1 | /*------------------------------------------------------------------------------ |
2 | * Copyright (C) 2003-2006 Ben van Klinken and the CLucene Team |
3 | * |
4 | * Distributable under the terms of either the Apache License (Version 2.0) or |
5 | * the GNU Lesser General Public License, as specified in the COPYING file. |
6 | * |
7 | * Changes are Copyright (C) 2009 Nokia Corporation and/or its subsidiary(-ies). |
8 | ------------------------------------------------------------------------------*/ |
9 | #ifndef lucene_stdheader_h |
10 | #define lucene_stdheader_h |
11 | |
12 | #if defined(OVERRIDE_DEFAULT_CLCONFIG) |
13 | #include "AltCLConfig.h" |
14 | #else |
15 | #include "CLucene/CLConfig.h" |
16 | #endif |
17 | |
18 | //first inclusion of compiler.h (it will be called again later) |
19 | #include "CLucene/config/compiler.h" |
20 | |
21 | extern void _lucene_shutdown(); |
22 | extern int _lucene_counter_break; //can set a watch on this |
23 | #if defined(LUCENE_ENABLE_MEMLEAKTRACKING) |
24 | extern bool _lucene_disable_debuglogging; //if LUCENE_ENABLE_CONSTRUCTOR_LOG is on, dont do log if this is true |
25 | #endif |
26 | |
27 | //////////////////////////////////////////////////////// |
28 | // default includes |
29 | //////////////////////////////////////////////////////// |
30 | #ifndef LUCENE_DISABLE_INCLUDES |
31 | |
32 | #include <stdio.h> |
33 | |
34 | #if defined(_CL_STDC_HEADERS) |
35 | #include <stdlib.h> |
36 | #include <stddef.h> |
37 | #else |
38 | #if defined(_CL_HAVE_STDLIB_H) |
39 | #include <stdlib.h> |
40 | #endif |
41 | #endif |
42 | |
43 | #if defined(_CL_HAVE_STRING_H) |
44 | #if !defined(_CL_STDC_HEADERS) && defined(_CL_HAVE_MEMORY_H) |
45 | #include <memory.h> |
46 | #endif |
47 | #include <string.h> |
48 | #elif defined(_CL_HAVE_STRINGS_H) |
49 | //note: as a side note, strtok is not thread-safe.. so be careful where you use it! |
50 | #error "strtok replacement for BSD has not been implemented" |
51 | #include <strings.h> |
52 | #if !defined(_CL_HAVE_STRCHR) |
53 | #define strchr index |
54 | #define strrchr rindex |
55 | #endif |
56 | #endif |
57 | |
58 | #if defined(_CL_HAVE_UNISTD_H) |
59 | #include <unistd.h> |
60 | #elif defined(_CL_HAVE_IO_H) && defined(_CL_HAVE_DIRECT_H) |
61 | #ifndef UNDER_CE |
62 | #include <io.h> |
63 | #include <direct.h> |
64 | #endif |
65 | #else |
66 | #error "Neither unistd.h or (io.h & direct.h) were available" |
67 | #endif |
68 | |
69 | #ifndef _CL_DISABLE_NATIVE_EXCEPTIONS |
70 | #ifdef _CL_HAVE_STDEXCEPT |
71 | #include <stdexcept> |
72 | #else |
73 | #error "CLucene can't compile with exception handling on because <stdexcept> header is not available" |
74 | #endif |
75 | #endif |
76 | |
77 | #if defined(_CL_STAT_MACROS_BROKEN) |
78 | #error "Haven't implemented STAT_MACROS_BROKEN fix yet" |
79 | #elif defined(_CL_HAVE_SYS_STAT_H) |
80 | #ifdef UNDER_CE |
81 | #include <types.h> |
82 | #else |
83 | #include <sys/stat.h> |
84 | #endif |
85 | #else |
86 | #error "Haven't implemented platforms with no sys/stat.h" |
87 | #endif |
88 | |
89 | #if defined(_CL_HAVE_STDARG_H) |
90 | #include <stdarg.h> |
91 | #else |
