/// <summary>
/// 文本转HTML
/// </summary>
/// <param name="value">文本值</param>
/// <returns>HTML编码</returns>
public unsafe void ToHtml(ref subString value)
{
if (value.Length != 0)
{
int length = value.Length;
fixed(char *valueFixed = value.value)
{
char *start = valueFixed + value.StartIndex, end = start + length;
int count = encodeCount(start, end);
if (count != 0)
{
string newValue = fastCSharp.String.FastAllocateString(length += count << 2);
fixed(char *data = newValue) toHtml(start, end, data);
value.UnsafeSet(newValue, 0, newValue.Length);
}
}
}
}
internal void SetString(string value)
{
String.UnsafeSet(value, 0, value.Length);
Type = type.String;
}
/// <summary>
/// 文本分词
/// </summary>
/// <param name="text">文本</param>
/// <param name="length">文本长度</param>
/// <returns>分词结果</returns>
private unsafe list <subString> getWords(string text, int length)
{
fixed(char *textFixed = text)
{
simplified.Format(textFixed, length);
int count = (length + 7) >> 3;
byte * match = stackalloc byte[count];
fixedMap matchMap = new fixedMap(match, count, 0);
list <subString> words = typePool <list <subString> > .Pop();
if (words == null)
{
words = new list <subString>();
}
else if (words.Count != 0)
{
words.Clear();
}
list <keyValue <int, int> > matchs = typePool <list <keyValue <int, int> > > .Pop() ?? new list <keyValue <int, int> >();
byte * charTypes = charTypePointer.Byte;
subString matchWord = default(subString);
for (char *start = textFixed, end = textFixed + length; start != end;)
{
if (*start == ' ')
{
*end = '?';
while (*++start == ' ')
{
;
}
}
else
{
* end = ' ';
char *segment = start;
if ((uint)(*start - 0x4E00) <= 0X9FA5 - 0x4E00)
{
while ((uint)(*++start - 0x4E00) <= 0X9FA5 - 0x4E00)
{
;
}
if ((length = (int)(start - segment)) == 1)
{
words.Add(subString.Unsafe(text, (int)(segment - textFixed), 1));
}
else
{
int startIndex = (int)(segment - textFixed);
matchs.Empty();
matchWord.UnsafeSet(text, startIndex, length);
wordTrieGraph.LeftRightMatchs(ref matchWord, matchs);
if ((count = matchs.Count) != 0)
{
foreach (keyValue <int, int> value in matchs.UnsafeArray)
{
words.Add(subString.Unsafe(text, value.Key, value.Value));
matchMap.Set(value.Key, value.Value);
if (--count == 0)
{
break;
}
}
}
int index = startIndex;
for (int endIndex = startIndex + length; index != endIndex; ++index)
{
if (matchMap.Get(index))
{
if ((count = index - startIndex) != 1)
{
words.Add(subString.Unsafe(text, startIndex, count));
}
startIndex = index;
}
else
{
words.Add(subString.Unsafe(text, index, 1));
}
}
if ((index -= startIndex) > 1)
{
words.Add(subString.Unsafe(text, startIndex, index));
}
}
}
else
{
byte type = charTypes[*start];
if (type == (byte)charType.OtherLetter)
{
while (charTypes[*++start] == (byte)charType.OtherLetter)
{
;
}
}
else
{
char *word = start;
for (byte newType = charTypes[*++start]; newType >= (byte)charType.Letter; newType = charTypes[*++start])
{
if (type != newType)
{
if (type != (byte)charType.Keep)
{
words.Add(subString.Unsafe(text, (int)(word - textFixed), (int)(start - word)));
}
type = newType;
word = start;
}
}
}
words.Add(subString.Unsafe(text, (int)(segment - textFixed), (int)(start - segment)));
}
}
}
typePool <list <keyValue <int, int> > > .PushNotNull(matchs);
if ((count = words.Count) == 0)
{
typePool <list <subString> > .PushNotNull(words);
return(null);
}
return(words);
}
}