/// <summary>
/// 解析HTML节点
/// </summary>
/// <param name="html"></param>
private void create(string html)
{
int length = html.Length;
children = new list<htmlNode>();
if (length < 2)
{
children.Add(new htmlNode { nodeText = new htmlText { FormatHtml = html }, Parent = this });
}
else
{
int nextIndex, nodeCount;
htmlNode nextNode;
fixed (char* htmlFixed = html + "<")
{
fixedMap spaceFixedMap = new fixedMap(spaceMap.Map);
fixedMap spaceSplitFixedMap = new fixedMap(spaceSplitMap);
fixedMap tagNameFixedMap = new fixedMap(tagNameMap);
fixedMap tagNameSplitFixedMap = new fixedMap(tagNameSplitMap);
fixedMap attributeSplitFixedMap = new fixedMap(attributeSplitMap);
fixedMap attributeNameSplitFixedMap = new fixedMap(attributeNameSplitMap);
int startIndex, tagNameLength;
string name, htmlValue;
char* startChar = htmlFixed, currentChar = htmlFixed, endChar = htmlFixed + length, scriptChar;
char splitChar;
while (currentChar != endChar)
{
for (*endChar = '<'; *currentChar != '<'; ++currentChar) ;
if (currentChar != endChar)
{
if ((*++currentChar & 0xff80) == 0)
{
if (tagNameFixedMap.Get(*currentChar))
{
while ((*startChar & 0xffc0) == 0 && spaceFixedMap.Get(*startChar)) ++startChar;
if (startChar != currentChar - 1)
{
for (scriptChar = currentChar - 2; (*scriptChar & 0xffc0) == 0 && spaceFixedMap.Get(*scriptChar); --scriptChar) ;
children.Add(new htmlNode { nodeText = new htmlText { FormatHtml = html.Substring((int)(startChar - htmlFixed), (int)(scriptChar - startChar) + 1) } });
}
if (*currentChar == '/')
{
#region 标签回合
startChar = currentChar - 1;
if (++currentChar != endChar)
{
while ((*currentChar & 0xffc0) == 0 && spaceFixedMap.Get(*currentChar)) ++currentChar;
if (currentChar != endChar)
{
if ((uint)((*currentChar | 0x20) - 'a') <= 26)
{
for (*endChar = '>'; (*currentChar & 0xffc0) != 0 || !tagNameSplitFixedMap.Get(*currentChar); ++currentChar) ;
TagName = html.Substring((int)((startChar += 2) - htmlFixed), (int)(currentChar - startChar)).toLower();
for (startIndex = children.Count - 1; startIndex >= 0 && (children[startIndex].nodeText.FormatHtml != null || children[startIndex].TagName != TagName); --startIndex) ;
if (startIndex != -1)
{
for (nextIndex = children.Count - 1; nextIndex != startIndex; --nextIndex)
{
nextNode = children[nextIndex];
if (nextNode.nodeText.FormatHtml == null)
{
if (web.html.MustRoundTagNames.Contains(nextNode.TagName) && (nodeCount = (children.Count - nextIndex - 1)) != 0)
{
nextNode.children = new list<htmlNode>(children.GetSub(nextIndex + 1, nodeCount), true);
children.RemoveRange(nextIndex + 1, nodeCount);
foreach (htmlNode value in nextNode.children) value.Parent = nextNode;
}
}
else if (nextNode.nodeText.FormatHtml.Length == 0) nextNode.nodeText.FormatHtml = null;
}
nextNode = children[startIndex];
if ((nodeCount = children.Count - ++startIndex) != 0)
{
nextNode.children = new list<htmlNode>(children.GetSub(startIndex, nodeCount), true);
children.RemoveRange(startIndex, nodeCount);
foreach (htmlNode value in nextNode.children) value.Parent = nextNode;
}
nextNode.nodeText.FormatHtml = string.Empty;//已回合标识
}
while (*currentChar != '>') ++currentChar;
if (currentChar != endChar) ++currentChar;
}
else
{
for (*endChar = '>'; *currentChar != '>'; ++currentChar) ;
if (currentChar != endChar) ++currentChar;
htmlValue = html.Substring((int)(startChar - htmlFixed), (int)(currentChar - startChar));
children.Add(new htmlNode { TagName = "/", nodeText = new htmlText { FormatHtml = htmlValue, FormatText = htmlValue } });
}
startChar = currentChar;
}
}
#endregion
}
else if (*currentChar != '!')
