src/common/intl.cpp

   1 /////////////////////////////////////////////////////////////////////////////
   2 // Name:        intl.cpp
   3 // Purpose:     Internationalization and localisation for wxWindows
   4 // Author:      Vadim Zeitlin
   5 // Modified by:
   6 // Created:     29/01/98
   7 // RCS-ID:      $Id$
   8 // Copyright:   (c) 1998 Vadim Zeitlin <zeitlin@dptmaths.ens-cachan.fr>
   9 // Licence:     wxWindows license
  10 /////////////////////////////////////////////////////////////////////////////
  11
  12 // ============================================================================
  13 // declaration
  14 // ============================================================================
  15
  16 // ----------------------------------------------------------------------------
  17 // headers
  18 // ----------------------------------------------------------------------------
  19
  20 #ifdef __GNUG__
  21     #pragma implementation "intl.h"
  22 #endif
  23
  24 // For compilers that support precompilation, includes "wx.h".
  25 #include "wx/wxprec.h"
  26
  27 #ifdef __BORLANDC__
  28     #pragma hdrstop
  29 #endif
  30
  31 // standard headers
  32 #include  <locale.h>
  33 #include  <ctype.h>
  34
  35 // wxWindows
  36 #include "wx/defs.h"
  37 #include "wx/string.h"
  38 #include "wx/intl.h"
  39 #include "wx/file.h"
  40 #include "wx/log.h"
  41 #include "wx/utils.h"
  42
  43 #include <stdlib.h>
  44
  45 // ----------------------------------------------------------------------------
  46 // simple types
  47 // ----------------------------------------------------------------------------
  48
  49 // FIXME adjust if necessary
  50 typedef unsigned char size_t8;
  51 typedef unsigned long size_t32;
  52
  53 // ----------------------------------------------------------------------------
  54 // constants
  55 // ----------------------------------------------------------------------------
  56
  57 // magic number identifying the .mo format file
  58 const size_t32 MSGCATALOG_MAGIC    = 0x950412de;
  59 const size_t32 MSGCATALOG_MAGIC_SW = 0xde120495;
  60
  61 // extension of ".mo" files
  62 #define MSGCATALOG_EXTENSION  ".mo"
  63
  64 // ----------------------------------------------------------------------------
  65 // global functions
  66 // ----------------------------------------------------------------------------
  67
  68 // suppress further error messages about missing translations
  69 // (if you don't have one catalog file, you wouldn't like to see the
  70 //  error message for each string in it, so normally it's given only
  71 //  once)
  72 void wxSuppressTransErrors();
  73
  74 // restore the logging
  75 void wxRestoreTransErrors();
  76
  77 // get the current state
  78 bool wxIsLoggingTransErrors();
  79
  80 static wxLocale *wxSetLocale(wxLocale *pLocale);
  81
  82 // ----------------------------------------------------------------------------
  83 // wxMsgCatalog corresponds to one disk-file message catalog.
  84 //
  85 // This is a "low-level" class and is used only by wxLocale (that's why
  86 // it's designed to be stored in a linked list)
  87 // ----------------------------------------------------------------------------
  88
  89 class wxMsgCatalog
  90 {
  91 public:
  92   // ctor & dtor
  93   wxMsgCatalog();
  94  ~wxMsgCatalog();
  95
  96   // load the catalog from disk (szDirPrefix corresponds to language)
  97   bool Load(const char *szDirPrefix, const char *szName);
  98   bool IsLoaded() const { return m_pData != NULL; }
  99
 100   // get name of the catalog
 101   const char *GetName() const { return m_pszName; }
 102
 103   // get the translated string: returns NULL if not found
 104   const char *GetString(const char *sz) const;
 105
 106   // public variable pointing to the next element in a linked list (or NULL)
 107   wxMsgCatalog *m_pNext;
 108
 109 private:
 110   // this implementation is binary compatible with GNU gettext() version 0.10
 111
 112   // an entry in the string table
 113   struct wxMsgTableEntry
 114   {
 115     size_t32   nLen;           // length of the string
 116     size_t32   ofsString;      // pointer to the string
 117   };
 118
 119   // header of a .mo file
 120   struct wxMsgCatalogHeader
 121   {
 122     size_t32  magic,          // offset +00:  magic id
 123               revision,       //        +04:  revision
 124               numStrings;     //        +08:  number of strings in the file
 125     size_t32  ofsOrigTable,   //        +0C:  start of original string table
 126               ofsTransTable;  //        +10:  start of translated string table
 127     size_t32  nHashSize,      //        +14:  hash table size
 128               ofsHashTable;   //        +18:  offset of hash table start
 129   };
 130
 131   // all data is stored here, NULL if no data loaded
 132   size_t8 *m_pData;
 133
 134   // data description
 135   size_t32          m_numStrings,   // number of strings in this domain
 136                     m_nHashSize;    // number of entries in hash table
 137   size_t32         *m_pHashTable;   // pointer to hash table
 138   wxMsgTableEntry  *m_pOrigTable,   // pointer to original   strings
 139                    *m_pTransTable;  //            translated
 140
 141   const char *StringAtOfs(wxMsgTableEntry *pTable, size_t32 index) const
 142     { return (const char *)(m_pData + Swap(pTable[index].ofsString)); }
 143
 144   // utility functions
 145     // calculate the hash value of given string
 146   static inline size_t32 GetHash(const char *sz);
 147     // big<->little endian
 148   inline size_t32 Swap(size_t32 ui) const;
 149
 150   // internal state
 151   bool HasHashTable() const // true if hash table is present
 152     { return m_nHashSize > 2 && m_pHashTable != NULL; }
 153
 154   bool          m_bSwapped;   // wrong endianness?
