private void MergeKeywords(string[] keys, int id, string keyword, List <Tuple <string, string[]> > list)
{
if (id >= keys.Length)
{
list.Add(Tuple.Create(keyword, keys));
//list[keyword.Substring(1)] = keys;
//list.Add(keyword.Substring(1));
return;
}
var key = keys[id];
if (key[0] >= 0x3400 && key[0] <= 0x9fd5)
{
var all = PinyinDict.GetAllPinyin(key[0]);
var fpy = new HashSet <char>();
foreach (var item in all)
{
fpy.Add(item[0]);
}
foreach (var item in fpy)
{
MergeKeywords(keys, id + 1, keyword + item, list);
}
}
else
{
MergeKeywords(keys, id + 1, keyword + key[0], list);
}
}
/// <summary>
/// 设置关键字，注：索引会被清空
/// </summary>
/// <param name="keywords"></param>
public void SetKeywords(ICollection <string> keywords)
{
_keywords.AddRange(keywords);
for (int i = 0; i < _keywords.Count; i++)
{
var text = _keywords[i];
var pys = PinyinDict.GetPinyinList(text);
string fpy = "";
ulong hash = 0;
for (int j = 0; j < pys.Length; j++)
{
pys[j] = pys[j].ToUpper();
fpy += pys[j][0];
hash = BuildHashByChar(hash, pys[j][0]);
}
if (Regex.IsMatch(text, "[^a-zA-Z0-9]", RegexOptions.Compiled))//关键字中有中文
{
hash = BuildHashByChar(hash, ' ');
}
_keywordsPinyin.Add(pys);
_keywordsFirstPinyin.Add(fpy);
hash = BuildHashByLength(hash, text.Length);
_hash.Add(hash);
}
_indexs = null;
}
public void TestOnData(NGramModelBase model)
{
PinyinDict pydict = model.PinyinDict;
var inputer = new NGramInputer(model);
var tester = new InputerTester(inputer);
using (var inputFile = File.OpenText(testInputPath))
//using (var outputFile = File.CreateText(testOutputPath))
tester.TestData(inputFile, Console.Out);
// Assert("Check the output.");
}
public static string GetPinyinFast(string text, bool tone = false)
{
StringBuilder sb = new StringBuilder();
for (int i = 0; i < text.Length; i++)
{
var c = text[i];
sb.Append(PinyinDict.GetPinyinFast(c, tone ? 1 : 0));
}
return(sb.ToString());
}
public string GetPinYinFast(string text)
{
StringBuilder sb = new StringBuilder();
for (int i = 0; i < text.Length; i++)
{
var c = text[i];
sb.Append(PinyinDict.GetPinYinFast(c));
}
return(sb.ToString());
}
/// <summary>
/// 设置关键字，注：索引会被清空
/// </summary>
/// <param name="keywords"></param>
public void SetKeywords(ICollection <string> keywords)
{
_keywords = keywords.ToArray();
_keywordsFirstPinyin = new string[_keywords.Length];
_keywordsPinyin = new string[_keywords.Length][];
for (int i = 0; i < _keywords.Length; i++)
{
var text = _keywords[i];
var pys = PinyinDict.GetPinyinList(text);
string fpy = "";
for (int j = 0; j < pys.Length; j++)
{
pys[j] = pys[j].ToUpper();
fpy += pys[j][0];
}
_keywordsPinyin[i] = pys;
_keywordsFirstPinyin[i] = fpy;
}
_indexs = null;
}
/// <summary>
/// 添加关键字
/// </summary>
/// <param name="keyword">关键字</param>
public void AddKeyword(string keyword)
{
_keywords.Add(keyword);
var pys = PinyinDict.GetPinyinList(keyword);
string fpy = "";
ulong hash = 0;
for (int j = 0; j < pys.Length; j++)
{
pys[j] = pys[j].ToUpper();
fpy += pys[j][0];
hash = BuildHashByChar(hash, pys[j][0]);
}
if (Regex.IsMatch(keyword, "[^a-zA-Z0-9]", RegexOptions.Compiled))//关键字中有中文
{
hash = BuildHashByChar(hash, ' ');
}
_keywordsPinyin.Add(pys);
_keywordsFirstPinyin.Add(fpy);
hash = BuildHashByLength(hash, keyword.Length);
_hash.Add(hash);
}
/// <summary>
/// 获取姓名拼音,中文字符集为[0x3400,0x9FD5],[0x20000-0x2B81D]，注：偏僻汉字很多未验证
/// </summary>
/// <param name="name">姓名</param>
/// <param name="tone">是否带声调</param>
/// <returns></returns>
public static string GetPinyinForName(string name, bool tone = false)
{
return(string.Join("", PinyinDict.GetPinyinForName(name, tone ? 1 : 0)));
}
/// <summary>
