-- Prosody IM

-- Copyright (C) 2021 Kim Alvefur

--

-- This project is MIT/X11 licensed. Please see the

-- COPYING file in the source package for more information.

--

-- XEP-0363: HTTP File Upload

-- Again, from the top!

local t_insert = table.insert;

local jid = require "util.jid";

local st = require "util.stanza";

local url = require "socket.url";

local dm = require "core.storagemanager".olddm;

local jwt = require "util.jwt";

local errors = require "util.error";

local dataform = require "util.dataforms".new;

local urlencode = require "util.http".urlencode;

local dt = require "util.datetime";

local hi = require "util.human.units";

local cache = require "util.cache";

local lfs = require "lfs";

local unknown = math.abs(0/0);

local unlimited = math.huge;

local namespace = "urn:xmpp:http:upload:0";

module:depends("disco");

module:add_identity("store", "file", module:get_option_string("name", "HTTP File Upload"));

module:add_feature(namespace);

local uploads = module:open_store("uploads", "archive");

local persist_stats = module:open_store("upload_stats", "map");

-- id, <request>, time, owner

local secret = module:get_option_string(module.name.."_secret", require"util.id".long());

local external_base_url = module:get_option_string(module.name .. "_base_url");

local file_size_limit = module:get_option_number(module.name .. "_size_limit", 10 * 1024 * 1024); -- 10 MB

local file_types = module:get_option_set(module.name .. "_allowed_file_types", {});

local safe_types = module:get_option_set(module.name .. "_safe_file_types", {"image/*","video/*","audio/*","text/plain"});

local expiry = module:get_option_number(module.name .. "_expires_after", 7 * 86400);

local daily_quota = module:get_option_number(module.name .. "_daily_quota", file_size_limit*10); -- 100 MB / day

local total_storage_limit = module:get_option_number(module.name.."_global_quota", unlimited);

local access = module:get_option_set(module.name .. "_access", {});

if not external_base_url then

	module:depends("http");

end

module:add_extension(dataform {

	{ name = "FORM_TYPE", type = "hidden", value = namespace },

	{ name = "max-file-size", type = "text-single", datatype = "xs:integer" },

}:form({ ["max-file-size"] = file_size_limit }, "result"));

local upload_errors = errors.init(module.name, namespace, {

	access = { type = "auth"; condition = "forbidden" };

	filename = { type = "modify"; condition = "bad-request"; text = "Invalid filename" };

	filetype = { type = "modify"; condition = "not-acceptable"; text = "File type not allowed" };

	filesize = { type = "modify"; condition = "not-acceptable"; text = "File too large";

		extra = {tag = st.stanza("file-too-large", {xmlns = namespace}):tag("max-file-size"):text(tostring(file_size_limit)) };

	};

	filesizefmt = { type = "modify"; condition = "bad-request"; text = "File size must be positive integer"; };

	quota = { type = "wait"; condition = "resource-constraint"; text = "Daily quota reached"; };

	outofdisk = { type = "wait"; condition = "resource-constraint"; text = "Server global storage quota reached" };

});

local upload_cache = cache.new(1024);

local quota_cache = cache.new(1024);

local total_storage_usage = unknown;

local measure_upload_cache_size = module:measure("upload_cache", "amount");

local measure_quota_cache_size = module:measure("quota_cache", "amount");

local measure_total_storage_usage = module:measure("total_storage", "amount", { unit = "bytes" });

do

	local total, err = persist_stats:get(nil, "total");

	if not err then

		total_storage_usage = tonumber(total) or 0;

	end

end

module:hook_global("stats-update", function ()

	measure_upload_cache_size(upload_cache:count());

	measure_quota_cache_size(quota_cache:count());

	measure_total_storage_usage(total_storage_usage);

end);

local buckets = {};

for n = 10, 40, 2 do

	local exp = math.floor(2 ^ n);

	table.insert(buckets, exp);

	if exp >= file_size_limit then break end

end

local measure_uploads = module:measure("upload", "sizes", {buckets = buckets});

