setup_native/source/ulfconv/ulfconv.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: ulfconv.cxx,v $
  10  * $Revision: 1.10 $
  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 #include <stdlib.h>
  32 #include <stdio.h>
  33 #include <fcntl.h>
  34 #include <errno.h>
  35 #include <string.h>
  36 #include <unistd.h>
  37 #include <ctype.h>
  38 #include <sal/alloca.h>
  39
  40 #include <rtl/ustring.hxx>
  41
  42 #include <map>
  43 #include <string>
  44
  45 /*****************************************************************************
  46  * typedefs
  47  *****************************************************************************/
  48
  49 typedef std::map< const std::string, rtl_TextEncoding > EncodingMap;
  50
  51 struct _pair {
  52     const char *key;
  53     rtl_TextEncoding value;
  54 };
  55
  56 static int _pair_compare (const char *key, const _pair *pair);
  57 static const _pair* _pair_search (const char *key, const _pair *base, unsigned int member );
  58
  59
  60 const _pair _ms_encoding_list[] = {
  61     { "0",       RTL_TEXTENCODING_UTF8        },
  62     { "1250",    RTL_TEXTENCODING_MS_1250     },
  63     { "1251",    RTL_TEXTENCODING_MS_1251     },
  64     { "1252",    RTL_TEXTENCODING_MS_1252     },
  65     { "1253",    RTL_TEXTENCODING_MS_1253     },
  66     { "1254",    RTL_TEXTENCODING_MS_1254     },
  67     { "1255",    RTL_TEXTENCODING_MS_1255     },
  68     { "1256",    RTL_TEXTENCODING_MS_1256     },
  69     { "1257",    RTL_TEXTENCODING_MS_1257     },
  70     { "1258",    RTL_TEXTENCODING_MS_1258     },
  71     { "874",     RTL_TEXTENCODING_MS_874      },
  72     { "932",     RTL_TEXTENCODING_MS_932      },
  73     { "936",     RTL_TEXTENCODING_MS_936      },
  74     { "949",     RTL_TEXTENCODING_MS_949      },
  75     { "950",     RTL_TEXTENCODING_MS_950      }
  76 };
  77
  78
  79 /*****************************************************************************
  80  * fgets that work with unix line ends on Windows
  81  *****************************************************************************/
  82
  83 char * my_fgets(char *s, int n, FILE *fp)
  84 {
  85     int i;
  86     for( i=0; i < n-1; i++ )
  87     {
  88         int c = getc(fp);
  89
  90         if( c == EOF )
  91             break;
  92
  93         s[i] = (char) c;
  94
  95         if( s[i] == '\n' )
  96         {
  97             i++;
  98             break;
  99         }
 100     }
 101
 102     if( i>0 )
 103     {
 104         s[i] = '\0';
 105         return s;
 106     }
 107     else
 108     {
 109         return NULL;
 110     }
 111 }
 112
 113 /*****************************************************************************
 114  * compare function for binary search
 115  *****************************************************************************/
 116
 117 static int
 118 _pair_compare (const char *key, const _pair *pair)
 119 {
 120     int result = rtl_str_compareIgnoreAsciiCase( key, pair->key );
 121     return result;
 122 }
 123
 124 /*****************************************************************************
 125  * binary search on encoding tables
 126  *****************************************************************************/
 127
 128 static const _pair*
 129 _pair_search (const char *key, const _pair *base, unsigned int member )
 130 {
 131     unsigned int lower = 0;
 132     unsigned int upper = member;
 133     unsigned int current;
 134     int comparison;
 135
 136     /* check for validity of input */
 137     if ( (key == NULL) || (base == NULL) || (member == 0) )
 138         return NULL;
 139
 140     /* binary search */
 141     while ( lower < upper )
 142     {
 143         current = (lower + upper) / 2;
 144         comparison = _pair_compare( key, base + current );
 145         if (comparison < 0)
 146             upper = current;
 147         else
 148         if (comparison > 0)
 149             lower = current + 1;
 150         else
 151             return base + current;
 152     }
 153
 154     return NULL;
 155 }
 156
 157
 158 /************************************************************************
 159  * read_encoding_table
 160  ************************************************************************/
 161
 162 void read_encoding_table(char * file, EncodingMap& aEncodingMap)
 163 {
 164     FILE * fp = fopen(file, "r");
 165     if ( ! fp  ) {
 166         fprintf(stderr, "ulfconv: %s %s\n", file, strerror(errno));
 167         exit(2);
 168     }
 169
 170     char buffer[512];
 171     while ( NULL != my_fgets(buffer, sizeof(buffer), fp) ) {
 172
 173         // strip comment lines
 174         if ( buffer[0] == '#' )
 175             continue;
 176
 177         // find end of language string
 178         char * cp;
 179         for ( cp = buffer; ! isspace(*cp); cp++ )
 180             ;
 181         *cp = '\0';
 182
 183         // find start of codepage string
 184         for ( ++cp; isspace(*cp); ++cp )
 185             ;
 186         char * codepage = cp;
 187
 188         // find end of codepage string
 189         for ( ++cp; ! isspace(*cp); ++cp )
 190             ;
 191         *cp = '\0';
 192
 193         // find the correct mapping for codepage
 194         const unsigned int members = sizeof( _ms_encoding_list ) / sizeof( _pair );
