public string ExportLine(WordLibrary wl)
{
//StringBuilder sb = new StringBuilder();
string str = wl.GetPinYinString("'", BuildType.None) + "," + wl.Word;
return str;
}
public string ExportLine(WordLibrary wl)
{
var sb = new StringBuilder();
sb.Append(wl.GetPinYinString("", BuildType.None));
sb.Append(" ");
sb.Append(wl.Word);
return sb.ToString();
}
/// <summary>
/// 将一行纯文本转换为对象
/// </summary>
/// <param name="line"></param>
/// <returns></returns>
public virtual WordLibraryList ImportLine(string line)
{
var py = pinyinFactory.GetCodeOfString(line);
var wl = new WordLibrary();
wl.Word = line;
wl.PinYin = ToArray(py);
var wll = new WordLibraryList();
wll.Add(wl);
return wll;
}
public string ExportLine(WordLibrary wl)
{
var sb = new StringBuilder();
sb.Append(wubiFactory.GetCodeOfString(wl.Word)[0]);
sb.Append(" ");
sb.Append(wl.Word);
return sb.ToString();
}
public string ExportLine(WordLibrary wl)
{
var sb = new StringBuilder();
sb.Append(WubiHelper.GetStringWubi86Code(wl.Word));
sb.Append(" ");
sb.Append(wl.Word);
return sb.ToString();
}
public WordLibraryList ImportLine(string line)
{
string[] c = line.Split('\t');
var wl = new WordLibrary();
wl.Word = c[0];
wl.Count = Convert.ToInt32(c[1]);
wl.PinYin = c[2].Split(new[] {' '}, StringSplitOptions.RemoveEmptyEntries);
var wll = new WordLibraryList();
wll.Add(wl);
return wll;
}
public WordLibraryList ImportLine(string line)
{
string py = line.Split(',')[0];
string word = line.Split(',')[1];
var wl = new WordLibrary();
wl.Word = word;
wl.Count = 1;
wl.PinYin = py.Split(new[] {'\''}, StringSplitOptions.RemoveEmptyEntries);
var wll = new WordLibraryList();
wll.Add(wl);
return wll;
}
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);
}
}
public void TestChar2TerraPinyin(string word, string pinyin)
{
WordLibrary wl = new WordLibrary()
{
Word = word, Rank = 123, CodeType = CodeType.NoCode
};
generater.GetCodeOfWordLibrary(wl);
foreach (var py in wl.Codes.ToCodeString(" "))
{
Debug.WriteLine(py);
}
}
public WordLibraryList ImportLine(string line)
{
string[] lineArray = line.Split('\t');
string py = lineArray[1];
string word = lineArray[0];
var wl = new WordLibrary();
wl.Word = word;
wl.Count = Convert.ToInt32(lineArray[2]);
wl.PinYin = py.Split(new[] {' '}, StringSplitOptions.RemoveEmptyEntries);
var wll = new WordLibraryList();
wll.Add(wl);
return wll;
}
/// <summary>
/// 将一行纯文本转换为对象
/// </summary>
/// <param name="line"></param>
/// <returns></returns>
public virtual WordLibraryList ImportLine(string line)
{
//IList<string> py = pinyinFactory.GetCodeOfString(line);
var wl = new WordLibrary();
wl.Word = line;
wl.CodeType = CodeType;
//wl.PinYin = CollectionHelper.ToArray(py);
var wll = new WordLibraryList();
wll.Add(wl);
return(wll);
}
public WordLibraryList ImportLine(string line)
{
string[] wp = line.Split('\t');
string word = wp[0];
var wl = new WordLibrary();
wl.Word = word;
wl.Count = Convert.ToInt32(wp[1]);
wl.PinYin = new string[] {};
var wll = new WordLibraryList();
wll.Add(wl);
