基于lucene.net 和ICTCLAS2014的站内搜索的实现1

Lucene.net是一个搜索引擎的框架，它自身并不能实现搜索。须要我们自己在当中实现索引的建立，索引的查找。全部这些都是依据它自身提供的API来实现。Lucene.net本身是基于java的，可是经过翻译成.ne版本号的，能够在ASP.net中使用这个来实现站内搜索。

要实现基于汉语的搜索引擎，首先的要实现汉语的分词。眼下网上大部分都是利用已经有的盘古分词来实现的分词系统。可是盘古分词效果不太好。在这里我把最新的ICTCLAS2014嵌入到Lucene.net中。Lucene.net中全部的分词系统都是基于Analyzer类来继承实现的。所以假设要使用ICTCLAS2014嵌入到Lucene.net中。就必要要继承Analyzer类实现自己的分词类。

1 ICTCLAS的引入

首先我们要把ICTCLAS的dll引入到C#文件里 ，由于这个dll不是在C#中建立的类库，所以无法直接将其增加到C#的引用中。我们考虑使用以下的方法来实现。为了方便，我们把引入的函数以及结构体放入一个类中。例如以下所看到的：

using System;
using System.Collections.Generic;
using System.Linq;
using System.Text;
using System.Runtime.InteropServices;
using Lucene.Net.Analysis;

namespace Lucene.Net.Analysis.DChinese
{
    [StructLayout(LayoutKind.Explicit)]
    public struct result_t
    {
        [FieldOffset(0)]
        public int start;
        [FieldOffset(4)]
        public int length;
        [FieldOffset(8)]
        public int sPos1;
        [FieldOffset(12)]
        public int sPos2;
        [FieldOffset(16)]
        public int sPos3;
        [FieldOffset(20)]
        public int sPos4;
        [FieldOffset(24)]
        public int sPos5;
        [FieldOffset(28)]
        public int sPos6;
        [FieldOffset(32)]
        public int sPos7;
        [FieldOffset(36)]
        public int sPos8;
        [FieldOffset(40)]
        public int sPos9;
        [FieldOffset(44)]
        public int sPos10;
        //[FieldOffset(12)] public int sPosLow;
        [FieldOffset(48)]
        public int POS_id;
        [FieldOffset(52)]
        public int word_ID;
        [FieldOffset(56)]
        public int word_type;
        [FieldOffset(60)]
        public double weight;
    }
    public class SplitWord
    {
        const string path = @"NLPIR.dll";//设定dll的路径

        //对函数进行申明
        [DllImport(path, CharSet = CharSet.Ansi, EntryPoint = "NLPIR_Init", CallingConvention = CallingConvention.Cdecl)]
        public static extern bool NLPIR_Init(String sInitDirPath, int encoding = 0, String sLicenceCode = null);

        //特别注意，C语言的函数NLPIR_API const char * NLPIR_ParagraphProcess(const char *sParagraph,int bPOStagged=1);必须相应以下的申明
        [DllImport(path, CharSet = CharSet.Ansi, CallingConvention = CallingConvention.Cdecl, EntryPoint = "NLPIR_ParagraphProcess")]
        public static extern IntPtr NLPIR_ParagraphProcess(String sParagraph, int bPOStagged = 1);

        [DllImport(path, CharSet = CharSet.Ansi, EntryPoint = "NLPIR_Exit", CallingConvention = CallingConvention.Cdecl)]
        public static extern bool NLPIR_Exit();

        [DllImport(path, CharSet = CharSet.Ansi, EntryPoint = "NLPIR_ImportUserDict", CallingConvention = CallingConvention.Cdecl)]
        public static extern int NLPIR_ImportUserDict(String sFilename);

        [DllImport(path, CharSet = CharSet.Ansi, EntryPoint = "NLPIR_FileProcess", CallingConvention = CallingConvention.Cdecl)]
        public static extern bool NLPIR_FileProcess(String sSrcFilename, String sDestFilename, int bPOStagged = 1);

