private void 制任务ToolStripMenuItem_Click(object sender, EventArgs e)
{
int ID = Get_DataViewID();
DALTask dal = new DALTask();
int currentMaxId = dal.GetMaxId();
ModelTask model = dal.GetModelSingleTask(ID);
string currentTaskName = model.TaskName + "复制" + DateTime.Now.Millisecond;
model.ID = currentMaxId;
model.TaskName = currentTaskName;
dal.Add(model);
DALTaskLabel dalLable = new DALTaskLabel();
DataTable dt = dalLable.GetList(" TaskId=" + ID).Tables[0];
if (dt != null && dt.Rows.Count > 0)
{
foreach (DataRow dr in dt.Rows)
{
ModelTaskLabel modelLabel = dalLable.GetModel(StringHelper.Instance.SetNumber(dr["Id"].ToString()));
modelLabel.TaskID = currentMaxId;
dalLable.Add(modelLabel);
}
}
CreateDataFile(currentTaskName, currentMaxId);
this.ClassID = model.TaskClassID.Value;
Bind_DataList();
}
/// <summary>
/// 更新一条数据
/// </summary>
public bool Update(ModelTaskLabel model)
{
StringBuilder strSql = new StringBuilder();
strSql.Append("update [S_TaskLabel] set ");
strSql.Append("LabelName=@LabelName,");
strSql.Append("IsLoop=@IsLoop,");
strSql.Append("TestViewUrl=@TestViewUrl,");
strSql.Append("LabelNameCutRegex=@LabelNameCutRegex,");
strSql.Append("LabelHtmlRemove=@LabelHtmlRemove,");
strSql.Append("LabelRemove=@LabelRemove,");
strSql.Append("LabelReplace=@LabelReplace,");
strSql.Append("TaskID=@TaskID,");
strSql.Append("GuidNum=@GuidNum,");
strSql.Append("OrderID=@OrderID,");
strSql.Append("CreateTime=@CreateTime,");
strSql.Append("SpiderLabelPlugin=@SpiderLabelPlugin,");
strSql.Append("IsDownResource=@IsDownResource,");
strSql.Append("DownResourceExts=@DownResourceExts");
strSql.Append(" where ID=@ID ");
int rowsAffected = DbHelper.Execute(CommonHelper.SQLiteConnectionString, strSql.ToString(), model);
if (rowsAffected > 0)
{
return(true);
}
else
{
return(false);
}
}
/// <summary>
/// 测试单个标签
/// </summary>
/// <param name="itemLabel"></param>
/// <param name="pageContent"></param>
/// <returns></returns>
public string TestSingleLabel(ModelTaskLabel itemLabel, string pageContent)
{
var CutContent = string.Empty;
CutContent += "【" + itemLabel.LabelName + "】： ";
CutContent += GetLabelContent("测试", 10, "", itemLabel, pageContent, true);
return(CutContent);
}
public List <ModelTaskLabel> GetModelByTaskID(int TaskID)
{
List <ModelTaskLabel> list = new List <ModelTaskLabel>();
DataTable dt = this.GetList(" TaskID=" + TaskID + " Order by OrderID asc").Tables[0];
foreach (DataRow dr in dt.Rows)
{
ModelTaskLabel model = this.GetModel(int.Parse("0" + dr["ID"]));
list.Add(model);
}
return(list);
}
/// <summary>
/// 增加一条数据
/// </summary>
public int Add(ModelTaskLabel model)
{
StringBuilder strSql = new StringBuilder();
strSql.Append("insert into [S_TaskLabel] (");
strSql.Append("LabelName,IsLoop,TestViewUrl,LabelNameCutRegex,LabelHtmlRemove,LabelRemove,LabelReplace,TaskID,GuidNum,OrderID,CreateTime,SpiderLabelPlugin,IsDownResource,DownResourceExts)");
strSql.Append(" values (");
strSql.Append("@LabelName,@IsLoop,@TestViewUrl,@LabelNameCutRegex,@LabelHtmlRemove,@LabelRemove,@LabelReplace,@TaskID,@GuidNum,@OrderID,@CreateTime,@SpiderLabelPlugin,@IsDownResource,@DownResourceExts)");
object obj = DbHelper.Execute(CommonHelper.SQLiteConnectionString, strSql.ToString(), model);
if (obj == null)
{
return(0);
}
else
{
return(Convert.ToInt32(obj));
}
}
private void btnSubmit_Click(object sender, EventArgs e)
{
int ID = int.Parse("0" + this.txtID.Text);
ModelTaskLabel model = new ModelTaskLabel();
model.TaskID = TaskID;
model.LabelName = this.txtLabelName.Text;
model.LabelNameCutRegex = this.txtLabelNameCutRegex.Text.Replace("'", "''");
model.LabelRemove = string.Empty;