92 | #error "CLucene can compile, but some extras may not work" |
93 | #endif |
94 | |
95 | #if defined(_CL_HAVE_MATH_H) |
96 | #include <math.h> |
97 | #else |
98 | #error "CLucene can't compile without <math.h>" |
99 | #endif |
100 | |
101 | #if defined(_CL_HAVE_MAP) |
102 | #include <map> |
103 | #else |
104 | #error "CLucene can't compile without the map header" |
105 | #endif |
106 | |
107 | #if defined(_CL_HAVE_LIST) |
108 | #include <list> |
109 | #else |
110 | #error "CLucene can't compile without the list header" |
111 | #endif |
112 | |
113 | #if defined(_CL_HAVE_SET) |
114 | #include <set> |
115 | #else |
116 | #error "CLucene can't compile without the set header" |
117 | #endif |
118 | |
119 | #if defined(_CL_HAVE_VECTOR) |
120 | #include <vector> |
121 | #else |
122 | #error "CLucene can't compile without the vector header" |
123 | #endif |
124 | |
125 | #if !defined(LUCENE_DISABLE_HASHING) && defined(_CL_HAVE_HASH_MAP) && defined(_CL_HAVE_HASH_SET) |
126 | //hashing is all or nothing! |
127 | #include <hash_map> |
128 | #include <hash_set> |
129 | #elif !defined(LUCENE_DISABLE_HASHING) && defined(_CL_HAVE_EXT_HASH_MAP) && defined(_CL_HAVE_EXT_HASH_SET) |
130 | #include <ext/hash_map> |
131 | #include <ext/hash_set> |
132 | #elif !defined(LUCENE_DISABLE_HASHING) |
133 | #define LUCENE_DISABLE_HASHING |
134 | #endif |
135 | #if !defined(LUCENE_DISABLE_HASHING) && !defined(CL_NS_HASHING) |
136 | #define CL_NS_HASHING(func) std::func |
137 | #endif |
138 | |
139 | #if defined(_CL_HAVE_ALGORITHM) |
140 | # include <algorithm> |
141 | #else |
142 | # error "Can't compile clucene without <algorithm>" |
143 | #endif |
144 | |
145 | #if defined(_CL_HAVE_FUNCTIONAL) |
146 | # include <functional> |
147 | #else |
148 | # error "Can't compile clucene without <functional>" |
149 | #endif |
150 | |
151 | #if !defined(_CL_HAVE_PRINTF) |
152 | #error "CLucene can't compile without printf, replacements have not been implemented" |
153 | #endif |
154 | |
155 | #if !defined(_CL_HAVE_SNPRINTF) && !defined(_CL_HAVE__SNPRINTF) |
156 | #error "CLucene can't compile without snprintf, replacements have not been implemented" |
157 | #elif !defined(_CL_HAVE__SNPRINTF)&& defined(_CL_HAVE_SVNPRINTF) |
158 | #define _snprintf snprintf |
159 | #endif |
160 | |
161 | #if defined(_UCS2) |
162 | #if defined(_CL_HAVE_WCHAR_H) |
163 | #include <wchar.h> |
164 | #else |
165 | //actually the repl_wchar.h replacements header will |
166 | //always be included. It replaces some functions |
167 | //that are missing in some wchar.h headers. |
168 | #endif |
169 | #endif |
170 | |
171 | #if defined(_UCS2) && defined(_CL_HAVE_WCTYPE_H) |
172 | #include <wctype.h> |
173 | #elif defined(_ASCII) && defined(_CL_HAVE_CTYPE_H) |
174 | #include <ctype.h> |
175 | #undef LUCENE_USE_INTERNAL_CHAR_FUNCTIONS |
176 | #elif defined(_UCS2) |
177 | //must be in _UCS2 to use internal char functions |
178 | #undef LUCENE_USE_INTERNAL_CHAR_FUNCTIONS |
179 | #define LUCENE_USE_INTERNAL_CHAR_FUNCTIONS |
180 | #else |
181 | #error "Cannot compile in _ASCII without ctype.h" |
182 | #endif |
183 | |
184 | //always include replacement, some missing tchar defines |
185 | #include "CLucene/config/repl_tchar.h" |
