gchar *
gedit_convert_from_utf8 (const gchar *content,
gsize len,
const GeditEncoding *encoding,
gsize *new_len,
GError **error)
{
GError *conv_error = NULL;
gchar *converted_contents = NULL;
gsize bytes_written = 0;
gedit_debug (DEBUG_UTILS);
g_return_val_if_fail (content != NULL, NULL);
g_return_val_if_fail (g_utf8_validate (content, len, NULL), NULL);
g_return_val_if_fail (encoding != NULL, NULL);
if (len < 0)
len = strlen (content);
if (encoding == gedit_encoding_get_utf8 ())
return g_strndup (content, len);
converted_contents = g_convert (content,
len,
gedit_encoding_get_charset (encoding),
"UTF-8",
NULL,
&bytes_written,
&conv_error);
if (conv_error != NULL)
{
gedit_debug_message (DEBUG_UTILS, "Cannot convert from UTF-8 to %s",
gedit_encoding_get_charset (encoding));
if (converted_contents != NULL)
{
g_free (converted_contents);
converted_contents = NULL;
}
g_propagate_error (error, conv_error);
}
else
{
if (new_len != NULL)
*new_len = bytes_written;
}
return converted_contents;
}
static GCharsetConverter *
guess_encoding (GeditDocumentOutputStream *stream,
const void *inbuf,
gsize inbuf_size)
{
GCharsetConverter *conv = NULL;
if (inbuf == NULL || inbuf_size == 0)
{
stream->priv->is_utf8 = TRUE;
return NULL;
}
if (stream->priv->encodings != NULL &&
stream->priv->encodings->next == NULL)
{
stream->priv->use_first = TRUE;
}
/* We just check the first block */
while (TRUE)
{
const GeditEncoding *enc;
if (conv != NULL)
{
g_object_unref (conv);
conv = NULL;
}
/* We get an encoding from the list */
enc = get_encoding (stream);
/* if it is NULL we didn't guess anything */
if (enc == NULL)
{
break;
}
gedit_debug_message (DEBUG_UTILS, "trying charset: %s",
gedit_encoding_get_charset (stream->priv->current_encoding->data));
if (enc == gedit_encoding_get_utf8 ())
{
gsize remainder;
const gchar *end;
if (g_utf8_validate (inbuf, inbuf_size, &end) ||
stream->priv->use_first)
{
stream->priv->is_utf8 = TRUE;
break;
}
/* Check if the end is less than one char */
remainder = inbuf_size - (end - (gchar *)inbuf);
if (remainder < 6)
{
stream->priv->is_utf8 = TRUE;
break;
}
continue;
}
conv = g_charset_converter_new ("UTF-8",
gedit_encoding_get_charset (enc),
NULL);
/* If we tried all encodings we use the first one */
if (stream->priv->use_first)
{
break;
}
/* Try to convert */
if (try_convert (conv, inbuf, inbuf_size))
{
break;
}
}
if (conv != NULL)
{
g_converter_reset (G_CONVERTER (conv));
}
return conv;
}
static void
update_menu (GeditEncodingsComboBox *menu)
{
GtkListStore *store;
GtkTreeIter iter;
GSList *encodings, *l;
gchar *str;
const GeditEncoding *utf8_encoding;
const GeditEncoding *current_encoding;
store = menu->priv->store;
/* Unset the previous model */
g_signal_handler_block (menu, menu->priv->changed_id);
gtk_list_store_clear (store);
gtk_combo_box_set_model (GTK_COMBO_BOX (menu),
NULL);
utf8_encoding = gedit_encoding_get_utf8 ();
current_encoding = gedit_encoding_get_current ();
if (!menu->priv->save_mode)
{
gtk_list_store_append (store, &iter);
gtk_list_store_set (store, &iter,
NAME_COLUMN, _("Automatically Detected"),
ENCODING_COLUMN, NULL,
ADD_COLUMN, FALSE,
-1);
gtk_list_store_append (store, &iter);
gtk_list_store_set (store, &iter,
NAME_COLUMN, "",
ENCODING_COLUMN, NULL,
ADD_COLUMN, FALSE,
-1);
}
if (current_encoding != utf8_encoding)
str = gedit_encoding_to_string (utf8_encoding);
else
str = g_strdup_printf (_("Current Locale (%s)"),
gedit_encoding_get_charset (utf8_encoding));
gtk_list_store_append (store, &iter);
gtk_list_store_set (store, &iter,
NAME_COLUMN, str,
ENCODING_COLUMN, utf8_encoding,
ADD_COLUMN, FALSE,
-1);
g_free (str);
if ((utf8_encoding != current_encoding) &&
(current_encoding != NULL))
{
