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);
}
public WordLibraryList Import(string path)
{
var pyAndWord = new WordLibraryList();
var fs = new FileStream(path, FileMode.Open, FileAccess.Read);
fs.Position = 0x44;
CountWord = BinFileHelper.ReadInt32(fs);
int segmentCount = BinFileHelper.ReadInt32(fs); //分为几段
CurrentStatus = 0;
for (int i = 0; i < segmentCount; i++)
{
try
{
fs.Position = 0xC00 + 1024*i;
var segment = new Segment(fs);
pyAndWord.AddWordLibraryList(segment.WordLibraryList);
CurrentStatus += segment.WordLibraryList.Count;
}
catch (Exception e)
{
Debug.WriteLine(e.Message);
}
}
return pyAndWord;
}
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;
}
public WordLibraryList ImportText(string str)
{
var wlList = new WordLibraryList();
string[] lines = str.Split(new[] { '\r', '\n' }, StringSplitOptions.RemoveEmptyEntries);
CountWord = lines.Length;
for (int i = 0; i < lines.Length; i++)
{
string line = lines[i];
CurrentStatus = i;
if (line.Length > 0 && line[0] != '#')//注释行
{
try
{
wlList.AddWordLibraryList(ImportLine(line));
}
catch (Exception ex)
{
var log = string.Format("[{0}] 不是一个有效的谷歌拼音词库格式，解析时抛出异常：{1}", line, ex.Message);
throw new Exception(log);
}
}
}
return(wlList);
}
//{0x05 2word
//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 = 0x44;
CountWord = BinFileHelper.ReadInt32(fs);
var segmentCount = BinFileHelper.ReadInt32(fs); //分为几段
CurrentStatus = 0;
for (int i = 0; i < segmentCount; i++)
{
try
{
fs.Position = 0xC00 + 1024 * i;
Segment segment = new Segment(fs);
pyAndWord.AddWordLibraryList(segment.WordLibraryList);
CurrentStatus += segment.WordLibraryList.Count;
}
catch (Exception e)
{
Debug.WriteLine(e.Message);
}
}
return(pyAndWord);
}
public WordLibraryList ImportText(string str)
{
var wlList = new WordLibraryList();
string[] lines = str.Split(new[] {'\r','\n'}, StringSplitOptions.RemoveEmptyEntries);
CountWord = lines.Length;
for (int i = 1; i < lines.Length; i++)
{
string line = lines[i];
CurrentStatus = i;
wlList.AddWordLibraryList(ImportLine(line));
}
return wlList;
}
public WordLibraryList Import(string path)
{
var wll = new WordLibraryList();
string txt = ParseQpyd(path);
foreach (string line in txt.Split('\n'))
{
if (line != "")
{
wll.AddWordLibraryList(ImportLine(line));
}
}
return wll;
}
public WordLibraryList ImportText(string str)
{
var wlList = new WordLibraryList();
string[] lines = str.Split(new[] { "\r\n" }, StringSplitOptions.RemoveEmptyEntries);
CountWord = lines.Length;
for (int i = 0; i < lines.Length; i++)
{
string line = lines[i];
CurrentStatus = i;
wlList.AddWordLibraryList(ImportLine(line));
}
return(wlList);
}
public WordLibraryList Import(string path)
{
var wll = new WordLibraryList();
string txt = ParseQpyd(path);
foreach (string line in txt.Split('\n'))
{
if (line != "")
{
wll.AddWordLibraryList(ImportLine(line));
}
}
return(wll);
}
public WordLibraryList ImportText(string str)
{
var wlList = new WordLibraryList();
string[] lines = str.Split(new[] {"\r\n"}, StringSplitOptions.RemoveEmptyEntries);
CountWord = lines.Length;
for (int i = 0; i < lines.Length; i++)
{
string line = lines[i];
CurrentStatus = i;
if (line.IndexOf("'") == 0)
{
wlList.AddWordLibraryList(ImportLine(line));
}
}
return wlList;
}
public WordLibraryList ImportText(string str)
{
var wlList = new WordLibraryList();
string[] lines = str.Split(new[] { '\r', '\n' }, StringSplitOptions.RemoveEmptyEntries);
CountWord = lines.Length;
for (int i = 0; i < lines.Length; i++)
{
string line = lines[i];
CurrentStatus = i;
if (line.Length > 0 && line[0] != '#')//注释行
{
wlList.AddWordLibraryList(ImportLine(line));
}
}
return(wlList);
}
public WordLibraryList ImportText(string text)
{
var list = new WordLibraryList();
string[] lines = text.Split(new[] { '\n', '\r' }, StringSplitOptions.RemoveEmptyEntries);
CountWord = lines.Length;
CurrentStatus = 0;
foreach (string s in lines)
{
CurrentStatus++;
if (IsContent(s))
{
list.AddWordLibraryList(ImportLine(s));
}
}
return(list);
}
public WordLibraryList ImportText(string str)
{
var wlList = new WordLibraryList();
string[] words = str.Split(new[] { '\r', '\n' }, StringSplitOptions.RemoveEmptyEntries);
CountWord = words.Length;
for (int i = 0; i < words.Length; i++)
{
CurrentStatus = i;
try
{
wlList.AddWordLibraryList(ImportLine(words[i]));
}
catch
{
}
}
return(wlList);
}
public WordLibraryList ImportText(string str)
{
var wlList = new WordLibraryList();
string[] lines = str.Split(new[] { "\n" }, StringSplitOptions.RemoveEmptyEntries);
CountWord = lines.Length;
for (int i = 1; i < lines.Length; i++)
{
string line = lines[i];
CurrentStatus = i;
try
{
wlList.AddWordLibraryList(ImportLine(line));
}
catch
{
SendImportLineErrorNotice("无效的词条，解析失败：" + line);
}
}
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 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 WordLibraryList ImportText(string str)
{
var wlList = new WordLibraryList();
string[] lines = str.Split(new[] {"\r", "\n"}, StringSplitOptions.RemoveEmptyEntries);
CountWord = lines.Length;
for (int i = 0; i < lines.Length; i++)
{
string line = lines[i];
CurrentStatus = i;
if (line.StartsWith("#"))
{
continue;
}
wlList.AddWordLibraryList(ImportLine(line));
}
return wlList;
}
public WordLibraryList ImportText(string text)
{
var list = new WordLibraryList();
string[] lines = text.Split(new[] {'\n', '\r'}, StringSplitOptions.RemoveEmptyEntries);
CountWord = lines.Length;
CurrentStatus = 0;
foreach (string s in lines)
{
CurrentStatus++;
if (IsContent(s))
{
list.AddWordLibraryList(ImportLine(s));
}
}
return list;
}
public WordLibraryList ImportText(string str)
{
var wlList = new WordLibraryList();
string[] words = str.Split(new[] {'\r', '\n'}, StringSplitOptions.RemoveEmptyEntries);
CountWord = words.Length;
for (int i = 0; i < words.Length; i++)
{
CurrentStatus = i;
try
{
wlList.AddWordLibraryList(ImportLine(words[i]));
}
catch
{
}
}
return wlList;
}
