#!/usr/bin/sed

#let us know we've been translated
s/<[Hh][Tt][Mm][Ll]>/[translated from html with h2a]/

#do away with tags </P> </DIV> </HTML> </A> & any more <HTML>
s:</[Pp].*>::g
s:</[Dd][Ii][Vv].*>::g
s:</[Hh][Tt][Mm][Ll].*>::g
s:</[Aa].*>::g
s:<[Hh][Tt][Mm][Ll].*>::g

#substitute nl for <BR> <DIV>
s/<[Bb][Rr].*>/\
/g
s/<[Dd][Ii][Vv].*>/\
/g

#substitue nl nl for <P>
s/<[Pp].*>/\
\
/g

#<TD>, etc
#

#<hr>
s/<[Hh][Rr].*>/\
----------------------------------------------------------\
/g

#substitute "[image]" for images
s/<[Ii][Mm][Gg].*>/[image:&]/g
#should actually be <IMG SRC="..." *> becomes [image ...]

#display hrefs, mailtos in parens
s/<[Aa] [Hh][Rr][Ee][Ff].*>/(link:&)/g
#right now, the & will disappear when s/<.*>//g is executed

#<B> <U> <I> <EM>
s:<[Bb]>:*:g
s:</[Bb]>:*:g
s:<[Ii]>:*:g
s:</[Ii]>:*:g
s:<[Ee][Mm]>:*:g
s:</[Ee][Mm]>:*:g
s:<[Uu]>:_:g
s:</[Uu]>:_:g

#&gt; etc
s/&gt;/>/g
s/&lt;/</g
s/&nbsp;/ /g
#s/&eacute;/e/g
#s/&egrave;/e/g
#s/&uuml;/u/g

#remove all other html tags
#s/<.*>//g

#limit line width to 70 char
#