From 005be9fbf58197ba2a726bb92f31884aba1e68a2 Mon Sep 17 00:00:00 2001
From: Lars Mueller <appgurulars@gmx.de>
Date: Tue, 4 Jan 2022 13:42:46 +0100
Subject: [PATCH] Add web.html.[un]escape

---
 build/html_entities.lua | 19 +++++++++++++++++++
 init.lua                |  2 ++
 test.lua                |  8 ++++++++
 web.lua                 | 28 ++++++++++++++++++++++++++++
 4 files changed, 57 insertions(+)
 create mode 100644 build/html_entities.lua
 create mode 100644 web.lua

diff --git a/build/html_entities.lua b/build/html_entities.lua
new file mode 100644
index 0000000..57ebf04
--- /dev/null
+++ b/build/html_entities.lua
@@ -0,0 +1,19 @@
+-- Generate lookup table for HTML entities out of https://html.spec.whatwg.org/entities.json
+-- Requires https://github.com/brunoos/luasec to fetch the JSON
+local https = require 'ssl.https'
+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))
+end
+local entries = {}
+for entity, chars in pairs(entity_map) do
+	table.insert(entries, ("[%q] = %q"):format(entity, chars))
+end
+local serialized = [[-- HTML entity lookup table generated by build/html_entities.lua. Do not edit.
+return {]] .. table.concat(entries, ", ") .. "}"
+local loaded = assert(loadstring(serialized))
+setfenv(loaded, {})
+assert(modlib.table.equals(entity_map, loaded()))
+modlib.file.write(modlib.mod.get_resource("modlib", "web", "html", "entities.lua"), serialized)
\ No newline at end of file
diff --git a/init.lua b/init.lua
index b915cd4..0d43015 100644
--- a/init.lua
+++ b/init.lua
@@ -153,6 +153,8 @@ if minetest then
 	modlib.conf.build_setting_tree()
 end
 
+-- Run build scripts
+-- dofile(modlib.mod.get_resource("modlib", "build", "html_entities.lua"))
 
 --[[
 --modlib.mod.include"test.lua"
diff --git a/test.lua b/test.lua
index 2c17c06..d1560f9 100644
--- a/test.lua
+++ b/test.lua
@@ -324,6 +324,14 @@ do
 	end)
 end
 
+do
+	local text = "<tag> & '\""
+	local escaped = web.html.escape(text)
+	assert(web.html.unescape(escaped) == text)
+	assert(web.html.unescape("&#42;") == _G.string.char(42))
+	assert(web.html.unescape("&#x42;") == _G.string.char(0x42))
+end
+
 if not _G.minetest then return end
 
 assert(minetest.luon:read_string(minetest.luon:write_string(ItemStack"")))
diff --git a/web.lua b/web.lua
new file mode 100644
index 0000000..dafcbd2
--- /dev/null
+++ b/web.lua
@@ -0,0 +1,28 @@
+-- TODO incorporate https://github.com/minetest/minetest/pull/11578
+return {
+	html = setmetatable({
+		escape = function(text)
+			return text:gsub(".", {
+				["<"] = "&lt;",
+				[">"] = "&gt;",
+				["&"] = "&amp;",
+				["'"] = "&apos;",
+				['"'] = "&quot;",
+			})
+		end
+	}, {__index = function(self, key)
+		if key == "unescape" then
+			local func = assert(loadfile(modlib.mod.get_resource("modlib", "web", "html", "entities.lua")))
+			setfenv(func, {})
+			local named_entities = assert(func())
+			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
+			end
+			self.unescape = unescape
+			return unescape
+		end
+	end})
+}
\ No newline at end of file