Recipe 20.3 Extracting URLs

20.3.1 Problem

You want to extract all URLs from an HTML file. For example, you have downloaded a page that lists the MP3 files downloadable from some site. You want to extract those MP3s' URLS so you can filter the list and write a program to download the ones you want.

20.3.2 Solution

Use the HTML::LinkExtor module from CPAN:

use HTML::LinkExtor; $parser = HTML::LinkExtor->new(undef, $base_url); $parser->parse_file($filename); @links = $parser->links; foreach $linkarray (@links) {     my @element = @$linkarray;     my $elt_type = shift @element;                  # element type     # possibly test whether this is an element we're interested in     while (@element) {         # extract the next attribute and its value         my ($attr_name, $attr_value) = splice(@element, 0, 2);         # ... do something with them ...     } }

20.3.3 Discussion

You can use HTML::LinkExtor in two different ways: either by calling links to get a list of all links in the document once it is completely parsed, or by passing a code reference in the first argument to new. The referenced function is called on each link as the document is parsed.

The links method clears the link list, so call it only once per parsed document. It returns a reference to an array of elements. Each element is itself an array reference with an HTML::Element object at the front followed by a list of attribute name and attribute value pairs. For instance, the HTML:

<A HREF="http://www.perl.com/">Home page</A> <IMG src="/books/2/106/1/html/2/images/big.gif" LOWsrc="/books/2/106/1/html/2/images/big-lowres.gif">

would return a data structure like this:

[   [ a,   href   => "http://www.perl.com/" ],   [ img, src    => "images/big.gif",          lowsrc => "images/big-lowres.gif" ] ]

Here's an example of how to use $elt_type and $attr_name to print out and anchor an image:

if ($elt_type eq 'a' && $attr_name eq 'href') {     print "ANCHOR: $attr_value\n"          if $attr_value->scheme =~ /http|ftp/; } if ($elt_type eq 'img' && $attr_name eq 'src') {     print "IMAGE:  $attr_value\n"; }

To extract links only to MP3 files, you'd say:

foreach my $linkarray (@links) {      my ($elt_type, %attrs) = @$linkarray;      if ($elt_type eq 'a' && $attrs{'href'} =~ /\.mp3$/i) {        # do something with $attr{'href'}, the URL of the mp3 file      }  }

Example 20-2 is a complete program that takes as its arguments a URL, such as file:///tmp/testing.html or http://www.ora.com/, and produces on standard output an alphabetically sorted list of unique URLs linked from that site.

Example 20-2. xurl

  #!/usr/bin/perl -w   # xurl - extract unique, sorted list of links from URL   use HTML::LinkExtor;   use LWP::Simple;      $base_url = shift;   $parser = HTML::LinkExtor->new(undef, $base_url);   $parser->parse(get($base_url))->eof;   @links = $parser->links;   foreach $linkarray (@links) {       my @element  = @$linkarray;       my $elt_type = shift @element;       while (@element) {           my ($attr_name , $attr_value) = splice(@element, 0, 2);           $seen{$attr_value}++;       }   }   for (sort keys %seen) { print $_, "\n" }

This program does have a limitation: if the get of $base_url involves a redirection, links resolve using the original URL instead of the URL after the redirection. To fix this, fetch the document with LWP::UserAgent and examine the response code to find out whether a redirection occurred. Once you know the post-redirection URL (if any), construct the HTML::LinkExtor object accordingly.

Here's an example of the run:

% xurl http://www.perl.com/CPAN ftp://[e-mail]/CPAN/CPAN.html http://language.perl.com/misc/CPAN.cgi http://language.perl.com/misc/cpan_module http://language.perl.com/misc/getcpan http://www.perl.com/index.html http://www.perl.com/gifs/lcb.xbm

In mail or Usenet messages, you may see URLs written as:

<URL:http://www.perl.com>

This is supposed to make it easy to pick URLs from messages:

@URLs = ($message =~ /<URL:(.*?)>/g);

20.3.4 See Also

The documentation for the CPAN modules LWP::Simple, HTML::LinkExtor, and HTML::Entities; Recipe 20.1