transex3/source/gsiconv.cxx

   1 /*************************************************************************
   2  *
   3  * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
   4  *
   5  * Copyright 2008 by Sun Microsystems, Inc.
   6  *
   7  * OpenOffice.org - a multi-platform office productivity suite
   8  *
   9  * $RCSfile: gsiconv.cxx,v $
  10  * $Revision: 1.8 $
  11  *
  12  * This file is part of OpenOffice.org.
  13  *
  14  * OpenOffice.org is free software: you can redistribute it and/or modify
  15  * it under the terms of the GNU Lesser General Public License version 3
  16  * only, as published by the Free Software Foundation.
  17  *
  18  * OpenOffice.org is distributed in the hope that it will be useful,
  19  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  20  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  21  * GNU Lesser General Public License version 3 for more details
  22  * (a copy is included in the LICENSE file that accompanied this code).
  23  *
  24  * You should have received a copy of the GNU Lesser General Public License
  25  * version 3 along with OpenOffice.org.  If not, see
  26  * <http://www.openoffice.org/license.html>
  27  * for a copy of the LGPLv3 License.
  28  *
  29  ************************************************************************/
  30
  31 // MARKER(update_precomp.py): autogen include statement, do not remove
  32 #include "precompiled_transex3.hxx"
  33 #include <stdio.h>
  34 #include <tools/fsys.hxx>
  35 #include <tools/stream.hxx>
  36
  37 // local includes
  38 #include "utf8conv.hxx"
  39
  40 #define GSI_FILE_UNKNOWN                0x0000
  41 #define GSI_FILE_OLDSTYLE               0x0001
  42 #define GSI_FILE_L10NFRAMEWORK  0x0002
  43
  44 /*****************************************************************************/
  45 USHORT GetGSIFileType( SvStream &rStream )
  46 /*****************************************************************************/
  47 {
  48     USHORT nFileType = GSI_FILE_UNKNOWN;
  49
  50     ULONG nPos( rStream.Tell());
  51     rStream.Seek( STREAM_SEEK_TO_BEGIN );
  52
  53     ByteString sLine;
  54     while( !rStream.IsEof() && !sLine.Len())
  55         rStream.ReadLine( sLine );
  56
  57     if( sLine.Len()) {
  58         if( sLine.Search( "($$)" ) != STRING_NOTFOUND )
  59             nFileType = GSI_FILE_OLDSTYLE;
  60         else
  61             nFileType = GSI_FILE_L10NFRAMEWORK;
  62     }
  63
  64     rStream.Seek( nPos );
  65
  66     return nFileType;
  67 }
  68
  69 /*****************************************************************************/
  70 ByteString GetGSILineId( const ByteString &rLine, USHORT nFileType )
  71 /*****************************************************************************/
  72 {
  73     ByteString sId;
  74     switch ( nFileType ) {
  75         case GSI_FILE_OLDSTYLE:
  76             sId = rLine;
  77             sId.SearchAndReplaceAll( "($$)", "\t" );
  78             sId = sId.GetToken( 0, '\t' );
  79           break;
  80
  81         case GSI_FILE_L10NFRAMEWORK:
  82             sId = rLine.GetToken( 0, '\t' );
  83             sId += "\t";
  84             sId += rLine.GetToken( 1, '\t' );
  85             sId += "\t";
  86             sId += rLine.GetToken( 4, '\t' );
  87             sId += "\t";
  88             sId += rLine.GetToken( 5, '\t' );
  89           break;
  90     }
  91     return sId;
  92 }
  93
  94 /*****************************************************************************/
  95 ByteString GetGSILineLangId( const ByteString &rLine, USHORT nFileType )
  96 /*****************************************************************************/
  97 {
  98     ByteString sLangId;
  99     switch ( nFileType ) {
 100         case GSI_FILE_OLDSTYLE:
 101             sLangId = rLine;
 102             sLangId.SearchAndReplaceAll( "($$)", "\t" );
 103             sLangId = sLangId.GetToken( 2, '\t' );
 104           break;
 105
 106         case GSI_FILE_L10NFRAMEWORK:
