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 #if wxUSE_INTL
  32
  33 // standard headers
  34 #include  <locale.h>
  35 #include  <ctype.h>
  36
  37 // wxWindows
  38 #include "wx/defs.h"
  39 #include "wx/string.h"
  40 #include "wx/intl.h"
  41 #include "wx/file.h"
  42 #include "wx/log.h"
  43 #include "wx/debug.h"
  44 #include "wx/utils.h"
  45
  46 #include <stdlib.h>
  47
  48 // ----------------------------------------------------------------------------
  49 // simple types
  50 // ----------------------------------------------------------------------------
  51
  52 // this should *not* be wxChar, this type must have exactly 8 bits!
  53 typedef unsigned char size_t8;
  54
  55 #ifdef __WXMSW__
  56     #if defined(__WIN16__)
  57         typedef unsigned long size_t32;
  58     #elif defined(__WIN32__)
  59         typedef unsigned int size_t32;
  60     #else
  61         // Win64 will have different type sizes
  62         #error "Please define a 32 bit type"
  63     #endif
  64 #else // !Windows
  65     // SIZEOF_XXX are defined by configure
  66     #if defined(SIZEOF_INT) && (SIZEOF_INT == 4)
  67         typedef unsigned int size_t32;
  68     #elif defined(SIZEOF_LONG) && (SIZEOF_LONG == 4)
  69         typedef unsigned long size_t32;
  70     #else
  71         // assume sizeof(int) == 4 - what else can we do
  72         typedef unsigned int size_t32;
  73
  74         // ... but at least check it during run time
  75         static class IntSizeChecker
  76         {
  77         public:
  78             IntSizeChecker()
  79             {
  80                 // Asserting a sizeof directly causes some compilers to
  81                 // issue a "using constant in a conditional expression" warning
  82                 size_t                   intsize = sizeof(int);
  83
  84                 wxASSERT_MSG( intsize == 4,
  85                               "size_t32 is incorrectly defined!" );
  86             }
  87         } intsizechecker;
  88     #endif
  89 #endif // Win/!Win
  90
  91 // ----------------------------------------------------------------------------
  92 // constants
  93 // ----------------------------------------------------------------------------
  94
  95 // magic number identifying the .mo format file
  96 const size_t32 MSGCATALOG_MAGIC    = 0x950412de;
  97 const size_t32 MSGCATALOG_MAGIC_SW = 0xde120495;
  98
  99 // extension of ".mo" files
 100 #define MSGCATALOG_EXTENSION  ".mo"
 101
 102 // ----------------------------------------------------------------------------
 103 // global functions
 104 // ----------------------------------------------------------------------------
 105
 106 // suppress further error messages about missing translations
 107 // (if you don't have one catalog file, you wouldn't like to see the
 108 //  error message for each string in it, so normally it's given only
 109 //  once)
 110 void wxSuppressTransErrors();
 111
 112 // restore the logging
 113 void wxRestoreTransErrors();
 114
 115 // get the current state
 116 bool wxIsLoggingTransErrors();
 117
 118 static wxLocale *wxSetLocale(wxLocale *pLocale);
 119
 120 // ----------------------------------------------------------------------------
 121 // wxMsgCatalog corresponds to one disk-file message catalog.
