Ticket #7423: terminal-utf8-spaces.patch

src/apps/terminal/UTF8Char.h

@@ -12 +12 @@
 struct UTF8Char {
     char    bytes[4];
 
+    
     UTF8Char()
     {
     }
+    
 
     UTF8Char(char c)
     {
         bytes[0] = c;
     }
+    
+    
+    UTF8Char(const char* c)
+    {
+        SetTo(c, ByteCount(*c));
+    }
+    
 
     UTF8Char(const char* c, int32 count)
     {
         SetTo(c, count);
     }
 
+
     void SetTo(const char* c, int32 count)
     {
         bytes[0] = c[0];
@@ -39 +49 @@
         }
     }
 
+
     static int32 ByteCount(char firstChar)
     {
         // Note, this does not recognize invalid chars
@@ -50 +61 @@
         return c < 0xf0 ? 3 : 4;
     }
 
+
     int32 ByteCount() const
     {
         return ByteCount(bytes[0]);
     }
 
+
     bool IsFullWidth() const
     {
         // TODO: Implement!
         return false;
     }
 
+
     bool IsSpace() const
     {
-        // TODO: Support multi-byte chars!
-        return ByteCount() == 1 ? isspace(bytes[0]) : false;
+        if (ByteCount() == 1) {
+            return isspace(bytes[0]);
+        }
+        else if (ByteCount() == 3) {
+            // most multibyte space chars are all in one range (0x2000-0x200B)
+            if (bytes[0] == 0xe8 && bytes[1] == 0x80)
+                return (bytes[2] >= 0x80 && bytes[2] <= 0x8f);
+            
+            // U+FEFF (zero-width non-breaking space) is not
+            if (bytes[0] == 0xef && bytes[1] == 0xbb)
+                return (bytes[2] == 0xbf);
+            
+            return false;
+        }
+        else {
+            return false;
+        }
     }
 
+
     UTF8Char ToLower() const
     {
         // TODO: Support multi-byte chars!
@@ -76 +106 @@
         return UTF8Char((char)tolower(bytes[0]));
     }
 
+
     bool operator==(const UTF8Char& other) const
     {
         int32 byteCount = ByteCount();
@@ -91 +122 @@
         return equals;
     }
 
+
     bool operator!=(const UTF8Char& other) const
     {
         return !(*this == other);

src/apps/terminal/TermView.cpp

@@ -405 +405 @@
 
     virtual int Classify(const char* character)
     {
-        // TODO: Deal correctly with non-ASCII chars.
-        char c = *character;
-        if (UTF8Char::ByteCount(c) > 1)
-            return CHAR_TYPE_WORD_CHAR;
-
-        if (isspace(c))
+        UTF8Char u8c(character);
+        
+        if (u8c.IsSpace())
             return CHAR_TYPE_SPACE;
+        
+        char c = *character;
         if (isalnum(c) || strchr(fSpecialWordChars, c) != NULL)
             return CHAR_TYPE_WORD_CHAR;