-- SPDX-FileCopyrightText: 2025 toastal <toastal@posteo.net>
-- SPDX-License-Identifier: MPL-2.0
--
-- address-url-encoding.lua
--
-- This plugin percent-encodes certain URI scheme which prevents lazy amount
-- of scrapers

Plugin.require_version("4.0.0")

function url_encode_fn(c)
	return format("%%%02X", ascii(c))
end

function split_url_encode(elem)
	local href = HTML.get_attribute(elem, "href")
	local splat = Regex.split(String.url_decode(href), ":")
	local scheme = splat[1]
	local addr = ""
	local idx = 2
	while splat[idx] do
		local part = splat[idx]
		if addr == "" then
			addr = part
		else
			addr = addr .. ":" .. part
		end
		idx = idx + 1
	end
	local addr_chars = Regex.split(addr, "")
	Table.apply_to_values(url_encode_fn, addr_chars)
	-- URL encodes *all* characters—including ASCII
	local new_href = scheme .. ":" .. String.join("", addr_chars)
	HTML.set_attribute(elem, "href", new_href)
end

local address_likes = HTML.select_all_of(page, {
	"[href^=\"mailto:\"]",
	"[href^=\"matrix:\"]",
	"[href^=\"tel:\"]",
	"[href^=\"xmpp:\"]"
})

Table.iter_values(split_url_encode, address_likes)