I now think that this may not work out at all. It looks like if we want proper

Unicode handling we'll need to implement a series of codecs and converters as
well as tables of codepages and lookups.  It'll be interesting, I guess, but
it makes me care a lot less about proper encoding.  Anyway, UtfString uses
shorts instead of chars, so it's a step in the right direction, but still not
enough to be able to handle proper UTF-16 encoding, maybe UCS-2 encoding, but...
...that's lame.  Bu::FBasicString has been generalized a bit with optimizations
from libc for char based strings.  It also, unfortunately, still uses char-only
functions in several places, those all rely on char casting strings at the
moment just to get the thing to compile.  Basically, it's not a good UTF-16
solution yet, and it may never be and remain compatible with char based strings.

1 files changed, 29 insertions, 0 deletions

diff --git a/src/utfstring.cpp b/src/utfstring.cpp
new file mode 100644
index 0000000..ae5efaf
--- /dev/null
+++ b/src/utfstring.cpp
@@ -0,0 +1,29 @@
+#include "bu/utfstring.h"
+
+template class Bu::FBasicString<short>;
+
+template<> uint32_t Bu::__calcHashCode<Bu::UtfString>( const Bu::UtfString &k )
+{
+        long j, sz = k.getSize()*2;
+        const char *s = (const char *)k.getStr();
+
+        long nPos = 0;
+        for( j = 0; j < sz; j++, s++ )
+        {
+                nPos = *s + (nPos << 6) + (nPos << 16) - nPos;
+        }
+
+        return nPos;    
+}
+
+template<> bool Bu::__cmpHashKeys<Bu::UtfString>(
+                const Bu::UtfString &a, const Bu::UtfString &b )
+{
+        return a == b;
+}
+
+template<> void Bu::__tracer_format<Bu::UtfString>( const Bu::UtfString &v )
+{
+        printf("(%ld)\"%s\"", v.getSize(), (const char *)v.getStr() );
+}
+