return wll;
}
public string ExportLine(WordLibrary wl)
{
var sb = new StringBuilder();
sb.Append(wl.Word);
sb.Append("\t");
sb.Append(wl.Rank);
sb.Append("\t");
sb.Append(wl.GetPinYinString(" ", BuildType.None));
return(sb.ToString());
}
public WordLibraryList ImportLine(string line)
{
string[] c = line.Split('\t');
var wl = new WordLibrary();
wl.Word = c[0];
wl.Rank = Convert.ToInt32(c[1]);
wl.PinYin = c[2].Split(new[] { ' ' }, StringSplitOptions.RemoveEmptyEntries);
var wll = new WordLibraryList();
wll.Add(wl);
return(wll);
}
public IList <string> GetCodeOfWordLibrary(WordLibrary str, string charCodeSplit = "")
{
if (str.CodeType == CodeType.Pinyin)
{
return(new List <string>()
{
str.GetPinYinString("", BuildType.None)
});
}
else
{
return(CollectionHelper.Descartes(str.Codes));
}
}
public WordLibraryList ImportLine(string line)
{
string py = line.Split('\t')[1];
string word = line.Split('\t')[0];
var wl = new WordLibrary();
wl.Word = word;
wl.Count = 1;
wl.PinYin = py.Split(new[] { '\'' }, StringSplitOptions.RemoveEmptyEntries);
var wll = new WordLibraryList();
wll.Add(wl);
return(wll);
}
public override void GetCodeOfWordLibrary(WordLibrary wl)
{
if (wl.CodeType == CodeType.English)
{
wl.SetCode(CodeType.UserDefinePhrase, wl.Word);
}
else if (wl.CodeType == CodeType.Pinyin)
{
wl.SetCode(CodeType.UserDefinePhrase, wl.GetPinYinString("", BuildType.None));
}
var codes = CollectionHelper.Descartes(wl.Codes);
wl.SetCode(CodeType.UserDefinePhrase, codes[0]);
}
public string ExportLine(WordLibrary wl)
{
var sb = new StringBuilder();
sb.Append(wl.Word);
sb.Append("\t");
if (!wl.IsEnglish)
{
sb.Append(wl.GetPinYinString("'", BuildType.RightContain));
sb.Append("\t");
}
sb.Append(wl.Count);
return(sb.ToString());
}
public override void GetCodeOfWordLibrary(WordLibrary wl)
{
base.GetCodeOfWordLibrary(wl);
for (int i = 0; i < wl.Codes.Count; i++)
{
var row = wl.Codes[i];
for (int j = 0; j < row.Count; j++)
{
string s = row[j];
string zy = ZhuyinHelper.GetZhuyin(s);
wl.Codes[i][j] = zy;
}
}
}
public string ExportLine(WordLibrary wl)
{
var sb = new StringBuilder();
sb.Append(wl.Word);
sb.Append("\t");
if (!wl.IsEnglish)
{
sb.Append(wl.GetPinYinString("'", BuildType.RightContain));
sb.Append("\t");
}
sb.Append(wl.Count);
return sb.ToString();
}
public string ExportLine(WordLibrary wl)
{
var sb = new StringBuilder();
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 WordLibraryList ImportLine(string line)
{
string code = line.Split(' ')[0];
string word = line.Split(' ')[1];
var wl = new WordLibrary();
wl.Word = word;
wl.Count = DefaultRank;
wl.PinYin = ToArray(pinyinFactory.GetCodeOfString(word));
var wll = new WordLibraryList();
if (wl.PinYin.Length > 0)
{
wll.Add(wl);
}
return wll;
}
public WordLibraryList ImportLine(string line)
{
string[] c = line.Split(' ');
var wl = new WordLibrary();
string code = c[0];
wl.Word = c[1];
wl.Count = DefaultRank;
wl.PinYin = CollectionHelper.ToArray(pyGenerater.GetCodeOfString(wl.Word));
wl.SetCode(CodeType, code);
