hmmmm
On 13 Mar 2002, (Jason E. Stewart) wrote:
> "John Utz" <[EMAIL PROTECTED]> writes:
>
> > I think that i have created the correct actual_cast and reinterpret_class,
> > but i just realized that this doesnt get propogated to the methods,
> >
> > i see there is some code that has some 'transmogrification' regexp work
> > and does this work on the lsit @dom_node_methods.
> >
> > do i need to do something with this? do i need to create a
> > @grammar_methods? what goes into this list and what doesnt go into this
> > list, or am i barking up the wrong tree?
>
> You're barking pretty loud. Chances are it's the wrong tree ;-)
really? i am working from the example of the DOM stuff, and the methods
that are listed in the @dom_node_methods all end up in Xerces.pm with
the following line added:
# automatically convert to base class
$result = $result->actual_cast();
> At about line 609 in postModule.pl, there's the begining of a HERE
> doc:
>
> my $extra = <<'EXTRA';
>
> which goes until almost the end of the file at line 1234. The last few
> lines of the script print out $extra to the TEMP filehandle, and then
> the temp file gets renamed to Xerces.pm.
>
> The moral of this story is that anything that you insert between the
> start of the HERE doc and the end will get printed out verbatim into
> Xerces.pm.
>
> The only thing you have to watch out for is the 'package'
> declarations, like:
>
> package XML::Xerces::DOM_Node;
>
> every method defined after that line will be in the
> XML::Xerces::DOM_Node package, until a new package declaration occurs.
>
> So if you want stuff to appear in the XML::Xerces::Grammar package,
> you'll need to stick it right before one of my package defs like so:
>
> package XML::Xerces::Grammar;
>
> # insert you methods here
>
> package XML::Xerces::DOM_Node;
>
> # define methods for DOM_Node...
>
> Also, for actual_cast() you don't need the following:
>
> sub actual_cast {
> return undef unless _isa( ref($_[0]), 'XML::Xerces::DOM_Node' );
> return $_[0] if $_[0]->isNull;
>
> This should work:
>
> sub actual_cast {
> my ($self) = @_;
> return undef unless $self->isa('XML::Xerces::Grammar');
>
> The isNull() stuff is only for DOM_Node subclasses, and I have no idea
> why the _isa() method exists. As far as I can tell, it is just a very
> inefficient reimplementation of UNIVERSAL::isa(). If Harmon or
> Frederick Paul are listening maybe they can enlighten me.
well, i thing that i actually had all those parts sorted out yesterday,
before i sent the mail:
package XML::Xerces::Grammar;
sub _isa {
return 0 unless defined $_[0];
my %package_hash = ($_[0] => 1);
my @package_array = ($_[0]);
my $base = $_[1];
while(my $class = shift @package_array) {
return 1 if "$class" eq "$base";
foreach my $inherit (eval "\@$class\::ISA;") {
unless ($package_hash{$inherit}) {
$package_hash{$inherit} = 1;
push @package_array, $inherit;
}
}
}
return 0;
};
sub _reinterpret_cast {
return undef unless ref $_[1];
bless $_[1], $_[0];
my($raw_ref, $tied) = ($_[1] =~ m[=(.*)\(]);
if ($raw_ref eq 'SCALAR') { $tied = tied ${$_[1]}; }
elsif ($raw_ref eq 'ARRAY') { $tied = tied @{$_[1]}; }
elsif ($raw_ref eq 'HASH') { $tied = tied %{$_[1]}; }
return $_[1] unless $tied;
bless $tied, $_[0];
return $_[1];
};
sub actual_cast {
return undef unless _isa( ref($_[0]), 'XML::Xerces::Grammar' );
return $_[0] if $_[0]->isNull;
my $grammar_type = $_[0]->getGrammarType;
return _reinterpret_cast('XML::Xerces::SchemaGrammar', $_[0])
if $grammar_type == $XML::Xerces::SchemaGrammar;
return _reinterpret_cast('XML::Xerces::Grammar', $_[0])
if $grammar_type == $XML::Xerces::Grammar;
return $_[0];
}
but i dont think this is sufficient, because it appears to me that the
call to actual_cast has to be inserted in the methods in the pm file.
thats what i assume this code is doing:
# I?DOM_Node: automatically convert to base class
if (grep {/$CURR_CLASS/} @dom_node_methods) {
if (my ($sub) = /^sub\s+([\w_]+)/) {
$sub = "$ {CURR_CLASS}::$sub";
if (grep {/$sub$/} @dom_node_methods) {
my $fix = <<'EOT';
# automatically convert to base class
$result = $result->actual_cast();
EOT
fix_method(\*FILE,
\*TEMP,
qr/return undef/,
$fix,
1);
next;
}
}
}
so, am i still wrong? i am content to be wrong. :-)
10 days ago i knew zip about SWIG, and this was probably not the project i
would have picked for getting my tootsies wet :-)
but the help you've provided has more than made up for the steepness of
the learning curve...
> HTH,
> jas.
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: [EMAIL PROTECTED]
> For additional commands, e-mail: [EMAIL PROTECTED]
>
>
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
