gio/gcharsetconverter.c

   1 /* GIO - GLib Input, Output and Streaming Library
   2  *
   3  * Copyright (C) 2009 Red Hat, Inc.
   4  *
   5  * This library is free software; you can redistribute it and/or
   6  * modify it under the terms of the GNU Lesser General Public
   7  * License as published by the Free Software Foundation; either
   8  * version 2 of the License, or (at your option) any later version.
   9  *
  10  * This library is distributed in the hope that it will be useful,
  11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  13  * Lesser General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU Lesser General
  16  * Public License along with this library; if not, see <http://www.gnu.org/licenses/>.
  17  *
  18  * Author: Alexander Larsson <alexl@redhat.com>
  19  */
  20
  21 #include "config.h"
  22
  23 #include "gcharsetconverter.h"
  24
  25 #include <errno.h>
  26
  27 #include "ginitable.h"
  28 #include "gioerror.h"
  29 #include "glibintl.h"
  30
  31
  32 enum {
  33   PROP_0,
  34   PROP_FROM_CHARSET,
  35   PROP_TO_CHARSET,
  36   PROP_USE_FALLBACK
  37 };
  38
  39 /**
  40  * SECTION:gcharsetconverter
  41  * @short_description: Convert between charsets
  42  * @include: gio/gio.h
  43  *
  44  * #GCharsetConverter is an implementation of #GConverter based on
  45  * GIConv.
  46  */
  47
  48 static void g_charset_converter_iface_init          (GConverterIface *iface);
  49 static void g_charset_converter_initable_iface_init (GInitableIface  *iface);
  50
  51 /**
  52  * GCharsetConverter:
  53  *
  54  * Conversions between character sets.
  55  */
  56 struct _GCharsetConverter
  57 {
  58   GObject parent_instance;
  59
  60   char *from;
  61   char *to;
  62   GIConv iconv;
  63   gboolean use_fallback;
  64   guint n_fallback_errors;
  65 };
  66
  67 G_DEFINE_TYPE_WITH_CODE (GCharsetConverter, g_charset_converter, G_TYPE_OBJECT,
  68                          G_IMPLEMENT_INTERFACE (G_TYPE_CONVERTER,
  69                                                 g_charset_converter_iface_init);
  70                          G_IMPLEMENT_INTERFACE (G_TYPE_INITABLE,
  71                                                 g_charset_converter_initable_iface_init))
  72
  73 static void
  74 g_charset_converter_finalize (GObject *object)
  75 {
  76   GCharsetConverter *conv;
  77
  78   conv = G_CHARSET_CONVERTER (object);
  79
  80   g_free (conv->from);
  81   g_free (conv->to);
  82   if (conv->iconv)
  83     g_iconv_close (conv->iconv);
  84
  85   G_OBJECT_CLASS (g_charset_converter_parent_class)->finalize (object);
  86 }
  87
  88 static void
  89 g_charset_converter_set_property (GObject      *object,
  90                                   guint         prop_id,
  91                                   const GValue *value,
  92                                   GParamSpec   *pspec)
  93 {
  94   GCharsetConverter *conv;
  95
  96   conv = G_CHARSET_CONVERTER (object);
  97
  98   switch (prop_id)
  99     {
 100     case PROP_TO_CHARSET:
 101       g_free (conv->to);
 102       conv->to = g_value_dup_string (value);
 103       break;
 104
 105     case PROP_FROM_CHARSET:
 106       g_free (conv->from);
 107       conv->from = g_value_dup_string (value);
 108       break;
 109
 110     case PROP_USE_FALLBACK:
 111       conv->use_fallback = g_value_get_boolean (value);
 112       break;
 113
 114     default:
 115       G_OBJECT_WARN_INVALID_PROPERTY_ID (object, prop_id, pspec);
 116       break;
 117     }
 118
 119 }
 120
 121 static void
