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, write to the
  17  * Free Software Foundation, Inc., 59 Temple Place, Suite 330,
  18  * Boston, MA 02111-1307, USA.
  19  *
  20  * Author: Alexander Larsson <alexl@redhat.com>
  21  */
  22
  23 #include "config.h"
  24
  25 #include "gcharsetconverter.h"
  26
  27 #include <errno.h>
  28
  29 #include "ginitable.h"
  30 #include "gioerror.h"
  31 #include "glibintl.h"
  32
  33
  34 enum {
  35   PROP_0,
  36   PROP_FROM_CHARSET,
  37   PROP_TO_CHARSET,
  38   PROP_USE_FALLBACK
  39 };
  40
  41 /**
  42  * SECTION:gcharsetconverter
  43  * @short_description: Convert between charsets
  44  * @include: gio/gio.h
  45  *
  46  * #GCharsetConverter is an implementation of #GConverter based on
  47  * GIConv.
  48  */
  49
  50 static void g_charset_converter_iface_init          (GConverterIface *iface);
  51 static void g_charset_converter_initable_iface_init (GInitableIface  *iface);
  52
  53 /**
  54  * GCharsetConverter:
  55  *
  56  * Conversions between character sets.
  57  */
  58 struct _GCharsetConverter
  59 {
  60   GObject parent_instance;
  61
  62   char *from;
  63   char *to;
  64   GIConv iconv;
  65   gboolean use_fallback;
  66   guint n_fallback_errors;
  67 };
  68
  69 G_DEFINE_TYPE_WITH_CODE (GCharsetConverter, g_charset_converter, G_TYPE_OBJECT,
  70                          G_IMPLEMENT_INTERFACE (G_TYPE_CONVERTER,
  71                                                 g_charset_converter_iface_init);
  72                          G_IMPLEMENT_INTERFACE (G_TYPE_INITABLE,
  73                                                 g_charset_converter_initable_iface_init))
  74
  75 static void
  76 g_charset_converter_finalize (GObject *object)
  77 {
  78   GCharsetConverter *conv;
  79
  80   conv = G_CHARSET_CONVERTER (object);
  81
  82   g_free (conv->from);
  83   g_free (conv->to);
  84   if (conv->iconv)
  85     g_iconv_close (conv->iconv);
  86
  87   G_OBJECT_CLASS (g_charset_converter_parent_class)->finalize (object);
  88 }
  89
  90 static void
  91 g_charset_converter_set_property (GObject      *object,
  92                                   guint         prop_id,
  93                                   const GValue *value,
  94                                   GParamSpec   *pspec)
  95 {
  96   GCharsetConverter *conv;
  97
  98   conv = G_CHARSET_CONVERTER (object);
  99
 100   switch (prop_id)
 101     {
 102     case PROP_TO_CHARSET:
 103       g_free (conv->to);
 104       conv->to = g_value_dup_string (value);
 105       break;
 106
 107     case PROP_FROM_CHARSET:
 108       g_free (conv->from);
 109       conv->from = g_value_dup_string (value);
 110       break;
 111
 112     case PROP_USE_FALLBACK:
 113       conv->use_fallback = g_value_get_boolean (value);
 114       break;
 115
 116     default:
 117       G_OBJECT_WARN_INVALID_PROPERTY_ID (object, prop_id, pspec);
 118       break;
 119     }
 120
 121 }
 122
 123 static void
 124 g_charset_converter_get_property (GObject    *object,
 125                                   guint       prop_id,
 126                                   GValue     *value,
 127                                   GParamSpec *pspec)
 128 {
 129   GCharsetConverter *conv;
 130
 131   conv = G_CHARSET_CONVERTER (object);
 132
 133   switch (prop_id)
 134     {
 135     case PROP_TO_CHARSET:
 136       g_value_set_string (value, conv->to);
 137       break;
 138
 139     case PROP_FROM_CHARSET:
 140       g_value_set_string (value, conv->from);
 141       break;
 142
 143     case PROP_USE_FALLBACK:
 144       g_value_set_boolean (value, conv->use_fallback);
 145       break;
 146
 147     default:
 148       G_OBJECT_WARN_INVALID_PROPERTY_ID (object, prop_id, pspec);
 149       break;
 150     }
 151 }
 152
 153 static void
 154 g_charset_converter_class_init (GCharsetConverterClass *klass)
 155 {
 156   GObjectClass *gobject_class = G_OBJECT_CLASS (klass);
 157
 158   gobject_class->finalize = g_charset_converter_finalize;
 159   gobject_class->get_property = g_charset_converter_get_property;
 160   gobject_class->set_property = g_charset_converter_set_property;
