++ed by:

2 non-PAUSE users.

Author image Andrea Guzzo


XML::TinyXML - Little and efficient Perl module to manage xml data.


  use XML::TinyXML;

  # First create an XML Context
  $xml = XML::TinyXML->new();
  # and add a root node
  $xml->addRootNode('nodelabel', 'somevalue', { attr1 => v1, attr2 => v2 });

  # if you want to reuse a previously created node object ...
  # ( you can create one calling :
  #    * %attrs = ( attr1 => v1, attr2 => v2 );
  #    * $node = XML::TinyXML::Node->new('nodelabel', param => 'somevalue', attrs => \%attrs);
  # )
  # the new XML Context can be created giving the root node directly to the constructor
  $xml = XML::TinyXML->new($node);

  # A better (and easier) option is also to let the context constructor create the rootnode for you:
  $xml = XML::TinyXML->new('rootnode', param => 'somevalue', attrs => { attr1 => v1, attr2 => v2 });

  # an empty root node can be created as well:
  $xml = XML::TinyXML->new('rootnode');

  # You can later obtain a reference to the node object using the getNode() method
  $node = $xml->getNode('/rootnode'); 

  # the leading '/' is optional ... since all paths will be considered absolute and 
  # first element is assumed to be always a root node
  $node = $xml->getNode('rootnode');

  # xpath-like predicates are also supported so you can also do something like
  $node->addChildNode('child', 'value1');
  $node->addChildNode('child', 'value2');
  $child2 = $xml->getNode('/rootnode/child[2]');

  # or even 
  $node->addChildNode('child', 'value3', { attr => 'val' });
  # the 3rd 'child' has an attribute
  $child3 = $xml->getNode('/rootnode/child[@attr="val"]');

  #### IMPORTANT NOTE: #### 
  # this is not xpath syntax. use XML::TinyXML::Selector::XPath 
  # if you want to use an xpath-compliant syntax

  # see XML::TinyXML::Node documentation for further details on possible
  # operations on a node reference

  ########                                            #########
  ########## hashref2xml and xml2hashref facilities ###########
  ########                                            #########
  # An useful facility is loading/dumping of hashrefs from/to xml
  # for ex:
  $hashref = { some => 'thing', someother => 'thing' };
  my $xml = XML::TinyXML->new($hashref, param => 'mystruct');

  # or to load on an existing XML::TinyXML object
  $xml->loadHash($hashref, 'mystruct');

  # we can also create and dump to string all at once :
  my $xmlstring = XML::TinyXML->new($hashref, param => 'mystruct')->dump;

  # to reload the hashref back
  my $hashref = $xml->toHash;


Since in some environments it could be desirable to avoid installing Expat, XmlParser and blahblahblah, needed by most XML-related perl modules, my main scope was to obtain a fast xml library usable from perl (so with the possibility to expose a powerful api) but without the need to install a lot of other modules (or even C libraries) to have it working.

The actual version of XML::TinyXML (0.18 when I'm writing this) implements an xpath selector and supports encodings (through iconv).

The OO Tree-based api allows to :

    - create a new document programmaticaly
    - import an existing document from a file
    - import an existing document by parsing a memory buffer
    - modify the loaded/created document programmatically
    - export the document to an xml file
    - bidirectional conversion between xml documents and perl hashrefs

There are other "no-dependencies" tiny xml implementations on CPAN. Notably : XML::Easy and XML::Tiny but both are still missing some key features which are actually implemented in this module and are required by the projects where I use it, in particular :

    - an OO api to allow changing and re-exporting the xml document,
    - an easy-to-use bidirectional xml<->hashref conversion 
    - encoding-conversion capabilities
    - xpath selectors
    - small memory footprint (well... it's always a perl module)

The underlying xml implementation resides in txml.c and is imported in the perl context through XS. It uses a linkedlist implementation took out of freebsd kernel and is supposed to be fast enough to represent and access 'not-huge' xml trees. If at some stage there will be need for more preformance in accessing the xml data (especially when using the xpath selector) an hash-table based index could be built on top of the tree structure to speed-up access to inner parts of the tree when using paths.

An Event-based api is actually missing, but will be provided in future releases.


  • new ($arg, %params)

    Creates a new XML::TinyXML object.

    $root can be any of :


    and if present will be used as first root node of the newly created xml document.

    %params is an optional hash parameter used only if $arg is an HASHREF or a scalar

        %params = (
            param =>  * if $root is an hashref, this will be
                          name of the root node (it will be passed 
                          as second argument to loadHash())
                      * if $root is a scalar, this will be 
                          the value of the root node.
            attrs =>  attributes of the 'contextually added' $root node 
            encoding => output encoding to use (among iconv supported ones)
  • addNodeAttribute ($node, $key, $value)

    Adds an attribute to a specific $node

    $node MUST be an XML::TinyXML::Node object.

    $key is the name of the attribute $value is the value of the attribute

    This method is just an accessor. See XML::TinyXML::Node::addAttributes() instead.

  • removeNodeAttribute ($node, $index)

    Removes from $node the attribute at $index if present.

    $node MUST be a XML::TinyXML::Node object.

    This method is just an accessor. See XML::TinyXML::Node::removeAttribute() instead.

  • addRootNode ($name, $val, $attrs)

    Adds a new root node. (This can be considered both as a new tree in the forest represented in the xml document or as a new branch in the xml tree represented by the document itself)

  • addChildNode ($parent, $name, $val, $attrs)

    Adds a new child node. This method is exactly like addRootNode but first argument must be a valid XML::TinyXML::Node which will be the parent of the newly created node

  • dump ()

    Returns a stringified version of the XML structure represented internally

  • loadFile ($path)

    Load the xml structure from a file

  • loadHash ($hash, $root)

    Load the xml structure from an hashref (AKA: convert an hashref to an xml document)

    if $root is specified, it will be the entity name of the root node in the resulting xml document.