 122 g_charset_converter_get_property (GObject    *object,
 123                                   guint       prop_id,
 124                                   GValue     *value,
 125                                   GParamSpec *pspec)
 126 {
 127   GCharsetConverter *conv;
 128
 129   conv = G_CHARSET_CONVERTER (object);
 130
 131   switch (prop_id)
 132     {
 133     case PROP_TO_CHARSET:
 134       g_value_set_string (value, conv->to);
 135       break;
 136
 137     case PROP_FROM_CHARSET:
 138       g_value_set_string (value, conv->from);
 139       break;
 140
 141     case PROP_USE_FALLBACK:
 142       g_value_set_boolean (value, conv->use_fallback);
 143       break;
 144
 145     default:
 146       G_OBJECT_WARN_INVALID_PROPERTY_ID (object, prop_id, pspec);
 147       break;
 148     }
 149 }
 150
 151 static void
 152 g_charset_converter_class_init (GCharsetConverterClass *klass)
 153 {
 154   GObjectClass *gobject_class = G_OBJECT_CLASS (klass);
 155
 156   gobject_class->finalize = g_charset_converter_finalize;
 157   gobject_class->get_property = g_charset_converter_get_property;
 158   gobject_class->set_property = g_charset_converter_set_property;
 159
 160   g_object_class_install_property (gobject_class,
 161                                    PROP_TO_CHARSET,
 162                                    g_param_spec_string ("to-charset",
 163                                                         P_("To Charset"),
 164                                                         P_("The character encoding to convert to"),
 165                                                         NULL,
 166                                                         G_PARAM_READWRITE | G_PARAM_CONSTRUCT_ONLY |
 167                                                         G_PARAM_STATIC_STRINGS));
 168   g_object_class_install_property (gobject_class,
 169                                    PROP_FROM_CHARSET,
 170                                    g_param_spec_string ("from-charset",
 171                                                         P_("From Charset"),
 172                                                         P_("The character encoding to convert from"),
 173                                                         NULL,
 174                                                         G_PARAM_READWRITE | G_PARAM_CONSTRUCT_ONLY |
 175                                                         G_PARAM_STATIC_STRINGS));
 176   g_object_class_install_property (gobject_class,
 177                                    PROP_USE_FALLBACK,
 178                                    g_param_spec_boolean ("use-fallback",
 179                                                          P_("Fallback enabled"),
 180                                                          P_("Use fallback (of form \\<hexval>) for invalid bytes"),
 181                                                          FALSE,
 182                                                          G_PARAM_READWRITE |
 183                                                          G_PARAM_CONSTRUCT |
 184                                                          G_PARAM_STATIC_STRINGS));
 185 }
 186
 187 static void
 188 g_charset_converter_init (GCharsetConverter *local)
 189 {
 190 }
 191
 192
 193 /**
 194  * g_charset_converter_new:
 195  * @to_charset: destination charset
 196  * @from_charset: source charset
 197  * @error: #GError for error reporting, or %NULL to ignore.
 198  *
 199  * Creates a new #GCharsetConverter.
 200  *
 201  * Returns: a new #GCharsetConverter or %NULL on error.