 122 //
 123 // This is a "low-level" class and is used only by wxLocale (that's why
 124 // it's designed to be stored in a linked list)
 125 // ----------------------------------------------------------------------------
 126
 127 class wxMsgCatalog
 128 {
 129 public:
 130   // ctor & dtor
 131   wxMsgCatalog();
 132  ~wxMsgCatalog();
 133
 134   // load the catalog from disk (szDirPrefix corresponds to language)
 135   bool Load(const wxChar *szDirPrefix, const wxChar *szName);
 136   bool IsLoaded() const { return m_pData != NULL; }
 137
 138   // get name of the catalog
 139   const wxChar *GetName() const { return m_pszName; }
 140
 141   // get the translated string: returns NULL if not found
 142   const char *GetString(const char *sz) const;
 143
 144   // public variable pointing to the next element in a linked list (or NULL)
 145   wxMsgCatalog *m_pNext;
 146
 147 private:
 148   // this implementation is binary compatible with GNU gettext() version 0.10
 149
 150   // an entry in the string table
 151   struct wxMsgTableEntry
 152   {
 153     size_t32   nLen;           // length of the string
 154     size_t32   ofsString;      // pointer to the string
 155   };
 156
 157   // header of a .mo file
 158   struct wxMsgCatalogHeader
 159   {
 160     size_t32  magic,          // offset +00:  magic id
 161               revision,       //        +04:  revision
 162               numStrings;     //        +08:  number of strings in the file
 163     size_t32  ofsOrigTable,   //        +0C:  start of original string table
 164               ofsTransTable;  //        +10:  start of translated string table
 165     size_t32  nHashSize,      //        +14:  hash table size
 166               ofsHashTable;   //        +18:  offset of hash table start
 167   };
 168
 169   // all data is stored here, NULL if no data loaded
 170   size_t8 *m_pData;
 171
 172   // data description
 173   size_t32          m_numStrings,   // number of strings in this domain
 174                     m_nHashSize;    // number of entries in hash table
 175   size_t32         *m_pHashTable;   // pointer to hash table
 176   wxMsgTableEntry  *m_pOrigTable,   // pointer to original   strings
 177                    *m_pTransTable;  //            translated
 178
 179   const char *StringAtOfs(wxMsgTableEntry *pTable, size_t32 index) const
 180     { return (const char *)(m_pData + Swap(pTable[index].ofsString)); }
 181
 182   // utility functions
 183     // calculate the hash value of given string
 184   static inline size_t32 GetHash(const char *sz);
 185     // big<->little endian
 186   inline size_t32 Swap(size_t32 ui) const;
 187
 188   // internal state
 189   bool HasHashTable() const // true if hash table is present
 190     { return m_nHashSize > 2 && m_pHashTable != NULL; }
 191
 192   bool          m_bSwapped;   // wrong endianness?
 193
 194   wxChar       *m_pszName;    // name of the domain
 195 };
 196
 197 // ----------------------------------------------------------------------------
 198 // global variables
 199 // ----------------------------------------------------------------------------
 200
 201 // the list of the directories to search for message catalog files
 202 static wxArrayString s_searchPrefixes;
 203
 204 // ============================================================================
 205 // implementation
 206 // ============================================================================
 207
 208 // ----------------------------------------------------------------------------
 209 // wxMsgCatalog class
 210 // ----------------------------------------------------------------------------
 211
 212 // calculate hash value using the so called hashpjw function by P.J. Weinberger
 213 // [see Aho/Sethi/Ullman, COMPILERS: Principles, Techniques and Tools]
 214 size_t32 wxMsgCatalog::GetHash(const char *sz)
 215 {
 216   #define HASHWORDBITS 32     // the length of size_t32
 217
 218   size_t32 hval = 0;
