Move some unicode handling stuff to a separate file.

c/jv_unicode.c

@@ -0,0 +1,30 @@
+#include <assert.h>
+#include "jv_unicode.h"
+int jvp_utf8_encode_length(int codepoint) {
+  if (codepoint <= 0x7F) return 1;
+  else if (codepoint <= 0x7FF) return 2;
+  else if (codepoint <= 0xFFFF) return 3;
+  else return 4;
+}
+
+int jvp_utf8_encode(int codepoint, char* out) {
+  assert(codepoint >= 0 && codepoint <= 0x10FFFF);
+  char* start = out;
+  if (codepoint <= 0x7F) {
+    *out++ = codepoint;
+  } else if (codepoint <= 0x7FF) {
+    *out++ = 0xC0 + ((codepoint & 0x7C0) >> 6);
+    *out++ = 0x80 + ((codepoint & 0x03F));
+  } else if(codepoint <= 0xFFFF) {
+    *out++ = 0xE0 + ((codepoint & 0xF000) >> 12);
+    *out++ = 0x80 + ((codepoint & 0x0FC0) >> 6);
+    *out++ = 0x80 + ((codepoint & 0x003F));
+  } else {
+    *out++ = 0xF0 + ((codepoint & 0x1C0000) >> 18);
+    *out++ = 0x80 + ((codepoint & 0x03F000) >> 12);
+    *out++ = 0x80 + ((codepoint & 0x000FC0) >> 6);
+    *out++ = 0x80 + ((codepoint & 0x00003F));
+  }
+  assert(out - start == jvp_utf8_encode_length(codepoint));
+  return out - start;
+}