public AnonyConverter()
{
var dllPath = System.Reflection.Assembly.GetExecutingAssembly().Location;
var dicPath = Path.Combine(Path.GetDirectoryName(dllPath), "IpaDic");
this.tagger = MeCabIpaDicTagger.Create(dicPath);
}
static void UseSoftWakachi()
{
Console.WriteLine("----------------------------------------------------------------------");
Console.WriteLine("Example of using Soft-Wakachi :");
Console.WriteLine();
using (var tagger = MeCabIpaDicTagger.Create())
{
var theta = 1f / 800f / 2f; // 温度パラメータ
var nodes = tagger.ParseSoftWakachi("本部長", theta); // ソフトわかち解を取得
foreach (var node in nodes.Where(n => n.Prob > 0.1f)) // 周辺確率＞0.1の形態素ノードだけを処理
{
Console.WriteLine($"表層形 ：{node.Surface}");
Console.WriteLine($"読み ：{node.Reading}");
Console.WriteLine($"品詞 ：{node.PartsOfSpeech}");
Console.WriteLine($"周辺確率：{node.Prob}");
Console.WriteLine();
}
// さらに、周辺確率の上位から表層形の異なるものの5件までを取得
var searchWords = nodes.OrderByDescending(n => n.Prob)
.Select(n => n.Surface)
.Distinct()
.Take(5);
Console.WriteLine($"上位ワード：{string.Join(",", searchWords)}");
}
}
public void ParseTest()
{
const string dicDir = "../../../../../dic/ipadic";
using (var tagger = MeCabIpaDicTagger.Create(dicDir))
{
var nodes = tagger.Parse("東京へ行け");
var node1 = nodes[0];
Assert.Equal("名詞", node1.PartsOfSpeech);
Assert.Equal("固有名詞", node1.PartsOfSpeechSection1);
Assert.Equal("地域", node1.PartsOfSpeechSection2);
Assert.Equal("一般", node1.PartsOfSpeechSection3);
Assert.Equal("*", node1.ConjugatedForm);
Assert.Equal("*", node1.Inflection);
Assert.Equal("東京", node1.OriginalForm);
Assert.Equal("トウキョウ", node1.Reading);
Assert.Equal("トーキョー", node1.Pronounciation);
var node2 = nodes[2];
Assert.Equal("動詞", node2.PartsOfSpeech);
Assert.Equal("自立", node2.PartsOfSpeechSection1);
Assert.Equal("*", node2.PartsOfSpeechSection2);
Assert.Equal("*", node2.PartsOfSpeechSection3);
Assert.Equal("五段・カ行促音便", node2.ConjugatedForm);
Assert.Equal("命令ｅ", node2.Inflection);
Assert.Equal("行く", node2.OriginalForm);
Assert.Equal("イケ", node2.Reading);
Assert.Equal("イケ", node2.Pronounciation);
}
}
public void OneTimeSetUp()
{
// NOTE : フォルダ移動の可能性があるので，"UniBagOfWords"のパスから見つける
var topDirectoryPath = Directory.GetDirectories("Assets", "*", SearchOption.AllDirectories)
.FirstOrDefault(path => Path.GetFileName(path) == "UniBagOfWords");
var dicDir = $"{topDirectoryPath}/Scripts/dic/ipadic";
_tagger = MeCabIpaDicTagger.Create(dicDir);
}
public NMeCabMorphAnalyzer(NMeCabSetting setting)
{
if (string.IsNullOrEmpty(setting.DicDir))
{
throw new ArgumentException("dicDirが正しく設定されていません");
}
DicDir = setting.DicDir;
_tagger = MeCabIpaDicTagger.Create(DicDir);
}
public void IpaDic()
{
using var tagger = MeCabIpaDicTagger.Create("../../../../../dic/ipadic");
var node = tagger.Parse("すもも")[0];
Assert.Equal("名詞", node.PartsOfSpeech);
Assert.Equal("一般", node.PartsOfSpeechSection1);
Assert.Equal("*", node.PartsOfSpeechSection2);
Assert.Equal("*", node.PartsOfSpeechSection3);
Assert.Equal("*", node.ConjugatedForm);
Assert.Equal("*", node.Inflection);
Assert.Equal("すもも", node.OriginalForm);
Assert.Equal("スモモ", node.Reading);
Assert.Equal("スモモ", node.Pronounciation);
}