 161
 162   g_object_class_install_property (gobject_class,
 163                                    PROP_TO_CHARSET,
 164                                    g_param_spec_string ("to-charset",
 165                                                         P_("To Charset"),
 166                                                         P_("The character encoding to convert to"),
 167                                                         NULL,
 168                                                         G_PARAM_READWRITE | G_PARAM_CONSTRUCT_ONLY |
 169                                                         G_PARAM_STATIC_STRINGS));
 170   g_object_class_install_property (gobject_class,
 171                                    PROP_FROM_CHARSET,
 172                                    g_param_spec_string ("from-charset",
 173                                                         P_("From Charset"),
 174                                                         P_("The character encoding to convert from"),
 175                                                         NULL,
 176                                                         G_PARAM_READWRITE | G_PARAM_CONSTRUCT_ONLY |
 177                                                         G_PARAM_STATIC_STRINGS));
 178   g_object_class_install_property (gobject_class,
 179                                    PROP_USE_FALLBACK,
 180                                    g_param_spec_boolean ("use-fallback",
 181                                                          P_("Fallback enabled"),
 182                                                          P_("Use fallback (of form \\<hexval>) for invalid bytes"),
 183                                                          FALSE,
 184                                                          G_PARAM_READWRITE |
 185                                                          G_PARAM_CONSTRUCT |
 186                                                          G_PARAM_STATIC_STRINGS));
 187 }
 188
 189 static void
 190 g_charset_converter_init (GCharsetConverter *local)
 191 {
 192 }
 193
 194
 195 /**
 196  * g_charset_converter_new:
 197  * @to_charset: destination charset
 198  * @from_charset: source charset
 199  * @error: #GError for error reporting, or %NULL to ignore.
 200  *
 201  * Creates a new #GCharsetConverter.
 202  *
 203  * Returns: a new #GCharsetConverter or %NULL on error.
 204  *
 205  * Since: 2.24
 206  **/
 207 GCharsetConverter *
 208 g_charset_converter_new (const gchar  *to_charset,
 209                          const gchar  *from_charset,
 210                          GError       **error)
 211 {
 212   GCharsetConverter *conv;
 213
 214   conv = g_initable_new (G_TYPE_CHARSET_CONVERTER,
 215                          NULL, error,
 216                          "to-charset", to_charset,
 217                          "from-charset", from_charset,
 218                          NULL);
 219
 220   return conv;
 221 }
 222
 223 static void
 224 g_charset_converter_reset (GConverter *converter)
 225 {
 226   GCharsetConverter *conv = G_CHARSET_CONVERTER (converter);
 227
 228   if (conv->iconv == NULL)
 229     {
 230       g_warning ("Invalid object, not initialized");
 231       return;
 232     }
 233
 234   g_iconv (conv->iconv, NULL, NULL, NULL, NULL);
 235   conv->n_fallback_errors = 0;
 236 }
 237
 238 static GConverterResult
 239 g_charset_converter_convert (GConverter *converter,
 240                              const void *inbuf,
 241                              gsize       inbuf_size,
 242                              void       *outbuf,
 243                              gsize       outbuf_size,
 244                              GConverterFlags flags,
 245                              gsize      *bytes_read,
 246                              gsize      *bytes_written,
 247                              GError    **error)
 248 {
 249   GCharsetConverter  *conv;
 250   gsize res;
 251   GConverterResult ret;
 252   gchar *inbufp, *outbufp;
 253   gsize in_left, out_left;
 254   int errsv;
 255   gboolean reset;
