Convert TAG (program name) from source encoding to locale encoding if source

encoding is specified and other than locale encoding.

4 files changed, 103 insertions, 39 deletions

diff --git a/daemon/conf.c b/daemon/conf.c
index 2255369..78f95c2 100644
--- a/daemon/conf.c
+++ b/daemon/conf.c
@@ -676,6 +676,64 @@ parsed:
 }
 
 /******************************************************************************
+ * create_conversion_descriptors
+ *
+ * helper function for resolve_logpaths;
+ * creates charset conversion descriptors for message and program name
+ */
+static void create_conversion_descriptors( struct logpath* logpath )
+{
+    /* create message charset conversion descriptor */
+    if( logpath->source->encoding || logpath->destination->encoding )
+    {
+        char *source_encoding = logpath->source->encoding?
+                                    logpath->source->encoding : "UTF-8";
+        char *destination_encoding = logpath->destination->encoding?
+                                    logpath->destination->encoding : "UTF-8";
+
+        if( strcasecmp( source_encoding, destination_encoding ) != 0 )
+        {
+            logpath->message_cd = g_iconv_open( destination_encoding, source_encoding );
+            if( logpath->message_cd == (GIConv) -1 )
+            {
+                ERR( "Cannot convert messages from %s to %s\n",
+                     source_encoding, destination_encoding );
+            }
+            else
+            {
+                TRACE( "Log path %s-%s: converting messages from %s to %s\n",
+                       logpath->source->name, logpath->destination->name,
+                       source_encoding, destination_encoding );
+            }
+        }
+    }
+
+    /* create charset conversion descriptor for program name */
+    if( logpath->source->encoding )
+    {
+        const gchar *destination_encoding;
+
+        g_get_charset( &destination_encoding );
+
+        if( strcasecmp( logpath->source->encoding, destination_encoding ) != 0 )
+        {
+            logpath->program_cd = g_iconv_open( destination_encoding, logpath->source->encoding );
+            if( logpath->program_cd == (GIConv) -1 )
+            {
+                ERR( "Cannot convert program names from %s to %s\n",
+                     logpath->source->encoding, destination_encoding );
+            }
+            else
+            {
+                TRACE( "Log path %s-%s: converting program name from %s to %s\n",
+                       logpath->source->name, logpath->destination->name,
+                       logpath->source->encoding, destination_encoding );
+            }
+        }
+    }
+}
+
+/******************************************************************************
  * resolve_logpaths
  *
  * replace logpath_names structures
@@ -695,6 +753,7 @@ static void resolve_logpaths()
         logpath = g_malloc( sizeof(struct logpath) );
 
         logpath->message_cd = (GIConv) -1;
+        logpath->program_cd = (GIConv) -1;
 
         /* find source */
         for( item = sources; item; item = item->next )
@@ -741,30 +800,7 @@ static void resolve_logpaths()
                 logpath->filter = NULL;
         }
 
-        /* create message charset conversion descriptor */
-        if( logpath->source->encoding || logpath->destination->encoding )
-        {
-            char *source_encoding = logpath->source->encoding?
-                                        logpath->source->encoding : "UTF-8";
-            char *destination_encoding = logpath->destination->encoding?
-                                        logpath->destination->encoding : "UTF-8";
-
-            if( strcasecmp( source_encoding, destination_encoding ) != 0 )
-            {
-                logpath->message_cd = g_iconv_open( destination_encoding, source_encoding );
-                if( logpath->message_cd == (GIConv) -1 )
-                {
-                    ERR( "Cannot convert messages from %s to %s\n",
-                         source_encoding, destination_encoding );
-                }
-                else
-                {
-                    TRACE( "Log path %s-%s: converting messages from %s to %s\n",
-                           logpath->source->name, logpath->destination->name,
-                           source_encoding, destination_encoding );
-                }
-            }
-        }
+        create_conversion_descriptors( logpath );
 
         /* add item to paths */
         paths = g_list_append( paths, logpath );
diff --git a/daemon/syslogd.c b/daemon/syslogd.c
index e4467b3..0a3b2b6 100644
--- a/daemon/syslogd.c
+++ b/daemon/syslogd.c
@@ -180,6 +180,30 @@ done:
 }
 