/// 获取姓名首字母拼音,中文字符集为[0x3400,0x9FD5]，注：偏僻汉字很多未验证
/// </summary>
/// <param name="name">姓名</param>
/// <param name="splitSpan">分隔符</param>
/// <returns></returns>
public static string GetFirstPinyinForName(string name, string splitSpan)
{
return(string.Join(splitSpan, PinyinDict.GetPinyinForName(name)));
}
/// <summary>
/// 获取姓名首字母拼音,中文字符集为[0x3400,0x9FD5]，注：偏僻汉字很多未验证
/// </summary>
/// <param name="name">姓名</param>
/// <returns></returns>
public static string GetFirstPinyinForName(string name)
{
return(string.Join("", PinyinDict.GetPinyinForName(name)));
}
/// <summary>
/// 获取拼音首字母,支持多音,中文字符集为[0x3400,0x9FD5]，注：偏僻汉字很多未验证
/// </summary>
/// <param name="text">原文本</param>
/// <returns></returns>
public static string[] GetFirstPinyinList(string text)
{
return(PinyinDict.GetPinyinList(text));
}
/// <summary>
/// 获取首字母，中文字符集为[0x3400,0x9FD5]，注：偏僻汉字很多未验证
/// </summary>
/// <param name="text">原文本</param>
/// <returns></returns>
public static string GetFirstPinyin(string text)
{
return PinyinDict.GetFirstPinyin(text, 0);
}
/// <summary>
/// 查询
/// </summary>
/// <param name="keywords"></param>
/// <returns></returns>
public List <T> Find(string keywords)
{
if (_keywordsFunc == null)
{
throw new Exception("请先使用SetKeywordsFunc方法。");
}
keywords = keywords.ToUpper().Trim();
if (string.IsNullOrEmpty(keywords))
{
return(null);
}
List <T> result = new List <T>();
var hasPinyin = Regex.IsMatch(keywords, "[a-zA-Z]");
if (hasPinyin == false)
{
foreach (var item in _list)
{
var keyword = _keywordsFunc(item);
if (keyword.Contains(keywords))
{
result.Add(item);
}
}
return(result);
}
var pykeys = SplitKeywords(keywords);
var minLength = int.MaxValue;
List <Tuple <string, string[]> > list = new List <Tuple <string, string[]> >();
foreach (var pykey in pykeys)
{
var keys = pykey.Split((char)0);
if (minLength > keys.Length)
{
minLength = keys.Length;
}
MergeKeywords(keys, 0, "", list);
}
PinyinSearch search = new PinyinSearch();
search.SetKeywords(list);
foreach (var item in _list)
{
var keyword = _keywordsFunc(item);
if (keyword.Length < minLength)
{
continue;
}
string fpy = "";
string[] pylist;
if (_pinyinFunc == null)
{
pylist = PinyinDict.GetPinyinList(keyword);
}
else
{
pylist = _pinyinFunc(item).Split(_splitChar);
}
for (int j = 0; j < pylist.Length; j++)
{
pylist[j] = pylist[j].ToUpper();
fpy += pylist[j][0];
}
if (search.Find(fpy, keyword, pylist))
{
result.Add(item);
}
}
return(result);
}
/// <summary>
/// 得到完整的拼音
/// </summary>
/// <param name="text">文本信息</param>
/// <returns></returns>
public string GetPinYin(string text)
{
return(PinyinDict.GetPinYin(text));
}
/// <summary>
/// 清理缓存
/// </summary>
public static void ClearCache()
{
PinyinDict.ClearCache();
}
/// <summary>
/// 获取姓名拼音,中文字符集为[0x3400,0x9FD5]，注：偏僻汉字很多未验证
/// </summary>
/// <param name="name">姓名</param>
/// <param name="tone">是否带声调</param>
/// <returns></returns>
public static List<string> GetPinyinListForName(string name, bool tone = false)
{
return PinyinDict.GetPinyinForName(name, tone ? 1 : 0);
}
/// <summary>
/// 获取姓名拼音,中文字符集为[0x3400,0x9FD5]，注：偏僻汉字很多未验证
/// </summary>
/// <param name="name">姓名</param>
/// <param name="splitSpan">分隔符</param>
/// <param name="tone">是否带声调</param>
/// <returns></returns>
public static string GetPinyinForName(string name, string splitSpan, bool tone = false)
{
return string.Join(splitSpan, PinyinDict.GetPinyinForName(name, tone ? 1 : 0));
}
/// <summary>
/// 获取拼音全拼,支持多音,中文字符集为[0x3400,0x9FD5]，注：偏僻汉字很多未验证
/// </summary>
/// <param name="text">原文本</param>
/// <param name="tone">是否带声调</param>
/// <returns></returns>
public static string[] GetPinyinList(string text, bool tone = false)
{