 219   size_t32 g;
 220   while ( *sz != '\0' ) {
 221     hval <<= 4;
 222     hval += (size_t32)*sz++;
 223     g = hval & ((size_t32)0xf << (HASHWORDBITS - 4));
 224     if ( g != 0 ) {
 225       hval ^= g >> (HASHWORDBITS - 8);
 226       hval ^= g;
 227     }
 228   }
 229
 230   return hval;
 231 }
 232
 233 // swap the 2 halves of 32 bit integer if needed
 234 size_t32 wxMsgCatalog::Swap(size_t32 ui) const
 235 {
 236   return m_bSwapped ? (ui << 24) | ((ui & 0xff00) << 8) |
 237                       ((ui >> 8) & 0xff00) | (ui >> 24)
 238                     : ui;
 239 }
 240
 241 wxMsgCatalog::wxMsgCatalog()
 242 {
 243   m_pData   = NULL;
 244   m_pszName = NULL;
 245 }
 246
 247 wxMsgCatalog::~wxMsgCatalog()
 248 {
 249   wxDELETEA(m_pData);
 250   wxDELETEA(m_pszName);
 251 }
 252
 253 // small class to suppress the translation erros until exit from current scope
 254 class NoTransErr
 255 {
 256 public:
 257     NoTransErr() { wxSuppressTransErrors(); }
 258    ~NoTransErr() { wxRestoreTransErrors();  }
 259 };
 260
 261 // return all directories to search for given prefix
 262 static wxString GetAllMsgCatalogSubdirs(const wxChar *prefix,
 263                                         const wxChar *lang)
 264 {
 265     wxString searchPath;
 266
 267     // search first in prefix/fr/LC_MESSAGES, then in prefix/fr and finally in
 268     // prefix (assuming the language is 'fr')
 269     searchPath << prefix << wxFILE_SEP_PATH << lang << wxFILE_SEP_PATH
 270                          << wxT("LC_MESSAGES") << wxPATH_SEP
 271                << prefix << wxFILE_SEP_PATH << lang << wxPATH_SEP
 272                << prefix << wxPATH_SEP;
 273
 274     return searchPath;
 275 }
 276
 277 // construct the search path for the given language
 278 static wxString GetFullSearchPath(const wxChar *lang)
 279 {
 280     wxString searchPath;
 281
 282     // first take the entries explicitly added by the program
 283     size_t count = s_searchPrefixes.Count();
 284     for ( size_t n = 0; n < count; n++ )
 285     {
 286         searchPath << GetAllMsgCatalogSubdirs(s_searchPrefixes[n], lang)
 287                    << wxPATH_SEP;
 288     }
 289
 290     // then take the current directory
 291     // FIXME it should be the directory of the executable
 292     searchPath << GetAllMsgCatalogSubdirs(wxT("."), lang) << wxPATH_SEP;
 293
 294     // and finally add some standard ones
 295     searchPath
 296         << GetAllMsgCatalogSubdirs(wxT("/usr/share/locale"), lang) << wxPATH_SEP
 297         << GetAllMsgCatalogSubdirs(wxT("/usr/lib/locale"), lang) << wxPATH_SEP
 298         << GetAllMsgCatalogSubdirs(wxT("/usr/local/share/locale"), lang);
 299
 300     return searchPath;
 301 }
 302
 303 // open disk file and read in it's contents
 304 bool wxMsgCatalog::Load(const wxChar *szDirPrefix, const wxChar *szName0)
 305 {
 306    /* We need to handle locales like  de_AT.iso-8859-1
 307       For this we first chop off the .CHARSET specifier and ignore it.
 308       FIXME: UNICODE SUPPORT: must use CHARSET specifier!
 309    */
 310    wxString szName = szName0;
 311    if(szName.Find('.') != -1) // contains a dot
 312       szName = szName.Left(szName.Find('.'));
 313
 314   // FIXME VZ: I forgot the exact meaning of LC_PATH - anyone to remind me?
 315   // KB: search path where to find the mo files, probably : delimited
 316 #if 0
 317   const wxChar *pszLcPath = wxGetenv("LC_PATH");
 318   if ( pszLcPath != NULL )
 319       strPath += pszLcPath + wxString(szDirPrefix) + MSG_PATH;
 320 #endif // 0
 321
 322   wxString searchPath = GetFullSearchPath(szDirPrefix);
 323   const wxChar *sublocale = wxStrchr(szDirPrefix, wxT('_'));
 324   if ( sublocale )
 325   {
 326       // also add just base locale name: for things like "fr_BE" (belgium
 327       // french) we should use "fr" if no belgium specific message catalogs
 328       // exist
 329       searchPath << GetFullSearchPath(wxString(szDirPrefix).
