public WordLibraryList Import(string path)
{
WordLibraryList re = new WordLibraryList();
FileStream fp = File.OpenRead(path);
int user_word_base = 0x2400;
//get word num
byte[] bytes = new byte[50];
fp.Seek(12, SeekOrigin.Begin);
fp.Read(bytes, 0, 4);
int cnt = bytesToIntLittle(bytes, 0, 4);
//get each word
for (int i = 0; i < cnt; i++)
{
int cur_idx = user_word_base + i * 60;
//get word len
fp.Seek(cur_idx + 10, SeekOrigin.Begin);
fp.Read(bytes, 0, 1);
int wordLen = bytesToIntLittle(bytes, 0, 1);
//get word
fp.Seek(cur_idx + 12, SeekOrigin.Begin);
fp.Read(bytes, 0, wordLen * 2);
string word = Encoding.Unicode.GetString(bytes, 0, wordLen * 2);
re.Add(new WordLibrary()
{
Word = word, CodeType = this.CodeType,
});
}
fp.Close();
return(re);
}
public override WordLibraryList ImportLine(string line)
{
var wlList = new WordLibraryList();
string[] strs = line.Split(' ');
for (int i = 1; i < strs.Length; i++)
{
string oriWord = strs[i];
string word = oriWord.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 = oriWord;
//if (IsWubi)
//{
// wl.SetCode(CodeType.Wubi, strs[0]);
//}
//wl.PinYin = CollectionHelper.ToArray(list);
wl.SetCode(CodeType, strs[0]);
wlList.Add(wl);
//}
}
return(wlList);
}
public WordLibraryList Import(string path)
{
//IWordCodeGenerater pinyinFactory = new PinyinGenerater();
IList <string> words = Parse(path);
var wll = new WordLibraryList();
foreach (string word in words)
{
var wl = new WordLibrary();
//词典转换，不进行注音操作，以提高速度
//if (IsChinese(word)) //是中文就要进行注音
//{
// var list = pinyinFactory.GetCodeOfString(word);
// wl.PinYin = CollectionHelper.ToArray(list);
//}
//else
{
wl.IsEnglish = true;
}
wl.Word = word;
wl.Count = DefaultRank;
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++)
{
var oriWord = strs[i];
string word = oriWord.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 = oriWord;
//if (IsWubi)
//{
// wl.SetCode(CodeType.Wubi, strs[0]);
//}
//wl.PinYin = CollectionHelper.ToArray(list);
wl.SetCode(this.CodeType,strs[0]);
wlList.Add(wl);
//}
}
return wlList;
}
public WordLibraryList ImportText(string str)
{
var xmlDoc = new XmlDocument();
xmlDoc.LoadXml(str);
var wlList = new WordLibraryList();
XmlNodeList xns = xmlDoc.SelectNodes("//plist/array/dict");
CountWord = xns.Count;
for (int i = 0; i < xns.Count; i++)
{
XmlNode xn = xns[i];
var nodes = xn.SelectNodes("string");
var wl = new WordLibrary();
wl.Word = nodes[0].InnerText;
wl.Rank = 1;
wl.SetPinyinString(nodes[1].InnerText);
CurrentStatus = i;
wlList.Add(wl);
}
return(wlList);
}
public WordLibraryList ImportLine(string line)
{
string[] c = line.Split('\t');
var wl = new WordLibrary();
wl.Word = c[0];
wl.Count = DefaultRank;
string zhuyin = c[1];
var pys = new List <string>();
foreach (string zy in zhuyin.Split(','))
{
try
{
string py = ZhuyinHelper.GetPinyin(zy);
pys.Add(py);
}
catch (Exception ex)
{
Debug.WriteLine(ex.Message);
}
}
wl.PinYin = pys.ToArray();
var wll = new WordLibraryList();
wll.Add(wl);
return(wll);
}
public void TestGeneratePinyinThen2String()
{
ParsePattern parser = new ParsePattern()
{
IsPinyinFormat = true,
CodeSplitType = BuildType.FullContain,
CodeSplitString = "~",
ContainCode = true,
ContainRank = true,
SplitString = "|",
CodeType = CodeType.Pinyin,
LineSplitString = "\r",
Sort = new List <int>()
{
2, 1, 3
}
};
WordLibraryList wll = new WordLibraryList();
WordLibrary wl = new WordLibrary()
{
Word = "深蓝", Rank = 123, CodeType = CodeType.UserDefine
};
wl.Codes = new Code();
wl.Codes.Add(new[] { "sn" });
wl.Codes.Add(new[] { "ln" });
wll.Add(wl);
selfDefining.UserDefiningPattern = parser;
var str = selfDefining.Export(wll);
Assert.AreEqual(str, "深蓝|~shen~lan~|123\r");
}
public WordLibraryList ImportLine(string line)
{
if (line.Length > 0 && line[0] == ';')
{
return(null);
}
string[] sp = line.Split(' ');
string word = sp[0];
string[] py = new string[word.Length];
for (var i = 0; i < word.Length; i++)
{
py[i] = sp[i + 1];
}
var wl = new WordLibrary();
wl.Word = word;
wl.Count = 1;
wl.PinYin = py;
var wll = new WordLibraryList();
