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 "gcontenttypeprivate.h"
  30 #include "ginitable.h"
  31 #include "gioerror.h"
  32 #include "glibintl.h"
  33
  34
  35 enum {
  36   PROP_0,
  37   PROP_FROM_CHARSET,
  38   PROP_TO_CHARSET,
  39   PROP_USE_FALLBACK
  40 };
  41
  42 /**
  43  * SECTION:gcharsetconverter
  44  * @short_description: Convert between charsets
  45  * @include: gio/gio.h
  46  *
  47  * #GCharsetConverter is an implementation of #GConverter based on
  48  * GIConv.
  49  */
  50
  51 static void g_charset_converter_iface_init          (GConverterIface *iface);
  52 static void g_charset_converter_initable_iface_init (GInitableIface  *iface);
  53
  54 /**
  55  * GCharsetConverter:
  56  *
  57  * Conversions between character sets.
  58  */
  59 struct _GCharsetConverter
  60 {
  61   GObject parent_instance;
  62
  63   char *from;
  64   char *to;
  65   GIConv iconv;
  66   gboolean use_fallback;
  67   guint n_fallback_errors;
  68 };
  69
  70 G_DEFINE_TYPE_WITH_CODE (GCharsetConverter, g_charset_converter, G_TYPE_OBJECT,
  71                          G_IMPLEMENT_INTERFACE (G_TYPE_CONVERTER,
  72                                                 g_charset_converter_iface_init);
  73                          G_IMPLEMENT_INTERFACE (G_TYPE_INITABLE,
  74                                                 g_charset_converter_initable_iface_init))
  75
  76 static void
  77 g_charset_converter_finalize (GObject *object)
  78 {
  79   GCharsetConverter *conv;
  80
  81   conv = G_CHARSET_CONVERTER (object);
  82
  83   g_free (conv->from);
  84   g_free (conv->to);
  85   if (conv->iconv)
  86     g_iconv_close (conv->iconv);
  87
  88   G_OBJECT_CLASS (g_charset_converter_parent_class)->finalize (object);
  89 }
  90
  91 static void
  92 g_charset_converter_set_property (GObject      *object,
  93                                   guint         prop_id,
  94                                   const GValue *value,
  95                                   GParamSpec   *pspec)
  96 {
  97   GCharsetConverter *conv;
  98
  99   conv = G_CHARSET_CONVERTER (object);
 100
 101   switch (prop_id)
 102     {
 103     case PROP_TO_CHARSET:
 104       g_free (conv->to);
 105       conv->to = g_value_dup_string (value);
 106       break;
 107
 108     case PROP_FROM_CHARSET:
 109       g_free (conv->from);
 110       conv->from = g_value_dup_string (value);
 111       break;
 112
 113     case PROP_USE_FALLBACK:
 114       conv->use_fallback = g_value_get_boolean (value);
 115       break;
 116
 117     default:
 118       G_OBJECT_WARN_INVALID_PROPERTY_ID (object, prop_id, pspec);
 119       break;
 120     }
 121
 122 }
 123
 124 static void
 125 g_charset_converter_get_property (GObject    *object,
 126                                   guint       prop_id,
 127                                   GValue     *value,
 128                                   GParamSpec *pspec)
 129 {
 130   GCharsetConverter *conv;
 131
 132   conv = G_CHARSET_CONVERTER (object);
 133
 134   switch (prop_id)
 135     {
 136     case PROP_TO_CHARSET:
 137       g_value_set_string (value, conv->to);
 138       break;
 139
 140     case PROP_FROM_CHARSET:
 141       g_value_set_string (value, conv->from);
 142       break;
 143
 144     case PROP_USE_FALLBACK:
 145       g_value_set_boolean (value, conv->use_fallback);
 146       break;
 147
 148     default:
 149       G_OBJECT_WARN_INVALID_PROPERTY_ID (object, prop_id, pspec);
 150       break;
 151     }
 152 }
 153
 154 static void
 155 g_charset_converter_class_init (GCharsetConverterClass *klass)
 156 {
 157   GObjectClass *gobject_class = G_OBJECT_CLASS (klass);
 158
 159   gobject_class->finalize = g_charset_converter_finalize;
 160   gobject_class->get_property = g_charset_converter_get_property;
 161   gobject_class->set_property = g_charset_converter_set_property;
 162
 163   g_object_class_install_property (gobject_class,
 164                                    PROP_TO_CHARSET,