-- Convenience wrapper for logging file sizes

local function B(bytes)

	if bytes ~= bytes then

		return "unknown"

	elseif bytes == unlimited then

		return "unlimited";

	end

	return hi.format(bytes, "B", "b");

end

local function get_filename(slot, create)

	return dm.getpath(slot, module.host, module.name, "bin", create)

end

function get_daily_quota(uploader)

	local now = os.time();

	local max_age = now - 86400;

	local cached = quota_cache:get(uploader);

	if cached and cached.time > max_age then

		return cached.size;

	end

	local iter, err = uploads:find(nil, {with = uploader; start = max_age });

	if not iter then return iter, err; end

	local total_bytes = 0;

	local oldest_upload = now;

	for _, slot, when in iter do

		local size = tonumber(slot.attr.size);

		if size then total_bytes = total_bytes + size; end

		if when < oldest_upload then oldest_upload = when; end

	end

	-- If there were no uploads then we end up caching [now, 0], which is fine

	-- since we increase the size on new uploads

	quota_cache:set(uploader, { time = oldest_upload, size = total_bytes });

	return total_bytes;

end

function may_upload(uploader, filename, filesize, filetype) -- > boolean, error

	local uploader_host = jid.host(uploader);

	if not ((access:empty() and prosody.hosts[uploader_host]) or access:contains(uploader) or access:contains(uploader_host)) then

		return false, upload_errors.new("access");

	end

	if not filename or filename:find"/" then

		-- On Linux, only '/' and '\0' are invalid in filenames and NUL can't be in XML

		return false, upload_errors.new("filename");

	end

	if not filesize or filesize < 0 or filesize % 1 ~= 0 then

		return false, upload_errors.new("filesizefmt");

	end

	if filesize > file_size_limit then

		return false, upload_errors.new("filesize");

	end

	if total_storage_usage + filesize > total_storage_limit then

		module:log("warn", "Global storage quota reached, at %s / %s!", B(total_storage_usage), B(total_storage_limit));

		return false, upload_errors.new("outofdisk");

	end

	local uploader_quota = get_daily_quota(uploader);

	if uploader_quota + filesize > daily_quota then

		return false, upload_errors.new("quota");

	end

	if not ( file_types:empty() or file_types:contains(filetype) or file_types:contains(filetype:gsub("/.*", "/*")) ) then

		return false, upload_errors.new("filetype");

	end

	return true;

end

function get_authz(slot, uploader, filename, filesize, filetype)

local now = os.time();

	return jwt.sign(secret, {

		-- token properties

		sub = uploader;

		iat = now;

		exp = now+300;

		-- slot properties

		slot = slot;

		expires = expiry >= 0 and (now+expiry) or nil;

		-- file properties

		filename = filename;

		filesize = filesize;

		filetype = filetype;

	});

end

function get_url(slot, filename)

	local base_url = external_base_url or module:http_url();

	local slot_url = url.parse(base_url);

	slot_url.path = url.parse_path(slot_url.path or "/");

	t_insert(slot_url.path, slot);

	if filename then

		t_insert(slot_url.path, filename);

		slot_url.path.is_directory = false;

	else

		slot_url.path.is_directory = true;

	end

	slot_url.path = url.build_path(slot_url.path);

	return url.build(slot_url);

end

function handle_slot_request(event)

	local stanza, origin = event.stanza, event.origin;

	local request = st.clone(stanza.tags[1], true);

	local filename = request.attr.filename;

	local filesize = tonumber(request.attr.size);

	local filetype = request.attr["content-type"] or "application/octet-stream";

	local uploader = jid.bare(stanza.attr.from);

	local may, why_not = may_upload(uploader, filename, filesize, filetype);

	if not may then

		origin.send(st.error_reply(stanza, why_not));

		return true;

	end

	module:log("info", "Issuing upload slot to %s for %s", uploader, B(filesize));

	local slot, storage_err = errors.coerce(uploads:append(nil, nil, request, os.time(), uploader))

	if not slot then

		origin.send(st.error_reply(stanza, storage_err));

		return true;

	end

	total_storage_usage = total_storage_usage + filesize;

	module:log("debug", "Total storage usage: %s / %s", B(total_storage_usage), B(total_storage_limit));

	local cached_quota = quota_cache:get(uploader);

	if cached_quota and cached_quota.time > os.time()-86400 then

		cached_quota.size = cached_quota.size + filesize;

		quota_cache:set(uploader, cached_quota);

	end

	local authz = get_authz(slot, uploader, filename, filesize, filetype);

	local slot_url = get_url(slot, filename);

	local upload_url = slot_url;

	local reply = st.reply(stanza)

