From 86bfec58c7e282b4d3b8c4015f20becba8a86546 Mon Sep 17 00:00:00 2001
From: roker <roker@pep-project.org>
Date: Thu, 7 Oct 2021 20:35:39 +0200
Subject: [PATCH] single-char isNFC() removed. Makes no sense. Add freestanding
 operator+() for nfc_string

---
 src/nfc.cc | 27 ++++++++++++++-------------
 src/nfc.hh | 41 +++++++++++++++++++++++++++++++++++++++--
 2 files changed, 53 insertions(+), 15 deletions(-)

diff --git a/src/nfc.cc b/src/nfc.cc
index 3e6879c..c52f204 100644
--- a/src/nfc.cc
+++ b/src/nfc.cc
@@ -554,24 +554,25 @@ std::u32string createNFC(std::u32string nfd)
 }
 
 
-template<>
-IsNFC UTF<char>::isNFC(char c)
+template<class CharT>
+bool UTF<CharT>::is_safe_NFC_start(std::basic_string_view<CharT> s)
 {
-	if( c & 0x80 )
-		throw illegal_utf("Single octet >0x80 is invalid UTF-8");
+	if(s.empty() || (s[0] & 0x80)==0 ) // shortcut for empty string or starts with ASCII char
+	{
+		return true;
+	}
+	
+	const CharT* begin = s.data();
+	const CharT* const end = s.data() + s.size();
 	
-	return IsNFC::Yes; // all ASCII characters are valid NFC.
+	const uint32_t u = parse(begin, end);
+	if(NFC_No.count(u)) return false;
+	if(NFC_Maybe.count(u)) return false;
+	
+	return true;
 }
 
 
-template<>
-IsNFC UTF<char16_t>::isNFC(char16_t c)
-{
-	if(NFC_No.count(c)) return IsNFC::No;
-	if(NFC_Maybe.count(c)) return IsNFC::Maybe;
-	return IsNFC::Yes;
-}
-
 
 template<class CharT>
 IsNFC UTF<CharT>::isNFC_quick_check(std::basic_string_view<CharT> s)
diff --git a/src/nfc.hh b/src/nfc.hh
index 3139c3f..8a8a0bc 100644
--- a/src/nfc.hh
+++ b/src/nfc.hh
@@ -47,9 +47,9 @@ public:
 	void generate(const char32_t c, OutIter& out);
 	
 
-	/// returns the NFC class of a single character
+	/// returns whether the sequence starts with IsNFC==Yes char
 	static
-	IsNFC isNFC(CharT c);
+	bool is_safe_NFC_start(std::basic_string_view<CharT> s);
 
 	/// returns No or Maybe, if at least one character with NFC_Quickcheck class is "No" or "Maybe"
 	/// might throw illegal_utf exception
@@ -205,6 +205,10 @@ public:
 			return s.find( std::forward<Args>(args)... );
 		}
 		
+		/// might throw illegal_utf, if a multi-char sequence is clipped.
+		nfc_string substr(std::size_t pos=0, std::size_t count=npos) const;
+		
+		
 		
 	private:
 		std::basic_string<CharT> s;
@@ -214,6 +218,39 @@ public:
 	};
 };
 
+/// can be more efficient than the operator+() below.
+template<class CharT>
+typename
+UTF<CharT>::nfc_string operator+(
+		typename UTF<CharT>::nfc_string left,
+		const typename UTF<CharT>::nfc_string& right);
+
+template<class CharT, class T>
+inline
+typename
+UTF<CharT>::nfc_string operator+(typename UTF<CharT>::nfc_string left, const T& right)
+{
+	return left+=right;
+}
+
+template<class CharT, class T>
+inline
+typename
+UTF<CharT>::nfc_string operator+(typename UTF<CharT>::nfc_string&& left, const T& right)
+{
+	return left+=right;
+}
+
+
+template<class CharT, class T>
+inline
+typename
+UTF<CharT>::nfc_string operator+(const T& left, const typename UTF<CharT>::nfc_string& right)
+{
+	UTF<CharT> left_s{left};
+	return left_s+=right;
+}
+
 /// convenient alias names:
 using UTF8 = UTF<char>;
 using UTF16 = UTF<char16_t>;