var wll = new WordLibraryList();
wll.Add(wl);
return(wll);
}
public WordLibraryList ImportLine(string line)
{
string[] lineArray = line.Split('\t');
string py = lineArray[1];
string word = lineArray[0];
var wl = new WordLibrary();
wl.Word = word;
wl.Count = Convert.ToInt32(lineArray[2]);
wl.PinYin = py.Split(new[] { ' ' }, StringSplitOptions.RemoveEmptyEntries);
var wll = new WordLibraryList();
wll.Add(wl);
return(wll);
}
public WordLibraryList ImportLine(string line)
{
line = line.Split(',')[0]; //如果有逗号，就只取第一个
string[] sp = line.Split(' ');
string py = sp[0];
string word = sp[1];
int count = Convert.ToInt32(sp[2]);
var wl = new WordLibrary();
wl.Word = word;
wl.Count = count;
wl.PinYin = py.Split(new[] {'\''}, StringSplitOptions.RemoveEmptyEntries);
var wll = new WordLibraryList();
wll.Add(wl);
return wll;
}
public override WordLibraryList ImportLine(string line)
{
string[] c = line.Split(' ');
var wl = new WordLibrary();
string code = c[0];
wl.Word = c[1];
wl.Rank = DefaultRank;
wl.SetCode(CodeType.Cangjie, pyGenerater.GetCodeOfString(wl.Word));
wl.SetCode(CodeType, code);
var wll = new WordLibraryList();
wll.Add(wl);
return(wll);
}
public override WordLibraryList ImportLine(string line)
{
string[] wp = line.Split('\t');
string word = wp[0];
var wl = new WordLibrary();
wl.Word = word;
wl.Rank = Convert.ToInt32(wp[1]);
wl.PinYin = new string[] {};
var wll = new WordLibraryList();
wll.Add(wl);
return(wll);
}
//public Dictionary<char,string > CharAndPinyin=new Dictionary<char, string>();
//private void AddWordAndPinyin(char word,string pinyin)
//{
// if (!CharAndPinyin.ContainsKey(word))
// {
// CharAndPinyin.Add(word,pinyin);
// }
//}
/// <summary>
/// 读取一个词语，格式为：
/// 4字节的长度len
/// len*2字节的拼音（对于每个字，声母、韵母各1字节）
/// len*2字节的汉字（Unicode编码）
/// </summary>
/// <param name="fs"></param>
/// <returns></returns>
private WordLibrary ImportWord(FileStream fs)
{
int show = 0;
var wordLibrary = new WordLibrary();
var temp = new byte[4];
fs.Read(temp, 0, 4);
int len = BitConverter.ToInt32(temp, 0);
if (len == 0)
{
Debug.WriteLine(fs.Position);
return(null);
//return SpecialWord(fs);
}
var pinyinList = new List <string>();
for (int i = 0; i < len; i++)
{
temp = new byte[2];
fs.Read(temp, 0, 2);
try
{
string sm = Shengmu[temp[0]];
string ym = Yunmu[temp[1]];
pinyinList.Add(sm + ym);
}
catch (Exception e)
{
Debug.WriteLine(e.Message);
show = temp[0];
}
}
wordLibrary.PinYin = pinyinList.ToArray();
temp = new byte[2 * len];
fs.Read(temp, 0, 2 * len);
wordLibrary.Word = Encoding.Unicode.GetString(temp);
//for (var i = 0; i < wordLibrary.Word.Length;i++ )
//{
// AddWordAndPinyin(wordLibrary.Word[i], wordLibrary.PinYin[i]);
//}
if (show > 0)
{
Debug.WriteLine(show + " " + wordLibrary.Word + "----" + wordLibrary.PinYinString);
}
return(wordLibrary);
}
//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());
}
public WordLibraryList Import(string path)
{
var wordLibraryList = new WordLibraryList();
var fs = new FileStream(path, FileMode.Open, FileAccess.Read);