        [DllImport(path, CharSet = CharSet.Ansi, EntryPoint = "NLPIR_FileProcessEx", CallingConvention = CallingConvention.Cdecl)]
        public static extern bool NLPIR_FileProcessEx(String sSrcFilename, String sDestFilename);

        [DllImport(path, CharSet = CharSet.Ansi, EntryPoint = "NLPIR_GetParagraphProcessAWordCount", CallingConvention = CallingConvention.Cdecl)]
        public static extern int NLPIR_GetParagraphProcessAWordCount(String sParagraph);

        //NLPIR_GetParagraphProcessAWordCount
        [DllImport(path, CharSet = CharSet.Ansi, EntryPoint = "NLPIR_ParagraphProcessAW", CallingConvention = CallingConvention.Cdecl)]
        public static extern void NLPIR_ParagraphProcessAW(int nCount, [Out, MarshalAs(UnmanagedType.LPArray)] result_t[] result);

        [DllImport(path, CharSet = CharSet.Ansi, EntryPoint = "NLPIR_AddUserWord", CallingConvention = CallingConvention.Cdecl)]
        public static extern int NLPIR_AddUserWord(String sWord);

        [DllImport(path, CharSet = CharSet.Ansi, EntryPoint = "NLPIR_SaveTheUsrDic", CallingConvention = CallingConvention.Cdecl)]
        public static extern int NLPIR_SaveTheUsrDic();

        [DllImport(path, CharSet = CharSet.Ansi, EntryPoint = "NLPIR_DelUsrWord", CallingConvention = CallingConvention.Cdecl)]
        public static extern int NLPIR_DelUsrWord(String sWord);

        [DllImport(path, CharSet = CharSet.Ansi, EntryPoint = "NLPIR_NWI_Start", CallingConvention = CallingConvention.Cdecl)]
        public static extern bool NLPIR_NWI_Start();

        [DllImport(path, CharSet = CharSet.Ansi, EntryPoint = "NLPIR_NWI_Complete", CallingConvention = CallingConvention.Cdecl)]
        public static extern bool NLPIR_NWI_Complete();

        [DllImport(path, CharSet = CharSet.Ansi, EntryPoint = "NLPIR_NWI_AddFile", CallingConvention = CallingConvention.Cdecl)]
        public static extern bool NLPIR_NWI_AddFile(String sText);

        [DllImport(path, CharSet = CharSet.Ansi, EntryPoint = "NLPIR_NWI_AddMem", CallingConvention = CallingConvention.Cdecl)]
        public static extern bool NLPIR_NWI_AddMem(String sText);

        [DllImport(path, CharSet = CharSet.Ansi, CallingConvention = CallingConvention.Cdecl, EntryPoint = "NLPIR_NWI_GetResult")]
        public static extern IntPtr NLPIR_NWI_GetResult(bool bWeightOut = false);

        [DllImport(path, CharSet = CharSet.Ansi, EntryPoint = "NLPIR_NWI_Result2UserDict", CallingConvention = CallingConvention.Cdecl)]
        public static extern uint NLPIR_NWI_Result2UserDict();

        [DllImport(path, CharSet = CharSet.Ansi, EntryPoint = "NLPIR_GetKeyWords", CallingConvention = CallingConvention.Cdecl)]
        public static extern IntPtr NLPIR_GetKeyWords(String sText, int nMaxKeyLimit = 50, bool bWeightOut = false);

        [DllImport(path, CharSet = CharSet.Ansi, CallingConvention = CallingConvention.Cdecl, EntryPoint = "NLPIR_GetFileKeyWords")]
        public static extern IntPtr NLPIR_GetFileKeyWords(String sFilename, int nMaxKeyLimit = 50, bool bWeightOut = false);
    }
}

这个类里面包括了全部的ICTCLAS的API函数，包括初始化，加入词语，加入词典。词典保存，分词等各种API。

而且都是STATIC函数。

2 分词类DChineseAnalyzer的建立

分词类的建立我们參考StandarAnalyzer分词的实现，再次基础上实现了DChineseAnalyzer类。在分词类中实现必要的构造函数，以及

public overrideTokenStream TokenStream(System.String fieldName, System.IO.TextReader
reader)

public overrideTokenStream ReusableTokenStream(System.String fieldName, System.IO.TextReader
reader)