		:tag("slot", { xmlns = namespace })

			:tag("get", { url = slot_url }):up()

			:tag("put", { url = upload_url })

				:text_tag("header", "Bearer "..authz, {name="Authorization"})

		:reset();

	origin.send(reply);

	return true;

end

function handle_upload(event, path) -- PUT /upload/:slot

	local request = event.request;

	local authz = request.headers.authorization;

	if authz then

		authz = authz:match("^Bearer (.*)")

	end

	if not authz then

		module:log("debug", "Missing or malformed Authorization header");

		event.response.headers.www_authenticate = "Bearer";

		return 401;

	end

	local authed, upload_info = jwt.verify(secret, authz);

	if not (authed and type(upload_info) == "table" and type(upload_info.exp) == "number") then

		module:log("debug", "Unauthorized or invalid token: %s, %q", authed, upload_info);

		return 401;

	end

	if not request.body_sink and upload_info.exp < os.time() then

		module:log("debug", "Authorization token expired on %s", dt.datetime(upload_info.exp));

		return 410;

	end

	if not path or upload_info.slot ~= path:match("^[^/]+") then

		module:log("debug", "Invalid upload slot: %q, path: %q", upload_info.slot, path);

		return 400;

	end

	if request.headers.content_length and tonumber(request.headers.content_length) ~= upload_info.filesize then

		return 413;

		-- Note: We don't know the size if the upload is streamed in chunked encoding,

		-- so we also check the final file size on completion.

	end

	local filename = get_filename(upload_info.slot, true);

	do

		-- check if upload has been completed already

		-- we want to allow retry of a failed upload attempt, but not after it's been completed

		local f = io.open(filename, "r");

		if f then

			f:close();

			return 409;

		end

	end

	if not request.body_sink then

		module:log("debug", "Preparing to receive upload into %q, expecting %s", filename, B(upload_info.filesize));

		local fh, err = io.open(filename.."~", "w");

		if not fh then

			module:log("error", "Could not open file for writing: %s", err);

			return 500;

		end

		function event.response:on_destroy() -- luacheck: ignore 212/self

			-- Clean up incomplete upload

			if io.type(fh) == "file" then -- still open

				fh:close();

				os.remove(filename.."~");

			end

		end

		request.body_sink = fh;

		if request.body == false then

			if request.headers.expect == "100-continue" then

				request.conn:write("HTTP/1.1 100 Continue\r\n\r\n");

			end

			return true;

		end

	end

	if request.body then

		module:log("debug", "Complete upload available, %s", B(#request.body));

		-- Small enough to have been uploaded already

		local written, err = errors.coerce(request.body_sink:write(request.body));

		if not written then

			return err;

		end

		request.body = nil;

	end

	if request.body_sink then

		local final_size = request.body_sink:seek();

		local uploaded, err = errors.coerce(request.body_sink:close());

		if final_size ~= upload_info.filesize then

			-- Could be too short as well, but we say the same thing

			uploaded, err = false, 413;

		end

		if uploaded then

			module:log("debug", "Upload of %q completed, %s", filename, B(final_size));

			assert(os.rename(filename.."~", filename));

			measure_uploads(final_size);

			upload_cache:set(upload_info.slot, {

					name = upload_info.filename;

					size = tostring(upload_info.filesize);

					type = upload_info.filetype;

					time = os.time();

				});

			return 201;

		else

			assert(os.remove(filename.."~"));

			return err;

		end

	end

end

local download_cache_hit = module:measure("download_cache_hit", "rate");

local download_cache_miss = module:measure("download_cache_miss", "rate");

function handle_download(event, path) -- GET /uploads/:slot+filename

	local request, response = event.request, event.response;

	local slot_id = path:match("^[^/]+");