private void btnConvertTest_Click(object sender, EventArgs e)
{
if (SelectedParsePattern == null)
{
MessageBox.Show("请点击右上角按钮选择匹配规则");
return;
}
IWordCodeGenerater factory = null;
if (string.IsNullOrEmpty(txbFilePath.Text))
{
factory = new WordPinyinGenerater();
}
else
{
factory = new SelfDefiningCodeGenerater();
UserCodingHelper.FilePath = txbFilePath.Text;
}
SelectedParsePattern.Factory = factory;
rtbTo.Clear();
string[] fromList = rtbFrom.Text.Split(new[] { '\r', '\n' }, StringSplitOptions.RemoveEmptyEntries);
foreach (string str in fromList)
{
string s = str.Trim();
var wl = new WordLibrary {
Word = s
};
string result = SelectedParsePattern.BuildWLString(wl);
rtbTo.AppendText(result + "\r\n");
}
}
public virtual WordLibraryList ImportText(string str)
{
pinyinFactory = new PinyinGenerater();
var wlList = new WordLibraryList();
string[] words = str.Split(new[] {'\r', '\n'}, StringSplitOptions.RemoveEmptyEntries);
CountWord = words.Length;
CurrentStatus = 0;
for (int i = 0; i < words.Length; i++)
{
try
{
string word = words[i].Trim();
if (word != string.Empty)
{
wlList.AddWordLibraryList(ImportLine(word));
}
}
catch (Exception ex)
{
Debug.WriteLine(ex.Message);
}
CurrentStatus++;
}
return wlList;
}
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 Export(WordLibraryList wlList)
{
codeGenerater = CodeTypeHelper.GetGenerater(CodeType);
var sb = new StringBuilder();
for (int i = 0; i < wlList.Count; i++)
{
sb.Append(ExportLine(wlList[i]));
sb.Append("\r\n");
}
return sb.ToString();
}
public IList<string> Export(WordLibraryList wlList)
{
codeGenerater = CodeTypeHelper.GetGenerater(CodeType);
var sb = new StringBuilder();
for (int i = 0; i < wlList.Count; i++)
{
sb.Append(ExportLine(wlList[i]));
sb.Append(lineSplitString);
}
return new List<string>() { sb.ToString() };
}
public void CodingString(WordLibrary wl, IWordCodeGenerater factory)
{
var codes = new List <string>();
foreach (char c in wl.Word)
{
string code = factory.GetCodeOfChar(c);
codes.Add(code);
}
wl.PinYin = codes.ToArray();
}
public string ExportLine(WordLibrary wl)
{
var sb = new StringBuilder();
if (codeGenerater == null)
{
codeGenerater = CodeTypeHelper.GetGenerater(CodeType);
}
if (codeGenerater.Is1CharMutiCode)
{
IList <string> codes = codeGenerater.GetCodeOfString(wl.Word);
int i = 0;
foreach (string code in codes)
{
sb.Append(wl.Word);
sb.Append("\t");
sb.Append(code);
sb.Append("\t");
sb.Append(wl.Count);
i++;
if (i != codes.Count)
{
sb.Append("\r\n");
}
}
}
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
{
if (codeGenerater.Is1Char1Code)
{
sb.Append(CollectionHelper.ListToString(codeGenerater.GetCodeOfString(wl.Word), " "));
}
else
{
sb.Append(CollectionHelper.ListToString(codeGenerater.GetCodeOfString(wl.Word)));
}
}
sb.Append("\t");
sb.Append(wl.Count);
}
return(sb.ToString());
}
public string Export(WordLibraryList wlList)
{
codeGenerater = CodeTypeHelper.GetGenerater(CodeType);
var sb = new StringBuilder();
for (int i = 0; i < wlList.Count; i++)
{
sb.Append(ExportLine(wlList[i]));
sb.Append(lineSplitString);
}
return(sb.ToString());
}
public string ExportLine(WordLibrary wl)
{
var sb = new StringBuilder();
if (codeGenerater == null)
{
codeGenerater = CodeTypeHelper.GetGenerater(CodeType);
}
if (codeGenerater.Is1CharMutiCode)
{
IList<string> codes = codeGenerater.GetCodeOfString(wl.Word);
int i = 0;
foreach (string code in codes)
{
sb.Append(wl.Word);
sb.Append("\t");
sb.Append(code);
sb.Append("\t");
sb.Append(wl.Count);
i++;
if (i != codes.Count)
sb.Append("\r\n");
}
}
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
{
if (codeGenerater.Is1Char1Code)
{
sb.Append(CollectionHelper.ListToString(codeGenerater.GetCodeOfString(wl.Word), " "));
}
else
{
sb.Append(CollectionHelper.ListToString(codeGenerater.GetCodeOfString(wl.Word)));
}
}
