delos.uoi.gr scraper

delos-scraper.pl

#!/usr/bin/perl
# Last tested: 2017-01-22
use strict;
use warnings;

my $lesson_id = $ARGV[0];

for (my $page_id = 1; ; ++$page_id) {
    my $url = "http://delos.uoi.gr/opendelos/search?crs=$lesson_id&sa=$page_id";
    my $page = qx{curl --silent "$url"};
    my $had_matches = 0;

    while ($page =~ /[?]rid=([[:xdigit:]]{8})" class="lecture-title">\s*<strong>\s*([^<]+)[^)]+[)]\s*<[\/]span>\s*<br>\s*<[\/]div>\s*<[\/]div>\s*([^\n\r]+)/g) {
        my $url = "http://delos.uoi.gr/delosrc/resources/vl/$1/$1.mp4";
        my $title = $2;
        my $description = $3;
        $description =~ s/(^\s+|\s+$)//g;
        $title =~ s/(^\s+|\s+$)//g;

        print "$url\n";
        print "$title ($description)\n\n";

        $had_matches = 1;
    }

    if (!$had_matches) {
        last;
    }
}