model.LabelReplace = string.Empty;
#region Html¹ýÂË
model.LabelHtmlRemove = string.Empty;
if (this.chkAllHtml.Checked)
{
model.LabelHtmlRemove += "all||||";
}
if (this.chkTable.Checked)
{
model.LabelHtmlRemove += "table||||";
}
if (this.chkFont.Checked)
{
model.LabelHtmlRemove += "font<span>||||";
}
if (this.chkhref.Checked)
{
model.LabelHtmlRemove += "a||||";
}
if (this.chkScript.Checked)
{
model.LabelHtmlRemove += "script||||";
}
if (model.LabelHtmlRemove.Trim() != "")
{
model.LabelHtmlRemove = model.LabelHtmlRemove.Remove(model.LabelHtmlRemove.Length - 4);
}
#endregion
#region ÄÚÈÝÅųý
foreach (ListViewItem item in this.listViewContentRemove.Items)
{
model.LabelRemove += item.SubItems[0].Text + "||" + item.SubItems[1].Text + "$$$$";
}
if (model.LabelRemove.Trim() != "")
{
model.LabelRemove = model.LabelRemove.Remove(model.LabelRemove.Length - 4);
}
#endregion
#region ÄÚÈÝÌæ»»
foreach (ListViewItem item in this.listViewContentReplace.Items)
{
model.LabelReplace += item.SubItems[0].Text + "||" + item.SubItems[1].Text + "$$$$";
}
if (model.LabelReplace.Trim() != "")
{
model.LabelReplace = model.LabelReplace.Remove(model.LabelReplace.Length - 4);
}
#endregion
model.IsLoop = this.chkLabelIsLoop.Checked ? 1 : 0;
model.SpiderLabelPlugin = this.cmbSpiderPlugin.Text;
model.IsDownResource = this.chkDownResource.Checked ? 1 : 0;
model.DownResourceExts = this.txtDownResourceExt.Text;
if (ID > 0)
{
model.ID = ID;
dal.Update(model);
}
else if (ID == 0)
{
int OrderID = dal.GetMaxOrderID(TaskID) + 1;
model.OrderID = OrderID;
ID = dal.Add(model);
}
ViewLabel?.Invoke(this, ev);
this.Hide();
this.Close();
}
private void btnTest_Click(object sender, EventArgs e)
{
this.txtLogView.Clear();
this.btnTest.Enabled = false;
var model = new ModelTaskLabel();
model.TestViewUrl = this.txtTestUrl.Text;
model.TaskID = TaskID;
model.LabelName = this.txtLabelName.Text;
model.LabelNameCutRegex = this.txtLabelNameCutRegex.Text.Replace("'", "''");
model.LabelRemove = string.Empty;
model.LabelReplace = string.Empty;
#region Html¹ýÂË
model.LabelHtmlRemove = string.Empty;
if (this.chkAllHtml.Checked)
{
model.LabelHtmlRemove += "all||||";
}
if (this.chkTable.Checked)
{
model.LabelHtmlRemove += "table||||";
}
if (this.chkFont.Checked)
{
model.LabelHtmlRemove += "font<span>||||";
}
if (this.chkhref.Checked)
{
model.LabelHtmlRemove += "a||||";
}
if (this.chkScript.Checked)
{
model.LabelHtmlRemove += "script||||";
}
if (model.LabelHtmlRemove.Trim() != "")
{
model.LabelHtmlRemove = model.LabelHtmlRemove.Remove(model.LabelHtmlRemove.Length - 4);
}
#endregion
#region ÄÚÈÝÅųý
foreach (ListViewItem item in this.listViewContentRemove.Items)
{
model.LabelRemove += item.SubItems[0].Text + "||" + item.SubItems[1].Text + "$$$$";
}
if (model.LabelRemove.Trim() != "")
{
model.LabelRemove = model.LabelRemove.Remove(model.LabelRemove.Length - 4);
}
#endregion
#region ÄÚÈÝÌæ»»
foreach (ListViewItem item in this.listViewContentReplace.Items)
{
model.LabelReplace += item.SubItems[0].Text + "||" + item.SubItems[1].Text + "$$$$";
}
if (model.LabelReplace.Trim() != "")
{
model.LabelReplace = model.LabelReplace.Remove(model.LabelReplace.Length - 4);
}
#endregion
model.IsLoop = this.chkLabelIsLoop.Checked ? 1 : 0;
model.SpiderLabelPlugin = this.cmbSpiderPlugin.Text;
model.IsDownResource = this.chkDownResource.Checked ? 1 : 0;
model.DownResourceExts = this.txtDownResourceExt.Text;
var task = new System.Threading.Tasks.TaskFactory().StartNew(() => {
var pageContent = CommonHelper.getPageContent(model.TestViewUrl, PageEncode);
var spider = new SpiderViewHelper();
var s = spider.TestSingleLabel(model, pageContent);
