public void TestPinyin2TerraPinyin()
{
WordLibrary wl = new WordLibrary()
{
Word = "深蓝", Rank = 123, PinYin = new [] { "shen", "lan" }, CodeType = CodeType.Pinyin
};
generater.GetCodeOfWordLibrary(wl);
foreach (var py in wl.Codes)
{
Debug.WriteLine(py);
}
}
private void GenerateDestinationCode(WordLibraryList wordLibraryList, CodeType codeType)
{
IWordCodeGenerater generater = CodeTypeHelper.GetGenerater(codeType);
if (generater == null) //未知编码方式，则不进行编码。
{
return;
}
countWord = wordLibraryList.Count;
currentStatus = 0;
foreach (WordLibrary wordLibrary in wordLibraryList)
{
currentStatus++;
processMessage = "生成目标编码：" + currentStatus + "/" + countWord;
if (wordLibrary.CodeType == codeType)
{
continue;
}
try
{
generater.GetCodeOfWordLibrary(wordLibrary);
}
catch (Exception ex)
{
Debug.WriteLine("生成编码失败" + ex.Message);
}
if (codeType != CodeType.Unknown)
{
wordLibrary.CodeType = codeType;
}
}
}
public string ExportLine(WordLibrary wl)
{
if (lineFormat == "")
{
BuildLineFormat();
}
var lines = new List <string>();
//需要判断源WL与导出的字符串的CodeType是否一致，如果一致，那么可以采用其编码，如果不一致，那么忽略编码，
//调用CodeGenerater生成新的编码，并用新编码生成行
IList <string> codes = null;
if (wl.CodeType != CodeType)
{
codeGenerater.GetCodeOfWordLibrary(wl);
}
string word = wl.Word;
int rank = wl.Rank;
foreach (string code in wl.Codes.ToCodeString(UserDefiningPattern.CodeSplitString, UserDefiningPattern.CodeSplitType))
{
string line = String.Format(lineFormat, code, word, rank);
lines.Add(line);
}
return(String.Join(UserDefiningPattern.LineSplitString, lines.ToArray()));
}
//private RimeConfigForm form;
public string ExportLine(WordLibrary wl)
{
var sb = new StringBuilder();
if (codeGenerater == null)
{
codeGenerater = CodeTypeHelper.GetGenerater(CodeType);
}
codeGenerater.GetCodeOfWordLibrary(wl);
if (codeGenerater.Is1CharMutiCode)
{
IList <string> codes = codeGenerater.GetCodeOfString(wl.Word).ToCodeString(" ");
int i = 0;
foreach (string code in codes)
{
sb.Append(wl.Word);
sb.Append("\t");
sb.Append(code);
sb.Append("\t");
sb.Append(wl.Rank);
i++;
if (i != codes.Count)
{
sb.Append(lineSplitString);
}
}
}
else
{
sb.Append(wl.Word);
sb.Append("\t");
if (CodeType == CodeType.Pinyin || CodeType == CodeType.TerraPinyin)
{
sb.Append(wl.GetPinYinString(" ", BuildType.None));
}
else if (CodeType == wl.CodeType)
{
sb.Append(wl.Codes[0][0]);
}
else
{
sb.Append(wl.Codes.ToCodeString(" ")[0]);
}
sb.Append("\t");
sb.Append(wl.Rank);
}
return(sb.ToString());
}
private void GenerateDestinationCode(WordLibraryList wordLibraryList, CodeType codeType)
{
if (wordLibraryList.Count == 0)
{
return;
}
if (wordLibraryList[0].CodeType == CodeType.NoCode && codeType == CodeType.UserDefinePhrase)
{
codeType = CodeType.Pinyin;
}
IWordCodeGenerater generater = CodeTypeHelper.GetGenerater(codeType);
if (generater == null) //未知编码方式，则不进行编码。
{
return;
}
countWord = wordLibraryList.Count;
currentStatus = 0;
foreach (WordLibrary wordLibrary in wordLibraryList)
{
currentStatus++;
processMessage = "生成目标编码：" + currentStatus + "/" + countWord;
if (wordLibrary.CodeType == codeType)
{
continue;
}
if (wordLibrary.CodeType == CodeType.English)
{
wordLibrary.SetCode(CodeType.English, wordLibrary.Word.ToLower());
continue;
}
try
{
generater.GetCodeOfWordLibrary(wordLibrary);
}
catch (Exception ex)
{
Debug.WriteLine("生成编码失败" + ex.Message);
}
if (codeType != CodeType.Unknown)
{
wordLibrary.CodeType = codeType;
}
}
}
private void GenerateDestinationCode(WordLibraryList wordLibraryList, CodeType codeType)
{
if (wordLibraryList.Count == 0)
{
return;
}
if (wordLibraryList[0].CodeType == CodeType.NoCode && codeType == CodeType.UserDefinePhrase)
{
codeType = CodeType.Pinyin;
}
IWordCodeGenerater generater = CodeTypeHelper.GetGenerater(codeType);
if (generater == null) //未知编码方式，则不进行编码。
{
return;
}
countWord = wordLibraryList.Count;
currentStatus = 0;
Regex spaceRegex = new Regex("(?=[^a-zA-Z])\\s+");