 165                                    g_param_spec_string ("to-charset",
 166                                                         P_("To Charset"),
 167                                                         P_("The character encoding to convert to"),
 168                                                         NULL,
 169                                                         G_PARAM_READWRITE | G_PARAM_CONSTRUCT_ONLY |
 170                                                         G_PARAM_STATIC_STRINGS));
 171   g_object_class_install_property (gobject_class,
 172                                    PROP_FROM_CHARSET,
 173                                    g_param_spec_string ("from-charset",
 174                                                         P_("From Charset"),
 175                                                         P_("The character encoding to convert from"),
 176                                                         NULL,
 177                                                         G_PARAM_READWRITE | G_PARAM_CONSTRUCT_ONLY |
 178                                                         G_PARAM_STATIC_STRINGS));
 179   g_object_class_install_property (gobject_class,
 180                                    PROP_USE_FALLBACK,
 181                                    g_param_spec_boolean ("use-fallback",
 182                                                          P_("Fallback enabled"),
 183                                                          P_("Use fallback (of form \\<hexval>) for invalid bytes"),
 184                                                          FALSE,
 185                                                          G_PARAM_READWRITE |
 186                                                          G_PARAM_CONSTRUCT |
 187                                                          G_PARAM_STATIC_STRINGS));
 188 }
 189
 190 static void
 191 g_charset_converter_init (GCharsetConverter *local)
 192 {
 193 }
 194
 195
 196 /**
 197  * g_charset_converter_new:
 198  * @to_charset: destination charset
 199  * @from_charset: source charset
 200  * @error: #GError for error reporting, or %NULL to ignore.
 201  *
 202  * Creates a new #GCharsetConverter.
 203  *
 204  * Returns: a new #GCharsetConverter or %NULL on error.
 205  *
 206  * Since: 2.24
 207  **/
 208 GCharsetConverter *
 209 g_charset_converter_new (const gchar  *to_charset,
 210                          const gchar  *from_charset,
 211                          GError       **error)
 212 {
 213   GCharsetConverter *conv;
 214
 215   conv = g_initable_new (G_TYPE_CHARSET_CONVERTER,
 216                          NULL, error,
 217                          "to-charset", to_charset,
 218                          "from-charset", from_charset,
 219                          NULL);
 220
 221   return conv;
 222 }
 223
 224 static void
 225 g_charset_converter_reset (GConverter *converter)
 226 {
 227   GCharsetConverter *conv = G_CHARSET_CONVERTER (converter);
 228
 229   if (conv->iconv == NULL)
 230     {
 231       g_warning ("Invalid object, not initialized");
 232       return;
 233     }
 234
 235   g_iconv (conv->iconv, NULL, NULL, NULL, NULL);
 236   conv->n_fallback_errors = 0;
 237 }
 238
 239 static GConverterResult
 240 g_charset_converter_convert (GConverter *converter,
 241                              const void *inbuf,
 242                              gsize       inbuf_size,
 243                              void       *outbuf,
 244                              gsize       outbuf_size,
 245                              GConverterFlags flags,
 246                              gsize      *bytes_read,
 247                              gsize      *bytes_written,
 248                              GError    **error)
 249 {
 250   GCharsetConverter  *conv;
 251   gsize res;
 252   GConverterResult ret;
 253   gchar *inbufp, *outbufp;
 254   gsize in_left, out_left;
 255   int errsv;
 256   gboolean reset;
 257
 258   conv = G_CHARSET_CONVERTER (converter);
 259
 260   if (conv->iconv == NULL)