str = g_strdup_printf (_("Current Locale (%s)"),
gedit_encoding_get_charset (current_encoding));
gtk_list_store_append (store, &iter);
gtk_list_store_set (store, &iter,
NAME_COLUMN, str,
ENCODING_COLUMN, current_encoding,
ADD_COLUMN, FALSE,
-1);
g_free (str);
}
encodings = gedit_prefs_manager_get_shown_in_menu_encodings ();
for (l = encodings; l != NULL; l = g_slist_next (l))
{
const GeditEncoding *enc = (const GeditEncoding *)l->data;
if ((enc != current_encoding) &&
(enc != utf8_encoding) &&
(enc != NULL))
{
str = gedit_encoding_to_string (enc);
gtk_list_store_append (store, &iter);
gtk_list_store_set (store, &iter,
NAME_COLUMN, str,
ENCODING_COLUMN, enc,
ADD_COLUMN, FALSE,
-1);
g_free (str);
}
}
g_slist_free (encodings);
if (gedit_prefs_manager_shown_in_menu_encodings_can_set ())
{
gtk_list_store_append (store, &iter);
/* separator */
gtk_list_store_set (store, &iter,
NAME_COLUMN, "",
ENCODING_COLUMN, NULL,
ADD_COLUMN, FALSE,
-1);
gtk_list_store_append (store, &iter);
gtk_list_store_set (store, &iter,
NAME_COLUMN, _("Add or Remove..."),
ENCODING_COLUMN, NULL,
ADD_COLUMN, TRUE,
-1);
}
/* set the model back */
gtk_combo_box_set_model (GTK_COMBO_BOX (menu),
GTK_TREE_MODEL (menu->priv->store));
gtk_combo_box_set_active (GTK_COMBO_BOX (menu), 0);
g_signal_handler_unblock (menu, menu->priv->changed_id);
}
gchar *
gedit_convert_to_utf8 (const gchar *content,
gsize len,
const GeditEncoding **encoding,
gsize *new_len,
GError **error)
{
gedit_debug (DEBUG_UTILS);
g_return_val_if_fail (content != NULL, NULL);
g_return_val_if_fail (encoding != NULL, NULL);
if (len < 0)
len = strlen (content);
if (*encoding != NULL)
{
const gchar* charset;
charset = gedit_encoding_get_charset (*encoding);
g_return_val_if_fail (charset != NULL, NULL);
return gedit_convert_to_utf8_from_charset (content,
len,
charset,
new_len,
error);
}
else
{
/* Automatically detect the encoding used */
GSList *encodings;
GSList *start;
gchar *ret = NULL;
gedit_debug_message (DEBUG_UTILS,
"Automatically detect the encoding used");
encodings = gedit_prefs_manager_get_auto_detected_encodings ();
if (encodings == NULL)
{
gedit_debug_message (DEBUG_UTILS, "encodings == NULL");
if (g_utf8_validate (content, len, NULL))
{
gedit_debug_message (DEBUG_UTILS, "validate OK.");
if (new_len != NULL)
*new_len = len;
return g_strndup (content, len);
}
else
{
gedit_debug_message (DEBUG_UTILS, "validate failed.");
g_set_error (error, GEDIT_CONVERT_ERROR,
GEDIT_CONVERT_ERROR_AUTO_DETECTION_FAILED,
"gedit was not able to automatically determine "
"the encoding of the file you want to open.");
return NULL;
}
}
gedit_debug_message (DEBUG_UTILS, "encodings != NULL");
start = encodings;
while (encodings != NULL)
{
const GeditEncoding *enc;
const gchar *charset;
gchar *utf8_content;
enc = (const GeditEncoding *)encodings->data;
gedit_debug_message (DEBUG_UTILS, "Get charset");
charset = gedit_encoding_get_charset (enc);
g_return_val_if_fail (charset != NULL, NULL);
gedit_debug_message (DEBUG_UTILS, "Trying to convert %lu bytes of data from '%s' to 'UTF-8'.",
(unsigned long) len, charset);
utf8_content = gedit_convert_to_utf8_from_charset (content,
len,
charset,
new_len,
NULL);
if (utf8_content != NULL)
{
*encoding = enc;
ret = utf8_content;
break;
}
encodings = g_slist_next (encodings);
}
if (ret == NULL)
{
gedit_debug_message (DEBUG_UTILS, "gedit has not been able to automatically determine the encoding of "
"the file you want to open.");
g_set_error (error, GEDIT_CONVERT_ERROR,
GEDIT_CONVERT_ERROR_AUTO_DETECTION_FAILED,
"gedit was not able to automatically determine "
"the encoding of the file you want to open.");
}
g_slist_free (start);
return ret;
}
g_return_val_if_reached (NULL);
}
