locale/po2strings.c

   1 /*
   2  * Convert a GNU gettext .po file to an Apple .strings file.
   3  *
   4  * Copyright 2007-2017 by Apple Inc.
   5  *
   6  * Licensed under Apache License v2.0.  See the file "LICENSE" for more information.
   7  *
   8  * Usage:
   9  *
  10  *   po2strings filename.strings filename.po
  11  *
  12  * Compile with:
  13  *
  14  *   gcc -o po2strings po2strings.c `cups-config --libs`
  15  */
  16
  17 #include <cups/cups-private.h>
  18
  19
  20 /*
  21  * The .strings file format is simple:
  22  *
  23  * // comment
  24  * "msgid" = "msgstr";
  25  *
  26  * The GNU gettext .po format is also fairly simple:
  27  *
  28  *     #. comment
  29  *     msgid "some text"
  30  *     msgstr "localized text"
  31  *
  32  * The comment, msgid, and msgstr text can span multiple lines using the form:
  33  *
  34  *     #. comment
  35  *     #. more comments
  36  *     msgid ""
  37  *     "some long text"
  38  *     msgstr ""
  39  *     "localized text spanning "
  40  *     "multiple lines"
  41  *
  42  * Both the msgid and msgstr strings use standard C quoting for special
  43  * characters like newline and the double quote character.
  44  */
  45
  46 static char     *normalize_string(const char *idstr, char *buffer, size_t bufsize);
  47
  48
  49 /*
  50  *   main() - Convert .po file to .strings.
  51  */
  52
  53 int                                     /* O - Exit code */
  54 main(int  argc,                         /* I - Number of command-line args */
  55      char *argv[])                      /* I - Command-line arguments */
  56 {
  57   int                   i;              /* Looping var */
  58   const char            *pofile,        /* .po filename */
  59                         *stringsfile;   /* .strings filename */
  60   cups_file_t           *po,            /* .po file */
  61                         *strings;       /* .strings file */
  62   char                  s[4096],        /* String buffer */
  63                         *ptr,           /* Pointer into buffer */
  64                         *temp,          /* New string */
  65                         *msgid,         /* msgid string */
  66                         *msgstr,        /* msgstr string */
  67                         normalized[8192];/* Normalized msgid string */
  68   size_t                length;         /* Length of combined strings */
  69   int                   use_msgid;      /* Use msgid strings for msgstr? */
  70
  71
  72  /*
  73   * Process command-line arguments...
  74   */
  75
  76   pofile      = NULL;
  77   stringsfile = NULL;
  78   use_msgid   = 0;
  79
  80   for (i = 1; i < argc; i ++)
  81   {
  82     if (!strcmp(argv[i], "-m"))
  83       use_msgid = 1;
  84     else if (argv[i][0] == '-')
  85     {
  86       puts("Usage: po2strings [-m] filename.po filename.strings");
  87       return (1);
  88     }
  89     else if (!pofile)
  90       pofile = argv[i];
  91     else if (!stringsfile)
  92       stringsfile = argv[i];
  93     else
  94     {
  95       puts("Usage: po2strings [-m] filename.po filename.strings");
  96       return (1);
  97     }
  98   }
  99
 100   if (!pofile || !stringsfile)
 101   {
 102     puts("Usage: po2strings [-m] filename.po filename.strings");
 103     return (1);
 104   }
 105
 106  /*
 107   * Read strings from the .po file and write to the .strings file...
 108   */
 109
 110   if ((po = cupsFileOpen(pofile, "r")) == NULL)
 111   {
 112     perror(pofile);
 113     return (1);
 114   }
 115
 116   if ((strings = cupsFileOpen(stringsfile, "w")) == NULL)
 117   {
 118     perror(stringsfile);
 119     cupsFileClose(po);
 120     return (1);
 121   }
 122
 123   msgid = msgstr = NULL;
 124
 125   while (cupsFileGets(po, s, sizeof(s)) != NULL)
 126   {
 127     if (s[0] == '#' && s[1] == '.')
 128     {
 129      /*
 130       * Copy comment string...
 131       */
 132
 133       if (msgid && msgstr)
 134       {
 135        /*
 136         * First output the last localization string...
 137         */
 138
 139         if (*msgid)
 140           cupsFilePrintf(strings, "\"%s\" = \"%s\";\n", msgid,
 141                          (use_msgid || !*msgstr) ? msgid : msgstr);
 142
 143         free(msgid);
 144         free(msgstr);
 145         msgid = msgstr = NULL;
 146       }
 147
 148       cupsFilePrintf(strings, "//%s\n", s + 2);
 149     }
 150     else if (s[0] == '#' || !s[0])
 151     {
 152      /*
 153       * Skip blank and file comment lines...
 154       */
 155
 156       continue;
 157     }
 158     else
 159     {
 160      /*
 161       * Strip the trailing quote...
 162       */
 163
 164       if ((ptr = strrchr(s, '\"')) == NULL)
 165         continue;
 166
 167       *ptr = '\0';
 168
 169      /*
 170       * Find start of value...
 171       */
 172
 173       if ((ptr = strchr(s, '\"')) == NULL)
 174         continue;
 175
 176       ptr ++;
 177
 178      /*
 179       * Create or add to a message...
 180       */
 181
 182       if (!strncmp(s, "msgid", 5))
 183       {
 184        /*
 185         * Output previous message as needed...
 186         */
 187
 188         if (msgid && msgstr)
 189         {
 190           if (*msgid)
 191             cupsFilePrintf(strings, "\"%s\" = \"%s\";\n", msgid, normalize_string((use_msgid || !*msgstr) ? msgid : msgstr, normalized, sizeof(normalized)));
 192         }
 193
 194         if (msgid)
 195           free(msgid);
 196
 197         if (msgstr)
 198           free(msgstr);
 199
 200         msgid  = strdup(ptr);
 201         msgstr = NULL;
 202       }
 203       else if (s[0] == '\"' && (msgid || msgstr))
 204       {
 205        /*
 206         * Append to current string...
