Changeset 453 for branches/thune/thread_safe/encoding.c

Timestamp:

08/25/07 22:49:09 (15 months ago)

Author:

krobillard

Message:

UTF-8 strings can now be read.
ur_makeString() now takes UCell argument.

Files:

• branches/thune/thread_safe/encoding.c (modified) (5 diffs)

branches/thune/thread_safe/encoding.c

@@ -1 +1 @@
 /*============================================================================
     Urlan Interpreter
-    Copyright (C) 2005-2006  Karl Robillard
+    Copyright (C) 2005-2007  Karl Robillard
 
     This library is free software; you can redistribute it and/or
@@ -31 +31 @@
    Returns number of characters copied.
 */
-int copyUtf16ToAscii( char* dest, const uint16_t* src, int len )
+int copyUcs2ToUtf8( char* dest, const uint16_t* src, int len )
+{
+    // TODO: Prevent overflow of dest.
+    const char* start;
+    const uint16_t* end;
+    uint16_t c;
+
+    start = dest;
+    end = src + len;
+
+    while( src != end )
+    {
+        c = *src++;
+        if( c > 127 )
+        {
+            if( c > 0x07ff )
+            {
+                *dest++ = 0xE0 | (c >> 12);
+                *dest++ = 0x80 | ((c >> 6) & 0x3f);
+                c = 0x80 | (c & 0x3f);
+            }
+            else
+            {
+                *dest++ = 0xC0 | (c >> 6);
+                c = 0x80 | (c & 0x3f);
+            }
+        }
+        *dest++ = (char) c;
+    }
+    return dest - start;
+}
+
+
+#if 0
+int copyUcs2ToAscii( char* dest, const uint16_t* src, int len )
 {
     const uint16_t* end;
@@ -47 +81 @@
     return len;
 }
+#endif
 
 
-#ifndef UR_CONFIG_UDS
 /*
    Returns number of characters copied.
@@ -87 +121 @@
                     case UR_ENC_UTF16:
                         count >>= 1;
-                        strN = ur_makeBinary( count, &bin ); 
-                        bin->used = copyUtf16ToAscii( bin->ptr.c,
-                                                      (uint16_t*) cpA, count );
+                        strN = ur_makeBinary( count * 2, &bin ); 
+                        bin->used = copyUcs2ToUtf8( bin->ptr.c,
+                                                    (uint16_t*) cpA, count );
                         enc = UR_ENC_ASCII;
                         goto set_result;
@@ -156 +190 @@
     }
 }
-#endif