#!/usr/bin/perl -w sub download_detail { my ($dl) = @_; open(PIN, "wget -q \"http://careo.ucalgary.ca/$dl\" -O - |") || die "Unable to open wget command: $!\n"; my $detail_html=""; while () { $detail_html .= $_; } close(PIN); return $detail_html } sub download_lom_html { my ($wosid,$object,$detail_html) = @_; my @links = ($detail_html =~ m//sg); my @lom_links = grep { $_ =~ m/1\.1\.1\.1\.1/ } @links; foreach my $ll (@lom_links) { open(PIN, "wget -q \"http://careo.ucalgary.ca/$ll\" -O - |") || die "Unable to open wget command: $!\n"; my $lom_html=""; while () { $lom_html .= $_; } close(PIN); my ($lom_xml) = ($lom_html =~ m/(<lom.*\/lom>)/s); $lom_xml =~ s/<//g; $lom_xml =~ s/&/&/g; $lom_xml =~ s/"/"/g; print "Saving lom as $object.xml\n"; open(HOUT,">$object.xml") || die "Unable to open file out for writing: $!\n"; print HOUT $lom_xml; close(HOUT); } } sub browse_category { my ($cat_url) = @_; open(PIN, "wget -q \"$cat_url\" -O - |") || die "Unable to open wget command: $!\n"; my $file=""; while () { $file .= $_; } close(PIN); my @links = ($file =~ m//sg); my @detail_links = grep { $_ =~ m/details\?/ } @links; foreach my $dl (@detail_links) { $dl =~ s/&/&/g; my ($wosid,$object) = ($dl =~ m/wosid=(\w+).*object=(\w+)$/); print "$wosid, $object\n"; ## print "$dl\n"; my $detail_html = download_detail($dl); download_lom_html($wosid,$object,$detail_html); } } sub browse_arts { my $arts_url = "http://careo.ucalgary.ca/cgi-bin/WebObjects/CAREO.woa/1/wo/tUkzqf9mt6KIbfa9Dmbd0g/40.0.6.7.5.1.7"; browse_category($arts_url); } sub browse_science { my $science_url = "http://careo.ucalgary.ca/cgi-bin/WebObjects/CAREO.woa/1/wo/BeWs1oNbUyFqlcTa4wWTuM/8.0.5.3.1.1.1.0.2"; browse_category($science_url); } browse_arts(); browse_science(); #http://careo.ucalgary.ca/cgi-bin/WebObjects/CAREO.woa/1/wo/Nvlr48HkBbktBsDlEKDZWM/34.0.6.1.3.0.1.1.1.1.1.0 #http://careo.ucalgary.ca/cgi-bin/WebObjects/CAREO.woa/1/wo/Nvlr48HkBbktBsDlEKDZWM/34.0.6.1.3.0.1.1.1.1.1.0 #http://careo.ucalgary.ca/cgi-bin/WebObjects/CAREO.woa/1/wa/details?theme=careo&wosid=tUkzqf9mt6KIbfa9Dmbd0g&object=651479 #http://careo.ucalgary.ca/cgi-bin/WebObjects/CAREO.woa/1/wa/discuss?theme=careo&wosid=tUkzqf9mt6KIbfa9Dmbd0g&object=651479