这两个函数。

他们的作用是在函数中调用用分词器Tokenizer的派生类来实现分词。

在现有的版本号中通常是在使用分词类的时候。直接调用ReusableTokenStream函数，而不是调用TokenStream函数，这样能够做到一个分词类对象的建立可供多个分词文本的使用。

从而降低内存的浪费，提高效率。

以及一些字段。利用这些字段。我们能够增加一些停用词，用户自己的词典。

3 分词器DChineseTokenizer的建立

这个类是分词的核心关键所在。我们要在当中调用ICTCLAS中的分词。在这里面要注意的一个函数是publicoverride
bool IncrementToken()

它是我们获取下一个分词结果要用到的函数，假设想要遍历分词结果，就要建立一个循环。不断的调用IncrementToken函数。

整个分词系统代码例如以下所看到的：

using System;
using System.Collections.Generic;
using System.Linq;
using System.Text;
using System.IO;
using System.Runtime.InteropServices;
using Lucene.Net.Analysis;
using Lucene.Net.Analysis.Standard;
using Lucene.Net.Util;
using Lucene.Net.Documents;
using Lucene.Net.Analysis.Tokenattributes;
using Version = Lucene.Net.Util.Version;

namespace Lucene.Net.Analysis.DChinese
{
    public class DChineseAnalyzer : Analyzer
    {
        private ISet<string> stopSet;
        public static readonly ISet<string> STOP_WORDS_SET;
        private Version matchVersion;
        private bool replaceInvalidAcronym;
        private bool enableStopPositionIncrements;

        public DChineseAnalyzer(Version version, ISet<string> stopWords)
        {
            stopSet = stopWords;
            replaceInvalidAcronym = false;
            enableStopPositionIncrements = StopFilter.GetEnablePositionIncrementsVersionDefault(version);
            replaceInvalidAcronym = matchVersion.OnOrAfter(Version.LUCENE_24);
            this.matchVersion = version;
        }

        public DChineseAnalyzer(Version version)
            : this(version, STOP_WORDS_SET)
        {
        }

        public DChineseAnalyzer(Version version, System.IO.FileInfo stopWords)
            : this(version, WordlistLoader.GetWordSet(stopWords))
        {
        }

        static DChineseAnalyzer()
        {
            STOP_WORDS_SET = StopAnalyzer.ENGLISH_STOP_WORDS_SET;
        }
        public override TokenStream TokenStream(System.String fieldName, System.IO.TextReader reader)
        {
            TokenStream result = new DChineseTokenizer(matchVersion, reader);
            result = new LowerCaseFilter(result);
            result = new StopFilter(enableStopPositionIncrements, result, stopSet);
            result = new PorterStemFilter(result);
            return result;
        }

        private class SavedStreams
        {
            protected internal DChineseTokenizer source;
            protected internal TokenStream result;
        };
        public override TokenStream ReusableTokenStream(System.String fieldName, System.IO.TextReader reader)
        {
            SavedStreams streams = (SavedStreams)PreviousTokenStream;
            if (streams == null)
            {
                streams = new SavedStreams();
                streams.source = new DChineseTokenizer(matchVersion, reader);
                streams.result = new LowerCaseFilter(streams.source);
                streams.result = new StopFilter(enableStopPositionIncrements, streams.result, stopSet);
                streams.result = new PorterStemFilter(streams.result);
                PreviousTokenStream = streams;
            }
            else
            {
                streams.source.Reset(reader);
            }

            streams.source.SetReplaceInvalidAcronym(replaceInvalidAcronym);
            return streams.result;
        }

