X-Git-Url: https://harrygodden.com/git/?a=blobdiff_plain;f=common.h;h=c0a5f3599187fa39904489082b25954c1f13d847;hb=b93c61c54e7ac56f6808b9a563d3e4221ca8482e;hp=a36dc0a2460528ff70aad8c2fb1d9d7da0d9fd3e;hpb=a98ce96041b248580c0447bf87d4c6106483cade;p=carveJwlIkooP6JGAAIwe30JlM.git

diff --git a/common.h b/common.h
index a36dc0a..c0a5f35 100644
--- a/common.h
+++ b/common.h
@@ -9,6 +9,7 @@
 #define VG_3D
 #define VG_FRAMEBUFFER_RESIZE 1
 #include "vg/vg.h"
+#include "anyascii/anyascii.h"
 
 typedef struct ray_hit ray_hit;
 struct ray_hit
@@ -20,4 +21,63 @@ struct ray_hit
 
 static int network_scores_updated = 0;
 
+static u32 utf8_byte0_byte_count( u8 char0 )
+{
+   for( u32 k=2; k<4; k++ )
+   {
+      if( !(char0 & (0x80 >> k)) )
+         return k;
+   }
+
+   return 0;
+}
+
+static void str_utf8_collapse( const char *str, char *buf, u32 length )
+{
+   u8 *ustr = (u8 *)str;
+   u32 utf32_code = 0x00000000;
+   u32 i=0, j=0, utf32_byte_ct=0;
+
+   for(;i < length-1;)
+   {
+      if( ustr[i] == 0x00 )
+         break;
+      
+      if( ustr[i] & 0x80 )
+      {
+         if( utf32_byte_ct )
+         {
+            utf32_byte_ct --;
+            utf32_code |= (ustr[i] & 0x3F) << (utf32_byte_ct*6);
+
+            if( !utf32_byte_ct )
+            {
+               const char *match;
+               size_t chars = anyascii( utf32_code, &match );
+
+               for( u32 k=0; k<VG_MIN(chars, length-1-j); k++ )
+               {
+                  buf[ j++ ] = (u8)match[k];
+               }
+            }
+         }
+         else
+         {
+            utf32_byte_ct = utf8_byte0_byte_count( ustr[i] )-1;
+            utf32_code = ustr[i] & (0x3F >> utf32_byte_ct);
+            utf32_code <<= utf32_byte_ct*6;
+         }
+      }
+      else
+      {
+         utf32_byte_ct = 0x00;
+         buf[j ++] = str[i];
+      }
+
+      i++;
+   }
+
+   buf[j] = 0x00;
+}
+
 #endif /* COMMON_H */