Convert all parts of logged message to destination encoding.

This is required if destination encoding is UCS-2 or UCS-4 for example.
Otherwise the logfile will not be uniform.
Fixed a bug in charset conversion function: divide by zero when
cannot convert the first character and calculating new size for the output
buffer.

5 files changed, 80 insertions, 33 deletions

diff --git a/daemon/conf.c b/daemon/conf.c
index 78f95c2..9be1c18 100644
--- a/daemon/conf.c
+++ b/daemon/conf.c
@@ -679,11 +679,9 @@ 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?
@@ -693,8 +691,8 @@ static void create_conversion_descriptors( struct logpath* logpath )
 
         if( strcasecmp( source_encoding, destination_encoding ) != 0 )
         {
-            logpath->message_cd = g_iconv_open( destination_encoding, source_encoding );
-            if( logpath->message_cd == (GIConv) -1 )
+            logpath->src2dest_cd = g_iconv_open( destination_encoding, source_encoding );
+            if( logpath->src2dest_cd == (GIConv) -1 )
             {
                 ERR( "Cannot convert messages from %s to %s\n",
                      source_encoding, destination_encoding );
@@ -706,9 +704,23 @@ static void create_conversion_descriptors( struct logpath* logpath )
                        source_encoding, destination_encoding );
             }
         }
+
+        if( strcasecmp( "ASCII", destination_encoding ) != 0 )
+        {
+            logpath->ascii2dest_cd = g_iconv_open( destination_encoding, source_encoding );
+            if( logpath->ascii2dest_cd == (GIConv) -1 )
+            {
+                ERR( "Cannot convert messages from ASCII to %s\n", destination_encoding );
+            }
+            else
+            {
+                TRACE( "Log path %s-%s: converting messages from ASCII to %s\n",
+                       logpath->source->name, logpath->destination->name,
+                       destination_encoding );
+            }
+        }
     }
 
-    /* create charset conversion descriptor for program name */
     if( logpath->source->encoding )
     {
         const gchar *destination_encoding;
@@ -717,15 +729,15 @@ static void create_conversion_descriptors( struct logpath* logpath )
 
         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 )
+            logpath->src2locale_cd = g_iconv_open( destination_encoding, logpath->source->encoding );
+            if( logpath->src2locale_cd == (GIConv) -1 )
             {
-                ERR( "Cannot convert program names from %s to %s\n",
+                ERR( "Cannot convert locale parts from %s to %s\n",
                      logpath->source->encoding, destination_encoding );
             }
             else
             {
-                TRACE( "Log path %s-%s: converting program name from %s to %s\n",
+                TRACE( "Log path %s-%s: converting locale parts from %s to %s\n",
                        logpath->source->name, logpath->destination->name,
                        logpath->source->encoding, destination_encoding );
             }
@@ -752,8 +764,9 @@ static void resolve_logpaths()
         g_free( logpath );
         logpath = g_malloc( sizeof(struct logpath) );
 
-        logpath->message_cd = (GIConv) -1;
-        logpath->program_cd = (GIConv) -1;
+        logpath->src2dest_cd = (GIConv) -1;
+        logpath->ascii2dest_cd = (GIConv) -1;
+        logpath->src2locale_cd = (GIConv) -1;
 