186 | |
187 | #if defined(_CL_HAVE_ERRNO_H) |
188 | #ifndef UNDER_CE |
189 | #include <errno.h> |
190 | #endif |
191 | #else |
192 | #error "Haven't implemented platforms with no errno.h" |
193 | #endif |
194 | |
195 | #if defined(_CL_HAVE_FCNTL_H) |
196 | #ifndef UNDER_CE |
197 | #include <fcntl.h> |
198 | #endif |
199 | #else |
200 | #error "Haven't implemented platforms with no fcntl.h" |
201 | #endif |
202 | |
203 | #if defined(_CL_HAVE_WINDOWS_H) |
204 | #include <windows.h> |
205 | #endif |
206 | |
207 | #endif //LUCENE_DISABLE_INCLUDES |
208 | // |
209 | //////////////////////////////////////////////////////// |
210 | |
211 | //second inclusion of compiler.h |
212 | //this gives CompilerXXX.h a chance to include other headers |
213 | #include "CLucene/config/compiler.h" |
214 | // |
215 | //////////////////////////////////////////////////////// |
216 | |
217 | |
218 | //////////////////////////////////////////////////////// |
219 | // Character functions. |
220 | // Here we decide whose character functions to use |
221 | //////////////////////////////////////////////////////// |
222 | #if defined(LUCENE_USE_INTERNAL_CHAR_FUNCTIONS) |
223 | #define stringCaseFold cl_tcscasefold |
224 | #define stringCaseFoldCmp cl_tcscasefoldcmp |
225 | |
226 | #undef _istspace |
227 | #undef _istdigit |
228 | #undef _istalnum |
229 | #undef _istalpha |
230 | #undef _totlower |
231 | #undef _totupper |
232 | #define _istalnum cl_isalnum |
233 | #define _istalpha cl_isletter |
234 | #define _istspace cl_isspace |
235 | #define _istdigit cl_isdigit |
236 | #define _totlower cl_tolower |
237 | #define _totupper cl_toupper |
238 | |
239 | //here are some functions to help deal with utf8/ucs2 conversions |
240 | //lets let the user decide what mb functions to use... we provide pure utf8 ones no matter what. |
241 | /*#undef _mbtowc |
242 | #undef _mbstowcs |
243 | #undef _wctomb |
244 | #undef _wcstombs |
245 | #define _mbtowc lucene_mbstowc |
246 | #define _mbsstowcs lucene_mbstowcs |
247 | #define _wctomb lucene_wcto_mb |
248 | #define _wcstombs lucene_wcstombs*/ |
249 | #else |
250 | //we are using native functions |
251 | //here are some functions to help deal with utf8/ucs2 conversions |
252 | /*#define _mbtowc mbtowc |
253 | #define _wctomb wctomb |
254 | #define _mbstowcs mbstowcs |
255 | #define _wcstombs wcstombs*/ |
256 | |
257 | //we are using native character functions |
258 | #if defined(_ASCII) |
259 | #undef _istspace |
260 | #undef _istdigit |
261 | #undef _istalnum |
262 | #undef _istalpha |
263 | #undef _totlower |
264 | #undef _totupper |
265 | #define _istspace(x) isspace((unsigned char)x) |
266 | #define _istdigit(x) isdigit((unsigned char)x) |
267 | #define _istalnum(x) isalnum((unsigned char)x) |
268 | #define _istalpha(x) isalpha((unsigned char)x) |
269 | #define _totlower(x) tolower((unsigned char)x) |
270 | #define _totupper(x) toupper((unsigned char)x) |
271 | #endif |
272 | #endif |
273 | |
274 | //the methods contained in gunichartables.h |
275 | typedef unsigned long clunichar; |
276 | bool cl_isletter(clunichar c); |
277 | bool cl_isalnum(clunichar c); |
