From 29b56ef4e4113bcd091b19d6926f18814162ca53 Mon Sep 17 00:00:00 2001
From: Nicholas Noll <nbnoll@eml.cc>
Date: Tue, 26 Oct 2021 21:01:41 -0700
Subject: Feat(libunicode): Added an explicit unicode library

Refactored code to pull out utf8 functions from base into a standalone
library. Also left the required function inside arg.c so that code that
calls ARG_BEGIN doesn't have to link to libunicode.
---
 sys/base/string.c | 174 ------------------------------------------------------
 1 file changed, 174 deletions(-)

(limited to 'sys/base/string.c')

diff --git a/sys/base/string.c b/sys/base/string.c
index 8973a4e..1ef9068 100644
--- a/sys/base/string.c
+++ b/sys/base/string.c
@@ -10,180 +10,6 @@ typedef struct Hdr
     byte  buf[];
 } Hdr;
 
-// -------------------------------------------------------------------------
-// UTF-8 functions
-
-#define Bit(i) (7-(i))
-/* N 0's preceded by i 1's e.g. T(Bit(2)) is 1100 0000 */
-#define Tbyte(i) (((1 << (Bit(i)+1))-1) ^ 0xFF)
-/* 0000 0000 0000 0111 1111 1111 */
-#define	RuneX(i) ((1 << (Bit(i) + ((i)-1)*Bitx))-1)
-
-enum
-{
-    Bitx  = Bit(1),
-    Tx    = Tbyte(1),
-    Rune1 = (1 << (Bit(0)+0*Bitx)) - 1,
-
-    Maskx = (1 << Bitx) - 1, /* 0011 1111 */
-    Testx = Maskx ^ 0xff,    /* 1100 0000 */
-
-    SurrogateMin = 0xD800,
-    SurrogateMax = 0xDFFF,
-    Bad = RuneErr,
-};
-
-int
-utf8·bytetorune(rune* r, byte* s)
-{
-    int c[UTFmax], i;
-    rune l;
-
-    c[0] = *(ubyte*)(s);
-    if(c[0] < Tx) {
-        *r = c[0];
-        return 1;
-    }
-
-    l = c[0];
-    for(i = 1; i < UTFmax; i++) {
-        c[i]  = *(ubyte*)(s+i);
-        c[i] ^= Tx;
-        if (c[i] & Testx) goto bad;
-
-        l = (l << Bitx) | c[i];
-        if(c[0] < Tbyte(i + 2)) {
-            l &= RuneX(i + 1);
-            if (i == 1) {
-                if (c[0] < Tbyte(2) || l <= Rune1)
-                    goto bad;
-            } else if (l <= RuneX(i) || l > RuneMax)
-                goto bad;
-            if (i == 2 && SurrogateMin <= l && l <= SurrogateMax)
-                goto bad;
-
-            *r = l;
-            return i + 1;
-        }
-    }
-bad:
-    *r = RuneErr;
-    return 1;
-}
-
-int
-utf8·runetobyte(byte* s, rune* r)
-{
-    int i, j;
-    rune c;
-
-    c = *r;
-    if (c <= Rune1) {
-        s[0] = c;
-        return 1;
-    }
-
-	for (i = 2; i < UTFmax + 1; i++){
-		if (i == 3){
-			if (c > RuneMax)
-				c = RuneErr;
-			if (SurrogateMin <= c && c <= SurrogateMax)
-				c = RuneErr;
-		}
-		if (c <= RuneX(i) || i == UTFmax) {
-			s[0] = Tbyte(i) |  (c >> (i - 1)*Bitx);
-			for(j = 1; j < i; j++)
-				s[j] = Tx | ((c >> (i - j - 1)*Bitx) & Maskx);
-			return i;
-		}
-	}
-
-    return UTFmax;
-}
-
-int
-utf8·runelen(rune r)
-{
-    byte s[10];
-    return utf8·runetobyte(s, &r);
-}
-
-int
-utf8·fullrune(byte* s, int n)
-{
-    int  i;
-    rune c;
-
-    if (n <= 0) return 0;
-    c = *(ubyte*) s;
-    if (c < Tx) return 1;
-
-    for (i = 3; i < UTFmax + 1; i++) {
-        if (c < Tbyte(i)) return n >= i - 1;
-    }
-
-    return n >= UTFmax;
-}
-
-byte*
-utf8·findrune(byte* s, long c)
-{
-    long c1;
-    rune r;
-    int  n;
-
-    if (c < RuneSync) return strchr(s, c);
-
-    for (;;) {
-        c1 = *(ubyte*)s;
-        if (c1 < RuneSelf) {
-            if (c1 == 0) return nil;
-            if (c1 == c) return s;
-            s++;
-            continue;
-        }
-        n = utf8·bytetorune(&r, s);
-        if (r == c) return s;
-        s += n;
-    }
-
-    return nil;
-}
-
-byte*
-utf8·findrrune(byte* s, long c)
-{
-    long c1;
-    rune r;
-    byte *l;
-
-    if (c < RuneSync) 
-        return strrchr(s, c);
-
-    l = nil;
-    for (;;) {
-        c1 = *(ubyte*)s;
-        if (c1 < RuneSelf) {
-            if (c1 == 0) return l;
-            if (c1 == c) l = s;
-            s++;
-            continue;
-        }
-        c1 = utf8·bytetorune(&r, s);
-        if (r == c) 
-            l = s;
-        s += c1;
-    }
-
-    return nil;
-}
-
-#undef Bit
-#undef Tbyte
-#undef RuneX
-
-#include ".generated/utf8.c"
-
 // -------------------------------------------------------------------------
 // Dynamic string functions
 
-- 
cgit v1.2.1