From 3f7c3de28871a1eb133e5025448fedd8745be333 Mon Sep 17 00:00:00 2001
From: bumbread <thebumboni@gmail.com>
Date: Mon, 18 Jul 2022 02:09:26 +1100
Subject: [PATCH] mb <-> c32 functions

---
 src/code/uchar.c | 80 +++++++++---------------------------------------
 1 file changed, 14 insertions(+), 66 deletions(-)

diff --git a/src/code/uchar.c b/src/code/uchar.c
index 3a01b97..30aa02b 100644
--- a/src/code/uchar.c
+++ b/src/code/uchar.c
@@ -100,46 +100,18 @@ size_t mbrtoc32(
     mbstate_t  *restrict ps
 ) {
     if(s == NULL) {
-        *ps = (mbstate_t) {0};
         return 0;
     }
-    size_t nbytes;
-    
-    // Decode the first byte of UTF-8 sequence
-    unsigned byte0 = *s;
-    if     (0x00 <= byte0 && byte0 < 0x80) nbytes = 1;
-    else if(0xc0 <= byte0 && byte0 < 0xe0) nbytes = 2;
-    else if(0xe0 <= byte0 && byte0 < 0xf0) nbytes = 3;
-    else if(0xf0 <= byte0 && byte0 < 0xf8) nbytes = 4;
-    else goto encoding_error;
-    unsigned nbytesreq = nbytes;
-    if(n < nbytesreq) {
-        return (size_t)(-2);
+    char32_t code_point;
+    int mblen = utf8_chdec((char8_t *)s, n, &code_point);
+    if(mblen == UNI_ESTRLN) return (size_t)(-2);
+    if(mblen <= 0) {
+        errno = EILSEQ;
+        return (size_t)(-1);
     }
-    char32_t cp = byte0;
-    switch(nbytesreq) {
-        case 2: cp &= 0x1f; break;
-        case 3: cp &= 0x0f; break;
-        case 4: cp &= 0x07; break;
-    }
-    while(--nbytesreq)
-        cp |= (cp << 6) | ((*++s) & 0x3f);
-    if(0xdc00 <= cp && cp <= 0xe000)
-        goto encoding_error;
-    // Overloing seqs
-    if(cp < 0x80    && nbytes > 1) goto encoding_error;
-    if(cp < 0x800   && nbytes > 2) goto encoding_error;
-    if(cp < 0x10000 && nbytes > 3) goto encoding_error;
-    if(cp > 0x10ffff) goto encoding_error;
-    
-    if(pc32 != NULL) *pc32 = cp;
-    if(cp == 0)
-        return 0;
-    else
-        return nbytes;
-encoding_error:
-    errno = EILSEQ;
-    return (size_t)(-1);
+    *pc32 = code_point;
+    if(code_point == 0) return 0;
+    return (size_t)mblen;
 }
 
 size_t c32rtomb(
@@ -151,34 +123,10 @@ size_t c32rtomb(
         *ps = (mbstate_t) {0};
         return 0;
     }
-    unsigned cp = c32;
-    if(cp >= 0x10ffff) goto encoding_error;
-    size_t nbytes = 4;
-    if(cp < 0x10000) nbytes = 3;
-    if(cp < 0x800)   nbytes = 2;
-    if(cp < 0x80)    nbytes = 1;
-    switch(nbytes) {
-        case 1: {
-            s[0] = cp;
-        } break;
-        case 2: {
-            s[0] = 0xc0 | (cp >> 6);
-            s[1] = 0x80 | ((cp >> 0)  & 0x3f);
-        } break;
-        case 3: {
-            s[0] = 0xe0 | (cp >> 12);
-            s[1] = 0x80 | ((cp >> 6)  & 0x3f);
-            s[2] = 0x80 | ((cp >> 0)  & 0x3f);
-        } break;
-        case 4: {
-            s[0] = 0xf0 | (cp >> 18);
-            s[1] = 0x80 | ((cp >> 12) & 0x3f);
-            s[2] = 0x80 | ((cp >> 6)  & 0x3f);
-            s[3] = 0x80 | ((cp >> 0)  & 0x3f);
-        } break;
+    int mblen = utf8_enc(s, 4, c32);
+    if(mblen <= 0) {
+        errno = EILSEQ;
+        return (size_t)(-1);
     }
-    return nbytes;
-encoding_error:
-    errno = EILSEQ;
-    return (size_t)(-1);
+    return (size_t)mblen;
 }