Lua Xml

lua-users home
wiki

The following is some sample code for handling XML. It is divided into four sections:

Credit to the authors is mentioned where appropriate.

Toolkits

LazyKit is a collection of XML processing tools. Its primary purpose is to provoke discussion of XML tools in Lua.

PenlightLibraries provides an XML module [See Docs] which uses the LOM defined by LuaExpat? and provides pretty-printing, template matching and Orbit-style 'htmlfication'. It will use LuaExpat? if available, otherwise falls back on a pure Lua parser based on Roberto's (see below).

Lua-only XML parsers

Lua XML Parser

LuaXML-0.0.0, From Paul Chakravarti [1] (original link broken; see: [2]) (Lua 5 version available at [3])

There is a bug with parsing DTD. Replace the line with

self:_parseDTD(string,pos)
to
self:_parseDTD(str,pos)

The module implements a non-validating XML stream parser with a handler based event api (conceptually similar to SAX) which can be used to post-process the event data as required (eg into a tree).

The current functionality is -

The limitations are -

The distribution also includes sample event handlers to convert the SAX event stream into a Lua table -

SLAXML

Another pure-Lua non-validating SAX-like streaming processor. It also includes an implementation of a simple DOM parser (parse to hierarchy of tables).

https://github.com/Phrogz/SLAXML

Features:

Classic Lua-only version

From: Roberto Ierusalimschy

I have this basic skeleton that parses the "main" part of an XML string (it does not handle meta-data like "<?" and "<!"...). -- Roberto

[!] VersionNotice: The below code pertains to an older Lua version, Lua 4. It does not run as is under Lua 5.
[!] This implementation apparently does not correctly parse colons (:), as used for XML namespaces. See below for a proposed fix.

function parseargs (s)
  local arg = {}
  gsub(s, "(%w+)=([\"'])(.-)%2", function (w, _, a)
    %arg[w] = a
  end)
  return arg
end

function collect (s)
  local stack = {n=0}
  local top = {n=0}
  tinsert(stack, top)
  local ni,c,label,args, empty
  local i, j = 1, 1
  while 1 do
    ni,j,c,label,args, empty = strfind(s, "<(%/?)([%w:]+)(.-)(%/?)>", i)
    if not ni then break end
    local text = strsub(s, i, ni-1)
    if not strfind(text, "^%s*$") then
      tinsert(top, text)
    end
    if empty == "/" then  -- empty element tag
      tinsert(top, {n=0, label=label, args=parseargs(args), empty=1})
    elseif c == "" then   -- start tag
      top = {n=0, label=label, args=parseargs(args)}
      tinsert(stack, top)   -- new level
    else  -- end tag
      local toclose = tremove(stack)  -- remove top
      top = stack[stack.n]
      if stack.n < 1 then
        error("nothing to close with "..label)
      end
      if toclose.label ~= label then
        error("trying to close "..toclose.label.." with "..label)
      end
      tinsert(top, toclose)
    end 
    i = j+1
  end
  local text = strsub(s, i)
  if not strfind(text, "^%s*$") then
    tinsert(stack[stack.n], text)
  end
  if stack.n > 1 then
    error("unclosed "..stack[stack.n].label)
  end
  return stack[1]
end


-- example

x = collect[[
     <methodCall kind="xuxu">
      <methodName>examples.getStateName</methodName>
      <params>
         <param>
            <value><i4>41</i4></value>
            </param>
         </params>
      </methodCall>
]]


updated for 5.1
[!] This implementation apparently does not correctly parse colons (:), as used for XML namespaces. See below for a proposed fix.
function parseargs(s)
  local arg = {}
  string.gsub(s, "([%-%w]+)=([\"'])(.-)%2", function (w, _, a)
    arg[w] = a
  end)
  return arg
end
    