 330                                       Left((size_t)(sublocale - szDirPrefix)))
 331                  << wxPATH_SEP;
 332   }
 333
 334   wxString strFile = szName;
 335   strFile += MSGCATALOG_EXTENSION;
 336
 337   // don't give translation errors here because the wxstd catalog might
 338   // not yet be loaded (and it's normal)
 339   //
 340   // (we're using an object because we have several return paths)
 341
 342   NoTransErr noTransErr;
 343   wxLogVerbose(wxT("looking for catalog '%s' in path '%s'."),
 344                szName.c_str(), searchPath.c_str());
 345
 346   wxString strFullName;
 347   if ( !wxFindFileInPath(&strFullName, searchPath, strFile) ) {
 348     wxLogWarning(_("catalog file for domain '%s' not found."), szName.c_str());
 349     return FALSE;
 350   }
 351
 352   // open file
 353   wxLogVerbose(_("using catalog '%s' from '%s'."),
 354              szName.c_str(), strFullName.c_str());
 355
 356   wxFile fileMsg(strFullName);
 357   if ( !fileMsg.IsOpened() )
 358     return FALSE;
 359
 360   // get the file size
 361   off_t nSize = fileMsg.Length();
 362   if ( nSize == wxInvalidOffset )
 363     return FALSE;
 364
 365   // read the whole file in memory
 366   m_pData = new size_t8[nSize];
 367   if ( fileMsg.Read(m_pData, nSize) != nSize ) {
 368     wxDELETEA(m_pData);
 369     return FALSE;
 370   }
 371
 372   // examine header
 373   bool bValid = (size_t)nSize > sizeof(wxMsgCatalogHeader);
 374
 375   wxMsgCatalogHeader *pHeader = (wxMsgCatalogHeader *)m_pData;
 376   if ( bValid ) {
 377     // we'll have to swap all the integers if it's true
 378     m_bSwapped = pHeader->magic == MSGCATALOG_MAGIC_SW;
 379
 380     // check the magic number
 381     bValid = m_bSwapped || pHeader->magic == MSGCATALOG_MAGIC;
 382   }
 383
 384   if ( !bValid ) {
 385     // it's either too short or has incorrect magic number
 386     wxLogWarning(_("'%s' is not a valid message catalog."), strFullName.c_str());
 387
 388     wxDELETEA(m_pData);
 389     return FALSE;
 390   }
 391
 392   // initialize
 393   m_numStrings  = Swap(pHeader->numStrings);
 394   m_pOrigTable  = (wxMsgTableEntry *)(m_pData +
 395                    Swap(pHeader->ofsOrigTable));
 396   m_pTransTable = (wxMsgTableEntry *)(m_pData +
 397                    Swap(pHeader->ofsTransTable));
 398
 399   m_nHashSize   = Swap(pHeader->nHashSize);
 400   m_pHashTable  = (size_t32 *)(m_pData + Swap(pHeader->ofsHashTable));
 401
 402   m_pszName = new wxChar[wxStrlen(szName) + 1];
 403   wxStrcpy(m_pszName, szName);
 404
 405   // everything is fine
 406   return TRUE;
 407 }
 408
 409 // search for a string
 410 const char *wxMsgCatalog::GetString(const char *szOrig) const
 411 {
 412   if ( szOrig == NULL )
 413     return NULL;
 414
 415   if ( HasHashTable() ) {   // use hash table for lookup if possible
 416     size_t32 nHashVal = GetHash(szOrig);
 417     size_t32 nIndex   = nHashVal % m_nHashSize;
 418
 419     size_t32 nIncr = 1 + (nHashVal % (m_nHashSize - 2));
 420
 421 #if defined(__VISAGECPP__)
 422 // VA just can't stand while(1) or while(TRUE)
 423     bool bOs2var = TRUE;
 424     while(bOs2var) {
 425 #else
 426     while (1) {
 427 #endif
 428       size_t32 nStr = Swap(m_pHashTable[nIndex]);
 429       if ( nStr == 0 )
 430         return NULL;
 431
 432       if ( strcmp(szOrig, StringAtOfs(m_pOrigTable, nStr - 1)) == 0 )
 433         return StringAtOfs(m_pTransTable, nStr - 1);
 434
 435       if ( nIndex >= m_nHashSize - nIncr)
 436         nIndex -= m_nHashSize - nIncr;
 437       else
 438         nIndex += nIncr;
 439     }
 440   }
 441   else {                    // no hash table: use default binary search
 442     size_t32 bottom = 0,
 443            top    = m_numStrings,
 444            current;
 445     while ( bottom < top ) {
 446       current = (bottom + top) / 2;
 447       int res = strcmp(szOrig, StringAtOfs(m_pOrigTable, current));
 448       if ( res < 0 )
 449         top = current;
 450       else if ( res > 0 )
 451         bottom = current + 1;
 452       else    // found!