 261     {
 262       g_set_error_literal (error, G_IO_ERROR, G_IO_ERROR_NOT_INITIALIZED,
 263                            _("Invalid object, not initialized"));
 264       return G_CONVERTER_ERROR;
 265     }
 266
 267   inbufp = (char *)inbuf;
 268   outbufp = (char *)outbuf;
 269   in_left = inbuf_size;
 270   out_left = outbuf_size;
 271   reset = FALSE;
 272
 273   /* if there is not input try to flush the data */
 274   if (inbuf_size == 0)
 275     {
 276       if (flags & G_CONVERTER_INPUT_AT_END ||
 277           flags & G_CONVERTER_FLUSH)
 278         {
 279           reset = TRUE;
 280         }
 281       else
 282         {
 283           g_set_error_literal (error, G_IO_ERROR, G_IO_ERROR_PARTIAL_INPUT,
 284                                _("Incomplete multibyte sequence in input"));
 285           return G_CONVERTER_ERROR;
 286         }
 287     }
 288
 289   if (reset)
 290     /* call g_iconv with NULL inbuf to cleanup shift state */
 291     res = g_iconv (conv->iconv,
 292                    NULL, &in_left,
 293                    &outbufp, &out_left);
 294   else
 295     res = g_iconv (conv->iconv,
 296                    &inbufp, &in_left,
 297                    &outbufp, &out_left);
 298
 299   *bytes_read = inbufp - (char *)inbuf;
 300   *bytes_written = outbufp - (char *)outbuf;
 301
 302   /* Don't report error if we converted anything */
 303   if (res == (gsize) -1 && *bytes_read == 0)
 304     {
 305       errsv = errno;
 306
 307       switch (errsv)
 308         {
 309         case EINVAL:
 310           /* Incomplete input text */
 311           g_set_error_literal (error, G_IO_ERROR, G_IO_ERROR_PARTIAL_INPUT,
 312                                _("Incomplete multibyte sequence in input"));
 313           break;
 314
 315         case E2BIG:
 316           /* Not enough destination space */
 317           g_set_error_literal (error, G_IO_ERROR, G_IO_ERROR_NO_SPACE,
 318                                _("Not enough space in destination"));
 319           break;
 320
 321         case EILSEQ:
 322           /* Invalid code sequence */
 323           if (conv->use_fallback)
 324             {
 325               if (outbuf_size < 3)
 326                 g_set_error_literal (error, G_IO_ERROR, G_IO_ERROR_NO_SPACE,
 327                                      _("Not enough space in destination"));
 328               else
 329                 {
 330                   const char hex[] = "0123456789ABCDEF";
 331                   guint8 v = *(guint8 *)inbuf;
 332                   guint8 *out = (guint8 *)outbuf;
 333                   out[0] = '\\';
 334                   out[1] = hex[(v & 0xf0) >> 4];
 335                   out[2] = hex[(v & 0x0f) >> 0];
 336                   *bytes_read = 1;
 337                   *bytes_written = 3;
 338                   in_left--;
 339                   conv->n_fallback_errors++;
 340                   goto ok;
 341                 }
 342             }
 343           else
 344             g_set_error_literal (error, G_IO_ERROR, G_IO_ERROR_INVALID_DATA,
 345                                  _("Invalid byte sequence in conversion input"));
 346           break;
 347
 348         default:
 349           g_set_error (error, G_IO_ERROR, G_IO_ERROR_FAILED,
 350                        _("Error during conversion: %s"),
 351                        g_strerror (errsv));
 352           break;
 353         }
 354       ret = G_CONVERTER_ERROR;
 355     }
 356   else
 357     {
 358     ok:
 359       ret = G_CONVERTER_CONVERTED;
 360
 361       if (reset &&
 362           (flags & G_CONVERTER_INPUT_AT_END))
 363         ret = G_CONVERTER_FINISHED;
 364       else if (reset &&
 365                (flags & G_CONVERTER_FLUSH))
 366         ret = G_CONVERTER_FLUSHED;
 367     }
 368
 369   return ret;
 370 }
 371
 372 /**
 373  * g_charset_converter_set_use_fallback:
 374  * @converter: a #GCharsetConverter
 375  * @use_fallback: %TRUE to use fallbacks
 376  *
 377  * Sets the #GCharsetConverter:use-fallback property.
