XML::DT - a package for down translation of XML to strings


    use XML::DT;
    %xml=( 'music'    => sub{"Music from: $c\n"},
           'lyrics'   => sub{"Lyrics from:$c\n (the value of attribute
                               IN is:$v{IN}\n)"},
           'title'    => sub{ uc($c) },
           '-default' => sub{"$q:$c"},
           '-outputenc' => 'ISO-8859-1');

    print dt($filename,%xml);
    print dtstring("<arq>
                    <title>Vejam Bem</title>
                    <music>Zeca Afonso</music>
    ctxt(1)       /* the father element */


This module processes XML files with an approach similar to OMNIMARK.

Down translation function dt receives a filename and a set of expressions (functions) defining the processing and associated values for each element.

dtstring is similar but takes input from a string instead of a file.

pathdt function

The pathdt function uses a subset of XPath as key in the handler. Example:

   %handler = (
        "article/title" => sub{ toxml("h1",{},$c) },
        "section/title" => sub{ toxml("h2",{},$c) },
        "title"         => sub{ $c },
        "//image[@type='jpg']" => sub{ "JPEG: <img src=\"$c\">" },
        "//image[@type='bmp']" => sub{ "BMP: sorry, no bitmaps on the web" },

Here are some examples of valid XPath expressions under XML::DT:

  //ccc                           - ccc somewhere (same as "ccc")
  //*                             - same as "-default"
  /aaa[@id]                       - aaa with an attribute id
  /*[@*]                          - root with an attribute
  /aaa[not(@name)]                - aaa with no attribute "name"
  //bbb[@name='foo']              - ... attribute "name" = "foo"
  //*[name()='bbb']               - complex way of saying "//bbb"
  //*[starts-with(name(),'aa')]   - an element named "aa.*"
  //*[contains(name(),'c')]       - an element       ".*c.*"
  //aaa[string-length(name())=4]  -                  "...."
  //aaa[string-length(name())&lt;4]                  ".{1,4}"
  //aaa[string-length(name())&gt;5]                  ".{5,}"

For more information, visit or try a tutorial under

inctxt function

inctxt(pattern) is true if the actual element path matches the provided pattern. This function is meant to be used in the element functions in order to achieve context dependent processing.

User provided element processing functions

The user must provide an HASH with a function for each element, that computes element output. Functions can use the element name $q, the element content $c and the attribute values hash %v.

All those global variables are defined in $CALLER::.

Each time an element is find the associated function is called.

Content is calculated by concatenation of element contents strings and interior elements return values.

-default function

When a element has no associated function, the function associated with -default called. If no -default function is defined the default function returns a XML like string for the element.

When you use /-type definitions, you often need do set -default function to return just the contents: sub{$id}.

-outputenc option

-outputenc defines the output encoding (default is Unicode UTF8).

-inputenc option

-inputenc forces a input encoding type. Whenever that is possible, define the input encoding in the XML file:

   <?xml version='1.0' encoding='ISO-8859-1'?>

-pcdata function

-pcdata function is used to define transformation over the contents. Typically this function should look at context (see inctxt function)

The default -pcdata function is the identity

-begin function

Function to be executed before processing XML file.

Example of use: initialization of side-effect variables

-end function

Function to be executed after processing XML file. I can use $c content value. The value returned by -end will be the dt return value.

Example of use: post-processing of returned contents

toxml function

This is the default ``-default'' function. It can be used to generate xml based on $c $q and %v variables. Example: add a new attribute to element ele1 without changing it:

  %handler=( ...
             ele1 => sub { $v{at1} = "v1"; toxml(); },

toxml can also be used with 3 arguments: tag, attrigutes and contents

  toxml("a",{href=> "http://local/f.html";}, "example")


  <a href='http://local/f.html'>example</a>

Elements with values other than strings (-type)

By default all elements return strings, and contents ($c) is the concatenation of the strings returned by the sub-elements.

In some situations the XML text contains values that are better processed as a structured type.

The following types (functors) are available:

     STR  -> concatenates all the subelements returned values (DEFAULT)
          all the subelement should return strings to be concatenated
     SEQ  -> makes an ARRAY with all the sub elements contents; attributes are
          ignored (they should be processed in the subelement). (returns a ref)
     SEQH -> makes an ARRAY of HASH with all the sub elements (returns a ref);
          for each subelement: 
                 -q  => element name
                 -c  => contents
                 at1 => at value1    for each attribute
     MAP  -> makes an HASH with the sub elements; keys are the sub-element
          names, values are their contents. Attributes are ignored. (they should
          be processed in the subelement) (returns a ref)
     MULTIMAP -> makes an HASH of ARRAY; keys are the sub-element names;
         values are lists of contents; attributes are ignored (they should be
         processed in the subelement); (returns a ref)
     MMAPON(elementlist) -> makes an HASH with the subelements; 
          keys are the sub-element names, values are their contents; 
          attributes are ignored (they should be processed in the subelement);
          for all the elements contained in the elementlist, it is created 
          an ARRAY with their contents. (returns a ref)
     ZERO -> don't process the subelements; return ""

When you use /-type definitions, you often need do set -default function returning just the contents sub{$id}.

An example:

   use XML::DT;
   %handler = ( contacts => sub{ [ split(";",$c)] },
                -default => sub{$c},
                -type    => { institution => 'MAP',
                              degrees     =>  MMAPON('name')
                              tels        => 'SEQ' }
   $a = dt ("f.xml", %handler);

with the following f.xml

      <name>University of Minho</name>
      <contacts>J.Joao; J.Rocha; J.Ramalho</contacts>
     <name>Computer science</name>
     <name>Informatica </name>
     <name> history </name>

would make $a

  { 'name' => [ 'Computer science',
                'Informatica ',
                ' history ' ],
    'institution' => { 'tels' => [ 1111,
                                   1113 ],
                       'name' => 'University of Minho',
                       'where' => 'Portugal',
                       'id' => 'U.M.',
                       'contacts' => [ 'J.Joao',
                                       ' J.Rocha',
                                       ' J.Ramalho' ] } };

DT Skeleton generation

It is possible to build an initial processor program based on an example

To do this use the function mkdtskel(filename).


  perl -MXML::DT -e 'mkdtskel "f.xml"' >

DTD skeleton generation

It makes a naive DTD based on an example(s).

To do this use the function mkdtdskel(filename*).


  perl -MXML::DT -e 'mkdtdskel "f.xml"' > f.dtd


This section is out of date...


Jose Joao,

Alberto Simoes <>

thanks to

  Michel Rodriguez <>
  Josť Carlos Ramalho <>

NAME - module for unicode utf8 to latin1 translation


   $latin1string = lat1::utf8($utf8string)


Translating the latin1 subset of unicode utf8 is very simples and needs no tables.

If you need more complex translation, see the perl modules about unicode and the recode command.