 453         return StringAtOfs(m_pTransTable, current);
 454     }
 455   }
 456
 457   // not found
 458   return NULL;
 459 }
 460
 461 // ----------------------------------------------------------------------------
 462 // wxLocale
 463 // ----------------------------------------------------------------------------
 464
 465 wxLocale::wxLocale()
 466 {
 467   m_pszOldLocale = NULL;
 468   m_pMsgCat = NULL;
 469 }
 470
 471 // NB: this function has (desired) side effect of changing current locale
 472 bool wxLocale::Init(const wxChar *szName,
 473                     const wxChar *szShort,
 474                     const wxChar *szLocale,
 475                     bool        bLoadDefault)
 476 {
 477   m_strLocale = szName;
 478   m_strShort = szShort;
 479
 480   // change current locale (default: same as long name)
 481   if ( szLocale == NULL )
 482   {
 483     // the argument to setlocale()
 484     szLocale = szShort;
 485   }
 486   m_pszOldLocale = wxSetlocale(LC_ALL, szLocale);
 487   if ( m_pszOldLocale == NULL )
 488     wxLogError(_("locale '%s' can not be set."), szLocale);
 489
 490   // the short name will be used to look for catalog files as well,
 491   // so we need something here
 492   if ( m_strShort.IsEmpty() ) {
 493     // FIXME I don't know how these 2 letter abbreviations are formed,
 494     //       this wild guess is surely wrong
 495     m_strShort = tolower(szLocale[0]) + tolower(szLocale[1]);
 496   }
 497
 498   // save the old locale to be able to restore it later
 499   m_pOldLocale = wxSetLocale(this);
 500
 501   // load the default catalog with wxWindows standard messages
 502   m_pMsgCat = NULL;
 503   bool bOk = TRUE;
 504   if ( bLoadDefault )
 505     bOk = AddCatalog(wxT("wxstd"));
 506
 507   return bOk;
 508 }
 509
 510 void wxLocale::AddCatalogLookupPathPrefix(const wxString& prefix)
 511 {
 512     if ( s_searchPrefixes.Index(prefix) == wxNOT_FOUND )
 513     {
 514         s_searchPrefixes.Add(prefix);
 515     }
 516     //else: already have it
 517 }
 518
 519 // clean up
 520 wxLocale::~wxLocale()
 521 {
 522     // free memory
 523     wxMsgCatalog *pTmpCat;
 524     while ( m_pMsgCat != NULL ) {
 525         pTmpCat = m_pMsgCat;
 526         m_pMsgCat = m_pMsgCat->m_pNext;
 527         delete pTmpCat;
 528     }
 529
 530     // restore old locale
 531     wxSetLocale(m_pOldLocale);
 532     wxSetlocale(LC_ALL, m_pszOldLocale);
 533 }
 534
 535 // get the translation of given string in current locale
 536 const wxMB2WXbuf wxLocale::GetString(const wxChar *szOrigString,
 537                                      const wxChar *szDomain) const
 538 {
 539   if ( wxIsEmpty(szOrigString) )
 540       return szDomain;
 541
 542   const char *pszTrans = NULL;
 543 #if wxUSE_UNICODE
 544   const wxWX2MBbuf szOrgString = wxConvCurrent->cWX2MB(szOrigString);
 545 #else // ANSI
 546   #define szOrgString szOrigString
 547 #endif // Unicode/ANSI
 548
 549   wxMsgCatalog *pMsgCat;
 550   if ( szDomain != NULL ) {
 551     pMsgCat = FindCatalog(szDomain);
 552
 553     // does the catalog exist?
 554     if ( pMsgCat != NULL )
 555       pszTrans = pMsgCat->GetString(szOrgString);
 556   }
 557   else {
 558     // search in all domains
 559     for ( pMsgCat = m_pMsgCat; pMsgCat != NULL; pMsgCat = pMsgCat->m_pNext ) {
 560       pszTrans = pMsgCat->GetString(szOrgString);
 561       if ( pszTrans != NULL )   // take the first found
 562         break;
 563     }
 564   }
 565
 566   if ( pszTrans == NULL ) {
 567     if ( wxIsLoggingTransErrors() ) {
 568       // suppress further error messages if we're not debugging: this avoids
 569       // flooding the user with messages about each and every missing string if,
 570       // for example, a whole catalog file is missing.
