From 825599f2d09ec4473f1441e985ab235a604e4799 Mon Sep 17 00:00:00 2001
From: Lars Mueller <appgurulars@gmx.de>
Date: Fri, 9 Sep 2022 14:17:18 +0200
Subject: [PATCH] Move text.utf8 to utf8.char

---
 build/html_entities.lua |  2 +-
 init.lua                |  1 +
 json.lua                |  4 ++--
 text.lua                | 21 ---------------------
 utf8.lua                | 25 +++++++++++++++++++++++++
 web/html.lua            |  4 ++--
 6 files changed, 31 insertions(+), 26 deletions(-)
 create mode 100644 utf8.lua

diff --git a/build/html_entities.lua b/build/html_entities.lua
index 57ebf04..60083f1 100644
--- a/build/html_entities.lua
+++ b/build/html_entities.lua
@@ -5,7 +5,7 @@ local res, code = https.request"https://html.spec.whatwg.org/entities.json"
 assert(code == 200)
 local entity_map = {}
 for entity, chars in pairs(assert(modlib.json:read_string(res))) do
-	entity_map[entity:sub(2, #entity - 1)] = table.concat(modlib.table.map(chars.codepoints, modlib.text.utf8))
+	entity_map[entity:sub(2, #entity - 1)] = table.concat(modlib.table.map(chars.codepoints, modlib.utf8.char))
 end
 local entries = {}
 for entity, chars in pairs(entity_map) do
diff --git a/init.lua b/init.lua
index 755e9d2..c0109cf 100644
--- a/init.lua
+++ b/init.lua
@@ -45,6 +45,7 @@ for _, file in pairs{
 	"table",
 	"vararg",
 	"text",
+	"utf8",
 	"vector",
 	"quaternion",
 	"trie",
diff --git a/json.lua b/json.lua
index e4f9bb3..1168d07 100644
--- a/json.lua
+++ b/json.lua
@@ -101,7 +101,7 @@ for i = 0, 5 do
 end
 
 -- TODO SAX vs DOM
-local utf8 = modlib.text.utf8
+local utf8_char = modlib.utf8.char
 function read(self, read_)
 	local index = 0
 	local char
@@ -149,7 +149,7 @@ function read(self, read_)
 		end
 	end
 	local function utf8_codepoint(codepoint)
-		return syntax_assert(utf8(codepoint), "invalid codepoint")
+		return syntax_assert(utf8_char(codepoint), "invalid codepoint")
 	end
 	local function string()
 		local chars = {}
diff --git a/text.lua b/text.lua
index 5574c0f..6a1c859 100644
--- a/text.lua
+++ b/text.lua
@@ -138,27 +138,6 @@ magic_charset = "[" .. table.concat(magic_charset) .. "]"
 
 function escape_magic_chars(text) return text:gsub("(" .. magic_charset .. ")", "%%%1") end
 
-function utf8(number)
-	if number <= 0x007F then
-		-- Single byte
-		return string.char(number)
-	end
-	if number < 0x00A0 or number > 0x10FFFF then
-		-- Out of range
-		return
-	end
-	local result = ""
-	local i = 0
-	while true do
-		local remainder = number % 64
-		result = string.char(128 + remainder) .. result
-		number = (number - remainder) / 64
-		i = i + 1
-		if number <= 2 ^ (8 - i - 2) then break end
-	end
-	return string.char(256 - 2 ^ (8 - i - 1) + number) .. result
-end
-
 local keywords = modlib.table.set{"and", "break", "do", "else", "elseif", "end", "false", "for", "function", "if", "in", "local", "nil", "not", "or", "repeat", "return", "then", "true", "until", "while"}
 keywords["goto"] = true -- Lua 5.2 (LuaJIT) support
 
diff --git a/utf8.lua b/utf8.lua
new file mode 100644
index 0000000..061298f
--- /dev/null
+++ b/utf8.lua
@@ -0,0 +1,25 @@
+local string_char = string.char
+
+local utf8 = {}
+
+function utf8.char(codepoint)
+	if codepoint <= 0x007F then
+		-- Single byte
+		return string_char(codepoint)
+	end
+	if codepoint < 0x00A0 or codepoint > 0x10FFFF then
+		-- Out of range
+		return -- TODO (?) error instead
+	end
+	local result = ""
+	local i = 0
+	repeat
+		local remainder = codepoint % 64
+		result = string_char(128 + remainder) .. result
+		codepoint = (codepoint - remainder) / 64
+		i = i + 1
+	until codepoint <= 2 ^ (8 - i - 2)
+	return string_char(0x100 - 2 ^ (8 - i - 1) + codepoint) .. result
+end
+
+return utf8
\ No newline at end of file
diff --git a/web/html.lua b/web/html.lua
index f1b9aad..9b141d4 100644
--- a/web/html.lua
+++ b/web/html.lua
@@ -6,8 +6,8 @@ local html = setmetatable({}, {__index = function(self, key)
 		local function unescape(text)
 			return text
 				:gsub("&([A-Za-z]+);", named_entities) -- named
-				:gsub("&#(%d+);", function(digits) return modlib.text.utf8(tonumber(digits)) end) -- decimal
-				:gsub("&#x(%x+);", function(digits) return modlib.text.utf8(tonumber(digits, 16)) end) -- hex
+				:gsub("&#(%d+);", function(digits) return modlib.utf8.char(tonumber(digits)) end) -- decimal
+				:gsub("&#x(%x+);", function(digits) return modlib.utf8.char(tonumber(digits, 16)) end) -- hex
 		end
 		self.unescape = unescape
 		return unescape