo(int uid)
{
string sql = "SELECT * FROM ReleaseInfo WHERE uid={0}";
sql = string.Format(sql, uid);
try
{
DataBaseServer.MySqlCmd dbobj = new MySqlCmd();
return dbobj.GetTabel(sql);
}
catch (Exception ex)
{
throw new Exception("查询失败,位置:SelReleaseInfo.原因:" + ex.Message);
}
}
/// <summary>
/// 查询采集信息(按网址)
/// </summary>
/// <returns></returns>
public DataTable GetReleaseInfoFormat()
{
string sql = "SELECT uid,Title,Contexts,ReleaseDate,InfoSource,KeyWords,ReleaseName,CollectDate,Snapshot,webName,pid,part,comments,reposts FROM ReleaseInfo WHERE 1=2";
try
{
DataBaseServer.MySqlCmd dbobj = new MySqlCmd();
return dbobj.GetTabel(sql);
}
catch (Exception ex)
{
throw new Exception("查询失败,位置:GetReleaseInfoFormat.原因:" + ex.Message);
}
}
public int GetMaxUid()
{
string sql = "SELECT max(uid) FROM ReleaseInfo";
try
{
DataBaseServer.MySqlCmd dbobj = new MySqlCmd();
DataTable dt = new DataTable();
dt = dbobj.GetTabel(sql);
try
{
return int.Parse(dt.Rows[0][0].ToString());
}
catch (Exception)
{
return 0;
}
}
catch (Exception ex)
{
throw new Exception("查询失败,位置:GetMaxUid.原因:" + ex.Message);
}
}
public void timer_Elapsed(object source, System.Timers.ElapsedEventArgs e)
{
if (!Program.ProClose)
{
getSinaClientAuto();
cmd = new MySqlCmd();
dt_kw = cmd.GetTabel("select KeyWord from KeyWords group by KeyWord");
dt_event = cmd.GetTabel("select name,keyword,kid from KeyWords group by name,keyword,kid");
dt_partWord = cmd.GetTabel("select word,part from partword");
getPublic_Timeline_sina(sina, 200);
getFriends_Timeline_sina(sina, 100);
}
}
/// <summary>
/// 得到网站的新闻类数据
/// </summary>
private void GetWebNewsInfo()
{
lbWeb.Text = "";
lbWeb.Visible = true;
//相似链接
string Similar = "";
DataBaseServer.MySqlCmd cmd = new MySqlCmd();
//得到相似表
DataTable dtXs = new DataTable();
dtXs = cmd.GetTabel("Select * from WebAddress WHERE pid=0");
dtParts = cmd.GetTabel("SELECT * FROM partword");
DataTable dtkey = new DataTable();
dtkey = cmd.GetTabel("select * from Keywords");
//相似表中的被抓取网址
string webInfo = "";
//要过滤链接中首页的正则
string strTopFormat = "http://.+/";
List<string> strTop = new List<string>();
sb = new StringBuilder();
sb.Append("");
string filterStr = "";
#region 读取相似度表中的数据据,循环抓取
for (int xs = 0; xs < dtXs.Rows.Count; xs++)
{
lbWeb.ForeColor = Color.DarkBlue;
lbWeb.Text = "正在搜索:" + dtXs.Rows[xs]["name"].ToString();
//读取相似表中要抓取的网址
webInfo = getHtml(dtXs.Rows[xs]["url"].ToString(), "");
//读取相似链接
Similar = dtXs.Rows[xs]["likeurl"].ToString();
//取出
//string[] strA = HtmlUtil.GetElementsByTagName(webInfo, "a");
List<string> strList = HtmlUtil.GetElementsByTagNameList(webInfo, "a");
string strURLformat = "http://.[^\"]+";
TbReleaseInfo ri = new TbReleaseInfo();
string[] strA = GetLIstDate(strList.Distinct());
#region 逐个链接判断
//循环时判断是否要验证
bool isThere = false;
for (int i = 0; i < strA.Length; i++)
{
if (Program.ProClose == true) break;
Application.DoEvents();
try
{
//得到目标网址中的所有链接,如果未得到,那么就继续读取下一个
