I wouldn't start any project with XML::Parser, it's a bit antique, and XML::LibXML is more feature rich and much faster parser to start any project with.

Parsing RSS is a real pain, as it's often not well formed, so anything using a proper XML parser will die. XML::RSS and XML::RSS::Tools get round this by having a pre-filter in them that cleans up well know bad code, before attempting to pass the file onto the XML parser.

The XML::RSS::Tools module (which I wrote) uses XML::RSS for parsing RSS, one of several HTTP clients for getting RSS feeds, and the XML::LibXSLT module for converting the feed into something else.

Some useful nodes:

• Converting RSS file to HTML
• XML::RSS
• Using RSS
• How do I clean RSS feeds to make them usable?
• Parsing badly formed RSS or XML
• Parsing XML At All Costs

rss feed grabber

Hi DK. I'm trying to figure out what your objective here is. Are you trying to learn how XML::Parser works? Or are you trying to do something with RSS? I mean, if its the latter then I would do it like this:

#!/usr/bin/perl -w
use strict;
use XML::Simple;
use LWP::Simple;
use Data::Dump::Streamer;

$|++;
my $ticker=['http://perlmonks.org/index.pl?node_id=30175&xmlstyle=rss'
+,
            "http://rss.news.yahoo.com/rss/science"]->[rand 2];
print "Getting RSS from $ticker\n";
my $feed = get($ticker);
print "Parsing RSS...\n";
my $ref = XMLin($feed);
print "Dumping Parse Tree...\n";
Dump $ref;
[download]

If its the former then I can't really help much beyond pointing out that what you are doing with the lexical var "$feed" in there scares the willies out of me.

And in today's "making Perl work a lot harder than you need to do", let's nominate the following entry:

my $ticker=['http://perlmonks.org/index.pl?node_id=30175&xmlstyle=rss'
+,
            "http://rss.news.yahoo.com/rss/science"]->[rand 2];
[download]

So, we've asked Perl to construct an array, take a reference to it, then dereference that reference to pick out one of the items, then discard the reference, which then garbage-collects the array. All when we could have written that this way:

my $ticker=('http://perlmonks.org/index.pl?node_id=30175&xmlstyle=rss'
+,
            "http://rss.news.yahoo.com/rss/science")[rand 2];
[download]

saving two characters of typing, and all that mess of creating the new array and reference and garbage collecting. We're simply constructing a list, then picking out an element of that list with a literal slice (a construct I suggested for Perl 3, by the way {grin}).

To optimize this further, I'd go with a qw for that first list:

my $ticker=(qw(http://perlmonks.org/index.pl?node_id=30175&xmlstyle=rs
+s
            http://rss.news.yahoo.com/rss/science))[rand 2];
[download]

And in recent versions of Perl, you can even drop that outer set of parens:

my $ticker=qw(http://perlmonks.org/index.pl?node_id=30175&xmlstyle=rss
            http://rss.news.yahoo.com/rss/science)[rand 2];
[download]

I saw bracket-arrow-bracket as a "cute syntax" once. I'm trying to stomp it out, because there's an equivalent construct (as I showed) that is a lot less work for Perl. Please don't propogate "cute syntax" that is more expensive.

-- Randal L. Schwartz, Perl hacker
Be sure to read my standard disclaimer if this is a reply.

perl -MBenchmark=cmpthese -e'cmpthese -1, { cute => sub { [0,1]->[rand
+ 2] }, list => sub { qw(0 1)[rand 2] } }'
[download]

          Rate cute list
cute  768000/s   -- -74%
list 2899719/s 278%   --
[download]

Yes, the list slice is much faster. But for something that probably runs only once in every 5 minutes, isn't 768000 per second fast enough? Optimizing seems premature here. For things like this, I am against choosing a particular language or syntax for its speed.

I'm not saying that your reply is useless. It's important to know what code does and this information will certainly help some of the readers when they do have to optimize. But the code is written now and not much is gained by changing it, so I'd just let it be. Programmer time is still much more expensive than computer time.

Juerd # { site => 'juerd.nl', plp_site => 'plp.juerd.nl', do_not_use => 'spamtrap' }

Yep, guilty as charged. But lighten up a little. It was just a snippet to make things a little more interesting.



---
demerphq


First they ignore you, then they laugh at you, then they fight you, then you win.
-- Gandhi

Flux8

at least you have your disclaimer to guard against anyone thinking that you're an ass...

The XML is not properly formed - probably a missing closing tag of an element. It would probably help if you could post the source of your RSS.

/J\

bobf asked me to expand on my scary comment in my original reply. Here goes.

#!/usr/bin/perl -w

use strict;
use LWP::Simple 'get';
use XML::RAI;

my $rss = XML::RAI->parse(get(shift||die"please enter rss uri"));
my $title = $rss->channel->title;
my $link  = $rss->channel->link;

print "$title\n$link\n\n";

for my $item (@{$rss->items}) {
  $title = $item->title;
  $link  = $item->link;
  print "$title\n$link\n";
}
[download]

Check out XML::Feed if you haven't. I've been really happy with it.

...
use WWW::Mechanize;
my $url = shift; # any .xml RSS feed url

my $mech = WWW::Mechanize->new();
$mech->get( $url );
my @content = split /\n/, $mech->content;
my $title_pattern = "<title>(.*?)</title>";
my $description_pattern = "<description>(.*?)</description>";
my @titletags = grep s/$title_pattern/$1/i, @content;
my @descriptiontags = grep s/$description_pattern/$1/i, @content;

my $thetitle=$titletags[0];
if ( $thetitle !~ s/<\!\[CDATA\[//g ) {}
if ( $thetitle !~ s/Librivox\://g ) {}
if ( $thetitle !~ s/]]>//g ) {}
print "$thetitle\n";

my $thedescription=$descriptiontags[0];
if ($thedescription !~ s/<\!\[CDATA\[//g ) {}
if ($thedescription !~ s/]]>//g ) {}
print "$thedescription\n";
[download]