XML::LibXML::PrettyPrint - add pleasant whitespace to a DOM tree

     my $document = XML::LibXML->new->parse_file('in.xml');
     my $pp = XML::LibXML::PrettyPrint->new(indent_string => "  ");
     $pp->pretty_print($document); # modified in-place
     print $document->toString;

    Long XML files can be daunting for humans to read. Of course, XML is
    really designed for computers to read - not people - but there are times
    when mere mortals do need to read and edit XML by hand. For example, if
    your application stores its configuration in XML, or you need to dump some
    XML to STDOUT for debugging purposes.

    Syntax highlighting helps, but to really make sense of some XML, proper
    indentation can be vital. Hence `XML::LibXML::PrettyPrint` - it can be
    applied to an XML::LibXML DOM tree to reformat it into a more readable

    Pretty-printing XML is not as CPU-efficient as dumping it out sloppily, so
    unless you're pretty sure that a human is going to need to make sense of
    your XML, you should probably not use this module.

        Constructs a pretty-printer object.


        *   indent_string - The string to use to indent each line. Defaults to
            a single tab character. Setting it to a non-whitespace character
            is allowed, but will carp a warning.

        *   new_line - The string to use to begin a new line. Defaults to

        *   element - A hashref of element categorisations. Each
            categorisation is a reference to an array of element names or
            callback functions. Element names may use Clark notation.

              my $callback = sub {
                my $node = shift;
                return 1 if $node->hasAttribute('is_block');
                return undef;
              my $pp = XML::LibXML::PrettyPrint->new(
                  element => {
                      inline   => [qw/span strong em b i a/],
                      block    => [qw/p div body html head/, $callback],
                      compact  => [qw/title caption li dd dt th td/],
                      preserves_whitespace => [qw/pre script style/],

            Callbacks should return 1 (true), 0 (false) or undef (dunno).

        Constructs a pretty printer object pre-configured to be suitable for
        HTML and XHTML. The indent_string and new_line options are supported.

    If you just need to use a default configuration (no options passed to the
    constructor, then you can call these as class methods, unless otherwise

        Strips superfluous whitespace from an `XML::LibXML::Document` or

        Whitespace just before, just after or leading/trailing within an
        inline element is not considered superfluous. Runs of multiple
        whitespace characters are replaced with a single space. Whitespace is
        not changed within an element that preserves whitespace.

        The node is modified in place.

    `indent($node, $level)`
        Indents the node to a certain indentation level, and its direct
        children to `$level + 1`, grandchildren to `$level + 2`, etc.
        Typically you'd just want to indent the root node to level 0.

        The node is modified in place.

        Elements that preserve whitespace are not changed.

    `pretty_print($node, $level)`
        Strip whitespace and indent. The node is modified in place and

        Example use as a class method:

         print XML::LibXML::PrettyPrint

        Returns the string that would be used to indent something to a
        particular level. Descendent classes could override this method to do
        funky indentation, such as having varying levels of indentation.

        Returns the string that would be used to begin a new line.

        Returns EL_INLINE, EL_BLOCK, EL_COMPACT or undef.

        Boolean indicating whether the contents of the element have
        significant whitespace that needs preserving.

        Returns undef if $node is not an `XML::LibXML::Element`.

    `print_xml $xml`
        Given an XML string or an XML::LibXML::Node object, prints it nicely.

        This function is not exported by default, but can be requested:

         use XML::LibXML::PrettyPrint 0.001 qw(print_xml);

        Use like this:

         print_xml '<foo> <bar> </bar> </foo>';

    `IO::Handle::print_xml($handle, $xml)`
        Partly experimental, partly mental. You can enable this feature like

         use XML::LibXML::PrettyPrint 0.001 qw(-io);

        And that will allow stuff like this to work:

         open LOG, '>mylog.xml';
         print_xml LOG '<foo> <bar> </bar> </foo>';
         close LOG;

         open my $log, '>otherlog.xml';
         print_xml $log '<foo> <bar> </bar> </foo>';
         close $log;

         print_xml STDERR '<foo> <bar> </bar> </foo>';

    These can be exported:

     use XML::LibXML::PrettyPrint 0.001 qw(:constants);


    There are three categories of element: inline, block and compact.

    For inline elements the presence of whitespace (though not the amount of
    whitespace) is considered significant just before the element, just after
    the element, or just within the element.

    In XHTML, consider the difference between the block element `<div>`:

     <div>Will</div><div>Carlton</div> <div>Ashley</div>

    and the inline element `<span>`:

     <span>Spider</span>-<span>Man</span> <span>lives</span>

    The space or lackthereof between `<div>` elements does not matter one
    whit. The lack of spaces between the first two `<span>` elements allows
    them to be read as a single (in this case, hyphenated) word, whereas the
    space before the third `<span>` separates out the word "lives".

    In terms of indentation, inline elements do not start a new indented line,
    unless they are the first element within their block, or are preceded by a
    block or compact element.

    Block elements always start a new line, and cause their child nodes to be
    indented to the next level.

    Compact elements are somewhere in-between. When it comes to whitespace
    stripping, they're treated as block elements. In terms of indentation,
    they always start a new line, but they only cause their child nodes to be
    indented to the next level if they have block descendents. If we imagine
    that in HTML, `<ul>` is a block element, `<i>` is an inline element, and
    `<li>` is a compact element:

       <li>Will Smith - Will Smith</li>
       <li>Carlton Banks - Alfonso Ribeiro</li>
         Vivian Banks:
           <li>Janet Hubert-Whitten <i>(seasons 1-3)</i></li>
           <li>Daphne Maxwell Reid <i>(seasons 3-6)</i></li>

    The third `<li>` element is indented like a block element because it
    contains a block `<ul>` element. The other `<li>` elements do not have
    their contents indented, because they contain only inline content.

    Elements default to being block, but you can specify particular elements
    as inline or compact by passing node names or callbacks to the
    constructor. Elements default to not preserving whitespace unless they
    have an `xml:space="preserve"` attribute, but again you can use the
    constructor to change this.

    Comments and processing instructions default to being compact, but you can
    make particular comments or PIs inline or block by passing appropriate
    callbacks to the constructor. Whitespace within comments and PIs is always
    preserved. (There is rarely any reason to make comments and processing
    instructions block, but making them inline can occasionally be useful, as
    it will mean that the presence of whitespace just before or just after the
    comment is treated as significant.)

    Text nodes are always inline.

    Please report any bugs to

    Related: XML::LibXML, HTML::HTML5::Writer.

    XML::Tidy - similar, but based on XML::XPath. Doesn't differentiate
    between inline and block elements.

    XML::Filter::Reindent - similar again, based on XML::Parser. Doesn't
    differentiate between inline and block elements.

    Sermon: <>. Read it.

    Toby Inkster <>.

    This software is copyright (c) 2011-2014 by Toby Inkster.

    This is free software; you can redistribute it and/or modify it under the
    same terms as the Perl 5 programming language system itself.