strA[i] = HtmlUtil.GetListByHtml(dtXs.Rows[xs]["url"].ToString(), strA[i], strURLformat)[0];
//处理含有单引号的链接
strA[i] = UrlCl(strA[i]);
//处理单引号的链接
if (strA[i].IndexOf("'") != -1)
{
strA[i] = GetstringByHtmlArray(strA[i], "http://.[^\']+");
}
}
catch (Exception)
{
continue;
}
//得到相似值,大于0.70的认为相同,并开始抓取
if (HtmlUtil.getSimilarDegree(Similar, strA[i]) >= 0.60)
{
//判断这个链接是否已经在库中或者列表中,如果存在,此次就不再执行
strTop = HtmlUtil.GetListByHtmlArray(strA[i], strTopFormat);
if (strTop.Count != 0)
{
//if (strTop[0] == "http://blog.sohu.com/")
continue;//同新闻,如果将首页去掉
}
if (isThere)
{
continue;
}
else
{
//if (strA[i] == "http://news.ifeng.com/mainland/detail_2013_10/18/30459577_0.shtml'>[详细]</a>")
//{
// strA[i] = strA[i];
//}
if (UrlThereare(strA[i], this.dtnewsinfo, dtWebNewsInfo, true) != 0) { isThere = true; continue; }
}
//得到此链接的源码
webInfo = getHtml(strA[i], "");
if (webInfo.Length == 0) { continue; }
//创建数据对象
ModelReleaseInfo newsInfo = new ModelReleaseInfo();
try
{
//流水+1
newsInfo.Uid = this.dvAll.Rows.Count + 1;
//标题
string[] strT = HtmlUtil.GetElementsByTagName(webInfo, "title");
if (strT.Length == 0)
{
continue;
}
else
{
newsInfo.Title = HtmlUtil.NoHTML(HtmlUtil.GetElementsByTagName(webInfo, "title")[0]);
}
//得到正文,以P标签来区分
string[] strContext = HtmlUtil.GetElementsByTagName(webInfo, "p");
newsInfo.Contexts = "";
for (int j = 0; j < strContext.Length; j++)
{
//循环累加正文信息
newsInfo.Contexts += HtmlUtil.NoHTML(strContext[j]);
}
//如果正文信息为空,那么将无法做关键字对照,此条数据舍弃
if (newsInfo.Contexts.Length == 0)
{
continue;
}
//网站链接
newsInfo.InfoSource = strA[i].Trim();
//关键字的设置
newsInfo.KeyWords = "";
for (int j = 0; j < dtkey.Rows.Count; j++)
{
Application.DoEvents();
if (newsInfo.Contexts.IndexOf(dtkey.Rows[j][1].ToString()) > 0)
{ newsInfo.KeyWords += dtkey.Rows[j][1].ToString() + ","; }
else
{
}
}
if (newsInfo.KeyWords.Length == 0) { continue; }
newsInfo.KeyWords = newsInfo.KeyWords.Substring(0, newsInfo.KeyWords.Length - 1);
//收集日期
newsInfo.CollectDate = string.Format(DateTime.Now.ToString("yyyy-MM-dd HH:mm:ss"));
//发布人和发布日期暂时无法取到,手工赋值为空
newsInfo.ReleaseDate = "";
newsInfo.ReleaseName = "";
//网页快照,这里为用户指定生成,如果未选择生成,那么为空
newsInfo.Snapshot = "";
//网站名
newsInfo.WebName = dtXs.Rows[xs]["Name"].ToString();
//pid
newsInfo.Pid = 0;
//part正负判断
newsInfo.Part = GetParts(newsInfo.Contexts);
//reposts
newsInfo.Reposts = 0;
//comments
newsInfo.Comments = 0;
//新建数据行
DataRow dr = dtnewsinfo.NewRow();
if (dvWeb.RowCount == 0)
{
dr[0] = 1;
}
else
{
dr[0] = int.Parse(dvWeb.Rows[dvWeb.RowCount - 1].Cells[0].Value.ToString()) + 1;
}
//dr[0] = newsInfo.Uid;
dr[1] = newsInfo.Title;
dr[2] = newsInfo.Contexts;
dr[3] = newsInfo.ReleaseDate;
dr[4] = newsInfo.InfoSource;
dr[5] = newsInfo.KeyWords;
dr[6] = newsInfo.ReleaseName;
dr[7] = newsInfo.CollectDate;
dr[8] = newsInfo.Snapshot;
dr[9] = newsInfo.WebName;
dr[10] = newsInfo.Pid;
dr[11] = newsInfo.Part;
dr[12] = newsInfo.Reposts;
dr[13] = newsInfo.Comments;
//把行加到DT中
dtnewsinfo.Rows.InsertAt(dr, 0);
//数据源刷新
if (dtnewsinfo.Rows.Count >= 500)
{