 155
 156   char         *m_pszName;    // name of the domain
 157 };
 158
 159 // ----------------------------------------------------------------------------
 160 // global variables
 161 // ----------------------------------------------------------------------------
 162
 163 // the list of the directories to search for message catalog files
 164 static wxArrayString s_searchPrefixes;
 165
 166 // ============================================================================
 167 // implementation
 168 // ============================================================================
 169
 170 // ----------------------------------------------------------------------------
 171 // wxMsgCatalog class
 172 // ----------------------------------------------------------------------------
 173
 174 // calculate hash value using the so called hashpjw function by P.J. Weinberger
 175 // [see Aho/Sethi/Ullman, COMPILERS: Principles, Techniques and Tools]
 176 size_t32 wxMsgCatalog::GetHash(const char *sz)
 177 {
 178   #define HASHWORDBITS 32     // the length of size_t32
 179
 180   size_t32 hval = 0;
 181   size_t32 g;
 182   while ( *sz != '\0' ) {
 183     hval <<= 4;
 184     hval += (size_t32)*sz++;
 185     g = hval & ((size_t32)0xf << (HASHWORDBITS - 4));
 186     if ( g != 0 ) {
 187       hval ^= g >> (HASHWORDBITS - 8);
 188       hval ^= g;
 189     }
 190   }
 191
 192   return hval;
 193 }
 194
 195 // swap the 2 halves of 32 bit integer if needed
 196 size_t32 wxMsgCatalog::Swap(size_t32 ui) const
 197 {
 198   return m_bSwapped ? (ui << 24) | ((ui & 0xff00) << 8) |
 199                       ((ui >> 8) & 0xff00) | (ui >> 24)
 200                     : ui;
 201 }
 202
 203 wxMsgCatalog::wxMsgCatalog()
 204 {
 205   m_pData   = NULL;
 206   m_pszName = NULL;
 207 }
 208
 209 wxMsgCatalog::~wxMsgCatalog()
 210 {
 211   wxDELETEA(m_pData);
 212   wxDELETEA(m_pszName);
 213 }
 214
 215 // small class to suppress the translation erros until exit from current scope
 216 class NoTransErr
 217 {
 218 public:
 219     NoTransErr() { wxSuppressTransErrors(); }
 220    ~NoTransErr() { wxRestoreTransErrors();  }
 221 };
 222
 223 // return all directories to search for given prefix
 224 static wxString GetAllMsgCatalogSubdirs(const char *prefix,
 225                                         const char *lang)
 226 {
 227     wxString searchPath;
 228
 229     // search first in prefix/fr/LC_MESSAGES, then in prefix/fr and finally in
 230     // prefix (assuming the language is 'fr')
 231     searchPath << prefix << wxFILE_SEP_PATH << lang << wxFILE_SEP_PATH
 232                          << "LC_MESSAGES" << wxPATH_SEP
 233                << prefix << wxFILE_SEP_PATH << lang << wxPATH_SEP
 234                << prefix << wxPATH_SEP;
 235
 236     return searchPath;
 237 }
 238
 239 // construct the search path for the given language
 240 static wxString GetFullSearchPath(const char *lang)
 241 {
 242     wxString searchPath;
 243
 244     // first take the entries explicitly added by the program
 245     size_t count = s_searchPrefixes.Count();
 246     for ( size_t n = 0; n < count; n++ )
 247     {
 248         searchPath << GetAllMsgCatalogSubdirs(s_searchPrefixes[n], lang)
 249                    << wxPATH_SEP;
 250     }
 251
 252     // then take the current directory
 253     // FIXME it should be the directory of the executable
 254     searchPath << GetAllMsgCatalogSubdirs(".", lang) << wxPATH_SEP;
 255
 256     // and finally add some standard ones
 257     searchPath
 258         << GetAllMsgCatalogSubdirs("/usr/share/locale", lang) << wxPATH_SEP
 259         << GetAllMsgCatalogSubdirs("/usr/lib/locale", lang) << wxPATH_SEP
 260         << GetAllMsgCatalogSubdirs("/usr/local/share/locale", lang);
 261
 262     return searchPath;
 263 }
 264
 265 // open disk file and read in it's contents
 266 bool wxMsgCatalog::Load(const char *szDirPrefix, const char *szName)
 267 {
 268   // FIXME VZ: I forgot the exact meaning of LC_PATH - anyone to remind me?