 107             sLangId = rLine.GetToken( 9, '\t' );
 108           break;
 109     }
 110     return sLangId;
 111 }
 112
 113 /*****************************************************************************/
 114 void ConvertGSILine( BOOL bToUTF8, ByteString &rLine,
 115         rtl_TextEncoding nEncoding,     USHORT nFileType )
 116 /*****************************************************************************/
 117 {
 118     switch ( nFileType ) {
 119         case GSI_FILE_OLDSTYLE:
 120             if ( bToUTF8 )
 121                 rLine = UTF8Converter::ConvertToUTF8( rLine, nEncoding );
 122             else
 123                 rLine = UTF8Converter::ConvertFromUTF8( rLine, nEncoding );
 124         break;
 125
 126         case GSI_FILE_L10NFRAMEWORK: {
 127             ByteString sConverted;
 128             for ( USHORT i = 0; i < rLine.GetTokenCount( '\t' ); i++ ) {
 129                 ByteString sToken = rLine.GetToken( i, '\t' );
 130                 if (( i > 9 ) && ( i < 14 )) {
 131                     if( bToUTF8 )
 132                         sToken = UTF8Converter::ConvertToUTF8( sToken, nEncoding );
 133                     else
 134                         sToken = UTF8Converter::ConvertFromUTF8( sToken, nEncoding );
 135                 }
 136                 if ( i )
 137                     sConverted += "\t";
 138                 sConverted += sToken;
 139             }
 140             rLine = sConverted;
 141         }
 142         break;
 143     }
 144 }
 145
 146 /*****************************************************************************/
 147 void Help()
 148 /*****************************************************************************/
 149 {
 150     fprintf( stdout, "\n" );
 151     fprintf( stdout, "gsiconv (c)1999 by StarOffice Entwicklungs GmbH\n" );
 152     fprintf( stdout, "===============================================\n" );
 153     fprintf( stdout, "\n" );
 154     fprintf( stdout, "gsiconv converts strings in GSI-Files (Gutschmitt Interface) from or to UTF-8\n" );
 155     fprintf( stdout, "\n" );
 156     fprintf( stdout, "Syntax: gsiconv (-t|-f langid charset)|(-p n) filename\n" );
 157     fprintf( stdout, "Switches: -t   => conversion from charset to UTF-8\n" );
 158     fprintf( stdout, "          -f   => conversion from UTF-8 to charset\n" );
 159     fprintf( stdout, "          -p n => creates several files with ca. n lines\n" );
 160     fprintf( stdout, "\n" );
 161     fprintf( stdout, "Allowed charsets:\n" );
 162     fprintf( stdout, "          MS_932  => Japanese\n" );
 163     fprintf( stdout, "          MS_936  => Chinese Simplified\n" );
 164     fprintf( stdout, "          MS_949  => Korean\n" );
 165     fprintf( stdout, "          MS_950  => Chinese Traditional\n" );
 166     fprintf( stdout, "          MS_1250 => East Europe\n" );
 167     fprintf( stdout, "          MS_1251 => Cyrillic\n" );
 168     fprintf( stdout, "          MS_1252 => West Europe\n" );
 169     fprintf( stdout, "          MS_1253 => Greek\n" );
 170     fprintf( stdout, "          MS_1254 => Turkish\n" );
 171     fprintf( stdout, "          MS_1255 => Hebrew\n" );
 172     fprintf( stdout, "          MS_1256 => Arabic\n" );
 173     fprintf( stdout, "\n" );
 174     fprintf( stdout, "Allowed langids:\n" );
 175     fprintf( stdout, "          1  => ENGLISH_US\n" );
 176     fprintf( stdout, "          3  => PORTUGUESE \n" );
 177     fprintf( stdout, "          4  => GERMAN_DE (new german style)\n" );
 178     fprintf( stdout, "          7  => RUSSIAN\n" );
 179     fprintf( stdout, "          30 => GREEK\n" );
 180     fprintf( stdout, "          31 => DUTCH\n" );
 181     fprintf( stdout, "          33 => FRENCH\n" );
 182     fprintf( stdout, "          34 => SPANISH\n" );
 183     fprintf( stdout, "          35 => FINNISH\n" );
 184     fprintf( stdout, "          36 => HUNGARIAN\n" );
 185     fprintf( stdout, "          39 => ITALIAN\n" );
 186     fprintf( stdout, "          42 => CZECH\n" );
