public override void GetCodeOfWordLibrary(WordLibrary wl)
{
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 void GetCodeOfWordLibrary(WordLibrary wl)
{
if (wl.CodeType == CodeType.Pinyin)
{
var code = ChaoyinHelper.GetChaoyin(wl.PinYin);
wl.SetCode(CodeType.Chaoyin, code);
}
else
{
wl.SetCode(CodeType.Chaoyin, GetCodeOfString(wl.Word));
}
}
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 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 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);
}
private WordLibrary ReadOnePhrase(FileStream fs, int nextStartPosition)
{
WordLibrary wl = new WordLibrary();
var magic = BinFileHelper.ReadInt32(fs);
var hanzi_offset = BinFileHelper.ReadInt16(fs);
wl.Rank = fs.ReadByte();
var x6 = fs.ReadByte(); //不知道干啥的
var unknown8 = BinFileHelper.ReadInt64(fs); //新增的，不知道什么意思
var pyBytesLen = hanzi_offset - 18;
var pyBytes = BinFileHelper.ReadArray(fs, pyBytesLen);
var wubiStr = Encoding.Unicode.GetString(pyBytes);
var split = BinFileHelper.ReadInt16(fs); //00 00 分割拼音和汉字
var wordBytesLen = nextStartPosition - (int)fs.Position - 2; //结尾还有个00 00
var wordBytes = BinFileHelper.ReadArray(fs, wordBytesLen);
BinFileHelper.ReadInt16(fs); //00 00分割
var word = Encoding.Unicode.GetString(wordBytes);
wl.Word = word;
try
{
wl.SetCode(CodeType.Wubi98, wubiStr);
}
catch
{
return(null);
}
wl.CodeType = CodeType.Wubi98;
return(wl);
}
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)
{
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);
}
//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 override 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 WordLibrary ReadOnePhrase(FileStream fs, DictCodeType type)
{
WordLibrary wl = new WordLibrary();
var codeBytesLen = fs.ReadByte();
var wordBytesLen = fs.ReadByte();
var split = fs.ReadByte();
// 0x64对应正常词组（包含中英混拼，如"阿Q"）。
Debug.Assert(split.Equals(0x64) || split.Equals(0x32) ||
split.Equals(0x10) || split.Equals(0x66) ||
split.Equals(0x67)); // 0x67: "$X[计算器]calc"
var codeBytes = BinFileHelper.ReadArray(fs, codeBytesLen);
var codeStr = Encoding.ASCII.GetString(codeBytes);
var wordBytes = BinFileHelper.ReadArray(fs, wordBytesLen);
var word = Encoding.Unicode.GetString(wordBytes);
if (split.Equals(0x32)) // 如“醃(腌)”，后者是相应简化字？
{
word = word.Substring(0, 1); // 暂定只取首字
}
Debug.Assert(word.IndexOf("(") < 0);
wl.Word = word;
try
{
if (type == DictCodeType.Pinyin)
{
wl.CodeType = CodeType.Pinyin;
wl.SetPinyinString(codeStr);
}
else if (type == DictCodeType.Wubi98)
{
wl.CodeType = CodeType.Wubi98;
wl.SetCode(CodeType.Wubi98, codeStr);
}
}
catch
{
wl.CodeType = CodeType.NoCode;
ImportLineErrorNotice?.Invoke(wl.Word + " 的编码缺失");
}
return(wl);
}
/// <summary>
/// 根据Pattern设置的格式，对输入的一行该格式的字符串转换成WordLibrary
/// </summary>
/// <param name="line"></param>
/// <returns></returns>
public WordLibrary BuildWordLibrary(string line)
{
var wl = new WordLibrary();
wl.CodeType = UserDefiningPattern.CodeType;
string[] strlist = line.Split(new[] { UserDefiningPattern.SplitString }, StringSplitOptions.RemoveEmptyEntries);
var newSort = new List <int>(UserDefiningPattern.Sort);
newSort.Sort();
string code = "", word = "";
int rank = 0;
int index1 = UserDefiningPattern.Sort.FindIndex(i => i == newSort[0]); //最小的一个
if (index1 == 0 && UserDefiningPattern.ContainCode) //第一个是编码
{
code = strlist[0];
}
if (index1 == 1) //第一个是汉字
{
word = strlist[0];
}
if (index1 == 2 && UserDefiningPattern.ContainRank) //第一个是词频
{
rank = Convert.ToInt32(strlist[0]);
}
if (strlist.Length > 1)
{
int index2 = UserDefiningPattern.Sort.FindIndex(i => i == newSort[1]); //中间的一个
if (index2 == 0 && UserDefiningPattern.ContainCode) //一个是Code
{
code = strlist[1];
}
if (index2 == 1)
{
word = strlist[1];
}
if (index2 == 2 && UserDefiningPattern.ContainRank)
{
rank = Convert.ToInt32(strlist[1]);
}
}
if (strlist.Length > 2)
{
int index2 = UserDefiningPattern.Sort.FindIndex(i => i == newSort[2]); //最大的一个
if (index2 == 0 && UserDefiningPattern.ContainCode) //第一个是拼音
{
code = strlist[2];
}
if (index2 == 1)
{
word = strlist[2];
}
if (index2 == 2 && UserDefiningPattern.ContainRank)
{
rank = Convert.ToInt32(strlist[2]);
}
}
wl.Word = word;
wl.Rank = rank;
if (code != "")
{
if (UserDefiningPattern.IsPinyinFormat)
{
string[] codes = code.Split(new[] { UserDefiningPattern.CodeSplitString },
StringSplitOptions.RemoveEmptyEntries);
wl.SetCode(UserDefiningPattern.CodeType, new List <string>(codes), UserDefiningPattern.IsPinyinFormat);
}
else
{
wl.SetCode(UserDefiningPattern.CodeType, code);
}
}
return(wl);
}