 195         const _pair *encoding = _pair_search( codepage, _ms_encoding_list, members );
 196
 197         if ( encoding != NULL ) {
 198             const std::string language(buffer);
 199             aEncodingMap.insert( EncodingMap::value_type(language, encoding->value) );
 200         }
 201     }
 202 }
 203
 204 /************************************************************************
 205  * print_legacy_mixed
 206  ************************************************************************/
 207
 208 void print_legacy_mixed(
 209     FILE * ostream,
 210     const rtl::OUString& aString,
 211     const std::string& language,
 212     EncodingMap& aEncodingMap)
 213 {
 214     EncodingMap::iterator iter = aEncodingMap.find(language);
 215
 216     if ( iter != aEncodingMap.end() ) {
 217         fputs(OUStringToOString(aString, iter->second).getStr(), ostream);
 218     } else {
 219         fprintf(stderr, "ulfconv: WARNING: no legacy encoding found for %s\n", language.c_str());
 220     }
 221 }
 222
 223 /************************************************************************
 224  * print_java_style
 225  ************************************************************************/
 226
 227 void print_java_style(FILE * ostream, const rtl::OUString& aString)
 228 {
 229     int imax = aString.getLength();
 230     for (int i = 0; i < imax; i++) {
 231         sal_Unicode uc = aString[i];
 232         if ( uc < 128 ) {
 233             fprintf(ostream, "%c", (char) uc);
 234         } else {
 235             fprintf(ostream, "\\u%2.2x%2.2x", uc >> 8, uc & 0xFF );
 236         }
 237     }
 238 }
 239
 240 /************************************************************************
 241  * main
 242  ************************************************************************/
 243
 244 int main( int argc, char * const argv[] )
 245 {
 246     EncodingMap aEncodingMap;
 247
 248     FILE *istream = stdin;
 249     FILE *ostream = stdout;
 250
 251     char *outfile = NULL;
 252
 253     int errflg = 0;
 254     int argi;
 255
 256     for( argi=1; argi < argc; argi++ )
 257     {
 258         if( argv[argi][0] == '-' && argv[argi][2] == '\0' )
 259         {
 260             switch(argv[argi][1]) {
 261             case 'o':
 262                 if (argi+1 >= argc || argv[argi+1][0] == '-')
 263                 {
 264                     fprintf(stderr, "Option -%c requires an operand\n", argv[argi][1]);
 265                     errflg++;
 266                     break;
 267                 }
 268
 269                 ++argi;
 270                 outfile = argv[argi];
 271                 break;
 272             case 't':
 273                 if (argi+1 >= argc || argv[argi+1][0] == '-')
 274                 {
 275                     fprintf(stderr, "Option -%c requires an operand\n", argv[argi][1]);
 276                     errflg++;
 277                     break;
 278                 }
 279
 280                 read_encoding_table(argv[++argi], aEncodingMap);
 281                 break;
 282             default:
 283                 fprintf(stderr, "Unrecognized option: -%c\n", argv[argi][1]);
 284                 errflg++;
 285             }
 286         }
 287         else
 288         {
 289             break;
 290         }
 291     }
 292
 293     if (errflg) {
 294       fprintf(stderr, "Usage: ulfconv [-o <output file>] [-t <encoding table>] [<ulf file>]\n");
 295       exit(2);
 296     }
 297
 298     /* assign input file to stdin */
 299     if ( argi < argc )
 300     {
 301         istream = fopen(argv[argi], "r");
 302         if ( istream  == NULL ) {
 303             fprintf(stderr, "ulfconv: %s : %s\n", argv[argi], strerror(errno));
 304             exit(2);
 305         }
 306     }
 307
 308     /* open output file if any */
 309     if ( outfile )
 310     {
 311         ostream = fopen(outfile, "w");
 312         if ( ostream == NULL ) {
 313             fprintf(stderr, "ulfconv: %s : %s\n", outfile, strerror(errno));
 314             exit(2);
 315         }
 316     }
 317
 318     /* read line by line from stdin */
 319     char buffer[65536];
 320     while ( NULL != fgets(buffer, sizeof(buffer), istream) ) {
 321
 322         /* only handle lines containing " = " */
 323         char * cp = strstr(buffer, " = \"");
 324         if ( cp ) {
 325             rtl::OUString aString;
 326
 327             /* find end of lang string */
 328             int n;
 329             for ( n=0; ! isspace(buffer[n]); n++ )
 330                 ;
 331
 332             std::string line = buffer;
 333             std::string lang(line, 0, n);
 334
 335             cp += 4;
 336             rtl_string2UString( &aString.pData, cp, strrchr(cp, '\"') - cp,
 337                 RTL_TEXTENCODING_UTF8, OSTRING_TO_OUSTRING_CVTFLAGS );
 338
 339             fprintf(ostream, "%s = \"", lang.c_str());
 340
 341             if ( aEncodingMap.empty() ) {
 342                 print_java_style(ostream, aString);
 343             } else {
 344                 print_legacy_mixed(ostream, aString, lang, aEncodingMap);
 345             }
 346
 347             fprintf(ostream, "\"\n");
 348
 349
 350         } else {
 351             fputs(buffer, ostream);
 352         }
 353     }
 354 }