/// <summary>
/// 获取产品描述
/// </summary>
/// <param name="docHtml">html文件</param>
/// <param name="version">区分图书1和其他0</param>
/// <returns></returns>
public static string GetProDesc(string docHtml, int version)
{
string area = "";
try
{
docHtml = RegexHelper.ClearTrn(docHtml);
if (version == 0)
{
//增加 规格描述 -2012-02-29 shy
string pt = HtmlCls.GetHtmlByCss(docHtml, "Ptable").FirstOrDefault();
if (!string.IsNullOrEmpty(pt))
{
area += pt;
}
area += HtmlCls.GetHtmlByCss(docHtml, "content").FirstOrDefault();
}
else
{
//图书类
var list = HtmlCls.GetHtmlByCss(docHtml, "m m1");
area = list.Aggregate(area, (current, s) => current + s);
string listH = HtmlCls.GetHtmlByCss(area, "list-h").FirstOrDefault();
if (!string.IsNullOrEmpty(listH))
{
area = area.Replace(listH, "");
}
//去除【该作者其它作品】区域
listH = HtmlCls.GetHtmlById(area, "related-works");
if (!string.IsNullOrEmpty(listH))
{
area = area.Replace(listH, "");
}
string sum = HtmlCls.GetHtmlById(docHtml, "summary"); //加入图书信息
var sumList = RegexHelper.Matches(sum, "<li[^>]*>(.*?)</li>").Take(9);
sum = sumList.Aggregate("", (current, s) => current + "<div>" + s + "</div>");
sum = Regex.Replace(sum, "<a[^>]*href=[\"']([^'\"]+?)[\"'][^>]*>(.*?)</a>", "$2"); //排除a标签
area = sum + area;
}
//排除授权html
string red = HtmlCls.GetHtmlByAttr(area, "color=\"red\"").FirstOrDefault();
if (!string.IsNullOrEmpty(red))
{
area = area.Replace(red, "");
}
area = area.Replace("class=\"content\"", ""); //排除样式冲突
//area = Regex.Replace(area, "class=['\"][^'\"]*['\"]", "");//排除样式冲突.终极
area = Regex.Replace(area, "<a[^>]*href=[\"']([^'\"]+?)[\"'][^>]*>(.*?)</a>", "$2"); //排除a标签
area = Regex.Replace(area, "\\sstyle=(['\"])[^'\"]+?\\1", ""); //排除样式
area = Regex.Replace(area, "<script[^>]*>(.*?)</script>", ""); //排除script标签
area = Regex.Replace(area, "src\\d=", "src="); //显示src
area = Regex.Replace(area, "京东商城|京东", "本商城"); //排除京东字样
}
catch (Exception ex)
{
FileHelper.WriteException(ex);
}
return(area);
}
