slimpack/src/formats/xml.lua

--- @alias xml_node_raw { tag: string, attribs: { [string]: string }, [integer]: xml_element }
--- @alias xml_element string | xml_node

--- @class xml_node
--- @field tag string
--- @field attribs { [string]: string }
--- @field [integer] xml_element
local xml_node = {};
xml_node.__index = xml_node;

--- @param name string
function xml_node:get_all(name)
	--- @type xml_node[]
	local res = {};

	for _, el in ipairs(self) do
		if type(el) ~= "string" and el.tag == name then
			res[#res + 1] = el;
		end
	end

	return res;
end

--- @param name string
function xml_node:get(name)
	local res = self:get_all(name);
	if #res == 0 then
		error("node '" .. name .. "' not found");
	elseif #res > 1 then
		error("multiple nodes '" .. name .. "' exist");
	else
		return res[1];
	end
end

function xml_node:text()
	if #self == 0 then
		return "";
	elseif #self == 1 and type(self[1]) == "string" then
		return self[1];
	else
		error "not a text-only node";
	end
end

--- @param raw xml_node_raw
--- @return xml_node
function xml_node.new(raw)
	local res = setmetatable({
		tag = raw.tag,
		attribs = raw.attribs or {},
	}, xml_node);

	table.move(raw, 1, #raw, 1, res);

	return res;
end

local function skip_spaces(raw, i)
	local next_i = raw:find("[^%s]", i);
	if next_i then return next_i end

	if raw:find("^%s", i) then
		local match = raw:match("^%s*", i);
		if match then return i + #match end
	else
		return i;
	end
end

local function parse_tag(raw, i)
	i = skip_spaces(raw, i);

	local tag = raw:match("^[%w0-9%-]+", i);
	if tag == nil then error("expected tag name near '" .. raw:sub(i, i + 25) .. "'") end

	i = i + #tag;

	local attribs = {};

	while true do
		i = skip_spaces(raw, i);

		local all, key, _, val = raw:match("^(([%w0-9%-]-)%s*=%s*(['\"])(.-)%3%s*)", i);
		if all then
			attribs[key] = val;
			i = i + #all;
		else
			break;
		end
	end

	return { tag = tag, attribs = attribs }, i;
end

local function parse_part(raw, i, allow_version)
	i = skip_spaces(raw, i);

	repeat
		local comment_end;

		if raw:sub(i, i + 3) == "<!--" then
			i = i + 43;

			_, comment_end = raw:find("-->", i);
			i = comment_end or #raw;
			i = skip_spaces(raw, i);
		end
	until not comment_end;

	if i > #raw then
		return { type = "eof" }, i;
	elseif allow_version and raw:sub(i, i + 1) == "<?" then
		i = i + 2;

		local tag;
		tag, i = parse_tag(raw, i);

		if raw:sub(i, i + 1) ~= "?>" then
			error("malformed XML near '" .. raw:sub(i, i + 25) .. "'");
		end
		i = i + 2;

		return { type = "version", tag = tag.tag, attribs = tag.attribs }, i;
	elseif raw:sub(i, i + 1) == "</" then
		i = i + 2;
		i = skip_spaces(raw, i);

		local tag = raw:match("[%w0-9%-]+", i);
		if tag == nil then error("expected closing tag name near '" .. raw:sub(i, i + 25) .. "'") end

		i = i + #tag;

		if raw:sub(i, i) == ">" then
			i = i + 1;
			return { type = "end", tag = tag }, i;
		else
			error("malformed closing tag near '" .. raw:sub(i, i + 25) .. "'");
		end

	elseif raw:sub(i, i) == "<" then
		i = i + 1;

		local tag;
		tag, i = parse_tag(raw, i);


		if raw:sub(i, i + 1) == "/>" then
			i = i + 2;
			return { type = "small", tag = tag.tag, attribs = tag.attribs }, i;
		elseif raw:sub(i, i) == ">" then
			i = i + 1;
			return { type = "begin", tag = tag.tag, attribs = tag.attribs }, i;
		else
			error("malformed opening tag near '" .. raw:sub(i, i + 25) .. "'");
		end
	else
		local text_parts = {};

		while i <= #raw do
			local text_end = raw:find("<", i);

			local text_part = raw:sub(i, text_end and text_end - 1 or #raw):match "^%s*(.-)%s*$";
			if text_part ~= "" then
				text_parts[#text_parts + 1] = text_part;
			end

			if not text_end then break end

			i = text_end or #raw;

			local comment_end;

			if raw:sub(i, i + 3) == "<!--" then
				i = i + 4;

				_, comment_end = raw:find("-->", i);
				i = comment_end and comment_end + 1 or #raw;
				i = skip_spaces(raw, i);
			else
				break
			end
		end

		if #text_parts > 0 then
			return { type = "text", text = table.concat(text_parts, " ") }, i;
		elseif i > #raw then
			return { type = "eof" }, i;
		else
			error("malformed XML near '" .. raw:sub(i, i + 25) .. "'");
		end
	end

end

--- @param raw string
--- @return xml_element
local function parse(raw)
	--- @type xml_node
	local curr_node = xml_node.new { tag = "document", attribs = {} };
	local stack = { curr_node };
	local i = 1;
	local first = true;

	while true do
		local part;
		part, i = parse_part(raw, i, first);
		if part.type == "eof" then break end

		first = false;

		if part.type == "text" then
			if #stack == 1 then
				error "text may not appear outside a tag";
			else
				curr_node[#curr_node + 1] = part.text;
			end
		elseif part.type == "version" then
			curr_node.attribs.type = part.tag;
			curr_node.attribs.version = part.attribs.version;
		elseif part.type == "begin" then
			local new_node = xml_node.new { tag = part.tag, attribs = part.attribs };
			curr_node[#curr_node + 1] = new_node;
			curr_node = new_node;
			stack[#stack + 1] = new_node;
		elseif part.type == "end" then
			if part.tag ~= curr_node.tag then
				error("closing tag '" .. part.tag .. "' doesn't match most recent opening tag '" .. curr_node.tag .. "'");
			else
				table.remove(stack);
				curr_node = stack[#stack];
			end
		elseif part.type == "small" then
			curr_node[#curr_node + 1] = xml_node.new { tag = part.tag, attribs = part.attribs };
		else
			error "wtf";
		end
	end

	if #stack > 1 then
		error("tag '" .. curr_node.tag .. "' was left open");
	end

	return curr_node;
end

return {
	parse = parse,
	node = xml_node.new,
};