public void StartWrite()
{
Action actionG = () =>
{
List <ModelReleaseInfo> data;
while (true)
{
if (Program.ProClose)
{
break;
}
if (Queue.TryDequeue(out data))
{
//// 写入数据库
#region 数据入库
try
{
StringBuilder sb = new StringBuilder();
sb.Append("");
TbReleaseInfo tri = new TbReleaseInfo();
MySqlCmd cmd = new MySqlCmd();
foreach (var mri in data)
{
#region 2016.11.16 加入精确匹配的判断
string keywords = mri.KeyWords;
string title = mri.Title;
string context = mri.Contexts;
if (!string.IsNullOrEmpty(keywords))
{
bool isFundTitle = true;
bool isFundContext = true;
string[] keyw = keywords.Split(' ');
if (keyw != null && keyw.Count() > 0)
{
foreach (string key in keyw)
{
if (title.IndexOf(key) < 0)
{
isFundTitle = false;
}
if (context.IndexOf(key) < 0)
{
isFundContext = false;
}
}
}
if (!isFundTitle && !isFundContext)
{
//如果标题或者内容没有匹配全部关键字则去掉该条数据
continue;
}
}
#endregion
if (tri.GetReleaseInfoCount(mri.InfoSource, mri.KeyWords) > 0)
{
continue;
}
string sql = tri.GetInsertStr(mri);
if (!sql.Trim().EndsWith(";"))
{
sql += sql + ";";
}
sb.Append(sql);
}
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)
{
Comm.WriteErrorLog(ex.Message);
Comm.WriteErrorLog(ex.StackTrace);
}
#endregion
log.Info("数据层写入数据库成功");
}
else
{
System.Threading.Thread.Sleep(1000);
}
}
};
Parallel.Invoke(actionG, actionG);
}
public static ModelReleaseInfo CrawlHtmlSource(string html, string url, DataTable dtkey, string sheng, string shi, string xian, string webName, string webInfo, int pid)
{
//string strURLformat = "https?://.[^\"]+";
Dictionary <string, int> events = new Dictionary <string, int>();
//创建数据对象
ModelReleaseInfo newsInfo = new ModelReleaseInfo();
try
{
newsInfo.Title = HtmlUtil.NoHTML(html);
//newsInfo.Title = html;
for (int j = 0; j < dtkey.Rows.Count; j++)
{
string[] keys = dtkey.Rows[j][4].ToString().Split(new char[] { ' ' });
if (!events.ContainsKey(dtkey.Rows[j][1].ToString() + "-" + dtkey.Rows[j][6].ToString()))
{
events.Add(dtkey.Rows[j][1].ToString() + "-" + dtkey.Rows[j][6].ToString(), 1);
foreach (string k in keys)
{
if (!html.ToLower().Contains(k.ToLower()))
{
events.Remove(dtkey.Rows[j][1].ToString() + "-" + dtkey.Rows[j][6].ToString());
break;
}
}
}
}
foreach (KeyValuePair <string, int> ev in events)
{
if (ev.Value == 1)
{
newsInfo.KeyWords += "," + ev.Key.Split(new char[] { '-' })[0] + "-" + int.Parse(ev.Key.Split(new char[] { '-' })[1]);
}
}
newsInfo.Contexts = HtmlUtil.NoHTML(webInfo);
//网站链接
newsInfo.InfoSource = url;
//关键字的设置
if (newsInfo.KeyWords == null || newsInfo.KeyWords.Length == 0)
{
for (int j = 0; j < dtkey.Rows.Count; j++)
{
//Application.DoEvents();
string[] keys = dtkey.Rows[j][4].ToString().Split(new char[] { ' ' });
if (!events.ContainsKey(dtkey.Rows[j][1].ToString() + "-" + dtkey.Rows[j][6].ToString()))
{
events.Add(dtkey.Rows[j][1].ToString() + "-" + dtkey.Rows[j][6].ToString(), 1);
foreach (string k in keys)
{
if (!newsInfo.Contexts.ToLower().Contains(k.ToLower()))
{
events.Remove(dtkey.Rows[j][1].ToString() + "-" + dtkey.Rows[j][6].ToString());
break;
}
}
}
}
foreach (KeyValuePair <string, int> ev in events)
{
if (ev.Value == 1)
{
newsInfo.KeyWords += "," + ev.Key.Split(new char[] { '-' })[0] + "-" + int.Parse(ev.Key.Split(new char[] { '-' })[1]);
}
}
}
//if (newsInfo.KeyWords.Length == 0) { continue; }
if (newsInfo.KeyWords != null)
{
newsInfo.KeyWords = newsInfo.KeyWords.Substring(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 = "";
newsInfo.Sheng = sheng == null ? "" : sheng;
newsInfo.Shi = shi == null ? "" : shi;
newsInfo.Xian = xian == null ? "" : xian;
//网站名
newsInfo.WebName = webName == null ? "" : webName;
//pid
newsInfo.Pid = pid;
//part正负判断
newsInfo.Part = GetParts(newsInfo.Contexts);
//reposts
newsInfo.Reposts = 0;
//comments
newsInfo.Comments = 0;
}
catch (Exception ex)
{
Comm.WriteErrorLog(ex.StackTrace);
}
return(newsInfo);
}