 187     fprintf( stdout, "          44 => ENGLISH (UK)\n" );
 188     fprintf( stdout, "          45 => DANISH\n" );
 189     fprintf( stdout, "          46 => SWEDISH\n" );
 190     fprintf( stdout, "          47 => NORWEGIAN\n" );
 191     fprintf( stdout, "          49 => GERMAN (old german style)\n" );
 192     fprintf( stdout, "          55 => PORTUGUESE_BRAZILIAN\n" );
 193     fprintf( stdout, "          81 => JAPANESE\n" );
 194     fprintf( stdout, "          82 => KOREAN\n" );
 195     fprintf( stdout, "          86 => CHINESE_SIMPLIFIED\n" );
 196     fprintf( stdout, "          88 => CHINESE_TRADITIONAL\n" );
 197     fprintf( stdout, "          90 => TURKISH\n" );
 198     fprintf( stdout, "          96 => ARABIC\n" );
 199     fprintf( stdout, "          97 => HEBREW\n" );
 200     fprintf( stdout, "\n" );
 201 }
 202
 203 /*****************************************************************************/
 204 #if defined(UNX) || defined(OS2)
 205 int main( int argc, char *argv[] )
 206 #else
 207 int _cdecl main( int argc, char *argv[] )
 208 #endif
 209 /*****************************************************************************/
 210 {
 211     if (( argc != 5 ) && ( argc != 4 )) {
 212         Help();
 213         exit ( 0 );
 214     }
 215
 216     if ( argc == 4 ) {
 217         if ( ByteString( argv[ 1 ] ) == "-p" ) {
 218
 219             DirEntry aSource = DirEntry( String( argv[ 3 ], RTL_TEXTENCODING_ASCII_US ));
 220             if ( !aSource.Exists()) {
 221                 fprintf( stderr, "\nERROR: GSI-File %s not found!\n\n", ByteString( argv[ 3 ] ).GetBuffer());
 222                 exit ( 2 );
 223             }
 224
 225             DirEntry aOutput( aSource );
 226
 227             String sBase = aOutput.GetBase();
 228             String sExt = aOutput.GetExtension();
 229
 230             String sGSI( argv[ 3 ], RTL_TEXTENCODING_ASCII_US );
 231             SvFileStream aGSI( sGSI, STREAM_STD_READ  );
 232             if ( !aGSI.IsOpen()) {
 233                 fprintf( stderr, "\nERROR: Could not open GSI-File %s!\n\n", ByteString( argv[ 3 ] ).GetBuffer());
 234                 exit ( 3 );
 235             }
 236
 237             USHORT nFileType( GetGSIFileType( aGSI ));
 238
 239             ULONG nMaxLines = (ULONG) ByteString( argv[ 2 ] ).ToInt64();
 240             if ( !nMaxLines ) {
 241                 fprintf( stderr, "\nERROR: Linecount must be at least 1!\n\n" );
 242                 exit ( 3 );
 243             }
 244
 245             ByteString sGSILine;
 246             ByteString sOldId;
 247             ULONG nLine = 0;
 248             ULONG nOutputFile = 1;
 249
 250             String sOutput( sBase );
 251             sOutput += String( "_", RTL_TEXTENCODING_ASCII_US );
 252             sOutput += String::CreateFromInt64( nOutputFile );
 253             if ( sExt.Len()) {
 254                 sOutput += String( ".", RTL_TEXTENCODING_ASCII_US );
 255                 sOutput += sExt;
 256             }
 257             nOutputFile ++;
 258
 259             aOutput.SetName( sOutput );
 260             SvFileStream aOutputStream( aOutput.GetFull(), STREAM_STD_WRITE | STREAM_TRUNC );
 261
 262             while ( !aGSI.IsEof()) {
 263
 264                 aGSI.ReadLine( sGSILine );
 265                 ByteString sId( GetGSILineId( sGSILine, nFileType ));
 266
 267                 nLine++;
 268
 269                 if (( nLine >= nMaxLines ) && ( sId != sOldId )) {
 270                     aOutputStream.Close();
 271
 272                     ByteString sText( aOutput.GetFull(), gsl_getSystemTextEncoding());
 273                     sText += " with ";
 274                     sText += ByteString::CreateFromInt64( nLine );
 275                     sText += " lines written.";
 276
 277                     fprintf( stdout, "%s\n", sText.GetBuffer());
 278                     String sOutput1( sBase );
 279                     sOutput1 += String( "_", RTL_TEXTENCODING_ASCII_US );
 280                     sOutput1 += String::CreateFromInt64( nOutputFile );