fs.Position = 0x350;
do
{
try
{
WordLibrary wl = ImportWord(fs);
if (wl.Word != "" && wl.PinYin.Length > 0)
{
wordLibraryList.Add(wl);
}
}
catch (Exception ex)
{
Debug.WriteLine(ex.Message);
}
} while (fs.Position != fs.Length);
fs.Close();
//StreamWriter sw=new StreamWriter("D:\\py.txt",true,Encoding.Unicode);
//SinglePinyin singlePinyin=new SinglePinyin();
//foreach (var cpy in CharAndPinyin)
//{
// var py = "";
// try
// {
// py = singlePinyin.GetPinYinOfChar(cpy.Key)[0];
// }
// catch
// {
// Debug.Write(cpy.Key);
// }
// sw.WriteLine(cpy.Key+"\t"+ py+"\t"+cpy.Value);
//}
//sw.Close();
//wordLibraryList.ForEach(delegate(WordLibrary wl) { if(wl.Word==""||wl.PinYin.Length==0)
//{
// Debug.WriteLine(wl.ToDisplayString());
//}
//});
return(wordLibraryList);
}
public WordLibraryList ImportLine(string line)
{
var wl = new WordLibrary();
wl.Word = line.Split('\t')[1];
wl.CodeType = CodeType;
wl.IsEnglish = IsEnglish(wl.Word);
if (wl.IsEnglish)
{
wl.SetCode(CodeType.English, wl.Word);
}
var wll = new WordLibraryList();
wll.Add(wl);
return(wll);
}
public override WordLibraryList ImportLine(string line)
{
if (line.IndexOf("'") == 0)
{
string py = line.Split(' ')[0];
string word = line.Split(' ')[1];
var wl = new WordLibrary();
wl.Word = word;
wl.Rank = 1;
wl.PinYin = py.Split(new[] { '\'' }, StringSplitOptions.RemoveEmptyEntries);
var wll = new WordLibraryList();
wll.Add(wl);
return(wll);
}
return(null);
}
public string ExportLine(WordLibrary wl)
{
var sb = new StringBuilder();
string py = wl.GetPinYinString("'", BuildType.None);
sb.Append(py);
sb.Append(" ");
sb.Append(wl.Word);
sb.Append(" ");
sb.Append(number);
sb.Append(" Z, ");
sb.Append(py);
sb.Append(" ");
sb.Append(number);
return(sb.ToString());
}
public WordLibraryList ImportLine(string line)
{
string[] sp = line.Split(',');
string word = sp[0];
int count = Convert.ToInt32(sp[1]);
var wl = new WordLibrary();
wl.Word = word;
wl.Count = count;
wl.PinYin = new string[] {};
var wll = new WordLibraryList();
wll.Add(wl);
return(wll);
}
public void TestExport1()
{
WordLibraryList wl = new WordLibraryList();
var wl1 = new WordLibrary()
{
Word = "曾毅曾诚", PinYin = new string[] { "zeng", "yi", "zeng", "cheng" }, CodeType = CodeType.Pinyin
};
wl.Add(wl1);
var export = new Win10MsPinyinSelfStudy();
//export.ExportFilePath = "c:\\Temp\\win10selfstudy5.dat";
var filePath = export.Export(wl);
Debug.WriteLine(filePath[0]);
}
public WordLibraryList ImportLine(string line)
{
string code = line.Split(' ')[0];
string word = line.Split(' ')[1];
var wl = new WordLibrary();
wl.Word = word;
wl.Count = DefaultRank;
wl.PinYin = ToArray(pinyinFactory.GetCodeOfString(word));
var wll = new WordLibraryList();
if (wl.PinYin.Length > 0)
{
wll.Add(wl);
}
return(wll);
}
//4字节使用同一个拼音的词条数x，2字节拼音长度n，n字节拼音的编号，（2字节汉字的长度y，y*2字节汉字的内容Unicode编码，2字节词频，2字节未知，4字节未知）*x
#region IWordLibraryImport Members
public WordLibraryList Import(string path)
{
var pyAndWord = new WordLibraryList();
