#!/bin/bash

set -e

checksum() {
  [ -r $1 ] && md5sum $1 | cut -f 1 -d ' '
}

fetch() {
  curl -s -o tv.nrk.no.html.new http://tv.nrk.no/listobjects/recentlysent

  if [ -r tv.nrk.no.html -a "`checksum tv.nrk.no.html.new`" == "`checksum tv.nrk.no.html`" ]
  then
    rm -f tv.nrk.no.html.new
    echo "Already up to date."
    exit 0
  fi

  dos2unix -q tv.nrk.no.html.new
  xmllint --html --dropdtd --xmlout tv.nrk.no.html.new 2> /dev/null > tv.nrk.no.html.new2
  mv tv.nrk.no.html.new2 tv.nrk.no.html.new
  mv tv.nrk.no.html.new tv.nrk.no.html
}

fetch

mkdir -p tmp

cat tv.nrk.no.html | \
  xmlstarlet sel \
  -N strings=http://exslt.org/strings \
  -N fn=http://www.w3.org/2005/xpath-functions \
  -t -m '//a[@class="listobject-link"]' \
  -v '@href' -v '" "' \
  -v 'strings:replace(substring-after(normalize-space(@href), "http://tv.nrk.no/"), "/", "_")' -n \
  | while read url name
do
  if [ "$name" == "" -o -r "tmp/$name" ]
  then
    continue
  fi

  echo Url:  $url
  echo Name: $name

  curl -s -L "$url" -o "tmp/$name"
  rm -f "tmp/$name.html" "tmp/$name.xml"
  set +e
  cat "tmp/$name" | dos2unix | \
    tidy -utf8 -asxhtml -quiet -f /dev/null \
    --new-inline-tags "time" \
    --new-blocklevel-tags "article, hgroup, section, header, footer, mark, aside" \
    > "tmp/$name.html"
  set -e
  cat "tmp/$name.html" | xmllint --format --xmlout - >"tmp/${name}.xml"
done

cat tv.nrk.no.html | xmlstarlet tr ./tv.nrk.no.xsl > tv.nrk.no.atom