function collect(s)
  local stack = {}
  local top = {}
  table.insert(stack, top)
  local ni,c,label,xarg, empty
  local i, j = 1, 1
  while true do
    ni,j,c,label,xarg, empty = string.find(s, "<(%/?)([%w:]+)(.-)(%/?)>", i)
    if not ni then break end
    local text = string.sub(s, i, ni-1)
    if not string.find(text, "^%s*$") then
      table.insert(top, text)
    end
    if empty == "/" then  -- empty element tag
      table.insert(top, {label=label, xarg=parseargs(xarg), empty=1})
    elseif c == "" then   -- start tag
      top = {label=label, xarg=parseargs(xarg)}
      table.insert(stack, top)   -- new level
    else  -- end tag
      local toclose = table.remove(stack)  -- remove top
      top = stack[#stack]
      if #stack < 1 then
        error("nothing to close with "..label)
      end
      if toclose.label ~= label then
        error("trying to close "..toclose.label.." with "..label)
      end
      table.insert(top, toclose)
    end
    i = j+1
  end
  local text = string.sub(s, i)
  if not string.find(text, "^%s*$") then
    table.insert(stack[#stack], text)
  end
  if #stack > 1 then
    error("unclosed "..stack[#stack].label)
  end
  return stack[1]
end


Colon parsing fix (proposal)
The code above uses (%w+) to capture tag names and parameter names. However, in XML namespaces are frequently used which introduces a colon (:) in the name of the tag/parameter. To make this implementation able to handle these colons, try replacing the "(%w+)" part in the function parseargs with "([%w:]+)".

Original version

From: Yutaka Ueno

That is my test program, which is now revised [4] (link broken) But it is only tested for a few XML files used in biology. Probably Roberto's code provides a better skelton than mine, but there is a difference in xml-tag descriptions with Lua tables.

XML :           <methodCall kind="xuxu">
Lua by Roberto: { label="methodCall", args={kind="xuxu"} }
Lua by Ueno:    { xml="methodCall", kind="xuxu" }

Because the property name "xml" never appears in XML. This method is a bit better for terribly deep XML tags proposed in biology.


C bindings

Kino

From: Eckhart Koeppen

Well, there is something I programmed which might help, the Kino XML processor. It has wrappers for Tcl and Lua via SWIG. A Xt and an experimental Gtk widget for displaying XML with CSS are also available. Take a look at it at: [5] (link broken)

It is under constant development but tries to stick to the DOM, so I hope that the interface changes remain small.

libxml

[luagnome] (link broken, use [6]) includes the wrapping of libxml-1.8.x, as it is considered as a part of Gnome. It allows to parse and to generate XML files, with a simple api (object oriented).

[lua-xmlreader] is an implementation of the XmlReader API using libxml2.

Expat

For Lua 5.0/5.1, use [LuaExpat], which is full-featured.

For Lua 4.0:

From: Jay Carlson

I've put a simple binding of expat, James Clark's C stream-based XML parser up at [7]. No, not everything is bound, but it should be obvious how to bind more stuff to it.

LuaXML

[LuaXML] is a lean yet complete module for the direct mapping between XML data and Lua tables.

XML-DOM parser

PugXML is a C++ small, fast, non-validating DOM XML parser, contained in a single header, having no dependencies other than the standard C libraries, and <iostream> (KERNEL32.DLL with WIN32). This XML parser segments a given string in situ (like strtok), performing scanning/tokenization, and parsing in a single pass.

Here is an example of the parser use in Lua:

-- create xml_parser object
parser = pug.xml_parser( pug.xml_parser.parse_default, true, 4);

-- parse string
xml_string = '<xml><child>some data </child><child2 attr="value"/></xml>';
print('parsing string: ' .. xml_string );
parser:parse(xml_string, pug.xml_parser.parse_noset);
print( tostring(parser:document()) );

-- Testing xml_node
-- getting root
root=parser:document();

-- add a element child
child=root:append_child( pug.xml_node_type.element );
print( tostring(root) );

-- rename child to child
child:name('child');
print('child name is ' .. child:name() );
print( tostring(root) );

-- adding attributes
child:append_attribute('attribute','value');
child:append_attribute('attribute2','value2');

-- adding on children
child2=child:append_child( pug.xml_node_type.element );
child2:name('child2');
print( tostring(root) );

A wrapper around this parser has been written with [LuaBind] and is available at [8] (link broken). The original article about PugXML is located at [9].

TinyXML

For Lua 5.0:

From: Robert Noll

Just a plain "Parse File to lua array" function in c++, using the [TinyXML] (2.4.3) lib.

