FirteX－高性能全文索引和检索平台
API Documentation
|
FirteX－高性能全文索引和检索平台API Documentation |
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 : 2006/5/20 00013 // 00014 #ifndef _FXSTRING_H 00015 #define _FXSTRING_H 00016 00017 #include "typedefs.h" 00018 00019 #if _MSC_VER > 1000 00020 #pragma once 00021 #endif // _MSC_VER > 1000 00022 00023 #include <string> 00024 00025 using namespace firtex; 00026 00027 namespace firtex 00028 { 00029 namespace utility 00030 { 00031 template <typename CharT> 00032 class CFXString_base 00033 { 00034 public: 00035 typedef CharT char_type; 00036 public: 00037 CFXString_base(void); 00038 CFXString_base(const char_type* str); //will copy the string 00039 CFXString_base(char_type* str,bool bNoCopy); //will or will NOT copy the string,depends on bNoCopy 00040 CFXString_base(const char_type* str,size_t len); //will copy the string 00041 CFXString_base(char_type* str,size_t len,bool bNoCopy); //will or will NOT copy the string,depends on bNoCopy 00042 CFXString_base(const _str<char_type>& str); //will copy the string 00043 CFXString_base(const _str<char_type>& str,bool bNoCopy); //will or will NOT copy the string,depends on bNoCopy 00044 CFXString_base(const basic_string<char_type>& str); //will copy the string 00045 CFXString_base(const CFXString_base& src); 00046 ~CFXString_base(void); 00047 public: 00048 CFXString_base& operator=(const _str<char_type>& str); //will copy the string 00049 CFXString_base& operator=(const char_type* str); //will copy the string 00050 CFXString_base& operator=(const CFXString_base<char_type>& str); //will copy the string 00051 CFXString_base& operator=(const basic_string<char_type>& str); //will copy the string 00052 00053 operator char_type*(); 00054 operator _str<char_type>(); 00055 operator basic_string<char_type>(); 00056 00057 bool operator==(const char_type* str)const; 00058 bool operator==(const basic_string<char_type>& str)const; 00059 bool operator==(const CFXString_base<char_type>& right)const; 00060 00061 bool operator!=(const char_type* str)const; 00062 bool operator!=(const basic_string<char_type>& str)const; 00063 bool operator!=(const CFXString_base<char_type>& right)const; 00064 public: 00068 const char_type* c_str()const; 00069 00073 char_type* data(); 00074 00078 size_t length()const; 00079 00083 size_t maxLength()const{return m_maxLen;} 00084 00089 void reserve(size_t len); 00090 00095 void resize(size_t len); 00096 00103 void assign(char_type* data,size_t len,bool bNoCopy); 00104 00105 void assign (const char_type* str); 00106 void assign (const char_type* str,size_t len); 00107 void assign (const _str<char_type>& str); 00108 public: 00109 void append(char_type* str,size_t len); 00110 public: 00111 int compare(const CFXString_base<char_type>& right)const; 00112 int compare(const char_type* str)const; 00113 int compare(const basic_string<char_type>& str)const; 00114 int compare(const _str<char_type>& str)const; 00115 00116 void clear(); 00117 00118 protected: 00119 inline void init(); 00120 private: 00121 _str<char_type> m_str; 00122 size_t m_maxLen; 00123 bool m_bOwner; 00124 }; 00125 00127 //Inline Functions 00128 template <class CharT> 00129 inline CFXString_base<CharT>::CFXString_base(void) 00130 { 00131 init(); 00132 } 00133 template <class CharT> 00134 inline CFXString_base<CharT>::CFXString_base(const CharT* str) 00135 { 00136 init(); 00137 assign(str); 00138 } 00139 template <class CharT> 00140 inline CFXString_base<CharT>::CFXString_base(CharT* str,bool bNoCopy) 00141 { 00142 if(bNoCopy) 00143 { 00144 m_str.str = str; 00145 m_maxLen = m_str.length = (sizeof(CharT)==sizeof(char))?strlen(str):wcslen(str); 00146 m_bOwner = false; 00147 } 00148 else 00149 { 00150 init(); 00151 assign(str); 00152 } 00153 00154 } 00155 template <class CharT> 00156 inline CFXString_base<CharT>::CFXString_base(const CharT* str,size_t