XML::Filter::Reindent - Reformats whitespace for pretty printing XML
use XML::Handler::Composer;
use XML::Filter::Reindent;
my $composer = new XML::Handler::Composer (%OPTIONS);
my $indent = new XML::Filter::Reindent (Handler => $composer, %OPTIONS);
XML::Filter::Reindent is a sub class of the XML::Filter::DetectWS manpage.
XML::Filter::Reindent can be used as a PerlSAX filter to reformat an
XML document before sending it to a PerlSAX handler that prints it
(like the XML::Handler::Composer manpage.)
Like the XML::Filter::DetectWS manpage, it detects ignorable whitespace and blocks of
whitespace characters in certain places. It uses this information and
information supplied by the user to determine where whitespace may be
modified, deleted or inserted.
Based on the indent settings, it then modifies, inserts and deletes characters
and ignorable_whitespace events accordingly.
This is just a first stab at the implementation.
It may be buggy and may change completely!
This method determines whether children of a certain element
may be reformatted.
The default implementation checks the PreserveWS parameter of the specified
start_element event and returns 0 if it is set or MAYBE otherwise.
The value MAYBE (2) indicates that further investigation is needed, e.g.
by examining the element contents. A value of 1 means yes, indent the
child nodes, no further investigation is needed.
NOTE: the PreserveWS parameter is set by the parent class,
the XML::Filter::DetectWS manpage, when the element or one of its ancestors has
the attribute xml:space=``preserve''.
Override this method to tweak the behavior of this class.
This method determines whether a certain element may be re-indented.
The default implementation returns the value of the $parent_says_indent
parameter, which was set to the value returned by indent_children for the
parent element. In other words, the element will be re-indented if the
parent element allows it.
Override this method to tweak the behavior of this class.
I'm not sure how useful this hook is. Please provide feedback!
The current implementation puts all incoming Perl SAX events in a queue for
further processing. When determining which nodes should be re-indented,
it sometimes needs information from previous events, hence the use of the
queue.
The parameter (Compress => 1) is added to
matching start_element and end_element events with no events in between
This indicates to an XML printer that a compressed notation can be used,
e.g <foo/>.
If an element allows reformatting of its contents (xml:space=``preserve'' was
not active and indent_children returned MAYBE), the element
contents will be reformatted unless it only has one child node and that
child is a regular text node (characters event.)
In that case, the element will be printed as <foo>text contents</foo>.
If you want element nodes with just one text child to be reindented as well,
simply override indent_children to return 1 instead of MAYBE (2.)
This behavior may be changed or extended in the future.
This code is highly experimental!
It has not been tested well and the API may change.
The code that detects blocks of whitespace at potential indent positions
may need some work.
Send bug reports, hints, tips, suggestions to Enno Derksen at
<enno@att.com>.