// header

class lua_State;
	
/// register parser functions to lua
void	RegisterLuaXML (lua_State *L);


// sourcefile

#include "tinyxml.h"

extern "C" {
	#include "lua.h"
	#include "lauxlib.h"
	#include "lualib.h"
}

void LuaXML_ParseNode (lua_State *L,TiXmlNode* pNode) { PROFILE
	if (!pNode) return;
	// resize stack if neccessary
	luaL_checkstack(L, 5, "LuaXML_ParseNode : recursion too deep");
	
	TiXmlElement* pElem = pNode->ToElement();
	if (pElem) {
		// element name
		lua_pushstring(L,"name");
		lua_pushstring(L,pElem->Value());
		lua_settable(L,-3);
		
		// parse attributes
		TiXmlAttribute* pAttr = pElem->FirstAttribute();
		if (pAttr) {
			lua_pushstring(L,"attr");
			lua_newtable(L);
			for (;pAttr;pAttr = pAttr->Next()) {
				lua_pushstring(L,pAttr->Name());
				lua_pushstring(L,pAttr->Value());
				lua_settable(L,-3);
				
			}
			lua_settable(L,-3);
		}
	}
	
	// children
	TiXmlNode *pChild = pNode->FirstChild();
	if (pChild) {
		int iChildCount = 0;
		for(;pChild;pChild = pChild->NextSibling()) {
			switch (pChild->Type()) {
				case TiXmlNode::DOCUMENT: break;
				case TiXmlNode::ELEMENT: 
					// normal element, parse recursive
					lua_newtable(L);
					LuaXML_ParseNode(L,pChild);
					lua_rawseti(L,-2,++iChildCount);
				break;
				case TiXmlNode::COMMENT: break;
				case TiXmlNode::TEXT: 
					// plaintext, push raw
					lua_pushstring(L,pChild->Value());
					lua_rawseti(L,-2,++iChildCount);
				break;
				case TiXmlNode::DECLARATION: break;
				case TiXmlNode::UNKNOWN: break;
			};
		}
		lua_pushstring(L,"n");
		lua_pushnumber(L,iChildCount);
		lua_settable(L,-3);
	}
}

static int LuaXML_ParseFile (lua_State *L) { PROFILE
	const char* sFileName = luaL_checkstring(L,1);
	TiXmlDocument doc(sFileName);
	doc.LoadFile();
	lua_newtable(L);
	LuaXML_ParseNode(L,&doc);
	return 1;
}

void	RegisterLuaXML (lua_State *L) {
	lua_register(L,"LuaXML_ParseFile",LuaXML_ParseFile);
}

pugilua

a binding to the neat [pugixml], supporting DOM and XPath 1.0. The binding is made using [LuaBridge]

https://github.com/d-led/pugilua

Example:

require 'pugilua'


---- reading ----
local doc=pugi.xml_document()
local res=doc:load_file [[..\..\scripts\pugilua\pugilua.vcxproj]]

print(res.description)

local node1=doc:root():child('Project')
local query1=doc:root():select_nodes('Project/PropertyGroup')

local n=query1.size
for i=0,n-1 do
	local node=query1:get(i):node()
	local attribute=query1:get(i):attribute()
	print(node.valid,node.path)
	local a=node:first_attribute()
	while a.valid do
		print(a.name)
		a=a:next_attribute()
	end
end

---- creating ----
doc:reset()
--- from the tutorial
-- add node with some name
local node = doc:root():append_child("node")

-- add description node with text child
local descr = node:append_child("description")
descr:append(pugi.node_pcdata):set_value("Simple node")

-- add param node before the description
local param = node:insert_child_before("param", descr)

-- add attributes to param node
param:append_attribute("name"):set_value("version")
param:append_attribute("value"):set_value(1.1)
param:insert_attribute_after("type", param:attribute("name")):set_value("float")

doc:save_file("tutorial.xml")

xerceslua

As a supplement to pugilua there's an effort to provide a minimal binding to [xerces.apache.org/xerces-c/ Xerces-C++] to be able to validate xml documents:

https://github.com/d-led/xerceslua

assert(require 'xerceslua')

Example:

local parser=xerces.XercesDOMParser()
parser:loadGrammar("Employee.dtd",xerces.GrammarType.DTDGrammarType)
parser:setValidationScheme(xerces.ValSchemes.Val_Auto)
local log=parser:parse("Employeexy.xml")
print(log.Ok)
if not log.Ok then
    print(log.Count)
    for i=0,log.Count-1 do
        local err=log:GetLogEntry(i)
        print(err.SystemId..', l:'..err.LineNumber..', c:'..err.ColumnNumber..', e:'..err.Message,err.LogType)
    end
end


XML-based protocols

XML-RPC

For Lua 5.0/5.1, use [LuaXMLRPC] library developed by [The Kepler Project].