wll.Add(wl);
return(wll);
}
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);
}
public WordLibraryList Import(string path)
{
var pyAndWord = new WordLibraryList();
var fs = new FileStream(path, FileMode.Open, FileAccess.Read);
fs.Position = 0x10;
var phrase_offset_start = BinFileHelper.ReadInt32(fs);
var phrase_start = BinFileHelper.ReadInt32(fs);
var phrase_end = BinFileHelper.ReadInt32(fs);
var phrase_count = BinFileHelper.ReadInt32(fs);
fs.Position = phrase_offset_start;
var offsets = ReadOffsets(fs, phrase_count);
offsets.Add(phrase_end - phrase_start);
fs.Position = phrase_start;
for (var i = 0; i < phrase_count; i++)
{
var wl = ReadOnePhrase(fs, phrase_start + offsets[i + 1]);
if (wl != null)
{
pyAndWord.Add(wl);
}
}
return(pyAndWord);
}
private WordLibraryList Filter(WordLibraryList wlList)
{
var result = new WordLibraryList();
IReplaceFilter replace = null;
if (PinyinType != PinyinType.FullPinyin)
{
replace = new ShuangpinReplacer(PinyinType);
}
foreach (var wl in wlList)
{
if (replace != null)
{
replace.Replace(wl);
}
if (wl.GetPinYinLength() > 32)
{
continue;
}
if (wl.Word.Length > 64)
{
continue;
}
result.Add(wl);
}
return(result);
}
public WordLibraryList ImportLine(string line)
{
var wlList = new WordLibraryList();
WordLibrary wl = UserDefiningPattern.BuildWordLibrary(line);
wlList.Add(wl);
return wlList;
}
//private IWordCodeGenerater pyGenerater=new PinyinGenerater();
public override WordLibraryList ImportLine(string line)
{
string[] lineArray = line.Split('\t');
string word = lineArray[0];
string code = lineArray[1];
var wl = new WordLibrary();
wl.Word = word;
wl.Rank = Convert.ToInt32(lineArray[2]);
if (CodeType == CodeType.Pinyin)
{
wl.PinYin = code.Split(new[] { ' ' }, StringSplitOptions.RemoveEmptyEntries);
}
else
{
//wl.PinYin = CollectionHelper.ToArray(pyGenerater.GetCodeOfString(wl.Word));
wl.SetCode(CodeType, code);
}
var wll = new WordLibraryList();
wll.Add(wl);
return(wll);
}
public WordLibraryList ImportText(string str)
{
var xmlDoc = new XmlDocument();
xmlDoc.LoadXml(str);
var namespaceManager = new XmlNamespaceManager(xmlDoc.NameTable);
namespaceManager.AddNamespace("ns1", "http://www.microsoft.com/ime/dctx");
var wlList = new WordLibraryList();
XmlNodeList xns = xmlDoc.SelectNodes("//ns1:Dictionary/ns1:DictionaryEntry", namespaceManager);
CountWord = xns.Count;
for (int i = 0; i < xns.Count; i++)
{
XmlNode xn = xns[i];
string py = xn.SelectSingleNode("ns1:InputString", namespaceManager).InnerText;
string word = xn.SelectSingleNode("ns1:OutputString", namespaceManager).InnerText;
var wl = new WordLibrary();
wl.Word = word;
wl.Rank = 1;
wl.PinYin = py.Split(new[] { ' ', '1', '2', '3', '4' }, StringSplitOptions.RemoveEmptyEntries);
CurrentStatus = i;
wlList.Add(wl);
}
return(wlList);
}
public WordLibraryList ImportLine(string line)
{
var wlList = new WordLibraryList();
WordLibrary wl = BuildWordLibrary(line);
wlList.Add(wl);
return(wlList);
}
public WordLibraryList Import(string path)
{
int endPosition = 0;
var wordLibraryList = new WordLibraryList();
var fs = new FileStream(path, FileMode.Open, FileAccess.Read);
fs.Position = 0x60;
endPosition = BinFileHelper.ReadInt32(fs);
fs.Position = 0x350;
CurrentStatus = 0;
do
{
//CurrentStatus++;
try
{
WordLibrary wl = ImportWord(fs);
if (wl == null)
{
break;
}
if (wl.Word != "" && wl.PinYin.Length > 0)
{
wordLibraryList.Add(wl);
}
}
catch (Exception ex)
{
Debug.WriteLine(ex.Message);
}
} while (fs.Position != endPosition); //< 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);
}
/// <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;
}
/// <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[] c = line.Split('\t');
var wl = new WordLibrary();
wl.Word = c[0];
wl.Rank = Convert.ToInt32(c[2]);
wl.PinYin = c[1].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.Rank = 1;
wl.PinYin = py.Split(new[] {'\''}, StringSplitOptions.RemoveEmptyEntries);
var wll = new WordLibraryList();
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
