From 039a17f454c84ac5a8b733937cf8717ccc45276b Mon Sep 17 00:00:00 2001
From: DTB <trinity@trinity.moe>
Date: Mon, 4 Sep 2023 09:24:37 -0400
Subject: [PATCH] more work

---
 unicode/libunicode.c |  2 ++
 unicode/libunicode.h |  2 ++
 unicode/utf8.c       | 61 ++++++++++++++++++++------------------------
 3 files changed, 32 insertions(+), 33 deletions(-)

diff --git a/unicode/libunicode.c b/unicode/libunicode.c
index bed1533..6b78b4f 100644
--- a/unicode/libunicode.c
+++ b/unicode/libunicode.c
@@ -4,6 +4,8 @@ utf8_t utf8(utf32_t c){
 	unsigned char n;
 	utf8_t r;
 
+	if(c > UTF8_MAX)
+		c = 0;
 	r = 0;
 	switch(n = (c < 0x10000) + (c < 0x0800) + (c < 0x0080)){
 	case 0: r = 0xF0 + ((c >> 18) & 0x07); /* 11110xxx */
diff --git a/unicode/libunicode.h b/unicode/libunicode.h
index ea9679f..3d1fb48 100644
--- a/unicode/libunicode.h
+++ b/unicode/libunicode.h
@@ -10,5 +10,7 @@
  * strings; >=32b value */
 #define utf8_t unicode_codepoint_t
 
+#define UTF8_MAX 0x10FFFF
+
 /* encode UTF-32 value into UTF-8 */
 utf8_t utf8(utf32_t c);
diff --git a/unicode/utf8.c b/unicode/utf8.c
index 8632b40..9114c6b 100644
--- a/unicode/utf8.c
+++ b/unicode/utf8.c
@@ -16,44 +16,39 @@ int main(int argc, char *argv[]){
 	utf32_t codepoint;
 	utf8_t encoded;
 
-	c = '\0';
+	l = 0;
+
+init:	codepoint = 0;
 	i = 0;
-	l = 1;
-	while(c != EOF){
-		for(
-			i = 0, n = NULL, codepoint = 0;
-			(c = getc(stdin)) != '\n'
-			&& c != EOF
-			&& i < 10;
-			++i
-		){
-			if(
-				(i == 0 && c != 'U')
-				|| (i == 1 && c != '+')
-				|| i > 10
-				|| (i > 1 && (n = strchr(hex, c))
-					== NULL)
-			){
+	++l;
+	n = NULL;
+	while((c = getc(stdin)) != EOF){
+		if(c == '\n'){
+			if(i < 2 && i > 0) /* empty lines are fine */
 				fprintf(stderr, "%s: %s: Syntax error.\n",
 					argv[0], l);
-				SKIPLINE;
-				i = -1;
-				break;
+			else if(i >= 2){
+				encoded = utf8(codepoint);
+				for(i = 3; i >= 0; --i)
+					if((encoded >> 8 * i) > 0 || i == 0)
+						putc(encoded >> 8 * i, stdout);
 			}
-			if(n != NULL)
-				codepoint = (codepoint << 4) + (n - hex) % 16;
+			goto init;
 		}
-		if(i < 3){
-			if(c != '\n' && c != EOF)
-				SKIPLINE;
-			if(c == EOF)
-				return 0;
-			continue;
+		if(
+				(i == 0 && c != 'U')
+				|| (i == 1 && c != '+')
+				|| i > 8 /* strlen("U+10FFFF") */
+				|| (i > 1 && ((n = strchr(hex, c)) == NULL))
+		){
+				fprintf(stderr, "%s: %s: Syntax error.\n",
+					argv[0], l);
+				while((c = getc(stdin)) != '\n' && c != EOF);
+				++l;
+				continue;
 		}
-		encoded = utf8(codepoint);
-		for(i = 3; i >= 0; --i)
-			if((encoded >> 8 * i) > 0 || i == 0)
-				putc(encoded >> 8 * i, stdout);
-		++l;
+		if(n != NULL)
+			codepoint = (codepoint << 4) + (n - hex) % 16;
+		++i;
 	}
 }