 269 #if 0
 270   const char *pszLcPath = getenv("LC_PATH");
 271   if ( pszLcPath != NULL )
 272       strPath += pszLcPath + wxString(szDirPrefix) + MSG_PATH;
 273 #endif // 0
 274
 275   wxString searchPath = GetFullSearchPath(szDirPrefix);
 276   const char *sublocale = strchr(szDirPrefix, '_');
 277   if ( sublocale )
 278   {
 279       // also add just base locale name: for things like "fr_BE" (belgium
 280       // french) we should use "fr" if no belgium specific message catalogs
 281       // exist
 282       searchPath << GetFullSearchPath(wxString(szDirPrefix).
 283                                       Left((size_t)(sublocale - szDirPrefix)))
 284                  << wxPATH_SEP;
 285   }
 286
 287   wxString strFile = szName;
 288   strFile += MSGCATALOG_EXTENSION;
 289
 290   // don't give translation errors here because the wxstd catalog might
 291   // not yet be loaded (and it's normal)
 292   //
 293   // (we're using an object because we have several return paths)
 294   NoTransErr noTransErr;
 295
 296   wxLogVerbose(_("looking for catalog '%s' in path '%s'."),
 297                szName, searchPath.c_str());
 298
 299   wxString strFullName;
 300   if ( !wxFindFileInPath(&strFullName, searchPath, strFile) ) {
 301     wxLogWarning(_("catalog file for domain '%s' not found."), szName);
 302     return FALSE;
 303   }
 304
 305   // open file
 306   wxLogVerbose(_("using catalog '%s' from '%s'."),
 307              szName, strFullName.c_str());
 308
 309   wxFile fileMsg(strFullName);
 310   if ( !fileMsg.IsOpened() )
 311     return FALSE;
 312
 313   // get the file size
 314   off_t nSize = fileMsg.Length();
 315   if ( nSize == wxInvalidOffset )
 316     return FALSE;
 317
 318   // read the whole file in memory
 319   m_pData = new size_t8[nSize];
 320   if ( fileMsg.Read(m_pData, nSize) != nSize ) {
 321     wxDELETEA(m_pData);
 322     return FALSE;
 323   }
 324
 325   // examine header
 326   bool bValid = (size_t)nSize > sizeof(wxMsgCatalogHeader);
 327
 328   wxMsgCatalogHeader *pHeader = (wxMsgCatalogHeader *)m_pData;
 329   if ( bValid ) {
 330     // we'll have to swap all the integers if it's true
 331     m_bSwapped = pHeader->magic == MSGCATALOG_MAGIC_SW;
 332
 333     // check the magic number
 334     bValid = m_bSwapped || pHeader->magic == MSGCATALOG_MAGIC;
 335   }
 336
 337   if ( !bValid ) {
 338     // it's either too short or has incorrect magic number
 339     wxLogWarning(_("'%s' is not a valid message catalog."), strFullName.c_str());
 340
 341     wxDELETEA(m_pData);
 342     return FALSE;
 343   }
 344
 345   // initialize
 346   m_numStrings  = Swap(pHeader->numStrings);
 347   m_pOrigTable  = (wxMsgTableEntry *)(m_pData +
 348                    Swap(pHeader->ofsOrigTable));
 349   m_pTransTable = (wxMsgTableEntry *)(m_pData +
 350                    Swap(pHeader->ofsTransTable));
 351
 352   m_nHashSize   = Swap(pHeader->nHashSize);
 353   m_pHashTable  = (size_t32 *)(m_pData + Swap(pHeader->ofsHashTable));
 354
 355   m_pszName = new char[strlen(szName) + 1];
 356   strcpy(m_pszName, szName);
 357
 358   // everything is fine
 359   return TRUE;
 360 }
 361
 362 // search for a string
 363 const char *wxMsgCatalog::GetString(const char *szOrig) const
 364 {
 365   if ( szOrig == NULL )
 366     return NULL;
 367
 368   if ( HasHashTable() ) {   // use hash table for lookup if possible
 369     size_t32 nHashVal = GetHash(szOrig);
 370     size_t32 nIndex   = nHashVal % m_nHashSize;
 371
 372     size_t32 nIncr = 1 + (nHashVal % (m_nHashSize - 2));
 373
 374     while ( TRUE ) {
 375       size_t32 nStr = Swap(m_pHashTable[nIndex]);