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
if (pyArray.Length == word.Length)
{
var wl = new WordLibrary {
Rank = count, Word = word, PinYin = pyArray
};
pyAndWord.Add(wl);
}
else
{
Debug.WriteLine("Error data: word:[" + word + "] pinyin:[" + string.Join(",", pyArray) + "]");
}
CurrentStatus++;
}
}
return(pyAndWord);
}
/// <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 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)
{
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;
}
/// <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[] 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 WordLibraryList ImportLine(string line)
{
string[] c = line.Split('\t');
var wl = new WordLibrary();
wl.Word = c[0];
wl.Rank = Convert.ToInt32(c[2]);
wl.PinYin = c[1].Split(new[] { '\'' }, StringSplitOptions.RemoveEmptyEntries);
var wll = new WordLibraryList();
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.AddCode(CodeType, code);
var wll = new WordLibraryList();
wll.Add(wl);
return wll;
}
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.Rank = count;
wl.PinYin = new string[] {};
var wll = new WordLibraryList();
wll.Add(wl);
return wll;
}
public WordLibraryList ImportLine(string line)
{
string[] sp = line.Split(' ');
string py = sp[1];
string word = sp[0];
var wl = new WordLibrary {CodeType = CodeType.Pinyin};
wl.Word = word;
wl.Rank = DefaultRank;
wl.PinYin = py.Split(new[] {'\''}, StringSplitOptions.RemoveEmptyEntries);
var wll = new WordLibraryList();
wll.Add(wl);
return wll;
}
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 WordLibraryList Import(string path)
{
var pyAndWord = new WordLibraryList();
var fs = new FileStream(path, FileMode.Open, FileAccess.Read);
fs.Position = 0x00;
var headerstr = "Freeime Dictionary";
var header = Encoding.ASCII.GetString(BinFileHelper.ReadArray(fs, headerstr.Length));
Debug.Assert(header.Equals(headerstr));
DictCodeType curType;
fs.Position = 0x23;
var headerTypeBytes = BinFileHelper.ReadArray(fs, 4);
var headerTypeStr = Encoding.Unicode.GetString(headerTypeBytes);
if (headerTypeStr.Equals("拼音"))
{
curType = DictCodeType.Pinyin;
}
else if (headerTypeStr.Equals("五笔"))
{
curType = DictCodeType.Wubi98;
}
else
{
throw new NotImplementedException("未知词库，请在反馈中提交文件");
}
var phrase_start = 0x1B620; // 'a'词条所在
fs.Position = phrase_start;
while (true)
{
var wl = ReadOnePhrase(fs, curType);
if (wl != null)
{
pyAndWord.Add(wl);
}
if (fs.Length == fs.Position) //文件结束
{
fs.Close();
break;
}
}
return(pyAndWord);
}
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 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)
{
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 Filter(WordLibraryList list)
{
if (Percentage == 100)
{
return list;
}
int count = list.Count*Percentage/100;
list.Sort((a, b) => a.Rank - b.Rank);
var result = new WordLibraryList();
for (int i = 0; i < count; i++)
{
result.Add(list[i]);
}
return result;
}
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.Rank = 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);
}
private WordLibraryList Filter(WordLibraryList wlList)
{
var result = new WordLibraryList();
foreach (var wl in wlList)
{
if (wl.Word.Length > 12 || wl.Word.Length == 1)//最多支持12个字
{
continue;
}
result.Add(wl);
}
return(result);
}
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[] 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 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 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 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.Rank = DefaultRank;
wl.SetCode(CodeType.Wubi, code);
//wl.PinYin = CollectionHelper.ToArray(pinyinFactory.GetCodeOfString(word));
var wll = new WordLibraryList();
if (wl.PinYin.Length > 0)
{
wll.Add(wl);
}
return wll;
}
//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 WordLibraryList Import(string str)
{
WordLibraryList wlList = new WordLibraryList();
var lines = str.Split(new string[] { "\r\n" }, StringSplitOptions.RemoveEmptyEntries);