var fs = new FileStream(path, FileMode.Open, FileAccess.Read);
fs.Position = 0x18;
CountWord = BinFileHelper.ReadInt32(fs);
CurrentStatus = 0;
fs.Position = 0x30;
while (CurrentStatus < CountWord)
{
int samePyCount = BinFileHelper.ReadInt16(fs);
int unkown1 = BinFileHelper.ReadInt16(fs);
short pyLength = BinFileHelper.ReadInt16(fs);
var pyArray = new string[pyLength / 2];
for (int i = 0; i < pyLength / 2; i++)
{
short idx = BinFileHelper.ReadInt16(fs);
try
{
pyArray[i] = PinYinDic[idx];
}
catch
{
pyArray[i] = "--";
}
}
for (int i = 0; i < samePyCount; i++)
{
short wordByteLength = BinFileHelper.ReadInt16(fs);
var wordArray = new byte[wordByteLength];
fs.Read(wordArray, 0, wordByteLength);
string word = Encoding.Unicode.GetString(wordArray);
short count = BinFileHelper.ReadInt16(fs);
short count2 = BinFileHelper.ReadInt16(fs);
int unknown = BinFileHelper.ReadInt32(fs); //不知道干啥的
var wl = new WordLibrary {
Count = count, Word = word, PinYin = pyArray
};
pyAndWord.Add(wl);
CurrentStatus++;
}
}
return(pyAndWord);
}
//private IWordCodeGenerater pyGenerater = new PinyinGenerater();
public WordLibraryList ImportLine(string str)
{
var list = new WordLibraryList();
string[] words = str.Split(' ');
for (int i = 1; i < words.Length; i++)
{
string word = words[i];
var wl = new WordLibrary();
wl.Word = word;
wl.Count = DefaultRank;
wl.SetCode(CodeType, words[0]);
//wl.PinYin = CollectionHelper.ToArray(pyGenerater.GetCodeOfString(word));
list.Add(wl);
}
return(list);
}
public string ExportLine(WordLibrary wl)
{
var codes = wl.Codes;
if (IsShortCode)
{
codes = new Code();
foreach (var c in wl.Codes)
{
codes.Add(new List <string>()
{
c[0][0].ToString()
});
}
}
return(string.Format(PhraseFormat, wl.Word, CollectionHelper.Descartes(codes)[0], wl.Rank == 0?DefaultRank:wl.Rank));
}
public virtual string ExportLine(WordLibrary wl)
{
var sb = new StringBuilder();
if (string.IsNullOrEmpty(wl.WubiCode))
{
sb.Append(wubiFactory.GetCodeOfString(wl.Word)[0]);
}
else
{
sb.Append(wl.WubiCode);
}
sb.Append(" ");
sb.Append(wl.Word);
return(sb.ToString());
}
public WordLibraryList ImportLine(string line)
{
line = line.Split(',')[0]; //如果有逗号，就只取第一个
string[] sp = line.Split(' ');
string py = sp[0];
string word = sp[1];
int count = Convert.ToInt32(sp[2]);
var wl = new WordLibrary();
wl.Word = word;
wl.Count = count;
wl.PinYin = py.Split(new[] { '\'' }, StringSplitOptions.RemoveEmptyEntries);
var wll = new WordLibraryList();
wll.Add(wl);
return(wll);
}
public virtual WordLibraryList ImportLine(string line)
{
var wlList = new WordLibraryList();
string[] strs = line.Split(' ');
for (int i = 1; i < strs.Length; i++)
{
string word = strs[i].Replace("，", ""); //把汉字中带有逗号的都去掉逗号
var list = pinyinFactory.GetCodeOfString(word);
for (int j = 0; j < list.Count; j++)
{
var wl = new WordLibrary();
wl.Word = word;
wl.PinYin = ToArray(list);
wlList.Add(wl);
}
}
return wlList;
}
public override WordLibraryList ImportLine(string line)
{
var wll = new WordLibraryList();
var array1 = line.Split('(');
string word = array1[0];