 376       if ( nStr == 0 )
 377         return NULL;
 378
 379       if ( strcmp(szOrig, StringAtOfs(m_pOrigTable, nStr - 1)) == 0 )
 380         return StringAtOfs(m_pTransTable, nStr - 1);
 381
 382       if ( nIndex >= m_nHashSize - nIncr)
 383         nIndex -= m_nHashSize - nIncr;
 384       else
 385         nIndex += nIncr;
 386     }
 387   }
 388   else {                    // no hash table: use default binary search
 389     size_t32 bottom = 0,
 390            top    = m_numStrings,
 391            current;
 392     while ( bottom < top ) {
 393       current = (bottom + top) / 2;
 394       int res = strcmp(szOrig, StringAtOfs(m_pOrigTable, current));
 395       if ( res < 0 )
 396         top = current;
 397       else if ( res > 0 )
 398         bottom = current + 1;
 399       else    // found!
 400         return StringAtOfs(m_pTransTable, current);
 401     }
 402   }
 403
 404   // not found
 405   return NULL;
 406 }
 407
 408 // ----------------------------------------------------------------------------
 409 // wxLocale
 410 // ----------------------------------------------------------------------------
 411
 412 wxLocale::wxLocale()
 413 {
 414   m_pszOldLocale = NULL;
 415   m_pMsgCat = NULL;
 416 }
 417
 418 // NB: this function has (desired) side effect of changing current locale
 419 bool wxLocale::Init(const char *szName,
 420                     const char *szShort,
 421                     const char *szLocale,
 422                     bool        bLoadDefault)
 423 {
 424   m_strLocale = szName;
 425   m_strShort = szShort;
 426
 427   // change current locale (default: same as long name)
 428   if ( szLocale == NULL )
 429     szLocale = szName;
 430   m_pszOldLocale = setlocale(LC_ALL, szLocale);
 431   if ( m_pszOldLocale == NULL )
 432     wxLogError(_("locale '%s' can not be set."), szLocale);
 433
 434   // the short name will be used to look for catalog files as well,
 435   // so we need something here
 436   if ( m_strShort.IsEmpty() ) {
 437     // FIXME I don't know how these 2 letter abbreviations are formed,
 438     //       this wild guess is surely wrong
 439     m_strShort = tolower(szLocale[0]) + tolower(szLocale[1]);
 440   }
 441
 442   // save the old locale to be able to restore it later
 443   m_pOldLocale = wxSetLocale(this);
 444
 445   // load the default catalog with wxWindows standard messages
 446   m_pMsgCat = NULL;
 447   bool bOk = TRUE;
 448   if ( bLoadDefault )
 449     bOk = AddCatalog("wxstd");
 450
 451   return bOk;
 452 }
 453
 454 void wxLocale::AddCatalogLookupPathPrefix(const wxString& prefix)
 455 {
 456     if ( s_searchPrefixes.Index(prefix) == wxNOT_FOUND )
 457     {
 458         s_searchPrefixes.Add(prefix);
 459     }
 460     //else: already have it
 461 }
 462
 463 // clean up
 464 wxLocale::~wxLocale()
 465 {
 466     // free memory
 467     wxMsgCatalog *pTmpCat;
 468     while ( m_pMsgCat != NULL ) {
 469         pTmpCat = m_pMsgCat;
 470         m_pMsgCat = m_pMsgCat->m_pNext;
 471         delete pTmpCat;
 472     }
 473
 474     // restore old locale
 475     wxSetLocale(m_pOldLocale);
 476     setlocale(LC_ALL, m_pszOldLocale);
 477 }
 478
 479 // get the translation of given string in current locale
 480 const char *wxLocale::GetString(const char *szOrigString,
 481                                 const char *szDomain) const
 482 {
 483   if ( IsEmpty(szOrigString) )
 484       return szDomain;
 485
 486   const char *pszTrans = NULL;
 487
 488   wxMsgCatalog *pMsgCat;
 489   if ( szDomain != NULL ) {
 490     pMsgCat = FindCatalog(szDomain);
 491
 492     // does the catalog exist?
 493     if ( pMsgCat != NULL )
 494       pszTrans = pMsgCat->GetString(szOrigString);
 495   }
 496   else {
 497     // search in all domains
 498     for ( pMsgCat = m_pMsgCat; pMsgCat != NULL; pMsgCat = pMsgCat->m_pNext ) {
 499       pszTrans = pMsgCat->GetString(szOrigString);
 500       if ( pszTrans != NULL )   // take the first found
 501         break;
 502     }
 503   }
 504
 505   if ( pszTrans == NULL ) {
 506     if ( wxIsLoggingTransErrors() ) {
 507       // suppress further error messages if we're not debugging: this avoids
 508       // flooding the user with messages about each and every missing string if,
 509       // for example, a whole catalog file is missing.
 510
 511       // do it before calling LogWarning to prevent infinite recursion!
 512 #ifdef __WXDEBUG__
 513       NoTransErr noTransErr;
 514 #else // !debug
 515       wxSuppressTransErrors();
 516 #endif // debug/!debug
 517
 518       if ( szDomain != NULL )
 519       {
 520         wxLogWarning(_("string '%s' not found in domain '%s' for locale '%s'."),
 521                      szOrigString, szDomain, m_strLocale.c_str());
 522       }
 523       else
 524       {
 525         wxLogWarning(_("string '%s' not found in locale '%s'."),
 526                      szOrigString, m_strLocale.c_str());
 527       }
 528     }
 529
 530     return szOrigString;
 531   }
 532   else
 533     return pszTrans;
 534 }
 535
 536 // find catalog by name in a linked list, return NULL if !found
 537 wxMsgCatalog *wxLocale::FindCatalog(const char *szDomain) const
 538 {
 539 // linear search in the linked list
 540   wxMsgCatalog *pMsgCat;
 541   for ( pMsgCat = m_pMsgCat; pMsgCat != NULL; pMsgCat = pMsgCat->m_pNext ) {
 542     if ( Stricmp(pMsgCat->GetName(), szDomain) == 0 )
 543       return pMsgCat;
 544   }
 545
 546   return NULL;
 547 }
 548
 549 // check if the given catalog is loaded
 550 bool wxLocale::IsLoaded(const char *szDomain) const
 551 {
 552   return FindCatalog(szDomain) != NULL;
 553 }
 554
 555 // add a catalog to our linked list
 556 bool wxLocale::AddCatalog(const char *szDomain)
 557 {
 558   wxMsgCatalog *pMsgCat = new wxMsgCatalog;
 559
 560   if ( pMsgCat->Load(m_strShort, szDomain) ) {
 561     // add it to the head of the list so that in GetString it will
 562     // be searched before the catalogs added earlier
 563     pMsgCat->m_pNext = m_pMsgCat;
 564     m_pMsgCat = pMsgCat;
 565
 566     return TRUE;
 567   }
 568   else {
 569     // don't add it because it couldn't be loaded anyway
 570     delete pMsgCat;
 571
 572     return FALSE;
 573   }
 574 }
 575
 576 // ----------------------------------------------------------------------------
 577 // global functions and variables
 578 // ----------------------------------------------------------------------------
 579
 580 // translation errors logging
 581 // --------------------------
 582
 583 static bool gs_bGiveTransErrors = TRUE;
 584
 585 void wxSuppressTransErrors()
 586 {
 587   gs_bGiveTransErrors = FALSE;
 588 }
 589
 590 void wxRestoreTransErrors()
 591 {
 592   gs_bGiveTransErrors = TRUE;
 593 }
 594
 595 bool wxIsLoggingTransErrors()
 596 {
 597   return gs_bGiveTransErrors;
 598 }
 599
 600 // retrieve/change current locale
 601 // ------------------------------
 602
 603 // the current locale object
 604 static wxLocale *g_pLocale = NULL;
 605
 606 wxLocale *wxGetLocale()
 607 {
 608   return g_pLocale;
 609 }
 610
 611 wxLocale *wxSetLocale(wxLocale *pLocale)
 612 {
 613   wxLocale *pOld = g_pLocale;
 614   g_pLocale = pLocale;
 615   return pOld;
 616 }