From 1689f7822599a011434c0f0645fac96f4d4cb906 Mon Sep 17 00:00:00 2001
From: DTB <trinity@trinity.moe>
Date: Sun, 3 Sep 2023 02:23:32 -0400
Subject: [PATCH] might work now

---
 unicode/utf8.c | 91 ++++++++++++++++----------------------------------
 1 file changed, 29 insertions(+), 62 deletions(-)

diff --git a/unicode/utf8.c b/unicode/utf8.c
index bfe780f..24f6c7b 100644
--- a/unicode/utf8.c
+++ b/unicode/utf8.c
@@ -2,41 +2,29 @@
 #include <string.h> /* memset(3) */
 #include "ascii.h" /* ASCII_HEXADECIMAL_DIGITS_LOWER,
                     * ASCII_HEXADECIMAL_DIGITS_UPPER */
+#define SKIPLINE while((c = getc(stdin)) != '\n' && c != EOF)
 
 static char *hex = ASCII_HEXADECIMAL_DIGITS_UPPER
 	ASCII_HEXADECIMAL_DIGITS_LOWER;
 
-void print_hexascii(unsigned char *hexes, int n){
-	if(n % 2 != 0)
-		return;
-	while(n --> 0){
-		putc((char)(((hex - strchr(hex, hexes[0])) << 4)
-			+ (hex - strchr(hex, hexes[1]))), stdout);
-		++hexes;
-	}
-}
-
 int main(int argc, char *argv[]){
 	int c;
 	int i;
 	int l; /* line counter */
 	char *n;
-	unsigned char utf32_hex[8]; /* nybbles */
 	long int utf32_lit;
-	unsigned char utf8_hex[8]; /* nybbles */
-	long int utf8_lit;
+	unsigned char utf8_bytes[8];
 
 	c = '\0';
 	i = 0;
 	l = 1;
 	while(c != EOF){
-		memset(utf32_hex, 0, sizeof utf32_hex);
-		memset(utf8_hex, '0', sizeof utf8_hex);
+		memset(utf8_bytes, '0', sizeof utf8_bytes);
 		for(
-			i = 0, n = NULL, utf32_lit = 0, utf8_lit = 0;
+			i = 0, n = NULL, utf32_lit = 0;
 			(c = getc(stdin)) != '\n'
 			&& c != EOF
-			&& i < (sizeof utf32_hex) / (sizeof *utf32_hex);
+			&& i < 10;
 			++i
 		){
 			if(
@@ -48,58 +36,37 @@ int main(int argc, char *argv[]){
 			){
 				fprintf(stderr, "%s: %s: Syntax error.\n",
 					argv[0], l);
-				while((c = getc(stdin)) != '\n' && c != EOF);
+				SKIPLINE;
 				i = -1;
 				break;
 			}
-			if(n != NULL){
-				n -= 16;
-				utf32_hex[i - 2] = *(n -= 16 * (n - hex > 16));
-			}else if(i >= 2)
-				utf32_hex[i - 2] = c;
+			if(n != NULL)
+				utf32_lit = (utf32_lit << 4) + (n - hex) % 16;
 		}
-		if(i == -1 || i < 3)
+		if(i < 3){
+			if(c != '\n' && c != EOF)
+				SKIPLINE;
+			if(c == EOF)
+				return 0;
 			continue;
-		while(utf32_hex[7] == '\0'){ /* slow but easy */
-			for(i = 0; i < 7; ++i)
-				utf32_hex[i + 1] = utf32_hex[i];
-			utf32_hex[0] = '0';
 		}
-		/* this code is embarrassing */
-		for(i = 0; i < 8; ++i)
-			utf32_lit = (utf32_lit << 4)
-				+ strchr(hex, utf32_hex[i]) - hex;
-		if(utf32_lit < 128){
-			utf8_hex[7] = utf32_hex[7];
-			utf8_hex[6] = utf32_hex[6];
-			i = 6;
-			goto done;
-		}else{
-			utf8_hex[7] = hex[utf32_lit & 15];
-			utf8_hex[6] = hex[((utf32_lit >> 4) & 3) + 8];
+		switch(i =	(utf32_lit < 65536)
+			+ (utf32_lit < 2048)
+			+ (utf32_lit < 128)){
+		case 0: utf8_bytes[0] =
+			((utf32_lit >> 18) & 7) + 240; /* 11110xxx */
+		case 1:	utf8_bytes[1] = i == 1
+			? ((utf32_lit >> 12) & 15) + 224 /* 1110xxxx */
+			: ((utf32_lit >> 12) & 63) + 80; /* 10xxxxxx */
+		case 2:	utf8_bytes[2] = i == 2
+			? ((utf32_lit >> 6) & 31) + 192 /* 110xxxxx */
+			: ((utf32_lit >> 6) & 63) + 80; /* 10xxxxxx */
+		case 3:	utf8_bytes[3] = i == 3
+			? utf8_bytes[3] = utf32_lit & 127 /* 0xxxxxxx */
+			: (utf32_lit & 63) + 80; /* 10xxxxxx */
 		}
-		if(utf32_lit < 2048){
-			utf8_hex[5] = hex[(utf32_lit >> 6) & 15];
-			utf8_hex[4] = hex[((utf32_lit >> 10) & 1) + 12];
-			i = 4;
-			goto done;
-		}else{
-			utf8_hex[5] = hex[(utf32_lit >> 6) & 15];
-			utf8_hex[4] = hex[((utf32_lit >> 10) & 3) + 8];
-		}
-		if(utf32_lit < 65536){
-			utf8_hex[3] = hex[(utf32_lit >> 12) & 15];
-			utf8_hex[2] = 14;
-			i = 2;
-			goto done;
-		}else{
-			utf8_hex[3] = hex[(utf32_lit >> 12) & 15];
-			utf8_hex[2] = hex[((utf32_lit >> 16) & 3) + 8];
-			utf8_hex[1] = hex[(utf32_lit >> 21) & 3];
-			utf8_hex[0] = hex[15];
-			i = 0;
-		}
-done:		print_hexascii(utf8_hex + i, 8 - i);
+		for( ; i < 4; ++i)
+			putc(utf8_bytes[i], stdout);
 		++l;
 	}
 }