hare

commit 0806d8afc8ff8e22be9307098d432a499c67bfb2
parent 5c5cfb36e47cdaf80a3c292a32f6b9e9d9e48ab9
Author: Drew DeVault <sir@cmpwn.com>
Date:   Mon,  1 Feb 2021 17:39:31 -0500

encoding::utf8: various improvements

Diffstat:
M
encoding/utf8/decode.ha
 | 
14
+++++++-------
M
encoding/utf8/rune.ha
 | 
1
-

2 files changed, 7 insertions(+), 8 deletions(-)
diff --git a/encoding/utf8/decode.ha b/encoding/utf8/decode.ha
@@ -44,30 +44,30 @@ export fn next(d: *decoder) (rune | void | more | invalid) = {
 	let bytes = d.src[d.offs..d.offs+n];
 	d.offs += n;
 
-	let cp = 0u32;
+	let r = 0u32;
 	if (bytes[0] < 128u8) {
 		// ASCII
 		return bytes[0]: u32: rune;
 	};
 
 	const mask = masks[n - 1z];
-	cp = bytes[0] & mask;
+	r = bytes[0] & mask;
 	for (let i = 1z; i < len(bytes); i += 1z) {
-		cp <<= 6u8;
-		cp |= bytes[i] & 0x3Fu8;
+		r <<= 6u8;
+		r |= bytes[i] & 0x3Fu8;
 	};
-	return cp: rune;
+	return r: rune;
 };
 
 fn utf8sz(src: []u8) size = {
-	assert(sizes[0].octets == 1z);
-
 	assert(len(src) > 0z);
+
 	let c = src[0];
 	for (let i = 0z; i < len(sizes); i += 1z) {
 		if (c & sizes[i].mask == sizes[i].result) {
 			return sizes[i].octets;
 		};
 	};
+
 	return types::SIZE_MAX;
 };
diff --git a/encoding/utf8/rune.ha b/encoding/utf8/rune.ha
@@ -15,7 +15,6 @@ const sizes: [_]rsize = [
 	rsize { mask = 0xF8u8, result = 0xF0u8, octets = 4z },
 	rsize { mask = 0xFCu8, result = 0xF8u8, octets = 5z },
 	rsize { mask = 0xFEu8, result = 0xF8u8, octets = 6z },
-	rsize { mask = 0x80u8, result = 0x80u8, octets = types::SIZE_MAX },
 ];
 
 // Returns the size of a rune, in octets, when encoded as UTF-8.