public static String GetHiragana(String dicPathFromExe, String str)
{
MeCabParam param = new MeCabParam();
param.DicDir = dicPathFromExe;
MeCabTagger tagger = MeCabTagger.Create(param);
MeCabNode node = tagger.ParseToNode(str);
String hiragana = "";
while (node != null)
{
if (node.CharType > 0)
{
String[] splitStrArray = node.Feature.Split(',');
String splitStr;
if (splitStrArray.Length < 9)
{
splitStr = node.Surface;
}
else
{
splitStr = splitStrArray[7];
}
hiragana = hiragana + splitStr;
}
node = node.Next;
}
return hiragana;
}
/// <summary>
/// 形態素解析を行い、可能性があるすべての形態素を周辺確率付きで取得します。
/// </summary>
/// <param name="sentence">解析対象の文字列へのポインタ</param>
/// <param name="length">解析対象の文字列の長さ</param>
/// <param name="theta">ソフト分かち書きの温度パラメータ</param>
/// <returns>すべての形態素ノードの配列</returns>
public unsafe TNode[] ParseSoftWakachi(char *sentence, int length,
float theta = MeCabParam.DefaltTheta)
{
var param = new MeCabParam()
{
LatticeLevel = MeCabLatticeLevel.Two,
Theta = theta
}
;
return(this.ParseToLattice(sentence, length, param).GetAllNodes());
}
/// <summary>
/// コンストラクタ
/// </summary>
/// <param name="nodeAllocator"></param>
/// <param name="param"></param>
/// <param name="length"></param>
internal MeCabLattice(Func <TNode> nodeAllocator, MeCabParam param, int length)
{
this.nodeAllocator = nodeAllocator;
this.Param = param;
this.BeginNodeList = new TNode[length + 1];
this.EndNodeList = new TNode[length + 1];
var bosNode = CreateNewNode();
bosNode.IsBest = true;
bosNode.Stat = MeCabNodeStat.Bos;
this.EndNodeList[0] = bosNode;
this.BosNode = bosNode;
var eosNode = CreateNewNode();
eosNode.IsBest = true;
eosNode.Stat = MeCabNodeStat.Eos;
this.BeginNodeList[length] = eosNode;
this.EosNode = eosNode;
}
/// <summary>
/// MeCabTaggerを開く
/// </summary>
/// <param name="param">初期化パラメーター</param>
private void Open(MeCabParam param)
{
this.viterbi.Open(param);
this.writer.Open(param);
}
/// <summary>
/// 形態素解析を行い、結果をラティスとして取得します
/// </summary>
/// <param name="sentence">解析対象の文字列へのポインタ</param>
/// <param name="length">解析対象の文字列の長さ</param>
/// <param name="param">解析パラメータ</param>
/// <returns>ラティス</returns>
public unsafe MeCabLattice <TNode> ParseToLattice(char *sentence, int length, MeCabParam param)
{
this.ThrowIfDisposed();
if (length < 0)
{
throw new ArgumentOutOfRangeException(nameof(length));
}
var lattice = new MeCabLattice <TNode>(this.CreateNewNode, param, length);
this.viterbi.Analyze(sentence, length, lattice);
return(lattice);
}
private void InitializeMeCabTagger()
{
var param = new MeCabParam();
param.DicDir = Path.Combine(
Path.GetDirectoryName(Assembly.GetExecutingAssembly().Location),
@"dic\ipadic"
);
_meCabTagger = MeCabTagger.Create(param);
}
static void Main(string[] args)
{
Console.WriteLine("<html><head><title>小倉 - 自然科学系和英大辞典</title></head><body>");
var nmparam = new NMeCab.MeCabParam();
System.IO.Directory.SetCurrentDirectory(System.IO.Path.GetDirectoryName(System.Reflection.Assembly.GetExecutingAssembly().Location));
nmparam.DicDir = "dic/ipadic";
var meCabTagger = NMeCab.MeCabTagger.Create(nmparam);
var genreList = new Dictionary <string, int>();
while (true)
{
var title = Console.In.ReadLine();
if (title == null)