string py = array1[1].Split(')')[0];
var wl = new WordLibrary();
wl.Word = word;
wl.Rank = 1;
wl.PinYin = py.Split(new[] { '|' }, StringSplitOptions.RemoveEmptyEntries);
wll.Add(wl);
return(wll);
}
public string ExportLine(WordLibrary wl)
{
try
{
var sb = new StringBuilder();
string str = wl.Word;
for (int j = 0; j < str.Length; j++)
{
sb.Append(str[j] + wl.PinYin[j]);
}
return sb.ToString();
}
catch
{
return "";
}
}
public WordLibraryList ImportLine(string line)
{
var wl = new WordLibrary();
string[] array = line.Split('\t');
wl.Word = array[0];
if (array.Length == 2) //English
{
wl.IsEnglish = true;
wl.Count = Convert.ToInt32(array[1]);
}
else
{
string py = line.Split('\t')[1];
wl.PinYin = py.Split(new[] {'\''}, StringSplitOptions.RemoveEmptyEntries);
wl.Count = Convert.ToInt32(array[2]);
}
var wll = new WordLibraryList();
wll.Add(wl);
return wll;
}
public WordLibraryList Import(string path)
{
IWordCodeGenerater pinyinFactory = new WordPinyinGenerater();
IList<string> words = Parse(path);
var wll = new WordLibraryList();
foreach (string word in words)
{
var wl = new WordLibrary();
if (IsChinese(word)) //是中文就要进行注音
{
var list = pinyinFactory.GetCodeOfString(word);
wl.PinYin = ToArray(list);
}
else
{
wl.IsEnglish = true;
}
wl.Word = word;
wl.Count = DefaultRank;
wll.Add(wl);
}
return wll;
}
private WordLibrary ImportWord(FileStream fs)
{
var wordLibrary = new WordLibrary();
var temp = new byte[2];
fs.Read(temp, 0, 2);
short len = BitConverter.ToInt16(temp, 0);
fs.Read(temp, 0, 2); //what's the meaning of these 2 bytes?
var pinyinList = new List<string>();
for (int i = 0; i < len; i++)
{
temp = new byte[2];
fs.Read(temp, 0, 2);
pinyinList.Add(Shengmu[temp[0]] + Yunmu[temp[1]]);
}
wordLibrary.PinYin = pinyinList.ToArray();
temp = new byte[2*len];
fs.Read(temp, 0, 2*len);
wordLibrary.Word = Encoding.Unicode.GetString(temp);
//for (var i = 0; i < wordLibrary.Word.Length;i++ )
//{
// AddWordAndPinyin(wordLibrary.Word[i], wordLibrary.PinYin[i]);
//}
return wordLibrary;
}
public WordLibraryList ImportLine(string word)
{
string hz = "";
var py = new List<string>();
int j;
for (j = 0; j < word.Length - 1; j++)
{
hz += word[j];
if (word[j + 1] > 'z') //而且后面跟的不是拼音
{
py.Add(single.GetCodeOfChar(word[j]));
}
else //后面跟拼音
{
int k = 1;
string py1 = "";
while (j + k != word.Length && word[j + k] <= 'z')
{
py1 += word[j + k];
k++;
}
py.Add(py1);
j += k - 1; //减1是因为接下来会运行j++
}
}
if (j == word.Length - 1) //最后一个字是汉字
{
hz += word[j];
py.Add(single.GetCodeOfChar(word[j]));
}
var wl = new WordLibrary();
wl.PinYin = py.ToArray();
wl.Word = hz;
var wll = new WordLibraryList();
wll.Add(wl);
return wll;
}
public string ExportLine(WordLibrary wl)
{
return wl.Word + "\t" + (54999 + wl.Count);
}
public bool IsKeep(WordLibrary wl)
{
return !englishRegex.IsMatch(wl.Word);
}
public string ExportLine(WordLibrary wl)
{
string line = UserDefiningPattern.BuildWLString(wl);
return line;
}
private bool IsKeep(WordLibrary wordLibrary)
{
foreach (ISingleFilter filter in Filters)
