Beispiel #1
0
        public unsafe Token[] GetToken(DoubleArray.ResultPair n)
        {
            Token[] dist     = new Token[this.GetTokenSize(n)];
            int     tokenPos = n.Value >> 8;

#if MMF_DIC
            this.tokens.ReadArray <Token>(tokenPos * sizeof(Token), dist, 0, dist.Length);
#else
            Array.Copy(this.tokens, tokenPos, dist, 0, dist.Length);
#endif
            return(dist);
        }
Beispiel #2
0
        public unsafe DoubleArray.ResultPair ExactMatchSearch(char *key, int len, int nodePos = 0)
        {
            //if (this.encoding == Encoding.Unicode)
            //    return this.da.ExactMatchSearch((byte*)key, len, nodePos);

            //エンコード
            int   maxByteCount = this.encoding.GetMaxByteCount(len);
            byte *bytes        = stackalloc byte[maxByteCount];
            int   bytesLen     = this.encoding.GetBytes(key, len, bytes, maxByteCount);

            DoubleArray.ResultPair result = this.da.ExactMatchSearch(bytes, bytesLen, nodePos);

            //文字数をデコードしたものに変換
            result.Length = this.encoding.GetCharCount(bytes, result.Length);

            return(result);
        }
Beispiel #3
0
 public int GetTokenSize(DoubleArray.ResultPair n)
 {
     return(0xFF & n.Value);
 }
Beispiel #4
0
        public void Open(MeCabParam param)
        {
            this.dic = new MeCabDictionary[param.UserDic.Length + 1];

            string prefix = param.DicDir;

            this.property.Open(prefix);

            this.unkDic.Open(Path.Combine(prefix, UnkDicFile));
            if (this.unkDic.Type != DictionaryType.Unk)
            {
                throw new MeCabInvalidFileException("not a unk dictionary", this.unkDic.FileName);
            }

            MeCabDictionary sysDic = new MeCabDictionary();

            sysDic.Open(Path.Combine(prefix, SysDicFile));
            if (sysDic.Type != DictionaryType.Sys)
            {
                throw new MeCabInvalidFileException("not a system dictionary", sysDic.FileName);
            }
            this.dic[0] = sysDic;

            for (int i = 0; i < param.UserDic.Length; i++)
            {
                MeCabDictionary d = new MeCabDictionary();
                d.Open(Path.Combine(prefix, param.UserDic[i]));
                if (d.Type != DictionaryType.Usr)
                {
                    throw new MeCabInvalidFileException("not a user dictionary", d.FileName);
                }
                if (!sysDic.IsCompatible(d))
                {
                    throw new MeCabInvalidFileException("incompatible dictionary", d.FileName);
                }
                this.dic[i + 1] = d;
            }

            this.unkTokens = new Token[this.property.Size][];
            for (int i = 0; i < this.unkTokens.Length; i++)
            {
                string key = this.property.Name(i);
                DoubleArray.ResultPair n = this.unkDic.ExactMatchSearch(key);
                if (n.Value == -1)
                {
                    throw new MeCabInvalidFileException("cannot find UNK category: " + key, this.unkDic.FileName);
                }
                this.unkTokens[i] = this.unkDic.GetToken(n);
            }

            this.space = this.property.GetCharInfo(' ');

            this.bosFeature = param.BosFeature;
            this.unkFeature = param.UnkFeature;

            this.maxGroupingSize = param.MaxGroupingSize;
            if (this.maxGroupingSize <= 0)
            {
                this.maxGroupingSize = DefaltMaxGroupingSize;
            }
        }