From 1016cb7d48acda978809300f77b6c3dba335efd8 Mon Sep 17 00:00:00 2001
From: Waqas Hussain <waqas20@gmail.com>
Date: Sun, 2 Aug 2009 14:35:03 +0500
Subject: ejabberdsql2prosody: Added an XML parser

---
 tools/ejabberdsql2prosody.lua | 52 +++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 52 insertions(+)

diff --git a/tools/ejabberdsql2prosody.lua b/tools/ejabberdsql2prosody.lua
index 5b975269..c9f8beb6 100644
--- a/tools/ejabberdsql2prosody.lua
+++ b/tools/ejabberdsql2prosody.lua
@@ -156,6 +156,58 @@ return readFile(filename);
 ------
 end
 
+-- XML parser
+local parse_xml = (function()
+	local entity_map = setmetatable({
+		["amp"] = "&";
+		["gt"] = ">";
+		["lt"] = "<";
+		["apos"] = "'";
+		["quot"] = "\"";
+	}, {__index = function(_, s)
+			if s:sub(1,1) == "#" then
+				if s:sub(2,2) == "x" then
+					return string.char(tonumber(s:sub(3), 16));
+				else
+					return string.char(tonumber(s:sub(2)));
+				end
+			end
+		end
+	});
+	local function xml_unescape(str)
+		return (str:gsub("&(.-);", entity_map));
+	end
+	local function parse_tag(s)
+		local name,sattr=(s):gmatch("([^%s]+)(.*)")();
+		local attr = {};
+		for a,b in (sattr):gmatch("([^=%s]+)=['\"]([^'\"]*)['\"]") do attr[a] = xml_unescape(b); end
+		return name, attr;
+	end
+	return function(xml)
+		local stanza = st.stanza("root");
+		local regexp = "<([^>]*)>([^<]*)";
+		for elem, text in xml:gmatch(regexp) do
+			if elem:sub(1,1) == "!" or elem:sub(1,1) == "?" then -- neglect comments and processing-instructions
+			elseif elem:sub(1,1) == "/" then -- end tag
+				elem = elem:sub(2);
+				stanza:up(); -- TODO check for start-end tag name match
+			elseif elem:sub(-1,-1) == "/" then -- empty tag
+				elem = elem:sub(1,-2);
+				local name,attr = parse_tag(elem);
+				stanza:tag(name, attr):up();
+			else -- start tag
+				local name,attr = parse_tag(elem);
+				stanza:tag(name, attr);
+			end
+			if #text ~= 0 then -- text
+				stanza:text(xml_unescape(text));
+			end
+		end
+		return stanza.tags[1];
+	end
+end)();
+-- end of XML parser
+
 local arg, host = ...;
 local help = "/? -? ? /h -h /help -help --help";
 if not(arg and host) or help:find(arg, 1, true) then
-- 
cgit v1.2.3