 256
 257   conv = G_CHARSET_CONVERTER (converter);
 258
 259   if (conv->iconv == NULL)
 260     {
 261       g_set_error_literal (error, G_IO_ERROR, G_IO_ERROR_NOT_INITIALIZED,
 262                            _("Invalid object, not initialized"));
 263       return G_CONVERTER_ERROR;
 264     }
 265
 266   inbufp = (char *)inbuf;
 267   outbufp = (char *)outbuf;
 268   in_left = inbuf_size;
 269   out_left = outbuf_size;
 270   reset = FALSE;
 271
 272   /* if there is not input try to flush the data */
 273   if (inbuf_size == 0)
 274     {
 275       if (flags & G_CONVERTER_INPUT_AT_END ||
 276           flags & G_CONVERTER_FLUSH)
 277         {
 278           reset = TRUE;
 279         }
 280       else
 281         {
 282           g_set_error_literal (error, G_IO_ERROR, G_IO_ERROR_PARTIAL_INPUT,
 283                                _("Incomplete multibyte sequence in input"));
 284           return G_CONVERTER_ERROR;
 285         }
 286     }
 287
 288   if (reset)
 289     /* call g_iconv with NULL inbuf to cleanup shift state */
 290     res = g_iconv (conv->iconv,
 291                    NULL, &in_left,
 292                    &outbufp, &out_left);
 293   else
 294     res = g_iconv (conv->iconv,
 295                    &inbufp, &in_left,
 296                    &outbufp, &out_left);
 297
 298   *bytes_read = inbufp - (char *)inbuf;
 299   *bytes_written = outbufp - (char *)outbuf;
 300
 301   /* Don't report error if we converted anything */
 302   if (res == (gsize) -1 && *bytes_read == 0)
 303     {
 304       errsv = errno;
 305
 306       switch (errsv)
 307         {
 308         case EINVAL:
 309           /* Incomplete input text */
 310           g_set_error_literal (error, G_IO_ERROR, G_IO_ERROR_PARTIAL_INPUT,
 311                                _("Incomplete multibyte sequence in input"));
 312           break;
 313
 314         case E2BIG:
 315           /* Not enough destination space */
 316           g_set_error_literal (error, G_IO_ERROR, G_IO_ERROR_NO_SPACE,
 317                                _("Not enough space in destination"));
 318           break;
 319
 320         case EILSEQ:
 321           /* Invalid code sequence */
 322           if (conv->use_fallback)
 323             {
 324               if (outbuf_size < 3)
 325                 g_set_error_literal (error, G_IO_ERROR, G_IO_ERROR_NO_SPACE,
 326                                      _("Not enough space in destination"));
 327               else
 328                 {
 329                   const char hex[] = "0123456789ABCDEF";
 330                   guint8 v = *(guint8 *)inbuf;
 331                   guint8 *out = (guint8 *)outbuf;
 332                   out[0] = '\\';
 333                   out[1] = hex[(v & 0xf0) >> 4];
 334                   out[2] = hex[(v & 0x0f) >> 0];
 335                   *bytes_read = 1;
 336                   *bytes_written = 3;
 337                   in_left--;
 338                   conv->n_fallback_errors++;
 339                   goto ok;
 340                 }
 341             }
 342           else
 343             g_set_error_literal (error, G_IO_ERROR, G_IO_ERROR_INVALID_DATA,
 344                                  _("Invalid byte sequence in conversion input"));
 345           break;
 346
 347         default:
 348           g_set_error (error, G_IO_ERROR, G_IO_ERROR_FAILED,
 349                        _("Error during conversion: %s"),
 350                        g_strerror (errsv));
 351           break;
 352         }
 353       ret = G_CONVERTER_ERROR;
 354     }
 355   else
 356     {
 357     ok:
 358       ret = G_CONVERTER_CONVERTED;
 359
 360       if (reset &&
 361           (flags & G_CONVERTER_INPUT_AT_END))
 362         ret = G_CONVERTER_FINISHED;
 363       else if (reset &&
 364                (flags & G_CONVERTER_FLUSH))
 365         ret = G_CONVERTER_FLUSHED;
 366     }
 367
 368   return ret;
 369 }
 370
 371 /**
 372  * g_charset_converter_set_use_fallback:
 373  * @converter: a #GCharsetConverter
 374  * @use_fallback: %TRUE to use fallbacks
 375  *
 376  * Sets the #GCharsetConverter:use-fallback property.
