From 170a143d543bd2cc828f5678c7df9b38927901d2 Mon Sep 17 00:00:00 2001
From: Hudd <haddayn@gmail.com>
Date: Mon, 11 Mar 2024 21:14:29 +0400
Subject: [PATCH] unicode: add utf8::skip

Moving random bits of code into awlib. I remember needing this in the
past, so I'll add it here.
---
 utility/include/aw/utility/unicode/utf8.h | 37 +++++++++++++++--------
 utility/tests/utf8toutf16.c++             | 10 ++++++
 2 files changed, 35 insertions(+), 12 deletions(-)

diff --git a/utility/include/aw/utility/unicode/utf8.h b/utility/include/aw/utility/unicode/utf8.h
index e5f5f3e0..c9869b0a 100644
--- a/utility/include/aw/utility/unicode/utf8.h
+++ b/utility/include/aw/utility/unicode/utf8.h
@@ -140,18 +140,6 @@ struct codec {
 		return output;
 	}
 
-	/*!
-	 * Advance to the next sequence of code units
-	 */
-	template<typename Iterator>
-	static Iterator next(Iterator input, Iterator end)
-	{
-		while ((input < end) && !is_head(*input) && !is_ascii(*input))
-			++input;
-
-		return input;
-	}
-
 	/*!
 	 * Decode UTF-8 sequence, if valid
 	 */
@@ -207,6 +195,31 @@ struct codec {
 
 		return input;
 	}
+
+	/*!
+	 * Advance to the next sequence of code units
+	 */
+	template<typename Iterator>
+	static Iterator next(Iterator input, Iterator end)
+	{
+		while ((input < end) && !is_head(*input) && !is_ascii(*input))
+			++input;
+
+		return input;
+	}
+
+	/*!
+	 * Skip n code points
+	 */
+	template<typename Iterator>
+	static Iterator skip(Iterator input, Iterator end, size_t n)
+	{
+		while (input < end && n > 0) {
+			input = next(std::next(input), end);
+			--n;
+		}
+		return input;
+	}
 };
 } // namespace aw::unicode::utf8
 #endif//aw_utility_utf8_h
diff --git a/utility/tests/utf8toutf16.c++ b/utility/tests/utf8toutf16.c++
index 008b7782..088f3d60 100644
--- a/utility/tests/utf8toutf16.c++
+++ b/utility/tests/utf8toutf16.c++
@@ -4,6 +4,16 @@
 TestFile("utf8toutf16");
 
 namespace aw {
+Test(utf8skip) {
+	std::string s1 = "αβγ";
+	std::string s2 = "こんにちは";
+
+	auto it1 = aw::unicode::utf8::codec::skip(begin(s1), end(s1), 2);
+	auto it2 = aw::unicode::utf8::codec::skip(begin(s2), end(s2), 4);
+	TestEqual(std::string(it1, end(s1)), "γ");
+	TestEqual(std::string(it2, end(s2)), "は");
+}
+
 Test(utf8toutf16)
 {
 	const std::pair<std::u8string, std::u16string> utf_strings[] = {