if the file structure like the one you included you can extract the content by doing grep -v "^<" x
or more conveniant approach cat x|sed -e 's/<[^>]*>//g'|grep -v '^$'
or to do it the csplit way based on the comments below
you can do it lik this
cat doc.xml | egrep -v '<?xml version="1.0" \?>|<docs>|</docs>' | csplit -q -z - '/<doc/' '{*}' --prefix=out-