For Lua 4.0:

From: Jay Carlson

I've put an initial release of client/server bindings for Lua for XML-RPC at [10]. It contains my lxp expat binding, and uses LuaSocket for client transport.

For more information on XML-RPC, see [11].

Although the packaging and documentation are scant, this package successfully passes the validation tests at [12].

SOAP

[LuaSOAP] is a Lua library to ease the use of SOAP.

Lua only XmlParser?

For Lua 5.1:

From: Alexander Makeev

This XmlParser? allows build object like C# XmlDocument? with XmlNodes?. See example for details.

-----------------------------------------------------------------------------------------
-- LUA only XmlParser from Alexander Makeev
-----------------------------------------------------------------------------------------
XmlParser = {};

function XmlParser:ToXmlString(value)
	value = string.gsub (value, "&", "&amp;");		-- '&' -> "&amp;"
	value = string.gsub (value, "<", "&lt;");		-- '<' -> "&lt;"
	value = string.gsub (value, ">", "&gt;");		-- '>' -> "&gt;"
	--value = string.gsub (value, "'", "&apos;");	-- '\'' -> "&apos;"
	value = string.gsub (value, "\"", "&quot;");	-- '"' -> "&quot;"
	-- replace non printable char -> "&#xD;"
   	value = string.gsub(value, "([^%w%&%;%p%\t% ])",
       	function (c) 
       		return string.format("&#x%X;", string.byte(c)) 
       		--return string.format("&#x%02X;", string.byte(c)) 
       		--return string.format("&#%02d;", string.byte(c)) 
       	end);
	return value;
end

function XmlParser:FromXmlString(value)
  	value = string.gsub(value, "&#x([%x]+)%;",
      	function(h) 
      		return string.char(tonumber(h,16)) 
      	end);
  	value = string.gsub(value, "&#([0-9]+)%;",
      	function(h) 
      		return string.char(tonumber(h,10)) 
      	end);
	value = string.gsub (value, "&quot;", "\"");
	value = string.gsub (value, "&apos;", "'");
	value = string.gsub (value, "&gt;", ">");
	value = string.gsub (value, "&lt;", "<");
	value = string.gsub (value, "&amp;", "&");
	return value;
end
   
function XmlParser:ParseArgs(s)
  local arg = {}
  string.gsub(s, "(%w+)=([\"'])(.-)%2", function (w, _, a)
    	arg[w] = self:FromXmlString(a);
  	end)
  return arg
end

function XmlParser:ParseXmlText(xmlText)
  local stack = {}
  local top = {Name=nil,Value=nil,Attributes={},ChildNodes={}}
  table.insert(stack, top)
  local ni,c,label,xarg, empty
  local i, j = 1, 1
  while true do
    ni,j,c,label,xarg, empty = string.find(xmlText, "<(%/?)([%w:]+)(.-)(%/?)>", i)
    if not ni then break end
    local text = string.sub(xmlText, i, ni-1);
    if not string.find(text, "^%s*$") then
      top.Value=(top.Value or "")..self:FromXmlString(text);
    end
    if empty == "/" then  -- empty element tag
      table.insert(top.ChildNodes, {Name=label,Value=nil,Attributes=self:ParseArgs(xarg),ChildNodes={}})
    elseif c == "" then   -- start tag
      top = {Name=label, Value=nil, Attributes=self:ParseArgs(xarg), ChildNodes={}}
      table.insert(stack, top)   -- new level
      --log("openTag ="..top.Name);
    else  -- end tag
      local toclose = table.remove(stack)  -- remove top
      --log("closeTag="..toclose.Name);
      top = stack[#stack]
      if #stack < 1 then
        error("XmlParser: nothing to close with "..label)
      end
      if toclose.Name ~= label then
        error("XmlParser: trying to close "..toclose.Name.." with "..label)
      end
      table.insert(top.ChildNodes, toclose)
    end
    i = j+1
  end
  local text = string.sub(xmlText, i);
  if not string.find(text, "^%s*$") then
      stack[#stack].Value=(stack[#stack].Value or "")..self:FromXmlString(text);
  end
  if #stack > 1 then
    error("XmlParser: unclosed "..stack[stack.n].Name)
  end
  return stack[1].ChildNodes[1];
end

