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     while ( TRUE ) {
 422       size_t32 nStr = Swap(m_pHashTable[nIndex]);
 423       if ( nStr == 0 )
 424         return NULL;
 425
 426       if ( strcmp(szOrig, StringAtOfs(m_pOrigTable, nStr - 1)) == 0 )
 427         return StringAtOfs(m_pTransTable, nStr - 1);
 428
 429       if ( nIndex >= m_nHashSize - nIncr)
 430         nIndex -= m_nHashSize - nIncr;
 431       else
 432         nIndex += nIncr;
 433     }
 434   }
 435   else {                    // no hash table: use default binary search
 436     size_t32 bottom = 0,
 437            top    = m_numStrings,
 438            current;
 439     while ( bottom < top ) {
 440       current = (bottom + top) / 2;
 441       int res = strcmp(szOrig, StringAtOfs(m_pOrigTable, current));
 442       if ( res < 0 )
 443         top = current;
 444       else if ( res > 0 )
 445         bottom = current + 1;
 446       else    // found!
 447         return StringAtOfs(m_pTransTable, current);
 448     }
 449   }
 450
 451   // not found
 452   return NULL;
 453 }
 454
 455 // ----------------------------------------------------------------------------
 456 // wxLocale
 457 // ----------------------------------------------------------------------------
 458
 459 wxLocale::wxLocale()
 460 {
 461   m_pszOldLocale = NULL;
 462   m_pMsgCat = NULL;
 463 }
 464
 465 // NB: this function has (desired) side effect of changing current locale
 466 bool wxLocale::Init(const wxChar *szName,
 467                     const wxChar *szShort,
 468                     const wxChar *szLocale,
 469                     bool        bLoadDefault)
 470 {
 471   m_strLocale = szName;
 472   m_strShort = szShort;
 473
 474   // change current locale (default: same as long name)
 475   if ( szLocale == NULL )
 476   {
 477     // the argument to setlocale()
 478     szLocale = szShort;
 479   }
 480   m_pszOldLocale = wxSetlocale(LC_ALL, szLocale);
 481   if ( m_pszOldLocale == NULL )
 482     wxLogError(_("locale '%s' can not be set."), szLocale);
 483
 484   // the short name will be used to look for catalog files as well,
 485   // so we need something here
 486   if ( m_strShort.IsEmpty() ) {
 487     // FIXME I don't know how these 2 letter abbreviations are formed,
 488     //       this wild guess is surely wrong
 489     m_strShort = tolower(szLocale[0]) + tolower(szLocale[1]);
 490   }
 491
 492   // save the old locale to be able to restore it later
 493   m_pOldLocale = wxSetLocale(this);
 494
 495   // load the default catalog with wxWindows standard messages
 496   m_pMsgCat = NULL;
 497   bool bOk = TRUE;
 498   if ( bLoadDefault )
 499     bOk = AddCatalog(wxT("wxstd"));
 500
 501   return bOk;
 502 }
 503
 504 void wxLocale::AddCatalogLookupPathPrefix(const wxString& prefix)
 505 {
 506     if ( s_searchPrefixes.Index(prefix) == wxNOT_FOUND )
 507     {
 508         s_searchPrefixes.Add(prefix);
 509     }
 510     //else: already have it
 511 }
 512
 513 // clean up
 514 wxLocale::~wxLocale()
 515 {
 516     // free memory
 517     wxMsgCatalog *pTmpCat;
 518     while ( m_pMsgCat != NULL ) {
 519         pTmpCat = m_pMsgCat;
 520         m_pMsgCat = m_pMsgCat->m_pNext;
 521         delete pTmpCat;
 522     }
 523
 524     // restore old locale
 525     wxSetLocale(m_pOldLocale);
 526     wxSetlocale(LC_ALL, m_pszOldLocale);
 527 }
 528
 529 // get the translation of given string in current locale
 530 const wxMB2WXbuf wxLocale::GetString(const wxChar *szOrigString,
 531                                      const wxChar *szDomain) const
 532 {
 533   if ( wxIsEmpty(szOrigString) )
 534       return szDomain;
 535
 536   const char *pszTrans = NULL;
 537 #if wxUSE_UNICODE
 538   const wxWX2MBbuf szOrgString = wxConvCurrent->cWX2MB(szOrigString);
 539 #else // ANSI
 540   #define szOrgString szOrigString
 541 #endif // Unicode/ANSI
 542
 543   wxMsgCatalog *pMsgCat;
 544   if ( szDomain != NULL ) {
 545     pMsgCat = FindCatalog(szDomain);
 546
 547     // does the catalog exist?