 207         */
 208
 209         size_t ptrlen = strlen(ptr);    /* Length of string */
 210
 211         length = strlen(msgstr ? msgstr : msgid);
 212
 213         if ((temp = realloc(msgstr ? msgstr : msgid,
 214                             length + ptrlen + 1)) == NULL)
 215         {
 216           free(msgid);
 217           if (msgstr)
 218             free(msgstr);
 219           perror("Unable to allocate string");
 220           return (1);
 221         }
 222
 223         if (msgstr)
 224         {
 225          /*
 226           * Copy the new portion to the end of the msgstr string - safe
 227           * to use strcpy because the buffer is allocated to the correct
 228           * size...
 229           */
 230
 231           msgstr = temp;
 232
 233           memcpy(msgstr + length, ptr, ptrlen + 1);
 234         }
 235         else
 236         {
 237          /*
 238           * Copy the new portion to the end of the msgid string - safe
 239           * to use strcpy because the buffer is allocated to the correct
 240           * size...
 241           */
 242
 243           msgid = temp;
 244
 245           memcpy(msgid + length, ptr, ptrlen + 1);
 246         }
 247       }
 248       else if (!strncmp(s, "msgstr", 6) && msgid)
 249       {
 250        /*
 251         * Set the string...
 252         */
 253
 254         if (msgstr)
 255           free(msgstr);
 256
 257         if ((msgstr = strdup(ptr)) == NULL)
 258         {
 259           free(msgid);
 260           perror("Unable to allocate msgstr");
 261           return (1);
 262         }
 263       }
 264     }
 265   }
 266
 267   if (msgid && msgstr)
 268   {
 269     if (*msgid)
 270       cupsFilePrintf(strings, "\"%s\" = \"%s\";\n", msgid, normalize_string((use_msgid || !*msgstr) ? msgid : msgstr, normalized, sizeof(normalized)));
 271   }
 272
 273   if (msgid)
 274     free(msgid);
 275
 276   if (msgstr)
 277     free(msgstr);
 278
 279   cupsFileClose(po);
 280   cupsFileClose(strings);
 281
 282   return (0);
 283 }
 284
 285
 286 /*
 287  * 'normalize_string()' - Normalize a msgid string.
 288  *
 289  * This function converts ASCII ellipsis and double quotes to their Unicode
 290  * counterparts.
 291  */
 292
 293 static char *                           /* O - Normalized string */
 294 normalize_string(const char *idstr,     /* I - msgid string */
 295                  char       *buffer,    /* I - Normalized string buffer */
 296                  size_t     bufsize)    /* I - Size of string buffer */
 297 {
 298   char  *bufptr = buffer,               /* Pointer into buffer */
 299         *bufend = buffer + bufsize - 3; /* End of buffer */
 300   int   quote = 0,                      /* Quote direction */
 301         html = 0;                       /* HTML text */
 302
 303
 304   while (*idstr && bufptr < bufend)
 305   {
 306     if (!strncmp(idstr, "<A ", 3))
 307       html = 1;
 308     else if (html && *idstr == '>')
 309       html = 0;
 310
 311     if (*idstr == '.' && idstr[1] == '.' && idstr[2] == '.')
 312     {
 313      /*
 314       * Convert ... to Unicode ellipsis...
 315       */
 316
 317       *bufptr++ = (char)0xE2;
 318       *bufptr++ = (char)0x80;
 319       *bufptr++ = (char)0xA6;
 320       idstr += 2;
 321     }
 322     else if (!html && *idstr == '\\' && idstr[1] == '\"')
 323     {
 324       if (quote)
 325       {
 326        /*
 327         * Convert second \" to Unicode right (curley) double quote.
 328         */
 329
 330         *bufptr++ = (char)0xE2;
 331         *bufptr++ = (char)0x80;
 332         *bufptr++ = (char)0x9D;
 333         quote     = 0;
 334       }
 335       else if (strchr(idstr + 2, '\"') != NULL)
 336       {
 337        /*
 338         * Convert first \" to Unicode left (curley) double quote.
 339         */
 340
 341         *bufptr++ = (char)0xE2;
 342         *bufptr++ = (char)0x80;
 343         *bufptr++ = (char)0x9C;
 344         quote     = 1;
 345       }
 346       else
 347       {
 348        /*
 349         * Convert lone \" to Unicode double prime.
 350         */
 351
 352         *bufptr++ = (char)0xE2;
 353         *bufptr++ = (char)0x80;
 354         *bufptr++ = (char)0xB3;
 355       }
 356
 357       idstr ++;
 358     }
 359     else if (*idstr == '\'')
 360     {
 361       if (strchr(idstr + 1, '\'') == NULL || quote)
 362       {
 363        /*
 364         * Convert second ' (or ' used for a contraction) to Unicode right
 365         * (curley) single quote.
 366         */
 367
 368         *bufptr++ = (char)0xE2;
 369         *bufptr++ = (char)0x80;
 370         *bufptr++ = (char)0x99;
 371         quote     = 0;
 372       }
 373       else
 374       {
 375        /*
 376         * Convert first ' to Unicode left (curley) single quote.
 377         */
 378
 379         *bufptr++ = (char)0xE2;
 380         *bufptr++ = (char)0x80;
 381         *bufptr++ = (char)0x98;
 382         quote     = 1;
 383       }
 384     }
 385     else
 386       *bufptr++ = *idstr;
 387
 388     idstr ++;
 389   }
 390
 391   *bufptr = '\0';
 392
 393   return (buffer);
 394 }