this.Invoke(new MethodInvoker(() => {
this.btnTest.Enabled = true;
this.txtLogView.AppendText(s);
this.txtLogView.AppendText("\r\n");
}));
});
}
/// <summary>
/// ±à¼±êÇ©
/// </summary>
private void Bind_DataEdit()
{
DALTaskLabel dal = new DALTaskLabel();
ListView.SelectedListViewItemCollection item = (ListView.SelectedListViewItemCollection)EditItem;
int ID = int.Parse("0" + item[0].Tag);
ModelTaskLabel model = dal.GetModel(ID);
this.txtID.Text = ID.ToString();
this.txtLabelName.Text = model.LabelName;
this.txtLabelNameCutRegex.Text = model.LabelNameCutRegex;
this.chkScript.Checked = false;
this.chkhref.Checked = false;
if (!string.IsNullOrEmpty(model.LabelHtmlRemove))
{
string[] arr = model.LabelHtmlRemove.Split(new string[] { "||||" }, StringSplitOptions.RemoveEmptyEntries);
foreach (string str in arr)
{
if (str == "all")
{
this.chkAllHtml.Checked = true;
}
else if (str == "table")
{
this.chkTable.Checked = true;
}
else if (str == "font<span>")
{
this.chkFont.Checked = true;
}
else if (str == "a")
{
this.chkhref.Checked = true;
}
else if (str == "script")
{
this.chkScript.Checked = true;
}
}
}
#region ÄÚÈÝÅųý
if (!string.IsNullOrEmpty(model.LabelRemove))
{
foreach (string str in model.LabelRemove.Split(new string[] { "$$$$" }, StringSplitOptions.RemoveEmptyEntries))
{
string[] aa = str.Split(new string[] { "||" }, StringSplitOptions.None);
ListViewItem lvi = new ListViewItem(aa[0]);
lvi.SubItems.Add(aa[1]);
this.listViewContentRemove.Items.Add(lvi);
}
}
#endregion
#region ÄÚÈÝÌæ»»
if (!string.IsNullOrEmpty(model.LabelReplace))
{
foreach (string str in model.LabelReplace.Split(new string[] { "$$$$" }, StringSplitOptions.RemoveEmptyEntries))
{
string[] aa = str.Split(new string[] { "||" }, StringSplitOptions.None);
ListViewItem lvi = new ListViewItem(aa[0]);
lvi.SubItems.Add(aa[1]);
this.listViewContentReplace.Items.Add(lvi);
}
}
#endregion
this.chkLabelIsLoop.Checked = model.IsLoop == 1 ? true : false;
this.cmbSpiderPlugin.Text = model.SpiderLabelPlugin == string.Empty ? "²»Ê¹Óòå¼þ" : model.SpiderLabelPlugin;
this.chkDownResource.Checked = model.IsDownResource == 1 ? true : false;
this.txtDownResourceExt.Text = model.DownResourceExts;
}
/// <summary>
/// 增加一条数据
/// </summary>
public int Add(ModelTaskLabel model)
{
StringBuilder strSql = new StringBuilder();
StringBuilder strSql1 = new StringBuilder();
StringBuilder strSql2 = new StringBuilder();
if (model.LabelName != null)
{
strSql1.Append("LabelName,");
strSql2.Append("'" + model.LabelName + "',");
}
if (model.LabelSource != null)
{
strSql1.Append("LabelSource,");
strSql2.Append(model.LabelSource + ",");
}
if (model.LabelNameCutRegex != null)
{
strSql1.Append("LabelNameCutRegex,");
strSql2.Append("'" + model.LabelNameCutRegex + "',");
}
if (model.LabelHtmlRemove != null)
{
strSql1.Append("LabelHtmlRemove,");
strSql2.Append("'" + model.LabelHtmlRemove + "',");
}
if (model.LblHtmlRemove != null)
{
strSql1.Append("LblHtmlRemove,");
strSql2.Append("'" + model.LblHtmlRemove + "',");
}
if (model.LabelRemove != null)
{
strSql1.Append("LabelRemove,");
strSql2.Append("'" + model.LabelRemove + "',");
}
if (model.LabelReplace != null)
{
strSql1.Append("LabelReplace,");
strSql2.Append("'" + model.LabelReplace + "',");
}
if (model.TaskID != null)
{
strSql1.Append("TaskID,");
strSql2.Append("" + model.TaskID + ",");
}
if (model.GuidNum != null)
{
strSql1.Append("GuidNum,");
strSql2.Append("'" + model.GuidNum + "',");
}
if (model.OrderID != null)
{
strSql1.Append("OrderID,");
strSql2.Append("" + model.OrderID + ",");
}
if (model.CreateTime != null)
{
strSql1.Append("CreateTime,");
