public ChineseChar(char ch, ChineseTypes types, string[] pinyins)
{
Char = ch;
Pinyins = pinyins;
IsPolyphone = Pinyins.Length > 1;
Types = types;
}
private static string Match(ChineseLexicon lexicon, ChineseTypes chineseType, string part)
{
var isMatch = chineseType == ChineseTypes.Traditional
? lexicon.Words.Any(x => x.Traditional == part)
: lexicon.Words.Any(x => x.Simplified == part);
return(isMatch ? part : null);
}
/// <summary>
/// 获取分词结果。
/// </summary>
/// <param name="chineseType"></param>
/// <param name="chinese"></param>
/// <returns></returns>
public static string[] SplitWords(ChineseTypes chineseType, string chinese)
{
var scope = LexiconScope.Current ?? LexiconScope.Default;
var lexicon = scope.Lexicon;
if (lexicon is null)
{
return(chinese.Select(ch => ch.ToString()).ToArray());
}
var list = new LinkedList <string>();
var length = chinese.Length;
var maxOffset = Math.Min(chinese.Length, lexicon.WordMaxLength);
var ptext = length - maxOffset;
var maxLengthPerTurn = maxOffset;
int matchLength;
for (; ptext + maxLengthPerTurn >= 0; ptext -= matchLength)
{
matchLength = 1;
for (var i = ptext > 0 ? 0 : -ptext; i < maxLengthPerTurn; i++)
{
var part = chinese.Substring(ptext + i, maxLengthPerTurn - i);
if (part.Length == 1)
{
list.AddFirst(part);
break;
}
else
{
var match = Match(lexicon, chineseType, part);
if (match != null)
{
list.AddFirst(match);
matchLength = match.Length;
break;
}
}
}
}
return(list.ToArray());
}
/// <summary>
/// 获取中文拼音
/// </summary>
/// <param name="Chinese"></param>
/// <param name="format"></param>
/// <param name="chineseType"></param>
/// <returns></returns>
public static string CHNPinYin(string Chinese, PinyinFormat format, ChineseTypes chineseType = ChineseTypes.Simplified)
{
return(Pinyin.GetString(chineseType, Chinese, format));
}
/// <summary>
/// 获取指定类型字符串的拼音
/// </summary>
/// <param name="chineseTypes"></param>
/// <param name="chinese"></param>
/// <param name="format"></param>
/// <returns></returns>
public static string GetString(ChineseTypes chineseTypes, string chinese, PinyinFormat format = PinyinFormat.Default)
{
var scope = LexiconScope.Current ?? LexiconScope.Default;
var lexicon = scope.Lexicon;
IEnumerable <int> GetDefaultSteps()
{
foreach (var ch in chinese)
{
yield return(1);
}
}
var steps = lexicon is null?GetDefaultSteps() : ChineseTokenizer.SplitWords(chineseTypes, chinese).Select(x => x.Length);
if (!chinese.IsNullOrWhiteSpace())
{
var sb = new StringBuilder();
var insertSpace = false;
var ptext = 0;
foreach (var step in steps)
{
var word = chinese.Substring(ptext, step);
try
{
string pinyin = null;
if (pinyin is null && chineseTypes.HasFlag(ChineseTypes.Simplified))
{
pinyin = lexicon.Words.FirstOrDefault(x => x.Simplified == word)?.SimplifiedPinyin;
}
if (pinyin is null && chineseTypes.HasFlag(ChineseTypes.Traditional))
{
pinyin = lexicon.Words.FirstOrDefault(x => x.Traditional == word)?.TraditionalPinyin;
}
if (pinyin is null && chineseTypes.HasFlag(ChineseTypes.Simplified))
{
pinyin = Builtin.ChineseChars.FirstOrDefault(x => x.Char == word[0])?.SimplifiedPinyin;
}
if (pinyin is null && chineseTypes.HasFlag(ChineseTypes.Traditional))
{
pinyin = Builtin.ChineseChars.FirstOrDefault(x => x.Char == word[0])?.TraditionalPinyin;
}
if (pinyin is null)
{
throw new ArgumentException($"未能匹配文字（{word}）。");
}
if (format != PinyinFormat.InitialConsonant)
{
if (insertSpace)
{
sb.Append(" ");
}
}
switch (format)
{
case PinyinFormat.Default: sb.Append(pinyin); break;
case PinyinFormat.WithoutTone: sb.Append(GetPinyinWithoutTone(pinyin)); break;
case PinyinFormat.Phonetic: sb.Append(GetPhoneticSymbol(pinyin)); break;
case PinyinFormat.InitialConsonant: sb.Append(pinyin.First()); break;
}
insertSpace = true;
}
catch
{
sb.Append(word);
insertSpace = false;
}
ptext += step;
}
return(sb.ToString());
}
return(chinese);
}