Issue: #381: Try to handle X11 input and UTF-8 better.

In window_get_text_prop do conversion when input is of type STRING. (latin1) to utf8.
Dmenu: don't skip invalid lines, but try to display as much as possible.
Window mode: Double check all input from X.

8 files changed, 73 insertions, 13 deletions

diff --git a/include/helper.h b/include/helper.h
index eae59ce1..de6f1097 100644
--- a/include/helper.h
+++ b/include/helper.h
@@ -166,5 +166,10 @@ void cmd_set_arguments ( int argc, char **argv );
 char *rofi_expand_path ( const char *input );
 unsigned int levenshtein ( const char *needle, const char *haystack );
 
+/**
+ * Convert string to valid utf-8, replacing invalid parts with replacement character.
+ */
+char * rofi_force_utf8 ( gchar *data );
+char * rofi_latin_to_utf8_strdup ( const char *input, gssize length );
 /*@}*/
 #endif // ROFI_HELPER_H
diff --git a/include/mode.h b/include/mode.h
index 90bae0f3..aaf38566 100644
--- a/include/mode.h
+++ b/include/mode.h
@@ -7,7 +7,7 @@
  * The 'object' that makes a mode in rofi.
  * @{
  */
-typedef struct rofi_mode Mode;
+typedef struct rofi_mode   Mode;
 
 /**
  * Enum used to sum the possible states of ROFI.
diff --git a/include/x11-helper.h b/include/x11-helper.h
index a1435c3f..730724bc 100644
--- a/include/x11-helper.h
+++ b/include/x11-helper.h
@@ -36,6 +36,8 @@ void window_set_atom_prop ( xcb_window_t w, xcb_atom_t prop, xcb_atom_t *atoms,
 #define EWMH_ATOMS( X )           \
     X ( _NET_WM_WINDOW_OPACITY ), \
     X ( I3_SOCKET_PATH ),         \
+    X ( UTF8_STRING ),            \
+    X ( STRING ),                 \
     X ( WM_WINDOW_ROLE )
 
 enum { EWMH_ATOMS ( ATOM_ENUM ), NUM_NETATOMS };
diff --git a/source/dialogs/dmenu.c b/source/dialogs/dmenu.c
index 02594572..2ca1540e 100644
--- a/source/dialogs/dmenu.c
+++ b/source/dialogs/dmenu.c
@@ -92,10 +92,8 @@ static char **get_dmenu ( DmenuModePrivateData *pd, FILE *fd, unsigned int *leng
             data[l - 1] = '\0';
             l--;
         }
-        if (  !g_utf8_validate ( data, l, NULL ) ) {
-            fprintf ( stderr, "String: '%s' is not valid utf-8\n", data );
-            continue;
-        }
+        data = rofi_force_utf8 ( data );
+        printf ( "data: %s\n", data );
 
         retv[( *length )] = data;
         data              = NULL;
diff --git a/source/dialogs/window.c b/source/dialogs/window.c
index 5c2bdc11..57684151 100644
--- a/source/dialogs/window.c
+++ b/source/dialogs/window.c
@@ -53,8 +53,6 @@
 
 #define WINLIST             32
 
-#define CLIENTTITLE         100
-#define CLIENTCLASS         50
 #define CLIENTSTATE         10
 #define CLIENTWINDOWTYPE    10
 #define CLIENTROLE          50
@@ -285,7 +283,8 @@ static client* window_client ( xcb_window_t win )
     cky = xcb_icccm_get_wm_class ( xcb->connection, c->window );
     xcb_icccm_get_wm_class_reply_t wcr;
     if ( xcb_icccm_get_wm_class_reply ( xcb->connection, cky, &wcr, NULL ) ) {
-        c->class = g_strdup ( wcr.class_name );
+        c->class = rofi_latin_to_utf8_strdup ( wcr.class_name, -1 );
+        c->name  = rofi_latin_to_utf8_strdup ( wcr.instance_name, -1 );
         xcb_icccm_get_wm_class_reply_wipe ( &wcr );
     }
 
@@ -295,6 +294,14 @@ static client* window_client ( xcb_window_t win )
         c->hint_flags = r.flags;
     }
 
