plugins/storage/xmlparse.lib.lua

1
local st = require "util.stanza";
2
 
3
-- XML parser
4
local parse_xml = (function()
5
	local entity_map = setmetatable({
6
		["amp"] = "&";
7
		["gt"] = ">";
8
		["lt"] = "<";
9
		["apos"] = "'";
10
		["quot"] = "\"";
11
	}, {__index = function(_, s)
12
			if s:sub(1,1) == "#" then
13
				if s:sub(2,2) == "x" then
14
					return string.char(tonumber(s:sub(3), 16));
15
				else
16
					return string.char(tonumber(s:sub(2)));
17
				end
18
			end
19
		end
20
	});
21
	local function xml_unescape(str)
22
		return (str:gsub("&(.-);", entity_map));
23
	end
24
	local function parse_tag(s)
25
		local name,sattr=(s):gmatch("([^%s]+)(.*)")();
26
		local attr = {};
27
		for a,b in (sattr):gmatch("([^=%s]+)=['\"]([^'\"]*)['\"]") do attr[a] = xml_unescape(b); end
28
		return name, attr;
29
	end
30
	return function(xml)
31
		local stanza = st.stanza("root");
32
		local regexp = "<([^>]*)>([^<]*)";
33
		for elem, text in xml:gmatch(regexp) do
34
			if elem:sub(1,1) == "!" or elem:sub(1,1) == "?" then -- neglect comments and processing-instructions
35
			elseif elem:sub(1,1) == "/" then -- end tag
36
				elem = elem:sub(2);
37
				stanza:up(); -- TODO check for start-end tag name match
38
			elseif elem:sub(-1,-1) == "/" then -- empty tag
39
				elem = elem:sub(1,-2);
40
				local name,attr = parse_tag(elem);
41
				stanza:tag(name, attr):up();
42
			else -- start tag
43
				local name,attr = parse_tag(elem);
44
				stanza:tag(name, attr);
45
			end
46
			if #text ~= 0 then -- text
47
				stanza:text(xml_unescape(text));
48
			end
49
		end
50
		return stanza.tags[1];
51
	end
52
end)();
53
-- end of XML parser
54
 
55
return parse_xml;