static void UseNotBeAwareOfDictionaly()
{
Console.WriteLine("----------------------------------------------------------------------");
Console.WriteLine("Example of using not be aware of dictionary :");
Console.WriteLine();
using (var tagger = MeCabIpaDicTagger.Create()) // Taggerインスタンスを生成
{
var nodes = tagger.Parse("皇帝の新しい心"); // 形態素解析を実行
foreach (var node in nodes) // 形態素ノード配列を順に処理
{
Console.WriteLine($"表層形：{node.Surface}");
Console.WriteLine($"読み ：{node.Reading}");
Console.WriteLine($"品詞 ：{node.PartsOfSpeech}");
Console.WriteLine();
}
}
}
static void UseWithUserPreparedDictionaly2()
{
Console.WriteLine("----------------------------------------------------------------------");
Console.WriteLine("Example of using user prepared dictionaly 2 (for IPAdic) :");
Console.WriteLine();
var dicDir = "../../../../../dic/ipadic"; // 辞書のパス
using (var tagger = MeCabIpaDicTagger.Create(dicDir)) // IPAdic形式用のTaggerインスタンスを生成
{
var nodes = tagger.Parse("皇帝の新しい心"); // 形態素解析を実行
foreach (var node in nodes) // 形態素ノード配列を順に処理
{
Console.WriteLine($"表層形：{node.Surface}");
Console.WriteLine($"読み ：{node.Reading}"); // 個別の素性
Console.WriteLine($"品詞 ：{node.PartsOfSpeech}"); // 〃
Console.WriteLine();
}
}
}
static void UseNBest()
{
Console.WriteLine("----------------------------------------------------------------------");
Console.WriteLine("Example of using N-Best :");
Console.WriteLine();
using (var tagger = MeCabIpaDicTagger.Create())
{
var results = tagger.ParseNBest("東京大学"); // Nベスト解を取得
foreach (var nodes in results.Take(5)) // 上位から5件までの解を処理
{
foreach (var node in nodes) // 形態素ノード配列を順に処理
{
Console.WriteLine($"表層形：{node.Surface}");
Console.WriteLine($"読み ：{node.Reading}");
Console.WriteLine($"品詞 ：{node.PartsOfSpeech}");
Console.WriteLine();
}
Console.WriteLine("----------------");
}
}
}
public MeCabIpaDicWrapper(string dicDir, string[]?userDirDics = null)
{
IpaDicTagger = MeCabIpaDicTagger.Create(dicDir, userDirDics);
}
static void Main()
{
var dicDir = "../../../../../dic/ipadic";
var targetFile = "kokoro.txt";
var encoding = Encoding.UTF8;
var sw = new Stopwatch();
//開始指示を待機
Console.WriteLine("Press Enter key to start.");
Console.ReadLine();
Console.WriteLine("\t\t\tProcessTime\tTotalMemory");
//解析準備処理
GC.Collect();
sw.Start();
var tagger = MeCabIpaDicTagger.Create(dicDir);
sw.Stop();
Console.WriteLine("OpenTagger:\t\t{0:0.000}sec\t{1:#,000}byte",
sw.Elapsed.TotalSeconds, GC.GetTotalMemory(false));
//ファイル読込だけの場合
using (var reader = new StreamReader(targetFile, encoding))
{
sw.Reset();
GC.Collect();
sw.Start();
for (string line = reader.ReadLine(); line != null; line = reader.ReadLine())
{
}
sw.Stop();
}
Console.WriteLine("ReadLine:\t\t{0:0.000}sec\t{1:#,000}byte",
sw.Elapsed.TotalSeconds, GC.GetTotalMemory(false));
//解析処理（Nodeの出力）
using (var reader = new StreamReader(targetFile, encoding))
{
sw.Reset();
GC.Collect();
sw.Start();
for (string line = reader.ReadLine(); line != null; line = reader.ReadLine())
{
var node = tagger.Parse(line);
}
sw.Stop();
}
Console.WriteLine("ParseToNode:\t\t{0:0.000}sec\t{1:#,000}byte",
sw.Elapsed.TotalSeconds, GC.GetTotalMemory(false));
//解析処理（素性文字列分解）
using (var reader = new StreamReader(targetFile, encoding))