strSql2.Append("'" + model.CreateTime + "',");
}
//=============================================2012 2-6
if (model.IsLoop != null)
{
strSql1.Append("IsLoop,");
strSql2.Append("" + model.IsLoop + ",");
}
if (model.IsNoNull != null)
{
strSql1.Append("IsNoNull,");
strSql2.Append("" + model.IsNoNull + ",");
}
if (model.IsLinkUrl != null)
{
strSql1.Append("IsLinkUrl,");
strSql2.Append("" + model.IsLinkUrl + ",");
}
if (model.IsPager != null)
{
strSql1.Append("IsPager,");
strSql2.Append("" + model.IsPager + ",");
}
if (model.LabelValueLinkUrlRegex != null)
{
strSql1.Append("LabelValueLinkUrlRegex,");
strSql2.Append("'" + model.LabelValueLinkUrlRegex + "',");
}
if (model.LabelValuePagerRegex != null)
{
strSql1.Append("LabelValuePagerRegex,");
strSql2.Append("'" + model.LabelValuePagerRegex + "',");
}
//
if (model.SpiderLabelPlugin != null)
{
strSql1.Append("SpiderLabelPlugin,");
strSql2.Append("'" + model.SpiderLabelPlugin + "',");
}
//
if (model.IsDownResource != null)
{
strSql1.Append("IsDownResource,");
strSql2.Append("" + model.IsDownResource + ",");
}
if (model.DownResourceExts != null)
{
strSql1.Append("DownResourceExts,");
strSql2.Append("'" + model.DownResourceExts + "',");
}
strSql.Append("insert into S_TaskLabel(");
strSql.Append(strSql1.ToString().Remove(strSql1.Length - 1));
strSql.Append(")");
strSql.Append(" values (");
strSql.Append(strSql2.ToString().Remove(strSql2.Length - 1));
strSql.Append(")");
strSql.Append(";select LAST_INSERT_ROWID()");
object obj = SQLiteHelper.ExecuteScalar(dbStr, strSql.ToString());
if (obj == null)
{
return(0);
}
else
{
return(Convert.ToInt32(obj));
}
}
/// <summary>
/// 得到一个对象实体
/// </summary>
public ModelTaskLabel GetModel(int ID)
{
StringBuilder strSql = new StringBuilder();
strSql.Append("select ");
strSql.Append(" * ");
strSql.Append(" from S_TaskLabel ");
strSql.Append(" where ID=" + ID + "");
ModelTaskLabel model = new ModelTaskLabel();
DataSet ds = SQLiteHelper.Query1(dbStr, strSql.ToString());
if (ds.Tables[0].Rows.Count > 0)
{
if (ds.Tables[0].Rows[0]["ID"].ToString() != "")
{
model.ID = int.Parse(ds.Tables[0].Rows[0]["ID"].ToString());
}
if (ds.Tables[0].Rows[0]["LabelName"] != null)
{
model.LabelName = ds.Tables[0].Rows[0]["LabelName"].ToString();
}
if (ds.Tables[0].Rows[0]["LabelSource"] != null)
{
model.LabelSource = int.Parse(ds.Tables[0].Rows[0]["LabelSource"].ToString());
}
if (ds.Tables[0].Rows[0]["LabelNameCutRegex"] != null)
{
model.LabelNameCutRegex = ds.Tables[0].Rows[0]["LabelNameCutRegex"].ToString();
}
if (ds.Tables[0].Rows[0]["LabelHtmlRemove"] != null)
{
model.LabelHtmlRemove = ds.Tables[0].Rows[0]["LabelHtmlRemove"].ToString();
}
if (ds.Tables[0].Rows[0]["LblHtmlRemove"] != null)
{
model.LblHtmlRemove = ds.Tables[0].Rows[0]["LblHtmlRemove"].ToString();
}
if (ds.Tables[0].Rows[0]["LabelRemove"] != null)
{
model.LabelRemove = ds.Tables[0].Rows[0]["LabelRemove"].ToString();
}
if (ds.Tables[0].Rows[0]["LabelReplace"] != null)
{
model.LabelReplace = ds.Tables[0].Rows[0]["LabelReplace"].ToString();
}
if (ds.Tables[0].Rows[0]["TaskID"].ToString() != "")
{
model.TaskID = int.Parse(ds.Tables[0].Rows[0]["TaskID"].ToString());
}
if (ds.Tables[0].Rows[0]["GuidNum"] != null)
{
model.GuidNum = ds.Tables[0].Rows[0]["GuidNum"].ToString();
}
if (ds.Tables[0].Rows[0]["OrderID"].ToString() != "")
{
model.OrderID = int.Parse(ds.Tables[0].Rows[0]["OrderID"].ToString());
}
if (ds.Tables[0].Rows[0]["CreateTime"] != null)
{
model.CreateTime = ds.Tables[0].Rows[0]["CreateTime"].ToString();
}
//=================================2012 2-6
if (ds.Tables[0].Rows[0]["IsLoop"] != null && ds.Tables[0].Rows[0]["IsLoop"].ToString() != "")