 571
 572       // do it before calling LogWarning to prevent infinite recursion!
 573 #ifdef __WXDEBUG__
 574       NoTransErr noTransErr;
 575 #else // !debug
 576       wxSuppressTransErrors();
 577 #endif // debug/!debug
 578
 579       if ( szDomain != NULL )
 580       {
 581         wxLogWarning(_("string '%s' not found in domain '%s' for locale '%s'."),
 582                      szOrigString, szDomain, m_strLocale.c_str());
 583       }
 584       else
 585       {
 586         wxLogWarning(_("string '%s' not found in locale '%s'."),
 587                      szOrigString, m_strLocale.c_str());
 588       }
 589     }
 590
 591     return (wxMB2WXbuf)(szOrigString);
 592   }
 593   else
 594   {
 595     return wxConvertMB2WX(pszTrans); // or preferably wxCSConv(charset).cMB2WX(pszTrans) or something,
 596                                      // a macro similar to wxConvertMB2WX could be written for that
 597   }
 598
 599   #undef szOrgString
 600 }
 601
 602 // find catalog by name in a linked list, return NULL if !found
 603 wxMsgCatalog *wxLocale::FindCatalog(const wxChar *szDomain) const
 604 {
 605 // linear search in the linked list
 606   wxMsgCatalog *pMsgCat;
 607   for ( pMsgCat = m_pMsgCat; pMsgCat != NULL; pMsgCat = pMsgCat->m_pNext ) {
 608     if ( wxStricmp(pMsgCat->GetName(), szDomain) == 0 )
 609       return pMsgCat;
 610   }
 611
 612   return NULL;
 613 }
 614
 615 // check if the given catalog is loaded
 616 bool wxLocale::IsLoaded(const wxChar *szDomain) const
 617 {
 618   return FindCatalog(szDomain) != NULL;
 619 }
 620
 621 // add a catalog to our linked list
 622 bool wxLocale::AddCatalog(const wxChar *szDomain)
 623 {
 624   wxMsgCatalog *pMsgCat = new wxMsgCatalog;
 625
 626   if ( pMsgCat->Load(m_strShort, szDomain) ) {
 627     // add it to the head of the list so that in GetString it will
 628     // be searched before the catalogs added earlier
 629     pMsgCat->m_pNext = m_pMsgCat;
 630     m_pMsgCat = pMsgCat;
 631
 632     return TRUE;
 633   }
 634   else {
 635     // don't add it because it couldn't be loaded anyway
 636     delete pMsgCat;
 637
 638     return FALSE;
 639   }
 640 }
 641
 642 // ----------------------------------------------------------------------------
 643 // global functions and variables
 644 // ----------------------------------------------------------------------------
 645
 646 // translation errors logging
 647 // --------------------------
 648
 649 static bool gs_bGiveTransErrors = TRUE;
 650
 651 void wxSuppressTransErrors()
 652 {
 653   gs_bGiveTransErrors = FALSE;
 654 }
 655
 656 void wxRestoreTransErrors()
 657 {
 658   gs_bGiveTransErrors = TRUE;
 659 }
 660
 661 bool wxIsLoggingTransErrors()
 662 {
 663   return gs_bGiveTransErrors;
 664 }
 665
 666 // retrieve/change current locale
 667 // ------------------------------
 668
 669 // the current locale object
 670 static wxLocale *g_pLocale = NULL;
 671
 672 wxLocale *wxGetLocale()
 673 {
 674   return g_pLocale;
 675 }
 676
 677 wxLocale *wxSetLocale(wxLocale *pLocale)
 678 {
 679   wxLocale *pOld = g_pLocale;
 680   g_pLocale = pLocale;
 681   return pOld;
 682 }
 683
 684 #endif // wxUSE_INTL
 685