+    /** Do UTF-8 Check, should not be needed, does not hurt here to be paranoid. */
+    {
+        c->title = rofi_force_utf8 ( c->title );
+        c->class = rofi_force_utf8 ( c->class );
+        c->name  = rofi_force_utf8 ( c->name );
+        c->role  = rofi_force_utf8 ( c->role );
+    }
+
     winlist_append ( cache_client, c->window, c );
     g_free ( attr );
     return c;
diff --git a/source/helper.c b/source/helper.c
index 2485e802..955939e6 100644
--- a/source/helper.c
+++ b/source/helper.c
@@ -659,3 +659,42 @@ unsigned int levenshtein ( const char *needle, const char *haystack )
     }
     return column[needlelen];
 }
+
+char * rofi_latin_to_utf8_strdup ( const char *input, gssize length )
+{
+    gsize slength = 0;
+    return g_convert_with_fallback ( input, length, "UTF-8", "latin1", "\uFFFD", NULL, &slength, NULL );
+}
+
+char * rofi_force_utf8 ( gchar *start )
+{
+    if ( start == NULL ) {
+        return NULL;
+    }
+    const char *data = start;
+    const char *end;
+    gsize      length = strlen ( data );
+    GString    *string;
+
+    if ( g_utf8_validate ( data, length, &end ) ) {
+        return start;
+    }
+    string = g_string_sized_new ( length + 16 );
+
+    do {
+        /* Valid part of the string */
+        g_string_append_len ( string, data, end - data );
+        /* Replacement character */
+        g_string_append ( string, "\uFFFD" );
+        length -= ( end - data ) + 1;
+        data = end + 1;
+    } while ( !g_utf8_validate ( data, length, &end ) );
+
+    if ( length ) {
+        g_string_append_len ( string, data, length );
+    }
+
+    // Free input string.
+    g_free ( start );
+    return g_string_free ( string, FALSE );
+}
diff --git a/source/rofi.c b/source/rofi.c
index eba030ba..7f708f3b 100644
--- a/source/rofi.c
+++ b/source/rofi.c
@@ -626,8 +626,8 @@ int main ( int argc, char *argv[] )
     // 2 the binary that executed is called dmenu (e.g. symlink to rofi)
     else{
         // Get the base name of the executable called.
-        char *base_name = g_path_get_basename ( argv[0] );
-        const char * const dmenu_str = "dmenu";
+        char               *base_name = g_path_get_basename ( argv[0] );
+        const char * const dmenu_str  = "dmenu";
         dmenu_mode = ( strcmp ( base_name, dmenu_str ) == 0 );
         // Free the basename for dmenu detection.
         g_free ( base_name );
diff --git a/source/x11-helper.c b/source/x11-helper.c
index 9a98d507..f56122ab 100644
--- a/source/x11-helper.c
+++ b/source/x11-helper.c
@@ -39,6 +39,7 @@
 #include "xcb-internal.h"
 #include "xcb.h"
 #include "settings.h"
+#include "helper.h"
 
 #include <rofi.h>
 #define OVERLAP( a, b, c,                          \
@@ -90,9 +91,17 @@ char* window_get_text_prop ( xcb_window_t w, xcb_atom_t atom )
     xcb_get_property_reply_t  *r = xcb_get_property_reply ( xcb->connection, c, NULL );
     if ( r ) {
         if ( xcb_get_property_value_length ( r ) > 0 ) {
-            char *str = g_malloc ( xcb_get_property_value_length ( r ) + 1 );
-            memcpy ( str, xcb_get_property_value ( r ), xcb_get_property_value_length ( r ) );
-            str[xcb_get_property_value_length ( r )] = '\0';
+            char *str = NULL;
+            if ( r->type == netatoms[UTF8_STRING] ) {
+                str = g_strndup ( xcb_get_property_value ( r ), xcb_get_property_value_length ( r ) );
+            }
+            else if ( r->type == netatoms[STRING] ) {
+                str = rofi_latin_to_utf8_strdup ( xcb_get_property_value ( r ), xcb_get_property_value_length ( r ) );
+            }
+            else {
+                str = g_strdup ( "Invalid encoding." );
+            }
+
             free ( r );
             return str;
         }