len) 00157 { 00158 init(); 00159 assign(str,len); 00160 } 00161 template <class CharT> 00162 inline CFXString_base<CharT>::CFXString_base(CharT* str,size_t len,bool bNoCopy) 00163 { 00164 init(); 00165 if(bNoCopy) 00166 { 00167 m_str.str = str; 00168 m_maxLen = m_str.length = len; 00169 m_bOwner = false; 00170 } 00171 else assign(str,len); 00172 } 00173 template <class CharT> 00174 inline CFXString_base<CharT>::CFXString_base(const basic_string<CharT>& str) 00175 { 00176 init(); 00177 assign(str.c_str(),str.length()); 00178 } 00179 template <class CharT> 00180 inline CFXString_base<CharT>::CFXString_base(const CFXString_base& src) 00181 { 00182 init(); 00183 if(src.m_bOwner) 00184 assign(src.m_str.str,src.m_str.length); 00185 else 00186 { 00187 m_str = src.m_str; 00188 } 00189 m_maxLen = src.m_maxLen; 00190 } 00191 template <class CharT> 00192 inline CFXString_base<CharT>::~CFXString_base(void) 00193 { 00194 clear(); 00195 } 00196 template <class CharT> 00197 inline void CFXString_base<CharT>::init() 00198 { 00199 m_str.str = NULL; 00200 m_str.length = 0; 00201 m_bOwner = false; 00202 m_maxLen = 0; 00203 } 00204 00205 template<class CharT> 00206 inline void CFXString_base<CharT>::clear() 00207 { 00208 if(m_bOwner) 00209 { 00210 if(m_str.str) 00211 delete[] m_str.str; 00212 } 00213 m_str.str = NULL; 00214 m_str.length = 0; 00215 m_maxLen = 0; 00216 m_bOwner = false; 00217 } 00218 00219 template <> 00220 inline void CFXString_base<char>::assign (const char* str) 00221 { 00222 m_str.length = strlen(str); 00223 if(!m_bOwner) 00224 { 00225 m_maxLen = m_str.length; 00226 m_str.str = new char[m_str.length + 1]; 00227 } 00228 else 00229 { 00230 if((m_str.length > m_maxLen)) 00231 { 00232 if(m_str.str) 00233 delete[] m_str.str; 00234 m_str.str = new char[m_str.length + 1]; 00235 m_maxLen = m_str.length; 00236 } 00237 } 00238 strcpy(m_str.str,str); 00239 m_bOwner = true; 00240 } 00241 template <> 00242 inline void CFXString_base<wchar_t>::assign (const wchar_t* str) 00243 { 00244 m_str.length = wcslen(str); 00245 if(!m_bOwner) 00246 { 00247 m_maxLen = m_str.length; 00248 m_str.str = new wchar_t[m_str.length + 1]; 00249 } 00250 else 00251 { 00252 if((m_str.length > m_maxLen)) 00253 { 00254 if(m_str.str) 00255 delete[] m_str.str; 00256 m_str.str = new wchar_t[m_str.length + 1]; 00257 m_maxLen = m_str.length; 00258 } 00259 } 00260 wcscpy(m_str.str,str); 00261 m_bOwner = true; 00262 } 00263 00264 template <> 00265 inline void CFXString_base<char>::assign (const char* str,size_t len) 00266 { 00267 if(len == 0) 00268 len = strlen(str); 00269 m_str.length = len; 00270 if(!m_bOwner) 00271 { 00272 m_maxLen = len; 00273 m_str.str = new char[len + 1]; 00274 } 00275 else 00276 { 00277 if((len > m_maxLen)) 00278 { 00279 if(m_str.str) 00280 delete[] m_str.str; 00281 m_str.str = new char[len + 1]; 00282 m_maxLen = len; 00283 } 00284 } 00285 strncpy(m_str.str,str,len); 00286 m_str.str[len] = 0; 00287 m_bOwner = true; 00288 00289 } 00290 template <> 00291 inline void CFXString_base<wchar_t>::assign (const wchar_t* str,size_t len) 00292 { 00293 if(len == 0) 00294 len = wcslen(str); 00295 m_str.length = len; 00296 if(!m_bOwner) 00297 { 00298 m_maxLen = len; 00299 m_str.str = new wchar_t[len + 1]; 00300 } 00301 else 00302 { 00303 if((len > m_maxLen)) 00304 { 00305 if(m_str.str) 00306 delete[] m_str.str; 00307 m_str.str = new wchar_t[len + 1]; 00308 m_maxLen = len; 00309 } 00310 } 00311 wcsncpy(m_str.str,str,len); 00312 m_str.str[len] = 0; 00313 m_bOwner = true; 00314 } 00315 00316 template <class CharT> 00317 inline CFXString_base<CharT>& CFXString_base<CharT>::operator=(const _str<CharT>& str) 00318 { 00319 assign(str.str,str.length); 00320 return (*this); 00321 } 00322 template <class CharT> 00323 inline CFXString_base<CharT>& CFXString_base<CharT>::operator=(const CharT* str) 00324 { 00325 assign(str); 