#!/bin/sed -f

# xsltproc --html doesnt understand html5
s:\(</*\)\(section\|header\|nav\|footer\|article\):\1div:g

/data-href/s/&/&amp;/g

# Fix incorrect < and > inside p tags, that is by allowing only
# known tag be surrounded by < and >.
ta
:a
s/<p>\(.*\)<\/p>/\1/
tfix
b

:fix
s/</\&lt;/g
s/>/\&gt;/g

s/&lt;span\([^;]*\)&gt;/<span\1>/g
s/&lt;\/span&gt;/<\/span>/g

s/\&lt;em\&gt;/<em>/g
s/\&lt;\/em\&gt;/<\/em>/g

s:.*:<p>&</p>: