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