-- sitemap Soupault Plugin (post-build hook)
-- SPDX-FileCopyrightText: 2026 toastal <https://toast.al/contact>
-- SPDX-License-Identifier: MPL-2.0
--
-- Generates a sitemap.xml by walking the build output. A filesystem walk is
-- used rather than site_index because Soupault omits index pages (/, /section/)
-- from the index, which are precisely the URLs a sitemap must include.
--
-- Intended for use as:
--   hooks.post-build.file = "hook/sitemap.lua"
Plugin.require_version("5.0.0")

custom_options = soupault_config["custom_options"]

if Value.is_nil(custom_options) or not Table.has_key(custom_options, "site_url") then
	Plugin.fail([[custom_options["site_url"] option is required for sitemap generation]])
end

site_url = custom_options["site_url"]
output_file = Table.get_key_default(config, "output_file", "sitemap.xml")
exclude_path_regexes = Table.get_key_default(config, "exclude_path_regexes", {})
-- NOTE: this Lua dialect has no true/false literals (they evaluate to nil),
-- & the TOML→Lua bridge drops explicit false values entirely, so an absent
-- clean_urls is indistinguishable from clean_urls = false. Fall back to 1,
-- which matches Soupault core’s own default (clean_urls = true).
clean_urls = soupault_config["settings"]["clean_urls"]
if Value.is_nil(clean_urls) then
	clean_urls = 1
end

function is_excluded(web_path)
	local idx = 1
	while exclude_path_regexes[idx] do
		if Regex.match(web_path, exclude_path_regexes[idx]) then
			return 1
		end
		idx = idx + 1
	end
	return nil
end

function str_compare(a, b)
	if a < b then
		return -1
	end
	if a > b then
		return 1
	end
	return 0
end

-- Explicit stack (each item is a table with fs/web paths) to avoid recursion
-- (no TCO in Lua 2.5).
stack = {}
stack[1] = { fs = build_dir, web = "/" }
top = 1

urls = {}

sitemap_path = Sys.join_path(build_dir, output_file)

while top >= 1 do
	cur = stack[top]
	top = top - 1
	entries = Sys.list_dir(cur.fs)
	local idx = 1
	while entries[idx] do
		-- NOTE: Sys.list_dir returns full paths (meaning "build/about"), not
		-- basenames, so use entries directly as filesystem paths.
		local fpath = entries[idx]
		local name = Sys.basename(fpath)
		if not String.starts_with(name, ".") and fpath ~= sitemap_path then
			if Sys.is_dir(fpath) then
				local web = nil
				if cur.web == "/" then
					web = "/" .. name .. "/"
				else
					web = cur.web .. name .. "/"
				end
				top = top + 1
				stack[top] = { fs = fpath, web = web }
			elseif clean_urls and name == "index.html" then
				if not is_excluded(cur.web) then
					urls[Table.length(urls) + 1] = cur.web
				end
			elseif not clean_urls and Sys.has_extension(name, "html") then
				local web = nil
				if cur.web == "/" then
					web = "/" .. name
				else
					web = cur.web .. name
				end
				if not is_excluded(web) then
					urls[Table.length(urls) + 1] = web
				end
			end
		end
		idx = idx + 1
	end
end

if Table.is_empty(urls) then
	Plugin.fail("Sitemap generation found no pages in build output.")
end
Table.sort(str_compare, urls)

default_template = [[<?xml version="1.0" encoding="UTF-8"?>
<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">
{%- for url in urls %}
	<url>
		<loc>{{ url | escape }}</loc>
	</url>
{%- endfor -%}
</urlset>
]]

template = default_template
if not Value.is_nil(config["template_file"]) then
	template = Sys.read_file(config["template_file"])
end

function make_loc(web_path)
	return site_url .. web_path
end

locs = Table.copy(urls)
Table.apply_to_values(make_loc, locs)
data = {}
data["urls"] = locs
sitemap = String.render_template(template, data)

Sys.write_file(Sys.join_path(build_dir, output_file), sitemap)
Log.info(format("Wrote sitemap with %d URLs to %s", Table.length(urls), Sys.join_path(build_dir, output_file)))