00326 return (*this); 00327 } 00328 template <class CharT> 00329 inline CFXString_base<CharT>& CFXString_base<CharT>::operator=(const CFXString_base<CharT>& str) 00330 { 00331 assign(str.m_str.str,str.m_str.length); 00332 m_maxLen = str.m_maxLen; 00333 00334 return (*this); 00335 } 00336 template <class CharT> 00337 inline CFXString_base<CharT>& CFXString_base<CharT>::operator=(const basic_string<CharT>& str) 00338 { 00339 assign(str.c_str(),str.length()); 00340 return (*this); 00341 } 00342 template <class CharT> 00343 inline CFXString_base<CharT>::operator CharT*() 00344 { 00345 return m_str.str; 00346 } 00347 template <class CharT> 00348 inline CFXString_base<CharT>::operator _str<CharT>() 00349 { 00350 return m_str; 00351 } 00352 template <class CharT> 00353 inline CFXString_base<CharT>::operator basic_string<CharT>() 00354 { 00355 basic_string<CharT> s; 00356 s.append(m_str.str,m_str.length); 00357 return s; 00358 } 00359 00360 00361 template <class CharT> 00362 inline void CFXString_base<CharT>::assign(CharT* data,size_t len,bool bNoCopy) 00363 { 00364 m_str.length = len; 00365 if(bNoCopy) 00366 { 00367 if(m_bOwner) 00368 delete[] m_str.str; 00369 m_str.str = data; 00370 m_bOwner = false; 00371 } 00372 else 00373 { 00374 if(!m_bOwner) 00375 m_str.str = new CharT[len + 1]; 00376 else 00377 { 00378 if((len > m_maxLen)) 00379 { 00380 if(m_str.str) 00381 delete[] m_str.str; 00382 m_str.str = new CharT[len + 1]; 00383 m_maxLen = len; 00384 } 00385 } 00386 memcpy(m_str.str,data,len*sizeof(CharT)); 00387 m_str.str[len] = 0; 00388 m_bOwner = true; 00389 } 00390 } 00391 00392 template <class CharT> 00393 inline bool CFXString_base<CharT>::operator==(const CharT* str)const 00394 { 00395 return (compare(str)==0); 00396 } 00397 template <class CharT> 00398 inline bool CFXString_base<CharT>::operator==(const basic_string<CharT>& str)const 00399 { 00400 return (compare(str)==0); 00401 } 00402 template <class CharT> 00403 inline bool CFXString_base<CharT>::operator==(const CFXString_base<CharT>& right)const 00404 { 00405 return (compare(right)==0); 00406 } 00407 00408 template <class CharT> 00409 inline bool CFXString_base<CharT>::operator!=(const CharT* str)const 00410 { 00411 return (compare(str) != 0); 00412 } 00413 template <class CharT> 00414 inline bool CFXString_base<CharT>::operator!=(const basic_string<CharT>& str)const 00415 { 00416 return (compare(str) != 0); 00417 } 00418 template <class CharT> 00419 inline bool CFXString_base<CharT>::operator!=(const CFXString_base<CharT>& right)const 00420 { 00421 return (compare(right)!=0); 00422 } 00423 00424 template<> 00425 inline int CFXString_base<char>::compare(const CFXString_base<char>& right)const 00426 { 00427 if(m_str.length == 0 || right.m_str.length == 0) 00428 { 00429 return (int)(m_str.length - right.m_str.length); 00430 } 00431 int ret = strncmp(m_str.str,right.m_str.str,(m_str.length > right.m_str.length)?right.m_str.length:m_str.length); 00432 if (ret != 0) return ret; 00433 else return (int)(m_str.length - right.m_str.length); 00434 } 00435 00436 template<> 00437 inline int CFXString_base<wchar_t>::compare(const CFXString_base<wchar_t>& right)const 00438 { 00439 if(m_str.length == 0 || right.m_str.length == 0) 00440 { 00441 return (int)(m_str.length - right.m_str.length); 00442 } 00443 int ret = wcsncmp(m_str.str,right.m_str.str,(m_str.length > right.m_str.length)?right.m_str.length:m_str.length); 00444 if (ret != 0) return ret; 00445 else return (int)(m_str.length - right.m_str.length); 00446 } 00447 00448 template<> 00449 inline int CFXString_base<char>::compare(const char* str)const 00450 { 00451 size_t rlen = strlen(str); 00452 if(m_str.length == 0 || rlen == 0) 00453 { 00454 return (int)(m_str.length - rlen); 00455 } 00456 int ret = strncmp(m_str.str,str,(m_str.length > rlen)?rlen:m_str.length); 00457 if (ret != 0) return