Argument.h

00001 //
00002 // Copyright(C) 2005--2006 Institute of Computing Tech, Chinese Academy of Sciences. 
00003 // All rights reserved.
00004 // This file is part of FirteX (www.firtex.org)
00005 //
00006 // Use of the FirteX is subject to the terms of the software license set forth in 
00007 // the LICENSE file included with this software, and also available at
00008 // http://www.firtex.org/license.html
00009 //
00010 // Author       : 郭瑞杰(GuoRuijie)
00011 // Email        : ruijieguo@software.ict.ac.cn,ruijieguo@gmail.com
00012 // Created      : 2006/8/3
00013 //
00014 #ifndef _ARGUMENT_H
00015 #define _ARGUMENT_H
00016 
00017 #if _MSC_VER > 1000
00018 #pragma once
00019 #endif // _MSC_VER > 1000
00020 
00021 #include "StdHeader.h"
00022 #include "../document/Field.h"
00023 #include <vector>
00024 using namespace std;
00025 using namespace firtex::document;
00026 
00027 namespace firtex
00028 {
00029         namespace utility
00030         {
00031                 class CXMLElement;
00032                 class CArgument
00033                 {
00034                 public:
00035                         CArgument(void);
00036                         ~CArgument(void);
00037                 protected:              
00038                         typedef vector<pair<tstring,TermVector_> > termvector_array;
00039                 public:
00074                         void    parseArg(const tchar* argFile);
00075 
00081                         void    saveArg(const tchar* argFile = NULL);
00082 
00087                         const tstring&  getFileName(){return m_filename;}
00088                 public:
00089                         struct scollection
00090                         {
00091                                 tstring identifier;                     
00092                                 struct smodifieddate 
00093                                 {
00094                                         tstring field;
00095                                         bool    store;
00096                                         bool    index;
00097                                 }modifieddate;
00098                         }collection;
00099 
00100                         struct sparser
00101                         {
00102                                 tstring                         identifier;                     
00103                                 termvector_array        termvectors;            
00104                         }parser;
00105 
00106                         struct sanalyzer
00107                         {
00108                                 tstring identifier;                     
00109                                 tstring stopwordFile;           
00110                         }analyzer;
00111 
00112                         struct sindex
00113                         {
00114                                 tstring srcfile;                        
00115                                 tstring dstindex;                       
00116                                 bool    append;                         
00117                                 tstring mergemode;                      
00118                                 int32_t maxIndexTerms;          
00119                                 int64_t memory;                         
00120                         }index;
00121 
00122                         struct sindexmerger 
00123                         {
00124                                 int32_t mergeFactor;            
00125                                 int32_t maxIndexBarrels;        
00126                         }indexmerger;
00127 
00128                         struct slog
00129                         {
00130                                 tstring level;                          
00131 
00132                         }log;
00133                         
00134                 public:
00138                         void    replaceStdArg();
00139 
00143                         void    restoreStdArg();
00144 
00148                         void    resetArg();
00149                 protected:
00150                         //保存标准参数以供恢复
00151                         struct sstd_analyzer
00152                         {
00153                                 tstring stopwordFile;
00154                         }std_analyzer;
00155 
00156                         struct sstd_index
00157                         {
00158                                 int32_t maxIndexTerms;          
00159                                 int64_t memory;                         
00160                         }std_index;
00161 
00162                         struct sstd_indexmerger 
00163                         {
00164                                 int32_t mergeFactor;            
00165                                 int32_t maxIndexBarrels;        
00166                         }std_indexmerger ;
00167                 protected:
00168                         void    add(const tstring& sName, int64_t* pInt64, int64_t nDefault);
00169                         void    add(const tstring& sName, int32_t* pInt32, int32_t nDefault);
00170                         void    add(const tstring& sName, double* pFloat, double nDefault);
00171                         void    add(const tstring& sName, tstring* pString, const tstring& pszDefault);
00172                         void    add(const tstring& sName, bool* pBool, bool bDefault = true);
00173                         void    add(const tstring& sName, termvector_array* pTermVector);
00174 
00175                 
00176                 protected:
00177                         class CArgumentItem
00178                         {
00179                         public:
00180                                 CArgumentItem(const tstring& sName);
00181                                 CArgumentItem(const tstring& sName, int64_t* pInt64,int32_t* pInt32,double* pFloat, tstring* pString,bool* pBool);
00182                                 virtual void    load(CXMLElement* pXML);
00183                                 virtual void    save(CXMLElement* pXML);
00184                         public:
00185                                 tstring         m_sName;
00186                                 int32_t*        m_pInt32;
00187                                 int64_t*        m_pInt64;
00188                                 double*         m_pFloat;
00189                                 tstring*        m_pString;
00190                                 bool*           m_pBool;
00191                         };
00192                         class CTermVectorFieldsItem : public CArgumentItem
00193                         {
00194                         public:
00195                                 CTermVectorFieldsItem(const tstring& sName,termvector_array* tvs);
00196                                 virtual void    load(CXMLElement* pXML);
00197                                 virtual void    save(CXMLElement* pXML);
00198                         public:
00199                 termvector_array*       m_pTermVectors;
00200                         };
00201                 protected:
00202                         vector<CArgumentItem*>  m_items;
00203                         tstring                         m_filename;
00204                 };
00205         }
00206 }
00207 
00208 #endif