 548     if ( pMsgCat != NULL )
 549       pszTrans = pMsgCat->GetString(szOrgString);
 550   }
 551   else {
 552     // search in all domains
 553     for ( pMsgCat = m_pMsgCat; pMsgCat != NULL; pMsgCat = pMsgCat->m_pNext ) {
 554       pszTrans = pMsgCat->GetString(szOrgString);
 555       if ( pszTrans != NULL )   // take the first found
 556         break;
 557     }
 558   }
 559
 560   if ( pszTrans == NULL ) {
 561     if ( wxIsLoggingTransErrors() ) {
 562       // suppress further error messages if we're not debugging: this avoids
 563       // flooding the user with messages about each and every missing string if,
 564       // for example, a whole catalog file is missing.
 565
 566       // do it before calling LogWarning to prevent infinite recursion!
 567 #ifdef __WXDEBUG__
 568       NoTransErr noTransErr;
 569 #else // !debug
 570       wxSuppressTransErrors();
 571 #endif // debug/!debug
 572
 573       if ( szDomain != NULL )
 574       {
 575         wxLogWarning(_("string '%s' not found in domain '%s' for locale '%s'."),
 576                      szOrigString, szDomain, m_strLocale.c_str());
 577       }
 578       else
 579       {
 580         wxLogWarning(_("string '%s' not found in locale '%s'."),
 581                      szOrigString, m_strLocale.c_str());
 582       }
 583     }
 584
 585     return (wxMB2WXbuf)(szOrigString);
 586   }
 587   else
 588   {
 589     // FIXME it was
 590     // return (wxMB2WXbuf)(wxConvCurrent->cMB2WX(pszTrans));
 591     //       before, but we don't want to use wxConvCurrent explicitly to
 592     //       avoid linking unnecessary code in ANSI programs without MB
 593     //       support
 594     return (wxMB2WXbuf)(pszTrans);
 595   }
 596
 597   #undef szOrgString
 598 }
 599
 600 // find catalog by name in a linked list, return NULL if !found
 601 wxMsgCatalog *wxLocale::FindCatalog(const wxChar *szDomain) const
 602 {
 603 // linear search in the linked list
 604   wxMsgCatalog *pMsgCat;
 605   for ( pMsgCat = m_pMsgCat; pMsgCat != NULL; pMsgCat = pMsgCat->m_pNext ) {
 606     if ( wxStricmp(pMsgCat->GetName(), szDomain) == 0 )
 607       return pMsgCat;
 608   }
 609
 610   return NULL;
 611 }
 612
 613 // check if the given catalog is loaded
 614 bool wxLocale::IsLoaded(const wxChar *szDomain) const
 615 {
 616   return FindCatalog(szDomain) != NULL;
 617 }
 618
 619 // add a catalog to our linked list
 620 bool wxLocale::AddCatalog(const wxChar *szDomain)
 621 {
 622   wxMsgCatalog *pMsgCat = new wxMsgCatalog;
 623
 624   if ( pMsgCat->Load(m_strShort, szDomain) ) {
 625     // add it to the head of the list so that in GetString it will
 626     // be searched before the catalogs added earlier
 627     pMsgCat->m_pNext = m_pMsgCat;
 628     m_pMsgCat = pMsgCat;
 629
 630     return TRUE;
 631   }
 632   else {
 633     // don't add it because it couldn't be loaded anyway
 634     delete pMsgCat;
 635
 636     return FALSE;
 637   }
 638 }
 639
 640 // ----------------------------------------------------------------------------
 641 // global functions and variables
 642 // ----------------------------------------------------------------------------
 643
 644 // translation errors logging
 645 // --------------------------
 646
 647 static bool gs_bGiveTransErrors = TRUE;
 648
 649 void wxSuppressTransErrors()
 650 {
 651   gs_bGiveTransErrors = FALSE;
 652 }
 653
 654 void wxRestoreTransErrors()
 655 {
 656   gs_bGiveTransErrors = TRUE;
 657 }
 658
 659 bool wxIsLoggingTransErrors()
 660 {
 661   return gs_bGiveTransErrors;
 662 }
 663
 664 // retrieve/change current locale
 665 // ------------------------------
 666
 667 // the current locale object
 668 static wxLocale *g_pLocale = NULL;
 669
 670 wxLocale *wxGetLocale()
 671 {
 672   return g_pLocale;
 673 }
 674
 675 wxLocale *wxSetLocale(wxLocale *pLocale)
 676 {
 677   wxLocale *pOld = g_pLocale;
 678   g_pLocale = pLocale;
 679   return pOld;
 680 }
 681
 682 #endif // wxUSE_INTL
 683