| 1 |
sashby |
1.1 |
#
|
| 2 |
|
|
# Added XML support: SFA 26/07/05
|
| 3 |
|
|
#
|
| 4 |
|
|
# SimpleXMLURLDoc.pm. - Extends SimpleDoc with URL download functionality
|
| 5 |
|
|
#
|
| 6 |
|
|
# Originally Written by Christopher Williams
|
| 7 |
|
|
#
|
| 8 |
|
|
# Description
|
| 9 |
|
|
# -----------
|
| 10 |
|
|
#
|
| 11 |
|
|
# Interface
|
| 12 |
|
|
# ---------
|
| 13 |
|
|
# new(URLcache[,DocVersionTag] : A new SimpleXMLURLDoc object. You can also
|
| 14 |
|
|
# specify an alternative doc version tag
|
| 15 |
|
|
# addbasetags(parse) : Add Base Tags to the given parse
|
| 16 |
|
|
# urlget(urlstring[,location]) : get the given url - using the cache.
|
| 17 |
|
|
# Returns (url, filename)
|
| 18 |
|
|
# urldownload(urlstring[,location]) : get the given url ignoring any cached
|
| 19 |
|
|
# version. Returns (url, filename)
|
| 20 |
|
|
# expandurl(urlstring) : return a URLclass object of the given url expanded
|
| 21 |
|
|
# according to the base settings
|
| 22 |
|
|
# cache([cache]) : get/set the current URL cache
|
| 23 |
|
|
# doctype() : return the (type,version) of the document
|
| 24 |
|
|
# as specified by the DocVersionTag
|
| 25 |
|
|
package ActiveDoc::SimpleXMLURLDoc;
|
| 26 |
|
|
use ActiveDoc::SimpleXMLDoc;
|
| 27 |
|
|
use URL::URLhandler;
|
| 28 |
|
|
require 5.001;
|
| 29 |
|
|
@ISA=qw(ActiveDoc::SimpleXMLDoc);
|
| 30 |
|
|
|
| 31 |
|
|
sub new {
|
| 32 |
|
|
my $class=shift;
|
| 33 |
|
|
my $self={};
|
| 34 |
|
|
bless $self, $class;
|
| 35 |
|
|
$self->cache(shift);
|
| 36 |
|
|
$self->_initdoc("doc",@_);
|
| 37 |
|
|
return $self;
|
| 38 |
|
|
}
|
| 39 |
|
|
|
| 40 |
|
|
sub addbasetags
|
| 41 |
|
|
{
|
| 42 |
|
|
my $self=shift;
|
| 43 |
|
|
my $parse=shift;
|
| 44 |
|
|
|
| 45 |
|
|
$self->registerTag($parse, "base",
|
| 46 |
|
|
\&basetaghandler,
|
| 47 |
|
|
[ "url" ],
|
| 48 |
|
|
1);
|
| 49 |
|
|
}
|
| 50 |
|
|
|
| 51 |
|
|
sub cache {
|
| 52 |
|
|
my $self=shift;
|
| 53 |
|
|
if ( @_ ) {
|
| 54 |
|
|
$self->{cache}=shift;
|
| 55 |
|
|
$self->{urlhandler}=URL::URLhandler->new($self->{cache});
|
| 56 |
|
|
}
|
| 57 |
|
|
return $self->{cache};
|
| 58 |
|
|
}
|
| 59 |
|
|
|
| 60 |
|
|
sub expandurl {
|
| 61 |
|
|
my $self=shift;
|
| 62 |
|
|
my $urlstring=shift;
|
| 63 |
|
|
|
| 64 |
|
|
return $self->{urlhandler}->expandurl($urlstring);
|
| 65 |
|
|
}
|
| 66 |
|
|
|
| 67 |
|
|
sub urldownload {
|
| 68 |
|
|
my $self=shift;
|
| 69 |
|
|
my $urlstring=shift;
|
| 70 |
|
|
|
| 71 |
|
|
($fullurl,$filename)=$self->{urlhandler}->download($urlstring, @_);
|
| 72 |
|
|
if ( ( ! defined $filename ) || ( $filename eq "" ) ) {
|
| 73 |
|
|
$self->parseerror("Failed to get $fullurl");
|
| 74 |
|
|
}
|
| 75 |
|
|
return ($fullurl,$filename);
|
| 76 |
|
|
}
|
| 77 |
|
|
|
| 78 |
|
|
sub urlget {
|
| 79 |
|
|
my $self=shift;
|
| 80 |
|
|
my $urlstring=shift;
|
| 81 |
|
|
|
| 82 |
|
|
($fullurl,$filename)=$self->{urlhandler}->get($urlstring, @_);
|
| 83 |
|
|
|
| 84 |
|
|
if ( ( ! defined $filename ) || ( $filename eq "" ) ) {
|
| 85 |
|
|
$self->parseerror("Failed to get $fullurl");
|
| 86 |
|
|
}
|
| 87 |
|
|
return ($fullurl,$filename);
|
| 88 |
|
|
}
|
| 89 |
|
|
|
| 90 |
|
|
# ------------------------ Support Routines ---------------------------
|
| 91 |
|
|
|
| 92 |
|
|
# ------------------------ Tag Routines -------------------------------
|
| 93 |
|
|
sub basetaghandler()
|
| 94 |
|
|
{
|
| 95 |
|
|
my ($name, $hashref, $nesting)=@_;
|
| 96 |
|
|
# No action for Char handler:
|
| 97 |
|
|
return if ($nesting == 2);
|
| 98 |
|
|
|
| 99 |
|
|
# Closing tag:
|
| 100 |
|
|
if ($nesting == 1)
|
| 101 |
|
|
{
|
| 102 |
|
|
# Probably not needed as error handling (tags not closed etc.) is
|
| 103 |
|
|
# handled by the XML parser anyway:
|
| 104 |
|
|
if ( $#{$self->{basestack}} >= 0 )
|
| 105 |
|
|
{
|
| 106 |
|
|
my $type=pop @{$self->{basestack}};
|
| 107 |
|
|
$self->{urlhandler}->unsetbase($type);
|
| 108 |
|
|
}
|
| 109 |
|
|
}
|
| 110 |
|
|
else
|
| 111 |
|
|
{
|
| 112 |
|
|
my $url=$self->{urlhandler}->setbase($$hashref{'url'});
|
| 113 |
|
|
# Add store for url of the file currently being parsed. This info can
|
| 114 |
|
|
# then be extracted in Requirements objects
|
| 115 |
|
|
$self->{configurl}=$url;
|
| 116 |
|
|
push @{$self->{basestack}}, $url->type();
|
| 117 |
|
|
}
|
| 118 |
|
|
}
|