for (int i = 0; i < lines.Length; i++)
{
string line = lines[i];
var c = line.Split('\t');
WordLibrary wl = new WordLibrary();
wl.Word = c[0];
wl.Count = Convert.ToInt32(c[1]);
wl.PinYin = c[2].Split(new char[] { ' ' }, StringSplitOptions.RemoveEmptyEntries);
wlList.Add(wl);
}
return wlList;
}
public void TestExportExtCodeLots()
{
string str="深蓝词库转换测试代码";
var list = new WordLibraryList();
var ts = "";
foreach (var c in str)
{
ts += c;
list.Add(new WordLibrary() {Count = 10, IsEnglish = false, Word = ts});
}
export.UserDefiningPattern = InitPattern();
export.UserDefiningPattern.MappingTablePath = "Test\\array30.txt";
var x = export.Export(list);
Debug.WriteLine(x);
Assert.IsNotNullOrEmpty(str);
}
public WordLibraryList ImportLine(string line)
{
string[] sp = line.Split('\t');
string word = sp[0];
string py = sp[1];
int count = 1;
var wl = new WordLibrary();
wl.Word = word;
wl.Rank = count;
wl.PinYin = py.Split(new[] {'\''}, StringSplitOptions.RemoveEmptyEntries);
var wll = new WordLibraryList();
if (!string.IsNullOrEmpty(py))
{
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 WordLibraryList ImportLine(string line)
{
if (line.Length > 0 && line[0] == ';')
return null;
string[] sp = line.Split(' ');
string word = sp[0];
var py = new string[word.Length];
for (int i = 0; i < word.Length; i++)
{
py[i] = sp[i + 1];
}
var wl = new WordLibrary();
wl.Word = word;
wl.Rank = 1;
wl.PinYin = py;
var wll = new WordLibraryList();
wll.Add(wl);
return wll;
}
public WordLibraryList ImportLine(string line)
{
var wll = new WordLibraryList();
try
{
string py = line.Split(' ')[1];
string word = line.Split(' ')[0];
var wl = new WordLibrary();
wl.Word = word;
wl.Rank = 1;
wl.PinYin = py.Split(new[] {'|'}, StringSplitOptions.RemoveEmptyEntries);
wll.Add(wl);
}
catch (Exception ex)
{
Debug.WriteLine(line + "\t" + ex.Message);
}
return wll;
}
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;
}
public WordLibraryList Import(string path)
{
//IWordCodeGenerater pinyinFactory = new PinyinGenerater();
IList<string> words = Parse(path);
var wll = new WordLibraryList();
foreach (string word in words)
{
var wl = new WordLibrary();
//词典转换，不进行注音操作，以提高速度
//if (IsChinese(word)) //是中文就要进行注音
//{
// var list = pinyinFactory.GetCodeOfString(word);
// wl.PinYin = CollectionHelper.ToArray(list);
//}
//else
{
wl.IsEnglish = true;
}
wl.Word = word;
wl.Rank = DefaultRank;
wll.Add(wl);
}
return wll;
}
public void TestGeneratePinyinThen2String()
{
ParsePattern parser = new ParsePattern()
{
IsPinyinFormat = true,
CodeSplitType = BuildType.FullContain,
CodeSplitString = "~",
ContainCode = true,
ContainRank = true,
SplitString = "|",
CodeType = CodeType.Pinyin,
LineSplitString = "\r",
Sort = new List<int>() { 2, 1, 3 }
};
WordLibraryList wll = new WordLibraryList();
WordLibrary wl = new WordLibrary() { Word = "深蓝", Rank = 123, CodeType = CodeType.UserDefine };
wl.Codes = new Code();
wl.Codes.Add(new[] { "sn" });
wl.Codes.Add( new[] { "ln" });
wll.Add(wl);
selfDefining.UserDefiningPattern = parser;
var str = selfDefining.Export(wll);
Assert.AreEqual(str, "深蓝|~shen~lan~|123\r");
}
//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.AddCode(CodeType, words[0]);
//wl.PinYin = CollectionHelper.ToArray(pyGenerater.GetCodeOfString(word));
list.Add(wl);
}
return list;
}
public WordLibraryList ImportLine(string line)
{
string[] c = line.Split('\t');
var wl = new WordLibrary();
wl.Word = c[0];
wl.Rank = DefaultRank;
string zhuyin = c[1];
var pys = new List<string>();
foreach (string zy in zhuyin.Split(','))
{
try
{
string py = ZhuyinHelper.GetPinyin(zy);
pys.Add(py);
}
catch (Exception ex)
{
Debug.WriteLine(ex.Message);
}
}
wl.PinYin = pys.ToArray();
var wll = new WordLibraryList();
wll.Add(wl);
return wll;
}
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;
}
private WordLibraryList Filter(WordLibraryList list)
{
var result = new WordLibraryList();
foreach (WordLibrary wordLibrary in list)
{
if (IsKeep(wordLibrary))
{
result.Add(wordLibrary);
}
}
return result;
}