 281                     if ( sExt.Len()) {
 282                         sOutput1 += String( ".", RTL_TEXTENCODING_ASCII_US );
 283                         sOutput1 += sExt;
 284                     }
 285                     nOutputFile ++;
 286
 287                     aOutput.SetName( sOutput1 );
 288
 289                     aOutputStream.Open( aOutput.GetFull(), STREAM_STD_WRITE | STREAM_TRUNC );
 290                     nLine = 0;
 291                 }
 292
 293                 aOutputStream.WriteLine( sGSILine );
 294
 295                 sOldId = sId;
 296             }
 297
 298             aGSI.Close();
 299             aOutputStream.Close();
 300
 301             ByteString sText( aOutput.GetFull(), RTL_TEXTENCODING_ASCII_US );
 302             sText += " with ";
 303             sText += ByteString::CreateFromInt64( nLine );
 304             sText += " lines written.";
 305         }
 306         else {
 307             Help();
 308             exit( 1 );
 309         }
 310     }
 311     else {
 312         if ( ByteString( argv[ 1 ] ) == "-t" || ByteString( argv[ 1 ] ) == "-f" ) {
 313             rtl_TextEncoding nEncoding;
 314
 315             ByteString sCurLangId( argv[ 2 ] );
 316
 317             ByteString sCharset( argv[ 3 ] );
 318             sCharset.ToUpperAscii();
 319
 320             if          ( sCharset == "MS_932" )        nEncoding = RTL_TEXTENCODING_MS_932;
 321             else if ( sCharset == "MS_936" )    nEncoding = RTL_TEXTENCODING_MS_936;
 322             else if ( sCharset == "MS_949" )    nEncoding = RTL_TEXTENCODING_MS_949;
 323             else if ( sCharset == "MS_950" )    nEncoding = RTL_TEXTENCODING_MS_950;
 324             else if ( sCharset == "MS_1250" )   nEncoding = RTL_TEXTENCODING_MS_1250;
 325             else if ( sCharset == "MS_1251" )   nEncoding = RTL_TEXTENCODING_MS_1251;
 326             else if ( sCharset == "MS_1252" )   nEncoding = RTL_TEXTENCODING_MS_1252;
 327             else if ( sCharset == "MS_1253" )   nEncoding = RTL_TEXTENCODING_MS_1253;
 328             else if ( sCharset == "MS_1254" )   nEncoding = RTL_TEXTENCODING_MS_1254;
 329             else if ( sCharset == "MS_1255" )   nEncoding = RTL_TEXTENCODING_MS_1255;
 330             else if ( sCharset == "MS_1256" )   nEncoding = RTL_TEXTENCODING_MS_1256;
 331             else if ( sCharset == "MS_1257" )   nEncoding = RTL_TEXTENCODING_MS_1257;
 332             else if ( sCharset == "UTF8" )              nEncoding = RTL_TEXTENCODING_UTF8;
 333
 334             else {
 335                 Help();
 336                 exit ( 1 );
 337             }
 338
 339             DirEntry aSource = DirEntry( String( argv[ 4 ], RTL_TEXTENCODING_ASCII_US ));
 340             if ( !aSource.Exists()) {
 341                 fprintf( stderr, "\nERROR: GSI-File %s not found!\n\n", ByteString( argv[ 3 ] ).GetBuffer());
 342                 exit ( 2 );
 343             }
 344
 345             String sGSI( argv[ 4 ], RTL_TEXTENCODING_ASCII_US );
 346             SvFileStream aGSI( sGSI, STREAM_STD_READ );
 347             if ( !aGSI.IsOpen()) {
 348                 fprintf( stderr, "\nERROR: Could not open GSI-File %s!\n\n", ByteString( argv[ 3 ] ).GetBuffer());
 349                 exit ( 3 );
 350             }
 351             USHORT nFileType( GetGSIFileType( aGSI ));
 352
 353             ByteString sGSILine;
 354             while ( !aGSI.IsEof()) {
 355
 356                 aGSI.ReadLine( sGSILine );
 357                 ByteString sLangId( GetGSILineLangId( sGSILine, nFileType ));
 358                 if ( sLangId == sCurLangId )
 359                     ConvertGSILine(( ByteString( argv[ 1 ] ) == "-t" ), sGSILine, nEncoding, nFileType );
 360
 361                 fprintf( stdout, "%s\n", sGSILine.GetBuffer());
 362             }
 363
 364             aGSI.Close();
 365         }
 366         else {
 367             Help();
 368             exit( 1 );
 369         }
 370     }
 371     return 0;
 372 }