dtnewsinfo.Rows.RemoveAt(500);
}
dvWeb.Refresh();
}
catch (Exception ex)
{
StreamWriter sw = File.AppendText("log.txt");
sw.WriteLine(DateTime.Now.ToLongDateString());
sw.WriteLine("begin");
sw.WriteLine(ex.Message);
sw.WriteLine(sb.ToString());
sw.WriteLine("end");
sw.WriteLine("");
sw.Close();
}
////总表刷新
//dt.Rows.Add(dr);
//dvAll.Refresh();
//得到插入语句
try
{
if (isThere)
{
continue;
}
else
{
sb.Append(ri.GetInsString(newsInfo) + ";");
}
//每10次执行一次插入数据库
if (sb.ToString().Length != 0)
{
if (i % 10 == 0)
{
filterStr = sb.ToString();
filterStr = filterStr.Replace("[ ", "[");
filterStr = filterStr.Replace(" ]", "]");
//执行插入
cmd.ExecuteNonQuery(filterStr);
//清除插入字段串
sb.Clear();
filterStr = "";
}
}
}
catch (Exception ex)
{
StreamWriter sw = File.AppendText("log.txt");
sw.WriteLine(DateTime.Now.ToLongDateString());
sw.WriteLine("begin");
sw.WriteLine(ex.Message);
sw.WriteLine(sb.ToString());
sw.WriteLine("end");
sw.WriteLine("");
sw.Close();
}
}
}
#endregion
}
#endregion
try
{
if (sb.ToString().Length != 0)
{
filterStr = sb.ToString();
filterStr = filterStr.Replace("[ ", "[");
filterStr = filterStr.Replace(" ]", "]");
//执行插入
cmd.ExecuteNonQuery(filterStr);
//清除插入字段串
sb.Clear();
filterStr = "";
}
}
catch (Exception ex)
{
StreamWriter sw = File.AppendText("log.txt");
sw.WriteLine(DateTime.Now.ToLongDateString());
sw.WriteLine("begin");
sw.WriteLine(ex.Message);
sw.WriteLine(sb.ToString());
sw.WriteLine("end");
sw.WriteLine("");
sw.Close();
}
//执行完毕后,重新获取一次数据库的数据
dtWebNewsInfo = tri.SelReleaseInfo(DateTime.Now.ToString("yyyy-MM-dd HH:mm:ss"), DateTime.Now.AddDays(-15).ToString("yyyy-MM-dd HH:mm:ss"), "0 AND webName<>'百度'");
//MessageBox.Show("ok");
lbWeb.Text = "一轮搜索完毕！";
lbWeb.ForeColor = Color.Red;
}
private void TiebaWebSpider()
{
#region 处理关键字
MySqlCmd cmd = new MySqlCmd();
//得到关键字列表
DataTable dtkey;
DataTable dtParts;
if (selectKID == -1)
{
//全部
dtkey = cmd.GetTabel("select * from Keywords");
}
else
{
dtkey = cmd.GetTabel("select * from Keywords where kid=" + selectKID);
}
dtParts = cmd.GetTabel("SELECT * FROM partword");
#endregion
HtmlParse.Parse parse = new HtmlParse.Parse();
//parse.ReportCatchProcess += new HtmlParse.ReportCatchProcessEventHandler(Tieba_ReportCatchProcess);
//按关键字循环
for (int kw = 0; kw < dtkey.Rows.Count; kw++)
{
//处理关键字
if (selectKID != -1 && selectKwName != "全部")
{
if (dtkey.Rows[kw]["name"].ToString().Trim() != selectKwName) continue;
}
string keyword = dtkey.Rows[kw]["KeyWord"].ToString().Trim();
string keyTitle = dtkey.Rows[kw]["Name"].ToString().Trim();
int kid = 0;
int.TryParse(dtkey.Rows[kw]["kid"].ToString().Trim(), out kid);
#region 按关键字检索
//组成查询字串
//返回前60条数据，暂且不处理翻页
string url = "http://tieba.baidu.com/f/search/res?ie=utf-8&rn=60&qw=" + keyword;
string html = HtmlUtil.HttpGet(url, Encoding.Default);
#endregion
List<ModelReleaseInfo> mris = parse.ParseBaiduTieba(html, keyword, kid);
if (mris != null && mris.Count() > 0)
{
DataPersistenceControl.GetInstance().Add(mris);
}
