/* html2txt1.l
   flex file for converting html to txt (sort of...)
   Daniel Scharstein 2/18/08
   1st version - get rid of tags, compress whitespace, translate symbols
*/

ws       [ \t\n]|" "
tag      "<"[^>]*">" 

%%
{tag}     {}
{ws}+     {printf(" ");}
"&gt;"    {printf(">");}
"&lt;"    {printf("<");}
"&amp;"   {printf("&");}