return PinyinDict.GetPinyinList(text, tone ? 1 : 0);
}
/// <summary>
/// 获取拼音全拼,支持多音,中文字符集为[0x3400,0x9FD5]，注：偏僻汉字很多未验证
/// </summary>
/// <param name="text">原文本</param>
/// <param name="splitSpan">分隔符</param>
/// <param name="tone">是否带声调</param>
/// <returns></returns>
public static string GetPinyin(string text, string splitSpan, bool tone = false)
{
return string.Join(splitSpan, PinyinDict.GetPinyinList(text, tone ? 1 : 0));
}
/// <summary>
/// 获取拼音全拼,支持多音,中文字符集为[0x4E00,0x9FD5],[0x20000-0x2B81D]，注：偏僻汉字很多未验证
/// </summary>
/// <param name="text">原文本</param>
/// <param name="tone">是否带声调</param>
/// <returns></returns>
public static string GetPinyin(string text, bool tone = false)
{
return(string.Join("", PinyinDict.GetPinyinList(text, tone ? 1 : 0)));
}
/// <summary>
/// 查询，空格为通配符
/// </summary>
/// <param name="keywords"></param>
/// <returns></returns>
public List <T> FindWithSpace(string keywords)
{
if (_keywordsFunc == null)
{
throw new Exception("请先使用SetKeywordsFunc方法。");
}
keywords = keywords.ToUpper().Trim();
if (string.IsNullOrEmpty(keywords))
{
return(null);
}
if (keywords.Contains(" ") == false)
{
return(Find(keywords));
}
List <Tuple <string, string[]> > list = new List <Tuple <string, string[]> >();
List <int> indexs = new List <int>();
var minLength = 0;
int keysCount;
{
var keys = keywords.Split(new char[] { ' ' }, StringSplitOptions.RemoveEmptyEntries);
keysCount = keys.Length;
for (int i = 0; i < keys.Length; i++)
{
var key = keys[i];
var pykeys = SplitKeywords(key);
var min = int.MaxValue;
foreach (var pykey in pykeys)
{
var keys2 = pykey.Split((char)0);
if (min > keys2.Length)
{
min = keys2.Length;
}
MergeKeywords(keys2, 0, "", list, i, indexs);
}
minLength += min;
}
}
PinyinSearch search = new PinyinSearch();
search.SetKeywords(list);
search.SetIndexs(indexs.ToArray());
List <T> result = new List <T>();
foreach (var item in _list)
{
var keyword = _keywordsFunc(item);
if (keyword.Length < minLength)
{
continue;
}
string fpy = "";
string[] pylist;
if (_pinyinFunc == null)
{
pylist = PinyinDict.GetPinyinList(keyword);
}
else
{
pylist = _pinyinFunc(item).Split(_splitChar);
}
for (int j = 0; j < pylist.Length; j++)
{
pylist[j] = pylist[j].ToUpper();
fpy += pylist[j][0];
}
if (search.Find2(fpy, keyword, pylist, keysCount))
{
result.Add(item);
}
}
return(result);
}
/// <summary>
/// 获取拼音首字母,支持多音,中文字符集为[0x3400,0x9FD5]，注：偏僻汉字很多未验证
/// </summary>
/// <param name="text">原文本</param>
/// <param name="splitSpan">分隔符</param>
/// <returns></returns>
public static string GetFirstPinyin(string text, string splitSpan)
{
return(string.Join(splitSpan, PinyinDict.GetPinyinList(text)));
}
/// <summary>
/// 获取所有拼音,中文字符集为[0x3400,0x9FD5]，注：偏僻汉字很多未验证
/// </summary>
/// <param name="c">原文本</param>
/// <param name="tone">是否带声调</param>
/// <returns></returns>
public static List <string> GetAllFirstPinyin(char c)
{
return(PinyinDict.GetAllPinyin(c));
}
/// <summary>
/// 获取所有拼音,中文字符集为[0x3400,0x9FD5]，注：偏僻汉字很多未验证
/// </summary>
/// <param name="c">原文本</param>
/// <param name="tone">是否带声调</param>
/// <returns></returns>
public static List<string> GetAllPinyin(char c, bool tone = false)
{
return PinyinDict.GetAllPinyin(c, tone ? 1 : 0);
}
/// <summary>
/// 获取文字的全部拼音（多读音）
/// </summary>
/// <param name="text">文本信息</param>
/// <returns></returns>
public List <string> GetAllPinYin(char text)
{
return(PinyinDict.GetAllPinYin(text));
}
/// <summary>
/// 获取首字母，中文字符集为[0x3400,0x9FD5]，注：偏僻汉字很多未验证
/// </summary>
/// <param name="text">原文本</param>
/// <returns></returns>
public static string GetFirstPinyin(string text)
{
return(string.Join("", PinyinDict.GetPinyinList(text)));
}
