/// <summary>
/// 进行匹配判断Id或ClassName的函数，内部使用
/// </summary>
/// <param name="feature"></param>
/// <param name="Nodes"></param>
/// <returns></returns>
internal static Feature CheckforIdorClassName(Feature feature, IEnumerable <HtmlNode> Nodes, Feature stencilfeature, bool forListPage)
{
int nodestested = 0;
Dictionary <string, int> temp = new Dictionary <string, int>();
foreach (HtmlNode node in Nodes)
{
HtmlNode Node = node;
if (XPathUtility.isTopNode(Node))
{
break;
}
temp = new Dictionary <string, int>();
foreach (string key in stencilfeature.IdClassnameRecord.Keys)
{
if (stencilfeature.IdClassnameRecord[key] == 1)
{
temp.Add(key, 0);
}
}
for (int i = 1; i < 16; i++)
{
if (XPathUtility.isTopNode(Node.ParentNode))
{
break;
}
foreach (string key in stencilfeature.IdClassnameRecord.Keys)
{
if (stencilfeature.IdClassnameRecord[key] == 1 && temp[key] == 0 && XPathUtility.IDClassContain(Node, key))
{
temp[key] = i;
}
}
Node = Node.ParentNode;
}
foreach (string key in temp.Keys)
{
if (nodestested != 0 && feature.IdClassnameRecord[key] != temp[key])
{
feature.IdClassnameRecord[key] = 0;
}
feature.IdClassnameRecord[key] = temp[key];
}
nodestested++;
}
foreach (string key in temp.Keys)
{
if (feature.IdClassnameRecord[key] == 0)
{
feature.IdClassnameRecord[key] = 15;
}
}
return(feature);
}
/// <summary>
/// 对一个Node的ID和ClassName打分,越大越好
/// </summary>
/// <param name="Node"></param>
/// <param name="Must">如果命中则1分</param>
/// <param name="MustNot">如果命中则0分</param>
/// <param name="Matched">是否命中任何类名</param>
/// <param name="UpLevel">向上追溯几个级别（不含自身），上一级得分对折累加在自己的得分上</param>
/// <returns></returns>
static double IDClassNameScore(HtmlNode Node, IEnumerable <string> MustKeywords, IEnumerable <string> MustNotKeywords, out bool Matched, int UpLevel = 2)
{
Matched = false;
if (Node == null)
{
return(0);
}
int Level = 0;
double Score = 0;
double LevelScore = 1;
do
{
bool Must = XPathUtility.IDClassContain(Node, MustKeywords);
bool MustNot = XPathUtility.IDClassContain(Node, MustNotKeywords);
if (Must && !MustNot)
{
Score += LevelScore * 1.5; //命中正面关键词且无负面
}
if (Must && MustNot)
{
Score += LevelScore * 0.5; //正负面都命中
}
if (!Must && MustNot)
{
if (Level == 0) //只有负面
{
Matched = true;
return(0); //第一级则直接返回不可能
}
else
{
Score -= LevelScore;
}
}
if (Must || MustNot) //任何命中要有标志
{
Matched = true;
}
//向上一级
LevelScore *= 0.8;
Node = Node.ParentNode;
} while (!XPathUtility.isTopNode(Node) && Level++ < UpLevel);
//什么都没有命中保持1
return(Score);
}
/// <summary>
/// 检查Node的名称标签，对H1之类加权
/// </summary>
/// <param name="Node"></param>
/// <param name="UpLevel">向上追溯几个级别（不含自身）</param>
/// <returns>1表示没有命中,>1表示命中，及加分</returns>
static double HtmlTagNameScore(HtmlNode Node, int UpLevel = 2)
{
if (Node == null)
{
return(1);
}
int Level = 0;
while (!XPathUtility.isTopNode(Node) && Level++ < UpLevel)
{
if (HighLevelTagName.Contains(Node.Name.ToLower()))
{
return(1.5);
}
else
{
Node = Node.ParentNode;
}
}
return(1);
}