{
model.IsLoop = int.Parse(ds.Tables[0].Rows[0]["IsLoop"].ToString());
}
if (ds.Tables[0].Rows[0]["IsNoNull"] != null && ds.Tables[0].Rows[0]["IsNoNull"].ToString() != "")
{
model.IsNoNull = int.Parse(ds.Tables[0].Rows[0]["IsNoNull"].ToString());
}
if (ds.Tables[0].Rows[0]["IsLinkUrl"] != null && ds.Tables[0].Rows[0]["IsLinkUrl"].ToString() != "")
{
model.IsLinkUrl = int.Parse(ds.Tables[0].Rows[0]["IsLinkUrl"].ToString());
}
if (ds.Tables[0].Rows[0]["IsPager"] != null && ds.Tables[0].Rows[0]["IsPager"].ToString() != "")
{
model.IsPager = int.Parse(ds.Tables[0].Rows[0]["IsPager"].ToString());
}
if (ds.Tables[0].Rows[0]["LabelValueLinkUrlRegex"] != null)
{
model.LabelValueLinkUrlRegex = ds.Tables[0].Rows[0]["LabelValueLinkUrlRegex"].ToString();
}
if (ds.Tables[0].Rows[0]["LabelValuePagerRegex"] != null)
{
model.LabelValuePagerRegex = ds.Tables[0].Rows[0]["LabelValuePagerRegex"].ToString();
}
//
if (ds.Tables[0].Rows[0]["SpiderLabelPlugin"] != null)
{
model.SpiderLabelPlugin = ds.Tables[0].Rows[0]["SpiderLabelPlugin"].ToString();
}
//
if (ds.Tables[0].Rows[0]["IsDownResource"] != null && ds.Tables[0].Rows[0]["IsDownResource"].ToString() != "")
{
model.IsDownResource = int.Parse(ds.Tables[0].Rows[0]["IsDownResource"].ToString());
}
if (ds.Tables[0].Rows[0]["DownResourceExts"] != null)
{
model.DownResourceExts = ds.Tables[0].Rows[0]["DownResourceExts"].ToString();
}
return(model);
}
else
{
return(null);
}
}
/// <summary>
/// 更新一条数据
/// </summary>
public bool Update(ModelTaskLabel model)
{
StringBuilder strSql = new StringBuilder();
strSql.Append("update S_TaskLabel set ");
if (model.LabelName != null)
{
strSql.Append("LabelName='" + model.LabelName + "',");
}
if (model.LabelSource != null)
{
strSql.Append("LabelSource=" + model.LabelSource + ",");
}
if (model.LabelNameCutRegex != null)
{
strSql.Append("LabelNameCutRegex='" + model.LabelNameCutRegex + "',");
}
if (model.LabelHtmlRemove != null)
{
strSql.Append("LabelHtmlRemove='" + model.LabelHtmlRemove + "',");
}
if (model.LblHtmlRemove != null)
{
strSql.Append("LblHtmlRemove='" + model.LblHtmlRemove.Replace("'", "''") + "',");
}
if (model.LabelRemove != null)
{
strSql.Append("LabelRemove='" + model.LabelRemove + "',");
}
if (model.LabelReplace != null)
{
strSql.Append("LabelReplace='" + model.LabelReplace + "',");
}
if (model.TaskID != null)
{
strSql.Append("TaskID=" + model.TaskID + ",");
}
if (model.GuidNum != null)
{
strSql.Append("GuidNum='" + model.GuidNum + "',");
}
//=========================================2012 2-6
if (model.IsLoop != null)
{
strSql.Append("IsLoop=" + model.IsLoop + ",");
}
if (model.IsNoNull != null)
{
strSql.Append("IsNoNull=" + model.IsNoNull + ",");
}
if (model.IsLinkUrl != null)
{
strSql.Append("IsLinkUrl=" + model.IsLinkUrl + ",");
}
if (model.IsPager != null)
{
strSql.Append("IsPager=" + model.IsPager + ",");
}
if (model.LabelValueLinkUrlRegex != null)
{
strSql.Append("LabelValueLinkUrlRegex='" + model.LabelValueLinkUrlRegex + "',");
}
if (model.LabelValuePagerRegex != null)
{
strSql.Append("LabelValuePagerRegex='" + model.LabelValuePagerRegex + "',");
}
//
if (model.SpiderLabelPlugin != null)
{
strSql.Append("SpiderLabelPlugin='" + model.SpiderLabelPlugin + "',");
}
//
if (model.IsDownResource != null)
{
strSql.Append("IsDownResource=" + model.IsDownResource + ",");
}
if (model.DownResourceExts != null)
{
strSql.Append("DownResourceExts='" + model.DownResourceExts + "',");
}
int n = strSql.ToString().LastIndexOf(",");
strSql.Remove(n, 1);
strSql.Append(" where ID=" + model.ID + "");
int rowsAffected = SQLiteHelper.Execute(dbStr, strSql.ToString());