278 | bool cl_isdigit(clunichar c); |
279 | bool cl_isspace (clunichar c); |
280 | TCHAR cl_tolower (TCHAR c); |
281 | TCHAR cl_toupper (TCHAR c); |
282 | |
283 | int cl_tcscasefoldcmp(const TCHAR * dst, const TCHAR * src); |
284 | TCHAR* cl_tcscasefold( TCHAR * str, int len=-1 ); |
285 | |
286 | //we provide utf8 conversion functions |
287 | size_t lucene_utf8towc (wchar_t *ret, const char *s, size_t n); |
288 | size_t lucene_utf8towcs(wchar_t *, const char *, size_t maxslen); |
289 | size_t lucene_wctoutf8 (char * ret, const wchar_t str); |
290 | size_t lucene_wcstoutf8 (char *, const wchar_t *, size_t maxslen); |
291 | size_t lucene_utf8charlen(const char *p); |
292 | |
293 | ///a blank string... |
294 | extern TCHAR* _LUCENE_BLANK_STRING; |
295 | #define LUCENE_BLANK_STRING _LUCENE_BLANK_STRING |
296 | extern char* _LUCENE_BLANK_ASTRING; |
297 | #define LUCENE_BLANK_ASTRING _LUCENE_BLANK_ASTRING |
298 | |
299 | /* Converts a string into a form that is independent of case. The |
300 | * result will not correspond to any particular case, but can be |
301 | * compared for equality or ordered with the results of calling |
302 | * stringCaseFold() on other strings. |
303 | * |
304 | * If we did not define this elsewhere, then just convert to lower case |
305 | */ |
306 | #ifndef stringCaseFold |
307 | #define stringCaseFold _tcslwr |
308 | #endif |
309 | /* Compares 2 strings using case folding (if available) |
310 | * If we did not define this elsewhere, then just compare |
311 | * using normal method |
312 | */ |
313 | #ifndef stringCaseFoldCmp |
314 | #define stringCaseFoldCmp _tcsicmp |
315 | #endif |
316 | |
317 | //now that all the character routines are completed, include the |
318 | //wchar.h replacements. |
319 | #include "CLucene/config/repl_wchar.h" //always include replacements |
320 | |
321 | //a replacement for _tcsdup. This uses new TCHAR[] instead of malloc, so that we can use delete[] to free |
322 | #if defined(LUCENE_ENABLE_FILELINEINFO) |
323 | #define CL_FILELINE ,__FILE__,__LINE__ |
324 | #define CL_FILELINEREF ,file,line ///<for passing the reference along to another function |
325 | #define CL_FILELINEREF2 ,file,line ///<for passing the reference along to another function |
326 | #define CL_FILELINEPARAM ,char* file,int line |
327 | #else |
328 | #define CL_FILELINE |
329 | #define CL_FILELINEREF |
330 | #define CL_FILELINEREF2 ,NULL,-1 |
331 | #define CL_FILELINEPARAM |
332 | #endif |
333 | |
334 | char* lucenestrdup(const char* v CL_FILELINEPARAM); |
335 | #if defined(_UCS2) |
336 | wchar_t* lucenewcsdup(const wchar_t* v CL_FILELINEPARAM); |
337 | #define stringDuplicate(x) lucenewcsdup(x CL_FILELINE) //don't change this... uses [] instead of malloc |
338 | #else |
339 | #define stringDuplicate(x) lucenestrdup(x CL_FILELINE) //don't change this... uses [] instead of malloc |
340 | #endif |
341 | |
342 | #define STRCPY_AtoA(target,src,len) strncpy(target,src,len) |
343 | #define STRDUP_AtoA(x) lucenestrdup(x CL_FILELINE) |
344 | |
345 | #if defined(_UCS2) |
346 | #define STRDUP_WtoW(x) lucenewcsdup(x CL_FILELINE) |
347 | #define STRDUP_TtoT STRDUP_WtoW |
348 | #define STRDUP_WtoT STRDUP_WtoW |
