private void ProcessAfterSegment(string orginalText, SuperLinkedList <WordInfo> result)
{
//匹配同义词
if (_Options.SynonymOutput)
{
SuperLinkedListNode <WordInfo> node = result.First;
while (node != null)
{
List <string> synonyms = _Synonym.GetSynonyms(node.Value.Word);
if (synonyms != null)
{
foreach (string word in synonyms)
{
var sourceWord = node.Value.Word;
if (!(node.Value.WordType != WordType.Synonym))
{
sourceWord = node.Value.OriginalWord;
}
node = result.AddAfter(node, new WordInfo(word, node.Value.Position,
node.Value.Pos, node.Value.Frequency, _Parameters.SymbolRank,
WordType.Synonym, node.Value.WordType, sourceWord));
}
}
node = node.Next;
}
}
//通配符匹配
if (_Options.WildcardOutput)
{
SuperLinkedListNode <WordInfo> node = result.First;
while (node != null)
{
List <Dict.Wildcard.WildcardInfo> wildcards =
_Wildcard.GetWildcards(node.Value.Word);
if (wildcards.Count > 0)
{
for (int i = 0; i < wildcards.Count; i++)
{
Dict.Wildcard.WildcardInfo wildcardInfo = wildcards[i];
int count = wildcardInfo.Segments.Count;
if (!_Options.WildcardSegment)
{
count = 1;
}
for (int j = 0; j < count; j++)
{
WordInfo wi = wildcardInfo.Segments[j];
if (wi.Word == node.Value.Word)
{
continue;
}
wi.Rank = _Parameters.WildcardRank;
wi.Position += node.Value.Position;
result.AddBefore(node, wi);
}
}
}
node = node.Next;
if (node != null)
{
//过滤英文分词时多元分词重复输出的问题
if (node.Previous.Value.Word.ToLower() == node.Value.Word.ToLower())
{
node = node.Next;
}
}
}
}
//用户自定义规则
if (_Options.CustomRule)
{
ICustomRule rule = CustomRule.GetCustomRule(_Parameters.CustomRuleAssemblyFileName,
_Parameters.CustomRuleFullClassName);
if (rule != null)
{
rule.Text = orginalText;
rule.AfterSegment(result);
}
}
}
/// <summary>
/// 合并英文专用词。
/// 如果字典中有英文专用词如U.S.A, C++.C#等
/// 需要对初步分词后的英文和字母进行合并
/// </summary>
/// <param name="words"></param>
/// <param name="start"></param>
/// <param name="end"></param>
/// <returns></returns>
//private String MergeEnglishSpecialWord(CExtractWords extractWords, ArrayList words, int start, ref int end)
//{
// StringBuilder str = new StringBuilder();
// int i;
// for (i = start; i < words.Count; i++)
// {
// string word = (string)words[i];
// //word 为空或者为空格回车换行等分割符号，中断扫描
// if (word.Trim() == "")
// {
// break;
// }
// //如果遇到中文，中断扫描
// if (word[0] >= 0x4e00 && word[0] <= 0x9fa5)
// {
// break;
// }
// str.Append(word);
// }
// String mergeString = str.ToString();
// List<T_WordInfo> exWords = extractWords.ExtractFullText(mergeString);
// if (exWords.Count == 1)
// {
// T_WordInfo info = (T_WordInfo)exWords[0];
// if (info.Word.Length == mergeString.Length)
// {
// end = i;
// return mergeString;
// }
// }
// return null;
//}
private bool MergeEnglishSpecialWord(string orginalText, SuperLinkedList <WordInfo> wordInfoList, ref SuperLinkedListNode <WordInfo> current)
{
SuperLinkedListNode <WordInfo> cur = current;
cur = cur.Next;
int last = -1;
while (cur != null)
{
if (cur.Value.WordType == WordType.Symbol || cur.Value.WordType == WordType.English)
{
last = cur.Value.Position + cur.Value.Word.Length;
cur = cur.Next;
}
else
{
break;
}
}
if (last >= 0)
{
int first = current.Value.Position;
string newWord = orginalText.Substring(first, last - first);
WordAttribute wa = _WordDictionary.GetWordAttr(newWord);
if (wa == null)
{
return(false);
}
while (current != cur)
{
SuperLinkedListNode <WordInfo> removeItem = current;
current = current.Next;
wordInfoList.Remove(removeItem);
}
WordInfo newWordInfo = new WordInfo(new Dict.PositionLength(first, last - first,
wa), orginalText, _Parameters);
newWordInfo.WordType = WordType.English;
newWordInfo.Rank = _Parameters.EnglishRank;
if (_Options.EnglishSegment)
{
string lowerWord = newWordInfo.Word.ToLower();
if (lowerWord != newWordInfo.Word)
{
if (current == null)
{
wordInfoList.AddLast(newWordInfo);
}
else
{
wordInfoList.AddBefore(current, newWordInfo);
}
}
newWordInfo = new WordInfo(lowerWord, newWordInfo.Position, newWordInfo.Pos, newWordInfo.Frequency, _Parameters.EnglishLowerRank, newWordInfo.WordType,
newWordInfo.OriginalWordType);
}
else if (_Options.IgnoreCapital)
{
newWordInfo.Word = newWordInfo.Word.ToLower();
}
if (current == null)
{
wordInfoList.AddLast(newWordInfo);
}
else
{
wordInfoList.AddBefore(current, newWordInfo);
}
return(true);
}
return(false);
}
private SuperLinkedList <WordInfo> PreSegment(String text)
{
SuperLinkedList <WordInfo> result = GetInitSegment(text);
SuperLinkedListNode <WordInfo> cur = result.First;