//防止拉黑
Thread.Sleep(Interval50s);
}
}
private void GetBaiduInfo()
{
lbAll.Text = "";
lbAll.Visible = true;
MySqlCmd cmd = new MySqlCmd();
;
//得到关键字列表
DataTable dtkey = new DataTable();
dtkey = cmd.GetTabel("select * from Keywords");
dtParts = cmd.GetTabel("SELECT * FROM partword");
//链接的正则
string aa = "http://.[^\"]+";
string[] sDate;
sb = new StringBuilder();
sb.Append("");
//TbReleaseInfo ri = new TbReleaseInfo();
//按关键字循环
for (int kw = 0; kw < dtkey.Rows.Count; kw++)
{
lbAll.Text = "正在搜索关键字为<" + dtkey.Rows[kw]["KeyWord"].ToString().Trim() + ">的数据.";
lbAll.ForeColor = Color.DarkBlue;
//取得关键字
string keys = dtkey.Rows[kw]["KeyWord"].ToString().Trim();
//组成查询字串
string url = "http://www.baidu.com/s?wd=\"" + keys + "\"&rn=50";
//得到结果放在数组内
List<string> lis = new List<string>();
lis = HtmlUtil.GetElementsByClassList(getHtml(url, "utf-8"), "result");
//如果没取到,就结束本次循环
if (lis == null) return;
//webBrowser1.Navigate(url);
//循环时判断是否要验证
bool isThere = false;
for (int i = 0; i < lis.Count; i++)
{
if (Program.ProClose == true) break;
ModelReleaseInfo mri = new ModelReleaseInfo();
//发布日期的赋值
sDate = HtmlUtil.GetElementsByTagAndClass(lis[i], "span", "g");
if (sDate.Length <= 0) continue;
mri.ReleaseDate = HtmlUtil.NoHTML(sDate[0]);
mri.ReleaseDate = mri.ReleaseDate.Substring(mri.ReleaseDate.Length - 10, 10);
//判断日期
DateTime ddt;
if (DateTime.TryParse(mri.ReleaseDate, out ddt))
{
}
else
{
//百度的快照日期有时会是9位或8位,如果是这种情况,那么按规则去掉
mri.ReleaseDate = mri.ReleaseDate.Substring(1, 9);
if (DateTime.TryParse(mri.ReleaseDate, out ddt))
{
}
else
{
mri.ReleaseDate = mri.ReleaseDate.Substring(1, 8);
}
}
//处理日期
try
{
mri.ReleaseDate = DateTime.Parse(mri.ReleaseDate).ToString("yyyy-MM-dd HH:mm:ss");
}
catch (Exception ex)
{
StreamWriter sw = File.AppendText("log.txt");
sw.WriteLine(DateTime.Now.ToLongDateString());
sw.WriteLine("begin");
sw.WriteLine(ex.Message);
sw.WriteLine(sb.ToString());
sw.WriteLine("end");
sw.WriteLine("");
sw.Close();
}
//只拿取三天的内的数据
try
{
if (DateTime.Parse(mri.ReleaseDate) < DateTime.Now.AddDays(-3)) continue;
}
catch (Exception ex) { continue; }
try
{
//得到标题
mri.Title = HtmlUtil.NoHTML(HtmlUtil.GetElementsByTagName(lis[i], "h3")[0]);
string[] temp = HtmlUtil.GetElementsByClass(lis[i], "c-abstract");
//如果未取到内容部分,就跳出
if (temp.Length == 0)
continue;
mri.Contexts = HtmlUtil.NoHTML(temp[0]);
mri.InfoSource = HtmlUtil.GetListByHtml("", HtmlUtil.GetElementsByTagName(lis[i], "a")[0], aa)[0];
//去掉重复
if (isThere)
{
continue;
}
else
{
if (UrlThereare(mri.Title, this.dtqueryinfo, dtWebQueryInfo, false) != 0)
{ isThere = true; continue; }
}
mri.KeyWords = dtkey.Rows[kw]["KeyWord"].ToString().Trim();
mri.CollectDate = string.Format(DateTime.Now.ToString("yyyy-MM-dd HH:mm:ss"));
mri.Snapshot = "";
mri.ReleaseName = "";
mri.WebName = "百度";
mri.Pid = 0;
mri.Part = GetParts(mri.Contexts);
mri.Comments = 0;
mri.Reposts = 0;
DataRow dr = dtqueryinfo.NewRow();
if (dvAll.RowCount == 0)
{
dr[0] = 1;
}
else
{