    }

    public sealed class DChineseTokenizer : Tokenizer
    {

        private bool m_replaceInvalidAcronym;
        private int offset = 0;
        private int bufferIndex = 0;
        private int dataLen = 0;
        private const int MAX_WORD_LEN = 255;
        private const int IO_BUFFER_SIZE = 4096;
        private readonly char[] ioBuffer = new char[IO_BUFFER_SIZE];

        private ITermAttribute termAtt;
        private IOffsetAttribute offsetAtt;
        private IPositionIncrementAttribute posIncrAtt;

        private void Init(System.IO.TextReader input, Version matchVersion)
        {
            if (matchVersion.OnOrAfter(Version.LUCENE_24))
            {
                m_replaceInvalidAcronym = true;
            }
            else
            {
                m_replaceInvalidAcronym = false;
            }
            //this.input = input;
            this.input = ChangeInput(input);
            termAtt = AddAttribute<ITermAttribute>();
            offsetAtt = AddAttribute<IOffsetAttribute>();
            posIncrAtt = AddAttribute<IPositionIncrementAttribute>();
        }

        public DChineseTokenizer(Version matchVersion, System.IO.TextReader input)
            : base()
        {
            Init(input, matchVersion);
        }

        public DChineseTokenizer(Version matchVersion, System.IO.TextReader input, AttributeSource source)
            : base(source)
        {
            Init(input, matchVersion);
        }

        public DChineseTokenizer(Version matchVersion, System.IO.TextReader input, AttributeFactory factory)
            : base(factory)
        {
            Init(input, matchVersion);
        }

        public override bool IncrementToken()
        {
            ClearAttributes();
            int length = 0;
            int start = bufferIndex;
            char[] buffer = termAtt.TermBuffer();
            while (true)
            {

                if (bufferIndex >= dataLen)
                {
                    offset += dataLen;
                    dataLen = input.Read(ioBuffer, 0, ioBuffer.Length);
                    if (dataLen <= 0)
                    {
                        dataLen = 0;
                        if (length > 0)
                            break;
                        return false;
                    }
                    bufferIndex = 0;
                }

                char c = ioBuffer[bufferIndex++];

                if (!System.Char.IsWhiteSpace(c))
                {
                    if (length == 0)
                    {
                        start = offset + bufferIndex - 1;
                    }
                    else if (length == buffer.Length)
                    {
                        buffer = termAtt.ResizeTermBuffer(1 + length);
                    }

                    buffer[length++] = c;
                    if (length == MAX_WORD_LEN)
                        break;
                }
                else if (length > 0)
                    break;
            }

            termAtt.SetTermLength(length);
            offsetAtt.SetOffset(CorrectOffset(start), CorrectOffset(start + length));
            posIncrAtt.PositionIncrement = 1;
            return true;
        }

        public override void Reset()
        {
            base.Reset(input);
            bufferIndex = 0;
            offset = 0;
            dataLen = 0;
        }

        public override void Reset(TextReader input)
        {
            String inputString = input.ReadToEnd();
            IntPtr intPtr = SplitWord.NLPIR_ParagraphProcess(inputString, 0);
            string strResult = Marshal.PtrToStringAnsi(intPtr);
            this.input = new StringReader(strResult);
            bufferIndex = 0;
            offset = 0;
            dataLen = 0;
        }

        public override void End()
        {
            int finalOffset = CorrectOffset(offset);
            offsetAtt.SetOffset(finalOffset, finalOffset);
        }

        public void SetReplaceInvalidAcronym(bool replaceInvalidAcronym)
        {
            this.m_replaceInvalidAcronym = replaceInvalidAcronym;
        }

        private TextReader ChangeInput(TextReader input)
        {
            //string indexPath = System.Environment.CurrentDirectory;
            //string indexPath = GetType().Assembly.Location;
            //string indexPath = System.IO.Path.GetDirectoryName(Page.Request.PhysicalPath);
            //string dirParent = Directory.GetParent(indexPath).Parent.FullName;
            string dirParent = System.AppDomain.CurrentDomain.BaseDirectory;

            bool bInit = SplitWord.NLPIR_Init(dirParent, 0, null);
            if (!bInit)
            {
                return null;
            }
            String inputString = input.ReadToEnd();
            IntPtr intPtr = SplitWord.NLPIR_ParagraphProcess(inputString, 0);
            string strResult = Marshal.PtrToStringAnsi(intPtr);
            return new StringReader(strResult);
        }
    }
}