while (cur != null)
{
if (_Options.IgnoreSpace)
{
if (cur.Value.WordType == WordType.Space)
{
SuperLinkedListNode <WordInfo> lst = cur;
cur = cur.Next;
result.Remove(lst);
continue;
}
}
switch (cur.Value.WordType)
{
case WordType.SimplifiedChinese:
string inputText = cur.Value.Word;
WordType originalWordType = WordType.SimplifiedChinese;
if (_Options.TraditionalChineseEnabled)
{
string simplified = ChineseTranslate.Convert(cur.Value.Word, ChineseTranslateDirection.TraditionalToSimplified);
if (simplified != cur.Value.Word)
{
originalWordType = WordType.TraditionalChinese;
inputText = simplified;
}
}
Framework.AppendList <Dict.PositionLength> pls = _WordDictionary.GetAllMatchs(inputText, _Options.ChineseNameIdentify);
Match.ChsFullTextMatch chsMatch = new Match.ChsFullTextMatch(_WordDictionary);
chsMatch.Options = _Options;
chsMatch.Parameters = _Parameters;
SuperLinkedList <WordInfo> chsMatchWords = chsMatch.Match(pls.Items, cur.Value.Word, pls.Count);
SuperLinkedListNode <WordInfo> curChsMatch = chsMatchWords.First;
while (curChsMatch != null)
{
WordInfo wi = curChsMatch.Value;
wi.Position += cur.Value.Position;
wi.OriginalWordType = originalWordType;
wi.WordType = originalWordType;
if (_Options.OutputSimplifiedTraditional)
{
if (_Options.TraditionalChineseEnabled)
{
string newWord;
WordType wt;
if (originalWordType == WordType.SimplifiedChinese)
{
newWord = ChineseTranslate.Convert(wi.Word, ChineseTranslateDirection.TraditionalToSimplified);
wt = WordType.TraditionalChinese;
}
else
{
newWord = ChineseTranslate.Convert(wi.Word, ChineseTranslateDirection.SimplifiedToTraditional);
wt = WordType.SimplifiedChinese;
}
if (newWord != wi.Word)
{
WordInfo newWordInfo = new WordInfo(wi);
newWordInfo.Word = newWord;
newWordInfo.OriginalWordType = originalWordType;
newWordInfo.WordType = wt;
newWordInfo.Rank = _Parameters.SimplifiedTraditionalRank;
newWordInfo.Position = wi.Position;
chsMatchWords.AddBefore(curChsMatch, newWordInfo);
}
}
}
curChsMatch = curChsMatch.Next;
}
SuperLinkedListNode <WordInfo> lst = result.AddAfter(cur, chsMatchWords);
SuperLinkedListNode <WordInfo> removeItem = cur;
cur = lst.Next;
result.Remove(removeItem);
break;
case WordType.English:
cur.Value.Rank = _Parameters.EnglishRank;
List <string> output;
cur.Value.Word = ConvertChineseCapitalToAsiic(cur.Value.Word);
if (_Options.EnglishSegment)
{
string lower = cur.Value.Word.ToLower();
if (lower != cur.Value.Word)
{
result.AddBefore(cur, new WordInfo(lower, cur.Value.Position, POS.POS_A_NX, 1,
_Parameters.EnglishLowerRank, WordType.English, WordType.English));
}
string stem = GetStem(lower);
if (!string.IsNullOrEmpty(stem))
{
if (lower != stem)
{
result.AddBefore(cur, new WordInfo(stem, cur.Value.Position, POS.POS_A_NX, 1,
_Parameters.EnglishStemRank, WordType.English, WordType.English));
}
}
}
else if (_Options.IgnoreCapital)
{
cur.Value.Word = cur.Value.Word.ToLower();
}
if (_Options.EnglishMultiDimensionality)
{
bool needSplit = false;
foreach (char c in cur.Value.Word)
{
if ((c >= '0' && c <= '9') || (c == '_'))
{
needSplit = true;
break;
}
}
if (needSplit)
{
if (Framework.Regex.GetMatchStrings(cur.Value.Word, PATTERNS, true, out output))
{
int outputCount = 0;
foreach (string str in output)
{
if (!string.IsNullOrEmpty(str))
{
outputCount++;
if (outputCount > 1)
{
break;
}
}
}
if (outputCount > 1)
{
int position = cur.Value.Position;
foreach (string splitWord in output)
{
if (string.IsNullOrEmpty(splitWord))
{
continue;
}
WordInfo wi;
if (splitWord[0] >= '0' && splitWord[0] <= '9')
{
wi = new WordInfo(splitWord, POS.POS_A_M, 1);
wi.Position = position;
wi.Rank = _Parameters.NumericRank;
wi.OriginalWordType = WordType.English;
wi.WordType = WordType.Numeric;
}
else
{
wi = new WordInfo(splitWord, POS.POS_A_NX, 1);
wi.Position = position;
wi.Rank = _Parameters.EnglishRank;
wi.OriginalWordType = WordType.English;
wi.WordType = WordType.English;
}
result.AddBefore(cur, wi);
position += splitWord.Length;
}
}
}
}
}
if (!MergeEnglishSpecialWord(text, result, ref cur))
{
cur = cur.Next;
}
break;
case WordType.Numeric:
cur.Value.Word = ConvertChineseCapitalToAsiic(cur.Value.Word);
cur.Value.Rank = _Parameters.NumericRank;
if (!MergeEnglishSpecialWord(text, result, ref cur))
{
cur = cur.Next;
}
//cur = cur.Next;
break;
case WordType.Symbol:
cur.Value.Rank = _Parameters.SymbolRank;
cur = cur.Next;
break;
default:
cur = cur.Next;
break;
}
}
return(result);
}