{
sw.Reset();
GC.Collect();
sw.Start();
for (string line = reader.ReadLine(); line != null; line = reader.ReadLine())
{
var node = tagger.Parse(line);
foreach (var item in node)
{
var a = item.Surface;
var b = item.PartsOfSpeech;
var c = item.Reading;
}
}
sw.Stop();
}
Console.WriteLine("ParseToText:\t\t{0:0.000}sec\t{1:#,000}byte",
sw.Elapsed.TotalSeconds, GC.GetTotalMemory(false));
//解析処理（Best解5件のNodeの出力）
using (var reader = new StreamReader(targetFile, encoding))
{
sw.Reset();
GC.Collect();
sw.Start();
for (string line = reader.ReadLine(); line != null; line = reader.ReadLine())
{
int i = 0;
foreach (var node in tagger.ParseNBest(line))
{
if (++i == 5)
{
break;
}
}
}
sw.Stop();
}
Console.WriteLine("ParseNBestToNode:\t{0:0.000}sec\t{1:#,000}byte",
sw.Elapsed.TotalSeconds, GC.GetTotalMemory(false));
//対象の情報
using (var reader = new StreamReader(targetFile, encoding))
{
long charCount = 0;
long lineCount = 0;
long wordCount = 0;
for (string line = reader.ReadLine(); line != null; line = reader.ReadLine())
{
charCount += line.Length;
lineCount++;
var node = tagger.Parse(line);
wordCount += node.Length;
}
Console.WriteLine();
Console.WriteLine("Target: {0} {1:#,000}byte {2:#,000}char {3:#,000}line ({4:#,000}word)",
targetFile, reader.BaseStream.Position, charCount, lineCount, wordCount);
}
tagger.Dispose();
//終了したことを通知
Console.WriteLine();
Console.WriteLine("Finish!");
Console.WriteLine("Press Enter key to close.");
Console.ReadLine();
}
public KawazuConverter()
{
_tagger = MeCabIpaDicTagger.Create();
}
public KawazuConverter(string dicPath = null)
{
_tagger = MeCabIpaDicTagger.Create(dicPath);
}
public KawazuConverter(string dicdir)
{
_tagger = MeCabIpaDicTagger.Create(dicdir);
}
static void UseLattice()
{
Console.WriteLine("----------------------------------------------------------------------");
Console.WriteLine("Example of using Lattice :");
Console.WriteLine();
using (var tagger = MeCabIpaDicTagger.Create())
{
var prm = new MeCabParam()
{
LatticeLevel = MeCabLatticeLevel.Two,
Theta = 1f / 800f / 2f
};
var lattice = tagger.ParseToLattice("東京大学", prm); // ラティスを取得
// ラティスから、ベスト解を取得し処理
foreach (var node in lattice.GetBestNodes())
{
Console.Write(node.Surface);
Console.CursorLeft = 10;
Console.Write(node.Feature);
Console.WriteLine();
}
Console.WriteLine("--------");
// ラティスから、2番目と3番目のベスト解を取得し処理
foreach (var result in lattice.GetNBestResults().Skip(1).Take(2))
{
foreach (var node in result)
{
Console.Write(node.Surface);
Console.CursorLeft = 10;
Console.Write(node.Feature);
Console.WriteLine();
}
Console.WriteLine("----");
}
Console.WriteLine("--------");
// ラティスから、開始位置別の形態素を取得し処理
for (int i = 0; i < lattice.BeginNodeList.Length - 1; i++)
{
for (var node = lattice.BeginNodeList[i]; node != null; node = node.BNext)
{
if (node.Prob <= 0.001f)
{
continue;
}
Console.CursorLeft = i * 2;
Console.Write(node.Surface);
Console.CursorLeft = 10;
Console.Write(node.Prob.ToString("F3"));
Console.CursorLeft = 16;
Console.Write(node.Feature);
Console.WriteLine();
}
}
Console.WriteLine("--------");
// ラティスから、最終的な累積コストのみを取得し表示
Console.WriteLine(lattice.EosNode.Cost);
}
}