{
break;
}
var content = Console.In.ReadLine();
if (Console.In.ReadLine() != "")
{
throw new Exception();
}
content = content.TrimStart();
var titleBase = System.Text.RegularExpressions.Regex.Replace(title, @"\(.+?\)", "");
titleBase = System.Text.RegularExpressions.Regex.Replace(title, @"\[.+?\]", "");
var titleList = titleBase.Split(',');
var genreMc = System.Text.RegularExpressions.Regex.Matches(title, @"\[.+?\]");
var genres = new List <String>();
foreach (System.Text.RegularExpressions.Match match in genreMc)
{
genres.Add(match.Value.TrimStart('[').TrimEnd(']'));
}
var contentBase = System.Text.RegularExpressions.Regex.Replace(content, @"\[.+?\]", "");
var contentList = contentBase.Split(';');
for (int i = 0; i < contentList.Count(); i++)
{
contentList[i] = contentList[i].TrimStart();
contentList[i] = System.Text.RegularExpressions.Regex.Replace(contentList[i], @"^an? ", "");
contentList[i] = System.Text.RegularExpressions.Regex.Replace(contentList[i], @"^the ", "");
}
Console.WriteLine("<dl><dt>" + title + "</dt>");
foreach (var text in titleList)
{
Console.WriteLine("<key type=\"表記\">" + text + "</key>");
try
{
var node = meCabTagger.ParseToNode(text).Next;
string furigana = "";
while (node != null && node.Stat != MeCabNodeStat.Eos)
{
furigana += node.Feature.Split(',')[7];
node = node.Next;
}
Console.WriteLine("<key type=\"かな\">" + furigana + "</key>");
}
catch { }
}
foreach (var text in genres)
{
Console.WriteLine("<key type=\"複合\" name=\"ジャンル\">" + text + "</key>");
if (genreList.Keys.Contains(text))
{
genreList[text]++;
}
else
{
genreList.Add(text, 1);
}
}
foreach (var text in contentList)
{
var ttext = text.TrimStart();
ttext = ttext.TrimEnd();
Console.WriteLine("<key type=\"表記\">" + ttext + "</key>");
var words = ttext.Split(' ', '[', ']', '(', ')');
foreach (var word in words)
{
var w = word.TrimStart().TrimEnd();
Console.WriteLine("<key type=\"クロス\">" + word + "</key>");
}
}
Console.WriteLine("<dd>" + content + "</dd></dl>");
}
Console.WriteLine("</body></html>");
if (args.Count() > 0)
{
using (var sw = new System.IO.StreamWriter(args[0], false, System.Text.Encoding.GetEncoding("shift_jis")))
{
sw.Write("<?xml version=\"1.0\" encoding=\"Shift_JIS\"?><complex><group name=\"ジャンル検索\"><category name=\"ジャンル\">");
foreach (var kvp in genreList)
{
if (kvp.Value > 2)
{
sw.WriteLine("<subcategory name=\"" + kvp.Key + "\">");
}
}
sw.WriteLine("</category><keyword name=\"キーワード1\" /><keyword name=\"キーワード2\" /></group></complex>");
}
}
}
/// <summary>
/// MeCabTaggerを開く
/// </summary>
/// <param name="param">初期化パラメーター</param>
private void Open(MeCabParam param)
{
this.viterbi.Open(param);
this.writer.Open(param);
}
/// <summary>
/// MeCabTaggerのインスタンスを生成する
/// </summary>
/// <param name="param">初期化パラメーター</param>
/// <returns>MeCabTaggerのインスタンス</returns>
public static MeCabTagger Create(MeCabParam param)
{
MeCabTagger tagger = new MeCabTagger();
tagger.Open(param);
return tagger;
}
/// <summary>
/// MeCabTaggerのインスタンスを生成する
/// </summary>
/// <returns>MeCabTaggerのインスタンス</returns>
public static MeCabTagger Create()
{
MeCabParam param = new MeCabParam();
param.LoadDicRC();
return MeCabTagger.Create(param);
}