dr[0] = int.Parse(dvAll.Rows[dvAll.RowCount - 1].Cells[0].Value.ToString()) + 1;
}
dr[1] = mri.Title;
dr[2] = mri.Contexts;
dr[3] = mri.ReleaseDate;
dr[4] = mri.InfoSource;
dr[5] = mri.KeyWords;
dr[6] = mri.ReleaseName;
dr[7] = mri.CollectDate;
dr[8] = mri.Snapshot;
dr[9] = mri.WebName;
dr[10] = mri.Pid;
dr[11] = mri.Part;
dr[12] = mri.Reposts;
dr[13] = mri.Comments;
dtqueryinfo.Rows.InsertAt(dr, 0);
if (dtqueryinfo.Rows.Count >= 500)
{
dtqueryinfo.Rows.RemoveAt(500);
}
dvAll.Refresh();
}
catch (Exception ex)
{
StreamWriter sw = File.AppendText("log.txt");
sw.WriteLine(DateTime.Now.ToLongDateString());
sw.WriteLine("begin");
sw.WriteLine(ex.Message);
sw.WriteLine(sb.ToString());
sw.WriteLine("end");
sw.WriteLine("");
sw.Close();
}
try
{
//得到插入语句
if (isThere)
{
continue;
}
else
{
sb.Append(tri.GetInsString(mri) + ";");
}
//每10次执行一次插入数据库
if (sb.ToString().Length != 0)
{
if (i % 10 == 0)
{
//执行插入
cmd.ExecuteNonQuery(sb.ToString());
//清除插入字段串
sb.Clear();
}
}
}
catch (Exception ex)
{
StreamWriter sw = File.AppendText("log.txt");
sw.WriteLine(DateTime.Now.ToLongDateString());
sw.WriteLine("begin");
sw.WriteLine(ex.Message);
sw.WriteLine(sb.ToString());
sw.WriteLine("end");
sw.WriteLine("");
sw.Close();
}
}
}
try
{
if (sb.ToString().Length != 0)
{
//执行插入
cmd.ExecuteNonQuery(sb.ToString());
//清除插入字段串
sb.Clear();
}
}
catch (Exception ex)
{
StreamWriter sw = File.AppendText("log.txt");
sw.WriteLine(DateTime.Now.ToLongDateString());
sw.WriteLine("begin");
sw.WriteLine(ex.Message);
sw.WriteLine(sb.ToString());
sw.WriteLine("end");
sw.WriteLine("");
sw.Close();
}
//执行完毕后,重新获取一次数据库的数据
dtWebQueryInfo = tri.SelReleaseInfo(DateTime.Now.ToString("yyyy-MM-dd HH:mm:ss"), DateTime.Now.AddDays(-15).ToString("yyyy-MM-dd HH:mm:ss"), "0 AND webName='百度'");
//MessageBox.Show("ok");
lbAll.Text = "一轮搜索完毕！";
lbAll.ForeColor = Color.Red;
}
private void WeiboWebSpider()
{
#region 处理关键字
MySqlCmd cmd = new MySqlCmd();
//得到关键字列表
DataTable dtkey;
DataTable dtParts;
if (selectKID == -1)
{
//全部
dtkey = cmd.GetTabel("select * from Keywords");
}
else
{
dtkey = cmd.GetTabel("select * from Keywords where kid=" + selectKID);
}
dtParts = cmd.GetTabel("SELECT * FROM partword");
#endregion
HtmlParse.Parse parse = new HtmlParse.Parse();
//parse.ReportCatchProcess += new HtmlParse.ReportCatchProcessEventHandler(Weibo_ReportCatchProcess);
List<ModelReleaseInfo> webDatas = new List<ModelReleaseInfo>();
//按关键字循环
for (int kw = 0; kw < dtkey.Rows.Count; kw++)
{
//处理关键字
if (selectKID != -1 && selectKwName != "全部")
{
if (dtkey.Rows[kw]["name"].ToString().Trim() != selectKwName) continue;
}
string keyword = dtkey.Rows[kw]["KeyWord"].ToString().Trim();
string keyTitle = dtkey.Rows[kw]["Name"].ToString().Trim();
int kid = 0;
int.TryParse(dtkey.Rows[kw]["kid"].ToString().Trim(), out kid);
#region 新浪微博检索
String encodeKey = CrawlHtml.UrlEncode(keyword);
string url = "http://s.weibo.com/weibo/" + encodeKey + "?topnav=1&wvr=6&b=1&page=1";
string html = HtmlUtil.HttpGet(url, Encoding.UTF8);