  • toHash ()

    Dump the xml structure represented internally in the form of an hashref

  • loadBuffer ($buf)

    Load the xml structure from a preloaded memory buffer

  • getNode ($path)

    Get a node at a specific path.

    $path must be of the form: '/rootnode/child1/child2/leafnod' and the leading '/' is optional (since all paths will be interpreted as absolute)

    Returns an XML::TinyXML::Node object

  • getChildNode ($node, $index)

    Get the child of $node at index $index.

    Returns an XML::TinyXML::Node object

  • countChildren ()

    Returns the number of $node's children

  • removeNode ($path)

    Remove the node at specific $path, if present. See getNode() documentation for some notes on the $path format.

    Returns XML_NOERR (0) if success, error code otherwise.

    See Exportable constants for a list of possible error codes

  • getBranch ($index)

    alias for getRootNode

  • getRootNode ($index)

    Get the root node at $index.

    Returns an XML::TinyXML::Node object if present, undef otherwise

  • removeBranch ($index)

    Alias for removeRootNode

  • removeRootNode ($index)

    Remove the rootnode (and all his children) at $index.

  • countRootNodes

    Returns the number of root nodes within this context

  • rootNodes ()

    Returns an array containing all rootnodes. In scalar context returns an arrayref.

  • getChildNodeByName ($node, $name)

    Get the child of $node with name == $name.

    Returns an XML::TinyXML::Node object if there is such a child, undef otherwise

  • save ($path)

    Save the xml document represented internally into $path.

    Returns XML_NOERR if success, a specific error code otherwise

  • setOutputEncoding ($encoding)

    Sets the output encding to the specified one

    (any iconv-supported encoding is a valid argument for this method)

    Returns XML_NOERR if success, a specific error code otherwise

  • allowMultipleRootNodes ($bool)

    Allow to control if the document can contain multiple root nodes (out of xml spec) or if it will support only one single root node. Default is 0

  • ignoreBlanks ($bool)

    Controls the behaviour of both the parser and the dumper.

    - Parser :

    If ignoreBlanks is true any 'tab', 'newline' and 'carriage-return' ("\t", "\n", "\r"), between two tags will be ignored, unless surrounded by non-whitespace character. For example, considering the following node:


    ( literally : "<parent>\n\t<child>value</child>\n</parent>" )

    the parser will ignore the newlines and the tab between two nodes. So, the resulting structure would be :

        $node = {
                  value => undef,
                  children => [ { child => "value" } ]

    If ignoreBlanks is false, all characters between the node tags will be part of the value and the result would be :

        $node = {
                  value => "\n\t\n",
                  children => [ { child => "value" } ]

    - Dumper :

    If ignoreBlanks is true, both newlines and tabs will be used to prettify text formatting, so the node in the previous example will be printed out as:


    If ignoreBlanks is false, no extra characters will be added to the xml data (so no newlines, indentation or such). The previous example would be dumped as follows:


    Default is 1.

  • ignoreWhiteSpaces ($bool)

    Controls the behaviour of the parser.

    This flag includes the literal whitespace ' ' among the ignored characters commended by ignoreBlanks()

    consider the following examples:

        <child> </child> 
     - if true:
         value = ""
     - if false:
         value = " "

    The parser will ignore the whitespace if this flag is true.

        <child> a value </child>
        <child> a value</child>
        <child>a value </child>
     - if true:
         value = "a value"
     - if false:
         value = " a value "
                 " a value"
                 "a value "

    The whitespace will be part of the value if this flag is false.

    The dumper is not affected by this flag.

    Default is 1


None by default.

Exportable constants


Exportable C functions

  TXml *XmlCreateContext()
  void XmlDestroyContext(TXml *xml)
  XmlNode *XmlCreateNode(char *name,char *val,XmlNode *parent);
  char *XmlGetNodeValue(XmlNode *node);
  XmlErr XmlSetNodeValue(XmlNode *node,char *val);
  void XmlDestroyNode(XmlNode *node);
  int XmlAddAttribute(XmlNode *node, char *name, char *val)
  int XmlAddRootNode(TXml *xml, XmlNode *node)
  int XmlRemoveNode(TXml *xml,char *path);
  void XmlRemoveChildNodeAtIndex(XmlNode *node, unsigned long index);
  void XmlRemoveChildNode(XmlNode *parent, XmlNode *child);
  XmlErr XmlAddChildNode(XmlNode *parent,XmlNode *child);
  XmlNode *XmlGetChildNode(XmlNode *node, unsigned long index)
  XmlNode *XmlGetChildNodeByName(XmlNode *node,char *name);
  XmlNode *XmlGetNode(TXml *xml,  char *path)
  unsigned long XmlCountBranches(TXml *xml)
  int XmlRemoveBranch(TXml *xml, unsigned long index)
  XmlNode *XmlGetBranch(TXml *xml,unsigned long index);
  int XmlSubstBranch(TXml *xml,unsigned long index, XmlNode *newBranch);
  int XmlParseBuffer(TXml *xml, char *buf)
  int XmlSave(TXml *xml, char *path)
  char *XmlDump(TXml *xml, int *outlen)



You should also see libtinyxml documentation (mostly txml.h, redistributed with this module)

If you need a more complete and featureful XML implementation try looking at XML::LibXML XML::Parser XML::API XML::XPath or other modules based on either expat or libxml2


xant, <xant@cpan.org>


Copyright (C) 2008-2010 by xant

This library is free software; you can redistribute it and/or modify it under the same terms as Perl itself, either Perl version 5.8.8 or, at your option, any later version of Perl 5 you may have available.