if (rowsAffected > 0)
{
return(true);
}
else
{
return(false);
}
}
/// <summary>
/// 获取标签内容
/// </summary>
/// <param name="taskName"></param>
/// <param name="collectionContentStepTime"></param>
/// <param name="spiderViewUrl"></param>
/// <param name="itemTaskLebel"></param>
/// <param name="pageContent"></param>
/// <param name="isTest"></param>
/// <returns></returns>
private string GetLabelContent(string taskName, int collectionContentStepTime, string spiderViewUrl, ModelTaskLabel itemTaskLebel, string pageContent, bool isTest = false)
{
var remoteViewUrl = itemTaskLebel.TestViewUrl;
if (string.IsNullOrEmpty(itemTaskLebel.TestViewUrl))
{
remoteViewUrl = spiderViewUrl;
}
string regContent = HtmlHelper.Instance.ParseCollectionStrings(itemTaskLebel.LabelNameCutRegex);
regContent = CommonHelper.ReplaceSystemRegexTag(regContent);
string CutContent = CollectionHelper.Instance.CutStr(pageContent, regContent)[0];
#region 载资源
var imgTag = ImageDownHelper.GetImgTag(CutContent);
if (itemTaskLebel.IsDownResource == 1)
{
string[] imgExtArr = itemTaskLebel.DownResourceExts.Split(new string[] { ";" }, StringSplitOptions.RemoveEmptyEntries);
var downImgPath = AppDomain.CurrentDomain.BaseDirectory + "Data\\Collection\\" + taskName + "\\Images\\";
int ii = 1;
foreach (var img in imgTag)
{
var remoteImg = CollectionHelper.Instance.FormatUrl(remoteViewUrl, img);
var newImg = DateTime.Now.ToString("yyyyMMddHHmmssffffff") + "_" + ii + ".jpg";
if (!string.IsNullOrEmpty(itemTaskLebel.DownResourceExts))
{
var imgExt = remoteImg.Substring(remoteImg.LastIndexOf("."));
if (imgExtArr.SingleOrDefault(x => x.ToLower() == imgExt.ToLower()) != imgExt.ToLower())
{
continue;
}
}
CutContent = CutContent.Replace(img, downImgPath + newImg);
if (!isTest)
{
QueueImgHelper.AddImg(Model.ID, downImgPath + newImg, remoteImg, collectionContentStepTime);
}
ii++;
}
}
else
{
foreach (var img in imgTag)
{
var remoteImg = CollectionHelper.Instance.FormatUrl(remoteViewUrl, img);
CutContent = CutContent.Replace(img, remoteImg);
}
}
#endregion
#region 结果为循环
if (itemTaskLebel.IsLoop == 1)
{
string[] LabelString = CollectionHelper.Instance.CutStr(pageContent, regContent);
foreach (string s in LabelString)
{
CutContent += s + "$$$$";
}
int n = CutContent.LastIndexOf("$$$$");
CutContent = CutContent.Remove(n, 4);
}
#endregion
#region 过滤Html
if (!string.IsNullOrEmpty(itemTaskLebel.LabelHtmlRemove))
{
string[] arr = itemTaskLebel.LabelHtmlRemove.Split(new string[] { "||||" }, StringSplitOptions.RemoveEmptyEntries);
foreach (string str in arr)
{
if (str == "all")
{
CutContent = CollectionHelper.Instance.NoHtml(CutContent);
break;
}
else if (str == "table")
{
CutContent = CollectionHelper.Instance.ScriptHtml(CutContent, "table", 2);
}
else if (str == "font<span>")
{
CutContent = CollectionHelper.Instance.ScriptHtml(CutContent, "font", 3);
CutContent = CollectionHelper.Instance.ScriptHtml(CutContent, "span", 3);
}
else if (str == "a")
{
CutContent = CollectionHelper.Instance.ScriptHtml(CutContent, "a", 3);
}
}
}
#endregion
#region 排除字符
if (!string.IsNullOrEmpty(itemTaskLebel.LabelRemove))
{
foreach (string str in itemTaskLebel.LabelRemove.Split(new string[] { "$$$$" }, StringSplitOptions.RemoveEmptyEntries))
{
string[] ListStr = str.Split(new string[] { "||" }, StringSplitOptions.RemoveEmptyEntries);
if (ListStr[1] == "1")
{
CutContent = CollectionHelper.RemoveHtml(CutContent, ListStr[0]);
}
else
{
CutContent = CutContent.Replace(ListStr[0], "");
}
}
}
#endregion
#region 替换字符
if (!string.IsNullOrEmpty(itemTaskLebel.LabelReplace))
{