349 | #define STRDUP_TtoW STRDUP_WtoW |
350 | |
351 | #define STRDUP_AtoW(x) CL_NS(util)::Misc::_charToWide(x CL_FILELINE) |
352 | #define STRDUP_AtoT STRDUP_AtoW |
353 | |
354 | #define STRDUP_WtoA(x) CL_NS(util)::Misc::_wideToChar(x CL_FILELINE) |
355 | #define STRDUP_TtoA STRDUP_WtoA |
356 | |
357 | #define STRCPY_WtoW(target,src,len) _tcsncpy(target,src,len) |
358 | #define STRCPY_TtoW STRCPY_WtoW |
359 | #define STRCPY_WtoT STRCPY_WtoW |
360 | #define STRCPY_TtoT STRCPY_WtoW |
361 | |
362 | #define STRCPY_AtoW(target,src,len) CL_NS(util)::Misc::_cpycharToWide(src,target,len) |
363 | #define STRCPY_AtoT STRCPY_AtoW |
364 | |
365 | #define STRCPY_WtoA(target,src,len) CL_NS(util)::Misc::_cpywideToChar(src,target,len) |
366 | #define STRCPY_TtoA STRCPY_WtoA |
367 | #else |
368 | #define STRDUP_AtoT STRDUP_AtoA |
369 | #define STRDUP_TtoA STRDUP_AtoA |
370 | #define STRDUP_TtoT STRDUP_AtoA |
371 | |
372 | #define STRDUP_WtoT(x) xxxxxxxxxxxxxxx //not possible |
373 | #define STRCPY_WtoT(target,src,len) xxxxxxxxxxxxxxx //not possible |
374 | |
375 | #define STRCPY_AtoT STRCPY_AtoA |
376 | #define STRCPY_TtoA STRCPY_AtoA |
377 | #define STRCPY_TtoT STRCPY_AtoA |
378 | #endif |
379 | |
380 | // |
381 | //////////////////////////////////////////////////////// |
382 | |
383 | //////////////////////////////////////////////////////// |
384 | //namespace helper |
385 | //////////////////////////////////////////////////////// |
386 | #if defined(_LUCENE_DONTIMPLEMENT_NS_MACROS) |
387 | |
388 | #elif !defined(DISABLE_NAMESPACE) |
389 | // |
390 | // W A R N I N G |
391 | // ------------- |
392 | // |
393 | // adjustments here, need to be done in |
394 | // QTDIR/src/tools/assistant/lib/fulltextsearch/qclucene_global.h as well |
395 | // |
396 | # ifdef QT_NAMESPACE |
397 | # define CL_NS_DEF(sub) namespace QT_NAMESPACE { namespace lucene{ namespace sub{ |
398 | # define CL_NS_DEF2(sub,sub2) namespace QT_NAMESPACE { namespace lucene{ namespace sub{ namespace sub2 { |
399 | |
400 | # define CL_NS_END }}} |
401 | # define CL_NS_END2 }}}} |
402 | |
403 | # define CL_NS_USE(sub) using namespace QT_NAMESPACE::lucene::sub; |
404 | # define CL_NS_USE2(sub,sub2) using namespace QT_NAMESPACE::lucene::sub::sub2; |
405 | |
406 | # define CL_NS(sub) QT_NAMESPACE::lucene::sub |
407 | # define CL_NS2(sub,sub2) QT_NAMESPACE::lucene::sub::sub2 |
408 | # else |
409 | # define CL_NS_DEF(sub) namespace lucene{ namespace sub{ |
410 | # define CL_NS_DEF2(sub,sub2) namespace lucene{ namespace sub{ namespace sub2 { |
411 | |
412 | # define CL_NS_END }} |
413 | # define CL_NS_END2 }}} |
414 | |
415 | # define CL_NS_USE(sub) using namespace lucene::sub; |
416 | # define CL_NS_USE2(sub,sub2) using namespace lucene::sub::sub2; |
417 | |
418 | # define CL_NS(sub) lucene::sub |
419 | # define CL_NS2(sub,sub2) lucene::sub::sub2 |
420 | # endif |
421 | #else |
422 | # define CL_NS_DEF(sub) |
423 | # define CL_NS_DEF2(sub, sub2) |
424 | # define CL_NS_END |
425 | # define CL_NS_END2 |
426 | # define CL_NS_USE(sub) |
427 | # define CL_NS_USE2(sub,sub2) |
428 | # define CL_NS(sub) |
429 | # define CL_NS2(sub,sub2) |
430 | #endif |
431 | |