ret; 00458 else return (int)(m_str.length - rlen); 00459 } 00460 00461 template<> 00462 inline int CFXString_base<wchar_t>::compare(const wchar_t* str)const 00463 { 00464 size_t rlen = wcslen(str); 00465 if(m_str.length == 0 || rlen == 0) 00466 { 00467 return (int)(m_str.length - rlen); 00468 } 00469 int ret = wcsncmp(m_str.str,str,(m_str.length > rlen)?rlen:m_str.length); 00470 if (ret != 0) return ret; 00471 else return (int)(m_str.length - rlen); 00472 } 00473 template<> 00474 inline int CFXString_base<char>::compare(const basic_string<char>& str)const 00475 { 00476 size_t rlen = str.length(); 00477 if(m_str.length == 0 || rlen == 0) 00478 { 00479 return (int)(m_str.length - rlen); 00480 } 00481 int ret = strncmp(m_str.str,str.c_str(),(m_str.length > rlen)?rlen:m_str.length); 00482 if (ret != 0) return ret; 00483 else return (int)(m_str.length - rlen); 00484 } 00485 template<> 00486 inline int CFXString_base<wchar_t>::compare(const basic_string<wchar_t>& str)const 00487 { 00488 size_t rlen = str.length(); 00489 if(m_str.length == 0 || rlen == 0) 00490 { 00491 return (int)(m_str.length - rlen); 00492 } 00493 int ret = wcsncmp(m_str.str,str.c_str(),(m_str.length > rlen)?rlen:m_str.length); 00494 if (ret != 0) return ret; 00495 else return (int)(m_str.length - rlen); 00496 } 00497 template<> 00498 inline int CFXString_base<wchar_t>::compare(const _str<wchar_t>& str)const 00499 { 00500 size_t rlen = str.length; 00501 if(m_str.length == 0 || rlen == 0) 00502 { 00503 return (int)(m_str.length - rlen); 00504 } 00505 int ret = wcsncmp(m_str.str,str.str,(m_str.length > rlen)?rlen:m_str.length); 00506 if (ret != 0) return ret; 00507 else return (int)(m_str.length - rlen); 00508 } 00509 template <class CharT> 00510 inline const CharT* CFXString_base<CharT>::c_str()const 00511 { 00512 return m_str.str; 00513 } 00514 00515 template <class CharT> 00516 inline CharT* CFXString_base<CharT>::data() 00517 { 00518 return m_str.str; 00519 } 00520 00521 template <class CharT> 00522 inline size_t CFXString_base<CharT>::length()const 00523 { 00524 return m_str.length; 00525 } 00526 00527 template <class CharT> 00528 void CFXString_base<CharT>::reserve(size_t len) 00529 { 00530 if(len > m_maxLen) 00531 { 00532 CharT* tmp = new CharT[len + 1]; 00533 memset(m_str.str,0,sizeof(CharT)*(len+1)); 00534 if(m_str.length > 0) 00535 { 00536 memcpy(tmp,m_str.str,(m_str.length>len)?len:m_str.length); 00537 } 00538 if(m_bOwner && m_str.str) 00539 delete[] m_str.str; 00540 m_str.str = tmp; 00541 m_str.length = len; 00542 m_bOwner = true; 00543 m_maxLen = len; 00544 } 00545 } 00546 00547 template <class CharT> 00548 void CFXString_base<CharT>::resize(size_t len) 00549 { 00550 if(len > m_maxLen) 00551 { 00552 if(m_bOwner) 00553 reserve(len); 00554 else 00555 FIRTEX_THROW3(BAD_PARAMETER_ERROR,"void CFXString_base<CharT>::resize(size_t len):can't resize string,maybe the string is not owned by CFXString."); 00556 } 00557 m_str.str[len] = 0; 00558 m_str.length = len; 00559 } 00560 template <class CharT> 00561 void CFXString_base<CharT>::append(CharT* str,size_t len) 00562 { 00563 if((m_str.length + len) > m_maxLen) 00564 { 00565 if(m_bOwner) 00566 reserve(m_str.length + len); 00567 else 00568 FIRTEX_THROW3(BAD_PARAMETER_ERROR,"void CFXString_base<CharT>::append(CharT* str,size_t len):can't append string,maybe the string is not owned by CFXString."); 00569 } 00570 memcpy(m_str.str+m_str.length,str,len*sizeof(CharT)); 00571 m_str.length += len; 00572 m_str.str[m_str.length] = 0; 00573 } 00574 00575 00576 typedef CFXString_base<char> CFXStringA; 00577 typedef CFXString_base<wchar_t> CFXStringW; 00578 00579 #ifdef _UNICODE 00580 typedef CFXStringW CFXString; 00581 #else 00582 typedef CFXStringA CFXString; 00583 #endif 00584 00585 } 00586 } 00587 00588 #endif
http://www.firtex.org http://www.sourceforge.net/projects/firtex