public GeckofxWebbrowerFrm(TaskInfoEntity taskInfoEntity, DataGridView dgv, GeckofxWebbrowerType geckofxWebbrowerType, string URL = null)
{
InitializeComponent();
this.taskInfoEntity = taskInfoEntity;
this.dgv = dgv;
this.geckofxWebbrowerType = geckofxWebbrowerType;
this.URL = URL;
}
private string URL; //要执行的URL链接
//通用
public GeckofxWebbrower(Control control, DataGridView dgv, TaskInfoEntity taskInfoEntity, GeckofxWebbrowerType geckofxType, string URL)
{
//GeckoPreferences.User["gfx.font_rendering.graphite.enabled"] = false;
geckoWebBrowser = new GeckoWebBrowser();
geckoWebBrowser.Parent = control;
geckoWebBrowser.Dock = DockStyle.Fill;
this.taskInfoEntity = taskInfoEntity;
this.control = control;
this.dgv = dgv;
this.geckofxType = geckofxType; // 浏览类型
ruleStyle = new RuleStyle();
xpathHelper = new XpathHelper();
this.URL = URL;
init();
}
public void InsertDataGridRow(GeckofxWebbrowerType geckofxWebbrowerType, DataGridView dgv, GeckoWebBrowser geckoWebBrowser, params GeckoElement[] ele)
{
List <GeckoHtmlElement> geckoHtmlElementList = null;
//设置xpath 获取html信息
if (ele.Length == 1)
{
var xpathTxt = xpathHelper.GetSmallXpath(ele[0]);
var xresult = geckoWebBrowser.DomDocument.EvaluateXPath(xpathTxt);
var nodes = xresult.GetNodes();
var elements = nodes.Select(x => x as GeckoElement).ToArray();
geckoHtmlElementList = XpathHelper.FindHtmlTxt(true, elements);
}
else
{
geckoHtmlElementList = XpathHelper.FindHtmlTxt(true, ele);
}
if (geckofxWebbrowerType.Equals(GeckofxWebbrowerType.General))
{
foreach (GeckoHtmlElement element in geckoHtmlElementList)
{
DataGridViewRow row = new DataGridViewRow();
//文本
DataGridViewTextBoxCell fieldsNameCell = new DataGridViewTextBoxCell();
fieldsNameCell.Value = "字段" + (count++);
row.Cells.Add(fieldsNameCell);
DataGridViewTextBoxCell fieldsContentCell = new DataGridViewTextBoxCell();
fieldsContentCell.Value = element.TextContent;
row.Cells.Add(fieldsContentCell);
//下拉框
DataGridViewTextBoxCell fieldsTypeCell = new DataGridViewTextBoxCell();
fieldsTypeCell.Value = "抓取文本";
row.Cells.Add(fieldsTypeCell);
DataGridViewImageCell deleteCell = new DataGridViewImageCell();
deleteCell.Value = Image.FromFile(@"E:\Project\C#\SimpleCrawlProject\SimpleCrawlApp\Resources\1108658.png");
row.Cells.Add(deleteCell);
//隐藏浏览器句柄
DataGridViewTextBoxCell visibleGeckofxElementCell = new DataGridViewTextBoxCell();
visibleGeckofxElementCell.Value = element;
row.Cells.Add(visibleGeckofxElementCell);
//隐藏xpath
DataGridViewTextBoxCell visibleGeckofxXpathCell = new DataGridViewTextBoxCell();
visibleGeckofxXpathCell.Value = xpathHelper.GetSmallXpath(element);
row.Cells.Add(visibleGeckofxXpathCell);
//将元素插入datagridview控件中
dgv.Rows.Add(row);
}
}
else if (geckofxWebbrowerType.Equals(GeckofxWebbrowerType.ListDetails))
{
}
else if (geckofxWebbrowerType.Equals(GeckofxWebbrowerType.UrlModel))
{
foreach (GeckoHtmlElement element in geckoHtmlElementList)
{
DataGridViewRow row = new DataGridViewRow();
//文本
DataGridViewTextBoxCell fieldsNameCell = new DataGridViewTextBoxCell();
fieldsNameCell.Value = "字段" + (count++);
row.Cells.Add(fieldsNameCell);
DataGridViewTextBoxCell fieldsContentCell = new DataGridViewTextBoxCell();
fieldsContentCell.Value = element.TextContent;
row.Cells.Add(fieldsContentCell);
//下拉框
DataGridViewTextBoxCell fieldsTypeCell = new DataGridViewTextBoxCell();
fieldsTypeCell.Value = "抓取详情页链接";
row.Cells.Add(fieldsTypeCell);
DataGridViewImageCell deleteCell = new DataGridViewImageCell();
deleteCell.Value = Image.FromFile(@"E:\Project\C#\SimpleCrawlProject\SimpleCrawlApp\Resources\1108658.png");
row.Cells.Add(deleteCell);
//隐藏浏览器句柄
DataGridViewTextBoxCell visibleGeckofxElementCell = new DataGridViewTextBoxCell();
visibleGeckofxElementCell.Value = element;
row.Cells.Add(visibleGeckofxElementCell);
//隐藏xpath
DataGridViewTextBoxCell visibleGeckofxXpathCell = new DataGridViewTextBoxCell();
visibleGeckofxXpathCell.Value = xpathHelper.GetSmallXpath(element);
row.Cells.Add(visibleGeckofxXpathCell);
//将元素插入datagridview控件中
dgv.Rows.Add(row);
}
}
}
