NAME Role::Markup::XML - Moo(se) role for bolt-on lazy XML markup VERSION Version 0.02 SYNOPSIS package My::MarkupEnabled; use Moo; # or Moose, or something compatible with 'Role::Markup::XML'; # ...and this of course # write some other code... sub something_useful { my $self = shift; # put your XML-generating data structure here my %spec = ( -name => 'my:foo', # element name -content => { -name => 'my:bar' }, # element content hurr => 'durr', # attribute 'my:derp' => 'lulz', # namespaced attribute 'xmlns:my' => 'urn:x-bogus:foo', # namespaces go inline ); # create a document object to hang on to my $doc = $self->_DOC; # returns the last node generated, which we can chain my $stub = $self->_XML( doc => $doc, spec => \%spec, ); my @contents = ( # imagine a bunch of things in here ); # since these nodes will be appended to $stub, we aren't # interested in the output this time $self->_XML( parent => $stub, # owner document is derived spec => \@contents, # also accepts ARRAY refs args => $self->cb_args, # some useful state data ); # the rest of the ops are ordinary XML::LibXML return $doc->toString(1); } DESCRIPTION This is indeed yet another module for lazy XML markup generation. It exists because it is different: * It converses primarily in reusable, inspectable, and most importantly, *inert* Perl data structures, * It also ingests existing XML::LibXML nodes, * It enables you to generate markup *incrementally*, rather than all at once, * It Does the Right Thing™ around a bunch of otherwise tedious boilerplate operations, such as namespaces, XHTML, or flattening token lists in attributes, * It has a callback infrastructure to help you create modular templates, or otherwise override behaviour you don't like, * It is implemented as a Role, to be more conducive to modern Perl development. I began by using XML::LibXML::LazyBuilder. It is pretty good, definitely preferable to typing out reams of XML::LibXML DOM-like API any time I wanted to make some (guaranteed well-formed) XML. I even submitted a patch to it to make it better. Nevertheless, I have reservations about the general approach to terse markup-generating libraries, in particular about the profligate use of anonymous subroutines. (You also see this in lxml.etree for Python, Builder::XmlMarkup for Ruby, etc.) The main issue is that these languages aren't Lisp: it costs something at runtime to gin up a stack of nested anonymous subroutines, run them once, and then immediately throw them away. It likewise costs in legibility to have to write a bunch of imperative code to do what is essentially data declaration. It also costs in sanity to have to write function-generating-function-generating functions just to get the mess under control. The interim product is impossible to inspect or manipulate. This ostensibly time-saving pattern quickly hits a wall in both development, and at runtime. The answer? Use (in this case) Perl's elementary data structures to convey the requisite information: data structures which can be built up from bits and pieces, referenced multiple times, sliced, diced, spliced, frozen, thawed, inspected, and otherwise operated on by ordinary Perl routines. Provide mix-and-match capability with vanilla XML::LibXML, callbacks, and make the whole thing an unobtrusive mix-in that you can bolt onto your existing code. METHODS Methods in this module are named such as to stay out of the way of *your* module's interface. _DOC [$VERSION,] [$ENCODING] Generate a document node. _ELEM $TAG [, $DOC, \%NSMAP ] Generate a single XML element. Generates a new document unless $DOC is specified. Defaults to XHTML if no namespace map is provided. _XML $SPEC [, $PARENT, $DOC, $ARGS | @ARGS ] | %PARAMS Generate an XML tree according to the "specification format". Returns the *last node generated* by the process. Parameters are as follows: spec The node specification. Strictly speaking this is optional, but there isn't much of a point of running this method if there is no spec to run it over. doc The XML::LibXML::Document object intended to own the contents. Optional, however it is often desirable to supply a document object along with the initial call to this method. parent The XML::LibXML::Element object which is intended to be the parent node of the spec. Optional. args An "ARRAY" reference of arguments to be passed into "CODE" references embedded in the spec. Optional. Specification Format The building blocks of the spec are, unsurprisingly, "HASH" and "ARRAY" references. The former correspond to elements and other things, while the latter correspond to lists thereof. Literals are literals, and blessed objects will be treated like strings, so it helps if they have a string override. "CODE" references may be used just about anywhere, and will be dereferenced recursively using the supplied "args" until there is nothing left to dereference. It is up to you to keep these data structures free of cycles. Elements Special keys designate the name and content of an element spec. These are, unimaginitively, "-name" and "-content". They work like so: { -name => 'body', -content => 'hurr' } # produces hurr Note that "-content" can take any primitive: literal, "HASH", "ARRAY" or "CODE" reference, XML::LibXML::Node object, etc. Attributes Any key is not "-name" or "-content" will be interpreted as an attribute. { -name => 'body', -content => 'hurr', class => 'lolwut' } # produces hurr When references are values of attributes, they are flattened into strings: { -name => 'body', -content => 'hurr', class => [qw(one two three)] } # produces hurr Namespaces If there is a colon in either the "-name" key value or any of the attribute keys, the processor will expect a namespace that corresponds to that prefix. These are specified exactly as one would with ordinary XML, with the