List<ModelReleaseInfo> mris = parse.ParseSinaWeibo(html, keyword, kid);
if (mris != null && mris.Count() > 0)
{
DataPersistenceControl.GetInstance().Add(mris);
}
//防止微博拉黑
Thread.Sleep(Interval30s);
#endregion
#region 中搜检索
for (int i = 0; i < 10; i++)
{
//组成查询字串
url = "";
html = "";
mris = null;
url = string.Format("http://t.zhongsou.com/wb?w={0}&b={1}", keyword, i + 1);
html = HtmlUtil.HttpGet(url, Encoding.Default);
mris = parse.ParseZhongsouWeibo(html, keyword, kid);
if (mris != null && mris.Count() > 0)
{
//写入数据库
DataPersistenceControl.GetInstance().Add(mris);
}
else
{
break;
}
//防止拉黑
Thread.Sleep(Interval2m);
}
#endregion
}
}
private void WeixinWebSpider()
{
#region 处理关键字
MySqlCmd cmd = new MySqlCmd();
//得到关键字列表
DataTable dtkey;
//DataTable dtParts;
//dtkey = cmd.GetTabel("select * from Keywords");
//dtParts = cmd.GetTabel("SELECT * FROM partword");
if (selectKID == -1)
{
//全部
dtkey = cmd.GetTabel("select * from Keywords");
}
else
{
dtkey = cmd.GetTabel("select * from Keywords where kid=" + selectKID);
}
#endregion
HtmlParse.Parse parse = new HtmlParse.Parse();
//parse.ReportCatchProcess += new HtmlParse.Parse.ReportCatchProcessEventHandler(Weixin_ReportCatchProcess);
List<ModelReleaseInfo> webDatas = new List<ModelReleaseInfo>();
//按关键字循环
for (int kw = 0; kw < dtkey.Rows.Count; kw++)
{
//处理关键字
if (selectKID != -1 && selectKwName != "全部")
{
if (dtkey.Rows[kw]["name"].ToString().Trim() != selectKwName) continue;
}
string keyword = dtkey.Rows[kw]["KeyWord"].ToString().Trim();
//string keyTitle = dtkey.Rows[kw]["Name"].ToString().Trim();
int kid = 0;
int.TryParse(dtkey.Rows[kw]["kid"].ToString().Trim(), out kid);
#region 按关键字检索
//组成查询字串
string url = "http://weixin.sogou.com/weixin?type=2&query=" + keyword + "&ie=utf8";
//string html = HtmlUtil.HttpGet(url, Encoding.UTF8);
CookieContainer cookies = new CookieContainer();
string strCookie = "";
string html = HtmlUtil.HttpGet(url, Encoding.UTF8, cookies, "weixin.sogou.com", ref strCookie);
#endregion
List<ModelReleaseInfo> mris = parse.ParseSogouWeixin(html, keyword, kid, cookies, strCookie);
if (mris != null && mris.Count() > 0)
{
DataPersistenceControl.GetInstance().Add(mris);
//webDatas.AddRange(mris);
//刷新界面
//RefreshDataGridView(6);
}
//防止拉黑（2分钟）
Thread.Sleep(Interval2m);
}
}
public DataTable GetLatestData(int pid)
{
DataPage firstPage = Finder.util.Comm.GetPageInfo();
//string sql = "Select * From ReleaseInfo where pid={0} and deleted=0 order by collectdate desc limit 0,100";
string sql = @"select b.Name eventname, a.* from releaseinfo a left join keywords b on a.keywords=b.KeyWord
where b.Name is not null and a.pid={0} and a.deleted=0 and a.uid between " + firstPage.CurrenPageStartUid + " and " + firstPage.CurrenPageEndUid
+ " order by a.collectdate desc limit 0,50";
sql = string.Format(sql, pid);
try
{
DataBaseServer.MySqlCmd dbobj = new MySqlCmd();
return dbobj.GetTabel(sql);
}
catch (Exception ex)
{
return null;
}
}