 377  *
 378  * Since: 2.24
 379  */
 380 void
 381 g_charset_converter_set_use_fallback (GCharsetConverter *converter,
 382                                       gboolean           use_fallback)
 383 {
 384   use_fallback = !!use_fallback;
 385
 386   if (converter->use_fallback != use_fallback)
 387     {
 388       converter->use_fallback = use_fallback;
 389       g_object_notify (G_OBJECT (converter), "use-fallback");
 390     }
 391 }
 392
 393 /**
 394  * g_charset_converter_get_use_fallback:
 395  * @converter: a #GCharsetConverter
 396  *
 397  * Gets the #GCharsetConverter:use-fallback property.
 398  *
 399  * Returns: %TRUE if fallbacks are used by @converter
 400  *
 401  * Since: 2.24
 402  */
 403 gboolean
 404 g_charset_converter_get_use_fallback (GCharsetConverter *converter)
 405 {
 406   return converter->use_fallback;
 407 }
 408
 409 /**
 410  * g_charset_converter_get_num_fallbacks:
 411  * @converter: a #GCharsetConverter
 412  *
 413  * Gets the number of fallbacks that @converter has applied so far.
 414  *
 415  * Returns: the number of fallbacks that @converter has applied
 416  *
 417  * Since: 2.24
 418  */
 419 guint
 420 g_charset_converter_get_num_fallbacks (GCharsetConverter *converter)
 421 {
 422   return converter->n_fallback_errors;
 423 }
 424
 425 static void
 426 g_charset_converter_iface_init (GConverterIface *iface)
 427 {
 428   iface->convert = g_charset_converter_convert;
 429   iface->reset = g_charset_converter_reset;
 430 }
 431
 432 static gboolean
 433 g_charset_converter_initable_init (GInitable *initable,
 434                                    GCancellable *cancellable,
 435                                    GError  **error)
 436 {
 437   GCharsetConverter  *conv;
 438
 439   g_return_val_if_fail (G_IS_CHARSET_CONVERTER (initable), FALSE);
 440
 441   conv = G_CHARSET_CONVERTER (initable);
 442
 443   if (cancellable != NULL)
 444     {
 445       g_set_error_literal (error, G_IO_ERROR, G_IO_ERROR_NOT_SUPPORTED,
 446                            _("Cancellable initialization not supported"));
 447       return FALSE;
 448     }
 449
 450   conv->iconv =
 451     g_iconv_open (conv->to, conv->from);
 452
 453   if (conv->iconv == (GIConv)-1)
 454     {
 455       if (errno == EINVAL)
 456         g_set_error (error, G_IO_ERROR, G_IO_ERROR_NOT_SUPPORTED,
 457                      _("Conversion from character set '%s' to '%s' is not supported"),
 458                      conv->from, conv->to);
 459       else
 460         g_set_error (error, G_IO_ERROR, G_IO_ERROR_FAILED,
 461                      _("Could not open converter from '%s' to '%s'"),
 462                      conv->from, conv->to);
 463       return FALSE;
 464     }
 465
 466   return TRUE;
 467 }
 468
 469 static void
 470 g_charset_converter_initable_iface_init (GInitableIface *iface)
 471 {
 472   iface->init = g_charset_converter_initable_init;
 473 }