{
if (!filter.IsKeep(wordLibrary))
{
return false;
}
}
return true;
}
private void GenerateCode( WordLibrary wl)
{
var word = wl.Word;
if (SelectedParsePattern.IsPinyin&&SelectedParsePattern.IsPinyinFormat)
{
var py = pyFactory.GetCodeOfString(word, SelectedParsePattern.CodeSplitString);
wl.PinYin = CollectionHelper.ToArray(py);
}
else
{
if (!string.IsNullOrEmpty(SelectedParsePattern.MappingTablePath))
{
SelectedParsePattern.MappingTable = UserCodingHelper.GetCodingDict(SelectedParsePattern.MappingTablePath);
}
selfFactory.MappingDictionary = SelectedParsePattern.MappingTable;
selfFactory.Is1Char1Code = SelectedParsePattern.IsPinyinFormat;
selfFactory.MutiWordCodeFormat = SelectedParsePattern.MutiWordCodeFormat;
wl.SetCode(CodeType.UserDefine, selfFactory.GetCodeOfString(word, SelectedParsePattern.CodeSplitString));
}
}
public WordLibraryList Import(string str)
{
GlobalCache.CharList.Clear();
GlobalCache.Stackes.Clear();
GlobalCache.WordList.Clear();
var fs = new FileStream(str, FileMode.Open, FileAccess.Read);
ParseHeader(fs);
TouchPalChar rootChar = TouchPalChar.Load(fs); //载入第一个字
LoadTree(fs, rootChar);
fs.Close();
var wwl = new WordLibraryList();
foreach (int i in GlobalCache.WordList.Keys)
{
TouchPalWord w = GlobalCache.WordList[i];
var wl = new WordLibrary();
wl.Count = w.Count;
wl.PinYin = w.PinYin.ToArray();
wl.Word = w.ChineseWord;
//sb.AppendLine(py + "\t" + GlobalCache.WordList[i].ChineseWord + "\t" + GlobalCache.WordList[i].Count);
wwl.Add(wl);
}
return wwl;
}
/// <summary>
/// 将一个词加入到最后一个节点
/// </summary>
/// <param name="rootChar"></param>
/// <param name="wl"></param>
/// <param name="begin"></param>
/// <returns></returns>
private TouchPalChar AddWordLink2Char(TouchPalChar rootChar, WordLibrary wl, int begin)
{
var rootWord = new TouchPalWord {ChineseWord = wl.Word, Count = wl.Count};
TouchPalChar lastChar = rootChar;
if (begin > 0)
{
lastChar = lastChar.Word.Chars[begin - 1];
}
var chars = new TouchPalChar[wl.Word.Length];
rootWord.Chars = chars;
for (int i = 0; i < begin; i++)
{
chars[i] = rootChar.Word.Chars[i];
}
for (int i = begin; i < wl.Word.Length; i++)
{
char c = wl.Word[i];
string py = wl.PinYin[i];
var tpc = new TouchPalChar();
tpc.Char = c;
tpc.PinyinCode = GlobalCache.PinyinIndexMapping[py];
tpc.WordIndex = i + 1;
short s = tpc.IndexAndPinYin;
AddChar2Next(lastChar, tpc);
lastChar = tpc;
chars[i] = lastChar;
}
lastChar.Word = rootWord;
return lastChar;
}
private void btnTest_Click(object sender, EventArgs e)
{
var lines= rtbFrom.Text.Split(new char[] {'\r', '\n'}, StringSplitOptions.RemoveEmptyEntries);
StringBuilder sb=new StringBuilder();
foreach (var line in lines)
{
WordLibrary wl=new WordLibrary(){Word = line.Trim(),Count = 1234};
GenerateCode( wl);
sb.Append(SelectedParsePattern.BuildWlString(wl)+"\r\n");
}
rtbTo.Text = sb.ToString();
}
public WordLibraryList Import(string path)
{
var pyAndWord = new WordLibraryList();
var fs = new FileStream(path, FileMode.Open, FileAccess.Read);
fs.Position = 0x18;