foreach (string str in itemTaskLebel.LabelReplace.Split(new string[] { "$$$$" }, StringSplitOptions.RemoveEmptyEntries))
{
string[] ListStr = str.Split(new string[] { "||" }, StringSplitOptions.RemoveEmptyEntries);
CutContent = CutContent.Replace(ListStr[0], ListStr[1]);
}
}
#endregion
#region 加载插件
string SpiderLabelPlugin = itemTaskLebel.SpiderLabelPlugin;
if (SpiderLabelPlugin != "不使用插件" && !string.IsNullOrEmpty(SpiderLabelPlugin))
{
CutContent = PythonExtHelper.RunPython(PluginUtility.SpiderContentPluginPath + SpiderLabelPlugin, new object[] { remoteViewUrl, CutContent });
}
#endregion
return(CutContent);
}
private void btnSubmit_Click(object sender, EventArgs e)
{
int ID = int.Parse("0" + this.txtID.Text);
ModelTaskLabel model = new ModelTaskLabel();
model.TaskID = TaskID;
model.LabelName = this.txtLabelName.Text;
model.LabelSource = this.radioBtnSource.Checked ? 1 : 0;
model.LabelNameCutRegex = this.txtLabelNameCutRegex.Text.Replace("'", "''");
model.LabelRemove = string.Empty;
model.LabelReplace = string.Empty;
#region Html过滤
model.LabelHtmlRemove = string.Empty;
if (this.chkAllHtml.Checked)
{
model.LabelHtmlRemove += "all||||";
}
if (this.chkTable.Checked)
{
model.LabelHtmlRemove += "table||||";
}
if (this.chkFont.Checked)
{
model.LabelHtmlRemove += "font<span>||||";
}
if (this.chkhref.Checked)
{
model.LabelHtmlRemove += "a||||";
}
if (this.chkScript.Checked)
{
model.LabelHtmlRemove += "script||||";
}
if (model.LabelHtmlRemove.Trim() != "")
{
model.LabelHtmlRemove = model.LabelHtmlRemove.Remove(model.LabelHtmlRemove.Length - 4);
}
#endregion
#region 标签内容过滤
model.LblHtmlRemove = string.Empty;
if (this.chkTableLbl.Checked)
{
model.LblHtmlRemove += "table||||";
}
if (this.chkScriptLbl.Checked)
{
model.LblHtmlRemove += "script||||";
}
if (this.chkHrefLbl.Checked)
{
model.LblHtmlRemove += "a||||";
}
if (this.chkSpanLbl.Checked)
{
model.LblHtmlRemove += "span||||";
}
if (this.chkPLbl.Checked)
{
model.LblHtmlRemove += "p||||";
}
if (this.chkIFrameLbl.Checked)
{
model.LblHtmlRemove += "iframe||||";
}
if (this.chkFontLbl.Checked)
{
model.LblHtmlRemove += "font||||";
}
if (this.chkStyleLbl.Checked)
{
model.LblHtmlRemove += "style||||";
}
if (this.chkDivLbl.Checked)
{
model.LblHtmlRemove += "div||||";
}
if (this.chkRemarkLbl.Checked)
{
model.LblHtmlRemove += "remark||||";
}
if (model.LblHtmlRemove.Trim() != "")
{
model.LblHtmlRemove = model.LblHtmlRemove.Remove(model.LblHtmlRemove.Length - 4);
}
#endregion
#region 内容排除
foreach (ListViewItem item in this.listViewContentRemove.Items)
{
//model.LabelRemove += item.SubItems[0].Text + "$$$$";
model.LabelRemove += item.SubItems[0].Text + "||" + item.SubItems[1].Text + "$$$$";
}
if (model.LabelRemove.Trim() != "")
{
model.LabelRemove = model.LabelRemove.Remove(model.LabelRemove.Length - 4);
}
#endregion
#region 内容替换
foreach (ListViewItem item in this.listViewContentReplace.Items)
{
model.LabelReplace += item.SubItems[0].Text + "||" + item.SubItems[1].Text + "$$$$";
}
if (model.LabelReplace.Trim() != "")
{
model.LabelReplace = model.LabelReplace.Remove(model.LabelReplace.Length - 4);
}
#endregion
model.IsLoop = this.chkLabelIsLoop.Checked ? 1 : 0;
model.IsNoNull = this.chkLabelNoNull.Checked ? 1 : 0;
model.IsLinkUrl = this.chkLabelIsLinkUrl.Checked ? 1 : 0;
model.IsPager = this.chkLabelIsPager.Checked ? 1 : 0;
model.LabelValueLinkUrlRegex = this.txtLabelValueIsLinkUrlRegex.Text.Replace("'", "''");
model.LabelValuePagerRegex = this.txtLabelValueIsPagerRegex.Text.Replace("'", "''");
//
model.SpiderLabelPlugin = this.cmbSpiderPlugin.Text;
//下载文件
model.IsDownResource = this.chkDownResource.Checked ? 1 : 0;
model.DownResourceExts = this.txtDownResourceExt.Text;