 /******************************************************************************
+ * try_convert_string
+ *
+ * helper function for mux_message;
+ * try to convert encoding, return reference to the original string in case
+ * of failure
+ */
+static struct string* try_convert_string( struct string* s, GIConv cd )
+{
+    struct string *ret;
+    gchar *converted_str;
+            
+    converted_str = g_convert_with_iconv( s->gstr->str, -1, cd, NULL, NULL, NULL );
+    if( !converted_str )
+    {
+        TRACE( "conversion error\n" );
+        return string_addref( s );
+    }
+
+    ret = string_new( converted_str );
+    g_free( converted_str );
+    return ret;
+}
+
+/******************************************************************************
  * mux_message
  *
  * filter and multiplex message to destinations;
@@ -194,6 +218,7 @@ static void mux_message( struct message* msg )
     for( item = logpaths; item; item = item->next )
     {
         struct logpath *logpath = item->data;
+        gboolean need_message_iconv, need_program_iconv;
         struct message *converted_msg;
 
         if( logpath->source != msg->source )
@@ -202,27 +227,29 @@ static void mux_message( struct message* msg )
         if( !filter_message( msg, logpath->filter ) )
             continue;
 
-        /* convert message encoding if needed */
-        if( logpath->message_cd == (GIConv) -1 )
+        /* convert encoding if needed */
+        need_message_iconv = logpath->message_cd != (GIConv) -1;
+        need_program_iconv = logpath->program_cd != (GIConv) -1;
+
+        if( (!need_message_iconv) && (!need_program_iconv) )
         {
             converted_msg = msg;
             reference_message( msg );
         }
         else
         {
-            gchar *c_message = g_convert_with_iconv( msg->message->gstr->str, -1,
-                                                     logpath->message_cd,
-                                                     NULL, NULL, NULL );
-            if( !c_message )
+            converted_msg = duplicate_message( msg );
+
+            if( need_message_iconv )
             {
-                TRACE( "conversion error\n" );
-                c_message = g_strdup( msg->message->gstr->str );
+                string_release( converted_msg->message );
+                converted_msg->message = try_convert_string( msg->message, logpath->message_cd );
+            }
+            if( need_program_iconv )
+            {
+                string_release( converted_msg->program );
+                converted_msg->program = try_convert_string( msg->program, logpath->program_cd );
             }
-
-            converted_msg = duplicate_message( msg );
-            string_release( converted_msg->message );
-            converted_msg->message = string_new( c_message );
-            g_free( c_message );
         }
 
         /* put message to destination */
diff --git a/daemon/syslogd.h b/daemon/syslogd.h
index d2bb056..97b02ea 100644
--- a/daemon/syslogd.h
+++ b/daemon/syslogd.h
@@ -197,7 +197,8 @@ struct logpath
     struct source *source;
     struct filter *filter;
     struct destination *destination;
-    GIConv message_cd;
+    GIConv message_cd;  /* charset conversion descriptor for message */
+    GIConv program_cd;  /* charset conversion descriptor for program name */
 };
 
 extern GList *sources;
diff --git a/doc/src/configuration.xml b/doc/src/configuration.xml
index dc99510..f88059b 100644
--- a/doc/src/configuration.xml
+++ b/doc/src/configuration.xml
@@ -172,7 +172,7 @@ It may contain the following format characters:
 	<seglistitem><seg>%l</seg><seg>priority level in numeric form</seg></seglistitem>
 	<seglistitem><seg>%H</seg><seg>source host name (a <quote>device</quote>, according to <ulink url="http://www.ietf.org/rfc/rfc3164.txt">RFC 3164</ulink>)</seg></seglistitem>
 	<seglistitem><seg>%h</seg><seg>sender host name (datagram sender, which may be device or relay)</seg></seglistitem>
-	<seglistitem><seg>%P</seg><seg>program name</seg></seglistitem>
+	<seglistitem><seg>%P</seg><seg>program name; this is taken from the MSG part of message and converted to locale encoding if source encoding is specified</seg></seglistitem>
 	<seglistitem><seg>%%</seg><seg>% character</seg></seglistitem>
 </segmentedlist>
 						</para>