 202  *
 203  * Since: 2.24
 204  **/
 205 GCharsetConverter *
 206 g_charset_converter_new (const gchar *to_charset,
 207                          const gchar *from_charset,
 208                          GError      **error)
 209 {
 210   GCharsetConverter *conv;
 211
 212   conv = g_initable_new (G_TYPE_CHARSET_CONVERTER,
 213                          NULL, error,
 214                          "to-charset", to_charset,
 215                          "from-charset", from_charset,
 216                          NULL);
 217
 218   return conv;
 219 }
 220
 221 static void
 222 g_charset_converter_reset (GConverter *converter)
 223 {
 224   GCharsetConverter *conv = G_CHARSET_CONVERTER (converter);
 225
 226   if (conv->iconv == NULL)
 227     {
 228       g_warning ("Invalid object, not initialized");
 229       return;
 230     }
 231
 232   g_iconv (conv->iconv, NULL, NULL, NULL, NULL);
 233   conv->n_fallback_errors = 0;
 234 }
 235
 236 static GConverterResult
 237 g_charset_converter_convert (GConverter       *converter,
 238                              const void       *inbuf,
 239                              gsize             inbuf_size,
 240                              void             *outbuf,
 241                              gsize             outbuf_size,
 242                              GConverterFlags   flags,
 243                              gsize            *bytes_read,
 244                              gsize            *bytes_written,
 245                              GError          **error)
 246 {
 247   GCharsetConverter  *conv;
 248   gsize res;
 249   GConverterResult ret;
 250   gchar *inbufp, *outbufp;
 251   gsize in_left, out_left;
 252   int errsv;
 253   gboolean reset;
 254
 255   conv = G_CHARSET_CONVERTER (converter);
 256
 257   if (conv->iconv == NULL)
 258     {
 259       g_set_error_literal (error, G_IO_ERROR, G_IO_ERROR_NOT_INITIALIZED,
 260                            _("Invalid object, not initialized"));
 261       return G_CONVERTER_ERROR;
 262     }
 263
 264   inbufp = (char *)inbuf;
 265   outbufp = (char *)outbuf;
 266   in_left = inbuf_size;
 267   out_left = outbuf_size;
 268   reset = FALSE;
 269
 270   /* if there is not input try to flush the data */
 271   if (inbuf_size == 0)
 272     {
 273       if (flags & G_CONVERTER_INPUT_AT_END ||
 274           flags & G_CONVERTER_FLUSH)
 275         {
 276           reset = TRUE;
 277         }
 278       else
 279         {
 280           g_set_error_literal (error, G_IO_ERROR, G_IO_ERROR_PARTIAL_INPUT,
 281                                _("Incomplete multibyte sequence in input"));
 282           return G_CONVERTER_ERROR;
 283         }
 284     }
 285
 286   if (reset)
 287     /* call g_iconv with NULL inbuf to cleanup shift state */
 288     res = g_iconv (conv->iconv,
 289                    NULL, &in_left,
 290                    &outbufp, &out_left);
 291   else
 292     res = g_iconv (conv->iconv,
 293                    &inbufp, &in_left,
 294                    &outbufp, &out_left);
 295
 296   *bytes_read = inbufp - (char *)inbuf;
 297   *bytes_written = outbufp - (char *)outbuf;
 298
 299   /* Don't report error if we converted anything */
 300   if (res == (gsize) -1 && *bytes_read == 0)
 301     {
 302       errsv = errno;
 303
 304       switch (errsv)
 305         {
 306         case EINVAL:
 307           /* Incomplete input text */
 308           g_set_error_literal (error, G_IO_ERROR, G_IO_ERROR_PARTIAL_INPUT,
 309                                _("Incomplete multibyte sequence in input"));
 310           break;
 311
 312         case E2BIG:
 313           /* Not enough destination space */
 314           g_set_error_literal (error, G_IO_ERROR, G_IO_ERROR_NO_SPACE,
 315                                _("Not enough space in destination"));
 316           break;
 317
 318         case EILSEQ:
 319           /* Invalid code sequence */
 320           if (conv->use_fallback)
 321             {
 322               if (outbuf_size < 3)
 323                 g_set_error_literal (error, G_IO_ERROR, G_IO_ERROR_NO_SPACE,
 324                                      _("Not enough space in destination"));
 325               else
 326                 {
 327                   const char hex[] = "0123456789ABCDEF";
 328                   guint8 v = *(guint8 *)inbuf;
 329                   guint8 *out = (guint8 *)outbuf;
 330                   out[0] = '\\';
 331                   out[1] = hex[(v & 0xf0) >> 4];
 332                   out[2] = hex[(v & 0x0f) >> 0];
 333                   *bytes_read = 1;
 334                   *bytes_written = 3;
 335                   in_left--;
 336                   conv->n_fallback_errors++;
 337                   goto ok;
 338                 }
 339             }
 340           else
 341             g_set_error_literal (error, G_IO_ERROR, G_IO_ERROR_INVALID_DATA,
 342                                  _("Invalid byte sequence in conversion input"));
 343           break;
 344
 345         default:
 346           g_set_error (error, G_IO_ERROR, G_IO_ERROR_FAILED,
 347                        _("Error during conversion: %s"),
 348                        g_strerror (errsv));
 349           break;
 350         }
 351       ret = G_CONVERTER_ERROR;
 352     }
 353   else
 354     {
 355     ok:
 356       ret = G_CONVERTER_CONVERTED;
 357
 358       if (reset &&
 359           (flags & G_CONVERTER_INPUT_AT_END))
 360         ret = G_CONVERTER_FINISHED;
 361       else if (reset &&
 362                (flags & G_CONVERTER_FLUSH))
 363         ret = G_CONVERTER_FLUSHED;
 364     }
 365
 366   return ret;
 367 }
 368
 369 /**
 370  * g_charset_converter_set_use_fallback:
 371  * @converter: a #GCharsetConverter
 372  * @use_fallback: %TRUE to use fallbacks
 373  *
 374  * Sets the #GCharsetConverter:use-fallback property.