 378  *
 379  * Since: 2.24
 380  */
 381 void
 382 g_charset_converter_set_use_fallback (GCharsetConverter *converter,
 383                                       gboolean           use_fallback)
 384 {
 385   use_fallback = !!use_fallback;
 386
 387   if (converter->use_fallback != use_fallback)
 388     {
 389       converter->use_fallback = use_fallback;
 390       g_object_notify (G_OBJECT (converter), "use-fallback");
 391     }
 392 }
 393
 394 /**
 395  * g_charset_converter_get_use_fallback:
 396  * @converter: a #GCharsetConverter
 397  *
 398  * Gets the #GCharsetConverter:use-fallback property.
 399  *
 400  * Returns: %TRUE if fallbacks are used by @converter
 401  *
 402  * Since: 2.24
 403  */
 404 gboolean
 405 g_charset_converter_get_use_fallback (GCharsetConverter *converter)
 406 {
 407   return converter->use_fallback;
 408 }
 409
 410 /**
 411  * g_charset_converter_get_num_fallbacks:
 412  * @converter: a #GCharsetConverter
 413  *
 414  * Gets the number of fallbacks that @converter has applied so far.
 415  *
 416  * Returns: the number of fallbacks that @converter has applied
 417  *
 418  * Since: 2.24
 419  */
 420 guint
 421 g_charset_converter_get_num_fallbacks (GCharsetConverter *converter)
 422 {
 423   return converter->n_fallback_errors;
 424 }
 425
 426 static void
 427 g_charset_converter_iface_init (GConverterIface *iface)
 428 {
 429   iface->convert = g_charset_converter_convert;
 430   iface->reset = g_charset_converter_reset;
 431 }
 432
 433 static gboolean
 434 g_charset_converter_initable_init (GInitable *initable,
 435                                    GCancellable *cancellable,
 436                                    GError  **error)
 437 {
 438   GCharsetConverter  *conv;
 439
 440   g_return_val_if_fail (G_IS_CHARSET_CONVERTER (initable), FALSE);
 441
 442   conv = G_CHARSET_CONVERTER (initable);
 443
 444   if (cancellable != NULL)
 445     {
 446       g_set_error_literal (error, G_IO_ERROR, G_IO_ERROR_NOT_SUPPORTED,
 447                            _("Cancellable initialization not supported"));
 448       return FALSE;
 449     }
 450
 451   conv->iconv =
 452     g_iconv_open (conv->to, conv->from);
 453
 454   if (conv->iconv == (GIConv)-1)
 455     {
 456       if (errno == EINVAL)
 457         g_set_error (error, G_IO_ERROR, G_IO_ERROR_NOT_SUPPORTED,
 458                      _("Conversion from character set '%s' to '%s' is not supported"),
 459                      conv->from, conv->to);
 460       else
 461         g_set_error (error, G_IO_ERROR, G_IO_ERROR_FAILED,
 462                      _("Could not open converter from '%s' to '%s'"),
 463                      conv->from, conv->to);
 464       return FALSE;
 465     }
 466
 467   return TRUE;
 468 }
 469
 470 static void
 471 g_charset_converter_initable_iface_init (GInitableIface *iface)
 472 {
 473   iface->init = g_charset_converter_initable_init;
 474 }