Htgrep is a CGI script written in Perl that allows you to query any document accessible to your HTTP server on a paragraph-by-paragraph basis. Htgrep is a generic front-end to htgrep.pl, a Perl package that makes it relatively easy to implement simple search engines for the WWW. Htgrep allows you to pass all parameters to the search package as part of the URL, so a forms interface can be used to set the parameters. Usually, however, you will want to write your own CGI script by adapting htgrep so that most of the parameters are hard-wired.
Htgrep is no longer supported. A zipped distribution is still available, however.
The predecessor to htgrep, called parscan, was used to implement w3catalog, one of the first search engines for the web, in September 1993.