CountWord = BinFileHelper.ReadInt32(fs);
CurrentStatus = 0;
fs.Position = 0x30;
while (CurrentStatus < CountWord)
{
int samePyCount = BinFileHelper.ReadInt16(fs);
int unkown1 = BinFileHelper.ReadInt16(fs);
short pyLength = BinFileHelper.ReadInt16(fs);
var pyArray = new string[pyLength/2];
for (int i = 0; i < pyLength/2; i++)
{
short idx = BinFileHelper.ReadInt16(fs);
try
{
pyArray[i] = PinYinDic[idx];
}
catch
{
pyArray[i] = "--";
}
}
for (int i = 0; i < samePyCount; i++)
{
short wordByteLength = BinFileHelper.ReadInt16(fs);
var wordArray = new byte[wordByteLength];
fs.Read(wordArray, 0, wordByteLength);
string word = Encoding.Unicode.GetString(wordArray);
short count = BinFileHelper.ReadInt16(fs);
short count2 = BinFileHelper.ReadInt16(fs);
int unknown = BinFileHelper.ReadInt32(fs); //不知道干啥的
var wl = new WordLibrary {Count = count, Word = word, PinYin = pyArray};
pyAndWord.Add(wl);
CurrentStatus++;
}
}
return pyAndWord;
}
public string ExportLine(WordLibrary wl)
{
throw new NotImplementedException();
}
/// <summary>
/// 把一个词条一个字一个字的写入词库文件中
/// </summary>
/// <param name="fs"></param>
/// <param name="wl"></param>
/// <param name="isLastWord"></param>
/// <param name="from"></param>
/// <returns></returns>
public int WriteWord(FileStream fs, WordLibrary wl, bool isLastWord)
{
var beginPosition = (int) fs.Position;
int wordLength = wl.Word.Length;
int charIndex;
TouchPalChar stackChar = FindBeginPosition(wl.Word, out charIndex);
for (int i = charIndex; i < wordLength; i++)
{
var item = new TouchPalChar();
item.Char = wl.Word[i];
item.BeginPosition = (int) fs.Position;
string py = wl.PinYin[i];
int pyIndex = GlobalCache.PinyinIndexMapping[py];
var code = (short) (((i + 1) << 11) + pyIndex);
fs.Write(BitConverter.GetBytes(code), 0, 2);
int p1 = 0; //词频位置
if (i == wordLength - 1) //最后一个字
{
p1 = beginPosition + wordLength*26;
}
fs.Write(BitConverter.GetBytes(p1), 0, 4);
int p2 = 0; //下个字位置
if (i != wordLength - 1)
{
p2 = beginPosition + (i + 1)*26;
}
fs.Write(BitConverter.GetBytes(p2), 0, 4);
int p3 = 0; //跳转位置
if (!isLastWord && i == 0)
{
p3 = beginPosition + wordLength*28 + 5;
}
fs.Write(BitConverter.GetBytes(p3), 0, 4);
int p4 = 0; //上个字位置
if (charIndex == 0)
{
if (i == 0)
{
p4 = GlobalCache.JumpChar.BeginPosition;
GlobalCache.JumpChar = item;
}
else
{
p4 = beginPosition + (i - 1)*26;
}
}
else
{
p4 = stackChar.BeginPosition;
}
fs.Write(BitConverter.GetBytes(p4), 0, 4);
int p5 = 4;
if (charIndex == 0)
{
if (i != 0)
{
p5 = p4;
}
}
else
{
p5 = stackChar.PrevValidCharPosition;
}
item.PrevValidCharPosition = p5;
fs.Write(BitConverter.GetBytes(p5), 0, 4);
int p6 = 0;
fs.Write(BitConverter.GetBytes(p6), 0, 4);
GlobalCache.ExportStackes.Push(item);
}
int count = 96; // wl.Count;
fs.Write(BitConverter.GetBytes(count), 0, 4);
fs.WriteByte(0); //这个字节不知道干什么的
byte[] wordByte = Encoding.Unicode.GetBytes(wl.Word);
fs.Write(wordByte, 0, wordByte.Length);
return beginPosition;
}
public virtual string ExportLine(WordLibrary wl)
{
return wl.Word;
}