From ee59534aac4b07a54cf56fc0b2383c3b27185128 Mon Sep 17 00:00:00 2001
From: Trygve Laugstøl <trygvis@inamo.no>
Date: Fri, 6 Jul 2012 23:17:37 +0200
Subject: o Trying to include more stuff.

---
 update-feed.sh | 39 +++++++++++++++++++++++++++++++++++++--
 1 file changed, 37 insertions(+), 2 deletions(-)

(limited to 'update-feed.sh')

diff --git a/update-feed.sh b/update-feed.sh
index 4afd168..122db80 100755
--- a/update-feed.sh
+++ b/update-feed.sh
@@ -1,7 +1,6 @@
 #!/bin/bash
 
 set -e
-set -x
 
 checksum() {
   [ -r $1 ] && md5sum $1 | cut -f 1 -d ' '
@@ -22,4 +21,40 @@ fetch() {
 
 fetch
 
-cat tv.nrk.no.html | xmlstarlet tr ./tv.nrk.no.xsl > tv.nrk.no.atom
+mkdir -p tmp
+
+cat tv.nrk.no.html | \
+  xmlstarlet sel -N strings=http://exslt.org/strings -t -m '//a[@class="listobject-link"]' \
+  -v '@href' -v '" "' \
+  -v 'strings:encode-uri(normalize-space(.), true())' -n \
+  | while read url name
+do
+  if [ "$name" == "" -o -r "$name" ]
+  then
+    continue
+  fi
+
+  echo Url:  $url
+  echo Name: $name
+  if [ -r "tmp/$name" ]
+  then
+    z="tmp/$name"
+  else
+    z="Jan 1 1970"
+  fi
+
+  curl -v -s -z "$z" -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"
+  echo ret=$?
+  set -e
+  cat "tmp/$name.html" | xmllint --format --xmlout - >"tmp/${name}.xml"
+done
+
+cat tv.nrk.no.html | xmlstarlet tr ./tv.nrk.no.xsl | tee tv.nrk.no.atom
+#rm -rf tmp
-- 
cgit v1.2.3