 375  *
 376  * Since: 2.24
 377  */
 378 void
 379 g_charset_converter_set_use_fallback (GCharsetConverter *converter,
 380                                       gboolean           use_fallback)
 381 {
 382   use_fallback = !!use_fallback;
 383
 384   if (converter->use_fallback != use_fallback)
 385     {
 386       converter->use_fallback = use_fallback;
 387       g_object_notify (G_OBJECT (converter), "use-fallback");
 388     }
 389 }
 390
 391 /**
 392  * g_charset_converter_get_use_fallback:
 393  * @converter: a #GCharsetConverter
 394  *
 395  * Gets the #GCharsetConverter:use-fallback property.
 396  *
 397  * Returns: %TRUE if fallbacks are used by @converter
 398  *
 399  * Since: 2.24
 400  */
 401 gboolean
 402 g_charset_converter_get_use_fallback (GCharsetConverter *converter)
 403 {
 404   return converter->use_fallback;
 405 }
 406
 407 /**
 408  * g_charset_converter_get_num_fallbacks:
 409  * @converter: a #GCharsetConverter
 410  *
 411  * Gets the number of fallbacks that @converter has applied so far.
 412  *
 413  * Returns: the number of fallbacks that @converter has applied
 414  *
 415  * Since: 2.24
 416  */
 417 guint
 418 g_charset_converter_get_num_fallbacks (GCharsetConverter *converter)
 419 {
 420   return converter->n_fallback_errors;
 421 }
 422
 423 static void
 424 g_charset_converter_iface_init (GConverterIface *iface)
 425 {
 426   iface->convert = g_charset_converter_convert;
 427   iface->reset = g_charset_converter_reset;
 428 }
 429
 430 static gboolean
 431 g_charset_converter_initable_init (GInitable     *initable,
 432                                    GCancellable  *cancellable,
 433                                    GError       **error)
 434 {
 435   GCharsetConverter  *conv;
 436
 437   g_return_val_if_fail (G_IS_CHARSET_CONVERTER (initable), FALSE);
 438
 439   conv = G_CHARSET_CONVERTER (initable);
 440
 441   if (cancellable != NULL)
 442     {
 443       g_set_error_literal (error, G_IO_ERROR, G_IO_ERROR_NOT_SUPPORTED,
 444                            _("Cancellable initialization not supported"));
 445       return FALSE;
 446     }
 447
 448   conv->iconv = g_iconv_open (conv->to, conv->from);
 449
 450   if (conv->iconv == (GIConv)-1)
 451     {
 452       if (errno == EINVAL)
 453         g_set_error (error, G_IO_ERROR, G_IO_ERROR_NOT_SUPPORTED,
 454                      _("Conversion from character set '%s' to '%s' is not supported"),
 455                      conv->from, conv->to);
 456       else
 457         g_set_error (error, G_IO_ERROR, G_IO_ERROR_FAILED,
 458                      _("Could not open converter from '%s' to '%s'"),
 459                      conv->from, conv->to);
 460       return FALSE;
 461     }
 462
 463   return TRUE;
 464 }
 465
 466 static void
 467 g_charset_converter_initable_iface_init (GInitableIface *iface)
 468 {
 469   iface->init = g_charset_converter_initable_init;
 470 }