use of an "xmlns:foo" attribute>. (Prefix-free "xmlns" attributes likewise work as expected.) { -name => 'svg', xmlns => 'http://www.w3.org/2000/svg', 'xmlns:xlink' => 'http://www.w3.org/1999/xlink', -content => [ { -name => 'a', 'xlink:href' => 'http://some.host/' }, ], } # produces: # # # Other Nodes "-pi" Processing instructions are designated by the special key "-pi" and accept arbitrary pseudo-attributes: { -pi => 'xml-stylesheet', type => 'text/xsl', href => '/my.xsl' } # produces "-doctype" Document type declarations are designated by the special key "-doctype" and accept values for the keys "public" and "system": { -doctype => 'html' } # produces "-comment" Comments are designated by the special key "-comment" and whatever is in the value of that key: { -comment => 'hey you guyyyys' } # produces Callbacks Just about any part of a markup spec can be replaced by a "CODE" reference, which can return any single value, including another "CODE" reference. These are called in the context of $self, i.e., as if they were a method of the object that does the role. The "args" in the original method call form the subsequent input: sub callback { my ($self, @args) = @_; my %node = (-name => 'section', id => $self->generate_id); # ...do things to %node, presumably involving @args... return \%node; } sub make_xml { my $self = shift; my $doc = $self->_DOC; $self->_XML( doc => $doc, spec => { -name => 'p', -content => \&callback }, ); return $doc; } "CODE" references can appear in attribute values as well. _XHTML | %PARAMS Generate an XHTML+RDFa stub. Return the "" and the document when called in list context, otherwise return just the "" in scalar context (which can be used in subsequent calls to "_XML"). my ($body, $doc) = $self->_XHTML(%p); # or my $body = $self->_XHTML(%p); Parameters uri The "href" attribute of the "" element. ns A mapping of namespace prefixes to URIs, which by default will appear as *both* XML namespaces *and* the "prefix" attribute. prefix Also a mapping of prefixes to URIs. If this is set rather than "ns", then the XML namespaces will *not* be set. Conversely, if this parameter is defined but false, then *only* the contents of "ns" will appear in the conventional "xmlns:foo" way. title This can either be a literal title string, or "CODE" reference, or "HASH" reference assumed to encompass the whole "" element, or an "ARRAY" reference where the first element is the title and subsequent elements are predicates. link This can either be an "ARRAY" reference of ordinary markup specs, or a "HASH" reference where the keys are the "rel" attribute and the values are one or more (via "ARRAY" ref) URIs. In the latter form the following behaviour holds: * Predicates are grouped by "href", folded, and sorted alphabetically. * "<link>" elements are sorted first lexically by the sorted "rel", then by sorted "rev", then by "href". * A special empty "" hash key can be used to pass in another similar structure whose keys represent "rev", or reverse predicates. * A special "-about" key can be used to specify another "HASH" reference where the keys are subjects and the values are similar structures to the one described. { # ordinary links 'rel:prop' => [qw(urn:x-target:1 urn:x-target:2)], # special case for reverse links '' => { 'rev:prop' => 'urn:x-demo-subject:id' }, # special case for alternate subject -about => { 'urn:x-demo-subject:id' => { 'some:property' => 'urn:x-target' } }, } The "ARRAY" reference form is passed along as-is. meta Behaves similarly to the "link" parameter, with the following exceptions: * No "" or "-about" pseudo-keys, as they are meaningless for literals. * Literal values can be expressed as an "ARRAY" reference of the form "[$val, $lang, $type]" with either the second or third element "undef". They may also be represented as a "HASH" reference where the keys are the language (denoted by a leading "@") or datatype (everything else), and the values are the literal values. { 'prop:id' => ['foo', [2.3, undef, 'xsd:decimal']], 'exotic' => { '@en' => ['yo dawg', 'derp'] } } head This is an optional "ARRAY" reference of "<head>" elements that are neither "<link>" nor "<meta>" (or, if you want, additional unmolested "<link>" and "<meta>" elements). attr These attributes (including "-content") will be passed into the "<body>" element. content This parameter enables us to isolate the "<body>" content without additional attributes. Note that setting this parameter will cause the method to return the innermost, last node that is specified, rather than the "<body>". transform This is the URI of a (e.g. XSLT) transform which will be included in a processing instruction if supplied. args Same as "args" in "_XML". AUTHOR Dorian Taylor, "<dorian at cpan.org>" BUGS Please report any bugs or feature requests to "bug-role-markup-xml at rt.cpan.org", or through the web interface at <http://rt.cpan.org/NoAuth/ReportBug.html?Queue=Role-Markup-XML>. I will be notified, and then you'll automatically be notified of progress on your bug as I make changes. SUPPORT You can find documentation for this module with the perldoc command. perldoc Role::Markup::XML You can also look for information at: * RT: CPAN's request tracker (report bugs here) <http://rt.cpan.org/NoAuth/Bugs.html?Dist=Role-Markup-XML> * AnnoCPAN: Annotated CPAN documentation <http://annocpan.org/dist/Role-Markup-XML> * CPAN Ratings <http://cpanratings.perl.org/d/Role-Markup-XML> * Search CPAN <http://search.cpan.org/dist/Role-Markup-XML/> SEE ALSO * XML::LibXML::LazyBuilder * XML::LibXML * Moo LICENSE AND COPYRIGHT Copyright 2016 Dorian Taylor. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. You may obtain a copy of the License at <http://www.apache.org/licenses/LICENSE-2.0>. Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License.