diff --git a/scrape.rb b/scrape.rb index 847d0fc..95e9d67 100644 --- a/scrape.rb +++ b/scrape.rb @@ -17,16 +17,19 @@ doc = Nokogiri.parse(open(url).read) meeting = '' items = [] +items_this_meeting = 0 doc.at("#bodytext").children.each do |node| if node.inner_text.match(/\d{1,2}\s+\w+\s+\d{4}/) # eg 10 December 2012 meeting = node.inner_text.strip + items_this_meeting = 0 end node.children.each do |subnode| if subnode.name == 'a' && subnode['href'].match(/\.mp3$/i) + items_this_meeting += 1 items << { - :d => Time.parse(meeting), + :d => Time.parse(meeting) + ((items_this_meeting - 1) * 30 * 60), :href => subnode['href'].strip, :title => subnode.inner_text.strip }