Regex numberRegex = new Regex("[0-9０-９]+");
Regex englishRegex = new Regex("[a-zA-Zａ-ｚＡ-Ｚ]+");
Regex fullWidthRegex = new Regex("[\uff00-\uff5e]+");
// Regex fullWidthRegex = new Regex("[ａ-ｚＡ-Ｚ０-９]+");
// Regex punctuationRegex = new Regex("[-・·&%']");
Regex punctuationRegex = new Regex("[\u0021-\u002f\u003a-\u0040\u005b-\u0060\u007b-\u008f\u00a0-\u00bf\u00d7\u00f7\u2000-\u2bff\u3000-\u303f\u30a0\u30fb\uff01-\uff0f\uff1a-\uff20\uff5b-\uff65]");
foreach (WordLibrary wordLibrary in wordLibraryList)
{
currentStatus++;
processMessage = "生成目标编码：" + currentStatus + "/" + countWord;
if (wordLibrary.CodeType == codeType)
{
continue;
}
if (wordLibrary.CodeType == CodeType.English)
{
wordLibrary.SetCode(CodeType.English, wordLibrary.Word.ToLower());
continue;
}
try
{
string word_0 = wordLibrary.Word;
string word = wordLibrary.Word;
if (FilterConfig.FullWidth && fullWidthRegex.IsMatch(word))
{
char[] c = word.ToCharArray();
for (int i = 0; i < c.Length; i++)
{
if (c[i] <= 0xff5e && c[i] >= 0xff00)
{
c[i] = (char)(c[i] - 65248);
}
}
word = new String(c);
}
if (FilterConfig.KeepNumber_)
{
word = numberRegex.Replace(word, "");
}
if (FilterConfig.KeepEnglish_)
{
word = englishRegex.Replace(word, "");
}
if (FilterConfig.KeepSpace_)
{
if (FilterConfig.KeepSpace == false)
{
word = word.Replace(" ", "");
}
else
{
word = spaceRegex.Replace(word, "");
}
}
if (FilterConfig.KeepPunctuation_)
{
word = punctuationRegex.Replace(word, "");
}
if (FilterConfig.ChsNumber)
{
word = TranslateChineseNumber(word);
}
if ((englishRegex.IsMatch(word) && FilterConfig.KeepEnglish) || (numberRegex.IsMatch(word) && FilterConfig.KeepNumber) || (punctuationRegex.IsMatch(word) && FilterConfig.KeepPunctuation))
{
StringBuilder input = new StringBuilder();
List <IList <string> > output = new List <IList <string> >();
int clipType = -1; int type = 0;
foreach (char c in word)
{
if (c >= 0x30 && c <= 0x39)
{
type = 1;
}
else if (c >= 0x41 && c <= 0x5a)
{
type = 2;
}
else if (c >= 0x61 && c <= 0x7a)
{
type = 2;
}
else if (c == 0x20 && FilterConfig.KeepSpace && clipType == 2)
{
type = 2;
}
else if ("-・&%'".Contains(c))
{
type = 3;
}
else if (punctuationRegex.IsMatch(c.ToString()))
{
type = 3;
}
else
{
type = 0;
}
if (input.Length < 1)
{
clipType = type;
input.Append(c);
}
else if (type == clipType)
{
input.Append(c);
}
else
{
if (clipType == 2 && FilterConfig.KeepEnglish)
{
if (FilterConfig.needEnglishTag())
{
output.Add(new List <string> {
'_' + input.ToString()
});
}
else
{
output.Add(new List <string> {
input.ToString()
});
}
}
else if ((clipType == 1 && FilterConfig.KeepNumber) || (clipType == 3 && FilterConfig.KeepPunctuation))
{
output.Add(new List <string> {
input.ToString()
});
}
else
{
wordLibrary.Word = input.ToString();
wordLibrary.CodeType = CodeType.NoCode;
generater.GetCodeOfWordLibrary(wordLibrary);
output.AddRange(wordLibrary.Codes);
}
input.Clear();
input.Append(c);
clipType = type;
}
}
if (input.Length > 0)
{
if (clipType == 2 && FilterConfig.KeepEnglish)
{
if (FilterConfig.needEnglishTag())
{
output.Add(new List <string> {
'_' + input.ToString()
});
}
else
{
output.Add(new List <string> {
input.ToString()
});
}
}
else if ((clipType == 1 && FilterConfig.KeepNumber) || (clipType == 3 && FilterConfig.KeepPunctuation))
{
output.Add(new List <string> {
input.ToString()
});
}
else
{
wordLibrary.Word = input.ToString();
wordLibrary.CodeType = CodeType.NoCode;
generater.GetCodeOfWordLibrary(wordLibrary);
output.AddRange(wordLibrary.Codes);
}
}
wordLibrary.Word = word_0;
wordLibrary.Codes = new Code(output);
}
else
{
if (word.Equals(word_0))
{
generater.GetCodeOfWordLibrary(wordLibrary);
}
else
{
wordLibrary.Word = word;
generater.GetCodeOfWordLibrary(wordLibrary);
wordLibrary.Word = word_0;
}
}
}
catch (Exception ex)
{
Debug.WriteLine("生成编码失败" + ex.Message);
}
if (codeType != CodeType.Unknown)
{
wordLibrary.CodeType = codeType;
}
}
}