sb.Append("\t");
sb.Append(wl.Count);
}
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;
}
}
}
public IList <string> Export(WordLibraryList wlList)
{
codeGenerater = CodeTypeHelper.GetGenerater(CodeType);
var sb = new StringBuilder();
for (int i = 0; i < wlList.Count; i++)
{
var line = ExportLine(wlList[i]);
if (!string.IsNullOrEmpty(line))
{
sb.Append(line);
sb.Append(lineSplitString);
}
}
return(new List <string>()
{
sb.ToString()
});
}
public void Prepare()
{
codeGenerater = CodeTypeHelper.GetGenerater(this.UserDefiningPattern.CodeType);
if (UserDefiningPattern.CodeType == CodeType.UserDefine)
{
if (string.IsNullOrEmpty(UserDefiningPattern.MappingTablePath))
{
throw new Exception("未指定字符编码映射文件，无法对词库进行自定义编码的生成");
}
IDictionary <char, IList <string> > dict =
UserCodingHelper.GetCodingDict(UserDefiningPattern.MappingTablePath,
UserDefiningPattern.TextEncoding);
var g = codeGenerater as SelfDefiningCodeGenerater;
g.MappingDictionary = dict;
g.Is1Char1Code = UserDefiningPattern.IsPinyinFormat;
g.MutiWordCodeFormat = UserDefiningPattern.MutiWordCodeFormat;
}
BuildLineFormat();
}
public void Prepare()
{
codeGenerater = CodeTypeHelper.GetGenerater(this.UserDefiningPattern.CodeType);
if (UserDefiningPattern.CodeType == CodeType.UserDefine)
{
if (string.IsNullOrEmpty(UserDefiningPattern.MappingTablePath))
{
throw new Exception("未指定字符编码映射文件，无法对词库进行自定义编码的生成");
}
IDictionary<char, IList<string>> dict =
UserCodingHelper.GetCodingDict(UserDefiningPattern.MappingTablePath,
UserDefiningPattern.TextEncoding);
var g = codeGenerater as SelfDefiningCodeGenerater;
g.MappingDictionary = dict;
g.Is1Char1Code = UserDefiningPattern.IsPinyinFormat;
g.MutiWordCodeFormat = UserDefiningPattern.MutiWordCodeFormat;
}
BuildLineFormat();
}
public virtual WordLibraryList ImportText(string str)
{
pinyinFactory = new WordPinyinGenerater();
var wlList = new WordLibraryList();
string[] words = str.Split(new[] {'\r', '\n'}, StringSplitOptions.RemoveEmptyEntries);
for (int i = 0; i < words.Length; i++)
{
try
{
string word = words[i].Trim();
if (word != string.Empty)
{
wlList.AddWordLibraryList(ImportLine(word));
}
}
catch
{
}
}
return wlList;
}
public virtual WordLibraryList ImportText(string str)
{
pinyinFactory = new WordPinyinGenerater();
var wlList = new WordLibraryList();
string[] words = str.Split(new[] { '\r', '\n' }, StringSplitOptions.RemoveEmptyEntries);
for (int i = 0; i < words.Length; i++)
{
try
{
string word = words[i].Trim();
if (word != string.Empty)
{
wlList.AddWordLibraryList(ImportLine(word));
}
}
catch
{
}
}
return(wlList);
}
public void Setup()
{
generater = new PinyinGenerater();
}
//private RimeConfigForm form;
public string ExportLine(WordLibrary wl)
{
var sb = new StringBuilder();
if (this.CodeType == wl.CodeType && this.CodeType != CodeType.Pinyin && CodeType != CodeType.TerraPinyin)
{
return(wl.Word + "\t" + wl.Codes[0][0] + "\t" + wl.Rank);
}
if (codeGenerater == null)
{
codeGenerater = CodeTypeHelper.GetGenerater(CodeType);
}
try
{
codeGenerater.GetCodeOfWordLibrary(wl);
}
catch (Exception ex)
{
Debug.Fail(ex.Message);
return(null);
}
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());
}
public void CodingString(WordLibrary wl, IWordCodeGenerater factory)
{
var codes = new List<string>();
foreach (char c in wl.Word)
{
string code = factory.GetCodeOfChar(c);
codes.Add(code);
}
wl.PinYin = codes.ToArray();
}
public void Setup()
{
generater = new PinyinGenerater();
}
public void Setup()
{
generater = new ErbiGenerater();
}
public void Setup()
{
generater = new QingsongErbiGenerater();
}
public void Setup()
{
generater = new QingsongErbiGenerater();
}
public void Setup()
{
generater = new ZhuyinGenerater();
}
public void Setup()
{
generater = new ZhuyinGenerater();
}
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;
}
}
}
