public static string GetAllContent(TAGBlock block)
{
string ret = string.Empty;
if (block.Name == null)
{
return(ret);
}
if (!Excluded.Contains(block.Name))
{
if (Titles.ContainsKey(block.Name))
{
ret = DataResource.GetMarkFormatted(Titles[block.Name]);
}
ret = string.Format("{0}{1}", ret, block.content);
if (block.FirstInside != null)
{
ret = string.Format("{0}{1}", ret, GetAllContent(block.FirstInside));
}
}
if (block.NextBlock != null)
{
ret = string.Format("{0}{1}", ret, GetAllContent(block.NextBlock));
}
return(ret);
}
public static void SetNULL(TAGBlock block, string content)
{
if (block == null)
{
block = new TAGBlock();
}
block.Name = "NULL";
block.content = content;
}
//处理树状内容版块
public static Dictionary <string, string> GetLinks(TAGBlock block)
{
Dictionary <string, string> Links2Title = new Dictionary <string, string>();
if (block.Name == null)
{
return(Links2Title);
}
if (block.Name.Equals("link", StringComparison.CurrentCultureIgnoreCase) ||
block.Name.Equals("a", StringComparison.CurrentCultureIgnoreCase))
{
var properties = TAG.GetProperties(block.OrgHead);
if (properties != null)
{
if (properties.ContainsKey("href") && properties.ContainsKey("title"))
{
Links2Title.Add(properties["href"], properties["title"]);
}
}
}
if (block.FirstInside != null)
{
foreach (var innerL in GetLinks(block.FirstInside))
{
if (!Links2Title.ContainsKey(innerL.Key))
{
Links2Title.Add(innerL.Key, innerL.Value);
}
}
}
if (block.NextBlock != null)
{
foreach (var nextL in GetLinks(block.NextBlock))
{
if (!Links2Title.ContainsKey(nextL.Key))
{
Links2Title.Add(nextL.Key, nextL.Value);
}
}
}
return(Links2Title);
}
public static string FindTXTContent(TAGBlock block)
{
if (block == null || block.Name == null)
{
return(string.Empty);
}
string ret = string.Empty;
if (Included.Contains(block.Name) && !Excluded.Contains(block.Name))
{
ret += (SplitHtml.GetAllInsideContent(block));
}
else if (block.FirstInside != null && !Excluded.Contains(block.Name))
{
ret += FindTXTContent(block.FirstInside);
}
if (block.NextBlock != null)
{
ret += FindTXTContent(block.NextBlock);
}
return(ret);
}
public static TAGBlock SplitTo(string htmlOrgCode)
{
var TAGs = Regex.Matches(htmlOrgCode, @"<[^<>]+>", RegexOptions.None);
TAGBlock startNode = new TAGBlock();
TAGBlock current = startNode;
int end = -1;
if (TAGs.Count == 0)
{
TAGBlock.SetNULL(startNode, htmlOrgCode);
return(startNode);
}
for (int i = 0; i < TAGs.Count; i++)
{
if (TAGs[i].Index < end)
{
continue;
}
if (TAGs[i].Index > end)
{
TAGBlock.SetNULL
(current, htmlOrgCode.Substring
(end + 1, TAGs[i].Index - end - 1));
current.NextBlock = new TAGBlock();
current = current.NextBlock;
}
var type = TAG.GetType(TAGs[i].Value);
var name = TAG.GetName(TAGs[i].Value);
if (type == TAG.TAGType.PUREEND)
{
//throw new Exception("结构错误！");
end = TAGs[i].Index + TAGs[i].Length - 1;
continue;
}
else if (type == TAG.TAGType.FULLTAG)
{
current.Name = name;
current.OrgHead = TAGs[i].Value;
current.NextBlock = new TAGBlock();
current = current.NextBlock;
end = TAGs[i].Index + TAGs[i].Length - 1;
}
else
{
current.Name = name;
current.OrgHead = TAGs[i].Value;
var InsideContent = CatchFirstTag(TAGs[i].Index, out end, htmlOrgCode, name);
if (InsideContent != null)
{
current.FirstInside = SplitTo(InsideContent);
current.NextBlock = new TAGBlock();
current = current.NextBlock;
}
else
{
current.NextBlock = new TAGBlock();
current = current.NextBlock;
end = TAGs[i].Index + TAGs[i].Length - 1;
}
}
}
if (end < htmlOrgCode.Length - 1)
{
TAGBlock.SetNULL
(current, htmlOrgCode.Substring(end + 1, htmlOrgCode.Length - end - 2));
}
return(startNode);
}