{
#region 标签开始
startChar = currentChar;
children.Add(nextNode = new htmlNode());
for (*endChar = '>'; (*currentChar & 0xffc0) != 0 || !tagNameSplitFixedMap.Get(*currentChar); ++currentChar) ;
nextNode.TagName = html.Substring((int)(startChar - htmlFixed), (int)(currentChar - startChar)).toLower();
if (currentChar == endChar) startChar = endChar;
else
{
#region 属性解析
if (*currentChar != '>')
{
startChar = ++currentChar;
while (currentChar != endChar)
{
while ((*currentChar & 0xffc0) == 0 && attributeSplitFixedMap.Get(*currentChar)) ++currentChar;
if (*currentChar == '>')
{
if (currentChar != endChar)
{
if (*(currentChar - 1) == '/') nextNode.nodeText.FormatHtml = string.Empty;
startChar = ++currentChar;
}
break;
}
else
{
for (startChar = currentChar++; (*currentChar & 0xffc0) != 0 || !tagNameSplitFixedMap.Get(*currentChar); ++currentChar) ;
htmlValue = name = checkName(html.Substring((int)(startChar - htmlFixed), (int)(currentChar - startChar)).toLower());
if (currentChar != endChar && ((*currentChar & 0xffc0) != 0 || !attributeNameSplitFixedMap.Get(*currentChar)))
{
if (*currentChar != '=')
{
while ((*currentChar & 0xffc0) == 0 && spaceFixedMap.Get(*currentChar)) ++currentChar;
}
if (*currentChar == '=')
{
while ((*++currentChar & 0xffc0) == 0 && spaceFixedMap.Get(*currentChar)) ;
if ((splitChar = *currentChar) != '>')
{
if (splitChar == '"' || splitChar == '\'')
{
for (startChar = ++currentChar, *endChar = splitChar; *currentChar != splitChar; ++currentChar) ;
*endChar = '>';
}
else
{
for (startChar = currentChar++; (*currentChar & 0xffc0) != 0 || !spaceSplitFixedMap.Get(*currentChar); ++currentChar) ;
}
htmlValue = html.Substring((int)(startChar - htmlFixed), (int)(currentChar - startChar));
}
}
}
if (nextNode.attributes == null) nextNode.attributes = new Dictionary<string, htmlText>();
nextNode.attributes[name] = new htmlText { FormatHtml = htmlValue };
if (currentChar != endChar)
{
if (*currentChar == '>')
{
if (*(currentChar - 1) == '/') nextNode.nodeText.FormatHtml = string.Empty;
startChar = ++currentChar;
break;
}
startChar = ++currentChar;
}
}
}
}
else startChar = ++currentChar;
#endregion
#region 非解析标签
if (currentChar == endChar) startChar = endChar;
else if (web.html.NonanalyticTagNames.Contains(TagName = nextNode.TagName))
{
scriptChar = endChar;
tagNameLength = TagName.Length + 2;
fixed (char* tagNameFixed = TagName)
{
while ((int)(endChar - currentChar) > tagNameLength)
{
for (currentChar += tagNameLength; *currentChar != '>'; ++currentChar) ;
if (currentChar != endChar && *(int*)(currentChar - tagNameLength) == (('/' << 16) + '<'))
{
if (unsafer.String.EqualCase(currentChar - TagName.Length, tagNameFixed, TagName.Length))
{
scriptChar = currentChar - tagNameLength;
if (currentChar != endChar) ++currentChar;
break;
}
}
}
}
if (startChar != scriptChar)
{
nextNode.nodeText.FormatHtml = nextNode.nodeText.FormatText = html.Substring((int)(startChar - htmlFixed), (int)(scriptChar - startChar));
}
if (scriptChar == endChar) currentChar = endChar;
startChar = currentChar;
}
#endregion
}
#endregion
}
else
{
#region 注释
startChar = currentChar - 1;
if (++currentChar != endChar)
{
*endChar = '>';
if ((length = (int)(endChar - currentChar)) > 2 && *(int*)currentChar == (('-' << 16) + '-'))
{
for (currentChar += 2; *currentChar != '>'; ++currentChar) ;
while (currentChar != endChar && *(int*)(currentChar - 2) != (('-' << 16) + '-'))
{
if ((currentChar += 3) < endChar)
{
while (*currentChar != '>') ++currentChar;
}
else currentChar = endChar;
}
}
else if (length > 9
&& (*(int*)currentChar & 0x200000) == ('[' + ('c' << 16))
&& (*(int*)(currentChar + 2) & 0x200020) == ('d' + ('a' << 16))
&& (*(int*)(currentChar + 4) & 0x200020) == ('t' + ('a' << 16))
&& *(currentChar + 6) == '[')
{
for (currentChar += 9; *currentChar != '>'; ++currentChar) ;
while (currentChar != endChar && *(int*)(currentChar - 2) != ((']' << 16) + ']'))
{
if ((currentChar += 3) < endChar)
{
while (*currentChar != '>') ++currentChar;
}
else currentChar = endChar;
}
}
else
{
while (*currentChar != '>') ++currentChar;
}
if (currentChar != endChar) ++currentChar;
}
htmlValue = html.Substring((int)(startChar - htmlFixed), (int)(currentChar - startChar) + (*(currentChar - 1) == '>' ? 0 : 1));
children.Add(new htmlNode { TagName = "!", nodeText = new htmlText { FormatHtml = htmlValue, FormatText = htmlValue } });
startChar = currentChar;
#endregion
}
}
}
else ++currentChar;
}
}
if (startChar != endChar)
{
*endChar = '>';
while ((*startChar & 0xffc0) == 0 && spaceFixedMap.Get(*startChar)) ++startChar;
if (startChar != endChar)
{
for (scriptChar = endChar - 1; (*scriptChar & 0xffc0) == 0 && spaceFixedMap.Get(*scriptChar); --scriptChar) ;
children.Add(new htmlNode { nodeText = new htmlText { FormatHtml = html.Substring((int)(startChar - htmlFixed), (int)(scriptChar - startChar) + 1) } });
}
}
}
for (nextIndex = children.Count - 1; nextIndex != -1; nextIndex--)
{
nextNode = children[nextIndex];
if (nextNode.nodeText.FormatHtml == null)
{
if (web.html.MustRoundTagNames.Contains(nextNode.TagName)
&& (nodeCount = (children.Count - nextIndex - 1)) != 0)
{
nextNode.children = new list<htmlNode>(children.GetSub(nextIndex + 1, nodeCount), true);
children.RemoveRange(nextIndex + 1, nodeCount);
foreach (htmlNode value in children) value.Parent = nextNode;
}
}
else if (nextNode.nodeText.FormatHtml.Length == 0) nextNode.nodeText.FormatHtml = null;
}
foreach (htmlNode value in children) value.Parent = this;
}
}