/// <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 btnTestPython_Click(object sender, EventArgs e)
{
var s = PythonExtHelper.RunPython(@"Plugins\SpiderUrl\test.py", new object[] { "你好啊随碟附送大放送的" });
}