WOO IT WORKS A LOT BETTER WHEN I ACTUALLY READ

1 files changed, 17 insertions, 7 deletions

diff --git a/main.c b/main.c
index 69a3966..cf258c1 100644
--- a/main.c
+++ b/main.c
@@ -37,19 +37,29 @@ static void usage(int argc, char **argv, char *message, ...) {
     exit(1);
 }
 
-static inline size_t utf8_encode(uint8_t *buf, uint16_t codepoint) {
+static inline size_t utf8_encode(uint8_t *buf, uint32_t codepoint) {
     if ((codepoint & 0x007f) == codepoint) {
-        buf[0] = codepoint & 0xff;
+        buf[0] = codepoint & 0x007f;
+
         return 1;
     } else if ((codepoint & 0x07ff) == codepoint) {
-        buf[0] = (codepoint & 0x07c0) >> 6;
-        buf[1] =  codepoint & 0x003f;
+        buf[0] = 0xc0 | ((codepoint & 0x07c0) >> 6);
+        buf[1] = 0x80 |  (codepoint & 0x003f);
+
         return 2;
     } else if ((codepoint & 0xffff) == codepoint) {
-        buf[0] = (codepoint & 0xf000) >> 12;
-        buf[1] = (codepoint & 0x0fc0)  >>  6;
-        buf[2] =  codepoint & 0x003f;
+        buf[0] = 0xe0 | ((codepoint & 0xf000) >> 12);
+        buf[1] = 0x80 | ((codepoint & 0x0fc0) >>  6);
+        buf[2] = 0x80 |  (codepoint & 0x003f);
+
         return 3;
+    } else {
+        buf[0] = 0xf0 | ((codepoint & 0x1c0000) >> 18);
+        buf[1] = 0x80 | ((codepoint & 0x03f000) >> 12);
+        buf[2] = 0x80 | ((codepoint & 0x000fc0) >>  6);
+        buf[3] = 0x80 |  (codepoint & 0x00003f);
+
+        return 4;
     }
 }