         /* find source */
         for( item = sources; item; item = item->next )
diff --git a/daemon/dest_file.c b/daemon/dest_file.c
index c4d3cfe..d0b9ccd 100644
--- a/daemon/dest_file.c
+++ b/daemon/dest_file.c
@@ -181,8 +181,8 @@ static void write_message_to_logfile( struct file_writer* writer, struct message
     if( INVALID_HANDLE_VALUE == writer->fd )
         return;
     TRACE_2( "%p: %s\n", writer, (*msg)->message->gstr->str );
-    len = string_concat( &buffer, (*msg)->timestamp, space, (*msg)->hostname,
-                         space, (*msg)->message, line_feed, NULL );
+    len = string_concat( &buffer, (*msg)->timestamp, (*msg)->separator, (*msg)->hostname,
+                         (*msg)->separator, (*msg)->message, (*msg)->end_of_line, NULL );
     WriteFile( writer->fd, buffer, len, &written, NULL );
     g_free( buffer );
     release_message( *msg );
@@ -458,7 +458,7 @@ static void make_file_name( char* pattern, struct message* message, char* buffer
         case 'f': dest += sprintf( dest, "%d", message->facility ); break;
         case 'L': dest += sprintf( dest, "%s", get_priority_name( message->priority ) ); break;
         case 'l': dest += sprintf( dest, "%d", message->priority ); break;
-        case 'H': dest += sprintf( dest, "%s", message->hostname->gstr->str ); break;
+        case 'H': dest += sprintf( dest, "%s", message->hostname_in_locale->gstr->str ); break;
         case 'h': dest += sprintf( dest, "%s", message->sender->gstr->str ); break;
         case 'P': dest += sprintf( dest, "%s", message->program->gstr->str ); break;
         default: *dest++ = c; break;
diff --git a/daemon/dest_relay.c b/daemon/dest_relay.c
index 0ddf675..701c2c3 100755
--- a/daemon/dest_relay.c
+++ b/daemon/dest_relay.c
@@ -100,10 +100,11 @@ static void put_message_to_relay_dest( struct destination* destination, struct m
 
     pri = string_printf( "<%d>", LOG_MAKEPRI( msg->facility, msg->priority ) );
     if( destination->u.relay.omit_hostname )
-        len = string_concat( &buffer, pri, msg->timestamp, space, msg->message, line_feed, NULL );
+        len = string_concat( &buffer, pri, msg->timestamp, msg->separator,
+                             msg->message, msg->end_of_line, NULL );
     else
-        len = string_concat( &buffer, pri, msg->timestamp, space,
-                             msg->hostname, space, msg->message, line_feed, NULL );
+        len = string_concat( &buffer, pri, msg->timestamp, msg->separator, msg->hostname,
+                             msg->separator, msg->message, msg->end_of_line, NULL );
     string_release( pri );
 
     if( len > 1024 )
diff --git a/daemon/syslogd.c b/daemon/syslogd.c
index 5ff5401..6f981f1 100644
--- a/daemon/syslogd.c
+++ b/daemon/syslogd.c
@@ -83,8 +83,11 @@ struct message* create_message( struct source* source,
     g_free( ts );
 
     msg->hostname = string_addref( hostname );
+    msg->hostname_in_locale = string_addref( hostname );
     msg->program = string_addref( program );
     msg->message = string_addref( message );
+    msg->separator = string_addref( space );
+    msg->end_of_line = string_addref( line_feed );
 
     TRACE_LEAVE( "message=%p\n", msg );
     return msg;
@@ -109,8 +112,11 @@ struct message* duplicate_message( struct message* msg )
     new_msg->priority  = msg->priority;
     new_msg->timestamp = string_addref( msg->timestamp );
     new_msg->hostname  = string_addref( msg->hostname );
+    new_msg->hostname_in_locale = string_addref( msg->hostname_in_locale );
     new_msg->program   = string_addref( msg->program );
     new_msg->message   = string_addref( msg->message );
+    new_msg->separator = string_addref( msg->separator );
+    new_msg->end_of_line = string_addref( msg->end_of_line );
 
     TRACE_LEAVE( "new message=%p\n", new_msg );
     return new_msg;
@@ -143,8 +149,11 @@ void release_message( struct message* msg )
     string_release( msg->sender );
     string_release( msg->timestamp );
     string_release( msg->hostname );
+    string_release( msg->hostname_in_locale );
     string_release( msg->program );
     string_release( msg->message );
+    string_release( msg->separator );
+    string_release( msg->end_of_line );
     g_free( msg );
     TRACE_LEAVE( "done\n" );
 }
@@ -223,10 +232,14 @@ static struct string* try_convert_string( struct string* s, GIConv cd )
         {
         case E2BIG:
             /* guess the right output length */
-            outbytes = inbytes_left *
-                       /* average size of destination char: */
-                       converted_str_len / (s->gstr->len - inbytes_left)
-                       + 16; /* if the result of above expression too small */
+            if( s->gstr->len > inbytes_left )
+                outbytes = inbytes_left *
+                           /* average size of destination char: */
+                           converted_str_len / (s->gstr->len - inbytes_left)
+                           + 16; /* if the result of above expression too small */
+            else
+                /* cannot convert the first character; choose some initial value */
+                outbytes = 32;
             converted_str_allocated += outbytes;
             converted_str = g_realloc( converted_str, converted_str_allocated );
             outbuf = converted_str + converted_str_len;
@@ -269,7 +282,7 @@ 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;
+        gboolean need_dest_iconv, need_locale_iconv;
         struct message *converted_msg;
 