432 | #if defined(LUCENE_NO_STDC_NAMESPACE) |
433 | //todo: haven't actually tested this on a non-stdc compliant compiler |
434 | #define CL_NS_STD(func) ::func |
435 | #else |
436 | #define CL_NS_STD(func) std::func |
437 | #endif |
438 | // |
439 | //////////////////////////////////////////////////////// |
440 | |
441 | // |
442 | void CLDebugBreak(); //define a debugbreak function |
443 | |
444 | |
445 | //////////////////////////////////////////////////////////////// |
446 | // These are defines and functions used throughout clucene |
447 | //////////////////////////////////////////////////////////////// |
448 | #undef _T //remove any previously defined _T - required for ppc os |
449 | #if defined(_UCS2) |
450 | #define _T(x) L ## x |
451 | #else |
452 | #define _T(x) x |
453 | #endif |
454 | |
455 | //third inclusion of compiler.h |
456 | //this gives CompilerXXX.h a chance to fix any of the |
457 | //default settings |
458 | #include "CLucene/config/compiler.h" |
459 | |
460 | #if defined _MSC_VER && (_MSC_VER < 1300) |
461 | # define LUCENE_NO_STDC_NAMESPACE |
462 | #endif |
463 | |
464 | //use std namespace |
465 | #ifndef LUCENE_NO_STDC_NAMESPACE |
466 | #ifdef _LIBCPP_VERSION |
467 | // clang's libc++ pulls in cwchar, clucene pulls in wchar.h and the entire std |
468 | // namespace which causes ambiguity between ::wcschar() and std::wcschar(), but |
469 | // clucene is only using std::min() and std::max() |
470 | using std::min; |
471 | using std::max; |
472 | #else |
473 | using namespace std; |
474 | #endif |
475 | #endif |
476 | |
477 | |
478 | //////////////////////////////////////////////////////// |
479 | //misc shortcut defines |
480 | //////////////////////////////////////////////////////// |
481 | |
482 | //include the headers that we need practically everywhere |
483 | #include "CLucene/debug/error.h" //all delete/creation/mem debugging code |
484 | #include "CLucene/debug/condition.h" //conditional debugging (like assert) |
485 | #include "CLucene/debug/mem.h" //all delete/creation/mem debugging code |
486 | #include "LuceneThreads.h" //lucene threading support |
487 | |
488 | #include "CLucene/util/Misc.h" |
489 | #include "CLucene/util/Equators.h" |
490 | #include "CLucene/util/VoidList.h" |
491 | #include "CLucene/util/VoidMap.h" |
492 | |
493 | |
494 | CL_NS_DEF(util) |
495 | typedef CL_NS(util)::CLVector<TCHAR*> StringArray; |
496 | typedef CL_NS(util)::CLVector<TCHAR*, CL_NS(util)::Deletor::tcArray > StringArrayWithDeletor; |
497 | typedef CL_NS(util)::CLVector<const TCHAR*> StringArrayConst; |
498 | typedef CL_NS(util)::CLVector<const TCHAR*, CL_NS(util)::Deletor::tcArray > StringArrayConstWithDeletor; |
499 | |
500 | typedef CL_NS(util)::CLVector<char*> AStringArray; |
501 | typedef CL_NS(util)::CLVector<char*, CL_NS(util)::Deletor::acArray > AStringArrayWithDeletor; |
502 | typedef CL_NS(util)::CLVector<const char*> AStringArrayConst; |
503 | typedef CL_NS(util)::CLVector<const char*, CL_NS(util)::Deletor::acArray > AStringArrayConstWithDeletor; |
504 | CL_NS_END |
505 | |
506 | // |
507 | //////////////////////////////////////////////////////// |
508 | |
509 | #endif // STDHEADER_H |
510 | |