if (ID > 0)
{
model.ID = ID;
dal.Update(model);
}
else if (ID == 0)
{
//获取最大的排序ID
int OrderID = dal.GetMaxOrderID(TaskID) + 1;
model.OrderID = OrderID;
ID = dal.Add(model);
}
if (ViewLabel != null)
{
ViewLabel(this, ev);
}
this.Hide();
this.Close();
}
/// <summary>
/// 编辑标签
/// </summary>
private void Bind_DataEdit()
{
DALTaskLabel dal = new DALTaskLabel();
ListView.SelectedListViewItemCollection item = (ListView.SelectedListViewItemCollection)EditItem;
int ID = int.Parse("0" + item[0].Tag);
ModelTaskLabel model = dal.GetModel(ID);
this.txtID.Text = ID.ToString();
this.txtLabelName.Text = model.LabelName;
if (model.LabelSource == 1)
{
this.radioBtnSource.Checked = true;
}
else
{
this.radioBtnUrl.Checked = true;
}
this.txtLabelNameCutRegex.Text = model.LabelNameCutRegex;
this.chkScript.Checked = false;
this.chkhref.Checked = false;
if (!string.IsNullOrEmpty(model.LabelHtmlRemove))
{
string[] arr = model.LabelHtmlRemove.Split(new string[] { "||||" }, StringSplitOptions.RemoveEmptyEntries);
foreach (string str in arr)
{
if (str == "all")
{
this.chkAllHtml.Checked = true;
}
else if (str == "table")
{
this.chkTable.Checked = true;
}
else if (str == "font<span>")
{
this.chkFont.Checked = true;
}
else if (str == "a")
{
this.chkhref.Checked = true;
}
else if (str == "script")
{
this.chkScript.Checked = true;
}
}
}
if (!string.IsNullOrEmpty(model.LblHtmlRemove))
{
string[] arr = model.LblHtmlRemove.Split(new string[] { "||||" }, StringSplitOptions.RemoveEmptyEntries);
foreach (string str in arr)
{
if (str == "table")
{
this.chkTableLbl.Checked = true;
}
else if (str == "script")
{
this.chkScriptLbl.Checked = true;
}
else if (str == "a")
{
this.chkHrefLbl.Checked = true;
}
else if (str == "span")
{
this.chkSpanLbl.Checked = true;
}
else if (str == "p")
{
this.chkPLbl.Checked = true;
}
else if (str == "iframe")
{
this.chkIFrameLbl.Checked = true;
}
else if (str == "font")
{
this.chkFontLbl.Checked = true;
}
else if (str == "style")
{
this.chkStyleLbl.Checked = true;
}
else if (str == "div")
{
this.chkDivLbl.Checked = true;
}
else if (str == "remark")
{
this.chkRemarkLbl.Checked = true;
}
}
}
#region 内容排除
if (!string.IsNullOrEmpty(model.LabelRemove))
{
foreach (string str in model.LabelRemove.Split(new string[] { "$$$$" }, StringSplitOptions.RemoveEmptyEntries))
{
//ListViewItem lvi = new ListViewItem(str);
//this.listViewContentRemove.Items.Add(lvi);
string[] aa = str.Split(new string[] { "||" }, StringSplitOptions.None);
ListViewItem lvi = new ListViewItem(aa[0]);
lvi.SubItems.Add(aa[1]);
this.listViewContentRemove.Items.Add(lvi);
}
}
#endregion
#region 内容替换
if (!string.IsNullOrEmpty(model.LabelReplace))
{
foreach (string str in model.LabelReplace.Split(new string[] { "$$$$" }, StringSplitOptions.RemoveEmptyEntries))
{
string[] aa = str.Split(new string[] { "||" }, StringSplitOptions.None);
ListViewItem lvi = new ListViewItem(aa[0]);
lvi.SubItems.Add(aa[1]);
this.listViewContentReplace.Items.Add(lvi);
}
}
#endregion
this.chkLabelIsLoop.Checked = model.IsLoop == 1 ? true : false;
this.chkLabelNoNull.Checked = model.IsNoNull == 1 ? true : false;
this.chkLabelIsLinkUrl.Checked = model.IsLinkUrl == 1 ? true : false;
this.chkLabelIsPager.Checked = model.IsPager == 1 ? true : false;
this.txtLabelValueIsLinkUrlRegex.Text = model.LabelValueLinkUrlRegex;
this.txtLabelValueIsPagerRegex.Text = model.LabelValuePagerRegex;
this.cmbSpiderPlugin.Text = model.SpiderLabelPlugin == string.Empty ? "不使用插件" : model.SpiderLabelPlugin;
//下载资源
this.chkDownResource.Checked = model.IsDownResource == 1 ? true : false;
this.txtDownResourceExt.Text = model.DownResourceExts;
}