         if( logpath->source != msg->source )
@@ -279,10 +292,10 @@ static void mux_message( struct message* msg )
             continue;
 
         /* convert encoding if needed */
-        need_message_iconv = logpath->message_cd != (GIConv) -1;
-        need_program_iconv = logpath->program_cd != (GIConv) -1;
+        need_dest_iconv = (logpath->src2dest_cd != (GIConv) -1) && (logpath->ascii2dest_cd != (GIConv) -1);
+        need_locale_iconv = logpath->src2locale_cd != (GIConv) -1;
 
-        if( (!need_message_iconv) && (!need_program_iconv) )
+        if( (!need_dest_iconv) && (!need_locale_iconv) )
         {
             converted_msg = msg;
             reference_message( msg );
@@ -291,15 +304,31 @@ static void mux_message( struct message* msg )
         {
             converted_msg = duplicate_message( msg );
 
-            if( need_message_iconv )
+            if( need_dest_iconv )
             {
+                string_release( converted_msg->timestamp );
+                converted_msg->timestamp = try_convert_string( msg->timestamp, logpath->ascii2dest_cd );
+
+                string_release( converted_msg->hostname );
+                converted_msg->hostname = try_convert_string( msg->hostname, logpath->src2dest_cd );
+
                 string_release( converted_msg->message );
-                converted_msg->message = try_convert_string( msg->message, logpath->message_cd );
+                converted_msg->message = try_convert_string( msg->message, logpath->src2dest_cd );
+
+                string_release( converted_msg->separator );
+                converted_msg->separator = try_convert_string( msg->separator, logpath->ascii2dest_cd );
+
+                string_release( converted_msg->end_of_line );
+                converted_msg->end_of_line = try_convert_string( msg->end_of_line, logpath->ascii2dest_cd );
             }
-            if( need_program_iconv )
+            if( need_locale_iconv )
             {
+                string_release( converted_msg->hostname_in_locale );
+                converted_msg->hostname_in_locale = try_convert_string( msg->hostname_in_locale,
+                                                                        logpath->src2locale_cd );
+
                 string_release( converted_msg->program );
-                converted_msg->program = try_convert_string( msg->program, logpath->program_cd );
+                converted_msg->program = try_convert_string( msg->program, logpath->src2locale_cd );
             }
         }
 
diff --git a/daemon/syslogd.h b/daemon/syslogd.h
index 7b41db8..b7569bb 100644
--- a/daemon/syslogd.h
+++ b/daemon/syslogd.h
@@ -96,9 +96,12 @@ struct message
     int facility;
     int priority;
     struct string *timestamp;
-    struct string *hostname;
-    struct string *program;
+    struct string *hostname;            /* for logging */
+    struct string *hostname_in_locale;  /* for filename construction */
+    struct string *program;             /* in locale, for filename construction */
     struct string *message;
+    struct string *separator;
+    struct string *end_of_line;
 };
 
 extern struct message* create_message( struct source* source,
@@ -197,8 +200,9 @@ struct logpath
     struct source *source;
     struct filter *filter;
     struct destination *destination;
-    GIConv message_cd;  /* charset conversion descriptor for message */
-    GIConv program_cd;  /* charset conversion descriptor for program name */
+    GIConv src2dest_cd;    /* charset conversion descriptor for message */
+    GIConv ascii2dest_cd;  /* charset conversion descriptor for ascii parts of message */
+    GIConv src2locale_cd;  /* charset conversion descriptor for those parts that are used for filename construction */
 };
 
 extern GList *sources;