Use a proper HTML handling tool. XML::LibXML can handle HTML if it follows the standard decently. I usually use its wrapper XML::XSH2, in which you can extract the text simply as

open :F html input.html ;
for //p echo .//text() ;
[download]