FirteXCollection.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      : 2006/8/5
00013 //
00014 #ifndef _FIRTEXCOLLECTION_H
00015 #define _FIRTEXCOLLECTION_H
00016 
00017 #if _MSC_VER > 1000
00018 #pragma once
00019 #endif // _MSC_VER > 1000
00020 
00021 
00022 #include "Collection.h"
00023 
00024 namespace firtex
00025 {
00026         namespace collection
00027         {
00028                 class CFirteXCollection : public CCollection
00029                 {
00030                 public:
00031                         static const tstring identifier;
00032                 public:
00033                         CFirteXCollection(const tchar* location,CIndexWriter* pWriter);
00034                         virtual ~CFirteXCollection(void);
00035                 protected:
00036                         class CTagPair
00037                         {
00038                         public:
00039                                 CTagPair(schemaid_t sid,const tchar* tag)
00040                                         :schemaid(sid)
00041                                 {
00042                                         tagLen = (int32_t)_tcslen(tag);
00043                                         beginTag = new tchar[tagLen + 3];
00044                                         beginTag[0] = _T('<');
00045                                         _tcscpy(beginTag + 1,tag);
00046                                         beginTag[tagLen + 1] = _T('>');
00047                                         beginTag[tagLen + 2] = 0;
00048                                         endTag = new tchar[tagLen + 4];
00049                                         _tcscpy(endTag + 2,tag);
00050                                         endTag[0] = _T('<');
00051                                         endTag[1] = _T('/');
00052                                         endTag[tagLen + 2] = _T('>');
00053                                         endTag[tagLen + 3] = 0;
00054                                 }
00055                                 ~CTagPair()
00056                                 {
00057                                         delete[] beginTag;
00058                                         delete[] endTag;
00059                                 }
00060                         public:
00061                                 schemaid_t      schemaid;
00062                                 tchar*          beginTag;
00063                                 tchar*          endTag;
00064                                 int32_t         tagLen;
00065                         };
00066                 protected:
00070                         bool            scanInternal();
00071                 protected:
00072                         bool            readIndexFile(const tstring& indexfile);
00073                         bool            processTag(char*& buff,CTagPair* p,char*& value,size_t& valueLen,bool skip = false);
00074                 private:
00075                         tstring         m_sDirectory;
00076 
00077                         CTagPair**      m_tags;
00078                         int32_t         m_numTags;
00079                 };
00081                 inline bool CFirteXCollection::processTag(char*& buff,CTagPair* p,char*& value,size_t& valueLen,bool skip)
00082                 {                       
00083                         char* bTag = strstr(buff,p->beginTag);
00084                         if(bTag == NULL)
00085                         {
00086                                 buff += strlen(p->endTag);
00087                                 //FIRTEX_CLOG(level::warn) << "can't find begin tag: " << p->first<<endl;
00088                                 return false;
00089                         }
00090                         char* eTag = strstr(bTag,p->endTag);
00091                         if(eTag == NULL)
00092                         {
00093                                 FIRTEX_CLOG(level::warn) << "can't find end tag : " << p->endTag<<endl;
00094                                 return false;
00095                         }       
00096                         bTag += (p->tagLen + 2);
00097                         if(!skip)
00098                         {
00099                                 value = bTag;
00100                                 valueLen = eTag - bTag;
00101                                 buff = eTag + p->tagLen + 3;// strlen(p->endTag);
00102                         }
00103                         else 
00104                         {
00105                                 buff = bTag;
00106                         }
00107                         return true;
00108                 }               
00109         }
00110 }
00111 
00112 #endif