function XmlParser:ParseXmlFile(xmlFileName)
	local hFile,err = io.open(xmlFileName,"r");
	if (not err) then
		local xmlText=hFile:read("*a"); -- read file content
		io.close(hFile);
        return self:ParseXmlText(xmlText),nil;
	else
		return nil,err;
	end
end
------------------------------------------------------------------------------------------

example:

function dump(_class, no_func, depth)
	if(not _class) then 
		log("nil");
		return;
	end
	
	if(depth==nil) then depth=0; end
	local str="";
	for n=0,depth,1 do
		str=str.."\t";
	end
    
	log(str.."["..type(_class).."]");
	log(str.."{");
    
	for i,field in pairs(_class) do
		if(type(field)=="table") then
			log(str.."\t"..tostring(i).." =");
			dump(field, no_func, depth+1);
		else 
			if(type(field)=="number") then
				log(str.."\t"..tostring(i).."="..field);
			elseif(type(field) == "string") then
				log(str.."\t"..tostring(i).."=".."\""..field.."\"");
			elseif(type(field) == "boolean") then
				log(str.."\t"..tostring(i).."=".."\""..tostring(field).."\"");
			else
				if(not no_func)then
					if(type(field)=="function")then
						log(str.."\t"..tostring(i).."()");
					else
						log(str.."\t"..tostring(i).."<userdata=["..type(field).."]>");
					end
				end
			end
		end
	end
	log(str.."}");
end

--local obj,err = XmlParser:ParseXmlFile("test.xml");
--if(not err) then
--	dump(obj);
--else
--	log("ERROR: "..err);		
--end

local xmlTree=XmlParser:ParseXmlText([[<?xml version="1.0" encoding="utf-8"?>
<Config>
	<EntityList>
		<Entity value="1&quot;2&quot;3">innerText</Entity>	
		<Entity value="456"/>
	</EntityList>
</Config>
]])
for i,xmlNode in pairs(xmlTree.ChildNodes) do
	if(xmlNode.Name=="EntityList") then
		for i,subXmlNode in pairs(xmlNode.ChildNodes) do
			if(subXmlNode.Name=="Entity") then
				log("Entity value=\""..subXmlNode.Attributes.value.."\"");
				if(subXmlNode.Value) then
					log("   Content=\""..subXmlNode.Value.."\"");
				end
			end
		end
	end
end
dump(xmlTree)

result:

<log>Entity value="1"2"3"
<log>   Content="innerText"
<log>Entity value="456"
<log>	[table]
<log>	{
<log>		Attributes =
<log>		[table]
<log>		{
<log>		}
<log>		Name="Config"
<log>		ChildNodes =
<log>		[table]
<log>		{
<log>			1 =
<log>			[table]
<log>			{
<log>				Attributes =
<log>				[table]
<log>				{
<log>				}
<log>				Name="EntityList"
<log>				ChildNodes =
<log>				[table]
<log>				{
<log>					1 =
<log>					[table]
<log>					{
<log>						Value="innerText"
<log>						Attributes =
<log>						[table]
<log>						{
<log>							value="1"2"3"
<log>						}
<log>						Name="Entity"
<log>						ChildNodes =
<log>						[table]
<log>						{
<log>						}
<log>					}
<log>					2 =
<log>					[table]
<log>					{
<log>						Attributes =
<log>						[table]
<log>						{
<log>							value="456"
<log>						}
<log>						Name="Entity"
<log>						ChildNodes =
<log>						[table]
<log>						{
<log>						}
<log>					}
<log>				}
<log>			}
<log>		}
<log>	}

RecentChanges · preferences
edit · history
Last edited February 17, 2014 12:27 pm GMT (diff)