Skip to content


Subversion checkout URL

You can clone with
Download ZIP
Fetching contributors…
Cannot retrieve contributors at this time
290 lines (195 sloc) 8.14 KB
XML::SAX::Base is intended for use as a base class for SAX filter modules
and XML parsers generating SAX events.
If you simply wish to build a SAX handler class to 'consume' SAX events you
do not need to use XML::SAX::Base directly although you will need to install
This module used to be distributed as part of the XML:SAX distribution but
is now distributed separately and referenced as a dependency by XML::SAX.
via tarball:
% tar -zxvf XML-SAX-Base-xxx.tar.gz
% perl Makefile.PL
% make && make test
% make install
via CPAN shell:
% perl -MCPAN -e shell
% install XML::SAX::Base
The rest of this file consists of the fine XML::SAX::Base documentation
contributed by Robin Berjon.
=head1 NAME
XML::SAX::Base - Base class SAX Drivers and Filters
package MyFilter;
use XML::SAX::Base;
@ISA = ('XML::SAX::Base');
This module has a very simple task - to be a base class for PerlSAX
drivers and filters. It's default behaviour is to pass the input directly
to the output unchanged. It can be useful to use this module as a base class
so you don't have to, for example, implement the characters() callback.
The main advantages that it provides are easy dispatching of events the right
way (ie it takes care for you of checking that the handler has implemented
that method, or has defined an AUTOLOAD), and the guarantee that filters
will pass along events that they aren't implementing to handlers downstream
that might nevertheless be interested in them.
The Perl Sax API Reference is at L<>.
Writing SAX Filters is tremendously easy: all you need to do is
inherit from this module, and define the events you want to handle. A
more detailed explanation can be found at
Writing Drivers is equally simple. The one thing you need to pay
attention to is B<NOT> to call events yourself (this applies to Filters
as well). For instance:
package MyFilter;
use base qw(XML::SAX::Base);
sub start_element {
my $self = shift;
my $data = shift;
# do something
$self->{Handler}->start_element($data); # BAD
The above example works well as precisely that: an example. But it has
several faults: 1) it doesn't test to see whether the handler defines
start_element. Perhaps it doesn't want to see that event, in which
case you shouldn't throw it (otherwise it'll die). 2) it doesn't check
ContentHandler and then Handler (ie it doesn't look to see that the
user hasn't requested events on a specific handler, and if not on the
default one), 3) if it did check all that, not only would the code be
cumbersome (see this module's source to get an idea) but it would also
probably have to check for a DocumentHandler (in case this were SAX1)
and for AUTOLOADs potentially defined in all these packages. As you can
tell, that would be fairly painful. Instead of going through that,
simply remember to use code similar to the following instead:
package MyFilter;
use base qw(XML::SAX::Base);
sub start_element {
my $self = shift;
my $data = shift;
# do something to filter
$self->SUPER::start_element($data); # GOOD (and easy) !
This way, once you've done your job you hand the ball back to
XML::SAX::Base and it takes care of all those problems for you!
Note that the above example doesn't apply to filters only, drivers
will benefit from the exact same feature.
=head1 METHODS
A number of methods are defined within this class for the purpose of
inheritance. Some probably don't need to be overridden (eg parse_file)
but some clearly should be (eg parse). Options for these methods are
described in the PerlSAX2 specification available from
=over 4
=item * parse
The parse method is the main entry point to parsing documents. Internally
the parse method will detect what type of "thing" you are parsing, and
call the appropriate method in your implementation class. Here is the
mapping table of what is in the Source options (see the Perl SAX 2.0
specification for the meaning of these values):
Source Contains parse() calls
=============== =============
CharacterStream (*) _parse_characterstream($stream, $options)
ByteStream _parse_bytestream($stream, $options)
String _parse_string($string, $options)
SystemId _parse_systemid($string, $options)
However note that these methods may not be sensible if your driver class
is not for parsing XML. An example might be a DBI driver that generates
XML/SAX from a database table. If that is the case, you likely want to
write your own parse() method.
Also note that the Source may contain both a PublicId entry, and an
Encoding entry. To get at these, examine $options->{Source} as passed
to your method.
(*) A CharacterStream is a filehandle that does not need any encoding
translation done on it. This is implemented as a regular filehandle
and only works under Perl 5.7.2 or higher using PerlIO. To get a single
character, or number of characters from it, use the perl core read()
function. To get a single byte from it (or number of bytes), you can
use sysread(). The encoding of the stream should be in the Encoding
entry for the Source.
=item * parse_file, parse_uri, parse_string
These are all convenience variations on parse(), and in fact simply
set up the options before calling it. You probably don't need to
override these.
=item * get_options
This is a convenience method to get options in SAX2 style, or more
generically either as hashes or as hashrefs (it returns a hashref).
You will probably want to use this method in your own implementations
of parse() and of new().
=item * get_feature, set_feature
These simply get and set features, and throw the
appropriate exceptions defined in the specification if need be.
If your subclass defines features not defined in this one,
then you should override these methods in such a way that they check for
your features first, and then call the base class's methods
for features not defined by your class. An example would be:
sub get_feature {
my $self = shift;
my $feat = shift;
if (exists $MY_FEATURES{$feat}) {
# handle the feature in various ways
else {
return $self->SUPER::get_feature($feat);
Currently this part is unimplemented.
It would be rather useless to describe all the methods that this
module implements here. They are all the methods supported in SAX1 and
SAX2. In case your memory is a little short, here is a list. The
apparent duplicates are there so that both versions of SAX can be
=over 4
=item * start_document
=item * end_document
=item * start_element
=item * start_document
=item * end_document
=item * start_element
=item * end_element
=item * characters
=item * processing_instruction
=item * ignorable_whitespace
=item * set_document_locator
=item * start_prefix_mapping
=item * end_prefix_mapping
=item * skipped_entity
=item * start_cdata
=item * end_cdata
=item * comment
=item * entity_reference
=item * notation_decl
=item * unparsed_entity_decl
=item * element_decl
=item * attlist_decl
=item * doctype_decl
=item * xml_decl
=item * entity_decl
=item * attribute_decl
=item * internal_entity_decl
=item * external_entity_decl
=item * resolve_entity
=item * start_dtd
=item * end_dtd
=item * start_entity
=item * end_entity
=item * warning
=item * error
=item * fatal_error
=head1 TODO
- more tests
- conform to the "SAX Filters" and "Java and DOM compatibility"
sections of the SAX2 document.
=head1 AUTHOR
Kip Hampton ( did most of the work, after porting
it from XML::Filter::Base.
Robin Berjon ( pitched in with patches to make it
usable as a base for drivers as well as filters, along with other patches.
Matt Sergeant ( wrote the original XML::Filter::Base,
and patched a few things here and there, and imported it into
the XML::SAX distribution.
=head1 SEE ALSO
Jump to Line
Something went wrong with that request. Please try again.